Forensic Pharmacometrics: Part 2 - Deliverables for Regulatory Submission
Thaddeus H. Grasela* (1), Jill Fiedler-Kelly (1), Elizabeth A. Ludwig (1), Julie A. Passarell (1), Darcy J. Hitchcock (1)
(1) Cognigen Corporation, Buffalo, New York, USA
Introduction: As modeling and simulation results become increasingly integral to critical development-related decision-making and program outcomes, the consequences of poor documentation of pharmacometric analyses can jeopardize the role of pharmacometrics in contributing to the transition to model-based drug development. While the EMEA and FDA Population PK Guidance documents recommend pharmacometric report content, forensic assessment of analysis inputs and outputs may enable the development of standards to define measures of acceptability and support the continued evolution of these methods.
Objective: Define and apply a process for the prospective forensic assessment of regulatory deliverables to gain understanding of common problems.
Methods: A review of recent externally-generated pharmacometric analysis inputs (analysis-ready datasets, analysis plans) and outputs (models, final technical reports), intended for submission to regulatory authorities, was performed using a systematic process for forensic assessment. For each deliverable, descriptive statistics summarizing categories of common problems were generated.
Results: The process included the following steps: (1) initial review and identification of issues for further investigation, (2) request for additional supporting information, (3) verification of consistency of supporting information, and (4) suggested strategy for correction. For analysis-ready datasets, the supporting information may include source data collection methods, additional exploratory graphical displays, or a flowchart of the programming logic applied in the data manipulation process. For analysis plans, a series of questions addressing how likely scenarios would be handled might be generated. For models described in technical reports, consistency between output tables of results, NM-TRAN code, NONMEM® report files, and text describing results can easily be confirmed. Based on this process, the following types of common issues were identified: systematic errors in the creation of dosing histories, incomplete strategies for assumption violations, and numerous inconsistencies in the reporting of modeling results.
Conclusions: The process developed for this assessment can be used as a basis for independent validation of pharmacometric deliverables intended for regulatory submission, as well as in the development of standards for quality assurance activities for pharmacometric analyses.