51Cr-EDTA volume of distribution as a function of time and its implications for exponential and adaptive gamma variate plasma-clearance models

2030 Objectives Time varying volume models, V, are developed and tested for Sums of Exponential term (SET) models with one (E1), two (E2) and n (En) exponential terms and for adapted gamma variate (GV) functions. Methods Plasma clearance (CL) and V(t) models were validated using 51Cr-EDTA from 13 patients with liver failure (t = 5, 1440 min). Results GV models obtained by adaptive Tikhonov regularization (Tk-GV) are consistent with V being zero at t =0. E1 models produce both CL- and V-values that approach infinity at late times, as measured from E1 solutions for CL and V using temporally adjacent plasma samples. CL-values from E1 models applied to temporally adjacent samples are always larger than their corresponding CL(Tk-GV) values. From Tk-GV model predictions, the least inflated CL(E1)-values still average 15% too large. This explains the empirically observed Chantler constant. When the E1 and GV functions agree at two points, the Chantler type correction factor varies with 1) sampling times, 2) level of renal function, and 3) ratio between the inhomogeneous and homogeneous volumes of distribution of the system, the latter of which is difficult to calculate from an E1 model. E2 models also had significantly inflated CL-values. The E1 and E2 models have significantly more negative curvature than Cobs. Even En models incorporating V(t) have V >0 at t =0 and convergence to E1 for t large, Thus, En has inflated CL-values from underestimation of AUC. Conclusions GV functions from Tk-GV models exhibit V =0 at t =0, do not converge to E1 for t large, had the least root mean square error relative to Cobs (rRMS =7.7%) and were the best concentration models tested. The SET functions used had rRMS >10%, underestimated AUC, overestimated CL, and did not estimate CL as well as adapted GV models for 51Cr-EDTA.