Regulatory Submissions for Combination Products

Abstract

Regulatory submissions for combination products -- products comprising a drug, device, and/or biologic constituent part under a single regulatory application -- represent the most complex category of medical product regulatory filing, requiring simultaneous compliance with drug approval standards (NDA/BLA under 21 CFR Parts 314/601), device approval standards (PMA/510(k) under 21 CFR Part 814), and the coordination requirements of the FDA Office of Combination Products that assigns primary mode of action and lead review centre. Manufacturers navigating combination product submissions face multi-centre review, constituent part designation questions, divided technical standard requirements, and post-market surveillance obligations spanning both drug and device regulatory frameworks. This study presents the Combination Product Submission Assessment Framework (CPSAF), evaluating five submission strategy approaches -- constituent part designation with single-application NDA, cross-labelled constituent applications, master file-supported modular submission, streamlined IND-IDE co-submission, and AI-assisted submission readiness assessment -- across four combination product submission contexts: prefilled delivery system NDAs, drug-eluting implant PMAs, digital therapeutic-drug co-submissions, and biologic-device combination BLAs. Our Submission Effectiveness Score (SUBS) integrates review cycle efficiency, technical deficiency rate, cross-centre coordination success, post-market compliance integration, and first-cycle approval rate. AI-assisted submission readiness assessment achieved the highest SUBS (0.926) through NLP models trained on 3,200 FDA review letters that predicted technical deficiencies before submission with 88% accuracy, enabling pre-submission gap closure that reduced first-cycle deficiency letters by 46%, while master file-supported modular submissions achieved the highest cross-centre coordination success (0.955) by separating constituent part technical data into referenced master files that enabled parallel CDER/CDRH review.