Laura Boles1, Umberto Villani1,2, Alessandro Di Deo1,2, Jerry Nedelman3, Todd Black3, Ece Auffarth3, Dean Hickman3, Professor Oscar Della Pasqua1,2
1Clinical Pharmacology and Therapeutics (CPT) group, School of Pharmacy, University College London (UCL), 2Istituto per le Applicazioni del Calcolo, Consiglio Nazionale delle Ricerche, 3TB Alliance
Objectives Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), is the leading cause of infectious disease mortality, with 1.5 million deaths in 2020 alone [1]. Multidrug-resistant TB (MDR-TB) poses a major challenge, underscoring the need for drugs with novel mechanisms of action to combat the disease. As part of a broader European initiative, the ERA4TB consortium aims to accelerate development of novel TB infection models to evaluate candidate molecules and promote integration of preclinical and clinical data through in silico model-informed approaches. With the goal of guiding selection of novel multidrug regimens to be tested in clinical trials, establishing a robust rationale for translating preclinical findings into accurate predictions of drug effect is key. Currently in clinical development for TB, telacebec (Q203) is a first-in-class imidazopyrimidine amide derivative which suppresses Mtb growth by inhibiting cytochrome bc1 in oxidative phosphorylation [2]. Despite a promising safety and efficacy profile, its role in TB regimens remains under discussion. Accordingly, this work aimed to 1) characterise the pharmacokinetics (PK) and pharmacokinetic-pharmacodynamic (PKPD) relationship of telacebec and 2) explore translational steps required to predict antibacterial activity in humans along with the dose rationale for further investigation of early bactericidal activity (EBA) in a Phase IIa study. Methods The dose-exposure-response relationship in an infection model in mice was characterised as follows. A previously validated telacebec PK model [3] was used to simulate plasma concentration over time following regimens employed in two protocols: an acute C57BL/6J mice study featuring doses from 0.15 to 92 mg/kg and a chronic BALB/c mice study featuring doses of 0.4, 2 and 10 mg/kg. As killing kinetics showed biphasic bacterial growth, only data within 14 days of treatment initiation were considered. Importantly, PK- and disease-related differences between mice species were assumed to be negligible. Simulated PK profiles and CFU quantified in lung homogenate samples were used to develop a growth model and quantify the antibacterial activity of telacebec. Due to scarce pre-treatment samples, characterisation of bacterial growth was complemented by digitised data from a previous publication [4]. Gompertz and Velhurst models were considered for the characterisation of bacterial growth, whilst linear and sigmoidal EMAX models were investigated for the PKPD relationship of telacebec. Subsequently, a population PK (popPK) model was built from data from a single and multiple ascending dose Phase I study. Exploratory data analyses suggested a multiexponential decline during elimination, including nonlinear PK and food effect. As such, these factors were accounted for during model development. Models featuring different compartment numbers, absorption kinetics, and interindividual variability (IIV) components were tested. Multiple error models were also considered to describe residual variability. To translate antibacterial activity in mice into EBA estimates in patients, parameters describing bacterial growth and killing rates were used, accounting for system-specific differences. The parameterisation assumed a similar exposure-response relationship across species and, since telacebec is highly protein-bound in mice [5] and data for humans is unavailable, a similar fraction unbound in humans compared to mice. In addition, a first-order kinetics function accounted for tissue equilibration and CFU measurement error in sputum samples, characterised by a Gaussian distribution of CFU counts in pre-treatment samples. Whilst no further scaling of parameters was deemed necessary, a sensitivity analysis was performed to check the implications of the assumptions supporting the framework. To obtain EBA predictions, final estimates of the PKPD model were used to simulate a population of subjects with characteristics