Mortality prediction using modern peptide biomarkers in hemodialysis patients--a comparative analysis

Determination of peptide biomarkers such as troponins, natriuretic peptides or the recently reported FGF23 can be useful to identify hemodialysis patients with a high risk of mortality. However, it is desirable to focus on few robust parameters to warrant their routine application.

In a prospective cohort study with 239 prevalent hemodialysis patients we studied the prognostic significance of 10 simultaneously determined modern peptide biomarkers (high sensitive troponin I and T, NT-pro-BNP, BNP, MR-pro-ANP, MR-pro-ADM, CT-pro-ET1, copeptin, FGF23 and a-Klotho) and compared them with parameters traditionally associated with mortality (PTH, Ca, Pi, albumin, CRP, cholesterol, AP).

After a follow-up of 4 years, plasma concentration of troponins, natriuretic peptides, MR-pro-ADM, FGF23 as well as PTH, CRP, AP were significantly higher in deceased patients (n=95). Hazard ratios from cox regression on a continuous scale (doubling of plasma concentration) or relative in tertiles were highest for high sensitive troponins, followed by natriuretic peptides and MR-pro-ADM (1.6-2.0 and 2.3-5.5, resp.). C-indices were also highest for troponins (0.708-0.746), followed by natriuretic peptides (0.706-0.731). Traditional parameters had low c-indices (0.598-0.655). Stepwise cox regression revealed that among all parameters troponin I, NT-pro-BNP, PTH and CRP remained independent predictors of mortality and a composite score had the highest c-index (0.799 [0.740-0.849]).

Among peptide biomarkers high sensitive troponins and to a lesser extent natriuretic peptides are strong predictors of mortality in asymptomatic hemodialysis patients, followed by markers of mineral-bone disease and inflammation.