NeuroImage 52 (2010) 217–223

Contents lists available at ScienceDirect

NeuroImage

j ourna l homepage: www.e lsev ie r.com/ locate /yn img

DTI fiber tracking to differentiate demyelinating diseases from diffusebrain stem glioma

Carlo Giussani a, Andrew Poliakov a,b, Raymond T. Ferri c, Lauren L. Plawner c, Samuel R. Browd a,Dennis W.W. Shaw a,b, Tanya Z. Filardi a, Corrine Hoeppner d, J. Russell Geyer d, James M. Olson d,James G. Douglas e, Elisabeth H. Villavicencio d, Richard G. Ellenbogen a, Jeffrey G. Ojemann a,⁎a Department of Neurological Surgery, University of Washington School of Medicine, Seattle Children's Hospital, Seattle, WA, USAb Department of Radiology, University of Washington School of Medicine, Seattle Children's Hospital, Seattle, WA, USAc Department of Neurology, University of Washington School of Medicine, Seattle Children's Hospital, Seattle, WA, USAd Department of Hematology-Oncology, University of Washington School of Medicine, Seattle Children's Hospital, Seattle, WA, USAe Department of Radiation Oncology, University of Washington School of Medicine, Seattle Children's Hospital, Seattle, WA, USA

Abbreviations: ADEM, acute disseminated encephaloimaging; CSF, cerebral spinal fluid; FA, fractional anisoinversion recovery; IVIG, intravenous immunoglobulimaging; PPD, principal diffusion direction; WM, white⁎ Corresponding author. Department of Neurological

NE, Mailstop W-7729, Seattle, WA 98105, USA. Fax: +1E-mail address: jeff.ojemann@seattlechildrens.org (J

1053-8119/$ – see front matter © 2010 Elsevier Inc. Adoi:10.1016/j.neuroimage.2010.03.079

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 28 December 2009Revised 20 March 2010Accepted 29 March 2010Available online 2 April 2010

Keywords:DTI fiber trackingDiffuse brainstem gliomaDemyelinating diseases

Object: Intrinsic diffuse brainstem tumors and demyelinating diseases primarily affecting the brainstem canshare common clinical and radiological features, sometimes making the diagnosis difficult especially at thetime of first clinical presentation. To explore the potential usefulness of new MRI sequences in particulardiffusion tensor imaging fiber tracking in differentiating these two pathological entities, we review a series ofbrainstem tumors and demyelinating diseases treated at our institution.Material and methods: The clinical history including signs and symptoms and MRI findings of threeconsecutive demyelinating diseases involving the brainstem that presented with diagnostic uncertainty andthree diffuse intrinsic brainstem tumors were reviewed, along with a child with a supratentorial tumor forcomparison. Fiber tracking of the pyramidal tracts was performed for each patient using a DTI study at the

time of presentation. Additionally Fractional Anisotropy values were calculated for each patient in the ponsand the medulla oblongata.Results: Routine MR imaging was unhelpful in differentiating between intrinsic tumor and demyelination. Incontrast, retrospective DTI fiber tracking clearly differentiated the pathology showing deflection of thepyramidal tracts posteriorly and laterally in the case of intrinsic brainstem tumors and, in the case ofdemyelinating disease, poorly represented and truncated fibers. Regionalized FA values were variable and ofthemselves were not predictive either pathology.Conclusion: DTI fiber tracking of the pyramid tracts in patients with suspected intrinsic brainstem tumor ordemyelinating disease presents two clearly different patterns that may help in differentiating between thesetwo pathologies when conventional MRI and clinical data are inconclusive.

© 2010 Elsevier Inc. All rights reserved.

Introduction

Pediatric brainstem tumors remain a major challenge in neuro-oncological practice, accounting for the 10–20% of pediatric braintumors (Barkovich, 2000). Themajority of brainstem gliomas show aninfiltrative characteristic along the pons white matter fibers and areclassified as intrinsic diffuse glioma (mainly localized in the ventralpons) in comparison to the more localized pattern of growth seen

myelitis; DTI, diffusion tensortropy; FLAIR, fluid attenuationin; MRI, magnetic resonancematter.Surgery, 4800 Sand Point Way206 987 3925..G. Ojemann).

ll rights reserved.

with exophytic brainstem gliomas (Reddy, and Mapstone, 1994).Currently, intrinsic glioma are considered inoperable while a subset ofexophytic brainstem lesions are amenable to surgical resection andcarry a better prognosis (Fisher et al., 2000). The deep localization andthe pattern of growth of intrinsic diffuse brainstem glioma, inter-weaving between normal axons, render this group of tumorsunfavorable to surgery (Epstein and Constantini, 1996; Lesniaket al., 2003) leading to upfront conventional radiotherapy and/orchemotherapy without tissue biopsy (Albright et al., 1993; Cartmilland Punt, 1999; Boviatsis et al., 2001; Wagner et al., 2006; Jallo et al.,2003). Controversy occasionally arises when differentiating betweenan intrinsic brainstem glioma and a demyelinating process as thetreatment modalities for these entities are totally counter to oneanother. In our experience, diagnostic difficulty potentially ariseswhen the clinical course is monophasic and develops over days toweeks rather than acutely andwith variable neurological presentation

