www.thelancet.com/neurology Vol 11 May 2012 453

Review

Lancet Neurol 2012; 11: 453–66

Sanford-Burnham Medical Research Institute, La Jolla, CA, USA (Prof H H Freeze PhD, B G Ng BS); Section of Experimental Paediatrics, Department of Clinical Sciences, Lund University, Lund, Sweden (E A Eklund MD); and Division of Child and Adolescent Neurology, Mayo Clinic, Rochester, MN, USA (Prof M C Patterson MD)

Correspondence to:Dr Hudson H Freeze, Sanford-Burnham Medical Research Institute, 10901 North Torrey Pines Road, La Jolla, CA 92037, USAhudson@sanfordburnham.org

Neurology of inherited glycosylation disorders Hudson H Freeze, Erik A Eklund, Bobby G Ng, Marc C Patterson

Congenital disorders of glycosylation comprise most of the nearly 70 genetic disorders known to be caused by impaired synthesis of glycoconjugates. The eff ects are expressed in most organ systems, and most involve the nervous system. Typical manifestations include structural abnormalities (eg, rapidly progressive cerebellar atrophy), myopathies (including congenital muscular dystrophies and limb-girdle dystrophies), strokes and stroke-like episodes, epileptic seizures, developmental delay, and demyelinating neuropathy. Patients can also have neurological symptoms associated with coagulopathies, immune dysfunction with or without infections, and cardiac, renal, or hepatic failure, which are common features of glycosylation disorders. The diagnosis of congenital disorder of glycosylation should be considered for any patient with multisystem disease and in those with more specifi c phenotypic features. Measurement of concentrations of selected glyco conjugates can be used to screen for many of these disorders, and molecular diagnosis is becoming more widely available in clinical practice. Disease-modifying treatments are available for only a few disorders, but all aff ected individuals benefi t from early diagnosis and aggressive management.

IntroductionA hypotonic child presents with seizures, hypoglycaemia, mild liver fi brosis and high transaminase concentrations. Another child with intrauterine growth retardation and dysmorphic features, including a long, thin face with large protruding ears and a micropenis, presents with delayed speech and motor development. A third child is referred with retinitis pigmentosa and stroke-like episodes in the context of heart failure secondary to cardiomyopathy. An active athletic adult aged 25 years has recently developed peripheral neuropathy and progressive foot drop. These patients have very diff erent phenotypes, but they all have inherited defects in glycosylation—the process of adding complex sugar chains to proteins and lipids. Nearly 70 genetic disorders of glycosylation have been discovered, mostly within the past 15 years, and the catalogue continues to grow. A simple biochemical test can confi rm the general diagnosis in most cases, although a few disorders require more invasive procedures, and all require defi nitive genetic confi rmation. We present an overview of these diseases, with an emphasis on phenotypes, diagnostic approaches, and treatment.

Common features of glycosylation pathwaysThe glycome—all the sugar chain structures in a cell or organism—is orders of magnitude larger than the expressed genome. Its daunting complexity claims 1–2% of the genome to encode the known glycosylation machinery.1 Sugar chains (glycans) are added to mam-malian proteins and lipids through eight distinct pathways (table 1, fi gure 1). Each pathway requires a diff erent enzyme or transferase complex to initiate glycosylation. The fi rst sugar unit (monosaccharide) linked to the protein or lipid defi nes the pathway, to which a single sugar or a preformed sugar chain might be added (fi gure 1). All pathways require activated monosaccharides in the form of nucleotide sugars, which are delivered to correct locations in the endoplasmic reticulum (ER) or Golgi apparatus to enable glycan biosynthesis. Because pathway precursors

are shared, low concentrations or ineffi cient delivery could aff ect several pathways, although such eff ects have generally been studied in only one pathway at a time. Eff ects on multiple pathways have been reported in some instances and might be more common than is appreciated at present. Most of the eff ects of defects in the early steps of glycosylation are highly pathway specifi c, whereas those later in the process can aff ect multiple pathways. One protein can carry multiple glycans from diff erent pathways. The outcome is not pathway driven, and many factors determine the spectrum of glycan structures. Examples include protein structure, availability of donor substrates, and the amounts of diff erent sugar transferases and their kinetic constants. The eff ects of these factors on a particular glycan can exclude or enhance subsequent extension, or can place proteins in an environment where other transferases compete for a single glycan. Transferases are transcriptionally regu lated, but their localisation and effi ciency of recycling through the dynamic ER-Golgi network is crucial.2 The eff ect of faulty glycosylation on the function of any individual protein is unpredictable and ranges from trivial to essential. Eff ects must, therefore, be determined for each protein and for each function.3

The most recent nomenclature for glycosylation disorders proposes using the gene name followed by CDG to denote a congenital disorder of glycosylation.4 While this system is not the only one in use, we fi nd it useful and convenient, and we use it in this Review. Where relevant we also provide other common or traditional names, such as CDG-I or CDG-II.

Specifi c glycosylation pathwaysGenetic defects are known to occur in seven of the eight major ER-Golgi network glycan-generating pathways (table 1). The best known, and by far the most studied, is the N-linked glycosylation pathway, especially in terms of defects located in the ER (fi gure 1). Protein O-linked glycosylation is more diverse than N-linked glycosylation. Serine or threonine residues are linked to glycans

454 www.thelancet.com/neurology Vol 11 May 2012

Review

through N-acetylgalactosamine, xylose, mannose or fucose (fi gure 1). Chains are extended by specifi c glycosyltransferases, whereas terminal sugars are in many cases added by non-specifi c transferases that service multiple pathways.

N-linked glycosylationN-acetylglucosamine is bound to asparagine on nascent proteins in the ER lumen, but it is not added as a

single sugar. Rather, a universal 14-sugar precursor containing two N-acetylglucosamine, nine mannose, and three glucose units is assembled on a lipid carrier (dolichol) to form a lipid-linked oligosaccharide.5 The entire glycan is transferred to asparagine by the multisubunit oligo-saccharyltransferase complex.6,7 The glucose units and up to six of the mannose units are often removed after transfer, and N-acetylglucosamine, galactose, sialic acid, and fucose are added on to multiple branches. The order of remodelling

Number of disorders

Typical clients Functions

N-linked 38 Nearly all cell-surface receptors, ECM, and secreted proteins

Assistance in folding, stabilisation of target proteins, signalling

O-linked

O-xylose 10 ECM proteins, heparan and chondroitin sulphate Growth-factor binding, structure of ECM

O-mannose 6 α-Dystroglycan Bridging neuromuscular 1α-receptor junction

O-fucose 2 Notch and Notch ligands Notch signalling, developmental patterning

O-GalNAc 2 Mucins, leucocyte receptors Pathogen decoys, lubrication, protection of cell surface

O-glucose 0 Notch and Notch ligands Notch signalling, developmental patterning

GPI-anchor glycans 5 Proteins in lipid rafts, signalling molecules Organisation of plasma-membrane domains

Glycosphingolipid 1 Brain (highest concentration), synapses Membrane organisation, signalling

Multiple 13 Soluble and membrane-bound molecules to provide substrates or Golgi-ER homoeostasis

Traffi cking of Golgi resident proteins, synthesis of precursors

ECM=extracellular matrix. GalNAc=N-acetylgalactosamine. GPI=glycophosphatidylinositol. ER=endoplasmic reticulum.

Table 1: Overview of glycan types by pathway

Figure 1: Pathways of glycosylation in the endoplasmic reticulum-Golgi network of mammalian cellsThe main types of glycosylation are shown. Various representative sugar chain structures are given as examples. The grey shaded areas denote common core regions. Most of the glycosylation disorders that aff ect the nervous system involve alterations in N-linked and O-mannosylated glycoproteins. Some glycophosphatidylinositol-anchor and glycosphingolipid disorders also involve these alterations, but most proteoglycan and O-GalNAc defects do not. Other types of glycosylation exist, such as cytoplasmic O-GlcNAc and C-mannosylation, but are not shown. Glc=glucose. Gal=galactose. Man=mannose. GlcNAc=N-acetylglucosamine. GalNAc=N-acetylgalactosamine. GlcA=glucuronic acid. IdoA=Iduronic acid. Fuc=fucose. Xyl=xylose. Sia=sialic acid. S=sulphation. P=phosphorylation. Ac=acetylation. S/T=serine or threonine. Adapted from Stanley and colleagues,5 by permission of the Consortium of Glycobiology Editors, La Jolla, CA, USA.

S/T

Cytoplasm

Etn

NH2

P

InositolAc

S

S

S SS

OS/T

NAsn

NAsn

P

SS

S

Heparan sulphate

Chondroitin/dermatan sulphate4S 4S 4SP S

Ser-O-

6S 6S6S

NS NSNS

6S-O-Ser

4S

3S

P

Glycosphingolipids

Glycosylphosphatidylinositolanchor

P

S/T

S/T

O-linked chains

Glycoproteins

N-linked chains

2S

ProteoglycansGlcGalManGlcNAcGalNAcGlcAIdoAFucXylSia

www.thelancet.com/neurology Vol 11 May 2012 455

Review

of the chains is prescribed, but, as with all glycosylation, it is not template driven.

Essentially, all proteins (except albumin) that travel through the ER-Golgi network undergo N-linked glycosylation. Glycans promote protein folding, stability, traffi cking, localisation, and oligomerisation.5 They play vital parts in cell–cell interactions and intracellular signalling.8

The dolichol carrier lipid also carries mannose, which serves three other pathways: glycophosphatidylinositol anchors, O-mannosylation, and C-mannosylation. The fi rst two are described below. No C-mannosylation disorders have been reported.9,10

O-linked glycosylationO-linked protein glycosylation involves initial linkage between serine or threonine residues and mannose, xylose, N-acetylgalactosamine, fucose, or glucose (fi gure 1). The O-linked α-mannose glycans contain N-acetyl glucosamine, galactose, N-acetylgalactosamine, and sialic acid.11,12 α-Dystroglycan, which has a crucial role as a link between the extracellular matrix and the cytoskeleton in skeletal muscle cells, is the major identifi ed carrier for O-linked α-mannose glycans. Other proteins certainly carry them, but are yet to be identifi ed. Defects in this pathway often cause neurological defi cits.13

O-linked α-N-acetylgalactosamine-based glycans link to serine or threonine, after which one of four sugars is added to form a disaccharide. Sequential addition of galactose, N-acetylglucosamine, fucose, and sialic acid generates linear or multibranched chains14 that are found on secreted or cell-surface mucins of epithelial cells. These chains can lubricate and are eff ective pathogen decoys. None causes neurological pathology.

O-linked β-xylose glycans on serine generate glycosaminoglycans, such as heparan sulphate, heparin, chondroitin, and dermatan sulphates.15 Long repeating disaccharides of glucuronic acid-N-acetylgalactosamine (chondroitin and dermatan sulphates) or glucuronic acid-N-acetylglucosamine (heparin and heparan sul phate) are extended from a small core glycan.13 Some glucuronic acid is epimerised to iduronic acid. Sulphation occurs on de-N-acetylated N-acetylglucosamine NH2 groups or OH groups. Proteins carrying chondroitin are used for physical integrity and cushioning. Cell-surface heparin-sulphate chains bind growth factors (eg, fi broblast growth factors), cytokines, and morphogens during development to establish gradients of these molecules.16,17

O-linked α-fucose-based glycans occur in selected epidermal growth factor modules in Notch and Notch ligands, and are extended by Fringe family N-acetylglucosamine transferases and other glycosyl-transferases.18,19 The presence of these glycans has a strong eff ect on Notch signalling. Thrombospondin type I repeats can be O-fucosylated by a diff erent transferase, and fucose is extended with one or two glucose units.20

Lipid-bearing glycansGlycosphingolipids link glucose to ceramide. If galactose is also added, lactosylceramide is formed. This core can be variably extended to more complex glyco-sphingolipids, including the sialylated gangliosides.21 The greatest diversities in types and concentrations of glycosphingolipids occur in the brain and peripheral nervous system. Glyco sphingolipids bind to each other or to proteins, such as integrins, which enables them to aff ect signalling.22 Glyco phosphatidylinositol-anchor glycans substitute for transmembrane regions of many proteins. They contain mannose and glucosamine, and are assembled in the ER on a phosphatidylinositol backbone. The entire glycolipid is transferred to C-terminal regions of proteins.23 Glycophospha-tidylinositol anchors also have roles in membrane diff usion, intracellular protein sorting, and signalling.24,25 Defects in glyco sphingolipid and glycopho-sphatidylinositol pathways can cause neurological complications.26–29

Traffi cking and homoeostasisClient proteins travel through the dynamic ER-Golgi network, where glycosylation occurs. Genetic defects in proteins needed to recycle the glycosylation machinery between the ER and the Golgi apparatus can aff ect the process.2 Most of the known genes encode soluble cyto-plasmic proteins that transiently associate with the Golgi apparatus and help to guide vesicles containing the glycosylation machinery to their location. Most patients with these defects have neurological defi cits that, along with skeletal abnormalities and dysmorphic features, are probably due to their eff ects on multiple glycosylation pathways.30

Selected specifi c defectsSpace limitations prevent description of all 66 glyco-sylation disorders in this Review. In tables 2–4 we list the glycosylation disorders with neurological mani festations and their most common symptoms. A list of disorders without or with minor neurological compli cations is available online (appendix). Many of the listed congenital disorders of glycosylation, such as defi ciencies in sulphation, primarily aff ect glycosaminoglycan struc ture, and their classifi cation as glycosylation disorders is the subject of debate. In this section we highlight disorders that have been documented in at least ten patients.

PMM2-CDG (CDG-Ia)PMM2-CDG is the most frequent N-linked congenital disorder of glycosylation and accounts for around 80% of all diagnosed cases.31 Mutations in PMM2, which encodes the phosphomannomutase 2 enzyme that catalyses conversion of mannose-6-phosphate to mannose-1-phosphate, cause the disorder. The CNS and peripheral nervous system are prime targets, but multisystem abnormalities cause substantial residual

See Online for appendix

456 www.thelancet.com/neurology Vol 11 May 2012

Review

neurological defi cits and 20% of mortality in infancy. Milder phenotypes have been recognised.

Clinical presentationsThe classic presentation in infancy includes failure to thrive, inverted nipples, subcutaneous fat pads and other forms of lipodystrophy (fi gure 2). Aff ected children are hypotonic, developmentally delayed, and show evolving facial hypotonia and esotropia, which might progress to complete sixth-nerve paralysis.32–34 Roving eye movements are typical, and delayed or absent fi xation might be present. Overt pigmentary retinopathy might be seen in adolescent and adult patients, but is generally not

apparent in children. The subsequent evolution of this disease, which was initially described by Hagberg and colleagues,35 has four stages.

In the fi rst stage, during infancy, systemic symptoms dominate. These include susceptibility to infection, episodes of hepatic impairment or overt failure, episodes of bleeding or thrombosis, and hypertrophic cardio-myopathy, with or without pericardial eff usion.36

The second phase occurs in early childhood and is dominated by seizures and stroke-like episodes. These symptoms might be seen in the context of intercurrent illness and are sometimes misdiagnosed as febrile seizures, with or without Todd’s paralysis.34 Studies

Enzymatic defect Gene OMIM code Number of known patients

Neurological abnormalities

ID Brain Seizures Ocular

PMM2-CDG (CDG-Ia)* Phosphomannomutase conversion of Man-6-phosphate to Man-1-phosphate

PMM2 212065 >600 Yes Yes Yes Yes

ALG6-CDG (CDG-Ic)* α-1,3-glucosyltransferase ALG6 603147 57 Yes Yes Yes Yes

ALG3-CDG (CDG-Id)* α-1,3-mannosyltransferase ALG3 601110 ~10 Yes Yes Yes Yes

DPM1-CDG (CDG-Ie)* Dolichol-P-Man synthase DPM1 603503 14 Yes Yes Yes Yes

MPDU1-CDG (CDG-If)* Dolichol-P sugar utilisation protein MPDU1 609180 7 Yes Yes Yes Yes

ALG12-CDG (CDG-Ig)* α-1,6-mannosyltransferase ALG12 607143 11 Yes Yes Yes Yes

ALG8-CDG (CDG-Ih)* α-1,3-glucosyltransferase ALG8 608104 8 Yes Yes Yes Yes

ALG2-CDG (CDG-Ii)* α-1,3-mannosyltransferase ALG2 607906 1 Yes Yes Yes Yes

DPAGT1-CDG (CDG-Ij)* GlcNAc-1-P transferase DPAGT1 608093 11 Yes Yes Yes Yes

ALG1-CDG (CDG-Ik)* β-1,4-mannosyltransferase ALG1 608540 14 Yes Yes Yes Yes

ALG9-CDG (CDG-IL)* α-1,2-mannosyltransferase ALG9 608776 3 Yes Yes Yes Yes

DOLK-CDG (CDG-Im)* Dolichol kinase DOLK 610768 16 Yes Yes Yes Yes

RFT1-CDG (CDG-In)* Man5GlcNAc2 fl ippase RFT1 612015 6 Yes Yes Yes Yes

TUSC3-CDG (MRT7)* Subunit of the OST complex TUSC3 611093 4 Yes NR NR NR

MAGT1-CDG (MRX95)* Subunit of the OST complex MAGT1 300716 14 Yes NR NR NR

ALG11-CDG (CDG-Ip)* α-1,2-mannosyltransferase ALG11 613661 1 Yes Yes Yes NR

SRD5A3-CDG (CDG-Iq)* 5-α steroid reductase (polyprenol reductase) SRD5A3 612379 14 Yes Yes Yes Yes

DDOST-CDG (CDG-Ir)* Subunit of the OST complex DDOST 614507 1 Yes NR Yes NR

MGAT2-CDG (CDG-IIa)* N-acetylglucosaminyltransferase II MGAT2 212066 3 Yes NR NR NR

GCS1-CDG (CDG-IIb)* α-1,2-glucosidase GCS1 606056 1 NR NR Yes NR

SLC35C1-CDG (CDG-IIc)* GDP-fucose transporter FUCTI 266265 7 Yes Yes Yes NR

B4GALT1-CDG (CDG-IId)* β-1,4-galactosyltransferase B4GALTI 607091 2 Yes Yes Yes NR

COG1-CDG (CDG-IIg)* Conserved oligomeric Golgi subunit 1 COG1 611209 3 Yes Yes Yes NR

COG4-CDG (CDG-IIj)* Conserved oligomeric Golgi subunit 4 COG4 613489 2 Yes Yes Yes NR

COG5-CDG (CDG-Iii)* Conserved oligomeric Golgi subunit 5 COG5 613612 4 Yes Yes Yes NR

COG6-CDG (CDG-IIL)* Conserved oligomeric Golgi subunit 6 COG6 ·· 1 Yes Yes Yes NR

COG7-CDG (CDG-IIe)* Conserved oligomeric Golgi subunit 7 COG7 608779 9 Yes Yes Yes NR

COG8-CDG (CDG-IIh)* Conserved oligomeric Golgi subunit 8 COG8 611182 2 Yes Yes Yes NR

ATP6V0A2-CDG Golgi pH regulator ATP6V0A2 219200 >20 Yes Yes Yes NR

OMIM=Online Mendelian Inheritance in Man database. ID=intellectual development is impaired. Brain=brain malformations, such as cerebellar atrophy or hypoplasia, or thinning of the corpus callosum. Ocular=ocular abnormalities, such as cataracts, optic-nerve atrophy, or coloboma. Man=mannose. GlcNAc=N-acetylglucosamine. OST=oligosaccharyltransferase. NR=not reported. *Previous classifi cation provided in brackets.

Table 2: Neurological features of glycosylation disorders detectable by measurement of transferrin glycosylation

www.thelancet.com/neurology Vol 11 May 2012 457

Review

Facial dys-morph isms

Inverted nipples or lipodys-trophy

Skeletal defects

Failure to thrive

Liver dys function/raised ALT

Renal path ology

Ichthyosis or other skin disorder

Coagu-lopathy

Recurrent infection or low immuno-globulin

Cardio-myop athy

Deaf ness Endo crin-opathy or micropenis

Hypo gly-caemia

PLE or intes tinal dys func tion

PMM2-CDG (CDG-Ia)*

++ ++ ++ ++ ++ ++ NR ++ (+) (+) (+) ++ (+) (+)

ALG6-CDG (CDG-Ic)*

++ ++ + + ++ NR NR ++ NR (+) NR + NR (+)

ALG3-CDG (CDG-Id)*

++ (+) (+) ++ + NR NR ++ (+) NR NR (+) (+) +

DPM1-CDG (CDG-Ie)*

++ (+) + ++ ++ NR NR ++ (+) NR NR NR NR NR

MPDU1-CDG (CDG-If)*

++ NR NR (+) NR (+) ++ + NR NR NR NR NR (+)

ALG12-CDG (CDG-Ig)*

++ ++ + ++ ++ NR NR ++ ++ (+) (+) ++ ++ NR

ALG8-CDG (CDG-Ih)*

++ NR NR (+) ++ + NR ++ NR NR NR + NR ++

ALG2-CDG (CDG-Ii)*

(+) NR NR NR (+) NR NR (+) NR NR NR NR NR NR

DPAGT1-CDG (CDG-Ij)*

++ (+) (+) NR (+) NR NR NR NR NR NR NR NR NR

ALG1-CDG (CDG-Ik)*

++ NR NR + NR NR NR (+) (+) (+) NR (+) NR NR

ALG9-CDG (CDG-IL)*

++ ++ NR ++ ++ (+) NR (+) NR NR NR NR NR (+)

DOLK-CDG (CDG-Im)*

(+) NR NR + (+) NR ++ NR NR ++ NR NR (+) NR

RFT1-CDG (CDG-In)*

++ ++ NR ++ (+) NR NR ++ NR NR ++ NR NR (+)

ALG11-CDG (CDG-Ip)*

(+) ++ NR (+) NR NR NR (+) NR NR (+) (+) NR NR

SRD5A3-CDG (CDG-Iq)*

++ (+) NR + ++ NR ++ ++ NR NR NR (+) NR NR

DDOST-CDG (CDG-Ir)*

(+) NR (+) (+) (+) NR NR (+) NR NR NR NR NR NR

MGAT2-CDG (CDG-IIa)*

++ NR ++ (+) NR NR NR ++ (+) NR NR ++ NR ++

GCS1-CDG (+) NR (+) (+) (+) NR NR (+) (+) NR NR NR NR NR

SLC35C1-CDG ++ NR ++ NR NR NR NR NR ++ NR NR NR NR NR

B4GALT1-CDG (CDG-IId)*

NR NR NR NR (+) NR NR (+) NR NR NR (+) NR NR

COG1-CDG (CDG-IIg)*

++ NR ++ (+) (+) (+) NR NR (+) NR NR NR NR NR

COG4-CDG (CDG-IIj)*

(+) NR NR (+) (+) NR NR (+) (+) NR NR NR NR NR

COG5-CDG (CDG-Iii)*

NR NR NR NR NR NR NR NR NR NR NR NR NR NR

COG6-CDG (CDG-IIL)*

NR NR NR NR (+) NR NR (+) NR NR NR NR NR NR

COG7-CDG (CDG-IIe)*

++ (+) ++ ++ ++ (+) ++ (+) + NR NR NR NR +

COG8-CDG (CDG-IIh)*

(+) NR (+) (+) (+) NR NR NR NR NR NR NR NR (+)

ATP6V0A2-CDG

++ NR NR NR NR NR ++ NR NR NR NR NR NR NR

Non-neurological features are presented as a diagnostic aid. ++=very common fi nding (≥50 % of patients). +=common fi nding (<50% of patients). (+)=described but uncommon fi nding (<3 known patients). ALT=alanine aminotransferase. PLE=protein-losing enteropathy. NR=not reported. *Previous classifi cations provided in brackets.

Table 3: Non-neurological and muscular features of glycosylation disorders with neurological involvement detectable by measurement of transferrin glycosylation

458 www.thelancet.com/neurology Vol 11 May 2012

Review

suggest that the stroke-like episodes represent ischaemic stroke, but can also be mimicked by periods of otherwise subclinical focal status epilepticus or localised oedema after a seizure.37 The distinction cannot readily be made clinically, and MRI is required. Coagulation studies and MR angiography might also be useful. Seizures can be partial or generalised and persist beyond the stroke-like episode. They can be readily controlled in most cases.

The third stage is characterised by slowly progressive limb atrophy (refl ecting demyelinating peripheral neuropathy) combined with severe ataxia and intellectual defi ciency. Esotropia evolves into bilateral sixth-nerve palsies and visual loss associated with progressive pigmentary degeneration of the retina. In a series of 23 patients aged 10 months to 20 years, visual maturation was delayed in most, and at the last follow-up visit 16 had low vision and fi ve were legally blind and showed pallor of the optic discs; 18 had myopia.38

The fi nal stage occurs in adolescence and adulthood, when patients have fi xed neurological defi cits, including intellectual disability, severe cerebellar ataxia, peripheral neuropathy with distal wasting, and arefl exia. Patients who survive to adulthood commonly have pigmentary retinopathy, kyphoscoliosis, and endocrin opathies.39 Some patients have hypergonadotropic hypogonadism, and skeletal deformities have also been described.

Screening has been used increasingly and has identifi ed milder phenotypes. Two infants with absent or mild lipodystrophy were reported to have only subtle biochemical alterations and one of these children showed loss of purposeful hand movements and stereotypies similar to those reported in Rett’s disorder, plus intention tremor.40 Patients with mild phenotypes who survive to adulthood might have borderline cognitive impairment, with or without strabismus. In one report, all three aff ected adults were in full-time employment.41 One individual has been described who had only gastro intestinal dysfunction in childhood and normal neurodevelopment.42

PathologySeveral reports have described the neuropathology of PMM2-CDG. Antoun and colleagues43 described a patient who presented in early infancy with dysmorphism (high nasal bridge and large jaw and ears), failure to thrive, hypotonia, lipodystrophy, acquired microcephaly, and abnormal (roving) eye movements. The child died at age 7 months from haematemesis complicating hepatic failure. Total brain weight was normal, but the cerebellum was very small, particularly the anterior vermis and adjacent hemispheres. Microscopic examination showed substantial neuronal loss in the olives and internal granular and Purkinje-cell layers, accompanied by reactive gliosis. Slight neuronal loss was seen in the

Enzymatic defect Gene OMIM code

Neurological abnormalities

ID Brain Seizures Ocular

GPI-anchor disorders*

Autosomal recessive GPI-anchor defi ciency Synthesis of N-acetylglucosaminyl phosphatidylinositol PIGA 300868 NR Yes Yes NR

CHIME syndrome De-N-acetylation of N-acetylglucosaminyl phosphatidylinositol PIGL 280000 Yes Yes Yes Yes

Autosomal recessive GPI-anchor defi ciency First α-mannosyltransferase in GPI biosynthesis PIGM 610293 Yes NR Yes NR

Autosomal recessive GPI-anchor defi ciency Transfers phosphoethanolamine to fi rst mannose of GPI anchor PIGN 614080 Yes NR Yes NR

Hyperphosphatasia mental retardation syndrome Second α-mannosyltransferase in GPI biosynthesis PIGV 239300 Yes NR Yes NR

α-Dystroglycanopathies†

Walker-Warburg syndrome (MDDGA1 and MDDGA2)‡ O-mannosyltransferase POMT1- POMT2 236670 Yes Yes Yes Yes

Muscle-eye-brain disease (MDDGA3)‡ O-mannosyl glycan GlcNAc transferase POMGNT1 253280 Yes Yes Yes Yes

Fukuyama-type congenital muscular dystrophy (MDDGA4, some MDDGB4)‡

Putative glycosyltransferase FKTN 253800 Yes Yes Yes Yes

Congenital muscular dystrophy type 1C (MDDGA5)‡ Fukutin-related protein, putative glycosyltransferase FKRP 606612 Yes Yes NR Yes

Congenital muscular dystrophy type 1D (MDDGA6)‡ Glycosyltransferase LARGE 608840 Yes Yes NR NR

Disorders without biochemical markers

Amish infantile epilepsy Sia2,3Galβ1,4Glc-Cer synthase (GM3) SIAT9 609056 Yes NR Yes NR

Non-syndromic intellectual disability α-1,2-mannosidase MAN1B1 614202 Yes NR NR NR

Peters plus syndrome β-1,3-glucosyltransferase specifi c for O-linked fucose on thrombospondin type 1 repeats

B3GALTL 261540 Yes NR NR NR

I-cell disease GlcNAc-1-P transferase GNPTA 252500 Yes Yes NR Yes§

OMIM=Online Mendelian Inheritance in Man database. ID=intellectual development is impaired. Brain=brain malformations, such as cerebellar atrophy or hypoplasia, or thinning of the corpus callosum. Ocular=ocular abnormalities, such as cataracts, optic-nerve atrophy, or coloboma. GPI=glycophosphatidylinositol. NR=not reported. CHIME=coloboma, heart defects, ichthyosis, mental insuffi ciency and ear defects. MDDG=muscular dystrophy-dystroglycanopathy. GlcNAc=N-acetylglucosamine. *Detectable by measurement of CD59 and fl uorescently labelled aerolysin. †Detectable by measurement of antibodies to glycosylated α-dsytroglycans. ‡In the MDDG nomenclature system, the letter relates to severity (A, severe; B, intermediate; C, mild) and the number to the gene involved. §Cataracts.

Table 4: Neurological features of glycosylation disorders detectable by methods other than measurement of transferrin glycosylation

www.thelancet.com/neurology Vol 11 May 2012 459

Review

pontine nuclei, but the dentate nuclei were preserved. The cerebrum was spared. An earlier report had described two siblings who had neonatal olivoponto cerebellar atrophy with similar clinical course and autopsy fi ndings, although those patients exhibited macrocephaly.44

Sural-nerve pathology has been described in nine patients aged 6 months to 15 years.45 The chief fi ndings included thin myelin sheaths, distorted periaxonal Schmidt-Lanterman incisures, and periaxonal dense bodies. Schwann cells contained multivacuolar bodies, and in some cases foamy inclusions. Fibrillary inclusions were noted in fi broblasts. These eff ects were seen at presentation in all patients except the youngest, in whom they appeared later and became more prominent with increasing age. Onion bulb formation and axonal regeneration was also noted in most patients. Nordborg and colleagues45 speculated that myelin-associated glycoprotein dysfunction had a crucial role because its distribution coincided with abnormalities seen in the region of the Schmidt-Lanterman incisures.

NeurophysiologyDetailed neurophysiology was reported in a boy aged 8 years with PMM2-CDG.46 Motor-nerve conduction gradually slowed over time, with a concomitant increase in latency, a decrease in amplitude of somatosensory evoked potentials, and progressive impairment of response on electroretinography and of visual evoked potentials that were consistent with pigmentary retinopathy. Selective rod involvement was noted on electroretinography.38 Auditory brainstem responses showed late changes suggestive of sensorineural hearing loss, and electroencephalography showed progressive slowing of background activity and the presence of sharp waves. A study of seven children with PMM2-CDG confi rmed the slowing of motor-nerve conduction with sparing of sensory-nerve conduction.47

NeuroimagingMost children with PMM2-CDG have notable atrophy of the superior vermis and cerebellar hemispheres at birth, but in mild cases initial MRI fi ndings might be normal, with subsequent progressive atrophy (fi gure 3).48 Pathological fi ndings suggest a very-early-onset atrophic process, rather than primary hypoplasia.43,44 PMM2-CDG is an important cause of cerebellar ataxia, as was shown in a study of posterior fossa imaging in 158 children with ataxia.49 Of 84 children with global cerebellar atrophy, 21 had PMM2-CDG, compared with 18 who had respiratory-chain defects. In general, the cerebral hemi-spheres are spared in PMM2-CDG.

ManagementNo defi nitive therapy is yet available for PMM2-CDG, but symptomatic management of seizures, stroke-like episodes, and systemic manifestations improves quality of life and can be life-saving.

ALG6-CDG (CDG-Ic)ALG6-CDG is the second most common N-linked congenital disorder of glycosylation. More than 50 pa tients have been identifi ed.31 An initial report of eight patients showed that the phenotype of ALG6-CDG is milder than that of PMM2-CDG.50 Like PMM2-CDG, clinical features include developmental delay, axial hypotonia, strabismus, and seizures. Ataxia is less common in ALG6-CDG than in PMM2-CDG, and usually the cerebellum is spared. Dysmorphic features are seldom seen in patients with ALG6-CDG, episodes are typically not life-threatening, and hepatic and renal involvement is rare. Patients with ALG6-CDG might, however, have severe coagulopathy, which is thought to have contributed to or caused a grade III intraventricular haemorrhage in one neonate.51 The overall neurological outcome in ALG6-CDG is generally better than in typical PMM2-CDG, although at least fi ve children have died from complications.31

Additional features have included severe protein-losing enteropathy during rotavirus infections and in the context of gastroenteritis and skeletal dysplasia.52,53 One woman with confi rmed ALG6 defi ciency has skeletal anomalies, virilisation, and intellectual disability, and has experienced deep-vein thrombosis and benign intracranial hyper-tension.54 By contrast, normal puberty was reported in another woman with ALG6-CDG.55 Benign intracranial hypertension has been reported in another case, in combination with optic atrophy and pigmentary retinopathy.56 Furthermore, one patient with ALG6-CDG presented with dilated cardiomyopathy.57

Most patients are compound heterozygotes for point mutations, but a Japanese child with sagittal

Figure 3: Cranial MRI in a patient with PMM2-CDG (CDG-1a)Images were obtained from a girl aged 17 years. Notable atrophy of the cerebellar vermis and hemispheres can be seen in (A) sagittal T1, (B) axial T2, and (C) coronal fl uid-attenuated inversion recovery images.

BA C

Figure 2: Clinical features of PMM2-CDG (CDG-1a)Children frequently present with (A) facies and esotropia, (B) nipple inversion, and (C) supragluteal fat pads.

CBA

460 www.thelancet.com/neurology Vol 11 May 2012

Review

cranio synostosis, a hypoplastic corpus callosum, absent septum pellucidum, and grade III intraventricular haemorrhage had a novel point mutation in the maternal allele, and a deletion that included the ALG6 locus in the paternal allele.51

SRD5A3-CDG (CDG-Iq)In 2010, Cantagrel and colleagues58 used homozygosity mapping of a consanguineous family and identifi ed a candidate reductase suspected of being involved in testosterone metabolism. The “long-sought polyprenol reductase”, SRD5A3, was shown to be encoded by SRD5A3 and to be crucial to proper glycosylation. Interestingly, most of the mutations identifi ed in SRD5A3 led to protein truncations and probably to complete loss of function. Additional proteins might be involved in the reduction of polyprenol, as Srd5a3 null mice can synthesise small amounts of dolichol.58

Patients with SRD5A3-CDG reported so far have had similar clinical presentations. All patients have shown intellectual disability, neurological manifestations (brain malformations or structural defects), including cerebellar vermis atrophy or hypoplasia and cerebellar ataxia. Ophthalmological defects have included forms of coloboma, nystagmus, optic-nerve atrophy or hypoplasia, cataracts, and glaucoma.59

ALG1-CDG (CDG-lk)ALG1 encodes an enzyme that acts early in the lipid-linked oligosaccharide biosynthesis pathway. Seven cases of ALG1-CDG had been described by 2009,29 Dupré and colleagues60 later identifi ed an additional fi ve cases, and a further ten cases were indentifi ed in eight families by a targeted analysis of unsolved cases in the USA (Freeze HH, unpublished). The latter fi nding emphasises the usefulness of a targeted approach.61

The patients described so far have all had severe neurological phenotypes. The children presented with muscular hypotonia, usually accompanied by facial dysmorphisms. Half of the patients had microcephaly and severe intellectual impairment, and all had epileptic seizures that were frequently intractable. MRI revealed cerebellar hypoplasia and cortical and subcortical atrophy in some cases, although results were normal in a few patients. Several patients were blind at presentation. In the initial reports, four of seven patients died in early childhood. Of the 15 identifi ed since 2009, however, 13 survived beyond the fi rst year of life. Thus, ALG1-CDG is more common and has a much broader clinical spectrum than was initially suspected.

Defects of the conserved oligomeric Golgi complexProper glycosylation depends on a bi-directional endosomal transportation system (secretory pathway) between the ER and the Golgi apparatus. Several protein complexes regulate transportation within and between these organelles, such as the conserved oligomeric Golgi (COG)

complex, which is an eight-subunit complex believed to regulate late and early endosomal retrograde transport to the trans-Golgi network.62,63 In 2004, mutations detected in COG complex subunit 7 led to the discovery of a new subgroup of congenital disorders of glycosylation, termed the COG-CDGs.64 Two siblings presented with multiple facial dysmorphias, generalised hypotonia, hepatomegaly, seizures, and cardiac insuffi ciency. Biochemical analysis showed combined N-linked and O-linked defi ciencies, with hypoglycosylation of trans ferrin and of apolipoprotein C-III, which is one of the few serum proteins with exclusively O-linked glycosylation. Defi ciencies in COGs 1, 4, 5, 6, and 8 have since been described.65–70

The severity of disease and degree of organ involvement diff er between patients, but psychomotor developmental delay (mild to severe) has been seen in all, along with micro cephaly and hypotonia, and often with epilepsy. Facial dysmorphias and failure to thrive are common, as are hepatomegaly and cardiac involve ment (heart failure, hypertrophy, or congenital heart defects). Brain imaging can be normal, but some patients have cerebral and cere-bellar atrophy. The diagnostic procedure for COG-CDGs involves analysis of transferrin and apolipoprotein C-III, western blot analysis of the subunits, and sequencing.

Cutis laxa syndromesCutis laxa is a dermatological disorder that can be localised or generalised, and is characterised by loose, sagging, inelastic skin. This disorder was seen in the fi rst COG7-CDG patient and was accompanied by severe neurological defi cits.64 Kornak and colleagues71 found that some patients with autosomal recessive cutis laxa II, or wrinkled skin syndrome, had mutations in ATP6V0A2, which encodes the α2 subunit of vacuolar-type proton-ATPase. Patients showed combined N-linked and O-linked glycosylation defi ciencies, which led to the creation of the classifi cation ATP6V0A2-CDG.71

20 patients with ATP6V0A2-CDG have been described. All have wrinkly skin and, typically, a large fontanelle with delayed closure, down-slanting palpebral fi ssures, joint laxity, muscular hypotonia, and eye anomalies (strabismus and myopia). Many patients have micro cephaly. Intellectual disability and developmental delay are variable. Half of the patients have brain abnor malities, such as partial pachygyria or cobblestone-like lissencephaly, but these are less severe than in the α-dystroglycanopathies, which are discussed below. Seizures are uncommon, but have been reported in older children. The early feeding diffi culties, growth retardation, and skin condition resolve with age in most patients.72,73

Although most patients have abnormal transferrin glycosylation, this feature was normal in some patients who presented at young ages, whereas O-linked glycosylation was abnormal.74 Nevertheless, some patients were only positive for defi cient N-linked glycosylation.72 Testing for both N-linked and O-linked hypoglycosylation is recommended if ATP6V0A2-CDG is suspected.

www.thelancet.com/neurology Vol 11 May 2012 461

Review

α-DystroglycanopathiesO-mannosylation has been proven to occur on α-dystroglycan, but other O-mannosylated proteins probably exist in the brain and in skeletal muscle. Defective O-mannosylation of α-dystroglycan causes a set of heterogeneous disorders called α-dystro glycanopathies. Manifestations range from early-onset muscular dystrophy, to severe brain and eye malfor mations, to late-onset muscle weakness with normal intelligence. The most severe disorders include muscle-eye-brain disease, Walker-Warburg syndrome, and Fukuyama congenital muscular dystrophy. Milder forms include limb-girdle muscular dystrophies. Intermediate presentations might be seen with or without intellectual disability (eg, congenital muscular dystrophy type 1D).13 Mutations in several genes have been associated with the traditional clinical syndromes, termed muscular dystrophy-dystroglycan-opathies (MDDGs; table 4), which has led to a nomenclature based on clinical severity and genetic cause. The severity classifi cations are A (severe), B (intermediate), and C (mild). The subtypes are numbered one to six according to the genetic cause, in the following order: POMT1, POMT2, POMGNT1, FKTN, FKRP, and LARGE. These genes encode known (POMT1, POMT2, POMGNT1, LARGE) or putative (FKTN and FKRP) glycosyltransferases.75 Thus, for example, Walker-Warburg syndrome due to POMT1 mutations is referred to as MDDGA1.76 More genes are likely to be discovered in this pathway.12

Patients with muscle-eye-brain disease (mainly MDDGA3) have generalised muscle weakness from birth, intellectual disability, and eye disorders, such as congenital myopia, pallor of the optic discs, and glaucoma. CNS malformations include frontoparietal pachygyria, polymicrogyria, cerebellar hypoplasia, and a fl attened pons and brainstem. The disease progresses slowly and most patients survive into adulthood. Most of the patients reported so far have been Finnish.77

Walker-Warburg syndrome (mainly MDDGA1 and MDDGA2) progresses rapidly and patients generally succumb by age 3 years.78 Muscular dystrophy is severe, with no motor develop ment, as are ophthalmopathies (congenital cataracts, microphthalmia, and buphthalmos). Imaging often reveals severe hydrocephalus, agenesis of the corpus callosum, agyria or cobblestone lissencephaly, and cerebellar hypoplasia.77

Fukuyama congenital muscular dystrophy (mainly MDDGA4) overlaps in phenotype with muscle-eye-brain disease. Patients present with general ised muscle weakness and hypotonia from early infancy. They have intellectual defi cits and some have seizures. The brain malformations include cerebral and cerebellar micropolygyria, hydrocephalus, and hypoplasia of the corticospinal tracts. Milder forms (MDDGB4) are well known. Most patients with Fukuyama congenital muscular dystrophy are of Japanese ancestry.77

The N-linked glycan and O-linked mannose pathways require dolichol phosphate-mannose, and defi ciency of

this carrier might aff ect both pathways. Discovery of a patient with DPM3-CDG, who had exclusively late-onset muscular dystrophy and cardiomyopathy underscores this point.79 The DPM3 mutation reduces concentrations of functional dolichol phosphate-mannose, which leads to hypoglycosylation of α-dystroglycan and muscle symptoms. Furthermore, 11 patients with dilated cardio-myopathy as their major symptom had DOLK-CDG.80 This congenital disorder of glycosylation is normally a severe multisystem disorder,81 but in this sub-group, heart biopsies revealed defi cient α-dystroglycan O-mannosylation. Thus, a transferrin test should be done in dystroglycanopathies where mutations cannot be found in the six known MDDG genes.

Glycosylphosphatidylinositol-anchor defectsTypical of glycosylation pathway disorders, glycosyl-phosphatidylinositol-anchor defects display a broad clinical spectrum. The fi rst defect to be identifi ed causes paroxysmal nocturnal haemoglobinuria owing to somatic mutations of the PIGA gene, exclusively in haemopoietic precursors.82 Inherited mutations in PIGA, PIGM, PIGN, PIGV, and PIGL systemically disrupt this pathway. Mutations in PIGM cause venous thrombosis and seizures; however, in one patient targeted butyrate treatment reversed intractable seizures because the mutations occur in an Sp1-promoter binding site.27,83

Mutations in PIGN led to severe neurological impairment, dysmorphia, chorea, seizures, and early death in one consanguineous family with seven aff ected children.28 Whole-exome sequencing showed that mutations in PIGV caused hyperphosphatasia mental retardation syndrome, with seizures of variable severity and muscular involvement.29 These patients have increased serum alkaline phosphatase concentrations with normal calcium and phosphate levels, brachy telephalangy, and facial dysmorphism. A germ-line mutation in PIGA resulted in a lethal disorder in three infants in one family. The children had neonatal seizures, thin corpus callosum, white-matter immaturity, no sep tum pellucidum, and a small cerebellum.84 Defects in PIGL have recently been shown to cause CHIME (coloboma, heart defects, ichthyosis, mental insuffi ciency and ear defects) syndrome (also called Zunich-Kaye syndrome).85

Diagnosis is challenging owing to the variable clinical presentations. All patients exhibit defi ciencies of glyco-phosphatidylinositol-anchored proteins on the cell surface. CD59 and fl uorescently labelled aerolysin are reliable surface markers on leucocytes that can quickly and easily identify glycophosphatidylinositol defi ciencies (table 4). At least 11 separate steps and more than 20 known proteins are involved in the formation of a functional glyco phosphatidylinositol anchor on proteins and, therefore, other inherited disorders that involve glyco pho s phatidylinositol biosynthesis will doubtless be found.23,86

462 www.thelancet.com/neurology Vol 11 May 2012

Review

Disorders in general substratesMutations in genes encoding proteins at the neuromuscular junction cause congenital myasthenic syndromes by impairing the safety margin for neuro-muscular transmission.87 In a study that exploited genetic linkage, Senderek and colleagues88 found 18 diff erent biallelic mutations in GFPT1 in 13 unrelated Pakistani families with myasthenic syndromes. The gene en codes glutamine-fructose-6-phosphate transaminase 1.88 This enzyme catalyses the fi rst step of hexosamine biosynthesis and has a rate-limiting eff ect that controls the fl ux of glucose into the pathway, which ultimately produces uridine diphosphate N-acetylglucosamine—a donor in nearly all glycosylation pathways. Atlhough no specifi c glycosylation pathways have been shown to cause the pathology, knock down of the GFPT1 orthologue in zebrafi sh embryos altered muscle fi bre morphology and impaired neuromuscular-junction development. Some mutations reduced enzymatic activity, but many did not, which suggests that the mutations could alter binding of the enzyme to other proteins to effi ciently direct substrate use.

DiagnosisCurrent statusBiochemical and genetic approaches are both required to identify a defective gene and to show that the mutations impair function. Biochemical analysis initially guided gene identifi cation, but high-capacity, low-cost DNA sequencing and powerful informatics computer programs can now identify possible candidate genes. Until all known and likely candidates have been queried, however, each candidate must be proven to cause a biochemical or biological abnormality.89,90

The discovery of many glycosylation disorders in the past 15 years is largely due to the ease of measurement of glycosylated transferrin concentrations in serum as a marker of N-linked glycosylation fi delity (table 2). Mass spectrometry is the preferred method and can indicate the absence of entire N-linked glycan chains, incomplete sculpting of those chains, or both. Isoelectric focusing analysis of transferrin measures loss of sialic acids and can, therefore, serve as a less expensive method for assessment of N-linked glycosylation disorders. Other diagnostic methods that measure glycosylated transferrin concentrations include capillary electrophoresis, high-performance liquid chromatography, and ion-exchange chromatography. Repeated analyses, sialidase digestion, or analysis of other serum glycoproteins might be required to confi rm results. In the traditional nomen-clature for congenital disorders of glycosylation, absence of entire glycans was designated type I, and loss of one or more monosaccharides as type II. These patterns are not gene specifi c, but they help to identify the defective gene by localisation in the relevant pathway.

False-positive results occur in the context of uncon-trolled hereditary fructose intolerance, galactosaemia,

heavy alcohol consumption, certain hepatic pathologies, and rare mutations at the transferrin glycosylation site.91 Conversely, preterm infants might have normal patterns of transferrin glycosylation at birth that later become abnormal. In patients who present with N-linked glycosylation defi ciencies, initially abnormal transferrin glycosylation sometimes normalises without improve-ments in symptoms, and initially normal transferrin glycosylation has been reported in patients who were later proven to have glycosylation disorders.39,91–93 These fi ndings suggest that a window of opportunity exists for diagnosis made on the basis of transferrin measurement. Data from thousands of transferrin tests done at the Mayo Clinic, Rochester, MN, USA, clearly show that the period between the ages of 6 months and 18 months is optimum for diagnosis (unpublished). Moreover, abnormal transferrin glyco sylation is detected three to four times as often as are aminoacid, organic acid, fatty acid, mitochondrial, and peroxisomal disorders (Raymond K, Mayo Clinic, Rochester, MN, USA, unpublished), which suggests that the frequency of glycosylation disorders is notably higher than is currently appreciated. Genetic confi rm ation of diagnoses was not available and, therefore, we cannot distinguish prevalence from a phenotypic clinical bias. A congenital disorder of glycosylation should be suspected in any patient with developmental delay of undetermined cause, particularly when associated with symptoms and signs presented in tables 2 and 3. Measurement of additional markers, such as transaminases, creatine kinase, thyroxine-binding globulin, coagulation factors, and lysosomal enzymes can be useful, along with liver, muscle, and kidney function tests. These tests are frequently clinically indicated for other reasons.

As mentioned above, a normal result in transferrin glycosylation testing does not exclude a glycosylation disorder. For instance, normal results have been reported in patients with genetic disorders involving the oligosaccharyltransferase complex subunits encoded by TUSC3 and MAGT1.94,95 By contrast, mutations in another subunit, encoded by DDOST, have led to abnormal transferrin glyco sylation.85,96 The reason for this diff erence is unclear, but additional disorders involving the other subunits of the complex are likely to be identifi ed.

Not all suspected congenital disorders of glycosylation have convenient markers, although some tests can indicate pathways, albeit without identifi cation of specifi c genes: the α-dystroglycanopathies can be investigated by measurement of monoclonal antibodies to the O-mannosylated glycan in muscle biopsy samples,97 and measurement of CD59 concentrations can be useful to detect glycophosphatidylinositol-anchor defi ciencies (table 4).27,29 Testing of targeted genes that is approved through Clinical Laboratory Improvement Amendments is available.61 An analysis of about 35 patients with abnormal transferrin glycosylation showed damaging mutations in only seven known genes (Freeze H, Ng B,

www.thelancet.com/neurology Vol 11 May 2012 463

Review

unpublished). Barring the unlikely event that these patients have mutations in the promoter, deep intronic regions, or exons that were not amplifi ed, the results suggest that many glycosylation disorders result from mutations in as yet unidentifi ed genes.

Next-generation sequencingDiscovery of most congenital disorders of glycosylation has relied on initial biochemical analysis of fi broblasts and serum glycans to identify mutated genes. Many of the expected candidate genes were identifi ed, but unexpected genes were found, as some of the examples above show, and more will certainly be found in the future. Genetic mapping techniques have been very useful, especially in consanguineous families, but the availability of whole-exome (or entire-genome) sequencing is the future for diagnosis of glycosylation and other rare disorders. Falling costs, improvements in informatics analysis, and increased availability of technology puts such approaches within reach. However, fi ndings must be considered in conjunction with initial biochemical analysis and evidence of the glycosylation-related pathology for the optimum paradigm. In the past 3 years, fi ve defects were solved with traditional biochemical analysis and ten by homozygosity mapping. Within the past year, six defects have been assessed with state-of-the-art whole-exome analysis.

The coming years will require the dedicated collaboration of physicians and scientists to confi rm the putative candidate genes in patients with glycosylation disorders. This approach will be extremely important to enable translation of the results into useful information for families, who are the biggest stakeholders in the fi eld of rare diseases.

Prospects for therapyAlthough early in-vitro studies showed that PMM2 defi ciency could be corrected in patient donor cells by the addition of mannose to the culture medium, this treatment was not successful in vivo for PMM2-CDG patients.98,99 In-vitro studies aimed at redirecting exogenous mannose from catabolism towards glycosylation have shown positive results in some patient-derived cells, but not in others.100,101 A recent study in Pmm2 hypomorphic mice is noteworthy.102 Compound heterozygous mice carrying the two most common human mutations in Pmm2 all die in utero at about 10·5 days’ gestation. Provision to pregnant dams of small amounts of mannose in their drinking water during gestation prevented embryonic death. Pups did not require mannose after weaning and seemed to be healthy and have a normal lifespan. This remarkable fi nding suggests that only slight increases in mannose availability and use can have substantial benefi ts for this specifi c genotype. Whether this result will be applicable to PMM2-CDG patients remains to be seen and we believe it would be premature to view mannose

intake by at-risk pregnant mothers as a potential benefi t for a PMM2-CDG fetus.

The antidiabetic drug, metformin, has been proposed as a glycosylation enhancer.103 MPI-CDG patients who have a non-neurological phenotype (appendix) have shown almost complete remission with dietary mannose sup-plementation.104,105 Fucose therapy is eff ective for a few patients with CDG-IIc, which is a defi ciency of Golgi GDP-fucose transporter, and butyrate is eff ective in patients with PIGM mutations.83,106 Whether this approach can be extended to other disorders is unknown, but provision of N-acetylglucosamine (although not glucosamine) has shown some effi cacy in a mouse model of multiple sclerosis.107,108 Sialic acid and N-acetylmannosamine sup-plements might be useful in the treatment of hereditary inclusion-body myopathy, which has adult onset.109,110 These examples show that simple monosaccharide therapy can be useful for treating selected disorders.

The repurposing of approved drugs has not been tried to any great extent. This approach has obvious advantages, but there is currently insuffi cient information to aid selection of candidate drugs. Another approach is to identify small molecules that enhance substrate fl ux into depleted pathways or stabilise mutated proteins.111,112 The former approach has been used to inhibit the activity of mannose-6-phosphate isomerase and direct the fl ux of mannose through PMM2 to increase glycosylation. The latter approach, in the form of enzyme enhancement or chaperone therapy, should be applicable to most disorders that involve hypomorphic mutations, and has received substantial attention,113,114 although it has not been applied to glycosylation disorders. Stem-cell therapy might ultimately provide a therapeutic solution, but is not yet a viable option. Gene therapy was used in a case of adult-onset inclusion-body myopathy with promising results.115

Even in the absence of defi nitive disease-modifying therapies for congenital disorders of glycosylation, substantial benefi ts can be expected from early recognition and management. Early diagnosis saves patients and their families from a diagnostic odyssey and enables accurate genetic counselling for members of the extended family, and patients can receive appropriate treatment for symptoms and proactive management in view of the types of complications that are likely to occur.

Aggressive symptomatic therapy can yield impressive benefi ts in disorders for which there are no defi nitive treatments. The best example is cystic fi brosis. In the past 50 years, the lifespan of patients with cystic fi brosis has increased substantially, despite the absence of disease-modifying therapy.116 The institution of treatment guidelines based on evidence and experience, and the establishment of specialist clinical centres, combined with aggressive symptomatic therapy explain this impressive result. There is no reason why this approach should not be employed with equal effi cacy in other rare diseases, including congenital disorders of glycosylation.

464 www.thelancet.com/neurology Vol 11 May 2012

Review

Conclusions and future perspectivesIn the past decade the number of genes identifi ed as having causal roles in congenital disorders of glycosylation has far surpassed that for any other subgroup of congenital errors of metabolism. The classic PMM2-CDG phenotype—the fi rst to be recognised—describes only a portion of the patients, and defective glycosylation should be suspected in any patient with symptoms that involve many organs. Because many of these disorders can be detected with a simple and inexpensive blood test, there is every incentive to exclude or confi rm the diagnosis. Despite vast growth in our knowledge of congenital disorders of glycosylation, treatment options are non-existent in most cases, and substantially more research needs to be done.

In the next few years, inexpensive whole-exome and whole-genome sequencing are likely to reveal candidate genes for scores, if not hundreds, of rare disorders with neurological consequences. Confi rmation of the patho genicity of candidate genes might be straight-forward in cases where no gene product is made, but is not typically the case. More often, to determine the pathogenicity of mutations, cooperation will be required between clinicians and basic scientists. For glycosylation dis orders, participation of a consultant glycobiologist will also be important.

ContributorsHHF originated and organised the paper, EAE and BGN prepared tables,

and MCP planned the Review. All authors wrote portions of the article

and reviewed drafts of the paper.

Confl icts of interestMCP acts as a consultant for Shire HGT and is Chair of the data

monitoring committee for Stem Cells Inc. He has received travel

expenses from the National Niemann-Pick Disease Foundation and the

US Institute of Medicine (as a member of the Committee on Adverse

Eff ects of Vaccines), and is a member of the WHO Topic Advisory Group

(Neurology) – Revision of ICD-10, for which he receives no

compensation or expenses. The other authors declare that they have no

confl icts of interest.

AcknowledgmentsHHF is supported by The Rocket Fund, National Institutes of Health

(R01DK55615), and a Sanford Professorship; EAE is supported by Avtal

om läkarutbildning och forskning and Crafoordska stiftelsen; and MCP

is supported by NINDS (U54 NS065768), National MS Society, Actelion

Pharmaceuticals, and Merck-Serono. We thank Kimiyo Raymond for

allowing us to cite her unpublished work, Ping He,

Search strategy and selection criteria

We searched PubMed for papers published from 1978 to Feb 10, 2012, with the terms “glycosylation”, “carbohydrate defi cient glycoprotein syndrome”, “congenital disorder(s) of glycosylation”, “glycosaminoglycan”, “glycosylphosphatidylinositol anchors”, “glycolipid”, “glycosphingolipid”, and “transferrin”. Articles were also identifi ed by consultation with other experts and by manual searches of the reference lists of retrieved articles. We only reviewed papers published in English. The fi nal reference list was generated on the basis of relevance to the topics covered in this Review. In addition, we reviewed Online Mendelian Inheritance in Man (OMIM) for each of the disorders mentioned. We placed emphasis on congenital disorders of glycosylation discovered in the past 2–3 years, those for which ten or more patients have been described, and those being seen with increasing frequency in the clinic.

Mariam Rodriguez Lee, Marie-Estelle Losfeld, and Vandana Sharma for

critical review of the paper, and Amy Zimmon for her expert assistance

in preparation of the Review. We also thank Gert Matthijs, Rafael Artuch

Iriberri, Donna Krasnewich, Flemming Skovby, and Nathalie Seta for

providing information on the numbers of patients with specifi c

congenital disorders of glycosylation.

References1 Freeze HH. Genetic defects in the human glycome. Nat Rev Genet

2006; 7: 537–51.

2 Ungar D. Golgi linked protein glycosylation and associated diseases. Semin Cell Dev Biol 2009; 20: 762–69.

3 Varki A. Biological roles of oligosaccharides: all of the theories are correct. Glycobiology 1993; 3: 97–130.

4 Jaeken J, Hennet T, Freeze HH, Matthijs G. On the nomenclature of congenital disorders of glycosylation (CDG). J Inherit Metab Dis 2008; 31: 669–72.

5 Stanley P, Schachter H, Taniguchi N. N-Glycans. In: Varki A, Cummings RD, Esko JD, et al, eds. Essentials of glycobiology, 2nd edn. New York, NY: Cold Spring Harbor Laboratory Press, 2009.

6 Zielinska DF, Gnad F, Wisniewski JR, Mann M. Precision mapping of an in vivo N-glycoproteome reveals rigid topological and sequence constraints. Cell 2010; 141: 897–907.

7 Li H, Chavan M, Schindelin H, Lennarz WJ. Structure of the oligosaccharyl transferase complex at 12 A resolution. Structure 2008; 16: 432–40.

8 Dennis JW, Nabi IR, Demetriou M. Metabolism, cell surface organization, and disease. Cell 2009; 139: 1229–41.

9 Furmanek A, Hofsteenge J. Protein C-mannosylation: facts and questions. Acta Biochim Pol 2000; 47: 781–89.

10 Ihara Y, Manabe S, Ikezaki M, et al. C-Mannosylated peptides derived from the thrombospondin type 1 repeat interact with Hsc70 to modulate its signaling in RAW264.7 cells. Glycobiology 2010; 20: 1298–310.

11 Yoshida-Moriguchi T, Yu L, Stalnaker SH, et al. O-mannosyl phosphorylation of alpha-dystroglycan is required for laminin binding. Science 2010; 327: 88–92.

12 Stalnaker SH, Aoki K, Lim JM, et al. Glycomic analyses of mouse models of congenital muscular dystrophy. J Biol Chem 2011; 286: 21180–90.

13 Wang CH, Bonnemann CG, Rutkowski A, et al. Consensus statement on standard of care for congenital muscular dystrophies. J Child Neurol 2010; 25: 1559–81.

14 Brockhausen I, Schachter H, Stanley P. O-GalNAc glycans. In: Varki A, Cummings RD, Esko JD, et al, eds. Essentials of glycobiology, 2nd edn. New York, NY: Cold Spring Harbor Laboratory Press, 2009.

15 Esko JD, Kimata K, Lindahl U. Proteoglycans and sulfated glycosaminoglycans. In: Varki A, Cummings RD, Esko JD, et al, eds. Essentials of glycobiology, 2nd edn. New York, NY: Cold Spring Harbor Laboratory Press, 2009.

16 Lander AD. Morpheus unbound: reimagining the morphogen gradient. Cell 2007; 128: 245–56.

17 Zhang YT, Lander AD, Nie Q. Computational analysis of BMP gradients in dorsal-ventral patterning of the zebrafi sh embryo. J Theor Biol 2007; 248: 579–89.

18 Stanley P, Okajima T. Roles of glycosylation in Notch signaling. Curr Top Dev Biol 2010; 92: 131–64.

19 Takeuchi H, Haltiwanger RS. Role of glycosylation of Notch in development. Semin Cell Dev Biol 2010; 21: 638–45.

20 Freeze HH, Haltiwanger RS. Other classes of ER/Golgi-derived glycans. In: Varki A, Cummings RD, Esko JD, et al, eds. Essentials of glycobiology, 2nd edn. New York, NY: Cold Spring Harbor Laboratory Press, 2009.

21 Schnaar RL, Suzuki A, Stanley P. Glycosphingolipids. In: Varki A, Cummings RD, Esko JD, et al, eds. Essentials of glycobiology, 2nd edn. New York, NY: Cold Spring Harbor Laboratory Press, 2009.

22 Hakomori S. Carbohydrate-to-carbohydrate interaction, through glycosynapse, as a basis of cell recognition and membrane organization. Glycoconj J 2004; 21: 125–37.

23 Ferguson MAJ, Kinoshita T, Hart GW. Glycosylphosphatidylinositol anchors. In: Varki A, Cummings RD, Esko JD, et al, eds. Essentials of glycobiology. New York, NY: Cold Spring Harbor Laboratory Press, 2009.

www.thelancet.com/neurology Vol 11 May 2012 465

Review

24 Hwa KY. Glycosyl phosphatidylinositol-linked glycoconjugates: structure, biosynthesis and function. Adv Exp Med Biol 2001; 491: 207–14.

25 Hancock JF. GPI-anchor synthesis: Ras takes charge. Dev Cell 2004; 6: 743–45.

26 Simpson MA, Cross H, Proukakis C, et al. Infantile-onset symptomatic epilepsy syndrome caused by a homozygous loss-of-function mutation of GM3 synthase. Nat Genet 2004; 36: 1225–29.

27 Almeida AM, Murakami Y, Layton DM, et al. Hypomorphic promoter mutation in PIGM causes inherited glycosylphosphatidylinositol defi ciency. Nat Med 2006; 12: 846–51.

28 Maydan G, Noyman I, Har-Zahav A, et al. Multiple congenital anomalies-hypotonia-seizures syndrome is caused by a mutation in PIGN. J Med Genet 2011; 48: 383–89.

29 Krawitz PM, Schweiger MR, Rodelsperger C, et al. Identity-by-descent fi ltering of exome sequence data identifi es PIGV mutations in hyperphosphatasia mental retardation syndrome. Nat Genet 2010; 42: 827–29.

30 Freeze HH, Ng BG. Golgi glycosylation and human inherited diseases. Cold Spring Harb Perspect Biol 2011; 3: a005371.

31 Haeuptle MA, Hennet T. Congenital disorders of glycosylation: an update on defects aff ecting the biosynthesis of dolichol-linked oligosaccharides. Hum Mutat 2009; 30: 1628–41.

32 Pearl PL, Krasnewich D. Neurologic course of congenital disorders of glycosylation. J Child Neurol 2001; 16: 409–13.

33 Miossec-Chauvet E, Mikaeloff Y, Heron D, et al. Neurological presentation in pediatric patients with congenital disorders of glycosylation type Ia. Neuropediatrics 2003; 34: 1–6.

34 Grunewald S. The clinical spectrum of phosphomannomutase 2 defi ciency (CDG-Ia). Biochim Biophys Acta 2009; 1792: 827–34.

35 Hagberg BA, Blennow G, Kristiansson B, Stibler H. Carbohydrate-defi cient glycoprotein syndromes: peculiar group of new disorders. Pediatr Neurol 1993; 9: 255–62.

36 de Lonlay P, Seta N, Barrot S, et al. A broad spectrum of clinical presentations in congenital disorders of glycosylation I: a series of 26 cases. J Med Genet 2001; 38: 14–19.

37 Ishikawa N, Tajima G, Ono H, Kobayashi M. Diff erent neuroradiological fi ndings during two stroke-like episodes in a patient with a congenital disorder of glycosylation type Ia. Brain Dev 2009; 31: 240–43.

38 Jensen H, Kjaergaard S, Klie F, Moller HU. Ophthalmic manifestations of congenital disorder of glycosylation type 1a. Ophthalmic Genet 2003; 24: 81–88.

39 Krasnewich D, O’Brien K, Sparks S. Clinical features in adults with congenital disorders of glycosylation type Ia (CDG-Ia). Am J Med Genet C Semin Med Genet 2007; 145C: 302–06.

40 Casado M, O’Callaghan MM, Montero R, et al. Mild clinical and biochemical phenotype in two patients with PMM2-CDG (congenital disorder of glycosylation Ia). Cerebellum 201; published online Oct 20. DOI:10.1007/s12311-011-0313-y.

41 Coman D, McGill J, MacDonald R, et al. Congenital disorder of glycosylation type 1a: three siblings with a mild neurological phenotype. J Clin Neurosci 2007; 14: 668–72.

42 Perez-Duenas B, Garcia-Cazorla A, Pineda M, et al. Long-term evolution of eight Spanish patients with CDG type Ia: typical and atypical manifestations. Eur J Paediatr Neurol 2009; 13: 444–51.

43 Antoun H, Villeneuve N, Gelot A, Panisset S, Adamsbaum C. Cerebellar atrophy: an important feature of carbohydrate defi cient glycoprotein syndrome type 1. Pediatr Radiol 1999; 29: 194–98.

44 Horslen SP, Clayton PT, Harding BN, Hall NA, Keir G, Winchester B. Olivopontocerebellar atrophy of neonatal onset and disialotransferrin developmental defi ciency syndrome. Arch Dis Child 1991; 66: 1027–032.

