Delamanid Resistance: Update and Clinical Management

Abstract

Delamanid, a-first-in-class bicyclic nitroimidazole, was recently approved for multidrug-resistant tuberculosis treatment. Pitted against the hope for improving treatment outcomes is the threat of the rapid resistance emergence. This review provides information on the mechanisms of action, resistance emergence, and drug susceptibility testing (DST) for delamanid. Delamanid resistance has already been reported in both in vitro experiments and clinical settings. Although mutations conferring delamanid resistance have been identified in fbiA, fbiB, fbiC, ddn, and fgd1 genes of Mycobacterium tuberculosis, knowledge about the molecular resistance mechanisms is limited, and there remains no standardized DST method. The rapid acquisition of delamanid resistance emphasizes the need for optimal use of new drugs, the need for drug resistance surveillance, and a comprehensive understanding of drug resistance mechanisms. Further studies are necessary to investigate genetic and phenotypic changes that determine clinically relevant delamanid resistance to help develop a rapid delamanid DST.

The drug development landscape for tuberculosis (TB) treatment has evolved significantly with the introduction of bedaquiline and delamanid. Delamanid, a first-in-class bicyclic nitroimidazole, was conditionally approved by the European Medicines Agency based on promising phase IIb trial results and medical need for the MDR-TB (multidrug-resistant tuberculosis: resistant to at least isoniazid and rifampicin) treatment in 2014 [1]. Delamanid has been made available to over 100 low- and middle-income countries eligible for financing through the Global Fund to Fight AIDS, TB, and Malaria [2].

Delamanid resistance has been already reported in the context of inadequate MDR-TB and XDR-TB (extensively drug-resistant TB: MDR-TB that is also resistant to any fluoroquinolone [FQ] and any of the second-line injectable agents) treatment regimens [3]. The surprisingly rapid acquisition of delamanid resistance observed toward a novel medication emphasizes the need for appropriate use of new drugs and underlines the significance of drug resistance surveillance. Using delamanid in combination with other active anti-TB agents is recommended to prevent acquired resistance [4, 5]. This review aims to provide an overview of the mechanisms of action, identified resistance mechanisms reported in clinical settings to date, the status of drug susceptibility testing (DST) methods, and provide recommendations on how to prevent the emergence of delamanid resistance.

METHODS

A nonsystematic literature review of English literature was performed in PubMed and Web of Science using the keywords: tuberculosis, delamanid, mechanism of action, resistance, drug susceptibility testing, mutations, pharmacokinetics, and pharmacodynamics. References of retrieved articles were screened as well. Evidence resulting from the literature search was summarized and recommendations provided.

Mechanisms of Action

Delamanid inhibits the synthesis of methoxy-mycolic and keto-mycolic acid, cell wall components of mycobacteria [6]. The pro-drug delamanid is activated by Mycobacterium tuberculosis (MTB) reductive metabolism to produce an active free radical via the mycobacterial F₄₂₀-dependent nitroreductase coenzyme system. This system includes deazaflavin-dependent nitroreductase (Ddn) encoded by the gene ddn (Rv3547) [4], a glucose-6-phosphate dehydrogenase (G6PD) encoded by fgd1 (Rv0407), and 3 proteins in the F₄₂₀ biosynthetic pathway, FbiA encoded by fbiA (Rv3261), FbiB encoded by fbiB (Rv3262), and FbiC encoded by fbiC (Rv1173) genes, together responsible for the synthesis and the reactivation of the cofactor F₄₂₀. FbiC catalyzes the transfer of the hydroxybenzyl group from 4-hydroxy-phenylpyruvate pyrimidinedione to 5-amino-6-ribitylamino-2,4(1H,3H)- pyrimidinedione to form 7,8-didemethyl-8-hydroxy-5-deazariboflavin (FO). FO is further modified by FbiA and FbiB [7, 8]. Thus, Ddn metabolizes delamanid into its active form. In addition, G6PD recycles F420 into the reduced form [4]. Ddn also converts delamanid into a desnitro form not active against MTB. Likely, additional mechanisms, for example, interruption of cellular respiration, due to reactive intermediates in the metabolic pathway of the bicyclic nitroimidazoles contribute to delamanid efficacy [4, 5, 9]. The genes involved in the bioactivation of delamanid are shown in Figure 1.

