Validation of non-compartmental analysis (NCA) and bioequivalence results of PKanalix with respect to Phoenix WinNonLin and SAS.
Blaise Pasquiers (1), Virginie Gualano (1), Monika Twarogowska (2), Pauline Traynard (2), Géraldine Cellière(2)
(1) Phinc Development, Massy, France; (2) Lixoft, Antony, France
Objectives: Non-compartmental analysis (NCA) is used at all stages of drug development and is a key method to understand the pharmacokinetic properties of a compound. Bioequivalence analysis of the NCA parameters is a critical step to investigate generic formulations, food effects or drug-drug interactions.
PKanalix, which is part of the MonolixSuite, is dedicated to NCA and CA (compartmental analysis) and includes a bioequivalence module starting from the 2021 version. As NCA and bioequivalence are commonly part of analyses submitted to the regulatory agencies, tools implementing these methods must be appropriately validated. For historical reasons, Phoenix WinNonLin is considered as the standard for NCA and SAS the standard for bioequivalence. In this work, we rigorously compare the results of PKanalix with those of WinNonLin and SAS.
Methods:
For the NCA, the comparison included:
- real PK datasets from literature
- small datasets generated by hand to verify specific implementation aspects
- large simulated datasets using 1-, 2- and 3- compartment models, with oral, iv bolus or iv infusion administration, with or without absorption lag time, with or without BLQ data and with linear or nonlinear elimination. The population parameters varied several folds around a set of typical values. The proportional residual error was set to 4%, 10% and 20%. Urine data was also included.
All datasets were analyzed in PKanalix and WinNonLin. All NCA settings, including the interpolation method, the rule to choose the points to include for the lambda_Z, the rule for the BLQ data, were tested. The calculated NCA parameters calculated with both softwares were then compared.
For the Bioequivalence, the comparison included:
- datasets specifically designed for bioequivalence validation and provided in [1] and [2]
- real PK datasets from literature
- large datasets simulated assuming bioequivalence or bioinequivalence. Parallel, crossover and repeated crossover designs were considered.
For all datasets, the AUC and Cmax NCA parameters were first calculated with PKanalix. These values were then analyzed for bioequivalence in PKanalix and SAS using the PROC GLM routine. For parallel designs, the formulation was included as a fixed factor. For crossover design, all factors were considered as fixed effects by default: sequence, subject, period and formulation. The point estimates and their confidence interval obtained by both software were then compared.
Results: The NCA parameters for all individuals of all datasets are identical (to precision 1e-8) between PKanalix and WinNonLin. All bioequivalence point estimates and confidence intervals are identical (to precision 1e-8) between PKanalix and SAS.
Conclusion: PKanalix provided industry-standard NCA and bioequivalence calculations, with results strictly identical to the two reference softwares. PKanalix is thus a reliable alternative and can be used for submissions to the regulatory agencies. Advantages of PKanalix include a user-friendly interface, interactive tables and plots to explore the results, and interconnection with other applications of the MonolixSuite for population modeling or simulation.
References:
[1] Schütz et al. “Reference Datasets for 2-Treatment , 2-Sequence , 2-Period Bioequivalence Studies.”, 16, 1292–1297 (2014).
[2] Fuglsang et al. “Reference Datasets for Bioequivalence Trials in a Two-Group Parallel Design.”, AAPS J. 17, 400–404 (2015).