Serum albumin in continuous peritoneal dialysis: Its predictors and relationship to urea clearance

Abstract

We investigate the predictors of serum albumin and the relationship between serum albumin and urea kinetic indices in continuous peritoneal dialysis (CPD). In a training set (TS) of 143 urea kinetic studies performed in 92 CPD patients, stepwise logistic regression identified high/high- average peritoneal solute transport, diabetes, advanced age and high daily drain volume normalized by body water as predictors of low serum albumin (< 35 g/liter). This analysis was then substantiated in a validation set (VS) of 187 kinetic studies performed in another 102 CPD patients. The calculated area under the receiver operating characteristic (ROC) curve by this logistic regression model was 0.782 (95% CI, 0.745 to 0.819). Logistic regression was repeated in the TS using only the first kinetic study from each patient, and it identified high/high-average peritoneal solute transport, diabetes, and advanced age as predictors of low albumin. Using only the first kinetic study from each patient in the VS, the second logistic regression model calculated an area under the ROC curve equal to 0.850 (95% CI, 0.810 to 0.890). The relative risk (RR) of serum albumin < 35 g/liter was computed for all kinetic studies after combining the TS and the VS and using non-diabetic CPD subjects aged ≤ 61 years with low/low average peritoneal solute transport as the reference group. The RR with only one risk factor present ranged from 1.076 (age > 61 years) to 6.792 (high/high-average transport). The RR with two risk factors present ranged from 5.200 to 9.729. The RR with all three risk factors present was 9.100 (95% CI, range 3.923 to 21.111). A subset of 37 CPD patients had a second urea kinetic study 8 ± 5 months after an increase in the amount of dialysis due to low urea clearance and/or uremic symptoms. The weekly KT/V urea increased from 1.40 ± 0.24 to 2.10 ± 0.31 after the increase in the CPD dose. With the increase in dialysis, the protein catabolic rate increased substantially; however, the mean serum albumin remained stable (from 33.9 ± 4.6 to 33.3 ± 6.2 g/liter; decrease 18; increase 15; same 4). In comparison to the subjects who had a decrease in serum albumin after the increase in KT/V, those with the increase in serum albumin were younger (44.2 ± 11.2 vs 54.3 ± 16.2 years, P = 0.044) and has as higher serum urea after the increase in the dose of CPD (22.4 ± 7.8 vs. 17.0 ± 6.0 mmol/liter, P = 037). We conclude that the major predictors of low serum albumin CPD are advanced age, diabetes, and high/high-average peritoneal solute transport, but not urea kinetic studies. An increase in the dose of dialysis does not cause a consistent rise in serum albumin in underdialyzed CPD subjects. However, a subset of younger patients may be able to increase their serum albumin in response to the increase in KT/V.