Pharmacokinetic Profiling of Drug-Device Combination Products

Abstract

Drug-device combination products (DDCPs) -- drug-eluting stents, drug-coated balloons, implantable drug delivery systems, prefilled injectors, and transdermal patch-device hybrids -- present unique pharmacokinetic (PK) challenges because drug release, absorption, distribution, and elimination are governed by both the drug's molecular properties and the device's engineering characteristics. The PK profile of a drug-eluting stent depends on polymer degradation kinetics; an implantable pump's delivery profile depends on reservoir design and catheter geometry; a transdermal patch's absorption depends on adhesive formulation and skin contact area. Regulatory agencies (FDA, EMA) require PK characterisation that demonstrates both local tissue concentrations (efficacy) and systemic exposure (safety), yet standard PK modelling approaches designed for oral or injectable drugs do not capture device-mediated release mechanisms. We present the Drug-Device Pharmacokinetic Assessment Framework (DDPAF), evaluating five PK modelling approaches -- compartmental PK with device release functions, physiologically-based PK (PBPK) with device modules, finite element drug transport models, population PK with device covariates, and machine learning PK predictors-- across four DDCP categories (drug-eluting implants, drug-coated surfaces, implantable infusion devices, and transdermal delivery systems). Our PK Modelling Performance Score (PKPS) measures local concentration prediction, systemic exposure prediction, inter-patient variability capture, regulatory acceptance, and computational efficiency. PBPK with device modules achieves the highest PKPS (0.924) through mechanistic integration of device release physics with whole-body pharmacology, while ML PK predictors achieve the highest inter-patient variability capture (0.960) through learning patient-specific PK parameters from sparse clinical sampling.