Fifty-three French families were typed for alleles at seven loci of the HLA complex (HLA-A, -B, -C, -DR, -Bf, -C2 and -GLO) and 212 haplotypes were demonstrated. Eleven recombinations were observed (two A/B, two A/C, two B/Bf, one Bf/D and four D/GLO). The linkage disequilibrium was calculated not only between two alleles (delta) but between three, four...seven alleles (D). In order to compare the intensity of D values in the various haplotypes, the influence of the differences in gene frequencies was eliminated by the introduction of the standardized Ds (Ds = D/D max). The number of haplotypes in disequilibrium is relatively limited since most of the significant Ds involved about 17 haplotypes. For some haplotypes, the disequilibrium covered the whole distance from A to GLO but the stronger disequilibrium concerns the C to Bf or C to DR segment. Three hypotheses (isolation, admixture of population and selection) concerning the formation and maintenance of the disequilibria are discussed.

To determine whether genetic markers can improve the predictive value of islet cell antibodies for development of Type 1 (insulin-dependent) diabetes mellitus, 536 siblings aged 2-29 years were consecutively enrolled in a 8-year prospective survey. The risk of developing diabetes was estimated, using life-table methods, by years of follow-up and age, according to genetic factors (shared HLA-haplotypes, DR antigens, C4 allotypes) and islet cell antibody status. Fifteen siblings (2.8%) developed Type 1 diabetes during the study period (risk 4.4% after 8 years, 4% by age 22 years). DR3,4 heterozygosity identified higher risk (16% after 8 years, 12% by age 22 years, p less than 10(-5)) than HLA-identity (10% and 7%, respectively, p less than 0.01); risks for DR3 or DR4 positive and for haplo-identical siblings were low (4%, 3% and 4.4%, respectively, NS). C4BQO also conferred significant risk (11% vs 3% in non-C4BQO siblings, p less than 0.01). The predictive value of genetic markers alone was poor (12% for DR3,4, 7% for HLA-identity, 9% for C4BQO) compared with that of islet cell antibody levels greater than 4 Juvenile Diabetes Foundation units (41%, risk 56% after 8 years, p less than 10(-7)). HLA markers significantly contributed to risk prediction in combination with islet cell antibodies: islet cell antibody-positive DR3,4+ subjects had the highest risk (70% after 8 years, predictive value 58%, p less than 10(-7)) compared with islet cell antibody-positive DR3,4- (37% and 20%, respectively) and islet cell antibody-negative DR3,4+ (5% and 3.6%, respectively).(ABSTRACT TRUNCATED AT 250 WORDS)

HLA genotypes were ascertained in 53 French Caucasian families, comprising 68 juvenile onset insulin-dependent diabetic siblings. Among the 49 alleles detected at different loci in the HLA complex (A, C, B, Bf, DR) 4 appeared to occur at a significantly higher frequency among the 53 index cases than in a control series of 116 healthy individuals: HLA-B18 (p < 10(-3)), DRw3, DRw4 and BfF1 (p < 10(-6)). The excess of HLA identical affected siblings confirms genotype disequilibrium and supports the hypothesis of an HLA-linked gene(s) conferring susceptibility. There was no increase of homozygosity for HLA DRw3 and DRw4 whereas there was a marked excess heterozygosity for HLA DRw3/DRw4 in diabetic patients (32% versus 0% in the control series, p < 0.001). These data provide evidence for the existence of two cooperating genes, linked to each of the HLA DR alleles.