Phoenix version 8

Trastuzumab data were simulated using WinNonlin and NLME with Phoenix Version 8.3 (Certara USA, Inc.). We used a published two-compartment population pharmacokinetic model with parallel linear and nonlinear elimination that included the following covariates: body weight (in kg), AST, and albumin^{24 (link)}. An R Software package (Version 4.0, http://www.r-project.org/) was used for data assembly, exploratory data analysis, model diagnosis, and simulation.

The mean, median, standard deviation, 95% confidence interval, and percentiles of simulated trastuzumab plasma concentrations were calculated using Phoenix Version 8.3 (Certara USA, Inc.) and R Software package (Version 4.0, http://www.r-project.org/).

PK/PD parameters, which enable a description of the whole-time course of bacterial kill and growth, can be determined in an in silico model (Nielsen and Friberg, 2013 (link)). Dose fractions of 0, 2.5, 5, 10, 15, 20, and 25 mg/kg for single administration and twice-daily administration were inputted in the PBPK/PD model. Concentration–time curves of ADP and bacterial count–time curves can be simulated. PK/PD parameters (AUC/MIC and %T > MIC) are directly calculated by the PBPK model (Mi et al., 2023 (link)). The cumulative area under the curve of the total bacterial count over 24 h (AUC_{0-24h (Bacterial count)}) was used as a bacteriological effect (Pelligand et al., 2019 (link); Ronaghinia et al., 2021 (link)). PK/PD parameters (AUC/MIC and %T > MIC) versus the bacteriological effect, AUC_{0-24h (Bacterial count)}, were fitted with an inhibitory sigmoid model (Eq. 4). Curve fitting was performed in Phoenix (version 8.3, Certara, United States), and the best PK/PD parameter related to bacteriological effect was selected. $E=E_0-\frac{I_{\mathit{max}}\cdot {INDEX}^N}{{INDEX}^N+{{INDEX}_{50}}^N},$
where E₀ was the effect under a drug concentration of zero. The maximum possible observed effect is I_max. INDEX was the value of the PK/PD parameters (AUC_24h/MIC or %T > MIC). INDEX₅₀ was the value of AUC_24h/MIC or %T > MIC, producing a 50% reduction in I_max, and N was the Hill coefficient that described the steepness of the curve.