TheRoleofMagneticResonanceImaginginPredictingSurgicalOutcomeinPatientswithDegenerativeCervical

MyelopathyBy

AriaNouri,MD

AthesissubmittedinconformitywiththerequirementsforthedegreeofMasterofScience

InstituteofMedicalScienceUniversityofToronto

©CopyrightbyAriaNouri(2015)

ii

TheRoleofMagneticResonanceImaginginPredictingSurgicalOutcomeinPatientswithDegenerative

CervicalMyelopathy

AriaNouri,MDMasterofScience

InstituteofMedicalScienceUniversityofToronto

2015

AbstractMagnetic resonance imaging (MRI) is commonly used to confirm a diagnosis of

degenerative cervicalmyelopathy (DCM), however, its role in predicting outcome

following surgical treatment remains unclear. Herein, it is hypothesized that

quantitative preoperative MRI analysis can predict surgical outcome. Using data

derivedfromaprospectiveandmulticentercohortofpatientsundergoingsurgical

treatment for DCM, multiple MRI parameters were assessed, including

presence/absenceofsignalchangeonT1andT2,T2signalquantitativefactors,and

anatomical measurements. A dichotomized postoperative modified Japanese

orthopedic association score at 6‐monthswas then used to characterize patients

with mild myelopathy (≥16) and those with substantial residual neurological

impairment (<16). Using multivariate logistic regression analysis, the results

indicate that assessment of maximum canal compromise, T1 signal change and

baseline functional severity provide superior prognostic value than baseline

functional severity alone. It is therefore concluded that MRI has a statistically

significantroleinpredictingsurgicaloutcome.

iii

AcknowledgementsThesuccessofanacademicisoftenafunctionofhissurroundingsandthosewho

haveprovidedthesupportandmeanstopermitthepursuitoftheirscientific

enquiry.ThishasbeennodifferentformeandthusIwouldliketoacknowledgethe

graduatecoordinators,particularlyDrs.HowardMountandCindiMorshead,aswell

asthestaffattheInstituteofMedicalSciencefortheirsupportandguidance.I

wouldalsoliketoacknowledgemyprogramadvisorycommitteemembers,Drs.

AlbertYee,AileenM.DavisandDavidMikulisfortheirvaluableinputanddirection.

Aswell,Iwouldliketorecognizethemanymembersoftheclinicalandbasic

scienceresearchteam,withparticularacknowledgementtoLindsayTetreault,Dr.

JuanZamorano,Dr.KristianDalzell,andDr.AnoushkaSinghfortheircontributions

tothisthesis.Lastly,Iwouldliketoextendmygreatestgratitudetomymentorand

thesissupervisor,Dr.MichaelG.Fehlings,forhissupportaswellasguidance,and

withoutwhomtheculminationofthisworkwouldnothavebeenachievable.

AlessonfromananecdoteIshallneverforget:

Indiscussingthestrugglesofresidentsinmanagingacomplexsacralfractureina

homelesswoman,Dr.Fehlingsremindedhisresidents,“Youshouldfeelhonoredto

takecareofher,becauseitisaprivilegetoservethemostvulnerableinour

society”.

Idedicatethisworktomyparents,mybrotherandmysister,whohavebeenmy

strongestsupportersandmygreatestteachers.

AcknowledgmentofFundingIwouldliketoacknowledgetheUniversityofTorontofortheirfundingsupportvia

theUniversityofTorontoOpenFellowshipAwardaswellasTravelGrant.Additionally,IwouldliketoacknowledgeAOSpineforfundingthecollectionofthe

studydatausedtoconducttheresearchinthisthesis.

iv

ContributionsAriaNouriPrimaryinvestigatorofallMRimaging.Conceptualizedstudydesignandstructuredmethodology.Interpretedresults,drafted,reviewed,andfinalizedallrelevantresearchcontained.Solewriterofthisthesis.MichaelG.FehlingsMentorandthesissupervisor.Providedsignificantguidance,contributions,andconceptualdesignforallpartsofthisthesis.LindsayTetreaultContributionsinchapter1forthesectiononepidemiologyofDCMandtable1.4.Significantcontributionstochapter3intermsofstatisticalanalysis,draftingandreviewingcontents,figuresandtables.KristianDalzellSignificantcontributionstoMRIanalysisandquantitativeMRImethodology.Furthercontributionsincludedraftingandreviewingofcontentinchapter3.JuanJoseZamoranoSignificantcontributionstoMRIanalysisandquantitativeMRImethodology.Furthercontributionsincludedraftingandreviewingofcontentinchapter3.AnoushkaSinghSpecificcontributionsrelatedtoepidemiologyandetiologyofDCMinchapter1.SpyridonKaradimasSpecificcontributiontothediscussionofthepathobiologyofspinalcordcompressioninchapter1.MadeleineO'HigginsSignificanteditorialsupportforvariousworkswithinthisthesis.ProgramAdvisoryCommittee‐AlbertYee,AileenM.DavisandDavidMikulisProvidedsignificantguidanceandassistanceforapproachingthesisaims.Providedexperienceandconceptualsupport,mostnotablyforchapter3.

v

TableofContents Abstract iiAcknowledgements iiiContributions ivTableofContents v‐viiListofFigures viiiListofTables ixDescriptionofAbbreviations x‐xi

Preamble 1‐2

CHAPTER1–DegenerativeCervicalMyelopathy:AComprehensiveReviewofLiterature 3‐78

1.1. Introduction 4‐51.2. AnatomyoftheSpine 5‐8

1.2.1. AnatomyoftheSpinalCord 6‐81.2.2. AnatomyoftheCervicalSpine 8

1.3. DegenerativeCervicalMyelopathy(DCM) 9‐231.3.1. CervicalSpondyloticMyelopathyandDegenerativeDisc

Disease 11‐41.3.2. Ossification,Hypertrophy,andCalcificationofSpinalCanal

Ligaments 15‐71.3.3. PathobiologyofSpinalCordCompression 18‐201.3.4. DynamicInjuryMechanismsandDegenerative

Spondylolisthesis 21‐21.3.5. DegenerativeCervicalSpineDeformityandMyelopathy 22‐3

1.4. EpidemiologyofDCM 24‐301.4.1. EstimationofIncidenceandPrevalenceofDCMfromNon‐

TraumaticSpinalCordInjuryData 281.4.2. EstimationofDCMProgressionRatesfromStudieson

OsteoarthritisintheSpine 291.4.3. EstimationbyHospitalAdmissionRates 29‐301.4.4. Survival 30

1.5. RiskFactorsAssociatedwithDCM 31‐441.5.1. HereditaryandGeneticcauses 31‐7

1.5.1.1. CervicalSpondyloticMyelopathyandDegenerativeDiscDisease 33‐4

1.5.1.2. OssificationofthePosteriorLongitudinalLigament 35‐6

1.5.1.3. OssificationoftheLigamentumFlavum 37

vi

1.5.2. Occupational,PersonalandOtherAssociations 38‐441.5.2.1. PhysicalTransportationofGoods 38‐91.5.2.2. CervicalSpineDegenerationinHighPerformance

AviatorsandAstronauts 39‐401.5.2.3. CervicalSpineDegenerationinAthletes 40‐11.5.2.4. PersonalFactors 41‐21.5.2.5. MotorVehicleAccidentInjuryandDCM 42‐31.5.2.6. OtherAssociations 43‐4

1.5.3. CongenitalDisordersassociatedwithDCM 45‐91.5.3.1. Down’sSyndrome 451.5.3.2. Klippel‐FeilSyndrome 46‐71.5.3.3. CongenitalSpinalStenosis 481.5.3.4. OtherCongenitalAssociations 49

1.6. ClinicalAssessment 50‐631.6.1. ClinicalAssessmentToolsforEvaluatingDiseaseSeverity 50‐31.6.2. MedicalImaging 54‐63

1.6.2.1. ConventionalRadiography 54‐51.6.2.2. ComputerizedTomography 56‐71.6.2.3. ConventionalMRI 58‐611.6.2.4. AdvancedMRItechniques 62‐3

1.7. TreatmentofDegenerativeCervicalMyelopathy 64‐701.7.1. Non‐operativeManagement 64‐51.7.2. SurgicalManagement 65‐81.7.3. PharmacologicalTreatment 69‐70

1.7.3.1. Riluzole:APotentialTherapeuticAdjuncttoSurgicalManagement 70

1.8. Prognosis:PredictingSurgicalandNon‐SurgicalOutcome 71‐81.8.1. PredictorsofMyelopathyDevelopmentinpatientswithCervical

SpinalCordCompression,CanalStenosisand/orOPLL 72‐31.8.2. ComparisonofOutcomesinNon‐operativeManagementvs.

SurgicallyManagedDCMPatients 731.8.3. ClinicalPredictorsofOutcomesofSurgicalTreatment 74‐51.8.4. ImagingPredictorsofSurgicalOutcome 75‐8

CHAPTER2–Hypothesis,ResearchObjectives,&Rationale79‐80

CHAPTER3–TheRoleofQuantitativeMRIAnalysisinPredictingSurgicalOutcomeinPatientswithCervicalSpondyloticMyelopathy 81‐101

3.1.Introduction 823.2.MaterialsandMethods 83‐92

3.2.1.StudyDataandDesign 83‐4

vii

3.2.2.QuantitativeMRIanalysis 85‐63.2.3.T2‐WIMCCandMSCC 87‐83.2.4.T2‐WISignalMeasurementsandSignalChangeRatio 89‐903.2.5.PrimaryClinicalSeverityMeasure 913.2.6.StatisticalAnalysis 91‐2

3.3.Results 93‐73.4.Discussion 98‐1013.5.Limitations 101

CHAPTER4–GeneralDiscussion,ClinicalImplications,andKnowledgeTranslation 102‐119

4.1. KeyFindings 1034.2. BasisforMRIAnalysis 103‐44.3. InterpretationandMeaningofMRIFindingsof

DegenerativeChanges 104‐64.4.ApproachandRationalefortheUseofMRI

MeasurementTechniques 106‐114.5. CreatingaMultivariateModelforPredictingOutcome 111‐24.6. ClinicalImplications 113‐44.7. KnowledgeTranslation:BringingResearchIntoPractice114‐6

4.7.1.Identifyingknowledgeusers 114‐54.7.2.Disseminatingtheknowledge 115‐64.7.3.Optimizingknowledgetranslation 116

4.8. Limitations 1174.9.Conclusion 118‐9

CHAPTER5–FutureDirections 120‐7

5.1.Introduction 1215.2. AssessmentofReliabilityandExternalValidity 1215.3. CombiningthePredictiveCapacityofImagingandClinical

Parameters 121‐25.4. AnApproachforDefiningT1andT2WeightedMRISignal

ChangesQuantitatively 122‐45.5. AdvancedMRITechniquesandtheFutureofDCM

Assessment 125‐75.5.1.Dynamic/Kinematic&UprightMRI 125‐65.5.2.AdvancedMRItechniques:DiffusionTensorMRI,BOLD,andNuclearMagneticResonance(NMR)Spectroscopy 126‐7

REFERENCES 128‐143APPENDIX 144

viii

ListofFigures

Figure1.1–AconceptualbreakdownofDegenerativeCervicalMyelopathypathoetiologicalconstituents.Figure1.2–AT2‐weightedMRIdepictinggeneraldegenerativechangesofthecervicalspineinapatientwithconfirmedDCM.Figure1.3–AnartisticdepictionofthemultipleanatomicalchangesthatmaypresentinthecervicalspineofpatientswithDCM.Figure1.4–LigamentouschangescommonlyobservedonmedicalimagingofpatientswithDCM.Figure1.5–MRIsandplainradiographsof3patientswithKFS.Figure1.6–Aradiographshowingthecanaltovertebralbodyratiomethodforevaluationofcervicalspinestenosis.Figure1.7–Measurementofthemaximumcanalcompromise(MCC)ofthecervicalspineonCT.Figure1.8–T1andT2weightedMRIsofthesamepatientwiththepresenceofsignalchangesandaconfirmeddiagnosisofDCM.Figure3.1–Aconsortdiagramindicatingtheclinicalandimagingdataavailableatbaselineanduponfollow‐upat6‐months.Figure3.2–MSCCandMCCmeasurementtechniquesillustratedonT2‐WIMR.Figure3.3–QuantitativesignalchangemeasurementsillustratedonT2‐WIMR.Figure3.4–Areceiveroperatorcharacteristiccurveofthefinalmodelbasedon3predictorsofoutcome.Figure5.1–AconceptualillustrationofhowT1‐WIandT2‐WIsignalchangecouldbedefinedquantitatively.

ix

ListofTables

Table1.1–Ascendingpathwaysandthesensoryinformationthattheymediate.Table1.2–Descendingpathwaysandthemotorfunctiontheymediate.Table1.3–Keypathobiologicalchangesoccurringinthesettingspinalcordcompressionsecondarytodegenerativechangesinthecervicalspine.Table1.4–Trendsofpatientcharacteristicsandregionalincidencesofthevariousformsofdegenerativepathologies.Table1.5–SummaryofgeneticresearchconductedonCSM,DDD,OPLLaswellasOLF.Table1.6–AdescriptionofthemodifiedJapaneseOrthopedicAssociationscore(mJOA)scoreandhowitistabulated.Table1.7–NurickGradefortheassessmentofmyelopathyseverityTable1.8–CorrelationbetweenMRIcharacteristicandpathobiologicalchangesaswellascurrentthoughtsrelatedtostructurerecoverypotentialafterdecompressionsurgery.Table1.9–AlistofconservativetherapeuticapproachesforthemanagementofDCM.Table1.10–AlistofcommonlyperformedanteriorandsurgicalproceduresforthetreatmentofDCM.Table1.11–Evidence‐basedrecommendationsregardingthecomparativeeffectivenessofanteriorandposteriorsurgicaltechniques.Table1.12–AsummaryofrecentreviewsthathaveevaluatedtheroleofMRIcharacteristicsinpredictingsurgicaloutcomeinpatientstreatedforDCM.Table3.1–Thetableoutlines,describesanddiscussesthevariousquantitativemeasuresevaluatedinthepatientpopulation.Respectivevalidationofthemethodshasalsobeendescribed.Table3.2–GeneralandMRIcharacteristicsofpatientsforwhichMRIanalysiswasconducted.Table3.3–UnivariateandmultivariaterelationofMRIparametersandbaselinemJOAwithadichotomizedmJOAoutcome(≥16,<16)at6months.

x

DescriptionofAbbreviations

AOSpine‐NA AOSpine‐NorthAmericaAIC AkaikeInformationCriterionASP AdjacentSegmentPathologyAUC AreaUndertheReceiverOperatingCharacteristicCurveBIC BayesianInformationCriterionBOLD Blood‐Oxygen‐LevelDependentBSCB BloodSpinalCordBarrierCLF CalcificationofLigamentumFlavumCSF CerebrospinalFluidCSM CervicalSpondyloticMyelopathyCT ComputerizedTomographydtMRI DiffusionTensorMRIDCM DegenerativeCervicalMyelopathyDDD DegenerativeDiscDiseasedkMRI DynamicKinematicMRIDSL DegenerativeSpondylolisthesisDS Down’sSyndromeEDS Ehlers–DanlosSyndromefMRI FunctionalMRIGz G‐Force,Gravitational‐ForceKFS Klippel‐FeilSyndromeICC IntraclassCorrelationCoefficientIVD IntervertebralDiscLF LigamentumFlavumMCC MaximumCanalCompromisemJOA ModifiedJapaneseOrthopedicAssociationscoreMR MagneticResonanceMRI MagneticResonanceImagingMSCC MaximumSpinalCordCompressionNMR NuclearMagneticResonanceMVA MotorVehicleAccidentntSCI Non‐TraumaticSpinalCordInjuryOA OsteoarthritisOLF OssificationofLigamentumFlavumOPLL OssificationofPosteriorLongitudinalLigamentPLL PosteriorLongitudinalLigamentpMRI Upright‐NeutralPositionMRIROI RegionofInterest

xi

SCI SpinalCordInjurySCR SignalChangeRatioSEP SensoryEvokedPotentialsSI SignalIntensitySN SpinalNerveTA TransverseAreaT1‐WI T1‐WeightedImageT2‐WI T2‐WeightedImageVDR VitaminDReceptor

1

PREAMBLELetusbeginbyconsideringaclinicalvignette.

Mr.Smith isahealthy68‐year‐oldCaucasianmalewhohasrecentlyretired

after long career as a carpenter. He frequents his family physician regularly and

prides himself for working hard to stay in shape and living a well‐balanced life.

Upon his recent visit for a regular check‐up, he was noted to walk with less

confidence, using a wider gait pattern, and requiring the use of the handrail for

takingafewstepsupthestairs.Despiteacknowledgingthesefindings,Mr.Smithdid

not complain of any problems. His medical history has been otherwise

unremarkable, with the exception of a diagnosis of urinary retention due to

suspected benign prostatic hypertrophy at his last check‐up; however, a digital

rectalexamwasunremarkable.Uponquestioning,heindicatedthathehasbecome

increasinglybotheredbyneckpainthatbeganapproximately5yearsagoandthatis

becoming less manageable with over‐the‐counter painkillers. He stated that he

madethechoicetoretire,asheno longer feltcomfortablehandlingtoolsandthat

thishadimpactedthequalityofhiswork.Healsostatedthathefeltthatthenatural

agingprocesshascaughtupwithhimandthathewishedtoenjoytheremainderof

hislife.Concernedoverhisprecipitousdeteriorationfromhislastvisit2yearsago,

when this constellation of findings were absent, the family physician ordered

imagingoftheneck.Amagneticresonanceimagingexamwasremarkablefordiffuse

degenerative changes in the cervical spine, with severe narrowing of the spinal

canal and frank compression of the spinal cord. Specifically, the report indicated

that therearesignal changesonT2‐weighted imaging in thespinal cordspanning

two vertebral levels in height. On the basis of these findings a referral to a spine

surgeonwasarranged.

WhileMr.Smith isa fictionalcharacterandonlyservestoexemplifyacase

scenario, such a presentation in the elderly is very real; indeed, spinal cord

compressiondue to degenerative changes in the spine represents the commonest

causeof spinal corddysfunction in thispopulation segmentworldwide.From this

casewe see that the disability arising from cervical spinal cord compression can

2

significantly impact thequalityof life. Ifwe lookevencloser,wecanalsosee that

early symptoms indicative of degenerative cervical myelopathy (DCM), such as

problems voiding, began before the patient deteriorated to his current level and

remainedundetected–aproblemthatcurrentlyprevails foranumberofreasons,

includingagenerallackofawarenessinthemedicalcommunitytoscreenforsigns

ofDCMandattributionoffunctionaldeclinetothenaturalagingprocessbypatients.

ForMr.Smith,animportantdecisionwillhavetobemadeafterconsultation

with a spine surgeon. Conservative management has limited utility in managing

mildDCM,but this cannotbe saidwithmoreadvanced levelsofdisease,which is

moreconsistentwithhiscase.ItisthuslikelythatMr.Smithwillberecommended

to undergo surgery to decompress the spinal cord. At this point, one important

question ariseswhen considering the possibility of surgery: Doctor,what aremy

futureprospectsofhealthaftertreatment?

Althoughonewouldhope that theanswer to thisquestion shouldbe fairly

clear‐cut,itisnot.Inrecognitionofthis,oneoftheobjectivesoftherecentAOSpine‐

North American prospective and multicenter study on cervical spondylotic

myelopathywastoclosethiscriticalknowledgegap.Indeed,notlongago,aclinical

predictionmodel derived from the AOSpine study data emerged andwill help to

addressMr.Smith’squestion.However,thismodelwasbasedonclinicalparameters

anddidnotfullyconsidertheroleofMRIanalysis.Accordingly,itistheobjectiveof

thisthesistoexploretherolethatpreoperativeMRIplaysinpredictinghowpatients

suchasMr.Smithcanexpecttofareaftersurgicaltreatment.

Asour study represents theonlyprospectiveandmulticenter investigation

on the subject to date, the findings herein provide the highest level of evidence

currentlyavailableonthesubjectandthereforehavedirectimplicationsoncurrent

practice. Consequently, it is the ultimate objective of this thesis to provide

informationthatwillhelpanswerMr.Smith’squestion.

3

CHAPTER1DegenerativeCervicalMyelopathy:

AComprehensiveReviewofLiterature

Thecontentsofsections1.3‐1.5ofthischapterderivefromthemanuscript:

Nouri A, Tetreault L, Singh A, Karadimas SK, Fehlings MG. Degenerative CervicalMyelopathy:Epidemiology,Genetics, andPathogenesis. Spine (PhilaPa 1976). 2015Jun15;40(12):E675‐93.

4

1.1IntroductionDegenerative CervicalMyelopathy (DCM) is an age‐related disease of the cervical

spine and represents one of themost common causes of spinal cord dysfunction

worldwide(TracyandBartleson,2010).Thoughgeneticfactorsmaybeimplicated

in its development, it is generally recognized that the degenerative process is a

function of two fundamental factors: advancing age, and the accumulative wear‐

and‐tearonthecervicalspineanatomy.Asthesetwofactorsoccurinallhumans,it

is quite understandable that a large portion of the elderly population, whether

clinicallymyelopathicornot,willpresentwithevidenceofstructuralchangesofthe

cervical spineonmedical imaging (TracyandBartleson,2010).Fromadiagnostic

perspective, it has been particularly challenging to determinewhen degenerative

changes should be considered significantly pathological and treated. Those who

have symptoms due to spinal cord compression, present clinically with variable

degrees of motor and sensory deficits. However, waiting for the onset of

myelopathic symptoms to emerge before treatment is initiatedmay only prevent

further deterioration and result in irreversible neurological deficits. Conversely,

prophylactictreatmentmayresultinunnecessaryriskstothepatient.Cervicalspine

surgery isan invasiveprocedurewithsignificant risksof injury to thespinal cord

anddeathattheextremeendofthespectrum,despitethefactthatitisacommonly

performedprocedurethathasbeenconsistentlyperformedwithoutsuchsequelae.

ToguidethemanagementofDCM,clinicalandimagingpredictorsofsurgical

outcome may be investigated to understand more about how these patients can

expecttofarewhentreatmentisinstigated.ThespecificroleofpreoperativeMRIin

this regard, which is the central role of this thesis, is further explored in the

following chapters. However, in beginning this discussion, it is appropriate to

provide a comprehensive background on various topics surrounding DCM to set

context.

In the following sections of this chapter, the anatomy and natural

degeneration of the cervical spine, as well as DCM pathogenesis, epidemiology,

clinicalpresentation,functionalassessment,medicalimaging,treatment,andfinally,

5

currentknowledgeonpredictorsofsurgicaloutcomewillbeexplored.Collectively,

thesesubjectswillsetthebasisofcurrentknowledgesurroundingDCM.

1.2AnatomyoftheSpine

The spine is the centralmusculoskeletal structure in thehumanbody, connecting

theheadandupper limbssuperiorly, to thepelvisand lower limbs inferiorly.The

typical spine is composedof 24articulatingvertebrae that are further segregated

into 7 cervical, 12 thoracic and 5 lumbar vertebrae (Moore, 2009). In the adult,

thesesitontopofanother9fused,or“false”,sacralandcoccygealvertebrae.Asthe

spinedescendscaudally, thevertebrae increase in size to support theprogressive

increaseinweightbearing(Moore,2009).

Though the spine is often referred to as a column, it is actually shaped in

curves,withanatural lordoticpredisposition inthecervicaland lumbarsegments

andkyphosisinthethoracicregion.

Allnon‐fusedvertebrallevelsofthespineareconnectedtothenextlevelby

anintervertebraldisc(IVD).IVDsarecomposedofacentralnucleuspulposusthatis

comprised of gelatinous like tissue, and a surrounding annulus fibrosus,which is

composed of several layers of fibrous cartilage. However, the cervical, thoracic,

lumbar,aswellasthefirstsacralvertebraealsoformsynovialzygapophysialjoints

via the inferior articularprocessof superior vertebraewith the superiorarticular

process of the inferior vertebrae.With exception to the above, it should benoted

thatC1articulatesatitssuperiorarticularsurfacewiththeoccipitalcondylesatthe

baseoftheskull.

Thoughoneofthecorefunctionsofthespineistoprovidestructuralsupport

andmobilityof the trunk, italsoconfersprotectionof thespinalcord,whichruns

through the spinal canal. The spinal canal is demarcated by the posterior

longitudinal ligament (PLL) anteriorly, and by the vertebral arches, vertebral

foraminaaswellasligamentumflavalaterallyandposteriorly(Moore,2009).

6

1.2.1AnatomyoftheSpinalCord

The spinal cord is an extension of the brain that descends through the foramen

magnumattheinferioraspectoftheskullandcontinuescaudallythroughthespinal

canal,commonlyterminatingatL1‐L2inadults.Duringitsdescent,thespinalcord

gives rise to 31 spinal nerves (SN) that exit the spine through intervertebral

foraminabilaterally.EachoneoftheseSNsiscomprisedofadorsalandventralroot,

constitutingprimarilysensoryinputsandmotoroutputs,respectively.

Grossly,thespinalcordvariesinsizeandwidenslaterallyinthecervicaland

lumbosacralregions.Theseenlargementsareattributabletotheincreasednumber

of lower motor neurons that constitute the origins of the nerves of the brachial

plexusoftheupperextremitiesandthelumbosacralplexusofthelowerextremities

(Waxman,2009).

On transverse‐section, the spinal cord shows anH‐shaped internalmassof

grey matter that is surrounded by white matter (Waxman, 2009). Located at its

center is thecentralcanal,whichcontainsCSFand is linedbyependymalcells.As

withthebrain,thespinalcordislinedatitsperipherybythemeninges,comprised

frominsidetooutside,thepia,arachnoidandduralayers,respectively.

Withinthewhitematter,therearefunctionallyrelatedbundlesofaxonsthat

formtracts,whicharefurtherbrokendownintoascendinganddescending(Rosset

al., 2003). The ascending tracts represent the pathways that take sensory

information from the body to the cerebrum or cerebellum. Disturbances in their

functiondue tomyelopathy can result in the lossof sensationandproprioception

below the level of injury.Most of these pathways decussate or cross to the other

sideofthespinalcordastheytravelrostral intheneuroaxis.Table1.1providesa

listofthemajorascendingpathwaysandthesensoryinformationthattheymediate.

Descending tracts of the spinal cord carry autonomic as well as voluntary

motorfibersfromthebraintothevisceraandmusclesofthebody.Disturbancesin

their function due to myelopathy will result in problems with motor control

necessary for both voluntary and involuntary movement, as well as movement

7

coordination below the site of injury. Table 1.2 presents a list of the major

descendingpathwaysandthemotorfunctiontheymediate.

Thegraymatterofthespinalcordissegregatedintothreehornsorcolumns:

Dorsal (Posterior)Horn,Lateral (Intermediate)HornandVentral (Anterior)Horn.

Within these, nerve bodies that are functionally related group together to form

nuclei called Rexed laminae (Ross et al., 2003, Fix, 2001). Posterior horn gray

matterreceiveandprocesssensoryinputs,intermediaryhorngraymatter(present

only inthoraxandlumbarregion)receiveviscerosensory input,andanteriorhorn

graymattercontainpredominatelymotornuclei(Fix,2001).

Table1.1Ascendingpathways and some of the sensory information that theymediate (Fix,2001,Blumenfeld,2010).

AscendingTracts Mediates

MedialLemniscusPathway Tactilediscrimination,vibration,formrecognition,jointmusclesensationandconsciousproprioception

VentralSpinothalamicTract Lighttouch(perceptionfromgentlecottonstrokes)

LateralSpinothalamicTract TemperatureandpainsensationDorsalSpinocerebellarTract Unconsciousproprioceptiontothecerebellum

Finecoordinationofpostureandmovementofindividualmusclesofthelowerextremity

VentralSpinocerebellarTract Unconsciousproprioceptiontothecerebellum Coordinatedmovementandpostureofentirelowerextremity

CuneocerebellarTract Upper‐extremityequivalentofthedorsalspinocerebellartract

SpinoreticularTract Emotionalandarousalaspectsofpain

8

Table1.2Descending pathways and some of themotor information that theymediate (Fix,2001,Blumenfeld,2010).

1.2.2AnatomyoftheCervicalSpine

Theanatomyof thespinevariesbetween thecervical, thoracic, lumbarandsacral

divisions,butcanalsodiffersignificantlywithintheseregions.Thecervicalportion

is composed of 7 segments, which comprise the smallest of the articulating

vertebrae. The cervical region is also naturally lordotic and gives rise to 8 SNs

between the cervical vertebrae, with the exception of the first SN, which arises

betweentheoccipitusandC1.Asthereare7cervicallevelsbut8SNsarisingoutof

thecervicalregion,SNsarenumberedstartingat1accordingtothecorresponding

vertebrae levelbelowtheirorigination,except forSN8,whicharisesbetweenC7‐

T1.Thefirst2cervicalvertebrae,theatlasandtheaxis,aremoreuniquelyshaped

thantherestoftheircounterpartsastheybridgetheconnectionbetweenthehead

andthespine.

Anatomical variations between cervical levels and their articulations with

adjacent vertebrae are related to function andpermit varyingdegreesofmobility

between segments.This is important, asdislocations are relativelymoreprone to

occurinregionsofgreaterflexibility.

DescendingTracts Mediates

LateralCorticospinal(pyramidal)pathway

Somaticandvisceralmotoractivitiesarisingfromthecerebralcortexorbrainstem

VentralCorticospinalTract Controlofaxialmuscles

RubrospinalTract Controlofflexortone

VestibulospinalTract Controlofextensortone,balanceandpositioningoftheheadandneck

DescendingAutonomicTracts Projectionstosympathetic(T1‐L3)andparasympathetic(S2‐S4)centersinspinalcord

Innervationofciliospinalcenter(T1‐T2)

9

1.3DegenerativeCervicalMyelopathy

Degenerationoftissueanywhereinthebodyoccursprincipallyasafunctionofuse

intensityovertime.Musculoskeletalstructuresthatbearsignificantstructuralloads,

however, may experience accelerated deterioration. In the cervical spine, these

degenerative changes canbedivided into spondylotic (orosteoarthritic) andnon‐

osteoarthritic changes,with further subtype categorization (Figure1.1).However,

while these pathological changes have been segregated into separate clinical

entities, there is a loose divergence between them in practical terms, as they are

highlyinterrelatedandoftentimesmanifestconcomitantly.Ultimately,theprincipal

unifying problem is the propensity of degenerative changes to cause spinal canal

stenosis, leadtocompressionofthespinalcord,andeventually,resultindisability

duetothedevelopmentofmyelopathy.

Pathophysiologically, symptomatic degenerative cervical myelopathies can

result from static compression of the spinal cord, spinalmalalignment leading to

altered cord tension and vascular supply, and repeated dynamic injury owing to

segmentalhypermobility(BaptisteandFehlings,2006a,Amesetal.,2013).Interms

of the latter, it has also been recognized thatunstable spine segments may be

responsible for chronic repetitive microtrauma upon the spinal cord not

significantly largeenoughtoberecognizedas traumaticSCI.Additionally,patients

withspinal stenosiswhohaveexperiencedminor trauma to theneckmaybeata

substantial risk of developing myelopathy or experiencing deterioration of

preexistingmyelopathy (Yoo et al., 2010). Accordingly, some degree of trauma is

likelytobecontributorytothenaturalhistoryofDCMdevelopment.

In the following section, the pathology of CSM and DDDwill be discussed

together, and separated from the discussion of the ossification, hypertrophy and

calcificationof spinal ligaments.Thecurrentunderstandingof thepathobiologyof

spinalcordcompressionwillthenbediscussedasaconsequenceofthese.Following

this,degenerativespondylolisthesis(DSL)andevidencesupportingdynamicinjury

mechanisms are briefly reviewed. Finally, emerging evidence to support cervical

spinemalalignmentasacontributortoDCMpathogenesiswillbediscussed.

10

Figure1.1AconceptualbreakdownofthepathoetiologicalconstituentsofDegenerativeCervicalMyelopathy(DCM).Predominantfeaturesformthebasisofdiagnosis,butmorethanoneofthesepathologiesfrequentlypresentconcomitantly.Inaddition,congenitalanomaliesmaypredisposetoacceleratedmanifestationofDCM.CSS=CongenitalSpinalStenosis,KFS=Klippel‐Feilsyndrome,DS=Down’ssyndrome,OPLL=Ossificationofposteriorlongitudinalligament;OLF=Ossificationofligamentumflavum.

11

1.3.1CervicalSpondyloticMyelopathyandDegenerativeDisc

Disease

Eachintervertebraldisc(IVD)iscomprisedofanouterlayer,theannulusfibrosus,

and an inner layer called the nucleus pulposus. The nucleus pulposus is largely

comprisedofproteoglycanswhich,duetotheirhydrophilicnature,becomeencased

inwatermolecules.Theresultingviscoelasticnucleuspulposusconvertsaxial load

into hoop stress and is contained by the dense connective tissue of the annulus

fibrosuswhich serves toprovide structural integrity to the IVD.This construction

provides a unique capacity to withstand high compressive loads and transfer of

these forces longitudinally onto the cartilage endplates of vertebral bodies and

radially onto the surrounding annulus fibrosus (Galbusera et al., 2014). With

repeated use of everyday living, periods of trauma and excessive use, and other

nutritionalandenvironmentalfactors,theintervertebraldiscsbegintodegenerate

and the uncovertebral processes of the vertebrae become flattened, altering the

weight‐bearingandload‐transferringfunctionsoftheintervertebraljoint(Baptiste

andFehlings,2006b).Asaresult,thereisincreasedstressonthearticularcartilage

endplates and hypermobility in adjacent segments (Baptiste and Fehlings, 2006b,

Galbusera et al., 2014). Uneven pressure forces exerted upon the vertebrae as a

consequenceofthesestructuralchangesarethoughttoencouragetheformationof

osteophyticspursinanadaptiveremodelingprocessmeanttostabilizeanunstable

spinesegment(Galbuseraetal.,2014).

As DDD generally precedes the onset of osteophytosis (Karadimas et al.,

2013b),severecasesofdiskdegenerationthatresultinsignificantmigrationofdisk

elementsintothecanalcanbesolelyimplicatedinthedevelopmentofmyelopathy

(Jacobs et al., 2011). However, more subtle forms of disk degeneration that are

accompanied by progressive changes in the anatomical architecture of entire

cervicaljoint,asisfrequentlyobservedintheelderly,presentwithaconstellationof

findingsmoreconsistentwithCSM.AT2‐weightedMRIofapatientwithCSMand

12

DDDshowingthesechangesispresentedinFigure1.2.Additionally,thesechanges

arefurtherillustratedinFigure1.3.

