genetic epidemiology of multiple sclerosisdistribution of multiple sclerosis. whereas studies...

3rournal ofNeurology, Neurosurgery, and Psychiatry 1997;62:553-561

NEUROEPIDEMIOLOGY

Genetic epidemiology of multiple sclerosis

Alastair Compston

Epidemiological studies of multiple sclerosishave been performed on almost an industrialscale over the past 90 years. Morbidity statis-tics have been used to generate aetiologicalhypotheses, to assess local needs for the provi-sion of services and the allocation of resources,and to define the natural history of the disease.Methodological factors have limited the extentto which the many surveys have yielded defini-tive conclusions in any one of these contexts.Most vulnerable have been the comparisons ofprevalence between regions and the serialstudies of single places.

In planning an epidemiological study inmultiple sclerosis, the usual practice is toretrieve cases from lists of those already knownto be affected. Because in most parts of theworld the diagnosis is coordinated throughhospital clinics, scrutiny of departmental andoffice notes provides the best source of infor-mation. In some situations, a case can bemade for retrospectively adjusting statistics toinclude those people who would have featuredin a population based survey if their where-abouts or clinical status had been known at thetime (onset adjusted prevalence'); their exclu-sion can then be regarded as an error ofadministration, recognition, disease expres-sion, or any one of the quirks which makes oneperson seek medical advice in advance ofanother. Rigid application of criteria for inclu-sion, and the decision to omit suspected cases,will vary depending on the purposes of thestudy. For surveys examining biological fea-tures, the error should be towards inclusion ofthose who probably have the disease processeven if this is not yet clinically declared. Inother contexts, it is essential to restrict the reg-ister to those who definitely have the disease.

Cases of different racial origin should not begrouped because they may differ for importantcharacteristics. Sociohistorical factors areknown to create significant differences in riskstatus even across quite small regions; con-versely, some questions relating to the epi-demiology of multiple sclerosis, which involvecohort studies, can only be answered by com-paring specifically different locations. It makeslittle sense to plan a study requiring therecruitment of significant numbers of patientswith a rare manifestation of multiple sclerosis,such as twinning or familial disease, in a com-munity which has a low overall prevalence of

the disease. Similarly, a ubiquitous but biolog-ically important feature may not differ signifi-cantly between groups in places wheremultiple sclerosis is relatively common. It fol-lows that there are usually better opportunitesfor identifying risk factors which make a signif-icant contribution to the disease but are fre-quent in the at risk population by working inareas of low prevalence; conversely, risk fac-tors for multiple sclerosis which are not over-represented in the normal population will beidentified more easily in high prevalenceregions.The extent to which complete ascertain-

ment is achieved depends much on the struc-ture, size, and distribution of the denominator,and whether the population has previouslybeen surveyed. The few patients with multiplesclerosis in a medium to low prevalence islandcommunity with a demographically stablepopulation of around 20 000 and a normal agestructure, can easily be ascertained but whensurveying the disease in one at risk group liv-ing within a large metropolitan but ethnicallymixed community, it may prove impossible toascertain with confidence either the numeratoror denominator, especially if recent populationcensuses are not available. Improved provisionof facilities for the disabled inflates both preva-lence and mortality; and the arrival of aninvestigator with a special interest in the dis-ease abruptly increases morbidity estimatesalthough these will plateau once ascertainmentis saturated. Underestimating the absolutenumber of cases may not affect the definitionof geographical gradients in the distribution ofmultiple sclerosis but it does matter in serialstudies of a single region where a rise in preva-lence resulting from reduced mortality andaltered diagnostic criteria has to be distin-guished from a real increase in incidence. Itfollows that investigator vigilance is a majorconfounding factor in comparative epidemiol-ogy.

Distribution of multiple sclerosisBy the beginning of the 20th century, multiplesclerosis-a disease that merited individualcase reports 25 years previously-had becomeone of the commonest reasons for admissionto a neurological ward. The period 1900 to1950 saw a gradual evolution of methods

University ofCambridge ClinicalSchool, Addenbrooke'sHospital, Hills Road,Cambridge CB2 2QQ,UK and The MRCCambridge Centrefor Brain Repair,University Forvie Site,Robinson Way,CambridgeCB2 2PY, UKA CompstonCorrespondence to:Professor A Compston,University of CambridgeClinical School,Addenbrooke's Hospital,Hills Road, CambridgeCB2 2QQ, UK.Received 16 January 1997and in revised form28 February 1997Accepted 3 March 1997

553

on July 1, 2020 by guest. Protected by copyright.

http://jnnp.bmj.com

/J N

eurol Neurosurg P

sychiatry: first published as 10.1136/jnnp.62.6.553 on 1 June 1997. Dow

nloaded from

http://jnnp.bmj.com/

Compston

Figure 1 Distribution ofmultiple sclerosis inScandinavia (Finland,Sweden, Norway,Denmark, and Iceland),central Europe (Holland,Switzerland, Austria,Germany, Hungary,Slovakia, Poland),France, Spain andPortugal, Italy, Malta,Greece, Albania, Croatia,Romania, and Bulgaria;figures are prevalencellO5of the population.

required for accurate definition of populationbased statistics; over the past four decades,surveys from most parts of the world haveestablished the big picture on the geography ofthe disease and have allowed speculation onthe reasons for its chacteristic distrbution.Twenty years ago John Kurtzke, who has

worked hardest to make sense of the epidemio-logical facts relating to multiple sclerosis, col-lated the published surveys of prevalence andsuggested that the distribution could broadlybe classified into bands of low, medium, andhigh prevalence.2A4 High risk (>30/105) wasfound throughout northern Europe, the north-ern United States, Canada, southernAustralia, and new Zealand; medium risk(5-30/105) was found in southern Europe, thesouthern United States and northernAustralia; and low risk (<5/105) areas included

134 Shetland° 287 Orkney

Figure 2 Distribution ofmultiple sclerosis in theUnited Kingdom; figuresare prevalence/l 05 of thepopulation.