218 C. Giussani et al. / NeuroImage 52 (2010) 217–223

for any given pathological entities. Moreover, demyelinating disorderssuch as acute disseminated encephalomyelitis (ADEM) (Hynson et al.,2001) often have multifocal abnormalities on MRI, however isolatedinvolvement has been demonstrated as exemplified by tumefactivemultiple sclerosis (Sagar et al., 1982;McAdam et al., 2002) and variantssuch as Balo concentric sclerosis (Revel et al., 1993; Gharagozloo et al.,1994) which can have a single lesion that mimics a neoplastic process.Reliable radiological technique to differentiate between intrinsictumor and demyelination is critical to offer the proper treatment,especially if a diagnostic biopsy is not obtained prior to treatment.These lesions can be large and often have edema and mass effect(Luchinetti et al., 2008). MRI techniques such as magnetizationtransfer imaging, and MR Spectroscopy have not been able todefinitively distinguish between demyelination and tumor. Progres-sion on serial imaging and in the clinical course can be helpful inmaking a diagnosis; however, a delay in treatment can have significantramification when the lesion involves the brainstem (Enzinger et al.,2005). Schwartz et al. (2006) showed that if the MRI abnormalityshows ring enhancement, the pattern of ring enhancement and T2hypointensity can differentiate between neoplasms, demyelination,and abscesses, but the MR patterns frequently overlap (Schwartz et al.,2006). Moreover, lumbar puncture for spinal fluid analysis is notalways conclusive or even practical at clinical presentation (Link andHuang, 2006; Joseph et al., 2009).

Diffusion tensor imaging (DTI) and white matter fiber tractogra-phy are promising techniques for estimating the course, extent, andconnectivity patterns of the white matter (WM) structures in thebrain (Chen et al., 2007; Engelbrecht et al., 2002). In particular, DTIfiber tracking technique has been shown in pediatric patients to bereliable in distinguishing the involvement of white matter fibers indiffuse pontine glioma and focal brain stem tumors (Helton et al.,2006; Helton et al., 2008; Phillips et al., 2005; Rollins, 2007; Chen etal., 2007). Furthermore, the DTI features of different disordersaffecting central nervous structures with a high concentration ofwhite matter fibers, as demyelinating diseases or traumatic diffuseaxonal injury, have been recently described (Rollins, 2007; Rutgers etal., 2008; Schneider et al., 2003; Yu et al., 2006; Filippi et al., 2001;Engelbrecht et al., 2002).

In fact, the motion of water molecules can be altered by thepresence of structural obstacles at a cellular or subcellular level.Pathologic processes that modify the tissue organization by decreas-ing or increasing the number of barriers to water molecular diffusion,or that alter the permeability of the barriers, cause abnormal waterdiffusivity with a consequent impact on the rendering of the DTI fibertractography of the white matter bundles passing through that area.

The current study investigated the potential DTI fiber trackingpattern differences of diffuse pontine glioma andmidbrain/brainstemdemyelinating processes to understand a reliable role of thistechnique in differentiating these diseases once the clinical andradiological features are not conclusive especially at clinicalpresentation.

We present retrospectively three proven cases of pediatricbrainstem demyelination, initially thought to represent intrinsicglioma showing the difficult differential diagnosis that can accompanythese lesions. A comparison of standard MRI and DTI fiber trackingfindings is made between two of these cases and three cases ofintrinsic diffuse brainstem glioma. In the context of the current study,the recent literature regarding DTI and brain stem tumors in childrenis reviewed.

Material and methods

Patient population

The study was conducted at Seattle Children's Hospital afterobtaining standard IRB approval for a retrospective study review. Of

the approximate 80 brain tumors treated annually at our institution,only a few represent diffuse intrinsic brainstem gliomas. Ourtreatment strategy follows national standards whereby intrinsicbrainstem lesions that appear typical on imaging studies proceed toadjuvant therapy without a diagnostic brain biopsy. Pertinent to thecurrent study, three patients with multifaceted symptomatology andMRI findings atypical for the classic appearance of a diffuse brainstemtumor who were referred to our Hospital over a period of one yearwere evaluated. Given the atypical nature of the original MR imaging,adjuvant therapy was delayed and upon serial imaging, radiographicfeatures changed consistent with what was a final clinical andlaboratory diagnosis of demyelinating diseases.

To understand the potential role of MRI sequences, including DTIfiber tracking, in differentiating a demyelinating disease from anintrinsic diffuse brain stem tumor we retrospectively compared theMRI (three patients) and DTI fiber tracking (two patients) features ofthree cases of demyelinating disease involving the brainstem to threedifferent cases of diffuse brainstem glioma with pathognomonicstandard MRI features, and to a patient with a supratentorial tumorand a normal brainstem as a case control. We specifically analyzed thepyramidal tracts which are commonly involved in the clinicalmanifestation of both tumor progression and pontine demyelination.

