Assessing age-related etiologic heterogeneity in the onset of islet autoimmunity

Abstract

Type 1 diabetes (T1D), a chronic autoimmune disease, is often preceded by a preclinical phase of islet autoimmunity (IA) where the insulin-producing beta cells of the pancreas are destroyed and circulating autoantibodies can be detected. The goal of this study was to demonstrate methods for identifying exposures that differentially influence the disease process at certain ages by assessing age-related heterogeneity. The Diabetes Autoimmunity Study in the Young (DAISY) has followed 2,547 children at increased genetic risk for T1D from birth since 1993 in Denver, Colorado, 188 of whom developed IA. Using the DAISY population, we evaluated putative determinants of IA, including non-Hispanic white (NHW) ethnicity, maternal age at birth, and erythrocyte membrane n-3 fatty acid (FA) levels, for age-related heterogeneity. A supremum test, weighted Schoenfeld residuals, and restricted cubic splines were used to assess nonproportional hazards, that is, an age-related association of the exposure with IA risk. NHW ethnicity, maternal age, and erythrocyte membrane n-3 FA levels demonstrated a significant age-related association with IA risk. Assessing heterogeneity in disease etiology enables researchers to identify associations that may lead to better understanding of complex chronic diseases.

Figure 1

Flow chart illustrating the formation of the cohorts for the investigation of age-related heterogeneity.

Figure 2

The weighted Schoenfeld residual plots are displayed for non-Hispanic white ethnicity (NHW) (a) and a 5-year difference in maternal age (b) in the prospective DAISY cohort. The x-axis represents age in years and the y-axis represents the coefficient estimate for non-Hispanic white ethnicity in (a) and coefficient for maternal age in (b). The dots represent the residuals for each individual. The solid line is a smoothing-spline fit to the plot, with the dashed lines representing the 95% confidence interval. The global PH test P values based on the Schoenfeld residuals are 0.02 and 0.01 for NHW ethnicity and a 5-year difference in maternal age, respectively, indicating a violation of the PH assumption.

Figure 3

A restricted cubic spline model was used to estimate the hazard ratio as a function of age. The restricted cubic spline for non-Hispanic white ethnicity exhibits an increased risk of islet autoimmunity (IA) early on that then becomes protective in the older ages (a). The restricted cubic spline for a 5-year difference in maternal age exhibits a slightly elevated risk of IA in early childhood that becomes protective in adolescence (b). The x-axis represents age in years and the y-axis represents the hazard ratio on the log scale. The solid line represents the hazard ratio and the dashed lines represent the pointwise 95% confidence bands.

Figure 4

The restricted cubic spline function for erythrocyte membrane n-3 fatty acid levels and islet autoimmunity development in the prospective DAISY cohort displays a linear decrease in risk across childhood. The x-axis represents age in years and the y-axis represents the hazard ratio on the log scale. The solid line represents the hazard ratio and the dashed lines represent the pointwise 95% confidence bands.