matching those from the available trial (n = 52, 500 iterations). Model development and evaluation were conducted with first order conditional estimation with interaction (FOCE-I). Model selection was based on statistical and graphical diagnostic criteria, including goodness-of-fit, changes in objective function value, visual predictive checks, and plausibility and precision of estimates. Modelling and simulation were implemented in NONMEM 7.5 and PsN 5.2.6. Graphical analyses were performed in R 4.1.2. Results For the PKPD model describing the antibacterial activity of telacebec in mice, a Velhurst model adequately characterised natural Mtb growth and an EMAX model described the concentration-effect relationship. Net growth rate (Knet) was estimated to be 0.766 day-1 and EC50 and Emax were estimated to be 28.18 ng/mL and 0.762 day-1, respectively. Parameterisation of killing rates for fast- and slow-growing subpopulations was, however, not possible. The PK of telacebec in healthy subjects was described by a three-compartment model with sequential zero-first absorption kinetics. For a 70 kg individual, estimates of clearance (CL) and intercompartmental clearances (Q3 and Q4) were 17.2, 23.6, and 30.1 L/h, respectively, with central volume (V2) and peripheral volumes (V3 and V4) estimated to be 57.5, 3140, and 302 L, respectively. IIV was identified for CL, Q3, Q4, V3, duration of zero-order absorption (D1), and absorption lag time (ALAG1). Significant covariate effects included weight, fed state, and dose level when fasted. Residual error was described by a combined proportional and additive model. The final model yielded estimates with acceptable precision, without evidence of significant bias in diagnostic plots. Importantly, the VPCs showed adequate model performance, with accurate prediction of Phase IIa trial data. The final PKPD parameterisation was used to predict treatment response in the EBA study. Assuming infection was well established upon diagnosis, the initial bacterial burden (B0) and carrying capacity (Bmax) were set to 5.64 CFU/mL, the median value in patients prior to treatment. For each dose tested, numerical predictive checks showed that the median EBA values (5th-95th C.I. for median ?CFU/mL predictions, EBA100mg = [0.47 -1.07]; EBA200mg = [0.59 – 1.22]; EBA300mg = [0.62 – 1.21]) were in agreement with median observations in the trial (?CFU/mL, EBA100mg = 0.62; EBA200mg = 0.88; EBA300mg = 1.04). Conclusions This work highlights the value of a translational PKPD framework within the ERA4TB consortium, acting as a critical connecting element to enable integration of in vitro and in vivo data across species, as well as translation to humans. In contrast to empirical choices, our approach provided estimates of the effect of telacebec on bacterial killing, allowing prediction of the treatment duration required for eradication. With oncoming safety data, these results support the recommendation for the inclusion of telacebec in trials featuring novel multidrug regimens for MDR-TB.
1. Dheda K, Mirzayev F, Cirillo DM, Udwadia Z, Dooley KE, Chang KC, Omar SV, Reuter A, Perumal T, Horsburgh Jr CR, Murray M. Multidrug-resistant tuberculosis. Nature Rev Disease Primers. 2024;10(1):22. 2. Lamprecht DA, Finin PM, Rahman MA, Cumming BM, Russell SL, Jonnala SR, Adamson JH, Steyn AJ. Turning the respiratory flexibility of Mycobacterium tuberculosis against itself. Nature communications. 2016; 7(1):12393. 3. Kim, J., Kim, T.H., Choi, J., Nam, K., Shin, B.S. and Shin, S. Novel modeling approach integrating population pharmacokinetics and interspecies scaling to predict human pharmacokinetics of the new anti-tuberculosis agent telacebec (Q203). Biomed Pharmacother 2023;167:115441. 4. Zhang N, Strydom N, Tyagi S, Soni H, Tasneen R, Nuermberger EL, Savic RM. Mechanistic modeling of Mycobacterium tuberculosis infection in murine models for drug and vaccine efficacy studies. Antimicrobial agents and chemotherapy. 2020; 64(3):10-128. 5. Almeida DV, Converse PJ, Omansen TF, Tyagi S, Tasneen R, Kim J, Nuermberger EL. Telacebec for ultrashort treatment of Buruli ulcer in a mouse model. Antimicrob Agents Chemother 2020; 64(6):10-128.
Reference: PAGE 33 (2025) Abstr 11349 [www.page-meeting.org/?abstract=11349]
Poster: Drug/Disease Modelling - Other Topics