A Phase 2a Study of the Early Bactericidal Activity of Rifampicin in Combination With Meropenem Plus Amoxicillin/Clavulanate Among Adults With Rifampicin-Resistant Pulmonary Tuberculosis.
Dec 24, 2025DOI: 10.1093/cid/ciaf276Journal ArticleCID
Written by: Veronique de JagerLaurynas MockeliunasHuifang YouUlrika S H SimonssonTarryn CollingsCaryn M UptonJohannes A BezuidenhoutNikhil GupteSusan E DormanEric NuermbergerGyanu LamichhaneAndreas H DiaconKelly E Dooley
The COMRADE (Co-amoxiclav Meropenem Rifampicin Antimycobacterial Determination of Efficacy) trial (ClinicalTrials.gov identifier NCT03174184) was a phase 2a prospective, open-label, randomized clinical trial to determine the bactericidal activity of meropenem plus amoxicillin/clavulanate with vs without rifampicin in patients with pulmonary TB. The trial included patients with rifampicin-susceptible TB (RS-TB) (results previously reported [10]) and RR-TB (the focus of this study) based on XpertMTB/RIF (Cepheid, Sunnyvale, CA) results. The trial was conducted at a single site in Cape Town, South Africa, with enrollment occurring between August 2017 and May 2022. The trial received local regulatory authority and ethics committee approval, with all participants providing signed informed consent prior to the conduct of any trial-related procedures. Participants were adults aged 18 to 65 years with newly diagnosed untreated pulmonary TB, a grading of at least 1+ acid-fast bacilli on sputum smear microscopy or a semiquantitative result of “medium” or “high” on the XpertMTB/RIF assay. Participants with human immunodeficiency virus (HIV) and a CD4 count greater than 100 cells/mm3 were eligible. Participants were admitted to the trial hospital for the screening and treatment periods, after which they were transferred back to their local primary care facility to receive full TB treatment. A follow-up outpatient visit was conducted 14 days after study treatment completion.Participants with RR-TB were randomized 3:2 to receive meropenem (M) plus amoxicillin/clavulanate (AC) with rifampicin (R) (M-AC-R) vs meropenem plus amoxicillin/clavulanate without rifampicin (M-AC), respectively. The randomization ratio implemented was to account for the anticipated increased variability in treatment response in the rifampicin-containing arm (M-AC-R). Meropenem was administered intravenously at a dose of 2 g every 8 hours (3 times/day) and amoxicillin/clavulanate 500 mg/125 mg was given orally 3 times per day with each dose of meropenem. Rifampicin 20 mg/kg was administered orally once daily at the initiation of the first daily meropenem infusion. The dose of rifampicin was chosen to optimize its bactericidal effect, while maintaining an acceptable safety and tolerability profile. Each meropenem infusion lasted approximately 30 minutes. Study treatment was administered for 14 days.Adverse events (AEs) were collected daily by study investigators from the time a participant received the first dose of study treatment through the final visit. Severity was graded according to the Division of AIDS (DAIDS) Table for Grading the Severity of Adult and Pediatric Adverse Events, version 2.1, July 2017 [11]. All AEs of grade 2 or higher were collected. Blood samples for laboratory-based safety assessments were collected at screening and on treatment days 7, 14, and 28 and in the event of an early withdrawal from the trial.
On treatment day 13, pharmacokinetic (PK) sampling was performed pre-dose and at 0.5, 1, 1.5, 2, 3, 4, 6, and 8 hours post-dose for meropenem and rifampicin, and additionally at 12 and 24 hours post-dose for rifampicin. Plasma rifampicin and meropenem concentrations were determined using reversed-phase high-performance liquid chromatography with tandem mass spectrometry using electrospray ionization. Quantification ranges were 0.1–50 μg/mL and 0.5–256 μg/mL for rifampicin and meropenem, respectively.