DTI and fiber tracking techniques

All patients underwent standard MR imaging of the brain withand without contrast, our current protocols include diffusion tensorsequences. MRI data were acquired on a 3 T Siemens Trio scannerwith a Siemens 8-channel head coil. The DTI parameters were TR/TE=5800/96 ms, b=1000 s/mm2, 10 diffusion directions repeated2–4 times, 1.8 × 1.8 mm in-plane resolution, 3 mm (skip 0.5 mm)slice thickness. Analysis of DTI data was performed using FSL/FDTsoftware (FMRIB Image Analysis Group, University of Oxford,http://www.fmrib.ox.ac.uk/fsl/) and included Eddy current correc-tion, fitting of diffusion tensors and estimation of diffusionparameters including Fractional Anisotropy (FA), Principal DiffusionDirection (PDD), Mean Diffusivity and others. FA has emerged asthe de facto standard for measuring microstructural tissue organi-zation in clinical practice (Basser, 1995). To visualize and evaluatethe tracks, we also used images commonly referred to as color FAmap, which shows color-coded principle diffusion direction modu-lated by FA values. With this approach different primary colors areused to represent the components of the orientation of the fibers(Jones, 2005). Using Color FA maps helps identify adjacent fibertracts that may have similar FA values since they may have quitedifferent direction. One can gain an impression of the trajectory andintegrity of cortico-spinal tract by viewing the direction of thispathway and following it from one slice to the next. In fibertracking or tractography, algorithms can be used to perform asimilar task (Mori and van Zijl, 2002). Tractography was performedusing MedINRIA software package (Asclepios Research Project,http://www-sop.inria.fr/asclepios/software/MedINRIA/). A mini-mum FA value of 0.2 was used as the fiber termination criterion,in agreement with generally accepted practice based on typical FAvalues observed in gray and white matter (Mori and van Zijl, 2002).It should be noted that although we used software developed forresearch purposes and off-line analysis, similar functionality is nowavailable commercially as part software provided by vendors of MRIequipment.

Case illustrations

Normal brainstem

A 15-year-old boy with a history of complex seizures and normalmotor exam was operated on for a left temporal choroid plexus

Fig. 1. Patient without brainstem pathology (temporal lobe tumor) shown for comparison. (A) Sagittal T1 weighted MRI with contrast shows no pathology in the brainstem.(B) Coronal and (C) sagittal images show large corticospinal tracts. The blue color indicates a strong superior–inferior orientation of the large tract running in the brainstem.

219C. Giussani et al. / NeuroImage 52 (2010) 217–223

papilloma (Fig. 1A). He underwent preoperative MRI of the brain withDTI technique. The DTI fiber tracking of the cortico-spinal tractsshowed a very accurate representation of the pyramidal tracts thatrun compactly through the brainstem without any distortion (Fig. 1Band C). The FA was calculated in the pons and in the medullaoblungata showing respectively a value of 0.72 and 0.39.

Demyelination (Cases 1–3)

Case 1 is a 17-year-old girl who presented to another hospitalwith a three week history of double vision, right cranial nerve sixpalsy and rapidly progressing left-side sensory symptoms. A CT-scanwas reported as normal and CSF analysis for oligoclonal bands was

Fig. 2. Brainstem anatomical and DTI images showing demyelinating brainstem lesions. (A) Cmild mass effect, (B) Case 2 with diffuse FLAIR signal change in the lower brainstem (red ccoronal and sagittal planes (see text) and (C) Case 3 with an expanding pontine lesion (redtracking in the corticospinal tracts bilaterally. Compare to the robust tracts evident in the b

negative for a demyelinating disease. Once transferred to ourhospital, she underwent an MRI of the brain that showed a pontinelesion which was initially concerning for a brainstem tumor: theMRI showed a patchy increased T2 signal in the pons and cephaladportion of medulla with some associated decreased T1 signal andscant linear enhancement in the pons. The lesion measured 2 cmtransverse, 2.2 cm anterior–posterior, and 2.6 cm craniocaudal (Fig.2A). No gadolinium enhancement was present. No DTI techniquewas applied. Because of persistent diagnostic uncertainty, a secondlumbar puncture was performed after a new evaluation at thetumor board. The CSF returned positive for oligoclonal bands. Visualevoked potentials did show normal response in the left eye, butthe right eye showed a borderline high latency and decreased

ase 1 with FLAIR (axial) and T2 (coronal) changes in the brainstem (red circle) showingircle) with truncated cortical spinal tracts seen superimposed on the B0 images in thecircle) in FLAIR (axial) and T2 (sagittal) sequences and either truncated or diminishedrainstem tumor cases of Fig. 1.

