Anthropometric measures, cytokines and survival in haemodialysis patients

Background. Lower serum albumin concentration (sAlb) and higher levels of pro-inflammatory cytokines have been reported to predict death in patients treated with haemodialysis (HD). SAlb, along with anthropometric measures, has been used as a surrogate marker for nutritional status in patients with chronic disease. Though adequate nutrition has been considered an important factor for patients treated with HID, it has not been established if any nutritional markers other than lower serum albumin and lower body mass index (BMI) predict death. Furthermore, it has not been shown whether anthropometric measures other than BMI are associated with predictors of mortality. Methods. At the outset of the study, patients were assessed using demographic and anthropometric indices including arm fat area (AFA), arm muscle area (AMA), BMI, per cent ideal weight (PIW), predialysis sAlb, and circulating levels of tumour necrosis factor-alpha (TNF-alpha), IL-1 and IL-6. A severity index, previously demonstrated to be a mortality marker, was used to grade medical co-morbidity. Results. Two-hundred and forty patients entered the study. The mean age was 55.1 +/- 14.3 years, mean sAlb 2 3.76 +/- 0.60 mg/dl, mean AFA 1742 +/- 1225 mm(2), mean AMA 5464 +/-1817 mm(2), mean PIW 101.0 +/- 21.3% and 2 mean BMI 24.9 +/- 5.6 kg/m(2). PIW, BMI, AFA and AMA were, as expected, all highly correlated with one another. SAlb correlated with serum transferrin; however, neither sAlb nor serum transferrin concentration correlated with circulating cytokine levels. Circulating cytokines and sAlb did not correlate with PIW, BMI, AFA or AMA. In Cox regression analyses using multiple control variables, IL-6 predicted survival, while the anthropometric measures did not. Conclusions. Pro-inflammatory cytokines and sAlb are robust predictors of death in patients treated with HD. PIW and BMI correlate well with other anthropometric measures in patients treated with HID, but these measures do not correlate with markers of inflammation. Anthropometric measures are poor predictors of survival compared with measures linked to the acute-phase response.