Asia, South America, and many unchartedregions. Systematic updating of these figuresshows that the absolute number of cases iden-tified in different parts of the world has risensteadily since the 1950s.57 Rather than cata-logue all the surveys (now running to manyhundreds) or describe the laborious evolutionof the ideas on causation which flow fromthem, figs 1-4 depict surveys for the regionalprevalences of multiple sclerosis in 1997 inEurope, the United Kingdom, North Americaand Canada, and Australasia: there is a rela-tive paucity of information on incidence; indi-vidual sources of information are not cited;nor are 95% confidence intervals (95% CIs)shown as these are not always provided in theoriginal publications and some of the figuresare a best guess. The figures for the UnitedKingdom in fig 1 are regional approximationsfor those given in fig 2. What emerges from themore recent surveys is that many of the puta-tive claims for latitudinal gradients were over-stated but it remains the case that the diseasedoes show variations in its distribution overquite small distances which may be informa-tive with respect to ideas on the aetiology.

Multiple sclerosis is evidently a commondisorder of young adults in northern Europe,continental North America and Australasiabut not in the Orient, the Arabian peninsula,Africa, continental South America, and India(not all shown in the figures). Nevertheless, innorthern Europe, prevalence (and incidence)are higher in southern Scandinavia, northernGermany, the United Kingdom and parts ofItaly, than in northern Scandinavia, France,Spain, and the eastern Mediterranean coun-tries. Within the United Kingdom, the diseaseis more prevalent in north east Scotland, andthe Orkney and Shetland islands than in otherparts of England, Wales, and Ireland.8 In Italy,

554


http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



Figure 3 Distribution of multiple sclerosis in the United States and Canada; figures areprevalencell0/ of the population.

38~~2

77

Figure 4 Distribution of multiple sclerosis in Australia and New Zealand;figures areprevalencell0/ of the population.

pronounced differences in prevalence existbetween regions and islands that are geograph-ically close but differ in their genetic and cul-tural histories.9 In north America, there is adiagonal gradient in frequency with the high-est rates in the midwest and the lowest in theMissisippi delta.'0 The comprehensive surveyfrom Australia, in which four regions were sur-veyed simultaneously using comparable meth-ods and working to a common prevalence datealso shows a latitudinal gradient for the whiteAustralian population with higher rates in thesouth than in the north."I

Multiple sclerosis in migrantsPopulations are neither geographically norsocially stable and there have been manymigrations involving relatively large numbersof people which seem to have influenced thedistribution of multiple sclerosis. Whereasstudies of stable indigenous populations illus-trate differences in genetic susceptibility, the

effect of migration has been to define multiplesclerosis as an acquired exogenous disorder.

Attention was first paid to comparisons ofthe frequency of multiple sclerosis betweenracial groups in a single geographical setting inSouth Africa during the 1950s. Dean showedthat age corrected frequency of the disease washighest in immigrants from Europe, low inAfrikaaners, and intermediate in SouthAfrican English for both prevalence and inci-dence'2; the absolute absence of multiple scle-rosis in African black people was confirmedbut a slightly higher rate was seen in the CapeColoured population, which has mixedAfrican and European ancestry. Within theEnglish speaking white population, those mov-ing from northern Europe to southern Africaas adults took with them the high frequency ofthe country of origin, whereas those migratingyounger than 15 years showed the lower ratescharacteristic of native born inhabitants ofsouthern Africa.No less influential has been the study of

United Kingdom born children of immigrantsfrom the Indian subcontinent, Africa, and theWest Indies,'3 14 which showed that the preva-lence of multiple sclerosis in the UnitedKingdom born children of West Indian,African, and Asian immigrants approximatesto that seen in similar age groups among thewhite population. Although methodologicaldifficulties arise in this study of an ethnicminority living in a large metropolis, there areseveral reasons for suspecting that the numberof patients with multiple sclerosis, born in theUnited Kingdom of parents who weremigrants from the West Indies, Africa, or Asia,has actually been underascertained. However,the danger of extrapolating from studiesinvolving a small numerator living in anunusual environment is well illustrated by thestudy of multiple sclerosis among immigrantsfrom Vietnam to Paris, France.'5 Three caseswere identified among a cohort of around3400 persons born of Vietnamese motherswho came to France under the age of 20 years.The fact that the cumulative 18 year risk ofmultiple sclerosis, by 1975, was 89/105 (95%CI 18-260), with an age specific prevalence of169/105 (95% CI 94-135) in the third decadeactually tells us nothing about the shift in risk ofthe disease consequent on movement of peo-ple from southern Asia to northern France as,apart from the wide confidence intervals, hav-ing a French born father was a requirement forimmigration; genetic admixture thereforeintroduced an essential bias into this study.

Another classic series of epidemiologicalstudies compared the frequency of multiplesclerosis among Japanese in Hawaii, the westcoast of North America, and Japan. In Hawaii,the prevalence among Japanese was 7/105compared with 34/105 in white immigrants toHawaii'6S8; these rates were virtually identicalfor Japanese and white people living inCalifornia'9 and can be compared with theexpected rate of 2/105 for native Japanese.20Here, the evidence favours a strong protectiveeffect for the Japanese irrespective of environ-ment.

555


http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


Compston

The other location where migration hasoccurred on a sufficient scale to show impor-tant age related differences in prevalence ofmultiple sclerosis is Israel. The original studyin immigrants reported a difference in preva-lence between migrants from northern Europe(Ashkenazi Jews) and from Asia and Africa(Sephardic Jews).21 22 The higher frequency inAshkenazi than Sephardic Jews also showed anage at migration effect, in that there were veryfew Ashkenazis in the cohort arriving in Israelbefore adolescence. Although crude ratesretained the difference seen in the parentalgroups, prevalence in the Israeli born childrenof Ashkenazi and Sephardic Jews was the sameafter age adjustment to the population of theUnited States. These important studies haverecently been updated23; depending on placeof paternal birth, prevalence (age adjusted tothe Israeli population of 1960) was estimatedat 32 (fathers born in Israel), 38 (fathers bornin Europe or North America), and 29 (fathersborn in Africa or Asia) compared with 14/105in immigrants. Higher rates were found inJerusalem (61, 68, and 51/105 for Israelis withIsraeli, European/American, and African/Asian fathers respectively) than in other partsof the country. The implication is that, at leastfor Ashkenazi and Sephardic Jews, raciallydetermined differences in risk for multiplesclerosis are modified by environment.Movements within one continent or country

are also informative for the assessment of riskdepending on time spent in regions of differingprevalence. A study of United States citizensshowed that mortality for southern bornpatients dying in the north was 0-68/105/yearcompared with 0A46/105/year for those remain-ing in the south. The mortality ratio forUnited States army veterans born in- the highfrequency northern states and entering mili-tary service from the middle zone droppedfrom 1-48 to 1-27, and to 0 74 for those enter-ing from the southern states. Those born inthe medium risk zone showed a ratio increasein frequency of multiple sclerosis to 1-4 ifentering military service from the north and areduced ratio of 0-73 if entering from thesouth.24

