Letters to the editor chi-square value of 5 x 10 -4, (not significant). Comparison of the distribution of Cw7, B8, DR3, DQ2 haplotypes yields a chi-square value of 6 x 10 -3 (not significant). If HLA-A really does contribute to diabetes susceptibility in addition to the other loci on the extended HLA haplotype, a much larger study would be required to demonstrate statistical significance.
The authors state correctly that a powerful method of analysing whether HLA-A alleles predispose to IDDM independently of class II ',alleles is to examine the distribution of class I alleles in those diabetic subjects who possess neither of the high-risk class II alleles encoding DR3 and DR4. The authors found that 47 of 55 diabetic subjects possess two copies of the high-risk HLA-A alleles compared with 40 of 58 control subjects, X 2 = 4.33, (p ,,-0.05). This weak association is interesting but the authors do not show the data for the remaining subjects who possess one or none high-risk HLA-A alleles. Unless there is any reason to suggest that high-risk HLA-A alleles predispose to IDDM recessively we believe that the entire distribution of high-risk A alleles should be examined before any