How much should we still worry about QTc prolongation in rifampicin-resistant tuberculosis? ECG findings from TB-PRACTECAL clinical trial

Abstract

Regimens for the treatment of rifampicin-resistant tuberculosis currently rely on the use of QT-prolonging agents. Using data from the randomized controlled trial, TB-PRACTECAL, we investigated differences in QTcF among participants in the three interventional arms: BPaL (bedaquiline, pretomanid, and linezolid), BPaLC (BPaL with clofazimine), and BPaLM (BPaL with moxifloxacin). Additionally, we assessed whether age, body mass index, and country were causally associated with QTcF prolongation. The trial included participants from South Africa, Uzbekistan, and Belarus. A post hoc analysis of electrocardiogram data was undertaken. Random effects regression was used to model QTcF longitudinally over 24 weeks and causal frameworks guided the analysis of non-randomized independent variables. 328 participants were included in BPaL-based arms. The longitudinal analysis of investigational arms showed an initial QTcF steep increase in the first week. QTcF trajectories between weeks 2 and 24 differed slightly by regimen, with highest mean peak for BPaLC (QTcF 446.5 ms). Overall, there were 397 QTcF >450 ms (of 3,744) and only one QTcF >500 ms. The odds of QTcF >450 ms among participants in any investigational arm, was 8.33 times higher in Uzbekistan compared to Belarus (95% confidence interval: 3.25-21.33). No effect on QTcF prolongation was found for baseline age or body mass index (BMI). Clinically significant QTc prolongation was rare in this cohort of closely monitored participants. Across BPaL-based regimens, BPaLC showed a slightly longer and sustained effect on QTcF prolongation, but the differences (both in magnitude of change and trajectory over time) were clinically unimportant. The disparity in the risk of QTc prolongation across countries would be an important factor to further investigate when evaluating monitoring strategies.

Clinical trials: This study is registered with ClinicalTrials.gov as NCT02589782.

Keywords: BPaL/M; DR-TB; bedaquiline; cardiotoxicity; drug-resistant tuberculosis.

Fig 1

Observed and modeled QTcF values over the 24-week follow-up time. Top panel: Observed QTcF summarised as box plots. Time points from 0 to 24 weeks are equally spaced and do not reflect the true spacing between visits and the y-axis extends from 330 to 550 ms. The asterisks indicate the mean. The symbol ‘+’ on the x-axis represents a change of scale from 1 to 4 weeks. These data, plotted as spaghetti plots for x-axis as continuous scale, from 0 to 108 weeks, are included in (Appendix A5). Bottom panel: Modeled QTcF values, with the x-axis extending from 0 to 24 weeks (with time points reflecting the true spacing between visits) and the y-axis extending from 350 to 470 ms. Strong evidence for interaction between BPaLC and the second term for time [ln(time)] (P value < 0.001), is indicated by non-parallel trajectories of QTcF for BPaLC after the turning point.