All of these studies proved enormouslyinfluential in shaping ideas on the contributionof environmental factors to the aetiology ofmultiple sclerosis. Having established from themigration studies and analyses of epidemicsthat environmental factors probably do alterthe risks of developing multiple sclerosis, andmay override racial susceptibility, it becameimportant to establish at what age these influ-ences occur. Clearly, the critical period isbefore clinical onset and the studies fromSouth Africa and the United States veterans'survey suggest that, in all probability, the dis-ease process is established in childhood butfew would be confident about confining risk toa particular calender age.As part of a national survey of multiple scle-

rosis in France, 246 persons were identifiedwho had migrated from north Africa in thefirst quinquennium of the 1960s after theAlgerian war for independence, among 8000

cases ascertained overall in France.5Excluding the 27 patients who had multiplesclerosis before, or at the time of migration,86% of these 246 probands were European inorigin and the remainder Arab or Berber.There was no apparent age or sex adjusted dif-ference in frequency or mean age at onsetbetween these and native French cases andthis has been interpreted as indicating that theprovocative exogenous factors are ubiquitousand that multiple sclerosis is acquired by thesame age in each group. As far as this study isconcerned, Kurtzke considers that matchingby age has introduced a confounding factor asthis will have restricted individual subjects tothose with the same age at onset; he prefersthe interpretation that in this cohort there is afixed interval between migration and clinicalonset, regardless of age, and takes the study toprovide evidence for susceptibility extendingto people in the mid-40s.7 However, a range ofmore than 30 years could be considered not toprovide much insight into the critical age ofexposure to the putative agent which causesmultiple sclerosis.

Epidemics of multiple sclerosisThose who espouse the environmental doc-trine of multiple sclerosis are naturally enthu-siastic about epidemics of the disease. Thearguments put forward by Kurtzke7 for pointsource epidemics, especially that proposed forthe Faroe Islands, have not been universallyaccepted and others take the view that theseare epidemics of recognition reflecting thearrival of specialist medical services in islandcommunities rather than a genuine change inincidence arising from the introduction oftransmissible aetiological factors into virginpopulations.

In the first survey of Iceland, 168 cases ofmultiple sclerosis were identified with onsetbetween 1900 and 1975.26 Annual incidencerates seemed to rise around 1922; they thenstabilised until a further increase occurred in1945, heralding a steady decline from the mid-1950s. Each quinquennial rate for incidencefrom 1900 was lower than between 1945 and1954, during which age at onset was alsoyounger than before or after this period. Thisled to the conclusion that there had been apostwar epidemic of multiple sclerosis inIceland26 27 but opinions differ both withrespect to the facts and their interpretation.John Benedikz, an Icelandic neurologist whohas taken a particular interest in the epidemi-ology of multiple sclerosis, favours the viewthat any change in frequency of multiple scle-rosis in Iceland during the 20th century shouldbe attributed to improved recognition anddiagnostic procedures rather than increasedincidence.' 28 In support of this interpretation,Benedikz et al have reviewed 323 patients withonset of symptoms attributed to multiple scle-rosis after 1 January 1900 of whom 252 werestill living in December 1989.29 Incidence rateswere <1/105/year up until the 1930s but thenincreased to 2-5/105/year, coinciding with thearrival of two neurologists. With waning

556


http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



enthusiasm, so the analysis goes, there wasthen a lull between 1945 and 1955, when inci-dence increased to 3-3/105/year after the firstsystematic survey of the disease. With nineneurologists in practice from 1975, there was afurther increase to a peak incidence of4 1/105/year, which is being maintained.Benedicz et al argue that case ascertainment inthe era before 1950 was far from adequate asshown by the long interval at that timebetween onset and diagnosis, with a paucity ofmilder registered cases, and comparisons ofdisability in the affected cohorts showing anoverloading of those with severe multiple scle-rosis before 1950. The abrupt reduction ininterval between onset and diagnosis after1940 is seen as further evidence for the impactof neurological expertise on early diagnosisand improved case ascertainment.The original findings on multiple sclerosis

in the Faroe islands303' showed fewer casesthan expected from comparisons with neigh-bouring Orkney and Shetland. Kurtzke7searched many places in his attempt to tracecases back to before the second world war butin neither the initial survey3' nor later assess-ments32<-3 was any patient identified with anestimated date of onset earlier than 1943.There were then 16 cases with onset between1943 and 1949, a further 16 developing clinicalmanifestations between 1950 and 1973 and,by 1986, 41 patients had been ascertained ofwhom nine had lived abroard for three ormore years and so were not considered directlyto have been part of the epidemic. Kurtzkeconcludes that the critical factor determiningthe Faroes experience of multiple sclerosis wasoccupation by British troops between 1940and 1945, the development of multiple sclero-sis showing both a temporal and spatial rela-tion in that villages where people lived whocontributed to each of the incidence peakswere also those where troops were billeted.But some are not convinced by this analysis3637despite robust counter claims from the mainprotagonists with respect to specific criticismsconcerning validity of the diagnoses, exclu-sions, case ascertainment, definition of epi-demics, and the putative role of the Britishoccupation in the genesis of this cluster.735

In the Orkney and Shetland Islands, theincidence and prevalence of multiple sclerosiswere at one time higher, almost by an order ofmagnitude, than in other regions. Estimates ofprevalence carried out on four occasionsbetween 1954 and 1974 showed a steady risein frequency from 11 1/105 in 1954 to 309/105in 1974 for Orkney, and from 134/105 to184/105 in Shetland over the same period.838Comparison of incidence and mortality con-firms the impression that there has been nosignificant alteration in statistics for the dis-ease other than those attributable to changesin classification and ascertainment. Over theperiod of these studies, systematic depopula-tion in Orkney and Shetland left an older pop-ulation less at risk of multiple sclerosis and, aselsewhere, the rise in prevalence can best beattributed to increased survival (from 26 to 40years in Orkney and from 24-34 years in