13

Figure1.2AT2‐weightedMRIdepictinggeneraldegenerativechangesofthecervicalspineinapatientwithconfirmedDCM.Thecervicalvertebrabodies,mostprominentlyseenincervicalvertebra“4”,havelosttheirheightandhaveincreasedanteriortoposteriorlength. Intervertebraldisc elementshavemigrated in the spinal canalC3‐C4 (thinarrow) as well as C4‐C5 and are contributing to spinal cord compression.HyperintensitysignalchangeofthespinalcordisalsoclearlyvisibleextendingfromC3‐C5(thickarrow).

14

Figure1.3AnartisticdepictionofthemultipleanatomicalchangesthatmaypresentinthecervicalspineofpatientswithDCM.ConceptualdesignbyAriaNouri,editsbyMichaelG.FehlingsandmedicalillustrationbyDianaKryski.SeeAppendix.

15

1.3.2 Ossification,HypertrophyandCalcificationofSpinalCanal

Ligaments

Age‐related changes to spinal ligaments implicated in thedevelopmentof cervical

myelopathy involve theposterior longitudinal ligament (PLL)and the ligamentum

flavum (LF). While hypertrophy and ossification of spinal ligaments may be

influencedsignificantlybyunderlyinggenetics, theirmanifestation inolderageas

well as presumed multifactorial pathogenesis suggests a progressive and

degenerativecausefortheirdevelopment(Ogataetal.,2002,Inamasuetal.,2006,

Stapletonetal.,2011,Stetleretal.,2011,Kudoetal.,2011,Hayashietal.,1997).

Stiffening and buckling of the LF has been suggested to occur as a

consequence of natural cervical joint degenerative changes, including loss of IVD

height(Kalsi‐Ryanetal.,2013,BaptisteandFehlings,2006b).Similarly,prolapseof

thenucleuspulposushasbeenimplicatedasaninitiatingstimulusforhypertrophy

ofPLL(Inamasuetal.,2006).Althoughthissuggeststhatstiffeningorhypertrophy

maydevelop as a consequenceof cervical jointdegeneration, the recognition that

theirprevalencevariesbetweenspinalregions[OLFismorecommoninthethorax

(Guo et al., 2010)] and the frequentmanifestationwithout signs of DDD or CSM,

supports additional pathomechanistic processes in their development. A T2‐

weightedMRI of spinal cord compression due to hypertrophy of the ligamentum

flavumisdisplayedinFigure1.4A.

It remains unclear how a normal ligament transitions into an ossified

ligament, but it has been suggested that hypertrophy may be a precursor for

ossification (Inamasu et al., 2006). Histologically, hypertrophy of PLL has been

definedas“hyperplasiaofnonfibrouscartilagewithstrongmetaplasiainthePLLin

more than two intervertebral spaces,withcapillaryhyperplasiaand infiltrationof

the connective tissue” (Inamasu et al., 2006). It has been hypothesized that

hypertrophy of PLL may be replaced by lamellar bone to become ossified when

hypertrophy has potential systematic or genetic factors to induce secondary

ossification (Mizuno et al., 2001). The role of genetic factors is reinforced by the

frequentcoexistenceofcervicalandthoracicOPLLandOLFdisease(referredtoas

16

tandem ossification) in which different segmental stresses still exhibit similar

pathologicalchangesinnearlyone‐thirdofpatients(Parketal.,2008).Thisfinding

is also interesting as it suggests that although OPLL and OLFmay have different

naturalhistories,theysharesomeformofpathologicalrelationship(Inamasuetal.,

2006, Ono et al., 1999, Yoshida et al., 1992). Figure 1.4B displays a CT scan of a

patientwithOPLL.

It has been stated that calcification of the ligamentum flavum (CLF) is a

commonfindingonCT(Meyeretal.,2011);however,muchremainsunknownabout

this clinical entity. Given the limited amount of research on CLF, it has been

challenging to definitively classify its common pathological features; however,

reports have supported different histopathology fromOLF (Miyasaka et al., 1983,

Inoue et al., 2013). CLF has been described as occurring in degenerated and

thickened ligaments and that the calcification has no continuity with the lamina

(Miyasakaetal.,1983).Moreover,thesuperficialanddeeplayersoftheligamentum

flavumarerelativelypreserved (Miyasakaetal.,1983).This is incontrast toOLF,

which has been described as a metaplastic process in which endochondral

ossification leads to lamellar bone formation (Ben Hamouda et al., 2003). More

specifically, Miyasaka et al. (1983) state that in OLF cartilage forms an ossified

bridgethatextendsfromtheupperandloweredgesoftwoadjacentlaminae.

Reports indicatethatCLFmaybemorecommoninfemalesandthat itmay

be a manifestation of pseudogout (calcium pyrophosphate dihydrate deposition

disease) (Meyeretal.,2011,Miyasakaetal.,1983,Cabreetal.,2001).However,a

number of other potential associations have also been noted, including

hypercalcemia, hyperparathyroidism, hemochromatosis and renal failure (Ruiz

Santiagoetal.,1997).Despitethesereports,thereremainsapaucityofresearchon

thesubjectandfurtherinvestigationisnecessary.

17

Figure1.4LigamentouschangescommonlyobservedonmedicalimagingofpatientswithDCM.A)AT2‐WIMRIofapatientwithLFhypertrophythatisresultinginsignificantposteriorcompressionofthespinalcord(shortarrow).B)Asagittalreconstructedcomputerizedtomography(CT)imageofapatientwithOPLL(longarrows).

18

1.3.3PathobiologyofSpinalCordCompression

Incomparisontotraumaticformsofspinalcordinjury(SCI),thepathophysiological

sequelaeofSCI inDCMhavebeenconsiderably less investigated.Havingsaid this,

recent studies onCSM rodentmodels (Karadimas et al., 2013d, Klironomos et al.,

2011)aswellasonhumanspinalcords(Yuetal.,2011)ofpatientswithCSMhave

uncoveredanumberofunderlyingphenomenarelated tochronicandprogressive

compression and have implicated chronic stretching in the development of

important pathophysiological events. In particular, it has been demonstrated that

chronic cervical spinal cord compression results in chronic reduction of the

intraparenchymalspinalcordbloodflow(Karadimasetal.,2013d,Karadimasetal.,

2014). The resulting chronic ischemic injury, in conjunction with mechanical

stretch, has been found to activate some key biological events and cause neural

degeneration.

There is emerging evidence supporting that chronic intraparenchymal

ischemia generates a unique immune response (Beattie and Manley, 2011,

Karadimas et al., 2013c). Persistent activation of microglia and macrophage

accumulationatthesiteofthecompressionisthemainknowncomponentsofthis

neuroinflammatory reaction (Karadimaset al., 2013b,Yuet al., 2011,Moonet al.,

2014). However, the role of microglia/macrophage activation under the chronic

compression of the cervical spinal cord has yet to be fully elucidated. Hirai et al

(2013)demonstratedthatboththeM1andM2microgliaphenotypeareactivatedin

CSM,indicatingthatmicrogliamaybeinvolvedinpromotingneuraldamageandin

the repairingprocess. Interestingly,Moonet al (2014)demonstrated that riluzole

attenuatedtheneuropathicpaininarodentmodelofCSMthatwasassociatedwith

decreased levels of microglia activation. The neuroinflammatory story becomes

more complex since it has recently been shown that chronic endothelial cell

dysfunction and blood spinal cord barrier (BSCB) disruption are occurring even

duringthelatestagesofthecompression(Karadimasetal.,2013d),phenomenathat

inevitably exaggerate the existing inflammatory process being propagated by

peripheralimmunecellinfiltration.

19

Thispersistenthypoxicinsultandtheongoingneuroinflammatoryresponse

duringchronicspinalcordcompressionmaybeimplicatedinprogressiveneuronal

andoligodendroglialcelldeaththroughactivationofapoptoticpathways(Yuetal.,

2011,Yuetal.,2009,Kalsi‐Ryanetal.,2013,Karadimasetal.,2013b,Karadimaset

al.,2010).Indeed,thelevelsofneuronalandoligodendroglialinjuryhavebeenwell

associated with neurobehavioural dysfunction. As previously indicated by

histological studies on human and experimental CSM tissue, the compression‐

mediated activation of the pathophysiological cascades described previously can

lead to development of gliosis, cavity formation, degeneration of the main

corticospinaltracts,interneuronalloss,andatrophyoftheanteriorhornsassociated

withmotoneuronal loss, theconstellationofwhich isconsistentwiththeprincipal

neuroanatomical features ofmyelopathy (Karadimas et al., 2013a). A summary of

thekeypathobiologicalchangesisoutlinedinTable1.3.

20

Table1.3Keypathobiologicalchangesoccurringinthesettingspinalcordcompressionsecondarytodegenerativechangesinthecervicalspine(Karadimasetal.,2014,Yuetal.,2011,Karadimasetal.,2013d,Moonetal.,2014).

Pathobiologicalconsequencesofspinalcordcompression Chronicreductioninspinalcordbloodflow Disruptionofthemicrovascularnetwork:

EndothelialcelldysfunctionDisruptionofthevascularbasementmembraneLossofblood‐spinalcordbarrierintegrity

Glutamateexcitotoxicity Microgliaactivationregulatedby:

CX3CR1:CX3CL1axisCD200:CD200Raxis

Neuronalandoligodendroglialapoptosis Demyelination Astrogliosis Axonaldegeneration

21

1.3.4DynamicInjuryMechanismsandDegenerative

Spondylolisthesis

Inadditiontostretchingandstaticcompression,itisalsorecognizedthatthespinal

cordmaybe injured throughdynamic injurymechanisms(Matsunagaetal.,2002,

Hayashi et al., 2014, Fengbin et al., 2013, Fujiyoshi et al., 2010). It has been

proposed that dynamic injury may occur through instability (Jiang et al., 2011),

increased range of motion (Fujiyoshi et al., 2010, Matsunaga et al., 2002), and

throughminortraumainthesettingofpreexistingDCM(Fengbinetal.,2013).

Degenerative spondylolisthesis (DSL) results in regional instability of the

spine that can manifest as an anterior or posterior subluxation of a vertebral

segment and is due to a number of degenerative changes, including facet joint

arthropathy,discdegeneration,and increasedstretchofdiscsaswellas ligaments

(Jianget al., 2011,Chaputet al., 2013).DSLoccursmost frequentlybetweenC3‐4

andC4‐5,representing46%and49%ofobservedoccurrences,respectively(Jiang

et al., 2011). It hasbeen suggested that this regionmaybepreferentially affected

due to greater degenerative changes occurring in the lower cervical spine,which

resultinexcessivecompensatorymobilityintheuppercervicalsegments(Kawasaki

etal.,2007).Althoughthereisnocleargradingsystemfordelineatingseverity,itis

generallyacceptedthatpatientswith3‐3.5mmofhorizontaltranslationhavesevere

DSL(Kawasakietal.,2007,Suzukietal.,2013).Whensevereandunstable,DSLcan

resultinnarrowingofthespinalcanal(Jiangetal.,2011)andpotentiallyrecurrent

compressionofthespinalcord.Indeed,arecentsystematicreviewhasreportedthat

myelopathyormyeloradiculopathywaspresentin64%ofDSLpatientsreferredfor

surgery (Jiang et al., 2011). However, static imaging can make it challenging to

discovermovement dependent subluxation and therefore patientswith suspected

DSLmay benefit from kinematic MRI examination. Hayashi et al (2014) recently

reportedthatspinalcordcompressioninthecervicalspineofsymptomaticpatients

(neckpainwithorwithoutneurologicalsigns)wasobservedin5.3%intheneutral

position,butthatmisseddynamicstenosiswasdiscoveredin8.3%inextensionand

1.6%inflexionusingkinematicMRI.

22

Ithasalsobeensuggestedthatanincreasedrangeofmotioninthecervical

spinemayresult indynamic injurytothecord.Matsunagaetal (2002)previously

showed thatOPLLpatientswith critical stenosis (<6mm) all hadmyelopathy and

thatnomyelopathywaspresentinpatientswithcanaldiameters≥14mm.However,

when spinal canal size was >6mm but less <14mm in diameter, myelopathy

preferentiallydevelopedinthosewithincreasedrangeofmotion(Matsunagaetal.,

2002). Given these results, the authors suggest that static compression cannot be

solelyimplicatedinmyelopathydevelopmentandhypothesizethatdynamicinjury

mechanismsplayarole.

Recently,ithasbeenshownthatdynamicinjurytothespinalcordmayalso

occur in CSM patients in the setting of minor trauma. Fengbin et al (2013) has

shownthatpatientswithminortrauma,particularlyinthesettingoflowercervical

instability, had a greater incidence of neurological deterioration and experienced

lesspostoperativeimprovementthanthosewithouttrauma.

It isclear,however,thatfurtherresearchondynamicinjurymechanismsin

thesettingofadegeneratedcervicalspineiswarranted.

1.3.5DegenerativeCervicalSpineDeformityandMyelopathy

Thecervicalspinehasanatural lordoticdisposition,andage‐relateddegeneration

may result in hyperlordosis, scoliosis, and most notably kyphosis. Indeed, the

cervicalalignmentinDCMpatientsmustbetakenintoconsiderationduringsurgical

planning, particularly to prevent the development or progression of kyphosis

postoperatively. Interestingly, however, there is emerging evidence that cervical

alignmentalsocontributestothepathogenesisandseverityofcervicalmyelopathy

(Ames et al., 2013, Mohanty et al., 2015, Smith et al., 2013). Ames et al (2013)

hypothesize that themechanismbehindmyelopathydevelopment inpatientswith

kyphosis can be attributed to the deformity forcing the spinal cord against the

vertebralbodies.They further state that this results inanterior cordpathologyas

well as longitudinal cord tension due to the cord being tethered by the dentate

ligaments and cervical nerve roots (Ames et al., 2013). Ultimately, spinal cord

tethering is believed to increase intramedullary pressure that results in neuronal

23

loss,demyelination,anddecreasedbloodsupplyduetotheflatteningofsmallblood

vessels(Amesetal.,2013).Thesefindingsindicatethatspinalcordcompressiondue

to canal stenosis may not be the only pathoetiological factor contributing to

myelopathy. Furthermore, this supports the idea that the absence of frank spinal

cordcompressiononmedicalimagingcannotexcludeadiagnosisofDCM.

Recent research on patients with kyphotic deformity using kinematic MRI

hasalsoindicatedthatsubtypesofkyphosisaffectcervicalspinemobilitydifferently

and that kinematicMRImay be better than standard supineMRI in investigating

boththedegreeof instabilityandspinalcordcompression(Ruangchainikometal.,

2014). The same authors also found that in patients with kyphosis, spinal cord

compressionwasmostlylocatedatC4‐6,theapexinC‐type,andthetransitionzone

ofS‐typeandR‐typekyphoticdeformities.However,theapexofthefocalkyphotic

deformities was also noted as a high‐risk area for spinal cord compression

(Ruangchainikometal.,2014).

24

1.4EpidemiologyofDegenerativeCervicalMyelopathy

Although degenerative cervical conditions are generally regarded as the most

commoncauseofspinalcordimpairmentintheelderly,theirglobalepidemiological

trends remain challenging to evaluate. There are 3main reasons for this: (1) the

prevailingclassificationofdegenerativeconditionsintoseparateclinicalentitieshas

resultedinsegregationoftheirpreviousepidemiologicalinvestigation;(2)thereisa

general paucity of literature on the topic; and, (3) investigations that have been

conducted have focused on particular world regions or populations. Current

epidemiological data on the incidenceandprevalenceof degenerativepathologies

from around the world are summarized in Table 1.4. In the following section

incidence, prevalence, progression rates as well as survival will be further

discussed.

25

Table1.4EpidemiologicalTrendsandPatientCharacteristicsoftheVariousFormsofDegenerativePathologies.NotethatregardingOPLLregional incidence, authors have frequentlyusedprevalence and incidence interchangeably. †Protrusionbeyond the vertebralbodycausingcordcompression.AP=Anterior‐Posterior

TypeofDegenerativePathology

TrendsinPatientCharacteristics RegionalIncidence

Spondylosis Male:Femaleratiois3:2(TracyandBartleson,2010).Malespresentwithagreaterincidenceofcanalstenosis,morevertebrallevelsandmoreseverestenosisattheprimarylevelofpathology(HukudaandKojima,2002,Northoveretal.,2012).TheheightandAPdiameterofthevertebralbodyarelargerinmales(HukudaandKojima,2002,Northoveretal.,2012).Femalespresentatayoungerageandhavealargercanalbodyratio(HukudaandKojima,2002,Northoveretal.,2012).

TheincidenceofosteophytedevelopmentinthecervicalvertebraofskeletonswasgreaterinCaucasiansthannativeAfricansatallcervicallevelsstudiedinSouthAfrica.NativeAfricans,however,hadsmallermidsagittalandtransversediameteratallcervicallevels(Taitz,1999).

DiscDegeneration Asymptomaticvolunteers(Ernstetal.,2005):Milddiscdegeneration:50%insubjects<30years,75%inthoseaged31‐45yearsand100%involunteers>40years.Severedischerniation:16.6%insubjects<30years,8.3%inthoseaged31‐45years,56%inthoseaged46‐60yearsand100%involunteers>60years.Japaneseasymptomaticvolunteers(Matsumotoetal.,1998):Evidenceofdiscdegenerationin17%ofmalesand12%offemalesintheir20s,risingto86%and89%,respectively,inpatientsovertheageof60years.

Asymptomaticvolunteers(Ernstetal.,2005):Annulartears:37%Bulgingdiscs:73%Protrusions:50%Discextrusions:rareRadiologicalsignsofmedullarycompression:13.3%Japaneseasymptomaticvolunteers(Matsumotoetal.,1998):Discdegenerationwasthemostfrequentfinding.Accompaniedbyeitheranterior(78%)orposterior(71%)protrusions(80%bulges,20%prolapsed).Grade2†posteriordiscprotrusionwasidentifiedin7.6%.

26

Table1.4(Continued)

TypeofDegenerativePathology

TrendsinPatientCharacteristics RegionalIncidences

Ossificationoftheligamentumflavum(OLF)

ThoracicOLFsaremorelikelytopresentasymptomaticallythancervical(Al‐Jarallahetal.,2012).SoutheastAsian:Believedtobecommonlyafflicted,particularlymalesbetween40‐60yearsofage(Wanietal.,2012).Kuwait:63.2%presentwithsingle‐leveldisease.57.7%involvedsegmentsinthecervicalspine(Al‐Jarallahetal.,2012).SouthChina(volunteers):Femalepredominant.Agedependent,withprevalencerisingfrom0.5%insubjects<35years,to5%inthe35‐45yearsagegroupandto>7%involunteers>45yearsofage(Guoetal.,2010).

Kuwait:18.6%ofpatientswithlowbackpainhadevidenceofOLF(Al‐Jarallahetal.,2012).SouthChina(volunteers):3.8%(Guoetal.,2010).

Calcificationoftheligamentumflavum

Japan:Caseseriesof3femaleswithCLFpredominatelyinthecervicalspineatC5‐C7(Miyasakaetal.,1983).FrenchWestIndies:Acaseserieson6patientsidentified5femalesand1male;averageage71.7years(Cabreetal.,2001).

Rare(Cabreetal.,2001).PredominantlyreportedintheJapanese(Cabreetal.,2001).

27

Table1.4(Continued)

TypeofDegenerativePathology

TrendsinPatientCharacteristics RegionalIncidences

Ossificationoftheposteriorlongitudinalligament(OPLL)

Malepredominant:Atleasttwotimesmorecommoninmalesthanfemales(Ohtsukaetal.,1987,Harshetal.,1987,Jeonetal.,2012,Kimetal.,2008,Saetiaetal.,2011,Smithetal.,2011,Trojanetal.,1992,Wuetal.,2011).Mayoccurin20‐25%ofpatientstreatedforcervicalmyelopathyinNorthAmerica(Epstein,1994).OPLLsymptomatology:50%inJapanese,14%innon‐JapaneseAsiansand50.5%inIndianCaucasians(Jayakumaretal.,1996).Japan:Prevalencewas4.3%inmales,2.4%infemales(Ohtsukaetal.,1987).Prevalencewas2.6%forpatientsintheir50sand4.5%forpatientsintheir60s(Ohtsukaetal.,1987).Peakincidenceinthe50‐60yearagegroup(Harshetal.,1987,Jeonetal.,2012,Kimetal.,2008,Ohtsukaetal.,1987,Tsuyama,1984).

Japan:1‐4.3%(Jayakumaretal.,1996,Ohtsukaetal.,1987,Onoetal.,1999,Trojanetal.,1992,Inamasuetal.,2006,MatsunagaandSakou,2012)UnitedStatesofAmerica:0.1‐1.3%(MatsunagaandSakou,2012)NorthAmerica:0.12%(Trojanetal.,1992)NewYorkCity:0.7%(Trojanetal.,1992)WesternCaucasians:0.16%(Jayakumaretal.,1996)Non‐JapaneseAsians:2‐4%(Jayakumaretal.,1996)IndianCaucasians:1‐8%(Jayakumaretal.,1996)Korea:0.6‐3.6%(Kimetal.,2008,Inamasuetal.,2006,MatsunagaandSakou,2012,Smithetal.,2011)Taiwan:2.1‐3.0%(Smithetal.,2011,MatsunagaandSakou,2012)Singapore:0.8%(MatsunagaandSakou,2012)HongKong:0.4%(MatsunagaandSakou,2012)Philippines:1.5%(MatsunagaandSakou,2012)Mongolia:1.5%(Tsuyama,1984)Non‐Asian:0.16%vs.Asia:2.4%(Saetiaetal.,2011)WestGermany:0.1%(Trojanetal.,1992)Italy:1.7%(MatsunagaandSakou,2012)

28

1.4.1EstimationofIncidenceandPrevalenceofDCMfromNon‐

TraumaticSpinalCordInjuryData

OnepotentialwayofestimatingtheincidenceandprevalenceofDCMisbylooking

atthereportedratesofSCIs,whicharegenerallybrokendownintotraumaticand

non‐traumatic forms. DCMs represent non‐traumatic forms of spinal cord injury

(ntSCI)and, therefore,are includedinsuchestimates.Thisclassification,however,

alsoincludesmotorneuron,infectious,inflammatory,andneoplasticdiseaseaswell

as various other conditions (Farry and Baxter, 2010). In a recent review on non‐

traumatic spinal cord injury epidemiology, New et al (2014) indicated that

degenerativediseaseofthespinemakeup59%ofntSCIinJapan,54%intheUSA,

31%inEurope,22%inAustraliaandbetween4‐30%inAfrica.Inthissamereview

itwasestimatedthattheregionalincidenceofntSCIinNorthAmerica,Europeand

Australia were 76, 26 and 6 permillion, respectively, and that the prevalence in

Canadais1,120/millionandKashmirregionis2,310/million.Fromthesenumbers,

itcanbeestimatedthattheincidenceandprevalenceofntSCIrelatedtoDCMinthe

North American region is at a minimum 41 and 605 per million, respectively. A

numberof important limitationsof this estimatehave tobe considered, however.

Theseincludethegenerallowlevelofqualityofliteraturethatwasatthedisposalto

make these approximations, and the fact thatpatients recorded asntSCI injury in

countries with higher quality data, such as Canada (Farry and Baxter, 2010),

includedonlythefindingsofthosewithparaplegiaandquadriplegiaandnotthose

withlessseveredisability.Therecognitionthatmanymyelopathicpatientswillhave

a milder clinical presentation indicates that the aforementioned figure is

significantlyunderestimatedand likely representsonly thoseat the severe endof

thespectrumofdisease.Ontheotherhand,itshouldalsobenotedthatnotallcases

ofntSCIarecervicalinnature.

29

1.4.2EstimationofDCMProgressionRatesfromStudieson

OsteoarthritisintheSpine

Despitethefindingthatradiographicevidenceofosteoarthritis(OA)presentsinthe

majority of people by 65 years of age and in about 80% of those over 75, the

literaturehasprimarilyfocusedonOAratesofthehands,kneeandhip(Ardenand

Nevitt,2006),thusmakingitdifficulttoderiveestimatesforratesofspinalOA.One

exceptiontothishasbeenareportontheradiographiccervicalspineosteoarthritis

progression rates byWilder et al (2011). In their longitudinal study, the authors

found that progression of OA (as defined by the Kellgren and Lawrence ordinal

scale; an increased grade denotes worse degeneration)(Kellgren and Lawrence,

1963),increasedinmalesatallagesatagreaterratethaninfemales.Furthermore,

theauthorsreportthatthehighestprogressionrateswaspresentinmenbetween

70‐79yearsofageat12.5casesper100patient‐yearsandinwomenovertheageof

80at9.3casesper100patient‐years.Unfortunately,intermsofDCM,thisestimate

is limited by the fact that it indicates patients at risk of myelopathy rather than

clinicallymyelopathicpatientsandwouldalsolikelyexcludepatientsconsideredto

have OPLL and OLF. For those inwhom progression leads to asymptomatic cord

compression, clinically evidence of myelopathy has been shown to develop at

approximately8%at1‐yearfollow‐upand23%atamedianof44monthoffollow‐

up(Wilsonetal.,2013a).

1.4.3EstimationbyHospitalAdmissionRates

Recently, 2 studies have emerged that estimated epidemiological trends for CSM

based on hospital admission data. Boogaarts and Bartels (2013) estimated a

prevalence of CSM of 1.6/100,000 inhabitants based on surgical cases at their

hospital in the Netherlands. Wu et al (2013) retrospectively assessed a 12‐year

nationwide database and estimated an overall incidence of CSM‐related

hospitalizationof4.04/100,000person‐years.

When interpreting these findings,anumberof limitationshave tobe taken

into consideration, including the geographical basis (referral area), aswell as the

30

fact that patients undergoing surgerymost likely represented those at the severe

end of the spectrum of disease; therefore, it is likely that these estimations

undervalue actual epidemiological trends. Furthermore, while Wu et al (2013)

included OPLL and disc pathologies as part of their CSM rate estimations, it is

unclearifthiswasthecasewiththestudybyBoogaartsandBartels(2013).

In addition to these studies, it has been proposed that the rate of surgical

treatmentforCSMisrising.Ladetal(2009)examinednationaltrends inoutcome

for CSM in the USA and reported that annual admissions increased from 9,623

patientsin1993to19,212in2002(approximately3.73to7.88per100,000)based

on theUSnational inpatientdatabase.Moreover, theauthors found thatanear7‐

foldincreaseinthenumberofspinalfusionsforCSMtookplacebetween1993and

2002intheUS,representinganincreasefrom0.6to4.1per100,000people(Ladet

al.,2009).Itshouldbenotedthatincreasesinfusionratesmightbeareflectionof

thetransitionintheunderstandingofthepathophysiologyofDCMfromonesimply

based on static compression of the cord to one recognizing malalignment and

dynamicinjuryaspartofthediseaseprocess.

1.4.4Survival

A study from Israel by Ronen et al (2004) assessed the survival of patientswith

ntSCI between 1962‐2000. Though these data included all forms of ntSCI,

informationregardingthelocationandtypeofetiologyforntSCIwereprovided.The

authors found that of 1066 patientswith ntSCI, 32%were cervical in nature and

that 62% of those with spinal stenosis were cervical. Patients with ntSCI of the

cervicalspineofanyetiologyhadamediansurvivalof22.7yearsandpatientswith

spinalstenosisatanysitehadasurvivalof17.6years.PatientswithntSCIduetoa

herniateddiskhada survivalof29years;however,mostof thesewere lumbar in

nature.

31

1.5RiskFactorsAssociatedwithDegenerationCervical

Myelopathy

Though the anatomical restructuring resulting from degeneration is a natural

process of aging, the timing of itsmanifestationmay be influenced by hormonal,

metabolic,geneticandpersonalfactorsaswellascongenitalpredispositions(Singh

et al., 2012, Tracy and Bartleson, 2010, Ogata et al., 2002, Inamasu et al., 2006).

Despite the fact that many of these risk factors have been proposed and seem

logicallyrelevant,thereisapaucityofhighqualityevidenceassociatingthemwith

DCM.Inthefollowingsections,genetic,occupational,personal,congenitalaswellas

otherassociatedriskfactorsthathavebeendiscussedinliteraturearepresented.

1.5.1HereditaryandGeneticCauses

Anumberofstudiesintheliteraturediscussthegeneticbasisofdegenerativespinal

disease. Much of this academic enquiry however, has focused on OPLL and DDD,

with relatively little research being conducted on CSM. Moreover, most studies

seeking to elucidate anunderlying genetic predispositionhavebeen conducted in

Asia, limitingglobalgeneralizability. It isevident fromthe literaturehowever that

OPLL, OLF, DDD and CSM all may entail some form of genetic or hereditary

etiological component; albeit, the degree of evidence varies between the types.

Someof thestrongestpotentialgenetic factors implicated include thoserelated to

MMP‐2andcollagenIXforDDD(ErwinandFehlings,2013),andCollagenVIandXI

forOPLL(Wilsonetal.,2013c).Additionally,someofthecandidategeneticvariants

haveshowntocausepredispositiontothedevelopmentofmorethanoneofthese

conditions,forexample,geneticstudiesontheVitaminDreceptor(VDR)havebeen

associatedwithbothDDDandCSM(ErwinandFehlings,2013,Wangetal.,2010b).

Likewise, it hasbeen found that the collagenVI gene is potentially linked toboth

OPLL andOLF (Kong et al., 2007). These findings indicate thepossibility of some

pathomechanical alignment between these conditions. Table 1.5 summarizes

currentgenes/geneticproductsthathavebeenstudied.

32

Table1.5SummaryofgeneticresearchconductedonCSM,DDD,OPLLaswellasOLF.VDR=VitaminDReceptor;MMP=MatrixMetalloproteinase;ADAMTS=adisintegrinandmetalloproteinasewiththrombospondinmotifsBMP=BoneMorphogeneticProteins;TSG‐6=TumorNecrosisFactoralphastimulatedgene6;IL=Interleukin.

DCMTypes StudiedGenes/Geneproducts

CervicalSpondyloticMyelopathy(CSM)(Setzeretal.,2008,Wangetal.,2012,Wangetal.,2010b)

Structural: CollagenIXHormonal:VDROther:ApolipoproteinE

DegenerativeDiscDisease(DDD)(ErwinandFehlings,2013)

Structural:CollagenIXEnzymes:MMP‐2,MMP‐3,MMP‐9,ADAMTSHormonal:VDRImmunological:IL‐β,IL‐1receptor

OssificationofPosteriorLongitudinalLigament(OPLL)(Inamasuetal.,2006,Wilsonetal.,2013c,Ogataetal.,2002,Kudoetal.,2011,Kimetal.,2012,Ikedaetal.,2005,Yoshizawaetal.,2004,Iwasawaetal.,2006,Ohishietal.,2003)

Structural: CollagenVI andXIHormonal:RetinoicXreceptor,VDR,ParathyroidHormone,LeptinReceptor,EstrogenReceptorImmunological:TSG‐6,TNF‐β1,IL‐1αandβ,IL‐15,Interferon‐γTranscriptionfactors:RUNX2,PromyeloticLeukemiaZincFinger,Msx2,GrowthFactors:Insulin‐likegrowthfactor,Connectivetissuegrowthfactor,BMP2and4,Growth‐hormone‐bindingprotein,platelet‐derivedgrowthfactorOther:NucleotidePyrophosphatase,Endothelin‐1,ProstaglandinI2

OssificationofLigamentumFlavum(OLF)(Liuetal.,2010,Kongetal.,2007,Kishiyaetal.,2008)

Structural: CollagenVIHormonal:VDRGrowthFactors:BMP‐2Transcriptionfactors:RUNX2

33

1.5.1.1CervicalSpondyloticMyelopathyandDegenerativeDisc

Disease

Earlyreportssuggestingageneticsusceptibility for thedevelopmentofCSMwere

published by Bull el at (1969) and Palmer et al (1984). In their research, both

groupsassessedandcomparedcervicalradiographsoftwinsiblings.ThoughBullel

at(1969)foundthattherewasamarkedresemblanceinthedegenerativeprocess

between twinson imaging,Palmeretal (1984)observed theopposite, concluding

insteadthatthoughanatomicalsimilaritiesintwinsmaypredisposethemtosimilar

degeneration over time, the incongruence of pathological findings amongst their

subjectsindicatethatnumerousotherfactorsarelikelyatplay,andthusprecludea

simple genetic causation. One potential reason for discrepancy between their

findings may be related to epigenetic factors, specifically, the transcriptional

alterations in cells over time related to non‐hereditary influences such as the

environment.

Morerecently,Setzeretal(2008),Wangetal(2010b)andWangetal(2012)

have soughtparticular genetic determinantsof CSM susceptibility, implicating the

potential polymorphism of Apolipoprotein E, Vitamin D receptor (VDR) and

Collagen9genes, respectively.Wilson et al (2013c) in their systematic reviewon

thetopiccautionthatthoughthesestudiesdidfindsignificantgeneticassociations

between single nucleotide polymorphisms (SNPs) and CSM, none of the findings

havebeenvalidatedinseparatestudies.

Todate,thestrongestindicationforageneticetiologicalcomponenthasbeen

published by Patel et al (2012) who assessed a genealogical database of over 2

million Utah (USA) residents for excessive familial clustering, reporting the

presenceofsignificant(p<0.001)relatednessofthosewithCSM.Mostnotably,they

determined that there was a significant relative risk of 5.21 (p<0.001) for the

developmentofCSMamongfirst‐degreerelatives(Pateletal.,2012).

IndiscussinggeneticsusceptibilityforCSM,itisalsoimportanttoaccountfor

genesthathavebeenassociatedwiththedevelopmentofDDD,becauseithasbeen

recognized that the pathogenesis of CSM is preceded by disk degeneration

34

(Karadimas et al., 2013b). Erwin and Fehlings (2013) recently summarized a

number of such genes, including those related tometalloproteinases, collagen IX,

VDR,andinterleukins.Interestingly,genesrelatedtoVDRandcollagenIXhavebeen

associated with both CSM and DDD, suggesting that these may be of particular

importanceinthenaturalcourseofCSMdevelopment.Therecognitionthatmultiple

genesmay be involved in the occurrence of these conditions indicates that there

maybesubstantialheterogeneitybetweenpatients,theirclinicalmanifestation,and

thenaturalcourseoftheirprogression.

