LAUSR

Epidemiological studies suggest that coffee intake (CI) is protective against body weight gain. We explored whether genetic determinants of CI are associated with obesity-related phenotypic traits, primarily body mass Index (BMI). We leveraged information from ∼354,000 individuals in the UKBB database (https://genetics.opentargets.org/) searching for genetic variants associated with CI (cutoff P < 0.5E-8). We further explored the association of these variants with BMI and other obesity-related traits (body fat percentage-BFP, obesity or waist circumference-WC) using summarized data from Neale's lab (http://www.nealelab.is/uk-biobank/). Twenty seven variants were significantly associated with CI, including rs2472297-CYP1A1/2 (P = 3.4E-116, beta(b) = 0.047) and rs4410790-AHR- (P = 3.2E-95, b = 0.039), which were previously reported to be associated with CI. Seventeen variants showed significant associations with BMI in the same direction (i.e., rs2472297- CYP1A1/2, P = 1.2E-6, b = 0.06; rs4410790-AHR, P = 1.7E-4, b = 0.04; rs589500-SEC16B, P = 2.5E-59, b = 0.22; rs1260326-GCKR, P = 8.0E-6, b = 0.05; rs3814424-MEF2C, P = 6.3E-20, b = 0.14; rs1189470082-AL355997.1, P = 4.3E-8, b = 0.07; rs9398171-LINC00222/FOXO3, P = 6.1E-10, b = 0.08; rs370535199-KBTBD2, P = 9.2E-5, b = 0.045; rs1057868-POR, P = 9.9E-6, b = 0.055; rs56094641-FTO, P = 2.1E-219, b = 0.26). The remaining 9 variants showed no associations with any obesity-related trait. One variant (rs57918684-MED13) showed a marginal and opposite effect. Beta coefficients for CI and BMI were significantly correlated (Spearman R: 0.69, P < 0.0001), which is compatible with a significant genetic correlation between both traits (rg = 0.24 ± 0.02, P = 4.06E-23). The positive association between CI and BMI is biologically supported by genetic correlations between CI and food intake (rg = 0.26 ± 0.07, P = 1.0E-4), BFP (rg = 0.16 ± 0.02, P = 2.54E-13) and WC (rg = 0.23 ± 0.02, P = 2.0E-22). Variants associated with CI present direct pleiotropic effects on obesity-related traits such as BMI, BFP, and WC. If these are not causal relations, then from a Mendelian Randomization point of view, CI has an undesirable effect. Supported by grants from the National Agency for Scientific and Technological Promotion and the National Scientific and Technical Research Council (Argentina).