Shetland between 1954 and 1974) andimproved case recognition. The question ofserial change in incidence for multiple sclero-sis in Orkney has since been revisited;although there had been a steady reductionfrom 1964, prevalence figures for 1994 were287/105 (probable and definite cases) and134/105 for Orkney and Shetland respectively(S Cook, personal communication). Thesestatistics show that although there has beensignificantly less multiple sclerosis in Shetlandsince 1965, this is no longer true for Orkney.An epidemic has been claimed for Key

West, a tropical island off the west coast ofFlorida where 37 patients with peak onset inand around 1977-9 were identified in 1984(prevalence 140/1 05)-an increase that couldnot be attributed to alterations in clinical vigi-lance or differential migration of symptomaticpersons to a more favourable climate.39

Factors which predict the clinical courseApart from surveys which have assessed highlyselected populations of patients, multiple scle-rosis is almost always found to be more com-mon in females than in males. The reportedfigures vary but a sex ratio of two females: onemale is usual, irrespective of ethnicity40 41; inchildren, the excess of females is even morepronounced42 whereas multiple sclerosis pre-senting in or beyond the fifth decade morecommonly affects males.40

At diagnosis, many patients with multiplesclerosis want information on the prognosis.So-called benign multiple sclerosis is charac-terised by young age at onset, usually occur-ring in females with infrequent sensoryepisodes which recover fully; conversely, theprognosis is more predictably gloomy in maleswith older age of onset and a progressivecourse involving motor systems. The influenceof attack rate early in the disease on outcomehas been somewhat less clear. Whereas allneurologists will be familiar with patients inwhom a brisk start to the disease with severalnasty attacks settled down into a much morefavourable middle and long term course thanhad at first seemed probable, the large popula-tion based studies have shown that attack ratein the early years is of some prognosticvalue.40 41 43Trauma as a trigger of disease activity in

multiple sclerosis has been much considered,and still features in the law courts, whereplaintiffs may claim that an accident has pro-voked the first appearance of multiple sclerosisor altered the course of pre-existing manifesta-tions. Sibley et a144 prospectively studied dis-ease activity by questionnaire and physicalexamination for eight years. Taking either thethree or six month period after each event as atrisk, only electrical trauma showed an associa-tion with new episodes; all other noted formsof trauma were negatively correlated both withclinical exacerbations and disease progression.Siva et al,45 using the Mayo Clinic cohort, alsoconclude that disease exacerbations are nomore frequent in the six months after limbfracture than at other times.

557


http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


Compston

The situation with respect to the risk fromanaesthesia is unsatisfactory in that no epi-demiologically based studies have been per-formed and the evidence is entirely anecdotal;some neurologists advise patients to avoidelective interventions while remaining sensibleabout treatments or procedures which justifythe small risk of increasing disease activity-if itactually exists.46 47

Several authors have shown prospectivelythat new episodes of demyelination increaseafter (presumed) viral exposure but no singleagent has been implicated48 49; 9% of presumedinfections are followed by relapse and 27% ofnew episodes are related to infection; the rela-tive risk for relapse in the four week periodafter upper respiratory (especially adenovirus)or gastrointestinal infections is 1-3.

Anecdotal evidence on whether pregnancyaffects the immediate or long term course ofmultiple sclerosis has now been supplementedby prospective surveys which indicate that theonset of multiple sclerosis does not clusteraround pregnancy, and that having childrendoes not alter the long term course of the dis-ease but there is an increase in relapse rateduring the puerperium. However, a majorconfounder in the detailed interpretation ofthese studies is the decision by women withsevere disability not to embark on pregnancyand the corresponding preparedness of thosewith mild disease to start or extend their fami-lies. The prospective studies indicate a roughlythreefold higher risk in the three to six monthsafter term than during pregnancy,5051 and sug-gest that the attacks may be more severe.52There is less agreement on whether or not therelapse rate is maintained or falls during thepregnancy itself, as was suggested in several ofthe retrospective surveys.3 In the most com-prehensive epidemiological analysis of issuesrelating to multiple sclerosis and pregnancy,Runmarker and Andersen56 studied an incep-tion cohort in Goteborg, Sweden and disposedof the hypothesis that the onset of multiplesclerosis is influenced by pregnancy; there wasa conspicuous absence of onset bouts duringpregnancy compared with non-pregnantepochs including the puerperal eight months.Fecundity was reduced in women with multi-ple sclerosis, presumably by choice, especiallyin the context of significant disability and this isthe probable explanation for the conclusionthat pregnancy after onset is associated with alower risk of progression.

Familial multiple sclerosisMultiple sclerosis has a familial recurrence rateof about 15% and it is usually assumed thatthis is due to coinheritance of susceptibilityfactors, but the altemative hypothesis is thatthis results from common exposure to environ-mental factors in childhood. The most com-

prehensive study of recurrence (from Canada)takes as its baseline a lifetime risk of 0-2% forthe entire population, and shows an increase to3% in other first degree relatives (relative risk20) and 1% in second degree relatives (relativerisk 5 5).57

Comparable studies from the UnitedKingdom confirm that the highest age adjustedrecurrence rate is for sisters (4A4%) and brothers(3 2%), compared with parents (2-1%) andoffspring (1 8%). Overall, the reduction in riskchanges from 2-8% (relative risk 9 2) in firstdegree relatives to 1-0% (relative risk 3 4) and0 9% (relative risk 2 9) in second and thirddegree relatives, respectively, compared with abackground age adjusted risk in this popula-tion of 0 3%.5 In Flanders, the recurrencerisks are 10-fold to 1 2-fold for first degree andthreefold for second degree relatives.59With some variations in methodology, three

recent studies approximate to a populationbased series of multiple sclerosis in twins.60 63Two show remarkable consistency in demon-strating a higher clinical concordance rate inmonozygotic twins (about 25%) than dizygoticpairs (about 3%); the French study is excep-tional in showing no significant differencebetween monozygotic and dizygotic twins butcritics have argued that this result is within theconfidence limits of the other surveys. The rel-ative risk for multiple sclerosis in the monozy-gotic twin partner of an affected proband istherefore about 190.