45 Nordborg C, Hagberg BA, Kristiansson B. Sural nerve pathology in the carbohydrate-defi cient glycoprotein syndrome. Acta Paediatr 1991; 80: 39–49.

46 Veneselli E, Biancheri R, Di Rocco M, Tortorelli S. Neurophysiological fi ndings in a case of carbohydrate-defi cient glycoprotein (CDG) syndrome type I with phosphomannomutase defi ciency. Eur J Paediatr Neurol 1998; 2: 239–44.

47 Pascual-Castroviejo I, Pascual-Pascual SI, Quijano-Roy S, et al. Cerebellar ataxia of Norman-Jaeken. Presentation of seven Spanish patients. Rev Neurol 2006; 42: 723–28 (in Spanish).

48 Mader I, Dobler-Neumann M, Kuker W, Stibler H, Krageloh-Mann I. Congenital disorder of glycosylation type Ia: benign clinical course in a new genetic variant. Childs Nerv Syst 2002; 18: 77–80.

49 Boddaert N, Desguerre I, Bahi-Buisson N, et al. Posterior fossa imaging in 158 children with ataxia. J Neuroradiol 2010; 37: 220–30.

50 Imbach T, Grunewald S, Schenk B, et al. Multi-allelic origin of congenital disorder of glycosylation (CDG)-Ic. Hum Genet 2000; 106: 538–45.

51 Eklund EA, Sun L, Yang SP, Pasion RM, Thorland EC, Freeze HH. Congenital disorder of glycosylation Ic due to a de novo deletion and an hALG-6 mutation. Biochem Biophys Res Commun 2006; 339: 755–60.

52 Westphal V, Schottstadt C, Marquardt T, Freeze HH. Analysis of multiple mutations in the hALG6 gene in a patient with congenital disorder of glycosylation Ic. Mol Genet Metab 2000; 70: 219–23.

53 Drijvers JM, Lefeber DJ, de Munnik SA, et al. Skeletal dysplasia with brachytelephalangy in a patient with a congenital disorder of glycosylation due to ALG6 gene mutations. Clin Genet 2010; 77: 507–09.

54 Sun L, Eklund EA, Van Hove JL, Freeze HH, Thomas JA. Clinical and molecular characterization of the fi rst adult congenital disorder of glycosylation (CDG) type Ic patient. Am J Med Genet A 2005; 137: 22–26.

55 Miller BS, Freeze HH, Hoff mann GF, Sarafoglou K. Pubertal development in ALG6 defi ciency (congenital disorder of glycosylation type Ic). Mol Genet Metab 2011; 103: 101–03.

56 Kahook MY, Mandava N, Bateman JB, Thomas JA. Glycosylation type Ic disorder: idiopathic intracranial hypertension and retinal degeneration. Br J Ophthalmol 2006; 90: 115–16.

57 Al-Owain M, Mohamed S, Kaya N, Zagal A, Matthijs G, Jaeken J. A novel mutation and fi rst report of dilated cardiomyopathy in ALG6-CDG (CDG-Ic): a case report. Orphanet J Rare Dis 2010; 5: 7.

58 Cantagrel V, Lefeber DJ, Ng BG, et al. SRD5A3 is required for converting polyprenol to dolichol and is mutated in a congenital glycosylation disorder. Cell 2010; 142: 203–17.

59 Morava E, Wevers RA, Cantagrel V, et al. A novel cerebello-ocular syndrome with abnormal glycosylation due to abnormalities in dolichol metabolism. Brain 2010; 133: 3210–20.

60 Dupré T, Vuillaumier-Barrot S, Chantret I, et al. Guanosine diphosphate-mannose:GlcNAc2-PP-dolichol mannosyltransferase defi ciency (congenital disorders of glycosylation type Ik): fi ve new patients and seven novel mutations. J Med Genet 2010; 47: 729–35.

61 Jones MA, Bhide S, Chin E, et al. Targeted polymerase chain reaction-based enrichment and next generation sequencing for diagnostic testing of congenital disorders of glycosylation. Genet Med 2011; 13: 921–32.

62 Reynders E, Foulquier F, Annaert W, Matthijs G. How Golgi glycosylation meets and needs traffi cking: the case of the COG complex. Glycobiology 2011; 21: 853–63.

63 Laufman O, Hong W, Lev S. The COG complex interacts directly with Syntaxin 6 and positively regulates endosome-to-TGN retrograde transport. J Cell Biol 2011; 194: 459–72.

64 Wu X, Steet RA, Bohorov O, et al. Mutation of the COG complex subunit gene COG7 causes a lethal congenital disorder. Nat Med 2004; 10: 518–23.

65 Foulquier F, Vasile E, Schollen E, et al. Conserved oligomeric Golgi complex subunit 1 defi ciency reveals a previously uncharacterized congenital disorder of glycosylation type II. Proc Natl Acad Sci USA 2006; 103: 3764–69.

66 Reynders E, Foulquier F, Leao Teles E, et al. Golgi function and dysfunction in the fi rst COG4-defi cient CDG type II patient. Hum Mol Genet 2009; 18: 3244–56.

67 Paesold-Burda P, Maag C, Troxler H, et al. Defi ciency in COG5 causes a moderate form of congenital disorders of glycosylation. Hum Mol Genet 2009; 18: 4350–56.

68 Lubbehusen J, Thiel C, Rind N, et al. Fatal outcome due to defi ciency of subunit 6 of the conserved oligomeric Golgi complex leading to a new type of congenital disorders of glycosylation. Hum Mol Genet 2010; 19: 3623–33.

69 Foulquier F, Ungar D, Reynders E, et al. A new inborn error of glycosylation due to a Cog8 defi ciency reveals a critical role for the Cog1-Cog8 interaction in COG complex formation. Hum Mol Genet 2007; 16: 717–30.

466 www.thelancet.com/neurology Vol 11 May 2012

Review

70 Kranz C, Ng BG, Sun L, et al. COG8 defi ciency causes new congenital disorder of glycosylation type IIh. Hum Mol Genet 2007; 16: 731–41.

71 Kornak U, Reynders E, Dimopoulou A, et al. Impaired glycosylation and cutis laxa caused by mutations in the vesicular H+-ATPase subunit ATP6V0A2. Nat Genet 2008; 40: 32–34.

72 Hucthagowder V, Morava E, Kornak U, et al. Loss-of-function mutations in ATP6V0A2 impair vesicular traffi cking, tropoelastin secretion and cell survival. Hum Mol Genet 2009; 18: 2149–65.

73 Morava E, Lefeber DJ, Urban Z, et al. Defi ning the phenotype in an autosomal recessive cutis laxa syndrome with a combined congenital defect of glycosylation. Eur J Hum Genet 2008; 16: 28–35.

74 Morava E, Guillard M, Lefeber DJ, Wevers RA. Autosomal recessive cutis laxa syndrome revisited. Eur J Hum Genet 2009; 17: 1099–110.

75 Hewitt JE. Abnormal glycosylation of dystroglycan in human genetic disease. Biochim Biophys Acta 2009; 1792: 853–61.

76 Amberger J, Bocchini C, Hamosh A. A new face and new challenges for Online Mendelian Inheritance in Man (OMIM®). Hum Mutat 2011; 32: 564–67.

77 Clement E, Mercuri E, Godfrey C, et al. Brain involvement in muscular dystrophies with defective dystroglycan glycosylation. Ann Neurol 2008; 64: 573–82.

78 Vajsar J, Schachter H, Orphanet J. Walker-Warburg syndrome. Rare Dis 2006; 1: 29.

79 Lefeber DJ, Schonberger J, Morava E, et al. Defi ciency of Dol-P-Man synthase subunit DPM3 bridges the congenital disorders of glycosylation with the dystroglycanopathies. Am J Hum Genet 2009; 85: 76–86.

80 Lefeber DJ, de Brouwer AP, Morava E, et al. Autosomal recessive dilated cardiomyopathy due to DOLK mutations results from abnormal dystroglycan O-mannosylation. PLoS Genet 2011; 7: e1002427.

81 Kranz C, Jungeblut C, Denecke J, et al. A defect in dolichol phosphate biosynthesis causes a new inherited disorder with death in early infancy. Am J Hum Genet 2007; 80: 433–40.

82 Miyata T, Yamada N, Iida Y, et al. Abnormalities of PIG-A transcripts in granulocytes from patients with paroxysmal nocturnal hemoglobinuria. N Engl J Med 1994; 330: 249–55.

83 Almeida AM, Murakami Y, Baker A, et al. Targeted therapy for inherited GPI defi ciency. N Engl J Med 2007; 356: 1641–47.

84 Johnston JJ, Gropman AL, Sapp JC, et al. The phenotype of a germline mutation in PIGA: the gene somatically mutated in paroxysmal nocturnal hemoglobinuria. Am J Hum Genet 2012; 90: 295–300.

85 Ng BG, Hackmann K, Jones MA, et al. Mutations in the glycosylphosphatidylinositol gene PIGL cause CHIME syndrome. Am J Hum Genet (in press).

86 Maeda Y, Kinoshita T. Structural remodeling, traffi cking and functions of glycosylphosphatidylinositol-anchored proteins. Prog Lipid Res 2011; 50: 411–24.

87 Farrugia ME. Myasthenic syndromes. J R Coll Physicians Edinb 2011; 41: 43–47; quiz 48.

88 Senderek J, Muller JS, Dusl M, et al. Hexosamine biosynthetic pathway mutations cause neuromuscular transmission defect. Am J Hum Genet 2011; 88: 162–72.

89 Zelinger L, Banin E, Obolensky A, et al. A missense mutation in DHDDS, encoding dehydrodolichyl diphosphate synthase, is associated with autosomal-recessive retinitis pigmentosa in Ashkenazi Jews. Am J Hum Genet 2011; 88: 207–15.

90 Gahl WA, Markello TC, Toro C, et al. The National Institutes of Health Undiagnosed Diseases Program: insights into rare diseases. Genet Med 2012; 14: 51–59.

91 Perez-Cerda C, Quelhas D, Vega AI, Ecay J, Vilarinho L, Ugarte M. Screening using serum percentage of carbohydrate-defi cient transferrin for congenital disorders of glycosylation in children with suspected metabolic disease. Clin Chem 2008; 54: 93–100.

92 Vermeer S, Kremer HP, Leijten QH, et al. Cerebellar ataxia and congenital disorder of glycosylation Ia (CDG-Ia) with normal routine CDG screening. J Neurol 2007; 254: 1356–58.

93 Freeze HH. Congenital disorders of glycosylation: CDG-I, CDG-II, and beyond. Curr Mol Med 2007; 7: 389–96.

94 Garshasbi M, Hadavi V, Habibi H, et al. A defect in the TUSC3 gene is associated with autosomal recessive mental retardation. Am J Hum Genet 2008; 82: 1158–64.

95 Molinari F, Foulquier F, Tarpey PS, et al. Oligosaccharyltransferase-subunit mutations in nonsyndromic mental retardation. Am J Hum Genet 2008; 82: 1150–57.

96 Jones MA, Ng BG, Bhide S, et al. DDOST mutations identifi ed by whole-exome sequencing are implicated in congenital disorders of glycosylation. Am J Hum Genet 2012; 90: 363–68.

97 Muntoni F, Torelli S, Wells DJ, Brown SC. Muscular dystrophies due to glycosylation defects: diagnosis and therapeutic strategies. Curr Opin Neurol 2011; 24: 437–42.

98 Panneerselvam K, Freeze HH. Mannose corrects altered N-glycosylation in carbohydrate-defi cient glycoprotein syndrome fi broblasts. J Clin Invest 1996; 97: 1478–87.

99 Mayatepek E, Schröder M, Kohlmüller D, Bieger WP, Nützenadel W. Continuous mannose infusion in carbohydrate-defi cient glycoprotein syndrome type I. Acta Paediatr 1997; 86: 1138–40.

100 Sharma V, Ichikawa M, He P, et al. Phosphomannose isomerase inhibitors improve N-glycosylation in selected phosphomannomutase defi cient fi broblasts. J Biol Chem 2011; 286: 39431–38.

101 Dahl R, Bravo Y, Sharma V, et al. Potent, selective, and orally available benzoisothiazolone phosphomannose isomerase inhibitors as probes for congenital disorder of glycosylation Ia. J Med Chem 2011; 54: 3661–68.

102 Schneider A, Thiel C, Rindermann J, et al. Successful prenatal mannose treatment for congenital disorder of glycosylation-Ia in mice. Nat Med 2011; 18: 71–73.

103 Shang J, Lehrman MA. Metformin-stimulated mannose transport in dermal fi broblasts. J Biol Chem 2004; 279: 9703–12.

104 de Lonlay P, Seta N. The clinical spectrum of phosphomannose isomerase defi ciency, with an evaluation of mannose treatment for CDG-Ib. Biochim Biophys Acta 2009; 1792: 841–43.

105 Harms HK, Zimmer KP, Kurnik K, Bertele-Harms RM, Weidinger S, Reiter K. Oral mannose therapy persistently corrects the severe clinical symptoms and biochemical abnormalities of phosphomannose isomerase defi ciency. Acta Paediatr 2002; 91: 1065–72.

106 Eklund EA, Freeze HH. The congenital disorders of glycosylation: a multifaceted group of syndromes. NeuroRx 2006; 3: 254–63.

107 Mkhikian H, Grigorian A, Li CF, et al. Genetics and the environment converge to dysregulate N-glycosylation in multiple sclerosis. Nat Commun 2011; 2: 334.

108 Grigorian A, Araujo L, Naidu NN, Place D, Choudhury B, Demetriou M. N-acetylglucosamine inhibits T-helper 1 (Th1)/T-helper 17 (Th17) responses and treats experimental autoimmune encephalomyelitis. J Biol Chem 2011; 286: 40133–41.

109 Malicdan MC, Noguchi S, Nishino I. A preclinical trial of sialic acid metabolites on distal myopathy with rimmed vacuoles/hereditary inclusion body myopathy, a sugar-defi cient myopathy: a review. Ther Adv Neurol Disord 2010; 3: 127–35.

110 Galeano B, Klootwijk R, Manoli I, et al. Mutation in the key enzyme of sialic acid biosynthesis causes severe glomerular proteinuria and is rescued by N-acetylmannosamine. J Clin Invest 2007; 117: 1585–94.

111 Freeze HH. Towards a therapy for phosphomannomutase 2 defi ciency, the defect in CDG-Ia patients. Biochim Biophys Acta 2009; 1792: 835–40.

112 Hutt D, Balch WE. Cell biology. The proteome in balance. Science 2010; 329: 766–67.

113 Hutt DM, Herman D, Rodrigues AP, et al. Reduced histone deacetylase 7 activity restores function to misfolded CFTR in cystic fi brosis. Nat Chem Biol 2010; 6: 25–33.

114 Powers ET, Morimoto RI, Dillin A, Kelly JW, Balch WE. Biological and chemical approaches to diseases of proteostasis defi ciency. Annu Rev Biochem 2009; 78: 959–91.

115 Nemunaitis G, Jay CM, Maples PB, et al. Hereditary inclusion body myopathy: single patient response to intravenous dosing of GNE gene lipoplex. Hum Gene Ther 2011; 22: 1331–41.

116 Cohen-Cymberknoh M, Shoseyov D, Kerem E. Managing cystic fi brosis: strategies that increase life expectancy and improve quality of life. Am J Respir Crit Care Med 2011; 183: 1463–71.

www.thelancet.com/neurology Vol 11 May 2012 405

Articles

Effi cacy and tolerability of lasmiditan, an oral 5-HT1F receptor agonist, for the acute treatment of migraine: a phase 2 randomised, placebo-controlled, parallel-group, dose-ranging studyMarkus Färkkilä, Hans-Christoph Diener, Gilles Géraud, Miguel Láinez, Jean Schoenen, Nadja Harner, Alison Pilgrim, Uwe Reuter, for the COL MIG-202 study group*

SummaryBackground Lasmiditan (COL-144) is a novel, centrally acting, highly selective 5-HT1F receptor agonist without vasoconstrictor activity that seemed eff ective when given as an intravenous infusion in a proof-of-concept migraine study. We aimed to assess the effi cacy and safety of oral lasmiditan for the acute treatment of migraine.

Methods In this multicentre, double-blind, parallel-group, dose-ranging study in 43 headache centres in fi ve European countries, patients with migraine with and without aura and who were not using prophylaxis were randomly assigned (1:1:1:1:1) to treat one moderate or severe attack at home with 50 mg, 100 mg, 200 mg, or 400 mg lasmiditan, or placebo. Study drug and placebo were supplied in identical numbered tablet packs. The randomisation code was generated by an independent statistician. Patients and investigators were masked to treatment allocation. The primary endpoint was dose response for headache relief (moderate or severe becoming mild or none) at 2 h. The primary analysis was done in the modifi ed intention-to-treat population. This study is registered with ClinicalTrials.gov, number NCT00883051.

Findings Between July 8 2009, and Feb 18, 2010, 512 patients were randomly assigned to treatment, 391 of whom received treatment. 86 patients received placebo (81 included in primary analysis) and 305 received lasmiditan (50 mg n=79, 100 mg n=81, 200 mg n=69, and 400 mg n=68 included in primary analysis). There was a linear association between headache response rate at 2 h and lasmiditan dose (Cochran-Armitage test p<0·0001). Every lasmiditan treatment dose signifi cantly improved headache response at 2 h compared with placebo (lasmiditan 50 mg: diff erence 17·9%, 95% CI 3·9–32·1, p=0·022; 100 mg: 38·2%, 24·1–52·4, p<0·0001; 200 mg: 28·8%, 9·6–39·9, p=0·0018; 400 mg: 38·7%, 23·9–53·6, p<0·0001). The proportion of patients with treatment-emergent adverse events increased with increasing doses (53/82 [65%], 59/82 [72%], 61/71 [86%], and 59/70 [84%] for lasmiditan 50, 100, 200, and 400 mg, respectively vs 19/86 [22%] for placebo). Most adverse events were mild or moderate in intensity, with 16 of 82 (20%), 23 of 82 (28%), 28 of 71 (39%), and 31 of 70 (44%) of patients on lasmiditan 50, 100, 200, and 400 mg, respectively reporting a severe adverse event compared with fi ve of 86 (6%) on placebo. The most common adverse events were CNS related and included dizziness, fatigue, vertigo, paraesthesia, and somnolence.

Interpretation Oral lasmiditan seems to be safe and eff ective in the acute treatment of migraine. Further assessment in larger placebo-controlled and triptan-controlled trials are needed to assess the potential role of lasmiditan in acute migraine therapy.

Funding CoLucid Pharmaceuticals.

IntroductionMigraine is one of the most common neurological disorders and is ranked by WHO as one of the 20 most debilitating disorders.1 Although the introduction of 5-HT1B/1D receptor agonists (triptans) has greatly improved acute treatment of migraine, the American Migraine Prevalence and Prevention study2 revealed that 40% of episodic migraineurs still have unmet treatment needs. Headache-related disability (19%) and dissatisfaction with present drugs (15%) were the most frequent complaints.2 In clinical trials, over 35% of patients do not benefi t from treatment with oral triptan formulations.3,4 Because of potential vasoconstriction, patients with cardiovascular disease, uncontrolled hyper tension, and

certain forms of migraine (eg, hemiplegic migraine) cannot use triptans,4,5 and side-eff ects such as chest tightness, throat discomfort, muscle pain, and paraesthesia lead some patients to avoid them.6 There-fore, eff ective treatment options for patients who do not achieve adequate headache relief with triptans or who cannot or will not take them remains a considerable area of unmet clinical need.

5-HT1F receptor agonists are a potential treatment alternative to triptans.7 The expression of 5-HT1F receptor mRNA in neurons of the trigeminal ganglia led to the suggestion that 5-HT1F receptors could be a therapeutic target for migraine.8 Lasmiditan, a highly selective 5-HT1F agonist, has 470 times higher affi nity for 5-HT1F

Lancet Neurol 2012; 11: 405–13

Published OnlineMarch 28, 2012 DOI:10.1016/S1474-4422(12)70047-9

See Comment page 383

*Members listed at end of paper

University of Helsinki, Department of Neurology, Helsinki, Finland (Prof M Färkkilä MD); University of Duisburg–Essen, Department of Neurology, Essen, Germany (Prof H-C Diener MD); CHU Toulouse, Pôle des Neurosciences, Service de Neurologie, Hôpital Rangueil, Toulouse, France (Prof G Géraud MD); Hospital Clínico Universitario University of Valencia Department of Neurology, Valencia, Spain (Prof M Láinez MD); University of Liège, CHR Citadelle, Department of Neurology, Liège, Belgium (Prof J Schoenen MD); FGK Clinical Research, Munich, Germany (N Harner MSc); CoLucid Pharmaceuticals, Durham, NJ, USA (A Pilgrim BM); and Charité Universitätsmedizin Berlin, Department of Neurology, Berlin, Germany (U Reuter MD)

Correspondence to:Dr Uwe Reuter, Charité Universitätsmedizin Berlin, Department of Neurology, Charitéplatz 1, 10117 Berlin, Germanyuwe.reuter@charite.de

Articles

406 www.thelancet.com/neurology Vol 11 May 2012

receptors than for vasoconstrictor 5-HT1B receptors.9 Administration of lasmiditan inhibited neurogenic infl ammation in the dura and decreased c-Fos expression in the trigeminal nucleus caudalis after stimulation of the trigeminal ganglion in rats; unlike triptans, lasmiditan did not cause constriction of rabbit saphenous vein—an assay predictive of human coronary artery vasoconstriction.9

A proof-of-concept randomised, multicentre, placebo-controlled trial with 130 patients showed that intravenous doses of lasmiditan of 20 mg and higher provided eff ective headache relief at 2 h of an acute migraine attack.10 However, migraine is usually self-treated on an outpatient basis. Therefore, an oral formulation of lasmiditan was developed. In otherwise healthy patients, oral lasmiditan doses up to 400 mg were well tolerated without clinically signifi cant eff ects on vital signs, electrocardiogram (ECG), or laboratory param eters.11 We therefore undertook a dose-ranging study to assess the effi cacy and safety of oral lasmiditan for the acute treatment of migraine.

MethodsPatientsWe undertook a randomised, double-blind, placebo-controlled, multi centre, parallel-group, dose-ranging out patient study in patients with acute migraine from 43 headache centres in fi ve European countries. Men or women (18–65 years) who had at least a 1-year history of

migraine with or without aura (according to International Headache Society criteria 1.1 and 2.1)12 with onset before the age of 50 years and one to eight migraine attacks per month were eligible for enrolment. Exclusion criteria included patients taking prescription or herbal migraine prophylaxis, vasoactive drugs, serotonin reuptake inhibitors, or known cyto chrome P450 inhibitors. Prescription preventative migraine drugs were dis continued at least 15 days (fl unarizine 30 days) before screening. By pharma cokinetic/pharma-codynamic (PK/PD) modelling, we selected doses for the study, with 50 mg predicted to have minimal effi cacy and 400 mg to have both high effi cacy and a rapid onset of eff ect.13 The rapidly disintegrating lasmiditan tablets used in this study achieve maximum plasma con-centrations at 2·0–2·5 h.

The study was approved by the relevant authorities and independent ethics committees. This study was done in accordance with the Declaration of Helsinki and internationally accepted standards of Good Clinical Practice. All patients gave written informed consent before enrolment.

Randomisation and maskingUsing a randomisation code generated by an independent statistician, patients were randomly assigned (1:1:1:1:1) to 50 mg, 100 mg, 200 mg, or 400 mg lasmiditan, or placebo in blocks of fi ve. Treatment was double-blind, with all patients receiving numbered drug packs that were

103 patients assigned placebo

17 did not use study drug

86 patients included in safety analysis

2 did not consume all 4 tablets1 severity at baseline not moderate or severe2 used other drugs before study drug

81 patients included in primary analysis (MITT)

106 patients assigned 50 mg lasmiditan

24 did not use study drug

82 patients included in safety analysis

1 did not consume all 4 tablets1 severity at baseline not moderate or severe1 used other drugs before study drug

79 patients included in primary analysis (MITT)

104 patients assigned 100 mg lasmiditan

22 did not use study drug

82 patients included in safety analysis

1 used other drugs before study drug

81 patients included in primary analysis (MITT)

100 patients assigned 200 mg lasmiditan

29 did not use study drug

71 patients included in safety analysis

1 did not consume all 4 tablets1 severity at baseline not moderate or severe

69 patients included in primary analysis (MITT)

99 patients assigned 400 mg lasmiditan

29 did not use study drug

70 patients included in safety analysis

2 used other drugs before study drug

68 patients included in primary analysis (MITT)

534 patients screened

22 failed screening

512 randomly assigned

Figure 1: Trial profi leMITT=modifi ed intention-to-treat population.

Articles

www.thelancet.com/neurology Vol 11 May 2012 407

identical in appearance. All patients and investigators, excluding the independent statistician, were masked to treatment allocation.

ProceduresAt screening, medical and migraine history were taken, and physical examination, ECG, and laboratory tests were done outside a migraine attack for all patients. Eligible

patients were instructed to treat their next migraine attack within 4 h of onset, providing that any aura symptoms had resolved and their headache was moderate or severe. Rescue drugs (excluding triptans or ergotamine) could be taken after 2 h. Patients were allowed 8 weeks to treat an attack.

The primary objectives were to assess the lasmiditan dose relation for headache relief at 2 h after intake of the

Placebo (n=81); Response n (%; 95% CI)

Lasmiditan

50 mg (n=79) 100 mg (n=81) 200 mg (n=69) 400 mg (n=68)

Response n (%; 95% CI)

Diff erence (95% CI)

p value*

Response n (%; 95% CI)

Diff erence (95% CI)

p value*

Response n (%; 95% CI)

Diff erence (95% CI)

p value*

Response n (%; 95% CI)

Diff erence (95% CI)

p value*

Headache response at 2 h

21 (25·9%; 16·8–36·9)

34 (43%; 31·9–54·7)

17·9% (3·9–32·1)

0·022 52 (64%; 52·8–74·6)

38·2% (24·1–52·4)

<0·0001 35 (51%; 38·4–63·0)

28·8% (9·6–39·9)

0·0018 44 (65%; 52·2–75·9)

38·7% (23·9–53·6)

<0·0001

Pain free at 2 h 6 (7·4%; 2·8–15·4)

11 (14%; 7·2–23·5)

6·5% (3·0–16·0)

0·18 11 (14%; 7·1–23·3)

6·3% (3·1–15·8)

0·19 13 (19%; 10·6–30·5)

11·7% (0·1–22·6)

0·032 19 (28%; 18·0–40·7)

21·9% (8·8–33·1)

0·0007

Headache recurrence within 24 h

12 (57·1%; 34·0–78·2)

19 (56%; 37·9–78·8)

1·33% (–25·7 to 28·2)

0·93 30 (58%; 43·2–71·3)

–0·5% (–25·6 to 24·5)

0·97 22 (63%; 48·1–82·0)

9·5% (–35·8 to 16·8)

0·48 22 (50%; 34·6–65·4)

7·1% (–18·6 to 32·9)

0·59

Rescue drug 2–24 h

55 (68·8%; 57·4–78·7)

42 (55%; 42·8–65·9)

14·2% (–0·9 to 29·3)

0·067 42 (52%; 40·5–63·1)

16·9% (2·0–31·8)

0·029 41 (61%; 48·5–72·9)

7·6% (–7·9–23·0)

0·34 28 (42%; 29·8–54·5)

26·7% (11·4–42·5)

0·001

Patients’ global impression (much or very much better) at 2 h

13 (16·0%; 8·8–25·9)

18 (23%; 14·1–33·6)

–6·74% (–18·9 to 5·5)

0·28 29 (36%; 25·4–47·2)

–19·7% (–32·9 to –6·6)

0·0041 19 (28%; 17·5–39·6)

–11·5% (–24·7 to 1·7)

0·087 23 (34%; 23·2–46·9)

–18·3% (–32·1 to 4·4)

0·0099

Clinical disability score at 2 h†

81 (2·0; 1·7–2·2)

79 (1·5; 1·3–1·8)

0·4 (0·1–0·7)

0·01 78 (1·4; 1·1–1·6)

0·6 (0·3–0·9)

0·0002 66 (1·5; 1·2–1·8)

0·5 (0·1–0·8)

0·0081 63 (1·5; 1·2–1·7)

0·5 (0·2–0·8) 0·0039

Headache severity at 2 h†

81 (2·1; 1·9–2·41)

79 (1·7; 1·5–1·9)

0·4 (0·1–0·7)

0·014 80 (1·3; 1·1–1·5)

0·8 (0·5–1·1)

<0·0001 68 (1·5; 1·3–1·8)

0·6 (0·3–0·9)

0·0003 67 (1·2; 0·9–1·4)

0·9 (0·6–1·2)

<0·0001

*For comparison with placebo. †For clinical disability and headache severity, data are mean (SD; 95% CI). Because the eff ects after 2 h were central to the primary endpoint, we only show data for this timepoint.

Table 2: Primary and secondary endpoints

Placebo (n=86) Lasmiditan

50 mg (n=82) 100 mg (n=82) 200 mg (n=71) 400 mg (n=70)

Age (years) 40·5 (10·3; 19–66) 40·4 (12·5; 18–65) 42·0 (10·6; 20–65) 39·5 (10·3; 18–57) 38·7 (10·3; 20–60)

Female sex 75 (87%) 69 (84%) 68 (83%) 65 (92%) 65 (93%)

White ethnic origin 86 (100%) 81 (99%) 81 (99%) 70 (99%) 69 (99%)

Migraine frequency (past 3 months) 3·1 (1·7) 3·3 (1·6) 3·3 (1·7) 3·3 (1·9) 3·1 (1·6)

Duration of treated attack before use of study drug (h) 2·2 (0·0–31·8) 1·8 (0·0–19·0) 2·8 (0·0–15·0) 2·3 (0·0–15·0) 2·1 (0·0–19·8)

Duration of moderate-to-severe headache before treatment (h)

0·2 (0·0–7·5) 0·1 (0·0–2·1) 0·2 (0·0–3·1) 0·3 (0·0–1·9) 0·1 (0·0–2·3)

Accompanying aura*†

No 79 (92%) 70 (85%) 76 (93%) 68 (96%) 62 (89%)

Yes 6 (7%) 11 (13%) 5 (6%) 3 (4%) 8 (11%)

Severity*

Moderate 51 (59%) 49 (60%) 49 (60%) 36 (51%) 39 (56%)

Severe 34 (40%) 32 (39%) 33 (40%) 34 (48%) 31 (44%)

Data are mean (SD; range), number (%), mean (SD), or median (range). *Some percentages do not add up to 100% because of missing data. †If the migraine attack was accompanied by aura, the study drug was not taken until the aura had resolved.

Table 1: Baseline demographics and clinical characteristics

Articles

408 www.thelancet.com/neurology Vol 11 May 2012

study drug and to assess the safety of lasmiditan over 24 h. Secondary endpoints were headache response over time, proportion of patients who were pain free (ie, absence of headache) at 2 h, associated symptoms, time to meaningful pain relief, headache recurrence within 24 h, clinical disability within 24 h, use of rescue drugs between 2 and 24 h, and patients’ global impression at 2 h. The safety objective was to assess the safety and tolerability of lasmiditan in terms of adverse events, physical examination, vital signs, laboratory tests, and ECGs.

Patients recorded migraine symptoms in a standardised paper diary immediately before and 0·5, 1, 1·5, 2, 3, 4, and 24 h after intake of study drug. Headache severity and clinical disability were rated on a four-point scale (none, mild, moderate, and severe). Headache response was defi ned as a reduction of moderate or severe pain to mild or no pain. Patients also recorded the date and time when they experienced meaningful relief of migraine.

Associated symptoms (nausea, vomiting, phono-phobia, and photophobia) were each rated as present or absent. Patients recorded their global impression at 2 h after study drug intake on a seven-point scale (very much better, much better, a little better, no change, a little worse, much worse, and very much worse). Patients also recorded any unusual symptom (possible adverse event) within the treatment period (24 h). Adverse events were graded mild, moderate, or severe according to the judgment of the investigator, and a causal relation was assessed by the investigator.

At follow-up within 14 days after treatment, patients returned their completed diary card and study drug pack. A physical examination, vital signs, 12-lead ECG, and

laboratory assessments were done and adverse events, concomitant drugs, and rescue drugs were recorded.

Because each patient received only one dose of lasmiditan within a range that had been well tolerated in phase 1 studies, the principal investigators and ethics committees did not deem a data safety monitoring board to be necessary. The trial was expected to recruit so rapidly that by the time a signifi cant amount of data were available for review by a data safety monitoring board the trial would be near to completion. Instead, to safeguard patient safety all serious adverse events were submitted urgently to a medical monitor for review and action.

Statistical analysisThe sample size was estimated assuming a response rate of 40% in the placebo group and 65% in the 400 mg lasmiditan group on the basis of data from previous intravenous studies.10,13 We assumed that the treatment groups were equally spaced (i.e. the dose response was linear) and that the response odds ratios (ORs) between pairs of adjacent dose groups were equal, and thus estimated the sample size needed to test for a linear association by the method of Nam.14 Based on a 1:1:1:1:1 randomisation, a total sample size of 330 evaluable patients (66 per group) was needed for 90% power, on the basis of a two-sided test at the 5% level of signifi cance.

Patients who did not take study drugs because of occurrence of no or mild headaches, did not record baseline headache severity, did not take all study drug, or took other migraine drugs fi rst were excluded from the modifi ed intention-to-treat population, as prespecifi ed for the primary analysis. All patients who received any study drugs were included in the safety analysis.

For all tests, a two-sided signifi cance level of 5% was applied. A hierarchical test procedure was done for the primary analysis: we used the Cochran-Armitage test for trend to calculate whether there was a linear association between response rate and dose and, if signifi cant, we analysed individual between-treatment diff erences with Pearson’s χ² tests starting with lasmiditan 400 mg versus placebo, then 200 mg and 100 mg versus placebo, followed by 50 mg versus 400 mg, and fi nally 50 mg versus placebo. Each test was done only if the previous test was statistically signifi cant.

Patients who took rescue drugs within the fi rst 2 h or failed to record headache severity at 2 h were assumed to have had no headache response. All secondary endpoint analyses were exploratory and were done with a two-sided test at the 5% level of signifi cance.

Headache freedom and associated symptoms were analysed by similar methods to the primary analysis except that the comparison of 50 mg versus 400 mg lasmiditan was not done. For headache severity, clinical disability, and patients’ global impression at 2 h after treatment, we used the Cochran-Mantel-Haenszel (CMH) mean score test to compare placebo and each dose of

Number at riskPlacebo 81 80 76 67 32 19 15 11 11

50 mg lasmiditan 79 78 72 53 33 24 20 15 13 100 mg lasmiditan 81 77 66 45 22 14 6 4 3 200 mg lasmiditan 69 66 56 36 22 12 10 8 8 400 mg lasmiditan 68 61 50 35 20 9 6 3 3

Dist

ribut

ion

func

tion

1·0

0·8

0·6

0·4

0·2

0

Time to meaningful pain relief after dosing (h)

10842 60 11953 71 12

Placebo50 mg lasmiditan100 mg lasmiditan200 mg lasmiditan400 mg lasmiditan

0·5

Figure 2: Time to meaningful pain relief

Articles

www.thelancet.com/neurology Vol 11 May 2012 409

lasmiditan. We made comparisons across multiple dose levels with the CMH correlation test. To analyse and display time to meaningful pain relief, we did a Kaplan-Meier analysis. Diff erences among treatment groups and diff erences between placebo and each dose of lasmiditan were compared with the log-rank test.

This study is registered with ClinicalTrials.gov, number NCT00883051.

Role of the funding sourceThe study was designed by the principal investigators together with the sponsor (CoLucid Pharmaceuticals). The sponsor participated in data collection, data analysis, data interpretation, and the writing of the report. All authors had full access to all the data in the study and reviewed the paper. The corresponding author had fi nal responsibility for the decision to submit for publication.

ResultsBetween July 8, 2009, and Dec 22, 2009, 534 patients were screened, of whom 512 were randomly assigned to treatment. The study fi nished on Feb 18, 2010. 121 patients did not use the study drug and one patient used the study drug but was lost to follow-up. The remaining 390 patients fi nished the study. 13 patients were excluded from the primary analysis for pre-specifi ed protocol violations (fi gure 1). Table 1 shows patient demographics and features of the treated migraine attacks. Baseline headache characteristics were broadly similar across groups. However, the proportion of patients with severe headache was higher in the 200 mg lasmiditan group than in all other active treatment groups.

There was a signifi cant linear association between headache response rate and lasmiditan dose (Cochran-Armitage test, p<0·0001). Every lasmiditan treatment dose signifi cantly improved headache response at 2 h compared with placebo (table 2; appendix). Signifi cantly more patients in the 400 mg lasmiditan group than in the 50 mg group reported a headache response at 2 h (diff erence 21·7%, 95% CI 5·9–37·4; p=0·0087).

A linear association was also noted between headache-free rates at 2 h and lasmiditan dose (Cochran-Armitage test, p=0·0006). Both the 200 mg (diff erence 11·7%) and 400 mg (21·9%) doses of lasmiditan were superior

Patie

nts (

%)

25

20

15

5

10

0

Time (h)

2·01·00·5 1·50·0

2·01·00·5 1·50·0

2·01·00·5 1·50·0

2·01·00·5 1·50·0

Patie

nts (

%)

80

60

20

40

10

90

70

30

50

B Photophobia

A Nausea

D Vomiting

C Phonophobia

0Pa

tient

s (%

)

80

60

20

40

10

90

70

30

50

0

Patie

nts (

%)

80

60

20

40

10

90

70

30

50

0

Placebo50 mg lasmiditan100 mg lasmiditan200 mg lasmiditan400 mg lasmiditan

*

†

‡

¶**||§

††‡‡

§§

¶¶ ||||***

Figure 3: Migraine-associated symptoms*p=0·034 for comparison between 100 mg lasmiditan and placebo at 2 h.

†p=0·0005 for comparison between 100 mg lasmiditan and placebo at 1·5 h. ‡p=0·015 for comparison between 400 mg lasmiditan and placebo at 1·5 h.

§p=0·018 for comparison between 50 mg lasmiditan and placebo at 2 h. ¶p<0·0001 for comparison between 100 mg lasmiditan and placebo at 2 h.

||p=0·031 for comparison between 200 mg lasmiditan and placebo at 2 h. **p=0·0006 for comparison between 400 mg lasmiditan and placebo at 2 h. ††p=0·018 for comparison between 100 mg lasmiditan and placebo at 1·5 h. ‡‡p=0·0013 for comparison between 100 mg lasmiditan and placebo at 2 h.

§§p=0·019 for comparison between 400 mg lasmiditan and placebo at 2 h. ¶¶p=0·018 for comparison between 100 mg lasmiditan and placebo at 1·5h.

||||p=0·0088 for comparison between 400 mg lasmiditan and placebo at 1·5 h. ***p=0·0027 for comparison between 100 mg lasmiditan and placebo at 2 h.

Articles

410 www.thelancet.com/neurology Vol 11 May 2012

to placebo (200 mg p=0·032; 400 mg p=0·0007; Pearson’s χ² test).

Lasmiditan reduced headache severity starting as early as 30 min in the 400 mg group versus placebo (CMH mean score test, p=0·0137). After 1 h, all but the lowest dose of lasmiditan (50 mg) were superior to placebo, and from 1·5 to 4 h all lasmiditan groups were superior to placebo. Likewise, there were statistically signifi cant diff erences between each lasmiditan dose group and placebo for time to meaningful pain relief (50 mg p=0·0294, 100 mg p<0·0001, 200 mg p=0·0003, 400 mg p<0·0001; log-rank test). Figure 2 shows data for the fi rst 12 h after dosing, whereas the log-rank test was based on the full data up to 24 h.

There was a dose-related reduction in use of rescue drugs in the lasmiditan groups (CMH for linear association of rescue drug intake with dose p=0·0093). A global impression rating of much or very much better was obtained from 16·0% of patients in the placebo group compared with 22·8–35·8% of patients in the lasmiditan groups (linear association with dose; CMH correlation test, p=0·0162). Similar rates of headache recurrence were reported in all groups (50–63%; table 2). In a post-hoc analysis, there was a statistically signifi cant linear association between the decrease in severity of clinical disability and increasing dose of lasmiditan from 2 h after treatment onwards (CMH correlation test, p=0·0036).

Nausea, phonophobia, and photophobia decreased in all treatment groups within 2 h after intake of study drug, with the smallest decrease in the placebo group (fi gure 3). The greatest improvements after 2 h were achieved for phonophobia and photophobia with the 100 mg and 400 mg doses of lasmiditan. The proportion of patients with vomiting was low in all groups (about 0–10%) and therefore diff erences over time and between groups for this symptom must be interpreted with caution.

The study drug or placebo was taken by 391 patients, who were all included in the safety analysis. In general,

lasmiditan was well tolerated. There were no deaths in the study and ECGs, vital signs, and laboratory assessments did not show any clinically relevant drug-related changes.

The proportion of patients who reported at least one adverse event and the proportion of patients with treatment-emergent adverse events were higher in the active treatment groups than in the placebo group. Treatment-emergent adverse events increased with increasing doses (53/82 [65%], 59/82 [72%], 61/71 [86%], 59/70 [84%] for lasmiditan 50, 100, 200, and 400 mg, respectively vs 19/86 [22%] for placebo). The most frequently reported treatment-emergent adverse events (table 3) were associated with the CNS (eg, dizziness, paraesthesia) or the vestibular system (eg, vertigo). The appendix lists treatment-emergent adverse events by country.

Most adverse events were mild or moderate in intensity, with 16 of 82 (20%), 23 of 82 (28%), 28 of 71 (39%), and 31 of 70 (44%) patients in the lasmiditan 50, 100, 200, and 400 mg groups, respectively, reporting a severe adverse event compared with fi ve of 86 (6%) on placebo. Dizziness was the most frequently reported severe adverse event.

A 46-year-old woman reported moderate dizziness 30 min after taking 200 mg lasmiditan. Because this led to an overnight hospital admission, the adverse event was classifi ed as serious. Her ECGs showed sinus bradycardia 1·5 and 4 h after study drug intake but no other abnormalities. She received a saline infusion and had recovered completely by the next day.

DiscussionThis trial with oral lasmiditan confi rms the results of the previous proof-of-concept trial10 with the intravenous formulation, suggesting that 5-HT1F receptor activation can dose-dependently improve acute migraine (panel). Dose-dependent effi cacy was also noted in a phase 2 study with a less selective 5-HT1F agonist, LY334370.15 However, a vascular eff ect contributing to increased effi cacy of high doses of LY334370 could not be entirely ruled out because of the affi nity of LY334370 itself and of its major metabolite for the 5-HT1B receptor. The affi nity

Placebo (n=86) Lasmiditan

Treatment emergent

Severe 50 mg (n=82) 100 mg (n=82) 200 mg (n=71) 400 mg (n=70)

Treatment emergent

Severe Treatment emergent

Severe Treatment emergent

Severe Treatment emergent

Severe

Sensation of heaviness 1 (1%) 1 (1%) 4 (5%) 3 (4%) 4 (5%) 1 (1%) 7 (10%) 2 (3%) 5 (7%) 3 (4%)

Nausea 0 (0%) 0 (0%) 4 (5%) 2 (2%) 8 (10%) 0 (0%) 2 (3%) 1 (1%) 5 (7%) 0 (0%)

Paraesthesia 2 (2%) 0 (0%) 2 (2%) 1 (1%) 9 (11%) 2 (2%) 12 (17%) 4 (6%) 14 (20%) 5 (7%)

Somnolence 2 (2%) 1 (1%) 8 (10%) 3 (4%) 10 (12%) 2 (2%) 8 (11%) 2 (3%) 8 (11%) 2 (3%)

Vertigo 1 (1%) 0 (0%) 8 (10%) 1 (1%) 12 (15%) 3 (4%) 12 (17%) 3 (4%) 16 (23%) 7 (10%)

Fatigue 2 (2%) 1 (1%) 10 (12%) 5 (6%) 17 (21%) 7 (9%) 15 (21%) 11 (15%) 16 (23%) 7 (10%)

Dizziness 0 (0%) 0 (0%) 19 (23%) 1 (1%) 21 (26%) 8 (10%) 27 (38%) 11 (15%) 26 (37%) 12 (17%)

Data are number (%).

Table 3: Most commonly reported treatment-emergent and severe adverse events

See Online for appendix

Articles

www.thelancet.com/neurology Vol 11 May 2012 411

of lasmiditan for the 5-HT1B receptor is signifi cantly lower than that of LY334370 and surrogate assays did not show vasoconstrictive activity, although the precise mechanism of action remains to be identifi ed.9 5-HT1F receptors are expressed in trigeminal ganglion neurons, and activation of these neurons might contribute to the inhibition of protein leakage from venous blood vessels, probably owing to the blockade of neuropeptide release.16 The blockade of secondary trigeminal neuron activation within the CNS might also contribute to this inhibition.9 Further sites of action remain speculative because 5-HT1F receptor distribution in the brain is diffi cult to map owing to the absence of a specifi c antagonist. However, studies with ³H-LY334370 show that in human beings the 5-HT1F receptor is present mainly in cortical areas (frontal, temporal, parietal, and occipital cortices) and in the granule cell layer of the cerebellum.17 Some of these cortical areas have been linked to acute pain, but whether 5-HT1F receptor binding of lasmiditan in these regions contributes to the anti-migraine activity is unclear.

The primary endpoint of this study was met: lasmiditan improved moderate-to-severe migraine headache to mild or none at 2 h in a dose-dependent manner. Although the sample size was small, which is a shortcoming of this trial, statistically signifi cant diff erences between each lasmiditan dose and placebo were noted. The benefi cial eff ects of lasmiditan on migraine were supported by the secondary endpoints pain freedom, headache intensity, associated symptoms, and patients’ global impression. The eff ect of 200 mg lasmiditan was lower for the primary endpoint and for some secondary endpoints than that of the 100 mg dose, which might be due to small sample sizes and random variation in migraine attack severity and response. Headache severity at baseline was higher in the 200 mg group than in all other groups. However, in a post-hoc logistic regression analysis for the primary endpoint, with treatment group and severe migraine headache just before dosing (yes/no) as covariates, there was a signifi cant eff ect of severe headache (OR estimates for no headache response: 1·75 [1·13–2·70]), but adjustment for this result did not explain the lower eff ect of 200 mg lasmiditan (OR estimates for 50 mg, 100 mg, 200 mg, and 400 mg doses vs placebo 0·46 [0·23–0·90], 0·19 [0·096–0·37], 0·32 [0·16–0·64], and 0·18 [0·09–0·37], respectively).

Analyses of secondary endpoints were exploratory and we did not adjust for multiple comparisons. The results should therefore be interpreted with caution. Lasmiditan reduced migraine-associated symptoms (nausea, photo-phobia, and phonophobia) at 2 h, with the strongest eff ects with the 100 mg and 400 mg doses. Relief of these symptoms might be underestimated in this analysis because the use of rescue drugs was conservatively treated as failure, which might have contributed to the absence of statistical signifi cance versus placebo for some parameters and timepoints, especially after 2 h.

In a study of the 5-HT1F receptor agonist LY334370,15 CNS side-eff ects were dose-dependent and seemed to be more frequent with lasmiditan than with triptans.18 However, chest, neck, and jaw heaviness, tightness, or pain, reported by up to a quarter of patients taking an oral triptan,19 were uncommon and no more frequent after lasmiditan than placebo. Dizziness was the main treatment-emergent complaint attributed to the CNS, followed by vertigo and fatigue. Vertigo and dizziness might be related to the activation of 5-HT1F receptors in the lateral vestibular nucleus, temporoparietal cortex, and cerebellum, because 5-HT1F receptor expression has been detected in these areas in rodents.20,21 In the human brain, radioactive ligands bind signifi cantly to 5-HT1F receptors in the cerebellum, a structure that is strongly linked to the vestibular system.17 However, the diff erences between countries in the rates of vertigo and dizziness suggest that cultural and linguistic factors might have led patients to confuse the two events, with possible over-reporting of vertigo. Modifi cation of the adverse event data collection procedure in future studies should resolve this issue. The CNS side-eff ects are unlikely to have been mediated by 5-HT1A receptor activation, as has been suggested for LY334370,15 because the 5-HT1A affi nity of lasmiditan is extremely low.9

This study provides important information for dose selection for phase 3 clinical trials. The lowest dose of lasmiditan in this trial was 50 mg and was expected to be ineff ective or only marginally eff ective. Based on PK/PD modelling of the intravenous data described by Ferrari and colleagues,10 the peak plasma concentration with this oral dose is about 30 ng/mL and higher plasma concentrations were expected to be necessary to achieve effi cacy.10,13 However, 50 mg lasmiditan seems superior to placebo in this study, suggesting that lower oral doses might be

Panel: Research in context

Systematic reviewWe searched Medline (1950 to December, 2011), the Cochrane Central Register of Controlled Trials (The Cochrane Library issue 12, 2011), and Embase (1988 to December, 2011) with the search terms “placebo-controlled, randomised, double-blind clinical trials”, “acute migraine”, “migraine treatment”, “5-HT1F agonist”; “CGRP receptor antagonist”, “triptans”, “adverse events”, and “triptan” alone and in several combinations. We included results from placebo-controlled, randomised, double-blind clinical trials with 5-HT1F receptor agonists, calcitonin gene-related peptide receptor antagonists, and triptans for the acute treatment of migraine. We also assessed meta-analyses of controlled triptan trials and articles that discussed the adverse-event profi les of triptans.

InterpretationOur study confi rms that selective activation of 5-HT1F receptors with oral or intravenously administered agonists without vasoconstrictive activity reduces headache severity in migraine attacks compared with placebo. Both effi cacy and nervous system-related adverse eff ects showed a clear dose response. The adverse-event profi le of lasmiditan in this trial is similar to those of a previous study with an intravenous formulation and a study with a less selective 5-HT1F agonist (LY334370), and is distinctly diff erent from that of triptans.10,15 However, long-term safety needs to be established.

Articles

412 www.thelancet.com/neurology Vol 11 May 2012

suffi cient for pain relief in some patients. The signifi cant headache response at 2 h and the favourable adverse-event profi le compared with higher doses supports the use of a lasmiditan dose of 100 mg for future clinical trials.

We have shown that in migraineurs selective 5-HT1F agonism with lasmiditan results in a greater reduction in headache response than placebo. All oral lasmiditan doses seemed more eff ective than placebo and the 100 mg dose produced headache response rates comparable with established treatment options. The placebo-subtracted headache response rate of lasmiditan is comparable to that reported with oral triptans.3 Also, pain relief after 2 h seems to be similar to results from trials with oral calcitonin gene-related peptide receptor antagonists.22,23

The neural site of action and absence of vasoconstrictor activity might be of benefi t in clinical practice, where a substantial proportion of patients are unable to take triptans or are poorly or inconsistently responsive to them. Adverse events with lasmiditan were qualitatively diff erent from those reported with triptans. The typical triptan sensations such as chest or neck pain, tightness, or heaviness were rare and occurred with similar frequency after placebo and lasmiditan. The extent to which the CNS adverse events might limit use, and the place of lasmiditan in treatment relative to triptans, need to be studied in larger comparator trials that more closely resemble clinical practice. Our results suggest that non-vascular mechanisms are suffi cient for the treatment of acute migraine and thereby support the notion of migraine as a neuronal rather than a vascular disease.

ContributorsAll authors contributed to the study design, the study analysis, and the

interpretation of the results. MF, H-CD, GG, ML, JS, and UR treated

patients. UR wrote the fi rst draft of the manuscript. UR and AP wrote

most of the manuscript. NH was responsible for the statistical analysis

plan and analysis.

COL MIG-202 study groupBelgium N de Klippel (Hospital Hasselt, Hasselt), L Herroelen

(University Leuven, Leuven), B Vandersmissen, M Pandolfo (Hospital

Erasme, Brussels), J Schoenen (University Liege, Liege),

M Van Zandijcke (Sint Jan AV, Bruges). France G Géraud (principal

investigator [PI]) M Barege (University of Toulouse, Toulouse), P Giraud

(Hospital Region Anneciennce, Pringy Cedex), E Guegan-Massadier

(Hospital Charles Nicolle, Rouen), M Lanteri-Minet (PI), H Alchaar

(University of Nice, Nice), C Lucas (Hospital Salengro, CHRU Lille),

D Valade (Hospital Lariboisiere, Paris). Finland M Färkkila (PI),

M Kallela (Meilahti Hospital, Helsinki), M Ilmavirta (Central Hospital,

Jyväskylö), T Jolma (Suomen Erikoisneurologiakeskus, Pori),

E Kinnunen (Lääkärikeskus Pipetti, Hyvinkää), J Liukkonen (Mikkelin

Neurologipalvelu, Mikkeli), E Säko (PI), M Nissila (Suomen Terveystalo

Clinical Research Oy, Turku), M L Sumelahti (Suomen Terveystalo

Operon, Tampere). Germany A Beckmann-Reinholdt (private practice,

Hamburg), R Brandt (Migraine Clinic, Königstein), H-C Diener (PI),

C Gaul (University of Duisburg-Essen, Essen), S Evers (University of

Münster, Münster), H Kaube (University of Freiburg, Freiburg),

K Längler (private practice, Erkelenz), A May (University of Hamburg,

Hamburg), T Nolte, R Brosius (private practice, Wiesbaden), A Peikert

(private practice, Bremen), V Pfaff enrath (private practice, Munich),

M Ribbat (private practice, Itzehoe), C Riemasch-Becker (private

practice, Wiesbaden), U Reuter (PI), J Hoff mann, (Charité

Universitätsmedizin Berlin, Berlin), H Staudenmayer (private practice,

Göttingen), A Straube (University of Munich, Munich), T M Wallasch

(private practice, Berlin). Spain S Diaz-Insa (PI), G Belandria (Hospital

Francesc de Borja, Valencia), T Temprano Fernandez (Hospital Central

de Asturias, Oviedo), J G Menacho (University Hospital Sant Joan,

Reus), P Irimia (University of Navarra, Pamplona), D Jimenez (Virgen

del Rocío Hospital, Sevilla), M Láinez (University of Valencia), R Leira

(Hospital Clínico Universitario, Santiago de Compostella), J Pareja

(Hospital Universitario Fundación Alcorcón, Alcorcón), P Pozo-Rosich

(Hospital Universitario Vall d’Hebron, Barcelona).

Confl icts of interestMF has received consultation fees and travel grants from CoLucid.

H-CD has received honoraria for participation in clinical trials,

contribution to advisory boards, or oral presentations from Addex

Pharma, Allergan, Almirall, AstraZeneca, Bayer Vital, Berlin Chemie,

Boehringer Ingelheim, Bristol-Myers Squibb, Coherix, CoLucid,

GlaxoSmithKline, Grünenthal, Janssen-Cilag, Lilly, Roche, 3M Medica,

Medtronic, Menarini, Minster, MSD, Neuroscore, Novartis, Johnson and

Johnson, Pierre Fabre, Pfi zer, Schaper and Brümmer, Sanofi , St Jude, and

Weber and Weber. H-CD has received fi nancial support for research

projects from Allergan, Almirall, AstraZeneca, Bayer, GlaxoSmithKline,

Janssen-Cilag, MSD, Pfi zer, German Research Council (DFG), BMBF,

and the EU. GG has received consulting honoraria and travel grants from

AstraZeneca, Allergan SA, Pfi zer, MSD, and Ménarini. ML has received

compensation or research support for activities with Allergan, Almirall,

ATI, Boston Scientifi c, CoLucid, GlaxoSmithKline, Ferrer International,

Janssen-Cilag, Merck and Co, Medtronic, Pfi zer, and Servier SA. JS is a

consultant for Autonomic Technologies and STX-Med. JS has contributed

to advisory boards for CoLucid, Allergan, Bristol-Myers Squibb, St Jude,

and ATI. NH is employed by FGK Clinical Research. AP is a consultant

for CoLucid Pharmaceuticals. UR has received honoraria for participation

in clinical trials, contribution of advisory boards, and oral presentations

from Addex Pharma, Allergan, Almirall, Boehringer Ingelheim, CoLucid,

and Jansen-Cilag. Headache research in the Department of Neurology at

Charité Universitätsmedizin Berlin is supported by the BMBF, Johnson

and Johnson, and Vasopharm.

References1 WHO. World Health Report 2001. Geneva: World Health

Organization, 2001.

2 Lipton RB, Buse DC, Serrano D, Ng-Mak DS, Pearlman SH, Reed M. Examination of unmet treatment needs among persons with episodic migraine: results of the American Migraine Prevalence and Prevention study (P1342). Eur J Neurol 2011; 18 (suppl 2): 66–343.

3 Ferrari MD, Goadsby PJ, Roon KI, Lipton RB. Triptans (serotonin 5-HT1B/1D agonists) in migraine: detailed results and methods of a meta-analysis of 53 trials. Cephalalgia 2002; 22: 633–58.

4 Tepper SJ, Millson D. Safety profi le of the triptans. Expert Opin Drug Saf 2003; 2: 123–32.

5 MaasenVanDenBrink A, Reekers M, Bax WA, Ferrari MD, Saxena PR. Coronary side-eff ect potential of current and prospective antimigraine drugs. Circulation 1998; 98: 25–30.

6 Gallagher RM, Kunkel R. Migraine medication attributes important for patient compliance: concerns about side eff ects may delay treatment. Headache 2003; 43: 36–43.

7 Neeb L, Meents J, Reuter U. 5-HT1F receptor agonists: a new treatment option for migraine attacks? Neurotherapeutics 2010; 7: 176–82.

8 Adham N, Bard JA, Zgombick JM, et al. Cloning and characterization of the guinea pig 5-HT1F receptor subtype: a comparison of the pharmacological profi le to the human species homolog. Neuropharmacology 1997; 36: 569–76.

9 Nelson DL, Phebus LA, Johnson KW, et al. Preclinical pharmacological profi le of the selective 5-HT1F receptor agonist lasmiditan. Cephalalgia 2010; 30: 1159–69.

10 Ferrari MD, Färkkilä M, Reuter U, et al. Acute treatment of migraine with the selective 5-HT1F receptor agonist lasmiditan—a randomised proof-of-concept trial. Cephalalgia 2010; 30: 1170–78.

11 Pilgrim AJ, Dussault B, Rupniak N, et al. COL-144, an orally bioavailable selective 5-HT1F receptor agonist for acute migraine therapy. Cephalalgia 2009; 29 (suppl 1): 24–25.

12 International Classifi cation of Headache Disorders: 2nd Edition. Cephalalgia 2004; 24 (suppl 1): 1–160.

Articles

www.thelancet.com/neurology Vol 11 May 2012 413

13 Liefaard L, Drenth H-J, Pilgrim AJ, et al. Prediction of therapeutically eff ective dose of COL-144 based on relationship between plasma concentrations and headache response (poster). Cephalalgia 2009; 29 (suppl 1): 24.

14 Nam JM. A simple approximation for calculating sample sizes for detecting linear trend in proportions. Biometrics 1987; 43: 701–05.

15 Goldstein DJ, Roon KI, Off en WW, et al. Selective serotonin 1F (5-HT1F) receptor agonist LY334370 for acute migraine: a randomized controlled trial. Lancet 2001; 358: 1230–34.

16 Bouchelet I, Cohen Z, Case B, Seguela P, Hamel E. Diff erential expression of sumatriptan-sensitive 5-hydroxytryptamine receptors in human trigeminal ganglia and cerebral blood vessels. Mol Pharmacol 1996; 50: 219–223.

17 Lucaites VL, Krushinski JH, Schaus JM, Audia JE, Nelson DL. [3H]LY334370, a novel radioligand for the 5-HT1F receptor. II. Autoradiographic localization in rat, guinea pig, monkey and human brain. Arch Pharmacol 2005; 371: 178–84.

18 Dodick DW, Martin V. Triptans and CNS side-eff ects: pharmacokinetic and metabolic mechanisms. Cephalalgia 2004; 24: 417–24.

19 Loder L. Triptan therapy in migraine. N Engl J Med 2010; 363: 63–70.

20 Bruinvels AT, Landwehrmeyer B, Gustafson EL et al. Localization of 5-HT1B, 5-HT1D alpha, 5-HT1E and 5-HT1F receptor messenger RNA in rodent and primate brain. Neuropharmacology 1994; 33: 367–86.

21 Ahn S-K, Khalmuratova R, Jeon S-Y, et al. Colocalization of 5-HT1F receptor and calcitonin gene-related peptide in rat vestibular nuclei. Neurosci Lett 2009; 465: 151–56.

22 Ho TW, Ferrari MD, Dodick DW, et al. Effi cacy and tolerability of MK-0974 (telcagepant), a new oral antagonist of calcitonin gene-related peptide receptor, compared with zolmitriptan for acute migraine: a randomised, placebo-controlled, parallel-treatment trial. Lancet 2008; 372: 2115–23.

23 Diener HC, Barbanti P, Dahlöf C, et al. BI 44370 TA, an oral CGRP antagonist for the treatment of acute migraine attacks: results from a phase II study. Cephalalgia 2011; 31: 573–84.

www.thelancet.com/neurology Vol 11 May 2012 429

Review

Treatment of motor and non-motor features of Parkinson’s disease with deep brain stimulationAlfonso Fasano, Antonio Daniele, Alberto Albanese

Deep brain stimulation (DBS) is an established procedure for the symptomatic treatment of Parkinson’s disease. Several deep brain nuclei have been stimulated, producing a wide range of eff ects on the motor and non-motor symptoms of Parkinson’s disease. Long-term, high-quality evidence is available for stimulation of the subthalamic nucleus and globus pallidus internus, both of which uniformly improve motor features, and for stimulation of the thalamic ventralis intermedius, which improves tremor. Short-term data are available for stimulation of other deep brain targets, such as the pedunculopontine nucleus and the centremedian/parafascicular thalamic complex. Some non-motor symptoms improve after DBS, partly because of motor benefi t or reduction of drug treatment, and partly as a direct eff ect of stimulation. More evidence on the eff ects of DBS on non-motor symptoms is needed and specifi cally designed studies are warranted.

IntroductionParkinson’s disease is a progressive neurodegenerative disorder that aff ects several regions of the central and peripheral nervous system.1 The symptoms of Parkin son’s disease encompass the classic parkinsonian triad (tremor, bradykinesia, and rigidity) associated with dopaminergic denervation, other motor signs associated with non-dopaminergic transmission (postural instability and impairment of gait, speech, and posture), and non-motor symptoms (NMS).

Surgical treatments for Parkinson’s disease were developed before the introduction of levodopa2 and re-emerged as a means to overcome diffi culties in the medical management of motor complications in patients with advanced Parkinson’s disease. After pioneering experiments on various CNS targets, stereo-tactic abla tions focused on the pallidothalamic pathway, including the globus pallidus, its outfl ow pathways, and the thalamus (table 1). Lesions in the globus pallidus internus (GPi) consistently improved dyskinesias and parkinsonian motor symptoms.32 How ever, there was a risk of inducing permanent neuro logical defi cits with pallidotomy (especially when bilateral). Lesions of the subthalamic region also improved parkinsonian symp-toms, but caused hemiballism in some patients.33

Deep brain stimulation (DBS) was historically used to check the area to be lesioned in a given functional target34 and later became an adjustable and reversible alternative procedure to stereotactic ablation,5 which was an important advancement in the treatment of tremor. Sub-sequently, GPi DBS was successfully introduced for the management of bradykinesia and rigidity.35 After the discovery of the key part played by hyperactivity of the subthalamic nucleus (STN) in the pathophysiology of Parkinson’s disease,36 STN lesions were shown to improve experimental parkinsonism,37 and the fi rst experiences in patients with Parkinson’s disease27 highlighted that STN DBS could become the surgical treatment of choice for Parkinson’s disease. However, experimental lesions of the pedunculopontine nucleus (PPN) induced akinesia38 and PPN DBS has not provided consistent motor benefi ts

in patients with Parkinson’s disease.39 The main anatomical structures that are targeted by DBS are shown in fi gure 1. In this Review, we aim to address the available evidence on the eff ect of DBS on motor aspects of Parkinson’s disease and particularly on NMS of the disorder, and to highlight the emerging role of new stimulation targets.

Motor featuresMotor control is the main treatment goal for patients with Parkinson’s disease. The motor eff ects of DBS are usually assessed by comparing the eff ects of stimulation with or without added drug treatment,40 as measured on the unifi ed Parkinson’s disease rating scale (UPDRS) motor score. After STN DBS, patients’ motor condition slowly deteriorates41 and often becomes unacceptable. Obser-vations for up to 1 h have shown incomplete motor decay in patients who have had STN stimulation for 10 years.42 No study has specifi cally assessed the re appearance of motor signs after switching off GPi DBS; fi ndings from patients assessed while not receiving drug treatment and with the stimulator turned off showed a gradual return of Parkinson’s disease signs, similar to that seen after STN DBS.40,43,44 By contrast, hyperkinetic features recur more quickly after withdrawal of thalamic or GPi stimulation, which enables assessment of the reappearance of tremor45 or dyskinesias induced by dopamine replacement therapy (DRT).46

The eff ects of STN and GPi implants on the motor features of Parkinson’s disease have been extensively assessed in class 4 studies, and a few randomised controlled trials have provided a higher class of evidence (appendix). The most robust data are for short-term (1–2 years) follow-up after surgery. STN DBS induces many of the antiparkinsonian eff ects of DRT, and preoperative response to levodopa contributes to pre-diction of the outcome after STN DBS.47 Fewer studies, which had short follow-up, are available for DBS of the GPi and other nuclei. Long-term UPDRS-based data are available for STN DBS (10 years);42 medium-term data are available for GPi DBS (5–6 years)44 and thalamic ventralis

Lancet Neurol 2012; 11: 429–42

Istituto di Neurologia, Università Cattolica del Sacro Cuore, Rome, Italy (A Fasano MD, A Daniele MD, Prof A Albanese MD); Associazione per la Ricerca Biomedica Fatebenefratelli, Rome, Italy (A Fasano); and Fondazione Istituto Neurologico Carlo Besta, Milan, Italy (A Albanese)

Correspondence to:Prof Alberto Albanese, Fondazione Istituto Neurologico Carlo Besta, Via Celoria, 11, 20133 Milan, Italyalberto.albanese@unicatt.it

See Online for appendix

430 www.thelancet.com/neurology Vol 11 May 2012

Review

intermedius nucleus (Vim) DBS (5 years);48 and short-term data are available for PPN DBS (2 years).49 Over the past 5 years, a signifi cant improvement in parkinsonian motor features has been reported in selected patients after unilateral DBS of either the STN or GPi.50–52

Several medium-term53–59 and some long-term studies42,60 have confi rmed that STN DBS improves motor fl uctuations, dyskinesias, and the cardinal motor manifestations of Parkinson’s disease, with less consistent eff ects on bradykinesia in the on-treatment condition. Moreover, after STN implant, the levodopa-equivalent dose (LED) is readily reduced on average by 55·9%,61 and a trade-off between LED and the total energy delivered by DBS can be also measured 5 years,56 8 years,60 or 10 years after surgery.42 By contrast, the medium-term eff ects of GPi DBS are less consistent, with some studies reporting stable43,44 or reduced bene fi cial eff ects62 up to 5 years after surgery.