Pre-existing and Emergence of Delamanid Resistance

Recently, the in vitro delamanid susceptibility of 90 clinical XDR MTB isolates from delamanid-naive patients in China was tested in liquid Middlebrook 7H9 medium [11]. A raised minimum inhibitory contribution (MIC) was found in 4 (4.4%) isolates (1 with a MIC of 0.5 mg/L, 3 with MICs of 32 mg/L) [11]. In a study on 420 clinical strains with no previous delamanid-exposure collected in Korea [12], delamanid resistance was found in 41 strains (9.76%) using a critical concentration of 0.2 mg/L as suggested by Otsuka [13]. Using the same breakpoint, another in vitro study on 220 clinical MDR/XDR strains with no previous delamanid-exposure in China detected 7 (3.18%) delamanid-resistant isolates [14].

In Trial 204 (phase II), delamanid resistance was observed at baseline in 2 of 316 participants (0.63%) and acquired during treatment in 4 of 205 participants (1.95%) [15]. In Trial 213 (phase III), delamanid baseline resistance was observed in 2 of 511 (0.39%) participants; acquired delamanid resistance was observed in 1.17% (4/341) of participants receiving delamanid for 6 months but in none of the 170 intent-to-treat participants in the placebo arm [16]. The 4 participants with acquired delamanid resistance received only 2 other antimycobacterial drugs in addition to delamanid [16].

Delamanid resistance was also documented after compassionate use in a single patient, with no previous TB treatment. The initial pre-XDR MTB isolate had 9 mutations in genes associated with resistance to 7 drugs [3]. The presence of a compensatory mutation in rpoC suggests a mature pre-XDR strain that evolved under antimicrobial drug pressure. Resistance rapidly developed to both bedaquiline and delamanid, despite personalized care in a well-resourced healthcare setting. The monitoring using next generation sequencing (NGS) demonstrated that multiple heterogeneous populations of MTB were involved [3].

Mechanisms of Delamanid Resistance

As delamanid is a prodrug, any mutations in the mycobacterial genome that reduce or prevent the conversion of delamanid to desnitroimidazooxazole lead to low- or high-level delamanid resistance, respectively [4]. This results in a high frequency of spontaneous mutations in vitro [5] as any viable mutations disrupting the function of these genes may lead to a rise in the MIC.

In laboratory conditions, mutations in 5 coenzyme F420 genes (fbiA, fbiB, fbiC, fgd, and Rv3547) were shown to result in the loss of the capacity to metabolize and activate delamanid in spontaneous delamanid-resistant M. bovis BCG Tokyo mutants [5]. As delamanid MICs and the F420-dependent bioactivation pathway in MTB and M. bovis Tokyo are comparable, it is expected that they share the mechanisms of delamanid resistance [5]. Indeed, in clinical delamanid-resistant MTB isolates, mutations in these 5 genes have also been observed (Table 1). For example, a few months after the addition of delamanid to the treatment regimen of an XDR-TB patient with clofazimine and bedaquiline resistance, 2 mutations in fbiA and fgd1 genes increased in frequency, coinciding with the emergence of phenotypic resistance to delamanid [3]. The frequency of the fgd1 mutation decreased thereafter, suggesting the presence of multiple delamanid-resistant clones [3]. The fbiA D49A mutation is not involved in substrate binding but was considered the most likely variant conferring successful delamanid resistance [17, 18].

A genetic analysis of 159 delamanid unexposed clinical strains from 4 different geographical regions revealed only 4 mutations in the ddn and fbiA genes leading to the following amino substitutions, Ddn Trp-88 → STOP and FbiA Lys-250→STOP. All 4 strains with mutations were delamanid-resistant; no mutations in these genes were observed in any of the 155 susceptible isolates investigated [19].