Adopted persons who subsequently developmultiple sclerosis, and affected people whohave themselves adopted children provide anunusual but informative resource for studyingthe relative contribution of genes and the envi-ronment in the aetiology of multiple sclerosis.Considering those with multiple sclerosis whoare adopted before the age of one year, andthose with multiple sclerosis who throughadoption have non-biological siblings or chil-dren, the frequency of multiple sclerosis innon-biological parents, siblings, and children ismore or less identical to the population preva-lence and lifetime risk for Europeans, and sig-nificantly lower than that expected from thestudy of recurrence risks in the biological rela-tives of index cases.64 Half siblings offer yet onemore variant on the familial multiple sclerosistheme in that they share a proportion ofparental genes and divide into those who arereared together and apart, at least during theperiod which is thought critical for the devel-opment of multiple sclerosis. The age adjustedrisk for half siblings is significantly lower thanfor full siblings and there is no difference inrisk for half siblings reared together andapart.65

Conjugal pairs with multiple sclerosis whohave children provide a special opportunity forassessing the contribution made to susceptibilityby genetic factors. Five of 86 offspringfrom 45 conjugal pairs living in the UnitedKingdom were shown also to have multiplesclerosis and a further five had either charac-teristic imaging abnormalities or clinical symp-toms consistent with demyelination but did notmeet the criteria for clinically definite disease;the recurrence risk (crude 1 in 17: age-adjusted1 in 5) is obviously much higher than the riskfor the children of single affected patients(crude 1 in 200: age adjusted 1 in 50).66Conjugal pairs can also be used to assess theinfluence of environmental factors in determin-

558


http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



Northern Europeans (1:600)000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000

Child [one affected parent] (1:200)000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000o000000000000000000000000000000000000000000000oo00000000000000000000000000000000000000

Affected sibling/dizygotic twin (1:40)oo0000000000000000000000o000000000000000

Child [conjugal parents] (1:17)00000000000000000

Affected monozygotic twin (1:3)000

Class 2 MHC Association (DRw15/DQw6)or

T cell receptor V beta 8 polymorphismor

Ig Vh region polymorphism (each c1:150)

00000000000000000000000000000000000000000000000000oooo0ooooooooooooooooooooooooooooooooooooooooooooo00000000000000000000000000000000000000000000000000

Class 2 MHC Association (DRw15/DQw6)and

TCR V beta 8 polymorphism (c1:60)000000000000000000000000000000000000000000000000000000000000

New regions from genome screens (1:25)000000000000000000000000

Figure 5 Scheme to show the reduction in crude risk for multiple sclerosis depending onrelation to the proband and the presence of defined susceptibility factors. Comparable ageadjustedfigures are: children ofsingle affected parents 1 in 50; siblings 1 in 30; children ofconjugal pairs 1 in 5; monozygotic twins 1 in 2.

ing the onset and course of multiple sclerosis.In the United Kingdom survey, there was noevidence for clinical concordance, clustering atyear of onset, or distortion of the expected pat-tern of age at onset in the second affectedspouse from 33 pairs in whom these compar-isons could be made.66 Figure 5 summarisesthese risks.

Markers of genetic susceptibilityThe change in recurrence risk seen betweentwins and first and second degree relatives sug-gests the independent or epistatic effects ofmore than one gene; the modest trend foraffected pairs to be female hints at a contribu-tion of sex independent of genes; the low con-cordance rate, even among monozygotic pairs,indicates a significant independent or modify-ing effect of the environment on expression ofgenetic susceptibility.

These findings have stimulated attempts toidentify and locate the genes which confer sus-ceptibility to the disease. Population studieshave shown an association between theclass II MHC alleles DR15 and DQ6 andtheir corresponding genotypes DRB1*1501,DRB5*0101 and DQA1*0102,DQB2*0602.67 An extensive search, usingassociation and linkage studies, has onlyyielded additional putative candidate genes inthe VH2-5 immunoglobulin heavy chain andthe T cell receptor I8 chain variable regions. 68-71But the contribution to susceptibility made bythe genes which have provisionally been identi-fied, even if their effects are interactive, canonly account for a proportion of the increasedrisk of multiple sclerosis implicated by familystudies; and it seems likely that other geneswhich make an even greater contribution tosusceptibility remain to be identified.

Three groups of investigators have nowundertaken a systematic search of the genomein an attempt to locate additional susceptibil-ity genes using affected family members-usu-ally identity by descent analysis in siblingpairs. Genotyping was completed on cohortseach of between 75 and 225 families, togetherinvolving in excess of 1000 members, for eachof between 257-443 microsatellite markers.These markers were chosen to have an averagespacing of around 10 cM giving enough powerto identify regions encoding a major suscepti-bility gene; and they are sufficiently polymor-phic to make a high proportion of the availablefamilies fully informative. Superficially, theresults show a disappointing lack of overlap.The importance of HIA is confirmed but ofthe other new regions of interest, several areclearly unique to each screen and so may befalse positives. The regions of interest emerg-ing from the United Kingdom genomescreen72 are 1 cen, 5 cen, 6p, 7p, 14q, 17q,19q, and Xp; they are 2p, 3p, 5p, 1 1 q, and Xpin the Canadian series;73 and 6p, 7q, 1 lp, 12q,and 19q in the United States/French survey.74Despite inconsistencies between the samples,it remains possible that meta-analysis will pro-vide stronger evidence implicating one ormore of these provisional areas of linkage, andthe hope is that many of the false positive leadswill be eliminated while larger clinicalresources are deployed and new strategies pur-sued for detecting both linkage and new associ-ations within families.

Analysing the epidemiological pattern inmultiple sclerosisThe complex interplay of nature and nurtureis reflected in the distribution of many dis-eases; some, such as malaria and the haemo-globinopathies, are reasonably wellunderstood, but factors which determine thegeography of complex traits remain muchmore enigmatic. As parts of Europe and NorthAmerica were repeatedly surveyed for inci-dence and prevalence of multiple sclerosis overseveral decades, neuroepidemiologists beganto polarise their views on the emerging pat-terns around the race versus place debate.