Bradykinesia and rigidityIn a meta-analysis of 38 short-term studies from 34 neurosurgical centres in 13 countries,63 STN DBS improved rigidity and bradykinesia by 63% and 52%, respectively, after 12 months. With the addition of DRT, these improvements increased to 73% and 69% respectively.63 GPi DBS reduced rigidity and bradykinesia 1–2 years after implantation,40,64,65 to the same extent as that reported after STN DBS.66 Whether bradykinesia and

rigidity are also improved by stimulation of other targets is unclear. The subthalamic region contains pallidal outfl ow pathways that can be infl uenced by stimulation in concert with the STN.67 Stimulation of its posterior part (including the zona incerta [Zi] and the prelemniscal radiation) improved contralateral rigidity by 92·7% and contralateral akinesia by 65·7%.24 By contrast, thalamic stimulation does not improve rigidity and bradykinesia,45 and the eff ects of PPN stimulation are still disputed.

Evidence suggests that the initial benefi t of STN DBS on akinesia decreases over time (appendix) and that the symptomatic eff ects of stimulation and drug treatment do not necessarily add up in the long term.53–55 8 years after STN DBS, improvement of rigidity was retained with or without additional drug treatment, whereas bradykinesia was improved only partially by stimulation alone (25·1% compared with baseline) and worsened by 21·6% when patients received stimulation and drug treatment (compared with the drug treatment alone at baseline).60 This fi nding, which was confi rmed at 10 years,42 is probably due to the progression of Parkinson’s disease and the appearance of drug-resistant and stimulation-resistant symptoms. Similarly, a reduction of benefi cial eff ects after GPi DBS has been reported at 5 years.62 The dramatic reduction in LED noted after STN DBS has not been reported for GPi DBS (appendix). Because of the size of the GPi, stimulation must deliver more energy to the GPi than the STN, leading to shorter battery life.64 STN DBS improves bradykinesia more than GPi stimulation:68 70–80% compared with 30–40% according to retrospective comparison’s.69 Findings from other studies suggest that the effi cacy of GPi stimulation on akinesia is lost in the early post-implant phase62 or later.70 Some patients who had GPi DBS successfully underwent subsequent STN DBS.62,70 The GPi is large and contains discrete segregated output pathways; individual variability of subnuclear location of the stimulating electrode accounts at least in part for a lower effi cacy compared with STN DBS. Stimulation in the anteromedial-ventral GPi is associated with a greater improvement in rigidity than stimulation in the central-dorsal GPi, whereas those located in the central-dorsal GPi are more eff ective on bradykinesia than stimulation in the anteromedial-ventral GPi.46 Conversely, stimulation of a smaller target than the GPi, such as the STN, might be associated with a greater predictability of eff ective outcome, but can result in a higher incidence of adverse eff ects.44,52,66

TremorParkinsonian tremor is thought to result from oscillating networks within basal ganglia circuits, and various nuclei within and outside the basal ganglia are potential targets for managing tremor. According to a traditional symptom-based approach, lesions or DBS of the thalamic Vim relieve tremor.5 Common DBS-related adverse events are paraesthesia and, in patients with bilateral implants, dysarthria and balance diffi culties.5 Although STN or GPi

Historical indications for ablation

Indications for deep brain stimulation

Thalamus

Ventralis intermedius nucleus

Tremor in Parkinson’s disease and other movement disorders (thalamotomy)3,4

Essential tremor and tremor in Parkinson’s disease5*Symptomatic tremors6

Orthostatic tremor7

Dystonia8

Nuclei ventro-oralis anterior and posterior

Dystonia9 Dystonia10

Centre median nucleus/parafascicular complex

Tourette’s syndrome11

Other movement disorders12

Parkinson’s disease13

Tourette’s syndrome14*

Globus pallidus Parkinson’s disease and other movement disorders (pallidotomy)15,16

Parkinson’s disease (GPi)17*Dystonia (GPi)18*Huntington’s disease (GPi)19

Tourette’s syndrome (GPi)20

Ansa lenticularis Parkinson’s disease (ansotomy)21 None

Forel’s fi elds Parkinson’s disease and other movement disorders (campotomy)22

Parkinson’s disease (caudal zona incerta)23

Parkinson’s disease (prelemniscal radiation)24

Tremor25

Subthalamic nucleus Parkinson’s disease (subthalamotomy†)26

Parkinson’s disease27*Dystonia28

Essential tremor29

Pedunculopontine nucleus None Parkinson’s disease30,31

GPi=globus pallidus internus. *Consolidated indication. †The term subthalamotomy, originally developed to describe stereotactic ablations of the subthalamic region, is now used to describe stereotactic lesions confi ned to the subthalamic nucleus.

Table 1: Identifi cation of targets for stereotactic ablation and present indications for deep brain stimulation in movement disorders

www.thelancet.com/neurology Vol 11 May 2012 431

Review

stimulations also improve Parkinson’s disease tremor, thalamic DBS remains a valuable surgical option for treatment of disabling tremor—eg, in patients with advanced age when other targets are not practicable.71,72 Stimulation of the caudal Zi produced a 93% improvement in tremor compared with 86% improvement after stimulation of the dorsal border of STN and 61% after stimulation of the STN itself.73 Unilateral Zi stimulation is also eff ective for treatment of contralateral Parkinson’s disease tremor.24 The centre median/parafascicular (CM/Pf) thalamic complex has also been proposed as a successful target for control of tremor.13,74 Long-term effi cacy in tremor management has been reported for STN,53,60 GPi62 (appendix), and thalamic DBS, as noted in a multicentre study with a 5-year follow-up that enrolled patients with either unilateral or bilateral Vim implants.48

Gait and balanceGait and postural diffi culties usually occur in the late stages of Parkinson’s disease, on average 10–15 years after onset, and represent a substantial problem in the management of Parkinson’s disease symptoms that might be particularly resistant to both DRT and DBS. A meta-analysis showed that, during the fi rst year after implantation, STN DBS improved postural instability gait diffi culty (PIGD) complex, roughly equalling the pre-operative eff ects of drug treatment.75 The addition of drug treatment provides further improvement in the short term.76 Findings from several studies have shown that off -period freezing is improved by STN DBS whereas freezing resistant to DRT is not,53,77 although this can rarely be improved.78 Gait analysis study fi ndings consistently showed that STN DBS and levodopa independently have a similar positive eff ect on spatio temporal gait parameters early after implantation.79 However, individual patients might show poor or no gait improvement after STN implantation, even in the short term.75

Inaccurate positioning of the stimulating electrode within the STN can cause stimulation-induced freezing.80 Furthermore, spread of current to the substantia nigra, Zi, or other adjacent regions can cause stimulation-related akinesia, as confi rmed by the negative eff ect on gait induced by a voltage increase.81 Reduction of the frequency of stimulation can improve gait and freezing,81 although the benefi t might not be sustained over time.82 In the long term, axial motor features decline despite STN stimulation.53,54,56 5 years after STN implantation, gait problems that respond poorly to STN DBS arise in 15–40% of patients.55,58 In a patient cohort with excellent preoperative gait improvement with DRT, continuous STN stimulation for 8 years improved gait by 41% compared with the preoperative condition.60 The long-term effi cacy of GPi DBS is less well documented (appendix). Findings from some studies suggest that GPi is less effi cacious than STN DBS on axial features,68 but a recent meta-regression analysis revealed that PIGD

initially improved after DBS of either the STN or GPi and gradually declined to presurgery values 2 years after implantation in the STN but not the GPi.83

Up to 35% of patients have a clinically meaningful worsening of postural stability between 5 and 8 years after implantation.60 In a 10-year follow-up study42 there was no diff erence between baseline and last visit in UPDRS postural stability scores in the practically defi ned off -condition, although the on-condition score greatly worsened compared with baseline.

PPN DBS has been proposed for patients with Parkinson’s disease who have severe axial signs that are unresponsive to drug treatment. Initial reports described an improvement of gait with stimulation at low frequencies (10–25 Hz) and a worsening at higher frequencies (>80 Hz).30,31,84 A synergistic eff ect was reported in patients with bilateral simultaneous STN and PPN implants, with PPN stimulation more eff ective on axial signs and STN

BA

DC

GPi

Vim

PPN

STN

ABC

Figure 1: Main anatomical structures targeted by deep brain stimulation, as they appear on T2-weighted brain MRIAxial sections correspond to the level of anterior commissure (A), superior colliculus (B), and inferior colliculus (C). Locations of sections are shown in D The target nuclei are shown by the green circles. The STN is observed as a small lens-shaped hypointense nucleus ventral to the red nucleus; the GPi appears as a hypointense region located laterally to the anterior limb of internal capsule. Vim cannot be observed with brain MRI and has been traditionally identifi ed during surgery on a conscious patient by recording its physiological signature (so-called tremor cells). Conventional MRI sequences are unsuitable for clear visualisation of the PPN and there are no hallmarks that allow a clear identifi cation of its boundaries. Other nuclei not displayed are the centremedian/parafascicular complex (medial to Vim in A), Forel’s fi elds and the zona incerta (surrounding the STN in B). GPi=globus pallidus internus. PPN=pedunculopontine nucleus. STN=subthalamic nucleus. Vim=ventralis intermedius nucleus.

432 www.thelancet.com/neurology Vol 11 May 2012

Review

stimulation more eff ective on limb features.31,84 Findings from studies suggest a small eff ect of PPN stimulation on some motor signs, particularly gait and balance, despite large interindividual variability (appendix).39,85,86

SpeechThe eff ect of STN DBS on hypokinetic dysarthria is limited87 (appendix). STN DBS has produced clinically signifi cant improvements in speech intelli gibility,88 phonation, or articulation.89,90 However, these positive eff ects might weaken over time89 or not be clinically meaningful.90,91 A consistent retrospective fi nding is that speech worsens after STN implantation, with 56% of patients with worsening speech at 1 year after implantation,92 70% at 3 years,55 57% at 5 years,58 and 90% at 8 years.60 In a prospective controlled study, loudness increased 1 year after STN DBS but speech intelligibility deteriorated by a mean of 14·2% (compared with 3·6% in the control group; p<0·05).93 Speech rate and rhythm are aff ected in patients with Parkinson’s disease and stuttering can recur or be aggravated after STN DBS.94,95

Delayed speech worsening 5–6 years after implantation and stimulation-induced dysarthria were reported in patients with GPi implants, albeit less commonly than after STN DBS.44 Vim stimulation does not improve hypokinetic dysarthria.71

Motor fl uctuations and dyskinesiasClinical trials and meta-analyses61,63 have assessed the benefi cial eff ects of STN DBS in reducing motor fl uctuations (appendix), with stable benefi ts that last for several years after surgery.60 STN DBS does not have an appreciable antidyskinetic eff ect and can even induce dyskinesias (which prevent increase of stimu lation during programming).27 Notwithstanding, dyskinesia reduction has been consistently reported after STN implantation, owing to the reduction of postoperative DRT by an average 60%,56,61 as confi rmed by the fi nding that acute levodopa administration can still provoke dyskinesias after STN implantations.96 Additionally, a further decrease of on-period dyskinesias can be induced by an overall stabilisation of basal ganglia networks and striatal synaptic function after STN DBS.59 Finally, at least in some patients and depending on the electrode trajectory, surrounding stimulation diff using outside the STN can also infl uence the surrounding subthalamic region, particularly the ansa lenticularis and the lenticular fasciculus, mimicking the antidyskinetic eff ect of GPi stimulation (fi gure 2).

After GPi DBS there is negligible long-term reduction in DRT, as confi rmed by two large multicentre STN-GPi comparative studies40,66 that reported a reduction in drug doses only in the STN group. However, GPi DBS has a direct and acute antidyskinetic eff ect, especially when stimulation is delivered through the ventral regions:46 apomorphine-induced dyskinesias are almost abolished by GPi DBS, in a similar way whereas they remain unchanged after STN stimulation.97 In addition to the

direct eff ect of stimulation, GPi DBS might produce long-term plastic changes that further contribute to dyskinesia reduction.98 Finally, GPi DBS might also induce dyskinesias when stimulation is delivered through the dorsal contacts.46

Preliminary data suggest that stimulation of the caudal Zi might aff ect dyskinesia scores and drug reduction to STN DBS.23 No eff ect on motor fl uctuations and dyskinesias has been noted after stimulation of the PPN or thalamic nuclei.

Non-motor symptomsNMS of Parkinson’s disease encompass various clinical manifestations, including cognitive dysfunction, behav-ioural changes, hyposmia, dysautonomia, and sleep dysfunction.99 These features are often more disabling and resistant to treatment than motor symptoms and are key determinants of quality of life. Behavioural disorders might be substantial in patients treated by STN DBS,100–102 whereas the few data available for implants in other targets (ie, Vim, GPi, or PPN) suggest low non-motor morbidity.52,103–106 Table 2 summarises the interactions between stimulation at diff erent targets and NMS.

CognitionStudies of the eff ects of STN DBS on cognition have consistently reported a postoperative decline on phono-logical and semantic verbal fl uency tasks,107,108 which was detectable a few months after surgery and gradually increased in the long term (up to 8 years).60,109 Besides a postoperative decline on a phonological verbal fl uency task, long-term cognitive follow-up revealed a slight but signifi cant decline in tasks of episodic memory, executive function, and abstract reasoning.60 Recent studies in patients with Parkinson’s disease who were treated with STN DBS compared with those given drug treatment showed that 1 year110 and 3 years after implantation111 the STN groups had a greater decline only on a phonological verbal fl uency task. The decline that has been detected shortly after STN DBS surgery might be caused by surgical microlesions aff ecting the cortical-basal ganglia circuits that are involved in word retrieval processes.112 Alternatively, STN stimulation might cause decreased activity of inferior frontal and temporal cortical areas in the left cerebral hemisphere, resulting in decreased verbal fl uency.113 Finally, because withdrawal of dopa-minergic drugs can aff ect performance of patients with Parkinson’s disease on verbal fl uency tasks,114 a post-operative reduction in DRT might also play a part in the decline in verbal fl uency after STN DBS. Overall, STN DBS is safe from a cognitive standpoint when strict inclusion criteria are used,115 although some studies have reported cognitive decline even when patients are subject to strict inclusion criteria.116

Bilateral GPi DBS has low cognitive morbidity, with some studies reporting a mild decline in semantic

www.thelancet.com/neurology Vol 11 May 2012 433

Review

verbal fl uency,62 and no signifi cant eff ect on cognitive functioning occurred 6 months after surgery in patients with advanced Parkinson’s disease.104 GPi DBS has lower cognitive morbidity than STN DBS, as shown by a greater decline on tasks of phonological verbal fl uency,52 overall cognition,44 and visuomotor processing speed in patients treated with STN DBS.66 A meta-analysis of reports on STN and GPi DBS over 10 years concluded that cognitive and behavioural adverse events were more common in the STN group than the GPi group.117

Cognitive eff ects of PPN DBS have been assessed in a few unmasked studies on a small number of patients from one centre. Bilateral PPN implants reduced reaction time in tests assessing executive function and working memory, and improved performance on delayed recall and verbal fl uency.105,106 Such an improvement might be mediated by

activation of ascending cholinergic neurons to the CM/Pf thalamic complex, leading to widespread activation mediated by the intralaminar thalamic nuclei. PET studies have reported an increase in fl uorodeoxyglucose consumption in prefrontal areas, suggesting a modulation of thalamic metabolism after PPN DBS.118 Vim DBS is thought to have a low cognitive morbidity, although this has not been extensively investigated.103

Impulse control disordersUp to 13·6% of patients with Parkinson’s disease develop impulse control disorders (ICDs).119 DRT might play an important pathogenic part in ICDs by over stimulating mesolimbic dopaminergic circuits that are involved in motivation and response to reward.120 STN DBS variably infl uences pre-existing ICD features. In most studies,

Figure 2: Organisation of the eff erent projection from the basal gangliaSimplifi ed anatomical structures and pathways (A) and the theoretical position of DBS electrode placement in the STN (B) are shown. A discrete number of subcortical nuclei, all involved in the wide basal ganglia circuitry, have been targeted in patients with Parkinson’s disease who have had stereotactic surgery. The STN is a glutamatergic nucleus located ventral to the thalamus. The globus pallidus is a large GABAergic nucleus composed of two functionally segregated subparts: the GPe, which receives inputs from the neostriatum (caudate nucleus and putamen) and the STN and in turn projects to STN and GPi (not shown); and the GPi, which is the main output structure of the basal ganglia (the other being the SNr) and projects to the nuclei of the motor thalamus (VA and VL), the CM/Pf complex, and the PPN. There are at least two diff erent functional regions within the GPi, due to the segregation of pallidofugal fi bres that ventrally form the ansa lenticularis (conveying projections from the outer portion of the GPi) and dorsally give rise to the lenticular fasciculus that conveys projections from the inner portion of the GPi. The subthalamic region is a white matter area abutting the STN and encompassing the Zi, Forel’s fi elds, and the prelemniscal radiation. The ansa lenticularis and the fasciculus lenticularis surround the STN before reaching Forel’s H2 fi eld, where they merge into the thalamic fasciculus. This crosses Forel’s H1 fi eld before distributing to the thalamus. Stimulation of STN can also infl uence the diff erent fi bre tracks surrounding the nucleus (B: the grey shadow represents the rough size of the electrical fi eld under unipolar stimulation of the second contact). The thalamic Vim is an ill-defi ned anatomical structure located posterior to the VL and anterior to the VP (which is involved in sensory processing). Its main function is to relay aff erents from the cerebellar nuclei. CM/Pf=centremedian/parafascicular thalamic complex. GPi=globus pallidus internus. GPe=globus pallidus externus. IC=internal capsule. PPN=pedunculopontine nucleus. SNc=substantia nigra pars compacta. SNr=substantia nigra pars reticulata. STN=subthalamic nucleus. VA=ventralis anterior thalamic nucleus. Vim=ventralis intermedius nucleus. VL=ventralis lateralis thalamic nucleus. VP=ventralis posterior thalamic nucleus. Zi=zona incerta. H1=Forel’s H1 fi eld. H2=Forel’s H2 fi eld.

Caudate nucleus

PutamenGPe

GPi

IC

CM/PfVA, VL

Zi

H1

H2

STN

SNrSNc

PPN

Thalamic fasciculus

Lenticular fasciculus

Ansa lenticularis

BA

Caudate nucleus

PutamenGPe

GPi

IC

CM/PfVA, VL

Zi

H1

H2

STN

SNrSNc

PPN

434 www.thelancet.com/neurology Vol 11 May 2012

Review

ICDs markedly improved or disappeared after STN DBS in patients with Parkinson’s disease.121–123 This eff ect might be due to the reduction of DRT after implantation, resulting in decreased stimulation of mesolimbic dopaminergic circuits,122 or to the direct inhibition of the ascending dopaminergic and serotonergic pathways that are involved in reward.121

A few studies have reported onset of ICDs (pathological gambling, hypersexuality, and compulsive eating)123,124 in patients with Parkinson’s disease after STN DBS despite a postsurgical reduction of DRT.125 A cross-sectional study that compared patients with Parkinson’s disease treated with STN DBS to patients treated with drugs alone reported a higher incidence of impulsivity in the DBS group.126 However, no preoperative data were available from this study. STN DBS might disrupt the activity of limbic circuits within the STN or the neighbouring fi bre tracts, resulting in increased impul sivity.127 Additionally, STN DBS might alter the coupling between the prefrontal cortex and basal ganglia during decision-making processes, resulting in impulsive behaviour during high-confl ict situations.100,128 Finally, STN DBS might mimic the action of DRT, thus facilitating the onset of ICDs, particularly in patients taking high doses of DRT. ICDs have been associated with

oscillatory theta-alpha activity in the ventral STN, which suggests that the limbic ventral STN might be involved in the development of ICDs.129

The eff ects of GPi DBS on ICDs are still poorly known: in two men with Parkinson’s disease, preoperative hyper-sexuality did not improve after surgery.123 In patients with Parkinson’s disease treated with STN DBS, pre-existing ICDs improved postoperatively, with a signifi cant reduction in DRT.

Dopamine dysregulation syndrome and pundingPatients with dopamine dysregulation syndrome (DDS) develop an addictive pattern of DRT use. In a series of 21 patients with Parkinson’s disease who underwent bilateral STN DBS, symptoms improved or resolved in 29% of the patients with preoperative DDS; in two patients symptoms of DDS appeared only after surgery (in one case after an 8-year latency).123 Resolution of symptoms has been associated with motor improvement and LED reduction after STN DBS.121,123

Punding is a stereotyped behaviour that is triggered by DRT; it is characterised by intense fascination with complex, excessive, non-goal-oriented, repetitive activ-ities, and is linked to dyskinesia severity, DDS, and occurrence of other ICDs.130,131 Punding can worsen or even arise after STN DBS surgery, despite DRT reduction.123,132,133

ApathySeveral studies have reported a worsening of apathy in patients with Parkinson’s disease after STN DBS.109,134 In a prospective study of patients with STN implants, apathy occurred after a mean of 4·7 months in 54% of patients and was reversible in half of them at 1 year.134 Apathy might be associated with insuffi cient DRT after DBS, resulting in a postoperative deactivation of dopaminergic receptors within the mesocortical and mesolimbic path ways.134 Accordingly, in patients with Parkinson’s disease who developed apathy after complete withdrawal of DRT after successful STN DBS, a 6-week trial of ropinirole induced reversal of apathy.135 In another study, apathy was assessed in patients with Parkinson’s disease who received unilateral GPi or STN implants and in a control group of drug-treated patients.136 Apathy was unchanged in the drug-treated group, whereas it progressively increased during the fi rst 6 months after implantation in both DBS groups, with no relation to postsurgical drug changes.

Mood disorders and anxietyPostoperative mood disorders (depression or mania) can occur after STN implantation, either as acute and transient or chronic and persistent disorders.102,125,137,138 In patients with bilateral chronic STN stimulation, depressive features improved,108 remained unchanged,60,117 or even worsened compared with the preoperative condition.138 Postoperative improvement of depression might result from a

Subthalamic nucleus

Globus pallidus internus

Pedunculopontine nucleus

Cognition

Memory 0/+* 0 0

Executive functions –*† 0 +†

Mood disorders

Apathy –*† 0 0

Depression –*† 0/–† 0

Anxiety –/+*†‡ 0/+† 0

Behaviour and other psychiatric issues

Impulse control disorders 0/+* 0 0

Delusions and hallucinations 0/+* 0 0

Dopamine dysregulation syndrome –/+*† 0/+† 0

Punding –/+*† 0/+† 0

Autonomic dysfunction

Sweating +‡ 0/+‡ 0

Urinary function +†‡ 0/+‡ 0

Bowel function 0/+*‡ 0/+‡ 0

Cardiovascular dysautonomia 0/+*‡ 0 0

Sleep

Quality +*†‡ +‡ +†

Architecture 0 0 +†

Rapid eye movement sleep behaviour disorder 0/+* 0 0/+†

Restless leg syndrome –* 0 0

Daytime sleepiness +* 0 +†

Pain +†‡ 0/+‡ 0

0=no eff ect. +=improvement. –=worsening. 0/–=no eff ect or worsening. 0/+=no eff ect or improvement. *Secondary to drug reduction. †Due to direct eff ect of stimulation. ‡Secondary to motor improvement.

Table 2: Synopsis of the eff ects of deep brain stimulation at diff erent targets on Parkinson’s disease non-motor symptoms

www.thelancet.com/neurology Vol 11 May 2012 435

Review

psychological response to the alleviation of disabling motor symptoms139 or from the eff ects of STN stimulation on neural circuits involved in mood.125 Suicidal tendencies have been reported in some patients with Parkinson’s disease after STN DBS.101,102 A retrospective study aimed at identifying the suicide rate after STN DBS in a large sample of patients with Parkinson’s disease reported a 0·9% rate of attempted suicide and a 0·45% rate of successful suicides.102 Suicide rates were higher during the fi rst postoperative year than at any other time. Various factors (postoperative depression, being single, previous history of ICDs, or compulsive drug use) were associated with attempted suicide risk; social and cultural variables might also play a part.140 Various mechanisms might be involved in the pathophysiology of postoperative depression after STN DBS, such as tapering DRT too fast or an indirect inhibition of the activity of ascending serotonergic neurons,141 possibly exerted by projections from the basal ganglia to the dorsal raphe nucleus.

Manic symptoms occur in about 4% of patients with Parkinson’s disease with bilateral STN implants,138 sometimes in the immediate postoperative period.109,125 By contrast, 7 months after surgery, no overt mood variations were noted in patients with unilateral GPi or STN DBS.52 Manic symptoms can last for hours or a few days and might be closely linked to STN stimulation.125,142 Stimulation of the most ventral contacts within the STN can generate mood abnormalities, which are seldom suppressed by switching off .143 More rarely, stimulation of the substantia nigra pars reticulata142 or of axons arising from the medial (limbic) portion of the STN and entering the medial forebrain bundle can give rise to DBS-induced reversible acute hypomania.144 In patients with stimulation-induced manic symptoms, PET shows increased regional cerebral blood fl ow during the manic state, mainly in the right cerebral hemisphere in the anterior cingulate and medial prefrontal cortex.142 Re adjusting the stimulation settings143 or switching to another stimulation target145 can resolve manic symptoms in some patients.

GPi and thalamic implants can also occasionally aff ect mood. Recurrent manic and hypomanic episodes, each lasting several days, were reported in one patient treated with bilateral GPi DBS.146 Manic symptoms have not been reported after thalamic implants, but improvement of mood was reported in 23% of patients after CM/Pf DBS.74 and in a small sample of patients with Parkinson’s disease with unilateral Vim DBS.103

Various studies reported a postoperative improvement of anxiety in patients with Parkinson’s disease after STN DBS;108,147 others have reported no change116 or even the appearance or worsening of pre-existing anxiety.43 In a short-term comparison trial (STN DBS vs DRT), anxiety was reduced in the DBS group.148 In the long term, no signifi cant changes in anxiety levels compared with baseline have been reported.60 Postoperative worsening of anxiety might result from a dopamine withdrawal syndrome.134 Variations in postoperative management of

DRT and individual variations of mesolimbic dopamin-ergic denervation might explain the variability in mood, anxiety, and motivation after STN DBS.134 Improvement of motor symptoms also contributes to a reduction in anxiety after STN DBS.109

PsychosisIn a series of patients with Parkinson’s disease treated with STN DBS, short-lasting transient hallucinations and delusions were noted shortly after surgery.125 Whether patients with a history of hallucin ations are appropriate candidates for STN DBS is still debated. Pre-existing severe drug-induced hallucinations or delusions dis-appeared postoperatively in eight of ten patients with bilateral STN DBS after a reduction of DRT.149 In the remaining two patients, hallucinations and delusions worsened immediately after surgery, despite complete DRT withdrawal, and disappeared after a few months of treatment with antipsychotic drugs. Another study investigated the eff ects of STN DBS on pre-existing hallucinations in 18 patients with advanced Parkinson’s disease and noted a signifi cant postoperative improvement of hallucination severity 6 months after DBS compared with baseline.150 These fi ndings suggest that a history of hallucinations does not formally contraindicate STN DBS in patients with advanced Parkinson’s disease.

There have been few studies on the occurrence of hallucinations and delusions in patients treated by GPi DBS. Preliminary evidence suggests that the incidence of visual hallucinations might be lower after GPi DBS than STN DBS.151 In a 6-year follow-up multicentre study of 38 patients with Parkinson’s disease treated by Vim DBS, the occurrence of cognitive and psychiatric adverse events was low, with one case of hallucinations reported among all centres.71

Autonomic dysfunctionAlthough orthostatic dizziness, bladder dysfunction (urge, incontinence, and frequency), hyperhidrosis, and erectile dysfunction are common NMS of Parkinson’s disease, only a few class 4 studies have addressed these features. After STN DBS, an improvement of dysautonomia after reduction of DRT (as suggested for bowel function)152 or an improvement in motor functioning (as for excessive sweating secondary to dyskinesias) might occur.152 Accordingly, the sympathetic skin response does not change after STN implantation, although dyshidrosis is improved by 66·7% compared with before surgery.153 Furthermore, a direct eff ect of stimulation on autonomic regions might explain the improvement of urinary symptoms after STN DBS,152,154 by an increase of bladder capacity and refl ex volume155 and improved integration of aff erent bladder signals by the basal ganglia, with subsequent modulation of activity of the lateral frontal and anterior cingulate cortex.154,156

STN DBS seems to have little eff ect on cardiovascular dysautonomia.157 One study noted that STN stimulation

436 www.thelancet.com/neurology Vol 11 May 2012

Review

increases peripheral vasoconstriction and barorefl ex sensitivity and stabilises blood pressure, thereby improving postural hypotension.158

SleepBilateral STN DBS improves objective measures of sleep on polysomnography, decreases nocturnal and early morning dystonia, and increases sleep effi ciency in the on-stimulation condition.159 Around-the-clock stimu lation improves nocturnal mobility, continuous sleep time, and sleep effi ciency compared with before sur gery.160,161 The duration of slow wave sleep and rapid eye movement (REM) sleep is increased after STN DBS, but the relative percentage of sleep stages does not vary; there is no association with motor improvement.161 A subjective benefi t of STN DBS on sleep quality has also been reported.160,162 In a 2-year follow-up study, the total sleep time increased after bilateral STN DBS; these changes were associated with an improvement in bradykinesia.163 The reported improvement in nocturia after STN DBS was consistent with the noted increase in bladder capacity. Other factors can infl uence sleep quality, such as DRT reduction and the ensuing improvement in daytime somnolence. No improvement in REM sleep behaviour disorder or periodic limb movements of sleep was detected after STN DBS.160 Some studies have reported benefi t in restless legs syndrome,162 whereas fi ndings from others suggested that restless legs syndrome might occur postoperatively, possibly due to reduction in DRT.164

A few studies have addressed the eff ects of GPi DBS on sleep quality in patients with Parkinson’s disease and reported subjective improvement of daytime sleepiness even though these patients did not reduce DRT.165 However, Vim DBS does not infl uence sleep architecture or sleep spindles.166

Experimental studies have shown that the PPN is involved in sleep functions. Polysomnographic studies reported a signifi cant increase in the absolute or relative duration of REM sleep after PPN DBS.105,167,168 The observation that REM behaviour disorder is improved after PPN DBS49 has not been confi rmed.168 Daytime polysomnography during diff erent stimulating condi-tions revealed that low-frequency stimulation (10–25 Hz) promotes alertness, whereas high-frequency pulses induce light sleep (stages N1 and N2).169

Pain and sensory symptomsSensory symptoms (pain and paraesthesia) might represent unwanted side-eff ects of stimulation at diff erent targets (Vim, STN, or PPN) if the current from the stimulating electrode reaches the medial lemniscus or the internal capsule. By contrast, little is known about the variations of Parkinson’s disease-related sensory symptoms after DBS. STN DBS can improve pain,170,171 particularly during off periods. Objective pain sensitivity was unchanged in patients who reported

pain improvement with STN DBS or drug treatment, suggesting that these treatment options do not directly infl uence central pain processing.172

Emerging issuesWith a rapidly growing body of evidence on DBS for Parkinson’s disease, new clinical issues have emerged. These have not yet been systematised in clinical practice, but are relevant for making appropriate clinical decisions.

Target choiceThe traditional anatomoclinical approach of stereotactic surgery (ie, one symptom equals one target) has its quintessential hallmark in tremor surgery, since appropriate Vim targeting has been consistently shown to provide immediate and long-lasting relief of contralateral tremor.71 However, the choice of the most suitable DBS target for each patient with Parkinson’s disease cannot be made solely on the basis of symptoms, because each target infl uences the activity of multiple brain structures within the basal ganglia network (fi gure 3).

There are no guidelines for the choice of DBS target in Parkinson’s disease. Randomised studies have provided evidence that there are no diff erences in short-term motor outcome after unilateral or bilateral implants in the STN or GPi52,66 (appendix), although non-motor outcome favours the GPi. However, long-term open-label results favour STN stimulation, because of the decay in motor effi cacy reported in the few available GPi studies.62,70 Target choice might also depend on technical reasons. Easier targeting in the larger GPi and easier medical management (with no need to adjust DRT) favour GPi implants, whereas the possibility to also infl uence the subthalamic region (which contains pallidofugal fi bres) and lower energy consumption favour the STN. The patient’s age might support the choice of one nucleus over the other: the STN should be chosen in younger patients who have prominent akinesia and tremor, who might otherwise have to have rapid DRT increases and could be exposed to the potential side-eff ects of antiparkinsonian drugs. Accordingly, mono genic early-onset Parkinson’s disease has been successfully treated with STN DBS.173–175

PPN was initially selected as a target in patients with Parkinson’s disease who had severe axial symptoms resistant to DRT.84,85 This target has also been stimulated in combination with others to achieve an additive symptomatic eff ect: bilateral four-electrode implants have been used in the STN and PPN84 or in the caudal Zi and PPN.176 However, the indications for PPN targeting are controversial and outcomes are highly variable. After initial enthusiasm, there has been a decline in optimism106,177 and at present there is no suggestion to propose PPN DBS as a primary option.

Unilateral stereotactic surgery has been traditionally done in patients with unilateral tremor by targeting the Vim contralateral to the tremulous body side.178 Implants

www.thelancet.com/neurology Vol 11 May 2012 437

Review

in the STN or GPi are usually done bilaterally, although unilateral DBS has been proposed recently either as a defi nitive procedure50–52 or as part of a staged approach. Logistic regression analysis of the COMPARE (cognition and mood in Parkinson’s disease in subthalamic nucleus versus globus pallidus internus deep brain stimulation) trial revealed that the odds of proceeding to bilateral DBS were 5·2 times higher in patients with unilateral STN implants than in those with unilateral GPi DBS,179 suggesting that STN DBS ends up being bilateral in most cases.

Quality of life and psychosocial functioningSTN and GPi stimulation represent two consolidated treatment options with known indications and adequate follow-up of functional variables,66 although high quality data have been mostly collected in patients with STN DBS. A recent meta-analysis reported a seven-point average functional improvement after STN DBS com pared with DRT alone, as measured by a 39-item Parkinson’s dis ease questionnaire.180 Additionally, dis abling motor com-plications that are not successfully managed by drug treatment are better managed after bilateral STN or GPi DBS compared with DRT alone.180–182

Quality of life and psychosocial functioning are important measures for therapeutic intervention in Parkinson’s disease. Although there is no formal age limit for DBS, age is inversely associated with improvement of motor function182,183 and positively associated with perioperative complications.182 To only use effi cacious surgical interventions, such as DBS, as a last resort once patients have experienced psychosocial decline is not of great help for the patients. In such situations, restoration of mobility through DBS does not necessarily restore quality of life.184 At present, the mean delay before neurosurgery is 14 years after diagnosis,61 but is expected to be reduced as evidence on earlier surgery is gathered.

Timing for surgeryAge and disease duration at time of surgery are important factors to take into account when selecting patients for DBS. Younger patients might have fewer cognitive complications,185 less deterioration of axial signs over time,186 and better improvement of rigidity,187 and there is evidence that, despite the expected motor improve ment, quality of life improves only in younger patients.188 Patients with early-onset genetic Parkinson’s disease benefi t from STN DBS and have a much younger age at implant (49·6 years in a series of patients with PARK2 mutations173 compared with 61·2 years in a non-genetically caused Parkinson’s disease cohort).42 A recent retrospective study concluded that undertaking surgery in patients with short disease duration might delay functional impair ment187 and an 18-month prospective pilot trial favoured early DBS (after average disease duration of 7 years) over medical therapy alone in quality of life measures.147

At present, there is no consensus for timing of stereotactic surgery after disease onset; core assessment program for surgical interventional therapies in Parkin-son’s disease (CAPSIT-PD) recommendations189 suggest that disease dur ation should be at least 5 years before DBS is considered. Controlled trials are needed to ascertain whether undertaking surgery in earlier disease stages is advantageous or even ethical. Two such trials, a German–French multicentre study (EARLYSTIM [The Eff ect of Deep Brain Stimulation of the Subthalamic Nucleus on Quality of Life in Comparison to Best Medical Treatment in Patients With Complicated Parkinson’s Disease and Preserved Psychosocial Competence], ClinicalTrials.gov NCT00354133) and a North American single centre trial (ClinicalTrials.gov NCT00282152) are underway.

Conclusions and outlookDBS is an established procedure that can be applied to diff erent brain targets to treat patients with Parkin son’s disease. Vim DBS is an accepted treatment for Parkinson’s disease-related tremor; its indications have been largely replaced by STN and GPi DBS, which also improve other Parkinson’s disease symptoms. PPN DBS has to be regarded still as an experimental option, which potentially

Figure 3: Synoptic diagram of the diff erent motor and non-motor eff ects of deep brain stimulation at various targetsFor each Parkinson’s disease feature, a prominent eff ect of deep brain stimulation is shown by a long radial distance from the centre of the polygon. Non-motor features are shown on the left side of the graph and motor features are on the right side. Stimulation of some targets (eg, the STN, Zi, or GPi) infl uences various features, particularly bradykinesia and rigidity, tremor, PIGD, fl uctuations, and dyskinesias. By contrast, Vim stimulation selectively aff ects tremor. STN implants also have a moderate eff ect on mood and apathy and a mild eff ect on cognition, whereas PPN implants infl uence PIGD, sleep, and cognition. STN=subthalamic nucleus. Zi=zona incerta. GPi=globus pallidus interna. PPN=pedunculopontine nucleus. Vim=ventralis intermedius nucleus. PIGD=postural instability gait diffi culty. ICD=impulse control disorders.

Autonomicdysfunction

Fluctuations

Dyskinesias

Mood and apathy

Cognition

ICD

Sleep

Bradykinesia and rigidity

Tremor

PIGD

Non-motor features Motor features

STNZiGPiPPNVim

438 www.thelancet.com/neurology Vol 11 May 2012

Review

infl uences PIGD. Other nuclei, such as the caudal Zi and the CM/Pf nucleus, are under investigation. The available evidence on the stimulation of targets diff erent from the STN and GPi are mostly from class 4 studies.

The rapidly growing body of evidence highlighted in this Review provides a synoptic picture of the eff ects of DBS on motor and non-motor features of Parkinson’s disease. Integrating clinical evidence with preclinical research allows future treatment scenarios to be identifi ed and issues that still need to be addressed to be focused on. First, bilateral DBS represents the standard procedure, whereas unilateral or staged implants can be considered in individual cases; furthermore, there is no evidence that implanting into multiple targets has a clinical advantage; rather, this method exposes patients with Parkinson’s disease to the risk of highly invasive surgery. Furthermore, despite widespread use of DBS, the mechanisms through which it alleviates the symptoms of Parkinson’s disease are not fully under stood; further research is needed on this important topic. Moreover, the present data show that the amount of improvement after DBS implants depends on relevant individual variations: there is a cogent need to associate the precise electrode location with surgical outcome as well as to search for predictive factors of long-term outcome after DBS. Careful patient selection is a key variable for improvement of outcome after DBS.190 Because more than 30% of DBS failures can be ascribed to an inappropriate indication for surgery,191 a refi ne ment of patient selection criteria is needed. Finally, a few electrode models are used for nearly all DBS applications, despite substantial anatomical diff erences among tar geted nuclei. Constant-current STN DBS has proven eff ective in a recent controlled trial,192 and future trials should compare constant-current with voltage-controlled stimulation. DBS technology will evolve through the implementation of multicontact electrodes and sensing capabilities, allowing modulation of DBS by monitoring motor and non-motor conditions.

Search strategy and selection criteria

We searched PubMed from January 2004, to January 2012 with the search terms “Parkinson disease”[MH] AND “deep brain stimulation”[MH] AND “English”[LA], which yielded 1179 papers. Data or additional articles were also recovered from other sources, such as recent reviews, reference lists of relevant publications, and a search of the authors’ own reference database, which yielded an additional 123 papers (covering also the period 1947–2003). From the retrieved papers, we selected only meta-analyses and randomised controlled trials on ventralis intermedius nucleus, subthalamic nucleus, or globus pallidus internus stimulation and all the available studies (including open-label trials) on less studied targets (eg, the centremedian/parafascicular complex, pedunculopontine nucleus, and zona incerta) or non-motor symptoms. We referred only to papers with broad-term outcomes on ventralis intermedius nucleus (≥5 years), subthalamic nucleus (>5 years), or globus pallidus internus (>3 years) stimulation. In total, 377 papers were taken into account for this Review.

ContributorsAF did the literature search, drew the fi gures, collected data, and wrote

the fi rst draft. AD did the data analysis and interpretation and reviewed

the manuscript. AA designed the study outline, ideated the fi gures, and

reviewed and fi nalised the manuscript.

Confl icts of interestAF has received speaker’s fees from Abbott, Medtronic, Chiesi

Farmaceutici, and UCB. AD has received honoraria from Pfi zer,

Novartis, and Eli-Lilly. AA has received honoraria from Abbott, Eisai,

Ipsen, and Merz.

AcknowledgmentsThis work was supported in part by a grant from the Italian Ministry of

Health to AA.

References1 Braak H, Del Tredici K. Invited article: nervous system pathology in

sporadic Parkinson disease. Neurology 2008; 70: 1916–25.

2 Blomstedt P, Hariz MI. Deep brain stimulation for movement disorders before DBS for movement disorders. Parkinsonism Relat Disord 2010; 16: 429–33.

3 Spiegel EA, Wycis HT, Marks M, Lee AJ. Stereotaxic apparatus for operations on the human brain. Science 1947; 106: 349–50.

4 Hassler R, Riechert T. A special method of stereotactic brain operation. Proc R Soc Med 1955; 48: 469–70.

5 Benabid AL, Pollak P, Gervason C, et al. Long-term suppression of tremor by chronic stimulation of the ventral intermediate thalamic nucleus. Lancet 1991; 337: 403–06.

6 Brice J, McLellan L. Suppression of intention tremor by contingent deep-brain stimulation. Lancet 1980; 1: 1221–22.

7 Espay AJ, Duker AP, Chen R, et al. Deep brain stimulation of the ventral intermediate nucleus of the thalamus in medically refractory orthostatic tremor: preliminary observations. Mov Disord 2008; 23: 2357–62.

8 Trottenberg T, Meissner W, Kabus C, et al. Neurostimulation of the ventral intermediate thalamic nucleus in inherited myoclonus-dystonia syndrome. Mov Disord 2001; 16: 769–71.

9 Siegfried J, Crowell R, Perret E. Cure of tremulous writer’s cramp by stereotaxic thalamotomy. Case report. J Neurosurg 1969; 30: 182–85.

10 Mundinger F. [New stereotactic treatment of spasmodic torticollis with a brain stimulation system (author’s transl)]. Med Klin 1977; 72: 1982–86 (in German).

11 Hassler R, Dieckmann G. Traitement stéréotoxique des tics et cris inarticulés ou coprolaliques considérés comme phénomène d’obsession motrice au cours de la maladie de Gilles de la Tourette. Rev Neurol 1970; 123: 89–91.

12 Adams JE, Rutkin BB. Lesions of the centrum medianum in the treatment of movement disorders. Confi n Neurol 1965; 26: 231–45.

13 Krauss JK, Pohle T, Weigel R, Burgunder JM. Deep brain stimulation of the centre median-parafascicular complex in patients with movement disorders. J Neurol Neurosurg Psychiatry 2002; 72: 546–48.

14 Vandewalle V, van der Linden C, Groenewegen HJ, Caemaert J. Stereotactic treatment of Gilles de la Tourette syndrome by high frequency stimulation of thalamus. Lancet 1999; 353: 724.

15 Meyers R. Surgical experiments in the therapy of certain ‘extrapyramidal’ diseases: a current evaluation. Acta Psychiatr Neurol Suppl 1951; 67: 1–42.

16 Cooper IS. Intracerebral injection of procaine into the globus pallidus in hyperkinetic disorders. Science 1954; 119: 417–18.

17 Laitinen LV, Bergenheim AT, Hariz MI. Leksell’s posteroventral pallidotomy in the treatment of Parkinson’s disease. J Neurosurg 1992; 76: 53–61.

18 Coubes P, Roubertie A, Vayssiere N, Hemm S, Echenne B. Treatment of DYT1-generalised dystonia by stimulation of the internal globus pallidus. Lancet 2000; 355: 2220–21.

19 Moro E, Lang AE, Strafella AP, et al. Bilateral globus pallidus stimulation for Huntington’s disease. Ann Neurol 2004; 56: 290–94.

20 Diederich NJ, Kalteis K, Stamenkovic M, Pieri V, Alesch F. Effi cient internal pallidal stimulation in Gilles de la Tourette syndrome: a case report. Mov Disord 2005; 20: 1496–99.

www.thelancet.com/neurology Vol 11 May 2012 439

Review

21 Spiegel EA, Wycis HT. Ansotomy in paralysis agitans. AMA Arch Neurol Psychiatry 1954; 71: 598–614.

22 Spiegel EA, Wycis HT, Szekely EG, Adams DJ, Flanagan M, Baird HW 3rd. Campotomy in various extrapyramidal disorders. J Neurosurg 1963; 20: 871–84.

23 Plaha P, Ben-Shlomo Y, Patel NK, Gill SS. Stimulation of the caudal zona incerta is superior to stimulation of the subthalamic nucleus in improving contralateral parkinsonism. Brain 2006; 129: 1732–47.

24 Kitagawa M, Murata J, Uesugi H, et al. Two-year follow-up of chronic stimulation of the posterior subthalamic white matter for tremor-dominant Parkinson’s disease. Neurosurgery 2005; 56: 281–89.

25 Murata J, Kitagawa M, Uesugi H, et al. Electrical stimulation of the posterior subthalamic area for the treatment of intractable proximal tremor. J Neurosurg 2003; 99: 708–15.

26 Andy OJ, Jurko MF, Sias FR Jr. Subthalamotomy in treatment of parkinsonian tremor. J Neurosurg 1963; 20: 860–70.

27 Benabid AL, Pollak P, Gross C, et al. Acute and long-term eff ects of subthalamic nucleus stimulation in Parkinson’s disease. Stereotact Funct Neurosurg 1994; 62: 76–84.

28 Chou KL, Hurtig HI, Jaggi JL, Baltuch GH. Bilateral subthalamic nucleus deep brain stimulation in a patient with cervical dystonia and essential tremor. Mov Disord 2005; 20: 377–80.

29 Lind G, Schechtmann G, Lind C, Winter J, Meyerson BA, Linderoth B. Subthalamic stimulation for essential tremor. Short- and long-term results and critical target area. Stereotact Funct Neurosurg 2008; 86: 253–58.

30 Mazzone P, Lozano A, Stanzione P, et al. Implantation of human pedunculopontine nucleus: a safe and clinically relevant target in Parkinson’s disease. Neuroreport 2005; 16: 1877–81.

31 Plaha P, Gill SS. Bilateral deep brain stimulation of the pedunculopontine nucleus for Parkinson’s disease. Neuroreport 2005; 16: 1883–87.

32 Lang AE, Lozano AM, Montgomery E, Duff J, Tasker R, Hutchinson W. Posteroventral medial pallidotomy in advanced Parkinson’s disease. N Engl J Med 1997; 337: 1036–42.

33 Alvarez L, Macias R, Pavon N, et al. Therapeutic effi cacy of unilateral subthalamotomy in Parkinson’s disease: results in 89 patients followed for up to 36 months. J Neurol Neurosurg Psychiatry 2009; 80: 979–85.

34 Albe-Fessard D. Electrophysiological methods for the identifi cation of thalamic nuclei. Z Neurol 1973; 105: 15–28.

35 Siegfried J, Lippitz B. Bilateral chronic electrostimulation of ventroposterolateral pallidum: a new therapeutic approach for alleviating all parkinsonian symptoms. Neurosurgery 1994; 35: 1126–30.

36 DeLong MR, Crutcher MD, Georgopoulos AP. Primate globus pallidus and subthalamic nucleus: functional organization. J Neurophysiol 1985; 53: 530–43.

37 Bergman H, Wichmann T, DeLong MR. Reversal of experimental parkinsonism by lesions of the subthalamic nucleus. Science 1990; 249: 1436–38.

38 Munro-Davies LE, Winter J, Aziz TZ, Stein JF. The role of the pedunculopontine region in basal-ganglia mechanism of akinesia. Exp Brain Res 1999; 129: 511–17.

39 Ferraye MU, Debu B, Fraix V, et al. Eff ects of pedunculopontine nucleus area stimulation on gait disorders in Parkinson’s disease. Brain 2010; 133: 205–14.

40 Deep-Brain Stimulation for Parkinson’s Disease Study Group. Deep-brain stimulation of the subthalamic nucleus or the pars interna of the globus pallidus in Parkinson’s disease. N Engl J Med 2001; 345: 956–63.

41 Temperli P, Ghika J, Villemure JG, Burkhard PR, Bogousslavsky J, Vingerhoets FJ. How do parkinsonian signs return after discontinuation of subthalamic DBS? Neurology 2003; 60: 78–81.

42 Castrioto A, Lozano AM, Poon YY, Lang AE, Fallis M, Moro E. Ten-year outcome of subthalamic stimulation in Parkinson disease: a blinded evaluation. Arch Neurol 2011; 68: 1550–56.

43 Rodriguez-Oroz MC, Obeso JA, Lang AE, et al. Bilateral deep brain stimulation in Parkinson’s disease: a multicentre study with 4 years follow-up. Brain 2005; 128: 2240–49.

44 Moro E, Lozano AM, Pollak P, et al. Long-term results of a multicenter study on subthalamic and pallidal stimulation in Parkinson’s disease. Mov Disord 2010; 25: 578–86.

45 Tarsy D, Scollins L, Corapi K, O’Herron S, Apetauerova D, Norregaard T. Progression of Parkinson’s disease following thalamic deep brain stimulation for tremor. Stereotact Funct Neurosurg 2005; 83: 222–27.

46 Krack P, Pollak P, Limousin P, et al. Opposite motor eff ects of pallidal stimulation in Parkinson’s disease. Ann Neurol 1998; 43: 180–92.

47 Welter ML, Houeto JL, Tezenas du Montcel S, et al. Clinical predictive factors of subthalamic stimulation in Parkinson’s disease. Brain 2002; 125: 575–83.

48 Pahwa R, Lyons KE, Wilkinson SB, et al. Long-term evaluation of deep brain stimulation of the thalamus. J Neurosurg 2006; 104: 506–12.

49 Mazzone P, Sposato S, Insola A, Scarnati E. The deep brain stimulation of the pedunculopontine tegmental nucleus: towards a new stereotactic neurosurgery. J Neural Transm 2011; 118: 1431–51.

50 Alberts JL, Hass CJ, Vitek JL, Okun MS. Are two leads always better than one: an emerging case for unilateral subthalamic deep brain stimulation in Parkinson’s disease. Exp Neurol 2008; 214: 1–5.

51 Tabbal SD, Ushe M, Mink JW, et al. Unilateral subthalamic nucleus stimulation has a measurable ipsilateral eff ect on rigidity and bradykinesia in Parkinson disease. Exp Neurol 2008; 211: 234–42.

52 Okun MS, Fernandez HH, Wu SS, et al. Cognition and mood in Parkinson’s disease in subthalamic nucleus versus globus pallidus interna deep brain stimulation: the COMPARE trial. Ann Neurol 2009; 65: 586–95.

53 Krack P, Batir A, Van Blercom N, et al. Five-year follow-up of bilateral stimulation of the subthalamic nucleus in advanced Parkinson’s disease. N Engl J Med 2003; 349: 1925–34.

54 Schupbach WM, Chastan N, Welter ML, et al. Stimulation of the subthalamic nucleus in Parkinson’s disease: a 5 year follow up. J Neurol Neurosurg Psychiatry 2005; 76: 1640–44.

55 Piboolnurak P, Lang AE, Lozano AM, et al. Levodopa response in long-term bilateral subthalamic stimulation for Parkinson’s disease. Mov Disord 2007; 22: 990–97.

56 Romito LM, Contarino MF, Vanacore N, Bentivoglio AR, Scerrati M, Albanese A. Replacement of dopaminergic medication with subthalamic nucleus stimulation in Parkinson’s disease: long-term observation. Mov Disord 2009; 24: 557–63.

57 Wider C, Pollo C, Bloch J, Burkhard PR, Vingerhoets FJ. Long-term outcome of 50 consecutive Parkinson’s disease patients treated with subthalamic deep brain stimulation. Parkinsonism Relat Disord 2008; 14: 114–19.

58 Gervais-Bernard H, Xie-Brustolin J, Mertens P, et al. Bilateral subthalamic nucleus stimulation in advanced Parkinson’s disease: fi ve year follow-up. J Neurol 2009; 256: 225–33.

59 Simonin C, Tir M, Devos D, et al. Reduced levodopa-induced complications after 5 years of subthalamic stimulation in Parkinson’s disease: a second honeymoon. J Neurol 2009; 256: 1736–41.

60 Fasano A, Romito LM, Daniele A, et al. Motor and cognitive outcome in patients with Parkinson’s disease 8 years after subthalamic implants. Brain 2010; 133: 2664–76.

61 Kleiner-Fisman G, Herzog J, Fisman DN, et al. Subthalamic nucleus deep brain stimulation: summary and meta-analysis of outcomes. Mov Disord 2006; 21 (suppl 14): S290–304.

62 Volkmann J, Allert N, Voges J, Sturm V, Schnitzler A, Freund HJ. Long-term results of bilateral pallidal stimulation in Parkinson’s disease. Ann Neurol 2004; 55: 871–75.

63 Hamani C, Richter E, Schwalb JM, Lozano AM. Bilateral subthalamic nucleus stimulation for Parkinson’s disease: a systematic review of the clinical literature. Neurosurgery 2005; 56: 1313–24.

64 Volkmann J, Allert N, Voges J, Weiss PH, Freund HJ, Sturm V. Safety and effi cacy of pallidal or subthalamic nucleus stimulation in advanced PD. Neurology 2001; 56: 548–51.

65 Rodrigues JP, Walters SE, Watson P, Stell R, Mastaglia FL. Globus pallidus stimulation improves both motor and nonmotor aspects of quality of life in advanced Parkinson’s disease. Mov Disord 2007; 22: 1866–70.

66 Follett KA, Weaver FM, Stern M, et al. Pallidal versus subthalamic deep-brain stimulation for Parkinson’s disease. N Engl J Med 2010; 362: 2077–91.

67 Herzog J, Fietzek U, Hamel W, et al. Most eff ective stimulation site in subthalamic deep brain stimulation for Parkinson’s disease. Mov Disord 2004; 19: 1050–54.

440 www.thelancet.com/neurology Vol 11 May 2012

Review

68 Anderson VC, Burchiel KJ, Hogarth P, Favre J, Hammerstad JP. Pallidal vs subthalamic nucleus deep brain stimulation in Parkinson disease. Arch Neurol 2005; 62: 554–60.

69 Krack P, Pollak P, Limousin P, et al. Subthalamic nucleus or internal pallidal stimulation in young onset Parkinson’s disease. Brain 1998; 121: 451–57.

70 Allert N, Lehrke R, Sturm V, Volkmann J. Secondary failure after ten years of pallidal neurostimulation in a patient with advanced Parkinson’s disease. J Neural Transm 2010; 117: 349–51.

71 Hariz MI, Krack P, Alesch F, et al. Multicentre European study of thalamic stimulation for parkinsonian tremor: a 6 year follow-up. J Neurol Neurosurg Psychiatry 2008; 79: 694–99.

72 Fasano A, Herzog J, Deuschl G. Selecting appropriate tremor patients for DBS. In: Bain P, Aziz T, Liu X, Nandi D, eds. Deep brain stimulation. Oxford: Oxford University Press, 2009.

73 Plaha P, Khan S, Gill SS. Bilateral stimulation of the caudal zona incerta nucleus for tremor control. J Neurol Neurosurg Psychiatry 2008; 79: 504–13.

74 Stefani A, Peppe A, Pierantozzi M, et al. Multi-target strategy for Parkinsonian patients: the role of deep brain stimulation in the centromedian-parafascicularis complex. Brain Res Bull 2009; 78: 113–18.

75 Bakker M, Esselink RA, Munneke M, Limousin-Dowsey P, Speelman HD, Bloem BR. Eff ects of stereotactic neurosurgery on postural instability and gait in Parkinson’s disease. Mov Disord 2004; 19: 1092–99.

76 Bejjani BP, Gervais D, Arnulf I, et al. Axial parkinsonian symptoms can be improved: the role of levodopa and bilateral subthalamic stimulation. J Neurol Neurosurg Psychiatry 2000; 68: 595–600.

77 Maurer C, Mergner T, Xie J, Faist M, Pollak P, Lucking CH. Eff ect of chronic bilateral subthalamic nucleus (STN) stimulation on postural control in Parkinson’s disease. Brain 2003; 126: 1146–63.

78 Ferraye MU, Debu B, Fraix V, et al. Eff ects of subthalamic nucleus stimulation and levodopa on freezing of gait in Parkinson disease. Neurology 2008; 70: 1431–37.

79 Stolze H, Klebe S, Poepping M, et al. Eff ects of bilateral subthalamic nucleus stimulation on parkinsonian gait. Neurology 2001; 57: 144–46.

80 Tommasi G, Lopiano L, Zibetti M, et al. Freezing and hypokinesia of gait induced by stimulation of the subthalamic region. J Neurol Sci 2007; 258: 99–103.

81 Moreau C, Defebvre L, Destee A, et al. STN-DBS frequency eff ects on freezing of gait in advanced Parkinson disease. Neurology 2008; 71: 80–84.

82 Ricchi V, Zibetti M, Angrisano S, et al. Transient eff ects of 80 Hz stimulation on gait in STN DBS treated PD patients: a 15 months follow-up study. Brain Stimul 2011; published online July 31. DOI:10.1016/j.brs.2011.07.001.

83 St George RJ, Nutt JG, Burchiel KJ, Horak FB. A meta-regression of the long-term eff ects of deep brain stimulation on balance and gait in PD. Neurology 2010; 75: 1292–99.

84 Stefani A, Lozano AM, Peppe A, et al. Bilateral deep brain stimulation of the pedunculopontine and subthalamic nuclei in severe Parkinson’s disease. Brain 2007; 130: 1596–607.

85 Moro E, Hamani C, Poon YY, et al. Unilateral pedunculopontine stimulation improves falls in Parkinson’s disease. Brain 2010; 133: 215–24.

86 Thevathasan W, Silburn PA, Brooker H, et al. The impact of low-frequency stimulation of the pedunculopontine nucleus region on reaction time in parkinsonism. J Neurol Neurosurg Psychiatry 2010; 81: 1099–104.

87 Deuschl G, Herzog J, Kleiner-Fisman G, et al. Deep brain stimulation: postoperative issues. Mov Disord 2006; 21 (suppl 14): S219–37.

88 Klostermann F, Ehlen F, Vesper J, et al. Eff ects of subthalamic deep brain stimulation on dysarthrophonia in Parkinson’s disease. J Neurol Neurosurg Psychiatry 2008; 79: 522–29.

89 Pinto S, Gentil M, Fraix V, Benabid AL, Pollak P. Bilateral subthalamic stimulation eff ects on oral force control in Parkinson’s disease. J Neurol 2003; 250: 179–87.

90 Rousseaux M, Krystkowiak P, Kozlowski O, Ozsancak C, Blond S, Destee A. Eff ects of subthalamic nucleus stimulation on parkinsonian dysarthria and speech intelligibility. J Neurol 2004; 251: 327–34.

91 Dromey C, Kumar R, Lang AE, Lozano AM. An investigation of the eff ects of subthalamic nucleus stimulation on acoustic measures of voice. Mov Disord 2000; 15: 1132–38.

92 Thobois S, Mertens P, Guenot M, et al. Subthalamic nucleus stimulation in Parkinson’s disease: clinical evaluation of 18 patients. J Neurol 2002; 249: 529–34.

93 Tripoliti E, Zrinzo L, Martinez-Torres I, et al. Eff ects of subthalamic stimulation on speech of consecutive patients with Parkinson disease. Neurology 2011; 76: 80–86.

94 Burghaus L, Hilker R, Thiel A, et al. Deep brain stimulation of the subthalamic nucleus reversibly deteriorates stuttering in advanced Parkinson’s disease. J Neural Transm 2006; 113: 625–31.

95 Toft M, Dietrichs E. Aggravated stuttering following subthalamic deep brain stimulation in Parkinson’s disease—two cases. BMC Neurol 2011; 11: 44.

96 Elia AE, Dollenz C, Soliveri P, Albanese A. Motor features and response to oral levodopa in patients with Parkinson’s disease under continuous dopaminergic infusion or deep brain stimulation. Eur J Neurol 2012; 19: 76–83.

97 Peppe A, Pierantozzi M, Bassi A, et al. Stimulation of the subthalamic nucleus compared with the globus pallidus internus in patients with Parkinson disease. J Neurosurg 2004; 101: 195–200.

98 Bejjani BP, Arnulf I, Demeret S, et al. Levodopa-induced dyskinesias in Parkinson’s disease: is sensitization reversible? Ann Neurol 2000; 47: 655–58.

99 Chaudhuri KR, Odin P. The challenge of non-motor symptoms in Parkinson’s disease. Prog Brain Res 2010; 184: 325–41.

100 Frank MJ, Samanta J, Moustafa AA, Sherman SJ. Hold your horses: impulsivity, deep brain stimulation, and medication in parkinsonism. Science 2007; 318: 1309–12.

101 Soulas T, Gurruchaga JM, Palfi S, Cesaro P, Nguyen JP, Fenelon G. Attempted and completed suicides after subthalamic nucleus stimulation for Parkinson’s disease. J Neurol Neurosurg Psychiatry 2008; 79: 952–54.

102 Voon V, Krack P, Lang AE, et al. A multicentre study on suicide outcomes following subthalamic stimulation for Parkinson’s disease. Brain 2008; 131: 2720–28.

103 Woods SP, Fields JA, Lyons KE, et al. Neuropsychological and quality of life changes following unilateral thalamic deep brain stimulation in Parkinson’s disease: a one-year follow-up. Acta Neurochir (Wien) 2001; 143: 1273–78.

104 Rouaud T, Dondaine T, Drapier S, et al. Pallidal stimulation in advanced Parkinson’s patients with contraindications for subthalamic stimulation. Mov Disord 2010; 25: 1839–46.

105 Alessandro S, Ceravolo R, Brusa L, et al. Non-motor functions in parkinsonian patients implanted in the pedunculopontine nucleus: focus on sleep and cognitive domains. J Neurol Sci 2010; 289: 44–48.

106 Costa A, Carlesimo GA, Caltagirone C, et al. Eff ects of deep brain stimulation of the peduncolopontine area on working memory tasks in patients with Parkinson’s disease. Parkinsonism Relat Disord 2010; 16: 64–67.

107 Ardouin C, Pillon B, Peiff er E, et al. Bilateral subthalamic or pallidal stimulation for Parkinson’s disease aff ects neither memory nor executive functions: a consecutive series of 62 patients. Ann Neurol 1999; 46: 217–23.

108 Daniele A, Albanese A, Contarino MF, et al. Cognitive and behavioural eff ects of chronic stimulation of the subthalamic nucleus in patients with Parkinson’s disease. J Neurol Neurosurg Psychiatry 2003; 74: 175–82.

109 Contarino MF, Daniele A, Sibilia AH, et al. Cognitive outcome 5 years after bilateral chronic stimulation of subthalamic nucleus in patients with Parkinson’s disease. J Neurol Neurosurg Psychiatry 2007; 78: 248–52.

110 Castelli L, Rizzi L, Zibetti M, Angrisano S, Lanotte M, Lopiano L. Neuropsychological changes 1-year after subthalamic DBS in PD patients: a prospective controlled study. Parkinsonism Relat Disord 2010; 16: 115–18.

111 Zangaglia R, Pacchetti C, Pasotti C, et al. Deep brain stimulation and cognitive functions in Parkinson’s disease: a three-year controlled study. Mov Disord 2009; 24: 1621–28.

112 De Gaspari D, Siri C, Di Gioia M, et al. Clinical correlates and cognitive underpinnings of verbal fl uency impairment after chronic subthalamic stimulation in Parkinson’s disease. Parkinsonism Relat Disord 2006; 12: 289–95.

www.thelancet.com/neurology Vol 11 May 2012 441

Review

113 Schroeder U, Kuehler A, Lange KW, et al. Subthalamic nucleus stimulation aff ects a frontotemporal network: a PET study. Ann Neurol 2003; 54: 445–50.

114 Gotham AM, Brown RG, Marsden CD. Frontal cognitive function in patients with Parkinson’s disease ‘on’ and ‘off ’ levodopa. Brain 1988; 111: 299–321.

115 Parsons TD, Rogers SA, Braaten AJ, Woods SP, Troster AI. Cognitive sequelae of subthalamic nucleus deep brain stimulation in Parkinson’s disease: a meta-analysis. Lancet Neurol 2006; 5: 578–88.

116 York MK, Dulay M, Macias A, et al. Cognitive declines following bilateral subthalamic nucleus deep brain stimulation for the treatment of Parkinson’s disease. J Neurol Neurosurg Psychiatry 2008; 79: 789–95.