35

1.5.1.2OssificationofthePosteriorLongitudinalLigament

The search for an underlying genetic basis for OPLL has been carried out by a

numberofinvestigators.Suspicionforageneticetiologicalcomponentissupported

by the fact that OPLL presents significantlymore frequently in Asian populations

thaninCaucasians(Inamasuetal.,2006,MatsunagaandSakou,2012,Jayakumaret

al., 1996, Ono et al., 1999, Saetia et al., 2011, Smith et al., 2011). Indeed, a

nationwide survey in Japan revealed that 24% of 2nd degree or closer blood

relativesand30%ofsiblingsofpatientswithOPLLhadradiographicallydetectable

OPLL(MatsunagaandSakou,2012).

Atthepresenttime,anumberofcandidategenesincludingthoserelatedto

collagen VI and XI, BMP 2 and 4, TNF‐alpha, TNF‐Beta 1 and 3, nucleotide

pyrophosphatase,retinoicXreceptor,VDR,parathyroidhormone, IL‐1BandIL‐15,

RUNX2, promyelotic leukemia zinc finger, connective tissue growth factor,

prostaglandinI2,endothelin‐1,leptinreceptor,estrogenreceptor1,andinterferon‐

gammahavebeeninvestigated(Inamasuetal.,2006,Kudoetal.,2011,Kongetal.,

2007, Liu et al., 2010, Kim et al., 2012). However, despite the numerous

investigations, a recent systematic review on the topic by Wilson et al (2013c)

determinedanoveralllowstrengthofevidenceinsupportforageneticassociation

forOPLLdevelopment.

Investigation of OPLL as a single disease may be hindering attempts to

attribute specific genetic factors with its development. Based on its pathological

distribution, OPLL has been classified into 4 subtypes: localized (bridged or

circumscribed);segmental;continuousandmixed,withsegmentalconsideredasthe

mostcommontype(Stapletonetal.,2011,MizunoandNakagawa,2006).Recently,

Kudo et al (2011) looked at the genetic differences amongst these subtypes and

proposed 2 categories for OPLL: continuous (include continuous andmixed) and

segmental(includesegmentalandcircumscribed).Theyconcludedthatcellsofthe

OPLL continuous group had higher osteogenic differentiation potency in

comparison to segmental group cells, and that a different genetic background

between the groups also exists (Kudo et al., 2011). This finding is supported by

36

previous research that has found that patients with continuous andmixed OPLL

types frequently experience progression of their ossification postoperatively, and

that contrarily, progression is rare in patients with segmental OPLL (Hori et al.,

2006,Kawaguchietal.,2001).ThisisworthyofnotesinceKawaguchietal(2001)

indicated that 3 of their 45 patients developed neurological deterioration due to

OPLLprogressionat10‐yearfollow‐up.Thus,therecognitionofpotentialetiological

differencesbetweenOPLLtypesmayhavesignificantclinicalramifications.

Agenerallimitationoftheinterpretationofthegeneticstudies,asisthecase

withOLF,isthatstudiesassessingcandidategeneshavebeenconductedinAsia,and

thereforemaynotbeetiologicallytranslatabletootherpopulations.

37

1.5.1.3OssificationoftheLigamentumFlavum

Atthepresenttimethereisinsufficientevidenceavailabletosupportageneticbasis

forthedevelopmentofOLFinisolation.Havingsaidthis,agreaterfrequencyofthe

diseaseinsomeregions,suchasAsia,hasgivensupporttotheideaofsomeformof

genetic and/orenvironmental susceptibility (Onoet al.,1999).Additional support

for an isolated genetic connection has also been demonstrated recently using a

mouse model with homozygous mutation in the NPPS gene, which has been

specificallylinkedwiththedevelopmentofOLFatC2/3(Yuetal.,2009).However,

few studies have looked at OLF in humans and have frequently studied OPLL in

parallel. For example, Kong et al (2007) found that intron 33 (+20) as well as

promoter(‐572)SNPsofCOL6A1wereassociatedwithbothOPLLaswellasOLFin

the Chinese Han population, although they also found that Haplotype 4 was

associatedwithOLFbutnotwithOPLL,andthatHaplotype1wasassociatedwith

OPLL but not with OLF. Other work by Liu et al (2010), has given support to a

previously described genetic association related to the RUNX2 gene in OPLL

patients (Kishiya et al., 2008), and has also found a similar associationwith OLF

patients,concludingthatgeneticvariantsinthegenemayberesponsibleforectopic

boneformationinspinalligaments.

Itisclearthatfurtherresearchneedstobeconductedtoestablishwhethera

genetic susceptibility exists and if such susceptibility is genetically separate from

OPLL as well as conditions with generalized spinal ligament ossification such as

DiffuseIdiopathicSkeletalHyperostosis(DISH).

38

1.5.2Occupational,PersonalandOtherAssociations

ThedevelopmentofDCMhasbeenrelatedtoanumberofpersonalandoccupational

risk factors. These factors may accelerate degeneration of the spine through

mechanical stress (bothdynamicandstatic) aswell asunderlyingpathobiological

processes.Also, theassociationbetweenindividualrisk factorshasbeenshownto

vary between DCM types, potentially providing insight into differences in their

pathomechanics.

1.5.2.1PhysicalTransportationofGoods

Porters and coolies in less developed countries transport substantial amountsof

goodsbybearingweighton theirheadsandbacks.Anumberof researchershave

investigated the potential for this mode of transportation to accelerate cervical

spondylosis(JumahandNyame,1994,EcharriandForriol,2005,Jageretal.,1997,

Joosabetal.,1994).However,asthetypesofloadbearingandtechniquesmaydiffer

amongthevariouspopulations,itisdifficulttoextractacollectiveconclusiononthe

precise biomechanical factors that may be at play. There is a general consensus

among these studies that degenerative changes of the cervical spine in weight

bearers is significantly more pronounced than what is observed in the general

population, and that the C5‐6 region is particularly affected (Jumah and Nyame,

1994,EcharriandForriol,2005,Jageretal.,1997,Joosabetal.,1994).Additionally,

weightbearersmorecommonlycomplainedofneckstiffnessandpain(Echarriand

Forriol,2005).

BistaandRoka(2008)hypothesizedthatNepaleseportersusinga“Namlo”

(aparticular typeofcarriage inNepalwhereweightbearingoccursprincipallyon

backs)wouldpromotecervicalspondylosisaswell.However,itwasfoundthatthe

prevalence of spondylosiswas actually lower than that of controls (non‐porters).

The authors suggest that using thismethod, the orientation of the cervical spine

requires isometric flexion of the neck and thereforemay significantly reduce the

riskofspondylosiscomparedtonon‐porters.

39

Given these findings, it appears that themethodof transportationofgoods

cansignificantlyimpacttherateofspondylosisandthisrecognitionmaybeusefulin

reducingthisoccupationalhazard.

1.5.2.2CervicalSpineDegenerationinHigh‐PerformanceAviators

andAstronauts

The potential of an increased susceptibility for degenerative changes in fighter

pilotsemergedafterreportsofacuteneckinjuriesduringflight,andtherecognition

thatGz (G‐force)place substantial axial forceon the cervical spine (Schall, 1989).

Uponneckmovementduringflightmaneuvering,theabilityofthecervicalspineto

withstand high compressive forces is substantially reduced and the IVDs are

predominatelyaffected,astheyarethechiefcomponentsofthespinethatadaptto

stress and have viscoelastic properties (Schall, 1989, Hamalainen et al., 1993).

Investigations on fighter pilots have mostly supported that accelerated DDD and

osteophytosis takes place (Petren‐Mallmin and Linder, 2001, Petren‐Mallmin and

Linder,1999,Hamalainenetal.,1993,HendriksenandHolewijn,1999,Landauetal.,

2006).

One of the more comprehensive studies on the subject was conducted by

Petrén‐Mallmin and Linder (1999) & (2001). These authors studied an

asymptomatic group of fighter pilots against controls (non‐fighter pilots) for

evidenceofspinaldegenerationandthenassessedthemagainata5‐yearfollowup.

Duringtheirinitialanalysisofasymptomaticsubjects,itwasevidentthatsignificant

acceleration indegenerationoccurred inolder fighterpilots compared toyounger

fighterpilotsaswellasage‐matchedcontrolswithoutexposure toGz; specifically,

experienced pilots had significantly increased degeneration in the form of

osteophytes, disk protrusions/herniations, cord compression and foraminal

stenosis.Uponfollow‐upat5years,theauthorsnotedthatthedifferencebetween

the groups diminished significantly, and they concluded that high performance

flying results in cervical spine degeneration at a younger age compared with

40

controls;however,with increasingage, thedifferencebetweenpilotsandcontrols

diminishes(Petren‐MallminandLinder,2001).

Astronauts frequently complain of back pain in microgravity (Sayson and

Hargens,2008)andhavebeenrecognizedtobeatriskfordegenerativediseaseof

the cervical spine (Johnston et al., 2010). It has been proposed that this may be

attributable to the physiological changes as a result of weightlessness, including

spineelongationandlossofthoracicandlordoticcurvatures(Johnstonetal.,2010).

Recently, it was reported thatastronauts have a significantly higher incidence of

nucleuspulposusherniationthancontrols(Johnstonetal.,2010).Theincidenceof

herniationwasreportedtobehighestwithinthefirstyearafterreturnfromaspace

mission, a striking35.9 times the risk compared tomatched (age, sex, bodymass

index) controls (Johnston et al., 2010). Itwas alsoparticularly interesting tonote

that the incidence of cervical herniation was 21.4 times higher in astronauts

(Johnston et al., 2010). The authors postulate that prolonged abnormal posture

during space missions, exposure to both low and high Gz environments,

predispositiontoinjuryduetopostflightdysfunctionoftheneurovestibularsystem,

and changes within the disc structures may be responsible for these findings

(Johnstonetal.,2010).

1.5.2.3CervicalSpineDegenerationinAthletes

Athletesengaging insignificantphysical contacthavebeenshowntopresentwith

accelerated cervical degeneration. This has been demonstrated in rugby players

whohavebeenstudiedbyBergeetal(1999),Scher(1990)andSilver(2002).There

isageneralconsensusthatacceleratedcervicaldegenerationinthisgroupislikely

attributable to thehigh forceplacedupon thecervical spine,particularly in rugby

forwards. Scher (1990) postulates that the forces in the rugby scrum which can

generate up to 1.5 tons upon the cervical spine, coupled with rapid changes in

directionandintensityofthrust,likelyresultsinrepeatedminortrauma.Inaddition

to this, in a series of 47 rugby players and 40 age‐matched controls, Berge et al

(1999)notedincreasedratesofdischerniation,discdegenerationandreductionof

spinal canal diameter as well as other factors amongst rugby players. Much like

41

rugby players, American football players are also exposed to potentially injurious

energy inputs directed at the cervical spine (Nyland and Johnson, 2004). Though

protectiveequipmentsuchashelmetsmaydiminishsomeoftheseforces,playersin

sports inwhich great force is repeatedly exerted on the cervical spinewill likely

experienceaccelerateddegeneration.

Cervical spine pain is also a common complaint across other high

performanceathletesincludingcyclistandtriathlonparticipants(Villavicencioetal.,

2007). Chronic neck pain (≥3 months) was reported in 15.4% of triathlon

respondentsinarecentstudy,andithasbeensuggestedthatthisispossiblydueto

discogenic injury and overuse (Villavicencio et al., 2007). While it would seem

intuitive to believe that such athletes may be predisposed to accelerated

degenerative disease of the spine due to increased work intensity, research on

weight‐lifters and participants of various other sports has shown that high

performanceathleticactivitiesarenotharmful (Mundtetal.,1993). Indeed, ithas

been postulated that participation in most sports may exert a protective effect

againstIVDherniations(Mundtetal.,1993).

1.5.2.4PersonalFactors

Patient factors have been implicated in the development of both CSM as well as

OPLL. Wang et al (2012) assessed the impact of smoking status on disease

development in theirstudyof172Chinesesubjects.They foundthatpatientswho

smokedwereatincreasedriskofCSMdevelopment,andthattheriskincreasedin

patients who carried tryptophan alleles of collagen IX. Indeed, it has been well

reported that smoking impacts bonemetabolism, decreases bonemineral density

andincreasesfracturerisk(Yoonetal.,2012).Furthermore,theeffectonboneshas

been shown to be influenced by dose and duration of smoking as well as body

weight (Yoon et al., 2012). Animal studies have shown that smoking has a

considerableeffectoncollagenstructure, increaseschondrocytedegeneration,and

isrelatedtocellnecrosisandfibrosis inthenucleuspulposus(Abateetal.,2013).

However, given that other research has indicated that cigarette smoking may

42

actually reduce the occurrence of osteoarthritis (Abate et al., 2013, Felson et al.,

1997),furtherresearchisnecessarytosubstantiateapotentialrelationship.

Earlyreportsindicatingthatthereisahighincidenceofdiabetesinpatients

with OPLL were discussed by Tsuyama (1984) and Takeuchi et al (1989). More

recently, Kobashi et al (2004) in their sex‐ and age‐matched study in Japanese

subjects, found that excessive weight gain between 20 and 40 years of age and

diabetes were independent risk factors for the development of OPLL. It has also

beenreportedthatgeneralossificationofspinal ligaments isassociatedwithnon–

insulin dependent diabetes mellitus (Kobashi et al., 2004). Though it has been

suggested that insulinhas implicationsonbone formation, theprecisemechanism

bywhichdiabetesandweightinfluencethedevelopmentOPLL,ifatall,remainsto

beinvestigated(Karasugietal.,2013).

Recently,2 reportshavedescribeddeficiencyofvitaminB12asapotential

contributortothepathologyinCSMpatients(Xuetal.,2013,Miyazakietal.,2014).

Indeed, vitamin B12 deficiency by itself can result in subacute combined

degeneration of the spinal cord,which can presentwith an inverse V‐shapedT2‐

weightedMRI hyperintensity on axial view (Miyazaki et al., 2014). Although it is

likely tobea rarephenomenon, this findingdoes suggest thatwhenMRI findings

areequivocal,anassessmentofvitaminB12deficiencyshouldbeconsidered.

1.5.2.5MotorVehicleAccidentInjuryandDCM

Motor vehicle accidents (MVAs) represent one of the most frequent causes of

traumaticSCIworldwide(Singhetal.,2014),andithasbeensuggestedthatspinal

canalstenosisanddegenerativediscdiseaserenderindividualsmoresusceptibleto

neckinjuryandreducesinjurythresholds(Petterssonetal.,1995,Zhouetal.,2010).

However, the long‐termeffectsof soft tissue injurysurrounding thecervical spine

includingneck strains (muscle‐tendonoverloading) and sprains (ligamentous and

capsulardamage)thatfrequentlyoccurwithMVAsintheabsenceofacutedamage

toneuralelementsremainslessclear(Zhouetal.,2010,Otremskietal.,1989).The

biomechanicaleffectsonthecervicalspineduringMVAsarecomplexanddependon

the type of accident (e.g. rear‐end, side impact) and, therefore, multiple injury

43

mechanismare likely relevant. In termsofwhiplash injuries, ithasbeen reported

thatdiscpathologymayberesponsibleforpersistentsymptomswhichmayleadto

neurological impairment later on (Jonsson et al., 1994, Pettersson et al., 1997).

However, Matsumoto et al (2010) in their 10‐year prospective follow‐up on

whiplashpatientsandcontrolsfoundthatwhilesomeevidenceofdiscdegeneration

wassignificantlymorecommoninpatientswithwhiplashinjury,nopatientsintheir

studydevelopedmyelopathyorrequiredsurgery.

1.5.2.6OtherAssociations

ThereareanumberoflessdefinedassociationswithDCM.Inastudyof30patients

withschizophrenia,Matsunagaetal(2008a)reportedafindingofOPLLin20%of

thesepatients,notingthatthisrepresentsnearly5timesthereportedincidencein

the general Japanese population. However, it is unclear how these patients were

selected and thus these findings may be subject to bias. The authors suggest a

potentialconnectionbetweentheconditionsviacalcineurin,whichhasbeenlinked

withsusceptibilityforschizophreniadevelopmentandhasbeenshowntopartakein

osteogenesisbyosteoblastactivation(Matsunagaetal.,2008a).Althoughthiswould

suggestabiochemicaletiology,itisunclearwhetherperhapsaberrantkineticssuch

asdystoniasorothermovementabnormalities,whichmayappear insomepeople

withschizophreniabecauseofneurolepticuse,maybecontributory. Indeed,other

aberrantkinetics,asareseeninpatientswithParkinson’sdisease,forexample,may

contribute to accelerated degeneration as well. Parkinson’s patients who have

chronic dystonia and who have not benefited from medical or botulinum toxin

treatment should be reassessed for complex cervical dystonia with tonic posture

and phasic movements (Krauss et al., 2002). These patients can rapidly develop

progressive myelopathy secondary to their dyskinesias and dystonia with

generalizedmovement(Kraussetal.,2002).Additionalnoteworthyconditionswith

associated movement abnormalities that have been implicated with accelerated

cervical spondylosis were described by Wong et al (2005) and include cerebral

palsy,torticollisandTourettesyndrome.Inparticular,patientswithcerebralpalsy

who exhibit severe involuntary movements of the head and neck have shown a

44

propensity to develop significantdegenerative changes in the cervical spine at an

earlyage(Kimetal.,2014).

Additional potential population groups associatedwith OPLL development

include patients with hypoparathyroidism, hypophosphatemic rickets, and short

sleepinghours(Inamasuetal.,2006).

45

1.5.3CongenitalDisordersAssociatedwithDegenerativeCervical

Myelopathy

Although congenital disorders do not generally cause DCM directly, they can

indirectly accelerate the pathological process or result in earlier clinical

manifestation. This may occur through various gross as well as microstructural

anatomicalaberrationsofthecervicalspine, includingcongenitallyfusedvertebral

segments (Pizzutillo et al., 1994, Nouri et al., 2015), congenital spinal stenosis

(Singhetal.,2012),orstructuralabnormalitiesleadingtohypermobilitysyndromes

(McKayetal.,2012).

1.5.3.1Down’sSyndrome

Patientswithtrisomy21orDown’ssyndrome(DS)havebeenrecognizedtobeata

predisposition for developing degenerative cervical disease (Bosma et al., 1999,

Oliveetal.,1988,Alietal.,2006).Thiscanoccurduetoalitanyofcongenitalfactors

including atlantoaxial instability, odontoid abnormalities, atlanto‐occipital

abnormalities and hypoplasia of the posterior arch of C1 (Bosma et al., 1999).

Atlantoaxial instability has been estimated to occur in 10‐20% of DS patients, of

which1‐2%presentwith symptomatic spinal cord compression (Ali et al., 2006).

Unfortunately, while it has been recognized that spondylosis is common in DS

patients,specificepidemiologicaldatadonotcurrentlyexist.Thesignificanceofthis

knowledge gap from a clinical perspective is evinced by the approximately 5400

childrenwithDSbornintheUSAperyearandtheirincreasinglifespan,whichhas

increased fromamedianageofdeath from25years in1983 to49years in1997

(Shinetal.,2009).Astheoccurrenceofdegenerativechangesinthecervicalspine

arenaturallyprogressiveandseemstomanifestatayoungerageinpatientswithDS

(Ali et al., 2006, Bosma et al., 1999), this trend suggests that an increase inDCM

arisingfromthispopulationcanbeanticipatedovertheyears.

46

1.5.3.2Klippel‐FeilSyndrome

Kilppel‐Feil syndrome (KFS) patients are diagnosed by the “hallmark” finding of

congenital fusionof cervicalvertebraeandhavebeendescribedbya clinical triad

encompassing a short neck, low posterior hairline and restriction of neckmotion

(Samartzis et al., 2006, Giampietro et al., 2013). Most occurrences of KFS are

thoughttooccursporadically,butautosomaldominant,autosomalrecessiveandX‐

linked forms have also been described (Giampietro et al., 2013). Recently, Rosti

(2013) implicatedmutations in theMEOX1 gene as the culprit behind autosomal

recessiveformsofKFSandTassabehjietal(2008)implicatedaGDF6genelocusin

familialandsporadiccasesofKFS.ExamplesofKFSonMRIandplainradiographs

aredemonstratedinFigure1.5.

PatientswithKFSmaybeatincreasedpredispositionforthedevelopmentof

cervical jointdegeneration(Pizzutilloetal.,1994,Nourietal.,2015).Althoughthe

mechanismforsucharelationshiphasnotbeenfullyelucidated,thepostulationthat

patientswho undergo surgical fusion of cervical vertebraemay be at risk for the

developmentofadjacentsegmentpathology(Bartolomeietal.,2005,Ishiharaetal.,

2004,Lawrenceetal.,2012),supportsthesuppositionthatincreasedbiomechanical

stressonnon‐fusedsegmentsmaybecontributory(Nourietal.,2015).Inarecent

prospectiveandmulticenterstudyonasurgicalcohortofpatientstreatedforCSM,it

wasdiscoveredthattheprevalenceofKFSamongsurgicalpatientswasnearly4%,

and that MRI features in patients with KFS were generally worse than in CSM

patientswithoutKFS (Nouri et al., 2015).Given that theprevalenceofKFS in the

generalpopulationhasbeenestimatedat0.71%(Brownetal.,1964),thissuggests

that patients with KFS may be at a substantial predisposition for DCM. Further

research will be necessary to fully uncover the relationship between these

conditions, andwhether the clinical approach to their management should differ

fromthegeneralpopulation.

47

Figure1.5MRIsandplainradiographsof3patientswithKFS.Thearrowheadspointtothefusedvertebrae.ThearrowsindicatethepresenceofhyperintensitychangesofthespinalcordonT2‐WIMRI.Theimagesshownherehavebeenadaptedandmodifiedfrom(Nourietal.,2015).

48

1.5.3.3CongenitalSpinalStenosis

Congenital cervical spinal stenosis has been widely recognized as a predisposing

factorforthedevelopmentofDCM(CounteeandVijayanathan,1979,Bernhardtet

al., 1993, Singh et al., 2012) and has been defined as a sagittal canal diameter of

<13mm (Bajwa et al., 2012), or a ratio of less than 0.82 using the Torg‐Pavlov

(Pavlov et al., 1987) measure (canal diameter/vertebral body diameter). It is

believed that a reduction of the spinal canal size will render the spinal cord

vulnerabletocompressionevenfromminorencroachmentoftissue intothecanal

(CounteeandVijayanathan,1979). Ithasalsobeensuggested thatanarrowcanal

reduces theeffect sizeof cerebrospinal fluidandwill impair cushioningofkinetic

energy in the setting of minor trauma (Del Bigio and Johnson, 1989), and

presumably other dynamic injury mechanisms. However, variability in the

transverse area of the spinal cord between individuals (Kameyama et al., 1994)

indicates thatcervicalcanalsizeshouldberelatedwith thesizeof thespinalcord

whenevaluatingriskforinjury.

Ithasbeenestimatedthatspinalcanalstenosis ispresent inapproximately

250,000‐500,000 people in the USA, of which 20‐25% are cervical (Bajwa et al.,

2012). Moreover, the prevalence in the USA is expected to increase substantially

over the next decade (Bajwa et al., 2012). These patients, and in particular those

withconditionsthatareknowntopresentwithapredilectionfortheoccurrenceof

cervical spinal stenosis, such as achondroplasia (King et al., 2009), should be

monitoredregularlyfortheearlyonsetofmyelopathicsymptoms.

49

1.5.3.4OtherCongenitalAssociations

Itisimportanttorecognizethatanumberofothercongenitalconditionsresultingin

gross structural malformations of the spine may be of importance in DCM

development.Giampietroetal(2013)havepublishedanextensivelistofconditions

related to congenital vertebral malformations that may be highly relevant. In

addition,moregeneralizedcongenitalaberrationssuchasthoserelatedtocollagen

structure may be of importance in DCM development as well. A particularly

noteworthy group includes patients with significant generalized hypermobility.

PatientswithEhlers‐DanlosSyndrome(EDS),forexample,experienceasubstantial

jointinstabilitysecondarytowidespreadligamentandtendonlaxity,andresulting

joint pain and orthopedic complications are common (Raff and Byers, 1996).

Furthermore, it has been suggested that chronic subluxation in these patients

rendersthemsusceptibletoosteoarthritis(RaffandByers,1996).Althoughlimited,

thereissomeevidencelinkingEDSwithatlantoaxialinstability,inparticular(Halko

etal.,1995).Giventhesefindings,itseemsreasonabletoassumethatEDSpatients

mayhaveasubstantialriskforbothacceleratedspinedegenerationandspinalcord

trauma. In terms of surgical treatment, this emphasizes that stabilization of the

spinetocorrecthypermobilityshouldbeakeyobjectiveifmyelopathyissuspected.

However,discussionontheimpactofhypermobilitysyndromesonDCMandspinal

cordtraumaingeneralrequiressubstantiallymoreinvestigation.

50

1.6ClinicalAssessment

Myelopathy secondary to degenerative changes in the cervical spine is a clinical

phenomenon almost exclusive to the elderly patient. Indeed, the average age of

patientspresentingwithCSMhasbeenestimatedat64(Kalsi‐Ryanetal.,2013).To

arriveatadiagnosisofDCMacarefulhistoryshouldbetaken,followedbyadetailed

clinicalexamination.

AswithOAinanyotherjointsinthebody,degenerativechangesincervical

spine articulations will impact mobility and anatomical function. As a result,

patientsmaycomplainofdecreasedrangeofmotion,stiffness,andpainrelatedto

neckmovement.Additionally,degenerativechangessuchasthoseoutlinedearlierin

section(1.3)havethepropensitytoreducespinalcanalspace,inflictneuraldamage,

and ultimately result in motor, sensory as well as autonomic nervous system

dysfunction. Clinically, patients may therefore present with variable degrees of

weakness, paresthesia and complaints of urinary dysfunction. Movement

coordination may also be impaired and patients may present with difficulty in

performing everyday tasks such as buttoning‐up their shirt, difficulty handling

utensils and walking. Clinical examination may also elicit a positive L’hermitte,

BabinskiandHoffmansign.Ultimately,thesemanifestationshaveaprofoundeffect

onthepatient’squalityoflife.

1.6.1ClinicalAssessmentToolsEvaluatingDiseaseSeverity

Althoughaclinicalexaminationmayelicittheconstellationofsignsandsymptoms

consistentwith a diagnosis ofmyelopathy, there is a spectrum of severitywithin

whichpatientsmayfall.Somepatientsmayhavesubtlesymptomswhileothersmay

suffer from more serious debilitation. It is therefore necessary to determine the

levelofimpairmentandthiscanbeachievedthroughtheuseofassessmenttools.In

addition to this, the application of assessment tools over time provides an

opportunity to evaluate progression of signs and symptoms of DCM, and if

treatment is instigated, serveas ameansbywhich the relativeefficacyofvarious

treatmentoptionscanbeobjectivelycompared.

51

In the absence of a gold standard measure, multiple tools have been

employedtoassesspatientswithDCM.Someassessment toolssuchas thepatient

reportedShortForm‐36,provideageneralassessmentofwell‐beingandhavebeen

usedtoevaluatehowmuchtheneurological impairment isaffecting thequalityof

life of DCM patients. Though these measures are subjective, they provide

informationregardingthegeneralhealthofthepatient.Morespecifictoolssuchas

themodifiedJapaneseOrthopedicAssociationScore(mJOA)aswellasNurickgrade

are utilized by physicians, and evaluate the level of impairment related to

neurological dysfunction (Table 1.6 and Table 1.7). Indeed, the mJOA score is

becoming widely accepted as the standard for assessing the functional status of

DCMpatients(Tetreaultetal.,2013c).Itisan18‐pointscalethatscoresupperand

lower extremitymotor function, sensation, and sphincter control. Although it has

beenwidely used, there has been little investigation to evaluate itsmeasurement

properties until recently. Kopjar et al (2014) found that themJOA hadmoderate

internal consistency (Cronbach’s alpha = 0.63), correlated moderately with the

Nurickgrade(r=0.62),andwasresponsivetochangeasdemonstratedbyaCohen’s

effect size of 1. The authors did not however investigate inter‐ and intra‐rater

reliabilityofthemJOAandindicatethatthisremainsacurrentknowledgegap.The

authors concluded that their findings validate existing literature and that there is

justification for the use of this assessment tool in research studies going forward

(Kopjaretal,2014).

52

Table1.6AdescriptionofthemodifiedJapaneseOrthopedicAssociationscore(mJOA)scoreandhowitistabulated.ThemJOAisamodifiedversionoftheJOAscaleandwasdevelopedbyBenzeletal(1991).ItassessestheleveloffunctionalimpairmentinDCMpatientsbyevaluatingmotorfunctionsoftheupperandlowerlimbs,sensoryfunctionoftheupperextremities,andurinaryfunctionbyassessingsphinctercontrol.

mJOADefinitions Score

MotorDysfunctionScoreoftheUpperExtremities Inabilitytomovehands 0Inabilitytoeatwithaspoon,butabletomovehands 1Inabilitytobuttonshirt,butabletoeatwithaspoon 2Abletobuttonshirtwithgreatdifficulty 3Abletobuttonshirtwithslightdifficulty 4Nodysfunction 5 MotorDysfunctionScoreoftheLowerExtremities Completelossofmotorandsensoryfunction 0Sensorypreservationwithoutabilitytomovelegs 1Abletomovelegs,butunabletowalk 2Abletowalkonflatfloorwithawalkingaid(i.e.,caneorcrutch) 3Abletowalkupand/ordownstairswithhandrail 4Moderatetosignificantlackofstability,butabletowalkup

5and/ordownstairswithouthandrailMildlackofstabilitybutwalkswithsmoothreciprocationunaided 6Nodysfunction 7 SensoryDysfunctionScoreoftheUpperExtremities Completelossofhandsensation 0Severesensorylossorpain 1Mildsensoryloss 2Nosensoryloss 3 SphincterDysfunctionScore Inabilitytomicturate 0Markeddifficultywithmicturition 1Mildtomoderatedifficultwithmicturition 2Normalmicturition 3 TotalMaxscore 18

53

Table1.7NurickGradefortheassessmentofmyelopathyseverity.Ahighergradedenotesincreasingneurologicalimpairment(Nurick,1972).

GradeSpinalRootSigns

SpinalCordDisease

WalkingImpairment SocialIndependence

AbilitytoWorkFull‐

time0(LeastSevere) Yes No Unremarkable Yes Yes1 Yes Yes Unremarkable Yes Yes2 Yes Yes Mild Yes Yes3 Yes Yes Substantial MildlyImpaired No

4 Yes YesRequiresAssistance

ModeratelyImpaired No

5(MostSevere) Yes Yes Chair/Bedridden SeverelyImpaired NoNurickGrade,asoriginallypublishedin1972Grade0:Signsorsymptomsofrootinvolvementbutwithoutevidenceofspinalcorddisease.Grade1:Signsofspinalcorddiseasebutnodifficultyinwalking.Grade2:Slightdifficultyinwalkingwhichdidnotpreventfull‐timeemployment.Grade3:Difficultyinwalkingwhichpreventedfull‐timeemploymentortheabilitytodoallhousework,butwhichwasnotsosevereastorequiresomeoneelse'shelptowalk.Grade4:Abletowalkonlywithsomeoneelse'shelporwiththeaidofaframe.Grade5:Chairboundorbedridden.

54

1.6.2MedicalImaging

Inthesettingofsuspectedmyelopathy,patientsinitiallyundergoathoroughclinical

examination. However, since a number of conditions can result in neurological

impairmentandmaymanifest similarly,medical imaging isultimately required to

confirm a preliminary clinical diagnosis. In the following section, the various

neuroimagingmodalitiesavailablefortheassessmentofDCMarediscussed.

1.6.2.1ConventionalRadiography

Radiographsaretheoldestformofmedicalimagingavailabletothephysician.They

arecreatedbythedetectionofionizingradiationthathaseitherpassedorfailedto

pass through (absorbed) the patient upon external emission of X‐rays. Radiation

that is absorbed by the patient is not detected on the radiograph and appears as

white;conversely,X‐raysthatarenotabsorbedbypatientpassthroughandappear

black. Generally, most of the radiation passes through human tissue, with the

notableexceptionofbones,duetotheirlargecalciumcomposition.Radiographsare

usuallythefirstmodalityusedatthegeneralpractitionersofficefortheassessment

ofpatientswithsuspectedneurologicaldysfunctionoftheneck.Theseimagesallow

fortheanalysisofbonestructuresandmaybeabletodetectaberrantossificationof

ligamentswithin the spine, listhesis, anduncover congenitalmalformations.Their

depiction of the spine is limited to 2‐dimensional space and therefore multiple

radiographs are required to observe different spatial orientations. Oftentimes,

lateralviewcervicalspineflexionandextensionradiographsareordered;however,

this procedure may not be appropriate in patients with severely unstable spine

segments.

Inadditiontoviewingtheradiographs,specificmeasurementstoassessDCM

maybeperformed.Oneofthemostrelevantmeasurements, thecanaltovertebral

body ratio,was developed by Torg et al (1986) andPavlov et al (1987), and is a

measureofcervicalspinalcanalstenosis.Figure1.6displaysanexampleofhowthe

canaltovertebralbodyratioismeasured.Radiographsarelimited,however,since

imaginginanygivenplanesuperimposesallradiodensestructuresfromoneplane

55

intoasingleimageandbecausetheydonotallowfortheeffectiveassessmentofsoft

tissuestructures.

Figure1.6Aradiographshowingthecanaltovertebralbodyratiomethodforevaluationofcervicalspinestenosis[DevelopedbyTorgetal(1986)andPavlovetal(1987)].Aratioof1isconsideredtonormal;aratiooflessthan0.82indicatessignificantcervicalspinestenosis.

56

1.6.2.2ComputerizedTomography

Computerized tomography (CT) is a technique that utilizes tomographic data to

formulate a 3‐dimensional view of the spine, and additionally, permits the

individualanalysisoftransverse,sagittalandcoronalanatomicalplanes.Incontrast

to simple radiographs that superimpose all radiodense structures fromone plane

into a single image, CT generates individual “slices” of adjustable thickness. This

allows for the careful analysisof spine restructuringmanifesting inDCMpatients,

including osteophytosis, vertebral hourglass reshaping, anterior or posterior

vertebraldisplacement,aswellaspotentialossificationofsofttissuestructures(e.g.

ligamentum flavum, posterior longitudinal ligament ‐ see Figure 1.4B). As with

conventional radiography, contrast between tissues is dependent on the electron

densityoftissue(hinderingthepassageofelectromagneticradiation).OnCT,thisis

measuredintermsofHounsfieldunitswherewaterisassignedthevalue0;denser

values (such as bone) rangeupwards of +500ormore; and less dense structures

(suchasfatandair)rangedownwardsof‐500orless(Novelline,2004).Thehigher

the Hounsfield unit the brighter the tissue on imaging, such that dense tissue

appears white, less dense tissue appears dark, and air appears black (Novelline,

2004).Despitethis,softtissuestructuressuchasthespinalcordarenotdelineated

wellonCT.

