Fagerholm, U.

Introduction: Solubility/dissolution and permeability are essential determinants of gastrointestinal absorption of drugs. In vitro aqueous solubility (S) and apparent permeability (Papp) are commonly used as measurements and predictors of in vivo fraction absorbed (fa) and BCS-classing in humans. The objective of this study was to explore the relationships between in vitro aqueous S and Dose number (Do) and in vivo fa and in vitro Papp and in vivo fa and the predictive power of in vitro aqueous S, Do and Papp. Methods: In vitro and in vivo data were taken from studies in the literature and correlated. In vitro S data were produced in various laboratories and with different methodologies. In vitro Papp data were produced using Caco-2 and MDCK cells in various laboratories and Caco-2 and RRCK cells in one laboratory each. Do was estimated as oral dose / (S x 250 mL). Results: 452 S data and 1480 Papp data were found and used. There was no correlation (R2=0.0) between in vitro log S and Do vs in vivo fa, not even at S<1 mg/L or not for compounds with <90 % and <30 % in vivo fa. A R2 of 0.43 was found between log Caco-2 Papp and in vivo fa. The corresponding R2 for Caco-2 from one laboratory was 0.65. The interlaboratory R2 for the Caco-2 model was 0.48. R2-estimates for Caco-2 vs MDCK and Caco-2 vs RRCK Papp were 0.23 and 0.21, respectively. Discussion and Conclusion: Aqueous S appears to have no predictive value of in vivo fa in humans, not even at low S or after correction for dose. The shows that one should not base human biopharmaceutical predictions based on aqueous S. Log Caco-2 Papp explains about half of the variance of in vivo fa in humans. Higher predictive accuracy can be achieved by avoiding mixing Papp data from different sources. The poor correlations found between Caco-2 and the two other Papp-models (MDCK and RRCK) demonstrate considerable methodological differences. The unexplained variance does not appear to be explained by S and dose, but rather by in vitro-in vivo difference in permeability and poor/absent relationship between in vitro S and in vivo dissolution potential.