117 Videnovic A, Metman LV. Deep brain stimulation for Parkinson’s disease: prevalence of adverse events and need for standardized reporting. Mov Disord 2008; 23: 343–49.

118 Ceravolo R, Brusa L, Galati S, et al. Low frequency stimulation of the nucleus tegmenti pedunculopontini increases cortical metabolism in Parkinsonian patients. Eur J Neurol 2011; 18: 842–49.

119 Weintraub D, Koester J, Potenza MN, et al. Impulse control disorders in Parkinson disease: a cross-sectional study of 3090 patients. Arch Neurol 2010; 67: 589–95.

120 Weintraub D, Siderowf AD, Potenza MN, et al. Association of dopamine agonist use with impulse control disorders in Parkinson disease. Arch Neurol 2006; 63: 969–73.

121 Witjas T, Baunez C, Henry JM, et al. Addiction in Parkinson’s disease: impact of subthalamic nucleus deep brain stimulation. Mov Disord 2005; 20: 1052–55.

122 Ardouin C, Voon V, Worbe Y, et al. Pathological gambling in Parkinson’s disease improves on chronic subthalamic nucleus stimulation. Mov Disord 2006; 21: 1941–46.

123 Lim SY, O’Sullivan SS, Kotschet K, et al. Dopamine dysregulation syndrome, impulse control disorders and punding after deep brain stimulation surgery for Parkinson’s disease. J Clin Neurosci 2009; 16: 1148–52.

124 Zahodne LB, Susatia F, Bowers D, et al. Binge eating in Parkinson’s disease: prevalence, correlates and the contribution of deep brain stimulation. J Neuropsychiatry Clin Neurosci 2011; 23: 56–62.

125 Romito LM, Raja M, Daniele A, et al. Transient mania with hypersexuality after surgery for high frequency stimulation of the subthalamic nucleus in Parkinson’s disease. Mov Disord 2002; 17: 1371–74.

126 Halbig TD, Tse W, Frisina PG, et al. Subthalamic deep brain stimulation and impulse control in Parkinson’s disease. Eur J Neurol 2009; 16: 493–97.

127 Demetriades P, Rickards H, Cavanna AE. Impulse control disorders following deep brain stimulation of the subthalamic nucleus in Parkinson’s disease: clinical aspects. Parkinsons Dis 2011; 2011: 658415.

128 Cavanagh JF, Wiecki TV, Cohen MX, et al. Subthalamic nucleus stimulation reverses mediofrontal infl uence over decision threshold. Nat Neurosci 2011; 14: 1462–67.

129 Rodriguez-Oroz MC, Lopez-Azcarate J, Garcia-Garcia D, et al. Involvement of the subthalamic nucleus in impulse control disorders associated with Parkinson’s disease. Brain 2011; 134: 36–49.

130 Silveira-Moriyama L, Evans AH, Katzenschlager R, Lees AJ. Punding and dyskinesias. Mov Disord 2006; 21: 2214–17.

131 Fasano A, Petrovic I. Insights into pathophysiology of punding reveal possible treatment strategies. Mol Psychiatry 2010; 15: 560–73.

132 Pallanti S, Bernardi S, Raglione LM, et al. Complex repetitive behavior: punding after bilateral subthalamic nucleus stimulation in Parkinson’s disease. Parkinsonism Relat Disord 2010; 16: 376–80.

133 Fasano A, Ricciardi L, Pettorruso M, Bentivoglio AR. Management of punding in Parkinson’s disease: an open-label prospective study. J Neurol 2011; 258: 656–60.

134 Thobois S, Ardouin C, Lhommee E, et al. Non-motor dopamine withdrawal syndrome after surgery for Parkinson’s disease: predictors and underlying mesolimbic denervation. Brain 2010; 133: 1111–27.

135 Czernecki V, Schupbach M, Yaici S, et al. Apathy following subthalamic stimulation in Parkinson disease: a dopamine responsive symptom. Mov Disord 2008; 23: 964–69.

136 Kirsch-Darrow L, Zahodne LB, Marsiske M, Okun MS, Foote KD, Bowers D. The trajectory of apathy after deep brain stimulation: from pre-surgery to 6 months post-surgery in Parkinson’s disease. Parkinsonism Relat Disord 2011; 17: 182–88.

137 Bejjani BP, Damier P, Arnulf I, et al. Transient acute depression induced by high-frequency deep-brain stimulation. N Engl J Med 1999; 340: 1476–80.

138 Temel Y, Kessels A, Tan S, Topdag A, Boon P, Visser-Vandewalle V. Behavioural changes after bilateral subthalamic stimulation in advanced Parkinson disease: a systematic review. Parkinsonism Relat Disord 2006; 12: 265–72.

139 Jahanshahi M, Ardouin CM, Brown RG, et al. The impact of deep brain stimulation on executive function in Parkinson’s disease. Brain 2000; 123: 1142–54.

140 Albanese A, Piacentini S, Romito LM, et al. Suicide after successful deep brain stimulation for movement disorders. Neurology 2005; 65: 499–500.

141 Temel Y, Tan S, Visser-Vandewalle V, Sharp T. Parkinson’s disease, DBS and suicide: a role for serotonin? Brain 2009; 132: e126; author reply e7.

142 Ulla M, Thobois S, Llorca PM, et al. Contact dependent reproducible hypomania induced by deep brain stimulation in Parkinson’s disease: clinical, anatomical and functional imaging study. J Neurol Neurosurg Psychiatry 2011; 82: 607–14.

143 Mandat TS, Hurwitz T, Honey CR. Hypomania as an adverse eff ect of subthalamic nucleus stimulation: report of two cases. Acta Neurochir (Wien) 2006; 148: 895–98.

144 Coenen VA, Honey CR, Hurwitz T, et al. Medial forebrain bundle stimulation as a pathophysiological mechanism for hypomania in subthalamic nucleus deep brain stimulation for Parkinson’s disease. Neurosurgery 2009; 64: 1106–15.

145 Raucher-Chene D, Charrel CL, de Maindreville AD, Limosin F. Manic episode with psychotic symptoms in a patient with Parkinson’s disease treated by subthalamic nucleus stimulation: improvement on switching the target. J Neurol Sci 2008; 273: 116–17.

146 Miyawaki E, Perlmutter JS, Troster AI, Videen TO, Koller WC. The behavioral complications of pallidal stimulation: a case report. Brain Cogn 2000; 42: 417–34.

147 Schupbach WM, Maltete D, Houeto JL, et al. Neurosurgery at an earlier stage of Parkinson disease: a randomized, controlled trial. Neurology 2007; 68: 267–71.

148 Witt K, Daniels C, Reiff J, et al. Neuropsychological and psychiatric changes after deep brain stimulation for Parkinson’s disease: a randomised, multicentre study. Lancet Neurol 2008; 7: 605–14.

149 Umemura A, Oka Y, Okita K, Matsukawa N, Yamada K. Subthalamic nucleus stimulation for Parkinson disease with severe medication-induced hallucinations or delusions. J Neurosurg 2011; 114: 1701–05.

150 Yoshida F, Miyagi Y, Kishimoto J, et al. Subthalamic nucleus stimulation does not cause deterioration of preexisting hallucinations in Parkinson’s disease patients. Stereotact Funct Neurosurg 2009; 87: 45–49.

151 Terao T, Okiyama R, Takahashi H, et al. [Comparison and examination of stereotactic surgical complications in movement disorders]. No Shinkei Geka 2003; 31: 629–36 (in Japanese).

152 Halim A, Baumgartner L, Binder DK. Eff ect of deep brain stimulation on autonomic dysfunction in patients with Parkinson’s disease. J Clin Neurosci 2011; 18: 804–06.

153 Trachani E, Constantoyannis C, Sirrou V, Kefalopoulou Z, Markaki E, Chroni E. Eff ects of subthalamic nucleus deep brain stimulation on sweating function in Parkinson’s disease. Clin Neurol Neurosurg 2010; 112: 213–17.

154 Herzog J, Weiss PH, Assmus A, et al. Subthalamic stimulation modulates cortical control of urinary bladder in Parkinson’s disease. Brain 2006; 129: 3366–75.

155 Finazzi-Agro E, Peppe A, D’Amico A, et al. Eff ects of subthalamic nucleus stimulation on urodynamic fi ndings in patients with Parkinson’s disease. J Urol 2003; 169: 1388–91.

156 Herzog J, Weiss PH, Assmus A, et al. Improved sensory gating of urinary bladder aff erents in Parkinson’s disease following subthalamic stimulation. Brain 2008; 131: 132–45.

157 Erola T, Haapaniemi T, Heikkinen E, Huikuri H, Myllya V. Subthalamic nucleus deep brain stimulation does not alter long-term heart rate variability in Parkinson’s disease. Clin Auton Res 2006; 16: 286–88.

158 Stemper B, Beric A, Welsch G, Haendl T, Sterio D, Hilz MJ. Deep brain stimulation improves orthostatic regulation of patients with Parkinson disease. Neurology 2006; 67: 1781–85.

442 www.thelancet.com/neurology Vol 11 May 2012

Review

159 Arnulf I, Bejjani BP, Garma L, et al. Improvement of sleep architecture in PD with subthalamic nucleus stimulation. Neurology 2000; 55: 1732–34.

160 Iranzo A, Valldeoriola F, Santamaria J, Tolosa E, Rumia J. Sleep symptoms and polysomnographic architecture in advanced Parkinson’s disease after chronic bilateral subthalamic stimulation. J Neurol Neurosurg Psychiatry 2002; 72: 661–64.

161 Monaca C, Ozsancak C, Jacquesson JM, et al. Eff ects of bilateral subthalamic stimulation on sleep in Parkinson’s disease. J Neurol 2004; 251: 214–18.

162 Chahine LM, Ahmed A, Sun Z. Eff ects of STN DBS for Parkinson’s disease on restless legs syndrome and other sleep-related measures. Parkinsonism Relat Disord 2011; 17: 208–11.

163 Lyons KE, Pahwa R. Eff ects of bilateral subthalamic nucleus stimulation on sleep, daytime sleepiness, and early morning dystonia in patients with Parkinson disease. J Neurosurg 2006; 104: 502–05.

164 Kedia S, Moro E, Tagliati M, Lang AE, Kumar R. Emergence of restless legs syndrome during subthalamic stimulation for Parkinson disease. Neurology 2004; 63: 2410–12.

165 Volkmann J, Albanese A, Kulisevsky J, et al. Long-term eff ects of pallidal or subthalamic deep brain stimulation on quality of life in Parkinson’s disease. Mov Disord 2009; 24: 1154–61.

166 Arnulf I, Bejjani BP, Garma L, et al. Eff ect of low and high frequency thalamic stimulation on sleep in patients with Parkinson’s disease and essential tremor. J Sleep Res 2000; 9: 55–62.

167 Romigi A, Placidi F, Peppe A, et al. Pedunculopontine nucleus stimulation infl uences REM sleep in Parkinson’s disease. Eur J Neurol 2008; 15: e64–65.

168 Lim AS, Moro E, Lozano AM, et al. Selective enhancement of rapid eye movement sleep by deep brain stimulation of the human pons. Ann Neurol 2009; 66: 110–14.

169 Arnulf I, Ferraye M, Fraix V, et al. Sleep induced by stimulation in the human pedunculopontine nucleus area. Ann Neurol 2010; 67: 546–49.

170 Kim HJ, Paek SH, Kim JY, et al. Chronic subthalamic deep brain stimulation improves pain in Parkinson disease. J Neurol 2008; 255: 1889–94.

171 Samura K, Miyagi Y, Morioka T, et al. Intractable facial pain in advanced Parkinson’s disease alleviated by subthalamic nucleus stimulation. J Neurol Neurosurg Psychiatry 2008; 79: 1410–11.

172 Gierthmuhlen J, Arning P, Binder A, et al. Infl uence of deep brain stimulation and levodopa on sensory signs in Parkinson’s disease. Mov Disord 2010; 25: 1195–202.

173 Romito LM, Contarino MF, Ghezzi D, Franzini A, Garavaglia B, Albanese A. High frequency stimulation of the subthalamic nucleus is effi cacious in Parkin disease. J Neurol 2005; 252: 208–11.

174 Lohmann E, Welter ML, Fraix V, et al. Are parkin patients particularly suited for deep-brain stimulation? Mov Disord 2008; 23: 740–43.

175 Moro E, Volkmann J, Konig IR, et al. Bilateral subthalamic stimulation in Parkin and PINK1 parkinsonism. Neurology 2008; 70: 1186–91.

176 Khan S, Mooney L, Plaha P, et al. Outcomes from stimulation of the caudal zona incerta and pedunculopontine nucleus in patients with Parkinson’s disease. Br J Neurosurg 2011; 25: 273–80.

177 Stefani A, Pierantozzi M, Ceravolo R, Brusa L, Galati S, Stanzione P. Deep brain stimulation of pedunculopontine tegmental nucleus (PPTg) promotes cognitive and metabolic changes: a target-specifi c eff ect or response to a low-frequency pattern of stimulation? Clin EEG Neurosci 2010; 41: 82–86.

178 Lyons KE, Koller WC, Wilkinson SB, Pahwa R. Long term safety and effi cacy of unilateral deep brain stimulation of the thalamus for parkinsonian tremor. J Neurol Neurosurg Psychiatry 2001; 71: 682–84.

179 Taba HA, Wu SS, Foote KD, et al. A closer look at unilateral versus bilateral deep brain stimulation: results of the National Institutes of Health COMPARE cohort. J Neurosurg 2010; 113: 1224–29.

180 Williams A, Gill S, Varma T, et al. Deep brain stimulation plus best medical therapy versus best medical therapy alone for advanced Parkinson’s disease (PD SURG trial): a randomised, open-label trial. Lancet Neurol 2010; 9: 581–91.

181 Deuschl G, Schade-Brittinger C, Krack P, et al. A randomized trial of deep-brain stimulation for Parkinson’s disease. N Engl J Med 2006; 355: 896–908.

182 Weaver FM, Follett K, Stern M, et al. Bilateral deep brain stimulation vs best medical therapy for patients with advanced Parkinson disease: a randomized controlled trial. JAMA 2009; 301: 63–73.

183 Merola A, Zibetti M, Artusi CA, et al. Subthalamic nucleus deep brain stimulation outcome in young onset Parkinson’s disease: a role for age at disease onset? J Neurol Neurosurg Psychiatry 2012; 83: 25–27.

184 Schupbach M, Gargiulo M, Welter ML, et al. Neurosurgery in Parkinson disease: a distressed mind in a repaired body? Neurology 2006; 66: 1811–16.

185 Saint-Cyr JA, Trepanier LL, Kumar R, Lozano AM, Lang AE. Neuropsychological consequences of chronic bilateral stimulation of the subthalamic nucleus in Parkinson’s disease. Brain 2000; 123: 2091–108.

186 Russmann H, Ghika J, Villemure JG, et al. Subthalamic nucleus deep brain stimulation in Parkinson disease patients over age 70 years. Neurology 2004; 63: 1952–54.

187 Parent B, Awan N, Berman SB, et al. The relevance of age and disease duration for intervention with subthalamic nucleus deep brain stimulation surgery in Parkinson disease. J Neurosurg 2011; 114: 927–31.

188 Derost PP, Ouchchane L, Morand D, et al. Is DBS-STN appropriate to treat severe Parkinson disease in an elderly population? Neurology 2007; 68: 1345–55.

189 Defer GL, Widner H, Marie RM, Remy P, Levivier M. Core assessment program for surgical interventional therapies in Parkinson’s disease (CAPSIT-PD). Mov Disord 1999; 14: 572–84.

190 Bronstein JM, Tagliati M, Alterman RL, et al. Deep brain stimulation for Parkinson disease: an expert consensus and review of key issues. Arch Neurol 2011; 68: 165.

191 Okun MS, Tagliati M, Pourfar M, et al. Management of referred deep brain stimulation failures: a retrospective analysis from 2 movement disorders centers. Arch Neurol 2005; 62: 1250–55.

192 Okun MS, Gallo BV, Mandybur G, et al. Subthalamic deep brain stimulation with a constant-current device in Parkinson’s disease: an open-label randomised controlled trial. Lancet Neurol 2012; 11: 140–49.

420 www.thelancet.com/neurology Vol 11 May 2012

Articles

Lancet Neurol 2012; 11: 420–28

Published OnlineApril 10, 2012

DOI:10.1016/S1474-4422(12)70056-X

See Comment page 386

*Members listed in appendix

University of British Columbia Hospital Multiple Sclerosis

Clinic, Vancouver, BC, Canada (V Devonshire MD); Department

of Neurology, First Faculty of Medicine, Charles University in Prague, Prague, Czech Republic (E Havrdova MD); Departments

of Neurology and Biomedicine, University Hospital, Basel,

Switzerland (Prof E W Radue MD,

Prof L Kappos MD); St Michael’s Hospital, Toronto, ON, Canada

(P O’Connor MD); Novartis, Basel, Switzerland

(L Zhang-Auberson MD, C Agoropoulou PhD,

D A Häring PhD); and Novartis, East Hanover, NJ, USA

(G Francis MD)

Correspondence to:Prof Ludwig Kappos,

Departments of Neurology and Biomedicine, University Hospital,

Basel, Switzerlandlkappos@uhbs.ch

See Online for appendix

Relapse and disability outcomes in patients with multiple sclerosis treated with fi ngolimod: subgroup analyses of the double-blind, randomised, placebo-controlled FREEDOMS studyVirginia Devonshire, Eva Havrdova, Ernst Wilhelm Radue, Paul O’Connor, Lixin Zhang-Auberson, Catherine Agoropoulou, Dieter Adrian Häring, Gordon Francis, Ludwig Kappos, for the FREEDOMS study group*

SummaryBackground Fingolimod 0·5 mg once daily is approved for treatment of relapsing multiple sclerosis (MS). In the phase 3, 2-year FREEDOMS (FTY720 Research Evaluating Eff ects of Daily Oral therapy in MS) study, fi ngolimod signifi cantly reduced annualised relapse rates (ARRs) and the risk of confi rmed disability progression compared with placebo. We aimed to investigate whether the benefi cial treatment eff ect reported for the overall population is consistent in subgroups of patients with diff erent baseline characteristics.

Methods We did subgroup analyses of ARRs (primary outcome) and confi rmed disability progression (a secondary outcome) over 24 months in the FREEDOMS study, a randomised, double-blind study that included 1272 patients with relapsing-remitting MS who were assigned 1:1:1 to fi ngolimod (0·5 mg or 1·25 mg) or placebo once daily for 24 months. Subgroups were predefi ned, predefi ned and slightly modifi ed, or defi ned post hoc, by demographic factors (including sex and age), disease characteristics (including baseline disability scores, relapse rates, and lesion parameters), and response to previous therapy (including analyses in patients eligible for fi ngolimod treatment according to the European label). Data were analysed by intention to treat. The FREEDOMS study is registered with ClinicalTrials.gov, number NCT00289978.

Findings Treatment with fi ngolimod 0·5 mg was associated with signifi cantly lower ARRs versus placebo across all subgroups except for patients aged over 40 years. ARR ratios ranged from 0·76 (95% CI 0·54–1·09; p=0·13) in patients aged over 40 years to 0·29 (0·16–0·52; p<0·0001) in patients who had relapse activity despite receiving interferon beta during the year before study enrolment. Hazard ratios for confi rmed disability progression over 24 months with fi ngolimod 0·5 mg versus placebo ranged from 0·85 (95% CI 0·53–1·36; p=0·50) in patients with a T2 lesion volume of 3300 mm³ or less to 0·32 (0·14–0·73; p=0·0066) in patients with an EDSS over 3·5. In patients who relapsed and had lesion activity despite treatment with interferon beta in the previous year, the ARR ratio for fi ngolimod 0·5 mg versus placebo was 0·38 (95% CI 0·21–0·68, p=0·0011), and for treatment-naive patients with rapidly evolving severe disease it was 0·33 (0·18–0·62, p=0·0006). Hazard ratios for confi rmed disability progression over 24 months were 0·68 (0·29–1·62; p=0·39) and 0·73 (0·25–2·07; p=0·55), respectively, in these groups.

Interpretation Patients with relapsing-remitting MS with a wide spectrum of clinical and MRI features including subgroups specifi ed by the European label can potentially benefi t from treatment with 0·5 mg fi ngolimod.

Funding Novartis.

IntroductionFingolimod 0·5 mg once daily is approved for the treatment of relapsing multiple sclerosis (MS) in many countries1 and in the European Union (EU) for the treatment of patients with high disease activity despite treatment with interferon beta or those with rapidly evolving severe relapsing-remitting MS (RRMS).2 Fingolimod has shown superior effi cacy on clinical and MRI outcomes compared with both an approved fi rst-line MS therapy—intramuscular interferon beta-1a—and placebo.1–6 In TRANSFORMS (Trial Assessing Injectable Interferon versus FTY720 Oral in RRMS),4 fi ngolimod 0·5 mg reduced the annualised relapse rate (ARR) by 52% compared with intramuscular interferon beta-1a over 1 year. In the 2-year FREEDOMS (FTY720 Research

Evaluating Eff ects of Daily Oral therapy in MS) study,3 fi ngolimod 0·5 mg signifi cantly reduced the ARR by 54%, from 0·40 (95% CI 0·34–0·47) in patients receiving placebo to 0·18 (0·15–0·22); the risk of disability progression after 3 and 6 months was reduced by 30% and 37%, respectively, over 24 months compared with placebo. Kaplan-Meier estimates for the proportions of patients free from 3-month confi rmed disability progression were 75·9% (95% CI 71·7–80·2%) in patients receiving placebo and 82·3% (78·6–86·1%) in patients treated with fi ngolimod 0·5 mg.3

As well as overall effi cacy, effi cacy in subgroups of patients who have high disease activity, and are therefore in urgent need of therapy, is of specifi c interest to clinicians. In MS, for which several treatment options

Articles

www.thelancet.com/neurology Vol 11 May 2012 421

are available, subgroup analyses can inform patients and physicians as they make treatment choices. A substantial proportion (41%) of patients included in the FREEDOMS study3 had received previous disease-modifying therapy (DMT) and the study thus provided an opportunity to assess the effi cacy of fi ngolimod treatment in patients who responded suboptimally to other available treat-ments. This assessment is important because up to 36% of patients who receive fi rst-line treatments for MS discontinue owing to perceived limited effi cacy.7,8

Here, we report subgroup analyses of ARRs (primary outcome) and disability progression (key secondary outcome) in the FREEDOMS study. The main objective of these analyses was to investigate whether the bene-fi cial treatment eff ect reported for the overall population is consistent in subgroups of patients with diff erent baseline characteristics and to describe potential clin-ically meaningful heterogeneities in treatment eff ect.

MethodsPatientsIn the FREEDOMS study,3 a randomised, double-blind, placebo-controlled, phase 3 study, 1272 patients were randomly assigned (1:1:1) to receive oral fi ngolimod capsules (0·5 mg or 1·25 mg) or matching placebo once daily for 24 months. Patients were aged 18–55 years, had RRMS diagnosed according to the 2005 McDonald criteria,9 had one or more docu mented relapses in the previous year or two or more in the previous 2 years, and had a score of 0–5·5 on the expanded disability status scale (EDSS). Key exclusion criteria were relapse or corticosteroid treatment within 30 days before randomisation, active infection, immune suppression (drug-induced or disease-induced), or clin ically signifi cant systemic disease. Interferon beta or glatiramer acetate therapy had to have been stopped 3 months or more before randomisation.

Patients were randomly assigned treatment at 138 centres in 22 countries from January, 2006, to August, 2007. Study protocols were approved by institutional review boards at each site and all patients gave written informed consent.

Randomisation and maskingRandomisation was done centrally, with the use of a validated system and stratifi cation according to site, with a block size of six within each site. Assessments with potential for unmasking were done by independent personnel: a specially trained and certifi ed examining neurologist determined all EDSS scores, and all MRI scans were analysed at a central MRI evaluation centre (University Hospital, Basel, Switzerland) by radiologists.

ProceduresThe primary analysis of the FREEDOMS study was a comparison of the ARRs (defi ned as the number of confi rmed relapses per year) between treatment groups in the overall intention-to-treat (ITT) population. Con-

fi rmed relapses were classifi ed based on an objective change on neurological examination, defi ned as an increase of at least 0·5 points on the EDSS, an increase of 1 point on two diff erent functional systems of the EDSS, or an increase of 2 points on one of the functional systems (excluding bowel, bladder or cerebral systems). The key secondary effi cacy outcome was time to disability progression confi rmed after 3 months for the overall ITT population. Disability progression was defi ned as an increase of 1 point on the EDSS from baseline (or 0·5 points if the baseline EDSS score was at least 5·5); disability progression confi rmed after 3 months required this EDSS criterion to be met at the time of onset of disability progression, at the 3-month confi rmation visit, and for any inter vening EDSS assessments.

Panel 1: Subgroup defi nitions

Baseline demographic factors and treatment history• Sex (men or women)*• Age (>40 years or ≤40 years)†‡• Treatment-naive or previously treated for multiple sclerosis (with any

disease-modifying therapy [DMT] at any time before study enrolment)*

Baseline disease characteristics• Number of relapses in the year before the study (≤1 relapse or >1 relapse)*• Number of relapses in the 2 years before the study (1, 2, or >2 relapses)†§• Baseline disability (expanded disability status scale score 0–3·5 or >3·5)†¶• Number of gadolinium-enhancing lesions at baseline (0 or ≥1)†||• T2 lesion volume (≤3300 mm³ or >3300 mm³)**

Disease activity in treatment-naive or previously treated patients• Group A: patients who received interferon beta during the year before study

enrolment but who had as many or more relapses in the year immediately before the study than in the 2 years before the study††

• Group B: patients who received any DMT during the year before study enrolment but who had as many or more relapses in the year immediately before the study than in the 2 years before the study††‡‡

• Group C: patients who received interferon beta during the year before study enrolment and had at least one relapse in the previous year plus at least either one gadolinium-enhancing T1 lesion or nine T2 lesions at baseline††

• Group D: patients who received any DMT during the year before study enrolment and had at least one relapse in the previous year plus at least either one gadolinium-enhancing T1 lesion or nine T2 lesions at baseline††‡‡

• Group E: treatment-naive rapidly evolving severe relapsing-remitting multiple sclerosis: ≥2 relapses within the year before baseline and ≥1 gadolinium-enhancing lesion at baseline††‡‡

*Predefi ned subgroup defi nition. †Predefi ned subgroup defi nition subsequently modifi ed. ‡Predefi ned subgroups were ≤37 and >37 years, and were changed to use an easier to remember number and to be comparable with the analysis for TRANSFORMS (Trial Assessing Injectable Interferon versus FTY720 Oral in RRMS).4 §Predefi ned subgroups were 0, 1, 2–3, 4–5, and >5 relapses, which were changed to combine groups with few patients. ¶Predefi ned subgroups were scores <3 and ≥3, which were changed to defi ne a group of more severely aff ected patients. ||Predefi ned subgroups were 0, 1–2, and ≥3 lesions, which were changed to group patients with or without infl ammation at baseline. **Defi ned post hoc. 3300 mm³ represents the approximate median T2 lesion burden in the fi ngolimod 0·5 mg group (median, 3304 mm³) and in the overall population (median, 3453 mm³). ††Defi ned post hoc at the request of the European Medicines Agency (EMA). ‡‡EMA defi nition modifi ed to include all disease-modifying drugs (not only interferon beta).

Articles

422 www.thelancet.com/neurology Vol 11 May 2012

We defi ned subgroups by demographic factors and treatment history, baseline disease characteristics, or disease activity in treatment-naive patients and previously treated patients. Of the seven predefi ned subgroups, four were subsequently modifi ed after database lock; the original defi nitions and the reasons for the change are described in panel 1. Six subgroups were defi ned post hoc, fi ve of which were defi ned at the request of the European Medicines Agency (EMA; panel 1).

Statistical analysesData were analysed according to the ITT principle. All patients who were randomly assigned and received at least one dose of study drug were included in the analysis. Patients who discontinued prematurely were kept in the study and relapses were counted regardless of whether they occurred on or off the study drug. No imputations were done for patients who prematurely discontinued from the study. In the overall FREEDOMS study population, the ARRs (primary analysis), corres-ponding 95% CIs, ARR ratios, and p values were estimated by a negative binomial regression model adjusted for treatment, country, number of relapses within 2 years before baseline, and baseline EDSS score. The log of the time on study (in years) was used as the off set variable to account for the varying lengths of time for which patients were in the study. The negative binomial regression model used is a natural and recommended way to model over-dispersed count data such as MS lesion counts or MS relapses.10 Compared with other possible methods to analyse relapse data, the negative binomial model has the advantages that it takes the recurrent nature of MS relapses into account and that it can be used for count data that are skewed, dispersed, or both.10

For the subgroup analyses, a less complex model was chosen because of the small size of some of the subgroups and to avoid convergence and collinearity problems between baseline adjustments and the sub-group variable.11 ARRs and 95% CIs, ARR ratios, and p values were estimated by a negative binomial regression model adjusted for treatment, subgroup variable, and treatment by subgroup variable interaction; log time on study was used as an off set variable.

In the overall FREEDOMS study population and sub-group populations, disability progression (key secondary outcome) was analysed with the log-rank test to compare time to disability progression curves for each treatment group generated by the Kaplan-Meier method for data up to 24 months. The time to disability progression was calculated from the date of the fi rst dose to the onset of disability progression. For patients with censored data, we used the time from fi rst dose to the last scheduled visit. We calculated Kaplan-Meier estimates and 95% CIs of proportions of patients with disability progression confi rmed after 3 months. As supportive analyses, we generated hazard ratios and corresponding

p values for time to disability progression confi rmed after 3 months with the Cox proportional hazards model adjusted for treatment, country, baseline EDSS score, and age for the overall FREEDOMS study population. Subgroup analyses were done at each level of the subgroup factor, rather than for the overall data, including Kaplan-Meier estimates and 95% CIs as well as hazard ratios from Cox proportional hazard models with only treatment in the linear predictor.

The FREEDOMS study is registered with ClinicalTrials.gov, number NCT00289978.

Fingolimod 0·5 mg (n=425)

Placebo (n=418)

Baseline demographic factors and treatment history

Sex

Men 0·18 (0·13–0·26) 0·56 (0·44–0·71)

Women 0·23 (0·19–0·28) 0·45 (0·39–0·53)

Age

>40 years 0·28 (0·21–0·37) 0·37 (0·29–0·46)

≤40 years 0·18 (0·15–0·23) 0·56 (0·47–0·66)

Treatment history*

Previously treated 0·28 (0·22–0·36) 0·53 (0·43–0·65)

Treatment-naive 0·17 (0·13–0·21) 0·46 (0·38–0·54)

Baseline disease characteristics

Number of relapses in year before study

>1 relapse 0·26 (0·20–0·34) 0·70 (0·58–0·85)

≤1 relapse 0·19 (0·15–0·23) 0·36 (0·30–0·43)

Number of relapses in 2 years before study

>2 relapses 0·31 (0·23–0·41) 0·61 (0·49–0·78)

2 relapses 0·21 (0·16–0·27) 0·47 (0·38–0·58)

1 relapse 0·14 (0·10–0·20) 0·38 (0·30–0·49)

Baseline disability

EDSS score >3·5 0·24 (0·16–0·37) 0·71 (0·53–0·96)

EDSS score 0–3·5 0·21 (0·17–0·25) 0·44 (0·38–0·51)

Number of gadolinium-enhancing lesions at baseline

≥1 0·28 (0·22–0·36) 0·70 (0·57–0·85)

0 0·18 (0·14–0·22) 0·36 (0·30–0·43)

T2 lesion volume

>3300 mm³ 0·28 (0·22–0·34) 0·58 (0·49–0·70)

≤3300 mm³ 0·15 (0·12–0·20) 0·38 (0·31–0·47)

Disease activity in treatment-naive or previously treated patients

Group A† 0·19 (0·12–0·31) 0·66 (0·46–0·94)

Group B† 0·22 (0·15–0·32) 0·57 (0·42–0·77)

Group C† 0·24 (0·15–0·38) 0·63 (0·44–0·91)

Group D† 0·26 (0·18–0·38) 0·54 (0·40–0·73)

Group E† 0·24 (0·15–0·40) 0·74 (0·49–1·11)

Data are annualised relapse rate (95% CI). EDSS=expanded disability status scale. Annualised relapse rates were estimated with a negative binomial regression model adjusted for treatment, subgroup variable, and treatment by subgroup variable interaction; log time on study was used as an offset variable. *Patients were categorised according to whether they were treatment-naive or had previously received treatment for multiple sclerosis with any disease-modifying therapy at any time before study enrolment. †Definitions are provided in panel 1.

Table 1: Annualised relapse rates

Articles

www.thelancet.com/neurology Vol 11 May 2012 423

Role of the funding sourceAn independent steering committee consisting of academic investigators collaborated with the sponsor (Novartis, Basel, Switzerland) to design the study and monitor its conduct. Data were collected by the inves-tigators; data management and analysis were done by the study sponsors. All authors had full access to all data in the study, participated in writing the manuscript, and approved the full manuscript. The corresponding author took fi nal responsibility for the decision to submit for publication.

ResultsBaseline characteristics have been reported previously,3 including mean age (placebo, 37·2 years [SD 8·6]; fi ngolimod 0·5 mg, 36·6 years [8·8]; fi ngolimod 1·25 mg, 37·4 years [8·9]), mean EDSS score (placebo, 2·5 [1·3]; fi ngolimod 0·5 mg, 2·3 [1·3]; fi ngolimod 1·25 mg, 2·4 [1·4]), mean number of gadolinium-enhancing lesions (placebo, 1·3 [2·9]; fi ngolimod 0·5 mg, 1·6 [5·6]; fi ngolimod 1·25 mg, 1·8 [4·7]) and number of relapses within the past 2 years (placebo, 2·2 [1·2]; fi ngolimod 0·5 mg, 2·1 [1·1]; fi ngolimod

Figure 1: Annualised relapse ratesDotted lines show the data points (placebo ARR estimate and ARR ratio) for the overall population. ARR=annualised relapse rate. RRMS=relapsing-remitting multiple sclerosis. EDSS=expanded disability status scale. *ARRs and ARR ratios were calculated with a negative binomial regression model adjusted for treatment, country, number of relapses within 2 years before baseline, and baseline EDSS score. †ARRs and ARR ratios were estimated with a negative binomial regression model adjusted for treatment, subgroup variable, and treatment by subgroup variable interaction; log time on study was used as an offset variable. ‡Patients were categorised according to whether they were treatment-naive or had previously received treatment for multiple sclerosis with any disease-modifying therapy at any time before study enrolment. §Of the total number of patients who could be categorised in this group. ¶Definitions are provided in panel 1.

Overall population* 425 418 0·46 (0·37–0·57) <0·0001Baseline demographic factors and treatment history†Sex Men 129 120 0·33 (0·22–0·50) <0·0001 Women 296 298 0·50 (0·39–0·65) <0·0001 Age >40 years 143 156 0·76 (0·54–1·09) 0·13 ≤40 years 282 262 0·33 (0·25–0·43) <0·0001 Treatment history‡ Previously treated 181 169 0·54 (0·39–0·74) <0·0001 Treatment naive 244 249 0·36 (0·27–0·49) <0·0001

Baseline disease characteristics†Number of relapses in year before study§ >1 160 155 0·37 (0·27–0·51) <0·0001 ≤1 265 263 0·52 (0·39–0·69) <0·0001 Number of relapses in 2 years before study§ >2 109 117 0·50 (0·34–0·72) 0·0003 2 188 172 0·45 (0·32–0·63) <0·0001 1 127 129 0·37 (0·24–0·58) <0·0001 Baseline disability EDSS score >3·5 62 72 0·34 (0·20–0·58) <0·0001 EDSS score 0–3·5 363 346 0·48 (0·38–0·60) <0·0001Number of gadolinium-enhancing lesions§ ≥1 161 154 0·40 (0·29–0·55) <0·0001 0 263 262 0·48 (0·36–0·65) <0·0001 T2 lesion volume§ >3300 mm³ 212 210 0·47 (0·36–0·63) <0·0001 ≤3300 mm³ 212 206 0·40 (0·29–0·57) <0·0001

Disease activity in treatment-naive or previously treated patients†Group A¶ 57 52 0·29 (0·16–0·52) <0·0001Group B¶ 84 80 0·38 (0·24–0·62) <0·0001Group C¶ 60 54 0·38 (0·21–0·68) 0·0011Group D¶ 90 79 0·49 (0·31–0·78) 0·0028Group E¶ 48 37 0·33 (0·18–0·62) 0·0006

Fingolimod Placebo ARR estimate in placebo ( ) Ratio (95% CI) p value 0·5 mg and fingolimod 0·5 mg ( ) groups

0·1 3·210·4 12·8

Favours fingolimod Favours placebo

0 1·00·5 0·80·2 1·6 6·4

Number of patients ARR (fingolimod 0·5 mg vs placebo)

Articles

424 www.thelancet.com/neurology Vol 11 May 2012

1·25 mg, 2·1 [1·3]). 520 of 1272 (41%) patients had previously received one or more DMT;3 of those who had been previously treated, 367 of 520 (71%) had a history of interferon beta use.

Fingolimod 0·5 mg signifi cantly reduced ARRs over 2 years versus placebo in all subgroups (table 1, fi gure 1) except for patients aged over 40 years (24% decrease in ARR; p=0·13). In the other patient groups defi ned by baseline demographic factors or treatment history, ARRs were signifi cantly reduced by fi ngolimod 0·5 mg by up to 67% versus placebo over 24 months. The greatest reductions were noted in men and in patients aged 40 years or less (67% and p<0·0001 for both).

Across the subgroups defi ned by baseline disease characteristics, ARRs were signifi cantly reduced by fi ngolimod 0·5 mg by 48–66% versus placebo over 24 months: the lowest reduction was seen in patients who had experienced a maximum of one relapse in the year before the study (48%; p<0·0001; fi gure 1), and the greatest reduction occurred in patients with baseline EDSS scores of more than 3·5 (66%; p<0·0001).

For subgroups defi ned by disease activity, those with high disease activity despite treatment with interferon beta in the previous year (groups A and C) and representative of the population of patients for whom fi ngolimod treatment is approved in the EU, had sig-nifi cant ARR reductions (of 71% and 62%, respectively) with fi ngolimod 0·5 mg over 24 months (table 1; fi gure 1). Similarly, in patients who had high disease activity despite treatment with any DMT in the previous year (groups B and D), ARRs were signifi cantly reduced by fi ngolimod 0·5 mg (by 62% and 51%, respectively) over 24 months. In treatment-naive patients with rapidly evolving severe RRMS, ARR was signifi cantly reduced by fi ngolimod 0·5 mg by 67% versus placebo over 24 months.

The ARR results in subgroups treated with fi ngolimod 1·25 mg were consistent with those for fi ngolimod 0·5 mg (appendix), but with fi ngolimod 1·25 mg the ARR was also signifi cantly reduced (by 49%) in patients aged over 40 years.

The risk of disability progression was numerically lower in the fi ngolimod 0·5 mg group versus placebo by 15–68% across all subgroups over 24 months (table 2; fi gure 2), although the eff ects were not statistically signifi cant in most cases. Least evidence of an eff ect was noted in patients with low disease activity at baseline (one relapse in the 2 years before the study, 16%; p=0·56) and low T2 volume (≤3300 mm³, 15%; p=0·50). The largest eff ect sizes were seen in men (57%; p=0·0097), in patients who had more than two relapses in the 2 years before the study (60%; p=0·0058), and in patients with advanced disability (EDSS score >3·5, 68%; p=0·0066). Signifi cant diff erences in treatment eff ects were also noted in treatment-naive patients (37%; p=0·030) and patients with a high baseline T2 lesion volume (41%; >3300 mm³; p=0·015).

In patients who had high disease activity despite treatment with interferon beta in the previous year (groups A and C), the risk of disability progression was numerically, but not signifi cantly, lower by 36% and 32%, respectively, in the fi ngolimod 0·5 mg group versus the placebo group over 24 months (table 2; fi gure 2). In patients who had high disease activity despite treatment with any DMT in the previous year (groups B and D), this same risk was numerically, but not signifi cantly, lower by 41% and 46%, respectively. In treatment-naive patients with rapidly evolving severe RRMS, the risk of disability progression was also numerically, but not signifi cantly, lower for fi ngolimod 0·5 mg by 27% versus

Fingolimod 0·5 mg (n=425)

Placebo (n=418)

Baseline demographic factors and treatment history

Sex

Men 86·7 (80·7–92·8) 75·0 (67·0–83·1)

Women 80·6 (76·0–85·1) 76·3 (71·3–81·3)

Age

>40 years 78·6 (71·7–85·6) 71·3 (64·1–78·6)

≤40 years 84·2 (79·9–88·5) 78·9 (73·7–84·1)

Treatment history*

Previously treated 80·1 (74·1–86·1) 76·3 (69·6–83·0)

Treatment-naive 84·0 (79·3–88·6) 75·7 (70·2–81·2)

Baseline disease characteristics

Number of relapses in year before study

>1 relapse 83·0 (77·1–89·0) 75·5 (68·4–82·6)

≤1 relapse 81·9 (77·2–86·7) 76·2 (70·9–81·5)

Number of relapses in 2 years before study

>2 relapses 85·6 (78·8–92·3) 70·7 (62·1–79·2)

2 relapses 81·0 (75·3–86·8) 75·5 (68·8–82·2)

1 relapse 81·4 (74·5–88·3) 81·2 (74·3–88·2)

Baseline disability

EDSS score >3·5 84·3 (74·9–93·8) 63·7 (52·0–75·4)

EDSS score 0–3·5 82·0 (77·9–86·0) 78·4 (73·9–82·9)

Number of gadolinium-enhancing lesions at baseline

≥1 83·0 (77·1–89·0) 73·8 (66·5–81·1)

0 81·9 (77·2–86·7) 77·0 (71·7–82·3)

T2 lesion volume

>3300 mm³ 81·6 (76·2–87·0) 71·7 (65·4–78·0)

≤3300 mm³ 83·1 (77·9–88·2) 80·1 (74·4–85·8)

Disease activity in treatment-naive or previously treated patients

Group A† 84·3 (74·8–93·7) 74·6 (62·0–87·1)

Group B† 84·7 (77·0–92·4) 74·9 (64·8–85·0)

Group C† 83·4 (73·5–93·3) 75·1 (62·8–87·4)

Group D† 85·0 (77·2–92·9) 73·4 (63·4–83·5)

Group E† 84·7 (74·3–95·2) 78·9 (64·9–92·8)

Data are Kaplan-Meier estimates of patients free from disease progression (95% CI). EDSS=expanded disability status scale. *Patients were categorised according to whether they were treatment-naive or had previously received treatment for multiple sclerosis with any disease-modifying therapy at any time before study enrolment. †Defi nitions are provided in panel 1.

Table 2: Freedom from disability progression confi rmed after 3 months

Articles

www.thelancet.com/neurology Vol 11 May 2012 425

placebo. Consistent results were noted for patients treated with fi ngolimod 1·25 mg (appendix).

DiscussionHere we report subgroup analyses of the eff ects of fi ngolimod versus placebo on relapse rates and disability progression in patient subgroups in the FREEDOMS study.3 Overall, our analyses show broadly consistent eff ects of fi ngolimod across most subgroups of patients.

Subgroup analyses provide an appropriate means to investigate the generalisability and validity of the overall

conclusions of a clinical trial and allow identifi cation of patient groups in which the treatment has lower or higher than average effi cacy. However, some methodo-logical limitations exist: the FREEDOMS study was not prospectively designed, or powered, to test for treatment diff erences within subgroups or to test formally for heterogeneity between subgroups. As a consequence, these analyses might have been underpowered, which is a potential source of false negative results. Additionally, we did many subgroup analyses and did not make any formal adjustments to correct for multiplicity, in

Figure 2: Disability progression confi rmed after 3 monthsDotted lines show the datapoints (placebo Kaplan-Meier estimate and hazard ratio) for the overall population. EDSS=expanded disability status scale. RRMS=relapsing-remitting multiple sclerosis. *Hazard ratios calculated by Cox proportional hazard models adjusted for treatment, country, baseline EDSS score, and age. †Hazard ratios calculated by Cox proportional hazard models with only treatment in the linear predictor. ‡Patients were categorised according to whether they were treatment-naive or had previously received treatment for multiple sclerosis with any disease-modifying therapy at any time before study enrolment. §Of the patients who could be categorised in this group. ¶Defi nitions are provided in panel 1.

Overall population* 425 418 0·70 (0·52–0·96) 0·024Baseline demographic factors and treatment history†Sex Men 129 120 0·43 (0·22–0·81) 0·0097 Women 296 298 0·77 (0·53–1·10) 0·15 Age >40 years 143 156 0·74 (0·46–1·19) 0·22 ≤40 years 282 262 0·68 (0·45–1·02) 0·062 Treatment history‡ Previously treated 181 169 0·70 (0·43–1·14) 0·15 Treatment naive 244 249 0·63 (0·41–0·95) 0·030

Baseline disease characteristics†Number of relapses in year before study§ >1 160 155 0·62 (0·37–1·05) 0·075 ≤1 265 263 0·70 (0·47–1·03) 0·070 Number of relapses in 2 years before study§ >2 109 117 0·40 (0·21–0·77) 0·0058 2 188 172 0·71 (0·44–1·13) 0·15 1 127 129 0·84 (0·46–1·52) 0·56 Baseline disability EDSS score >3·5 62 72 0·32 (0·14–0·73) 0·0066 EDSS score 0–3·5 363 346 0·77 (0·55–1·09) 0·15Number of gadolinium-enhancing lesions§ ≥1 161 154 0·62 (0·37–1·04) 0·071 0 263 262 0·75 (0·50–1·11) 0·14 T2 lesion volume§ >3300 mm³ 212 210 0·59 (0·38–0·90) 0·015 ≤3300 mm³ 212 206 0·85 (0·53–1·36) 0·50

Disease activity in treatment-naive or previously treated patients†Group A¶ 57 52 0·64 (0·27–1·51) 0·31Group B¶ 84 80 0·59 (0·29–1·20) 0·14Group C¶ 60 54 0·68 (0·29–1·62) 0·39Group D¶ 90 79 0·54 (0·26–1·10) 0·092Group E¶ 48 37 0·73 (0·25–2·07) 0·55

Fingolimod Placebo Kaplan-Meier estimate with disability Ratio (95% CI) p value 0·5 mg progression (%) in placebo ( ) and fingolimod 0·5 mg ( ) groups

Number of patients Disability progression (fingolimod 0·5 mg vs placebo)

0·1 3·210·4 12·8

Favours fingolimod Favours placebo

0 10050 0·80·2 1·6 6·4

Articles

426 www.thelancet.com/neurology Vol 11 May 2012

accordance with previous practice,12 which potentially introduced a risk of false positive results. For these reasons, the p values should be interpreted with caution and over-interpretation avoided; neverthe less, inter-pretation of the model estimates of activity within treatment arms and of the estimates of between-treatment eff ects is valid, although the 95% CIs were wide in some of the small subgroups.

Results from predefi ned subgroups are sometimes deemed more meaningful than those from post-hoc defi ned subgroups, although the aforementioned con-cerns related to power and multiplicity are unrelated to whether or not a subgroup has been predefi ned. In this study, analyses were predefi ned, predefi ned but slightly

modifi ed, or defi ned post hoc (including post-hoc analyses requested by the EMA). All subgroups were defi ned on the basis of clinical-scientifi c criteria of interest and not in a data-driven manner. Any modifi -cations to predefi ned subgroups were made either to amalgamate subgroups with very few patients or to adjust the cutoff value, not to aff ect the overall conclusions. The fi ve post-hoc subgroups, which refer to persistence of disease activity despite previous use of interferon beta or any DMT and to treatment-naive patients with rapidly evolving severe RRMS, were defi ned after suggestions by the EMA as part of the submission or post-approval processes for fi ngolimod.

Notwithstanding the general limitations of subgroup analyses, the eff ects reported with the licensed 0·5 mg dose of fi ngolimod in subgroups of patients are broadly consistent with the reductions in relapse rate and risk of disability progression in the overall study population. In the overall population, ARR was reduced by 54%,3 the risk of disability progression was reduced by 30% by fi ngolimod 0·5 mg and the Kaplan-Meier estimates for the proportions of patients free from disability pro-gression were 75·9% in patients receiving placebo and 82·3% in patients treated with fi ngolimod 0·5 mg.

In patients previously treated with interferon beta (refl ecting the populations of patients for whom fi ngolimod therapy is approved in the EU; groups A, C, and E) the eff ect sizes of fi ngolimod 0·5 mg exceeded those in the overall population. Across subgroups, mean ARRs for the placebo groups ranged between 0·36 and 0·74, compared with between 0·14 and 0·31 in patients treated with fi ngolimod 0·5 mg. The general consistency of the reductions in ARRs with fi ngolimod 0·5 mg across most subgroups, despite the small size of some of the groups, shows the robustness of these results.

There was little evidence of heterogeneity of the treatment eff ect across subgroups. However, patients aged over 40 years or more did not have a signifi cant reduction in relapse rate with fi ngolimod 0·5 mg treatment versus placebo by contrast with all other patient subgroups. The reduced eff ect size in patients aged over 40 years was noted with both doses, but the eff ect of fi ngolimod 1·25 mg in reducing relapse rates in this subgroup was signifi cant. Although we did not directly compare the doses for any of the subgroups studied, this fi nding could suggest that the fi ngolimod 1·25 mg dose is more eff ective in reducing relapses in patients aged over 40 years than the lower dose.

Possible diff erences in treatment benefi t by age have been reported for both natalizumab and cladribine, for which treatment eff ect was greater in younger than in older patients.13,14 DMT has previously been suggested to have greater treatment benefi ts in younger than in older patients because of the higher rate of relapse activity reported in younger populations with MS.15 However, in the present study, the diff erential response to study treatment could not be explained by the higher relapse

Panel 2: Research in context

Systematic reviewWe did a PubMed search on Nov 7, 2011 for “multiple sclerosis AND (subgroup OR subpopulation OR subset)”, restricted to the titles of papers and to papers published in English. No date restrictions were set. Results were screened to identify reports about treatment eff ects of disease-modifying therapies (DMTs) in patients with multiple sclerosis (MS). Of 39 articles retrieved, fi ve assessed the eff ects of DMTs (BG-12, alemtuzumab, interferon beta, cladribine, and natalizumab) in patient subgroups. Of these, only two studies were placebo controlled and used relapse and disability as outcome measures; these studies were the AFFIRM13 (Natalizumab Safety and Effi cacy in Relapsing Remitting Multiple Sclerosis Study) trial and the CLARITY (Cladribine Tablets Treating Multiple Sclerosis Orally) trial,14 which were used as the most relevant literature comparisons.

InterpretationFingolimod 0·5 mg once daily is approved as a first-line treatment for relapsing MS in some countries and for patients with relapsing-remitting MS (RRMS) with inadequate response to interferon beta or with high disease activity in other countries, including some in the EU. Fingolimod 0·5 mg reduced the annualised relapse rate (ARR) by 54% and the risk of 3-month (and 6-month) confirmed disability progression by 30% (and 37%), relative to placebo over 2 years3 and also reduced the ARR over 1 year relative to intramuscular interferon beta-1a by 52%.4 Therefore, fingolimod 0·5 mg compares favourably with other treatment options available for relapsing MS. In our subgroup analysis of the placebo-controlled, 2-year FREEDOMS (FTY720 Research Evaluating Effects of Daily Oral therapy in MS) study, this efficacy in terms of ARRs was confirmed across most subgroups of patients with RRMS and seemed to be more pronounced in patients with high pre-baseline disease activity, including those who fulfilled the definition of highly active RRMS used on the European label for fingolimod.

Articles

www.thelancet.com/neurology Vol 11 May 2012 427

activity in younger patients alone (multiple regression modelling; data not shown). When adjusted for baseline relapse activity in the last 2 years, patients who were aged 40 years or less had a stronger reduction in ARR (68%) when treated with fi ngolimod 0·5 mg versus placebo than patients who were aged over 40 years (22%; ARR ratio for ≤40 years 0·323, 95% CI 0·246–0·425; ARR ratio for >40 years 0·779, 0·549–1·107).

The highest relative reductions in relapse rates (>65%) were noted in subgroups with higher than average relapse activity in the placebo group, including patients with EDSS scores of more than 3·5, patients who were aged 40 years or less, treatment-naive patients with rapidly evolving severe RRMS, and patients with high disease activity despite previous treatment with inter-feron beta. The reduction in ARR with 0·5 mg fi ngolimod was noted for men and women, although men seemed to have a higher relative reduction in ARR than women and also seemed to have higher relapse rates in the placebo group. Treatment eff ects on relapse outcomes favouring men over women have been described for interferon beta treatment,16 but a post-hoc analysis of pooled data from 1406 patients (1027 women and 379 men) enrolled in fi ve clinical studies of intramuscular interferon beta-1a did not show signifi -cant treatment-by-sex interactions.17 The eff ects of natalizumab treatment in subgroup analyses favoured women over men and those with lower baseline EDSS scores (<3·5) over those with higher scores.13

Considering the width of the 95% CIs we report, and the inconsistencies within published work regarding subgroup eff ects, some of the suggested diff erential treatment eff ects are probably due to chance. Finding about one signifi cant subgroup-by-treatment interaction after undertaking 13 independent statistical interaction tests (α level of 0·05) in subgroups would have been expected by chance if the null hypothesis (no eff ect of subgroups in reality) was true.

The risk of disability progression was numerically lower in the fi ngolimod treatment group versus placebo in all patient subgroups, although this was not signifi cant in most cases. Considering the level of uncertainty (ie, the width of the 95% CIs), the subgroup results are broadly consistent with the average reduction in disability progression of 30% reported for the overall study. Among the overall FREEDOMS study population, disability progression was moderate during the 24-month study (94/418 [22%] of the placebo-treated patients and 72/425 [17%] of fi ngolimod 0·5 mg-treated patients had 3 month-confi rmed disability progression at month 24) and therefore the number of events noted in some of the subgroups was very low. Thus, that the disability progression fi ndings were often not signifi cant in subgroups is not surprising. Nevertheless, numerical diff erences in certain patient subgroups are more apparent, particularly in men, patients who had more than two relapses in the 2 years before the study,

patients with baseline EDSS scores of more than 3·5, patients with high disease activity despite treatment with any DMT (groups B and D), and patients with baseline T2 lesion volume of more than 3300 mm³. A greater eff ect of DMT on disability progression in patients with high baseline relapse numbers and T2 lesion counts has been reported for natalizumab.13 Also, in cladribine-treated patients a greater eff ect on disease activity (a composite measure of relapse and disability outcomes) was described in patient subgroups with high relapse or lesion activity at baseline.14 The eff ect size of natalizumab and cladribine 5·25 mg/kg on the risk of disability progression was smaller in patients with high baseline EDSS scores (≥3·5) than in patients with low EDSS scores (<3·5).13,14

Overall, our analyses suggest that patients with RRMS with a wide spectrum of clinical and MRI features can potentially benefi t from this oral treatment (panel 2).

ContributorsVD analysed and interpreted the data, and wrote and reviewed the

manuscript. EH collected and interpreted the data and reviewed the

manuscript. EWR and PO’C collected, analysed, and interpreted the data

and reviewed the manuscript. LZ-A and CA analysed and interpreted the

data and reviewed the manuscript. DAH analysed and interpreted the

data, helped to prepare the fi gures, and wrote and reviewed the

manuscript. GF analysed and interpreted the data and planned, wrote,

and reviewed the manuscript. LK designed the study, collected, analysed,

and interpreted the data, did the literature searches, and wrote and

reviewed the manuscript.

Confl icts of interestVD has received fees for consultancy or speaking in the past 3 years

from Novartis. EH has received fees for board membership,

consultancy or speaking, or grants in the past 3 years from Bayer,

Biogen Idec, Genzyme, Merck Serono, Novartis, and Teva. EWR has

received fees for board membership, consultancy, or speaking in the

past 3 years from Bayer Schering, Biogen Idec, Merck Serono, and

Novartis. PO’C has received fees for board membership or consultancy,

or grants in the past 3 years from Actelion, Biogen Idec, Celgene, EMD

Serono, Novartis, Sanofi Genzyme, Receptos, and Teva. The University

Hospital Basel, as employer of LK, has received and dedicated to

research support fees for board membership, consultancy or speaking,

or grants in the past 3 years from Actelion, Advancell, Allozyne, Bayer,

Bayhill, Biogen Idec, BioMarin, CSL Behring, Eli Lilly, EU, Genmab,

GeNeuro SA, Gianni Rubatto Foundation, Glenmark, Merck Serono,

MediciNova, Mitsubishi Pharma, Novartis, Novartis Research

Foundation, Novonordisk, Peptimmune, Roche, Roche Research

Foundation, Santhera, Sanofi -Aventis, Swiss MS Society, Swiss

National Research Foundation, Teva, UCB, and Wyeth. CA, LZ-A, DAH

and GF are employees of Novartis.

AcknowledgmentsWe thank the patients who participated in the FREEDOMS study, the

study site personnel, and Louise Verrall (Oxford Pharmagenesis,

funded by Novartis) and Hashem Salloukh (Novartis) for editorial

assistance. EH has been supported by the Czech Ministry of

Education, Research program MSM 0021620849. LK is supported by

the Swiss MS Society. PO’C receives support from the Canadian MS

Society in his role as Medical Advisor.

References1 US Food and Drug Administration. Gilenya prescribing

information, 2011. http://www.accessdata.fda.gov/drugsatfda_docs/label/2011/022527s002lbl.pdf (accessed Nov 30, 2011).

2 European Medicines Agency. Gilenya summary of product characteristics, 2011. http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Product_Information/human/002202/WC500104528.pdf (accessed Nov 30, 2011).

Articles

428 www.thelancet.com/neurology Vol 11 May 2012

3 Kappos L, Radue EW, O’Connor P, et al. A placebo-controlled trial of oral fi ngolimod in relapsing multiple sclerosis. N Engl J Med 2010; 362: 387–401.

4 Cohen JA, Barkhof F, Comi G, et al. Oral fi ngolimod or intramuscular interferon for relapsing multiple sclerosis. N Engl J Med 2010; 362: 402–15.

5 Comi G, O’Connor P, Montalban X, et al. Phase II study of oral fi ngolimod (FTY720) in multiple sclerosis: 3-year results. Mult Scler 2010; 16: 197–207.

6 Kappos L, Antel J, Comi G, et al. Oral fi ngolimod (FTY720) for relapsing multiple sclerosis. N Engl J Med 2006; 355: 1124–40.

7 Portaccio E, Zipoli V, Siracusa G, Sorbi S, Amato MP. Long-term adherence to interferon beta therapy in relapsing-remitting multiple sclerosis. Eur Neurol 2008; 59: 131–35.

8 Zwibel HL. Glatiramer acetate in treatment-naive and prior interferon-beta-1b-treated multiple sclerosis patients. Acta Neurol Scand 2006; 113: 378–86.

9 Polman CH, Reingold SC, Edan G, et al. Diagnostic criteria for multiple sclerosis: 2005 revisions to the “McDonald Criteria”. Ann Neurol 2005; 58: 840–46.

10 Wang YC, Meyerson L, Tang YQ, Qian N. Statistical methods for the analysis of relapse data in MS clinical trials. J Neurol Sci 2009; 285: 206–11.

11 Wang R, Lagakos SW, Ware JH, Hunter DJ, Drazen JM. Statistics in medicine—reporting of subgroup analyses in clinical trials. N Engl J Med 2007; 357: 2189–94.

12 Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology 1990; 1: 43–46.

13 Hutchinson M, Kappos L, Calabresi PA, et al. The effi cacy of natalizumab in patients with relapsing multiple sclerosis: subgroup analyses of AFFIRM and SENTINEL. J Neurol 2009; 256: 405–15.

14 Giovannoni G, Cook S, Rammohan K, et al. Sustained disease-activity-free status in patients with relapsing-remitting multiple sclerosis treated with cladribine tablets in the CLARITY study: a post-hoc and subgroup analysis. Lancet Neurol 2011; 10: 329–37.

15 Tremlett H, Zhao Y, Joseph J, Devonshire V. Relapses in multiple sclerosis are age- and time-dependent. J Neurol Neurosurg Psychiatry 2008; 79: 1368–74.

16 Trojano M, Pellegrini F, Paolicelli D, et al. Post-marketing of disease modifying drugs in multiple sclerosis: an exploratory analysis of gender eff ect in interferon beta treatment. J Neurol Sci 2009; 286: 109–13.

17 Rudick RA, Kappos L, Kinkel R, et al. Gender eff ects on intramuscular interferon beta-1a in relapsing-remitting multiple sclerosis: analysis of 1406 patients. Mult Scler 2011; 17: 353–60.

www.thelancet.com/neurology Vol 11 May 2012 397

Articles

Lancet Neurol 2012; 11: 397–404

Published OnlineApril 11, 2012DOI:10.1016/S1474-4422(12)70057-1

See Comment page 382

Department of Neurology (S Walter MD, P Kostopoulos MD, Prof A Haass MD, I Keller MD, M Lesmeister, S Helwig, Y Liu MD, Prof K Fassbender MD), Department of Neuroradiology (C Roth MD, P Papanagiotou MD, J Viera, H Körner, M Alexandrou, U Yilmaz MD, K Ziegler MD, K Schmidt MD, R Dabew, Prof W Reith MD), and Department of Anaesthesiology (D Kubulus MD, Prof T Volk MD), University Hospital of the Saarland, Homburg, Germany; Emergency Service Saarland, Germany (T Schlechtriemen MD); Department of Acute Vascular Imaging Centre, John Radcliff e Hospital, Oxford, UK (Prof I Grunwald MD); Boehringer Ingelheim, Ingelheim, Germany (H Schumacher PhD); Interdisciplinary Centre for Clinical Trials (IZKS), Mainz, Germany (K Kronfeld MD, C Ruckes PhD); and Department of Clinical Chemistry, Nuremberg Hospital, Nuremberg, Germany (T Bertsch MD)

Correspondence to:Prof Klaus Fassbender, Department of Neurology, University of the Saarland, 66424 Homburg, Germanyklaus.fassbender@uks.eu

Diagnosis and treatment of patients with stroke in a mobile stroke unit versus in hospital: a randomised controlled trialSilke Walter, Panagiotis Kostopoulos, Anton Haass, Isabel Keller, Martin Lesmeister, Thomas Schlechtriemen, Christian Roth, Panagiotis Papanagiotou, Iris Grunwald, Helmut Schumacher, Stephan Helwig, Julio Viera, Heiko Körner, Maria Alexandrou, Umut Yilmaz, Karin Ziegler, Kathrin Schmidt, Rainer Dabew, Darius Kubulus, Yang Liu, Thomas Volk, Kai Kronfeld, Christian Ruckes, Thomas Bertsch, Wolfgang Reith, Klaus Fassbender

SummaryBackground Only 2–5% of patients who have a stroke receive thrombolytic treatment, mainly because of delay in reaching the hospital. We aimed to assess the effi cacy of a new approach of diagnosis and treatment starting at the emergency site, rather than after hospital arrival, in reducing delay in stroke therapy.

Methods We did a randomised single-centre controlled trial to compare the time from alarm (emergency call) to therapy decision between mobile stroke unit (MSU) and hospital intervention. For inclusion in our study patients needed to be aged 18–80 years and have one or more stroke symptoms that started within the previous 2·5 h. In accordance with our week-wise randomisation plan, patients received either prehospital stroke treatment in a specialised ambulance (equipped with a CT scanner, point-of-care laboratory, and telemedicine connection) or optimised conventional hospital-based stroke treatment (control group) with a 7 day follow-up. Allocation was not masked from patients and investigators. Our primary endpoint was time from alarm to therapy decision, which was analysed with the Mann-Whitney U test. Our secondary endpoints included times from alarm to end of CT and to end of laboratory analysis, number of patients receiving intravenous thrombolysis, time from alarm to intravenous thrombolysis, and neurological outcome. We also assessed safety endpoints. This study is registered with ClinicalTrials.gov, number NCT00153036.

Findings We stopped the trial after our planned interim analysis at 100 of 200 planned patients (53 in the prehospital stroke treatment group, 47 in the control group), because we had met our prespecifi ed criteria for study termination. Prehospital stroke treatment reduced the median time from alarm to therapy decision substantially: 35 min (IQR 31–39) versus 76 min (63–94), p<0·0001; median diff erence 41 min (95% CI 36–48 min). We also detected similar gains regarding times from alarm to end of CT, and alarm to end of laboratory analysis, and to intravenous thrombolysis for eligible ischaemic stroke patients, although there was no substantial diff erence in number of patients who received intravenous thrombolysis or in neurological outcome. Safety endpoints seemed similar across the groups.

Interpretation For patients with suspected stroke, treatment by the MSU substantially reduced median time from alarm to therapy decision. The MSU strategy off ers a potential solution to the medical problem of the arrival of most stroke patients at the hospital too late for treatment.

Funding Ministry of Health of the Saarland, Germany, the Werner-Jackstädt Foundation, the Else-Kröner-Fresenius Foundation, and the Rettungsstiftung Saar.

IntroductionStroke is a main cause of death worldwide and is one of the most common causes of disability in developed countries.1 About 90% of all strokes are due to cerebral ischaemia, with the remainder due to cerebral haemor rhage.2 The only approved treatment for ischaemic stroke is recanalisation of occluded arteries by thrombolysis with alteplase within the very fi rst hours of symptom onset.3–5 However, imple-mentation of recanalising therapy within this narrow therapeutic window is diffi cult to achieve in clinical practice because neurological examination, imaging, and laboratory analyses are needed so that haemorrhagic stroke and other contraindications to thrombolysis can be excluded.4,5 An additional time-sensitive intervention for patients with acute stroke is blood-pressure management, which has been associated with improved outcome.6,7

Less than 15–40% of patients with acute stroke arrive at the hospital early enough to receive thrombolytic treatment,8–10 and only 2–5% of patients actually receive it.11,12 Of those patients who do receive state-of-the-art stroke treatment, outcome is closely related to the time to treatment.13–15 Management of acute stroke must be reconfi gured if we are to overcome the problem of patients arriving at the hospital too late for treatment.