559


http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


560

Especially influential were the serial surveys ofstable populations, and the epidemiologicallyless robust studies of small island populations.It was rapidly pointed out that the gradients infrequency within northern Europe, NorthAmerica, and Australia compared with otherparts of the world implicate an environmentalfactor in the aetiology of multiple sclerosiswhich is not ubiquitously distributed. Thesurveys of multiple sclerosis in northernEuropeans migrating west to east providedcrucial information implicating the role ofenvironmental factors in shaping the distribu-tion of the disease because, with the exceptionof Canada, parts of the world colonised fromnorthern Europe show prevalence rates thatare lower than in the country of origin.Considerable effort was therefore devoted to a

systematic assessment of environmental fac-tors in the hope that a relatively simplecausative event would be identified leading to

a strategy for eradication of the disease butnone was found using either population serol-ogy or the most complex methods for virusdetection which molecular biology can offer.With time it became clear that more complexpatterns of distribution exist within continentsand countries, and even across small regions.The evolving epidemiological features of multi-ple sclerosis made it necessary for epidemiolo-gists to construct ever more elaboratehypotheses to account for the distribution ofthe disease. Thus there were claims for a rolefor climate, diet, geomagnetism, and toxins inaddition to the infective aetiology.The migration studies indicate that the risk

of multiple sclerosis in a single ethnic group

varies with place of residence during a criticalperiod in childhood. However, migration doesnot just involve the movement of people; gene

transfer and population stratification also fol-low. Failure to define the environmental cause

of multiple sclerosis therefore led others tointerpret the distribution of multiple sclerosisas a function of genetic susceptibility.'07576Each of these commentators concluded thatmultiple sclerosis is to be found where thereare northern European genes. It was furthersuggested that the increased risk of the diseasein native people moving out of Africa to theUnited States correlates with the extent towhich white genes are introduced into theblack community77; and some at least of theaffected children of West Indian immigrants tothe United Kingdom are known to have whiteBritish or Irish ancestors.The best supporting evidence that markers

of susceptibility to multiple sclerosis showmuch the same geographical patterns as thedisease itself comes from the construction ofdetailed genetic maps which include informa-tion onclines for HLA-B7,75 known to beassociated in most populations with anincreased riskof multipnle sclerosis; presum-ably the same is true for DR15 and DQ6. As aconsequence of sociohistorical events andpopulation migrations, genes with a high fre-quency in the migrating populations necessarilybecome concentrated in small isolates whereasothers are excluded, thus adding relic groups

to the global pattern that exists for many poly-morphic genetic markers, and the phenotypi-cally more obvious distributions of languageand anthropometrics.By contrast with the interpretations offered

by genetic epidemiologists, backed by evi-dence from population genetics, environmen-talists must still leave the facts to speak forthemselves as the case for an environmentalagent as the dominant cause of multiple scle-rosis remains stubbornly circumstantial. Onthe basis of the putative epidemic of multiplesclerosis in the Faroes, Kurtzke7 concludesthat multiple sclerosis originated inScandinavia (central Norway or the south-central lake district of Sweden) in the early18th century and diffused across the Balticstates and northern Europe including theBritish Isles over the next 100 years. Fromthere, it was exported to north America andAustralasia, to South Africa and Italy. Thetheme is familiar but whereas for Kurtzke thefactors being distributed are germs, for othersthey are genes.79

1 Poser CM, Benedikz J, Hibberd PL. The epidemiology ofmultiple sclerosis: the Iceland model. Onset-adjustedprevalence rate and other methodological considerations._7 Neurol Sci 1992;111: 143-52.

2 Kurtzke JF. A reassessment of the distribution of multiplesclerosis. Part one. Acta Neurol Scand 1975;51:110-36.

3 Kurtzke JF, A reassessment of the distribution of multiplesclerosis. Part two. Acta Neurol Scand 1975;51:137-57.

4 KurtzkeJF. Geography in multiple sclerosis. 7 Neurol1977;215: 1-26.

5 Bauer HJ. Multiple sclerosis in Europe. Symposium report..7 Neurol 1987;234: 195-206.

6 Lauer K, Firnhaber W. Multiple sclerosis in Europe: an epi-demiological update. Darmstadt: Leuchturm-Verlag/LTVPress, 1994:350.

7 Kurtzke JF. Epidemiologic evidence for multiple sclerosisas an infection.Clin Microbiol Rev 1993;6:382-427.

8 Robertson NP, Compston DAS. Surveying multiple sclero-sis in the United Kingdom. _7 Neurol Neurosurg Psychiatry1995;58:2-6.

9 Rosati G. Descriptive epidemiology of multiple sclerosis inEurope in the 1980s: a critical overview. Ann Neurol1994;36(suppl 2):S164-74.

10 Bulman D, Ebers GC. The geography of multiple sclerosisreflects genetic susceptibility. Journal of Tropical andGeographical Neurology 1992;2:66-72.

11 Hammond SR, McLeod JG, Millingen KS, Stewart-WynneEG, English D, HollandJT, McCall MG. The epidemi-ology of multiple sclerosis in 3 Australian cities: Perth,Newcastle, and Hobart. Brain 1988;111:1-25.

12 Dean G. Annual incidence, prevalence, and mortality ofMS in white South African-born and in white immigrantsto South Africa. BM_7 1967;ii:724-30.

13 Elian M, Dean G. Multiple sclerosis among UnitedKingdom born children of immigrants from the WestIndies. 7 Neurol Neurosurg Psychiatry 1987;50:327-32.

14 Elian M, NightingaleS, Dean G. Multiple sclerosis amongUnited Kingdom-born children of immigrants from theIndian subcontinent, Africa, and the West Indies. _7Neurol Neurosurg Psychiatry 1990;53:906-1 1.

15 KurtzkeJF, Bui QH. Multiple sclerosis in a migrant popu-lation to half orientals immigrating in childhood. AnnNeurol 1980;8:256-60.

16 Alter M, Okihiro M, Rowley W, Morris T. Multiple sclero-sis among orientals and caucasians in Hawaii. Neurology197 1;21:122-30.

17 Lauer K. The risk of multiple sclerosis in the USA in relationto sociogeographic features: a factor-analytic study. .7 ClinEpidemiol 1994;47:43-8.

