Inflammation predicts all-cause and cardiovascular mortality in haemodialysis patients

Abstract

Among non-traditional cardiovascular risk factors both malnutrition and inflammation appear to be strong predictors of mortality and morbidity in haemodialysis (HD) patients. Our study objective was to determine predictors of all-cause and cardiovascular mortality, considering the nutritional and immunologic parameters, in a cohort of HD patients treated in a single haemodialysis centre. 216 patients on HD were analyzed for clinical, nutritional-serum albumen and BMI, immunologic-serum CRP (C-reactive protein) and fibrinogen and dialysis parameters -- ultrafiltration, length of dialysis in hours, HD dose (using spKt/V and eKt/V). Mortality was monitored prospectively over a two-year period. Fifty-five of the 216 HD patients died during the follow-up period and the main cause of death was cardiovascular disease (CVD) -- 33 patients out of 55 (60%), followed by infection/sepsis (13 pts, 24%). The patients who died were significantly older, had a significantly shorter duration of HD in hours, ultrafiltration was significantly less, HD doses were significantly lower, as were serum levels of albumin (36.06 +/- 4.17 vs. 39.74 +/- 3.31; p=0.000) and Hg (93.14 +/- 15.43 vs. 109,16 +/- 12,08; p=0.000), but they had significantly higher serum levels of CRP (40.26 +/- 34.75 vs. 8.71 +/- 7.68, p=0.000) and fibrinogen (5.28 +/- 1.28 vs. 4.42 +/- 0.97, p=0.000). Kaplan-Meier survival estimates showed that the group with the lowest levels of albumin (< 3.5 g/L), and with the greatest levels of CRP (>20 mg/l) and fibrinogen (>5 g/L) had the lowest survival (log-rank test p=0.0008, p=0.00000, p=0.0000). However, in the Cox proportional hazards model, a high CRP and low Hg level (chi-square=96.467, p=0.0000) were predictors of all-cause mortality, whereas serum level of albumin did not show to be predictive. When only cardiovascular mortality is entered into the Cox model, CRP and Hg levels are still more important in predicting mortality (chi-square=70.055, p=0.0000) and only if CRP is not taken into account in the multivariate analysis, serum albumin level remains, after Hg, the strongest predictor for both overall and cardiovascular mortality (chi-square=76,564, p=0.0000; chi-square 50.619 p=0.0000). It can be concluded that inflammation predicted all-cause and cardiovascular mortality in our study group, because high CRP, as a marker of inflammation and low haemoglobin, as a result of inflammation, remained powerful predictors of both overall and cardiovascular death.