As a potential solution to this problem, we fi rst designed16 and then studied in clinical practice17 the con-cept of bringing guideline-adherent stroke treatment directly to the emergency site, as has previously been made possible for patients with myocardial infarction.18 This strategy is based on a specialised ambulance (mobile stroke unit; MSU)16,17,19 equipped with a CT scanner, a point-of-care laboratory, and a telemedicine

Articles

398 www.thelancet.com/neurology Vol 11 May 2012

connection to the hospital. Apart from preliminary clinical observations, no systematic analysis of the benefi t of this MSU approach on stroke management has been done. We postulated that in clinical practice prehospital stroke treatment would signifi cantly reduce the detrimental delay in receipt of state-of-the-art stroke therapy. Hence we aimed to compare the times from alarm (emergency call) to therapy decision between the MSU and the optimised standard procedure.

MethodsParticipantsIn accordance with our protocol, between November, 2008, and July, 2011, we did a randomised, parallel-group, single-centre study at the University Hospital of the Saarland, Homburg, Germany. For inclusion in our study, patients needed to be aged 18–80 years, have one or more stroke symptoms according to the modifi ed recognition of stroke in the emergency room (ROSIER) scale20 (facial paresis, paresis of arm or leg, aphasia, or dysarthria) that had started within the previous 2·5 h, and have written informed consent provided by the patient or the patient’s legal representative. Our exclusion criteria were uncertain symptom onset (eg, after waking), no focal stroke-like symptoms, and pregnancy. We did not enrol patients if diagnosis and treatment options could not be off ered because of defective key equipment in the MSU or the hospital, if unstable medical conditions needed immediate treatment in the intensive care unit, or if patients were secondarily transferred from primary hospitals. We ran our study from 0800 h to 2200 h during the week and from 0800 h to 1800 h at weekends. Our trial protocol and the informed consent and participant information documents were approved by the Ethics Committee of the Medical Association of the Saarland, Germany (118/06, Aug 14, 2006). An independent clinical monitor

(Interdisciplinary Centre for Clinical Trials, Mainz, Germany) supervised the trial.

Randomisation and maskingFor organisational feasibility, the procedure to be applied to a patient (MSU or standard) was randomised week-wise—ie, all patients entered within a particular week received the same procedure. To achieve balance between both groups regarding potential confounders that might change during the year, such as weather, we chose a block size of 4 weeks to limit the length of time during which the same procedure was used. Our randomisation list was created by an independent statistician (HS) with the SAS procedure PLAN. The statistician also did the statistical analysis.

All emergency calls to the central emergency medical service (EMS) Coordinating Offi ce from a region of up to 30 km around Saarland University Hospital were assessed for reporting of stroke symptoms by the EMS dispatcher (a paramedic) with the modifi ed ROSIER scale20 and for inclusion and exclusion criteria. If the patient was eligible, either the MSU pathway (involving the MSU team in addition to the regular EMS) or the conventional pathway (involving the regular EMS combined with optimised in-hospital stroke management) was initiated in accordance with the pathway specifi ed for that week in the randomisation list. In the fi rst case, the EMS and the MSU team were notifi ed, and in the latter case, the EMS team and the in-hospital stroke team were notifi ed. Finally, the stroke physician in either the MSU or the hospital confi rmed the inclusion and exclusion criteria and obtained written informed consent before the patient was entered into the study. For feasibility of the integration of the MSU strategy into the routine EMS chain and because of the nature of these pathways, the procedure to be applied to a patient was randomly assigned without masking the allocation from the EMS dispatcher, the stroke physician, or the patients.

ProceduresOur MSU programme was integrated into the regular EMS system in a mixed urban and rural setting. The MSU response consisted of the combined dispatch of the MSU and the conventional EMS, which in Germany includes an emergency physician for critically ill patients. The MSU team included a paramedic, a stroke physician, and a neuroradiologist. The MSU team obtained the patient’s history; undertook a neurological examination, CT scan, and laboratory examinations; and, if the patient was eligible, gave thrombolysis directly at the emergency site. The MSU itself was an ambulance equipped, in addition to all standard equipment for primary care, with special equipment: an accumulator-driven and lead-shielded CT scanner (Tomoscan M, Philips, Andover, MA, USA; or, in the last study year, Ceretom, Neurologica, Danvers, MA, USA, allowing multimodal imaging with CT angiography and CT perfusion); a telemedicine

For the study protocol see http://www.uks.eu/msu-trial

Figure: Trial profi le

53 were assigned to receive the MSU pathway 47 were assigned to receive the optimised conventional pathway

100 were enrolled and entered the randomisation stage

361 patients screened

261 excluded 88 did not meet age criterion 25 had symptom onset >3 h 120 did not meet symptoms criteria 22 because of unavailable diagnostic equipment (12 CT, 10 point-of-care laboratory) 2 did not give informed consent 2 because of unstable ventilation with intubation 1 because of premedication 1 because of extensive vomiting

Articles

www.thelancet.com/neurology Vol 11 May 2012 399

system (Meytec, Werneuchen, Germany) enabling trans-mission of digital imaging and communication data obtained by CT scanning or video of clinical examination via the universal mobile tele communi cation system to the picture archiving and communi cation system of the hospital; and a point-of-care laboratory system.17 The laboratory system allowed the measurement of platelet count, leucocyte count, erythrocyte count, haemoglobin, and haematocrit (PocH 100i, Sysmex, Hamburg, Germany), international normalised ratio and activated partial thromboplastin time (Hemochron Jr, ITC, Edison, NY, USA), and γ-glutamyltransferase, p-amylase, and glucose (Refl otron plus, Roche Diagnostics Mannheim, Germany), as requested by the approval criteria for alteplase12 and by primary stroke management guide lines.4,5

To avoid comparing the MSU strategy with sub optimum intrahospital stroke management, patients in the control group received optimised conventional stroke manage-ment, which included point-of-care laboratory testing instead of testing by the centralised hospital laboratory—a change that achieved important time savings in stroke management.21 Thrombolysis was given in accordance with all inclusion and exclusion criteria for the approval of alteplase22 (including the upper age restriction of 80 years and the 3 h therapeutic window relevant during the trial) and primary stroke management guidelines.4,5

Patients were clinically assessed at the time of study enrolment (either before hospital arrival for the MSU group or in the hospital for the control group), after 24 h (±1 h), and at day 7 (±1 day, or last observation carried forward) after study inclusion. Clinical assessment included medical history, neurological examination, and analysis of incidence of safety endpoints and severe adverse events. For neurological examinations we used the National Institutes of Health Stroke Scale (NIHSS, with scores ranging from 0 to 42; higher scores suggest more severe disease), the modifi ed Rankin scale (mRS, with a range from 0, showing no residual symptoms, to 6, showing death), and the Barthel index (with a range from 0, showing complete dependence, to 100, showing no need for help in the activities of daily life). We assessed physiological variables, such as heart function, heart rate, and blood pressure, at baseline and, if necessary (ie, in the case of stroke or cardiopulmonary instability), continued to monitor these variables.

As further variables we recorded baseline epidemio-logical data, fi nal diagnoses, and further stroke manage-ment variables, such as time from symptom onset to alarm, time from alarm to arrival of the MSU or EMS at the scene, distances to hospital in km, and number of patients receiving diagnosis-specifi c blood-pressure management.

Our primary endpoint was time from alarm to therapy decision and our prespecifi ed secondary endpoints were number of patients with intravenous thrombolysis with alteplase, time from alarm to intravenous thrombolysis,

time from alarm to end of CT, and time from alarm to end of laboratory analysis. We assessed stroke manage-ment intervals with symptom onset rather than alarm as the starting point. Because intra-arterial recanalisation therapy developed during the years in which we did our study, we added post-hoc endpoints: number of patients with intravenous thrombolysis or intra-arterial recanalisation and time from alarm to intravenous thrombolysis or to intra-arterial recanal isation. We derived stroke management times from those recorded by the EMS Coordinating Offi ce, by the CT and laboratory equipment, and by the EMS protocols.

Further secondary endpoints were NIHSS (cutoff value ≤1 or ≥8 points improvement), Barthel index (≥95 points), and mRS scores (≤2) at days 1 and 7 for patients who had stroke.13,23

Our safety endpoints were survival at day 7, incidence of stroke-related or neurological death by day 7 (ie, fatal ischaemic stroke, fatal reinfarction, fatal primary or secondary intracerebral haemorrhage), symptomatic intracranial haemorrhage (defi ned as any haemorrhage

MSU group (n=53)

Control group (n=47)

Age (years) 72 (59–76) 71 (55–75)

Men 31 (58%) 32 (68%)

History

Hypertension* 41 (77%) 36 (77%)

Diabetes† 14 (26%) 20 (43%)

Atrial fl utter or fi brillation 12 (23%) 14 (30%)

Active smoking until onset of stroke symptoms

13 (25%) 17 (36%)

Hypercholesterolemia‡ 17 (32%) 23 (49%)

Previous stroke 16 (30%) 19 (40%)

NIHSS 5 (3–11) 6 (3–12)

Modifi ed Rankin scale 4 (2–5) 4 (2–5)

Symptom onset to alarm (min) 21 (8–65) 23 (7–67)

Distance to hospital (km) 6 (4–10) 8 (6–15)

Diagnoses

Ischaemic stroke 29 (55%) 25 (53%)

Transient ischaemic attack 8 (15%) 9 (19%)

Intracranial haemorrhage 4 (6%) 7 (15%)

Stroke mimics 12 (23%) 6 (13%)

Epileptic seizure 7 (13%) 3 (6%)

Migraine 0 2 (4%)

Shunt dislocation 1 (2%) 0

Vestibulopathy 0 1 (2%)

Encephalitis 1 (2%) 0

Cervical stenosis 1 (2%) 0

Syncope 2 (4%) 0

Data are median (IQR) or n (%). MSU=mobile stroke unit. NIHSS=National Institutes of Health Stroke Scale. *History of hypertension or antihypertensive treatment. †History of diabetes, antidiabetic drugs, or HbA1c >6·5%. ‡Serum cholesterol of ≥200 mg/dL, LDL ≥130mg/dL, or use of lipid-lowering drugs.

Table 1: Baseline demographic and clinical characteristics

Articles

400 www.thelancet.com/neurology Vol 11 May 2012

accompanied by neurological deterioration of at least 4 points in the NIHSS score),13 symptomatic peripheral haemorrhage, and rates of cerebral herniation and symptomatic brain oedema. We also recorded all other serious adverse events.

Statistical analysisWe prespecifi ed the statistical analyses of our primary and secondary endpoints. We expressed stroke manage-ment times as median (IQR) for comparison of our times with those previously reported. We used the Mann-Whitney U test to analyse the primary endpoint (time from alarm to therapy decision) and all secondary endpoints involving time intervals, because this test is robust against deviations from normality. A sensitivity analysis applying the t test was used to confi rm results. We calculated diff erences in medians with the Hodges-Lehmann estimator and respective distribution-free 95% CIs with Moses’ method. We used logistic regression for ordinal responses to analyse the mRS results at days 1 and 7, with baseline mRS as a covariate.

Our estimation of the required sample size was done on the basis of our assumption that data are normally distributed, because the power of the Mann-Whitney U test is slightly overestimated when the real distribution diverges from normality. With an SD of the time between emergency call and therapy decision of 30 min (based on unpublished historical data for the standard procedure in the hospital in which our study was done), a sample size of 100 patients per group would give the analysis about 90% power to detect an advantage of 14 min achieved by the MSU at an α level of 0·025. We scheduled a preplanned interim analysis to be done after the enrolment of 100 patients. With the application of O’Brien-Fleming boundaries, we planned to stop our study for futility if the p value of the interim analysis of the primary endpoint was greater than 0·4, or for proven superiority if the p value was less than 0·0015.24 We assessed all other endpoints only once. We used East software (version 4) for planning our interim analysis and SAS (version 9.2) for our other analyses. This study is registered with ClinicalTrials.gov, number NCT00153036.

Role of the funding sourceThe sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.

ResultsThe fi gure shows the trial profi le. Because we postulated that prehospital diagnostic work-up and stroke treatment would reduce the delay to therapy decision, we stopped our study when our predefi ned interim analysis showed the prespecifi ed superiority (p<0·0015) in the primary endpoint.

All patients we assigned to the MSU group gave informed consent but two who would have been assigned to the control group did not. We did not lose any patients from our fi nal analysis of our primary endpoint. Table 1

MSU group (n=53)

Control group (n=47)

p value Diff erence (95% CI)

Primary endpoint

Alarm to therapy decision (min) 35 (31–39) 76 (63–94) <0·0001 41 (36–48)

Secondary endpoints

Symptom onset to therapy decision (min) 56 (43–103) 104 (80–156) <0·0001 43 (30–58)

Number of patients with intravenous thrombolysis

12 (23%) 8 (17%) 0·30* ··

Alarm to intravenous thrombolysis (min) 38 (34–42) 73 (60–93) <0·0001 34 (23–54)

Symptom onset to intravenous thrombolysis (min)

72 (53–108) 153 (136–198) 0·0011 80 (40–115)

Number of patients with intravenous thrombolysis or intra-arterial recanalisation†

12 (23%) 11 (23%) 0·81* ··

Alarm to intravenous thrombolysis or intra-arterial recanalisation† (min)

38 (34–42) 78 (61–110) <0·0001 44 (27–73)

Symptom onset to intravenous thrombolysis or intra-arterial recanalisation† (min)

72 (53–108) 152 (135–209) <0·0001 80 (46–115)

Alarm to end of CT (min) 34 (30–38) 71 (62–87) <0·0001 38 (33–43)

Symptom onset to end of CT (min) 56 (43–103) 97 (74–156) <0·0001 39 (26–52)

Alarm to end of laboratory analysis (min) 28 (26–34) 69 (55–81) <0·0001 38 (32–44)

Symptom onset to end of laboratory analysis (min)

51 (40–95) 99 (70–140) <0·0001 39 (26–56)

NIHSS at day 1‡ 3 (1–10) 4 (2–12) 0·48 1 (–1 to 3)

NIHSS at day 7‡ 2 (1–8) 4 (0–8) 0·94 0 (–2 to 2)

Barthel index at day 1‡ 65 (25–85) 75 (0–95) 0·98 0 (–10 to 15)

Barthel index at day 7‡ 70 (30–95) 80 (25–95) 0·79 0 (–10 to 15)

mRS at day 1‡ 0·99 OR 1·00 (0·42–2·41)§

0 6 7 ·· ··

1 3 6 ·· ··

2 11 6 ·· ··

3 6 6 ·· ··

4 3 2 ·· ··

5 11 13 ·· ··

6 1 1 ·· ··

mRS at day 7‡ 0·77 OR 0·89 (0·39–2·00)§

0 8 8 ·· ··

1 9 8 ·· ··

2 3 5 ·· ··

3 7 5 ·· ··

4 3 7 ·· ··

5 5 6 ·· ··

6 6 2 ·· ··

Unless otherwise specifi ed, data are median (IQR) and tested with the Mann-Whitney U test as prespecifi ed in our protocol, because of unknown distribution of the time diff erences. We calculated diff erences in medians with the Hodges-Lehmann estimator and distribution-free CIs with Moses’ method. MSU=mobile stroke unit. NIHSS=National Institutes of Health Stroke Scale. mRS=modifi ed Rankin scale. OR=odds ratio. *Pearson’s χ² test of proportions. †Post-hoc endpoints that includes all types of recanalising stroke treatments. ‡Logistic regression with baseline mRS as categorical covariate. §Analysed in the stroke patient subgroup.

Table 2: Primary and secondary endpoints

Articles

www.thelancet.com/neurology Vol 11 May 2012 401

lists baseline demographic characteristics, distances to hospital, times from symptom onset to alarm, baseline disease severity, and diagnoses. No diagnosis of ischaemic or haemorrhagic stroke needed revision during further clinical follow-up. Median times from alarm to the MSU or EMS arriving at the scene were 12 min (IQR 9–16) in the MSU group and 8 min (6–11) in the control group. MSU-based stroke management roughly halved the time from alarm to therapy decision compared with the control group (table 2). Consistent with the fi ndings for our primary endpoint, the stroke management intervals of times from alarm to end of laboratory analysis and end of CT times were also reduced (p<0·0001; table 2). Our analysis showed that the MSU strategy reduced the time from symptom onset to therapy decision compared with the control group (table 2), which was less than 1 h for 30 (57%) of 53 patients in the MSU group but only for two (4%) of 47 patients in the control group.

Recanalising treatments given to patients in the MSU and control groups were intravenous alteplase (eight patients in the MSU group and seven in the control group), bridging of intravenous alteplase with later mechanical recanalisation (four in the MSU group and one in the control group), and mechanical recanalisation alone (none in the MSU group and three in the control group). Together, 12 patients in the MSU group were treated with intravenous thrombolysis in the fi eld.

Although the diff erences between groups in the numbers of patients with intravenous thrombolysis or with intravenous thrombolysis or intra-arterial recanal-isation were not statistically signifi cant, the times from alarm to intravenous thrombolysis and to intravenous thrombolysis or intra-arterial recanalisation were sub-stantially shorter in the MSU group (table 2). We obtained similar signifi cant diff erences when we used symptom onset rather than alarm as the starting point for these stroke management intervals (table 2).

Indicators of neurological outcome at days 1 and 7 did not substantially diff er between the groups (table 2). Six patients (11%) in the MSU group and three patients (6%) in the control group experienced events that were tracked as safety endpoints of the study. Five patients (9%) in the MSU group and two patients (4%) in the control group died of stroke-related or neurological causes (table 3). Altogether, three patients treated with alteplase died (all in the MSU group), and one of the untreated patients with ischaemic stroke (in the MSU group) died. Whereas non-fatal secondary intracranial haemorrhage was not evident in either study group, there was non-fatal cerebral herniation and oedema in one patient in the control group, and there was peripheral haemorrhage in one patient in the MSU group. Table 3 lists additional safety endpoints and severe adverse events.

By contrast with conventional stroke management (no patients in the control group received blood-pressure management), MSU-based stroke management allowed

eight patients to receive prehospital pharmaco logical intervention in blood pressure on the basis of information about the ischaemic or haemorrhagic cause.

Analysis of technical problems related to key equipment showed that CT scanning in the MSU group was unavailable for eight patients because of scanner defects, two patients because of handling problems (scanner dysfunction due to steep streets), and two patients because they were overweight. Intrahospital imaging options did not depend on a single device; therefore, the availability of CT was 100% for the control group. In the control group, the point-of-care laboratory (used also by other departments of the hospital and to assess samples from patients who were not part of our study), was unavailable for ten patients.

DiscussionOur main fi ndings are that the strategy of prehospital stroke diagnosis and treatment allows therapy decisions a median of 35 min after alarm in clinical reality. Median time from symptom onset to intravenous thrombolysis was 72 min, representing a new timescale in acute stroke management.

Stroke is a medical emergency for which “time is brain”;14 however, most patients still arrive at hospital too late to receive necessary treatment.8–10 We show that

MSU group (n=53)

Control group (n=47)

Prespecifi ed safety endpoints

Survival at day 7 47 (89%) 45 (96%)

Stroke-related or neurological death 5 (9%) 2 (4%)

Fatal ischaemic stroke 1 (2%) 0

Fatal reinfarction 1 (2%) 0

Fatal primary ICH 2 (4%) 2 (4%)

Fatal secondary ICH* 1 (2%) 0

Non-fatal secondary ICH (change in NIHSS ≥4) 0 0

Non-fatal cerebral herniation and symptomatic oedema (change in NIHSS ≥4)

0 1 (2%)

Peripheral haemorrhage† 1 (2%) 0

Other serious adverse events‡

None 46 (87%) 37 (79%)

Non-fatal reinfarction 3 (6%) 1 (2%)

Secondary ICH (change in NIHSS <4) 0 2 (4%)

Myocardial infarction§ 2 (4%) 1 (2%)

Pneumonia 2 (4%) 6 (13%)

Other infection 2 (4%) 1 (2%)

Data are n (%). MSU=mobile stroke unit. ICH=intracranial haemorrhage. NIHSS=National Institutes of Health Stroke Scale. *ICH was a subarachnoidal haemorrhage secondary to thrombolysis. †Haemorrhage was associated with rupture of an unknown aneurysm of the lienal artery one day after intravenous alteplase. ‡One patient in the control group concomitantly had ICH and pneumonia, one patient in the MSU group concomitantly had myocardial infarction, recurrent stroke, and pneumonia. §One patient in the MSU group with myocardial infarction died at day 2.

Table 3: Prespecifi ed safety endpoints and other serious adverse events

Articles

402 www.thelancet.com/neurology Vol 11 May 2012

MSU-based stroke management substantially breaks, to our knowledge, all reported times for stroke management: the median time from alarm to therapy decision was 41 min shorter than that for optimised conventional stroke management, with a median time from symptom onset to therapy decision of 56 min. The diff erences in times from alarm to intravenous thrombolysis and to intravenous thrombolysis or intra-arterial recanalisation between the MSU group and the control group in the subpopulation of patients with stroke were similar to those for the times to therapy decision for the overall population.

These times contrast with the times from arrival at hospital to treatment of 60 min that are still the stated goal of present guide lines4,5,15 and with the much larger intervals that are still clinical reality today. Although there is no previously published trial of prehospital stroke diagnosis and treatment, there are several reports of

prehospital delay in acute stroke care. The reported median times from symptom onset to arrival at hospital vary strongly, ranging from 3 h to 6 h24,25 and median times from arrival at hospital to thrombolysis ranging between 66 min and 78 min.26–28 For example, a recent large European multicentre registry study29 with 6483 patients showed that the median time from symptom onset to intravenous thrombolysis was 140 min, 68 min longer than the median time achieved by the MSU strategy. A further study30 showed that after prehospital and intra hospital stroke management had been strictly optimised the median time from symptom onset to therapy onset decreased from 149 min to 112 min, which is still roughly 40 min longer than that achieved for the MSU group.

According to the generally accepted “time is brain” concept,14,31–35 such a large reduction in delay should translate into improved outcome. Indeed, animal experi-mental33,34 and clinical31,32,35 evidence shows that time to treatment is the primary determinant of outcome. A recent meta-analysis of the major multicentre thromb-olysis trials showed that the number needed to treat for excellent outcome rapidly increases from fi ve in the range of 0–90 min after symptom onset to more than nine in the range of 91–180 min, and up to 15 in the range of 181–270 min (panel).35

The integration of our study into the routine EMS system, and the representative study population shown by the demographic variables in table 1, suggest our fi ndings can be generalised—eg, in countries with similar health-care systems. We chose time from alarm to therapy decision as our primary endpoint because it goes beyond the question of whether alteplase treatment is given, but includes decisions on mechanical recanal-isation (of large-vessel occlusion seen by prehospital CT angiography) or bridging therapy.

Despite the lack of evidence from randomised trials, an increasing number of fi ndings argue for the relevance of diff erential blood-pressure management for patients with ischaemic and haemorrhagic stroke and for the time sensitivity of this treatment.4–7 Recommendations of relevant guidelines for blood pressure are, indeed, diff erent for ischaemia (ie, systolic values as high as 185–220 mm Hg can be tolerated to enhance cerebral perfusion pressure4,5) and haemorrhage (intervening to target systolic values of 150–160 mm Hg36,37). Our fi ndings show that, in clinical reality, the MSU strategy provides an opportunity for cause-specifi c blood-pressure management well before hospital arrival.

Despite the substantial diff erences in stroke manage-ment times, our fi ndings show no signifi cant diff erences between the two study groups in the number of treated patients or in neurological outcomes. The limitations of our study are the lack of power for these and the other secondary endpoints in the subpopulation of stroke patients, the potential eff ects of previous disability, the relatively short follow-up time of outcome-related

Panel: Research in context

Systematic reviewWe searched Medline (1950–2012), the Cochrane Central Register of Controlled Trials in the Cochrane Library, and Embase (1988–2012), with the search terms “prehospital thrombolysis” and “prehospital treatment”. Our search was restricted to reports in English. We restricted our searches to all types of trials including at least three patients with acute ischaemic stroke. We also hand searched relevant journals and the reference lists of included reports.

InterpretationThere are no previous trials of prehospital thrombolysis to treat acute ischaemic stroke. Of the reported fi ndings on prehospital delay in acute stroke treatment, median times from symptom onset to arrival at hospital vary widely, with values ranging from 3 h to 6 h.24,25 Additionally, median times from arrival at hospital to thrombolysis range from 66 min to 78 min.26–28 For example, a recent large European multicentre registry29 with 6483 patients reported that the median time from symptom onset to therapy was 68–140 min longer than the median time achieved by the mobile stroke unit strategy. The stroke management times achieved with the mobile stroke unit (MSU) break the reported times for stroke management. Of interventional studies, eff ects of various measures (eg, public educational campaigns, regulations about transfer of patients to stroke units, preinformation of stroke teams by the emergency medical service, or optimisation of in-hospital processes) on improvement regarding times until treatment have been compared with historical data but not with control groups in a randomised design. For patients with suspected stroke, prehospital stroke treatment roughly halved the interval from alarm to therapy decision. The MSU strategy off ers a potential solution to the medical problem of the arrival of most stroke patients at the hospital too late for treatment.

Articles

www.thelancet.com/neurology Vol 11 May 2012 403

secondary endpoints, and the absence of masking in their assessment. Moreover, we cannot exclude with certainty a potential for bias due to our week-wise randomisation procedure.

Our analyses of safety endpoints and severe adverse events show that event rates for the two groups were within similar ranges. However, such analyses have low power. Although equipment and personnel in the MSU are the same as in the hospital, and although no previous studies suggested that earlier diagnosis and treatment of stroke increases the number of complications,31,32,35 safety aspects must be further assessed, in view of the recorded deaths in patients treated with alteplase in our study.

We enrolled into our study 28% of the patients we screened. This number could, in future trials, be substantially increased if thrombolysis were to be approved for patients older than 80 years. In 2011, the 3 h therapeutic window for use of alteplase, which was one of the main exclusion criteria in our study, was extended to 4·5 h in many countries. Furthermore, improved stroke symptom questionnaires and additional training of the EMS could contribute to more specifi c identifi cation of stroke.

The costs incurred by the MSU strategy are related to CT scanning, the point-of-care laboratory, and the tele-medicine equipment (about €300 000) and to the involvement of the core MSU team, which consists of a paramedic and a stroke physician. Both members of the team are active only in case of emergencies and otherwise work at their regular institutions; thus, the staff costs per treated patient are not substantially greater than for conventional procedures. It remains to be assessed whether the intervention costs are outweighed by reduced costs for the long-term care of stroke patients. The neuroradiologist was included in our study primarily for legal reasons, because in Germany the neurologist is at present not allowed to run the scanner alone. This person could be replaced by a radiology technician or by a neurologist with radiology training in the future. The MSU concept16 also implies that neuroradiological and even specifi c neurological expertise could in future be supplied to the regular EMS solely via telemedicine, thereby potentially facilitating integration of the MSU strategy into the diff erently organised EMS systems in various settings and countries.

In conclusion, although the eff ect on clinical outcome needs further study in larger (eg, multicentre) trials, the results of this fi rst randomised trial of the MSU strategy of bringing the hospital to the patient with stroke show that guideline-adherent diagnosis and therapy can reliably be delivered within the fi rst 35 min after alarm, thus speeding up acute stroke management.

ContributorsSW contributed to study conception, organisation, execution and

statistical review, data analysis, data interpretation, and writing of the

report. PK contributed to study conception, organisation and execution,

and review and critique of the report. AH contributed to study

conception, organisation, statistical review and critique, and review and

critique of the report. IK contributed to review and critique of the report.

ML contributed to statistical execution and review and critique of the

report. TS contributed to study conception and organisation, and review

and critique of the report. CRo, PP, HK, MA, and UY contributed to

study organisation and execution, and review and critique of the report.

IG contributed to study execution, statistical execution, and review and

critique of the report. HS contributed to statistical analysis, execution,

and review, and review and critique of the report. SH contributed to data

collection and review and critique of the report. JV, KZ, KS, and RD

contributed to study execution and review and critique of the report. DK

contributed to study organisation and review and critique of the report.

YL contributed to statistical execution, review, and critique, and review

and critique of the report. TV and WR contributed to study conception,

review, and critique, and review and critique of the report. KK

contributed to study conception, statistical design, review, and critique,

and review and critique of the report. CRu contributed to study

conception, statistical design, and review and critique of the report. TB

contributed to project conception and review and critique of the report.

KF contributed to study conception, organisation, statistical design,

execution, review, and critique, and writing, reviewing, and critique of

the report.

Confl icts of interestHS is an employee of Boehringer Ingelheim. All other authors declare

that they have no confl icts of interest.

AcknowledgmentsWe thank Uwe Feldmann for his expert advice on randomisation design;

Gernot Schulte-Altedorneburg, Stephan Ziegeler, Andreas Biedler, and

Stefan Kleinschmidt for their contribution to trial organisation; and

Christian Müller and his team, the German Red Cross, Homburg, and

the team of the EMS Coordination Offi ce, Saarland, for their

professional support in the integration of this trial into the routine

emergency medical service chain.

References1 Rothwell PM, Coull AJ, Silver LE, et al. Population-based study of

event-rate, incidence, case fatality, and mortality for all acute vascular events in all arterial territories (Oxford Vascular Study). Lancet 2005; 366: 1773–83.

2 Caplan LR, Hier DB, D’Cruz I, et al. Cerebral embolism in the Michael Reese Registry. Stroke 1983; 14: 530–40.

3 The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med 1995; 333: 1581–88.

4 Adams HP Jr, del Zoppo G, Alberts MJ, et al. Guidelines for the early management of adults with ischemic stroke: a guideline from the American Heart Association/American Stroke Association Stroke Council, Clinical Cardiology Council, Cardiovascular Radiology and Intervention Council, and the Atherosclerotic Peripheral Vascular Disease and Quality of Care Outcomes in Research Interdisciplinary Working Groups: The American Academy of Neurology affi rms the value of this guideline as an educational tool for neurologists. Circulation 2007; 115: e478–534.

5 European Stroke Organisation Executive Committee. Guidelines for management of ischaemic stroke and transient ischaemic attack 2008. Cerebrovasc Dis 2008; 25: 457–507.

6 Ahmed N, Wahlgren N, Brainin M, et al. Relationship of blood pressure, antihypertensive therapy, and outcome in ischemic stroke treated with intravenous thrombolysis: retrospective analysis from Safe Implementation of Thrombolysis in Stroke–International Stroke Thrombolysis Register (SITS-ISTR). Stroke 2009; 40: 2442–49.

7 Tsivgoulis G, Frey JL, Flaster M, et al. Pre-tissue plasminogen activator blood pressure levels and risk of symptomatic intracerebral hemorrhage. Stroke 2009; 40: 3631–34.

8 Katzan IL, Hammer MD, Hixson ED, Furlan AJ, Abou-Chebl A, Nadzam DM. Utilization of intravenous tissue plasminogen activator for acute ischemic stroke. Arch Neurol 2004; 61: 346–50.

9 Lichtman JH, Watanabe E, Allen NB, Jones B, Dostal J, Goldstein LB. Hospital arrival time and intravenous t-PA use in US academic medical centers, 2001–2004. Stroke 2009; 40: 3845–50.

Articles

404 www.thelancet.com/neurology Vol 11 May 2012

10 Reeves MJ, Broderick JP, Frankel M, et al. The Paul Coverdell National Acute Stroke Registry: initial results from four prototypes. Am J Prev Med 2006; 31: S202–09.

11 Albers GW, Olivot JM. Intravenous alteplase for ischaemic stroke. Lancet 2007; 369: 249–50.

12 California Acute Stroke Pilot Registry (CASPR) Investigators. Prioritizing interventions to improve rates of thrombolysis for ischemic stroke. Neurology 2005; 64: 654–59.

13 Hacke W, Kaste M, Bluhmki E, et al. Thrombolysis with alteplase 3 to 4·5 hours after acute ischemic stroke. N Engl J Med 2008; 359: 1317–29.

14 Saver JL. Time is brain—quantifi ed. Stroke 2006; 37: 263–66.

15 Marler JR, Winters-Jones P, Emr M. National Institutes of Neurological Disorders and Stroke. Proceedings of a national symposium on rapid identifi cation and treatment of acute stroke [NIH publication No 97-4239]. Bethesda, MD: National Institutes of Health, 1997: 157–58.

16 Fassbender K, Walter S, Liu Y, et al. “Mobile stroke unit” for hyperacute stroke treatment. Stroke 2003; 34: e44.

17 Walter S, Kostopoulos P, Haass A, et al. Bringing the hospital to the patient: fi rst treatment of stroke patients at the emergency site. PLoS One 2010; 5: e13758.

18 Wilcox RG, von der Lippe G, Olsson CG, Jensen G, Skene AM, Hampton JR. Eff ects of alteplase in acute myocardial infarction: 6-month results from the ASSET study. Anglo-Scandinavian Study of Early Thrombolysis. Lancet 1990; 335: 1175–78.

19 Kostopoulos P, Walter P, Haass A, et al. “Mobile Stroke Unit” for diagnosis-based triaging of persons with suspected stroke. Neurology (in press).

20 Nor AM, Davis J, Sen B, et al. The Recognition of Stroke in the Emergency Room (ROSIER) scale: development and validation of a stroke recognition instrument. Lancet Neurol 2005; 4: 727–34.

21 Walter S, Kostopoulos P, Haass A, et al. Point-of-care laboratory halves door-to-therapy-decision time in acute stroke. Ann Neurol 2011; 69: 581–86.

22 The electronic Medicines Compendium. Boehringer Ingelheim Limited: Actilyse—summary of product characteristics last updated on the eMC: 14/12/2009. Leatherhead: Datapharm, 2009.

23 Kwon S, Hartzema AG, Duncan PW, Min-Lai S. Disability measures in stroke: relationship among the Barthel Index, the Functional Independence Measure, and the Modifi ed Rankin Scale. Stroke 2004; 35: 918–23.

24 O’Brien PC, Fleming TR. A multiple testing procedure for clinical trial. Biometrics 1979; 35: 549–56.

25 Evenason KR, Rosamond WD, Morris DL. Prehospital and in-hospital delays in acute stroke care. Neuroepidemiology 2001; 20: 65–76.

26 Fonarow GC, Smith EE, Saver JL, et al. Timeliness of tissue-type plasminogen activator therapy in acute ischemic stroke: patient characteristics, hospital factors, and outcomes associated with door-to-needle times within 60 minutes. Circulation 2011; 123: 750–58.

27 Asplund K, Glader EL, Norrving B, et al. Eff ects of extending the time window of thrombolysis to 4.5 hours: observations in the swedish stroke register (Riks-Stroke). Stroke 2011; 42: 2492–97.

28 Price CI, Clement F, Gray J, Donaldson C, Ford GA. Systematic review of stroke thrombolysis service confi guration. Expert Rev Neurother 2009; 9: 211–33.

29 Wahlgren N, Ahmed N, Dávalos A, et al. Thrombolysis with alteplase for acute ischaemic stroke in the Safe Implementation of Thrombolysis in Stroke-Monitoring Study (SITS-MOST): an observational study. Lancet 2007; 369: 275–82.

30 Puolakka T, Väyrynen T, Häppölä O, Soinne L, Kuisma M, Lindsberg PJ. Sequential analysis of pretreatment delays in stroke thrombolysis. Acad Emerg Med 2010; 17: 965–69.

31 Marler JR, Tilley BC, Lu M, et al. Early stroke treatment associated with better outcome: the NINDS rt-PA stroke study. Neurology 2000; 55: 1649–55.

32 Hacke W, Donnan G, Fieschi C, et al. Association of outcome with early stroke treatment: pooled analysis of ATLANTIS, ECASS, and NINDS rt-PA stroke trials. Lancet 2004; 363: 768–74.

33 Astrup J, Siesjö BK, Symon L. Thresholds in cerebral ischemia—the ischemic penumbra. Stroke 1981; 12: 723–25.

34 Hossmann KA. Viability thresholds and the penumbra of focal ischemia. Ann Neurol 1994; 36: 557–65.

35 Lees KR, Bluhmki E, von Kummer R, et al. Time to treatment with intravenous alteplase and outcome in stroke: an updated pooled analysis of ECASS, ATLANTIS, NINDS, and EPITHET trials. Lancet 2010; 375: 1695–703.

36 Broderick J, Connolly S, Feldmann E, et al. Guidelines for the management of spontaneous intracerebral hemorrhage in adults: 2007 update: a guideline from the American Heart Association/American Stroke Association Stroke Council, High Blood Pressure Research Council, and the Quality of Care and Outcomes in Research Interdisciplinary Working Group. Circulation 2007; 116: e391–413.

37 Steiner T, Kaste M, Forsting M, et al. Recommendations for the management of intracranial haemorrhage—part I: spontaneous intracerebral haemorrhage. The European Stroke Initiative Writing Committee and the Writing Committee for the EUSI Executive Committee. Cerebrovasc Dis 2006; 22: 294–316.

www.thelancet.com/neurology Vol 11 May 2012 467

Review

Disability outcome measures in multiple sclerosis clinical trials: current status and future prospectsJeffrey A Cohen, Stephen C Reingold, Chris H Polman, Jerry S Wolinsky, for the International Advisory Committee on Clinical Trials in Multiple Sclerosis

Many of the available disability outcome measures used in clinical trials of multiple sclerosis are insensitive to change over time, inadequately validated, or insensitive to patient-perceived health status or quality of life. Increasing focus on therapies that slow or reverse disability progression makes it essential to refi ne existing measures or to develop new tools. Major changes to the expanded disability status scale should be avoided to prevent the loss of acceptance by regulators as a measure for primary outcomes in trials that provide substantial evidence of eff ectiveness. Rather, we recommend practical refi nements. Conversely, although substantial data support the multiple sclerosis functional composite as an alternative measure, changes to its component tests and scoring method are needed. Novel approaches, including the use of composite endpoints, patient-reported outcomes, and measurement of biomarkers, show promise as adjuncts to the current disability measures, but are insuffi ciently validated to serve as substitutes. A collaborative approach that involves academic experts, regulators, industry representitives, and funding agencies is needed to most eff ectively develop disability outcome measures.

IntroductionAt the fi rst comprehensive international meeting related to multiple sclerosis (MS) clinical trial design, which was sponsored by the US National MS Society and the MS Society of Canada and held in 1982, in Grand Island, NY, USA, randomised placebo-controlled trials were accepted as the gold standard.1 Subsequently, several disease-modifying agents were tested in pivotal phase 3 trials. Following regulatory approval of the fi rst agent in 1993, the National MS Society convened an international meeting in Charleston, SC, USA, on outcome measures in MS clinical trials. An important conclusion was that the available clinical outcome measures of disability, including the expanded disability status scale (EDSS), were not adequately responsive or sensitive.2

Spurred in large part by that meeting, a special task force of the National MS Society analysed disability outcomes used in past studies. They recommended further investigation of a new rating scale, the multiple sclerosis functional composite (MSFC).3,4 Validation in prospective studies was viewed as crucial before it could be adopted. To this end, the MSFC became used as a secondary outcome in many MS clinical trials, alongside the EDSS, and as the primary outcome in one study of secondary progressive MS.5 Despite its anticipated advantages, the MSFC did not gain wide acceptance by investigators or regulators as a primary MS disability outcome measure. Thus, irrespective of the perceived need for improved disability outcome measures and the continued dissatisfaction with the EDSS, consensus has not been reached on whether the MSFC as originally proposed3,4 was a suitable alternative.

With the current availability of multiple therapeutic agents that reduce relapse frequency in relapsing-remitting MS, future clinical trials will focus increasingly on prevention of CNS injury and promotion of recovery from damage already acquired, particularly in progressive forms of the disease. The need for sensitive and responsive but clinically meaningful disability outcome measures,

therefore, is increasing, as it is for surrogate endpoints that are highly predictive of future disability.

In this Review we summarise the conclusions of the International Conference on Disability Outcomes in Multiple Sclerosis, held in Washington, DC, USA, on May 19–20, 2011, under the auspices of the International Advisory Committee on Clinical Trials in MS (appendix). In addition to MS-specifi c, clinician-assessed disability measures, we consider composite endpoints, global measures of activities of daily living, and patient-reported outcomes for health status and quality of life. Finally, we review the status of MRI, optical coherence tomography (OCT), and other biomarkers as adjuncts to clinical disability outcomes, potential surrogate endpoints, or components of composite endpoints. Although regu latory acceptance of clinical outcome measures used in trials drives much of this discussion, other important uses are considered (panel 1).

Refi nement of existing disability outcome measuresDevelopment of outcome measures must take into account several features of MS: heterogeneous clinical manifestations; unpredictable relapse rates and severity; variable recovery from relapses, which leads to relapse-related disability accrual; and the generally slow rate of progression (non-relapse-related disability accrual; panel 2). Furthermore, as the mechanisms that underlie relapses and gradual progression may be distinct, although sometimes present concurrently, outcome measures must be sensitive to the biological processes targeted by the treatment under study. Finally, we acknowledge that discussion of these issues frequently blurs the distinction between impairment and disability, as defi ned by WHO.6

Expanded disability status scaleThe disability status scale was introduced 50 years ago and was revised 30 years ago into the EDSS (fi gure 1).7

Lancet Neurol 2012; 11: 467–76

Mellen Center, Neurological Institute, Cleveland Clinic, Cleveland, OH, USA (Prof J A Cohen MD); Scientifi c and Clinical Review Associates, Salisbury, CT, USA (S C Reingold PhD); Department of Neurology, Multiple Sclerosis Centre Amsterdam, Vrije Universiteit Medical Centre, Amsterdam, Netherlands (Prof C H Polman MD); and Department of Neurology, University of Texas Health Sciences Center at Houston, Houston, TX, USA (Prof J S Wolinsky MD)

Correspondence to:Prof Jeff rey A Cohen, Mellen Center U-10, Neurological Institute, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, OH 44195, USAcohenj@ccf.org

See Online for appendix

468 www.thelancet.com/neurology Vol 11 May 2012

Review

These measures have been the most widely used in MS clinical trials for clinician assessment of neurological disability. The EDSS has been used to characterise trial populations, defi ne objective changes in neurological function as a component of protocol-specifi c relapse determination, and measure disability accrual. The scale takes into account a wide range of neurological functions relevant to MS, enables comparison of features on a scale

of 0–10 for individual and groups of patients over time, and correlates, albeit weakly, with various other disease measures, including MRI.8 MS specialists are familiar with the EDSS and it is accepted by regulators as a measure of disease progression.

The EDSS does have shortcomings, which have been widely discussed.9 It is based on the standard neurological examination, which is inherently subjective. The original scoring rules for the component functional systems and the overall scale are ambiguous. As a result, the scale has poor reliability within and between raters.10–13 A version of the EDSS, available on CD-ROM or online from Neurostatus, was developed to improve reliability by standardisation of the method for doing the neurological examination, detailed defi nitions and scoring rules, a scoring sheet to be used as a source document, and certifi cation. This standardised version of the EDSS has been used in many MS trials.

How EDSS scores are calculated raises various issues. Scores of 4·0–7·5 are based primarily on the distance the patient can walk and the need for an assistive device (ambulatory endurance). Also, the EDSS captures MS-related cogntive impairment poorly. Finally, the scale used is a non-linear ordinal scale—ie, the clinical importance of a 1·0 point change varies according to the starting score. As a result, populations show a bimodal distribution of EDSS categories, and the rate of progression through the scale by individual patients over time varies as a function of baseline score.14 These factors limit the statistical approaches that can be used to analyse EDSS data and complicate interpretation of changes in score over time.

Owing partly to its acceptance by regulators, the EDSS will probably remain the most widely used MS disability outcome measure in the near future, despite the associated issues. Several refi nements might improve performance. First, in the Neurostatus version of the EDSS, evaluators are instructed to question patients to calculate the scores for several functional systems. Development of a standard script for examining clinicians is recom mended to improve reliability and lessen the risk of unblinding in clinical trials.

Second, the scoring rules in the Neurostatus version, although more explicit than in previous versions, have become quite complex and no evidence is available that supports improved precision. Simplifi cation of the scoring rules and studies to document intra-rater and inter-rater reliability are advised.

Third, studies have shown that improvement in EDSS scores after relapse continues beyond 3 months in a substantial proportion of patients,15 which suggests that confi rmation of worsening by assessment at 6 months would indicate long-term accrual of disability more reliably than confi rmation at 3 months. A reduced event rate would, however, be the trade-off for this change. Substantial data from published trials could be used to compare the usefulness of EDSS assessment at 3 months and 6 months in relapsing-remitting and progressive MS.

For Neurostatus see http://www.neurostatus.net

Panel 1: Potential uses of disability outcome measures for multiple sclerosis

• Drug development• Diff erent methods will be applicable for diff erent

stages of the drug development process, with highly sensitive measures being needed for early-phase studies and clinically meaningful measures for pivotal trials, although they might be less sensitive

• Measures to detect worsening and improvement of disease are needed

• Observational studies• Assessment of function of specifi c neurological pathways to

correlate with physiological, imaging, or other biomarkers• Measurement and monitoring of disability in clinical practice

Panel 2: Obstacles to the development of outcome measures for clinical trials of MS

• The range and severity of MS manifestations are heterogeneous across patients

• In individual patients, the range, severity, accrual, and resolution of clinical manifestations vary over time

• Clinically meaningful disability in MS typically develops over decades, but clinical trials typically last only 2–3 years

• The MS disease process is largely subclinical in the early stages when intervention might be most valuable, which makes meaningful changes on clinical disability measures diffi cult if not impossible to detect

• Disability is more apparent in progressive forms of MS than in non-progressive forms, but the pathobiology might be degenerative and less responsive to available (anti-infl ammatory) treatment strategies, which makes full validation of candidate outcome measures—specifi cally for confi rmation of treatment eff ects—diffi cult

• Measurement of disability in progressive forms of MS is particularly sensitive to confounding by functional changes due to factors that are not directly related to pathological changes (eg, spasticity, deconditioning, or metabolic derangements)

• A trade-off might be seen between sensitivity to change, treatment eff ects, or both, of an outcome measure and its clinical meaningfulness and interpretability, which creates a tension between measurement properties and clinical relevance

MS=multiple sclerosis.

www.thelancet.com/neurology Vol 11 May 2012 469

Review

Fourth, analysis of existing datasets to assess which functional systems most frequently contribute to confi rmed worsening at diff erent levels of disability (ie, which are informative) might enable streamlining of the EDSS. Consistency and validation across studies of various MS clinical phenotypes and therapeutic agents would, however, be required for simplifi cation of scoring to be appropriate.

Finally, modifi cation of the EDSS would be desirable to improve linearity of measurement, which would facilitate statistical analysis and clinical interpretation. How this refi nement could be accomplished without fundamentally changing the scale, though, is unclear.

Despite the recognised shortcomings of the EDSS, major changes should be avoided so that this scale’s historical legacy and principal advantage—acceptance by regulators—are not lost. Rather, we recommend that the practical features be refi ned as far as possible and that EDSS be used in conjunction with other measures that better assess the clinical domains it covers inadequately, such as cognition. We anticipate that in time other approaches to assessment of disability in MS will supplant the EDSS.

Multiple sclerosis functional compositeThe MSFC was originally proposed to address the perceived shortcomings of the EDSS and other clinical rating scales.3,4 By contrast with the EDSS, which is based on the standard neurological examination, the MSFC uses quantitative tests in three domains (table). The result on each component test is converted to a Z score, for which the mean and SD values for the pooled study population at baseline are typically used for reference. These three Z scores are averaged to yield one composite score. Advantages of the MSFC are that it can be done in 15–20 min by a trained technician, it is highly reliable, it covers three major MS clinical domains, and it yields a single score on a continuous scale. Because the results of the component tests are normalised, they can be compared directly within a study. Since the introduction of the MSFC, many data have accumulated that show cross-sectional and longitudinal correlations with disease stage,16 EDSS,17 MRI,18 patient-reported outcomes for health status and quality of life,19 employment status,20 and driving performance.21 Changes in MSFC score over a short time are predictive of evolution from relapsing-remitting MS to secondary progressive MS,22 worsening of EDSS score,4,22 decline in quality of life,19 and loss of whole-brain volume.23

The MSFC has not always been more sensitive than the EDSS to treatment eff ects in clinical trials, partly because it does not assess several important clinical domains, such as vision. Also, MSFC component tests can have fl oor or ceiling eff ects that depend on the study population’s disability level. Moreover, the MSFC has not been accepted as an alternative to the EDSS by regulators.

A vision test was not included when the MSFC was originally proposed because no test was at the time deemed adequate.3,4 The Sloan low-contrast letter acuity test has since been proposed as a candidate visual measure.17,24 Test charts are available in a standard format developed for the Early Treatment of Diabetic Retinopathy Study. The low-contrast charts show more sensitivity for MS-related visual impairment than do high-contrast (black on white) charts or the Pelli-Robson contrast sensitivity charts.24 The results of the Sloan low-contrast letter acuity test correlate with disease stage (ie, results are worse for patients with secondary progressive MS than for those with relapsing-remitting MS), thinning of the retinal-nerve-fi bre layer on OCT, MRI-defi ned brain lesion burden, and patient-reported visual impairment and quality of life. Changes in Sloan low-contrast letter acuity are clinically meaningful.25 The test detected disease worsening and proven treatment eff ects in two phase 3 trials of natalizumab in relapsing-remitting MS.26 Formal inclusion of the Sloan low-contrast letter acuity test in the MSFC seems appropriate, although prospective testing is needed to confi rm whether its addition improves MSFC performance overall. Add-itionally, the optimum contrast level and whether monocular or binocular testing is preferable need to be determined. Finally, Sloan low-contrast letter acuity sometimes worsens over time without a history of overt optic neuritis;27 whether this phenomenon refl ects episodes of subclinical optic neuritis or progression due to neurodegeneration remains unclear.

The cognitive test included in the original MSFC, the paced auditory serial addition task (PASAT),28 measures processing speed and working memory, which are two

Functional system scores: visual, brainstem, pyramidal,

cerebellar, sensory, bowel and bladder, cerebral

Neurologicalexamination

Ability to carry out activities of daily living

Ambulation distance and requirement of anassistance device

Expanded disability

status scale score (0–10)

Figure 1: Overview of the scoring of the expanded disability status scale7

Test Unit of measurement

Ambulation Timed walk over a distance of 25 ft Time to completion (seconds)

Upper-extremity function

Nine-hole peg test Time to completion (seconds)

Cognition Paced auditory serial addition test Number of correct responses (of a total of 60)

Table: The multiple sclerosis functional composite3,4

470 www.thelancet.com/neurology Vol 11 May 2012

Review

important features in MS-related cognitive impair ment.29,30 This test is reasonably reliable, sensitive to the presence of cognitive impairment, and responsive to change31 but is disliked by many patients, requires mathematical ability, and has prominent practice eff ects that are only partly addressed by use of run-in testing.5,32 The symbol digit modalities test (SDMT),33 which measures information processing speed for visually presented stimuli, has been proposed as an alternative to the PASAT.34 The SDMT is practical, better accepted than the PASAT by patients because it is self-paced, has at least equal reliability35 and sensitivity to the presence36 and worsening37 of MS-related cognitive impairment, and correlates better with MRI metrics.38,39 Disadvantages are its dependence on vision, which can be compromised in patients with MS, and that it also has practice eff ects, albeit less prominent than those for the PASAT. The SDMT is an attractive candidate to replace the PASAT in the MSFC. We recommend prospective confi rmation of its ability to detect treatment eff ects. Of note, however, is that a single cognitive test cannot capture the full range of potential cognitive manifestations of MS. Studies that focus on cognitive impairment, therefore, will need to use a range of tests.

The 25 ft walk in the MSFC primarily assesses ambulation speed. The 6 min walk test could be a useful alternative to measure walking endurance and, because it is more challenging, might be less subject to fl oor eff ects in studies enrolling populations with mild gait impairments.40 Validation of the 6 min walk test in other diseases is an advantage from a regulatory perspective.41 For studies enrolling patients with severe gait impair ment, the current walking test is likely to remain a better option.

The principal objection to the MSFC expressed by regulators relates to the abstract and dimensionless nature of the summary score. Many clinicians are unfamiliar with Z scores, and the reference population aff ects the absolute values for the components and their weighting.42 As a result, MSFC scores cannot be easily interpreted clinically or compared across studies. An alternative analytical approach is to defi ne worsening as a decrease in score by a prespecifi ed amount in any of the component tests, which can be determined reliably and has clinical relevance (eg, 20%),43 and to show worsening in the same component at two sequential time points. This approach is similar to that used to analyse EDSS data, although EDSS worsening typically does not require deterioration in the same functional system at both assessments. When this approach was applied post hoc to the Natalizumab Safety and Effi cacy in Relapsing-Remitting Multiple Sclerosis (AFFIRM) dataset,17 sensitivity to confi rmed worsening and benefi t of natalizumab treatment were similar for the MSFC and EDSS, and confi rmed MSFC worsening was associated with worsening of EDSS score, relapse rate, and quality of life.17 Nevertheless, although this adaptation might address the main concern of regulators about how clinically meaningful the MSFC score is, it might decrease

its sensitivity to change. One option is to use change in MSFC score as an endpoint in phase 2 studies and confi rmed worsening as the primary endpoint in phase 3 trials. Further work is needed to determine the optimum magnitude of worsening for each component test. To ensure clinical relevance, any such changes should be correlated with the clinical eff ects of patient-reported outcomes and changes in functional disabilities. We support continued eff orts to revise the MSFC to improve its performance and make it accetable to regulators.

Novel measures of disabilityEven if the EDSS and MSFC can be improved, additional methods are needed to measure disability accumulation. Several potential approaches are under development.

Composite endpointsThe use of composite endpoints can increase study sensitivity by enbling the capture of data for multiple clinical domains of MS. This approach can be particularly useful to assess events that occur too infrequently for an adequate sample size or duration of a trial to be feasible. Small or non-signifi cant treatment eff ects on individual clinical domains might add up to generate a signifi cant response when combined. This issue could become increasingly important in future MS trials, where low relapse rates and slow accumulation of disability are expected. If composite endpoints can provide a more complete picture of patients’ status than single endpoints, they might also be more clinically meaningful. Several important caveats must, however, be considered when developing any composite endpoint (panel 3).

Disease-activity-free status (DAFS) is the absence of measurable disease activity, and is a new composite outcome concept.44 DAFS, as used so far in clinical trials45–47 and observational studies of MS,48 is based on occurrence of relapses, confi rmed EDSS worsening, and MRI lesion activity (gadolinium-enhancing and new T2-hyperintense lesions). The parameters for disease activity can be tailored for the specifi c study design and adapted as more sensitive measures are developed and validated, including imaging measures and other biomarkers that relate to disability.

The specifi c procedures used to measure clinical and imaging outcomes, the frequency of assessments, and the duration of follow-up aff ect the sensitivity of DAFS to disease activity. Therefore, these parameters must be standardised to ensure the outcomes are comparable across studies, although they might be diffi cult to control in observational studies. Also, if one component is much more sensitive than the others, DAFS predominantly refl ects that outcome. As currently used in MS studies, DAFS is driven primarily by MRI-defi ned activity events and, therefore, might not show the clinically meaningful diff erential eff ects sought by regulatory agencies. Finally, as DAFS is a dichotomous outcome, some important data might be sacrifi ced in some settings when it is used.

www.thelancet.com/neurology Vol 11 May 2012 471

Review

Global measures of activities of daily livingThe EDSS and MSFC were developed as MS-specifi c measures. Wider assessments of function are potentially useful and could be acceptable to regulators, especially to indicate the clinical relevance of MS-specifi c outcomes and to enable comparisons across diseases. For example, the Rankin scale49,50 and Barthel index,51 which were originally developed to measure ability to perform activities of daily living after stroke, have been used as outcomes in a few MS studies, particularly those focused on rehabilition.52,53 The CombiRx trial was a 3-year study of 1000 patients with relapsing-remitting MS. It included change in Rankin scale score as an endpoint and will provide important data on its usefulness in clinical trials of MS.54 Whether such global measures of disability should be used as adjuncts to or could replace MS-specifi c measures will undoubtedly depend on the purpose of the study.

Patient-reported outcomesPatient-reported outcomes, which are defi ned as “any report of a patient’s health condition that comes directly from the patient, without interpretation of the patient’s response by a clinician or anyone else,”55 are being explored in several disorders. Studies are aided by the Patient Reported Outcomes Measurement Information System (PROMIS), the Neuro-QOL project, and by US Food and Drug Administration guidance.55 The European Medicines Agency also encourages the use of patient-reported outcomes,56 but has not provided specifi c regulatory guidance. A potential disadvantage of this approach is its susceptibility to unblinding and resultant expectation bias. Nevertheless, by the assessment of symptoms and function from the patient’s perspective, patient-reported outcomes used in association with clinician-determined measures can provide an indication of the clinical relevance of objective outcomes.

Patient-reported outcomes that are being tested for their applicability in MS include general instruments, such as the medical outcomes study short form-36 (assesses vitality, physical functioning, pain, general perceptions of health, physical, emotional and social functioning, and mental health),57 MS quality of life-54 (derived from short form-36),58 MS quality of life inventory,59 and MS impact scale-29.60,61 These multidimensional scales measure several domains, but patient-reported outcomes can focus on single domains, such as ambulation (MS walking scale-1262), depression (Beck depression inventory63 and patient health questionnaire-964), or fatigue (modifi ed fatigue impact scale65).

Initially, most questionnaires for patient-reported outcomes are fi xed in content and length—ie, all respondents are asked the same questions. The number of items can be reduced substantially by use of item-response theory and computer adaptive testing (eg, PROMIS) to target questions through an iterative process in which responses determine which items are

subsequently presented. Such tailored testing improves the precision and tolerability of assessments. Computer adaptive testing, however, requires development and validation of algorithms in addition to development and validation of the original questionnaire.

We support assessment of patient-reported outcomes in association with clinician-assessed objective disability outcomes because they involve participants in trials and can provide a valuable sense of the clinical eff ects of an intervention from patients’ perspectives, including the trade-off between effi cacy, practical issues, and adverse eff ects. The validation of patient-related outcomes and their roles in trials need further assessment and discussion with regulators.

Other exploratory outcomesSeveral electronic devices are being developed to quantify disability. For example, a glove analyser system enables researchers to record kinematic data from fi nger opposition movement and thereby to quantify arm and hand function.66 Applications for “smart” phones and other mobile devices are being studied that might provide ways to assess mobility on a near-real-time basis.67–69

Assessment of improvement in neurological functionMost observational studies and clinical trials in MS so far have focused on worsening disability. Treatment benefi ts related to confi rmed improvement in EDSS were,

For PROMIS see http://www.nihpromis.org

For Neuro-QOL see http://www.neuroqol.org

Panel 3: Caveats in the development of composite endpoints

• Interpretation might not be straightforward, especially if not tied to clinically relevant outcomes

• Component measures must be responsive to the expected biological eff ects of the agents to be tested

• Sensitivities frequently vary between measures for individual components because of intrinsic measurement properties, the sampling frequency, or the duration of the study, and, therefore, normalisation or weighting the components might be necessary, although it must not obscure the clinical meaning of the composite

• In some settings, the outcomes for the components of composite endpoints could move in opposite directions (improvement vs harm), which could illustrate a valid trade-off between benefi t and risk, but might obscure a straightfoward interpretation of effi cacy if component scores are averaged

• An increase in the number of outcome measures for a composite endpoint might or might not increase the sensitivity of the overall composite, especially if some components do not change

• Care should be taken to avoid outcome measures that yield redundant information, because false additives could cause the composite score to change disproportionately in patients who accrue defi cits in overlapping domains compared with that in patients with defi cits in distinct domains

• Increased sensitivity to change of the composite endpoint does not necessarily increase its sensitivity to treatment eff ects (eg, if the numbers of events in all treatment arms increase proportionally but the variability also increases, sensitivity to group diff erences may weaken)

472 www.thelancet.com/neurology Vol 11 May 2012

Review

however, reported in post-hoc analyses of clinical trials of alemtuzumab70 and natalizumab.71 Analysis of MSFC in this manner should also be possible and, in fact, fampridine was approved on the basis of improve ment in the walking component.72,73 Outcome measures for neurological improvement will be increasingly important as treatment strategies that promote repair and other restorative eff ects are developed. Longitudinal natural-history data on spontaneous improvement in MS will be essential to interpret these results.

BiomarkersMRIBecause MRI-detected lesion activity is seen more frequently than clinical relapses occur, and because MRI features might worsen before disability accrues, this type of imaging might be used to develop useful surrogate or composite endpoints for disease activity or status. Secondary endpoints based on imaging out-comes are well accepted and have been used in pivotal MS clinical trials. Additionally, activity measures, such as the number of gadolinium-enhancing lesions or newly appearing T2-hyperintense lesions on serial MRI, have proven indispensable in phase 1 and 2 clinical trials to help to establish the potential of anti-infl ammatory therapies to advance through the drug pipeline.74 Meta-analyses confi rm a strong correlation at trial level between treatment eff ects on MRI-defi ned activity measures and relapse activity. Benefi cial and adverse eff ects of therapy on subclinical lesional activity can been seen in advance of eff ects on clinical relapses.75 Data that have accumulated for some drug classes, such as interferon beta, suggest that MRI-monitored activity events could be predictive enough of treatment eff ects on clinical relapses to serve as true surrogates and be assessed during the development of new drugs within a class that has well established effi cacy in MS.60 Data to support this position have, however, not been presented in suffi cient detail for regulators to endorse this use.

The continuing work to validate changes in MRI-detected lesion activity as a trial-level surrogate endpoint for relapse in relapsing-remitting MS can serve as a model for development of its use as a surrogate for disability progression or reversal. The magnitude of therapeutic impact on MRI-defi ned acute activity also correlates at the group level with treatment eff ects on accumulated disability over the course of a typical 2-year trial in relapsing-remitting MS.8 These correlations, however, are weaker than those between therapeutic eff ects on MRI-detected lesion activity and clinical relapses. The known therapeutic eff ects on MRI-defi ned acute activity in trials of relapsing-remitting MS have so far only resulted in non-signifi cant trends for the proportion of patients accumulating clinical disability in trials of secondary progressive MS.76 Therefore, while substantial progress has been made in the development of surrogate endpoints for clinical relapses based on

MRI-defi ned acute activity, the MRI features that are adequately predictive of future disease progression and what magnitude of attenuation is required to predict consistently a benefi cial outcome in accumulated disability or transition from relapsing-remitting MS to secondary progressive MS remain unclear.

Before surrogate endpoints based on MRI features can be proposed for studies of treatment eff ects on disability progression, data are needed that defi ne, among other things, the relation with disability change in both relapsing-remitting and progressive MS. The kinetic features of this relation will be most crucial. Conventional MRI features (eg, lesion conversion to persistent T1-hypointense foci, global or regional brain atrophy, and spinal cord atrophy), so-called advanced MRI features (eg, magnetisation transfer imaging, diff usion tensor imaging, and chemically specifi c magnetic-resonance spectroscopy), and newly developed PET imaging of demyelination, axonal damage, and remyelination should be assessed.77,78 Tools to analyse CNS tissue damage and predict benefi ts on clinical disability are needed as trials of potential neuroprotective agents and those that promote repair are beginning.

Composite MRI endpoints have been used in some clinical trials in an eff ort to maximise detection of therapeutic eff ects when drug eff ects cannot be easily anticipated and the importance of individual MRI features is diffi cult to rank.79 Although this analytical approach is potentially powerful, the same caveats apply as for the interpretation of the dimensionless MSFC score and for composite measures in general.

Optical coherence tomographyOCT is a rapidly maturing technology that generates high-resolution retinal images (fi gure 2) and enables quantifi cation of pathology relevant to MS. The magnitude and kinetics of change in the retinal-nerve-fi bre layer following optic neuritis are well defi ned,27,80 and thinning correlates with poor Sloan low-contrast letter acuity test results27 and reduced patient-reported visual quality of life.81 Other OCT retinal features, including macular volume and segmented ganglion-cell layer thickness, are also under study in MS. This technique should be included in prospective therapeutic trials of relapsing-remitting and progressive MS to assess its ability to detect treatment eff ects and to determine whether it provides insight into tissue integrity distinct from MRI. Of note, however, is that recruitment has been slow in several trials of retinal-nerve-fi bre layer quantifi cation.

Other biomarkersBiomarkers, such as cerebrospinal fl uid IgM bands, neurofi lament light chains, and antibodies to glycans, might correlate with or predict disability status. Much work is, however, needed to validate outcome measures for such markers in trials.82

www.thelancet.com/neurology Vol 11 May 2012 473

Review

A path forwardWhile no new or revised measures can be recommended immediately for primary endpoints in clinical trials of MS, several strategies to improve assessment of MS-related disability can be proposed. Despite the shortcomings of the EDSS, it is the only disability measure currently accepted by regulators as a primary trial endpoint. Thus, fundamental changes, such as elimination of non-informative functional systems or modifi cations to improve linearity, need to be considered carefully. Our principal recommendations focus instead on practical refi nements. While the availability of standardised detailed instructions for the performance and scoring of EDSS are advantageous, we suggest further standardisation of scripts for questioning patients and assessment of simplifi ed instructions. The reliability and sensitivity to change over time of the Neurostatus version need to be documented.

For the the study of MS to advance, more informative disability measures are needed. Acceptance of the EDSS by regulators seems to be based largely on its historical legacy rather than on it satisfying the criteria required to validate potential substitute measures. These criteria, in fact, create a notable hindrance to the adoption of new outcomes. Substantial data that support the MSFC as a valid alternative to the EDSS already exist, and ways to improve its use by investigators and its acceptablility to regulators should be actively pursued. The major impediment to regulatory acceptance seems to be the uncertain clinical meaning of the dimensionless composite score. An alternative scoring approach based on confi rmed worsening in each component could address this weakness. Also, broadening of the range of clinical domains, particularly by inclusion of a vision test, and reconsideration of the currently used cognitive assessment tool are recommended, so long as added tests are sensitive to change in the population being studied and improve the performance of the MSFC.