Pretreatment spot sputum samples were collected for sputum smear microscopy, XpertMTB/RIF assay, and drug susceptibility testing. Sixteen-hour overnight sputum samples were collected from each participant for 2 days prior to initiation of study treatment (baseline), daily for 4 consecutive days thereafter (days 1–4), then on days 6, 8, 10, 12, and 14. Microbiological methods to determine bactericidal activity have been described previously [10]. In brief, individual overnight sputum samples for each participant were cultured on 7H11S agar plates for the determination of colony-forming units (CFU) and in the BD BactecMGIT 960 Mycobacteria Culture system for time-to-positivity (TTP).
The rifampicin MIC was determined using the European Committee on Antimicrobial Susceptibility Testing (EUCAST) broth microdilution reference method. Rifampicin MIC was determined without and with meropenem added for each individual participant pretreatment spot sputum sample, at a meropenem concentration of 10 μg/mL.
A noncompartmental analysis (NCA) to determine meropenem and rifampicin PK parameters (maximum plasma concentration [Cmax], time at which the maximum concentration was observed [tmax], area under the concentration-time curve from time zero to infinity [AUC0-∞], area under the concentration-time curve from time zero to the last measured concentration [AUC0-last], clearance [CL] and volume of distribution [VD] for meropenem, or apparent clearance [CL/F] and apparent volume of distribution [VD/F] for rifampicin, terminal phase elimination rate constant [ke], and the terminal phase elimination half-life [t½]) was performed using the linear-up log-down method for the PK profiles with at least 3 observations after tmax using the IQnca R package (version 1.2.0) [12]. Acceptance criteria for NCA results were set to adjusted R2 ≥0.80. An unpaired Student’s t test with a P value of .05 was performed to test for statistically significant differences in meropenem exposure between the 2 arms for t½, Cmax, AUC0-∞, and AUC0-last.
To determine EBA, a standardized pharmacometric model–based EBA analysis workflow was followed [13]. Mono- and bi-exponential models were developed for each biomarker separately, which was followed by covariate model building using stepwise covariate modeling with adaptive scope reduction [14]. Body mass index (BMI), age, self-reported race, sex, extent of disease and cavitary disease on chest X-ray, and HIV status were evaluated on the baseline parameter. BMI, baseline CFU or baseline TTP, age, self-reported race, sex, extent of disease and cavitary disease on chest X-ray, and HIV status were evaluated on the slope parameter(s) and the second intercept. The P value for the forward inclusion step was set to .05 and to .01 for the backward deletion step.
To determine if each of the treatments had quantifiable EBA, all parameters, except for baseline and covariates on baseline, were fixed to zero, and this model was compared with a model where the respective parameters were estimated for each treatment at a time. Early bactericidal activity was defined if there was a statistically significant difference in the slope parameters(s) compared with zero (P ≤ .05).
This was followed by PK-pharmacodynamic (PK-PD) modeling, where the PK parameters meropenem AUC0-last, meropenem Cmax, rifampicin AUC0-last, and rifampicin Cmax were evaluated as covariates on slope(s) and the second intercept using different linear and nonlinear relationships (P ≤ .05).
In order to determine if there was a difference in EBA between the 2 treatments, all inter-individual variability (IIV) and covariate effects on all parameters, except for baseline, were fixed to zero, and each of the treatment slopes was re-evaluated using a P value of .05. Predicted EBA was reported as a median and a 95% prediction interval (PI) of individual EBA values using empirical Bayes estimates.
Modeling was performed using the nonlinear mixed-effects modeling software NONMEM (version 7.5.1; ICON Plc, Gaithersburg, MD, USA) [15] on a Unix cluster using a GNU Fortran compiler (version 5.1.0) through the PsN toolkit (version 7.5.0; Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden) [16]. Throughout EBA analysis, statistical significance was determined using a likelihood ratio test assuming a chi-square distribution. The first-order conditional estimation with interaction (FOCE-I) estimation method was used for parameter estimation. R was used for data management and processing (version 4.2.2; R Foundation for Statistical Computing, Vienna, Austria) [17], while goodness-of-fit (GOF) plots were created using xpose (version 0.4.14; Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden) [18] and xpose4 (version 4.7.2; Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden) [16, 19].
