Predictive value of long-term changes of growth differentiation factor-15 over a 27-year-period for heart failure and death due to coronary heart disease

Background Growth differentiation factor-15 (GDF-15), Cystatin C and C-reactive protein (CRP) have been discussed as biomarkers for prediction of cardiac diseases. The aim of this study was to investigate the predictive value of single and repeated measurements of GDF-15 compared to Cystatin C and CRP for incidence of heart failure (HF) and death due to coronary heart disease (CHD) in the general population. Methods and results Levels of GDF-15, CRP and Cystatin C were determined in three repeated measurements collected 5 years apart in the DAN-MONICA (Danish-Multinational MONitoring of trends and determinants in Cardiovascular disease) cohort (participants at baseline n=3785). Cox regression models adjusted for cardiovascular risk factors revealed significantly increased hazard ratios (HR) for GDF-15 for incident HF 1.36 (HR per interquartile range (IQR) increase, 95% confidence interval (CI): 1.16; 1.59) and for death from CHD 1.51 (HR per IQR increase, 95% CI: 1.31, 1.75) (both with p<0.001). Joint modeling of time-to-event and longitudinal GDF-15 over a median 27-year follow-up period showed that the marker evolution was positively associated with death of CHD (HR per IQR increase 3.02 95% CI: (2.26, 4.04), p<0.001) and HF (HR per IQR increase 2.12 95% CI: (1.54, 2.92), p<0.001). However using Cox models with follow-up time starting at the time of the third examination, serial measurement of GDF-15, modeled as changes between the measurements, did not improve prediction over that of the most recent measurement. Conclusions GDF-15 is a promising biomarker for prediction of HF and death due to CHD in the general population, which may provide prognostic information to already established clinical biomarkers. Repeated measurements of GDF-15 displayed only a slight improvement in the prediction of these endpoints compared to a single measurement.