220 C. Giussani et al. / NeuroImage 52 (2010) 217–223

amplitude. At that point the lesion was treated as a demyelinatingdisease. Along with steroids, she received IVIG with a rapid partialresolution of her symptoms. Even if no DTI fiber tracking techniquewas applied, this case is presented to underline how frequentlypotentially catastrophic misdiagnosis can occur in the scenario ofbrainstem pathology with an acute/subacute clinical presentationand equivocal imaging findings.

Case 2 is a 19-year-old girl who presented with a 4 months historyof progressing weakness of the right side of the face, dizziness,vomiting and lack of gag reflex associated with weight loss. Atadmission she had progressed to developing left sided weakness.

She underwent, at an outside hospital, a lumbar CSF analysis thatwas normal. The patient was transferred to our ICU for thedevelopment of respiratory failure that necessitated intubation. Ahigh dose steroid treatmentwas started. AnMRI of the brain showed acervico-medullary lesion which was initially concerning for a braintumor: the exam showed a mild enlargement of the medulla withnon-enhancing abnormal patchy regions of hyperintense T2/FLAIRsignal involving also the upper cervical spine (Fig. 2B). DTI techniquewas applied. Because the clinical and radiological data were notunequivocally consistent with a diagnosis of brainstem glioma, asecond lumbar puncture was performed. The CSF returned positive foroligoclonal bands. She subsequently started to improve clinically andradiographically.

The DTI fiber tracking showed a severe decrease of the cortico-spinal tracts representation along brainstem with an amputation oftheir medial component without any sign of distortion (Fig. 2B). TheFA was calculated in the pons and in the medulla showing extremelylow values (respectively 0.12 and 0.17), under the software cut-off fortracking.

Case 3 is an 8-year-old female with 4 days of progressivelyworsening headache, diplopia upon extreme upward, downward andleftward gaze, and dizziness. Four days prior to admission, shecomplained of headache and severe nasal congestion as well as a drycough that were treated as a viral illness. At the time of admission toour hospital she presented with a multidirectional nystagmus, a leftfacial numbness, a left ptosis, cerebellar ataxia and dysmetria with apositive Romberg sign.

AnMRI of the brain showedmultifocal findings (Fig. 2C). The FLAIRsequences demonstrated an enlarged pons (2.6×3.6 cm in maximumaxial dimensions) with amild mass effect on the anterior aspect of thefourth ventricle as well as slight effacement of the inferior aspect ofthe basal cisterns. The pons showed a heterogeneous, patchyabnormal high T2 signal intensity with no enhancement aftergadolinium. Supratentorial gray matter also showed multilobar ill-defined patches of abnormal high FLAIR signal intensity with noassociated enhancement. Abnormalities of the thoracic spine wereobserved with abnormal high T2 signal intensity within the spinalcord at T3-T4, apparently affecting the central portion of the cord,with no definite associated abnormal enhancement. No gadoliniumenhancement was present. Though the multifocal appearancesuggested a more diffuse process, the imaging features weresuggestive of either a pontine glioma or acute disseminatedencephalomyelitis and thus non-conclusive.

The CSF findings (elevated IgG and no oligoclonal bands) were notdefinitely suggestive for a demyelinating disease. Further evaluationincluded a renal ultrasound and ophthalmologic exam to excludeTuberous Sclerosis. The patient underwent empirically a 2 week highdose steroids treatment that determined a rapid improvement of thesymptoms as well as a radiological improvement. The clinical andradiological course after the high dose steroids therapywas consistentwith an ADEM.

The DTI fiber tracking showed a severe decrease of the cortico-spinal tracts representation along the brainstem with an amputationof their medial and left components without any sign of distortion ofthe fibers (Fig. 2C). The FA was calculated in the pons and in the

medulla oblungata showing higher values in comparison to theprevious case (respectively 0.44 and 0.5).

Pontine glioma (Cases 4–6)

Case 4 is an 11-year-old boy with a two months history ofworsening headache nausea and vomiting with strabismus anddiplopia. His exam was consistent with an internuclear ophtalmople-gia associated with bilateral hypoacusia, severe balance problems aswell as a dysarthria. The patient presented with an asymmetrichyperreactivity of his deep tendon reflexes.

The MRI of the brain showed the presence of an expansive lesioncentered in the pons with a heterogeneous low T1 signal, and anelevated T2/FLAIR signal. The high T2/FLAIR signal extended from thepons to the left brachium pontes, and superiorly to the dorsalmidbrain and tectum. A cystic area with high T2 signal was present inthe posterior left pons. Postcontrast images demonstrated minimalcontrast enhancement along the periphery of the cystic region (Fig.3A). These MRI features were consistent with a pontine glioma. Thesevere clinical and radiological progression confirmed the oncologicalnature of the disease.

The DTI fiber tracking showed a mild decrease of the representationof the tracts that at the level of the pons were split in the coronal planeand bent posteriorly by the tumor in the sagittal plane (Fig. 3A). The FAwas calculated in the pons and in the medulla oblungata showing lowvalues very closed to the cut-off (respectively 0.27 and 0.24).

