Simon E Koele1 , Rob Aarnoutse1, Elin M Svensson1,2
1 Department of Pharmacy, Radboud Institute for Health Sciences, Radboud University Medical Center, Nijmegen, the Netherlands 2 Department of Pharmacy, Uppsala University, Uppsala, Sweden.
Background: A milestone in the development of novel anti-TB drugs is demonstrating the early bactericidal activity (EBA) in a phase IIa trial. No standardized method for EBA determination has been established but serial sputum measurements of the colony-forming units (CFU) on solid culture and/or time-to-positivity (TTP) in liquid culture over the first 14 days of treatment are commonly used in recent years. A bi-phasic decline of the CFU, and incline of TTP, over time on treatment is often observed. The EBA of a drug is thought to be determined by the exposure and is often essential to inform the selection of dose that will be carried forward into later-stage clinical trials. Therefore, adequate power to characterize the exposure-response relationship of a drug in an EBA study is imperative.
Objective: To investigate the power to detect the exposure-response relationship of a drug with model-based analysis of CFU or TTP data gathered in phase IIa clinical trials.
Methods: We performed an in silico simulation study of phase IIa TB clinical trial CFU and TTP measurements over 14 days. The choice of simulation models was based on EBA data gathered in the HIGHRIF1 trial (clinicaltrials.gov NCT01392911) (1, 2). Both CFU and TTP data were collected from sputum-smear-positive TB patients over 14 days of treatment. In total 796 CFU measurements (of which 61 negatives) and 800 TTP measurements (of which 8 negatives) were collected. Non-linear mixed-effects modeling was used to evaluate the most suitable pharmacodynamic models to describe the data. Linear, bi-linear, smooth bi-linear, and semi-mechanistic time-to-event models were evaluated (3, 4). The pharmacodynamic models were subsequently coupled to a one-compartmental pharmacokinetic model of four hypothetical drugs with the same pharmacokinetic profile and a range of efficacy profiles driven by the simulated drug exposures. Drug A) had a fast initial and fast terminal killing rate, drug B) had a fast initial and slow terminal killing rate, drug C) had a slow and fast terminal killing rate, and drug D) had a slow initial and slow terminal killing rate. To simulate phase IIa clinical trial EBA data, each of the four hypothetical drugs was dosed daily in four dosing groups, and sampling of CFU and TTP was simulated on days 0,1,2,3,4,5,7,10 and 14. An Emax model described the exposure-response relationships. Monte-Carlo mapped power analysis was used to identify the power of detecting the drug effect for each drug (5).
Results: The bi-linear model with estimated node-point was found to be the best model to describe both the CFU and TTP data from HIGHRIF1. This model structure and parameters were used for simulations. Coupling of CFU and TTP to each other was performed via shared killing rates and inter-individual variability of the killing rates of the underlying bacterial load and a patient. In the evaluated scenarios, TTP models showed higher power to detect the exposure-response relation than the CFU models. The power to detect an Emax exposure-response effect at 25 patients per arm for TTP and CFU respectively was 81% and 51% for drug A), 60% and 17% for drug B), 68% and 31% for drug C), and 29% and 1% for drug D).
Conclusions: Currently, most EBA studies include between 10-15 patients per arm, which results in limited power to identify an exposure-response relationship according to our results. We will investigate whether a multi-dimensional model, in which CFU and TTP are analyzed together, can increase the power to identify the exposure-response relationship.
References:
[1] Boeree MJ, Diacon AH, Dawson R, Narunsky K, du Bois J, Venter A, et al. A dose-ranging trial to optimize the dose of rifampin in the treatment of tuberculosis. Am J Respir Crit Care Med. 2015;191(9):1058-65.
[2] Te Brake LHM, de Jager V, Narunsky K, Vanker N, Svensson EM, Phillips PPJ, et al. Increased bactericidal activity but dose-limiting intolerability at 50 mg·kg(-1) rifampicin. Eur Respir J. 2021;58(1).
[3] Burger DA, Schall R. A Bayesian Nonlinear Mixed-Effects Regression Model for the Characterization of Early Bactericidal Activity of Tuberculosis Drugs. J Biopharm Stat. 2015;25(6):1247-71.
[4] Svensson EM, Karlsson MO. Modelling of mycobacterial load reveals bedaquiline’s exposure-response relationship in patients with drug-resistant TB. J Antimicrob Chemother. 2017;72(12):3398-405.
[5] Vong C, Bergstrand M, Nyberg J, Karlsson MO. Rapid sample size calculations for a defined likelihood ratio test-based power in mixed-effects models. Aaps j. 2012;14(2):176-86.
Reference: PAGE 30 (2022) Abstr 10171 [www.page-meeting.org/?abstract=10171]
Poster: Drug/Disease Modelling - Infection