RESULTS
In total, 52 participants were enrolled, 32 in the M-AC-R arm and 20 in the M-AC arm. The median age (interquartile range) was 37 years (28–46 years), 67% of study participants were male, and 27% were HIV positive (Table 1).
Participant Baseline Demographics of All Enrolled Participants and Per Treatment
Abbreviations: BMI, body mass index; CFU, colony-forming units; HIV, human immunodeficiency virus; IQR, interquartile range; M-AC, meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily (without rifampicin); M-AC-R, meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily with rifampicin 20 mg/kg once daily; TTP, time-to-positivity.
aFisher's exact test was used for comparing proportions and Kruskal–Wallis test was used for comparing medians.
bSputum smear microscopy grade based on the International Union Against Tuberculosis and Lung Disease scale.
cBaseline CFU and TTP based on sputum samples of 50 participants. Two randomized participants withdrew prior to initiating study treatment and thus did not have on-treatment or baseline sputum samples analyzed.
Of the 52 enrolled participants, 50 received at least 1 dose of study drug (31 participants in the M-AC-R arm and 19 participants in the M-AC arm), and 38 participants out of 50 completed treatment—74% and 79% in the M-AC-R and M-AC arms, respectively. Forty-three AEs of grade 2 (moderate) or grade 3 (severe) severity were reported (Table 2) in 25 participants, of whom 17 participants (55%) in the M-AC-R arm experienced 28 AEs (65% of all reported AEs), while 8 participants (42%) in the M-AC arm experienced 15 AEs (35% of all reported AEs). No serious or life-threatening (grade 4) AEs were reported. The most common AEs in the M-AC-R arm were vomiting (25%), nausea (21%), and diarrhea (7%), while in the M-AC arm these included diarrhea or loose stools (13%) and tachypnea (13%). Study investigators considered 68% of AEs reported in the M-AC-R arm to be related to study treatment, compared with 33% in the M-AC arm.
Summary of Reported Adverse Events
Data are presented as n (%).
Abbreviations: M-AC, meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily (without rifampicin); M-AC-R, meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily with rifampicin 20 mg/kg once daily.
Pharmacokinetic parameters of rifampicin and meropenem were determined in 38 participants (74% in the M-AC-R arm and 79% in the M-AC arm). Median (range) meropenem AUC0-last values were 170 (86.6–452) and 193 (114–271) μg × h/mL in the M-AC-R and M-AC arms, respectively (Table 3). Meropenem PK parameters were similar between the treatment groups.
Summarized Meropenem and Rifampicin Pharmacokinetic Parameters Reported as Median (Range) From Noncompartmental Analysis
Abbreviations: AUC0-last, area under concentration-time curve from time zero to time of last measured concentration; CL, clearance; CL/F, apparent clearance; Cmax, maximum observed plasma concentration; M-AC, meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily (without rifampicin); M-AC-R, meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily with rifampicin 20 mg/kg once daily; n.d., not determined; t1/2, terminal elimination half-life; tmax, time of maximum plasma concentration; VD, volume of distribution; VD/F, apparent volume of distribution.
aNumber of participants who underwent pharmacokinetic sampling.
Data from 47 participants (29 from the M-AC-R arm and 18 from the M-AC arm) were used in CFU data modeling. The CFU profiles of 3 participants were excluded from the analysis due to not having any positive on-treatment CFU observations. The final CFU model followed a mono-exponential pattern, indicating a linear decline in the log10 CFU count over time. The model included IIV in baseline CFU and the rate of decrease over time (slope). Only baseline CFU was a statistically significant covariate on the slope, where the slope increased by 41.1% per 1-unit increase in baseline. Both treatment arms showed EBA, but no statistically significant difference in EBA between the 2 treatments was found (P > .05). The median predicted 14-day EBA (95% PI) was 1.33 (−0.15 to 3.64) and 1.20 (0.05–2.66) log10 CFU/mL for the M-AC-R and M-AC arms, respectively (Figure 1A). The final CFU model parameter estimates are presented in Supplementary Table 1. A prediction-corrected visual predictive check of the final CFU model is shown in Supplementary Figure 1. The final model code is presented in Supplementary Text 1.
