12ournal ofNeurology, Neurosurgery, and Psychiatry 1993;56:1072-1077

Chiari malformation in adults: relation ofmorphological aspects to clinical features andoperative outcome

JM Stevens, W A D Serva, B E Kendall, A R Valentine, J R Ponsford

AbstractTo determine whether clinical featuresattributed to cerebellar ectopia could berelated to the severity of the malforma-tion, and if morphological features couldbe related to operative outcome, a retro-spective study of 141 patients with theadult Chiari malformation was carriedout, 81 receiving operative treatment.Morphological parameters derived frompreoperative clinical imaging were com-pared with presenting clinical featuresand postoperative outcomes. Patientswith the most severe cerebellar malfor-mation, defined as descent of the cere-bellar tonsils to or below the axis, haddisabling ataxia and nystagmus morefrequently. Those with brainstem com-pression had limb weakness and musclewasting more frequently. Operative out-come was significantly less favourable inpatients with severe cerebellar ectopia(12% improved, 690% deteriorated) thanin those with minor ectopia (50%improved, 17% deteriorated). Patientswith a distended cervical syrinx had amore favourable outcome than thosewithout. Morphological features helppredict operative risk.

(jNeurol Neurosurg Psychiatry 1993;56:1072-1077)

The NationalHospitals forNeurology andNeurosurgery, QueenSquare and MaidaVale, London, UKJM StevensWA D ServaB E KendallA R ValentineJ R PonsfordCorrespondence to:Dr JM Stevens,Department of Radiology,Maida Vale Hospital,London W9 ITL.Received 22 October 1992and in revised form8 January 1993.Accepted 19 January 1993

The term adult Chiari malformation is usedto describe descent of the hind brain into thecervical canal, in which meningomyelocoele isabsent and hydrocephalus rare, and patientsusually present in adult life.' The clinical fea-tures which may be associated are protean26and a causal relationship in a given case maynot be established with certainty, especially as

modern computed imaging suggests that min-imal protrusion of parts of the cerebellarhemispheres through the foramen magnum isrelatively common, and that even markedprotusion is often asymptomatic.7 The issue isimportant because it is often maintained thatdisability due to the adult Chiari malforma-tion usually is curable by operation.' 5

In the nine operative series which haveincluded more than 20 patients and in whichthe mean follow up was at least 6 months, themean proportion and patients with sustainedimprovement has been 56% (range 29-82%),the proportion who deteriorated 24% (range6-31%) and the perioperative mortality 3%(range 0_5-6%).1 4 The clinical featuresmost likely to be alleviated have been

headache and neck pain, and those mostrefractory were often ataxia and nystagmus.

Clearly not all patients with the adultChiari malformation benefit from operation.Attempts at risk stratification have beenmade, and recently a careful study byMenezes' group14 established three preopera-tive clinical features predictive of a poor oper-ative outcome-namely, ataxia, scoliosis, andmuscle atrophy. Attempts at relating theseverity of the Chiari lesion to individual clin-ical features or syndromes have been few andsome of the conclusions conflicting.15-'7 Therehas been no formal attempt at operative riskstratification based on morphological fea-tures.The present study was begun several years

ago to investigate if the severity of the hindbrain abnormality could be related to preop-erative clinical features and postoperative out-come. Because the design of the investigationwas retrospective, and large numbers ofpatients with detailed clinical data wererequired, it was necessary to use data mainlyfrom myelography and CT rather than MRIwhich only became available at our institutionsome time after the study began. Nevertheless,most of the criteria used in the study areequally applicable to MRI.

Materials and methodsThe study consisted of a retrospective reviewof 141 symptomatic adult patients diagnosedas having a Chiari malformation by myelogra-phy alone in 91 or in combination with CT in50 patients. At least five of these patients alsohad MRI performed on a Vista MR (PickerInternational) operating at 05 or 0-26 T.One patient only had hydrocephalus, andnone had a meningomyelocoele. Eighty onepatients were operated on in one of three sur-gical units; all were subjected to a posteriorsuboccipital and upper cervical decompres-sion, 12 had partial excision of herniatedcerebellar tonsils, and eight had an occlusionof the obex. Thirty one patients also had asyringostomy. Sixty patients did not receiveoperations at our institution: 32 were lost tofollow up and may have been operated onelsewhere, 18 had significant co-existing dis-ease such as multiple sclerosis (10 patients)and severe spondylotic spinal cord compres-sion (eight patients), eight refused operation,and it was unclear why no operation was per-formed in two.

