Longitudinal time to positivity is unable to predict clinical outcome for drug susceptible tuberculosis
Pieter Van Brantegem (1), Belén P. Solans (1), Rada Savic (1)
(1) University of California San Francisco, California, USA
Introduction
Sputum cultures, measuring bacterial load in liquid culture medium, are the gold standard biomarker to assess tuberculosis (TB) treatment success (1). Our goal was to investigate whether longitudinal time to positivity (TTP) can serve as a surrogate end point for the clinical outcome, defined as tuberculosis free survival at 12 months after randomization.
Methods
S31/A5349 is a randomized controlled phase 3 clinical trial comparing two 4-month regimens containing rifapentine with (HPZM) or without moxifloxacin (HPZE) to the 6-month standard of care regimen (HRZE) for drug susceptible TB (DSTB) (n = 2343). Sputum cultures were collected at each study visit (enrollment, 2, 4, 8, 12, 17, 22 and 26 weeks). Longitudinal TTP was measured up to 8 weeks since the start of treatment (2).
Nonlinear mixed effects models with linear and non-linear functions were explored to describe TTP. Parameters were estimated with NONMEM v7.5 using the Laplace method. Values above the upper limit of quantification of 44 days were censored and modeled accordingly using the M3 method. Models were evaluated using plausibility and precision of the parameter estimates, drop in objective function value (≥ 3.84 points at p ≤ 0.05) and diagnostic plots (200 simulations for visual predictive check).
Results
A linear biphasic function with three mixture groups for baseline TTP and the initial slope described the data best. In addition, interindividual variability on baseline TTP (24.2%, 20.6% and 22.1% for HRZE, HPZE and HPZM, resp.) and proportional residual error (32.1%, 31.2% and 31.5% for HRZE, HPZE and HPZM, resp.) was included. The three mixture groups described patients with (1) high baseline TTP value (12.4 [7.6%], 12.1 [4.2%] and 11.9 [3.9%] days for HRZE, HPZE and HPZM, resp.) and steep slope (24.0 [13.6%], 28.8 [11.6%] and 23.2 [9.9%] days to positivity per weeks on treatment [DPWT] for HRZE, HPZE and HPZM, resp.), (2) low baseline TTP value (8.7 [3.5%], 8.1 [2.5%] and 7.7 [3.2%] days for HRZE, HPZE and HPZM, resp.) and intermediate slope (8.4 [4.6%], 8.8 [2.7%] and 9.4 [4.1%] DPWT for HRZE, HPZE and HPZM, resp.), and (3) low baseline TTP value (7.6 [1.8%], 7.1 [2.3%] and 7.1 [2.3] days for HRZE, HPZE and HPZM, resp.) and low slope (2.3 [2.4%] , 3.6 [2.5%] and 4.0 [2.7%] DPWT for HRZE, HPZE and HPZM, resp.). The second slope was estimated to be 15.7 (48.6%), 13.9 (24.5%) and 18.7 (22.0%) DPWT for HRZE, HPZE and HPZM, resp. The inflection points for the change in slope were estimated to be 6.2 (16.7%), 6.1 (10.9%) and 6.2 (8.2) weeks on treatment for HRZE, HPZE and HPZM, resp.
Based on previous analyses of the clinical outcome, patients were categorized into clinical phenotypes. Of the hard-to-treat participants, 89.9%, 64.8% and 62.4% are in the low baseline TTP, low slope mixture group while 14.6%, 23.9% and 39.9% of the easy-to-treat participants are in the high baseline TTP, steep slope mixture group for HRZE, HPZE and HPZM, resp. Considering the proportion of participants per clinical phenotype assigned to each mixture group suggests that HPZE and HPZM both are superior to HPZE. Moderate-to-treat participants are mainly in the mixture groups with low baseline TTP value, and low or intermediate slope. Only 13.6%, 17.3% and 22.0% of the participants belong to the high baseline TTP with steep slope for each arm, resp.
The median TTP slopes of the raw data up to 6 weeks in easy-to-treat (4.3, 6.5 and 6.9 DPWT), moderate-to-treat (4.4, 6.0 and 6.2 DPWT) and hard-to-treat (2.8, 5.1 and 5.2 DPWT) participants for HRZE, HPZE and HPZM, resp, indicate a difference between the HRZE arm with respect to the other two arms (HPZE and HPZM), confirming that the later regimens are more efficacious when looking at longitudinal TTP.
Conclusion
We developed a model with a linear biphasic function describing the longitudinal course of TTP in participants with DSTB using three mixture groups. There is a partial overlap of the mixture groups and the easy, moderate or hard-to-treat clinical phenotype, indicating that other underlying factors may determine the course of TTP. Our results show that participants in HPZE and HPZM reach negative sputum cultures faster, irrespective of their clinical phenotype. As a result, TTP does not reflect clinical outcome which showed HPZM to be non-inferior to HRZE while HPZE is inferior (3). Therefore, TTP cannot serve as a surrogate end point for clinical outcome.
References:
[1] Falzon D, Jaramillo E, Schünemann HJ, Arentz M, Bauer M, Bayona J, et al. WHO guidelines for the programmatic management of drug-resistant tuberculosis: 2011 update. Eur Respir J. 2011 Sep;38(3):516–28.
[2] Dorman SE, Nahid P, Kurbatova EV, Goldberg SV, Bozeman L, Burman WJ, et al. High-dose rifapentine with or without moxifloxacin for shortening treatment of pulmonary tuberculosis: Study protocol for TBTC study 31/ACTG A5349 phase 3 clinical trial. Contemp Clin Trials. 2020 Mar;90:105938.
[3] 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 May 6;384(18):1705–18.