Lipoprotein Subfractions Predict All-cause and Cardiovascular Mortality in CKD Patients Undergoing Hemodialysis: A Prospective Cohort Study Based on NMR Metabolomics

Lipoprotein metabolism is markedly altered in chronic kidney disease (CKD). Studies on the association between nuclear magnetic resonance (NMR) derived lipoprotein subfraction signature and mortality in CKD patients undergoing hemodialysis are limited. NMR based metabolomics was performed on the baseline plasma samples from 368 maintenance hemodialysis patients. Survival analyses were used to investigate the effect of lipoproteins on all-cause mortality and cardiovascular disease (CVD) mortality. Prediction models were further developed using stepwise regression combined cox proportional-hazards model. During the average follow-up of 45.1 months, we observed 144 all-cause deaths and 67 CVD deaths. After adjustment for 14 important covariates, we identified 18 and 35 lipoprotein parameters associated with all-cause mortality and CVD mortality, separately. Cholesterol in total low-density lipoprotein (LDL-C) and total high-density lipoprotein (HDL-C) were correlated with neither all-cause death nor CVD death. For lipoprotein subfractions, triglyceride levels in large very-low density lipoprotein (VLDL) were positively correlated only with all-cause mortality. Lipids (triglyceride, cholesterol and phospholipid) in medium VLDL, cholesterol/total lipids in intermediate-density lipoprotein (IDL) and free cholesterol/total lipids in small high-density lipoprotein (HDL) were positively associated with both all-cause and CVD mortality. The addition of lipoprotein parameters to traditional risk factors significantly improved the mortality risk prediction: the area under the receiver operating characteristic curve (AUC) was improved from 0.811 to 0.842 (p-value = 0.020) for all-cause mortality and 0.806 to 0.854 (p-value = 0.005) for CVD mortality. Our results highlight the lipoprotein subfractions related to all-cause and CVD mortality of maintenance hemodialysis patients, and the lipoproteins-driven prediction models significantly outperform traditional risk factors.