OneofthemostrelevantmeasurementtechniquesusingCTinpatientswith

DCMistheassessmentofmaximumcanalcompromise(MCC)describedbyFehlings

et al (1999).Themethod for assessingMCCviaCT isdemonstrated inFigure1.7.

InformationregardingtheutilityofMCCisdiscussedinchapters3&4ofthethesis.

InadditiontousingconventionalCTfortheassessmentofDCM,contrastmay

be injected into the spine to provide better visual delineation between the spinal

cord and surrounding tissue – this method is referred to as a CT myelography.

Magnetic resonance imaging has largely replaced CT myelography. However, the

latter remains a useful alternative in patients with contraindications to MRI

examination,mostnotablythosewithferromagnetic implants(e.g.aneurysmclips,

pacemakers).

57

Figure1.7Measurementofthemaximumcanalcompromise(MCC)ofthecervicalspineonCT.MCCismeasuredbydividingthecanaldiameterattheregionofinterest(Di)bytheaverageofthenormalcanaldiametersizeabove(Da)andbelow(Db)(Fehlingsetal.,1999).MoreinformationaboutthecalculationisdescribedinTable3.1.Notethevertebraebodiesoflevel5and6appeartohaveafused.

58

1.6.2.3ConventionalMagneticResonanceImaging(MRI)

Whilemanyoftheanatomicalfeaturesconsistentwithadegeneratedspinecanbe

demonstratedbytheaforementionedimagingtechniques,theydonotallowforthe

clearinspectionofthespinalcord.MRIisthegoldstandardimagingmodalityforthe

analysisofsofttissuestructures,includingthenervoussystem,andistheprincipal

method for analysis of the spinal cord. Like CT, MRI can acquire multi‐planar

imagingoftissuethatfacilitatestheassessmentofregionsofinterestfrommultiple

angles.UnlikeCThowever,MRIusesradiofrequencyinsteadofionizingradiationto

formulatemedicalimaging(Rungeetal.,2014).

Althoughanumberofcomplexmethodshavebeendevelopedthatextendthe

utility of MRI beyond assessing anatomy [i.e. functional MRI (fMRI) techniques],

basic diagnostic MRI methods rendering T1‐weighted imaging (T1‐WI) and T2‐

weighted imaging (T2‐WI) remain thebasic techniquesapplied inclinicalpractice

fortheassessmentofsuspectedmyelopathy.MRIisalsousefulfortheassessmentof

anatomicalaberrationsinspinestructure,includingthoseidentifiablebyCTaswell

asdegenerativepathologiesinvolvingsofttissues(i.e.IVDsandligaments).

MRI technology is based on a complex application of both classical theory

and quantum physics that extend beyond the scope of the current discussion.

However,astheroleofMRI isbeingstudiedanddiscussedthroughoutthisthesis,

some background into how these images are created using this technology is

appropriate.

MRexploitstheinherentmagneticpropertiesofthemostabundantatomsin

ourbody,thehydrogenatom.IntheMRImachine,astaticmagneticfieldiscreated

that orients the hydrogen atoms in the body into parallel alignment with the

magnetic field. In this direction, the atoms wobble (or precess) at a frequency

knownastheLarmorfrequency.Whenaradiopulseisusedtoperturbthehydrogen

atoms at the same frequency (i.e. the Larmor frequency), hydrogen atoms absorb

energy,begintoorientthemselvesperpendicular(transverse)tothemagneticfield

andcoherewiththefrequencydirection(in‐phase).Whenthisoccurs,signalisgiven

offwhichisdetectedbyacoil.Oncethepulse isstopped,themagnetizationofthe

59

atomsbeginstofallbackintoalignmentwiththemagneticfieldduetolossofenergy

tosurroundingatomicnuclei.Thetimeittakesfor63%ofhydrogenatomstolose

their energy and reorient to the initialmagnetic field is termed theT1 relaxation

time.Coherenceofmagnetization fromthehydrogenatomsdecaysover time.The

timeittakestolose63%ofthiscoherenceiscalledtheT2relaxationtime.

Inorderforimagingtopossessclinicalutility,contrastbetweentissuesmust

bedemonstrated.MRachievesthisbydepictingareasofhighsignal(bright)versus

areasoflowsignal(dark).WithCTimaging,tissuesarediscriminatedbasedononly

onemetric,x‐rayattenuation,whichhasatightrangeforsofttissuesandmakessoft

tissuediscriminationpoor.Bycontrast,tissuediscriminationinMRIisbasedontwo

independentmetrics,T1andT2relaxation,whichhavewidevariationinsofttissues

andtherebyenablesuperiordiscrimination.

In individualswithout pathology, the spinal cord generally demonstrates a

homogenous signal intensity (grey appearance) on both T1‐WI and T2‐WI.

However,patientswithDCMmaypresentwithhypointense (dark) signal changes

on the spinal cord on T1‐WI, and more commonly, hyperintense (bright) signal

changesonT2‐WI.ExamplesofthesechangesarepresentedinFigure1.8.Thereare

multiplemethods to assess these signal changes, specifically viaT2‐WI, and these

arediscussedinfurtherdetailinthemethodssectionofchapter3.

Attempts have beenmade to correlate specific histopathology of DCM and

signal changes on MRI. Table 1.8 summarizes current thoughts on MRI

characteristicsseeninpatientswithDCMandtheirhistopathologicalcorrelation.

60

Figure1.8T1andT2weightedMRIsofthesamepatientwiththepresenceofsignalchangesandaconfirmeddiagnosisofDCM.A)AT1weightedMRIwithhypointensitychangesinthecervicalspinalcord.B)AT2weightedMRIwithhyperintensitychangesinthecervicalspinalcord.

61

Table1.8CorrelationbetweenMRIcharacteristicandpathobiologicalchangesaswellascurrentthoughtsrelatedtostructurerecoverypotentialafterdecompressionsurgery.T2‐WIsignalhyperintensitychangesarenonspecificandmayrepresentbothreversibleandirreversiblestructuralalterations.ItshouldbenotedthatpatientsmighthaveacombinationofbothweakandstrongT2‐WIsignalhyperintensityelements.T1‐WIsignalhypointensitychangesmorespecificallyrepresentfrankneuraltissueloss(Tetreaultetal.,2013a,Chenetal.,2001,Karpovaetal.,2013a,Al‐Meftyetal.,1988,Mehalicetal.,1990).

MRICharacteristicsLikelyPathobiological

CorrelationStructuralchange

T2‐WI

WeakSignalHyperintensity(brighterthanspinalcordbutdarkerthanCSFsignalintensity)appearing:Bright,fuzzy,diffuse,withoutclearborder/delineation

Edema WallerianDegeneration Demyelination Ischemia Gliosis

ConsideredReversible

StrongSignalHyperintensity(approachingisointensitywithCSF)appearing:Bright,focal,sharp,clearborder/delineation

Cavitation Neuraltissueloss Myelomalacia Necrosis Spongiformchangesin

graymatter

Consideredlargely

Irreversible

T1‐WI

Remarkablepresenceofsignalhypointensity(approachingisointensitywithCSF)appearing:Dark,focal,faint

Cavitation Neuraltissueloss Myelomalacia Necrosis Spongiformchangesin

graymatter

Consideredlargely

Irreversible

62

1.6.2.4AdvancedMRFunctionalImagingTechniques

TheadventoffunctionalimagingtechniquesbasedonMRItechnologyhavebrought

about a plethora of capabilities that may integrate into the diagnostic and

prognostic evaluation of patients with DCM in the future. Having said this,

accessibility to such potential technologies will likely be limited. Unlike

conventionalMRI,whichfacilitatesthedetailedinspectionofanatomicalstructures,

more advanced imaging techniques, allow for the evaluation of function, more

detailed anatomical structure, and physiology. A couple of these, diffusion tensor

MRI (dtMRI) and fMRI ‐ specifically the BOLD (Blood‐Oxygen‐Level Dependent)

sequence‐areparticularlyinterestingfortheinvestigationofthespinalcord.

DiffusionweightedMRIassessestheamountofdiffusionofwatermolecules

in various tissues and is the technology behind dtMRI. In areas with restricted

boundaries, such as membranes or ligaments, the motion of water molecules is

restricted.Thenetdisplacementofwatermoleculesacrossatissueareaovertime

(seconds)istermedtheapparentdiffusioncoefficient(Westbrooketal.,2011).The

diffusiongradientcanbeappliedinanydirectiontoassessthedirectionalmotility

ofwater.Thisenables3Dreconstructionoffibersandspecificvisualizationofwhite

mattertractsinthecentralnervoussystem.

Recent work using dtMRI has indicated diagnostic utility by showing that

damagetowhitemattertractscanbeidentifiedbeforesignalchangesappearona

T2‐WI(Jonesetal.,2013).Fractionalanisotropyvaluesand fiber tractratioshave

alsoshownsignificantcorrelationtosurgicaloutcome,indicatingaprognosticvalue

aswell(Tetreaultetal.,2013a,Nakamuraetal.,2012,Jonesetal.,2013)However,

the clinical application of dtMRI remains largely limited by long processing times

anduserdependencyoffibertracking(Rungeetal.,2014).

Blood‐Oxygen‐Level dependent MRI is another technology that has a

potential application in the evaluation of myelopathy. It takes advantage of the

different levelsofmagnetizationbetweenoxygen‐boundhemoglobinandunbound

hemoglobin in the blood. During BOLD examination, increased blood flow, and

subsequentoxygenation, isobservedduringneuronalactivationofmoving fingers

63

forexample(Rungeetal.,2014).Whenincreasedbloodflowstotheareaofinterest,

thedifference inmagnetizationintheblood, thoughsmall,canbeappreciatedand

evaluated.

ThoughBOLDhas great potential to become an important assessment tool

for uncovering pathobiological mechanisms underlying spinal cord injury, its

applicationatthepresenttimeremainsfundamentallyinvestigational.

64

1.7TreatmentofDegenerativeCervicalMyelopathy

Treatment of DCM can be approached either non‐operatively, or through surgical

managementthat is focusedonspinalcorddecompression.Therelativedifference

ofefficacybetweenthesetwooptionshasbeendifficulttoconclusivelydemonstrate

duetoadearthofliteratureonthesubjectandalackofrandomizedcontroltrials.

FromtheevidencethathasarisenthroughasystematicreviewbyRheeetal(2013),

it appears that non‐operative management for myelopathic patients can be

cautiously considered in those who are believed to be mildly myelopathic.

Conversely, inpatientswithmore severemyelopathy, non‐operativemanagement

hasbeenshowntobelesseffectivethansurgerywithrespecttofunctionaloutcome,

pain,andneurological function(Rheeetal.,2013).Ultimately, it istheobjectiveof

bothoftheseapproachestopreventfurtherneurologicaldeterioration,slowdisease

progression, and improve neurological function as well as quality of life. In the

followingsectionstheroleofbothnon‐operativeandsurgicalmanagementforthe

treatmentofDCMisdiscussed.

1.7.1Non‐operativeManagement

Non‐operative management includes exercise therapy, anti‐inflammatory

medicationandpromotionoflifestylechanges,suchasdiscouragementofhigh‐risk

activities and encouragement of intermittent bed rest. Table 1.9 outlines a more

extensivelistofthenon‐operativemanagementtechniquesreportedinliterature.

How rigorously non‐operative management is instituted may play an

importantroleintreatmentefficacy.Yoshimatsuetal(2001)reportedthatrigorous

management inthosewithseveremyelopathy(mJOA<13) ismoreefficaciousthan

nonrigorousmanagement.Theauthorsdefinedrigorousmanagementasaninitial1‐

3monthof traction(Good‐Samaritanmethod),cervicalspine immobilization,drug

aswellasexercisetherapy,andorthosis,thermalanddrugtherapyduringregular

visits to the outpatient clinic thereafter. Though presumably less intensive,

nonrigorous non‐operative therapy was not described. However, despite the

possibility to achieve more favorable results with more rigorous non‐operative

65

management, surgically treatedpatientsstillachievedbetterresults in theirstudy

(Yoshimatsuetal.,2001,Rheeetal.,2013).

Given that DCM is a progressive disorder, non‐operative management is

limited in that it does not adequately address the underlying issue, but there are

instancesinwhichsurgeryisnotaviableoption(i.e.riskstosurgeryoutweighthe

potential benefits). Such a situation may arise in patients with significant

comorbidities or advanced age. Other times, patients may simply refuse surgical

treatment. Accordingly, in addition to close follow‐up these patients may benefit

from non‐operative management. Having said this, however, current evidence‐

basedrecommendationsonthesubjectindicatethatthereisinsufficientevidenceto

supporttheuseofnon‐operativetreatment(Rheeetal.,2013).

Table1.9Alistofnon‐operativetherapeuticapproachesforthemanagementofDCM.NSAID=Nonsteroidalanti‐inflammatorydrug(Kadankaetal.,2002,Kadankaetal.,2011,Yoshimatsuetal.,2001,Sampathetal.,2000).

Non‐operativeManagementTechniques

MedicalTherapy Pharmacologicaltherapy(e.g.Steroids,NSAIDs,Narcotics) Thermaltherapy Orthosistherapy Exercisetherapy Cervicaltractiontherapy(e.g.Good‐Samaritanmethod) Localspinalinjections(i.e.nerveblocks,facetblocks,epiduralsteroids)

LifestyleAdjustments Bedrest Discouragementofhigh‐riskactivities Avoidanceofriskyenvironments(e.g.slipperysurfaces,spinemanipulation

therapies)

1.7.2SurgicalManagement

The purpose of surgical treatment of DCM is cord decompression, spine

stabilization,andpreventionoffurtherinsulttothecord.Althoughtheobjectiveis

66

clear, the means of achieving decompression is much more complex. As was

outlinedearlier inthepathogenesissection,patientswithDCMcanpresentwitha

widerangingsetofdegenerativechanges,andthus,therepresentsnoonesurgical

approachthatisappropriateforallpatients.Whilemultiplestudieshaveattempted

to synthesize an ideal algorithm that surgeonsmay use in the treatment of DCM

patients,theyhavefallenshort(Lawrenceetal.,2013b).Table1.10outlinessurgical

methodscommonlyusedtotreatDCM.

Thesurgicalmethodrequiresanumberofconsiderationsthathavebeenwell

described in a consensus statement of the surgical management of DCM by

Lawrenceetal(2013b).Theystatethatthefollowingfactorsshouldbeincludedin

surgicaldecision‐making:

Sagittalalignment Anatomicallocationofthecompressivepathology Numberoflevelsofcompression Presenceorabsenceofinstabilityorsubluxation Thetypeofcompressivepathology(e.g.Spondylosis,OPLL) Neckanatomy Bonequality Surgeonexperienceandpreference

Afterconsideringthesefactors,thesurgeonmustdecideiftheyaretoapproachthe

spinefromtheanterior,posterior,oracombinationofthetwo(two‐stepsurgery).A

posterior approach is often chosen when multiple spinal levels are involved

(Lawrence et al., 2013b). However, the choice of surgical approach is still largely

dependenton the locationof the compressive force.While there is an insufficient

levelofevidencetoindicatethatoneapproachissuperiortotheotherintermsof

JOAscore improvement, there ismoderatestrengthofevidence favoringposterior

surgery over anterior surgery with regards to increasing sagittal canal diameter

(Lawrenceetal.,2013a). In theirsystematicreview,Lawrenceetal (2013a)make

the following evidence‐based clinical recommendation: “We recommend a

individualized approach when treating patients with CSM accounting for

pathoanatomical variations (ventral vs. dorsal, focal vs. diffuse, sagittal, dynamic

67

instability), as there appears to be similar outcomes between the anterior and

posteriorapproachesinregardstoeffectivenessandsafety.”

Along with deciding the surgical approach, the surgeon must also decide

what technique to use. For posterior approaches, there are a fewmethods at the

disposal to the surgeon, but two procedures are most often performed:

Laminoplasty, and laminectomy with fusion. The relative efficacy between these

procedureswasassessedbyasystematicreviewbyYoonetal(2013).Foranterior

approaches,therearealsoanumberofpotentialproceduresthatcanbeperformed.

The comparative effectiveness between these anterior surgical options were

assessedbyShamjietal(2013)inaseparatesystematicreview.Theevidence‐based

recommendations based on the findings of these authors are presented in Table

1.11.

68

Table1.10A list of commonly performed anterior and posterior surgical procedures for thetreatmentofDCM.

Table1.11Evidence‐basedrecommendationsregardingthecomparativeeffectivenessofanteriorandposteriorsurgicaltechniques.

Anterior Posterior Discectomy Corpectomy Discectomy‐CorpectomyHybrid

Approach Arthroplasty ObliqueMinimallyDestructive

Corpectomy

Laminectomy LaminectomyandFusion Laminoplasty SkipLaminectomy

Evidence‐BasedClinicalRecommendations OutcomesAfterLaminoplastyvs.LaminectomyandFusion(Yoonetal.,2013).

Recommendation:“ForCSM,evidencesuggeststhatlaminoplastyandlaminectomy‐fusionprocedurescanbesimilarlyeffective.Wesuggestthatsurgeonsconsidereachcaseindividuallyandtakeintoaccounttheirownfamiliarityandexpertisewitheachprocedure.”StrengthofEvidence=LowStrengthofRecommendation=Weak

AnteriorSurgicalOptionsforCSM.(Shamjietal.,2013).

Recommendation:“Whenpathoanatomicallyappropriatewithminimalretrovertrebraldisease,werecommendtheselectionofmultiplediscectomyovercorpectomyordiscectomy‐corpectomyhybridprocedures.”StrengthofEvidence=LowStrengthofRecommendation=Strong

69

1.7.3PharmacologicalTreatment

Pharmacological treatment of DCM beyond pain control and anti‐inflammatory

medication does not presently exist. However, the growing understanding of the

underlying pathobiological factors involved in the development of SCI, as was

discussedingreaterdetailearlierinthechapter,haveopenedthepossibilityforthe

investigationoftwonotabletypesof therapeuticagents for thetreatmentofDCM:

(1)Neuroprotectiveagentsaimedatmitigatingsecondary injurymechanisms;and

(2)Neuroregenerativeagentsaimedatpromotingrepairandregenerationofaxons

aswellasthesupportingcellularmatrix(Wilsonetal.,2013b).

Neuroprotectivedrugs,suchasMethylprednisoloneandRiluzole(discussed

further in the following section),may be useful in patientswho are experiencing

ongoing compression of the spinal cord, or are at high risk of injury. The

mechanisms of these drugs differ and therefore their potential therapeutic

applicationandtreatmentwindowsmayvaryconsiderably.Ingeneralhowever,itis

the objective of neuroprotective agents to contain the epicenter of trauma, and

therefore, starting treatment either prior or immediately after injury is desirable

(Wilson et al., 2013b). Additionally, these drugsmay also play a role in reducing

reperfusioninjuryafterdecompressionsurgery.

Neuroregenerativeagentsmayplayausefulroleinpatientswithdebilitating

neurologicaldysfunctionandinwhomsubstantialneurologicalrecoveryisunlikely.

Neuroregenerativeagentsincludebothdrugs(e.g.GangliosidesorGM‐1)andstem

cells (various types includinghumanembryonicstemcells) (Wilsonetal.,2013b).

TheseagentshavebeenspecificallyinvestigatedforthetreatmentoftraumaticSCI,

butnodiscernableeffectivenesshascurrentlybeenshowninhumantrials.Having

said this, it could be conjectured that if such treatment was efficacious, it could

likewisebeuseful intreatingDCM.Despitetheirexcitingpotential,however,these

agentsremainlargelyinvestigationalforthetimebeing.

70

1.7.3.1Riluzole:APotentialTherapeuticAdjuncttoSurgical

Management

Riluzole is an FDA‐approved therapeutic agent for the treatment of Amyotrophic

Lateral Sclerosis and has been shown in animal models to exert its function by

decreasing sodium and glutamate‐mediated secondary injury (Fehlings et al.,

2013a). It is currentlybelieved thatpart of thepathobiologyofDCM is related to

excessive local release of the excitatory neurotransmitter glutamate and that this

resultsinlocalexcitotoxicityandsubsequentcelldeathinthespinalcord(Fehlings

etal.,2013a).Onthebasisofthesefindings,Riluzoleiscurrentlybeinginvestigated

inPhaseIIIclinicaltrials(CSM‐ProtectTrial1)foritssafetyandtherapeuticpotential

inpatientswithDCMundergoingsurgicaltreatment.Specifically,patientshavebeen

randomizedtoreceiveRiluzoleoraplacebo28daysbeforesurgicaldecompression.

1https://clinicaltrials.gov/ct2/show/NCT01257828

71

1.8Prognosis:PredictingSurgicalandNon‐SurgicalOutcome

Incounselingpatientsregardingtheirtreatment,anessentialpointofdiscussionis

howtheycanexpecttofaregiventheirtherapeuticchoice,andhowtheycanexpect

to fare if no treatment is instigated at all. Unfortunately, with regards to DCM,

answeringthesequestionsusinghighlevelsofevidencehasbeenagreatchallenge.

Althoughthereareclinicalaswellasimagingmeasuresthathavebeenshowntobe

predictiveofoutcome,manyofthemstillneedtobevalidated.

A general consensus on predictors of outcome does exist in the form of

expertopinionhowever.InarecentsurveyofmembersoftheinternationalAOSpine

community by Tetreault et al (2014), therewas consensus that both clinical and

imaging factors were predictive. Most notably, preoperative age and a lowmJOA

score were important factors that related to poor outcome. Moreover, members

fromallgeographicallocationswiththeexceptionofEuropeindicatedthatthetop

two predictors of outcome were duration of symptoms and the preoperative

severityscore(inEuropepresenceofmyelopathicsymptomswasratedhigherthan

baseline severity). In terms of imaging predictors, there was international

consensusthatMRIisanimportantprognostictoolandthatthepresence/absence

ofsignalchangeonthespinalcordonT2‐WIandT1/T2‐WIarethemostimportant

of these (Tetreault etal., 2014).However, findingsof the surveyhave tobe taken

intoconsiderationcautiously. Insomecases, thecollectiveopinionofrespondents

waserroneousbasedoncurrentliterature,asTechyandBenzel(2014)pointedout.

In the followingsection,discussionofpredictorsofoutcome in thesurgical

setting will be divided into those derived from clinical assessment and those

obtained throughmedical imaging, specifically viaMRI analysis. However, not all

patients with spinal cord compression present with myelopathy. Therefore,

predictorsof thenaturalprogressioninthenon‐operativesettingwillbeexplored

first.

72

1.8.1 Predictors of Myelopathy Development in Patients with

CervicalSpinalCordCompression,CanalStenosisand/orOPLL

Patientswithdegenerativechangesof thespineandcompressionof thecordmay

present with or without myelopathy. In those patients who do present with

symptoms, 20‐62% will deteriorate neurologically at 3‐6 years of follow‐up if

treated non‐operatively (Fehlings et al., 2013b). However, no specific disease or

patient characteristics have been shown to predict the course of this progression

reliably (Fehlings et al., 2013b). In nonmyelopathic patients with spinal cord

compression secondary to degenerative changes, the rate of myelopathy

development approximates 8% and 23% at 1 year and 4 years of follow‐up,

respectively(Fehlingsetal.,2013b).

In a recent systematic review, Wilson et al (2013a) discussed specific

predictors of neurological dysfunction in nonmyelopathic patients with

degenerative spine disease. The authors identified two studies by Bednarik et al

(2008)&(2011)onaprospectivecohortofCSMpatientswithoutclinicallyevident

myelopathy.Inthefirstofthese,thepresenceofradiculopathy,T2‐WIcervicalcord

hyperintensity, and prolonged somatosensory‐ andmotor‐evoked potentialswere

independent predictors of subsequent myelopathy development (Wilson et al.,

2013a, Bednarik et al., 2008). Interestingly, Bednarik et al (2008) found that the

absenceofT2‐WIspinalcordhyperintensityonMRIwasindependentlypredictive

of myelopathic development at ≤12 months follow‐up in subjects with

asymptomatic cervical spinal cord compression and that the presence of T2‐WI

signalchangewasrelatedwithlatedevelopment(>12months).Themeaningofthis

findingisdifficulttoexplainandexemplifiesthedifficultywithrelatingMRIfindings

withthestateofmyelopathicdisease. Intheirsecondstudy,Bednariketal(2011)

investigated the risk of symptomatic myelopathy after minor trauma in patients

with asymptomatic compression after conducting a questionnaire and data file

analysis.Theyconcluded, “SCIafterminor traumaof thecervicalspine inpatients

withasymptomaticcordcompressionappearedtobe lowprovidedriskyactivities

intheseindividualsisrestricted”.

73

Wilsonetal(2013a)alsoidentified3articlesonthefrequencyofmyelopathy

development in OPLL patients (Fujiyoshi et al., 2010, Matsunaga et al., 2008b,

Matsunagaetal.,2004),andfoundthattheprogressiontomyelopathyinthisgroup

ofpatientsrangedbetween0‐61.5%.Theauthorsofoneofthesearticles,Matsunaga

et al (2008b), identified canal stenosis (≥60%), laterally deviated OPLL, and

increased cervical range of motion as significant predictors of myelopathy

developmentintheirprospectivemulticenterinvestigation(Wilsonetal.,2013a).

1.8.2 Comparison ofOutcomes inNon‐operativeManagement vs.

SurgicallyManagedDCMPatients

In a recent systematic review, Rhee et al (2013) investigated the efficacy of non‐

operative treatment vs. surgery. Though their reviewwas limited by the relative

smallamountofresearchthathasbeenconductedonthesubject,someinteresting

findingswere reported. They indicate that low strength of evidence derived from

Kadankaetal(2002)&(2011)supportsthatpatientswithmildcervicalmyelopathy

treatednon‐operativelymayfareaswellas thosewhoundergosurgery‐ thiswas

demonstratedby a comparablemJOA scorebetween the groups throughout a10‐

year follow‐up period. Rhee et al (2013) caution however, that patients who

underwentsurgicaltreatmentdidnotimprovefromtheirbaselinestatusasmuchas

patients from other surgical studies, which could potentially explain the lack of

difference. In terms of patients withmoderate or severe DCM, Rhee et al (2013)

identifiedtwoseparatestudies(Sampathetal.,2000,Yoshimatsuetal.,2001)that

foundthatsurgicaltreatmentresultsinasignificantimprovementofoutcomewhen

compared to non‐operated patients. Both studies demonstrated not only that

patients with surgery improved from baseline but also that patients treated

conservatively deteriorated. Collectively, these findings provide low strength of

evidence that baseline severity seems to play a predictive role in terms of how

patientscanexpecttofarewithorwithoutsurgicalintervention.

74

1.8.3ClinicalPredictorsofOutcomesofSurgicalTreatment

TheroleofclinicalparametersinpredictingsurgicaloutcomeinpatientswithDCM

has been investigated by a number of researchers. Evidence derived from two

recentsystematicreviews(Hollyetal.,2009,Tetreaultetal.,2013b),andarecent

predictionmodel based on clinical parameters (Tetreault et al., 2013c), supports

theirutilityandformthebasisofcurrentknowledgeonthesubject.

Intheirreview,Hollyetal(2009)identifiedClassIIevidencetosupportthat

sensory evoked potentials (SEPs) have a role in predicting surgical outcome.

Specifically, they state “normal preoperative median nerve potentials and/or

normalization of potentials in the early decompression period appear to be

associatedwithamorefavorableoutcome”.Hollyetal(2009)alsoidentifiedClass

III evidence to support age, duration of symptoms as well as preoperative

neurologicalfunctionaspredictiveparameters.

In themore recent systematic review, Tetreault et al (2013b) determined

thatthemostimportantpredictorsofoutcomeswerepreoperativediseaseseverity

(establishedthroughclinicalassessmentmeasures)aswellasdurationofsymptoms

– indicating that patients with prolonged symptoms and greater preoperative

dysfunctionaremore likelytohaveanunfavorablesurgicaloutcome.Additionally,

theauthorspointedtoanumberofotherpotentialpredictors,includingage,specific

clinical manifestations (e.g. hyperreflexia, gait impairment), comorbidities (e.g.

diabetes, psychological impairment), and smoking status that merit further

investigation.

Informationderived from these reviewsprovided the scientific prelude for

the synthesis of a clinical prediction rule by Tetreault et al (2013c). Using

prospective andmulticenter population data derived from theAOSpine‐NA study,

the authors describe amodelwhich included 6 clinical and 1 imaging parameter,

including:Age(OR=0.97,95%CI=0.94‐0.99),durationofsymptoms(OR=0.78,

95%CI=0.61‐1.00),smokingstatus(OR=0.46,95%CI=0.21‐0.98),impairmentof

gait(OR=2.66,95%CI=1.17‐6.06),psychologicalcomorbidities(OR=0.33,95%CI

= 0.15‐0.69), baseline mJOA score (OR = 1.22, 95% CI = 1.05‐1.41), and the

75

preoperativetransverseareaofthespinalcordmeasuresonMRI(OR=1.02,95%CI

=0.99‐1.05)(Tetreaultetal.,2013c).Using these factors, theauthorsdevelopeda

model that discriminates a “successful” and “failed” postsurgical outcome (mJOA

≥16 andmJOA <16 as defined by the authors) at 1‐year with an area under the

receiveroperatingcurveof0.79.

1.8.4ImagingPredictorsofSurgicalOutcome

The potential of imaging factors to predict the outcome of patients with DCM

undergoing surgical decompression has been assessed by numerous investigators

over the years.Unfortunately, findings from these studieshavebeen inconsistent.

Thereasonsforthisaremultifoldbutcanbelargelyattributedto(1)theutilization

ofdifferentmethodsforanalyzingMRIs,(2)correlationofMRIfindingswithdiverse

clinical assessmentmeasures, (3) investigations at single centers, and (4) lack of

external validation. In an effort to provide clarity with regards to how imaging

factors can be utilized in prognostication, four recent reviews have evaluated

literatureassessingtheroleofMRIinpredictingsurgicaloutcome(Tetreaultetal.,

2013a,Lietal.,2011,Karpovaetal.,2013a,VedantamandRajshekhar,2013).The

summaryoftheirresultsisdisplayedinTable1.12.Collectively,theconclusionsof

thesepapersformthefoundationofthepresentknowledgeonthesubject.

In their systematic review onMRI predictors of outcome in DCM patients,

Karpova et al (2013a) conclude that there seems to be good evidence to suggest

transversearea(TA)ofthespinalcordcorrelateswithrecoveryratiobutnotwith

post‐operativefunctionalscoresasassessedbyJOA/mJOA.Theauthors’findingsare

alsosupportiveofthepredictivevalueofsignal intensity(SI)changes(intermsof

presenceonT2‐WI,extent,brightnessandpresenceonbothT1/T2‐WIs)onsurgical

outcomesasmeasuredbytheoriginalormodifiedversionofJOA,Nurickgrade,or

Neurosurgery Cervical Spine scale (Karpova et al., 2013a). However, in a

prospective single center study published by Karpova et al (2013b) in the same

year, neither TA nor signal changes (on T1‐WI and T2‐WI)were associatedwith

surgical outcome in patients with DCM – though TA did correlate with baseline

functional status. In the systematic review by Tetreault et al (2013a) on MRI

76

characteristics that affect treatment decision making and prediction of clinical

outcome,theauthorsconcludedbasedonlowstrengthofevidencethatthereare3

negative predictors of outcome: Number of SI segments on T2‐WI, presence of

combinedT1/T2‐WIsignalchange,andahighSIratioonT2.

Li et al (2011) conducted a meta‐analysis on the role of T2‐WI

hyperintensitytopredictsurgicaloutcome.Theauthorsconcludethatpatientswith

hyperintensity changes of the spinal cord on T2‐WI had worse postoperative

recovery (measured using the JOA recovery ratio) than patients without signal

change. However, they acknowledged that their conclusions are limited given the

relativelysmallnumber(n=5)ofarticlesavailableonthesubject.Theauthorsalso

caution that various other factors that could possibly affect prognosis were not

controlled for in their analysis, and that further high quality investigation is

necessarytosubstantiatetheirresults(Lietal.,2011).

Intheirrecentreview,Vedantam&Rajshekhar(2013)notedthechallengeof

interpretingresults fromstudiesassessing theroleofT2‐WIsignalhyperintensity

on surgical outcome, corroborating Karpova et al (2013a). However, the authors

concluded that Level II evidence supports that multisegmental T2‐WI

hyperintensityandsharp,intenseT2‐WIhyperintensityareassociatedwithpoorer

surgical outcome, and that regression of T2‐WI postoperatively correlates with

betterfunctionaloutcomes(VedantamandRajshekhar,2013).

Despite many attempts to elucidate the role of MRI in predicting surgical

outcome as evinced by the outlined reviews above, the findings have been

challenging to translate into practice. Specifically, at the present time the key

knowledgegapsinclude:

1) There remains no strong recommendation to guide how MRI factors can be

relatedtooutcome.

2) Therehavebeennomulticenterstudiesthathaveattemptedtoinvestigatethe

roleofMRIfactorsinpredictingoutcome.

3) MultipleMRI factorshavenotbeenadequately investigated inamultivariate

fashion.

