Asina Gijasi 1, Coen van Hasselt 1, Eric Nuermberger 2, Rada Savic 3, Rob van Wijk 1
1 Division of Systems Pharmacology and Pharmacy, Leiden Academic Centre for Drug Research, Leiden University (Leiden, The Netherlands), 2 Center for Tuberculosis Research, Department of Medicine, Johns Hopkins University School of Medicine (Baltimore, United States of America), 3 Department of Bioengineering and Therapeutic Sciences, University of California, San Francisco (San Francisco, United States of America)
Objectives: Tuberculosis (TB) remains the leading cause of infectious disease mortality worldwide [1]. The development of improved and shorter treatment regimens is therefore necessary to increase efficacy and safety and reduce the risk of relapse. Translational pharmacokinetic-pharmacodynamic (PKPD) models are essential for this purpose, as they facilitate the identification of promising combination regimens, inform clinical dosing, and accelerate clinical development. Already validated and applied in monotherapy Phase 2a clinical trials, translational PKPD models have demonstrated their utility in predicting early bactericidal activity and optimizing dosing strategies and trial design [2,3]. Yet, predicting regimen performance in later Phase 2b-c and Phase 3 clinical trials remains challenging, as reflected by various unsuccessful trials [4-7]. Therefore, we aim to develop a translational approach to predict later-phase clinical efficacy using back-translational modeling from clinical to preclinical. To address this, we developed a translational model to predict Phase 2b clinical efficacy of rifapentine regimens from Studies 29 and 29X [8,9], using a mouse TB model treated with a combination regimen, incorporating the mechanistic immune response and pharmacological translational factors.
Methods: Bacterial burden was quantified as colony-forming units (CFU) over time in a sub-acute infection model (14 days between infection and treatment) in Mycobacterium tuberculosis-infected BALB/c mice [10]. The mice received a combination therapy of rifapentine (P; 5, 7.5, and 20 mg/kg), isoniazid (H; 25 mg/kg), and pyrazinamide (Z; 150 mg/kg). The data were modelled with NONMEM (v 7.5.1, ICON Development Solutions, Maryland) using a previously developed translational PKPD framework [11,12]. PK was based on a previously published PK model for rifapentine [11]. Various PD models were explored and developed, including a monotherapy model for rifapentine, a combination of (interacting) rifapentine, isoniazid, and pyrazinamide monotherapy models, and re-estimation of Emax or EC50 relative to monotherapy with rifapentine-exposure driving combination efficacy as informed by exploratory data analysis. Additionally, distinctive PD models were tested for the first month of treatment and thereafter to account for differences between fast- and slow-replicating bacteria [13,14].
Translating mouse PKPD to clinical time-to-positivity (TTP) over time relied on imputed clinical exposure (AUC) and translational factors: (1) the efficacy (Emax) was translated from preclinical CFU to clinical (TTP) [15], and (2) the potency (EAUC50) was adjusted for differences in the fraction unbound to account for inter-species differences. Predictive performance was evaluated using a visual predictive check comparing simulated probabilities above the limit of quantification (TTP ≥ 44 days) with results from Study 29 and 29X [8,9]. Additionally, stable liquid culture conversion at 8 weeks was assessed, comparing observations and predictions. Finally, an integrated model combining both pre-clinical combination therapy data and clinical efficacy data was developed for simultaneous estimation of regimen efficacy.
Results: Dose-ranging PHZ combination therapy CFU data in Mycobacterium tuberculosis-infected BALB/c mice were best described by a direct sigmoidal Emax model driven by rifapentine exposures (AUC). PKPD relationships were incorporated for fast- and slow-replicating bacteria with distinct Emax values for the first month of treatment and after (0.543 and 0.272 day-1), and an EAUC50 of 3.54 mg∙d/L. The rifapentine-containing combination regimen showed two-fold higher efficacy in the first month compared to monotherapy (0.299 day-1 [11]).
Clinical predictions for 8-week liquid culture stable culture conversion using back-translational modeling were successful, with observed and median (95% confidence interval) at 66.7% (60.1-72.7%) and 70.2% (65.7-74.0%) for Study 29 and 75.2% (68.4-81.1%) and 82.3% (77.8-86.7%) for Study 29X, respectively.
