Synthetic routes to drug substances can result in the introduction of potentially mutagenic impurities (PMIs). The ICH M7 guideline offers a range of control options that assures that the level of this impurity in the drug substance and drug product is below the acceptable limit. Control option 4 leverages the use of predicted purge and outlines the method of utilizing chemical knowledge, literature evidence, and process knowledge to predict the purge of a PMI during drug substance synthesis. If the predicted levels of an impurity in the API are sufficiently lower than the acceptable limit, there will be no requirement for routine analytical testing. Mirabilis is an in silico tool that offers a standardized and conservative approach for the purge calculation of PMIs. The recent developments to methodology for assessing reactivity-based purges within Mirabilis aim to make predictions in a manner more consistent with how a chemist would assess the same situation. The condition-based approach considers the reactants and reagents present and how they may interact, bringing about significant improvements to the specificity and applicability of predictions versus the previous transformation-based approach. Purge predictions for reactivity are now available for 30 impurity types, including N-nitroso compounds and secondary aliphatic amines. Three case studies demonstrate how the new approach provides purge calculations that better align with expert users due to the increased specificity of the predictions.

Synthetic routes to drug products typically introduce several potentially mutagenic impurities (PMIs) which require controlling to a safe level in the final drug substance, generally directed by the control options within the ICH M7 guideline. These impurities are most commonly introduced due to their specific synthetic utility; however, the formation of a PMI can also occur indirectly from a combination of otherwise nonmutagenic sources, as was the case for NDMA within valsartan. Identifying these formation risks currently relies on manually assessing the synthetic route, a process requiring extensive knowledge and potentially liable to oversight. Herein we report on the development of functionality within an in silico risk assessment tool to facilitate the identification of synthetic stages which introduce the risk of formation of PMIs for industrial and/or regulatory users.