NCA, bioequivalence analysis and executed tables

• Cmax ratio T/R101.07% 90% CI 98.89 to 103.30
• AUC(0-t) ratio T/R103.99% 90% CI 98.52 to 109.75
• Bioequivalence set38 of 40 participants
• ConclusionBE shown on unrounded limits

NCA with PKNCA 0.12.1, configured per the SAP: actual times, linear-up/log-down AUC, conc.blq = list(first = "keep", middle = "keep", last = "keep") so BLQ counts as zero everywhere (ICH M13A), lambda z from at least 3 points after tmax. The SAP parameter rules are applied with PKNCA exclude(): AUC(0-inf) is flagged in 2 profiles with more than 20% extrapolated and lambda z-based parameters in 0 profiles with adjusted R-squared below 0.90; the values stay in ADPP (NCAEXRSN) and are not summarized or analyzed. The bioequivalence model is the ICH M13A general linear model (sequence, participant within sequence, period, formulation). Tables use rtables 0.6.17 and tern 0.9.13; RTF comes from formatters::export_as_rtf() (formatters 0.5.13 with r2rtf 1.3.1).

What the ICH M13A BLQ setting changes

PKNCA's default drops a BLQ value that sits between quantifiable concentrations; ICH M13A treats it as zero. One profile has such a value (scenario 6), so the same NCA was also run with the PKNCA default. The values that differ:

Cmax fell at the first post-dose sample in 2 evaluable profiles (first_point_cmax.csv). ICH M13A asks for a discussion of robustness, so a supplementary analysis removes those participants from the bioequivalence set; its result is in the report and in be_supplementary.csv.

Results report

report/report.html: the Quarto report (rendered by run_all.R) with the figures, every table, the exclusions and the ICH M13A acceptability checks.

Tables

Figures

Result files

nca_parameters.csv (one row per profile), nca_settings.json (options and parameter rules), blq_rule_effect.csv, be_results.csv, be_results.json, be_anova.csv, be_supplementary.csv, first_point_cmax.csv, pk_checks.csv, analysis_sets.csv. Programs: 01_nca.R, 02_be.R, 03_tlf.R, report/report.qmd.