The US Food and Drug Administration (FDA) released a new draft guidance on Population Pharmacokinetics (popPK) in July of 2019. The original popPK guidance was published in 1999, making it 20 years since the guidance had been updated. A wealth of published results have become available since 1999, much of it helping to clarify what factors can support having confidence in the results of a popPK analysis.
These published results help to avoid pitfalls of what can lead to bias, or simply wrong answers, as well as streamline the review process. These updates were highly anticipated and desired since the emphasis on popPK analyses in drug development has been steadily increasing since the original guidance was published.
PopPK analysis is frequently used to guide drug development and inform recommendations on therapeutic individualization. The FDA continues to acknowledge the importance of popPK as an essential component across all drug development programs.
The revised, 23-page draft specifically covers popPK analyses submitted as part of new drug applications (NDAs) and biologics license applications (BLAs). These revisions are intended to keep sponsors informed on the FDA’s current thinking on popPK data, analysis, and reporting requirements needed to support regulatory decisions and drug labeling recommendations. The draft guidance addresses the use of popPK analysis in drug development, with further discussions on sections including:
- Dose selection for clinical trials
- Deriving exposure metrics for guiding exposure-response analysis
- Pediatric study designs
- Drug-drug interactions (DDI)
Population Modeling, more formally called mixed effect modeling, was introduced in 1977. “Mixed effect” refers to the statistical idea that the prediction for the observed value (e.g., concentration) is a function of both “fixed effects” (the typical values for parameters in the population) and “random effects” (random variability in parameter values between people).
Frequently, there is not enough data collected on each person to fit a model to their individual data. In this case, popPK analysis is useful to obtain population estimates and an estimate of between-subject variability which minimize the weighted squared difference between the predicted and observed values (e.g., concentration).
Changes from the 1999 Guidance
Compared to the original 1999 guidance, the 2019 draft has been updated to more closely align with current best practices and provide more specific recommendations on data analysis and reporting that will result in more efficient and consistent review of popPK analyses. Some changes to note from the 1999 guidance include:
- Removing the discussion of two-stage approach
- Expanding the discussion of DDI with nested study design discussion
- Increasing emphasis on prespecified questions
- Adding specific report structure recommendations
Analysis Plan & Report Recommendations
The guidance outlines FDA’s recommendations on the general format and content of popPK analysis plans and reports. It also includes a description of the types of scientific and regulatory questions that are appropriate for popPK analysis and recommendations on any labeling statements informed by the results of popPK analyses.
Unlike the European Medicines Agency’s (EMA) 2008 guidance, the FDA guidance does not outline specifically what sections and content should be included in a popPK analysis plan. However, the newly revised FDA guidance states that the analysis plan should include prespecified questions that will be addressed by the popPK analysis and that certain aspects such as planned data omissions and reasons for declaring a data point to be an outlier should also be prespecified in the analysis plan.
The 2019 draft FDA guidance includes a very similar report structure recommendation to the EMA guidance, with the primary goal being efficient and consistent review of popPK analyses by the FDA. The FDA guidance also outlines the expected information to be included within each recommended section. Recommended sections include:
- Executive summary
The updated FDA guidance states that confidence in a given popPK analysis to support an intended objective is increased by the following:
- Understanding of the drug’s PK properties
- Prespecified questions in the study protocol or in the data analysis plan that will be addressed by a popPK analysis
- PK data of sufficient quantity and quality that represents the indicated population and relevant subpopulations of interest
- Good model performance (i.e., the model should describe the data with acceptable bias and precision) and valid for the intended purpose
Data Analysis Recommendations
The section on data analysis provides further direction on the methodological aspects of popPK analysis. The updates include a discussion of simulations based on popPK models, which use a combination of fixed-effect estimates, parameter uncertainty, and estimates of between-subject variability. The updated guidance specifically outlines guiding principles for:
- Preliminary examination of the data
- Model development
- Model validation
For data examination, the FDA recommends evaluating assumptions and checking for highly correlated covariates. The FDA acknowledges that model development methods are constantly evolving, and specifically advising how to conduct popPK analysis is outside of the scope of this guidance. However, it is important that model development methods are explicitly described in the report.
Sponsors should justify the approaches used, including methods for covariate analysis and prespecified criteria regarding missing data or outliers. Specifically, regarding outliers, the FDA recommends refitting the final model to the complete dataset if outliers are excluded and to investigate the influence of the outliers of the final parameter estimates.
For model validation, the guidance recommends a specific list of basic goodness-of-fit plots that should be generated, as well as simulation-based methods such as VPC, pcVPC, or NPC. Model validation is highly dependent on the overall objectives of the analysis, and a fit-for-purpose approach is recommended. The draft guidance also recommends evaluating model performance by numerical metrics such as precision of parameter estimates or condition number. Common methods for internal and external validation, as well as data splitting are mentioned, with a warning to consider the loss of data and power associated with these methods.
Recommendations for simulations depend on the purpose of the simulation, and include simulations based on fixed effects estimates, parameter uncertainty, and estimates of between-subject variability. Additionally, the importance of accounting for correlations between random effects and using reasonable covariate distributions are mentioned.
The new draft guidance on popPK contains more detail than the original guidance, including specific recommendations for data analysis and reporting that should result in a more efficient review process.
Although there is no specific structure recommended for the analysis plan, the newly revised guidance emphasizes the importance of including prespecified questions and other important prespecified methods in an analysis plan. The recommended popPK report structure is similar to that of the EMA guidance, with the exception of a few sections such as an executive summary, which are only recommended by the FDA.
The new draft FDA guidance also includes direction on data analysis methods, such as data examination, model development, model validation, and simulations, while also acknowledging that model development methods and best practice recommendations are constantly evolving.
With constantly evolving methods, it is great to have experts to help you along the way. Contact us to learn more about how Nuventra can help with your program’s popPK needs.