Effect modification by NT-proBNP on the association between adiponectin and total mortality in EPIC–Heidelberg: A case-cohort study

BackgroundNT-proBNP has been hypothesized as a possible explanation for the paradoxical association between adiponectin and cardiovascular and all-cause mortality. We examined the heterogeneities by NT-proBNP, sex, BMI, smoking status, hypertension and diabetes status in the association between adiponectin and cardiovascular disease risk and mortality. MethodsWe used a case-cohort design nested within the EPIC-Heidelberg cohort, including 1,387 incident cases of myocardial infarction or stroke, 582 deaths from cardiovascular causes and 2,352 total deaths. We estimated hazard ratios for the association between 1SD increase in log-transformed adiponectin levels and cardiovascular disease risk, cardiovascular mortality and total mortality using Prentice-weighted Cox-proportional hazard models and assessed heterogeneity of the associations across strata of covariates. ResultsOverall, adiponectin was significantly associated with all-cause mortality [HR = 1.09, 95%CI: 1.03-1.16, p = 0.004], the association with cardiovascular mortality did not reach statistical significance [1.10 (0.99-1.37), p = 0.073] whereas there was no association with cardiovascular disease risk. There was significant heterogeneity by NT-proBNP in the association between adiponectin and all-cause mortality (phet = 0.019) such that significant increase in hazards of mortality were restricted to participants in the highest tertile of NT-proBNP. Among participants in the highest tertile NT-proBNP, adiponectin showed a dose-response relationship with total mortality such that; compared to participants in the lowest quintile, those in the third, fourth and fifth were at 1.22 (0.87-1.70), 1.50 (1.07-2.11), and 1.59 (1.15-2.21) higher hazards of mortality respectively. ConclusionsSignificant association between adiponectin and mortality was only observed in the context of high NT-proBNP. Our findings provide further support for hypothesis that NT-proBNP may explain the adiponectin paradox.