Prediction of human clearance from animal data and molecular structural parameters using multivariate regression...

Area of Science:

• Pharmacokinetics and Drug Metabolism
• Computational Chemistry
• Interspecies Scaling

Background:

• Accurate prediction of human drug clearance is crucial for drug development.
• Current interspecies scaling methods, like allometry, have limitations in predicting human clearance for diverse drug types.
• Identifying reliable predictors for human clearance is essential for optimizing preclinical to clinical translation.

Purpose of the Study:

• To develop and validate a predictive model for human drug clearance.
• To compare the performance of regression methods against allometric approaches for interspecies scaling.
• To utilize animal clearance data and molecular descriptors for enhanced prediction accuracy.

Main Methods:

• Collected clearance data for 68 drugs across rats, dogs, and humans from literature.
• Employed molecular descriptors: molecular weight, c log P, and hydrogen bond acceptors.
• Applied multiple linear regression (MLR), partial least squares (PLS), and artificial neural network (ANN) analyses.
• Incorporated interaction and quadratic terms in MLR and PLS to capture nonlinear relationships.

Main Results:

• MLR and PLS models with quadratic terms achieved the best predictive performance, yielding an identical equation.
• The squared cross-validated correlation coefficient (q(2)) for the best model was 0.682.
• The developed regression methods demonstrated superior predictive performance compared to allometric approaches for the studied dataset.

Conclusions:

• Multiple linear regression using clearance data from two animal species and simple structural parameters provides a robust method for predicting human drug clearance.
• This approach is applicable to a wide range of drugs with varying characteristics.
• The developed method offers a valuable alternative to traditional allometric scaling for interspecies pharmacokinetic predictions.