Journal of Epidemiology and Community Health, 99, 33, 236-22 Polymorphisms and multiple sclerosis in Orkney DEREK F. ROBERTS AD SURIDER S. PAPIHA From the Department of Human Genetics, University of ewcastle upon Tyne DAVID C. POSKAZER From the eurology Service, Massachusetts General Hospital, Boston SUMMARY Study of the blood group, isoenzyme, and serum protein systems representing polymorphic variants at 23 loci, in a population of multiple sclerosis patients in Orkney, their relatives, and a control series, showed that patients were neither more homozygous nor more inbred than. Any possible association of the disorder with the ABO and rhesus blood groups was not directly causal, but was related to the families of the patients rather than to the patients themselves. In an endeavour to assess the importance of the genetic component in the aetiology, of multiple sclerosis (MS) in the Orkney Islands, which there has its highest reported prevalence, the genetic structure of the population was considered, and by direct methods the coefficients of inbreeding and of kinship in a series of patients and were calculated (Roberts et al., 99). At the same time, to provide an independent appraisal of the role of consanguinity and so to confirm or otherwise the results of the direct analysis, an attempt was made to examine inbreeding indirectly from the study of blood groups and other polymorphic variants. There was a second object of the study. Recent discussion of associations of MS with the histocompatibility antigens (Arnason et al., 92; Bertrams et al., 92; aito et al., 92; Compston et al., 96; Whitaker et al., 96; Opelz et al., 9; Poskanzer et al., 98) has reopened the question of whether patients possess a gene or genes rendering them genetically susceptible to the disease, and whether they represent a subsection of the population that is genetically identifiable. The association of the disease with blood groups has remained controversial throughout the years since it was first suggested (McAlpineetal., 965) but from recent studies it seems that MS patients may be characterised by a slightly higher frequency of blood groups and Rhesus ccddee. The Orkney population is of obvious relevance to this discussion in the light of its geographical situation at the extremes of the European gradients in frequency of these blood groups, and of its high prevalence of MS. The opportunity was therefore taken to examine in 236 Orkney the suggestion from cross-sectional studies that there may be blood group associations with MS, and also to inquire whether these occur within families, for no family study of this kind has yet been made. Material and methods All patients with MS alive on December 9 in the Orkney Islands were identified, and two more in whom the diagnosis was confirmed in 96. This gave a total of patients, of whom were diagnosed as probable cases and eight as possible cases, according to established international criteria. For each patient, except three who were not born in Orkney, an unrelated contiguous control was selected who was born in the same parish in the same year, and had lived in the same area for the first 5 years of life. At the same time a discontiguous control was selected who was of the same age and sex and who was born in an Orcadian parish not contiguous with the birthplace of the patient. All patients, contiguous, and, wherever possible, their close relatives, were examined; they included all available adult first-degree relatives, spouses, and any second-degree relatives who were present at the domiciliary visit, all of whom were unaffected. o relatives were included for the discontiguous. A blood specimen was obtained from each patient, from all except four pairs of, and from 32 family members (including first-degree relatives of patients and first-degree relatives of contiguous ). Comparisons can therefore be J Epidemiol Community Health: first published as.36/jech.33..236 on December 99. Downloaded from http://jech.bmj.com/ on 26 January 29 by guest.