Clinical data were gleaned from the casenotes and evaluated by one of two neurolo-

1072 on N

ovember 2, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.56.10.1072 on 1 O

ctober 1993. Dow

nloaded from

Chiari malformation in adults: relation of morphological aspects to clinicalfeatures and operative outcome

Table 1 Clinical groupings.

(a) General1 Headache: considered only if it was exacerbated by head

movement, exercise, or coughing.2 Drop attacks: unexplained falls with or without brief

loss of consciousness.3 Neck pain: with or without occipital radiation.(b) Functional anatomical1 Upper and brainstem (midbrain, pons): dysfunction of

cranial nerves m, IV, V, VI, VII, and VIII.2 Medulla oblongata: dysfunction of cranial nerves IX, X,

XI, xII.3 Brainstem-cerebelar connections: ataxia, vertigo,

nystagmnus.4 Peripheral sensory disturbances: impaired pain and

temperature sensation; loss of position and vibratorysense; subjective complaints ofnumbness ofparasthesiae in limb or trunk.

5 Peripheral motor disturbances: all forms of weakness inlimbs or trunk, with or without muscle wasting.

6 Syringomyelia: this category included only thosepatients in which there was a firm diagnosis ofsyringomyelia in the case notes, based on clinical andradiological findings.

(c) Postoperative clinical categories1 Original clinicalfeatures: The preoperative symptoms/

signs were classified postoperatively as either improved,unchanged or worse.

2 New clinicalfeatures: any complaints or signs appearingafter operation which was not present before, excludingthose directly related to the wound itself.

3 Headache: this feature was considered separately whenpresent in the preoperative clinical picture.

gists (WS and JP) not involved in the routinemanagement of the patients. A large volumeof data was obtained which, for the purposesof analysis, was simplified by arranging it intothe categories and groups indicated in table 1.Missing data in any individual case were

recorded as such, and the case excluded fromanalysis of that feature. Data sets were judgedto be adequate for 127 patients. The postop-erative assessments were divided into earlyand late. All early assessments were madewithin 1 month of operation and featuresreferable to the surgical wound wereexcluded, such as neck pain and stiffness.The late assessments were made, as far as

possible, from the last outpatient attendance.This varied from 3 months to 9 years afteroperation; but in only 18 did this periodexceed 2 years and the mean follow up periodwas 1 2 years. The available non-operatedgroup was considered too small and heteroge-neous to be used for a meaningful parallelgroup follow up study.The radiological features were assessed by

three experienced neuroradiologists (aS, BK,AV).

CEREBELLAR TONSILSThe level of the cerebellar tonsils wasassessed with head and spine either slightlyextended or slightly flexed but never hyper-flexed. The distance the most inferior part ofone or both cerebellar hemispheresdescended below the foramen magnum wasmeasured in millimetres on all availableimages, and the level also was noted relativeto easily recognisable anatomical landmarks.Magnification was not adequately standard-ised between examination types, so only therelative levels, reasonably consistent betweenmyelography, computed myelography, andMRI were considered in the final analysis.Patients were grouped into three categories:A, tonsils between foramen magnum andarch of the atlas; B, tonsils between the

arches of atlas and axis; and C, tonsils belowthe axis. The upper border of each neuralarch was taken as the reference point for eachgroup.

FORAMEN MAGNUMThe degree of obstruction at the foramenmagnum was also assessed. Contrast mediumflowed freely into the head from the spinalsubarachnoid space via the anterior cere-brospinal fluid pathway on myelography in allcases, although in some it was obstructedposteriorly. Two types of observation weremade:

1 The size of the cisterna magna was assessedboth above and at the level of the foramenmagnum and its mid-sagittal size determinedby measuring the minimum distance betweenthe internal occipital crest and posterior mar-gin of the foramen magnum and the cerebel-lum. CT or midline sagittal tomograms andMRI were used wherever possible for thisassessment. Accurate measurement was notpossible, but the cisterna magna was scoredas obliterated if it was not visible, small if itsmid-sagittal width was less than 2-3 mm, andnormal if it was greater than 2-3 mm.

