LAUSR

Due to the interplay of chronic kidney disease and the heart, it is common for myocardial damage and strain to be present in patients with end stage kidney disease. The cardiac biomarkers Troponin I (cTnI) and the N-terminal fragment of prohormone brain natriuretic peptide (NT-proBNP) are strongly predictive of heart failure in the general population, however the value of their routine measurement for stratification of risk in hemodialysis (HD) patients is unknown. International DOPPS data indicate that these cardiac biomarkers are measured in fewer than 2% of hemodialysis patients in real-world practice. We sought to test the associations of cTNI and NT-proBNP with death in a large HD cohort. We analyzed data from 1169 prevalent Japanese hemodialysis patients in DOPPS phase 5 (J-DOPPS; 2012-2015) where pre-dialysis levels of cTnI and NT-proBNP at study enrollment by protocol for all participants. We used Cox regression to test the association of the cardiac biomarkers with all-cause mortality, adjusting for potential confounders. We conducted stratified analyses to assess potential effect modification of individual clinical characteristics: age, systolic blood pressure, dialysis vintage, diabetes mellitus, cardiovascular disease, and heart failure (CHF). Median (interquartile range) cTnI and NT-proBNP levels were 0.018 [0.005, 0.04] ng/mL and 3432 [1580, 8017] pg/mL, respectively. There was a positive direct association between cTNI and NT-proBNP (Spearman correlation coefficient=0.58). We observed 174 deaths during a median [IQR] follow-up time of 2.8 [2.3, 2.9] years. Compared to the reference group of cTNI<0.01 the hazard ratio (HR) of death (95% CI) was 1.23 (0.57, 2.69) for cTNI ≥0.01 to 0.02; HR=1.94 (1.12, 3.37) for cTNI ≥0.02 to 0.04; and HR=3.65 (2.14, 6.23) forcTNI>0.04. Compared to the reference group of NT-proBNP <2000 pg/mL, HR= 1.20 (0.65, 2.23) for NT-proBNP ≥2000 to 4000 pg/mL; HR= 2.06 (1.22, 3.46) for NT-proBNP >4000 to 8000 pg/mL; and 3.23 (1.71, 6.09) for NT-proBNP > 8000pg/mL. For both variables associations with mortality were comparable in patients with and without heart failure (p values for interaction=0.07 and 0.19, for NT-proBNP and cTNI, respectively) and according to other stratification variables (the p values for interaction were >0.15 for all other stratified models). Routinely measured (not by indication) NT-proBNP and Troponin I are strongly associated with mortality among prevalent Japanese HD patients. Routine evaluation of these cardiac biomarkers in HD patients with or without a previous diagnosis of CHF may help identify patients at risk of death who may have undetected heart failure or cardiac strain due to fluid overload.