Polymorphisms and multiple sclerosis in Orkney made between the patients, 8 contiguous, 8 discontiguous, all the other unaffected (32) subjects, the first-degree relatives of patients, and the first-degree relatives of contiguous. The blood specimens were tested by standard serological procedures with the following antisera: Anti-A, A,, B, A+B, C, c, D, E, e, M,, S, s, K, k, P, Jka, Jkb, Fya, Fyb, Lua, Lub, and Lea. Red cell isoenzymes were identified by horizontal starch gel electrophoresis, using gels of 9% BDH hydrolised starch. Adenylate kinase was examined by the method of Fildes and Harris (966), acid phosphatase by the discontinuous buffer system of Hopkinson et al. (963), glucose-6-phosphate dehydrogenase by the method of Fildes and Parr (963). and phosphoglucomutase by using double the molar concentration of all ingredients of the buffer given by Spencer et al. (96) at ph -. Adenosine deaminase was typed by the method of Spencer et al. (968). For 6-phosphogluconate dehydrogenase, lactate dehydrogenase, and malate dehydrogenase, a phosphate buffer ph was used, the lower half of the gel slice stained using the reaction mixture of Fildes and Parr (963) and the upper half stained for LDH and MDH as indicated by Kirk et al. (969). Haptoglobin and transferrin were typed using the discontinuous buffer system of Poulik and Smithies (95) at volts for three to four hours in a cold room at C and staining with benzidine to visualise the phenotypes. Altogether, nine blood group systems (2 loci) and isoenzyme and serum protein systems ( loci) were examined. Gene frequencies were calculated on the assumption of Hardy Weinberg equilibrium. Comparison of phenotype frequencies in patients and the several control series was carried out by x2, or where necessary by exact probabilities. The intrafamilial analysis was carried out by the method of Clarke et al. (956) in which each blood group occurring in each affected individual is compared against his own sibs. For each family which is segregating for the blood group in question the chance is assessed of the propositus being of that blood group. In the simple case of an affected member of a sibship of four persons, two of whom are group and two group A, the chance of the patient being group is 5%, and this (the expected) is scored as '5. If in fact the propositus is of group, the observed score is, whereas if he is of group A, the observed score is. For each sibship, therefore, there is recorded an observed and an expected score which can be totalled for all the families being studied. By this method also a variance can be assigned to the expectation, against which the 23 difference between the observed and expected can be compared and its significance tested. The presence and extent of inbreeding may be examined by comparison of. the observed and expected phenotype frequencies in each system, either codominant or partially codominant, where heterozygotes are distinguishable. The F coefficient is estimated by F = - (H/2pq) where H is the observed proportion of heterozygotes, and p and q are the respective gene frequencies. If the population is inbred, this value will be positive. Results In Table are set out the phenotypes observed in the MS patients, compared with the numbers observed in the contiguous, the discontiguous, the two sets of first-degree relatives, and the total unaffected. On inspection of the data, there is only one barely significant difference between the affecteds and the unaffecteds, in the Lutheran-system where the affecteds appear to have an excess of the Lua gene. There is a suggestion that patients of blood group may be slightly in excess by comparison with the discontiguous and the large unaffected sample (9% compared with 3% and %); they share this elevation, however, with the contiguous (52%). In the rhesus system the phenotype ccddee occurs in 5% of affected, 8-% of discontiguous, and 9 % of other unaffecteds, again suggesting a slight elevation in the patients, and again this is shared with the contiguous (3%). It appears that the apparent differences may be due to local variation in ABO and rhesus frequencies within the Orkney Islands. The tendencies, however, are in the same direction as in the earlier suggestions of association. The intrafamilial analysis shows no significant association in any system. For both group and ccddee the observed number of affected is very close to that expected were the association with the disease random within sibships. The number of loci at which each patient was heterozygous was counted, and similarly the number at which he was homozygous. The mean number of loci at which the patients were heterozygous was 2 by comparison with that in contiguous (.2), discontiguous (.2), and all unaffected (.25). The mean number of loci at which homozygosity was established in the patients was, in contiguous -, in discontiguous -26, and in unaffecteds -25. It appears that patients have no excess of homozygosity, and hence that they are unlikely to be more inbred than the. The gene frequencies are set out in Table 2, and from these the expected phenotype frequencies in J Epidemiol Community Health: first published as.36/jech.33..236 on December 99. Downloaded from http://jech.bmj.com/ on 26 January 29 by guest.
