A new predictive model for insulin-dependent diabetes mellitus susceptibility based on combinations of molecular HLA-DRB1 and HLA-DQB1 pockets.

With a view to establishing an accurate evaluation of the genetic predisposition to insulin-dependent type I diabetes (IDDM), we have built a model based on the characteristics of the relevant pockets of HLA-DR and -DQ molecules. Three independent populations were investigated. Group I and group II were Caucasoids, while group III was Japanese, including a total of 1,166 IDDM patients and 2,391 healthy controls. We formulate the hypothesis that suceptibility to IDDM is not only explained by the absence of Aspartate 57 (negative charge) from pocket 9 of DQB1 (P9DQ), but also by the presence of an electric charge (+/- vs. neutral), generated by residues 70, 71 and 74 in pockets 4 of DRB1 (P4DR) and DQB1 (P4DQ) molecules. The respective weight of each pocket, was evaluated in a multivariate analysis based on the logistic regression method. The 4 components (2 loci and 2 pockets) were systematically analysed in the computer model. It was clearly shown that the structural characteristics of pockets P9DQ-P4DR and, to a lesser degree that of P4DQ, account for IDDM predisposition. On applying the model to the whole international series, it appears that the highest risk concerns individuals with P9DQ non-Asp 57 and both the charged P4 of DRB1 and P4 of DQB1, conferring a 80% prediction of susceptibility. Conversely, P9DQ Asp and neutral P4DR and P4DQ give the lowest risk with a predictive value of 5%. This model of risk susceptibility prediction fits remarkably well with the observed distribution in a worldwide study. It allows a better evaluation of the respective role of HLA-DR and -DQ molecules as a major component of susceptibility to IDDM.