Computational Chemistry Approach in the Assessment of Potential Acyl Glucuronide-Mediated Toxicity

Acyl glucuronides are common phase II metabolites of xenobiotics and can sometimes contribute to idiosyncratic toxicities. Their reactivity is primarily mediated through acyl migration and/or nucleophilic displacement, and shorter acyl glucuronide half-lives are associated with increased reactivity. This reactivity can lead to metabolite-induced toxicity, posing a significant risk during drug development.

We developed regression models trained on features derived from Density Functional Theory (DFT) calculations to predict the half-lives of acyl glucuronide metabolites. The aim was to provide a computational tool to guide the design of drug candidates with more stable acyl glucuronide metabolites.

The best-performing model achieved a strong correlation between predicted and experimental half-lives, with an R² of 0.67 on the test set. Predicted half-lives for drugs classified as clinically safe were longer than those for drugs in the warning and withdrawn categories, demonstrating a separation comparable to experimentally measured half-lives.

The model is sufficiently accurate to support the optimization of acyl glucuronides for longer half-lives. Further analysis indicated that acyl glucuronide stability can be modulated by electron-donating and electron-withdrawing groups, effects that are effectively captured by the model.

This modeling approach can be applied during drug discovery to reduce the risk of metabolite-related toxicity by enabling in silico screening of compound modifications and ranking them based on predicted effects on acyl glucuronide half-life.