2 Contrast medium visible in the fourth ventri-cle either during the myelogram, or on CTwithin 1 hour of the myelogram providedconcentration was similar to that in the poste-rior fossa cisterns, was recorded. This wastaken as evidence that either one or each ofthe foramena of Magendie or of Lushka werepatent. Arachnoiditis was considered presentif the cisterna magna was obliterated and thefourth ventricle failed to opacify on computedmyelography.

SYRINGOMYELIAThe upper spinal cord was also assessed forthe presence of syringomyelia. Syringomyeliawas considered probable if there was a centralaccumulation of contrast medium within thespinal cord on early or delayed post-myelography CT, or a circumscribed centralsignal change was visible on MRI, or if therewas a diffuse abnormality of cord size notexplained by compression. A mid-sagittaldiameter of the spinal cord of greater than 10mm was regarded as definitely enlarged, andof less than 6 mm as definitely small.Measurements between were regarded asnormal.

MEDULLA OBLONGATACompression of the medulla oblongata wasalso considered. This was often difficult toassess on myelography, and only assessmentsmade on adequate computed cisternogramsand MRI were analysed. The anterior andposterior surfaces were considered separately,and compression was considered presentwhen the whole or part of the contour wasflattened or concave on axial images. Theposterior surface of the medulla was assessedbelow the obex because the floor of the fourthventricle usually is concave. The compressingagent also was recorded, and was either theapex of the dens, an abnormality of the clivus,or the cerebellar tonsils acting either alone or

1073 on N

ovember 2, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.56.10.1072 on 1 O

ctober 1993. Dow

nloaded from

Stevens, Serva, KendaU, Valentine, Ponsford

Table 2 Clinicalfeatures in 141 patients*

Symptoms* Signs*

Headache 41Neck pain 46Drop attacks 19 -

Pons-midbrain 27 44Medulla oblongata 49 61Brainstem/cerebellar 48 73Peripheral sensory 67 67Peripheral motor 33 92

*Patients often had more than one clinical feature andadequate data sets were available in only 127 patients.

Table 3 Postoperative clinical assessments

Early Late

n % n %

Original Features:Improved* 20 25 23 34Unchanged 31 38 19 28Worse 30 37 25 37Total 81 100 67 100

New features: 23/81 28 21/67 31Headache:

Improved 21 68 17 63Unchanged 1 13 3 11Worse 9 29 7 26Total 31 100 27 100

*This implies that one or more cardinal features showed defi-nite improvement; usually it was only one. No case appeared tohave improved in one direction and deteriorated in another.

in combination. Compression was gradedvisually as mild, moderate, or severe-severeindicating that the mid-sagittal diameter ofthe neural axis at the point of maximal com-pression was reduced by 50% or more.

All the clinical and radiological featureswere assigned a numerical code and the datawere analysed using the Statistical Packagefor the Social Sciences (SPSS) version 7 pro-gram, on the University of London ComputerCentre Amdahl Computer. The incidence ofeach clinical feature was compared with thepresence or absence of each radiological fea-ture, considered independently. In someinstances a relationship was sought betweenindividual radiological features, but the analy-sis was concerned mainly with relationship ofradiological to clinical features, especiallyoperative outcome. X2 tests of associationwere used, except when an expectedfrequency was less than five, in which case aFisher's exact test was used.

Table 4 Clinicalfeatures related to tonsillar descent

Level of tonsils*

(0-Cl) (Cl-C2) (>C2) p Value

Medulla (symptoms) 16/53 14/49 14/25 0-04(30%) (29%) (56%)

Peripheral motor (sign) 28/53 32/49 21/25 0-02(53%) (65%) (84%)

Brainstem/Cerebellar 19/53 26/49 18/25 0-01(signs) (36%) (53%) (72%)Peripheral sensory 29/53 27/49 6/24 0-03(symptoms) (55%) (53%) (25%)Neck pain 24/53 11/48 5/25 0-02

(47%) (23%) (20%)Syringomyelia 28/49 31/44 6/25 0-00

(57%) (70%) (20%)

*0-C 1, tonsils between occiput and the atlas ring; C 1-2, tonsils between the atlas and axis;>C2, tonsils below the axis.