Case 5 is an 11-year-old boy with a 5 day history of occipitalheadache and progressive weakness of his left side associated to gaitand balance problems. TheMRI of the brain showed the presence of anexpansive lesion centered in the pons with abnormal T2/FLAIR signalinvolving the mesencephalic structures and the upper part of themedulla. The lesion in the pons measured 4 cm by 3 cm. Post-contrastimages demonstrated a ring-like enhancement of the pons (Fig. 3B).These MRI features were consistent with a pontine glioma. The severeclinical and radiological progression confirmed the oncological natureof the disease.

The DTI fiber tracking showed a strong representation of the tractsthat at the level of the pons were split in the coronal plane and weredistorted posteriorly by the tumor in the sagittal plane (Fig. 3B). TheFA was calculated in the pons and in the medulla oblungata showingvalues close to the normal subject (respectively 0.67 and 0.52).

Case 6 is a 5-year-old girl with a 1 week history of worseningdrooling, choking, slurred speech and gait imbalance, right sideweakness and urinary retention. She underwent an MRI of the brainthat showed a brainstem lesion with a high T2/FLAIR signal. The massmeasured approximately 4.8×3.0×3.8 cm and was involving theadjacent left cerebellar pedunclemeasuring. On sagittal T1 post contrastimaging, there was a small focus of enhancement apparently within thecentral portion of the lesion at the level of the fourth ventricle. Therewas a mass effect and a deformation of the fourth ventricle (Fig. 3C).The MRI findings and the clinical history were consistent with a diffusebrainstem glioma. The severe clinical and radiological progressionconfirmed the oncological nature of the disease.

The DTI fiber tracking showed a mild decrease of the representa-tion of the tracts that at the level of the pons were split in the coronalplane with an amputation on the left side. The tracts appeared bentposteriorly by the tumor in the sagittal plane (Fig. 3C). The FA wascalculated in the pons and in themedulla (respectively 0.43 and 0.39).

Discussion

Summary of the DTI fiber tracking data

Comparing the DTI fiber tracking data of the pyramidal fibers of thecontrol subject to the cases of diffuse brainstem tumors and the

Fig. 3. (A) Case 4, (B) Case 5, and (C) Case 6 showing expansive pontine lesions (first column – FLAIR and second column – post-contrast T1 images) and intact, but displaced,corticospinal tracts (third and fourth column). Tracts are displayed superimposed on the B0 images with the tumor evident on the sagittal view (red arrows).

221C. Giussani et al. / NeuroImage 52 (2010) 217–223

demyelinating disease, two different patterns of fiber representationare recognizable.

The patients with demyelinating diseases show an extremepaucity of the pyramidal fibers that can be truncated. However, thefibers are located in their normal anatomical position and are notdistorted. The FA values in these patients can be extremely low, underthe cut-off value.

The patients with diffuse brainstem gliomas show a slight decreasein the representation of the pyramidal fibers. The fibers are alwaysdistorted by the tumor mass in both the sagittal and coronal views.The pyramidal tracts are in fact bent posteriorly by the tumors in thesagittal view and are split and pushed laterally in the coronal view.The FA values in these patients can be close to normal or low withouttypically being under the cut-off value.

Discussion of the clinical and radiological data

A differential diagnosis between a diffuse brainstem tumor and ademyelination process localized in the brainstem carries importanttherapeutic and prognostic implications.

In fact patients with diffuse brainstem gliomas die of the diseasewithin 18 months of diagnosis, despite contemporary radiationtherapy and chemotherapy (Barkovich, 2000; Jallo et al., 2003).Demyelinating diseases, such as MS, ADEM and transverse myelitis,can affect the same category of patients but their prognosis, ifrecognized promptly, can be favorable due to currently availabletherapies, even in the presence of an extremely severe clinicalpresentation. The current diagnosis for diffuse brainstem tumors ismade clinically and by MR imaging due to the risks associated withstereotactic biopsy (Jallo et al., 2003; Cartmill and Punt, 1999;Boviatsis et al., 2001; Wagner et al., 2006). Unfortunately, demyelin-

ating diseases purely involving the brainstem can present with similarclinical and radiological features as diffuse brainstem tumors.

The similarity seen on imaging can delay treatment and thepotential for inappropriate treatment due to misdiagnosis is also aconcern. In the three recent cases referred to our pediatric oncologicservice the clinical and MRI appearance was sufficiently worrisomefor brainstem tumor that these patients were considered forradiotherapy while waiting for follow up imaging recommendedbecause of uncertainty in the diagnosis.

