Zhe Huang1, Jiawei Ye1, Emmanuelle Comets2,3, Maria C. Kjellsson1, Mats O. Karlsson1
1Department of Pharmacy, Uppsala University, 2Univ Rennes, Inserm, EHESP, Irset - UMR_S 1085, 3Université Paris Cité, IAME, Inserm
Introduction/Objectives: Pharmacokinetic/pharmacodynamic (PKPD) modeling and exposure-response (ER) analyses are essential in assessing dosing strategies, personalizing treatment, and optimizing clinical study designs. This review examines and evaluates key aspects of PKPD modeling and ER analysis based on recent publications in clinical pharmacology journals.
Method: A systematic search was conducted to identify articles related to PKPD or ER modelling of clinical data published between 2023 and 2025 in Clinical Pharmacology & Therapeutics (CPT), Clinical Pharmacology & Therapeutics: Pharmacometrics & Systems Pharmacology (CPT:PSP), and the Journal of Pharmacokinetics and Pharmacodynamics (JPKPD). A combined keyword-based and manual search was performed.
Results: A total of 96 articles were identified and included in this evaluation. Among the drugs analyzed, 54% were small molecules, while biologics, antibody-drug conjugates, and peptides accounted for 34%, 5%, and 3%, respectively. The remaining drugs included oligonucleotide, siRNA, and chimeric antigen receptor T-cell therapy (CAR-T). Regarding the clinical endpoints assessed, 47% of the publications focused on multiple endpoints, with the most common combination being efficacy and safety (78%). The remaining 22% of these studies focused on combinations involving either efficacy or safety biomarkers. Additionally, 38% of the studies examined efficacy alone, while 10% focused solely on safety. The remaining 5% of studies investigated biomarkers as their primary endpoint.
Regarding clinical trial phases, 47% of the publications analyzed data across multiple trial phases. Among the remaining 53% that focused on a single clinical trial phase, Phase 1 trials and investigator-initiated trials each accounted for 26% of the studies. Phase 2 and Phase 3 trials contributed 13% and 15%, respectively. The remaining studies utilized registry data or involved seamless clinical trial designs. Notably, only four publications applied dose-response analysis solely, whereas the majority relied on ER analysis.
A variety of exposure metrics were employed in ER analyses. Concentration over time (C(t)) was applied in 37% of the studies, while 26% of studies used a single secondary exposure metrics, such as AUC or AUCss (24%), Cav or C(av,ss) (44%), and other metrics including Cmax, Cmin, Ctrough. In contrast, 37% of the studies explored multiple exposure metrics, often combining AUC with Cmax or Cmin. With regards to the response, 41% of publications evaluated single PD observations per subject, and 59% used repeated PD observations.
The majority of publications (93%) employed the sequential modeling approach to PKPD analysis, either using fixed individual or population PK parameters, the latter approach including PK data in the PKPD analysis. Only 5% of publications used the simultaneous modeling approach and no study reported exploring random effect correlation between PK and PD models. 2% of publications did not apply a PK modeling approach. Logistic models were the most commonly employed for the analysis of binary and categorical PD outcomes. Indirect response models were most frequently utilized for continuous PD data, whereas Cox proportional hazards models were most often applied in the analysis of time-to-event PD outcomes.
Conclusion: It is clear that a majority of the published ER/PKPD modelling does not concern mixed effects modelling, even though such models in general were used to generate the exposure metrics. Sequential modelling dominates as well as exposure metrics such as C(t) and AUC.
Acknowledgement: INVENTS has received funding from the European Union’s Horizon Europe Research and Innovation programme under grant agreement 101136365.
Reference: PAGE 33 (2025) Abstr 11474 [www.page-meeting.org/?abstract=11474]
Poster: Methodology - Other topics