ResultsThe five radiological variables were tested foran association with each of the 15 clinicalvariables which included both early and latepostoperative assessments. There were 75sets of contingency tables of which 17 showedsignificant associations (24%). In addition,the six postoperative clinical variables in theearly and late assessments were comparedwith each of the preoperative clinical vari-ables, to create a further 90 contingencytables, of which only four indicated signifi-cant associations (4%).The duration of symptoms was less than 5

years in 46% and more than 10 years in 26%.Patients' ages ranged from 11 to 78 years, but54% were aged between 32 and 45 years.The distribution of preoperative clinical fea-tures in the sample is shown in table 2,although it is noted that adequate sets wereavailable in only 127 patients. The outcomesin the 81 operated patients are indicated intable 3. However, adequate sets of clinicaland radiological data were available for theearly postoperative assessment in only 73,and in 60 for the late assessment. Headachewas included only if specifically mentioned inthe postoperative notes, so that only 31 caseshad valid early assessments, and only 27 validlate assessments. Sixty five patients had radi-ological evidence of syringomyelia (55%).Basilar invagination or other significant bonyabnormalities were recognised in 17 cases.

Associations between several clinical fea-tures and radiological parameters whichyielded probability values less than 0-05 areindicated in tables 4 and 5. In addition, anenlarged spinal cord was associated with thepresence of peripheral sensory disturbanceswhich were present in 73% of 63 valid casesin whom the cord was enlarged, and in only17-5% of 40 valid cases in whom the cordwas normal or small (p = 0 0001). It was notpossible to investigate the effect of a smallspinal cord as opposed to a normal cord,since the numbers recorded as having a smallcord were too small. Similarly there was astrong positive association between abnormalcontrast accumulation in the cord on post-myelography CT and peripheral sensory dis-turbance (21 of 28 valid cases with contrastin the cord had sensory disturbance com-pared with seven of 22 without, p = 0 002).Interestingly, syringomyelia showed noassociation with obliteration of the cisternamagna or opacification of the fourth ventricle(p = 0-130).A significant association was found

between operative outcome, and the level ofcerebellar tonsils (table 6). A further associa-tion was found between clinical improvementand the presence of an enlarged spinal cord

Table S Clinicalfeatures related to size of cisternamagna

Normal Small Obliterated p Value

Medulla oblongata 5/18 31/74 10/12 0-02(signs) (28%) (42%) (83%)

Brainstem-cerebellar 3/18 39/74 10/12 0 00(signs) (16%) (53%) (83%)

1074 on N

ovember 2, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.56.10.1072 on 1 O

ctober 1993. Dow

nloaded from

Chiari malformation in adults: relation of morphological aspects to clinicalfeatures and operative outcome

Table 6 Operative outcome with respect to level of tonsils

Level of tonsils

Preoperative (0-Cl) (Cl-C2) (>C2)clinical features n % n % n %

Early:Worse 5 18 14 47 1 1 69Unchanged 13 48 1 1 37 3 19Improved 9 33 5 17 2 12Total 27 100 30 100 16 100

Late:Worse 3 17 7 26 9 60Unchanged 6 33 7 26 4 27Improved 9 50 13 48 2 13Total 18 100 27 100 15 100

By combining the categories of unchanged and improved, p values were calculated as 0-02 forearly and 0 04 for late assessment.

on early (p = 0-01) but not on late (p =0 082) assessment (table 7).The question of arachnoiditis could not be

addressed directly in this study. The combi-nation of obliteration of the cistema magnaand non-filling of the fourth ventricle mayhave been due to arachnoiditis in many cases.Both were associated with the cerebellar ton-sils lying below C2, but only non-filling of thefourth ventricle showed a relationship withpoor operative outcome. When the fourthventricle was opacified, two of 15 patientswere worse after operation compared with 28of 50 when the fourth ventricle did notopacify (p = 0-001). The role of arachnoiditisas opposed to apparent level of the cerebellartonsils could not be established by this analy-sis.

