Ancestry as a determinant of mean population C-reactive protein values: implications for cardiovascular risk prediction.

BACKGROUND

Eligibility for rosuvastatin treatment for cardiovascular disease prevention includes a C-reactive protein (CRP) concentration >2 mg/L. Most observational studies of CRP and cardiovascular disease have been in Europeans. We evaluated the influence of ancestry on population CRP concentration to assess the implications for statin targeting in non-Europeans.

METHODS AND RESULTS

In a systematic review and meta-analysis among 221 287 people from 89 studies, geometric mean CRP was 2.6 mg/L (95% credible interval, 2.27 to 2.96) in blacks resident in the United States (n=18 585); 2.51 mg/L (95% CI, 1.18 to 2.86) in Hispanics (n=5049); 2.34 mg/L (95% CI, 1.99 to 2.8) in South Asians (n=1053); 2.03 mg/L (95% CI, 1.77 to 2.3) in whites (n=104 949); and 1.01 mg/L (95% CI, 0.88 to 1.18) in East Asians (n=39 521). Differences were not explained by study design or CRP assay and were preserved after adjustment for age and body mass index. At age 60 years, fewer than half of East Asians but more than two thirds of Hispanics were estimated to have CRP values exceeding 2 mg/L. HapMap frequencies of CRP polymorphisms known to associate with CRP concentration but not coronary heart disease events differed by ancestry. In participant data from the Wandsworth Heart and Stroke Study including European, South Asian and African, and Caribbean-descent subjects, body mass index, systolic blood pressure, and smoking contributed to between-group differences in CRP, but the majority of the difference in CRP was unexplained.

CONCLUSIONS

Differences in CRP concentration in populations of diverse ancestry are sufficiently large to affect statin eligibility, based on a single CRP threshold of 2 mg/L, and only partially influenced by differences in variables related to cardiovascular risk. A single threshold value of CRP for cardiovascular risk prediction could lead to inequalities in statin eligibility that may not accurately reflect underlying levels of cardiovascular risk.