The recognition of a demyelination process depends on an aware-ness of this as a diagnostic possibility (Love, 2006). From a clinical pointof view, demyelinating diseases affecting the brainstem can mimic adiffuse glioma presenting with a polyhedric neurologic scenario due tothe complex neuroanatomical and neurophysiological architecture ofthe brainstem. A suggested difference between the two pathologicentities is the timingof thedevelopment of the symptoms: patientswithdemyelinating diseases typically present with a clinical history that isshorter (days to few weeks) than diffuse brainstem tumors (severalweeks to months). Nevertheless, the present series of patients confutesthis opinion: in fact, two brainstem tumor patients presented a clinicalhistory of 5 days (case 5) and, in the other, 1week duration (case 6) andtwo demyelinating patients presented with longer clinical histories –

one 3 weeks (case 1) and one of 4 month duration (case 2).CSF analysis has been reported to be sometimes inaccurate to

establish a diagnosis of demyelinating diseases (Link and Huang,2006; Selviaridis et al., 2007; Brinar, 2004; Schwartz and McCormick,2000). The present series confirms these findings. In fact in two casesof demyelinating disease the first CSF analysis was negative initiallyorienting the tumor board team to a diagnosis of brainstem tumor.

In the present study, the pattern of representation of the fibers inthe demyelinating diseases can be interpreted as the effect of the

222 C. Giussani et al. / NeuroImage 52 (2010) 217–223

demyelination on the water molecules diffusivity that is decreasedin the context of the disruption the myelinated fibers. This decreaseof the diffusivity negatively affects the DTI fiber tracking technique.It is important to emphasize that the meaning of this phenomenonis that the pyramid tracts fibers may still be in their normal locationbut the DTI technique cannot detect them. In fact, the results of theapplication of the DTI fiber tracking technique are a function of thecut-off value of the FA. The cut-off value of the FA is the result of acompromise between the capability to detect as much fibers aspossible and to obtain a consistent fiber tracking rendering. Theabsence of a bundle in the tracking technique rendering does notimply that the tract is anatomically destroyed. Nevertheless, usingthe same FA value, the presence of different DTI fiber trackingpatterns in two pathological entities is a function of how thepathological process affects the fibers. The effect of demyelinationon the water molecules diffusion determines in our study a specificDTI fiber tracking pattern at an established FA cut-off value that, ifconfirmed by larger numbers, could be a pathognomonic radiolog-ical data for demyelinating diseases in comparison to intrinsicglioma and vice versa. Unfortunately, in the present series ofdemyelinating diseases only the third case showed in theconventional FLAIR images an enlarged pons with just a mildmass effect on the anterior aspect of the fourth ventricle and aslight effacement of the inferior aspect of the basal cisterns. For thisreason it is not possible to infer the eventual DTI fiber trackingpattern of a demyelinating disease with an important tumefactiveaspect: paucity/truncation of the fibers versus their displacement asin brainstem glioma.

On the other hand, the pattern of representation of the fibers in thediffuse brainstem tumors can be explained by the mass effect that thetumor exercises on the fibers. Besides the distortion of the fibers, thetumor produces a decrease of the fibers representation due to adecrease of the diffusivity. In fact a feature of glioma cells is toinvading the normal tissue by migrating along the white matter fibers(Bello et al., 2004). This mechanism of tumor invasion can affect thediffusivity of the water molecules determining the mild decrease ofthe pyramid fibers represented by the DTI technique.

Limits of the present study

The current study uses DTI/tractography to distinguish demyelin-ating disease from a diffuse brainstem tumor in a pediatric populationthat is prone to both pathologic entities. The two different patterns ofthe DTI fiber tracking of pyramidal tracts representation for diffusebrainstem tumors and demyelinating diseases, even if quite appealingand explainable from a physiopathologic standpoint, need to befurther validated by a larger series study.

In fact, the major limit of this study is the small number of casesthat prevent a consistent generalization of the results in the clinical–radiological work-up of brainstem lesions of uncertain nature in thepediatric population. The present series suggests advantages to thedevelopment of DTI fiber tracking as a diagnostic tool in the scenarioof potential confusion between a demyelinating disease and anintrinsic brainstem tumor.

The other limit of this study is technical. DTI fiber tracking issoftware and operator dependent. This is a retrospective study ofavailable clinical data and, as such, based on the routine clinical DTIscans (some of them as short as 2 min) with only 10 diffusiondirections, not necessarily optimized for evaluation of a particularregion. These were not intended to perform complete, rigoroustractography studies (e.g. trace cortical-spinal tracts from the spinalcord to the cortex). It would be useful to confirm these results with alarger number of patients and by using different DTI software andparameters to eventually establish a standard MRI/DTI protocol inclinical scenario similar to those presented here.

Conclusion

MRI is a vital tool in identifying CNS abnormalities in children withnew onset focal neurologic deficits. However, in many cases thedifferential diagnosis includes neoplastic and demyelinating diseases,and the evaluation process can be problematic when the abnormalityis in the brainstem, and biopsy is felt unsafe. In this settingradiographic inference of a specific diagnosis without confirmativebiopsy can lead to chemotherapy and radiation that may beinappropriate.