Post-myelography CT studies all too fre-quently were of limited extent, and oftendelayed until the contrast density in the spinaland intracranial subarachnoid spaces wasmuch reduced, preventing detailed analysis.At the time the potential value of detailedassessment was not adequately appreciated, asituation which this study originally wasdesigned to redress. Therefore, only 28 casespermitted what we now regard as an adequateestimation of medullary compression. Themajority of attempted clinical associationswith compression failed to suggest any associ-ation but of the 21 who had some compres-sion 18 had peripheral motor signs (mainlyweakness and wasting), and only three ofseven without compression had such signs (p= 0 043, Fisher's exact test). Only 17 caseswere suitable for analysis of operative out-come, and although five patients with mildcompression were unchanged, whereas nineof 12 with moderate or severe compressionsuffered a deterioration in original signs andthree also developed new features, a Fisher'sexact test indicated a p value of 0-06 whichwas not significant. The major compressingagent was judged to be anterior in 18 (dens

Table 7 Operative outcome* with respect to spinal cordsize

Nornal EnlargedPreoperative clinicalfeatures (early) n % n %

Worse 10 59 8 23Unchanged/better 7 41 28 82Total 17 100 34 100

*Based on early assessment (p = 0-01).

17, clivus 1) and posterior in three (cerebellartonsils).

Analysis of other aspects of medullary mor-phology, such as length, level of obex anddorsal column nuclei, presence of spur orkink, were possible in only 16 cases withcomputed cisternograms and the five caseswho also had MRI. In seven there was a cer-vicomedullary spur just below the ectopiccerebellar tonsils over which the distal part ofthe fourth ventricle was prolonged, the obexpresumably lying under the spur. In all seventhe cerebellar tonsils were at or below C2(one was in group B, and six were in groupC). There were no independent clinical fea-tures distinguishing this small group ofpatients, though none had syringomyelia.As would be expected from previous find-

ings, poor operative outcome showed a rela-tionship with the presence of swallowingdifficulties, truncal ataxia, and nystagmus (p= 0-025 in each instance), all of which hadshown an association with marked tonsillardescent. Patient age and duration of clinicalsymptoms were considered, and no signifi-cant association was found with operativeoutcome on two sample t-tests. Mean ages inthe worse, unchanged, and improved groupswere 41-6, 41-1, and 32-0 years; and meanduration of symptoms in the same groupswere 5 6, 4-1, and 4 5 years.

DiscussionThe distribution of clinical features indicatedthat this sample of patients was similar to thatin other comparable studies, and the resultsof subocciptal decompression, when dueallowance was made for varying methods ofassessment, were generally compatible withthe consensus of published reports alreadycited. The mean follow up was relativelyshort, many other series having mean periodsof more than 2 years. The group of operatedpatients with mild cerebellar tonsillar descent(group A), who generally fared best, weremost frequently lost to follow up. Althoughthe proportion of positive associations wassimilar to that which might have beenexpected by chance, the positive findingswere broadly in accord with those of Elster etal"7 and Dyste et al14 and, where relevant,showed the interrelationships expected fromtrue as opposed to spurious associations. Forexample, marked tonsillar descent was associ-ated with the preoperative ataxia and pooroperative outcome, and poor operative out-come was associated with preoperative ataxia.The most frequent relationships involved

the level of the cerebellar tonsils. It has beenobserved at operation that sometimes the ton-sils are found at a lower level than suggestedby myelography.l9 Possibly the alteredhydrodynamics at the craniovertebral junc-tion caused by opening the dura explains atleast some such discrepancies, and hyperflex-ion of the head may explain others. MRI isnow the imaging method of choice and howestimates of tonsillar descent compared withthose from myelography in the same group ofpatients is unknown. The measurements

1075

on Novem

ber 2, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.56.10.1072 on 1 October 1993. D

ownloaded from

Stevens, Serva, Kendall, Valentine, Ponsford

made by Aboulezz et al7 on MRI comparedwith those made from myelography byO'Connor et a120 and Baker,21 suggest thatthey appear slightly lower on MRI.Observations from high definition computedcisternography have indicated that the medialparts of the biventral lobules are nearly alwaysinferior to the tonsils in normal cerebellarhemispheres and lie in the plane of the fora-men magnum posterolateral to the brainstem,very occasionally protruding slightly below.22Only when ectopic do the cerebellar tonsilsenter the foramen magnum or protrudebelow it, usually lying directly posterior to thebrainstem. Most MRI assessments are madefrom mid-sagittal images usually 3 to 5 mmthick in which this distinction often cannotreliably be made. It is even more difficult onmyelography unless thin section tomographyis used. In our cases, all that could reliably besaid was that some part of the cerebellarhemispheres was visible in the cervical part ofthe cisterna magna. In the cases with CT,protrusion of the actual cerebellar tonsilscould be confirmed and the relative degree ofprotrusion correlated reasonably on MRI andmyelographic assessment.Computed cisternography and MRI indi-