A delamanid MIC of 0.25 mg/L, just 3 dilution steps above the proposed epidemiological cutoff (ECOFF) of 0.03 mg/L, was reported from an XDR-TB patient with a Ddn G53D amino acid substitution [21]. The Ddn G53D did not disrupt the protein function entirely and resulted in less dramatic rise in MIC than mutations that inactivate the protein. The G53 amino acid is located in the conserved domain of the Ddn and thus may limit delamanid bioactivation. This results in low level but clinically relevant resistance, although preserving some useful enzymatic function [21], as is the case for the isoniazid resistance mutation in katG codon 315 [2].

Genetic analysis of 4 delamanid-resistant XDR-TB strains from patients in China revealed no mutations in ddn, fgd1, fbiA, or fbiB, but mutations in the codon 318 of fbiC gene were observed. These fbiC mutations were considered important contributors to delamanid resistance [11], although additional/alternative resistance mechanisms could not be excluded.

In 420 MTB clinical strains, fbiA or ddn mutations were found in 33 of 41 delamanid-resistant strains [12], but some mutations were different from those reported previously [3, 19, 22]. A Gly81Ser amino acid substitution in Ddn was observed with a high frequency of 75.6% in the resistant strains and with an even higher frequency (81%) in resistant strains with MIC > 0.4 mg/L. However, all 379 susceptible strains (MIC ≤ 0.0125 mg/L) also had the same mutation; it is thus unlikely that this mutation is directly involved in the resistance mechanism [12].

Recently, loss of function mutations in cofC (Rv2983) gene have been shown to cause pretomanid and delamanid resistance [23]. This gene was associated with all pretomanid resistance detected, although no mutation in the 5 previously reported genes were observed [23]. The CofC is a putative guanylyltransferase involved in the F420 synthesis and is suspected to play a similar role to CofC in the methanogen of Methadocalnococcus jannaschii in which it is required for F420 synthesis [24]. This is the first cofC homolog found in MTB, and it has been shown to catalyze an important step required for F420 biosynthesis in MTB. This observation raises the prospect that pretomanid exposure might result in selection of delamanid resistance and vice versa [23]. High-level resistance to pretonamid and delamanid in MTB was shown to be likely, at least in part, the result of a novel nonsynonymous mutation in the fbiA gene leading to the amino-acid substitution Glu249Lys in FbiA [14]. However, a lack of cross-resistance between these 2 drugs has also been reported, highlighting the need to explore new mechanisms of resistance [14].

Drug Susceptibility Testing

Several delamanid DST methods have been proposed; however, none are currently standardized, nor is there a commercially available test that explains the lack of data on programmatic evaluation of DST for delamanid.

Delamanid had equally potent in vitro activity against both sensitive and MDR-TB strains and was more potent than pretonamid [14], with MIC values between 0.006 and 0.012 μg/mL [4]. Recently, the World Health Organization (WHO) established critical concentrations for delamanid DST, 0.016 mg/L by middle brook 7H11, and 0.6 mg/L by MGIT liquid culture. However, DST is currently not widely implemented, and most laboratories do not have or have only recently started to develop protocols for routinely performing delamanid DST. MGIT (BD) is proposed as the reference method for delamanid DST at present [10].

The MIC data reported from different studies are presented in Tables 2, 3, and 4. Initially, Otsuka proposed the agar proportion method (APM) on Middlebrook 7H11 [9]. Using this APM, with the exception of 2 isolates, MICs of 460 isolates from MDR-TB patients ranged from 0.001 to 0.05 μg/mL with a median of 0.004 μg/mL [13] (Table 2).

DST using a semi-automated MGIT 960 system and EpiCenter software equipped with a TB eXiST module was proposed by Keller et al [25]. A mean wild-type delamanid MIC of 0.013 mg/L and an ECOFF of 0.04 mg/L was proposed, but this ECOFF value could not be confirmed [25] (Table 3).

