Fig 1.
Flow diagram showing the number of patients randomised to each group, and exclusions from the PK and PK/PD analyses.
Dark blue boxes indicate the primary analysis populations for the final PK and PK/PD analyses, following the exclusion of subjects with significant discrepancies between observed and predicted benznidazole concentrations, which may be explained by treatment misallocations, though the exact cause remains uncertain (n = 5). For the PK/PD sensitivity analysis, all subjects not meeting the PK/PD reliability criteria (PK outliers or uncertain exposure) were excluded from the mITT population, resulting in restricted population of 186 subjects. Abbreviations: EOT, end of treatment; BZN, benznidazole; E1224, fosravuconazole; wks, weeks; QD, once daily; BID, twice daily; mITT, modified intention-to-treat; PP, per-protocol.
Table 1.
Summary of baseline characteristics, compliance and adverse events affecting exposure of the study population.
Fig 2.
Graphical representation of the structural model describing the pharmacokinetics of benznidazole in the BENDITA study.
Absorption from the gut compartment is described by a transit absorption model (with one transit compartment), followed by a one-compartment disposition model. F is the relative oral bioavailability, VC is the apparent volume of distribution of the central compartment, and CL is the apparent elimination clearance. kTR is the rate constant between absorption compartments and was calculated from the mean absorption transit time (MTT) as kTR = (1 + n)/MTT where n is the number of transit compartments.
Table 2.
Parameter estimates of the final population PK model of benznidazole.
Fig 3.
Model diagnostics of the final population PK model for benznidazole in the PK analysis population (n = 175).
a) Goodness-of-fit. Observations are represented by grey circles. Solid black lines represent the line of identity (upper panel) or zero line (lower panel), and the dashed lines represent the local polynomial regression fitting for all observations. (b) Prediction-corrected visual predictive check. Open circles represent observed plasma benznidazole concentrations. The solid red line represents the 50th percentile (median) and the blue dashed lines represent the 5th and 95th percentiles of the observed data. The shaded area represents the 95% CI around the simulated 5th, 50th, and 95th percentiles.
Fig 4.
Distribution of selected exposure variables for benznidazole across treatment arms in the PK analysis population (n = 175).
Boxplots display the distribution of benznidazole exposures in terms of a) AUC∞, b) CMAX, and c) weeks of treatment (a week is counted if at least one dose of benznidazole was taken). Median (midline), interquartile range (IQR, box), and 1.5*IQR whiskers are shown. Abbreviations: BZN, benznidazole; QD, once daily; BID, twice daily; E1224, fosravuconazole.
Fig 5.
Simulated median benznidazole concentrations in DBS versus time for the BENDITA treatment arms.
Simulations were performed for 2000 patients (1000 women and 1000 men, body weight: 65 kg) using the final population PK model. The horizontal red line denotes the in vitro IC90, adjusted for protein binding and scaled to DBS concentrations. The grey shaded area marks the historically accepted therapeutic range: 3-6 mg/L in plasma, scaled to 2.5-5 mg/L in DBS samples.
Fig 6.
Proportions of patients with treatment failure (at least one post-treatment qPCR-positive result) in the mITT (n = 201) and PP population (n = 183).
Color of barplots by benznidazole monotherapy (light blue), co-administration of E1224 (dark blue), or placebo (grey). Numbers above bars indicate patients with treatment failure/ total patients, along with the proportion of patients with treatment failure (%) and 95% Clopper Pearson CIs. No formal statistical comparisons were made between active treatment arms due to small sample sizes and differences in the number of follow-up visits.
Fig 7.
qPCR positivity proportions in the modified ITT (mITT) population (n = 201).
Each dot represents an individual’s proportion of post-treatment qPCR-positive blood samples (‘qPCR positivity’). Symbol color indicates whether patients met all PK/PD reliability criteria (black) or had problematic PK data or uncertain drug exposure (orange). Closed symbols indicate ≥80% benznidazole compliance; open symbols indicate <80% compliance.
Table 3.
Univariate regression model diagnostics for the correlation between benznidazole exposure and qPCR positivity (mITT population).
Fig 8.
Predicted probabilities of qPCR positivity for a single follow-up visit, based on (beta) binomial regression.
Model fits are shown for the modified ITT population, including placebo (n = 201, grey lines and grey shaded area) and excluding placebo as well as one influential outlier (n = 170, red lines and red shaded area). The solid grey and red lines indicate the median, and the shaded areas represent the 95% CI around predicted probabilities. Points represent observed proportions of positive qPCR results after end of treatment (qPCR positivity) in the placebo arm (black) and active treatment arms. a) AUC∞, b) CMAX, c) weeks of treatment (a week is counted if at least one dose of benznidazole was taken). The insert in a) shows regression based on log-transformed AUC∞ (an increment of 1 was added to all AUC∞ values to allow for log-transformation of placebo data). The influential outlier, who took only four doses of 150 mg benznidazole over 8 days, was allocated to the 150 mg benznidazole once daily for 4 weeks treatment arm and is represented by the pink point with 100% qPCR positivity.