Abstract 16098: Additional Predictive Value of Plasma C24:0/C16:0 Ceramide, and Its Modulation by BMI, Over NT-proBNP in the Prediction of Mortality

Abstract

Introduction: NT-proBNP is a predictor of mortality in patients with or without HF. Ceramides are bioactive lipids involved in cell signaling and membrane structure. Our group showed that the ratio of specific plasma ceramides (C24:0/C16:0) is inversely related to death. However, NT-proBNP levels were not available for multivariate modeling of mortality in that cohort. Hypothesis: We hypothesized that plasma C24:0/C16:0 would add to a predictive model that includes NT-proBNP and classical cardiovascular risk factors (CVRs) of mortality. We also hypothesized that body mass index (BMI) modulates the predictive value of plasma C24:0/C16:0. Methods: We performed multivariable modeling on data from 2,760 Caucasian subjects in the SHIP (Study of Health in Pomerania) cohort from northern Europe. Plasma C24:0/C16:0 was measured using targeted liquid chromatograph/tandem mass spectrometry. CVRs in the modeling included: age, sex, systolic blood pressure, current smoking, diabetes, and total/high-density cholesterol. Model performance was evaluated using Akaike’s information criterion (AIC) and Harrel’s c-statistic. Results: There were 529 deaths over a median of 14.6 y. Model selection via AIC revealed that the best model with the least number of variables to predict mortality included plasma C24:0/16:0 and BMI (+ their interaction), CVRs, and NT-proBNP, P < 0.001. The evidence ratio showed that this best model is 540 times more likely to be the ideal model than the model that only includes CVRs. The best model was also much more informative than a model that added only NT-proBNP to the CVRs (ΔAIC = 19.865, Akaike weight < 0.001). With increasing BMI, the protective effect of plasma C24:0/16:0 on survival probability diminished. Post hoc analysis showed the best model predicted CVD mortality but not non-CVD mortality. Conclusions: A mortality prediction model that includes plasma C24:0/C16:0, NT-proBNP, and CVRs is superior to CVRs alone or CVRs + NT-proBNP and helps further stratify patients’ risk. BMI modifies the predictive value of plasma C24:0/C16:0. Plasma C24:0/C16:0 may be a novel target for modifying CVD mortality risk.