Bedaquiline's exposure-response relationship revealed through modeling of mycobacterial load
Elin M. Svensson, Mats O. Karlsson
Dept. Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden.
Objectives:
Pulmonary tuberculosis (TB) remains a dramatic health problem with an estimated incidence of 9.4 million cases in 2014, of which 0.5 million were caused by multi-drug resistant (MDR) TB (1). There is an acute lack of knowledge of how to best select regimens of second-line anti-TB drugs and this uncertainty is linked to poor description of dose-exposure-response relationships.
Bedaquiline (BDQ) was recently granted conditional approval for treatment of MDR-TB based on Phase II studies (2,3) and is now in use (4). The efficacy was primarily assessed on time to sputum culture conversion (TSCC, i.e. when mycobacteria can no longer be detected in sputum) and conversion status and cure rates at month 30. However, the performed analyses could not identify a relationship between BDQ exposure and any of these outcomes (2).
In this work we aimed to characterize an exposure-response relationship by modeling repeated measures of mycobacterial load (MBL), quantified by sputum cultures in a mycobacterial growth indicator tube system (MGIT). The results were discussed in relation to previously predicted (5–7) and recently confirmed drug-drug interactions (8).
Methods:
Data were obtained from a registration phase IIb study (TMC207-C208). The design was randomized, double-blind and placebo-controlled; enrolling newly diagnosed patients with pulmonary MDR-TB. All patients were treated with a background regimen of 5 second-line anti-TB drugs to which either placebo or BDQ was added. The duration of the addition was either 8 weeks (pilot) or 24 weeks (majority). BDQ was dosed at 400mg QD the first 2 weeks and thereafter 200mg 3 times per week. The study was conducted in accordance with Good Clinical Practice standards and received ethical approval from appropriate local authorities.
Triplicate sputum samples were collected at the day prior to treatment initiation, weekly until week 8 and bi-weekly until week 24. MGIT cultures were initiated from each sample and the time to positivity (TTP), i.e. a signal indicating presence of M. tuberculosis, was automatically recorded. TTP is a measure of MBL, with a shorter time indicating a higher bacterial burden. Samples without a signal at 42 days were classified as censored negative.
A model of MBL in patients was linked to the hazard in the time-to-event (TTE) model of TTP, and PK, interindividual variability (IIV) and covariate effects were evaluated on parameters of the MBL model. Individual secondary PK metrics were obtained from a previously developed model (9). The analysis was performed in NONMEM 7.3 with the Laplace estimation method. Parameter uncertainty was assessed with SIR (10).
Posterior predictive checks (PPC) of TSCC calculated based on observed and simulated datasets (n=100) were performed. The clinical importance of detected covariates, quantified by changes in TSCC and proportion without SCC at week 20, was assessed through simulations including parameter uncertainty (n=100) for a large dataset (2000 subjects with a specific set of characteristics per scenario).
Results:
TTP data from 206 patients were available; only samples collected during the intervention period were considered. After curation a dataset including 6330 observations (59.8% positive) from 193 individuals collected up to week 20 were used for model building.
The developed PD model included 3 simultaneously fitted components: (i) a longitudinal representation of MBL in patients over time on treatment, (ii) a model of the probability of bacterial presence in a sputum sample and (iii) a TTE model for TTP in MGIT. The MBL was described by a mono-exponential decline over time on treatment where the number of bacteria at start was informed by each individual’s observed baseline mean TTP. IIV with a Box-Cox transformed distribution was included on the half-life of the decline. The probability of bacterial presence was linked to MBL by an Emax-function including the maximal risk (Pmax) and the MBL value corresponding to 50% of Pmax. The hazard in the TTE model was proportional to the current number of bacteria in the growth tube. The number of bacteria in the tube over time was described by an inoculum size defined by the MBL and a logistic growth function.
The joint PD model described the TTP data well and the PPC demonstrated that also TSCC in the 2 study arms was well predicted. The model of probability of bacterial presence handled the increasing portion of negative samples, providing the characteristic shape of TTP survival curves reaching a plateau after about 25 days. Of previously described models the present model shows most similarity to (11). However, by including the probabilistic element, it was possible to use MBL inoculum size as the driver of TTP hazard without incorporating mechanistically unjustified changes in the bacterial growth processes in the MGIT system over time on treatment (11). The estimated initial doubling time in the MGIT tube was 33h (RSE 5%) which is in line with observed in vitro growth rates for M. tuberculosis.