Schena et al [19] studied 333 MTB isolates from patients never exposed to drugs and established DST protocols with pure delamanid using the resazurin microtiter assay (REMA) and the MGIT. Both assays showed bell-shaped MIC distributions and were confirmed to be in concordance with the APM. The ECOFF for the REMA and MGIT was 0.03 and 0.06 mg/L, respectively. Further studies should be done to confirm the proposed DST protocols [19] (Table 4).

Recently, the first microtiter plate assay UKMYC5 was designed to measure the MICs of 14 anti-TB agents including bedaquiline and delamanid [20]. The UKMYC5 plate was consistent with APM and MGIT in classification of resistance/susceptible strains, achieved indistinguishable interdealer agreement of 97.9%, intra- and interlaboratory reproducibility of 95.6 and 93.1%, respectively. This assay could enable large-scale DST and detect low levels of resistance providing opportunity to guide MDR-TB treatment [20] (Table 3).

Pharmacokinetics/Pharmacodynamics

To define the critical concentration that separates resistance and susceptibility and MIC distribution, pharmacokinetic (PK) and pharmacodynamic (PD) data are combined with clinical outcome data [26]. Based on the current data, delamanid showed dose-dependent killing in mice and in humans [9, 27].

In the phase II trial, no significant difference was seen between the 2 different dosages tested, 100 mg twice a day (bid) and 200 mg bid [15]. The exposure achieved with 200 mg bid was only moderately higher (50%) than with 100 mg bid and did not exclude comparable exposure among at least a part of the participants in both treatment arms. The average delamanid plasma concentrations at 100 mg bid were 0.372–0.562 mg/L, which was 2–3 dilution steps above the ECOFF, resulting in a proposed breakpoint of 0.125 mg/L for the REMA and MGIT [19]. PK data from participants in trial 204 receiving a dose of 100 mg bid showed that 95.8% of the participants had a C_max ≥ 0.2 mg/L and 79.8% had a C_min ≥ 0.2 mg/L, resulting in a proposed breakpoint of 0.2 μg/mL [13]. To date, PK/PD has not yet been evaluated in a programmatic setting.

DISCUSSION

Reports of delamanid-resistant TB infections soon after delamanid introduction highlight the potential for rapid acquired resistance in the presence of a weak MDR/XDR-TB treatment regimen emphasizes the need of systemic surveillance of drug resistance for new anti-TB agents [3]. Although the delamanid resistance mutation rate has not been estimated yet, the in vitro resistance frequencies reported range from 4.19 × 10⁻⁵ to 6.44 × 10⁻⁶ in 10 MTB H37Rv cultures. This rate is similar to the in vitro rates for isoniazid and pretomanid, but higher than those for rifampicin [5]. In addition, the potential to rapidly acquire resistance seriously hampers the efficacy of adding new drugs to cure pre-XDR and XDR strains especially when the prescribed regimen is weak [17, 21, 28, 29]. Moreover, reports of preexisting resistance to delamanid in a small proportion of unexposed isolates raise significant concerns regarding the current lack of routine DST.

To minimize the emergence of delamanid resistance, integrated approaches at diagnosis for the selection of the therapeutic regimen and monitoring during treatment are all important. Delamanid use should be based on individual assessment including drug susceptibility data, drug safety, and tolerability, risk-benefit analysis, and ethical considerations; this requires consultation with experts in MDR-TB management. Programmatic active drug safety monitoring (aDSM) has been recommended to detect and report adverse drug reactions to facilitate patient-centered management [30].

Restricting access to drugs for patients with sparse treatment options can result in weak therapeutic regimens [17]. The sequential use of the drugs rather than full review of a regimen increases the risk of acquired resistance and treatment failure [31]. To minimize risks, regulatory policy, competent MDR/XDR-TB management, active pharmacovigilance, patients’ informed consent, and appropriate consultation of experts via national and supranational TB consilia such as the WHO and GTN (Global Tuberculosis Network) consilium must be in place [31–33].

