Background: Cross sectional studies have established that the serum albumin level is dependent on serum levels of acute-phase proteins (APPs) or cytokine levels in hemodialysis patients. While the acute-phase response is generally associated with acute inflammatory events, a cross sectional analysis relating laboratory values to outcomes assumes these values to be unchanging. The longitudinal relationship among laboratory measurements and how they vary over time in a population of patients are unknown.
Methods: Patients who were enrolled in the HEMO Study were recruited into an ancillary longitudinal study to establish the predictive effect of temporal variation in the levels of APPs and of temporal variation in normalized protein catabolic rate (nPCR) on the serum albumin concentration. nPCR was measured monthly using a double-pool method. The positive APPs-C-reactive protein (CRP), alpha1 acid glycoprotein (alpha1-AG), and ceruloplasmin-and the negative APP-transferrin (Trf)-were measured in serum obtained before each dialysis session for six weeks and then monthly in 37 hemodialysis patients. A random coefficient regression analysis was used to assess the association of serum albumin with other measured parameters at each time point, as well as fixed patient characteristics.
Results: The within-subject coefficients of variation of albumin (median, range of 25th to 75th percentiles; median, 0.0614; range, 0.0485 to 0.0690) were significantly less than that of APPs (CRP, median, 0.878; range, 0.595 to 1.314, P < 0. 05; and alpha1 AG, median, 0.173; range, 0.116 to 0.247, P < 0.05). The levels of APPs and albumin varied considerably over time. The primary predictor of current albumin was the current CRP level (P = 0.0014). nPCR also was a significant predictor for albumin levels (P = 0.0440) after controlling for the effect of APPs, suggesting an effect of nPCR on serum albumin concentration irrespective of the acute-phase response. Age and the presence of an arteriovenous graft were significant predictors that were associated with reduced albumin.
Conclusions: The acute-phase response is intermittent and is not a continuous feature in individual dialysis patients. Levels of APPs are the most powerful predictors for the levels of albumin concentration in hemodialysis in a longitudinal setting. Since variations in albumin are small, measurement of variations in APPs may provide greater insight into the dynamics of clinically relevant processes.