77

Table1.12AsummaryofrecentreviewsthathaveevaluatedtheroleofMRIcharacteristicsinpredictingsurgicaloutcomeinpatientssurgicallytreatedforDCM.ReviewsonMRIcharacteristicsandtheirabilitytopredictsurgicaloutcome

MostImportantFindings

Systematicreviewofmagneticresonanceimagingcharacteristicsthataffecttreatmentdecisionmakingandpredictclinicaloutcomeinpatientswithcervicalspondyloticmyelopathy.(Tetreaultetal.,2013a)

Theauthorsidentified3imagingassessmentmeasuresthatdemonstratedanegativeassociationwithoutcomebasedonlow‐qualityevidence,including(1)multiplehyperintensitysegmentsonT2‐WI,(2)presenceofcombinedT1‐WIandT2‐WIsignalchange,and(3)deviationsinsignalchangeratios.WhileotherevaluatedMRIparametersmaybeimportant,therewasaninsufficientamountofevidencetoprovidearecommendation.Theauthorsprovidethefollowingevidence‐basedrecommendation:“T2signalmaybeausefulprognosticindicatorwhenusedincombinationwithlowSIchangeonT1‐WI,orasaratiocomparingcompressedwithnoncompressedsegments,orasaratioofT2comparedwithT1‐WI.WesuggestthatifsurgeonsuseMRIsignalintensitytoestimatetheriskofapooroutcomeaftersurgery,theyusehighSIchangeonT2WIincombinationwithothersignalintensityparameters,andnotinisolation.”StrengthofEvidence:LowStrengthofRecommendation:Weak

Assessmentofspinalcordcompressionbymagneticresonanceimaging‐‐canitpredictsurgicaloutcomesindegenerativecompressivemyelopathy?Asystematicreview.(Karpovaetal.,2013a)

Theauthorswerenotabletoconcludebasedontheirreviewthatcompressionofthecordintermsoftransverseareaorasacompressionratiowaspredictiveofsurgicaloutcome.Theauthorsconclude:“SIchangesdefinedby(1)itspresenceonT2‐WI,(2)itsextent(focalormultisegmental),(3)itsbrightness,and(4)itspresenceonbothT1‐/T2WIcanpredictsurgicaloutcomesindegenerativecompressivemyelopathy.”LevelofEvidenceforconclusion:2

78

Table1.12(continued)“Ameta‐analysisshowingthathighsignalintensityonT2‐weightedMRIisassociatedwithpoorprognosisforpatientswithcervicalspondyloticmyelopathy.”(Lietal.,2011)

Theauthorsconductedameta‐analysisusing5studiesanddeterminedthattheJOArecoveryratiowaslowerinpatientswithT2‐WIhyperintensitythanpatientswithoutT2‐WIsignalhyperintensity.TheirresultssuggeststhatT2‐WIisanegativepredictorofsurgicaloutcome.

“DoesthetypeofT2‐weightedhyperintensityinfluencesurgicaloutcomeinpatientswithcervicalspondyloticmyelopathy?Areview.”(VedantamandRajshekhar,2013)

Intheirreview,theauthorsconclude“BothmultisegmentalT2‐WISIandsharp,intenseT2‐WISIareassociatedwithpoorersurgicaloutcome.TheregressionofT2WISIpostoperativelycorrelateswithbetterfunctionaloutcomes”.LevelofEvidence:ClassII

79

CHAPTER2

Hypothesis,ResearchObjectives,andRationale

80

Hypothesis:

It is theprincipalhypothesis thatMRI featuresencountered inpatientswithDCM

preoperatively, including anatomical alterations (canal stenosis and cord

compression) aswell assignal intensity changes of the spinal cord on T1 and T2

weighted imaging (size and degree of deviation)will be predictive of surgical

outcomeasdeterminedbyfunctionalassessmentusingthemJOA.

ResearchObjective:

Using data derived from patients enrolled in the prospective, multicenter CSM‐

North America study, it is the objective to investigate the prognostic role of

preoperativeMRIfeaturesinpatientsundergoingsurgicaltreatmentforDCM.

SpecificObjectives:

1) To quantitatively assess pathological MRI features in DCM using three

approaches, (1) T1‐weighted signal analysis, (2) T2‐weighted signal analysis,

including size and intensity deviation, and (3) anatomical pathologies (canal

stenosisandspinalcordcompression),anddeterminetheassociationwithsurgical

outcomeasassessedbythemJOAscore.

2)Todevelopapredictionmodelthatcandiscriminatebetweenpatientsintothose

withmildmyelopathyandthosewithsubstantialresidualneurologicalimpairment

basedonadichotomized(>16,≤16)postoperativemJOAscore.

Rationale:

The constellation ofMRI characteristicsmanifesting in patientswith DCM can be

assessedquantitativelyusingmultiplemethods.WhiletheseMRImeasurementsare

useful in confirming the clinical diagnosis, their utility in predicting surgical

outcome remains unclear. The AOSpine‐NA study is the first prospective and

multicenterstudyassessing theefficacyof treatmentofsurgicaldecompression in

DCMpatients,andprovidesaplatformto investigate theroleofMRI inpredicting

surgicaloutcome.

81

CHAPTER3TheRoleofQuantitativeMRIAnalysisinPredictingSurgical

OutcomeinPatientswithCervicalSpondyloticMyelopathy

ThecontentsofthischapterhavebeenpublishedinthejournalSpine:

NouriA,Tetreault,L,Zamorano,JJ,DalzellK,Davis,AM,MikulisD,YeeA,Fehlings,MG.Roleofmagneticresonanceimaginginpredictingsurgicaloutcomeinpatientswithcervicalspondyloticmyelopathy.Spine(PhilaPa1976).2015Feb1;40(3):171‐8.

82

3.1Introduction

CervicalSpondyloticMyelopathy(CSM) isoneof the leadingcausesofDCMand is

the most common cause of spinal cord dysfunction in the elderly worldwide

(Karpova et al., 2013b). The development of CSM is associated with age‐related

degenerationof thecervicalspineanatomythatprogresseswithtime.Myelopathy

inthesepatientsarisesfromcompressiveforcesexerteduponthespinalcordfrom

the aberrant migration of tissue into the spinal canal. Although asymptomatic

degenerationof the cervical spine is common in theelderly (TracyandBartleson,

2010), when these changes lead to myelopathy, patients are at risk of motor,

sensory and autonomic dysfunction, as well as a reduction in quality of life

(Karadimasetal.,2013b).

Although CSM patients with mild symptoms are often managed

conservatively, surgical decompression is increasingly recommended as the

treatmentstrategyforthefullspectrumofmyelopathyseverity.Asdemonstratedby

Fehlings et al (2013a), surgical decompression is effective as it not only prevents

further neurological decline but also improves neurological function. However,

surgeryofanykind,particularlyinvolvingthecervicalspine,doesnotalwaysresult

inperfectoutcomeandisassociatedwithrisks.

Outcomepredictioninthissettingisthereforevaluableassurgeonscanuse

this information todiscusswithpatientshow theyare likely to fare after surgical

intervention and tomanage their expectations. In recent work by Tetreault et al

(2013c),aclinicalpredictionrulewasconstructedtopredictfunctionalstatusat1‐

year follow‐up using a combination of clinically relevant and routinely‐collected

variables. This model, however, did not assess the role of MRI in predicting

outcome.While a number of studies have attempted to assess this relationship, a

recent systematic review (Tetreault et al., 2013a) demonstrated that the overall

bodyofevidenceislowandthattheresultsdonotprovideclearguidanceregarding

theutilityofMRIinpredictingsurgicaloutcome.

83

3.2MaterialsandMethods

3.2.1StudyDataandDesign

Thecurrentstudyutilizeddata fromtheAOSpineCSM‐NAmulticenterstudy.That

study was primarily undertaken to assess the safety and efficacy of surgical

decompression inpatientswithmild,moderateorseveremyelopathy(Karpovaet

al.,2013a).Asecondaryobjectivewastoevaluatetheassociationbetweenvarious

preoperative MRI characteristics and postoperative functional status. Research

ethicsboardapprovalwasgrantedateachstudysite.

At 12 participating sites, 278 patients with clinically confirmed CSMwere

enrolledandunderwentsurgicaldecompressionofthecervicalspine.Theapproach,

numberofoperated levels,andwhetherornottouse instrumentationvariedcase

by case and was at the discretion of the surgeon. Extensive patient data were

collected at baseline and at 6, 12, and 24 months postoperatively, including

demographicinformation,surgicalsummary,medicalhistory,functionalstatusand

patient‐reportedqualityoflife.Asperstandardofcare,allpatientsalsoreceiveda

MRI or CT scan preoperatively. One hundred and fourteen patients had MRIs

available for research purposes, 102 of which had complete data required for

quantitativeanalysisandpredictivemodelingFigure3.1.

84

Figure3.1Aconsortdiagramindicatingtheclinicalandimagingdataavailableatbaselineanduponfollow‐upat6‐months.

85

3.2.2QuantitativeMRIAnalysis

MRIswere performed using 1.5 Teslamagnets. Image analysis of DICOM (Digital

Imaging and Communications in Medicine) formats was carried out using OsiriX

(open access software, http://www.osirix‐viewer.com).Quantitativemeasurement

techniques were primarily chosen based on findings from previously published

systematic reviews on imaging characteristics that predict surgical outcome

(Tetreault et al., 2013a, Karpova et al., 2013a). The images were reviewed by 3

investigators(AriaNouri–postdoctoraltrainee,KristianDalzell–orthopedicspine

fellow,JuanZamorano–orthopedicspinefellow)toidentifythemid‐sagittalsliceon

T2‐WI,thelevelofmaximumspinalcordcompression(MSCC)aswellasmaximum

canalcompromise(MCC),andthepresence/absenceofsignalchangeonT1‐WIand

T2‐WI.Iftwomid‐sagittalslicesapproximatedthemidlinewithequaldistance,the

mid‐sagittalslicewithgreatestcompressionontheaxialplaneonT2‐WIwasused

for measurement. Disagreement among the investigators was resolved with

consensus. Aria Nouri then conducted quantitative measurements and ratio

calculations. Table 3.1 summarizes the various quantitative methods of analysis

conducted.

86

QuantitativeMRIMeasurement Formula Description Reliability

MaximumSpinalCordCompression

Fehlingsetal(1999) MSCC 1–di

da db /2X100

Measuresthemostcompressedspinalcorddiameter(di)seenonmidsagittalMRIagainstnon‐compresseddiameterreferencesfromabove(da)andbelow(db).

Theintra‐ andinterobserverICCswerepreviouslyreportedas0.76+/‐0.08and0.79+/‐0.09forT2MSCC,respectively(Karpovaetal.,2013c).

MaximumCanalCompromise

Fehlingsetal(1999) MCC 1 Di

Da Db /2X100

Measuresthegreatestreductionofspinalcanaldiameter(Di)seenonmidsagittalMRIbasedonnon‐reduceddiameterreferencesfromabove(Da)andbelow(Db).

Theintra‐ andinterobserverICCswerepreviouslyreportedas0.88+/‐0.1and0.75+/‐0.04fortheT1MCC,respectively(Karpovaetal.,2013c).

SignalChangeRatios

Arvinetal(2011) SCR Arvin X 100

ComputesaratioofforROIsagainstaCSFsignalintensityreferenceobtainedfrombehindthedens.

Reliabilityunknown.

Wangetal(2010a)&Zhangetal(2010) SCR Wang/Zhang

.

ComputesaratiobasedonafixedROIof0.05cm2againstareferencepointonthespinalcordintheregionofC7/T1.

NewRatio SCR New.

/

ComputesaratiobasedonafixedROIof0.05cm2againstanaveragereferencepointonthespinalcordapproximatingC7/T1andC2.

Table3.1Thetableoutlines,describesanddiscussesthevariousquantitativemeasuresevaluatedinthepatientpopulation.Respectivevalidationofthemethodshasalsobeendescribed.MSCC=MaximumSpinalCordCompression,MCC=MaximumCanalCompromise,SCR=SignalChangeRatio,ROI=RegionofInterest,CSF=CerebralSpinalFluid,ICC=InterclassCorrelationCoefficients.

87

3.2.3T2‐WIMCCandMSCC

MCC and MSCC were measured using previously described methods originally

designed to assess patients with cervical SCI (Fehlings et al., 1999). Canal

compromiseandcordcompressionwerecomputedusingthemid‐sagittaldiameter

ofthemostcompressedregionofthecanal(Di)andcord(di).Thesediameterswere

then measured as a percentage of the non‐compressed areas (calculated using

references above and below the compression regions: Da, Db, and da, db for the

spinalcanalandthespinalcord,respectively),Figure3.2.Thebordersofthecanal

weredefinedby the limitsofvisibleCSFbothanteriorandposterior to thespinal

cord.

88

Figure3.2.MSCCandMCCmeasurementtechniquesillustratedonT2‐WIMR.A)T2‐WIofapatientwithclinicallyconfirmedCSMwithoutmeasurements.B)DisplaysthemeasurementsrequiredforMCCandMSCCcalculation.Di,DaandDbmeasurethediameterofthespinalcanalatthesiteofcompressionandatthenormalsiteaboveandbelow,respectively;di,daanddbindicatethediameterofthespinalcordatthesiteofcompressionandatthenormalsiteaboveandbelow,respectively.

89

3.2.4T2‐WISignalMeasurementsandSignalChangeRatio

ForpatientswithasignalhyperintensityonT2‐WI,theregionofinterest(ROI)area,

sagittalextent,andaverageintensitywithintheplanarROIweremeasured.Another

fixedareaof0.05cm2wasmeasuredforaverageintensityfromwithintheROIorat

the level of greatest cord compression in patientswithout a hyperintensity signal

change.BaselinemeasurementsweretakenfromtheCSF(betweenthedensandthe

spinalcord)andattwospinalcordsites(approximatingC2andC7/T1),measuring

anestimatedareaof0.05cm2,0.15cm2and0.15cm2,respectivelyFigure3.3.

T2‐WI signal change ratios (SCR)were computed by comparing the spinal

cordROIagainstabaselineintensitymeasurementtakenfromthesameimage(CSF

or“normalcord”dependingonthemethod).SCRsweremeasuredusing3different

methods;twoofthesehavebeenpreviouslypublishedandtheotherisamodified

methoddevelopedduringour investigation(Arvinetal.,2011,Wangetal.,2010a,

Zhangetal.,2010).

90

Figure3.3QuantitativesignalchangemeasurementsillustratedonT2‐WIMRI.A)T2‐WIofapatientwithclinicallyconfirmedCSMwithoutmeasurements.B)ROIindicatesthecircumscribedT2‐WIsignalchangefromwhichtheareaandaverageROIintensitywereobtained.AnadditionalcirclewithintheROIareaapproximating0.05cm2providestheROI(0.05cm2)utilizedforsignalratiocalculationsusingWang’sandtheauthor’smethod.Sagittalextentmeasuresthemaximumheightofthesignalchangeparalleltothespinalcord.RefA,RefBandCSFRefprovidetheaveragesignalintensityreferencepointsofthespinalcordaboveandbelow,andfromtheCSFbehindthedens,respectively.

91

3.2.5PrimaryClinicalSeverityMeasure

The mJOA scale was used as the primary measure to assess preoperative and

postoperativefunctionalstatus. It isan18‐pointCSM‐specificscalethatseparately

scores upper (5) and lower (7) extremity motor function, sensation (3) and

sphinctercontrol(3),andwasdiscussedingreaterdetailinthefirstchapter(Benzel

etal.,1991).AdichotomizedmJOAat6‐monthsfollow‐upwasusedtodiscriminate

between patients with mild myelopathy postoperatively (≥16) and those with

substantial residualneurological impairmentat6months following surgery (<16)

(Tetreaultetal.,2013c).The6‐monthsfollow‐upperiodwasusedasitrepresented

thefirstpostoperativeencounterwithpatients.

3.2.6StatisticalAnalysis

The distribution of all continuous variables was visualized using histograms and

normality was assessed using the Shapiro‐Wilk test. Descriptive statistics were

computed for all MRI parameters and relevant clinical variables: continuous

variables were assessed using means, standard deviations and ranges, including

SCRs,MCC,MSCCandbaselinemJOA,andcategoricalvariablesweredescribedusing

frequencies, including T1‐WI signal hypointensity (dichotomous), T2‐WI

hyperintensity (dichotomous), sagittal extent (cm)of signalhyperintensityonT2‐

WI (Ordinal: 0, <0.2, ≥0.2 and <0.35, ≥0.35) and ROI area (cm2) of signal

hyperintensity(Ordinal:0,<0.75,≥0.75and<1.50,≥1.50).

We conceptually divided the MRI variables into three groups: (1) T1‐WI

signal analysis; (2)T2‐WI signalanalysis, including signal sizeand intensity ratio;

and(3)anatomicalmeasurements(basedonT2‐WI).Univariateanalyseswereused

toevaluatetherelationshipbetweentheseMRIparametersandbaselinemJOAwith

functionaloutcomeat6‐months.AlongwithbaselinemJOA,thelowestp‐valuefrom

each group was then used to select variables for multivariate logistic regression

modeling (Group 2 was subdivided into 2 categories and therefore provided 2

variables:one forT2‐WIsignalsizeandone forsignal intensity).Thispreliminary

multivariatemodelwasthensimplifiedbymanualbackwardeliminationbasedon

92

practicalandstatisticalconsiderations(includingAkaikeinformationcriterion,AIC;

Bayesian information criterion, BIC) with an objective to create the most

parsimonious model. Collinearity of all variables was assessed by calculating

tolerance.

The finalmodelwas compared to amodel containing only baselinemJOA

usingalikelihoodratiotestasbaselineneurologicalfunctionhasbeenconsistently

shown to be predictive of outcome (Tetreault et al., 2013c, Holly et al., 2009,

Tetreaultetal.,2013b).

93

3.3Results

Our subset of patients with partial and complete MRI data analysis (n=114) had

similardemographicsasthecompletecohortasreportedbyFehlingsetal(2013a):

Averageage(55.75±11.84vs.56.33±11.71),femalesex(37.72%vs.40.65%),and

baselinemJOAseverity(12.81±2.74vs.12.85±0.32).

Review of MR imaging for signal changes revealed that T2‐WI signal

hyperintensitywaspresentin67.6%(75/111)andT1‐WIsignalhypointensitywas

presentin26.9%(29/108)ofpatientsatbaseline.T2‐WIsignalchangewaspresent

in86.2%(25/29)ofpatientswithT1‐WIhypointensities.Onaverage,patientshada

MSCC of 33.9% (‐4.60‐64.86) and MCC of 49.2% (16.12‐75.33) preoperatively.

CompletepatientcharacteristicsarepresentedinTable3.2.

Inunivariateanalyses,aworseoutcomeat6‐monthswasassociatedwitha

lowerormoreseverebaselinemJOA(p<0.001),greaterMCC(p=0.03),agreaterT2‐

WI hyperintensity area (p=0.04) and sagittal extent (p=0.03) (Table 3.3). In

addition,ahigherSCR,asassessedbyWang’sratio(Wangetal.,2010a)andournew

methodology also predicted a worse outcome, although neither of these

relationships reached statistical significance (p=0.07, p=0.11, respectively). T1‐WI

hypointensity (p=0.26), T2‐WI hyperintensity (p=0.17), Arvin’s ratio(Arvin et al.,

2011)(p=0.26)andMSCC(p=0.27)werenotrelatedtooutcomeunivariately.

A multivariate logistic regression model was constructed using a single

variabletodescribebothT1‐WIsignalchange(+/‐ofhypointensity)andanatomical

measurements (MCC), and two variables to describe T2‐WI signal characteristics

(signal intensity,Wang’s SCR (Wang et al., 2010a); and signal hyperintensity size,

sagittalextent).Thesefourimagingvariables,alongwithbaselinemJOA,yieldedan

areaunder thereceiveroperatingcharacteristiccurve(AUC)of0.85.Reductionof

variables to create parsimony resulted in a final model including T1‐WI

hypointensity (OR=0.24;CI=0.07‐0.87),MCC(OR=0.94;CI=0.90‐0.98)andbaseline

mJOA (OR=1.74; CI=1.35‐2.24). According to this image‐based model, a worse

outcomewasbestpredictedbypresenceofT1‐WIhypointensity,agreaterMCCand

amoreseverebaselinemJOAscore.ThismodelhadanAUCof0.84(Figure3.4)and

94

areducedAICandBICcomparedtotheinitialfive‐variablemodel(AIC=94.78vs.

98.29;BIC=104.78vs.113.29).

Incomparisontotheabovemodel,theAUCforthebaselinemJOA‐onlymodel

was0.81.Thelikelihood‐ratiotestindicatedsuperiorperformanceofthefullmodel

comparedwiththebaselinemJOA‐onlymodel(p<0.0001).

95

Table3.2GeneralandMRIcharacteristicsofpatientsforwhomMRIanalysiswasconducted.Continuousvariablesarepresentedasmeansandstandarddeviations,withtherangeinparentheses.MSCC=MaximumSpinalCordCompression;MCC=MaximumCanalCompromise.

GeneralCharacteristics(n=114)x̄±SD(Range),unlessotherwiseindicated

Age(years) 55.75±11.84(29‐86)Sex(M/F) 71/43Durationofsymptoms(months) 26.55±34.24(1‐240)BaselinemJOA 12.81±2.74(5‐18)Smokingstatus(Y/N) 30/84Imaging T2‐WIHyperintensity(Y/N)(n=111) 75/36T1‐WIHypointensity(Y/N)(n=108) 29/79Arvin'sSignalChangeRatio(n=111) 0.43±0.13(0.21‐0.92)Wang'sSignalChangeRatio(n=111) 1.49±0.41(0.91‐3.1)NewSignalChangeRatio(n=111) 1.42±0.38(0.89‐2.97)MSCC(%)(n=108) 33.91±15.34(‐4.60‐64.86)

MCC(%)(n=108) 49.24±12.95(16.12‐75.33)

SagittalExtentofT2‐signalchange(n=111)

Group1=0(n=36) 0Group2>0,≤0.75cm2(n=23) 0.58±0.12

Group3>0.75,≤1.50cm2(n=24) 1.15±0.20

Group4>1.50cm2(n=28) 2.17±0.74

AreaofT2‐signalchange(n=111) Group1=0(n=36) 0

Group2>0,≤0.20cm2(n=33) 0.13±0.04

Group3>0.20,≤0.35cm2(n=22) 0.26+0.04

Group4>0.35cm2(n=20) 0.51±0.16

Outcome mJOAat6‐months(n=101) 15.16±2.73(2‐18)≥16 55(54.46%)<16 46(45.54%)

96

Table3.3UnivariateandmultivariaterelationshipofMRIparametersandbaselinemJOAwithadichotomizedmJOAoutcome(≥16,<16)at6months.OddsratiosindicatetheoddsofhavinganmJOAof≥16inlightofthecorrespondingMRIfindings.

ImagingPredictor(Univariate) OddsRatio

T2‐WIHyperintensity(n=98) 0.55(CI:0.23,1.28)

T1‐WIHypointensity(n=95) 0.59(CI:0.23,1.50)

T2signalhyperintensitysagittalextent(n=98) 0.67(CI:0.48,0.95)

T2signalhyperintensityarea(n=98) 0.67(CI:0.46,0.99)

Newsignalchangeratio(n=98) 0.42(CI:0.15,1.21)

Wang’sRatio(n=98) 0.41(CI:0.16,1.09)

Arvin’sRatio(n=98) 0.18(CI:0.01,3.57)

Maximumspinalcordcompression,MSCC(n=96) 0.98(CI:0.96,1.01)

Maximumcanalcompromise,MCC(n=96) 0.96(CI:0.93,1.00)

FinalModelPredictors(Multivariate)n=90

BaselinemJOA 1.74(CI:1.35,2.24)

T1signalhypointensity 0.24(CI:0.07,0.87)

Maximumcanalcompromise,MCC 0.94(CI:0.90,0.98)

mJOA‐onlyModeln=90

BaselinemJOA 1.64(CI:1.33,2.04)

97

Figure3.4Areceiveroperatorcharacteristiccurveofthefinalmodelbasedon3predictorsofoutcome:T1hypointensity(OR=0.242;CI=0.07‐0.87),spinalcanalcompromise(OR=0.94;CI=0.90‐0.98)andbaselinemJOA,(OR=1.743;CI=1.35‐2.24).Theareaunderthecurveofthismodel(~0.84)indicatesa“verygood”abilitytodiscriminatebetweenadichotomizedmJOApostsurgicaloutcome(≥16,<16).

98

3.4Discussion

InthetypicalCSMpatient,MRIofthecervicalspineprovidesstaticimagesofspinal

cord compression. This manifestation is the consequence of degenerative

anatomical changes that have resulted in the aberrantmigration of soft tissue as

well as bone structures (i.e. osteophytes) into the spinal canal. Insult to the cord

maybefurtherexacerbatedbydynamiccompressionresultingfromhypermobility

and cervical joint instability. Myelopathic changes are frequently observable as

hyperintense changes on T2‐WI and have been suggested to represent necrosis,

microcavities, or spongiform changeswhenwell‐defined borders are present and

edema,demyelination,andWalleriandegenerationwhensignalchangesarediffuse

(Karpova et al., 2013a). Such changes were present in approximately 67.6% of

patientsinthepresentstudy.Lessfrequently,andgenerallypresentinmoresevere

case of CSM, patients may also display T1‐WI hypointensity changes, indicating

cavitation, ischemia and frank loss of neuronal tissue (Tetreault et al., 2013a).

Approximately26.9%ofpatientspresentedwithsuchchanges.

Numerous investigators have assessed the association between various

preoperativeMRIparametersandpostsurgicaloutcomeinCSMpatients;however,

theircollectiveresultshaveprovidedanunclearpictureintermsofpredictivevalue

andpracticalapplication(Tetreaultetal.,2013a,VedantamandRajshekhar,2013,Li

etal.,2011).Despitethis,internationalspinecareprofessionalsagreethatMRIisan

importantprognostic tool (Tetreault et al., 2014).Evidence to support thisnotion

hasrecentlybeenpublishedinasystematicreviewandindicatesthatfactors,such

as combined T1‐/T2‐signal change, multilevel signal change and SCRs, are

significantpredictorsofsurgicaloutcome(Tetreaultetal.,2013a).Havingsaidthis,

there are several limitations in methodology that impact the interpretation of

currentliterature.Issuesofparticularnoteincludevariabilityinapproachesusedto

analyze and quantify structural changes as well as a lack of accepted standard

protocolswith respect toMRI assessmentmethodology. Taking these factors into

consideration, in the present study, MRI analysis was conductedwith a focus on

99

quantitativeapproachesandwithagoalofestablishingareproducibleandeffective

protocolforMRIassessment.

Though most of the quantitative MRI parameters assessed in the present

studydisplayedatendencytowardssignificance(p<0.20),onlythreeofthese(MCC

and T2‐WI hyperintensity sagittal extent and area) were significant (p<0.05)

predictors of surgical outcome at 6 months following univariate analysis.

Interestingly, both measurements of T2‐WI signal hyperintensity size (sagittal

extentandarea)weresignificantlyassociatedwithsurgicaloutcome,indicatingthat

the larger the signal change the lower the odds of achieving an mJOA ≥16. The

finding that SCRs were not predictive of outcome indicates that perhaps such

analysis is dependent on specific MRI protocols, may be dependent on technical

factors,andmayvaryduetotruncation,chemicalshift,cerebrospinalfluidflow,or

motionartifacts(Karpovaetal.,2013c).Therefore,itcannotbeconcludedthatthese

methodshavenopracticalutility.Thefindingthattheaveragepatienthadamuch

greaterMCC(49.2%)thanMSCC(33.9%)isunderstandablegiventhatexcessspinal

canal space occupied by cerebral spinal fluid can allow for some migration of

structures into the canal without cord compression. Therefore, patients with a

narrowerspinalcanal,asseeninpatientswithcongenitalspinalstenosis,arelikely

tobeatapredispositionformyelopathydevelopment(Singhetal.,2012).

ItisnoteworthythatthemethodofanalysisforMCCwasamodifiedmethod

thatmayservetomoreaccuratelyassessthelevelofcompromiseinpatientswith

degenerativeconditions.TheoriginalmethodwascreatedfortraumaticSCIpatients

and bases the degree of MCC on the diameters from the posteriormargin of the

vertebralbodyanteriorlyandtheanteriorportionofthelaminaposteriorlyasseen

onCTorT1‐WIMRI(Fehlingsetal.,1999,Fehlingsetal.,2006).Compressionofthe

spinalcordindegenerativeconditionsmayhoweverariseasaresultofsofttissue

structures that are more clearly visible on T2‐WI, and therefore, assessment of

diameter sizebetweenbonestructuresmaypresenta falsepictureof canal size if

osteogenicrestructuringisnotinvolved.Itisthereforerecommendedthatpatients

withdegenerativecervicalmyelopathybeassessedwiththemodifiedmeasurement

asdescribedinthepresentstudy.

100

Whileindividualparametersareusefulinpredictingoutcome,amultivariate

approachisrequiredtoassessCSMpatientsduetotheheterogeneityofpathological

manifestations. Additionally, a combinatorial assessment of MRI parameters may

yield a greater predictive performance than individual parameters. Following this

view, a multivariate model was constructed and included the most predictive

parameters from threeMRI assessment groupsaswell asbaselinemJOAseverity.

ThegroupingofMRIparametersintothesecategoriesisbasedontheconceptthat:

(1)measurements therein are usually conducted separately, (2) parameters from

eachgrouprepresentdifferentpathologicalfindings,and(3)eachgroupisgenerally

investigated separately in literature. The initial model had an AUC of 0.85 and

included T1‐WI signal change,MCC,Wang’s SCR (Wang et al., 2010a) and T2‐WI

signalhyperintensitysagittalextent.However,moreparsimoniousmodelsarelikely

to reduce complexity and improve information gain and therefore themodelwas

simplified.Accordingly,assessmentofcomparableefficacy fromthisnestedmodel

indicated that T1‐WI signal analysis, MCC, and baseline mJOA severity provided

greater parsimony while maintaining a similar predictive capacity (AUC=0.84) to

the initial five‐variablemodel. The inclusionof thebaselinemJOA in themodel is

necessary to avoid regression to themean, also, neurological function is a strong

predictor of postsurgical outcome (Holly et al., 2009, Tetreault et al., 2013b), and

thereforeitnecessitatesacomparisonofthefinalMRI‐basedmodelwithonebased

solelyonmJOAbaselineseverity.Asmeasuredbythelikelihood‐ratiotestthefinal

modelprovidedagreaterpredictivecapacitythanthebaseline‐onlymodel.Thus,on

the basis of these findings, quantitative MRI analysis has a significant role in

predictingsurgicaloutcomeinpatientsundergoingdecompressivesurgeryforCSM.

Although external validation is necessary to confirm these results, recently

published systematic reviews have associated T1‐WI hypointensity with

postsurgical outcome (Tetreault et al., 2013a), and indicated that patients with

cervicalspinestenosisarepredisposedformyelopathydevelopment(Wilsonetal.,

2013a). These findings conceptually support the inclusion of both T1‐WI signal

changeandMCCinthefinalmodelandtheirindependentefficacyinimprovingthe

predictivecapacity.

101

Going forward, theapplicationofnewer techniquesandtechnologies in the

assessment of surgical outcome in CSM patients may allow for improvement of

predictionmodels.Methods described by Zhang et al (2011) for examplemay be

interestingtoevaluateinthefuture.Furthermore,newerMRItechnologiessuchas

diffusion tensor (Jones et al., 2013,Nakamuraet al., 2012) aswell asupright and

dynamic/kinematicMRI (Jinkins et al., 2005)mayuncover additional information

fromwhichprognosticinsightsmaybederived.

3.5Limitations

Thereareafewlimitationsofthisstudy.Thoughthemulticenterbasisofthisstudy

offersanopportunitytodeterminewhichMRIassessmentmethodsarevalidacross

differentcenters,SCRsare likelydependenton the typeofMRIprotocolusedand

therefore our insignificant findings for some these measurements may reflect

technical variability. As well, patients received different types of surgical

procedures, as is consistent with current practice, and this has prevented

standardizedtreatmentofpatients.Additionally,MRIdatawasonlyavailablefora

subsetoftheentirestudypopulation.Furthermore,MRIassessmentwasconducted

forasampleofpatientswhohadclinicallyconfirmedCSMandunderwentsurgical

treatment.Lastly, thecharacterizationofpatientswithmildmyelopathyandthose

with substantial neurological impairment using our dichotomized mJOA may

marginalizepatientsborderingthisdemarcation.Thesepatientsmayhavehadtheir

mJOAscoreimpactedbyadvancingageoralternativelymaybeconsideredtohave

moderateimpairment.

102

CHAPTER4GeneralDiscussion,ClinicalImplications,andKnowledge

Translation

103

4.1KeyFindings

Todate,therehavebeennoprospectivemulticenterstudiestoinvestigatetherole

ofMRI in predicting outcome in patients undergoing surgical treatment forDCM.

The research undertaken in the previous chapter was conducted to address this

knowledgegapandhasdemonstratedthefollowing:

(1) Onunivariateanalysis,MCC,T2‐WIhyperintensity(intermsofsagittalextent

as well as area), and baseline mJOA severity are statistically significantly

relatedwithmJOAoutcomeat6‐monthspostoperatively.

(2) AmultivariateregressionmodelofMRIfeaturesincorporatingassessmentof

T1‐WIsignalhypointensity,MCC,andbaselinemJOAseverityyieldedanAUC

of0.84andoffersaverygoodabilitytopredictwhetherpatientswillhavean

mJOAscoreof≥16or<16at6‐monthaftersurgery.

(3) Themultivariatemodelhadahigherpredictivecapacitythanamodelsolely

basedonbaselinemJOAseverity,indicatingthatwhilebaselineseverityisa

strong predictor of outcome, MRI assessment has a significant role in

predictingsurgicaloutcome.

4.2BasisforMRIAnalysis

Ithasbeenwellrecognizedthatwearoccursinthespineofallindividualsovertime.

This natural progression can be accelerated in some individuals, but is generally

considered a part of normal physiological and adaptive changes in response to

chronicusewithage(Galbuseraetal.,2014).Giventhis,imagingofelderlypatients

willgenerallydepictsomesignsofdegeneration. Indeed, inarecentreview,Tracy

andBartleson(2010)indicatedthatover50%oftheasymptomaticpopulationaged

40 and older may present with MRI evidence of disc degeneration or space

narrowing at 1 or more levels, and another 40% may present with bone spurs.

These changes can result in symptoms that are similar to those appreciatedwith

osteoarthriticchangesinotherbodysitessuchasthekneeorhand,andmayinclude

104

pain and restrictionofmovement.Themultiple articular layers that comprise the

spinalcolumn,however,donotsimplyfunctiontoconfermovementofthetorsoand

neck, but also harbor the passage of the nervous system from the brain to the

peripheryviathespinalcord.Asaconsequence,degenerativechangesofthespine

mayresultinthefailureof2functionsofthespine:Biomechanical,andprotectionof

thespinalcord.Compressionofneuralrootsor individualspinalnervescancause

substantial clinical symptoms, but compression of the spinal cord and resulting

myelopathy at the cervical level damages the neural pathway at an early course

awayfromthebrain,andcanresultindebilitationsevereenoughtopreventsocial

independence. These symptoms can include motor and sensory deficits of both

upper and lower limbs and also impact involuntarymotor function necessary for

urinary and bowel control. However, symptoms that present in patients with

degenerativecervicalmyelopathycanalsooccurinanumberofotherneurological

conditions and it is therefore necessary that a suspicion of the diagnosis be

confirmedviamedicalimaging,withthegoldstandardbeingMRI.

