#3827 BICARBONATE PROFILING IN HEMODIALYSIS: MATHEMATICAL MODELING OF DATA FROM A CLINICAL TRIAL

Abstract

Abstract Background and Aims During hemodialysis (HD), bicarbonate and other bases are added to the dialysis fluid to correct chronic acidosis that may develop in kidney-impaired patients, increasing plasma bicarbonate concentration to physiological levels. Previous studies have shown that both exceedingly high and low post-dialysis bicarbonate concentrations are associated with adverse clinical outcomes; however, no clear guidelines for the prescription of bicarbonate concentration in the dialysis fluid are available. Data from a recent controlled clinical trial, carried out at the Department of Nephrology, Dialysis and Internal Medicine, Medical University of Warsaw (Poland), were used to identify transport parameters of a mathematical model of acid-base homeostasis in HD patients, as the first step in the implementation of a tool to optimize bicarbonate profiling during HD. Method The study was carried out in 20 chronic patients, undergoing thrice weekly 4-hr HD. The patients underwent three treatment regimens: A) constant dialysis fluid bicarbonate concentration (HCO3,D) equal to 35 mEq/L; B) HCO3,D decreasing from 35 to 30 mEq/L after 2 hrs; C) HCO3,D increasing from 30 to 35 mEq/L after 2 hrs. The patients were divided in two cohorts: for the first (n = 11) treatment schedule was: treatment A during the first and third week, B during the second week and C during the fourth week. For the second cohort (n = 9) treatment C was carried out during the second week and treatment B during the fourth. Data were collected only during the second and third sessions of the week, and acid-base composition was determined using a blood gas analyzer. The mathematical model describes the transport of bicarbonate and dissolved CO2 from dialysis fluid to the patient and can predict changes in the acid-base equilibrium of arterial and venous plasma and erythrocytes, interstitial fluid, and tissue cells, using mass transport equations. The dialysances for bicarbonate (KHCO3) and dissolved CO2 (KCO2) were estimated via least square minimization. Results No significant difference was found comparing patients in the two cohorts, or comparing sessions of the same week. Pre-dialysis plasma bicarbonate, pH, or CO2 partial pressure (PCO2) were significantly lower for treatment B compared to treatment A. Average plasma bicarbonate concentrations for the different treatments are shown in Figure 1 for data and model simulations. Table 1 reports the average values of the estimated parameters for week. No statistically significant difference was found between sessions, cohorts, or treatments. The model fitted bicarbonate data with an average error of 2.5 ± 1.4%, and PCO2 with an error of 4.5 ± 3.6%. Simulations of mixed-venous blood, interstitial fluid, and tissue cells concentrations showed a similar response to different treatments as in arterial blood. Conclusion The model accurately describes plasma bicarbonate data with high accuracy regardless of the treatment conditions. The parameters estimated showed considerable variability between treatments (less so between the sessions of the same week), but without statistical significance. The model can be used to simulate the effect of dialysate bicarbonate profiling on acid-base equilibrium in HD patients.