238 Table Phenotypes observed System ABO A, A. B AlB AsB Rhesus CcDEE ccdee CCDEe CcDEe ccdee ccddee CCDee CcDce Ccddee ccdee ccddee MS MMSS MMSs MMss MSS MSs Mss SS Ss Affected 26 9-6 3-2 3-2 -9-9 2 22-6 -6 8-9 32-8 5-9 - 3 5-26- 5 28-3 ss 8 5- Fy a+b+ 2-3 a+ba-6+ 8 2-8 3 Jk a+b+ 2-3 a+ba-b+ 8 2-3 Lu a+b+ 8 5- a-b+ a+b- 8-9 p P+ p_ - 2 22-6 KeKl KK Kk kk 5 8 9-9-6 Lewis Le+ Le- 6 29-82-9 35 Condguous 2 2 52-2 26-3 5 6-5 -9 9 9-6 5-2 2-2 26-6 3-3 6-5 5-9 6 3-9 9-6 5 32-6 6 3-3-8 3 29-5 38-6 2-2 2-3 25-2 -5 2 95-5 89-5 -9 98 6 23-2 6-9 26 Derek F. Rdberts, Surinder S. Papiha and David C. Poskanzer Discondguous 3-5 32-6 5-2 5-2 i 2-2 5-2 5-9 2 26-5 32-6 8-5-2 6 3-2- 3-6 3-3- 2 26-3- 6 35-6 5 33-3 3-3 6-2 93-3 3 3-9 2 26-95- 2 25-6 5-8 ot affected, excluding contiguous and discondguous 32 29 6 22 32 3 6 25 23 3 32 5 3 S 66 82 3 6 32 39 3 39 2 82 92 36 9 29 36 265 56 32 2 296 32 8-66 23-3 9-9 8-23 6-9 9-2 relatives of propositi 38 2-2 2-3 6-5 25-6 -3 9-9 3 38-9 52-6. 9-38 5.3 5-6 23 9-3 5-6 2 2-6 26 25-5 33-9 2-2 2 9. 26 38 3-6 5 2-32-3 36 38-5 26-29- 6-93- 36 3 22 36 82-6 - 2-3 -5 92-2 -5 89-5 28 3 2 38 3-8 38 2-3 6-5 6-25 6-5 8-2 -5 5 8- -3 22-5 2 3-9 - 2 2-3 9-3 6- -9-5 2 8-8 25 23-9 2 - -5 3 8-8 26-3 26-2 32-6 39 3 36-8 5 32-25 3-2 3 3 9-6 89-9 - 3 82-6 9-25 9-89-9 relatves of condguous 8 5-2 9-8 52 59 33-2- 9-6 2- - - 2-6 -9-8 9-6 -9-2-6 - -8 3.5-8 8- - 2-23-5-9 2-6 22-6 -6 23-9 3-5 -8 22-3- 3.5 96-5 8-3 2- - 93- -9 88- J Epidemiol Community Health: first published as.36/jech.33..236 on December 99. Downloaded from http://jech.bmj.com/ on 26 January 29 by guest.
Polymorphisms and multiple sclerosis in Orkney Affected System ADA - -2 6PGD AA CA LDH MDH AK -2 2 AP AA BA BB CB CA PGMi - -2 2-2 PGM. - Hp - -2 2-2 eg. Tf cc CB 8 9-6 5 9- - 86-8 3-2 3-2 23 3-5 28-3 6-3 29 5-32- 3-2 - 5 9-8 28 5-9 8 35-3 5 - Contiguous 9-8-9 9-8 - - 9-8-9 6 3-3 22 8-9 3 28-9 3 6-2 6-3-8 3 6-8 - 2 26-9 -3 3- - Discontiguous 9-8-9 9-8 - - 38 8-5-6 5-22 3 8-9 28-9 3 6-3 68-2 9 2-5 - - 6 3-3 2 5-8 3 28-9 9-8 each system were calculated. Comparisons of the proportion of heterozygotes observed with that expected gave the F coefficients as estimated from each system, as set out in Table 3. For the 5 estimates in the patients only four are positive, and,none are* significant. In the contiguous six are positive, and two are significant (Duffy and ABO groups). In the discontiguous again six are positive and two are significant (Kidd and PGM), while in the three groups combined four are positive and three are significant (M, Duffy, and PGM). This distribution is reflected in the mean F (Table 3), ot affected, excluding contiguous and discontiguous 2 8-9 38 2-3 38 98-6 -9 3 3-3 - 28 29 3 33 5 3 9-5 9-2 -3 8 5-2 5 3-28 8-3 3-6 39 3 36 3 S 36 relatives of propositi ( 2 9-3 3 9-33 99-3 - - - 8-3 2-3- 2-2- 35-36-3 58 - -8 5 3-3-5 5-2 9- -3 35-8 -3 98- -6 9 59-9 3-9 5-3 32-22 6-5 2-5 5-3 - - which is negative in both the affected (--29) and the contiguous (-.) by comparison with the positive value in the discontiguous (+.352) and overall (+-3). Again it appears from this indirect analysis that the patients are not more inbred than the. Discussion relatives of contiguous 239 The indirect method of estimating inbreeding levels from relevant polymorphic systems is crude and, particularly when sample sizes are small, it is liable to 3 89-6 2-9 95-6 5 - - - 3 9-9-6 6-52 -6 39 3 3-9 2-6 3-5 6 5-36 6 3-9 -2 3-2 56 33 2-9- 29-8 - 96-5 3-5 J Epidemiol Community Health: first published as.36/jech.33..236 on December 99. Downloaded from http://jech.bmj.com/ on 26 January 29 by guest.