Predicted 14-day early bactericidal activity in (A) decrease in CFU and (B) increase in TTP. The circle represents the median, while the whiskers represent a 95% prediction interval. Numbers on the side represent the median and the 95% prediction interval for that arm. The M-AC-R arm represents meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily with rifampicin 20 mg/kg once daily, while the M-AC arm represents meropenem 2 g intravenous infusion 3 times daily and amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily (without rifampicin). Abbreviations: CFU, colony-forming units; M-AC, meropenem plus amoxicillin/clavulanate without rifampicin; M-AC-R, meropenem plus amoxicillin/clavulanate with rifampicin; TTP, time-to-positivity.
For TTP modeling, data from all 50 participants (31 from the M-AC-R arm and 19 from the M-AC arm) were used. The TTP was described by a mono-exponential function, indicating a linear increase in TTP over time on natural logarithm scale. Inter-individual variability was included on baseline TTP and slope. The presence of cavitary disease on chest X-ray was included on baseline TTP. Participants with no cavities or with the presence of single or multiple cavities with a diameter of less than 4 cm in aggregate had 52% higher baseline TTP compared with participants with single or multiple cavities of a diameter of 4 cm or greater in aggregate. Both treatment arms showed EBA, but no statistically significant difference between the 2 treatments was found. The median predicted 14-day EBA (95% PI) was 0.220 (0.098–0.551) and 0.216 (0.126–0.505) log10 hours for the M-AC-R and M-AC arms, respectively (Figure 1B). The final TTP model parameter estimates are presented in Supplementary Table 2, while a prediction-corrected visual predictive check of the final model is shown in Supplementary Figure 2. The final model code is presented in Supplementary Text 2. No statistically significant PK-PD relationships were identified between rifampicin or meropenem exposure and CFU or TTP.
The MIC was determined in sputum isolates from 30 and 16 participants in the M-AC-R and M-AC arms, respectively. Rifampicin MIC without meropenem ranged from 0.32 to more than 1.28 μg/mL. With the addition of meropenem, rifampicin MIC decreased in samples from 12 participants (8 [27%] in the M-AC-R arm and 4 [25%] in the M-AC arm) to a range of less than 0.01 to 0.64 μg/mL, while in the majority of samples (34 participants), rifampicin MIC remained greater than 1.28 μg/mL in the presence of meropenem (Figure 2). The rifampicin MIC at which 90% of the tested isolates were inhibited (MIC90) was more than 1.28 μg/mL, irrespective of the presence of meropenem for both arms (Table 4).
The MIC of rifampicin with and without meropenem. The size of the circle represents the number of participants with a specific MIC. The lines represent the change in MIC after addition of meropenem. The M-AC-R arm represents meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily with rifampicin 20 mg/kg once daily, while the M-AC arm represents meropenem 2 g intravenous infusion 3 times daily and amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily (without rifampicin). Abbreviations: M-AC, meropenem plus amoxicillin/clavulanate without rifampicin; M-AC-R, meropenem plus amoxicillin/clavulanate with rifampicin; MIC, minimum inhibitory concentration.
Summary of Rifampicin Minimum Inhibitory Concentrations in the Absence and Presence of Meropenem
Abbreviations: M-AC, meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily (without rifampicin); M-AC-R, meropenem 2 g intravenous infusion 3 times daily, amoxicillin/clavulanate 500 mg/125 mg orally 3 times daily with rifampicin 20 mg/kg once daily; MIC, minimum inhibitory concentration; MIC90, minimum inhibitory concentration at which 90% of isolates are inhibited.
DISCUSSION
The COMRADE clinical trial was performed to determine the EBA of meropenem plus amoxicillin/clavulanate with vs without rifampicin in patients with RR-TB and to assess if meropenem-rifampicin synergism could restore the bactericidal activity of rifampicin in this patient population. Bactericidal activity of the combination of meropenem with the β-lactamase inhibitor clavulanate against Mtb was demonstrated, but adding rifampicin, even at a higher than standard dose of 20 mg/kg, did not result in significant additional antimycobacterial activity.
