Abstract
BACKGROUND: Hispanics/Latinos are a heterogenous population with no validated atherosclerotic cardiovascular disease (ASCVD) risk estimation tool. We examined performance of the pooled cohort equation (PCE) across Hispanic/Latino background groups and quantiles of African, Amerindian, and European genetic ancestry. METHODS: The Multi-Ethnic Study of Atherosclerosis (MESA) was used to evaluate the performance of the non-Hispanic Black (NHB) and non-Hispanic White (NHW) PCE defined by predicted to observed (P/O) ratios of 10-year ASCVD events. Risk calibration was expressed as P/O ratios and risk discrimination was assessed with Harrell's C-statistic based on Cox models. RESULTS: Our study included 966 Hispanics/Latinos [age 58 years at baseline (SD=9); 52 % females], comprising 504 Hispanics/Latinos of Mexican descent (MX), 284 Hispanics/Latinos of Caribbean descent (CA) and 178 Other Hispanics (O). At 10-years, there were 54 ASCVD events: MX (33); CA (13) and O (8). The PCE overestimated ASCVD risk across disaggregated Hispanic/Latino background groups. Both NHW and NHB PCE models performed best with increasing European genetic ancestry (NHW PCE: P/O ratio decreasing from 1.5 to 1.0; NHB PCE: from 2.4 to 1.5 between the 20th and 80th quantile threshold of European genetic ancestry). In contrast, PCE performance worsened with increasing African genetic ancestry (NHW PCE: P/O ratio increasing from 1.5 to 2.6; NHB PCE: from 1.5 to 2.9 between the 20th and 80th quantile threshold of African genetic ancestry). CONCLUSIONS: Disaggregating Hispanics/Latinos by background and ancestry led to variability in PCE performance with greater overestimation of ASCVD risk for those of Caribbean background and those with increasing quantiles of African genetic ancestry.