FDA regulatory experience with drug manufacturing process models in approved applications

Drug manufacturing process models have been successfully applied to process development and control in regulatory submissions in at least 15 FDA-approved applications from 8 companies between 2012-2025. These submissions span drug substance and drug product manufacturing in approved NDAs, BLAs (including biosimilars), originals, and supplements. Eleven model-supported applications include continuous manufacturing, which requires an understanding of material residence time distribution (RTD) through a process. Models in approved submissions have been applied to design space establishment and real-time process control. Multiple models have often been used in a single process control strategy. FDA has encountered mechanistic (e.g., mass balance), empirical (e.g., NIR/Raman chemometric), and hybrid process models (e.g., RTD). The amount of information about a process model in a regulatory submission has generally been commensurate with its determined risk or impact level. Models in approved FDA applications cover risk levels of high (e.g., release decisions, parametric control), medium (e.g., in-process controls, non-conforming material diversions), and low (e.g., design space definition, investigations/monitoring). FDA's modeling experts are often consulted when evaluating an applicant's modeling and control strategy to support an application decision, which at times has included building in-house FDA models calibrated using submitted experimental data. Excluding three submissions with issues unrelated to modeling (e.g., clinical matters), drug submissions using process models were approved prior to their goal date by an average of 20.3 days, including several under accelerated review. Sharing FDA's regulatory experience can advance global understanding of drug manufacturing process model implementation to foster transparency and further innovation and policy development.