DTI fiber tracking of the pyramidal pathways through thebrainstem shows in the present study a different pattern ofrepresentation of the tracts that are postero-laterally dislocated incase of diffuse glioma and are less represented or even amputated inthe case of demyelinating diseases.

When conventional MRI findings are equivocal for these twoentities, we demonstrate that evaluating the DTI fiber tracking of thepyramidal tract can help to differentiate between diffuse brainstemglioma and demyelinating lesion. After a confirmation of these datawith a larger study, this distinction could guide the clinical decisionmaking, potentially limit the need for diagnostic biopsy in thissituation and more accurately determine the appropriate course oftherapy.

References

Albright, A.L., Packer, R.J., Zimmerman, R., Rorke, L.B., Boyett, J., Hammond, G.D., 1993.Magnetic resonance scans should replace biopsies for the diagnosis of diffuse brainstem gliomas: a report from the Children's Cancer Group. Neurosurgery 33,1026–1029.

Barkovich, A.J., 2000. Pediatric neuroimaging, In: Barkovich, A.J. (Ed.), Intracranial,Orbital, and Neck Tumors of Childhood, 3rd ed. Lippincott Williams & Wilkins,Philadelphia, Pa, pp. 462–470.

Basser, P., 1995. Inferring microstructural features and the physiological state of tissuesfrom diffusion-weighted images. Nucl. Magn. Reson. Biomed. 8, 333–344.

Bello, L., Giussani, C., Carrabba, G., Pluderi, M., Costa, F., Bikfalvi, A., 2004. Angiogenesisand invasion in gliomas. In: Kirsch, M., Black, P.M.c.L. (Eds.), Angiogenesis in BrainTumors. : Cancer Treat. Res. Kluwer Academic Publisher, Boston.

Boviatsis, E.J., Voumvourakis, K., Goutas, N., Kazdaglis, K., Kittas, C., Kelekis, D.A., 2001.Stereotactic biopsy of brainstem lesions. Minim. Invasive. Neurosurg. 44, 226–229.

Brinar, V.V., 2004. Non-MS recurrent demyelinating diseases [Review] Clin. Neurol.Neurosurg. 106, 197–210.

Cartmill, M., Punt, J., 1999. Diffuse brain stem glioma: a review of stereotactic biopsies.Childs. Nerv. Syst. 15, 235–237.

Chen, X., Weigel, D., Ganslandt, O., Buchfelder, M., Nimsky, C., 2007. Diffusion tensorimaging and white matter tractography in patients with brainstem lesions. ActaNeurochir. 149, 1117–1131.

Engelbrecht, V., Scherer, A., Rassek, M., Witsack, H.J., Mödder, U., 2002. Diffusion-weighted MR imaging in the brain in children: findings in the normal brain and inthe brain with white matter diseases. Radiology 222, 410–418.

Enzinger, C., Strasser-Fuchs, S., Ropele, S., Kapeller, P., Kleinert, R., Fazekas, F., 2005.Tumefactive demyelinating lesions: conventional and advanced magnetic reso-nance imaging. Mult. Scler. 11, 135–139.

Epstein, F., Constantini, S., 1996. Practical decisions in the treatment of pediatricbrainstem tumors. Pediatr. Neurosurg. 24, 24–34.

Filippi, M., Cercignani, M., Inglese, M., Horsfield, M.A., Comi, G., 2001. Diffusion tensormagnetic resonance imaging in multiple sclerosis. Neurology 56, 304–311.

Fisher, P.G., Breiter, S.N., Carson, B.S., Wharam, M.D., Williams, J.A., Weingart, J.D., et al.,2000. A clinicopathologic reappraisal of brain stem tumor classification. Identifi-cation of pilocystic astrocytoma and fibrillary astrocytoma as distinct entities.Cancer 89, 1569–1576.

Gharagozloo, A.M., Poe, L.B., Collins, G.H., 1994. Antemortem diagnosis of Baloconcentric sclerosis: correlative MR imaging and pathologic features. Radiology191, 817–819.

Helton, K.J., Phillips, N.S., Khan, R.B., Boop, F.A., Sanford, R.A., Zou, P., et al., 2006.Diffusion tensor imaging of tract involvement in children with pontine tumors.AJNR. Am. J. Neuroradiol. 27, 786–793.

Helton, K.J., Weeks, J.K., Phillips, N.S., Zou, P., Kun, L.E., Khan, R.B., et al., 2008. Diffusiontensor imaging of brainstem tumors: axonal degeneration of motor and sensorytracts. J. Neurosurg. Pediatr. 1, 270–276.

Hynson, J.L., Kornberg, A.J., Coleman, L.T., Shield, L., Harvey, A.S., Kean, M.J., 2001.Clinical and neuroradiologic features of acute disseminated encephalomyelitis inchildren. Neurology 56, 1308–1312.

Jallo, G.I., Freed, D., Roonprapunt, C., 2003. Current management of brainstem gliomas.Annals Neurosurg. 3, 1–17.

