An Exposure-Response Perspective on the Clinical Dose of Pretomanid

Abstract

Pretomanid was approved by the U.S. FDA, via the limited population pathway for antibacterial and antifungal drugs, as part of a three-drug regimen with bedaquiline and linezolid for the treatment of extensively drug-resistant and treatment-intolerant or nonresponsive multidrug-resistant tuberculosis. The recommended dose of pretomanid is 200 mg once daily with food. The objective of this work was to retrospectively evaluate this recommended dose by means of exposure-response (E-R) modeling applied to outcomes of both efficacy and safety. Cox proportional-hazards modeling was used, with the steady-state average pretomanid concentration as the exposure metric. The efficacy outcome was time to sputum culture conversion (TSCC) to negative. The safety outcomes were times to the first occurrence of adverse events in classes selected from either pretomanid's investigator brochure or the new drug application (NDA) submission as recognized safety signals for pretomanid based on preclinical as well as clinical experience. Significant E-R relationships were found for TSCC and two adverse-event classes, vomiting (a single preferred term) and gastrointestinal (GI) symptoms (a collection of related terms). No significant E-R relationships were found for the single preferred terms nausea, alanine aminotransferase (ALT) increased, aspartate aminotransferase (AST) increased, and headache and for the collections hepatic disorders, transaminases increased, skin and subcutaneous tissue disorders, and headache. The results suggest that the recommended dose of pretomanid, 200 mg given in the fed state, is appropriate over the range of pharmacokinetic exposures.

Keywords: antimicrobial agents; pretomanid; tuberculosis.

FIG 1

Kaplan-Meier plots of time to sputum culture conversion, by quantiles of C_avg.

FIG 2

Hazard ratios of TSCC versus C_avg and distribution of C_avg. Note that in the top panel, the solid curve shows estimated hazard ratios as a function of C_avg relative to the value at a C_avg of 2.6 μg/ml, where the ratio, by definition, is 1. Larger values of the hazard ratio imply a faster time to culture conversion. The dashed curves show the pointwise 95% confidence intervals for the hazard ratios.

FIG 3

Estimated probabilities of failure to culture convert by 8 weeks and 95% confidence intervals, by age and exposure.

FIG 4

Hazard ratios of exposure-related adverse events versus C_avg and distribution of C_avg. Note that in the top two panels, the solid curves show estimated hazard ratios as a function of C_avg relative to the value at a C_avg of 2.6 μg/ml, where the ratio, by definition, is 1. Larger values of the hazard ratio imply a faster time to adverse event occurrence. The dashed curves show the pointwise 95% confidence intervals for the hazard ratios. Conditions underlying the hazard ratios vary between the two AE types, depending on the optimal model found for each. Both are comparable to the data in Fig. 2 in that they apply to the BPaL regimen, although data for other regimens were used to build the model for vomiting, whereas the model for GI symptoms is based on only the BPaL regimen. See the text for more details.