Optimal longitudinal QT interval correction method considering changes in heart rate variability in patients treated for drug-resistant tuberculosis
Thanakorn Vongjarudech (1), Anne-Gaëlle Dosne (2), Bart Remmerie (2), Kelly E Dooley (4), James C M Brust (5), Gary Maartens (6), Graeme Meintjes (7), Mats O Karlsson (1), Elin M Svensson (1,3)
(1) Department of Pharmacy, Uppsala University, Sweden (2) Janssen R&D, Beerse, Belgium (3) Department of Pharmacy, Radboud university medical center, The Netherlands (4) Division of Infectious Diseases, Vanderbilt University Medical Center, Nashville, Tennessee, USA. (5) Albert Einstein College of Medicine and Montefiore Medical Center, Bronx, NY. (6) Division of Clinical Pharmacology, Department of Medicine, University of Cape Town, Cape Town, South Africa. (7) Wellcome Centre for Infectious Diseases Research in Africa, Institute of Infectious Disease and Molecular Medicine, University of Cape Town, Cape Town, South Africa.
Objective: Active tuberculosis (TB) is associated with tachycardia which generally diminishes with treatment. As the typical heart rate (HR) in a patient with TB changes over time on treatment, the QT-HR correlation also changes. Standard constant correction factors (CF) for QT interval, such as Bazett's (0.5)[2] and Fridericia's (0.33)[3], may result in sub-optimal correction in this population [1, 4, 5], making it challenging to assess drug effects on QT. A CF of 0.4081 has been proposed by Olliaro to optimally correct the QT interval in patients with TB pre-treatment [1]. This project aimed to establish a time-dependent correction method that optimally accounts for gradual changes in HR during the treatment period.
Methods: We developed an HR model and a time-varying correction method using data from 440 patients from two phase IIb clinical trials: C208 (2-stage, randomized, double-blind, placebo-controlled)[6, 7] and C209 (single-arm, open-label)[8]. Patients received bedaquiline (BDQ) or a placebo for 8 weeks (Stage 1, C208) or 24 weeks (C208 Stage 2 or C209) on top of a background regimen. Baseline and on-treatment ECG measurements were included. BDQ and M2 (main metabolite of BDQ) concentrations were predicted using a published pharmacokinetic model [9]. Initially, we developed an HR model to describe the change in HR during treatment. We then constructed the time-varying CF utilizing a parameter from the HR model describing the rate of change in HR. To evaluate the CF, we performed linear regression analyses in time intervals using QTc and HR, expecting the slope and r2 to be close to 0 for successful correction. Two independent studies were used for external validation: i) The A5343 DELIBERATE study (N=82), a phase 2, open-label, randomized, controlled trial, in patients receiving BDQ, delamanid, or both for 24 weeks in addition to background treatment [10] and ii) The PROBeX study (N=195), a prospective cohort study of patients in South Africa receiving a BDQ-containing regimen for 24 weeks [11]. Linear regression of the interval-specific slopes (the correlation between HR and QTc) accounting for standard error in the estimate was used to compare the performance of the new time-varying CF and established constant CFs, utilizing data from all studies. The model development and simulation were performed using NONMEM 7.5 aided with PsN 5.3.0, followed by visualization and analysis in R 4.2.2.
