HLA Antigens and Susceptibility to Juvenile Diabetes: do Additive Relative Risks Imply Genetic Heterogeneity?*

Abstract

The relative risk of B8/B15 heterozygotes for juvenile‐onset diabetes is higher than the risk for people having B8 or B15 alone. This has been cited as evidence for genetic heterogeneity in juvenile diabetes. However, the observed relative risks are compatible with a single susceptibility allele. If disease susceptibility is recessive, for example, then an individual with two disease associated antigens is more likely to be susceptible than an individual with only one associated antigen. The relative risk for an HLA heterozygote should be intermediate between that of the respective homozygotes, so that an interaction effect of two alleles can only be supported if the heterozygote risk is significantly greater than both homozygote risks. The estimated relative risks for B8 and B15 homozygotes, based on data from four different populations, is approximately equal to the risk for B8/B15 heterozygotes. Moreover, disease manifestations which are differentially associated with B8 and B15. such as antibody production to exogenous insulin, may be due to linkage disequilibrium between HLA and other loci which are not directly related to susceptibility of juvenile diabetes. Therefore, while the susceptibility to juvenile diabetes may have several genetic forms, there is no support for distinct B8‐associated and B15‐associated forms of susceptibility.

Population genetic analyses of insulin dependent diabetes mellitus using HLA allele frequencies

“…Falk & Rubinstein (1980) conclude, based on a recessive model and using gene frequencies and a measure of linkage disequilibrium as parameters, that a single susceptibility locus is indicated from the relative risks for IDDM ofthe B8, B15 heterozygote and homozygotes. Curie-Cohen (1981) finds similar results with more conservative calculations of the risks associated with the two antigens. His analyses of several population studies indicate that the disease gene has the same frequency on chromosomes bearing B8 or B15 and that the risk of the two homozygotes is about the same as the heterozygote.…”

“…It has been suggested that within IDDM there is heterogeneity based on the association of certain HLA antigens with specific characteristics of IDDM (Rotter & Rimoin 1978). As has been pointed out, most of these indications have been based on trends and not on significant associations (Curie-Cohen 1981). In our data we found B15 associated with an early age of onset.…”

“…In addition, no significant difference in risk was found between the B8 and B15 homozygotes when compared to each other. Using Curie-Cohen's (1981) method to test for increased risks for the heterozygotes, we failed to find any significant increases.…”

“…Here the numbers were small and may not give an accurate picture of the real associations. Using the methods of Curie-Cohen (1981) with the assumption of a recessive inheritance for the IDDM susceptibility gene in linkage disequilibrium with B8 and B15, we found that the risk for the BS/Bl5 heterozygote was not significantly greater than either homozygote. However, if there are two IDDM genes associated with HLA antigens, then these observations may be due to a low haplotype frequency of the associated B15 susceptibility gene, or a low penetrance of the IDDM gene associated with B15.…”

“…Again, a recessive mode of inheritance was assumed. Curie-Cohen's approach is appropriate to use, since the method is based on the risks relative to one another, and not, as with Falk & Rubinstein's approach on the comparison of the relative risks (Curie-Cohen 1981).…”