Prediction of liver volume - a population-based approach to meta-analysis of paediatric, adult and geriatric populations - an update.

Liver volume is a critical scaling factor for predicting drug clearance in physiologically based pharmacokinetic modelling and for both donor/recipient graft size estimation in liver transplantation. The accurate and precise estimation of liver volume is therefore essential. The objective here was to extend an existing meta-analysis using a non-linear mixed effects modelling approach for the estimation of liver volume to other race groups and paediatric and geriatric populations. Interrogation of the PubMed® database was undertaken using a text string query to ensure as objective a retrieval of liver volume data for the modelling exercise as possible. Missing body size parameters were estimated using simulations from the Simcyp Simulator V13R1 for an age and ethnically appropriate population. Non-linear mixed effect modelling was undertaken in Phoenix 1.3 (Certara) utilizing backward deletion and forward inclusion of covariates from fully parameterized models. Existing liver volume models based on body surface area (BSA) and body weight and height were implemented for comparison. The extension of a structural model using a BSA equation and incorporating the Japanese race and age as covariates and exponents on LV0 (θBaseline ) and body surface area (θBSA ), respectively, delivered a comparatively low objective function value. Bootstrapping of the original dataset revealed that the confidence intervals (2.5-97.5%) for the fitted (theta) parameter estimates were bounded by the bootstrapped estimates of the same. In conclusion, extension and re-parameterization of the existing Johnson model adequately describes changes in liver volume using the body surface area in all investigated populations. Copyright © 2017 John Wiley & Sons, Ltd.