Abstract Background Protein-energy wasting (PEW) or malnutrition is reportedly prevalent and consistently associated with increased risks of cardiovascular (CV) morbidity and mortality in end-stage renal disease (ESRD) patients. Geriatric nutritional risk index (GNRI), simply calculated from weight, height and albumin, has been established as a useful score to detect an early stage of the PEW. Recently, C-reactive protein (CRP) / albumin ratio has been developed as a newly surrogate marker of the PEW because the PEW results not only from an inadequate diet but also be induced by inflammatory processes. We compared the predictability of two nutritional risk scores at just starting haemodialysis (HD) therapy for CV morbidity and mortality in ESRD patients. Methods A total of 1,548 ESRD patients were enrolled and followed up for 10-year after starting HD therapy. Primary endpoint was CV events defined as hospitalization due to CV disease such as cardiac diseases, stroke and peripheral artery disease or CV death. We also compared the incremental value with C-index when the CRP/albumin ratio or the GNRI were added into a model with established risk factors. Results During follow-up period (median: 59 months), 512 cases experienced CV events (33.1%) including 165 cases of CV deaths (10.7%). After adjustment for all baseline variables, elevated CRP/albumin ratio and declined GNRI were identified as an independent predictor for CV events hazard ratio (HR) 1.02, 95% confidence interval (CI) 1.01-1.03, p=0.0080 and HR 1.01, 95% CI 1.01-1.02, p=0.020, respectively. When patients were divided into quartiles (Qs), 10-year freedom rate from CV events were 63.5%, 53.8%, 47.5% and 31.9%, and 37.3%, 43.5%, 54.8% and 60.1% in Q1, Q2, Q3 and Q4 of CRP/albumin ratio and GNRI, respectively. However, the C-index was significantly greater in the model with CRP/albumin ratio compared with GNRI (0.715 vs. 0.684, p=0.0035). Similar results were obtained for CV- and all-cause mortality (Table). Conclusion Both nutritional risk scores could independently predict and stratified the increasing risk of CV morbidity and mortality in incident HD patients. However, the predictability was significantly higher in the predicting model with CRP/albumin ratio compared with GNRI, with increasing C-statistics.
Takahashi et al. (Sat,) studied this question.