In addition to refi ning the EDSS and MSFC, several new strategies are being developed. Methods to assess composite endpoints, such as DAFS, show promise, although several important caveats must be borne in mind. The current statuses of global measures of function, patient-reported outcomes, advanced imaging techniques, and other biomarkers are insuffi cient to serve as primary endpoints in pivotal trials of disability worsening or improvement. Although a surrogate endpoint based on lesion activity on conventional MRI is being considered to assess treatment eff ects on relapse in specifi c settings, no MRI feature is suffi ciently validated to act as a substitute for clinical assessment of disability.

To attain these goals will require a collaborative approach that involves academic experts, regulators, industry representatives, and funding agencies. A model for such collaborative projects is outlined in the US Food and Drug Administration’s Critical Path Initiative.83,84

Eff orts should include systematic review of existing datasets to identify ways to refi ne existing measures or recommendations for new measures, followed by validation in prospective analyses for regulatory accept-ance. As an example, a task force of this type was appointed by the Research Programs Advisory Committee of the US National MS Society to review the MSFC approach, propose specifi c modifi cations to make it suitable as a primary disability outcome measure for registration trials, and develop a formal plan of action. Similar targeted eff orts will be needed to make progress in the refi nement, validation, and implementation of other disability-related outcomes.

ConclusionsSeveral guiding principles apply to all outcome measures. First, they must show clinically relevant eff ects. Second, consideration needs to be given to the fact that no single outcome measure will be applicable in all settings. Selection must be tailored to the disability level of the study population (to avoid fl oor and ceiling eff ects), the presumed action of the intervention, and the purpose of the study. Third, the necessary data for validation of new or revised outcome measures and comparison with existing measures should be obtained by incorporation into observational studies and into clinical trials as measures of secondary or exploratory endpoints. Finally, regulatory guidance as early as possible in the

Figure 2: Optical coherence tomographyThis image shows thinning of the peripapillary retinal-nerve-fi bre layer (thickness 69 μ) in a patient with multiple sclerosis. Reproduced by permission of Robert Bermel, Cleveland Clinic, Cleveland, OH, USA.

Search strategy and selection criteria

We searched PubMed for articles published from January, 2007, to January, 2012, with the terms “multiple sclerosis”, “outcomes”, “disability”, “impairment”, “MRI”, and “optical coherence tomography”. We also identifi ed articles by searching our own fi les, referenced in meeting presentations, or suggested by meeting participants. Only articles published in English were reviewed. The fi nal reference list was selected on the basis of originality and relevance to topic.

474 www.thelancet.com/neurology Vol 11 May 2012

Review

develop ment process is essential. Regulators largely defi ne what will be acceptable as a study outcome, and their input on issues such as specifi city, sensitivity, and clinical meaningfulness in any exploration of revised or new outcomes will help to guide the validation process.

ContributorsJAC, SCR, CHP, and JSW did the literature review, wrote and revised the

paper, and had fi nal responsibility for its content and the decision to

publish. The conference participants had the opportunity to review and

provide input on the manuscript.

Members of the International Advisory Committee on Clinical Trials in Multiple Sclerosis, 2011–12Chris H Polman (Chair), Department of Neurology, and Frederik Barkhof,

Diagnostic Radiology, VU Medical Centre, Amsterdam, Netherlands;

Laura J Balcer, Department of Neurology, University of Pennsylvania,

Philadelphia, PA, USA; Brenda Banwell, Division of Neurology, Hospital

for Sick Children, Toronto, ON, Canada; Peter Calabresi, Department of

Neurology, Johns Hopkins Hospital, Baltimore, MD, USA;

Michel Clanet,Fédération de Neurologie, CHU Hôpital Purpan, Toulouse,

France; Jeff rey A Cohen, Mellen Center for MS Treatment and Research,

Neurological Institute, Cleveland Clinic, Cleveland,OH, USA; Giancarlo

Comi, Institute of Experimental Neurology, Department of Neurology,

University Vita-Salute, Scientifi c Institute San Raff aele, Milan, Italy;

Gary R Cutter, Department of Biostatistics, University of Alabama at

Birmingham, Birmingham, AL, USA; Mark S Freedman, Department of

Medicine (Neurology), University of Ottawa and Ottawa Hospital Research

Institute, Ottawa, ON, Canada; Andrew D Goodman, Department of

Neurology, University of Rochester Medical Center, Rochester, NY, USA;

Ludwig Kappos, Department of Neurology, University Hospital Basel,

Basel, Switzerland; Bernd C Kieseier, Department of Neurology,

Heinrich-Heine-University, Düsseldorf, Germany; Catherine Lubetzki,

Department of Neurology, Hôpital de la Salpêtrière, University Paris VI

(UPMC), Paris, France; Fred D Lublin and Aaron Miller, Corinne

Goldsmith Dickinson Center for MS, Mount Sinai School of Medicine,

New York, NY, USA; Xavier Montalban, Unitat de Neuroimmunologia

Clinica, Hospital Universitari Vall d’Hebron, Barcelona, Spain;

Paul O’Connor, Division of Neurology, St Michael’s Hospital, Toronto, ON,

Canada; John Petkau, Department of Statistics, University of British

Columbia, Vancouver, BC, Canada; Carlo Pozzilli, Department of

Neurological Science, University of Rome “La Sapienza,” Rome, Italy;

Stephen C Reingold, Scientifi c and Clinical Review Association, New York,

NY, USA; Per Soelberg Sørensen, Danish Multiple Sclerosis Research

Center, Department of Neurology, Copenhagen University Hospital,

Rigshospitalet, Copenhagen, Denmark; Maria Pia Sormani, Unit of

Biostatistics, Health Sciences Department, University of Genoa, Genoa,

Italy; Olaf Stuve, Department of Neurology, University of Texas Health

Sciences Center, Dallas, TX, USA; Alan J Thompson, UCL Institute of

Neurology, National Hospital for Neurology and Neurosurgery, London,

UK; and and Jerry S Wolinsky, Department of Neurology, University of

Texas Health Sciences Center, Houston, TX, USA.

Confl icts of interestJAC has received consulting or speaking fees from Biogen Idec, Elan,

Eli Lilly, Novartis, Teva and Vaccinex, and receives grant support from the

US Department of Defense and the National Institutes of Neurological

Disorders and Stroke. SCR has received consulting fees and/or travel

funds for data safety monitoring board and other advisory activities from

Barofold, Bayer Healthcare, BioMarin Pharmaceutical Inc, Cleveland

Clinic Foundation, Coronado Biosciences Inc, Eli Lilly & Company, Merck

Serono, European Committee for Treatment and Research in Multiple

Sclerosis, Genentech, F Hoff mann-La Roche, INC Research, Ironwood

Pharmaceutical Company, Isis Pharmaceuticals Inc, MediciNova Inc,

National Multiple Sclerosis Society, Novartis Pharmaceuticals

Corporation, Sanofi -Aventis, SK Biopharmaceuticals, and Synthon

Pharmaceuticals Inc. CHP has received honoraria, consultation fees or

research support from Actelion, Biogen Idec, Bayer Schering,

GlaxoSmithKline, Merck Serono, Novartis, TEVA, UCB, and Roche. JSW

has received compensation for consultancy or other advisory activities

from Astellas, Bayer HealthCare, Bayer Multiple Sclerosis Council,

Celgene, Eli Lilly, F Hoff man La Roche, Novartis Pharmaceuticals,

Sanofi -Aventis, Teva Neuroscience, and Teva Pharmaceuticals, and has

received royalties for outlicensed monoclonal antibodies through the

UTHSCH to Millipore (Chemicon International) Corporation. JSW also

receives contractual support from Sanofi -Aventis, and grant support from

the US National MS Society and the National Institutes of Neurological

Disorders and Stroke.

AcknowledgmentsThe International Conference on Disability Outcomes in Multiple

Sclerosis, held in Washington, DC on May 19–20, 2011, was organised by

the International Advisory Committee on Clinical Trials in MS and

included additional experts in MS clinical trials, representatives of

pharmaceutical and biotechnology companies, and regulators from the

USA, Europe, and Canada (appendix). The meeting’s goal was to gather

expert opinion about disability outcomes currently used in MS clinical

trials and to make recommendations concerning potential revised or

new approaches to improve the effi ciency and practcality of future

studies. The activities of the International Advisory Committee on

Clinical Trials in Multiple Sclerosis are supported by the European

Committee for Treatment and Research in Multiple Sclerosis

(ECTRIMS) and the US National MS Society, and the 2011 conference

was supported by these organisations and the Americas Committee for

Treatment and Research in Multiple Sclerosis, the Multiple Sclerosis

Society of Canada, the Multiple Sclerosis International Federation, and

by the following pharmaceutical companies: Bayer HealthCare

Pharmaceuticals, Biogen Idec, F Hoff mann-La Roche, Genzyme

Corporatioon, Novartis Corporation, Sanofi -Aventis, and Teva

Pharmaceutical Industries. No non-industry participants received direct

support from the industry sponsors. We thank the members of the

International Advisory Committee on Clinical Trials in Multiple

Sclerosis, Ralph Benedict, Deborah Miller, Richard Rudick, and

Bernard Uitdehaag, and the other conference participants for reviewing

the paper and providing helpful suggestions.

References1 Herndon R. Proceedings of the International Conference on

Therapeutic Trials in Multiple Sclerosis. Grand Island, NY, April 23–24, 1982. Arch Neurol 1983; 40: 663–710.

2 Whitaker JN, McFarland HF, Rudge P, Reingold SC. Outcomes assessment in multiple sclerosis clinical trials: a critical analysis. Mult Scler 1995; 1: 37–47.

3 Rudick R, Antel J, Confavreux C, et al. Recommendations from the National Multiple Sclerosis Society Clinical Outcomes Assessment Task Force. Ann Neurol 1997; 42: 379–82.

4 Cutter GR, Baier ML, Rudick RA, et al. Development of a multiple sclerosis functional composite as a clinical trial outcome measure. Brain 1999; 122: 871–82.

5 Cohen JA, Cutter GR, Fischer JS, et al. Benefi t of interferon β-1a on MSFC progression in secondary progressive MS. Neurology 2002; 59: 679–87.

6 WHO. Disabilities. http://www.who.int/topics/disabilities/en/ (accessed March 15, 2012).

7 Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurology 1983; 33: 1444–52.

8 Sormani MP, Bonzano L, Roccatagliata L, Mancardi GL, Uccelli A, Bruzzi P. Surrogate endpoints for EDSS worsening in multiple sclerosis. A meta-analytic approach. Neurology 2010; 75: 302–09.

9 Hobart J, Freeman J, Thompson A. Kurtzke scales revisited: the application of psychometric methods to clinical intuition. Brain 2000; 123: 1027–40.

10 Amato MP, Fratiglioni L, Groppi C, Siracusa G, Amaducci L. Interrater reliability in assessing functional systems and disability on the Kurtzke scale in multiple sclerosis. Arch Neurol 1988; 45: 746–48.

11 Noseworthy JH, Vandervoort MK, Wong CJ, Ebers GC, the Canadian Cooperative MS Study Group. Interrater variability with the expanded disability status scale (EDSS) and functional systems (FS) in a multiple sclerosis clinical trial. Neurology 1990; 40: 971–75.

12 Francis DA, Bain P, Swan AV, Hughes RAC. An assessment of disability rating scales used in multiple sclerosis. Arch Neurol 1991; 48: 299–301.

www.thelancet.com/neurology Vol 11 May 2012 475

Review

13 Goodkin DE, Cookfair D, Wende K, et al. Inter- and intrarater scoring agreement using grades 1.0 to 3.5 of the Kurtzke Expanded Disability Status Scale (EDSS). Neurology 1992; 42: 859–63.

14 Weinshenker BG, Rice GPA, Noseworthy JH, Carriere W, Baskerville J, Ebers GC. The natural history of multiple sclerosis: a geographically based study. 4. Applications to planning and interpretation of clinical therapeutic trials. Brain 1991; 114: 1057–67.

15 Lublin FD, Baier M, Cutter G. Eff ect of relapses on development of residual defi cit in multiple sclerosis. Neurology 2003; 61: 1528–32.

16 Ozakbas S, Ormeci B, Idiman E. Utilization of the multiple sclerosis functional composite in follow-up: relationship to disease phenotype, disability and treatment strategies. J Neurol Sci 2005; 232: 65–69.

17 Rudick RA, Polman CH, Cohen JA, et al. Assessing disability progression with the multiple sclerosis functional composite. Mult Scler 2009; 15: 984–97.

18 Kalkers NF, Bergers L, de Groot V, et al. Concurrent validity of the MS Functional Composite using MRI as a biological disease marker. Neurology 2001; 56: 215–19.

19 Miller DM, Rudick RA, Cutter G, Baier M, Fischer JS. Clinical signifi cance of the multiple sclerosis functional composite. Relationship to patient-reported quality of life. Arch Neurol 2000; 57: 1319–24.

20 Honarmand K, Akbar N, Kou N, Feinstein A. Predicting employment status in multiple sclerosis patients: the utility of the MS functional composite. J Neurol 2011; 258: 244–49.

21 Shawaryn MA, Schultheis MT, Garay E, DeLuca J. Assessing functional status: exploring the relationship between the multiple sclerosis functional composite and driving. Arch Phys Med Rehabil 2002; 83: 1123–29.

22 Rudick RA, Cutter G, Baier M, et al. Use of the multiple sclerosis functional composite to predict disability in relapsing MS. Neurology 2001; 56: 1324–30.

23 Fisher E, Rudick RA, Cutter G, et al. Relationship between brain atrophy and disability: an 8-year follow up study of multiple sclerosis patients. Mult Scler 2000; 6: 373–77.

24 Balcer LJ, Baier ML, Cohen JA, et al. Contrast letter acuity as a visual component for the Multiple Sclerosis Functional Composite. Neurology 2003; 61: 1367–73.

25 Balcer LJ, Frohman EM. Evaluating loss of visual function in multiple sclerosis as measured by low-contrast letter acuity. Neurology 2010; 74 (suppl 3): S16–23.

26 Balcer LJ, Galetta SL, Calabresi PA, et al. Natalizumab reduces visual loss in patients with relapsing multiple sclerosis. Neurology 2007; 68: 1299–304.

27 Talman LS, Bisker ER, Sackel DJ, et al. Longitudinal study of vision and retinal nerve fi ber layer thickness in multiple sclerosis. Ann Neurol 2010; 67: 749–60.

28 Gronwall DMA. Paced auditory serial addition task: a measure of recovery from concussion. Percept Mot Skills 1977; 44: 367–73.

29 Chiaravalloti ND, DeLuca J. Cognitive impairment in multiple sclerosis. Lancet Neurol 2008; 7: 1139–51.

30 Benedict RHB, Zivadinov R. Risk factors for and management of cognitive dysfunction in multiple sclerosis. Nat Rev Neurol 2011; 7: 332–42.

31 Nagels G, Geentjens L, Kos D, et al. Paced visual serial addition test in multiple sclerosis. Clin Neurol Neurosurg 2005; 107: 218–22.

32 Cohen JA, Cutter GR, Fischer JS, et al. Use of the multiple sclerosis functional composite as an outcome measure in a phase 3 clinical trial. Arch Neurol 2001; 58: 961–67.

33 Smith A. Symbol digit modalities test: Manual. Torrance, CA: Western Psychological Services, 1982.

34 Drake AS, Weinstock-Guttman B, Morrow SA, Hojnacki D, Munchauer FE, Benedict RHB. Psychometrics and normative data for the multiple sclerosis functional composite: replacing the PASAT with the symbol digit modalities test. Mult Scler 2010; 16: 228–37.

35 Benedict RH. Eff ects of using same- versus alternate-form memory tests during short-interval repeated assessments in multiple sclerosis. J Int Neuropsychol Soc 2005; 11: 727–36.

36 Benedict RHB, Cookfair D, Gavett R, et al. Validity of the minimal assessment of cognitive function in multiple sclerosis (MACFIMS). J Int Neuropsychol Soc 2006; 12: 549–58.

37 Portaccio E, Goretti B, Zipoli V, et al. Reliability, practical eff ects, and change indices for Rao’s brief repeatable battery. Mult Scler 2010; 16: 611–17.

38 Benedict RH, Weinstock-Guttman B, Fishman I, Sharma J, Tjoa CW, Bakshi R. Prediction of neuropsychological impairment in multiple sclerosis: comparison of conventional magnetic resonance imaging measures of atrophy and lesion burden. Arch Neurol 2004; 61: 226–30.

39 Benedict RH, Bruce JM, Dwyer MG, et al. Neocortical atrophy, third ventricular width, and cognitive dysfunction in multiple sclerosis. Arch Neurol 2006; 63: 1301–06.

40 Goldman MD, Marrie RA, Cohen JA. Evaluation of the six-minute walk in multiple sclerosis subjects and healthy controls. Mult Scler 2008; 14: 383–90.

41 Duncan PW, Jorgensen HS, Wade DT. Outcome measures in acute stroke trials: a systematic review and some recommendations to improve practice. Stroke 2000; 31: 1429–38.

42 Fox RJ, Lee J-C, Rudick RA. Optimal reference population for the multiple sclerosis functional composite. Mult Scler 2007; 13: 909–14.

43 Schwid SR, Goodman AD, McDermott MP, Bever CF, Cook SD. Quantitative functional measures in MS: what is a reliable change? Neurology 2002; 58: 1294–96.

44 Lublin FD. Disease free activity status. Mult Scler Relat Dis 2012; 1: 6–7.

45 Havrdova E, Galetta S, Hutchinson M, et al. Eff ect of natalizumab on clinical and radiological disease activity in multiple sclerosis: a retrospective analysis of the Natalizumab Safety and Effi cacy in Relapsing-Remitting Multiple Sclerosis (AFFIRM) study. Lancet Neurol 2009; 8: 254–60.

46 Kappos L, Radue EW, O’Connor P, et al. Fingolimod treatment increases the proportion of patients who are free from disease activity in multiple sclerosis: results from a Phase 3, placebo-controlled study (FREEDOMS) (Poster PD6.002). Neurology 2011; 76 (suppl 4): A563.

47 Giovannoni G, Cook S, Rammohan K, et al. Sustained disease-activity-free status in patients with relapsing-remitting multiple sclerosis treated with cladribine tablets in the CLARITY study: a post-hoc and subgroup analysis. Lancet Neurol 2011; 10: 329–37.

48 Horga A, Castillo J, Rio J, et al. An observational study of the eff ectiveness and safety of natalizumab in the treatment of multiple sclerosis. Rev Neurol 2011; 52: 321–30.

49 Rankin J. Cerebral vascular accidents in patients over the age of 60. II. Prognosis. Scott Med J 1957; 2: 200–15.

50 Farrell B, Godwin J, Warlow C. The United Kingdom transient ischaemic attack (UK-TIA) aspirin trial: fi nal results. J Neurol Neurosurg Psychiatry 1991; 54: 1044–54.

51 Mahoney F, Barthel D. Functional evaluation: the Barthel index. Md State Med J 1965; 14: 61–65.

52 Smith SJ, Young CA. The role of aff ect on the perception of disability in multiple sclerosis. Clin Rehabil 2000; 14: 50–54.

53 Airlie J, Baker GA, Smith SJ, Young CA. Measuring the impact of multiple sclerosis on psychosocial functioning: the development of a new self-effi cacy scale. Clin Rehabil 2001; 15: 259–65.

54 Conwit R, Cofi eld SS, Cutter GR, Wolinsky JS, Lublin FD. The modifi ed Rankin scale, EDSS and MRI measures using CombiRx baseline assessments (P05.070). Neurology 2008; 70 (suppl 1): A270.

55 Food and Drug Administration. Guidance for industry: patient-reported outcome measures: use in medical product development to support labeling claims. http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM193282.pdf (accessed March 15, 2012).

56 European Medicines Agency. Refl ection paper on the regulatory guidance for the use of health-related quality of life (HRQL) meaures in the evaluation of medicinal products. http://www.ema.europa.eu/ema/pages/includes/document/open_document.jsp?webContentId=WC500003637 (accessed March 15, 2012).

57 Stewart AL, Hays R, Ware J. The MOS short-form general health survey. Reliability and validity in a patient population. Med Care 1988; 26: 724–35.

58 Vickrey BG, Hays RD, Harooni R, Myers LW, Ellison GW. A health related quality of life measure for multiple sclerosis. Qual Life Res 1995; 4: 187–206.

59 Fischer JS, LaRocca NG, Miller DM, Ritvo PG, Andrews H, Paty D. Recent developments in the assessment of quality of life in multiple sclerosis (MS). Mult Scler 1999; 5: 251–59.

476 www.thelancet.com/neurology Vol 11 May 2012

Review

60 Hobart J, Lamping D, Fitzpatrick R, Riazi A, Thompson A. The multiple sclerosis impact scale (MSIS-29): a new patient-based outcome measure. Brain 2001; 124: 962–73.

61 van der Linden FA, Kragt JJ, Klein M, van der Ploeg HM, Polman CH, Uitdehaag BM. Psychometric evaluation of the multiple sclerosis impact scale (MSIS-29) for proxy use. J Neurol Neurosurg Psychiatry 2005; 76: 1677–81.

62 Hobart JC, Riazi A, Lamping DL, et al. Measuring the impact of MS on walking ability: the 12-item MS walking scale (MSWS-12). Neurology 2003; 60: 31–36.

63 Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry 1961; 4: 561–71.

64 Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med 2001; 16: 606–13.

65 Fisk JD, Pontefract A, Ritvo PG, Archibald CJ, Murray TJ. The impact of fatigue on patients with multiple sclerosis. Can J Neurol Sci 1994; 21: 9–14.

66 Bonzano L, Sormani MP, Tacchino A, Abate L, Mancardi GL, Uccelli A. Validation of a new quantitative and objective tool for the assessment of hand motor disability in multiple sclerosis. Mult Scler 2011; 17 (suppl 10): S43.

67 Snook E, Morand A, Scott CL. A functional living outcome measure for multiple sclerosis: combining GPS and activity data (S34). Int J MS Care 2011; 13 (suppl 3): 26.

68 Sosnoff JJ, Socie MJ, Boes MK, Sandroff BM, Motl RW. Ambulatory mobility monitoring in people with multiple sclerosis (S125). Int J MS Care 2011; 13 (suppl 3): 69–70.

69 Weikert ML, Dlugonski D, Suh Y, Sandroff B, Motl RW. Accelerometry measures walking mobility, not physical activity, in multiple sclerosis (S36). Int J MS Care 2011; 13 (suppl 3): 27.

70 Coles AJ, Fox E, Vladic A, et al. Alemtuzumab versus interferon beta 1-a in early relapsing-remitting multiple sclerosis: post-hoc and subset analyses of clinical effi cacy outcomes. Lancet Neurol 2011; 10: 338–48.

71 Phillips JT, Giovannoni G, Lublin FD, et al. Sustained improvement in expanded disability status scale as a new effi cacy measure of neurologic change in multiple sclerosis: treatment eff ects with natalizumab in patients with relapsing multiple sclerosis. Mult Scler J 2011; 17: 970–79.

72 Goodman AD, Brown TR, Krupp LB, et al. Sustained-release oral fampridine in multiple sclerosis: a randomised, double-blind, controlled trial. Lancet 2009; 373: 732–38.

73 Goodman AD, Brown TR, Edwards KR, et al. A phase 3 trial of extended release oral dalfampridine in multiple sclerosis. Ann Neurol 2010; 68: 494–502.

74 Barkhof F, Simon JH, Fazekas F, et al. MRI monitoring of immunomodulation in relapse-onset multiple sclerosis trials. Nat Rev Neurol 2011; 8: 13–21.

75 Sormani MP, Bonzano L, Roccatagliata L, Cutter GR, Mancardi GL, Bruzzi P. Magnetic resonance imaging as a potential surrogate for relapses in multiple sclerosis: a meta-analytic approach. Ann Neurol 2009; 65: 268–75.

76 Kappos L, Weinshenker B, Pozzilli C, et al. Interferon beta-1b in secondary progressive MS. A combined analysis of the two trials. Neurology 2004; 63: 1779–87.

77 Barkhof F, Calabresi PA, Miller DH, Reingold SC. Imaging outcomes for neuroprotection and repair in multiple sclerosis trials. Nat Rev Neurol 2009; 5: 256–66.

78 Stankoff B, Freeman L, Aigrot MS, et al. Imaging central nervous system myelin by positron emission tomography in multiple sclerosis using [methyl-¹¹C]-2-(4’-methylaminophenyl)-6-hydroxybenzothiazole. Ann Neurol 2011; 69: 673–80.

79 Wolinsky JS, Narayana PA, Noseworthy JH, et al. Linomide in relapsing and secondary progressive MS. Part II: MRI results. Neurology 2000; 54: 1734–41.

80 Petzold A, de Boer JF, Schippling S, et al. Optical coherence tomography in multiple sclerosis: a systematic review and meta-analysis. Lancet Neurol 2010; 9: 921–32.

81 Sakai RE, Feller DJ, Galetta KM, Galetta SL, Balcer LJ. Vision in multiple sclerosis: the story, structure-function correlations, and models for neuroprotection. J Neuroophthalmol 2011; 31: 362–73.

82 Graber JJ, Dhib-Jalbut S. Biomarkers of disease activity in multiple sclerosis. J Neurol Sci 2011; 305: 1–10.

83 Food and Drug Administration. Innovation or stagnation: critical path opportunities report and list. http://www.fda.gov/downloads/ScienceResearch/SpecialTopics/CriticalPathInitiative/CriticalPathOpportunitiesReports/UCM077254.pdf (accessed March 15, 2012).

84 Woodcock J, Woosley R. The FDA critical path initiative and its infl uence on new drug development. Annu Rev Med 2008; 59: 1–12.

414 www.thelancet.com/neurology Vol 11 May 2012

Articles

Lancet Neurol 2012; 11: 414–19

Published OnlineApril 10, 2012

DOI:10.1016/S1474-4422(12)70060-1

See Comment page 385

Department of Neurology (Prof A A Rabinstein MD,

J E Fugate DO, E F M Wijdicks MD) and

Department of Biomedical Statistics and Informatics

(J Mandrekar PhD), Mayo Clinic, Rochester, MN, USA;

Department of Neurology, University of California, San

Francisco, CA, USA (A H Yee DO); Department of Intensive Care,

Erasmus MC University Medical Center, Rotterdam,

Netherlands (Y J de Groot MD, E J O Kompanje PhD);

Departments of Neurosurgery and Neurology, University of

Cincinnati, OH, USA (L A Shutter MD); Department

of Neurology, Mayo Clinic, Jacksonville, FL, USA

(W D Freeman MD); and Department of Neurology,

Washington University School of Medicine, St Louis, MO, USA

(M A Rubin MD)

Correspondence to:Prof Alejandro A Rabinstein,

200 First Street SW, Mayo W8B, Rochester, MN 55905, USA

rabinstein.alejandro@mayo.edu

Prediction of potential for organ donation after cardiac death in patients in neurocritical state: a prospective observational studyAlejandro A Rabinstein, Alan H Yee, Jay Mandrekar, Jennifer E Fugate, Yorick J de Groot, Erwin J O Kompanje, Lori A Shutter, W David Freeman, Michael A Rubin, Eelco F M Wijdicks

SummaryBackground Successful donation of organs after cardiac death (DCD) requires identifi cation of patients who will die within 60 min of withdrawal of life-sustaining treatment (WLST). We aimed to validate a straightforward model to predict the likelihood of death within 60 min of WLST in patients with irreversible brain injury.

Methods In this multicentre, observational study, we prospectively enrolled consecutive comatose patients with irreversible brain injury undergoing WLST at six medical centres in the USA and the Netherlands. We assessed four clinical characteristics (corneal refl ex, cough refl ex, best motor response, and oxygenation index) as predictor variables, which were selected on the basis of previous fi ndings. We excluded patients who had brain death or were not intubated. The primary endpoint was death within 60 min of WLST. We used univariate and multivariable logistic regression analyses to assess associations with predictor variables. Points attributed to each variable were summed to create a predictive score for cardiac death in patients in neurocritical state (the DCD-N score). We assessed performance of the score using area under the curve analysis.

Findings We included 178 patients, 82 (46%) of whom died within 60 min of WLST. Absent corneal refl exes (odds ratio [OR] 2·67, 95% CI 1·19–6·01; p=0·0173; 1 point), absent cough refl ex (4·16, 1·79–9·70; p=0·0009; 2 points), extensor or absent motor responses (2·99, 1·22–7·34; p=0·0168; 1 point), and an oxygenation index score of more than 3·0 (2·31, 1·10–4·88; p=0·0276; 1 point) were predictive of death within 60 min of WLST. 59 of 82 patients who died within 60 min of WLST had DCD-N scores of 3 or more (72% sensitivity), and 75 of 96 of those who did not die within this interval had scores of 0–2 (78% specifi city); taking into account the prevalence of death within 60 min in this population, a score of 3 or more was translated into a 74% chance of death within 60 min (positive predictive value) and a score of 0–2 translated into a 77% chance of survival beyond 60 min (negative predictive value).

Interpretation The DCD-N score can be used to predict potential candidates for DCD in patients with non-survivable brain injury. However, this score needs to be tested specifi cally in a cohort of potential donors participating in DCD protocols.

Funding None.

IntroductionDonation after cardiac death (DCD) protocols allow families of patients who are dying but not brain dead to donate organs. Such protocols have been implemented in many countries and might reduce the shortage of available organs for transplantation.1 After withdrawal of life-sustaining treatment (WLST), DCD allows organ procurement in an operating room after irreversible cessation of respiration and circulation has been declared.2 Although such donation contributes an increasing proportion of viable organs for trans-plantation, identifi cation of appropriate candidates is a restricting factor.

Patients with catastrophic, irreversible brain injury who do not meet criteria for brain death are the most frequent candidates for DCD,3–5 but about half of these patients continue to breathe and maintain circulation for more than 60 min after WLST.6 The success of DCD relies on identifi cation of patients who are most likely to die within

60 min of WLST. Prolongation of the withdrawal phase of warm ischaemia time (ie, the time between WLST and end of cardiopulmonary function) beyond 60 min can compromise organ function.2,7 Thus, most DCD protocols do not continue with organ retrieval if the patient is still alive 60 min after WLST.8 However, good outcomes have been reported with organ transplants (particularly kidneys) retrieved after up to 4 h of warm ischaemia.9,10

Available scores used by organ-procurement organ-isations to estimate the time to death after WLST, such as the Wisconsin criteria3 or the United Network for Organ Sharing (UNOS) criteria,4 include little information about the neurological status of the patient before WLST. Calculation of these scores requires temporary disconnection of the patient from the mechanical ventilator and the scores are tailored to assess the degree of pulmonary and circulatory support. These variables might be of less prognostic value in patients with catastrophic neurological injury who have not progressed

Articles

www.thelancet.com/neurology Vol 11 May 2012 415

to brain death than in patients with major non-neurological injury. This notion was reinforced by our previous analysis6 in which several respiratory and haemodynamic parameters were associated with death within 60 min of WLST on univariate analyses, but not on the multivariable analysis that included elements of a neurological examination. This single-centre study6 of 149 patients who were in coma with irreversible brain injury suggested that, after WLST, four clinical variables were associated with death within 60 min of extubation: absent corneal refl ex, absent cough refl ex, extensor or absent motor response, and higher oxygenation index. These associations were confi rmed subsequently in a smaller, independent cohort.11

To further validate this approach, we undertook a prospective study to produce a new model to predict death within 60 min of WLST in patients with catastrophic cerebral damage, which was based on the previously described clinical variables. We aimed to develop a practical score for assessment of potential candidates for DCD.

MethodsStudy design and participantsIn this multicentre, observational study, we prospectively obtained data from consecutive adult, comatose patients with irreversible brain damage who underwent WLST at the intensive care units of six participating centres in the USA and the Netherlands. We enrolled patients in the study if anticipated death was attributable directly to severe brain injury (eg, massive head trauma, intracranial haemorrhage, ischaemic stroke with malig nant oedema, or anoxic damage after cardiorespiratory arrest). We excluded patients without tracheal intubation or who fulfi lled criteria for brain death. The study protocol was

approved by the institutional review board of each centre and consent for data collection was obtained from the next of kin when requested by the individual board.

We selected variables for data collection on the basis of fi ndings from our previous comprehensive analysis.6 Data obtained for this study included age, sex, corneal refl ex (present or absent), cough refl ex (present or absent), motor response to pain (absent or extensor response or better response), oxygenation index, and time to death after WLST. We calculated oxygenation index with the following formula: 100 × (FiO2 × mean airway pressure in cm H2O)/PaO2 in torr), where mean airway pressure is half the combination of peak airway pressure in cm H2O and peak end expiratory pressure in torr.

We assessed these variables at the last examination before WLST, which occurred after discontinuation of sedation and opiate analgesia. The endpoint for the analysis was death within 60 min of WLST.

Statistical analysisWe used univariate and multivariable logistic regression analyses with death within 60 min as a binary outcome variable to assess the associations with predictor variables identifi ed in our previous study.6 The area under the receiver operating characteristic (ROC) curve was estimated as a measure of the ability of the model to discriminate between individuals who died within 60 min of WLST versus those who died after 60 min. An area under the ROC estimate of 0·7–0·8 was regarded as acceptable, 0·8–0·9 was regarded as excellent, and more than 0·9 was regarded as outstanding.12 We did sample size calculations with the assumption that oxygenation index would be the dichotomised variable that would need the maximum sample size.6 Assuming a rate of death within 60 min of 50% and a total sample size of 150 patients (ie, 75 per group), we would have 80% power to detect a diff erence of 45% versus 23% for the presence of higher oxygenation index level between patients who died within 60 min and those who died after 60 min, at a 5% level of signifi cance with the χ² test. We used NQuery advisor (version 6) for sample size and power calculations. To account for the possibility of missing or unusable data we aimed to recruit at least 15% more patients, meaning that our target enrolment was 175–180 patients. We used ROC curve analysis to identify the cutoff to dichotomise continuous variables of interest, such as oxygenation index, or the estimated score that increased the sum of sensitivity and specifi city to the highest amount. We also applied parameter estimates from the multivariable logistic regression model (fi tted to prospective data) to the previous retrospective data to calculate the predicted probabilities of health for every patient. We then used these predictive probabilities to estimate the area under the ROC. We did all analyses apart from the sample size calculation with SAS version 9.2.

Patients (n=178)

Age, years 63·4 (17·1, 17–94)

Sex, female 84 (47%)

Primary diagnosis

Intracerebral haemorrhage 60 (34%)

Ischaemic stroke 33 (19%)

Subarachnoid haemorrhage 30 (17%)

Head trauma* 26 (15%)

Others 26 (15%)

Active cancer 11 (6%)

Renal failure† 16 (9%)

Liver failure‡ 6 (3%)

Sepsis§ 5 (3%)

Data are mean (SD, range) or n (%). *Includes subdural haematoma. †Creatinine >221 μmol/L, creatinine clearance <0·5 mL/s, or renal replacement therapy. ‡Documented cirrhosis or acute hepatitis with altered protein synthesis (ie, increased international normalised ratio >1·5 without alternative explanation). §Sepsis syndrome with documented bacteraemia.

Table 1: Baseline characteristics

Articles

416 www.thelancet.com/neurology Vol 11 May 2012

Role of the funding sourceThere was no funding source for this study. All authors had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.

ResultsBetween March 30, 2010, and April 1, 2011, we assessed 178 patients. Table 1 summarises baseline characteristics of our study population. We enrolled 70 patients at the Mayo Clinic in Rochester (MN, USA), 39 at the Erasmus MC University Medical Center in Rotterdam (Netherlands), 30 at the University of Cincinnati (OH,

USA), 17 at the Mayo Clinic in Jacksonville (FL, USA), 12 at Washington University in St Louis (MO, USA), and ten at the University of California, San Francisco (CA, USA).

82 patients (46%) died within 60 min of WLST, 97 (54%) died within 2 h, and 109 (61%) died within 4 h. Absent corneal refl ex (p=0·0334), absent cough refl ex (p=0·0002), extensor or absent motor response to pain (p=0·0331), and higher oxygenation index (p=0·0183) were associated with death within 60 min of WLST in a multivariable logistic regression model including oxygenation index as a continuous variable.

Construction of a straight forward scoring system that allows quick estimation of probabilities of death within 60 min required all variables to be categorical or dichotomised. Thus, we explored the distribution of oxygenation index in the entire cohort and, using ROC analysis, we established that an oxygenation index of 3·0 had the best combination of sensitivity and specifi city to discriminate between patients who died within 60 min and those who died more than 60 min after WLST. Table 2 shows the distribution of the variables of interest in the two groups of patients and the strength of the associations.

We created a score to predict the chance of death within 60 min of WLST (ie, the DCD-N score) on the basis of odds ratios for every variable, assigning 2 points for absent cough refl ex and 1 point each for absent corneal refl exes, absent or extensor motor response to pain, and oxygenation index of more than 3·0. The probability of death within 60 min of WLST increased as the score increased (table 3). The area under the curve for the score was 0·81 (95% CI 0·75–0·87) for prediction of death within 60 min (fi gure), 0·77 (0·70–0·84) for death within 2 h, and 0·76 (0·69–0·83) for death within 4 h. Compared with a score of 0–2, a score of 3–5 had a sensitivity of 72%, a specifi city of 78%, a positive pre dictive value of 74%, and a negative predictive value of 77% to predict death within 60 min of WSLT. In other words, 59 of 82 (72%) patients who died within 60 min had a score of 3 or more (sensitivity) and 75 of 96 (78%) of patients who did not die within this interval had a score of 0–2 (specifi city). Taking into account the prevalence of death within 60 min in this population, a score of 3 or more translates into a 74% probability of death within 60 min (positive predictive value) whereas a score of 0–2 translates into a 77% probability of survival beyond 60 min (negative predictive value). Table 4 shows the probabilities of death within 60 min according to specifi c combinations of variables. We validated the performance of the score in our previously reported retrospective cohort of 149 patients (table 5). In this retrospective cohort,8 a cutoff score of 3 (scores 3–5 vs 1–2) had a sensitivity of 81%, a specifi city of 73%, a positive predictive value of 75%, and a negative predictive value of 79% to predict death within 60 min of WSLT.

Death within 60 min (n=82)

Death after 60 min (n=96)

Odds ratio (95% CI)

p value

Absent corneal refl ex 64 (74%) 31 (33%) 2·67 (1·19–6·01) 0·0173

Absent cough refl ex 45 (55%) 12 (13%) 4·16 (1·79–9·70) 0·0009

Extensor or absent motor response 72 (88%) 52 (54%) 2·99 (1·22–7·34) 0·0168

Oxygenation index >3·0 60 (73%) 50 (52%) 2·31 (1·10–4·88) 0·0276

Data are n (%) unless otherwise stated. Oxygenation index presented as a categorical variable (>3·0 vs ≤3·0).

Table 2: Distribution of variables of interest according to time to death after withdrawal of life-sustaining measures

Deaths within 60 min Deaths after 60 min

0 1 (5%) 18 (95%)

1 11 (27%) 30 (73%)

2 11 (29%) 27 (71%)

3 15 (52%) 14 (48%)

4 12 (80%) 3 (20%)

5 32 (89%) 4 (11%)

Overall 82 (46%) 96 (54%)

Table 3: Frequency of death after withdrawal of life-sustaining measures according to donation after cardiac death in patients in a neurocritical state (DCD-N) score

Figure: Receiver operating characteristic curve based on a multivariable model for prediction of death within 60 minDashed line shows equivalent sensitivity and 1–specifi city.

Sens

itivi

ty

1·0

0·8

0·6

0·4

0·2

0

1-specificity

1·00·80·40·2 0·60

Articles

www.thelancet.com/neurology Vol 11 May 2012 417

We also tested a previously described linear predictor, model with oxygenation index as a continuous variable.11 We adjusted the weight of the variables on the basis of the strength of the associations identifi ed in this prospective cohort, which was much larger than the one used to develop the linear predictor model. The resulting model used the following formula:

Logit = –2·49 + (0·90 × absent corneal refl ex) + (1·65 × absent cough refl ex) + (0·98 × extensor or absent motor response) + (0·12 × oxygenation index)

The equation Exp(logit)/1 + Exp(logit) can then be used to calculate the probability of death within 60 min of WLST for individual patients. Thus, for a patient with absent corneal refl ex, present cough refl ex, absent motor response to pain, and an oxygenation index of 6 (whose DCD-N score would be 3) the logit would be 0·11 and the predicted probability of death within 60 min of WLST would be 53%.

DiscussionWhen the decision to proceed with WLST is reached, medical teams in many countries contact local organ-procurement organ isations to address the possibility of organ donation. In these situations, reliable identifi cation of appropriate DCD candidates is essential for all concerned. Because catastrophic brain injury is the most common cause of death in patients who might be suitable for DCD, such methods should be especially applicable

to patients in neurocritical state. In this study, we validated the predictive value of a small set of variables that can be used to identify such patients confi dently (panel). Furthermore, we present a method—the DCD-N score—that can be used at the bedside to quantify the likelihood of death within the time required in most DCD protocols.

The variables included in the present study were identifi ed after a comprehensive analysis of various neurological and non-neurological parameters in a retrospective cohort of patients in neurocritical state.6 In that analysis, several other variables were associated with death within 60 min of WLST, including several haemodynamic and additional respiratory function measures. However, on multivariable analysis only three neurological variables (absent cough, absent corneal refl exes, and absent or extensor motor response to pain) and pulmonary function (oxygenation index) were independently associated with death within 60 min of extubation.

Prediction of time to death after WLST on the basis of clinical impression has proven inaccurate. In a recent prospective, multicentre, observational study of potential DCD donors, the clinical judgment of the treating intensive-care doctor had a fairly high sensitivity (73%) but a low specifi city (56%) to predict death within 60 min.13 Because of this restricted ability to predict the time of death reliably, the authors of that study suggested that the DCD procedure should be used for every potential donor to avoid loss of viable organs. However, this approach is troublesome for grieving families, labour-intensive for transplantation teams, and very expensive for hospitals. Instead, improvement of our ability to identify good candidates for donation should remain the objective.

Other scores to estimate early death after WLST are in use (eg, UNOS and the University of Wisconsin criteria) but have disadvantages.3,4 These scores incorporate neurological information only at the level of conscious-ness, which is not a reliable predictor of early death after terminal extubation in patients in neurocritical state.14 Moreover, their calculations require disconnection from mechanical ventilation for 10 min. Our scoring system has been specifi cally designed to be used in neurological

Absent corneal refl ex

Absent cough refl ex

Extensor or absent motor response

Oxygenation index >3·0

Score Probability

No No No No 0 0·08

No No No Yes 1 0·16

Yes No No No 1 0·18

No No Yes No 1 0·20

No Yes No No 2 0·26

Yes No No Yes 2 0·34

No No Yes Yes 2 0·37

Yes No Yes No 2 0·40

No Yes No Yes 3 0·45

Yes Yes No No 3 0·48

No Yes Yes No 3 0·51

Yes No Yes Yes 3 0·61

Yes Yes No Yes 4 0·68

No Yes Yes Yes 4 0·71

Yes Yes Yes No 4 0·74

Yes Yes Yes Yes 5 0·87

1 point was assigned for each of absent corneal refl ex, absent or extensor motor response to pain, and oxygenation index of more than 3·0. 2 points were assigned for an absent cough refl ex.

Table 4: Probabilities of death within 60 min according to the combinations of predictive variables

Death within 60 min Death after 60 min

0 0 14 (100%)

1 6 (12%) 21 (78%)

2 8 (30%) 19 (70%)

3 18 (62%) 11 (38%)

4 15 (71%) 6 (29%)

5 28 (90%) 3 (10%)

Overall 75 (50%) 74 (50%)

Table 5: Frequency of death after withdrawal of life-sustaining measures according to donation after cardiac death in patients in a neurocritical state (DCD-N) score in the retrospective cohort of 149 patients6

Articles

418 www.thelancet.com/neurology Vol 11 May 2012

patients with severe, irreversible brain injury and it can be fully assessed while the potential donor remains supported by mechanical ventilation. Further more, the information used to calculate the DCD-N score is usually available as part of the routine medical care of these patients.

To develop the DCD-N score, we transformed oxygenation index into a categorical variable. However, its predictive value is best when it is analysed as a continuous variable.11 The linear predictor model that incorporates the actual oxygenation index measurement is a valid alternative to the DCD-N score. Thus, the two approaches provide diff erent ways of assessing the chance of death within 60 min: the DCD-N score off ers cutoff s (eg, a DCD-N score of >2 suggests that 74% of patients will die within this time interval) and the linear predictor model expresses the likelihood of this outcome as a percentage. One model has the advantage of simplicity, the other delivers greater precision.

Our study has limitations. We prospectively collected data for only those variables that we had identifi ed previously as predictive of death within 60 min of WLST in a retrospective cohort.6 We regarded this approach as suffi cient because the previous study consisted of a comprehensive exploratory analysis of many neuro log i-cal and systemic variables, including those listed in presently used criteria. Furthermore, the validity of the predictive value of the variables we had identifi ed had been subsequently confi rmed in a diff erent cohort of patients with severe, irreversible brain injury.11 Never-theless, other variables not included in our predictive technique might also infl uence the likelihood of early death after extubation5,13,15 and the DCD-N score has not been compared against other predictive criteria. Although sedatives and opiates were stopped before assessment, a residual eff ect from medication might have had a confounding role. Moreover, our study was not restricted to patients deemed potential candidates for DCD and consequently some patients with advanced age, cancer, and severe infections were entered in the analysis. However, because our models rely mostly on neurological criteria, we would not expect their predictive performance to change substantially when used to screen patients in neurocritical state who are under consideration for DCD. Finally, the score we propose should be used only for patients dying from irreversible acute brain disease and should not be extrapolated to other candidates for DCD.

DCD protocols are used increasingly in the USA, UK, and Europe, although there are notable exceptions such as Germany.16 However, in some centres these protocols are underused because of concerns that the potential donor might not die within the accepted maximum time of warm ischaemia. We believe that the DCD-N score has good potential to advance the practice of DCD by improving the identifi cation of appropriate candidates. Its discriminative power is supported by the area under the curve noted in our ROC analysis. Nevertheless, the

score does not identify all patients who die within 60 min of WLST or at later times up to 4 h. Thus, future studies might help refi ne the score to reduce this under-recognition of potential donors. Such studies could explore combination of the DCD-N score with the Wisconsin or the UNOS criteria or incorporation of haemodynamic or additional respira tory variables. More importantly, the DCD-N score needs to be tested specifi cally in a prospective cohort of patients participating in DCD protocols as potential donors.

The DCD-N score provides a readily accessible estimate of the likelihood of death within 60 min of WLST in patients with critical brain injury who are dependent on artifi cial life support. The score needs to be tested in patients for whom consent of DCD has been obtained. If the reliability of its performance is confi rmed, this scoring technique could help guide resource allocation without compromising the availability of viable DCD donors.

ContributorsAAR, AHY, and EFMW designed the study with statistical assistance

from JM. JEF, AHY, YJdG, EJOK, LAS, WDF, and MAR collected the

data from their respective centres. AAR, AHY and EFMW supervised the

Panel: Research in context

Systematic reviewWe searched Medline and Embase databases up to Feb 29, 2012, for articles published in any language with the search terms “donation after cardiac death”, “donation after circulatory death”, and “prediction of time to death”. We also reviewed the reference lists of the papers identifi ed by this search. We assessed these articles for data for predictors of early death after withdrawal of life-sustaining treatment in patients with severe brain injury.

InterpretationThe fi ndings from our prospective, multicentre observational study suggest that the donation after cardiac death in patients in a neurocritical state (DCD-N) score predicts which patients with severe brain injury undergoing withdrawal of life-sustaining treatment will die within 60 min of extubation. This measure relies mainly on fi ndings from neurological examination and can be applied at the bedside without the need for transient disconnection from mechanical ventilation, unlike presently used predictive criteria (such as the United Network for Organ Sharing or Wisconsin criteria), which are more complex and might have reduced predictive value in patients with severe brain injury.3,4 Currently, donation after cardiac death protocols are sometimes underused because of concern that the potential donor will not die within the period of time compatible with an acceptable duration of warm ischaemia.10 The DCD-N score might be useful to identify the best candidates for donation among patients in a neurocritical state, thus reducing the chances of unsuccessful activation of retrieval teams and improved allocation of resources.

Articles

www.thelancet.com/neurology Vol 11 May 2012 419

study. JM did the statistical analysis. AAR, JM, AHY, and EFMW

interpreted the results of the analysis with subsequent substantial

contributions from all co-authors. AAR wrote the report. All authors

contributed with revisions and gave approval to the fi nal version of the

manuscript.

Confl icts of interestWe declare that we have no confl icts of interest.

AcknowledgmentsWe thank Maggie Brenner and Carolyn Koenig for their assistance with

enrolment of patients and data collection.

References1 Tuttle-Newhall JE, Krishnan SM, Levy MF, McBride V, Orlowski JP,

Sung RS. Organ donation and utilization in the United States: 1998–2007. Am J Transplant 2009; 9: 879–93.

2 Bernat JL, D’Alessandro AM, Port FK, et al. Report of a national conference on donation after cardiac death. Am J Transplant 2006; 6: 281–91.

3 Lewis J, Peltier J, Nelson H, et al. Development of the University of Wisconsin donation after cardiac death evaluation tool. Prog Transplant 2003; 13: 265–73.

4 De Vita MA, Mori Brooks M, Zawistowski C, Rudich S, Daly B, Chaitin E. Donors after cardiac death: validation of identifi cation criteria (DVIC) study for predictors of rapid death. Am J Transplant 2008; 8: 432–41.

5 Suntharalingam C, Sharples L, Dudley C, Bradley JA, Watson JE. Time to cardiac death after withdrawal of life-sustaining treatment in potential organ donors. Am J Transplant 2009; 9: 2157–65.

6 Yee AH, Rabinstein AA, Thapa P, Mandrekar J, Wijdicks EFM. Factors infl uencing time to death after withdrawal of life support in neurocritical patients. Neurology 2010; 74: 1380–85.

7 D’Alessandro AM, Fernandez LA, Chin LT, et al. Donation after cardiac death: the University of Wisconsin experience. Ann Transplant 2004; 9: 68–71.

8 Fugate JE, Stadtler M, Rabinstein AA, Wijdicks EFM. Variability in donation after cardiac death protocols: a national survey. Transplantation 2011; 91: 386–89.

9 Reid AWN, Harper S, Jackson CH, et al. Expansion of the kidney donor pool by using cardiac death donors with prolonged time to cardiorespiratory arrest. Am J Transplant 2011; 11: 995–1005.

10 Manara AR, Murphy PG, O’Callaghan G. Donation after circulatory death. Br J Anaesth 2012; 108 (suppl 1): i108–21.

11 de Groot YJ, Lingsma HF, Bakker J, Gommers DA, Steyerberg E, Kompanje EJ. External validation of a prognostic model predicting time of death after withdrawal of life support in neurocritical patients. Crit Care Med 2012; 40: 233–38.

12 Hosmer DW, Lemeshow S. Applied logistic regression, 2nd edn. Hoboken, NJ, USA: John Wiley and Sons, 2000.

13 Wind T, Snoeijs MG, Brugman CA, et al. Prediction of time of death after withdrawal of life-sustaining treatment in potential donors after cardiac death. Crit Care Med 2012; 40: 766–69.

14 Mayer SA, Kossoff SB. Withdrawal of life support in the neurological intensive care unit. Neurology 1999; 12: 1602–09.

15 Cooke CR, Hotchkin DL, Engelberg RA, Rubinson L, Curtis JR. Predictors of time to death after terminal withdrawal of mechanical ventilation in the ICU. Chest 2010; 138: 289–97.

16 Dominguez-Gil B, Haase-Kromwijk B, Van Leiden H, et al. Current situation of donation after circulatory death in European countries. Transpl Int 2011; 24: 676–86.

www.thelancet.com/neurology Vol 11 May 2012 443

Review

Childhood spinal muscular atrophy: controversies and challengesEugenio Mercuri, Enrico Bertini, Susan T Iannaccone

Spinal muscular atrophy is an autosomal recessive disorder characterised by degeneration of motor neurons in the spinal cord and is caused by mutations of the survival of motor neuron 1 gene SMN1. The severity of spinal muscular atrophy is highly variable and no cure is available at present. Consensus has been reached on several aspects of care, the availability of which can have a substantial eff ect on prognosis, but controversies remain. The development of standards of care for children with the disorder and the identifi cation of promising treatment strategies have changed the natural history of spinal muscular atrophy, and the prospects are good for further improvements in function, quality of life, and survival. A long-term benefi t for patients will be the development of eff ective interventions (such as antisense oligonucleotides), some of which are in clinical trials. The need to be prepared for clinical trials has been the impetus for a remarkable and unprecedented cooperation between clinicians, scientists, industry, government, and volunteer organisations on an international scale.

IntroductionSpinal muscular atrophy is a neuromuscular disease characterised by degeneration of α motor neurons in the spinal cord, resulting in progressive proximal muscle weakness and paralysis. The classical form of the disorder is caused by a genetic mutation1 in 5q11.2–q13.3, aff ecting the survival of motor neuron (SMN) gene.2 The severity of classical spinal muscular atrophy is highly variable and patients with heterogeneous clinical features can be classifi ed into four phenotypes on the basis of age of onset and maximum motor function achieved.3 No cure for spinal muscular atrophy is available but improved under-standing of the mechanisms underlying the disease has enabled the development of preclinical models to test potential therapeutic approaches.4 Some compounds are already being tested in clinical trials and others are likely to be available for testing in the next few years. These advances have provided the prospect of a possible cure for physicians and families who are willing to proactively manage the disease.

The increased attention to early diagnosis and to several aspects of management of spinal muscular atrophy has stimulated the development of clinical guidelines and standards of care,5,6 which have aff ected survival and the disease’s natural history. Studies of the most severe form of spinal muscular atrophy (type 1) have shown that survival beyond 1 year of age has improved as a result of the introduction of non-invasive ventilation and enteral feeding.7,8 By contrast with older investigations, which provided evidence of a progressive disorder, more recent studies of the course of type 2 and 3 spinal muscular atrophy have shown that motor and respiratory function do not change over a 12 month period.9,10 However, no consensus exists for numerous clinical aspects (eg, management of scoliosis or fundoplication in type 1 spinal muscular atrophy) of the disease.5

Our aim is not to provide a comprehensive Review of spinal muscular atrophy or therapeutic approaches—such reports are already available.5,6,11 Instead, we focus on the most recent advances in the fi eld of spinal muscular

atrophy, and particularly on aspects of care and methods of assessment that can be used both in clinical practice and as trial outcome measures. We discuss aspects of care for which consensus has been reached and those that are still controversial. We review natural history data for the diff erent forms of spinal muscular atrophy with paediatric onset, aiming to establish how earlier diagnosis, and improved understanding of the complications of the disorder and how to prevent and treat them, have aff ected patients with this debilitating disease.

Diagnosis and classifi cationThe fi rst diagnostic test for a patient with a normal or only mildly increased serum creatine kinase concen-tration who is suspected to have spinal muscular atrophy should be a search for a homozygous deletion in SMN1. The absence of SMN1 exon 7 (with or without deletion of exon 8) confi rms the diagnosis of spinal muscular atrophy. The test has 95% sensitivity and nearly 100% specifi city.5

If mutation analysis is negative, laboratory investi gations including electrophysiological tests such as electro-myography and nerve conduction tests should be done. If electromyography suggests a motor neuron disease, then further testing for SMN mutations should be undertaken. Genetic tests off er quick and reliable SMN1 gene copy-number testing by multiplex ligation-dependent probe amplifi cation or real-time PCR. Semi quantitative assays improve diagnostic sensitivity up to 98%.12,13 If the patient has one copy of SMN1, the coding region of the undeleted allele should be sequenced to identify the second causative mutation, usually subtle sequence variations such as point mutations, insertions, and deletions. Sequence analysis of SMN1 is recommended for patients who have typical symptoms, two copies of SMN1, and who are born to consanguineous parents or originate from genetic isolates. Patients homozygous for minor point mutations in SMN1 have been reported, albeit rarely.14

A large study15 based on data from 68 471 individuals showed that high-throughput population testing for carriers

Lancet Neurol 2012; 11: 443–52

Pediatric Neurology Unit, Catholic University, Rome, Italy (Prof E Mercuri MD); Department of Neuroscience, Unit of Neuromuscular Disorders, Laboratory of Molecular Medicine, Bambino Gesù Hospital, Rome, Italy (E Bertini MD); and Pediatric Neurology, Children’s Medical Center Ambulatory Care Pavilion in Dallas, University of Texas Southwestern Medical Center, Dallas, TX, USA (Prof S T Iannaccone MD)

Correspondence to:Prof Susan T Iannaccone, Professor of Neurology and Pediatrics, University of Texas, Southwestern Medical Center, Dallas, TX 75390-9063, USAsusan.iannaccone@utsouthwestern.edu

444 www.thelancet.com/neurology Vol 11 May 2012

Review

is feasible. The results showed a carrier frequency of one in 54 and a disease incidence of one in 11 000. Carrier frequency was slightly higher in white US (one in 47) and Taiwanese (one in 48) participants than in African Americans (one in 72).15,16

Spinal muscular atrophy is conventionally classifi ed into four phenotypes on the basis of age of onset and highest motor function achieved, with an additional phenotype (type 0) to describe the severe forms of antenatal-onset spinal muscular atrophy (table 1).3 This classifi cation has several clinical advantages but is not always adequate to provide prognostic information or to facilitate strati fi cation of patients in clinical trials.

In clinical trials, type 3 patients who have lost the ability to walk independently in childhood are often grouped as non-ambulant, or sitters, because they can be assessed with the same outcome measures. The number of copies of SMN2 might correlate with the severity of the phenotype and could be used as a biomarker in classifying patients for clinical trials. Although studies have shown that patients with a higher number of copies have, as a group, milder phenotypes, prediction of phenotype is not always accurate in individual cases.12,17,18

Standards of care and controversies in managementA consensus statement5 for standards of care was published in 2007, and is being updated. The document provides guidelines for aspects of diagnosis, assessment, and monitoring for which there was consensus among experts. Table 2 summarises aspects of assessment and management for which consensus was reached. Agreement has not been reached for several areas of care. The lack of agreement about some questions of assessment and management, such as management of scoliosis, is the result of diff erent approaches in diff erent countries and an absence of controlled comparative studies. For other aspects of care, such as the choice between palliative care and intervention in type 1 spinal muscular atrophy, the absence of consensus is related to diffi culties in defi ning standards of care without

considering caregiver burden and ethical and personal issues. The application of practice standards can be limited by cost or diff erent health-care delivery systems.

Palliative care in type 1 spinal muscular atrophyIntervention in type 1 spinal muscular atrophy increases survival,7 but the choice between inter vention and palliative care is not always easy. The best practice is to provide accurate information. The palliative care team should meet the family soon after diagnosis to help them to understand the diff erence between pro longing life and improving quality of life in terms relevant to their choices for intervention. The team can facilitate discussion between family members who might diff er in their opinions about what might be best for the child. They can liaise between the clinicians directing the child’s care and the family, who might have many questions but be reluctant to challenge the primary care provider. They might also help to put the child’s family in contact with other families who have been faced with similar challenges in the past and with advocacy support groups that can provide information for the families of newly diagnosed patients with spinal muscular atrophy.

Ventilatory support in type 1 spinal muscular atrophyThe natural history of type 1 spinal muscular atrophy has changed substantially in the past decade because of the availability of new technology and the implementation of an aggressive approach to improve survival and quality of life.19 Non-invasive ventilation can be used at a very young age—as early as the neonatal period—because interfaces are now available for small infants.20 Bi-level positive airway pressure ventilation can be provided at home with a face mask. Such technology enables the introduction of non-invasive ventilation while the infant is still reasonably healthy, as soon as or before paradoxical respirations are visible. Some infants might object to using a mask but most adjust quickly and sleep easily while being ventilated.21

As weakness progresses, the requirement for venti-lation might increase from 4–6 h per night to almost 24 h

Age of onset Maximum function achieved

Prognosis Proposed subclassifi cation SMN copy number

Type 0 (very severe) Neonatal with prenatal signs

Never sits If untreated, no survival beyond the fi rst months after birth

·· ··

Type 1 (severe) 0–6 months Never sits If untreated, life expectancy <2 years

1A, head control never achieved, signs in the neonatal period; 1B, head control never achieved, onset after neonatal period; 1C, head control achieved, onset after neonatal period

One or two copies of SMN2 in 80% of patients

Type 2 (intermediate) 7–18 months Sits but never stands Survival into adulthood Decimal classifi cation according to functional level, from 2·1 to 2·9

Three copies of SMN2 in >80% of patients

Type 3 (mild) >18 months Stands and walks Survival into adulthood 3A, onset of weakness before 3 years; 3B, onset of weakness after 3 years

Three or four copies of SMN2 in 96% of patients

Type 4 (adult) 10–30 years Stands and walks Survival into adulthood ·· Four or more copies of SMN2

Table 1: Classifi cation of spinal muscular atrophy

www.thelancet.com/neurology Vol 11 May 2012 445

Review

per day. With full-time ventilation the upper airway can become damaged, resulting in oedema, increased secretions, and bleeding. Such complications in the air-way or intercurrent infection can necessitate intubation. Whether to proceed to tracheostomy with long-term invasive ventilation is an individual choice for the child’s family. Many centres ask the palliative care team to off er support and advice to families who have to make diffi cult decisions. Most families opt to withdraw support when the child needs invasive ventilation. Children with type 1 spinal muscular atrophy who undergo tracheostomy and long-term ventilation remain dependent on ventilators for the rest of their lives. Some ventilator-dependent children attend school with the help of a carer. Some patients are at high risk of recurrent hypoxia, which is associated with brain injury.

The goal of intervention should always be to improve quality of life of the child, not to prolong life. Non-invasive ventilation in infants with spinal muscular atrophy can prevent or even reverse changes in the shape of the chest wall, increase lung growth, and slow the loss of chest wall compliance.22 The ultimate aim should be a decreased rate of infection and hospital admission, both of which can aff ect the quality of life of the child and his or her caregivers. The cost-eff ectiveness of these approaches has not yet been assessed.

Parenteral feeding in type 1 spinal muscular atrophyDecreased feeding is often the fi rst sign of progressive weakness. When breastfeeding, the child might have prolonged feeding times, cough while feeding, and tire quickly. Weight gain slows and then halts, and eventually weight is lost. Because poorly nourished children become fatigued and are more susceptible to infection, early gastrostomy is recommended. The surgery can be done soon after diagnosis and while the infant is healthy, and nocturnal feeding can be initiated to supplement

calories as oral feeding decreases. Even if the family decides against ventilation therapy, they might choose to proceed with gastrostomy. Gastrostomy can improve quality of life because the child will not be hungry even though he or she cannot safely eat. Fundoplication for type 1 patients at the time of gastrostomy placement is still controversial.

Cardiac fi ndings in type 1 spinal muscular atrophySome patients with severe type 1 spinal muscular atrophy (usually those who have one copy of SMN2) have heart defects,23,24 mostly atrial and ventricular septal defects, and possible involvement of the autonomic system, which might cause arrhythmia and sudden death. Some reports suggest that cardiac assessments should be done in patients with type 1 spinal muscular atrophy, but larger studies are needed to obtain more information about the prevalence, onset, and severity of cardiac symptoms.

Management of respiratory function in non-ambulant patientsRespiratory complications in type 2 or non-ambulant spinal muscular atrophy patients are less severe than in type 1. Implementation of standards of care and use of non-invasive ventilator support have substantially improved survival and quality of life. The methods of assessment and monitoring are quite similar to those used in type 1. Physical examination and assessment of cough eff ectiveness with respiratory muscle function tests should be routinely undertaken. Forced vital capacity should be measured in children older than 5 years. Overnight oximetry should be regularly done, especially in patients with severely reduced vital capacity (<65% predicted) or with clinical signs of nocturnal hypo-ventilation. Nocturnal hypoventilation should be treated with non-invasive ventilation.25

Pulmonary Gastrointestinal and nutritional Orthopaedic and rehabilitation

Assessment and monitoring Assistance and intervention Assessment and monitoring Assistance and intervention Assessment and monitoring Assistance and intervention

Non-sitters Assessment of cough eff ectiveness; respiratory muscle function tests; overnight oximetry; standard oximetry

Airway clearance; cough assistance; chest physiotherapy; nocturnal non-invasive ventilation (if nocturnal ventilatory failure); non-invasive ventilation (if daytime ventilatory failure)

Assessment of feeding (speech or occupational therapist); videofl uoroscopy (if indicated); search for signs of refl ux

Gastrostomy (if aspiration or poor effi ciency of feeding); Nissen fundoplication (if appropriate)

Physical and occupational therapy assessment (posture, contractures); hip and spine radiography; bone health

Equipment and devices for posture; splinting to preserve range of motion should be considered; no consensus on orthosis or surgery in scoliosis

Sitters and ambulant patients

Assessment of cough eff ectiveness; respiratory muscle function tests; forced vital capacity (patients >5 years); overnight oximetry

Airway clearance; cough assistance; chest physiotherapy; nocturnal non-invasive ventilation (if sleep-disorder breathing); immunisation and respiratory syncytial virus prophylaxis (when appropriate)

Assessment of feeding (speech or occupational therapist); videofl uoroscopy (if indicated); search for signs of refl ux

Optimise caloric intake with supplements (if not adequate intake but safe swallowing); gastrostomy (only if aspiration or poor effi ciency of feeding after calories supplemented orally); medical management (when appropriate)

Physical and occupational therapy assessment (posture, contractures, strength); assessment of power and manual mobility

Contracture management and exercise; orthoses

Table 2: Standards of care for pulmonary, gastrointestinal, nutritional, and orthopaedic care and rehabilitation

446 www.thelancet.com/neurology Vol 11 May 2012

Review

Airway clearance is important and there is evidence that assisted cough has a role in both prevention of infections and reduction of treatment duration. Avail-ability of mechanically assisted cough varies by country and although widely accepted as a part of the management of severe cases, its use in less severe cases is limited by cost. Further studies are needed to test the extent to which daily assisted cough can decrease the number of infections requiring hospital admission and to provide evidence for overall cost-eff ectiveness.

