Abstract P3049: Metabolic Measures and Incident Cardiovascular Disease and Mortality: A Pooled Analysis of Multi-ethnic NHLBI Cohorts

Abstract

Background: The added roles of metabolic measures in health outcomes in racial/ethnic groups in the US remain poorly understood. Pooling data from multi-ethnic cohorts can improve the power to conduct detailed assessment of health outcomes across different racial/ethnic groups. Objectives: We examined associations of metabolic measures (fasting plasma glucose, log-fasting insulin, log-HOMA-β, log-HOMA-IR and triglyceride glucose index [TGI]) with incident cardiovascular disease (CVD) and mortality in the US and examined the differences in strength of associations across racial/ethnic groups. Methods: We pooled individual-participants' data of 23,896 individuals (mean age 49.1years, 45.3% men) from 5 NHLBI cohorts (ARIC, CARDIA, CHS, JHS and MESA) without CVD or diabetes at baseline. Associations between metabolic measures and incident CVD were determined using Fine and Gray models adjusted for confounders. Results: Over a median period of 19 years (13.9-30.9), there were 4800 first-ever incident CVD and 7976 deaths. All metabolic measures showed positive log-linear association with incident CVD (aHR (95% CI) per SD: FPG 1.10 [1.06-1.15]; log-fasting insulin [1.18; 1.13-1.23]; logHOMA-β [1.15; 1.10-1.20]; logHOMA-IR [1.19; 1.14-1.24]; TGI [1.24; 1.19-1.29]) and all-cause mortality (aHR (95% CI) per SD: FPG 1.10 [1.07-1.13]; log-fasting insulin [1.14; 1.11-1.18]; logHOMA-β [1.11; 1.08-1.15]; logHOMA-IR [1.15; 1.11-1.19]; TGI [1.13; 1.09-1.16]). Associations with incident CVD were comparable across racial/ethnic groups. However, associations of each of FPG, log-fasting insulin and logHOMA-IR with incident mortality was stronger in Whites than other racial/ethnic groups. Adjustment for blood pressure, lipids, serum creatinine and adiposity did not explain the racial/ethnic differences in associations of these measures with mortality (FPG X 2 =7.6, p- Interaction=0.05); log-fasting insulin X 2 =8.8, p- Interaction=0.03) and logHOMA-IR X 2 =10.6, p- Interaction=0.01). Conclusion: There is need for further studies to investigate the putative biological mechanisms underlying the observed ethnic differences in association of metabolic measures with mortality in the US.