Results: The final HR model included a component describing an asymptotic change in heart rate with time from study start, 24 & 12-hour circadian rhythm cycles, effect of M2 (Emax-model), and patient covariates. The estimated baseline HR was 77.8 beats/min (95%CI 75.7-79.8), decreasing to a recovered HR of 70.7 beats/mins (95%CI 67.7-73.7) at a steady state. Although the M2 effect was statistically significant, the magnitude of the effect was mild (4% [95%CI 3-7] decrease at the maximum observed concentration). The estimated time to reach 50% of recovered HR from baseline HR(Tprog) was 7.74 weeks (95%CI 5.09-10.35). A sensitivity analysis showed that the estimated Tprog from a model without covariates was close to that of the final model (8.15 weeks, 95%CI 5.48-10.87). In the validation datasets, the estimated Tprog for A5343 and PROBeX studies were comparable, 8.04 (95%CI 5.52-10.55) and 8.63 (95%CI: 0.34-16.88) weeks, respectively. The estimated Tprog was utilized to construct the time-varying CF by assuming that the estimated rate of recovery for HR represents the rate of change for the CF as per the following formula: CF(t) = 0.4081 - (0.0781)*(1-e^(-log(2)*t/7.74)). Hence, the CF decreases asymptotically from 0.4081 (Olliaro’s)[1] towards 0.33 (Fridericia’s)[3] over the time on treatment (t). The evaluation showed that the overall slope derived from the time-varying CF was not different from 0 (-0.008 [95%CI -0.004-0.003]), whereas the slope derived from QTcF was 0.013 (95%CI 0.009-0.016).
Conclusion: The newly developed time-varying CF can capture the natural change in QT-HR correlation that occurs during TB treatment, enhancing accuracy in QT prolongation determination. It may alleviate the problem of QTcF underestimating the QT interval in early treatment and reduce the overestimation of QTc change from the baseline associated with QTcF. This could result in a more informative analysis of drug effects on QT in clinical trials and better treatment decisions in individual patients with TB.
References
[1] Olliaro PL, Merle C, Mthiyane T, et al. Effects on the QT Interval of a Gatifloxacin-Containing Regimen versus Standard Treatment of Pulmonary Tuberculosis. Antimicrob Agents Chemother 2017; 61: e01834-16.
[2] Bazett HC. AN ANALYSIS OF THE TIME-RELATIONS OF ELECTROCARDIOGRAMS. Annals of Noninvasive Electrocardiology 1997; 2: 177–194.
[3] Fridericia LS. The Duration of Systole in an Electrocardiogram in Normal Humans and in Patients with Heart Disease. Noninvasive Electrocardiol 2003; 8: 343–351.
[4] Department of Health and Human Services, Food and Drug Administration. Guidance for Industry E14 Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs. Fed Regist 2005; 70: 61134-61135.
[5] Li H, Salinger DH, Everitt D, et al. Long-Term Effects on QT Prolongation of Pretomanid Alone and in Combinations in Patients with Tuberculosis. Antimicrobial Agents and Chemotherapy 2019; 63(10): e00445-19.
[6] Diacon AH, Donald PR, Pym A, et al. Randomized pilot trial of eight weeks of bedaquiline (TMC207) treatment for multidrug-resistant tuberculosis: long-term outcome, tolerability, and effect on emergence of drug resistance. Antimicrob Agents Chemother 2012; 56: 3271–3276.
[7] Diacon AH, Pym A, Grobusch MP, et al. Multidrug-Resistant Tuberculosis and Culture Conversion with Bedaquiline. N Engl J Med 2014; 371: 723–732.
[8] Pym AS, Diacon AH, Tang S-J, et al. Bedaquiline in the treatment of multidrug- and extensively drug-resistant tuberculosis. Eur Respir J 2016; 47: 564–574.
[9] Svensson EM, Dosne AG, Karlsson MO. Population Pharmacokinetics of Bedaquiline and Metabolite M2 in Patients With Drug-Resistant Tuberculosis: The Effect of Time-Varying Weight and Albumin. CPT: pharmacometrics & systems pharmacology; 5. 2016; 5(12): 682-691.
[10] Dooley KE, Rosenkranz SL, Conradie F, et al. QT effects of bedaquiline, delamanid, or both in patients with rifampicin-resistant tuberculosis: a phase 2, open-label, randomised, controlled trial. The Lancet Infectious Diseases 2021; 21: 975–983.
[11] Brust JCM, Gandhi NR, Wasserman S, et al. Effectiveness and Cardiac Safety of Bedaquiline-Based Therapy for Drug-Resistant Tuberculosis: A Prospective Cohort Study. Clin Infect Dis 2021; 73: 2083–2092.
