Vincent Chang
University of California San Francisco
Objectives:
A 17-week rifapentine-moxifloxacin-based regimen has been recently shown to have noninferior efficacy to the standard 26-week regimen for the treatment of drug-susceptible pulmonary tuberculosis (TB), CDC Study 31/ NIH A5349. In the published results of the S31/A5349 trial, the 17-week rifapentine-moxifloxacin-based regimen had 5.7% TB-related unfavorable outcomes compared to 3.1% for the standard 26-week regimen [1]. Analyses of contemporary TB treatment shortening clinical trials, however, have shown that among subgroups of participants with varying degrees of disease severity, higher-risk subgroups experience unacceptably high rates of TB-related unfavorable outcomes [2]. Treating all patients with one-size-fits-all regimens is suboptimal. New clinical trials are needed to evaluate stratified medicine approaches for the treatment of TB.
Methods:
We used previously developed and integrated parametric time-to-event models to simulate individual patient clinical outcomes (favorable outcome: durable cure, or TB-related unfavorable outcome: treatment failure, relapse, or death) from patient demographics, baseline disease burden markers, and assumed treatment hazard ratios (based on regimen dosage and duration). We sampled patient phenotypes from the TB-ReFLECT database (N=3411) and stratified these into lower- and higher-risk groups based on patient demographics and baseline disease burden [3].
We conducted clinical trial simulations to assess receiver operating characteristics and sample size requirements for a phase 2 trial aimed at evaluating optimal durations of the rifapentine-moxifloxacin regimen within each risk stratum. We used a duration-randomization design, whereby participants are randomized to different durations of the same regimen or a fixed-duration control regimen. The objective of such a trial would be to demonstrate that a duration-response relationship exists, and to describe its shape. We adapted the MCP-Mod dose-finding methodology for use in duration-finding trials [4]. We explored nine candidate models for the duration-response relationship, including linear and Emax models.
Results:
In the TB-ReFLECT outcomes analysis, three different 17-week fluoroquinolone-based regimens with lower adherence than S31/A5349 had 19.1% and 31.2% TB-related unfavorable outcomes in lower- and higher-risk groups, respectively [2]. Further, the TB-ReFLECT treatment duration analysis suggested that among participants receiving isoniazid- and moxifloxacin- containing regimens, lower-risk participants can receive as little as 12 weeks of treatment and higher-risk participants require up to 26 weeks of treatment to achieve fewer than 5% TB-related unfavorable outcomes [3]. Based on these results, we propose a trial design to test a stratified approach to the 17-week rifapentine-moxifloxacin-based regimen that will evaluate five durations for the lower-risk stratum: 8, 10, 12, 14, and 16 weeks. A single duration of 26 weeks will be evaluated for the higher-risk stratum.
Assuming 12% TB-related unfavorable outcomes for the minimum duration of the lower-risk stratum (8 weeks) and 3% for the maximum duration (16 weeks), a sample size of 104 participants per duration in the lower-risk stratum (total lower-risk stratum sample size of 625, assuming 10% censoring and a 5:1 allocation ratio to the intervention arm versus the control arm) provides 80% power to detect a duration-response relationship.
For the higher-risk stratum, assuming 2% TB-related unfavorable outcomes with the 26-week rifapentine-moxifloxacin-based intervention and 7.5% for the standard 26-week control regimen, a sample size of 155 participants, assuming 10% censoring and a 5:1 allocation ratio, provides >90% power to demonstrate that the intervention has <7.5% TB-related unfavorable outcomes.
Conclusions:
We contend that stratified medicine approaches will be essential to ending the TB epidemic. Current TB regimen development pipelines focus solely on the development of novel regimens at a single treatment duration. We describe a feasible and practical duration-randomization phase 2 trial design to facilitate prospective evaluation of stratified treatment strategies.
References:
[1] Dorman SE, Nahid P, Kurbatova EV, Phillips PPJ, Bryant K, Dooley KE, et al. Four-Month Rifapentine Regimens with or without Moxifloxacin for Tuberculosis. N Engl J Med. 2021;384(18):1705-18.
[2] Imperial MZ, Nahid P, Phillips PPJ, Davies GR, Fielding K, Hanna D, et al. A patient-level pooled analysis of treatment-shortening regimens for drug-susceptible pulmonary tuberculosis. Nature Medicine. 2018;24(11):1708-15.
[3] Imperial MZ, Phillips PPJ, Nahid P, Savic RM. Precision-Enhancing Risk Stratification Tools for Selecting Optimal Treatment Durations in Tuberculosis Clinical Trials. Am J Respir Crit Care Med. 2021;204(9):1086-96.
[4] O’Quigley J, Iasonos A, Bornkamp B. Handbook of methods for designing and monitoring dose finding trials. Boca Raton: Chapman & Hall/CRC; 2017.
Reference: PAGE 30 (2022) Abstr 10198 [www.page-meeting.org/?abstract=10198]
Poster: Drug/Disease Modelling - Infection