Epistatic interactions of CDKN2B-TCF7L2 for risk of type 2 diabetes and of CDKN2B-JAZF1 for triglyceride/high-density lipoprotein ratio longitudinal change: evidence from the Framingham Heart Study

Fifteen known type 2 diabetes (T2D) gene variants were assessed for their associations with T2D status in 228 T2D families from the Framingham Heart Study (FHS) Original, Offspring, and Children Cohorts. Bayesian approach was used to test single-single-nucleotide polymorphism (SNP) association followed by logistic regression. Bayesian and logic regression approaches were used to test multiple SNP association searching for the best combinations of variants followed by logistic regression reconfirmation. The significant variants for T2D risk were also tested for their main and interacting effects on triglyceride (TG)/high-density lipoprotein (HDL) ratio change derived from four point measures across time. This slope phenotype was made available using mixed model growth curve approach from 155 T2D families in the FHS Offspring Cohort. RESULTS: CDKN2B rs10811661 (p = 0.042), TCF7L2 rs4506565 (p = 0.004), and JAZF1 rs864745 (p = 0.04) were individually associated with risk of T2D (OR = 1.0-2.0; effect size <1%). CDKN2B and TCF7L2 were found with significant main (p = 0.02, 0.01) and interacting (p = 0.05) effects for increased (OR = 3.0) risk of T2D. CDKN2B and JAZF1 were found with significant main (p = 0.0002 and 0.034) and interacting (p = 0.001) effects on increased (β = 0.42) TG/HDL ratio longitudinal change. These interacting effects were independent of effects of age and sex with effect sizes of 0.3-0.4% for risk of T2D or TG/HDL ratio longitudinal change. CONCLUSION: These synthetic approaches allowed for successful detection of CDKN2B and TCF7L2 interacting effect for T2D risk and CDKN2B and JAZF1 interacting effect on TG/HDL ratio increase over time among T2D families in the FHS. These interacting effects were consistent in conferring risk of T2D or progressive insulin resistance with modest effect sizes.