cate that patients with herniated cerebellartonsils often have an elongated and some-times even kinked medulla oblongata inwhich the obex lies below the foramen mag-num,2325 as was the case in at least seven ofour patients and probably also in many more.In many publications such a configurationwould be classified as Chiari type 1.f 13-17 24The association between increasing tonsillardescent and obliteration or smallness of thecisterna magna probably accounted for thelow incidence of filling of the fourth ventriclein group C. However, it was less clear whymore severe tonsillar descent should havebeen associated with a significantly low inci-dence of syringomelia. The explanation mayreside in the significance of abnormalities ofthe medulla oblongata such as the kink andspur, the incidence of which is increased inpatients with more severe malformations.25Such morphology is associated with a lowerincidence of syringomyelia, possibly becausethe obex lies below the obstructing tonsils.219The level of the cerebellar tonsils showed a

similar association with presenting clinicalfeatures to those found in a recent MRIstudy. 17 Features referable to the medullaoblongata were mainly difficulties with swal-lowing and phonation, and those referable tobrainstem cerebellar connections were mainlyataxia and nystagmus. These, together withperipheral motor system signs, appearedsignificant because of their relatively highfrequency in patients with severe tonsillardescent (group C). The relationship withperipheral sensory symptoms was mainly dueto the high frequency in patients with moder-ate tonsillar descent (group B), and there wasa close relationship between peripheral sen-sory disturbances and radiological signs ofsyringomyelia. It was no surprise perhaps tohave found that neck pain appeared a signifi-cant feature in the presentation of mild but

not severe malformations, the latter beingassociated with more disabling clinical mani-festations.

In this study, important associations alsowere found between tonsillar descent andoperative outcome. Unfavourable outcomeswere more frequent with increasing tonsillardescent. With severe descent (group C) only12% were improved and 69% had deterio-rated on late assessment, whereas with milddescent (group A) only 18-5% had deterio-rated. Clinical features least likely to improvewere those related to brainstem/cerebellarconnections and medulla oblongata, espe-cially ataxia, as was also the case in the recentreport by Menezes' group.'4 The differencebetween outcomes on late assessment ingroup A and B was not significant, so onceagain it was group C which was different.Outcome of both preoperative cough and

posture related headache showed no relationto tonsillar descent or any other imagingparameter including size of cisterna magna,yet the latter feature was significantlyimproved in 62-6% cases. Posture and coughrelated headache, like drop attacks, arethought to result from intermittent tonsillarimpaction in the foramen magnum27 whichsuboccipital craniectomy and duroplasty usu-ally relieve very well.26 Therefore the lack ofassociation of such features with small orobliterated cisterna magna or low lying tonsilsboth pre and postoperatively suggests that theorigin of these symptoms is more complex.

Small size or obliteration of the cisternamagna, like increasing tonsillar descent, wasrelated to clinical features localisable to themedulla oblongata and brainstem cerebellarconnections. The lack of a definite relationbetween the size of the cisterna magna andother clinical features which were related tolevel of the tonsils, such as operative out-come, suggests that the level of tonsils is themore important of these interrelated parame-ters. It was surprising that no association wasfound between the size of the cisterna magnaand syringomyelia, but once again the expla-nation in this study lies in group C where thecisterna magna usually was small or obliter-ated and syringomyelia uncommon. Howeverin a separate study using MRI, a lack of arelation between syringomyelia and the size ofthe CSF pathways across the foramen mag-num also was observed.25

Opacification of the fourth ventricle withwater soluble contrast medium nearly alwaysoccurs during examination of the cranioverte-bral junction in normal patients. It is a fea-ture which cannot be determined using MRIand was of potential interest in view of theo-ries still current about the causation of Chiariassociated syringomyelia. It was recorded inonly about 30% in this study which probablywas a sig-nificant departure from normal. Themaximum frequency of filling occurred ingroup B, which was the group with the high-est prevalence of syringomyelia. This was sur-prising perhaps, since it is widely acceptedthat obstruction of the outlets of the fourthventricle and cisterna magna by ectopic cere-bellar tonsils and arachnoiditis are central to