Feeding and nutritionFeeding and swallowing diffi culties are not uncommon in sitters and walkers. Gastro-oesophageal refl ux might be present but a questionnaire investigating subjective symptoms of refl ux did not support the routine diag-nostic use of oesophageal pH monitoring.26

Growth failure can occur for various reasons, from reduced intake because of masticatory muscle fatigue or prolonged meal times, to recurrent respiratory infections. Development of dysphagia because of progressive weakness of bulbar and oesophageal muscles can be insidious in adolescents with type 2 spinal muscular atrophy.27 Video swallow studies are helpful in the assessment of a patient’s chewing and swallowing.

At the other extreme, reduced activity and low energy expenditure increase the risk of obesity even in patients who have an apparently adequate nutritional intake for their age. Little evidence exists about how best to monitor these patients, partly because of diffi culties in obtaining accurate measures of height or body mass index in patients with spinal muscular atrophy in whom lean body mass is reduced.28–30 Further work is required to reach a consensus for the most appropriate methods with which to measure body composition and to defi ne optimum nutritional management,29 including the advantages and disadvantages of supplements and special formulas for patients with inadequate nutritional intake.

Scoliosis surgeryScoliosis occurs in almost all non-ambulant spinal muscular atrophy patients. Spinal fusion is the treatment of choice but diff erent centres have diff erent policies about when to do it and the role of orthotic management.5 No consensus exists for the effi cacy of braces or other trunk orthoses to slow the progression of scoliosis.31–33 Additionally, braces might compress the thoracic cage or have a negative eff ect on respiratory function.31,34

Scoliosis surgery should ideally be done in children older than 10 years. When progression of spinal curvature allows a delay until age 10 years or older, results are satisfactory for posture but no agreement among experts exists about the eff ects of surgery on pulmonary function.35–37 There is no consensus about children who develop severe and rapidly progressive scoliosis before age 5 years and new surgical treatments have been suggested for the management of skeletally immature patients.

Growing rods38 or vertical expandable prosthetic titanium ribs39,40 can be used to prevent progression of scoliosis in very young children if bracing is unsuccessful. These techniques have been eff ective in controlling progressive early-onset scoliosis before defi nitive spine fusion,41–43 although the specifi c role of each surgical technique has not been elucidated. Although growth rods can control early scoliotic curve and pelvic obliquity in young patients with spinal muscular atrophy, they do not halt rib collapse.43 Thus, a possible role for vertical expandable prosthetic titanium ribs might be to increase the space available for lung growth and chest compliance.

OsteoporosisIt is still a matter of debate whether reduced bone mineral density in spinal muscular atrophy is related to reduced mobility or to pathophysiological aspects of the disease. Comparative studies with other disorders such as Duchenne muscular dystrophy do not provide concordant results.44 In studies assessing dual-energy x-ray absorptiometry in children with type 2 and 3 spinal muscular atrophy, decreased bone mineral density seemed to be related more to increased age than to the ability to walk.44 Another comparative study of 79 children with diff erent neuromuscular disorders, particularly Duchenne muscular dystrophy and spinal muscular atrophy, showed that bone mineral density was lowest in patients with spinal muscular atrophy.45 Bone mineral density has been used as a secondary outcome in clinical trials in patients with spinal muscular atrophy.46

Results of a study of the Smn–/– SMN2 mouse model of spinal muscular atrophy showed a signifi cant decrease in the concentrations of osteoblast diff erentiation markers, and an increased rate of osteoclast formation and bone resorption capacity (46%) compared with wild-type mice, indicating that SMN function might be involved in bone remodelling and skeletal pathogenesis in spinal muscular atrophy.47 Improved understanding of the mechanisms of osteopenia is necessary to identify novel targets for therapeutic interventions and to establish standards of care, including clinical aspects of management, such as the promotion of weight bearing, that might help to prevent bone loss and reduce the risk of fracture.

Maintenance of independenceAbout half of patients with type 3 spinal muscular atrophy will lose independent ambulation by age 14 years.48,49 Only a small fraction are ambulatory throughout life—independent mobility can be severely aff ected by even the mild form of the disease. The decision to recommend wheelchair use depends on several factors, one of which is how often the patient falls. Daily falls often mean that the child needs a wheelchair but possibly only for part-time use. Another factor is fatigue, whether it occurs at the end of the day or is related to distance walked. A wheelchair might only be used for long distances such as for shopping. If the child is too weak to self-propel or

www.thelancet.com/neurology Vol 11 May 2012 447

Review

becomes easily fatigued operating the chair, then a motorised chair is recommended. In some cases, the motorised chair might be necessary only for school, while the child remains ambulant at home.

For adolescents, an important part of independence is the ability to drive a car. Technology to aid driving exists for all but the completely quadriplegic. Depending on the severity of paralysis, hand-activated or voice-activated controls make independent driving possible. The patient should be assessed by a qualifi ed occupational therapist with standardised assessment protocols and recom-mendations made about modifi cation of the steering wheel, accelerator, and brake controls. Such modifi cations are often expensive and not covered by insurance but many families are willing to pay the cost.

Transitional careTreatment and management for nearly all chronic diseases of childhood have improved greatly such that children are surviving into adulthood with diseases that are not well known to health-care providers for adults. Moreover, insurance coverage for childhood disease and its complications after the age of 21 years can be problematic. Children with severe chronic disease tend to be isolated, immature, and dependent on their parents.50 Therefore, the primary provider should take the lead in initiating the transition of care from paediatric clinics to an adult service. Around 16 years of age, patients can be placed in adolescent clinics, where they see other patients of their age and where they can learn to give their own interim histories and ask questions of their providers. In some centres, adult neuromuscular medicine and pulmonary specialists attend the adolescent clinics, making the transition from paediatric clinics easier. Patients should be encouraged to learn about their disease, to focus on life after school (especially higher education), and to discuss sexuality and family planning.

Pregnancy and childbirthAll patients with spinal muscular atrophy should be counselled about the risk of having aff ected children. Since the disease mutation is autosomal recessive, the risk is usually very low. However, patients should be aware of the risk that their partner might be a carrier of the gene and the partner should consider genetic testing before starting a family.

Little information exists about women with spinal muscular atrophy enduring pregnancy and labour. Loss of respiratory function and mobility are the most important risks in pregnancy. During labour, the choice of anaesthesia might be aff ected by previous scoliosis surgery, which can preclude epidural injections. General anaesthesia might be associated with a high risk of pulmonary complications for patients who need ventilation. Lung function should be regularly monitored during preg nancy. Most women with spinal muscular atrophy have a caesarean section and are more likely to have a premature baby.51

Progress and controversies in translational researchIn the past decade, clinical trials of several promising new therapeutic approaches have been undertaken in patients with spinal muscular atrophy,46,52–58 refl ecting the complexity of the mechanisms underlying the disorder (fi gure).4,11,63–66 The escalation of clinical trial planning has raised concerns about readiness for such trials. Are enough patients identifi ed who can be enrolled rapidly? Do we have reliable, valid, and sensitive outcome measures that can be implemented quickly and economically across many trial sites?67–69

Randomised double-blind placebo-controlled studies of rare disorders such as spinal muscular atrophy can only be done as large multicentre international trials. Numerous bottlenecks exist in undertaking such studies, related not only to diffi culties in enrolment and stratifi cation, but also to the need for common standards of care and consistent assessment methods and out-come measures to make data from diff erent centres comparable.67,70

Therapeutic targets Therapeutic approaches Trials completed or ongoing

Replacement ofSMN1

Gene replacementtherapy

..SMN1 genemutation

Inclusion of exon 7 Antisense oligonucleotides (newdrugs developed by PTCTherapeutics, tetracycline)

New drugs developed by ISIS Pharmaceuticals

Alternativesplicing ofSMN2 RNA

Increased amountsof SMN transcript

Histone deacetylaseinhibitors,quinazolones,RG3039,aminoglycerides,albuterol,prolactin

Phenylbutyrate (randomisedcontrolled trial) 52,53

Valproate (randomisedcontrolled trial) 46, 56, 58

Hydroxyurea (randomisedcontrolled trial) 59

Albuterol (open-label andongoing randomised controlled trial) 54, 60

Decreased full-length SMNtranscript

Stabilisation of SMNprotein

Indoprofen,proteasome inhibitors,polyphenols

..

Gabapentin (randomisedcontrolled trial) 61

Riluzole (open-label) 62

Olesoxime (TRO19622)

SMN proteindeficiency

Neuroprotection Neurotrophic factors

..Cell therapy Stem cells

Loss of motorneurons

.... ..Clinicalsymptoms

Figure: Relation between targets in the pathogenesis of spinal muscular atrophy and drug development

448 www.thelancet.com/neurology Vol 11 May 2012

Review

Enrolment, inclusion, and stratifi cationInternational registries for patients with spinal muscular atrophy have greatly increased the chances of identifying and recruiting patients for clinical trials. Recruitment and retention in clinical trials can be diffi cult because families can be discouraged by invasive delivery systems (such as intrathecal injection), frequent study visits, the number of tests done, study duration, and the possibility of receiving placebo. These challenges can be addressed by multicentre international studies that require a small number of patients per centre, and by creative randomisation—eg, 2:1 random-isation or interim ana lyses—which is more appealing to patients and their families.

In view of the clinical heterogeneity within each type of spinal muscular atrophy, clinical trial investigators suggest that stratifi cation should include functional measures to provide a fair representation of the diff erent levels of activities of patients in the treatment and placebo groups. Because the number of copies of SMN2 correlates with function,12,18 copy number could also be used for stratifi cation in clinical trials. Furthermore, in type 1 spinal muscular atrophy, survival and function diff er between infants who receive diff erent levels of care7 and should be considered at stratifi cation.

By contrast with early studies that provided evidence of a progressive disorder, more recent studies of patients who have received improved standards of care have shown that in type 2 and 3 spinal muscular atrophy motor and respiratory function do not change over a 12 month period.71 This result should be considered when designing studies in these groups.

Outcome measuresEff orts are being made to implement training for investigators and assessors in multicentre networks and to improve inter-rater reliability across centres with diff erent expertise.10,72 Collaboration within international networks such as the International Coordinating Committee or Translational Research in Europe—Assessment and Treatment of Neuromuscular Diseases has accelerated the development and validation of disease-specifi c outcome measures (table 3).67 One of the main advantages of this collaborative eff ort is that input from regulatory agencies and family advocacy groups has been taken into account, as well as the clinical views of investigators and statisticians, leading to the identifi cation of measures that are not only statistically robust, validated, and suitable for multicentre studies, but also clinically meaningful for patients and their caregivers. Although in type 1 spinal muscular atrophy survival or time to ventilation are the most important outcomes,88 in non-ambulant and ambulant patients functional motor scales seem to be the best methods to monitor clinically meaningful changes. Numerous clinical outcome measures have been promoted by investigators, including general scales that can be used across diff erent levels of severity63,76,89 and others developed specifi cally for non-ambulant or ambulant patients79–81,83,85,90,91 or for young infants.78,90 Neuro physiological techniques—such as compound muscle action potential—have proved to be reliable markers of the progression of motor neuron involvement.9,92–94 Dual-energy x-ray absorptiometry can be used to assess changes in lean mass.60 A paediatric questionnaire on quality of life in spinal muscular atrophy has been validated.88

The statistical robustness of functional motor scales currently in use is being determined, for example with Rasch analysis. Because scales are usually composed of diff erent items, using ordinal data, new methods of analysis are being used for neuromuscular functional scales to establish how the scale works and to identify possible gaps in the scale or redundant items, with the goal of converting ordinal data to linear measurement.95 The large Common Data Element project is underway—with funding and guidance from the US National Institute of Neurological Disorders and Stroke—in which case report forms and datapoints will be standardised according to disease. The fi rst goal of this project is to save time and money by having a central source of case report forms that would be immediately available once a clinical trial was ready to proceed. The second goal is to make data transferable and shareable for post-trial analysis.

Molecular biomarkersSMN gene products, either transcripts (both full-length and exon 7 deleted variants) or proteins, have been considered as biomarkers.96–98 Large multicentre studies,

Type of SMA Measure

Myometry61,73 2 and 3 Assessment of strength (Newtons)

Motor function measure (MFM)74,75 2 and 3 Functional scale, three domains

Gross motor function measure (GMFM)76 2 and 3 Functional scale

Children’s Hospital of Philadelphia infant test of neuromuscular disorders (CHOP-INTEND)77

Non-sitters or very weak sitters

Functional scale

Test of infant motor performance (TIMP)78 Non-sitters or very weak sitters

Functional scale

Hammersmith functional motor scale for SMA (HFMS)79

Sitters Functional scale

Upper limb module80 Sitters Functional scale

Expanded HFMS81,82 Sitters and ambulant patients

Functional scale with items from GMFM

Extended HFMS83,84 Sitters and ambulant

Functional scale with add-on fi ne, gross motor, and timed tests

6-minute walk test (6MWT)85 Ambulant Measure of endurance

Egen Klassifi cation (EK)86 Non-ambulant Questionnaire of functional abilities

PedsQL, neuromuscular module (NMM)87 Ambulant Quality-of-life questionnaire

Tests included in this table have been used and validated in studies of SMA (see references). SMA=spinal muscular atrophy.

Table 3: Outcome measures most commonly used in spinal muscular atrophy

www.thelancet.com/neurology Vol 11 May 2012 449

Review

such as BforSMA, have explored other candidate biomarkers. A molecular approach seems appropriate when therapy is based on increasing the amount of SMN protein produced by the residual SMN2 genes, through promoter activation or reduction of exon 7 alternative splicing, or both. However, the reliability of these molecular bio markers is still being investigated and none is ready for phase 3 trials.97

Family careGenetic counselling for couples who have one child with spinal muscular atrophy is commonly off ered. In most cases, both parents are carriers of one mutated allele. However, rarely, de novo point mutations can occur in one or both parents. If parents choose to pursue a successful preg nancy, they should be referred to a qualifi ed genetic counsellor associated with a high-risk pregnancy management team for a discussion of options, including carrier screening, prenatal diagnosis of the fetus, use of sperm or egg donors, and in-vitro fertilisation with pre-implantation testing.

Little information exists about the frequency of use of in-vitro fertilisation and pre-implantation genetic diagnosis in families with spinal muscular atrophy. One report from Poland describes fi ve families who had four healthy babies.99 This procedure is prohibitively costly for most families and is generally done by private organisations with little regulation or oversight, especially in the USA. Anecdotal reports exist of aff ected babies being born after such procedures. Whether negative outcomes are caused by poor practice or for biological reasons is unknown.

Testing asymptomatic siblings is controversial. Several professional organisations have suggested that genetic testing in asymptomatic children might be unethical.100 Many parents are anxious about the possibility of another child developing symptoms of the disease; since both severe and mild forms of spinal muscular atrophy can occur in siblings, their fear is not unfounded. Decisions about such testing should be made on an individual basis, taking into account each family’s needs. All should be counselled that asymptomatic siblings have a 50% risk of being a carrier and so they and their partners should consider carrier screening before starting a family. Screening newborn babies is also controversial.

ConclusionSpinal muscular atrophy is a rare disorder of infancy and childhood, the biology and pathophysiology of which have been extensively studied in the past decade. This scientifi c scrutiny has revolutionised our understanding of the disorder and has put an unprecedented focus on aff ected patients.

An immediate benefi t for patients has been the development and distribution of standard-of-care recommendations. Data for improved survival in the

most severely aff ected children with spinal muscular atrophy (type 1) are available. The long-term eff ect of the care guidelines on patients with type 2 and 3 spinal muscular atrophy is unclear, but some functional aspects—such as motor and respiratory function—are now reasonably stable over a 12 month period. A long-term benefi t for patients will be the development of eff ective interventions, some of which are now in clinical trials. The need for clinical trial readiness has been the impetus for remarkable cooperation between clinicians, scientists, industry, government, and volunteer organ-isations on an international scale. This collaboration has resulted in improved understanding of the barriers to clinical trial readiness and the identifi cation of possible strategies for overcoming them.

ContributorsThe authors contributed equally to the literature search and the writing

and formatting of the Review, and to critically reviewing the manuscript.

Confl icts of interestEM is a site principal investigator for the PTC Therapeutics (South

Plainfi eld, NJ, USA) extension study of ataluren in Duchenne muscular

dystrophy, for the TROPHOS (Marseille, France) clinical trial in spinal

muscular atrophy, and for a GlaxoSmithKline study of exon skipping. He

is also funded by Italian Telethon and SMA Europe for observational

studies of outcome measures. He has been on the advisory board for Shire

and PTC Therapeutics. EB is a site principal investigator for the PTC

extension study of ataluren in Duchenne muscular dystrophy, for the

TROPHOS clinical trial in spinal muscular atrophy, and for a

GlaxoSmithKline study of exon skipping. He is also funded by Italian

Telethon, the Italian Ministry of Health, and SMA Europe for observational

studies of outcome measures. STI is a site principal investigator for the

PTC extension study of ataluren in Duchenne muscular dystrophy and

receives funding from GlaxoSmithKline, ISIS (Carlsbad, CA, USA),

DuchEnne Muscular Dystrophy Long-term IdebenOne Study, and DART

(Great Barrington, MA, USA) for clinical studies of spinal muscular

atrophy and Duchenne muscular dystrophy. She has also received

expenses from these companies for attendance at investigator meetings.

She is co-principal investigator for the NeuroNEXT project funded by the

US National Institute of Neurological Disorders and Stroke, for which she

receives a salary. She is supported by the Muscular Dystrophy Association

for her neuromuscular clinics.

References1 Brzustowicz LM, Lehner T, Castilla LH, et al. Genetic mapping of

chronic childhood-onset spinal muscular atrophy to chromosome 5q11.2–13.3. Nature 1990; 344: 540–41.

2 Lefebvre S, Burglen L, Reboullet S, et al. Identifi cation and characterization of a spinal muscular atrophy-determining gene. Cell 1995; 80: 155–65.

3 Munsat TL, Davies KE. International SMA consortium meeting. (26–28 June 1992, Bonn, Germany). Neuromuscul Disord 1992; 2: 423–28.

Search strategy and selection criteria

We searched Medline, CINAHL, PsycINFO, and Embase for all publications containing the term “spinal muscular atrophy” from 1990, to February, 2012. We included all relevant reports published after 2000 but papers judged to be seminal by us were included irrespective of their publication date. Articles published before 1990 were identifi ed from PubMed and from our own fi les. Only papers published in English were reviewed.

450 www.thelancet.com/neurology Vol 11 May 2012

Review

4 Lorson CL, Rindt H, Shababi M. Spinal muscular atrophy: mechanisms and therapeutic strategies. Hum Mol Genet 2010; 19: R111–18.

5 Wang CH, Finkel RS, Bertini ES, et al. Consensus statement for standard of care in spinal muscular atrophy. J Child Neurol 2007; 22: 1027–49.

6 D’Amico A, Mercuri E, Tiziano FD, Bertini E. Spinal muscular atrophy. Orphanet J Rare Dis 2011; 6: 71.

7 Oskoui M, Levy G, Garland CJ, et al. The changing natural history of spinal muscular atrophy type 1. Neurology 2007; 69: 1931–36.

8 Boitano LJ. Equipment options for cough augmentation, ventilation, and noninvasive interfaces in neuromuscular respiratory management. Pediatrics 2009; 123 (suppl 4): S226–30.

9 Swoboda KJ, Prior TW, Scott CB, et al. Natural history of denervation in SMA: relation to age, SMN2 copy number, and function. Ann Neurol 2005; 57: 704–12.

10 Kaufmann P, McDermott MP, Darras BT, et al. Observational study of spinal muscular atrophy type 2 and 3: functional outcomes over 1 year. Arch Neurol 2011; 68: 779–86.

11 Lunn MR, Wang CH. Spinal muscular atrophy. Lancet 2008; 371: 2120–33.

12 Feldkotter M, Schwarzer V, Wirth R, Wienker TF, Wirth B. Quantitative analyses of SMN1 and SMN2 based on real-time lightCycler PCR: fast and highly reliable carrier testing and prediction of severity of spinal muscular atrophy. Am J Hum Genet 2002; 70: 358–68.

13 Arkblad EL, Darin N, Berg K, et al. Multiplex ligation-dependent probe amplifi cation improves diagnostics in spinal muscular atrophy. Neuromuscul Disord 2006; 16: 830–38.

14 Cusco I, Lopez E, Soler-Botija C, Jesus Barcelo M, Baiget M, Tizzano EF. A genetic and phenotypic analysis in Spanish spinal muscular atrophy patients with c.399_402del AGAG, the most frequently found subtle mutation in the SMN1 gene. Hum Mutat 2003; 22: 136–43.

15 Sugarman EA, Nagan N, Zhu H, et al. Pan-ethnic carrier screening and prenatal diagnosis for spinal muscular atrophy: clinical laboratory analysis of >72400 specimens. Eur J Hum Genet 2012; 20: 27–32.

16 Su YN, Hung CC, Lin SY, et al. Carrier screening for spinal muscular atrophy (SMA) in 107,611 pregnant women during the period 2005–2009: a prospective population-based cohort study. PLoS One 2011; 6: e17067.

17 Amara A, Adala L, Ben Charfeddine I, et al. Correlation of SMN2, NAIP, p44, H4F5 and Occludin genes copy number with spinal muscular atrophy phenotype in Tunisian patients. Eur J Paediatr Neurol 2012; 16: 167–74.

18 Tiziano FD, Bertini E, Messina S, et al. The Hammersmith functional score correlates with the SMN2 copy number: a multicentric study. Neuromuscul Disord 2007; 17: 400–03.

19 Schroth MK. Special considerations in the respiratory management of spinal muscular atrophy. Pediatrics 2009; 123 (suppl 4): S245–49.

20 Simonds AK. Respiratory support for the severely handicapped child with neuromuscular disease: ethics and practicality. Semin Respir Crit Care Med 2007; 28: 342–54.

21 Panitch HB. The pathophysiology of respiratory impairment in pediatric neuromuscular diseases. Pediatrics 2009; 123 (suppl 4): S215–18.

22 Roper H, Quinlivan R. Implementation of “the consensus statement for the standard of care in spinal muscular atrophy” when applied to infants with severe type 1 SMA in the UK. Arch Dis Child 2010; 95: 845–49.

23 Rudnik-Schoneborn S, Heller R, Berg C, et al. Congenital heart disease is a feature of severe infantile spinal muscular atrophy. J Med Genet 2008; 45: 635–38.

24 Shababi M, Habibi J, Yang HT, Vale SM, Sewell WA, Lorson CL. Cardiac defects contribute to the pathology of spinal muscular atrophy models. Hum Mol Genet 2010; 19: 4059–71.

25 Markstrom A, Cohen G, Katz-Salamon M. The eff ect of long term ventilatory support on hemodynamics in children with spinal muscle atrophy (SMA) type II. Sleep Med 2010; 11: 201–04.

26 Messina S, Pane M, De Rose P, et al. Feeding problems and malnutrition in spinal muscular atrophy type II. Neuromuscul Disord 2008; 18: 389–93.

27 Chen YS, Shih HH, Chen TH, Kuo CH, Jong YJ. Prevalence and risk factors for feeding and swallowing diffi culties in spinal muscular atrophy types II and III. J Pediatr 2012; 160: 447–451.

28 Sproule DM, Montes J, Dunaway S, et al. Adiposity is increased among high-functioning, non-ambulatory patients with spinal muscular atrophy. Neuromuscul Disord 2010; 20: 448–52.

29 Sproule DM, Montes J, Dunaway SL, et al. Bioelectrical impedance analysis can be a useful screen for excess adiposity in spinal muscular atrophy. J Child Neurol 2010; 25: 1348–54.

30 Sproule DM, Montes J, Montgomery M, et al. Increased fat mass and high incidence of overweight despite low body mass index in patients with spinal muscular atrophy. Neuromuscul Disord 2009; 19: 391–96.

31 Tangsrud SE, Carlsen KC, Lund-Petersen I, Carlsen KH. Lung function measurements in young children with spinal muscle atrophy; a cross sectional survey on the eff ect of position and bracing. Arch Dis Child 2001; 84: 521–24.

32 Granata C, Merlini L, Magni E, Marini ML, Stagni SB. Spinal muscular atrophy: natural history and orthopaedic treatment of scoliosis. Spine (Phila Pa 1976) 1989; 14: 760–62.

33 Fujak A, Ingenhorst A, Heuser K, Forst R, Forst J. Treatment of scoliosis in intermediate spinal muscular atrophy (SMA type II) in childhood. Ortop Traumatol Rehabil 2005; 7: 175–79.

34 Morillon S, Thumerelle C, Cuisset JM, Santos C, Matran R, Deschildre A. [Eff ect of thoracic bracing on lung function in children with neuromuscular disease]. Ann Readapt Med Phys 2007; 50: 645–50.

35 Chng SY, Wong YQ, Hui JH, Wong HK, Ong HT, Goh DY. Pulmonary function and scoliosis in children with spinal muscular atrophy types II and III. J Paediatr Child Health 2003; 39: 673–76.

36 Granata C, Cervellati S, Ballestrazzi A, Corbascio M, Merlini L. Spine surgery in spinal muscular atrophy: long-term results. Neuromuscul Disord 1993; 3: 207–15.

37 Robinson D, Galasko CS, Delaney C, Williamson JB, Barrie JL. Scoliosis and lung function in spinal muscular atrophy. Eur Spine J 1995; 4: 268–73.

38 Akbarnia BA, Boachie-Adjei O, Thompson AG, Asher MA. Dual growing rod technique for the treatment of progressive early-onset scoliosis: a multicenter study. Spine (Phila Pa 1976) 2005; 30 (17 suppl): S46–57.

39 Emans JB, Ordonez CL, Lee EY, Ciarlo M. The treatment of spine and chest wall deformities with fused ribs by expansion thoracostomy and insertion of vertical expandable prosthetic titanium rib: growth of thoracic spine and improvement of lung volumes. Spine (Phila Pa 1976) 2005; 30 (17 suppl): S58–68.

40 Hell AK, Campbell RM, Hefti F. The vertical expandable prosthetic titanium rib implant for the treatment of thoracic insuffi ciency syndrome associated with congenital and neuromuscular scoliosis in young children. J Pediatr Orthop B 2005; 14: 287–93.

41 Thompson GH. Growing rod techniques in early-onset scoliosis. J Pediatr Orthop 2007; 27: 354–61.

42 Chandran S, McCarthy J, Noonan K, Mann D, Nemeth B, Guiliani T. Early treatment of scoliosis with growing rods in children with severe spinal muscular atrophy: a preliminary report. J Pediatr Orthop 2011; 31: 450–54.

43 McElroy MJ, Shaner AC, Crawford TO, et al. Growing rods for scoliosis in spinal muscular atrophy: structural eff ects, complications, and hospital stays. Spine (Phila Pa 1976) 2011; 36: 1305–11.

44 Kinali M, Banks LM, Mercuri E, Manzur AY, Muntoni F. Bone mineral density in a paediatric spinal muscular atrophy population. Neuropediatrics 2004; 35: 325–28.

45 Khatri IA, Chaudhry US, Seikaly MG, Browne RH, Iannaccone ST. Low bone mineral density in spinal muscular atrophy. J Clin Neuromuscul Dis 2008; 10: 11–17.

46 Swoboda KJ, Scott CB, Reyna SP, et al. Phase II open label study of valproic acid in spinal muscular atrophy. PLoS One 2009; 4: e5268.

47 Shanmugarajan S, Tsuruga E, Swoboda KJ, Maria BL, Ries WL, Reddy SV. Bone loss in survival motor neuron (Smn(-/-) SMN2) genetic mouse model of spinal muscular atrophy. J Pathol 2009; 219: 52–60.

48 Rudnik-Schoneborn S, Hausmanowa-Petrusewicz I, Borkowska J, Zerres K. The predictive value of achieved motor milestones assessed in 441 patients with infantile spinal muscular atrophy types II and III. Eur Neurol 2001; 45: 174–81.

www.thelancet.com/neurology Vol 11 May 2012 451

Review

49 Russman BS, Buncher CR, White M, Samaha FJ, Iannaccone ST. Function changes in spinal muscular atrophy II and III. The DCN/SMA Group. Neurology 1996; 47: 973–76.

50 Crowley R, Wolfe I, Lock K, McKee M. Improving the transition between paediatric and adult healthcare: a systematic review. Arch Dis Child 2011; 96: 548–53.

51 Rudnik-Schoneborn S, Zerres K, Ignatius J, Rietschel M. Pregnancy and spinal muscular atrophy. J Neurol 1992; 239: 26–30.

52 Mercuri E, Bertini E, Messina S, et al. Pilot trial of phenylbutyrate in spinal muscular atrophy. Neuromuscul Disord 2004; 14: 130–35.

53 Mercuri E, Bertini E, Messina S, et al. Randomized, double-blind, placebo-controlled trial of phenylbutyrate in spinal muscular atrophy. Neurology 2007; 68: 51–55.

54 Pane M, Staccioli S, Messina S, et al. Daily salbutamol in young patients with SMA type II. Neuromuscul Disord 2008; 18: 536–40.

55 Tiziano FD, Lomastro R, Pinto AM, et al. Salbutamol increases survival motor neuron (SMN) transcript levels in leucocytes of spinal muscular atrophy (SMA) patients: relevance for clinical trial design. J Med Genet; 47: 856–58.

56 Kissel JT, Scott CB, Reyna SP, et al. SMA CARNIVAL Trial Part II: a prospective, single-armed trial of L-carnitine and valproic acid in ambulatory children with spinal muscular atrophy. PLoS One; 6: e21296.

57 Miller RG, Moore DH, Dronsky V, et al. A placebo-controlled trial of gabapentin in spinal muscular atrophy. J Neurol Sci 2001; 191: 127–31.

58 Swoboda KJ, Scott CB, Crawford TO, et al. SMA CARNIVAL trial part I: double-blind, randomized, placebo-controlled trial of L-carnitine and valproic acid in spinal muscular atrophy. PLoS One 2010; 5: e12140.

59 Chen TH, Chang JG, Yang YH, et al. Randomized, double-blind, placebo-controlled trial of hydroxyurea in spinal muscular atrophy. Neurology 2010; 75: 2190–97.

60 Kinali M, Mercuri E, Main M, et al. Pilot trial of albuterol in spinal muscular atrophy. Neurology 2002; 59: 609–10.

61 Merlini L, Solari A, Vita G, et al. Role of gabapentin in spinal muscular atrophy: results of a multicenter, randomized Italian study. J Child Neurol 2003; 18: 537–41.

62 Russman BS, Iannaccone ST, Samaha FJ. A phase 1 trial of riluzole in spinal muscular atrophy. Arch Neurol 2003; 60: 1601–03.

63 Bertini E, Burghes A, Bushby K, et al. 134th ENMC International Workshop: Outcome Measures and Treatment of Spinal Muscular Atrophy, 11–13 February 2005, Naarden, The Netherlands. Neuromuscul Disord 2005; 15: 802–16.

64 Lewelt A, Newcomb TM, Swoboda KJ. New therapeutic approaches to spinal muscular atrophy. Curr Neurol Neurosci Rep 2012; 12: 42–53.

65 Pruss RM, Giraudon-Paoli M, Morozova S, Berna P, Abitbol JL, Bordet T. Drug discovery and development for spinal muscular atrophy: lessons from screening approaches and future challenges for clinical development. Future Med Chem 2010; 2: 1429–40.

66 Shababi M, Mattis VB, Lorson CL. Therapeutics that directly increase SMN expression to treat spinal muscular atrophy. Drug News Perspect 2010; 23: 475–82.

67 Mercuri E, Mayhew A, Muntoni F, et al. Towards harmonisation of outcome measures for DMD and SMA within TREAT-NMD; report of three expert workshops: TREAT-NMD/ENMC workshop on outcome measures, 12th–13th May 2007, Naarden, The Netherlands; TREAT-NMD workshop on outcome measures in experimental trials for DMD, 30th June–1st July 2007, Naarden, The Netherlands; conjoint Institute of Myology TREAT-NMD meeting on physical activity monitoring in neuromuscular disorders, 11th July 2007, Paris, France. Neuromuscul Disord 2008; 18: 894–903.

68 Hirtz D, Iannaccone S, Heemskerk J, Gwinn-Hardy K, Moxley 3rd R, Rowland LP. Challenges and opportunities in clinical trials for spinal muscular atrophy. Neurology 2005; 65: 1352–57.

69 Swoboda KJ, Kissel JT, Crawford TO, et al. Perspectives on clinical trials in spinal muscular atrophy. J Child Neurol 2007; 22: 957–66.

70 Kaufmann P, Iannaccone ST. Clinical trials in spinal muscular atrophy. Phys Med Rehabil Clin N Am 2008; 19: 653–60.

71 Kaufmann P, McDermott MP, Darras BT, et al. Observational study of spinal muscular atrophy type 2 and 3: functional outcomes over 1 year. Arch Neurol 2011; 68: 779–86.

72 Mercuri E, Messina S, Battini R, et al. Reliability of the Hammersmith functional motor scale for spinal muscular atrophy in a multicentric study. Neuromuscul Disord 2006; 16: 93–98.

73 Merlini L, Bertini E, Minetti C, et al. Motor function-muscle strength relationship in spinal muscular atrophy. Muscle Nerve 2004; 29: 548–52.

74 Berard C, Fermanian J, Payan C. Outcome measure for SMA II and III patients. Neuromuscul Disord 2008; 18: 593–94.

75 Berard C, Payan C, Hodgkinson I, Fermanian J. A motor function measure for neuromuscular diseases. Construction and validation study. Neuromuscul Disord 2005; 15: 463–70.

76 Nelson L, Owens H, Hynan LS, Iannaccone ST. The gross motor function measure is a valid and sensitive outcome measure for spinal muscular atrophy. Neuromuscul Disord 2006; 16: 374–80.

77 Glanzman AM, McDermott MP, Montes J, et al. Validation of the Children’s Hospital of Philadelphia Infant Test of Neuromuscular Disorders (CHOP INTEND). Pediatr Phys Ther 2011; 23: 322–26.

78 Finkel RS, Hynan LS, Glanzman AM, et al. The test of infant motor performance: reliability in spinal muscular atrophy type I. Pediatr Phys Ther 2008; 20: 242–46.

79 Main M, Kairon H, Mercuri E, Muntoni F. The Hammersmith functional motor scale for children with spinal muscular atrophy: a scale to test ability and monitor progress in children with limited ambulation. Eur J Paediatr Neurol 2003; 7: 155–59.

80 Mazzone E, Bianco F, Martinelli D, et al. Assessing upper limb function in nonambulant SMA patients: development of a new module. Neuromuscul Disord 2011; 21: 406–12.

81 O’Hagen JM, Glanzman AM, McDermott MP, et al. An expanded version of the Hammersmith Functional Motor Scale for SMA II and III patients. Neuromuscul Disord 2007; 17: 693–97.

82 Glanzman AM, O’Hagen JM, McDermott MP, et al. Validation of the Expanded Hammersmith Functional Motor Scale in spinal muscular atrophy type II and III. J Child Neurol 2011; 26: 1499–507.

83 Krosschell KJ, Maczulski JA, Crawford TO, Scott C, Swoboda KJ. A modifi ed Hammersmith functional motor scale for use in multi-center research on spinal muscular atrophy. Neuromuscul Disord 2006; 16: 417–26.

84 Krosschell KJ, Scott CB, Maczulski JA, Lewelt AJ, Reyna SP, Swoboda KJ. Reliability of the Modifi ed Hammersmith Functional Motor Scale in young children with spinal muscular atrophy. Muscle Nerve 2011; 44: 246–51.

85 Montes J, McDermott MP, Martens WB, et al. Six-minute walk test demonstrates motor fatigue in spinal muscular atrophy. Neurology 2010; 74: 833–38.

86 Steff ensen BF, Lyager S, Werge B, Rahbek J, Mattsson E. Physical capacity in non-ambulatory people with Duchenne muscular dystrophy or spinal muscular atrophy: a longitudinal study. Dev Med Child Neurol 2002; 44: 623–32.

87 Iannaccone ST, Hynan LS, Morton A, Buchanan R, Limbers CA, Varni JW. The PedsQL in pediatric patients with spinal muscular atrophy: feasibility, reliability, and validity of the Pediatric Quality of Life Inventory Generic Core Scales and Neuromuscular Module. Neuromuscul Disord 2009; 19: 805–12.

88 Rudnik-Schoneborn S, Berg C, Zerres K, et al. Genotype-phenotype studies in infantile spinal muscular atrophy (SMA) type I in Germany: implications for clinical trials and genetic counselling. Clin Genet 2009; 76: 168–78.

89 Iannaccone ST, Hynan LS. Reliability of 4 outcome measures in pediatric spinal muscular atrophy. Arch Neurol 2003; 60: 1130–36.

90 Glanzman AM, Mazzone E, Main M, et al. The Children’s Hospital of Philadelphia Infant Test of Neuromuscular Disorders (CHOP INTEND): test development and reliability. Neuromuscul Disord; 20: 155–61.

91 Krosschell KJ, Scott CB, Maczulski JA, Lewelt AJ, Reyna SP, Swoboda KJ. Reliability of the Modifi ed Hammersmith Functional Motor Scale in young children with spinal muscular atrophy. Muscle Nerve 2011; 44: 246–51.

92 Bromberg MB, Swoboda KJ. Motor unit number estimation in infants and children with spinal muscular atrophy. Muscle Nerve 2002; 25: 445–47.

93 Bromberg MB, Swoboda KJ, Lawson VH. Counting motor units in chronic motor neuropathies. Exp Neurol 2003; 184 (suppl 1): S53–57.

452 www.thelancet.com/neurology Vol 11 May 2012

Review

94 Lewelt A, Krosschell KJ, Scott C, et al. Compound muscle action potential and motor function in children with spinal muscular atrophy. Muscle Nerve 2010; 42: 703–08.

95 Mayhew A, Cano S, Scott E, Eagle M, Bushby K, Muntoni F. Moving towards meaningful measurement: Rasch analysis of the North Star Ambulatory Assessment in Duchenne muscular dystrophy. Dev Med Child Neurol 2011; 53: 535–42.

96 Simard LR, Belanger MC, Morissette S, Wride M, Prior TW, Swoboda KJ. Preclinical validation of a multiplex real-time assay to quantify SMN mRNA in patients with SMA. Neurology 2007; 68: 451–56.

97 Tiziano FD, Neri G, Brahe C. Biomarkers in rare disorders: the experience with spinal muscular atrophy. Int J Mol Sci 2010; 12: 24–38.

98 Tiziano FD, Pinto AM, Fiori S, et al. SMN transcript levels in leukocytes of SMA patients determined by absolute real-time PCR. Eur J Hum Genet 2010; 18: 52–58.

99 Liss J, Bruszczynska A, Lukaszuk K. [Preimplantation genetic diagnosis in prevention of genetic diseases—diagnostic of spinal muscular atrophy (SMA)]. Ginekol Pol 2010; 81: 918–21 (in Polish).

100 Howard HC, Avard D, Borry P. Are the kids really alright? Direct-to-consumer genetic testing in children: are company policies clashing with professional norms? Eur J Hum Genet 2011; 19: 1122–16.

Comment

www.thelancet.com/neurology Vol 11 May 2012 383

treatment would be expected to result in about one fewer treated patient being dead or dependent at hospital discharge. In reality, no diff erence was evident in the clinical outcomes at 7 days in the trial (21/53 dead or dependent in the MSU group vs 20/47 in the control group). Moreover, of those patients with ischaemic stroke in the MSU group who were treated with thrombolysis, three died over the next few days whereas there were no deaths in patients with ischaemic stroke in the control group. Although these numbers are too small to allow reliable interpretation, they highlight the need for larger trials with suffi cient power to establish the eff ect of the MSU intervention on clinical outcome.

An important intermediate step in showing the potential for improved clinical outcomes with an MSU service would be to study the CT brain scans from patients scanned early by the MSU and to consider the “tissue window”, estimated with the Alberta Stroke Program Early CT (ASPECT) score,2 as well as the “time window”. If the patients scanned early had high ASPECT scores, suggesting a signifi cant potential to benefi t from thrombolysis, then that would provide some reassurance that better clinical outcomes could be expected.

The generalisability of the trial fi ndings to potential MSU services elsewhere will depend very much on the setting. This trial was set in an urban area with a median distance from the patient to the hospital of 7 km and median alarm to arrival times of 8 min for the standard ambulance versus 12 min for the MSU. The MSU would potentially work less well in rural areas in which locally based ambulances might be able to get patients to hospital about as quickly as a hospital-situated MSU could get out to the patient. Nevertheless, this trial has shown convincingly that in at least some settings an MSU-based service is feasible and can substantially reduce delays to treatment.

*Peter M Rothwell, Alastair M BuchanNuffi eld Department of Clinical Neuroscience (PMR) and Nuffi eld Department of Medicine (AMB), Oxford University, John Radcliff e Hospital, Headington, Oxford, UKpeter.rothwell@clneuro.ox.ac.uk

We declare that we have no confl icts of interest.

1 Walter S, Kostopoulos P, Haass A, et al. Diagnosis and treatment of patients with stroke in a mobile stroke unit versus in hospital: a randomised controlled trial. Lancet Neurol 2012; published online April 11. DOI:10.1016/S1474-4422(12)70057-1.

2 Barber PA, Demchuk AM, Zhang J, Buchan AM. Validity and reliability of a quantitative computed tomography score in predicting outcome of hyperacute stroke before thrombolytic therapy. ASPECTS Study Group. Alberta Stroke Programme Early CT Score. Lancet 2000; 355: 1670–74.

Defi ning and refi ning 5-HT receptor targets for migraineThe triptans are well established as effective treatments for migraine, but the specific mechanisms responsible for their therapeutic effects remain unknown. Triptans selectively activate 5-HT1 receptors, in particular the 1B, 1D, and 1F receptor subtypes.1 There has been extensive discussion about which of these 5-HT1 receptor subtypes are responsible for the therapeutic and adverse effects of treatment with triptans and about the locations in and around the brain where these effects might occur. In this issue of The Lancet Neurology, Färkkilä and colleagues2 describe a clinical trial of an oral formulation of lasmiditan—a non-triptan selective 5-HT1F receptor agonist—as an acute treatment for migraine. This study, in combination with previous studies of the clinical pharmacology and efficacy of lasmiditan in migraine,3,4 confirms the potential of lasmiditan as a treatment for acute migraine and also provides new

insights into the mechanisms of 5-HT1 receptors in migraine.

Lasmiditan is chemically distinct from triptans and has different pharmacological properties. Unlike triptans, lasmiditan has higher selectivity for the 5-HT1F receptor than for the 5-HT1B and 5-HT1D receptors.4 Lasmiditan was active in two in-vivo models that have been used to investigate migraine—plasma protein extravasation evoked by stimulation of the trigeminal ganglion and activation of c-fos in the trigeminal nucleus caudalis evoked by similar stimulation.4 Although these assays do not predict efficacy of migraine treatment, they do suggest functional effects in vivo, which in the case of lasmiditan are probably mediated by the 5-HT1F receptor.4 A previous study of intravenous lasmiditan suggested that it was an effective treatment for migraine.3

Published OnlineMarch 28, 2012 DOI:10.1016/S1474-4422(12)70062-5

See Articles page 405

Comment

384 www.thelancet.com/neurology Vol 11 May 2012

The present study investigated multiple doses of an oral preparation (50, 100, 200, and 400 mg) of lasmiditan versus placebo as a treatment for acute migraine in 391 patients. The results show dose-dependent improvements in the primary endpoint of headache response at 2 h after lasmiditan treatment, which are comparable to results that have been reported for several different triptans.5 For the harder to achieve secondary endpoint of proportion of patients who are pain free at 2 h, the results were somewhat less impressive, with only the 200 mg and 400 mg doses showing statistically significant superiority versus placebo. There was also a dose-dependent incidence of side-effects with lasmiditan, with dizziness, vertigo, and fatigue the most common. Few patients reported chest, neck, or jaw tightness or heaviness, which are commonly reported with triptans.

These results have several interesting and important implications. First, they confi rm that selective activation of 5-HT1F receptors has therapeutic eff ects in patients with migrane. 5-HT1F receptors are not widely expressed in the vasculature, and activation of these receptors does not have vascular eff ects in vitro.4 Although the therapeutic eff ects of triptans and calcitonin gene-related peptide antagonists have, for some time, been hypothesised to be independent of their eff ects on the vasculature, this hypothesis has been diffi cult to prove. Other eff ective treatments for acute migraine, such as aspirin or non-steroidal anti-infl ammatory drugs, do not seem to work via vascular mechanisms; however, because these drugs have analgesic eff ects that are not specifi c for migraine, they might be working downstream from primary migraine mechanisms. The eff ects of lasmiditan in the study by Färkkilä and colleagues2 provide strong and direct evidence for a non-vascular mechanism for migraine-specifi c acute treatment.

An additional important implication of the study is that the sites of the anti-migraine eff ects of lasmiditan are central rather than peripheral. Because 5-HT1F receptors are expressed in the trigeminal ganglion, a peripheral site of action cannot be excluded.6 However, receptor subtypes 1B, 1D, and 1F are all expressed in the brain and have anti-nociceptive functions at multiple central sites that might play a part in migraine.1 CNS sites of action have been proposed for most treatments

for acute migraine, but with the triptans a central site of action has been controversial because they have limited ability to cross the blood-brain barrier.7 The CNS symptoms caused by lasmiditan provide strong evidence that lasmiditan is indeed reaching the brain and suggest that it is treating migraine within the brain.

Whether or not lasmiditan will be able to compete with triptans as an eff ective and well-tolerated migraine treatment remains an open question. Although triptans are generally safe and the potential for adverse cardiac events is low for most patients,8 there is a subset of patients with cardiac issues for whom triptans are contraindicated and for whom lasmiditan might represent a suitable alternative. For patients without cardiac issues, one interesting question is whether 5-HT1F receptor activation alone is as eff ective as activation of 1B, 1D, and 1F subtypes together. Regardless of the answer to these questions, further basic and clinical studies with lasmiditan will likely provide additional important information about where and how acute migraine treatments are working.

Andrew CharlesHeadache Research and Treatment Program, Department of Neurology, David Geff en School of Medicine at UCLA, Los Angeles, CA 90095, USAacharles@ucla.edu

AC has served as a consultant for AGA Medical, Amgen, Bristol Myers Squibb, Eli Lilly, eNeura, MAP Pharmaceutical, Monosol Rx, Pfi zer, and Zogenix, and has received research grant support from MAP Pharmaceutical.

1 Goadsby PJ. Serotonin receptor ligands: treatments of acute migraine and cluster headache. Handb Exp Pharmacol 2007; 177: 129–43.

2 Färkkilä M, Diener HC, Géraud G, et al. Effi cacy and tolerability of lasmiditan, an oral 5-HT1F receptor agonist, for the acute treatment of migraine: a phase 2 randomised, placebo-controlled, parallel-group, dose-ranging study. Lancet Neurol 2012; published online March 28. DOI:10.1016/S1474-4422(12)70047-9

3 Ferrari MD, Farkkila M, Reuter U, et al. Acute treatment of migraine with the selective 5-HT1F receptor agonist lasmiditan—a randomised proof-of-concept trial. Cephalalgia 2010; 30: 1170–78.

4 Nelson DL, Phebus LA, Johnson KW, et al. Preclinical pharmacological profi le of the selective 5-HT1F receptor agonist lasmiditan. Cephalalgia 2010; 30: 1159–69.

5 Ferrari MD, Goadsby PJ, Roon KI, Lipton RB. Triptans (serotonin, 5-HT1B/1D agonists) in migraine: detailed results and methods of a meta-analysis of 53 trials. Cephalalgia 2002; 22: 633–58.

6 Classey JD, Bartsch T, Goadsby PJ. Distribution of 5-HT(1B), 5-HT(1D) and 5-HT(1F) receptor expression in rat trigeminal and dorsal root ganglia neurons: relevance to the selective anti-migraine eff ect of triptans. Brain Res 2010; 1361: 76–85.

7 Edvinsson L, Tfelt-Hansen P. The blood brain barrier in migraine treatment. Cephalalgia 2008; 28: 1245–58.

8 Dodick D, Lipton RB, Martin V, et al. Consensus statement: cardiovascular safety profi le of triptans (5-HT agonists) in the acute treatment of migraine. Headache 2004; 44: 414–25.

Comment

386 www.thelancet.com/neurology Vol 11 May 2012

refl ex, no better than extensor (decerebrate) posturing, and oxygen index. Each of these factors was assigned a score (1 point each for absent corneal refl ex, extensor or absent motor response, and oxygenation index of >3·0, and 2 points for absent cough refl ex) and the aggregate score was related directly to the likelihood of death within 1 h of WLST. For patients who had the highest score (5 points), the probability of cardiac arrest within 1 h of WLST was 0·87. However, in patients with scores of less than 5 points, the probability of death within 1 h of WLST varied with the combination of variables (eg, from 0·45 to 0·61 in patients with an aggregate score of 3 points). Thus, for aggregate scores of less than 5 points, specifi c factors need to be taken into consideration and not just the aggregate score.

Although the population in Rabinstein and colleagues’ study3 was small and heterogenous and their scoring system was not compared with the established Wisconsin criteria,4 their analysis provides an encouraging step in refi ning the prediction of cardiac arrest within 1–2 h of WLST and thus selection of patients for DCD. Further refi nements and validation studies are needed. In the meantime, reliance on the Wisconsin scoring system4 or that developed by Rabinstein and colleagues3 would be imprudent. Selection of the best potential DCD donors

will need good clinical judgment and incorporation and analysis of both scoring systems.

G Bryan Young, Michael D SharpeDivision of Critical Care, Western University, Room B10–106, University Hospital, 339 Windermere Road, London, Ontario, Canada N6A5A5bryan.young@lhsc.on.ca

We declare that we have no confl icts of interest.

1 Bernat JL. Contemporary controversies in the defi nition of death. Prog Brain Res 2009; 177: 21–31.

2 Hernandez-Alejandro R, Wall W, Jevnikar A, et al. Organ donation after cardiac death: donor and recipient outcomes after the fi rst three years of the Ontario experience. Can J Anaesth 2011; 58: 599–605.

3 Rabinstein AA, Yee AH, Mandrekar J, et al. Prediction of potential for organ donation after cardiac death in patients in neurocritical state: a prospective observational study. Lancet Neurol 2012; published online April 5. DOI:10.1016/S1474-4422(12)70060-1.

4 Lewis J, Peltier J, Nelson H, et al. Development of the University of Wisconsin donation after cardiac death evaluation tool. Prog Transplant 2003; 13: 265–73.

5 Wind J, Snoeijs MGJ, Brugman CA et al. Prediction of time of death after withdrawal of life-sustaining treatment in potential donors after cardiac death. Crit Care Med 2012; 40: 766–79.

6 Dharamsi S, Ho A, Spadafora SM, et al. The physician as health advocate: translating the quest for social responsibility into medical education and practice. Acad Med 2011; 86: 1108–13.

7 Yee AH, Rabinstein AA, Thapa P, Mandrekar J, Wijdicks EFM. Factors infl uencing time to death after withdrawal of life support in neurocritical patients. Neurology 2010; 74: 1380–85.

8 Wijdicks EF, Bamlet WR, Maramattom BV, et al. Validation of a new coma scale: the FOUR score. Ann Neurol 2005; 58: 585–93.

Published OnlineApril 10, 2012

DOI:10.1016/S1474-4422(12)70066-2

See Articles page 420

MS clinical trials: what can subgroup analyses teach us? Since the fi rst modern-day clinical trial in patients with multiple sclerosis (MS),1 the MS therapeutics speciality has matured. Eight disease-modifying therapies (DMTs) have been approved for relapsing forms of MS since 1993, and more are on the way. The clinical science that underlies trial design and analysis is also maturing. Clinical investigators and study sponsors have learned to acquire and analyse MRI data in multicentre studies, use time-to-disability survival analysis, in corporate patient-reported outcomes, and undertake exploratory studies of new clinical, imaging, and biomarker methods for monitoring treatment eff ects. The number of patients entering into clinical trials has increased over time, in part because recent trials have included patients with low MS disease activity. The fi rst placebo-controlled study1 to investigate a treatment for MS enrolled less than 100 participants per group and the trial that led to

regulatory approval for interferon beta-1b2 enrolled about 125 patients per group, whereas more recent placebo-controlled trials of natalizumab,3 cladribine,4 and fi ngolimod5 enrolled over 400 patients per group. Larger sample sizes provide greater opportunity for secondary analyses, including subgroup comparisons.

In that regard, in The Lancet Neurology, Devonshire and colleagues6 report subgroup analyses from the phase 3 FREEDOMS (FTY720 Research Evaluating Effects of Daily Oral therapy in MS) placebo-controlled trial of fingolimod for relapsing-remitting MS (RRMS).5 In the original trial, 1272 patients with RRMS were randomly assigned to placebo (n=418), fingolimod 0·5 mg/day (n=425), or fingolimod 1·25 mg/day (n=429). The approved dose of fingolimod, 0·5 mg/day, was associated with a 54% reduction in annualised relapse rate (ARR) and a 30% reduction in worsening of expanded disability

Comment

www.thelancet.com/neurology Vol 11 May 2012 387

status scale (EDSS) scores confirmed over 3 months. Subgroup analyses were done to assess the effects of fingolimod on ARR and confirmed EDSS worsening. Most of the subgroups were prespecified (sex, age, previous DMT treatment, pre-study relapses, baseline EDSS, and gadolinium-enhancing MRI activity at baseline). Other subgroups were requested by the European Medicines Agency (EMA; included groups based on disease activity in patients previously treated with DMTs).

The results from the subgroup analyses suggest that fi ngolimod 0·5 mg/day has consistent eff ects on ARR and EDSS in the specifi ed subgroups. One interesting fi nding was that the treatment eff ect of fi ngolimod 0·5 mg/day on ARR was attenuated in patients aged over 40 years. However, fi ngolimod 1·25 mg/day showed signifi cant benefi ts on ARR in patients in both age groups (>40 years or ≤40 years), which suggests that the approved dose might be inadequate to achieve benefi t on ARR in older patients. Other signifi cant fi ndings were that fi ngolimod 0·5 mg was eff ective in patients with breakthrough disease despite use of DMTs and for patients with highly active disease. These fi ndings are reassuring, since fi ngolimod is approved for patients with disease activity despite interferon beta or glatiramer acetate treatment and in patients with highly active disease.

Subgroup analyses have also been reported for placebo-controlled studies of natalizumab7 and cladribine.8 In the natalizumab trial,7 active treatment was associated with a relapse benefi t in most subgroups, including patients aged over 40 years. However, there was no treatment benefi t for older patients when confi rmed EDSS was used as the outcome. The cladribine study8 focused on “sustained freedom from disease activity”. For all subgroups tested, cladribine increased the proportion of patients who were disease-activity free throughout the study. As with fi ngolimod and natalizumab, the benefi t of cladribine on this outcome was also attenuated in patients over age 40 years. Taken together, fi ndings from the subgroup analyses from the placebo-controlled studies of fi ngolimod, cladribine, and natalizumab raise the possibility that the therapeutic eff ects of anti-infl ammatory DMTs might be lower in older patients.

There are well-known limitations of subgroup analyses.9,10 The most substantial limitation is that

the parent study is rarely designed or powered to answer questions about drug eff ects in subgroups. Consequently, the number of patients might be inadequate to answer the question, resulting in misleading false-negative results. Large sample sizes can to some degree mitigate this problem, particularly with highly eff ective treatments. There might also be signifi cant imbalances within the subgroups, particularly for subgroups with small numbers, which can lead to misleading false-positive results. Important precautions include prespecifi cation and accurate defi nition of subgroups, avoidance of over-reliance on p values, and inclusion of tests for interactions. Interaction testing can be used to assess directly whether treatment diff erence in an outcome depends on the patient’s subgroup. All of these precautions were included in the fi ngolimod subgroup report. However, even with these precautions, results from subgroup analyses must be viewed as exploratory and hypothesis generating.

Despite the caveats, subgroup analyses can be useful in assessing the consistency of treatment across clinically relevant patient groups. The results of the study by Devonshire and colleagues6 support the use of fingolimod in patients with disease activity despite treatment with first-line DMTs and suggest that treatment effects occur in most clinically relevant groups. Unanticipated findings are useful in generating hypotheses that need further testing: in the present study, the results provide a rationale for prospective dose-finding studies of fingolimod in patients aged over 40 years. They also support the emerging notion that pathogenic mechanisms and therapeutic responses change with MS disease evolution.

Richard A RudickMellen Center for Multiple Sclerosis Treatment and Research, Department of Neurology, Neurological Institute, Cleveland Clinic, Cleveland, OH, USArudickr@ccf.org

RAR has acted as a consultant for Biogen Idec, Bayhill Therapeutics, Pfi zer, Genzyme, and Novartis.

1 Rose AS, Kuzma JW, Kurtzke JF, Namerow NS, Sibley WA, Tourtellotte WW. Cooperative study in the evaluation of therapy in multiple sclerosis: ACTH vs placebo. Final Report. Neurology 1970; 20: 1–59.

2 The IFNB multiple sclerosis study group and the University of British Columbia MS/MRI analysis group. Interferon beta-1b in the treatment of multiple sclerosis: fi nal outcome of the randomized controlled trial. Neurology 1995; 45: 1277–85.

Comment

388 www.thelancet.com/neurology Vol 11 May 2012

3 Polman CH, O’Connor PW, Havrdova E, et al. A randomized, placebo-controlled trial of natalizumab for relapsing multiple sclerosis. N Engl J Med 2006; 354: 899–910.

4 Giovannoni G, Comi G, Cook S, et al. A placebo-controlled trial of oral cladribine for relapsing multiple sclerosis. N Engl J Med 2010; 362: 416–26.

5 Kappos L, Radue EW, O’Connor P, et al. A placebo-controlled trial of oral fi ngolimod in relapsing multiple sclerosis. N Engl J Med 2010; 362: 387–401.

6 Devonshire V, Havrdova E, Radu EW, et al. Relapse and disability outcomes in patients with multiple sclerosis treated with fi ngolimod: subgroup analyses of the double-blind, randomised, placebo-controlled FREEDOMS study. Lancet Neurol 2012; published online April 10. DOI:10.1016/S1474-4422(12)70056-X.

7 Hutchinson M, Kappos L, Calabresi PA, et al. The effi cacy of natalizumab in patients with relapsing multiple sclerosis: subgroup analyses of AFFIRM and SENTINEL. J Neurol 2009; 256: 405–15.

8 Giovannoni G, Cook S, Rammohan K, et al. Sustained disease-activity-free status in patients with relapsing-remitting multiple sclerosis treated with cladribine tablets in the CLARITY study: a post-hoc and subgroup analysis. Lancet Neurol 2011; 10: 329–37.

9 Pocock SJ, Assmann SE, Enos LE, Kasten LE. Subgroup analysis, covariate adjustment and baseline comparisons in clinical trial reporting: current practice and problems. Stat Med 2002; 21: 2917–30.

10 Hernandez AV, Boersma E, Murray GD, Habbema JD, Steyerberg EW. Subgroup analyses in therapeutic cardiovascular clinical trials: are most of them misleading? Am Heart J 2006; 151: 257–64.

ErratumByrne S, Elamin M, Bede P, et al. Cognitive and clinical characteristics of patients with amyotrophic lateral sclerosis carrying a C9orf72 repeat expansion: a population-based cohort study. Lancet Neurol 2012; 11: 232–40—The grant number in the fi rst sentence of the acknowledgments section should have read 259867. This correction has been made to the online version as of April 18, 2012.

Correspondence

390 www.thelancet.com/neurology Vol 11 May 2012

London, UK (AJS); Department of Neurosurgery, University of Pittsburgh, Pittsburgh, PA, USA (DOO); Department of Neurological Surgery, University of Miami, Miami, FL, USA (MRB)

1 Hartings JA, Bullock MR, Okonkwo DO, et al. Spreading depolarisations and outcome after traumatic brain injury: a prospective observational study. Lancet Neurol 2011; 10: 1058–64.

2 Hartings JA, Strong AJ, Fabricius M, et al. Spreading depolarizations and late secondary insults after traumatic brain injury. J Neurotrauma 2009; 26: 1857–66.

3 Dreier JP, Major S, Pannek HW, et al. Spreading convulsions, spreading depolarization and epileptogenesis in human cerebral cortex. Brain 2011; 135: 259–75.

4 Dreier JP, Major S, Manning A, et al. Cortical spreading ischaemia is a novel process involved in ischaemic damage in patients with aneurysmal subarachnoid haemorrhage. Brain 2009; 132: 1866–81.

5 Lauritzen M, Hansen AJ. The eff ect of glutamate receptor blockade on anoxic depolarization and cortical spreading depression. J Cereb Blood Flow Metab 1992; 12: 223–29.

6 Morris GF, Bullock R, Marshall SB, Marmarou A, Maas A, Marshall LF. Failure of the competitive N-methyl-D-aspartate antagonist Selfotel (CGS 19755) in the treatment of severe head injury: results of two phase III clinical trials. The Selfotel Investigators. J Neurosurg 1999; 91: 737–43.

7 Yurkewicz L, Weaver J, Bullock MR, Marshall LF. The eff ect of the selective NMDA receptor antagonist traxoprodil in the treatment of traumatic brain injury. J Neurotrauma 2005; 22: 1428–43.

8 Lo EH. A new penumbra: transitioning from injury into repair after stroke. Nat Med 2008; 14: 497–500.

9 Ikonomidou C, Turski L. Why did NMDA receptor antagonists fail clinical trials for stroke and traumatic brain injury? Lancet Neurol 2002; 1: 383–86.

blocking spreading depolarisations, it would occur only in the 56% of patients with depolarisations, or perhaps just the 20% with ISD. Thus any eff ect would be greatly diluted by broad inclusion, even within the surgical subgroup. Second, it is questionable whether suffi cient brain levels of Selfotel, a competitive antagonist, were achieved to compete with high glutamate concentrations (100–200 times more than normal) for receptor binding; there was no confi rmation of intended eff ect on the target mechanism. In a subsequent severe TBI trial with the non-competitive NR2B-selective antagon-ist traxoprodil, improvements in outcome up to 11·8% were reported in some subgroups.7 Third, the Selfotel trials were stopped in part for safety concerns in a concomitant stroke trial, so that suffi cient statistical power was never achieved.

We envision electrocorticography as another method in the arma-mentarium of neuromonitoring. Just as mannitol is given only to patients with elevated intracranial pressure, a therapy to block spreading depolarisations would be administered only to patients that have them and for as long as they persist. This will confi rm that the pathological processes targeted by therapeutics are active in individual patients. The advantages of this approach are selective inclusion, mechanistic targeting, and, not least, assessment of the eff ect of therapy on the targeted mechanism. Such a design diff ers dramatically from the failed trials in the past, but rather builds towards personalised treatment of this heterogeneous disease. More work is needed to determine which depolarisations should be targeted, which therapies are most eff ective, and to ensure that interventions do not interfere with repair processes.8,9 We declare that we have no confl icts of interest.

Jed A Hartings, Anthony J Strong, David O Okonkwo, M Ross Bullock Department of Neurosurgery, University of Cincinnati, Cincinnati, OH, USA (JAH); Department of Clinical Neuroscience, King’s College London,

after TBI. Depolarisations occurring in electrically silent cortex (or isoelectric spreading depolarisations [ISD]) accounted for 36% of all events and showed a similar time course to depolarisations with depression of spontaneous cortical activity (cortical spreading depression): a maximal activity in the initial 48 h, followed by a decline and then a delayed secondary peak on days 6–8.2,3 Thus, the association of ISD with poor outcome is not because ISD occur only as preterminal events associated with global brain death. The predominance of ISD within the initial days rather suggests that, as in ischaemic stroke, ISD are associated with acute lesion development. The association with mortality might be both correlative and causative, since isoelectricity could result from ischaemia or impaired neurovascular coupling,4 and depolarisations cause cell death under these conditions, probably through Ca2+-mediated pathways.

Assuming that NMDA-receptor antagonism inhibits all depolarisations, which is unlikely,5 does the failure of the Selfotel trials6 provide evidence against depolarisations as a promising therapeutic target? We don’t think so. First, if there was a benefi cial eff ect from

Depo

laris

atio

n ra

te (n

umbe

r/6

h/pa

tient

) 6

5

3

1

2

0

4

Num

ber of patients

40

30

10

20

0

Days after traumatic brain injury10842 60 953 71

Cortical spreading depressionIsoelectric spreading depolarisationNumber of patients

Figure: Time course of depolarisations after traumatic brain injuryThe total number of depolarisations in each 6 h bin was divided by the number of patients monitored with electrocorticography for the corresponding bin to calculate depolarisation rates than can be compared across time. This eliminates bias in the raw depolarisation counts created by the varying numbers of patients monitored through time. Rates for isoelectric spreading depolarisations and cortical spreading depression were computed separately and are stacked to show overall depolarisation rate. Non-integer values in the patient numbers result from start or termination of recordings within a 6-h time bin (eg, 4-h recording=0·67 patients). Data are derived from the 58 patients with depolarisations reported in our study.1

Epilepsy nurse specialists are a vital resourceI write in response to Mario Christodoulou’s feature1 about threats to the vital support provided by neurological nurse specialists in the UK. As the chief executive of Young Epilepsy, a UK charity working to improve health and education services for young people with epilepsy, I wholeheartedly support the arguments presented in the report, which highlighted the individualised care and cost benefi ts that specialist nurses bring to the health-care team.

The UK National Institute for Health and Clinical Excellence clinical

Correspondence

www.thelancet.com/neurology Vol 11 May 2012 391

guideline2 20 recommends access to epilepsy specialist nurses because this approach to care is good practice and is cost effi cient for the National Health Service (NHS). However, even in 2012, access to this kind of service is denied to many families. Additionally, as explained by Christodoulou, the positions of existing neurology nurse specialists face increasing pressure in the current fi nancial climate.

We know from The Child and Maternal Health Observatory review3 that for emergency hospital admissions (a key indicator of cost and quality of care) of children with epilepsy, the disparity between best and worst performing trusts is substantial, ranging between 19·1 and 181·2 admissions per 100 000 patients.

Research into the eff ect epilepsy specialist nurses have on these numbers is sparse, but a study done in the Nottingham area suggests that these nurses might reduce emergency admission by as much as 50%.4 Certainly further research is needed.

