Multi-Task ADME/PK prediction at industrial scale: leveraging large and diverse experimentaldatasets.

ADME (Absorption, Distribution, Metabolism, Excretion) properties are key parameters to judge whether a drug candidate exhibits a desired pharmacokinetic (PK) profile. In this study, we tested multi-task machine learning (ML) models to predict ADME and animal PK endpoints trained on in-house data generated at Boehringer Ingelheim. Models were evaluated both at the design stage of a compound (i. e., no experimental data of test compounds available) and at testing stage when a particular assay would be conducted (i. e., experimental data of earlier conducted assays may be available). Using realistic time-splits, we found a clear benefit in performance of multi-task graph-based neural network models over single-task model, which was even stronger when experimental data of earlier assays is available. In an attempt to explain the success of multi-task models, we found that especially endpoints with the largest numbers of data points (physicochemical endpoints, clearance in microsomes) are responsible for increased predictivity in more complex ADME and PK endpoints. In summary, our study provides insight into how data for multiple ADME/PK endpoints in a pharmaceutical company can be best leveraged to optimize predictivity of ML models.

Towards holistic Compound Quality Scores: Extending ligand efficiency indices with compound pharmacokinetic characteristics.

The suitability of small molecules as oral drugs is often assessed by simple physicochemical rules, the application of ligand efficiency scores or by composite scores based on physicochemical compound properties. These rules and scores are empirical and typically lack mechanistic background, such as information on pharmacokinetics (PK). We introduce new types of Compound Quality Scores (CQS, specifically called dose scores and cmax scores), which explicitly include predicted or, when available, experimental PK parameters and combine these with on-target potency. These CQS scores are surrogates for an estimated dose and corresponding cmax and allow prioritizing of compounds within test cascades as well as before synthesis. We demonstrate the complementarity and, in most cases, superior performance relative to existing efficiency metrics by project examples.

Mechanistic study on hydrodynamics in the mini-scale biphasic dissolution model and its influence on in vitro dissolution and partitioning.

Biphasic dissolution models were proposed to provide good predictive power for in vivo absorption kinetics. However, up to date the impact of hydrodynamics in mini-scale models are not well understood. Consequently, the aim of this work was to investigate different setups of a previously published mini-scale biphasic dissolution model (miBIdi-pH-II) to better understand the relevance of hydrodynamics for evaluating kinetic parameters and to simultaneously increase the robustness of the experimental model. As a first step, the hydrodynamics within the aqueous phase were characterized by in silico simulations of the flow patterns. Different settings, such as higher rotation speeds of the paddles, the implementation of a second propeller into the aqueous phase, and different shapes of aqueous stirrers were investigated. Second, to evaluate the results of the in silico simulations, in vitro experiments with glitter were carried out. Last, the same settings were applied in the miBIdi-pH-II using dipyridamole (DPD) as model compound to estimate kinetic parameters by applying a compartment-based modelling approach. Both in vitro experiments with glitter or DPD demonstrated the adequateness of the previous in silico hydrodynamic simulations. The use of higher rotation speeds and a second aqueous propeller resulted in more homogeneous mixing of the aqueous phase. This resulted in faster distribution of dissolved active pharmaceutical ingredient (API) into the octanol phase. A kinetic model was successfully applied to quantify the influence of hydrodynamics on the partitioning rate of the API into the octanol phase. In conclusion, the combination of in silico and in vitro methods was demonstrated to be powerful for investigating the flow patterns within the miBIdi-pH-II. A comprehensive understanding of the hydrodynamics and the respective influence on the dissolution and apparent partitioning into the octanol phase in the biphasic dissolution model was obtained and completed by using a compartmental kinetic model. This model allowed successful quantification of how the hydrodynamics influence the partitioning of API into the octanol phase.

Evolution of a mini-scale biphasic dissolution model: Impact of model parameters on partitioning of dissolved API and modelling of in vivo-relevant kinetics.

Biphasic dissolution models are proposed to have good predictive power for the in vivo absorption. The aim of this study was to improve our previously introduced mini-scale dissolution model to mimic in vivo situations more realistically and to increase the robustness of the experimental model. Six dissolved APIs (BCS II) were tested applying the improved mini-scale biphasic dissolution model (miBIdi-pH-II). The influence of experimental model parameters including various excipients, API concentrations, dual paddle and its rotation speed was investigated. The kinetics in the biphasic model was described applying a one- and four-compartment pharmacokinetic (PK) model. The improved biphasic dissolution model was robust related to differing APIs and excipient concentrations. The dual paddle guaranteed homogenous mixing in both phases; the optimal rotation speed was 25 and 75rpm for the aqueous and the octanol phase, respectively. A one-compartment PK model adequately characterised the data of fully dissolved APIs. A four-compartment PK model best quantified dissolution, precipitation, and partitioning also of undissolved amounts due to realistic pH profiles. The improved dissolution model is a powerful tool for investigating the interplay between dissolution, precipitation and partitioning of various poorly soluble APIs (BCS II). In vivo-relevant PK parameters could be estimated applying respective PK models.