2 Table System ABO Rhesus MS Fy Jk Lu P Kell ADA 6PGD LDH MDH AK AP PGM PGMs Hp Tf Gene combinatons Rhesus CDe cde cde cde Cde CDE cde MS MS S Ms s 2 Gene frequencies pi Ps q r C c D d E e m n s a b a b a b P p K k ADA' ADA' PGDA PGDC AK' AK' pa pb PGM' PGMI PGM, HpI Hp' Tfc Tfb Affected -8-26 -89 - -2-8 -62-388 -89-8 -55-3 -292-8 -3-566 -3 566-5 -925-52 -6 - -9-9 - -93-66 -368-55 -5-8 -292-33 -62-2 -89-39 -226 - -3 - Derek F. Roberts, Surinder S. Papiha and David C. Poskanzer Contiguous -56 - -2-28 - -55-639 -36-96 -8-58 -3-26 -39-5 -5-5 -89-23 -9-6 -33 - -989-956 - -989 - -956 - -389-56 - -3-22 -8-522 - -96-332 -2-2 -6-35 -35 give estimates grossly at variance with true population levels. For that reason little can be extracted from the results other than the overall impressions given here, that there is no evidence that Discontiguous -2-23 -23-5 -5-8 -6-33 -85-85 -598-2 -25-5 - -55-5 -89-33 -96-89 -5-22 -98-956 - -989 - -922-8 -38-56 -56-8 -26-22 -58-989 - -8-239 -283-26 - -3-3 ot affected, excluding contiguous and discontiguous -9-68 - -69-83 -5-65 -325-6 -8-5 -9-269 -3-9 -5-85 -55-33 -96-582 -8 - -96-9 -6-99 - -95-9 -36-599 - -3-25 -6-58 -992-8 -8-36 -3-6 - - -2-23 - -3-2 the patients in Orkney are more inbred than the. This interpretation is supported by the straight counts of loci at which each subsample is homozygous or heterozygous. These findings support J Epidemiol Community Health: first published as.36/jech.33..236 on December 99. Downloaded from http://jech.bmj.com/ on 26 January 29 by guest.
Polymorphisms and multiple sclerosis in Orkney 2 Table 3 Inbreeding estimates Affected Contiguous Discontiguous Combined F x' F X2 F X2 F X2 AB --9-56 -9 +-96 - --8 93 M --85-82 --66-2 --3-8 --6 3-3 Ss --9-3 +-2 2- + 5 +. Cc -- -5 +-6-2 - - - Ee +- --6-5 +-62 8 - Fya+b+ +-8-32 +-358 5 9 +-252 2-92 + 22-8 Jka+b+ + 8 32 + 9 + 288 3 8 + 2 83 Lua+b+ - 83 36-25 3 --36-6 --9-3 Kk --5-3 --8 - --25-3 --28. ADA 2- --5. --8. --8. --8 33 6PGD(CA) - --2 - --2. --6. AK 2- -- -26 --8. --8 33 --6 65 AP(BA) --6-9 --9 55 --2 93 --2-5 PGM 2- + 225 2-68 +-9 - +-396-2 +-2 8 35 Hp 2- --5-62 +-2-3 --23.9 -- -8 Mean F --29 --5 +-35 + the direct observations of consanguinity and inbreeding in the Orkney patients (Roberts et al., 99). There are few other data in the literature on gene frequencies in Orkney, and such as they are, these relate only to the ABO blood group system. Allan and Lewis (969) gave frequencies of the A gene, B gene, and gene respectively of 2-%, 9*6%, and 69-3%; Brown (965) gave 22-9%, -3%, and 65*8%; and Boyce et al. (93) gave 8-%, -%, and %. All show an elevation of the B gene frequency which distinguishes the Orcadians from the remainder of Britain. There is general agreement with these figures in the present series. Boyce also notes from unpublished data that there appear to be gene frequency differences between the different islands of Orkney, and this would partly explain the greater similarity of our patients to contiguous than to discontiguous. The first suggestion that susceptibility to MS may be associated with the patient's blood group was made in 965 by McAlpine et al., who drew together data on ABO blood groups of 38 patients from various centres in Britain, and showed a frequency of blood group among the MS patients -% higher than among the. An earlier study by Alexanderetal. (95) showed a similar slight excess of group of *9%. Simpson et al. (965) in the north-east of England were unable to detect any significant difference, but again in their series there was a slight excess of blood group amounting to -% overall. In a subsequent study in the north-east of England, MacDonald et