Early BDQ exposure (AUC0-24h at day 14) was found to significantly affect the half-life of MBL through an Emax-function. The maximal effect could not be estimated reliably due to the limited range of observed exposures and was therefore fixed to -100%. EC50 was estimated to 52 µg/mL*h (RSE 36%) and fell within the observed range of exposures (10-94 µg/mL*h). Simulations of the impact showed that the proportion of patients without SCC at week 20 is expected to decrease from 28% (95%CI 23-34) in the placebo arm to 19% (15-22), 14% (10-17) or 8% (5-11) in patients with half median, median (35 µg/mL*h) or double the median BDQ exposure, respectively. In addition, patients with (pre-) extensively drug resistant TB were found to clear mycobacteria 31% (RSE 22%) slower than patients with MDR-TB, leading to 2-4 weeks later median TSCC.
Conclusions:
A novel model describing MBL in MDR-TB patients with 3 linked sub-components was developed. In contrast to simpler analyses of secondary PD metrics based on the same data, our model could detect and describe an exposure-response relationship for BDQ.
The model shows that higher BDQ levels early during treatment lead to faster response. This may have implications on coadministration of BDQ with drugs that lower exposure such as efavirenz and rifampicin (which induce BDQ CL to 207% and 478% of normal, respectively), and strengthens the earlier recommendations against concomitant use at standard BDQ dose (5,7). Furthermore, the increased exposure when coadministered with lopinavir/ritonavir (inhibiting BDQ CL to 35% of normal (6,8)) could lead to faster SCC. The link between week 2 exposure and bacteriological response may also provide an opportunity for early assessment of individual patients’ chance of favorable outcome.
This characterization of BDQ’s exposure-response relationship could inform optimization of novel anti-TB regimens and their application also in TB-HIV coinfected patients.
References
[1] World Health Organization. Global Tuberculosis Report 2015.
[2] Diacon AH, Pym A, Grobusch MP, de los Rios JM, Gotuzzo E, Vasilyeva I, et al. Multidrug-resistant tuberculosis and culture conversion with bedaquiline. N Engl J Med. 2014;371(8):723–32.
[3] Pym AS, Diacon AH, Tang S-J, Conradie F, Danilovits M, Chuchottaworn C, et al. Bedaquiline in the treatment of multidrug- and extensively drug-resistant tuberculosis. Eur Respir J. 2015 Dec 2.
[4] Ndjeka N, Conradie F, Schnippel K, Hughes J, Bantubani N, Ferreira H, et al. Treatment of drug-resistant tuberculosis with bedaquiline in a high HIV prevalence setting: an interim cohort analysis. Int J Tuberc Lung Dis Off J Int Union Tuberc Lung Dis. 2015;19(8):979–85
[5] Svensson EM, Aweeka F, Park J-G, Marzan F, Dooley KE, Karlsson MO. Model-Based Estimates of the Effects of Efavirenz on Bedaquiline Pharmacokinetics and Suggested Dose Adjustments for Patients Coinfected with HIV and Tuberculosis. Antimicrob Agents Chemother. 2013;57(6):2780–7.
[6] Svensson EM, Dooley KE, Karlsson MO. Impact of Lopinavir-Ritonavir or Nevirapine on Bedaquiline Exposures and Potential Implications for Patients with Tuberculosis-HIV Coinfection. Antimicrob Agents Chemother. 2014;58(11):6406–12.
[7] Svensson EM, Murray S, Karlsson MO, Dooley KE. Rifampicin and rifapentine significantly reduce concentrations of bedaquiline, a new anti-TB drug. J Antimicrob Chemother. 2015;70(4):1106–14.
[8] Brill MJE, Svensson EM, Pandie M, Maartens G, Karlsson MO. Confirming model-predicted pharmacokinetic interactions between bedaquiline and lopinavir/ritonavir or nevirapine in patients with HIV and drug resistant tuberculosis. Abstract submitted to PAGE 2016.
[9] Svensson EM, Dosne A-G, Karlsson MO. Population pharmacokinetics of bedaquiline and metabolite M2 in drug-resistant tuberculosis patients – the effect of time-varying weight and albumin. Submitted.
[10] Dosne A-G, Bergstrand M, Karlsson MO. Application of Sampling Importance Resampling to estimate parameter uncertainty distributions. PAGE 22 (2013) Abstr 2907 [www.page-meeting.org/?abstract=2907]; Glasgow, Scotland.
[11] Chigutsa E, Patel K, Denti P, Visser M, Maartens G, Kirkpatrick CMJ, et al. A time-to-event pharmacodynamic model describing treatment response in patients with pulmonary tuberculosis using days to positivity in automated liquid mycobacterial culture. Antimicrob Agents Chemother. 2013;57(2):789–95.
Acknowledgement
The research leading to these results has received funding from the Innovative Medicines Initiative Joint Undertaking (www.imi.europa.eu) under grant agreement n°115337, resources of which are composed of financial contribution from the European Union's Seventh Framework Programme (FP7/2007-2013) and EFPIA companies' in kind contribution.
Disclosure
MK has received research grants from Janssen pharmaceuticals.
