Background: In patients with end-stage renal disease (ESRD), the ability of single and multiple biomarker monitoring to predict adverse outcomes has not been well established. This study determined the prognostic value of multiple biomarkers for all-cause death over 2 years in 399 ESRD patients.
Methods: The risk of all-cause death was determined by use of multiple biomarkers based on concentrations for a reference population (normal) and cutoffs based on tertile distributions in the ESRD group. Biomarkers studied included N-terminal pro-B-type natriuretic peptide (NT-proBNP), high-sensitivity C-reactive protein (hsCRP; Dade Behring and Roche assays), and cardiac troponin T (cTnT; Roche) and I (cTnI; Dade Behring and Beckman Coulter assays). Relative risks of death were estimated and survival curves computed.
Results: A total of 101 deaths occurred during 594 patient-years of follow-up. Increased NT-proBNP concentrations were not predictive of death on the basis of the normal cutoffs. However, tertile analysis of NT-proBNP was significantly predictive of death and had a ROC area under the curve equivalent to or better than any of the other biomarkers. Biomarkers independently predictive of survival were hsCRP (P <0.001, either assay), cTnT (P <0.05), and cTnI (Dade, P <0.05). Two-year mortality rates were 6% (n = 45) with normal hsCRP, cTnI, and cTnT concentrations; 19% (n = 173) with increased hsCRP or cTnT and normal cTnI; 44% (n = 160) with both hsCRP and cTnT increased and normal cTnI; 61% (n = 21) with increased cTnI (Dade) or 47% (n = 74) with increased cTnI (Beckman) regardless of hsCRP or cTnT concentrations. Defined by the normal cutoffs, increased concentrations of biomarkers were present in various proportions of the 399 patients with ESRD: NT-proBNP, 99%; hsCRP, 46% (both Roche and Dade assays); cTnT, 85%; cTnI, 19% (Beckman assay) and 5% (Dade assay).
Conclusions: Although mechanisms likely vary for causation, increased plasma hsCRP, cTnT, and cTnI above the cutoffs for our reference (normal) population were all independently predictive of subsequent death in ESRD patients. Tertile analysis for NT-proBNP also demonstrated prognostic value.