Specialist nurses can play an important part in deciding on approaches to care for individual patients, ensuring that patients are well informed and appropriately managed. They also make a valuable

contribution to the work of specialist teams at the tertiary level who identify young people suitable for epilepsy surgery. In a report5 presented at a Young Epilepsy Summit on paediatric epilepsy surgery in November 2010, the fi nancial eff ect of under-referral (only about one in four suitable patients are currently considered for surgery) was reported to be £280 million in state sector costs over 10 years.

The All Party Parliamentary Group on Epilepsy held an enquiry in 2007. The title of its subsequent report—Wasted Money, Wasted Lives6—sums up the situation then and little has changed in the past 5 years. Epilepsy is perhaps unusual in that charities working on behalf of people with epilepsy are not asking for more resources, but for the resources available to be better allocated.

Decisions about service provision taken under extreme pressure to save money can so easily lead to increased costs. Young Epilepsy supports the issues raised in Christodoulou’s piece, which should ring alarm bells for those commissioning services across the country.

Finally, we should remember—in this new patient-centred, outcomes-

based NHS—that the real cost of these decisions will be borne by patients, in this case young people with epilepsy, and their families.I am Chief Executive of Young Epilepsy. I declare that I have no confl icts of interest.

David Forddford@youngepilepsy.org.uk

Young Epilepsy, Lingfi eld, Surrey, RH7 6PW, UK

1 Christodoulou M. Neurological nurse specialists: a vital resource under threat. Lancet Neurol 2012; 11: 210–11.

2 NICE Clinical Guidance 20—The epilepsies: the diagnosis and management of the epilepsies in adults and children in primary and secondary care. http://guidance.nice.org.uk/CG20 (accessed March 30, 2012).

3 Child and Maternal Health Observatory (ChiMat). NHS Atlas of Variation in Healthcare for Children and Young People. RightCare, March 2012.

4 Johnson K, McGowan T, Dunkley C. A review of epilepsy specialist nurse’s clinical activity and impact on paediatric admissions. British Paediatric Neurology Association, 2010. http://www.cewt.org.uk/fi les/ESN%20BPNA%202010.pdf (accessed March 30, 2012).

5 Cruickshank A. Improving access to childhood epilepsy surgery—lifetime costs estimate. http://youngepilepsy.org.uk/what-we-do/campaigns/campaign-news-and-reports (accessed March 30, 2012).

6 Report by the All Party Parliamentary Group on Epilepsy—Wasted money, wasted lives—the human and economic cost of epilepsy in England. http://www.jointepilepsycouncil.org.uk/downloads/2011/Wasted%20Money,%20Wasted%20Lives.pdf (accessed March 30, 2012).

Comment

www.thelancet.com/neurology Vol 11 May 2012 385

Donation after cardiac death (DCD) is the process through which donation of solid organs takes place after the heart stops (usually within 5 min of cardiac arrest). Because the permanent cessation of heartbeat satisfi es the dead donor rule, which states that donor transplantation should not kill the patient but rather that the patient be already dead, DCD procedures have been approved in many countries.1 During the past decade, the disparity between the number of patients waiting for a transplant and the number of organs available has increased. As such, transplantation programmes in many countries have revisited the use of organs from donors after cardiac death. Previously, most programmes relied solely on organs from donors who were brain dead. As a result, DCD has greatly boosted organ donation beyond that which occurred after the declaration of brain death or neurological determination of death.2

In a study reported in this issue of The Lancet Neurology,3

Rabinstein and colleagues have tested a neurologically-based scoring system that should increase the accuracy of donor selection. Successful DCD requires selection of patients whose hearts are likely to stop within 1–2 h after withdrawal of life-sustaining treatment (WLST), to reduce anoxic injury that occurs during the withdrawal process and thus allow the organs to remain viable for transplantation. However, the DCD process needs substantial resources and therefore the ability to predict with reasonable certainty which patients will die within 1–2 h is key. The Wisconsin guidelines4 were developed by non-neurological intensive-care specialists to provide a measure of the probability of cardiac arrest within 1 h of WLST. The factors used in these guidelines include presence of spontaneous respiration after 10 min off the ventilator, body-mass index, use of vasopressors, age of the patient, use of tracheostomy versus use of endotracheal tube, and oxygen saturation after 10 min. Each of these variables was assigned a value and the aggregate score was used to predict cardiac death within 1 h of WLST. However, subsequent attempts to defi ne risk factors for early death more accurately have suggested that a reliable technique for prediction of time of death after WLST is not avaliable.5

Most patients who become DCD donors have severe CNS damage but are not brain dead. Although neurologists are at least one step removed from the DCD

process, such professionals might be better qualifi ed than other specialists to establish the neurological factors that predict cardiac arrest after WLST in patients who have brain damage. Since the withdrawal process includes the removal of the endotracheal tube in the operating room (and hence most patients die of asphyxic cardiac arrest), the presumption is that patients who are closer to brain death and are either not able to breathe adequately or protect the airway have a higher chance of prompt cardiac arrest. Whether or not health-care providers have qualms or philosophical concerns about DCD, when it is adopted as an accepted practice, as it currently is in many countries, the improvement of the process becomes a priority.6

Rabinstein and colleagues report a prospective observational study3 that aimed to validate the use of a neurological scoring system to assess likelihood of death within 60 min after WLST. The scoring system was built on a previous retrospective study7 by the investigators of patients who had brain injury and underwent WLST for compassionate reasons.6 Rabinstein and colleagues’ study3 included 178 patients with heterogenous disorders, but more than 85% of them had structural brain damage (intracranial haemorrhage, ischaemic stroke, or head injury). Borrowing from the FOUR score tabulation8 of neurological assessment of patients in coma, four factors were statistically associated with death within 1 h of WLST: absent corneal refl exes, absent cough

Donation after cardiac death: enter the neurologistPublished OnlineApril 10, 2012DOI:10.1016/S1474-4422(12)70068-6

See Articles page 414

Ria

Nov

osti/

Scie

nce

Phot

o Li

brar

y

Comment

386 www.thelancet.com/neurology Vol 11 May 2012

refl ex, no better than extensor (decerebrate) posturing, and oxygen index. Each of these factors was assigned a score (1 point each for absent corneal refl ex, extensor or absent motor response, and oxygenation index of >3·0, and 2 points for absent cough refl ex) and the aggregate score was related directly to the likelihood of death within 1 h of WLST. For patients who had the highest score (5 points), the probability of cardiac arrest within 1 h of WLST was 0·87. However, in patients with scores of less than 5 points, the probability of death within 1 h of WLST varied with the combination of variables (eg, from 0·45 to 0·61 in patients with an aggregate score of 3 points). Thus, for aggregate scores of less than 5 points, specifi c factors need to be taken into consideration and not just the aggregate score.

Although the population in Rabinstein and colleagues’ study3 was small and heterogenous and their scoring system was not compared with the established Wisconsin criteria,4 their analysis provides an encouraging step in refi ning the prediction of cardiac arrest within 1–2 h of WLST and thus selection of patients for DCD. Further refi nements and validation studies are needed. In the meantime, reliance on the Wisconsin scoring system4 or that developed by Rabinstein and colleagues3 would be imprudent. Selection of the best potential DCD donors

will need good clinical judgment and incorporation and analysis of both scoring systems.

G Bryan Young, Michael D SharpeDivision of Critical Care, Western University, Room B10–106, University Hospital, 339 Windermere Road, London, Ontario, Canada N6A5A5bryan.young@lhsc.on.ca

We declare that we have no confl icts of interest.

1 Bernat JL. Contemporary controversies in the defi nition of death. Prog Brain Res 2009; 177: 21–31.

2 Hernandez-Alejandro R, Wall W, Jevnikar A, et al. Organ donation after cardiac death: donor and recipient outcomes after the fi rst three years of the Ontario experience. Can J Anaesth 2011; 58: 599–605.

3 Rabinstein AA, Yee AH, Mandrekar J, et al. Prediction of potential for organ donation after cardiac death in patients in neurocritical state: a prospective observational study. Lancet Neurol 2012; published online April 5. DOI:10.1016/S1474-4422(12)70060-1.

4 Lewis J, Peltier J, Nelson H, et al. Development of the University of Wisconsin donation after cardiac death evaluation tool. Prog Transplant 2003; 13: 265–73.

5 Wind J, Snoeijs MGJ, Brugman CA et al. Prediction of time of death after withdrawal of life-sustaining treatment in potential donors after cardiac death. Crit Care Med 2012; 40: 766–79.

6 Dharamsi S, Ho A, Spadafora SM, et al. The physician as health advocate: translating the quest for social responsibility into medical education and practice. Acad Med 2011; 86: 1108–13.

7 Yee AH, Rabinstein AA, Thapa P, Mandrekar J, Wijdicks EFM. Factors infl uencing time to death after withdrawal of life support in neurocritical patients. Neurology 2010; 74: 1380–85.

8 Wijdicks EF, Bamlet WR, Maramattom BV, et al. Validation of a new coma scale: the FOUR score. Ann Neurol 2005; 58: 585–93.

Published OnlineApril 10, 2012

DOI:10.1016/S1474-4422(12)70066-2

See Articles page 420

MS clinical trials: what can subgroup analyses teach us? Since the fi rst modern-day clinical trial in patients with multiple sclerosis (MS),1 the MS therapeutics speciality has matured. Eight disease-modifying therapies (DMTs) have been approved for relapsing forms of MS since 1993, and more are on the way. The clinical science that underlies trial design and analysis is also maturing. Clinical investigators and study sponsors have learned to acquire and analyse MRI data in multicentre studies, use time-to-disability survival analysis, in corporate patient-reported outcomes, and undertake exploratory studies of new clinical, imaging, and biomarker methods for monitoring treatment eff ects. The number of patients entering into clinical trials has increased over time, in part because recent trials have included patients with low MS disease activity. The fi rst placebo-controlled study1 to investigate a treatment for MS enrolled less than 100 participants per group and the trial that led to

regulatory approval for interferon beta-1b2 enrolled about 125 patients per group, whereas more recent placebo-controlled trials of natalizumab,3 cladribine,4 and fi ngolimod5 enrolled over 400 patients per group. Larger sample sizes provide greater opportunity for secondary analyses, including subgroup comparisons.

In that regard, in The Lancet Neurology, Devonshire and colleagues6 report subgroup analyses from the phase 3 FREEDOMS (FTY720 Research Evaluating Effects of Daily Oral therapy in MS) placebo-controlled trial of fingolimod for relapsing-remitting MS (RRMS).5 In the original trial, 1272 patients with RRMS were randomly assigned to placebo (n=418), fingolimod 0·5 mg/day (n=425), or fingolimod 1·25 mg/day (n=429). The approved dose of fingolimod, 0·5 mg/day, was associated with a 54% reduction in annualised relapse rate (ARR) and a 30% reduction in worsening of expanded disability

Comment

382 www.thelancet.com/neurology Vol 11 May 2012

Published OnlineApril 11, 2012

DOI:10.1016/S1474-4422(12)70069-8

See Articles page 397

Mobile acute stroke units: bringing the hospital to the patientIt is well established that the benefi ts of thrombolysis for acute ischaemic stroke decline with delay from stroke onset to treatment. Furthermore, there is little evidence from studies in hospital settings to suggest that the risk of adverse eff ects of thrombolysis is greater in appropriately selected patients treated early within the accepted time window. Substantial eff ort is therefore put into designing acute stroke pathways in such a way that the delay from hospital admission to thrombolysis, the so-called door-to-needle time, is minimised.

The delay from stroke onset to thrombolysis is also dependent, however, on the time taken to get the patient to the hospital doors. Moreover, even in the most organised acute stroke services there is still sometimes a delay in getting brain imaging after the patient arrives at the hospital. It is therefore logical to take the acute stroke service and the CT brain scanner to the patient in the form of a specialised ambulance fi tted with a mobile scanner, particularly in regions where ambulance travel times to an appropriate hospital are likely to be long.

In this issue of The Lancet Neurology, Silke Walter and colleagues1 report the fi rst randomised trial of such an approach. In a single-centre trial, suspected stroke patients received either pre-hospital treatment in a mobile stroke unit (MSU—a specialised ambulance equipped with a CT scanner, point of care laboratory, and telemedicine connection) or conventional hospital-based treatment. The primary outcome, the median time

from alarm to treatment decision, was reduced from 76 min (IQR 63–94) in patients in the control group to 35 min (31–39) in patients in the MSU group. The trial was therefore stopped at a planned interim analysis after inclusion of 100 patients.

The main fi nding of the trial—the substantial reduction in alarm to treatment decision time in the MSU group—is probably valid. However, there are some methodological issues that could, at least in theory, have biased the secondary clinical outcomes. First, although the trial is described as randomised, the MSU intervention was actually allocated on a time period basis rather than on an individual patient basis. Some weeks were allocated the MSU service whereas others were allocated the standard hospital-based service. Although the weeks were allocated at random, this would not achieve the main purpose of randomisation, allocation concealment—ie, that the study investigators should ideally not know in advance which intervention each potential recruit would get before their inclusion (or not) in the trial. Potential trial patients were screened before inclusion by an ambulance dispatcher who knew which intervention was active that week. Since more patients were excluded at this stage than were included in the trial (261 vs 100) a small bias could have aff ected the characteristics of those patients included in particular weeks. Indeed, clinical characteristics were not particularly well balanced across the treatment groups, although this could simply relate to the relatively small size of the trial. Second, there was no masking of the initial treatment allocation from clinicians caring for patients later during their hospital stay and the assessment of clinical outcomes at 7 days was also unmasked. The design of future larger-scale trials should address these methodological issues.

Would such an MSU service be cost eff ective? The clinical outcomes are diffi cult to interpret in this trial because of small numbers, but the MSU did not signifi cantly increase the proportion of patients who received thrombolysis (12/53 vs 8/47 in the control group). However, the median time from alarm to thrombolysis was substantially shorter in the MSU group: 38 min versus 73 min. On the basis of the eff ect of thrombolysis on death and dependency in hospital-based randomised trials, this reduction in delay to M

ike

Devl

in/S

cienc

e Ph

oto

Libr

ary

Comment

www.thelancet.com/neurology Vol 11 May 2012 383

treatment would be expected to result in about one fewer treated patient being dead or dependent at hospital discharge. In reality, no diff erence was evident in the clinical outcomes at 7 days in the trial (21/53 dead or dependent in the MSU group vs 20/47 in the control group). Moreover, of those patients with ischaemic stroke in the MSU group who were treated with thrombolysis, three died over the next few days whereas there were no deaths in patients with ischaemic stroke in the control group. Although these numbers are too small to allow reliable interpretation, they highlight the need for larger trials with suffi cient power to establish the eff ect of the MSU intervention on clinical outcome.

An important intermediate step in showing the potential for improved clinical outcomes with an MSU service would be to study the CT brain scans from patients scanned early by the MSU and to consider the “tissue window”, estimated with the Alberta Stroke Program Early CT (ASPECT) score,2 as well as the “time window”. If the patients scanned early had high ASPECT scores, suggesting a signifi cant potential to benefi t from thrombolysis, then that would provide some reassurance that better clinical outcomes could be expected.

The generalisability of the trial fi ndings to potential MSU services elsewhere will depend very much on the setting. This trial was set in an urban area with a median distance from the patient to the hospital of 7 km and median alarm to arrival times of 8 min for the standard ambulance versus 12 min for the MSU. The MSU would potentially work less well in rural areas in which locally based ambulances might be able to get patients to hospital about as quickly as a hospital-situated MSU could get out to the patient. Nevertheless, this trial has shown convincingly that in at least some settings an MSU-based service is feasible and can substantially reduce delays to treatment.

*Peter M Rothwell, Alastair M BuchanNuffi eld Department of Clinical Neuroscience (PMR) and Nuffi eld Department of Medicine (AMB), Oxford University, John Radcliff e Hospital, Headington, Oxford, UKpeter.rothwell@clneuro.ox.ac.uk

We declare that we have no confl icts of interest.

1 Walter S, Kostopoulos P, Haass A, et al. Diagnosis and treatment of patients with stroke in a mobile stroke unit versus in hospital: a randomised controlled trial. Lancet Neurol 2012; published online April 11. DOI:10.1016/S1474-4422(12)70057-1.

2 Barber PA, Demchuk AM, Zhang J, Buchan AM. Validity and reliability of a quantitative computed tomography score in predicting outcome of hyperacute stroke before thrombolytic therapy. ASPECTS Study Group. Alberta Stroke Programme Early CT Score. Lancet 2000; 355: 1670–74.

Defi ning and refi ning 5-HT receptor targets for migraineThe triptans are well established as effective treatments for migraine, but the specific mechanisms responsible for their therapeutic effects remain unknown. Triptans selectively activate 5-HT1 receptors, in particular the 1B, 1D, and 1F receptor subtypes.1 There has been extensive discussion about which of these 5-HT1 receptor subtypes are responsible for the therapeutic and adverse effects of treatment with triptans and about the locations in and around the brain where these effects might occur. In this issue of The Lancet Neurology, Färkkilä and colleagues2 describe a clinical trial of an oral formulation of lasmiditan—a non-triptan selective 5-HT1F receptor agonist—as an acute treatment for migraine. This study, in combination with previous studies of the clinical pharmacology and efficacy of lasmiditan in migraine,3,4 confirms the potential of lasmiditan as a treatment for acute migraine and also provides new

insights into the mechanisms of 5-HT1 receptors in migraine.

Lasmiditan is chemically distinct from triptans and has different pharmacological properties. Unlike triptans, lasmiditan has higher selectivity for the 5-HT1F receptor than for the 5-HT1B and 5-HT1D receptors.4 Lasmiditan was active in two in-vivo models that have been used to investigate migraine—plasma protein extravasation evoked by stimulation of the trigeminal ganglion and activation of c-fos in the trigeminal nucleus caudalis evoked by similar stimulation.4 Although these assays do not predict efficacy of migraine treatment, they do suggest functional effects in vivo, which in the case of lasmiditan are probably mediated by the 5-HT1F receptor.4 A previous study of intravenous lasmiditan suggested that it was an effective treatment for migraine.3

Published OnlineMarch 28, 2012 DOI:10.1016/S1474-4422(12)70062-5

See Articles page 405

Editorial

www.thelancet.com/neurology Vol 11 May 2012 381

A decade at the forefront of clinical neurologyIn May, 2002, a new clinical neurology journal was launched as part of The Lancet family. The remit of The Lancet Neurology then, as it is today, was to provide authoritative articles, including topical reviews and lively discussion, and to become a nexus for practicing neurologists.

In the inaugural editorial, we speculated on how much the fi eld would change over the coming 10 years and posited that, with the advent of new technological developments in diagnosis and treatment, particularly in genetics and imaging, “it seems likely that neurologists will be able to predict, with some degree of accuracy, which individuals are at greatest risk of developing certain conditions long before the onset of symptoms”. In light of what has been achieved in the intervening 10 years, it would seem that this forecast was prescient.

The development of enhanced imaging techniques has increased the resources available to diagnose and monitor progression of neurological disorders. Likewise, the advent of large cohort and genome-wide association studies, coupled with improvements in sequencing technology and the development of integrated repositories and bioinformatics programs, has enabled identifi cation of common and rare genetic polymorphisms and mutations that increase the risk of developing neurological disorders; these advances have also brought with them the promise of new avenues for treatment. The burgeoning research eff ort into the identifi cation and validation of biomarkers will undoubtedly feed into this strategy and push back the frontiers of when a diagnosis can be made with certainty. With ongoing development, correct application, and careful interpretation of the results that are generated, all of these approaches will undoubtedly occupy an increasingly important position in the clinic in the years to come.

Despite great strides in our ability to defi ne and diagnose neurological disorders, there is still disparity in the availability of treatment options across disorders. There is also uncertainty about whether curative treatments or therapies that will halt progress and treat symptoms adequately will be available across the board in the foreseeable future. Neurologists today have a greater array of pharmacological options to off er their

patients than they had 10 years ago; however, these typically only delay progression or alleviate symptoms. Taken in tandem with recognition of the need to diagnose neurological disorders as early as possible, and thus begin interventions sooner, the net result is that many patients can maintain an acceptable quality of life for longer. But they cannot currently expect to be cured. A shift towards large collaborative studies and powerful randomised controlled trials can be seen as a step in the right direction to the provision of safe and tolerable treatment options. And the ongoing refi nement and upgrading of criteria for diagnosis and guidelines for treatment should ensure that research fi ndings and new treatment modalities are successfully and eff ectively translated into clinical practice.

What do the next 10 years hold for The Lancet Neurology? As the fi eld has changed over the past decade, we have striven to keep one step ahead; to continue in this vein, and in addition to our commitment to continue to provide the high-quality content you have come to expect, we are now seeking to expand the remit of the journal and to engage with the public health community. The imminent tipping of the global demographic scales towards an older population will bring with it new corollaries that will make healthy ageing a global priority. The eff ects of an ageing population will be particularly felt in low-income and middle-income countries, which might have insuffi cient medical infrastructure and fi nancial reserves to address the neurological burden that will accrue as life expectancy increases. At this time of unprecedented demographic change we feel it is important to bring more attention to the specifi c challenges faced by the global neurological community and highlight the wider implications associated with the provision of neurological care in low-resource settings.

One thing that we can predict with absolute certainty is that the continuing success of The Lancet Neurology lies in maintaining our close ties with the neurological community: our readers, our authors, and our reviewers. To all of you, we extend our gratitude and we look forward to a continuing and productive relationship over the coming 10 years and beyond. ■ The Lancet Neurology

For the inaugural editorial see Editorial Lancet Neurol 2002; 1: 1

For more on WHO World Health Day see http://www.who.int/world-health-day/2012/en/index.html

For Ageing well: a global priority see Editorial Lancet 2012; 379: 1274

Comment

388 www.thelancet.com/neurology Vol 11 May 2012

3 Polman CH, O’Connor PW, Havrdova E, et al. A randomized, placebo-controlled trial of natalizumab for relapsing multiple sclerosis. N Engl J Med 2006; 354: 899–910.

4 Giovannoni G, Comi G, Cook S, et al. A placebo-controlled trial of oral cladribine for relapsing multiple sclerosis. N Engl J Med 2010; 362: 416–26.

5 Kappos L, Radue EW, O’Connor P, et al. A placebo-controlled trial of oral fi ngolimod in relapsing multiple sclerosis. N Engl J Med 2010; 362: 387–401.

6 Devonshire V, Havrdova E, Radu EW, et al. Relapse and disability outcomes in patients with multiple sclerosis treated with fi ngolimod: subgroup analyses of the double-blind, randomised, placebo-controlled FREEDOMS study. Lancet Neurol 2012; published online April 10. DOI:10.1016/S1474-4422(12)70056-X.

7 Hutchinson M, Kappos L, Calabresi PA, et al. The effi cacy of natalizumab in patients with relapsing multiple sclerosis: subgroup analyses of AFFIRM and SENTINEL. J Neurol 2009; 256: 405–15.

8 Giovannoni G, Cook S, Rammohan K, et al. Sustained disease-activity-free status in patients with relapsing-remitting multiple sclerosis treated with cladribine tablets in the CLARITY study: a post-hoc and subgroup analysis. Lancet Neurol 2011; 10: 329–37.

9 Pocock SJ, Assmann SE, Enos LE, Kasten LE. Subgroup analysis, covariate adjustment and baseline comparisons in clinical trial reporting: current practice and problems. Stat Med 2002; 21: 2917–30.

10 Hernandez AV, Boersma E, Murray GD, Habbema JD, Steyerberg EW. Subgroup analyses in therapeutic cardiovascular clinical trials: are most of them misleading? Am Heart J 2006; 151: 257–64.

ErratumByrne S, Elamin M, Bede P, et al. Cognitive and clinical characteristics of patients with amyotrophic lateral sclerosis carrying a C9orf72 repeat expansion: a population-based cohort study. Lancet Neurol 2012; 11: 232–40—The grant number in the fi rst sentence of the acknowledgments section should have read 259867. This correction has been made to the online version as of April 18, 2012.

Correspondence

www.thelancet.com/neurology Vol 11 May 2012 389

Spreading depolarisations and traumatic brain injury: time course and mechanisms

In a recent research article in The Lancet Neurology,1 Jed A Hartings and colleagues reported their fi ndings of electrocorticography in 103 patients requiring intensive care and surgery for traumatic brain injury (TBI). They found an association between the presence and type of spreading depolarisations during acute ictus and survival and functional outcome 6 months later. The investigators suggest that, after further clinical studies, individualised care could include monitoring and treatment of spreading depolarisations. Two issues arise from this excellent report.

Patients were recruited into the study within 7 days of TBI and electrocorticography was done for the duration of other clinical invasive monitoring, or to a maximum of 7 days. Electrocorticography moni-toring continued for 67 h (IQR 40–102) but, theoretically, a patient could have been monitored as late as 14 days after injury. First, what was the time course of the spreading depolarisations? Were cortical spread ing depolarisations clustered in the early phase after injury, as in experimental middle cerebral artery occlusion?2 Alternatively, were isoelectric spreading depolarisations pre dominantly pre terminal phe-nom ena? In this regard, I note that isoelectric spreading depolarisations occurred in 12 of 28 patients that died compared with their occurrence in eight of 75 patients that survived (43% vs 11%, SD 3·6 diff erence in the hypothesis test for proportions, p<0·001).

In 1999, Morris and colleagues3 reported the failure of two phase 3 clinical trials of a competitive NMDA-receptor antagonist (Selfotel,

CGS 19755) in the treatment of severe TBI. In the international arm of the trial, there was a subgroup assessment of patients who underwent operation for mass lesions (which was similar to the types of operations and patients in Hartings and colleagues’ study1). On inspection of these data—in view of the limitations of small sample size and a quick post-hoc analysis using information from the tables—the use of intravenous Selfotel (5 mg/kg once a day) for 4 days did not seem to reduce the proportion of patients dying (more deaths in the Selfotel group, 31% vs 20%, SD 1·5, p>0·05). Also, in the Selfotel group, there was a smaller proportion of favourable outcomes (42% vs 67%, SD 2·9 diff erence, p<0·01) than in the placebo group.

The electrophysiological and ionic transients of cortical spreading depolarisations are NMDA-receptor mediated, and inhibition can be induced by agents acting at the NR2B, glycine, or non-competitive binding sites.4–7 The Selfotel trials3 can be considered a proof of concept or even a test of the hypothesis proposed by Hartings and colleagues’ in their conclusion—short-term inhibition of a cortical NMDA receptor-mediated process in severe TBI cases requiring neurosurgery fails to improve outcome. Therefore, the second point of discussion that I would like to raise is the type of tests of their hypothesis they envisage. For example, they could take the form of a longer duration of interventional therapy, up to 14 days—for which it would be important to know more about the clustering and preterminal information. Alternatively, tests could involve the use of NMDA-receptor antagonists that have both a neuronal and oligodendrocyte profi le.8,9 Such protection of white matter injury in severe TBI would indeed be inkeeping with outcome long after such injury.10 I declare that I have no confl icts of interest.

Robert C TaskerRobert.tasker@childrens.harvard.edu

Departments of Neurology and Anesthesiology (Pediatrics), Harvard Medical School and Children’s Hospital Boston, Boston, MA, USA

1 Hartings JA, Bullock MR, Okonkwo DO, et al. Spreading depolarisations and outcome after traumatic brain injury: a prospective observational study. Lancet Neurol 2011; 10: 1058–64.

2 Hartings JA, Rolli ML, Lu XC, Tortella FC. Delayed secondary phase of peri-infarct depolarizations after focal cerebral ischemia: relation to infarct growth and neuroprotection. J Neurosci 2003; 27: 11602–10.

3 Morris GF, Bullock R, Marshall SB, et al. Failure of the competitive N-methyl-D-aspartate antagonist Selfotel (CGS 19755) in the treatment of severe head injury: results of two phase III clinical trials. J Neurosurg 1999; 91: 737–43.

4 McLachlan RS. Suppression of spreading depression of Leao in neocortex by an N-methyl-D-aspartate receptor antagonist. Can J Neurol Sci 1992; 19: 487–91.

5 Nellgard B, Wieloch T. NMDA-receptor blockers but not NBQX, an AMPA-receptor antagonist, inhibit spreading depression in the rat brain. Acta Physiol Scand 1992; 146: 497–503.

6 Obrenovitch TP, Zilkha E. Inhibition of cortical spreading depression by L-701,324, a novel antagonist at the glycine site of the N-methyl-D-aspartate receptor complex. Br J Pharmacol 1996; 117: 931–37.

7 Somjen GG. Mechanisms of spreading depression and hypoxic spreading depression-like depolarization. Physiol Rev 2001; 81: 1065–96.

8 Bakiri Y, Hamilton NB, Karadottir R, Attwell D. Testing NMDA receptor block as a therapeutic strategy for reducing ischemic damage to CNS white matter. Glia 2008; 56: 233–40.

9 Giacino JT, Whyte J, Bagiella E, et al. Placebo-controlled trial of amantadine for severe traumatic brain injury. N Engl J Med 2012; 366: 819–26.

10 Tasker RC, Westland AG, White DK, Williams GB. Corpus callosum and inferior forebrain white matter microstructure are related to functional outcome from raised intracranial pressure in child traumatic brain injury. Dev Neurosci 2010; 32: 374–84.

Authors’ replyWe appreciate Robert C Tasker’s interest in our report1 and thank him for his interesting questions. His concerns relate to the temporal patterning of cortical spreading depolarisations and to the implications for design of therapeutic trials.

In the fi gure we present a diff erent display of the same data reported in our study to show the timing of electrocorticography and the spreading depolarisations. Recordings were obtained mainly in the fi rst week

Correspondence

www.thelancet.com/neurology Vol 11 May 2012 389

Spreading depolarisations and traumatic brain injury: time course and mechanisms

In a recent research article in The Lancet Neurology,1 Jed A Hartings and colleagues reported their fi ndings of electrocorticography in 103 patients requiring intensive care and surgery for traumatic brain injury (TBI). They found an association between the presence and type of spreading depolarisations during acute ictus and survival and functional outcome 6 months later. The investigators suggest that, after further clinical studies, individualised care could include monitoring and treatment of spreading depolarisations. Two issues arise from this excellent report.

Patients were recruited into the study within 7 days of TBI and electrocorticography was done for the duration of other clinical invasive monitoring, or to a maximum of 7 days. Electrocorticography moni-toring continued for 67 h (IQR 40–102) but, theoretically, a patient could have been monitored as late as 14 days after injury. First, what was the time course of the spreading depolarisations? Were cortical spread ing depolarisations clustered in the early phase after injury, as in experimental middle cerebral artery occlusion?2 Alternatively, were isoelectric spreading depolarisations pre dominantly pre terminal phe-nom ena? In this regard, I note that isoelectric spreading depolarisations occurred in 12 of 28 patients that died compared with their occurrence in eight of 75 patients that survived (43% vs 11%, SD 3·6 diff erence in the hypothesis test for proportions, p<0·001).

In 1999, Morris and colleagues3 reported the failure of two phase 3 clinical trials of a competitive NMDA-receptor antagonist (Selfotel,

CGS 19755) in the treatment of severe TBI. In the international arm of the trial, there was a subgroup assessment of patients who underwent operation for mass lesions (which was similar to the types of operations and patients in Hartings and colleagues’ study1). On inspection of these data—in view of the limitations of small sample size and a quick post-hoc analysis using information from the tables—the use of intravenous Selfotel (5 mg/kg once a day) for 4 days did not seem to reduce the proportion of patients dying (more deaths in the Selfotel group, 31% vs 20%, SD 1·5, p>0·05). Also, in the Selfotel group, there was a smaller proportion of favourable outcomes (42% vs 67%, SD 2·9 diff erence, p<0·01) than in the placebo group.

The electrophysiological and ionic transients of cortical spreading depolarisations are NMDA-receptor mediated, and inhibition can be induced by agents acting at the NR2B, glycine, or non-competitive binding sites.4–7 The Selfotel trials3 can be considered a proof of concept or even a test of the hypothesis proposed by Hartings and colleagues’ in their conclusion—short-term inhibition of a cortical NMDA receptor-mediated process in severe TBI cases requiring neurosurgery fails to improve outcome. Therefore, the second point of discussion that I would like to raise is the type of tests of their hypothesis they envisage. For example, they could take the form of a longer duration of interventional therapy, up to 14 days—for which it would be important to know more about the clustering and preterminal information. Alternatively, tests could involve the use of NMDA-receptor antagonists that have both a neuronal and oligodendrocyte profi le.8,9 Such protection of white matter injury in severe TBI would indeed be inkeeping with outcome long after such injury.10 I declare that I have no confl icts of interest.

Robert C TaskerRobert.tasker@childrens.harvard.edu

Departments of Neurology and Anesthesiology (Pediatrics), Harvard Medical School and Children’s Hospital Boston, Boston, MA, USA

1 Hartings JA, Bullock MR, Okonkwo DO, et al. Spreading depolarisations and outcome after traumatic brain injury: a prospective observational study. Lancet Neurol 2011; 10: 1058–64.

2 Hartings JA, Rolli ML, Lu XC, Tortella FC. Delayed secondary phase of peri-infarct depolarizations after focal cerebral ischemia: relation to infarct growth and neuroprotection. J Neurosci 2003; 27: 11602–10.

3 Morris GF, Bullock R, Marshall SB, et al. Failure of the competitive N-methyl-D-aspartate antagonist Selfotel (CGS 19755) in the treatment of severe head injury: results of two phase III clinical trials. J Neurosurg 1999; 91: 737–43.

4 McLachlan RS. Suppression of spreading depression of Leao in neocortex by an N-methyl-D-aspartate receptor antagonist. Can J Neurol Sci 1992; 19: 487–91.

5 Nellgard B, Wieloch T. NMDA-receptor blockers but not NBQX, an AMPA-receptor antagonist, inhibit spreading depression in the rat brain. Acta Physiol Scand 1992; 146: 497–503.

6 Obrenovitch TP, Zilkha E. Inhibition of cortical spreading depression by L-701,324, a novel antagonist at the glycine site of the N-methyl-D-aspartate receptor complex. Br J Pharmacol 1996; 117: 931–37.

7 Somjen GG. Mechanisms of spreading depression and hypoxic spreading depression-like depolarization. Physiol Rev 2001; 81: 1065–96.

8 Bakiri Y, Hamilton NB, Karadottir R, Attwell D. Testing NMDA receptor block as a therapeutic strategy for reducing ischemic damage to CNS white matter. Glia 2008; 56: 233–40.

9 Giacino JT, Whyte J, Bagiella E, et al. Placebo-controlled trial of amantadine for severe traumatic brain injury. N Engl J Med 2012; 366: 819–26.

10 Tasker RC, Westland AG, White DK, Williams GB. Corpus callosum and inferior forebrain white matter microstructure are related to functional outcome from raised intracranial pressure in child traumatic brain injury. Dev Neurosci 2010; 32: 374–84.

Authors’ replyWe appreciate Robert C Tasker’s interest in our report1 and thank him for his interesting questions. His concerns relate to the temporal patterning of cortical spreading depolarisations and to the implications for design of therapeutic trials.

In the fi gure we present a diff erent display of the same data reported in our study to show the timing of electrocorticography and the spreading depolarisations. Recordings were obtained mainly in the fi rst week

Correspondence

390 www.thelancet.com/neurology Vol 11 May 2012

London, UK (AJS); Department of Neurosurgery, University of Pittsburgh, Pittsburgh, PA, USA (DOO); Department of Neurological Surgery, University of Miami, Miami, FL, USA (MRB)

1 Hartings JA, Bullock MR, Okonkwo DO, et al. Spreading depolarisations and outcome after traumatic brain injury: a prospective observational study. Lancet Neurol 2011; 10: 1058–64.

2 Hartings JA, Strong AJ, Fabricius M, et al. Spreading depolarizations and late secondary insults after traumatic brain injury. J Neurotrauma 2009; 26: 1857–66.

3 Dreier JP, Major S, Pannek HW, et al. Spreading convulsions, spreading depolarization and epileptogenesis in human cerebral cortex. Brain 2011; 135: 259–75.

4 Dreier JP, Major S, Manning A, et al. Cortical spreading ischaemia is a novel process involved in ischaemic damage in patients with aneurysmal subarachnoid haemorrhage. Brain 2009; 132: 1866–81.

5 Lauritzen M, Hansen AJ. The eff ect of glutamate receptor blockade on anoxic depolarization and cortical spreading depression. J Cereb Blood Flow Metab 1992; 12: 223–29.

6 Morris GF, Bullock R, Marshall SB, Marmarou A, Maas A, Marshall LF. Failure of the competitive N-methyl-D-aspartate antagonist Selfotel (CGS 19755) in the treatment of severe head injury: results of two phase III clinical trials. The Selfotel Investigators. J Neurosurg 1999; 91: 737–43.

7 Yurkewicz L, Weaver J, Bullock MR, Marshall LF. The eff ect of the selective NMDA receptor antagonist traxoprodil in the treatment of traumatic brain injury. J Neurotrauma 2005; 22: 1428–43.

8 Lo EH. A new penumbra: transitioning from injury into repair after stroke. Nat Med 2008; 14: 497–500.

9 Ikonomidou C, Turski L. Why did NMDA receptor antagonists fail clinical trials for stroke and traumatic brain injury? Lancet Neurol 2002; 1: 383–86.

blocking spreading depolarisations, it would occur only in the 56% of patients with depolarisations, or perhaps just the 20% with ISD. Thus any eff ect would be greatly diluted by broad inclusion, even within the surgical subgroup. Second, it is questionable whether suffi cient brain levels of Selfotel, a competitive antagonist, were achieved to compete with high glutamate concentrations (100–200 times more than normal) for receptor binding; there was no confi rmation of intended eff ect on the target mechanism. In a subsequent severe TBI trial with the non-competitive NR2B-selective antagon-ist traxoprodil, improvements in outcome up to 11·8% were reported in some subgroups.7 Third, the Selfotel trials were stopped in part for safety concerns in a concomitant stroke trial, so that suffi cient statistical power was never achieved.

We envision electrocorticography as another method in the arma-mentarium of neuromonitoring. Just as mannitol is given only to patients with elevated intracranial pressure, a therapy to block spreading depolarisations would be administered only to patients that have them and for as long as they persist. This will confi rm that the pathological processes targeted by therapeutics are active in individual patients. The advantages of this approach are selective inclusion, mechanistic targeting, and, not least, assessment of the eff ect of therapy on the targeted mechanism. Such a design diff ers dramatically from the failed trials in the past, but rather builds towards personalised treatment of this heterogeneous disease. More work is needed to determine which depolarisations should be targeted, which therapies are most eff ective, and to ensure that interventions do not interfere with repair processes.8,9 We declare that we have no confl icts of interest.

Jed A Hartings, Anthony J Strong, David O Okonkwo, M Ross Bullock Department of Neurosurgery, University of Cincinnati, Cincinnati, OH, USA (JAH); Department of Clinical Neuroscience, King’s College London,

after TBI. Depolarisations occurring in electrically silent cortex (or isoelectric spreading depolarisations [ISD]) accounted for 36% of all events and showed a similar time course to depolarisations with depression of spontaneous cortical activity (cortical spreading depression): a maximal activity in the initial 48 h, followed by a decline and then a delayed secondary peak on days 6–8.2,3 Thus, the association of ISD with poor outcome is not because ISD occur only as preterminal events associated with global brain death. The predominance of ISD within the initial days rather suggests that, as in ischaemic stroke, ISD are associated with acute lesion development. The association with mortality might be both correlative and causative, since isoelectricity could result from ischaemia or impaired neurovascular coupling,4 and depolarisations cause cell death under these conditions, probably through Ca2+-mediated pathways.

Assuming that NMDA-receptor antagonism inhibits all depolarisations, which is unlikely,5 does the failure of the Selfotel trials6 provide evidence against depolarisations as a promising therapeutic target? We don’t think so. First, if there was a benefi cial eff ect from

Depo

laris

atio

n ra

te (n

umbe

r/6

h/pa

tient

) 6

5

3

1

2

0

4

Num

ber of patients

40

30

10

20

0

Days after traumatic brain injury10842 60 953 71

Cortical spreading depressionIsoelectric spreading depolarisationNumber of patients

Figure: Time course of depolarisations after traumatic brain injuryThe total number of depolarisations in each 6 h bin was divided by the number of patients monitored with electrocorticography for the corresponding bin to calculate depolarisation rates than can be compared across time. This eliminates bias in the raw depolarisation counts created by the varying numbers of patients monitored through time. Rates for isoelectric spreading depolarisations and cortical spreading depression were computed separately and are stacked to show overall depolarisation rate. Non-integer values in the patient numbers result from start or termination of recordings within a 6-h time bin (eg, 4-h recording=0·67 patients). Data are derived from the 58 patients with depolarisations reported in our study.1

Epilepsy nurse specialists are a vital resourceI write in response to Mario Christodoulou’s feature1 about threats to the vital support provided by neurological nurse specialists in the UK. As the chief executive of Young Epilepsy, a UK charity working to improve health and education services for young people with epilepsy, I wholeheartedly support the arguments presented in the report, which highlighted the individualised care and cost benefi ts that specialist nurses bring to the health-care team.

The UK National Institute for Health and Clinical Excellence clinical

In Context

392 www.thelancet.com/neurology Vol 11 May 2012

UK High Court case to reignite debate over assisted deathA paralysed man has won the right to argue in court that he should be able to request lawful euthanasia, in a hearing that could have repercussions for patients with dementia. David Holmes reports.

The debate about the rights and wrongs of assisted death is never far from the headlines, and so far 2012 has been no exception. In January the current legal status of assisted death in the UK was criticised for being “inadequate and incoherent” by the Commission on Assisted Dying: a report commissioned by the thinktank Demos. Many of those criticisms will now be aired in court in the coming months, after Tony Nicklinson, who was left paralysed from the neck down after a stroke, won the right to argue his case in the High Court of England and Wales that he should have the option of requesting euthanasia: that a physician be able to end his life without fear of prosecution.

At present, encouraging or assisting another person’s suicide is illegal in the UK under the Suicide Act of 1961, but the Director of Public Prosecutions will only bring a prosecution if they consider it in the public interest to do so. In 2010, the Director of Public Prosecutions clarifi ed this position by eff ectively stating that even in

cases for which there is clear evidence that an act of assisting a suicide has taken place, the Director of Public Prosecutions will not prosecute when assistance has been provided on compassionate grounds to a person who has made a considered and auto-nomous decision to end their life.

The ruling in the Nicklinson case could have unpredictable con sequences says Dan Hyde, of London-based legal fi rm Cubism Law. Although Hyde cautions that the odds are stacked against Nicklinson, “if things go in Mr Nicklinson’s favour” he says, “then people will be drilling down into that judgement to see whether they can then apply that to their case, and if it goes against Mr Nicklinson then there could be calls for further reviews”.

One group of people with an interest in the ruling will be the increasingly vocal set of campaigners who argue that patients who have a diagnosis of dementia should be able to opt for assisted death when their disease becomes terminal. Although Nicklinson’s case is very diff erent from that of a patient with dementia—there is no question of his capacity to make an informed and free choice—what is interesting, notes Hyde, is that he is raising a relatively new issue in the argument for assisted dying; the argument of legal necessity. This argument, says Hyde, as put by Nicklinson’s legal team, “eff ectively runs that if somebody is in a tragic situation and their quality of life will be so awful and ultimately they will have an unhappy prolonged life and suff er an awful death, then the old legal principle of necessity should prevail such that a person terminating their life or assisting their suicide, in accordance with their wishes, would have the defence that they acted out of necessity”.

“That would be a very dangerous precedent indeed”, say the Care not Killing alliance, a pro-life lobby group who are opposed to euthanasia and physician-assisted suicide. “The key point to grasp about this case”, they say, “is that Nicklinson, because he is not capable of killing himself even with assistance, is not seeking assisted suicide but euthanasia”. If successful, this would mean a change in the law greater than the proposals put forward by the Commission on Assisted Dying, which advised that under strictly defi ned circumstances, patients with a terminal illness and with less than 12 months to live could be assisted to die, assuming they could comply with a number of safeguards. A similar system is in place in several other

For the Commission on Assisted Dying report see http://www.

demos.co.uk/publications/thecommissiononassisteddying

“...if things go in Mr Nicklinson’s favour...then people will be drilling down into that judgement to see whether they can then apply that to their case...“

Assisted suicide for terminally ill patients is a controversial and emotive topic

Hen

ny A

llis/

Scie

nce

Phot

o Li

brar

y

In Context

www.thelancet.com/neurology Vol 11 May 2012 393

European countries and Oregon and Washington in the USA.

At the time of its publication, the Commission on Assisted Dying—chaired by the former Lord Chancellor Charles Falconer—was criticised by Care Not Killing and other groups opposed to a change in the current law for its perceived lack of independence. This criticism was partly motivated by the fact that the report was funded by the author Sir Terry Pratchett, who has been one of the most prominent advocates for assisted suicide to be legalised since he was diagnosed with a rare early-onset form of Alzheimer’s disease.

However, the report stopped far short of recommending that people with dementia be able to request an assisted death. The Commission did concede that it was sympathetic to people who are in the early stages of dementia who might appreciate the security of knowing they could specify in a legal document (a so-called living will) the circumstances in which they would like to be able to end their life once they had lost capacity. However, it considered “that the requirement of mental capacity is an essential safeguard for assisted dying legislation; therefore the Commission does not propose any legislation that might allow non-competent people to receive assistance in ending their lives”.

“That’s the furthest any panel or committee has ever gone, and even they excluded dementia”, notes Hyde, and were it not for the Nicklinson case the issue might well have faded from the public’s attention as it has in the USA. According to Jason Karlawish, an expert on bioethics at the University of Pennsylvania, since the furore that surrounded Oregon becoming the fi rst State to legalise assisted suicide 15 years ago, the “issue has really kind of disappeared from the national conversation. And I think, as well, it has disappeared from a lot of the conversations in the professional societies. With respect to dementia it is really rarely discussed.”

Instead, says Karlawish, there has been an increasing focus on trying to apply the clinical skills of palliative care to the care of people with dementia. Palliative medicine has emerged as a specialty of its own in the USA, he explains. “There are now fellowship programmes, there are journals, there’s a specialty society, they have a board certifi cation; little of that existed 15 years ago”. As in the UK, which can claim to be a world leader in palliative care for diseases such as cancer, the conversation about assisted suicide is seen by many as an unwelcome distraction from eff orts to improve end-of-life care across the disease continuum. “I respect that argument”, says Karlawish; “I do think we need to make better progress in treating the symptoms of patients with dementia, so probably it is kind of premature to say ‘oh, and let’s also talk about assisted suicide for these patients’”.

And there are other reasons that assisted suicide more generally is off limits in the USA. “In terms of the politics of the issue, assisted suicide is deeply contentious because in the States the issue of how to care for people who are dying has been bundled into issues around abortion, contraception, stem cell research, and a variety of other medical issues that are captured in this idea of the culture of life”, explains Karlawish.

However, Karlawish does foresee a time when a discussion about assisted death for patients with dementia will become necessary. “My speculation about assisted death for persons with dementia is that I think if we begin to develop drugs that competent professionals agree aff ect the rate of change of the disease, if those kinds of drugs become available and short of a cure people will still progress, I do think that we will start to have a conversation

about at what stage these drugs should be stopped, and I think companion to that conversation will be debates about death with dignity”, he says.

In Oregon—for which annual data for Death With Dignity Act prescription recipients and deaths has been compiled since the legislation was introduced in 1997—the number of Death With Dignity Act deaths has risen slowly from 16 in 1998, to 71 in 2011, accounting for just over two deaths per 1000 total deaths. In every year the three most frequently mentioned end-of-life concerns were a decreasing ability to participate in activities that made life enjoyable, loss of autonomy, and loss of dignity.

The more that the course of a disease can be controlled with drugs, Karlawish explains, the less natural death from that disease seems. As disease-modifying therapies are developed for Alzheimer’s disease, there might, Karlawish says, “be a resurgence of the conversation amongst people along the lines of ‘I took my therapy to try to preserve my quality of life, when the therapy is no longer preserving my quality of life I want to end it and also end the disease’. There’s that sense of control I think that people will feel over something that is no longer natural, but a manipulable problem”.

David Holmes

“...Karlawish does foresee a time when a discussion about assisted death for patients with dementia will become necessary.”

For more on Oregon’s Death With Dignity Act see http://public.health.oregon.gov/ProviderPartnerResources/EvaluationResearch/DeathwithDignityAct/Pages/index.aspx

Living wills enable patients to dictate their health care should they later lose the capacity to make informed decisions

Tek

Imag

e/Sc

ienc

e Ph

oto

Libr

ary

In Context

394 www.thelancet.com/neurology Vol 11 May 2012

“Eelco Wijdicks is the academic neurologist who has contributed the most to the fi eld of critical care over the last three decades” says Alejandro Rabinstein, who has worked with Wijdicks for the past few years in the Neurological Neurosurgical Intensive Care Unit of Saint Marys Hospital in Rochester (MN, USA). “Brain death, coma, subarachnoid haemorrhage, hemispheric ischaemic brain infarctions, massive intra cerebral haemorrhage, neuromuscular respiratory failure, and complications after organ trans plantations are just a few of the topics we understand more and better thanks to his work”, explains Rabinstein. And to that list can now be added the prediction of potential for organ donation in patients with non-survivable brain injury—a subject that Rabinstein and Wijdicks explore in this issue of The Lancet Neurology.

“Neurointensivists need to improve the outcome of critically ill neurologic patients, but they also have the obligation to fi nd the best possible closure, and in some patients this includes organ donation”, says Wijdicks, explaining the rationale for their latest study.

Organ donation is just one of the emotive issues that neurocritical care brings to the fore, many of which can take an emotional toll. “Some days are very rough and the clinical course is hopeless”, says Wijdicks, all of which has to be dealt with through the fi lter of sleep deprivation while on call. But Wijdicks, Rabinstein explains, “always manages to stay faithful to his principles; not only taking good care of patients, but also maintaining research and teaching as much a priority as always”.

The caring part of the equation, at least, comes naturally. Wijdicks was born in 1954, in Leiden, the Netherlands, where his father “was a deeply caring primary care physician”, he recalls. “Our life at home—with patients going in and out—totally adjusted to his practice. He taught me one critical medical principle: that when a patient calls, they need help.” But fi tting in the research and the teaching comes down to what Rabinstein calls Wijdicks’ “iron discipline”.

“When I was starting my fellowship, I asked him how was he capable of producing so much”, Rabinstein recounts. “I was hoping for a clue, a trick, a shortcut. I got an answer that admitted no further questions: ‘Work, work, work’”. “He is indefatigable” says Allan Ropper, one of three trailblazers in neurocritical care—along with Raymond Adams and C Miller Fisher—who made a lasting impression on Wijdicks during his time with them at the Massachusetts General Hospital in Boston (MA, USA). He arrived there in 1988, fresh from Erasmus University Medical Center in Rotterdam, the Netherlands, where an “extraordinarily talented mentor”—Rien Vermeulen—had taught Wijdicks how to see “both sides to a clinical

situation and eff ectively play the Devil’s advocate, and to constantly ask ‘show me the data’”. The Subarachnoid Hemorrhage Study Group of Hans van Crevel, Jan van Gijn, Rien Vermeulen, and Albert Hijdra at Rotterdam brought the subtleties of neurology back into care of patients with subarachnoid haemorrhage, Wijdicks explains, and it was there that he developed his interest in acute neurology.

At Boston, Wijdicks enthuses, he had never seen “so much dedication to patient care and clinical research in one place”. Ropper, he says, “taught me everything I know, showed me how to write eff ectively and how to say it clearly, and to look out for the things you see in the intensive care unit—they might be new observations”. His inquisitiveness and rigour undoubtedly endeared him to the faculty at Boston, qualities that have played no small part in his becoming, in Ropper’s view, “by far the most prolifi c and interesting person in the fi eld of critical care neurology”. But perhaps equally important was the fact that he was such “terrifi c fun to have as a fellow” says Ropper. For Wijdicks, living with his wife and two children next to a turkey farm in the suburbs of Boston and driving a “rusty stalling station wagon that looked like something from National Lampoon’s Vacation”, it was a transformative time; he recalls blissful days driving through “amber waves of grain” and under “spacious skies”.

In 1992, he took up the challenge of setting up the Neurocritical Care programme at the Mayo Clinic, becoming Chair of the Division of Critical Care Neurology in 2003. “Mayo Clinic has guided me to put my life into the care of patients” says Wijdicks; “there were spectacular opportunities to fi nally come to grips with the neurology of critical illness, critical care neurology, how to manage the golden hour in the emergency department, and to develop a programme”.

Wijdicks also likes to take the opportunity, when one presents itself, to share with his colleagues his passion outside medicine: the movies. “He could have been a movie critic in another life” says Rabinstein. “Unfailingly he invites trainees after each rotation for dinner and movies at his house.”

Perhaps, says Ropper, it is because “he is rarely critical of other people in the way many academics can be”, or perhaps, as his clinical fellow Jennifer Fugate says, it’s because his “constant curiosity and scepticism create a fantastic learning environment”, but Wijdicks inspires a rare and genuine warmth of feeling in his colleagues. “Above everything else I can say about him”, says Rabinstein, “he is a wonderful person. As a colleague and friend you can always, and I mean always, count on him”.

David Holmes

Profi leEelco Wijdicks: a neurocritical appraisal

See Articles page 414

www.thelancet.com/neurology Vol 11 May 2012 395

In Context

LifelineKlaus Fassbender has worked at the Department of Rheumatology, University Basel, Switzerland, the Max Planck Institute for Psychiatry, Munich, Germany, the Department of Neurology at the University of Mannheim, Germany, and at the University of Goettingen, Germany. In 2005, he became Chairman of the Department of Neurology, University of the Saarland, Germany, where he was also vice-president of the University in 2007–09, and was co-founder of the German Institute of Dementia Prevention in 2008. He works on defi ning the eff ects of innate immune responses in Alzheimer’s disease and on how to improve acute stroke management.

What do you think is the most neglected topic in science or medicine at the moment?Delivery of state-of-the-art treatment to patients in developing countries.

Which research article has had the biggest eff ect on your work, and why? The article by Lucia Meda and colleagues in Nature in 1995, which further increased my interest in the role of microglia in the pathophysiology of Alzheimer‘s disease.

If you had not entered your current profession, what would you have liked to do?Be a musician.

Who was your most infl uential teacher, and why?Michael Hennerici (University of Mannheim, Germany) and Konrad Beyreuther (University of Heidelberg, Germany), my teachers in clinical and basic neuroscience.

What would be your advice to a newly qualifi ed doctor?Never give up.

How do you relax?Travelling by train.

What, apart from your family, is the passion of your life?The combination of clinical work and research.

What is your greatest fear?War.

What item do you always carry with you?A pen.

If you were Bill Gates, how would you spend your fortune?I would try to ensure that all children in the world have the same chance of a good education.

Do you believe in ghosts? No.

If you knew you had a week to live, how would you live those days? Saying goodbye to the people important to me.

What keeps you awake at night? Thinking about the work in the clinic, sometimes.

Ten most wanted2002–20121 Alzheimer’s disease (Position Paper, August, 2007)

Dubois B, Feldman HH, Jacova C, et al. Research criteria for the diagnosis of Alzheimer’s disease: revising the NINCDS-ADRDA criteria. Lancet Neurol 2007; 6: 734–46.

2 Stroke (Review, January, 2003)Feigin VL, Lawes CMM, Bennett DA, Anderson CS. Stroke epidemiology: a review of population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Lancet Neurol 2003; 2: 43–53.

3 Alzheimer’s disease (Article, March, 2006)Hansson O, Zetterberg H, Buchhave P, Londos E, Blennow K, Minthon L. Association between CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: a follow-up study. Lancet Neurol 2006; 5: 228–34.

4 Alzheimer’s disease (Review, October, 2003)Blennow K, Hampel H. CSF markers for incipient Alzheimer’s disease. Lancet Neurol 2003; 2: 605–13.

5 Vascular cognitive impairment (Review, February, 2003)O’Brien JT, Erkinjuntti T, Reisberg B, et al. Vascular cognitive impairment. Lancet Neurol 2003; 2: 89–98.

6 Parkinson’s disease (Review, March, 2006)Chaudhuri KR, Healy DG, Schapira AHV. Non-motor symptoms of Parkinson’s disease: diagnosis and management. Lancet Neurol 2006; 5: 235–45.

7 Dementia (Review, June, 2004)Fratiglioni L, Paillard-Borg S, Winblad B. An active and socially integrated lifestyle in late life might protect against dementia. Lancet Neurol 2004; 3: 343–53.

8 Vascular dementia (Review, November, 2002)Román GC, Erkinjuntti T, Wallin A, Pantoni L, Chui HC. Subcortical ischaemic vascular dementia. Lancet Neurol 2002; 1: 426–36.

9 Malignant cerebral infarction (Article, March, 2007)Vahedi K, Hofmeijer J, Juettler E, et al. Early decompressive surgery in malignant infarction of the middle cerebral artery: a pooled analysis of three randomised controlled trials. Lancet Neurol 2007; 6: 215–22.

10 Cerebellar ataxias (Review, May, 2004)Schöls L, Bauer P, Schmidt T, Schulte T, Riess O. Autosomal dominant cerebellar ataxias: clinical features, genetics, and pathogenesis. Lancet Neurol 2004; 3: 291–304.

Most highly cited Lancet Neurology articles, according to Scopus.

See Articles page 397

For the article by Lucia Meda and colleagues see Nature 1995; 374: 647–50

www.thelancet.com/neurology Vol 11 May 2012 395

In Context

LifelineKlaus Fassbender has worked at the Department of Rheumatology, University Basel, Switzerland, the Max Planck Institute for Psychiatry, Munich, Germany, the Department of Neurology at the University of Mannheim, Germany, and at the University of Goettingen, Germany. In 2005, he became Chairman of the Department of Neurology, University of the Saarland, Germany, where he was also vice-president of the University in 2007–09, and was co-founder of the German Institute of Dementia Prevention in 2008. He works on defi ning the eff ects of innate immune responses in Alzheimer’s disease and on how to improve acute stroke management.

What do you think is the most neglected topic in science or medicine at the moment?Delivery of state-of-the-art treatment to patients in developing countries.

Which research article has had the biggest eff ect on your work, and why? The article by Lucia Meda and colleagues in Nature in 1995, which further increased my interest in the role of microglia in the pathophysiology of Alzheimer‘s disease.

If you had not entered your current profession, what would you have liked to do?Be a musician.

Who was your most infl uential teacher, and why?Michael Hennerici (University of Mannheim, Germany) and Konrad Beyreuther (University of Heidelberg, Germany), my teachers in clinical and basic neuroscience.

What would be your advice to a newly qualifi ed doctor?Never give up.

How do you relax?Travelling by train.

What, apart from your family, is the passion of your life?The combination of clinical work and research.

What is your greatest fear?War.

What item do you always carry with you?A pen.

If you were Bill Gates, how would you spend your fortune?I would try to ensure that all children in the world have the same chance of a good education.

Do you believe in ghosts? No.

If you knew you had a week to live, how would you live those days? Saying goodbye to the people important to me.

What keeps you awake at night? Thinking about the work in the clinic, sometimes.

Ten most wanted2002–20121 Alzheimer’s disease (Position Paper, August, 2007)

Dubois B, Feldman HH, Jacova C, et al. Research criteria for the diagnosis of Alzheimer’s disease: revising the NINCDS-ADRDA criteria. Lancet Neurol 2007; 6: 734–46.

2 Stroke (Review, January, 2003)Feigin VL, Lawes CMM, Bennett DA, Anderson CS. Stroke epidemiology: a review of population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Lancet Neurol 2003; 2: 43–53.

3 Alzheimer’s disease (Article, March, 2006)Hansson O, Zetterberg H, Buchhave P, Londos E, Blennow K, Minthon L. Association between CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: a follow-up study. Lancet Neurol 2006; 5: 228–34.

4 Alzheimer’s disease (Review, October, 2003)Blennow K, Hampel H. CSF markers for incipient Alzheimer’s disease. Lancet Neurol 2003; 2: 605–13.

5 Vascular cognitive impairment (Review, February, 2003)O’Brien JT, Erkinjuntti T, Reisberg B, et al. Vascular cognitive impairment. Lancet Neurol 2003; 2: 89–98.

6 Parkinson’s disease (Review, March, 2006)Chaudhuri KR, Healy DG, Schapira AHV. Non-motor symptoms of Parkinson’s disease: diagnosis and management. Lancet Neurol 2006; 5: 235–45.

7 Dementia (Review, June, 2004)Fratiglioni L, Paillard-Borg S, Winblad B. An active and socially integrated lifestyle in late life might protect against dementia. Lancet Neurol 2004; 3: 343–53.

8 Vascular dementia (Review, November, 2002)Román GC, Erkinjuntti T, Wallin A, Pantoni L, Chui HC. Subcortical ischaemic vascular dementia. Lancet Neurol 2002; 1: 426–36.

9 Malignant cerebral infarction (Article, March, 2007)Vahedi K, Hofmeijer J, Juettler E, et al. Early decompressive surgery in malignant infarction of the middle cerebral artery: a pooled analysis of three randomised controlled trials. Lancet Neurol 2007; 6: 215–22.

10 Cerebellar ataxias (Review, May, 2004)Schöls L, Bauer P, Schmidt T, Schulte T, Riess O. Autosomal dominant cerebellar ataxias: clinical features, genetics, and pathogenesis. Lancet Neurol 2004; 3: 291–304.

Most highly cited Lancet Neurology articles, according to Scopus.

See Articles page 397

For the article by Lucia Meda and colleagues see Nature 1995; 374: 647–50

396 www.thelancet.com/neurology Vol 11 May 2012

In Context

News in briefTenecteplase versus alteplaseA phase 2B trial has shown that tenecteplase improves clinical out come and perfusion compared with alteplase in acute ischaemic stroke (N Engl J Med 2012; 366: 1099–107). 75 patients were assigned to receive alteplase (0·9 mg per kg bodyweight) or tenecteplase, a genetically engineered mutant tissue plasminogen activator (0·1 mg or 0·25 mg per kg bodyweight). The primary endpoints were reperfusion and clinical improvement after 24 h. Mean reperfusion in patients who received tenecteplase was 79·3% (SD 28·8), versus 55·4% (38·7) in patients who received alteplase (p=0·004). Clinical improvement according to the National Institutes of Health stroke scale was 8·0 (5·5) in the tenecteplase group and 3·0 (6·3) in the alteplase group (p<0·001). Patients who received the high dose of tenecteplase had better perfusion and clinical outcomes at 24 h than did those who received the low dose. The researchers call for a phase 3 trial to test the effi cacy of tenecteplase in a large population.

Laquinimod for MSOral laquinimod modestly reduces relapse rate and might slow disease progression in patients with relapsing-remitting multiple sclerosis (RRMS). In the international phase 3 ALLEGRO trial, 1106 patients with RRMS were randomly assigned to receive

laquinimod 0·6 mg once daily or placebo for 2 years. The primary endpoint, mean annualised relapse rate over 24 months, was 0·30 (SE 0·02) in patients receiving the drug versus 0·39 (0·03) in those who received placebo (p=0·002). The risk of confi rmed disability progression, a secondary endpoint, was 11·1% in the laquinimod group versus 15·7% in the placebo group (hazard ratio for risk reduction 0·64, 95% CI 0·45–0·91, p=0·01). Elevated concentrations of liver enzymes were more frequent in the laquinimod group (N Engl J Med 2012; 366: 1000–09).

Treatments for advanced ADCholinesterase inhibitors, such as donepezil, can be eff ective for treating mild-to-moderate Alzheimer’s disease (AD), but it is unclear whether their benefi ts extend to patients with more advanced disease. Memantine can be benefi cial in patients with moderate-to-severe AD. In the DOMINO trial of donepezil and memantine in moderate-to-severe AD (N Engl J Med 2012; 366: 893–903), 295 patients with moderate or severe AD who were already being treated with donepezil were assigned to continue or discontinue donepezil, with or without starting memantine treatment, for 1 year. Patients who continued taking donepezil showed less cognitive decline on the standardised mini-mental state examination (SMMSE; p<0·001) and on the Bristol activities of daily living scale (BADLS; p<0·001) than did those who discontinued donepezil. Patients taking memantine also scored better on both tests than did those not taking memantine (p<0·001 for SSME; p=0·02 for BADLS), but the combination of both drugs did not produce a greater benefi t than treatment with donepezil alone. These results support the continued use of donepezil in patients with moderate-to-severe AD.

Autoimmune epilepsyAntiepileptic drugs (AEDs) are eff ective in only about half of patients with partial seizures, and the causes of AED-resistant partial epilepsy are often unclear. A new study points to the potential of early-initiated immunotherapy for patients in whom an autoimmune cause is suspected. In 27 patients with neural autoantibody, CSF, or MRI fi ndings suggestive of infl ammation, immuno therapy was tested for the treatment of persistent partial seizures (Arch Neurol 2012; published online Mar 26. DOI:10.1001/ archneurol.2011.2985). Patients received intravenous immune globulin (IVIg), methyl prednisolone (IVMP), or combinations of IVMP, IVIg, plasmapheresis, or cyclophosphamide. After a median of 17 months (range 3–72), seizure frequency had improved in 22 of the 27 patients; 18 were seizure free. The median time from seizure onset to the start of immunotherapy was 4 months for responders and 22 months for non-responders (p<0·05).

Network-based degenerationA new study takes us a step closer to understanding how neurodegenerative diseases spread from region to region to establish a network-based pattern of atrophy. The researchers explored the relationship between healthy brain connectivity, assessed by functional MRI, and neurodegenerative disease vulnerability, assessed by quantifying regional atrophy in patients (Neuron 2012; 73: 1216–27). For each of fi ve distinct neurodegenerative disorders, specifi c regions emerged as network “epicentres” whose connectivity profi les most resembled the disease-associated pattern of atrophy. Theoretical connectivity measures—including shorter functional paths to the disease-associated epicentres—predicted atrophy severity. The fi ndings best fi t a transneuronal-spread model of network-based vulnerability.AJ

Pho

to/S

cienc

e Ph

oto

Libr

ary