Development and validation of a time-varying correction factor for QT interval assessment in drug-resistant tuberculosis patients

Background Tachycardia associated with active tuberculosis (TB) often diminishes when patients recover from TB. Elevated heart rate (HR) may lead to suboptimal correction, complicating the assessment of QT prolongation when using standard correction factors (CFs), such as Fridericia's formula (QTcF). Olliaro has proposed a CF for QT interval correction in pretreatment TB patients. However, the QT-HR correlation changes as HR decreases during treatment, indicating the need for time-varying correction. Methods We developed an HR model to capture the HR normalisation during successful treatment. Subsequently, a time-varying CF was constructed using the estimated HR change rate. The performance of CFs to make corrected QT (QTc) independent from HR was evaluated by linear regression analyses of QTc versus HR within defined time bins. Results The final HR model included asymptotic change in HR attributed to time on treatment, circadian rhythm cycles, M2 (bedaquiline-metabolite) concentration, and patient covariates. The time-varying CF decreased from 0.4081 to 0.33, with a half-life of 7.74 weeks. The slope (QTc/HR vs. Time) derived from the time-varying correction was not significantly different from 0 (95% CI -0.003 to 0.002), and the intercept was not significantly different from 0 (95% CI -0.089 to 0.006), demonstrating successful QT correction from pretreatment to the end of treatment. Conclusion The time-varying CF effectively captures the dynamic QT-HR relationship during TB treatment, reducing the risk of misdiagnosing QT prolongation or unnecessary discontinuation of treatment. By addressing underestimation and overestimation issues in QT interval assessment, this method enhances drug evaluation in clinical trials and supports improved treatment decisions for TB patients.