1076 on N

ovember 2, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.56.10.1072 on 1 O

ctober 1993. Dow

nloaded from

Chiari malformation in adults: relation of morphological aspects to clinicalfeatures and operative outcome

the development of syringomyelia,2>32 andstrategies to correct this underpin operativetreatment. Once again we believe the expla-nation in this study was simply the low inci-dence of fourth ventricle filling and ofsyringomyelia in patients of group C, becausethe differences between groups A and B werenot significant. This probably also is the mainreason why patients with filling of the fourthventricle and an enlarged spinal cord had abetter operative outcome than those without,because most of the former were in groups Aand B.

Medullary compression could be evaluatedadequately in only a small cohort of patientsand we did not attempt to relate this to otherradiological variables such as level of the ton-sils. Although multivariate analysis would berequired to distinguish the roles of this, andother features in this study in relation to out-come, it did seem to be related to one clinicalfeature, namely motor peripheral signs, butonly when compression was severe. Thereseemed to be no relationship with syringo-myelia which differed from the conclusions ofSpinos et al24 but was similar to those ofClifton et al,25 both based on MRI. It wasexpected that severe medullary compressionalso may have been relevant to some pooroperative outcomes, which would be compat-ible with the known significance of severespinal cord compression in cervical spondylo-sis33 and atlanto-axial subluxation,34 but num-bers of patients in this study were too small topermit any firm conclusion. The major com-pression usually was anterior, where it waslikely to be made worse by flexing the headto facilitate a satisfactory posterior operativeapproach.35

ConclusionThe main outcome of this analysis has beento identify patients with tonsillar descentbelow the upper border of the neural arch ofC2 as a special group. They had the highestfrequency of disabling ataxia and nystagmus,the lowest frequency of syringomyelia, andsuboccipital decompression was most likely toresult in deterioration rather than improve-ment. On the other hand patients with themost favourable operative outcomes werethose with mild tonsillar descent and a dis-tended cervical syrinx. Clinical malfunctionin the cerebellar/brainstem connections prob-ably is related to the severity of the hind brainanomaly, and when the anomaly is severe,clinical features are most likely to be end-points. This conclusion is supported by con-tinuing observations at our institution whereMRI virtually is now the only imaging modal-ity used in preoperative assessment of thiscondition.

1 Paul KS, Lye RH, Strang FA, Dutton J. Arnold Chiarimalformation. Review of 71 cases. J Neurosurg 1983;58:183-7.

2 Mohr PD, Strang FA, Sambrook MA, Boddie HG.Clinical and surgical features in 40 patients with pri-mary cerebellar ectopia (Adult Chiari malformation).QJ7Med 1977;181:85-96.

3 Dobkin BH. Syncope in the adult Chiari anomaly.Neurology 1978;28:718-20.

4 Gordon D. Neurological syndromes associated with cran-iovertebral anomalies. Proc R Soc Med 1969;62:725-6.

5 Bronstein AM, Miller DH, Rudge P, Kendall BE. Downbeating nystagmus: magnetic resonance imaging andneurological findings. JNeurol Sci 1987;81:173-84.

6 Caetano de Barros M, Farias W, Ataide L, Lins S. Basilarimpression and Arnold Chiari malformation. A studyof 66 cases. J Neurol Neurosurg Psychiatry 1968;31:596-605.

7 Aboulezz AD, Sartor K, Geyer CA, Gado MH. Positionof cerebellar tonsils in the normal population and inpatients with Chiari malformation: a quantitativeapproach with MR imaging. Jf Comput Assis Tomogr1985;9: 1033-6.

8 Banerafi NK, Millar JHD. Chiari malformation presentingin adult life. The relationship to syringomyelia. Brain1974;97: 157-68.

9 Saez R, Onofrio B, Yanagihara T. Experience with ArnoldChiari malformation, 1960-1970. J Neurosurg 1976;45:416-22.

10 Garcia-Uria J, Leunda G, Carrillo R, Bravo G. Syringo-myelia: long-term results after posterior fossa decom-pression. J Neurosurg 1981;54:380-3.

11 Logue V, Edwards MR. Syringomyelia and its surgicaltreatment; an analysis of 75 patients. J Neurol NeurosurgPsychiaty 1981;44:273-85.

