Building a reproducible pharmacometrics environment to ensure quality, continuity, efficiency, and good communication
Oliver Ackaert (1), Marc Asghari (2), Jeike Biewenga (1), Corey Bishop (2), Emily Bozenhardt (2), Muriel Boulton (1), Laure Cougnaud (3), Anne-Gaelle Dosne (1), Nele Goeyvaerts (1), Alan Groot (3), Anne Kuemmel (4), Daniel Lill (4), Dandan Luo (2), Carlos Perez-Ruixo (1), Alberto Russu (5), Frederic Saad (1), Kim Stuyckens (1), Dominique Swerts (1), Huybrecht T’jollyn (1), Kirsten Van Hoorde (3), Xavier Woot de Trixhe (1), Martijn Wouters (1), Jason Zhang (2), Jessie Zhou (2), Wangda Zhou (2), Juan Jose Perez-Ruixo (1)
(1) Janssen R&D, Belgium; (2) Janssen R&D, USA; (3) Open Analytics, Belgium; (4) Intiquan, Switzerland; (5) Janssen-Cilag SpA, Italy
Introduction and objectives: Throughout recent years, several workstreams and initiatives have focused on bringing more structure and standardization in pharmacometric analyses.1,2,3 Although good progress has been made in this field, continuous improvement is required to standardize (parts of) pharmacometric analyses. This will create an environment which is less modeler-dependent and which allows the pharmacometrician to focus more on the scientific part of model-informed drug development, namely quantifying the interaction between drug and (diseased) host to aid efficient drug development and regulatory decision making.4 In addition, maintaining high quality analyses in an ever-changing environment requires pharmacometrics scientists to become more efficient, emphasizing the need for a framework that facilitates reproducible research. The objective of this initiative is to build an environment that allows performing recurrent pharmacometric analyses in a consistent, efficient, and reproducible manner.
Methods: In an initial phase, pharmacometric analyses were selected based on the frequency of their past and future occurrence and the ability to standardize (part of) these analyses. In a second phase, a framework was built for each type of analysis. Reproducibility was considered as a continuum from beginning to end of analyses, e.g., starting at the planning phase, continuing to the execution phase and finally to the reporting and submission phases. To this end, templates for analysis plans, trial specific data transfer agreements (tsDTA), and analysis reports were created. In conjunction, different R-packages were built to facilitate the analysis, post-processing, presentation, and reporting of results.
Results: In this framework, start-to-end workflows have been created for four analysis types: (1) population PK(-PD) analyses; (2) exposure-response analyses for time-to-event or dichotomous endpoints; (3) concentration-QT analyses according to the recently issued ICH-E14 guidelines5; and (4) pediatric dose selection based on matching adult exposures, assessment of the sample size and associated power for pediatric studies to evaluate doses, and exposure evaluation when pediatric PK data is available. This resulted in 5 analysis plan templates, 3 tsDTA templates, 5 report templates and 7 R-packages, including documentation (i.e. vignettes, help pages), tutorials, and Rmarkdown templates. The R-packages are integrated in our high-performance cluster-based platform at the core of our pharmacometrics computing environment. The framework has reduced the time to plan, conduct, QC, and report analyses for numerous projects. In addition, integration of these analyses in a general workflow improved reproducibility and traceability. Furthermore, standardization also has been beneficial to learn and understand the different analyses, and to improve communication with the different internal and external stakeholders such as clinical teams and service providers.
Conclusions: An environment to conduct pharmacometrics analyses in a generalized workflow, with integrated templates and internally developed R-packages, facilitates their execution in a more standardized and efficient manner. Moving forward, this framework will continue to evolve and expand, ensuring further integration in our pharmacometrics environment and enhancing drug development support.
References:
[1] Wilkins JJ, Jonsson EN (2013) Reproducible pharmacometrics [tutorial]. PAGE 22 Abstr 2774. www.page-meeting.org/?abstract=2774
[2] Radivojevic, A., Corrigan, B., Downie, N. et al. Data standards for model-informed drug development: an ISoP initiative. J Pharmacokinet Pharmacodyn 45, 659–661 (2018). https://doi.org/10.1007/s10928-018-9595-8
[3] https://www.intiquan.com/acop9-workshop-efficient-reporting/
[4] https://www.fda.gov/about-fda/center-drug-evaluation-and-research-cder/division-pharmacometrics#Overview
[5]: https://www.fda.gov/regulatory-information/search-fda-guidance-documents/e14-and-s7b-clinical-and-nonclinical-evaluation-qtqtc-interval-prolongation-and-proarrhythmic