P1061ACID-BASE MODELLING AND TREND IN CENTRAL VENOUS OXYGEN SATURATION IN HEMODIALYSIS

Abstract

Abstract Background and Aims Central venous oxygen saturation (sO2cv) is commonly used as a proxy of cardiac output in non-uremic populations. Recently, the decrease in sO2cv commonly observed in many hemodialysis (HD) sessions was correlated with ultrafiltration volume (UFV) and was suspected to be linked with a treatment-induced decrease in cardiac output (Zhang et al, NDT 2018). High variability in sO2cv has also been associated with all-cause mortality in this group of patients (Zhang et al, Blood Purif 2019). We adapted a comprehensive mathematical model of whole body oxygen (O2) and carbon dioxide (CO2) biochemistry, storage and transport and applied it to simulate the changes in sO2cv during bicarbonate hemodialysis to test their connection with changes in CO. Method The original model of biochemistry and transport of base excess, CO2 and O2 was originally developed by Rees and Andreassen (Crit Rev Biomed Eng 2005), and describes in detail the chemical and respiratory mechanisms that regulate the biochemical status in different physiological compartments of the patients’ body (venous blood, arterial blood, tissue cells, interstitial fluid). The model was modified to include the effect of the HD treatment on these systems, allowing to calculate the variations in venous and arterial sO2. Treatment perturbations included the removal via ultrafiltration of interstitial fluid and plasma volume (and consequent hemoconcentration), and infusion of bicarbonate. Simulation experiments of 4h HD sessions were carried out with bicarbonate dialysate concentration = 35 mmol/L and dialysance = 198 mL/min, and various ultrafiltration volumes (UFV); CO was assumed to decrease by 2% for each 1% change in blood volume. The initial biochemical status of the patient was defined assuming 20 mEq/L bicarbonate concentration, 7.4 pH, and 97% oxygen saturation in the arterial compartment, resulting in initial sO2cv equal to 66.6%. Both artero-venous and veno-venous access were simulated. Results The model predicted a negative linear trend for sO2cv with variable slope depending on UFV, as shown in the table: The trend in sO2cv was directly proportional to the decrease in CO during HD and inversely proportional to UFV. In case of constant CO, the simulated sO2cv would actually increase during the session. Although the type of vascular access determined changes in other simulated quantities, partial pressures of CO2 and O2 and oxygen saturation in both arterial and venous blood were practically unchanged. Conclusion Our model confirmed the correlation between the trend in sO2cv and UFV already highlighted in previous works. Moreover, it showed that the decrease in cardiac output is a major determinant of the decrease in sO2cv during the HD session. The developed mathematical model may provide physiological insights in the development of new technologies to monitor changes in cardiac performance during HD.