A Quantitative Efflux Ratio-Integrated Model for Passive Permeability Estimation from Caco-2 Assays

Reliable estimation of passive permeability in the presence of active efflux remains a key challenge in early drug discovery. In bidirectional Caco-2 assays, apparent permeability reflects both passive diffusion and transporter-mediated efflux, preventing direct determination of intrinsic passive permeability from baseline measurements. Existing approaches typically require additional experimental systems, increasing complexity and cost. This study aimed to develop a quantitative method to estimate passive permeability directly from standard Caco-2 data without additional assays.

An empirical Efflux Ratio Integrated Model (ERIM) was developed, incorporating Efflux Ratio (ER) into a correction framework applied to bidirectional apparent permeability data. The model was evaluated using a dataset of 125 in-house compounds and 13 reference compounds. Performance was assessed through repeated train-test validation, comparison with a Global Average Model (GAM), and classification-based error analysis.

GAM showed systematic overestimation of passive permeability that increased with ER, resulting in a high false-positive rate. ERIM eliminated efflux-dependent bias, substantially reduced false-positive classifications, and maintained low false-negative rates, with consistent performance across validation rounds.

The conventional averaging method (GAM) showed that efflux-driven distortion on passive permeability estimate is systematic and can be corrected using ER and a scaling factor. Results showed reduced bias without increasing false negatives, supporting its utility for early-stage screening.

ERIM provides a simple and practical framework for estimating passive permeability directly from a single Caco-2 assay by correcting efflux-related bias. This approach improves the interpretability of permeability data and supports more efficient compound prioritization in early-stage drug discovery.