12 Di Lorenzo N, Aldo F, Guidetti B. Craniovertebral junc-tion malformations. Clinico-radiological findings, long-term results and surgical indications in 63 cases.J Neurosurg 1982;57:603-8.

13 Eisenstat DDR, Bemstein M, Fleming JFR, et al. Chiarimalformation in adults. A review of 40 cases. Can JNeurol Sci 1986;13:221-8.

14 Dyste SN, Menezes AH, Van Gilder JC. SymptomaticChiari malformations. An analysis of presentation, man-agement and longterm outcome. Jf Neurosurg 1989;71:159-68.

15 Curnes JT, Oakes WJ, Boyko OB. MR imaging of hind-brain deformity in Chiari II patients with and withoutsymptoms of brainstem compression. AJNR 1989;10:293-302.

16 Wolpert SM, Scott RM, Platenberg C, Runge VM. Theclinical significance of hindbrain herniation and defor-mity as shown on MR images of patients with Chiari IImalformations. AJNR 1988;9:1075-8.

17 Elster AD, Chen MY, Chiari I malformations: clinical andradiological re-appraisal. Radiology 1992;183:347-53.

18 Batzdorf V. Chiari I malformation with syringomyelia.J Neurosurg 1988;68:726-30.

19 Rhoton AL. Microsurgery of the Arnold Chiari malforma-tion in adults with and without hydrocephalus.J Neurosurg 1976;45:473-83.

20 O'Connor S, du Boulay G, Logue V. The normal positionof the cerebellar tonsils as demonstrated by myelogra-phy. J Neurosurg 1973;39:387-9.

21 Baker H L. Myelographic examination of the posteriorfossa with positive contrast medium. Radiology 1963;81:791-801.

22 Stevens JM, Kendall E. Aspects of the anatomy of thecerebellum on computed tomography. Neuroradiology1985;27:390-5.

23 Naidich TP, McLone DS, Fulling KH. Chiari malforma-tion: Part IV. The hindbrain deformity. Neuroradiology1983;25: 179-97.

24 Spinos E, Laster DW, Moody P, et al. MR evaluationof the Chiari malformation at 0-15T. AJR 1985;144:1143-8.

25 Clifton AG, Stevens JM, Kendall BE. Idiopathic andChiari associated syringomyelia in adults: observationson cerebrospinal fluid pathways at the foramen magnumusing static MRI. Neuroradiology 199 1;33 (suppl): 167-9.

26 Clifton A, Stevens JM, Kendall BE. Morphological fea-tures and their interrelationships in the adult Chiarimalformation: an MRI study. Neuroradiology (in print).

27 Larson SJ, Sances A, Baker JB, Reigel DH. Herniatedcerebellar tonsils and cough syncope. J Neurosurg1974;40:524-8.

28 Williams B. Cerebrospinal fluid pressure changes inresponse to coughing. Brain 1976;99:331-46.

29 Barnett HJM, Foster JB, Hudgson P, eds. Syringomyelia.London: Saunders, 1973.

30 Newman PK, Terenty TR, Foster JB. Some observationson the pathogenesis of syringomyelia. JfNeurol NeurosurgPsychiatry 1981;44:964-9.

31 Newton EJ. Syringomyelia as a manifestation of defectivefourth ventricular drainage. Ann Roy Coll Surg Engl1969;44: 199-214.

32 du Boulay G, Shah S H, Currie J C, Logue V. The mech-anisms of hydromyelia in Chiari type I malformations.BrJRadiol 1973;47:579-87.

33 Fujiwara K, Yenenobu K, Ebara S, Yamashita K, Ono K.The prognosis of surgery for cervical compressionmyelopathy. An analysis of the factors involved. JfBoneJoint Surg 1989;71-B:393-8.

34 Hunter J, Stevens JM, Kendall BE, Moskovich R,Crockard HA. Radiological assessment for transoralsurgery in rheumatoid arthritis, using dynamic com-puted myelography. Neuroradiology 1991;33 (suppl):413-5.

35 Brieg A. Biomechanics of the central nervous system. Somebasic normal and pathological phenomena. Uppsala:Almquist and Wiksells, 1960.

1077 on N

ovember 2, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.56.10.1072 on 1 O

ctober 1993. Dow

nloaded from