4.3InterpretationandMeaningofMRIFindingsofDegenerative

Changes

Given that damage to the cord is preceded by changes in the surrounding spine

structure,itwouldbereasonabletoassumethatimagingevidencethatdepictsthe

cord under compression would justify intervention to prevent myelopathic

symptoms fromprogressing.Unfortunatelyhowever, given current literature, it is

not so clear that this is the case. Despite high prevalence rates of cervical spine

stenosis on postmortem examination due to spondylosis or OPLL, clinical

experience indicates that only a small percentage of these patients seek medical

careforsymptomaticmyelopathyandrequiresurgicalintervention(Leeetal.,2007,

Wilsonetal.,2013a).InarecentsystematicreviewbyWilsonetal(2013a),itwas

noted that22.6%ofnonmyelopathicpatientswithMRIsignsof cordcompression

duetospondylosiswilldevelopclinicalevidenceofmyelopathyatamedianof44‐

months. While some predictive factors where identified and may be used to

105

anticipate the development of myelopathy in these patients, the review also

recognizedthatthemajorityofpatientsremainstableandrequireonlyconservative

management.

Ambiguitywithregardstothemeaningofremarkabledegenerativechanges

on MRI is, however, not only limited to patients with asymptomatic spinal cord

compression.Aswas evident in our study, the degreeof spinal cord compression

variedwidely,withanumberofpatientswithoutfrankMRIevidenceofspinalcord

compressiondespiteaclinicaldiagnosisofCSM.Thereareanumberofreasonsfor

why this may be the case. Conventional MRI, though the most useful tool for

evaluating patients with suspected DCM, only provides static imaging. However,

hypermobile or dynamic factors may be additional mechanisms responsible for

compressingthespinalcord.Itisalsochallengingtodeterminetheextentofinsult

placedupon the spinal cordusing conventionalMRI due to thepatient’s position.

When lying supine, as is routinely the case, the patient is not bearing the same

weightuponthespineastheywouldduringanuprightorientation.Thismeansthat

during a typical MRI exam, there is decreased downward force placed upon the

intervertebraldiscsaswellasbetweenfacet joints.Accordingly, thismayresult in

less soft tissue migration into the spinal canal or less listhesis, and ultimately,

portrayaninaccuraterepresentationofcanalpatency.

In contrast to the unclear relationship between image evidence of spinal

cord compression and the state of myelopathy, when signal intensity changes

present on T1‐WI or T2‐WI it is clear that an intrinsic pathological process has

occurredinthespinalcord.HyperintensitychangesonT2‐WImayrepresentawide

rangeof pathologiesdependingon the characteristics andhavebeenproposed to

indicate (1) microcavities, spongiform changes when well‐defined borders are

present, and (2) edema, demyelination, and Wallerian degeneration when signal

changesarediffuse(Karpovaetal.,2013a,Mehalicetal.,1990,Chenetal.,2001).

Hypointensity changes on T1‐WI appear less frequently, are more difficult to

evaluate, and arebelieved to represent frank loss of neuronal tissueor cavitation

(Tetreaultetal.,2013a,Al‐Meftyetal.,1988).Whilemostpatientswillpresentwith

signal changes, as is supported with our findings, many do not; therefore, the

106

absence of signal intensity changes does not preclude a diagnosis of DCM. There

maybeafewreasonsforwhythismaybethecase.Compellingargumentscouldbe

made that patients without signal changemay represent those at early stages of

myelopathy,orperhapsrepresentthosewhoexperiencesymptomsduetodynamic

forces rather than persistent or static compression. These, however, remain

speculations.

Inarecentsurveyofspinecareprofessionalstherewasconsensusthatboth

clinicalandimagingfactorsarepredictiveofoutcomeinpatientssurgicallytreated

forCSM.Intermsoftheimagingpredictors,therewasinternationalconsensusthat

MRI isan importantprognostic toolandthat thepresenceofsignalchangeonthe

spinal cord on T2‐WI and combined T1/T2‐WI are the most important of these

(Tetreaultetal.,2014).

4.4ApproachandRationalefortheUseofMRIMeasurement

Techniques

Understandingthatdegenerationofthespinecanpresentwithgreatheterogeneity

setsthebasisforcreatingparametersthatassessthewidespectrumofchangesthat

mayappearonMRIofpatientswithDCM.Giventheliteratureandmethodsatour

disposal, MR imaging of these patients was investigated in 3 capacities: (1)

anatomicalchanges includingspinalcanalsizeaswellasspinalcordcompression,

(2)T1‐WIsignalintensitychange,and(3)T2‐WIsignalintensitychange,including

thequantitativeanalysisofitssizeandrelativeintensity.

Anatomicalchangesintermsofreductioninthecanalsizeandcompression

of the spinal cord have been effectively measured previously using methods

describedbyFehlingsetal(1999)fortheassessmentofpatientswithtraumaticSCI.

The original method assessed MCC using CT and MSCC using MRI. Subsequent

investigation by Fehlings et al (2006) assessed the intrarater and interrater

reliability of these methods and additionally assessed MCC using T1‐WI. Their

findings indicated that the interclass correlation coefficients (ICC) in terms of

intraraterreliabilitywerequitehigh,rangingbetween0.90‐0.97forMCCandMSCC.

107

Contrarily, the interrater reliability demonstrated disparity between investigator

measurements, ranging between 0.35‐0.56. More recent work by Karpova et al

(2013c)assessedthereliabilityofMSCCandMCCspecificallyinpatientswithDCM.

Theirfindingsshowedintra‐andinterobserverICCsof0.76±0.08and0.79±0.09

forT2‐WIMSCC, respectively, and intra‐ and interobserver ICCsof0.88±0.1and

0.75±0.04 forT1‐WIMCC, respectively.Overall, these findings indicate thatMCC

andMSCCmethodshaveacceptablemeasurementreliability.

In recognizing that these methods were originally devised for the

investigationofpatientswithtraumaticSCI,somemodificationtothemethodology

isnecessary, specifically for themeasurementofMCC.The canaldiametersonCT

and T1‐WI were originally measured from the posterior margin of the vertebral

body to the anterior portion of the lamina. Compression of the spinal cord in

degenerativeconditionsmayhoweverariseasaresultofsofttissuestructuresthat

aremoreclearlydemarcatedonT2‐WI,andtherefore,assessmentofdiametersize

between bone structures may present a false picture of canal size if soft tissue

structuresarechieflyinvolved.InthepublicationbyArvinetal(2011),T2‐WIwere

usedfortheanalysisofMCCinpatientswithCSM,anditappearsthatmeasurement

of canal sizewasdone in thismodified fashion, although thiswasnot specifically

indicated. Aswas described in chapter 3, our analysis ofMCCwas based on this

modification andwas conducted on T2‐WI. This is especially noteworthy asMCC

was one of the two MRI parameters included in our final prediction model (the

otherbeingT1‐WIhypointensity).Inourmodel,anincreasedMCCwasrelatedwith

aworsesurgicaloutcome,andspecifically,theoddsforagoodoutcome(mJOA≥16)

decreased6%forevery1%increaseinMCC.

In our analysis, the MSCC methods, contrary to MCC, were unaltered and

conducted on T2‐WI. In terms of its prognostic value, MSCC did not relate to

outcome univariately, andwas not included in our finalmodel. Indeed,while not

directlyaddressingourmethodofmeasuringMSCC,acurrentsystematicreviewby

Karpova et al (2013a) determined that the relationship between spinal cord

compression using variousmethods and surgical outcome remains unclear. Given

that compression of the spinal cord is a fundamental pathological component

108

responsible for the development of myelopathy, this finding is somewhat

perplexing,andperhapsservestoillustratethatthevariousmethodsusedtoassess

spinal cord compression do not reliability represent the pathological state of

patients.Karpovaetal (2013a)offeranumberofcompellingreasons forwhythis

maybethecaseincludingthatDCMmayappearasamultileveldiseasewithvarying

degreesofcordcompressionthroughthecourseofthecervicalspine.Theysuggest

thattechniquesassessingaccretiveeffectsofcompressionshouldbeconsideredfor

future investigationinsteadofassessmentatasingle level(Karpovaetal.,2013a).

Perhapsitshouldalsobeconsideredthatcompressionarisingfromdifferenttissue

might exert different degrees of force upon the spine. While this remains

speculative, compressive forces arising from soft tissuemay demonstrate greater

compliancethanforcesarisingfromsolidstructuressuchasbone.

AnothersetofMRIassessmenttechniquesmeasuresignalintensitychanges

present in thespinalcordonT1‐WIandT2‐WI. Inour investigation,T1‐WIsignal

hypointensitywas the second of twoMRI parameters (the other beingMCC) that

wasincludedinthefinalpredictionmodel.PresenceofT1‐WIsignalhypointensity

indicated that patients are likely to have a worse outcome than those without a

signal change. This prognostic effect was independent of T2‐WI signal

hyperintensity,sinceitwasnotrepresentedbyaparameterinthefinalmodel.The

recognitionthatT1‐WIindicatesafranklossofneuraltissue(Tetreaultetal.,2013a,

Al‐Mefty et al., 1988), suggests that these patients are likely to experience less

improvement in theirneurologicaldysfunction.Therefore, it isnot surprising that

the presence of T1‐WI signal hypointensity in our model was related with an

outcome that suggests substantial residual neurological impairment (mJOA <16)

aftersurgery.TherelativelylowprevalenceofT1‐WIsignalchangeinpatientswith

DCM(26.9%inourstudy)makesthisparameterparticularlyclinicallyrelevantfor

thoseinwhomitpresents.

T2‐WI signal hyperintensity on the spinal cord has been much more

thoroughlyinvestigatedbyresearchersthanT1‐WIsignalhypointensity.Ingeneral,

T2‐WIsignalhyperintensitycanbebrokendown in twocategoriesofparameters,

thosemeasuringsize,andthosemeasuringthedegreeorqualityofsignalintensity.

109

Sizemeasurementshavebeentypicallyconductedintermsofassessingtheareaand

thenumberofcervicallevelsinvolvedonsagittalview.Giventhatagreatersizeof

signalchangeindicatesmoreanatomicalinvolvement,itwouldseemreasonableto

assumethata largersizewouldindicatemoreextensivepathology. Indeed,during

our investigation, univariate analysis showed that both area and sagittal extent

(measuredincm)couldbeusedtopredictsurgicaloutcomeat6‐months;however,

neitherofthesecontributedtoadditionalpredictivecapacityofourfinalprediction

modelandthereforewerenotincluded.Recentreviewsthathavediscussedfindings

basedonthesize(intermsofareaandsagittalextent)ofT2‐WIsignalchangeshave

provided unclear conclusions regarding the utility of these parameters and their

abilitytopredictoutcome(Tetreaultetal.,2013a,Karpovaetal.,2013a,Vedantam

andRajshekhar,2013).Thestudiesincludedinthesereviewshavemostlyfocused

on evaluating whether patients have focal ormultisegmental (extendingmultiple

vertebral levels) T2‐WI signal change. However, in the absence of a quantitative

assessment of the sagittal extent, there remains ambiguity in the delineation

betweenfocalandmultisegmentalinvolvement.Inourinvestigation,sagittalextent

wasmeasuredquantitativelyincmratherthanclassifyingsignalchangesasfocalor

multisegmental. Our approach is recommended for future investigation as it

providesamoreobjectivemeasurementofextentandretainstheabilityforspecific

size categorization thereafter. Measurement of signal intensity size in terms of

sagittalareahasnotbeenroutinelyreported.Theremaybemanyreasons forthis

anditcanbespeculatedthatthismaybetheresultofdifficultyindemarcatingareas

of signal change that appear fuzzy or faint. Furthermore, it may be difficult to

establish which MRI sagittal slice to use as the signal hyperintensity may span

acrossmultipleslices.Therefore,ambiguityanddecreasedreliabilitymayarise.To

overcome such problems in our study, we devised a specific approach that was

discussed in section3.2.2, andused three investigatorswho reviewedandagreed

uponwhich slicewasmost appropriate beforemeasurementswere performed. It

wouldberecommendedthatfutureanalysisbyinvestigatorswouldlikewiseadopta

consistentapproachforsuchmeasurements,asitislikelytoincreasethereliability.

110

Though interesting work has come from the investigation of T2‐WI signal

hyperintensity size (sagittal area/extent), a large focus has also been placed on

assessing the degree of deviation of hyperintensity signal away from what is

considered to be normal, as well as the qualitative nature of the signal intensity

change. In a recent review by Vedantam and Rajshekhar (2013), a number of

qualitative categorizationmethods derived from other authorswere summarized.

While the categorization methods differed and have not been validated, there

appearedtobeageneralconsensusthatT2‐WIhyperintensitiescaneitherpresent

as highly intense, focal and well circumscribed, or less hyperintense, diffuse and

without a clear margin. In recognizing this distinction, many have proposed that

these changes not only represent different pathological states, but also indicate

different recovery potentials after spinal cord decompression (Tetreault et al.,

2013a). Asmentioned earlier in the thesis, less hyperintense and diffuse changes

areconsideredtorepresentmilderchangesandpotentiallypresentanopportunity

for regeneration. Conversely, focal and hyperintense changes are believed to

represent cavitation and necrosis (Tetreault et al., 2013a, Al‐Mefty et al., 1988).

TheseassociationsaresupportedbythefindingsthatveryhyperintenseT2‐WIfocal

changesappeartobeisointensewithCSFandthattheyarefrequentlyaccompanied

by hypointense T1‐WI changes,which are considered to represent cavitation and

lossofneuronal tissue.Definingand categorizing throughvisual inspectionwhich

T2‐WI signal changes are less or more hyperintense, and those that are focal or

diffuse, is however a subjective endeavor. As a consequence, the potential

subjectivity and differences in classification may serve as an explanation that

despitethemanyindicationsthatsignalchangescanberelatedtooutcomebysingle

center studies, a consensuson theirutilityhasnotbeendefinitively substantiated

upon replication. More recent research, including that undertaken by us, has

attemptedtoaddresstheseconcernsbyassessingT2‐WIsignalchangeobjectively

throughquantitativemethodsthatmeasurethedegreeofT2‐WIintensitydeviation

(Arvinetal.,2011,Wangetal.,2010a,Zhangetal.,2010,Zhangetal.,2011).Though

slightlydifferentmethodswereused,thepremisebehindmostofthesestudieswas

toassesstheT2‐WIsignalhyperintensityattheregionof interestagainstbaseline

111

values from various non‐pathological reference sites (spinal cord or CSF). In

addition to these methods, in our study we also recorded a parameter that

represented a simple absence orpresence of signal change.Despite these various

methods, none of the measures indicated a statistically significant univariate (or

multivariate)relationshipwithsurgicaloutcome,althoughWang’sandourmodified

SCRmethod approached statistical significance. There aremany reasons for why

this may be the case: (1) Signal changes may demonstrate different functional

impairmentbasedon their anatomical site; (2) signal intensitymeasurementsare

basedonasinglesliceandthereforedonottakeintoaccountvolume;(3)multiple

discontinuous signal intensity changes may present; and, (4) signal intensity

changesareunlikelytorepresentalinearrelationshipwithdiseaseseverity.These

factors make it challenging to relate signal intensity changes with a pathological

stateand,therefore,maybelimitedintheirabilitytopredictsurgicaloutcomeusing

currentapproaches.

4.5CreatingaMultivariatePredictionModelforPredicting

SurgicalOutcome

In order to evaluate the role of any of theMRI parameters in predicting surgical

outcome, it is necessary to objectively assess the functional change that has

occurred between the preoperative state, and postoperatively, when the effect of

surgerycanbemeasured. Indeed, therearea fewassessmenttools thatareatthe

disposaltocliniciansforevaluatingfunctionalstatusofDCMpatientsandanumber

of theseare frequentlyused inconcert in clinicalpractice.Of these, themJOAhas

emergedasawidelyacceptedstandardforassessingthefunctionalstatus(Tetreault

etal.,2013c).Theclinicalexaminationassessesmotor,sensoryaswellasautonomic

functionoutofamaximumscoreof18.Althoughpatientscantechnicallyrange in

scoresfrom18to0,substantialdebilitationarisesafterthelossofonlyafewpoints

onthescale.IntheirclinicalpredictionmodelofsurgicaloutcomeinCSMpatients,

Tetreaultetal(2013c)usedadichotomizedoutcometoindicateifpatientshadan

optimalor suboptimal surgical outcomeusinganmJOAscoredichotomizedat16,

112

and this method was likewise adopted in this study. The interpretation of this

outcomewasthatamJOAscoreof≥16representedpatientswithmildimpairment,

whereas a score of <16 represented those with substantial residual functional

impairment.However,sinceasinglepointcanswayapatientbetweengroups,there

remainsalimitationtosuchcategorization.

The construction of our predictionmodel was based on incorporating the

bestMRIparameters,whichwe considered to be the ones that had the lowest p‐

value on univariate analysis. The categorization of MRI parameters into three

groups of T1‐WI signal analysis, T2‐WI signal analysis, including signal size and

intensityratio,andanatomicalmeasurements(basedonT2‐WI)yieldedfourinitial

MRI parameters: T1‐WI hypointensity, Wang’s SCR (Wang et al., 2010a), sagittal

extent of T2‐WI andMCC. The initial model with the inclusion of baselinemJOA

provided an AUC of 0.85. While the AUC represents a very good ability to

discriminate between our dichotomized outcome, it is necessary to evaluate

whether themodel can be simplifiedwhile offering a similar predictive capacity.

Thisisbecausesimplermodelsareeasiertouseandarelesspronetomeasurement

error.WethereforeusedtheAkaikeandBayesianinformationcriterionstoassess

thetradeoffbetweenpredictivecapacityandcomplexityofvariables inthemodel.

Indoingso,wewereabletocreateamoreparsimoniousmodelthatwascomprised

of3 variables (MCC,T1‐WIhypointensity,mJOA0) thathad comparablepredictive

discriminationintermsofAUC(AUC=0.84vs.0.85)whilereducingboththeAICand

BIC(AIC=94.78vs.98.29;BIC=104.78vs.113.29).

In using the mJOA as the principal outcomemeasure, it was necessary to

includethepreoperativeseverityofpatientsintothepredictionmodel.Indoingso,

however, it also necessitated that our final model be compared to one that

consideredthepredictivecapacityofonlybaselinemJOA,sothatroleofMRIcould

be clearly demonstrated. Using the likelihood‐ratio test, we were able to

demonstrate that therewasasignificantstatisticaldifferencebetween themodels

and that indeeda greaterpredictive capacity existed for themodel includingMRI

features. This finding demonstrates that the MRI parameters had a distinct

prognosticrole.

113

4.6ClinicalImplications

The purpose of this researchwas to provide information thatwould help predict

howpatientscanexpecttofareaftersurgicaltreatment.Nowthatithasbeenshown

thatMRI can indeedprovide such information,what are the clinical implications?

Firstandforemost,theresearchhasprovidedclearsupportthatallpatients,inthe

absenceofcontraindications,shouldundergoanMRIexaminationof theircervical

spine.Secondly,atminimum,theMRIshouldbequantitativelyassessedtoprovide

theparametersthatwerefoundtobepredictiveofoutcomeinthemodelpresented

inchapter3.

When conducting theMRI assessment, the surgeon canobtain anobjective

estimationonhow thepatient canexpect to fare (basedon themJOAscore)after

surgical treatment. A suboptimal outcome (<16 mJOA) is best predicted by the

presenceofT1‐WIhypointensity,agreaterMCCandamoreseverebaselinemJOA

score.Patientswithfindingsindicatingthatasuboptimaloutcomeislikely,maybe

wearyof surgery, but shouldbe informed that they can still benefit from surgical

treatmentas itcansloworarrestthediseaseprocess.Ontheotherhand,patients

whoare likely toachieveanoptimaloutcomebasedontheirMRI findingsmaybe

moreinclinedtoundergosurgicaltreatment.Giventheseresults,surgeonsmayfeel

more confident in providing recommendations for these patients. However, it is

important to emphasize that the baseline severity of patients is an essential

componentofthemodelpresentedandstressesthatMRIfindingsneedtobeclosely

relatedwiththeclinicalcontext.

Indeed, given the strong predictive contribution of baseline severity to the

prediction model, it is worthy to also question the clinical relevance of the

additionalpredictiveperformancegarnered fromthe inclusionofMRI features. In

termsofprobability,theimprovementbetweenthebaselineseveritymodelandthe

model including MRI features (without rounding the AUC) was 0.807 to 0.845,

representinganimprovementinpredictiveprobabilitycloseto0.04(0.038).Viewed

anotherway, this suggests thatapproximately20%[(0.038/0.194)*100]or1 in5

patients who would have received incorrect guidance with themJOA‐onlymodel

114

could have received correct guidance using theMRI‐basedmodel. In terms of its

clinical significance, ifwewere to conservatively estimate that 20,000 individuals

received surgical treatment for DCM per year in North America and all were

evaluated by bothmodels, themJOA‐onlymodelwould correctly estimate 16,140

outcomes(20,000x0.807),whiletheMRIbasedmodelwouldcorrectlyestimatethe

outcomefor16,900(20,000x0.845).Thismeansthat760individualsperyearwho

wouldhavereceivedincorrectguidancebasedonthemJOA‐onlymodelcouldhave

receivedcorrectguidanceusingtheMRI‐basedmodel.

Ofcourse,itisclearthatourpredictionmodelcannotsubstitutegoodclinical

judgment,butthesetoolsdoprovideobjectiveestimatesthatcanbeusedtoinform

thesurgeonandhelpguidepatientexpectations.It isalsoevidentfrompreviously

reportedliteraturethatthegeneralpopulationwillpresentwithabnormalfindings

onMRIwithouthavingsymptomsofDCM,and therefore, it remainsessential that

MRIexaminationsbeundertakenafteranextensiveclinicalexamination.

Sincemostpatientswill receiveaMRIexamination fordiagnosticpurposes

andhaveitreadbyaradiologist, itwouldtakeperhapsanadditional5minutesto

performthequantitativemeasurementsoutlined.Therefore, theadoptionof these

findingsispracticalandrepresentsafeasiblemeansofimprovingpatientcare.

4.7KnowledgeTranslation:BringingResearchIntoPractice

Sincethisresearchcanpotentiallyimpactclinicalcarebyprovidinginformationthat

improves surgical decision‐making and management of patient expectations, a

naturalprogressionistotranslatethefindingsofthisresearchintoclinicalpractice.

Accordingly,inthefollowingsectionsfurtherinformationonidentifyingknowledge

users, methods for dissemination of the research, and optimizing knowledge

translationarediscussedingreaterdetail.

4.7.1IdentifyingKnowledgeUsers

Generallyspeaking,knowledgeusersofresearchcanencompassmanagersofhealth

careresources,policydecision‐makersandcliniciansamongstothers(CIHR,2014).

Inordertoeffectivelytranslatefindingsfromresearchtothebedsideitisessential

115

toidentifythecorrectindividualswhoarethegatekeepersfortheimplementation

andclinicalapplicationoftheknowledge.Theresearchthatwehavebroughtforth

is relevantmost specifically to individualswhoreadMRIsandare involved in the

medical care of DCM patients. Broadly speaking this group includes radiologist,

neurosurgeons, orthopedic surgeons, neurologists and rehabilitation specialists.

However,sincetheresearchalsoprovides insight intotheprognosisaftersurgical

treatment, healthcare funders may also have a keen interest in learning of the

impact of the research as it relates to determining the likelihood for treatment

efficacy.

4.7.2DisseminationofKnowledge

Disseminating the key findings brought forth in the thesis can be done through

multiple avenues. Conventionally, early on, research is presented at national and

internationalconferencesthatarefocusedonthemainsubjectareaofthework.For

ourresearch,suchconferenceswouldincludethosesurroundingtheareaofspine,

radiologyandneuroscience.Presentingtheresearchprovidesnotonlyameans to

disseminate the information to the most critical knowledge users (including

neurosurgeons, radiologists and orthopedic surgeons), but potentially also to

policymakersandmangersofhealthcareresources.

Commonly, research should then be peer‐reviewed and published to

communicatethepurpose,whathasbeenlearned,andhowitcanimpactstatusquo

medical care. Publication of work in a journal and listing on databases such as

PubMedprovidesaccesstotheresearchandallowsforitsdetailedreviewbythose

who are interested. Once the work is readily available, efforts to encourage

readership and application of the findings should follow. Thismay be achievable

throughtheengagementofmediaandthroughpromotionbyauthoritiesinthefield;

however, the efficacy of such promotionwould largely depend on interest in the

subject.Tostimulatethisinterest,disseminationthroughsocialmedia(e.g.twitter)

withlinkstotheresearchispossiblebutdependsonapreviouslydevelopedsocial

mediapresenceor throughsponsorship,withthe latter likelyrequiringsignificant

funds.Inadditiontothis,readershipmayalsobeimprovedthroughthepublication

116

of research as open‐access, which remains an option inmost journals even after

publication.

Lesscommonly,butalsopossible,isthecreationoflearningmodules.Forthe

research presented within this thesis, detailed teaching of the methods for the

quantitative analysis of MRI in DCM patients to knowledge users, namely

radiologists, neurosurgeons, neurologists and orthopedic surgeons,may be highly

effective in translating the research from academia into practice. In addition to

disseminating the key learning points, such sessions can also provide valuable

feedback from the knowledge user, which can be used to improve knowledge

translationandidentifyongoingknowledgegaps.

4.7.3OptimizingKnowledgeTranslation

Totranslateknowledgeintopracticeitisessentialtoconveytotheknowledgeusers

the potential for the research for improving patient care. To achieve this, it is

necessarytodescribethefindingsoftheresearchandhowitwillimpacttheclinical

decision‐making process. Our findings that T1‐WI hypointensity and MCC are

important in determining outcome, for example, should bewell communicated so

that reasons for the uptake of these assessment techniques into practice can be

appreciated. Unfortunately, it is however widely recognized that translating

evidence‐basedmedicineintopracticeisoftenmetwithgreatchallenges(Bradleyet

al., 2004).Therefore, to facilitate andoptimize the uptake of our research, itmay

servewell to start by introducing the additionalMRI analysismethods through a

pilot program. This may serve to uncover practical issues, deficiencies in

communicatingtheknowledgetransfer,andpotentialadministrativeroadblocksfor

theapplicationoftheresearchintheclinicalpractice.Thepilotprogrammayserve

as a valuable way to identify problems before efforts for wider knowledge

disseminationareundertaken.

117

4.8Limitations

Aspreviouslyoutlinedattheendofchapter3,thereareanumberoflimitationsthat

have to be taken into consideration. MR images may not be obtained routinely

worldwide and there are substantial costs associated with acquiring them.

Additionally, image acquisition and quality may differ between centers and may

result in technical challenges and hinder generalizability. Conventional MRIs are

alsolimitedintheirabilitytodetectdynamicpathoetiologiesanddonotaccountfor

degenerative changes in the spine that may occur when patients are standing

uprightandbearingtheweightoftheirhead.

Interpretationofthefindingsmustalsotakeintoconsiderationthatpatients

with DCM present with heterogeneous anatomical changes of the spine and

therefore multiple surgical treatment strategies were used. Furthermore, the

dichotomization of the mJOA outcome into two groups maymarginalize patients

that border the demarcation. Additional, noteworthy limitations include the

following:

1) ModelComparison ‐ The finalMRImodelwas only compared to amodel

containingbaselinemJOA,andalthoughitdemonstratedthatitissuperiorto

the mJOA‐only model, this assessment does not provide information on

whetheritwouldbesuperiortoamultivariableclinical‐basedmodel.

2) Inclusion of baseline mJOA – The final model was not only based on

imaging factors but also includedbaselinemJOA.While this is noteworthy,

the inclusion of baseline mJOA is necessary to address the statistical

phenomenonofregressiontothemean.

3) DichotomousmJOA outcome – Prediction of patient outcomes was not

assessed linearly in terms of change. Looking at differences in the mJOA

wouldbeusefultoobservewhethersubtlechangesnotlargeenoughtomove

patientsbetweenthedichotomousgroupscouldbepredicted.However,the

ceilingeffectofthemJOAandtherecognitionthatnotallpointsonthemJOA

score carry equal weight makes this a challenging endeavor when using

linearregressionmodeling.

118

4.9Conclusion

In counseling patients who are considering surgical treatment for DCM, it is

necessary to provide information on their anticipated recovery. Since MRI is

routinelyconductedtoconfirmadiagnosisofDCM,informationregardingsurgical

outcome derived from its assessment provides a unique opportunity to ascertain

additionalguidance.Inrecognizingthis,itwastheprincipalobjectiveofthisthesis

to investigate what role MR imaging plays in predicting the surgical outcome of

patientstreatedforDCM.Tostudythisrelationshipweevaluatedacohortofpatient

fromtheAOSpine‐NAprospectiveandmulticenterstudy,andhypothesizedthatMRI

characteristics would be prognostic of outcome. In conducting our analysis, we

created a protocol to facilitate the systematic review of MR images based on

previously published and modified methods. Through univariate analysis we

indicatedthatagreaterMCCandgreaterT2‐WIhyperintensitysize,intermsofboth

areaaswellassagittalextent,wassignificantlyassociatedwithsubstantialresidual

neurologicalimpairmentat6‐monthsaftersurgery.Wefurtherdemonstratedusing

multivariate logistic regression, that the presence of T1‐WI signal change and

degreeofMCC,inadditiontobaselinefunctionalimpairment,providedaverygood

ability(AUCof0.84)todiscriminatebetweenourdichotomousoutcome(mJOA<16,

≥16). Furthermore, this model demonstrated superior predictive capacity than a

modelbasedsolelyonbaselinefunctionalimpairmentandsupportsthatMRIhasa

separateanddistinctrole.

Ourstudyrepresents theonlyprospectiveandmulticenterstudytodateto

investigatetheroleofMRIinpredictingsurgicaloutcome.Wehaveutilizedalarge

number of MRImeasuring techniques and have determined throughmultivariate

analysis thatMRI factors have a distinct capacity to provide information on how

patientscanexpecttofare.Thisinformationispivotalforbothpatients,whomust

decidewhether to undergo surgical treatment, and for clinicians,who are tasked

withmakingtherapeuticrecommendations.Therefore, it isstronglyrecommended

that a detailed MRI analysis with particular emphasis onMCCmeasurement and

119

evaluationofT1‐WIsignalhypointensitybeconductedinallsurgicalcandidatesto

helpguideclinicaldecision‐makingandmanagepatientexpectations.

120

CHAPTER5FutureDirections

121

5.1Introduction

While considerable efforts have been made in this thesis to address the present

knowledge gap regarding the role of MRI in predicting surgical outcome, there

remains ample opportunity to investigate this subject further. In the following

sections,anumberofworthyfuturedirectionsofthisresearchareexplored.

5.2AssessmentofReliabilityandExternalValidity

It would be a valuable contribution to the field to establish the measurement

reliability of the quantitative MRI methods outlined in this thesis, as it will

determinewhether thesemeasurements are effective tools that can be translated

fromacademiccentersintoclinicalpractice.Inconductingthisresearch,itwouldbe

imperativetohaveateamofexpertsagreeonaprotocolforMRIassessment.

The reliability of the quantitativeMRImeasurements can be estimated by

assessing the intraclass and interclass correlation coefficients of multiple

investigators. Itwouldbepreferred that thisundertakingwould includeadiverse

groupof specialists coming fromdifferent global regions. The results can identify

methodsthatareeffectiveandindicatehowtheycanbeimprovedgoingforward.

In addition, if baseline clinical severity assessment (via the mJOA) were

collectedwhenconductingsuchresearch, therewouldbeauniqueopportunity to

externallyvalidatethepredictionmodelpresentedinthisthesis.

5.3CombiningthePredictiveCapacityofImagingandClinical

Parameters

Predicting surgical outcome of DCM patients can also be achieved using clinical

parameters.Asdiscussedearlierinthethesis,aclinicalpredictionmodelbasedon

the full AOSpine North America study cohort was published by Tetreault et al

(2013c) very recently. Given the findings presented in chapter 3, it would be

interesting to pursue future research that incorporated the combined predictive

capacity of clinical and imaging factors. Since such work will likely entail an

investigationofmultipleparameters,itmaybecomenecessarytoinvestigateamuch

122

larger cohort of patients to adequately present the diverse clinical pictures that

patientsmaymanifestwith.Additionally,asparsimoniousmodelsarepreferred,itis

plausible that some of the many previously established predictors would be

removedfromacombinedmodel.Thismaybebeneficialhowever,asitmayleadto

the retention of highly predictive parameters. A combinedmodelmay result in a

comparable or even superior predictive capacity than either a clinical or imaging

modelinisolation.

5.4AnApproachforDefiningT1andT2WeightedMRISignal

ChangesQuantitatively

CurrentassessmentfordeterminingthepresenceorabsenceofsignalchangeonT1‐

WIandT2‐WIislargelybasedonvisualinspection.Thoughtherearemeritstothis

methodasitcansavetime,particularlyincaseswheresignalintensityisdistinctly

contrasted fromadjacent tissueand iswellcircumscribed, itcanbechallengingto

classify a definitive presence of signal change when the intensity differences are

weakandappearfuzzyordiffuse. Thisdifficultybecomesclinicallyrelevantwhen

assessing patients for anatomical evidence of myelopathy, since prognostic

informationiscommonlyconveyedintermsofthebinaryclassificationofpresence

orabsence.

In lieu of visual inspection, it may be possible to develop quantitative

measurementtechniques.Thismaybeachievablebyanalyzingthesignal intensity

of the entire sagittal plane of the cervical spinal cord (C2‐C7) and plotting the

distributionofsignalintensity.Itwouldbeexpectedthatthesignalintensitywould

distribute normally and therefore deviations of intensity frequencies (i.e.

hyperintensity regions on T2‐WI and hypointensity regions on T1‐WI) could be

statisticallyevaluated.TheconceptisvisuallydescribedinFigure4.1.Moreover,this

process could serve as a potential tool to define a quantitative level of signal

intensitydeviationthatisnecessarytoconstituteadefinitiveregionofalteredsignal

intensity.Comparingthefrequencyofsignalchangevoxelstotheoverallamountof

voxelscouldalsoestimate thesizeof signal intensitychangerelative to theentire

123

cervical spinal cord.This could be calculatedby comparing thequantity of voxels

under the normal curve to the quantity of voxels under the distribution of the

abnormalsignalintensitycurve.

Finally,ifsignalintensitychangesareconvertedintosignalchangeratiosas

we have shown in chapter 3, there is an opportunity to assesswhether different

degrees of deviation represent different anatomical pathologies. Since intensity

changesdonotrepresentanatomicalchanges througha linearrelationship, future

analysis of signal change ratiosmay be better evaluated through cluster analysis

methods. Itmay also be appropriate to conduct such analysis using true T2MRI

sequencesinsteadofT2weightedsequences,asthiswouldreflectamoreaccurate

T2imageofthespinalcord.

