BDDCS class prediction for new molecular entities

Abstract

The Biopharmaceutics Drug Disposition Classification System (BDDCS) was successfully employed for predicting drug-drug interactions (DDIs) with respect to drug metabolizing enzymes (DMEs), drug transporters and their interplay. The major assumption of BDDCS is that the extent of metabolism (EoM) predicts high versus low intestinal permeability rate, and vice versa, at least when uptake transporters or paracellular transport is not involved. We recently published a collection of over 900 marketed drugs classified for BDDCS. We suggest that a reliable model for predicting BDDCS class, integrated with in vitro assays, could anticipate disposition and potential DDIs of new molecular entities (NMEs). Here we describe a computational procedure for predicting BDDCS class from molecular structures. The model was trained on a set of 300 oral drugs, and validated on an external set of 379 oral drugs, using 17 descriptors calculated or derived from the VolSurf+ software. For each molecule, a probability of BDDCS class membership was given, based on predicted EoM, FDA solubility (FDAS) and their confidence scores. The accuracy in predicting FDAS was 78% in training and 77% in validation, while for EoM prediction the accuracy was 82% in training and 79% in external validation. The actual BDDCS class corresponded to the highest ranked calculated class for 55% of the validation molecules, and it was within the top two ranked more than 92% of the time. The unbalanced stratification of the data set did not affect the prediction, which showed highest accuracy in predicting classes 2 and 3 with respect to the most populated class 1. For class 4 drugs a general lack of predictability was observed. A linear discriminant analysis (LDA) confirming the degree of accuracy for the prediction of the different BDDCS classes is tied to the structure of the data set. This model could routinely be used in early drug discovery to prioritize in vitro tests for NMEs (e.g., affinity to transporters, intestinal metabolism, intestinal absorption and plasma protein binding). We further applied the BDDCS prediction model on a large set of medicinal chemistry compounds (over 30,000 chemicals). Based on this application, we suggest that solubility, and not permeability, is the major difference between NMEs and drugs. We anticipate that the forecast of BDDCS categories in early drug discovery may lead to a significant R&D cost reduction.

Fig 1

Box plots showing the distribution of BDDCS classes in the linear discriminant space. Boxes delimit values between the 25^th and the 75^th percentile, the line contained in the box is the median, whiskers correspond to the maximum and the minimum values, and the dots are outliers.

Fig 2

Classification accuracy for EoM and FDA Solubility predictions. Accuracy in predicting FDAS is presented for i) model that ignores the HDS ii) the calculated dose number based on the VolSurf+ solubility and the HDS.

Fig 3

A) Distribution of BDDCS predicted classes for the validation set (green for true class having highest predicted BDDCS class membership score, yellow for true class having predicted BDDCS class membership score within the 1^st highest and the 2^nd highest, red for incorrect classifications). B) Comparison of the average class accuracy of different models for BDDCS class prediction.

Fig 4

Pie charts showing the proportion of predicted BDDCS classes for each actual BDDCS class; predictions are shown for the consensus model presented in this work, based on separate estimations of EoM and FDAS, in comparison to the classification approaches (RP, SVM and RF) presented in Khandelwal et al.

Fig 5

Single BDDCS class prediction for the validation set based on i) predicted BDDCS class membership scores to each BDDCS class (4-class models derived from EoM and FDAS predictions) ii) 2-class models (either SVM or RF) targeting one specific class.

Fig 6

Predicted BDDCS class distribution for marketed and clinical drugs (oral, non-oral, clinical phases 1, 2 and 3) and medicinal chemistry compounds (extracted from WOMBAT).

Fig 7

Box plots for measured and calculated plasma protein binding (PPB) percentages, Measured LogD7.4, Calculated LogD7.5, calculated Caco2 permeability and LgBB (predicted blood brain barrier permeability). The calculated parameters are VolSurf+ descriptors. See Fig. 1 for box plot description. Values for the 10^th, 25^th, 75^th and 90^th percentiles, together with number of molecules used, standard deviations, median and averages are available in Supporting Information Tables S13 – S18.