Jones, D.K., 2005. Fundamentals of diffusion MR imaging. In: Gillard, J., Waldmann, A.,Barker, P.B. (Eds.), Clinical MR Neuroimaging. Diffusion, Perfusion and Spectros-copy. Cambridge University Press, pp. 54–85.

223C. Giussani et al. / NeuroImage 52 (2010) 217–223

Joseph, F.G., Hirst, C.L., Pickersgill, T.P., Ben-Shlomo, Y., Robertson, N.P., Scolding, N.J.,2009. CSF oligoclonal band status informs prognosis in multiple sclerosis: a casecontrol study of 100 patients. J. Neurol. Neurosurg. Psychiatry. 80, 292–296.

Lesniak, M.S., Klem, J.M., Weingart, J., Carson, B.S., 2003. Surgical outcome followingresection of contrast-enhanced pediatric brainstem gliomas. Pediatr. Neurosurg.39, 314–322.

Link, H., Huang, Y.M., 2006. Oligoclonal bands in multiple sclerosis cerebrospinal fluid:an update on methodology and clinical usefulness [Review] J. Neuroimmunol. 180,17–28.

Love, S., 2006. Demyelinating diseases. J. Clin. Pathol. 59, 1151–1159.Luchinetti, C.F., Gavrilova, R.H., Metz, I., Parisi, J.E., Scheithauer, B.W., Weigand, S., et al.,

2008. Clinical and radiographic spectrum of pathologically confirmed tumefactivemultiple sclerosis. Brain 131, 1759–1775.

McAdam, L.C., Blaser, S.I., Banwell, B.L., 2002. Pediatric tumefactive demyelination: caseseries and review of the literature. Ped. Neurol. 26, 18–25.

Mori, S., van Zijl, P.C., 2002. Fiber tracking: principles and strategies—a technical review.NMR. Biomed. 15, 468–480.

Phillips, N.S., Sanford, R.A., Helton, K.J., Boop, F.A., Zou, P., Tekautz, T., 2005. Diffusiontensor imaging of intraaxial tumors at the cervicomedullary and pontomedullaryjunctions. Report of two cases. J. Neurosurg. 103 (6 Suppl), 557–562.

Reddy, A., Mapstone, T.B., 1994. Brainstem tumors. In: Bernstein, M., Berger, M.S. (Eds.),Neuro-Oncology: The Essentials. Thieme Medical, New York, pp. 352–362.

Revel, M.P., Valiente, E., Gray, F., Berges, C., Degos, J.D., Brugieres, P., Concentric, M.R.I.,1993. patterns in multiple sclerosis. Report of two cases. J. Neuroradiol. 20, 252–257.

Rollins, N.K., 2007. Clinical applications of diffusion tensor imaging and tractography inchildren. Pediatr. Radiol. 37, 769–780.

Rutgers, D.R., Toulgoat, F., Cazejust, J., Fillard, P., Lasjaunias, P., Ducreux, D., 2008. Whitematter abnormalities in mild traumatic brain injury: a diffusion tensor imagingstudy. AJNR Am. J. Neuroradiol. 29, 514–519.

Sagar, H.J., Warlow, C.P., Sheldon, P.W.E., Esiri, M.M., 1982. Multiple sclerosis withclinical and radiological features of cerebral brain tumor. J. Neurol. Neurosurg.Psychiatry 45, 802–808.

Schneider, J.F., Il'yasov, K.A., Boltshauser, E., Hennig, J., Martin, E., 2003. Diffusion tensorimaging in cases of adrenoleukodystrophy: preliminary experience as a marker forearly demyelination? AJNR Am. J. Neuroradiol. 24, 819–824.

Schwartz, T.H., McCormick, P.C., 2000. Non-neoplastic intramedullary pathology.Diagnostic dilemma: to Bx or not to Bx. Review. J. Neurooncol. 47, 283–292.

Schwartz, K.M., Erickson, B.J., Luchinetti, C., 2006. Pattern of T2 hypointensityassociated with ring enhancing brain lesions can help to differentiate pathology.Neuroradiology 48, 143–149.

Selviaridis, P., Zountsas, B., Chatzisotiriou, A., Zaraboukas, T., Gerdemeli, A., 2007.Demyelinating plaque imitates an intramedullary tumour. Clin. Neurol. Neurosurg.109, 905–909.

Wagner, S., Warmuth-Metz, M., Emser, A., Gnekow, A.K., Sträter, R., Rutkowski, S., et al.,2006. Treatment options in childhood pontine gliomas. J. Neurooncol. 79, 281–287.

Yu, C.S., Lin, F.C., Li, K.C., Jiang, T.Z., Zhu, C.Z., Qin, W., et al., 2006. Diffusion tensorimaging in the assessment of normal-appearing brain tissue damage in relapsingneuromyelitis optica. AJNR Am. J. Neuroradiol. 27, 1009–1015.