We quantified a clear exposure-response relationship. Specifically for Study 29X, observed and median predicted for low, medium and high rifapentine exposure were 54.8% (41.8-67.3%) and 69.5% (60.6-78.2%), 90.5% (79.7-96.1%) and 86.3% (76.5-92.2%), and 80.0% (67.9-88.5%) and 93.3% (87.5-98.1%), respectively. Further refinement of back-translational modeling was assessed using the integrated model combining both pre-clinical and clinical data.
Conclusions: Back-translation of clinical trial efficacy data for a rifapentine-containing combination regimen showed to be successful based on preclinical combination therapy PKPD data, improving translational modeling and predicting later-phase clinical outcomes. Our findings demonstrate the translatability of preclinical PKPD to clinical efficacy, an important step in improving drug combination regimen development in TB.
Acknowledgements: This research was supported by the PReDiCTR-TB consortium. Data used in the preparation of this abstract were obtained from the Critical Path – Data and Analytics Platform (CP-DAP) which is administered by Critical Path Institute.
References:
[1] Global tuberculosis report 2025. Geneva: World Health Organization; 2025. Licence: CC BY-NC-SA 3.0 IGO.
[2] Bartelink IH, et al. (2017). New Paradigm for Translational Modeling to Predict Long‐term Tuberculosis Treatment Response. CTS 10:366–79.
[3] Susanto BO, et al. (2020). Translational Model‐Informed Approach for Selection of Tuberculosis Drug Combination Regimens in Early Clinical Development. CPT 108:274–86.
[4] Jindani A, et al. (2014). High-Dose Rifapentine with Moxifloxacin for Pulmonary Tuberculosis. NEJM 371:1599–608.
[5] Gillespie SH, et al. (2014). Four-Month Moxifloxacin-Based Regimens for Drug-Sensitive Tuberculosis. NEJM 371:1577–87.
[6] Merle CS, et al. (2014). A Four-Month Gatifloxacin-Containing Regimen for Treating Tuberculosis. NEJM 371:1588–98.
[7] Cevik M, et al. (2024). Bedaquiline-pretomanid-moxifloxacin-pyrazinamide for drug-sensitive and drug-resistant pulmonary tuberculosis treatment: a phase 2c, open-label, multicentre, partially randomised controlled trial. LANCET 24:1003–14.
[8] Dorman SE, et al. (2012). Substitution of Rifapentine for Rifampin During Intensive Phase Treatment of Pulmonary Tuberculosis: Study 29 of the Tuberculosis Trials Consortium. JID 206:1030–40.
[9] Dorman SE, et al. (2021). Four-Month Rifapentine Regimens with or without Moxifloxacin for Tuberculosis. NEJM 384:1705–18.
[10] Rosenthal IM, et al. (2012). Dose-ranging comparison of rifampin and rifapentine in two pathologically distinct murine models of tuberculosis. AAC 56:4331-40.
[11] Ernest JP, et al. (2023). Translational predictions of phase 2a first-in-patient efficacy studies for antituberculosis drugs. ERJ 62:2300165.
[12] Strydom N, et al. (2024). Dose optimization of TBI-223 for enhanced therapeutic benefit compared to linezolid in antituberculosis regimen. NATCOM 15:1.
[13] Goh JJ, et al. (2024). Prospectively predicting BPaMZ phase IIb/III trial outcomes using a translational mouse-to-human platform. AAC 68:10.
[14] Zhang N, et al. (2020). Mechanistic Modeling of Mycobacterium tuberculosis Infection in Murine Models for Drug and Vaccine efficacy studies. AAC 64:3.
[15] Magombedze G, et al. (2021). Bacterial load slopes represent biomarkers of tuberculosis therapy success, failure, and relapse. COMBIOL 4:1.
Reference: PAGE 34 (2026) Abstr 12176 [www.page-meeting.org/?abstract=12176]
Poster: Drug/Disease Modelling - Infection