At present, knowledge on molecular mechanisms toward delamanid resistance is limited [17, 34]. Currently, reported resistance mechanisms mainly concern the mutations in the 5 genes involved in delamanid bioactivation. These genes have diverse dispersed distribution of mutations, which could be a hurdle for the development of a rapid molecular DST, requiring WGS [5]. Further studies are necessary to clarify the role of F420 biosynthesis in delamanid resistance and to elucidate the genetic and phenotypic changes in the resistant strains to guide the development of a rapid DST [5, 11]. Heteroresistance can be detected by WGS during treatment by identifying relevant mutations present in at least 10% of the reads when there is more than 100-fold average genome-wide coverage [21].

Currently, only phenotypic DSTs, which rely on the critical concentrations, are available for delamanid; thus, improving the phenotypic tests could result in a more informative DST [5]. Irrespective of baseline resistance, regular and repeated DST for delamanid at baseline and during treatment therapy [21] would help reduce the likelihood of transmitted resistance and rapidly detect acquired resistance. Clinical isolate collection from patients during treatment will provide a valuable resource for future studies [19]. If resistance is detected, the isolates should be stored and if possible, WGS should be performed to collect data on mutations conferring in vitro resistance [10]. Quantitative phenotypic DST data are needed to deduce phenotypes from genotypes but are unfortunately lacking for most of the reported genetic data sets. The links between genotypic and phenotypic drug resistance and outcome would help better understand the links between resistance mechanisms and clinical responses with the evolution and transmission of resistant MTB [6, 35]. Clinically, it is conceivable that delamanid resistance and treatment failure may be associated with fit strains with only slightly raised MIC levels. This needs to be considered for future comparisons between molecular and phenotypic DST results [21]. There is still a lack of data on validated media for MIC testing, PK/PD, and clinical data to define robust clinical breakpoints for delamanid according to EUCAST criteria [10, 36]. Although WHO already analyzed delamanid MIC data and set appropriate interim clinical concentrations for MGIT and 7H11 [10], there is an urgent need for the evaluation of programmatic use of DST for delamanid.

The risk of resistance emergence is likely to be unacceptably high if delamanid is not always coadministered with at least 1 active potent bactericidal drug or multiple other weakly bactericidal or bacteriostatic agents [4]. It was advised that delamanid should not be introduced into a regimen containing ineffective drugs and should not be added as a single drug to a failing regimen [2, 29, 37–39]. Although even prior to this recommendation the concurrent use of both bedaquiline and delamanid in XDR-TB patients was increasingly described; in a minority of cases, this combination was even exclusively prescribed by local clinicians [36]. The effective combination of bedaquiline and delamanid with good tolerance as part of MDR-/XDR-TB, rather than sequentially with washout to prevent drug resistance selection, was reported in the case of limited clinical options [3, 40, 41]. However, results should be evaluated cautiously and ideally confirmed in well-designed experimental studies.

The limited evidence of delamanid efficacy and the reported resistance emergence may raise doubts regarding the proposed posology of 100 mg twice daily as the optimal schedule [4]. Doses of new drugs used for regulatory approval should not necessarily be assumed to be optimal [42]. In case of delamanid, there is a paucity of literature on PK/PD, including real-life data that can be used to guide dose selection. It could be recommended to study PK/PD more extensively in the hollow fiber infection model and in operational research [43–45]. Such modeling would allow assessment of the effect of drug concentration on acquired resistance. Integrating human variability in PK, MIC distribution, kill kinetics from the hollow fiber model, and the use of Monte Carlo simulation could support dose optimization [46].

CONCLUSION

Mutations in fbiA, fbiB, fbiC, ddn, and fgd1 can result in delamanid resistance. The clinical introduction of delamanid should be accompanied by simultaneous establishment of a standardized and reproducible DST to avoid early development of drug resistance and transmission of resistant strains. Systematic resistance surveillance and aDSM support clinical management and the optimal use of delamanid in the context of patient centered care, which will help prevent resistance acquisition during treatment.

Note

Potential conflicts of interest. The authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.

References