Homogeneity of the age at diagnosis in sibs with type 2 diabetes: Implications for sib-pair analysis

Abstract

Homogeneity of the age at diagnosis of Type 2 diabetes was studied in 1,228 sibs in 300 unrelated families: 100 consecutive single-affected and 200 consecutive multiple-affected ones. There were 635 diabetic sibs. The mean and median age at diagnosis in all affected individuals was 50 years (range, 19-75 years). The mean age at diagnosis in the multiply affected families was 49 years (median 50); the between-sibs range of age at diagnosis within multiple-affected families was (mean and median) 17 years (range, 0-55), with 42% of these diagnosed within a 5-year age span, 66% within 10 years, and 90% within 13 years. When one parent had diabetes, it was more often the mother (79% P = 0.0023). In order to examine this apparent tendency toward homogeneity of age at diagnosis within families, with full regard for and parsimonious to right-censored data, we employed a Cox proportional-hazards survival analysis, with family as the explanatory variable. Deviance residuals resulting from that model were analyzed in a variance components, random effects model ANOVA which indicated a significant (P << 0.001) effect of family on age of diagnosis, with an intraclass correlation of 0.29. In many families the clustering of age at diagnosis appeared very tight, with single outliers, and in 20 families with the longest history, diabetes was diagnosed in 68 sibs within the span of 8 +/- 7 years, whereas 38 unaffected sibs remain free of diabetes 25 +/- 8 years later. The wide differences of the age at diagnosis between families and its intrafamilial homogeneity should be considered in planning genetic analysis of Type 2 diabetes.