18 Detels R, BrodyJF, Edgar AH. Multiple sclerosis amongAmerican, Japanese, and Chinese migrants to Californiaand Washington. _7 Chron Dis 1972;25:3-10.

19 Detels R, Visscher B, Malmgrem RM, Coulson AH, LuciaMV, DudleyJP. Evidence for lower susceptibility to mul-tiple sclerosis in Japanese-Americans. Am _7 Epidemiol1977;105:303-10.

20 Kuroiwa Y, Shibasaki H, Ikeda M. Prevalence of multiplesclerosis and its north-south gradient in Japan.Neuroepidemiology 1 983;2:62-9.

21 Alter M, Halpern L, Kurland LT, Bornstein V, Tikva P,Leibowitz U, Silberstein J. Multiple sclerosis in Israel:prevalence among immigrants and native inhabitants.Arch Neurol 1962;7:253-63.

22 Alter M, Kahana E, Loewenson R. Migration and risk ofmultiple sclerosis. Neurology 1978;28: 1089-93.

Compston


http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from



23 Kahana E, Zilber N, Abramson JH, Biton Y, Leibowitz Y,Abramsky 0. Multiple sclerosis: genetic versus environ-mental aetiology: epidemiology in Israel updated.Neurol 1994;241 :341-6.

24 Kurtzke JF, Kurland LT, Goldberg ID. Mortality andmigration in multiple sclerosis. Neurology 1971;21:1186-97.

25 Delasnerie-Laupretre N, Alperovitch A. Migration and ageat onset of multiple sclerosis in the pitfalls of migrantstudies. Acta Neurol Scand 1992;85:408-1 1.

26 Kurtzke JF, Gudmundsson KR, Bergmann S. Multiplesclerosis in Iceland. 1. Evidence of a post-war epidemic.Neurology 1982;32:143-50.

27 Cook SD, Gudmundsson G, Benedikz J, Dowling PC.Multiple sclerosis and distemper in Iceland. 1966-78.Acta Neurol Scand 1980;61:244-51.

28 Benedikz JG, Magnusson H, Poser CM, Benedikz E,Olafsdottir G, Gudmundsson G. Multiple sclerosis inIceland 1900-85. J7ournal of Tropical GeographicalNeurology 1991;1:16-22.

29 Benedikz JG, Magnusson H, Gudmundsson G. Multiplesclerosis in Iceland, with observations on the alleged epi-demic in the Faroe Islands. Ann Neurol 1994;36(suppl2):S 175-9.

30 Allison RS. Some neurologic aspects of medical geography.Proc R Soc Med 1963;56:71-6.

31 Fog T, Hyllested K. Prevalence of disseminated sclerosis inthe Faroes, the Orkneys, and Shetland. Acta Neurol Scand1966;42(suppl 19):9-11.

32 Kurtzke JF, Hyllested K. Multiple sclerosis in the FaroeIslands: 1. Clinical and epidemiological features. AnnNeurol 1979;5:6-21.

33 Kurtzke KF, Hyllested K. Multiple sclerosis in the FaroeIslands. II. Clinical update, transmission, and the natureof MS. Neurology 1986;36:307-28.

34 Kurtzke JF, Hyllested K. Multiple sclerosis in the FaroeIslands. III. An alternative assessment of the three epi-demics. Acta Neurol Scand 1987;76:317.

35 Kurtzke JF, Hyllested K. Validity of the epidemics of multi-ple sclerosis in the Faroe islands. Neuroepidemiology1988;7: 190-227.

36 Poser CM, Hibberd PL, Benedicz J, Gudmundsson G.Analysis of the "epidemic" of multiple sclerosis in theFaroe Islands. I. Clinical and epidemiological aspects.Neuroepidemiology 1988;7: 168-80.

37 Poser CM, Hibberd PL. Analysis of the "epidemic" of mul-tiple sclerosis in the Faroe Islands. II. Biostatisticalaspects. Neuroepidemiology 1988;7: 181-9.

38 Cook SD, Cromarty MB, Tapp W, Poskanzer D, WalkerJD, Dowling PC. Declining incidence of multiple sclerosisin the Orkney Islands. Neurology 1985;35:545-51.

39 Sheremata WA, Poskanzer DC, Withum DG, MacLeodCL, Whiteside ME. Unusual occurrence on a tropicalisland of multiple sclerosis [letter]. Lancet 1985;ii:618.

40 Weinshenker BG, Bass B, Rice GP, Noseworthy J, CarriereW, Baskerville J, Ebers GC. The natural history of multi-ple sclerosis: a geographicaly based study. 2. Predictivevalue of the early clinical course. Brain 1989;112:1419-28.

41 Weinshenker BG, Bass B, Rice GP, Noseworthy J, CarriereW, Baskerville J, Ebers GC. The natural history of multi-ple sclerosis: a geographically based study. 3. Multi-variate analysis of predictive factors and models ofoutcome. Brain 1991;114:1045-56.

42 Duquette P, Murray TJ, Pleines J, Ebers GC, Sadovnick D,Weldon P, et al. Multiple sclerosis in childhood: clinicalprofile in 125 patients. _Pediatr 1987;3:359-63.

43 Runmarker B, Andersen 0. Prognostic factors in a multiplesclerosis incident cohort with 25 years of follow up. Brain1993;1i6:117-34.

44 Sibley WA, Bamford CR, Clark K, Smith MS, Laguna JF. Aprospective study of physical trauma and multiple sclero-sis. _7 Neurol Neurosurg Psychiatry 199 1;54:584-9.

45 Siva A, Radhakrishnan K, Kurland LT, O'Brien PC,Swanson JW, Rodriguez M. Trauma and multiple sclero-sis: a population based cohort study from OlmstedCounty, Minnesota. Neurology 1993;43: 1878-82.

46 Baskett PJF, Armstrong R. Anaesthetic problems in multiplesclerosis. Anaesthesia 1970;25:397-401.

47 Siemkowicz E. Multiple sclerosis and surgery. Anaesthesia1976;31:1211-6.

48 Sibley, WA, Bamford CR, Clark K. Clinical viral infectionsand multiple sclerosis. Lancet 1985;i:1313-5.

49 Andersen 0, Lygner P-E, Berstrom T, Andersson M,Vahlne A. Viral infections trigger multiple sclerosisrelapses: a prospective seroepidemiological study. .7

Neurol 1991;240:417-22.50 Birk K, Ford C, Smeltzer S, Ryan D, Miller R, Rudick RA.

