This study was designed to examine the hypothesis that some environmental factors increase the risk for insulin-dependent diabetes mellitus. Data on dietary history was collected from 80 diabetic children from the Santiago de Chile Registry and from 85 nondiabetic control subjects who were comparable in terms of age, sex, and ethnic characteristics. Early exposure was defined as the ingestion of food sources other than maternal milk before 3 months of age. To define genetic susceptibility to insulin-dependent diabetes mellitus each subject was typed in terms of HLA DQA1 and DQB1, and the possible conformation of susceptible heterodimers was considered as a risk marker. Fewer children were exclusively breast fed in the diabetic group than in the control group (21.55 +/- 15.05 vs 33.95 +/- 20.40 weeks, P<0.01). In addition, exposure to cow's milk and solid foods occurred earlier in the diabetic group than in the control group (15.90 +/- 10.95 vs 21.15 13.65 and 16.85 +/- 10.25 vs 21.20 +/- 12.35 weeks, P<0.05). Our data show that a short duration of breast-feeding and early exposure to cow's milk and solid foods may be important factors in the development of insulin-dependent diabetes mellitus. The high relative risk observed in individuals genetically predisposed indicates an interaction effect between genetic and environmental components.

HLA-DQ alpha and beta alleles were chosen as the most sensitive Type 1 (insulin-dependent) diabetes mellitus susceptibility markers for evaluating the disease associations and Type 1 diabetes risk in a population-based registry from Madrid. The absence of aspartic acid in position 57 of the DQ beta chain (non-Asp 57), and the presence of arginine in position 52 of the DQ alpha chain (Arg 52) were found to be reliable markers of Type 1 diabetes susceptibility among the Spanish population, with significantly higher frequencies among the cases of Type 1 diabetes compared to randomly selected non-diabetic control subjects from the general Madrid population. While non-Asp 57 homozygosity conferred an absolute risk of 32.3 per 100,000 per year and Arg 52 of 31.5 per 100,000 per year, the risk for double homozygotes for both non-Asp 57 and Arg 52 was estimated as 101.7 per 100,000 per year. Individuals homozygous for only one of these alleles, and heterozygous at the other locus, had a markedly lower Type 1 diabetes risk (12.8 per 100,000 per year), approximating the general population incidence for Madrid. Thus, susceptibility to Type 1 diabetes in Spanish patients is associated, quantitatively, with non-Asp 57 DQ beta and Arg 52 DQ alpha alleles.