al. (96) showed an excess of that was significant compared with one normal series but not with another. Similarly, they showed a significant difference in the rhesus system, particularly striking in the excess of cde homozygotes. In the light of these findings, the current results are of interest in that the suggestions are in the same direction although they do not reach significance. Here, however, sample size is critical, for from the nature of blood group data there would need to be very gross differences in gene frequencies for them to be detected in a small sample of patients. However, the present study shows that it is not so much the features of the patients themselves that are responsible for the tendency. For when the first-degree relatives of propositi and the first-degree relatives of are selected from among the unaffected series (Table ), it is obvious that the former share elevated frequencies of and cde homozygous phenotypes by comparison with the first-degree relatives of. This suggests that it is the families themselves, rather than the patients, who are characterised by this excess. A similar finding comes from the intrafamilial analysis, which indicates that it is the sibships rather than the patients that are so characterised. These results suggest that the association, if it exists, between MS and the ABO or rhesus alleles is not directly causal. Instead, the association appears to be familial, so that the population consists of a number of family groups where blood group and a higher susceptibility to MS are associated together. This is compatible with a polygenic hypothesis of genetic involvement in MS in which the gene or a cde combination are pleiotropic and their presence increases the liability to develop the disease, as well as exerting their own serological effect. One possible interpretation of the previously suggested associations with and rhesus ccddee blood groups, that it reflects homozygosity, is eliminated by the present material, for it appears that patients have no excess of homozygosity, and hence that they are unlikely to be more inbred than the J Epidemiol Community Health: first published as.36/jech.33..236 on December 99. Downloaded from http://jech.bmj.com/ on 26 January 29 by guest.
22. This finding supports the direct observations of consanguinity and inbreeding in the Orkney patients (Roberts et al., 99). Despite the comprehensive nature of the present study, the possibility of association of MS with the ABO and rhesus blood groups remains neither established nor disproved. The results suggest, however, that if it in fact occurs, then it is at the familial rather than the individual level, that a number of family groups within the population are particularly liable to develop the disorder, and this is compatible with a polygenic hypothesis of genetic involvement in multiple sclerosis. We thank Mr. C. K. Creen, Miss R. A. ahar, Mrs. V. Lamb, and Mrs. A. Stephenson. This study was supported by the ational Institute of eurological and Communicative Disorders and Stroke (contracts IH-ICDS O-S--232 and O-S-6-233). Reprints from Professor D. F. Roberts, Department of Human Genetics, University of ewcastle upon Tyne, ewcastle upon Tyne EI RU. References Alexander, L., Loman, J., Lesses, H. F., and Green, I. (95). Blood groups and multiple sclerosis. Research Publications of the Association for Research in ervous and Mental Disease, 28, 9-2. Allan, T. M., and Lewis, B. (96). In The Distribution of the Human Blood Groups and Other Polymorphisms, second edition. Edited by A. E. Mourant, A. C. Kopec, and K. Domaniewska-Sobczak. Oxford University Press: London. Arnason, B. G. W., Fuller, T. C., Lerich, J. R., and Winn, H. J. (92). Leucocyte antigens (HL-A) in multiple sclerosis. In Transplantation Abstracts, th International Congress of the Transplantation Society, San Francisco. Edited by S. L. Kountz. Grune and Stratton: ew York. Bertrams, J., Kuwert, E., and Liedtke, U. (92). HL-A antigens and multiple sclerosis. Tissue Antigens, 2, 5-8. Boyce, A. J., Brothwell, D. R., and Holdsworth, V. M. L. (93). Demographic and genetic studies in the Orkney Islands. In Genetic Variation in Britain. Edited by D. F. Roberts and E. Sunderland. Taylor & Francis: London. Brown, E. S. (965). Distribution of the ABO and Rhesus (D) blood groups in the orth of Scotland. Heredity, 2, 289-33. Derek F. Roberts, Surinder S. Papiha and David C. Poskanzer Clarke, C. A., Edwards, J. W., Haddock, D. R. W., Howel-Evans, A. W., McConnell, R. B., and Sheppard, P. M. (956). ABO blood groups and secretor character in duodenal ulcer. Population and sibship studies. British Medical Journal, 2, 25-3. Compston, D. A. S., Batchelor, J. R., and McDonald, W. I. (96). B-lymphocyte alloantigens associated with multiple sclerosis. Lancet, 2, 26-265. Fildes, R. A. and Harris, H. (966). Genetically determined variation of adenylate kinase in man. ature, 29, 26-263. Fildes, R. A., and Parr, C. W. (963). Human red cell phosphogluconate dehydrogenase. ature, 2, 89-89. Hopkinson, D. A., Spencer, M., and Harris, H. (963). Red cell acid phosphatase variants: a new human polymorphism. ature, 99, 969-9. Kirk, R. L., Blake,. M., Lai, L. Y. C., and Cooke, D. R. (969). Population genetic studies. Australian Aborigines of the orthern Territory, The distribution of some serum protein and enzyme groups among the Malag of Elcho Island. Archaeology and Physical Anthropology in Oceania,, 238-25. McAlpine, D., Lumsden, C. E., and Acheson, E. D. (965). Multiple Sclerosis, a Reappraisal, second edition. Churchill Livingstone: Edinburgh and London. MacDonald, J. L., Roberts, D. F., Shaw, D. A., and Saunders, M. (96). Blood groups and other polymorphisms in multiple sclerosis. Journal of Medical Genetics, 3, 3-33. aito, S., amerow,., Mickey, M. R., and Terasaki, P. (92). I: Multiple sclerosis: association with HL-A3. Tissue Antigens, 3, -. Opelz, G., Terasaki, P., Myers, L., Ellison, G., Ebers, G., Zabriskie, J., Weiner, H., Kempe, H., and Sibley, W. (9). Association of HLA antigens A3, B, and DW2 with 33 multiple sclerosis patients in the United States. Tissue Antigens, 9, 5-58. Poskanzer, D. C. Terasaki, P. I., Park, M. S., Prenney, L. B., and Sheridan, J. L. (98). Multiple sclerosis in Orkney and Shetland. I: Histocompatibility determinants. (In preparation). Poulik, M. D., and Smithies,. (95). Starch gel electrophoresis in a discontinuous system of buffer. ature, 8,. Roberts, D. F., Roberts, M. J., and Poskanzer, D. C. (99). Genetic analysis of multiple sclerosis in Orkney. Journal of Epidemiology and Community Health, 33, 229-235. Simpson, C. A., Vejjajiva, A., Caspary, E. A., and Miller, H. (965). ABO blood groups in multiple sclerosis. Lancet,, 366-36. Spencer,., Hopkinson, D. A., and Harris, H. (96). Phosphoglucomutase polymorphism in man. ature, 2, 2-. Spencer,., Hopkinson, D. A., and Harris, H. (968). Adenosine deaminase polymorphism in man. Annals of Human Genetics, 32, 9-. Whitaker, J.., Herrmann, K. L., Rogentine, G.., Stein, S. F., and Killins, L. L. (96). Immunogenetic analysis and serum viral antibody titers in multiple sclerosis. Archives of eurology, 33, 399-3. J Epidemiol Community Health: first published as.36/jech.33..236 on December 99. Downloaded from http://jech.bmj.com/ on 26 January 29 by guest.