The clinical course of multiple sclerosis during pregnancyand the puerperium. Arch Neurol 1990;47:738-42.

51 Rouillet E, Verdier-Taillefer M-H, Amarenco P, Gharbi G,Alperovitch A, Marteau R. Pregnancy and multiple scle-

rosis: a longitudinal study of 125 remittent patients. .7Neurol Nurosurg Psychiatry 1993;56: 1062-5.

52 Worthington J, Jones R, Crawford M, Forti A. Pregnancyand multiple sclerosis-a 3 year prospective study. 7Neurol 1994;241:228-33.

53 Korn-Lubetzki I, Kahana E, Cooper G, Abramasky 0.Activity of multiple sclerosis during pregnancy and puer-perium. Ann Neurol 1984;16:228-31.

54 Frith JA, McLeod JG. Pregnancy and multiple sclerosis. ]Neurol Neurosurg Psychiany 1988;51:495-8.

55 Bemardi S, Grasso MG, Bertollini R, Orzi F, Fieschi C.The influence of pregnancy on relapses in multiple scle-rosis: a cohort study. Acta Neurol Scand 199 1;84:403-6.

56 Runmarker B, Andersen 0. Pregnancy is associated with alower risk of onset and a better prognosis in multiple scle-rosis. Brain 1995;118:253-61.

57 Sadovnick AD, Baird PA, Ward RH. Multiple sclerosis;updated risks for relatives. Am _7 Med Genet 1988;29:533-41.

58 Robertson NP, Fraser M, Deans J, Clayton D, CompstonDAS. Age adjusted recurrence risks for relatives ofpatients with multiple sclerosis. Brain 1996;119:449-55.

59 Carton H, Vlietinck R, Debruyne J, De Keyser J,D'Hooghe M-B, Loos R, et al. Recurrence risks of multi-ple sclerosis in relatives of patients in Flanders, Belgium.]Neurol Neurosurg Psychiatry 1997;62:329-33.

60 Ebers GC, Bulman DE, Sadovnick AD, Paty DW, WarrenS, Hader W, et al. A population based study of multiplesclerosis in twins. NEnglJ_Med 1986;315:1638-42.

61 Sadovnick AD, Armstrong H, Rice GPA, Bulman D,Hashimoto L, Paty DW, et al. A population-based study ofmultiple sclerosis in twins: update. Ann Neurol 1993;33:281-5.

62 French Research Group on Multiple Sclerosis. Multiplesclerosis in 54 twinships: concordance rate is indepen-dent of zygosity. Ann Neurol 1992;32:724-7.

63 Mumford CJ, NW Wood, HF Kellar-Wood, J Thorpe,Miller D, Compston DAS. The British Isles survey ofmultiple sclerosis in twins. Neurology 1994,44: 11-5.

64 Ebers GC, Sadovnick AD, Risch NJ. A genetic basis forfamilial aggregation in multiple sclerosis. Nature 1995;377:150-1.

65 Sadovnick AD, Ebers GC, Dyment DA, Risch N, theCanadian Collaborative Study Group. Evidence forgenetic basis of multiple sclerosis. Lancet 1996;347:1728-30.

66 Robertson NP, Clayton D, Fraser MB, Deans J, CompstonDAS. Conjugal multiple sclerosis. Lancet 1997 (in press).

67 Olerup 0, Hillert J. HLA class II-associated genetic suscep-tibility in multiple sclerosis: a critical evaluation. TissueAntigens 1991;38:1-15.

68 Seboun E, Robinson MA, Doolittle TH, Ciulla TA, KindtTJ, Hauser SL. A susceptibility locus for multiple sclero-sis is linked to the T cell receptor [3 chain complex. Cell1989;57: 1095-100.

69 Wood NW, Kellar-Wood HF, Holmans P, Clayton D,Robertson N, Compston DAS. The T-cell receptor /3locus and susceptibility to multiple sclerosis. Neurology1995;45: 1859-63.

70 Walter MA, Gibson WT, Ebers GC, Cox DW.Susceptibility to multiple sclerosis is associated with theproximal immunoglobulin heavy chain region. I ClinInvest 1991;87:1266-73.

71 Wood N, Sawcer SJ, Kellar-Wood H, Holmans P, ClaytonD, Robertson N, Compston DAS. A susceptibility genefor multiple sclerosis linked to the immunoglobulin heavychain variable region. I Neurol 1995;242:677-82.

72 Sawcer S, Jones HB, Feakes R, Gray J, Smaldon N,Chataway J, et al. A genome screen in multiple sclerosisreveals susceptibility loci on chromosome 6p2l and17q22. Nat Genet 1996;13:464-8.

73 Ebers GC, Kukay K, Bulman D, Sadovnick AD, Rice G,Anderson C, et al. A full genome search in multiple scle-rosis. Nat Genet 1996;13:472-6.

74 The Multiple Sclerosis Genetics Group. A completegenomic screen for multiple sclerosis underscores a rolefor the major histocompatibility complex. Nat Genet1996;13:469-71.

75 Davenport CB. Multiple sclerosis from the standpoint ofgeographic distribution and race. Arch Neurol 1922;8:51-8.

76 Sutherland JM. Observations on the prevalence of multiplesclerosis in northern Scotland. Brain 1956;79:635-54.

77 Chakraborty R, Kamboh, MI, Nwankwo M, Ferrell RE.Caucasian genes in American blacks: new data. Am 7Hum Genet 1992;50:145-55.

78 Cavalli-Sforza LL, Menozzi P, Piazza A. The history andgeography of human genes. Princeton: Princeton UniverstyPress, 1994:541.

79 Poser CM. The dissemination of multiple sclerosis: aViking saga? A historical essay. Ann Neurol 1994;36(suppl2):S231-43.

561


http://jnnp.bmj.com

/J N

eurol Neurosurg P


nloaded from


genetic epidemiology of multiple sclerosisdistribution of multiple sclerosis. whereas studies...

Documents