Serum alkaline phosphatase predicts mortality among maintenance hemodialysis patients

Abstract

Several observational studies have demonstrated that serum levels of minerals and parathyroid hormone (PTH) have U- or J-shaped associations with mortality in maintenance hemodialysis patients, but the relationship between serum alkaline phosphatase (AlkPhos) and risk for all-cause or cardiovascular death is unknown. In this study, a 3-yr cohort of 73,960 hemodialysis patients in DaVita outpatient dialysis were studied, and the hazard ratios for all-cause and cardiovascular death were higher across 20-U/L increments of AlkPhos, including within the various strata of intact PTH and serum aspartate aminotransferase. In the fully adjusted model, which accounted for demographics, comorbidity, surrogates of malnutrition and inflammation, minerals, PTH, and aspartate aminotransferase, AlkPhos > or =120 U/L was associated with a hazard ratio for death of 1.25 (95% confidence interval 1.21 to 1.29; P < 0.001). This association remained among diverse subgroups of hemodialysis patients, including those positive for hepatitis C antibody. A rise in AlkPhos by 10 U/L during the first 6 mo was incrementally associated with increased risk for death during the subsequent 2.5 yr. In summary, high levels of serum AlkPhos, especially >120 U/L, are associated with mortality among hemodialysis patients. Prospective controlled trials will be necessary to test whether serum AlkPhos measurements could be used to improve the management of renal osteodystrophy.

Figure 1.

Serum AlkPhos concentrations in the selected strata of the entire spectrum of serum iPTH and AST.

Figure 2.

Death hazard ratio for the entire range of serum AlkPhos (AP) in 73,960 MHD patients over 3 yr (July 2001 through June 2004). (A) All-cause mortality with fixed (baseline) covariate model. (B) Cardiovascular mortality with fixed (baseline) covariate model. (C) All-cause mortality with time-varying (calendar quarter) covariate model with one-quarter time lag. (D) All-cause mortality with time-varying (calendar quarter) covariate model with two-quarter time lag. Note that the fully (case mix and MICS) adjusted model also includes adjustment for serum minerals (calcium and phosphorus) and iPTH and AST concentrations. Hazard ratios in y axis are in logarithmic scale.

Figure 3.

Hazard ratio of all-cause mortality for the changes in serum AP from the baseline to subsequent calendar quarter (i.e., over 6 mo in 26,975 MHD patients), with subsequent survival over the subsequent 2.5 yr. All models are also adjusted for baseline AP. Note that the fully (case mix and MICS) adjusted model also includes adjustment for serum minerals (calcium and phosphorus) and iPTH and AST concentrations. Hazard ratios in y axis are in logarithmic scale.

Figure 4.

Hazard ratio of all-cause mortality for the dichotomized AP ≥120 U/L at baseline in various subgroups of 73,960 MHD patients over 3 yr. Note that the model is adjusted for all case mix and MICS markers, including serum minerals (calcium and phosphorus) and iPTH and AST concentrations. Note that hepatitis C antibody data were available only for 18% of the cohort (n = 13,664). Hazard ratios in y axis are in logarithmic scale.

Figure 5.

Hazard ratio of all-cause mortality for AlkPhos at baseline in three strata of serum iPTH in 73,960 MHD patients over 3 yr. (A) Low iPTH <150 pg/ml (n = 20,775). (B) K/DOQI recommended target zone for iPTH (between 150 and 300 pg/ml; n = 22,543). (C) High iPTH ≥300 pg/ml (n = 23,912). Note that the fully (case mix and MICS) adjusted model also includes adjustment for serum minerals (calcium and phosphorus) and iPTH and AST concentrations. Hazard ratios in y axis are in logarithmic scale.

Figure 6.

Hazard ratio of all-cause mortality for AlkPhos at baseline in two strata of serum AST in 73,960 MHD patients over 3 yr. (A) Low to normal AST <30 IU/L (n = 66,868). (B) High AST ≥ 30 IU/L (n = 6708). Note that the fully (case mix and MICS) adjusted model also includes adjustment for serum minerals (calcium and phosphorus) and iPTH and AST concentrations. Hazard ratios in y axis are in logarithmic scale.

Figure 7.

Comparing mortality predictability of deciles of serum AlkPhos and deciles of iPTH in 80,000 MHD patients who were followed for up to 3 yr (2001 through 2004). Note that these analyses are based on fixed covariate (non–time dependent) and fully adjusted models. The fifth decile serves as the reference group. The cutoff levels for AlkPhos deciles are 61, 71, 79, 87, 96, 107, 121, 142, and 186 U/L and for iPTH are 64, 106, 146, 184, 224, 274, 344, 459, and 697 pg/ml. Each decile includes approximately 7400 MHD patients. Hazard ratios in y axis are in logarithmic scale.