The EPIC-Norfolk Study
Urinary bisphenol A concentration and risk of future coronary artery disease in apparently healthy men and women.
Circulation 2012 ; 125: 1482-90.
Melzer D, Osborne NJ, Henley WE, Cipelli R, Young A, Money C, McCormack P, Luben R, Khaw KT, Wareham NJ, and Galloway TS
DOI : 10.1161/CIRCULATIONAHA.111.069153
PubMed ID : 22354940
PMCID : 0
Abstract
The endocrine-disrupting chemical bisphenol A (BPA) is widely used in food and beverage packaging. Higher urinary BPA concentrations were cross-sectionally associated with heart disease in National Health and Nutrition Examination Survey (NHANES) 2003-2004 and NHANES 2005-2006, independent of traditional risk factors.
We included 758 incident coronary artery disease (CAD) cases and 861 controls followed for 10.8 years from the European Prospective Investigation of Cancer-Norfolk UK. Respondents aged 40 to 74 years and free of CAD, stroke, or diabetes mellitus provided baseline spot urine samples. Urinary BPA concentrations (median value, 1.3 ng/mL) were low. Per-SD (4.56 ng/mL) increases in urinary BPA concentration were associated with incident CAD in age-, sex-, and urinary creatinine-adjusted models (n=1919; odds ratio=1.13; 95% confidence interval, 1.02-1.24; P=0.017). With CAD risk factor adjustment (including education, occupational social class, body mass index category, systolic blood pressure, lipid concentrations, and exercise), the estimate was similar but narrowly missed 2-sided significance (n=1744; odds ratio=1.11; 95% confidence interval, 1.00-1.23; P=0.058). Sensitivity analyses with the fully adjusted model, excluding those with early CAD (<3-year follow-up), body mass index >30, or abnormal renal function or with additional adjustment for vitamin C, C-reactive protein, or alcohol consumption, all produced similar estimates, and all showed associations at P≤0.05.
Associations between higher BPA exposure (reflected in higher urinary concentrations) and incident CAD during >10 years of follow-up showed trends similar to previously reported cross-sectional findings in the more highly exposed NHANES respondents. Further work is needed to accurately estimate the prospective exposure-response curve and to establish the underlying mechanisms.