We observed that meropenem plus amoxicillin/clavulanate had similar activity against RR-TB and RS-TB as found in previous studies and confirming that rifampicin resistance and its associated compensatory mutations do not meaningfully impact meropenem activity against Mtb [10, 20, 21]. Data from this study support the inclusion of the meropenem plus amoxicillin/clavulanate combination in the WHO list of medications as an additional anti-TB drug for patients with extensive resistance profiles against other more effective treatments, or for whom there are tolerability or drug delivery challenges with first-line drugs [6].
In our trial, nausea and vomiting were more commonly reported in the arm receiving rifampicin. Vomiting was also reported as one of the most frequent AEs within rifampicin groups in the PanACEA HIGHRIF1 trial, which evaluated increasing doses of rifampicin alone and with the addition of other anti-TB drugs [22]. While similar rifampicin exposure was achieved in our trial, compared with the rifampicin 20-mg/kg group in a HIGHRIF1 trial, we observed a larger range of both AUC0-last and Cmax, indicating higher variability. Overall, the tolerability of meropenem, with or without rifampicin, was suboptimal and may have been a significant contributor to the high dropout rate observed in this trial. In patient populations where the addition of meropenem as an anti-TB drug may add value, gastrointestinal side effects should be addressed pre-emptively and actively managed to facilitate treatment adherence.
Despite promising in vitro findings by Kaushik and colleagues [6], where the synergistic effects of rifampicin and meropenem against both rifampicin-susceptible and -resistant clinical isolates of Mtb were quantified, similar synergy was not identified in patients with RR-TB in this trial. While the addition of rifampicin to the meropenem-amoxicillin/clavulanate regimen significantly reduced bacterial burden in patients with RS-TB as assessed in both CFU and TTP [10], no difference was found between the treatment arms in our study conducted in patients with RR-TB. One can speculate that the synergism between meropenem and rifampicin may be strain-dependent, and it is only possible to restore rifampicin activity against specific Mtb strains, but this hypothesis is not possible to assess in this study, as bacterial strain typing was not performed. Additionally, it is possible that the short duration of the trial limited the ability to fully assess long-term meropenem-rifampicin synergism. Moreover, in vitro studies do not fully replicate the complex physiological conditions present in humans, including host immune response, PK-PD interactions, and microenvironmental factors. The absence of immune components, variations in drug metabolism, and the influence of host factors in vivo can lead to discrepancies between in vitro and clinical outcomes. In addition to the more well-characterized drug-resistance mechanisms, the differential expression of bacterial efflux pumps in the hostile immune-activated host environment may further contribute to the discrepant findings between human clinical trials and in vitro studies. Differences in study design, sample size, drug concentrations, and methodological approaches between in vitro and clinical settings may also impact the comparability of results.
Meropenem has poor oral bioavailability and is thus administered as an intravenous infusion. As expected, this poses significant challenges to treatment adherence and accessibility, particularly in resource-limited settings. It also impedes outpatient treatment strategies for TB. Exploring other carbapenems with favorable oral bioavailability is crucial to addressing these limitations. In a previous clinical trial involving oral carbapenems—namely, faropenem—Diacon et al [23] found the drug to be ineffective against TB, although exposures were unexpectedly low. Another oral carbapenem currently being investigated in a phase 2a trial is sanfetrinem (clinicaltrials.gov identifier: NCT05388448), for which enrollment has recently been completed. Our hope is that our results may contribute to the development of these new oral carbapenems. As adequate treatment adherence has consistently been identified as a predictor of mycobacterial clearance and successful treatment outcomes, future regimens including carbapenems should aim to facilitate adherence through improved tolerability and a reduced side-effect profile.
In conclusion, adding rifampicin to meropenem plus amoxicillin/clavulanate did not enhance short-term antibacterial activity in patients with RR-TB.