124

Figure5.1AconceptualillustrationofhowT1‐WIandT2‐WIsignalchangecouldbedefinedquantitatively.UsingasagittalMRimagethesignalintensityonT1‐WIandT2‐WIofthecervicalspineregioncanbemeasurefromC2‐C7.Additionally,aregionofinterestcanbecircumscribed.Whensignalintensityvaluesfromeachvoxelinthewholeareaareplottedintermsoffrequency,itwouldbeexpectedthattheywoulddistributenormally.Signalintensitychanges(hypointense,closerto0;hyperintense,furtherawayfrom0)wouldformseparatedistributionsandthedeviationsofthesemeansfromthenorm(M1,M2)canbestatisticallyassessed.

125

5.5AdvancedMRITechniquesandtheFutureofDCMAssessment

Overtheyears,withimprovementintechnologybothintermsofhardwareaswell

as software, additional novel application have emerged that have increased the

functional utility of MRI. These improvements include the ability to obtain

volumetricdata,toobtaindynamic/kinematicMRimaging,andtheabilityofMRto

providephysiologicalaswellaschemicalinformation.Thesetechnologiesmayoffer

anewsetofapproachesthatcanassistinamoresensitiveandspecificassessment

ofDCM.Itshouldbetakenintoconsiderationthatthesetechnologiesremainlargely

within the realmof research and their potential clinical applicationmaybemany

yearsaway.Inthefollowingsections,afewofthesetechnologiesandtheirpotential

utilityarebrieflydiscussed.

5.5.1Dynamic/Kinematic&UprightMRI

OneofthechieflimitationstoconventionalMRIinevaluatingDCMisthatitprovides

staticimaging.Inadditiontothis,imagesarealsoalmostexclusivelyobtainedwhile

thepatientislyingdown.ToovercometheselimitationsfutureanalysisofDCMmay

beconductedusingupright‐neutralposition(pMRI)aswellasdynamic/kinematic

MRimaging(dkMRI).

WhenusingapMRI,thepatientisuprightasopposedtolyingsupineduring

image acquisition, thereby allowing for the assessment of the spine during axial

loading. This is relevant for patients being assessed for DCM as IVDs function to

absorb considerable compressive loads from carrying structures above the spine

segments.InthesettingofDDD,whenadiscmayhavelostthestructuralintegrityof

theannulusfibrosis,potentialmigrationofdiscelementsintothecanalmaybeload‐

dependent. Accordingly, pMRImay serve to unmask occult disease (Jinkins et al.,

2005).

dkMRobtains imagingwhile thepatient ismovingandthereforeallows for

the assessment of kinematic‐dependent disease (Jinkins et al., 2005). During

movementinvolvingthecervicalspine,patientsmaydemonstratehypermobilityof

spine segments or listhesis, which may contribute to repetitive insult upon the

126

spinal cord.Suchapathomechanismwouldbe impossible tocaptureduringstatic

MRimaging.

Ultimately, these imaging techniquesmay provide essential pre‐ and post‐

operative information that have implications not only on improving prognostic

guidancebutalsoonsurgicalplanning.Additionally, these technologiesconfer the

abilitytoscanthepatientinthepositionofclinicallyrelevantsignsandsymptoms

(Jinkinsetal.,2005).

5.5.2AdvancedMRITechniques:DiffusionTensorMRI,Blood‐

Oxygen‐LevelDependent(BOLD),andNuclearMagneticResonance

(NMR)Spectroscopy

ThereareanumberofadvancedMRItechniquesthatmayofferadditionalmethods

bywhich patientswith DCMmay be assessed in the future. Themost exciting of

these,dtMRI,wasbrieflynotedinchapter1.TheabilityfordtMRItomapfibertracts

is of particular value forpatientswithDCMas it canhelp identify interruptionof

neuraltractwiringduetocavityformationorWalleriandegenerationforexample.

AparticularlyinterestingfindingdescribedbyJonesetal(2013)wastheirabilityto

discoverpathologicalchangesinthecordpriortothefindingofhyperintensityT2‐

WIsignalchanges.Thisfindingrequiresadditionalinquiryandvalidationasitmay

helpdeterminepatientsatriskforfurtherSCI.Also,knowledgeoftheevolutionof

signalchangewillallowforbetterunderstandingoftheunderlyingdiseaseprocess.

Itisthereforelikelythatsubstantiallymoreworkinthisareawillappearinthenear

future.

Another technique,whichwasalsodiscussedbriefly in chapter1, isBlood‐

Oxygen‐Level Dependent (BOLD) MRI. BOLD may provide useful information

regarding functional and damaged neural tissue in the spinal cord. Therefore,

assessingdifferencesinpre‐andpost‐operativeBOLDmaybeusefulindetermining

thepotential forneurologicalrecovery.However,whilethereisgreatpotential for

itsclinicalapplication,BOLDremainsintherealmofinvestigation.

127

Yet another technique with potential future application for the DCM

assessment is Nuclear Magnetic Resonance (NMR) Spectroscopy. Two commonly

usedmethodsforconductingNMRspectroscopyare:Singlevoxelspectroscopyand

chemical shift imaging. Both methods allow for the analysis of the chemical

environmentinaregionofinterestonMRI(Rungeetal.,2014).Atthepresenttime,

NMRspectroscopycanspecificallydetectandassesstheabundanceofanumberof

specific chemical compounds including glutamate, N‐acetylaspartate and choline

(Rungeetal.,2014).Giventhattraumatothespinalcordisthoughttobefollowed

by a secondary injury process involving glutamate excitotoxicity and other

molecular changes at the region of interest, NMR spectroscopy may serve as a

unique tool for providing chemical correlations with structural changes seen on

imaging.

AlthoughmanyMRI techniques arebeing investigated andothersmay still

emerge in the coming years,many challenges (e.g. economical and academic) for

their integration into clinical practice lie ahead. Therefore, excitement over these

innovationsshouldbeappreciatedcautiously.

128

REFERENCESABATE,M.,VANNI,D.,PANTALONE,A.&SALINI,V.2013.Cigarettesmokingand

musculoskeletaldisorders.MusclesLigamentsTendonsJ,3,63‐9.AL‐JARALLAH,K.,AL‐SAEED,O.,SHEHAB,D.,DASHTI,K.&SHEIKH,M.2012.

OssificationofligamentumflavuminMiddleEastArabs:ahospital‐basedstudy.MedPrincPract,21,529‐33.

AL‐MEFTY,O.,HARKEY,L.H.,MIDDLETON,T.H.,SMITH,R.R.&FOX,J.L.1988.Myelopathiccervicalspondyloticlesionsdemonstratedbymagneticresonanceimaging.JNeurosurg,68,217‐22.

ALI,F.E.,AL‐BUSTAN,M.A.,AL‐BUSAIRI,W.A.,AL‐MULLA,F.A.&ESBAITA,E.Y.2006.CervicalspineabnormalitiesassociatedwithDownsyndrome.IntOrthop,30,284‐9.

AMES,C.P.,BLONDEL,B.,SCHEER,J.K.,SCHWAB,F.J.,LEHUEC,J.C.,MASSICOTTE,E.M.,PATEL,A.A.,TRAYNELIS,V.C.,KIM,H.J.,SHAFFREY,C.I.,SMITH,J.S.&LAFAGE,V.2013.Cervicalradiographicalalignment:comprehensiveassessmenttechniquesandpotentialimportanceincervicalmyelopathy.Spine(PhilaPa1976),38,S149‐60.

ARDEN,N.&NEVITT,M.C.2006.Osteoarthritis:epidemiology.BestPractResClinRheumatol,20,3‐25.

ARVIN,B.,KALSI‐RYAN,S.,KARPOVA,A.,MERCIER,D.,FURLAN,J.C.,MASSICOTTE,E.M.&FEHLINGS,M.G.2011.Postoperativemagneticresonanceimagingcanpredictneurologicalrecoveryaftersurgeryforcervicalspondyloticmyelopathy:aprospectivestudywithblindedassessments.Neurosurgery,69,362‐8.

BAJWA,N.S.,TOY,J.O.,YOUNG,E.Y.&AHN,N.U.2012.Establishmentofparametersforcongenitalstenosisofthecervicalspine:ananatomicdescriptiveanalysisof1,066cadavericspecimens.EurSpineJ,21,2467‐74.

BAPTISTE,D.&FEHLINGS,M.2006a.Pathophysiologyofcervicalmyelopathy.Thespinejournal:officialjournaloftheNorthAmericanSpineSociety,6.

BAPTISTE,D.C.&FEHLINGS,M.G.2006b.Pathophysiologyofcervicalmyelopathy.SpineJ,6,190S‐197S.

BARTOLOMEI,J.C.,THEODORE,N.&SONNTAG,V.K.2005.Adjacentleveldegenerationafteranteriorcervicalfusion:aclinicalreview.NeurosurgClinNAm,16,575‐87,v.

BEATTIE,M.S.&MANLEY,G.T.2011.Tightsqueeze,slowburn:inflammationandtheaetiologyofcervicalmyelopathy.Brain,134,1259‐61.

BEDNARIK,J.,KADANKA,Z.,DUSEK,L.,KERKOVSKY,M.,VOHANKA,S.,NOVOTNY,O.,URBANEK,I.&KRATOCHVILOVA,D.2008.Presymptomaticspondyloticcervicalmyelopathy:anupdatedpredictivemodel.EurSpineJ,17,421‐31.

BEDNARIK,J.,SLADKOVA,D.,KADANKA,Z.,DUSEK,L.,KERKOVSKY,M.,VOHANKA,S.,NOVOTNY,O.,URBANEK,I.&NEMEC,M.2011.Aresubjectswithspondyloticcervicalcordencroachmentatincreasedriskofcervicalspinalcordinjuryafterminortrauma?JNeurolNeurosurgPsychiatry,82,779‐81.

129

BENHAMOUDA,K.,JEMEL,H.,HAOUET,S.&KHALDI,M.2003.Thoracicmyelopathycausedbyossificationoftheligamentumflavum:areportof18cases.JNeurosurg,99,157‐61.

BENZEL,E.C.,LANCON,J.,KESTERSON,L.&HADDEN,T.1991.Cervicallaminectomyanddentateligamentsectionforcervicalspondyloticmyelopathy.JSpinalDisord,4,286‐95.

BERGE,J.,MARQUE,B.,VITAL,J.M.,SENEGAS,J.&CAILLE,J.M.1999.Age‐relatedchangesinthecervicalspinesoffront‐linerugbyplayers.AmJSportsMed,27,422‐9.

BERNHARDT,M.,HYNES,R.A.,BLUME,H.W.&WHITE,A.A.,3RD1993.Cervicalspondyloticmyelopathy.JBoneJointSurgAm,75,119‐28.

BISTA,P.&ROKA,Y.B.2008.CervicalspondylosisinNepaleseporters.JNMAJNepalMedAssoc,47,220‐3.

BLUMENFELD,H.2010.NeuroanatomyThroughClinicalCases,Sunderland,Massachusetts,SinauerAssociates.

BOOGAARTS,H.D.&BARTELS,R.H.2013.Prevalenceofcervicalspondyloticmyelopathy.EurSpineJ.

BOSMA,G.P.,VANBUCHEM,M.A.,VOORMOLEN,J.H.,VANBIEZEN,F.C.&BROUWER,O.F.1999.CervicalspondylarthroticmyelopathywithearlyonsetinDown'ssyndrome:fivecasesandareviewoftheliterature.JIntellectDisabilRes,43(Pt4),283‐8.

BRADLEY,E.H.,SCHLESINGER,M.,WEBSTER,T.R.,BAKER,D.&INOUYE,S.K.2004.Translatingresearchintoclinicalpractice:makingchangehappen.JAmGeriatrSoc,52,1875‐82.

BROWN,M.W.,TEMPLETON,A.W.&HODGES,F.J.,3RD1964.TheIncidenceofAcquiredandCongenitalFusionsintheCervicalSpine.AmJRoentgenolRadiumTherNuclMed,92,1255‐9.

BULL,J.,ELGAMMAL,T.&POPHAM,M.1969.Apossiblegeneticfactorincervicalspondylosis.BrJRadiol,42,9‐16.

CABRE,P.,PASCAL‐MOUSSELLARD,H.,KAIDOMAR,S.,BUCKI,B.,BARDIN,T.,SMADJA,D.&ARFI,S.2001.SixcasesofcervicalligamentumflavumcalcificationinBlacksintheFrenchWestIndies.JointBoneSpine,68,158‐65.

CHAPUT,C.D.,ALLRED,J.J.,PANDORF,J.J.,SONG,J.&RAHM,M.D.2013.Thesignificanceoffacetjointcross‐sectionalareaonmagneticresonanceimaginginrelationshiptocervicaldegenerativespondylolisthesis.SpineJ,13,856‐61.

CHEN,C.J.,LYU,R.K.,LEE,S.T.,WONG,Y.C.&WANG,L.J.2001.IntramedullaryhighsignalintensityonT2‐weightedMRimagesincervicalspondyloticmyelopathy:predictionofprognosiswithtypeofintensity.Radiology,221,789‐94.

CIHR.2014.MoreAboutKnowledgeTranslationatCIHR[Online].GovernmentofCanada.Available:http://www.cihr‐irsc.gc.ca/e/39033.html[AccessedNovember3rd,20142014].

COUNTEE,R.W.&VIJAYANATHAN,T.1979.Congenitalstenosisofthecervicalspine:diagnosisandmanagement.JNatlMedAssoc,71,257‐64.

DELBIGIO,M.R.&JOHNSON,G.E.1989.Clinicalpresentationofspinalcordconcussion.Spine(PhilaPa1976),14,37‐40.

130

ECHARRI,J.J.&FORRIOL,F.2005.Influenceofthetypeofloadonthecervicalspine:astudyonCongolesebearers.SpineJ,5,291‐6.

EPSTEIN,N.E.1994.Thesurgicalmanagementofossificationoftheposteriorlongitudinalligamentin43northamericans.Spine(PhilaPa1976),19,664‐72.

ERNST,C.W.,STADNIK,T.W.,PEETERS,E.,BREUCQ,C.&OSTEAUX,M.J.2005.PrevalenceofannulartearsanddischerniationsonMRimagesofthecervicalspineinsymptomfreevolunteers.EurJRadiol,55,409‐14.

ERWIN,W.M.&FEHLINGS,M.G.2013.Intervertebraldiscdegeneration:genesholdthekey.WorldNeurosurg,80,e131‐3.

FARRY,A.&BAXTER,D.2010.TheincidenceandprevalenceofspinalcordinjuryinCanada:Overviewandestimatesbasedoncurrentevidence.RickHansenInstituteandUrbanFutures.

FEHLINGS,M.G.,FURLAN,J.C.,MASSICOTTE,E.M.,ARNOLD,P.,AARABI,B.,HARROP,J.,ANDERSON,D.G.,BONO,C.M.,DVORAK,M.,FISHER,C.,FRANCE,J.,HEDLUND,R.,MADRAZO,I.,NOCKELS,R.,RAMPERSAUD,R.,RECHTINE,G.,VACCARO,A.R.&SPINETRAUMASTUDY,G.2006.Interobserverandintraobserverreliabilityofmaximumcanalcompromiseandspinalcordcompressionforevaluationofacutetraumaticcervicalspinalcordinjury.Spine(PhilaPa1976),31,1719‐25.

FEHLINGS,M.G.,RAO,S.C.,TATOR,C.H.,SKAF,G.,ARNOLD,P.,BENZEL,E.,DICKMAN,C.,CUDDY,B.,GREEN,B.,HITCHON,P.,NORTHRUP,B.,SONNTAG,V.,WAGNER,F.&WILBERGER,J.1999.Theoptimalradiologicmethodforassessingspinalcanalcompromiseandcordcompressioninpatientswithcervicalspinalcordinjury.PartII:Resultsofamulticenterstudy.Spine(PhilaPa1976),24,605‐13.

FEHLINGS,M.G.,WILSON,J.R.,KARADIMAS,S.K.,ARNOLD,P.M.&KOPJAR,B.2013a.Clinicalevaluationofaneuroprotectivedruginpatientswithcervicalspondyloticmyelopathyundergoingsurgicaltreatment:designandrationalefortheCSM‐Protecttrial.Spine(PhilaPa1976),38,S68‐75.

FEHLINGS,M.G.,WILSON,J.R.,YOON,S.T.,RHEE,J.M.,SHAMJI,M.F.&LAWRENCE,B.D.2013b.Symptomaticprogressionofcervicalmyelopathyandtheroleofnonsurgicalmanagement:aconsensusstatement.Spine(PhilaPa1976),38,S19‐20.

FELSON,D.T.,ZHANG,Y.,HANNAN,M.T.,NAIMARK,A.,WEISSMAN,B.,ALIABADI,P.&LEVY,D.1997.Riskfactorsforincidentradiographickneeosteoarthritisintheelderly:theFraminghamStudy.ArthritisRheum,40,728‐33.

FENGBIN,Y.,DEYU,C.,XINWEI,W.,YU,C.,JINHAO,M.,XINYUAN,L.&XIAOWEI,L.2013.Trauma‐inducedspinalcordinjuryincervicalspondyloticmyelopathywithorwithoutlowercervicalinstability.JClinNeurosci,20,419‐22.

FIX,J.D.2001.BRSNeuroanatomy(BoardReviewSeries),LippincottWilliams&Wilkins.

FUJIYOSHI,T.,YAMAZAKI,M.,OKAWA,A.,KAWABE,J.,HAYASHI,K.,ENDO,T.,FURUYA,T.,KODA,M.&TAKAHASHI,K.2010.Staticversusdynamicfactorsforthedevelopmentofmyelopathyinpatientswithcervicalossificationoftheposteriorlongitudinalligament.JClinNeurosci,17,320‐4.

131

GALBUSERA,F.,VANRIJSBERGEN,M.,ITO,K.,HUYGHE,J.M.,BRAYDA‐BRUNO,M.&WILKE,H.J.2014.Ageinganddegenerativechangesoftheintervertebraldiscandtheirimpactonspinalflexibility.EurSpineJ,23Suppl3,S324‐32.

GIAMPIETRO,P.F.,RAGGIO,C.L.,BLANK,R.D.,MCCARTY,C.,BROECKEL,U.&PICKART,M.A.2013.Clinical,geneticandenvironmentalfactorsassociatedwithcongenitalvertebralmalformations.MolSyndromol,4,94‐105.

GUO,J.J.,LUK,K.D.,KARPPINEN,J.,YANG,H.&CHEUNG,K.M.2010.Prevalence,distribution,andmorphologyofossificationoftheligamentumflavum:apopulationstudyofonethousandsevenhundredthirty‐sixmagneticresonanceimagingscans.Spine(PhilaPa1976),35,51‐6.

HALKO,G.J.,COBB,R.&ABELES,M.1995.PatientswithtypeIVEhlers‐Danlossyndromemaybepredisposedtoatlantoaxialsubluxation.JRheumatol,22,2152‐5.

HAMALAINEN,O.,VANHARANTA,H.&KUUSELA,T.1993.Degenerationofcervicalintervertebraldisksinfighterpilotsfrequentlyexposedtohigh+Gzforces.AviatSpaceEnvironMed,64,692‐6.

HARSH,G.R.T.,SYPERT,G.W.,WEINSTEIN,P.R.,ROSS,D.A.&WILSON,C.B.1987.Cervicalspinestenosissecondarytoossificationoftheposteriorlongitudinalligament.JNeurosurg,67,349‐57.

HAYASHI,K.,ISHIDOU,Y.,YONEMORI,K.,NAGAMINE,T.,ORIGUCHI,N.,MAEDA,S.,IMAMURA,T.,KATO,M.,YOSHIDA,H.,SAMPATH,T.K.,TENDIJKE,P.&SAKOU,T.1997.Expressionandlocalizationofbonemorphogeneticproteins(BMPs)andBMPreceptorsinossificationoftheligamentumflavum.Bone,21,23‐30.

HAYASHI,T.,WANG,J.C.,SUZUKI,A.,TAKAHASHI,S.,SCOTT,T.P.,PHAN,K.,LORD,E.L.,RUANGCHAINIKOM,M.,SHIBA,K.&DAUBS,M.D.2014.Riskfactorsformisseddynamiccanalstenosisinthecervicalspine.Spine(PhilaPa1976),39,812‐9.

HENDRIKSEN,I.J.&HOLEWIJN,M.1999.DegenerativechangesofthespineoffighterpilotsoftheRoyalNetherlandsAirForce(RNLAF).AviatSpaceEnvironMed,70,1057‐63.

HIRAI,T.,UCHIDA,K.,NAKAJIMA,H.,GUERRERO,A.R.,TAKEURA,N.,WATANABE,S.,SUGITA,D.,YOSHIDA,A.,JOHNSON,W.E.&BABA,H.2013.Theprevalenceandphenotypeofactivatedmicroglia/macrophageswithinthespinalcordofthehyperostoticmouse(twy/twy)changesinresponsetochronicprogressivespinalcordcompression:implicationsforhumancervicalcompressivemyelopathy.PLoSOne,8,e64528.

HOLLY,L.T.,MATZ,P.G.,ANDERSON,P.A.,GROFF,M.W.,HEARY,R.F.,KAISER,M.G.,MUMMANENI,P.V.,RYKEN,T.C.,CHOUDHRI,T.F.,VRESILOVIC,E.J.,RESNICK,D.K.,JOINTSECTIONONDISORDERSOFTHE,S.,PERIPHERALNERVESOFTHEAMERICANASSOCIATIONOFNEUROLOGICAL,S.&CONGRESSOFNEUROLOGICAL,S.2009.Clinicalprognosticindicatorsofsurgicaloutcomeincervicalspondyloticmyelopathy.JNeurosurgSpine,11,112‐8.

132

HORI,T.,KAWAGUCHI,Y.&KIMURA,T.2006.Howdoestheossificationareaoftheposteriorlongitudinalligamentprogressaftercervicallaminoplasty?Spine(PhilaPa1976),31,2807‐12.

HUKUDA,S.&KOJIMA,Y.2002.Sexdiscrepancyinthecanal/bodyratioofthecervicalspineimplicatingtheprevalenceofcervicalmyelopathyinmen.Spine(PhilaPa1976),27,250‐3.

IKEDA,R.,YOSHIDA,K.,TSUKAHARA,S.,SAKAMOTO,Y.,TANAKA,H.,FURUKAWA,K.&INOUE,I.2005.ThepromyeloticleukemiazincfingerpromotesosteoblasticdifferentiationofhumanmesenchymalstemcellsasanupstreamregulatorofCBFA1.JBiolChem,280,8523‐30.

INAMASU,J.,GUIOT,B.H.&SACHS,D.C.2006.Ossificationoftheposteriorlongitudinalligament:anupdateonitsbiology,epidemiology,andnaturalhistory.Neurosurgery,58,1027‐39;discussion1027‐39.

INOUE,H.,SEICHI,A.,KIMURA,A.,ENDO,T.&HOSHINO,Y.2013.Multiple‐levelossificationoftheligamentumflavuminthecervicalspinecombinedwithcalcificationofthecervicalligamentumflavumandposterioratlanto‐axialmembrane.EurSpineJ,22Suppl3,S416‐20.

ISHIHARA,H.,KANAMORI,M.,KAWAGUCHI,Y.,NAKAMURA,H.&KIMURA,T.2004.Adjacentsegmentdiseaseafteranteriorcervicalinterbodyfusion.SpineJ,4,624‐8.

IWASAWA,T.,IWASAKI,K.,SAWADA,T.,OKADA,A.,UEYAMA,K.,MOTOMURA,S.,HARATA,S.,INOUE,I.,TOH,S.&FURUKAWA,K.I.2006.Pathophysiologicalroleofendothelininectopicossificationofhumanspinalligamentsinducedbymechanicalstress.CalcifTissueInt,79,422‐30.

JACOBS,W.,WILLEMS,P.C.,VANLIMBEEK,J.,BARTELS,R.,PAVLOV,P.,ANDERSON,P.G.&ONER,C.2011.Singleordouble‐levelanteriorinterbodyfusiontechniquesforcervicaldegenerativediscdisease.CochraneDatabaseSystRev,CD004958.

JAGER,H.J.,GORDON‐HARRIS,L.,MEHRING,U.M.,GOETZ,G.F.&MATHIAS,K.D.1997.Degenerativechangeinthecervicalspineandload‐carryingonthehead.SkeletalRadiol,26,475‐81.

JAYAKUMAR,P.N.,KOLLURI,V.R.,VASUDEV,M.K.&SRIKANTH,S.G.1996.OssificationoftheposteriorlongitudinalligamentofthecervicalspineinAsianIndians‐‐amultiracialcomparison.ClinNeurolNeurosurg,98,142‐8.

JEON,T.S.,CHANG,H.&CHOI,B.W.2012.Analysisofdemographics,clinical,andradiographicalfindingsofossificationofposteriorlongitudinalligamentofthecervicalspinein146Koreanpatients.Spine(PhilaPa1976),37,E1498‐503.

JIANG,S.D.,JIANG,L.S.&DAI,L.Y.2011.Degenerativecervicalspondylolisthesis:asystematicreview.IntOrthop,35,869‐75.

JINKINS,J.R.,DWORKIN,J.S.&DAMADIAN,R.V.2005.Upright,weight‐bearing,dynamic‐kineticMRIofthespine:initialresults.EurRadiol,15,1815‐25.

JOHNSTON,S.L.,CAMPBELL,M.R.,SCHEURING,R.&FEIVESON,A.H.2010.RiskofherniatednucleuspulposusamongU.S.astronauts.AviatSpaceEnvironMed,81,566‐74.

133

JONES,J.G.,CEN,S.Y.,LEBEL,R.M.,HSIEH,P.C.&LAW,M.2013.Diffusiontensorimagingcorrelateswiththeclinicalassessmentofdiseaseseverityincervicalspondyloticmyelopathyandpredictsoutcomefollowingsurgery.AJNRAmJNeuroradiol,34,471‐8.

JONSSON,H.,JR.,CESARINI,K.,SAHLSTEDT,B.&RAUSCHNING,W.1994.Findingsandoutcomeinwhiplash‐typeneckdistortions.Spine(PhilaPa1976),19,2733‐43.

JOOSAB,M.,TORODE,M.&RAO,P.V.1994.Preliminaryfindingsontheeffectofload‐carryingtothestructuralintegrityofthecervicalspine.SurgRadiolAnat,16,393‐8.

JUMAH,K.B.&NYAME,P.K.1994.RelationshipbetweenloadcarryingontheheadandcervicalspondylosisinGhanaians.WestAfrJMed,13,181‐2.

KADANKA,Z.,BEDNARIK,J.,NOVOTNY,O.,URBANEK,I.&DUSEK,L.2011.Cervicalspondyloticmyelopathy:conservativeversussurgicaltreatmentafter10years.EurSpineJ,20,1533‐8.

KADANKA,Z.,MARES,M.,BEDNANIK,J.,SMRCKA,V.,KRBEC,M.,STEJSKAL,L.,CHALOUPKA,R.,SURELOVA,D.,NOVOTNY,O.,URBANEK,I.&DUSEK,L.2002.Approachestospondyloticcervicalmyelopathy:conservativeversussurgicalresultsina3‐yearfollow‐upstudy.Spine(PhilaPa1976),27,2205‐10;discussion2210‐1.

KALSI‐RYAN,S.,KARADIMAS,S.K.&FEHLINGS,M.G.2013.Cervicalspondyloticmyelopathy:theclinicalphenomenonandthecurrentpathobiologyofanincreasinglyprevalentanddevastatingdisorder.Neuroscientist,19,409‐21.

KAMEYAMA,T.,HASHIZUME,Y.,ANDO,T.&TAKAHASHI,A.1994.Morphometryofthenormalcadavericcervicalspinalcord.Spine(PhilaPa1976),19,2077‐81.

KARADIMAS,S.,ERWIN,W.,ELY,C.,DETTORI,J.&FEHLINGS,M.2013a.Pathophysiologyandnaturalhistoryofcervicalspondyloticmyelopathy.Spine,38,36.

KARADIMAS,S.K.,ERWIN,W.M.,ELY,C.G.,DETTORI,J.R.&FEHLINGS,M.G.2013b.Pathophysiologyandnaturalhistoryofcervicalspondyloticmyelopathy.Spine(PhilaPa1976),38,S21‐36.

KARADIMAS,S.K.,GATZOUNIS,G.&FEHLINGS,M.G.2014.Pathobiologyofcervicalspondyloticmyelopathy.EurSpineJ.

KARADIMAS,S.K.,GIALELI,C.H.,KLIRONOMOS,G.,TZANAKAKIS,G.N.,PANAGIOTOPOULOS,E.,KARAMANOS,N.K.&GATZOUNIS,G.2010.Theroleofoligodendrocytesinthemolecularpathobiologyandpotentialmoleculartreatmentofcervicalspondyloticmyelopathy.CurrMedChem,17,1048‐58.

KARADIMAS,S.K.,KLIRONOMOS,G.,PAPACHRISTOU,D.J.,PAPANIKOLAOU,S.,PAPADAKI,E.&GATZOUNIS,G.2013c.ImmunohistochemicalprofileofNF‐kappaB/p50,NF‐kappaB/p65,MMP‐9,MMP‐2,andu‐PAinexperimentalcervicalspondyloticmyelopathy.Spine(PhilaPa1976),38,4‐10.

KARADIMAS,S.K.,MOON,E.S.,YU,W.R.,SATKUNENDRARAJAH,K.,KALLITSIS,J.K.,GATZOUNIS,G.&FEHLINGS,M.G.2013d.Anovelexperimentalmodelofcervicalspondyloticmyelopathy(CSM)tofacilitatetranslationalresearch.NeurobiolDis,54,43‐58.

134

KARASUGI,T.,NAKAJIMA,M.,IKARI,K.,GENETICSTUDYGROUPOFINVESTIGATIONCOMMITTEEONOSSIFICATIONOFTHESPINAL,L.,TSUJI,T.,MATSUMOTO,M.,CHIBA,K.,UCHIDA,K.,KAWAGUCHI,Y.,MIZUTA,H.,OGATA,N.,IWASAKI,M.,MAEDA,S.,NUMASAWA,T.,ABUMI,K.,KATO,T.,OZAWA,H.,TAGUCHI,T.,KAITO,T.,NEO,M.,YAMAZAKI,M.,TADOKORO,N.,YOSHIDA,M.,NAKAHARA,S.,ENDO,K.,IMAGAMA,S.,DEMURA,S.,SATO,K.,SEICHI,A.,ICHIMURA,S.,WATANABE,M.,WATANABE,K.,NAKAMURA,Y.,MORI,K.,BABA,H.,TOYAMA,Y.&IKEGAWA,S.2013.Agenome‐widesib‐pairlinkageanalysisofossificationoftheposteriorlongitudinalligamentofthespine.JBoneMinerMetab,31,136‐43.

KARPOVA,A.,ARUN,R.,CADOTTE,D.W.,DAVIS,A.M.,KULKARNI,A.V.,O'HIGGINS,M.&FEHLINGS,M.G.2013a.Assessmentofspinalcordcompressionbymagneticresonanceimaging‐‐canitpredictsurgicaloutcomesindegenerativecompressivemyelopathy?Asystematicreview.Spine(PhilaPa1976),38,1409‐21.

KARPOVA,A.,ARUN,R.,DAVIS,A.M.,KULKARNI,A.V.,MASSICOTTE,E.M.,MIKULIS,D.J.,LUBINA,Z.I.&FEHLINGS,M.G.2013b.Predictorsofsurgicaloutcomeincervicalspondyloticmyelopathy.Spine(PhilaPa1976),38,392‐400.

KARPOVA,A.,ARUN,R.,DAVIS,A.M.,KULKARNI,A.V.,MIKULIS,D.J.,SOOYONG,C.,RABIN,D.,CRACIUNAS,S.,SMITH,S.R.,HANSEN,M.A.,GEORGE,J.&FEHLINGS,M.G.2013c.Reliabilityofquantitativemagneticresonanceimagingmethodsintheassessmentofspinalcanalstenosisandcordcompressionincervicalmyelopathy.Spine(PhilaPa1976),38,245‐52.

KAWAGUCHI,Y.,KANAMORI,M.,ISHIHARA,H.,NAKAMURA,H.,SUGIMORI,K.,TSUJI,H.&KIMURA,T.2001.Progressionofossificationoftheposteriorlongitudinalligamentfollowingenbloccervicallaminoplasty.JBoneJointSurgAm,83‐A,1798‐802.

KAWASAKI,M.,TANI,T.,USHIDA,T.&ISHIDA,K.2007.Anterolisthesisandretrolisthesisofthecervicalspineincervicalspondyloticmyelopathyintheelderly.JOrthopSci,12,207‐13.

KELLGREN,J.H.&LAWRENCE,J.S.1963.Atlasofstandardradiographs:theepidemiologyofchronicrheumatism,Oxford,BlackwellScientific.

KIM,K.N.,AHN,P.G.,RYU,M.J.,SHIN,D.A.,YI,S.,YOONDO,H.&HA,Y.2014.Long‐termsurgicaloutcomesofcervicalmyelopathywithathetoidcerebralpalsy.EurSpineJ,23,1464‐71.

KIM,K.T.,KIM,D.H.,CHUNG,J.Y.,LEE,S.,JOO,J.,NAH,S.S.,SONG,H.Y.&KIM,H.J.2012.Associationofinterferongammapolymorphismwithossificationoftheposteriorlongitudinalligamentinthekoreanpopulation.ImmunolInvest,41,876‐87.

KIM,T.J.,BAE,K.W.,UHM,W.S.,KIM,T.H.,JOO,K.B.&JUN,J.B.2008.Prevalenceofossificationoftheposteriorlongitudinalligamentofthecervicalspine.JointBoneSpine,75,471‐4.

KING,J.A.,VACHHRAJANI,S.,DRAKE,J.M.&RUTKA,J.T.2009.Neurosurgicalimplicationsofachondroplasia.JNeurosurgPediatr,4,297‐306.

KISHIYA,M.,SAWADA,T.,KANEMARU,K.,KUDO,H.,NUMASAWA,T.,YOKOYAMA,T.,TANAKA,S.,MOTOMURA,S.,UEYAMA,K.,HARATA,S.,TOH,S.&

135

FURUKAWA,K.2008.AfunctionalRNAiscreenforRunx2‐regulatedgenesassociatedwithectopicboneformationinhumanspinalligaments.JPharmacolSci,106,404‐14.

KLIRONOMOS,G.,KARADIMAS,S.,MAVRAKIS,A.,MIRILAS,P.,SAVVAS,I.,PAPADAKI,E.,PAPACHRISTOU,D.J.&GATZOUNIS,G.2011.Newexperimentalrabbitanimalmodelforcervicalspondyloticmyelopathy.SpinalCord,49,1097‐102.

