Parameter estimation in six numeric models of transperitoneal transport of glucose

Six competing kinetic models of transperitoneal glucose transport were formulated and validated. The models were designed to elucidate the presence or absence of diffusive, nonlymphatic convective and lymphatic convective solute transport. The validation procedure included an assessment of theoretical and practical identifiability, goodness of fit, residual error analysis, and plausibility of parameter estimates. Experimental results were obtained from 21 patients without diabetes. The validation procedure demonstrated that the model that only included diffusion was superior to the other models. Theoretically, both nonlymphatic convective and lymphatic convective transports might exist. However, neither the ultrafiltration sieving coefficient nor the lymphatic flow rate were practically identifiable, probably because any amount of glucose transported by nonlymphatic convective and lymphatic convective transport mechanisms was negligible compared with the amount transported by diffusion. Based on these results, there appear to be problems measuring convective solute transport parameters when the solute transport is in the dialysate-to-blood direction while the fluid transport is in the blood-to-dialysate direction.