KOBASHI,G.,WASHIO,M.,OKAMOTO,K.,SASAKI,S.,YOKOYAMA,T.,MIYAKE,Y.,SAKAMOTO,N.,OHTA,K.,INABA,Y.,TANAKA,H.&JAPANCOLLABORATIVEEPIDEMIOLOGICALSTUDYGROUPFOREVALUATIONOFOSSIFICATIONOFTHEPOSTERIORLONGITUDINALLIGAMENTOFTHESPINE,R.2004.Highbodymassindexafterage20anddiabetesmellitusareindependentriskfactorsforossificationoftheposteriorlongitudinalligamentofthespineinJapanesesubjects:acase‐controlstudyinmultiplehospitals.Spine(PhilaPa1976),29,1006‐10.

KONG,Q.,MA,X.,LI,F.,GUO,Z.,QI,Q.,LI,W.,YUAN,H.,WANG,Z.&CHEN,Z.2007.COL6A1polymorphismsassociatedwithossificationoftheligamentumflavumandossificationoftheposteriorlongitudinalligament.Spine(PhilaPa1976),32,2834‐8.

KRAUSS,J.K.,LOHER,T.J.,POHLE,T.,WEBER,S.,TAUB,E.,BARLOCHER,C.B.&BURGUNDER,J.M.2002.Pallidaldeepbrainstimulationinpatientswithcervicaldystoniaandseverecervicaldyskinesiaswithcervicalmyelopathy.JNeurolNeurosurgPsychiatry,72,249‐56.

KUDO,H.,FURUKAWA,K.,YOKOYAMA,T.,ONO,A.,NUMASAWA,T.,WADA,K.,TANAKA,S.,ASARI,T.,UEYAMA,K.,MOTOMURA,S.&TOH,S.2011.Geneticdifferencesintheosteogenicdifferentiationpotencyaccordingtotheclassificationofossificationoftheposteriorlongitudinalligamentofthecervicalspine.Spine(PhilaPa1976),36,951‐7.

LAD,S.P.,PATIL,C.G.,BERTA,S.,SANTARELLI,J.G.,HO,C.&BOAKYE,M.2009.Nationaltrendsinspinalfusionforcervicalspondyloticmyelopathy.SurgNeurol,71,66‐9;discussion69.

LANDAU,D.A.,CHAPNICK,L.,YOFFE,N.,AZARIA,B.,GOLDSTEIN,L.&ATAR,E.2006.CervicalandlumbarMRIfindingsinaviatorsasafunctionofaircrafttype.AviatSpaceEnvironMed,77,1158‐61.

LAWRENCE,B.D.,HILIBRAND,A.S.,BRODT,E.D.,DETTORI,J.R.&BRODKE,D.S.2012.Predictingtheriskofadjacentsegmentpathologyinthecervicalspine:asystematicreview.Spine(PhilaPa1976),37,S52‐64.

LAWRENCE,B.D.,JACOBS,W.B.,NORVELL,D.C.,HERMSMEYER,J.T.,CHAPMAN,J.R.&BRODKE,D.S.2013a.Anteriorversusposteriorapproachfortreatmentofcervicalspondyloticmyelopathy:asystematicreview.Spine(PhilaPa1976),38,S173‐82.

LAWRENCE,B.D.,SHAMJI,M.F.,TRAYNELIS,V.C.,YOON,S.T.,RHEE,J.M.,CHAPMAN,J.R.,BRODKE,D.S.&FEHLINGS,M.G.2013b.Surgicalmanagementofdegenerativecervicalmyelopathy:aconsensusstatement.Spine(PhilaPa1976),38,S171‐2.

136

LEE,M.J.,CASSINELLI,E.H.&RIEW,K.D.2007.Prevalenceofcervicalspinestenosis.Anatomicstudyincadavers.JBoneJointSurgAm,89,376‐80.

LI,F.,CHEN,Z.,ZHANG,F.,SHEN,H.&HOU,T.2011.Ameta‐analysisshowingthathighsignalintensityonT2‐weightedMRIisassociatedwithpoorprognosisforpatientswithcervicalspondyloticmyelopathy.JClinNeurosci,18,1592‐5.

LIU,Y.,ZHAO,Y.,CHEN,Y.,SHI,G.&YUAN,W.2010.RUNX2polymorphismsassociatedwithOPLLandOLFintheHanpopulation.ClinOrthopRelatRes,468,3333‐41.

MATSUMOTO,M.,FUJIMURA,Y.,SUZUKI,N.,NISHI,Y.,NAKAMURA,M.,YABE,Y.&SHIGA,H.1998.MRIofcervicalintervertebraldiscsinasymptomaticsubjects.JBoneJointSurgBr,80,19‐24.

MATSUMOTO,M.,OKADA,E.,ICHIHARA,D.,CHIBA,K.,TOYAMA,Y.,FUJIWARA,H.,MOMOSHIMA,S.,NISHIWAKI,Y.,HASHIMOTO,T.,INOUE,T.,WATANABE,M.&TAKAHATA,T.2010.Prospectiveten‐yearfollow‐upstudycomparingpatientswithwhiplash‐associateddisordersandasymptomaticsubjectsusingmagneticresonanceimaging.Spine(PhilaPa1976),35,1684‐90.

MATSUNAGA,S.,KOGA,H.,KAWABATA,N.,KAWAMURA,I.,OTUSJI,M.,IMAKIIRE,T.&KOMIYA,S.2008a.Ossificationoftheposteriorlongitudinalligamentindizygotictwinswithschizophrenia:acasereport.ModRheumatol,18,277‐80.

MATSUNAGA,S.,KUKITA,M.,HAYASHI,K.,SHINKURA,R.,KORIYAMA,C.,SAKOU,T.&KOMIYA,S.2002.Pathogenesisofmyelopathyinpatientswithossificationoftheposteriorlongitudinalligament.JNeurosurg,96,168‐72.

MATSUNAGA,S.,NAKAMURA,K.,SEICHI,A.,YOKOYAMA,T.,TOH,S.,ICHIMURA,S.,SATOMI,K.,ENDO,K.,YAMAMOTO,K.,KATO,Y.,ITO,T.,TOKUHASHI,Y.,UCHIDA,K.,BABA,H.,KAWAHARA,N.,TOMITA,K.,MATSUYAMA,Y.,ISHIGURO,N.,IWASAKI,M.,YOSHIKAWA,H.,YONENOBU,K.,KAWAKAMI,M.,YOSHIDA,M.,INOUE,S.,TANI,T.,KANEKO,K.,TAGUCHI,T.,IMAKIIRE,T.&KOMIYA,S.2008b.Radiographicpredictorsforthedevelopmentofmyelopathyinpatientswithossificationoftheposteriorlongitudinalligament:amulticentercohortstudy.Spine(PhilaPa1976),33,2648‐50.

MATSUNAGA,S.&SAKOU,T.2012.Ossificationoftheposteriorlongitudinalligamentofthecervicalspine:etiologyandnaturalhistory.Spine(PhilaPa1976),37,E309‐14.

MATSUNAGA,S.,SAKOU,T.,TAKETOMI,E.&KOMIYA,S.2004.Clinicalcourseofpatientswithossificationoftheposteriorlongitudinalligament:aminimum10‐yearcohortstudy.JNeurosurg,100,245‐8.

MCKAY,S.D.,AL‐OMARI,A.,TOMLINSON,L.A.&DORMANS,J.P.2012.Reviewofcervicalspineanomaliesingeneticsyndromes.Spine(PhilaPa1976),37,E269‐77.

MEHALIC,T.F.,PEZZUTI,R.T.&APPLEBAUM,B.I.1990.Magneticresonanceimagingandcervicalspondyloticmyelopathy.Neurosurgery,26,217‐26discussion226‐7.

MEYER,C.A.,VAGAL,A.S.&SEAMAN,D.2011.Putyourbackintoit:pathologicconditionsofthespineatchestCT.Radiographics,31,1425‐41.

137

MIYASAKA,K.,KANEDA,K.,SATO,S.,IWASAKI,Y.,ABE,S.,TAKEI,H.,TSURU,M.,TASHIRO,K.,ABE,H.&FUJIOKA,Y.1983.Myelopathyduetoossificationorcalcificationoftheligamentumflavum:radiologicandhistologicevaluations.AJNRAmJNeuroradiol,4,629‐32.

MIYAZAKI,T.,SUDO,H.,HIRATSUKA,S.&IWASAKI,N.2014.Cervicalspondyloticmyelopathywithsubacutecombineddegeneration.SpineJ,14,381‐2.

MIZUNO,J.&NAKAGAWA,H.2006.Ossifiedposteriorlongitudinalligament:managementstrategiesandoutcomes.SpineJ,6,282S‐288S.

MIZUNO,J.,NAKAGAWA,H.&HASHIZUME,Y.2001.Analysisofhypertrophyoftheposteriorlongitudinalligamentofthecervicalspine,onthebasisofclinicalandexperimentalstudies.Neurosurgery,49,1091.

MOHANTY,C.,MASSICOTTE,E.M.,FEHLINGS,M.G.&SHAMJI,M.F.2015.Associationofpreoperativecervicalspinealignmentwithspinalcordmagneticresonanceimaginghyperintensityandmyelopathyseverity:analysisofaseriesof124cases.Spine(PhilaPa1976),40,11‐6.

MOON,E.S.,KARADIMAS,S.K.,YU,W.R.,AUSTIN,J.W.&FEHLINGS,M.G.2014.Riluzoleattenuatesneuropathicpainandenhancesfunctionalrecoveryinarodentmodelofcervicalspondyloticmyelopathy.NeurobiolDis,62,394‐406.

MOORE,K.,DALLEYAF,AGURAMR.2009.ClinicallyOrientatedAnatomy6thed,Philadelphia,USA,LippincottWilliams&Wilkin.

MUNDT,D.J.,KELSEY,J.L.,GOLDEN,A.L.,PANJABI,M.M.,PASTIDES,H.,BERG,A.T.,SKLAR,J.&HOSEA,T.1993.Anepidemiologicstudyofsportsandweightliftingaspossibleriskfactorsforherniatedlumbarandcervicaldiscs.TheNortheastCollaborativeGrouponLowBackPain.AmJSportsMed,21,854‐60.

NAKAMURA,M.,FUJIYOSHI,K.,TSUJI,O.,KONOMI,T.,HOSOGANE,N.,WATANABE,K.,TSUJI,T.,ISHII,K.,MOMOSHIMA,S.,TOYAMA,Y.,CHIBA,K.&MATSUMOTO,M.2012.Clinicalsignificanceofdiffusiontensortractographyasapredictoroffunctionalrecoveryafterlaminoplastyinpatientswithcervicalcompressivemyelopathy.JNeurosurgSpine,17,147‐52.

NEW,P.W.,CRIPPS,R.A.&BONNELEE,B.2014.Globalmapsofnon‐traumaticspinalcordinjuryepidemiology:towardsalivingdatarepository.SpinalCord,52,97‐109.

NORTHOVER,J.R.,WILD,J.B.,BRAYBROOKE,J.&BLANCO,J.2012.Theepidemiologyofcervicalspondyloticmyelopathy.SkeletalRadiol,41,1543‐6.

NOURI,A.,TETREAULT,L.,ZAMORANO,J.,MOHANTY,C.&FEHLINGS,M.2015.PrevalenceofKlippel‐FeilSyndromeinaSurgicalSeriesofPatientswithCervicalSpondyloticMyelopathy:AnalysisoftheProspective,MulticentreAOSpineNorthAmericaStudy.GlobalSpineJ.

NOVELLINE,R.A.2004.Squire'sFundamentalsofRadiology,Cambridge,Massachusetts,USA,HarvardUniversityPress.

NURICK,S.1972.Thepathogenesisofthespinalcorddisorderassociatedwithcervicalspondylosis.Brain,95,87‐100.

NYLAND,J.&JOHNSON,D.2004.CollegiateFootballPlayersDisplayMoreActiveCervicalSpineMobilityThanHighSchoolFootballPlayers.JAthlTrain,39,146‐150.

138

OGATA,N.,KOSHIZUKA,Y.,MIURA,T.,IWASAKI,M.,HOSOI,T.,SHIRAKI,M.,SEICHI,A.,NAKAMURA,K.&KAWAGUCHI,H.2002.Associationofbonemetabolismregulatoryfactorgenepolymorphismswithsusceptibilitytoossificationoftheposteriorlongitudinalligamentofthespineanditsseverity.Spine(PhilaPa1976),27,1765‐71.

OHISHI,H.,FURUKAWA,K.,IWASAKI,K.,UEYAMA,K.,OKADA,A.,MOTOMURA,S.,HARATA,S.&TOH,S.2003.RoleofprostaglandinI2inthegeneexpressioninducedbymechanicalstressinspinalligamentcellsderivedfrompatientswithossificationoftheposteriorlongitudinalligament.JPharmacolExpTher,305,818‐24.

OHTSUKA,K.,TERAYAMA,K.,YANAGIHARA,M.,WADA,K.,KASUGA,K.,MACHIDA,T.&MATSUSHIMA,S.1987.Aradiologicalpopulationstudyontheossificationoftheposteriorlongitudinalligamentinthespine.ArchOrthopTraumaSurg,106,89‐93.

OLIVE,P.M.,WHITECLOUD,T.S.,3RD&BENNETT,J.T.1988.LowercervicalspondylosisandmyelopathyinadultswithDown'ssyndrome.Spine(PhilaPa1976),13,781‐4.

ONO,K.,YONENOBU,K.,MIYAMOTO,S.&OKADA,K.1999.Pathologyofossificationoftheposteriorlongitudinalligamentandligamentumflavum.ClinOrthopRelatRes,18‐26.

OTREMSKI,I.,MARSH,J.L.,WILDE,B.R.,MCLARDYSMITH,P.D.&NEWMAN,R.J.1989.Softtissuecervicalspinalinjuriesinmotorvehicleaccidents.Injury,20,349‐51.

PALMER,P.E.,STADALNICK,R.&ARNON,S.1984.Thegeneticfactorincervicalspondylosis.SkeletalRadiol,11,178‐82.

PARK,J.Y.,CHIN,D.K.,KIM,K.S.&CHO,Y.E.2008.Thoracicligamentossificationinpatientswithcervicalossificationoftheposteriorlongitudinalligaments:tandemossificationinthecervicalandthoracicspine.Spine(PhilaPa1976),33,E407‐10.

PATEL,A.A.,SPIKER,W.R.,DAUBS,M.,BRODKE,D.S.&CANNON‐ALBRIGHT,L.A.2012.Evidenceofaninheritedpredispositionforcervicalspondyloticmyelopathy.Spine(PhilaPa1976),37,26‐9.

PAVLOV,H.,TORG,J.S.,ROBIE,B.&JAHRE,C.1987.Cervicalspinalstenosis:determinationwithvertebralbodyratiomethod.Radiology,164,771‐5.

PETREN‐MALLMIN,M.&LINDER,J.1999.MRIcervicalspinefindingsinasymptomaticfighterpilots.AviatSpaceEnvironMed,70,1183‐8.

PETREN‐MALLMIN,M.&LINDER,J.2001.Cervicalspinedegenerationinfighterpilotsandcontrols:a5‐yrfollow‐upstudy.AviatSpaceEnvironMed,72,443‐6.

PETTERSSON,K.,HILDINGSSON,C.,TOOLANEN,G.,FAGERLUND,M.&BJORNEBRINK,J.1997.Discpathologyafterwhiplashinjury.Aprospectivemagneticresonanceimagingandclinicalinvestigation.Spine(PhilaPa1976),22,283‐7;discussion288.

PETTERSSON,K.,KARRHOLM,J.,TOOLANEN,G.&HILDINGSSON,C.1995.Decreasedwidthofthespinalcanalinpatientswithchronicsymptomsafterwhiplashinjury.Spine(PhilaPa1976),20,1664‐7.

139

PIZZUTILLO,P.D.,WOODS,M.,NICHOLSON,L.&MACEWEN,G.D.1994.RiskfactorsinKlippel‐Feilsyndrome.Spine(PhilaPa1976),19,2110‐6.

RAFF,M.L.&BYERS,P.H.1996.Jointhypermobilitysyndromes.CurrOpinRheumatol,8,459‐66.

RHEE,J.M.,SHAMJI,M.F.,ERWIN,W.M.,BRANSFORD,R.J.,YOON,S.T.,SMITH,J.S.,KIM,H.J.,ELY,C.G.,DETTORI,J.R.,PATEL,A.A.&KALSI‐RYAN,S.2013.Nonoperativemanagementofcervicalmyelopathy:asystematicreview.Spine(PhilaPa1976),38,S55‐67.

RONEN,J.,GOLDIN,D.,BLUVSHTEIN,V.,FISHEL,B.,GELERNTER,I.&CATZ,A.2004.SurvivalafternontraumaticspinalcordlesionsinIsrael.ArchPhysMedRehabil,85,1499‐502.

ROSS,M.H.,KAYE,G.I.&PAWLINA,W.2003.Histology:ATextandAtlas:WithCellandMolecularBiology,LippincottWilliams&Wilkins.

ROSTI,R.O.2013.Ofmice,men,andKingTut:autosomalrecessiveKlippel‐FeilsyndromeiscausedbymutationsinMEOX1.ClinGenet,84,19.

RUANGCHAINIKOM,M.,DAUBS,M.D.,SUZUKI,A.,HAYASHI,T.,WEINTRAUB,G.,LEE,C.J.,INOUE,H.,TIAN,H.,AGHDASI,B.,SCOTT,T.P.,PHAN,K.H.,CHOTIVICHIT,A.&WANG,J.C.2014.Effectofcervicalkyphoticdeformitytypeonthemotioncharacteristicsanddynamicspinalcordcompression.Spine(PhilaPa1976),39,932‐8.

RUIZSANTIAGO,F.,ALCAZARROMERO,P.P.,LOPEZMACHADO,E.&GARCIAESPONA,M.A.1997.Calcificationoflumbarligamentumflavumandfacetjointscapsule.Spine(PhilaPa1976),22,1730‐4;discussion1734‐5.

RUNGE,V.M.,NITZ,W.R.,TRELLES,M.&GOERNER,F.L.2014.ThePhysicsofClinicalMRTaughtThroughImages,NewYork,USA,Thieme.

SAETIA,K.,CHO,D.,LEE,S.,KIM,D.H.&KIM,S.D.2011.Ossificationoftheposteriorlongitudinalligament:areview.NeurosurgFocus,30,E1.

SAMARTZIS,D.D.,HERMAN,J.,LUBICKY,J.P.&SHEN,F.H.2006.ClassificationofcongenitallyfusedcervicalpatternsinKlippel‐Feilpatients:epidemiologyandroleinthedevelopmentofcervicalspine‐relatedsymptoms.Spine(PhilaPa1976),31,E798‐804.

SAMPATH,P.,BENDEBBA,M.,DAVIS,J.D.&DUCKER,T.B.2000.Outcomeofpatientstreatedforcervicalmyelopathy.Aprospective,multicenterstudywithindependentclinicalreview.Spine(PhilaPa1976),25,670‐6.

SAYSON,J.V.&HARGENS,A.R.2008.Pathophysiologyoflowbackpainduringexposuretomicrogravity.AviatSpaceEnvironMed,79,365‐73.

SCHALL,D.G.1989.Non‐ejectioncervicalspineinjuriesdueto+Gzinhighperformanceaircraft.AviatSpaceEnvironMed,60,445‐56.

SCHER,A.T.1990.Prematureonsetofdegenerativediseaseofthecervicalspineinrugbyplayers.SAfrMedJ,77,557‐8.

SETZER,M.,HERMANN,E.,SEIFERT,V.&MARQUARDT,G.2008.ApolipoproteinEgenepolymorphismandtheriskofcervicalmyelopathyinpatientswithchronicspinalcordcompression.Spine(PhilaPa1976),33,497‐502.

SHAMJI,M.F.,MASSICOTTE,E.M.,TRAYNELIS,V.C.,NORVELL,D.C.,HERMSMEYER,J.T.&FEHLINGS,M.G.2013.Comparisonofanteriorsurgicaloptionsforthe

140

treatmentofmultilevelcervicalspondyloticmyelopathy:asystematicreview.Spine(PhilaPa1976),38,S195‐209.

SHIN,M.,BESSER,L.M.,KUCIK,J.E.,LU,C.,SIFFEL,C.,CORREA,A.,CONGENITALANOMALYMULTISTATE,P.&SURVIVAL,C.2009.PrevalenceofDownsyndromeamongchildrenandadolescentsin10regionsoftheUnitedStates.Pediatrics,124,1565‐71.

SILVER,J.R.2002.Theimpactofthe21stcenturyonrugbyinjuries.SpinalCord,40,552‐9.

SINGH,A.,TETREAULT,L.,FEHLINGS,M.G.,FISCHER,D.J.&SKELLY,A.C.2012.Riskfactorsfordevelopmentofcervicalspondyloticmyelopathy:resultsofasystematicreview.EvidBasedSpineCareJ,3,35‐42.

SINGH,A.,TETREAULT,L.,KALSI‐RYAN,S.,NOURI,A.&FEHLINGS,M.G.2014.Globalprevalenceandincidenceoftraumaticspinalcordinjury.ClinEpidemiol,6,309‐31.

SMITH,J.S.,LAFAGE,V.,RYAN,D.J.,SHAFFREY,C.I.,SCHWAB,F.J.,PATEL,A.A.,BRODKE,D.S.,ARNOLD,P.M.,RIEW,K.D.,TRAYNELIS,V.C.,RADCLIFF,K.,VACCARO,A.R.,FEHLINGS,M.G.&AMES,C.P.2013.Associationofmyelopathyscoreswithcervicalsagittalbalanceandnormalizedspinalcordvolume:analysisof56preoperativecasesfromtheAOSpineNorthAmericaMyelopathystudy.Spine(PhilaPa1976),38,S161‐70.

SMITH,Z.A.,BUCHANAN,C.C.,RAPHAEL,D.&KHOO,L.T.2011.Ossificationoftheposteriorlongitudinalligament:pathogenesis,management,andcurrentsurgicalapproaches.Areview.NeurosurgFocus,30,E10.

STAPLETON,C.J.,PHAM,M.H.,ATTENELLO,F.J.&HSIEH,P.C.2011.Ossificationoftheposteriorlongitudinalligament:geneticsandpathophysiology.NeurosurgFocus,30,E6.

STETLER,W.R.,LAMARCA,F.&PARK,P.2011.Thegeneticsofossificationoftheposteriorlongitudinalligament.NeurosurgFocus,30,E7.

SUZUKI,A.,DAUBS,M.D.,INOUE,H.,HAYASHI,T.,AGHDASI,B.,MONTGOMERY,S.R.,RUANGCHAINIKOM,M.,HU,X.,LEE,C.J.,WANG,C.J.,WANG,B.J.&NAKAMURA,H.2013.Prevalenceandmotioncharacteristicsofdegenerativecervicalspondylolisthesisinthesymptomaticadult.Spine(PhilaPa1976),38,E1115‐20.

TAITZ,C.1999.OsteophytosisofthecervicalspineinSouthAfricanblacksandwhites.ClinAnat,12,103‐9.

TAKEUCHI,Y.,MATSUMOTO,T.,TAKUWA,Y.,HOSHINO,Y.,KUROKAWA,T.,SHIBUYA,N.&OGATA,E.1989.Highincidenceofobesityandelevatedserumimmunoreactiveinsulinlevelinpatientswithparavertebralligamentousossification:Arelationshiptothedevelopmentofectopicossification.JBoneMinerMetab,7,17‐21.

TASSABEHJI,M.,FANG,Z.M.,HILTON,E.N.,MCGAUGHRAN,J.,ZHAO,Z.,DEBOCK,C.E.,HOWARD,E.,MALASS,M.,DONNAI,D.,DIWAN,A.,MANSON,F.D.,MURRELL,D.&CLARKE,R.A.2008.MutationsinGDF6areassociatedwithvertebralsegmentationdefectsinKlippel‐Feilsyndrome.HumMutat,29,1017‐27.

141

TECHY,F.&BENZEL,E.C.2014.Predictorsofoutcomeinpatientswithcervicalspondyloticmyelopathyundergoingsurgicaltreatment:theevidenceandtheinternationalcommonpractice.WorldNeurosurg,81,503‐7.

TETREAULT,L.A.,DETTORI,J.R.,WILSON,J.R.,SINGH,A.,NOURI,A.,FEHLINGS,M.G.,BRODT,E.D.&JACOBS,W.B.2013a.Systematicreviewofmagneticresonanceimagingcharacteristicsthataffecttreatmentdecisionmakingandpredictclinicaloutcomeinpatientswithcervicalspondyloticmyelopathy.Spine(PhilaPa1976),38,S89‐110.

TETREAULT,L.A.,KARPOVA,A.&FEHLINGS,M.G.2013b.Predictorsofoutcomeinpatientswithdegenerativecervicalspondyloticmyelopathyundergoingsurgicaltreatment:resultsofasystematicreview.EurSpineJ.

TETREAULT,L.A.,KOPJAR,B.,VACCARO,A.,YOON,S.T.,ARNOLD,P.M.,MASSICOTTE,E.M.&FEHLINGS,M.G.2013c.Aclinicalpredictionmodeltodetermineoutcomesinpatientswithcervicalspondyloticmyelopathyundergoingsurgicaltreatment:datafromtheprospective,multi‐centerAOSpineNorthAmericastudy.JBoneJointSurgAm,95,1659‐66.

TETREAULT,L.A.,NOURI,A.,SINGH,A.,FAWCETT,M.&FEHLINGS,M.G.2014.Predictorsofoutcomeinpatientswithcervicalspondyloticmyelopathyundergoingsurgicaltreatment:asurveyofmembersfromAOSpineInternational.WorldNeurosurg,81,623‐33.

TORG,J.S.,PAVLOV,H.,GENUARIO,S.E.,SENNETT,B.,WISNESKI,R.J.,ROBIE,B.H.&JAHRE,C.1986.Neurapraxiaofthecervicalspinalcordwithtransientquadriplegia.JBoneJointSurgAm,68,1354‐70.

TRACY,J.A.&BARTLESON,J.D.2010.Cervicalspondyloticmyelopathy.Neurologist,16,176‐87.

TROJAN,D.A.,POUCHOT,J.,POKRUPA,R.,FORD,R.M.,ADAMSBAUM,C.,HILL,R.O.&ESDAILE,J.M.1992.Diagnosisandtreatmentofossificationoftheposteriorlongitudinalligamentofthespine:reportofeightcasesandliteraturereview.AmJMed,92,296‐306.

TSUYAMA,N.1984.Ossificationoftheposteriorlongitudinalligamentofthespine.ClinOrthopRelatRes,71‐84.

VEDANTAM,A.&RAJSHEKHAR,V.2013.DoesthetypeofT2‐weightedhyperintensityinfluencesurgicaloutcomeinpatientswithcervicalspondyloticmyelopathy?Areview.EurSpineJ,22,96‐106.

VILLAVICENCIO,A.T.,HERNANDEZ,T.D.,BURNEIKIENE,S.&THRAMANN,J.2007.Neckpaininmultisportathletes.JNeurosurgSpine,7,408‐13.

WANG,L.F.,ZHANG,Y.Z.,SHEN,Y.,SU,Y.L.,XU,J.X.,DING,W.Y.&ZHANG,Y.H.2010a.UsingtheT2‐weightedmagneticresonanceimagingsignalintensityratioandclinicalmanifestationstoassesstheprognosisofpatientswithcervicalossificationoftheposteriorlongitudinalligament.JNeurosurgSpine,13,319‐23.

WANG,Z.C.,CHEN,X.S.,WANGDA,W.,SHI,J.G.,JIA,L.S.,XU,G.H.,HUANG,J.H.&FAN,L.2010b.ThegeneticassociationofvitaminDreceptorpolymorphismsandcervicalspondyloticmyelopathyinChinesesubjects.ClinChimActa,411,794‐7.

142

WANG,Z.C.,SHI,J.G.,CHEN,X.S.,XU,G.H.,LI,L.J.&JIA,L.S.2012.TheroleofsmokingstatusandcollagenIXpolymorphismsinthesusceptibilitytocervicalspondyloticmyelopathy.GenetMolRes,11,1238‐44.

WANI,A.A.,LAHERWAL,M.,RAMZAN,A.U.,MALIK,N.K.,ARIF,S.&ALAM,S.2012.CompressiveMyelopathyDuetoOssifiedLigamentumFlavum.NeurosurgeryQuarterly,22,119‐12210.1097/WNQ.0b013e31823cc336.

WAXMAN,S.2009.ClinicalNeuroanatomy,26thEd,USA,McGraw‐HillMedical.WESTBROOK,C.,ROTH,C.&TALBOT,J.2011.MRIinPractice,UK,Wiley‐Blackwell.WILDER,F.V.,FAHLMAN,L.&DONNELLY,R.2011.Radiographiccervicalspine

osteoarthritisprogressionrates:alongitudinalassessment.RheumatolInt,31,45‐8.

WILSON,J.R.,BARRY,S.,FISCHER,D.J.,SKELLY,A.C.,ARNOLD,P.M.,RIEW,K.D.,SHAFFREY,C.I.,TRAYNELIS,V.C.&FEHLINGS,M.G.2013a.Frequency,timing,andpredictorsofneurologicaldysfunctioninthenonmyelopathicpatientwithcervicalspinalcordcompression,canalstenosis,and/orossificationoftheposteriorlongitudinalligament.Spine(PhilaPa1976),38,S37‐54.

WILSON,J.R.,FORGIONE,N.&FEHLINGS,M.G.2013b.Emergingtherapiesforacutetraumaticspinalcordinjury.CMAJ,185,485‐92.

WILSON,J.R.,PATEL,A.A.,BRODT,E.D.,DETTORI,J.R.,BRODKE,D.S.&FEHLINGS,M.G.2013c.Geneticsandheritabilityofcervicalspondyloticmyelopathyandossificationoftheposteriorlongitudinalligament:resultsofasystematicreview.Spine(PhilaPa1976),38,S123‐46.

WONG,A.S.,MASSICOTTE,E.M.&FEHLINGS,M.G.2005.Surgicaltreatmentofcervicalmyeloradiculopathyassociatedwithmovementdisorders:indications,technique,andclinicaloutcome.JSpinalDisordTech,18Suppl,S107‐14.

WU,J.C.,KO,C.C.,YEN,Y.S.,HUANG,W.C.,CHEN,Y.C.,LIU,L.,TU,T.H.,LO,S.S.&CHENG,H.2013.Epidemiologyofcervicalspondyloticmyelopathyanditsriskofcausingspinalcordinjury:anationalcohortstudy.NeurosurgFocus,35,E10.

WU,J.C.,LIU,L.,CHEN,Y.C.,HUANG,W.C.,CHEN,T.J.&CHENG,H.2011.Ossificationoftheposteriorlongitudinalligamentinthecervicalspine:an11‐yearcomprehensivenationalepidemiologystudy.NeurosurgFocus,30,E5.

XU,Y.,CHEN,W.&JIANG,J.2013.CervicalspondyloticmyelopathywithvitaminBdeficiency:Twocasereports.ExpTherMed,6,943‐946.

YOO,D.S.,LEE,S.B.,HUH,P.W.,KANG,S.G.&CHO,K.S.2010.Spinalcordinjuryincervicalspinalstenosisbyminortrauma.WorldNeurosurg,73,50‐2;discussione4.

YOON,S.T.,HASHIMOTO,R.E.,RAICH,A.,SHAFFREY,C.I.,RHEE,J.M.&RIEW,K.D.2013.Outcomesafterlaminoplastycomparedwithlaminectomyandfusioninpatientswithcervicalmyelopathy:asystematicreview.Spine(PhilaPa1976),38,S183‐94.

YOON,V.,MAALOUF,N.M.&SAKHAEE,K.2012.Theeffectsofsmokingonbonemetabolism.OsteoporosInt,23,2081‐92.

143

YOSHIDA,M.,SHIMA,K.,TANIGUCHI,Y.,TAMAKI,T.&TANAKA,T.1992.Hypertrophiedligamentumflavuminlumbarspinalcanalstenosis.Pathogenesisandmorphologicandimmunohistochemicalobservation.Spine(PhilaPa1976),17,1353‐60.

YOSHIMATSU,H.,NAGATA,K.,GOTO,H.,SONODA,K.,ANDO,N.,IMOTO,H.,MASHIMA,T.&TAKAMIYA,Y.2001.Conservativetreatmentforcervicalspondyloticmyelopathy.predictionoftreatmenteffectsbymultivariateanalysis.SpineJ,1,269‐73.

YOSHIZAWA,T.,TAKIZAWA,F.,IIZAWA,F.,ISHIBASHI,O.,KAWASHIMA,H.,MATSUDA,A.,ENDO,N.&KAWASHIMA,H.2004.HomeoboxproteinMSX2actsasamoleculardefensemechanismforpreventingossificationinligamentfibroblasts.MolCellBiol,24,3460‐72.

YU,W.R.,BAPTISTE,D.C.,LIU,T.,ODROBINA,E.,STANISZ,G.J.&FEHLINGS,M.G.2009.Molecularmechanismsofspinalcorddysfunctionandcelldeathinthespinalhyperostoticmouse:implicationsforthepathophysiologyofhumancervicalspondyloticmyelopathy.NeurobiolDis,33,149‐63.

YU,W.R.,LIU,T.,KIEHL,T.R.&FEHLINGS,M.G.2011.HumanneuropathologicalandanimalmodelevidencesupportingaroleforFas‐mediatedapoptosisandinflammationincervicalspondyloticmyelopathy.Brain,134,1277‐92.

ZHANG,P.,SHEN,Y.,ZHANG,Y.Z.,DING,W.Y.&WANG,L.F.2011.SignificanceofincreasedsignalintensityonMRIinprognosisaftersurgicalinterventionforcervicalspondyloticmyelopathy.JClinNeurosci,18,1080‐3.

ZHANG,Y.Z.,SHEN,Y.,WANG,L.F.,DING,W.Y.,XU,J.X.&HE,J.2010.MagneticresonanceT2imagesignalintensityratioandclinicalmanifestationpredictprognosisaftersurgicalinterventionforcervicalspondyloticmyelopathy.Spine(PhilaPa1976),35,E396‐9.

ZHOU,S.‐W.,GUO,L.‐X.,ZHANG,S.‐Q.&TANG,C.‐Y.2010.StudyonCervicalSpineInjuriesinVehicleSideImpact.TheOpenMechanicalEngineeringJournal,4,29‐35.

144

APPENDIX