In vitro exposure to clofazimine can select for delamanid and pretomanid resistance in Mycobacterium tuberculosis

TL;DR

Exposure to clofazimine in lab experiments can lead to resistance to other drugs in tuberculosis bacteria.

Contribution

The study shows that clofazimine can select for resistance to delamanid and pretomanid in M. tuberculosis.

Findings

Clofazimine exposure led to resistance to delamanid and pretomanid in M. tuberculosis.

Resistance was associated with mutations in fbiA, fbiC, or fbiD genes.

Rv0678 mutations were observed in some cases.

Abstract

In vitro experiments with Mycobacterium tuberculosis showed that clofazimine exposure selected for delamanid and pretomanid resistance and mutations in fbiA, fbiC, or fbiD—after the acquisition of Rv0678 mutations where this could be determined. Whether this is also possible in vivo and in an Rv0678 wild-type background has to be studied further. Based on the available evidence, however, we propose that nitroimidazole resistance should not be considered an exclusion criterion for the use of clofazimine.

Tables

TABLE 1. Antimicrobial susceptibility testing results of in vitro selected mutants following serial exposure to clofazimine by Ismail et al. (7)h

Parental strain	Acquired mutations and allele frequencya	MGIT MIC (µg/mL) and corresponding categorical AST resultb
		CFZ		DLM		PMD
		Baseline	4th passage	Baseline	4th passage	Baseline	4th passage
H37Rvc	Rv0678 214dupCd (1.0), fbiC Glu142Asp (0.1),e fbiD 559_561delTGC (0.5)	0.125S	4R	0.125S	>1R	0.125S	>2R
R1f	Rv0678 Gln22Pro (0.4) and 172_183delACGGCGCTGGCG (0.6), fbiA 285_286delTGg (0.4) andTrp288Arg (0.3)	0.5S	4R	0.125S	>1R	0.25S	>2R

TABLE 2. Antimicrobial susceptibility testing results of in vitro selected mutants following serial exposure to clofazimine by Snobre et al. (9)i

Isolate	Acquired mutations and allele frequencya	Broth microdilution MIC in µg/mLb				Categorical MGIT resultc
		BDQ	CFZ	DLM	PMD	DLM	PMD
2002-01038(parentald)	None	0.06 (x1);0.125 (x1)	0.125 (x3)	0.008 (x1);0.016 (x1)	1 (x2)	S (x1)	Siu (x1)
2012-01331(1st passagee)	Rv0678 344delAf (0.9)	0.5 (x1);1 (x1)	1 (x2);2 (x1)	0.016 (x1)	0.5 (x1)	S (x1)	Siu (x1)
2013-02481(2nd passageg)	Rv0678 344delAf (1.0), fbiC Arg536Leu (1.0)	0.5 (x1);1 (x1)	2 (x1);4 (x2)	0.25 (x1)	>2 (x1)	R (x1)	R (x1)
H37Rv (controlh)	None	0.125 (x2);0.25 (x1)	0.06 (2x);0.03 (x1)	0.008 (x3)	0.125 (x3)	S (x3)	S (x3)

Full text

INTRODUCTION

Clofazimine is a group B drug for treating rifampicin-resistant Mycobacterium tuberculosis and is included in newly WHO-recommended 6- or 9-month all oral regimens (1). Mutations in Rv0678, a negative regulator of the MmpS5-MmpL5 efflux pump, represent the dominant clofazimine resistance mechanism known to date, whereas pepQ (Rv2535c) mutations are much rarer (both mechanisms confer cross-resistance to bedaquiline) (2). Although Rv1979c has been associated with clofazimine resistance in the literature, no mutations in this gene have been recognized as markers of clofazimine resistance by the WHO mutation catalog to date (2). Six non-essential genes (i.e., ddn [Rv3547], fbiA [Rv3261], fbiB [Rv3262], fbiC [Rv1173], fbiD [Rv2983], and fgd1 [Rv0407]) are required for the activation of both delamanid and pretomanid and, consequently, usually confer cross-resistance to both nitromidazoles when inactivated (2, 3). Waller et al. were the first and, so far, only group to demonstrate that clofazimine can select for high-level in vitro pretomanid resistance in M. tuberculosis due to mutations in fbiA and fbiC (4). We set out to investigate this property of clofazimine further.

Waller et al. used an avirulent auxotrophic strain derived from the lineage 4 H37Rv as the parental strain and obtained mutations in fbiA and fbiC after a single round of selection using three different concentrations of clofazimine, whereas Rv0678 remained wild-type (4). Based on CRISPR interference using the same Rv0678 wild-type parental auxotroph, they then found that transcriptional repression of fbiA, fbiB, fbiC, fbiD, and fgd1 confers approximately twofold MIC increases to clofazimine, whereas the pretomanid MIC increased between 5- to 40-fold depending on the gene (4). By contrast, the clofazimine MIC did not change when ddn expression was repressed, even though this repression correlated with a 46-fold MIC increase to pretomanid. These findings were mirrored in the bactericidal activity assays (4). The shared target of delamanid and pretomanid, dprE2 (Rv3791), and other more recently described potential resistance genes were not investigated (5, 6). Although not formally tested by Waller et al., it is likely that these fbiA and fbiC mutants were cross-resistant to delamanid (2, 3).

In Ismail et al., we reported the findings of in vitro selection experiments involving passages of increasing clofazimine concentrations (7). When analyzing the population derived from the lineage 4 H37Rv ATCC 27294 reference strain after the fourth passage, we found that a non-synonymous mutation in fbiC and an fbiD deletion had likely been selected after the acquisition of an Rv0678 frameshift (Table 1) (7). The equivalent experiment using the R1, a lineage 2 strain, yielded two different Rv0678 mutations as well as two fbiA mutations, for which the order of acquisition could not be inferred (Table 1). Because our focus in Ismail et al. had been on bedaquiline and clofazimine resistance, no antimicrobial susceptibility testing for nitroimidazoles was done at the time (7). Since then, we determined the delamanid and pretomanid MICs to confirm that these fbiA and fbiD mutations correlate with cross-resistance to both agents using the latest WHO critical concentrations (Table 1).

To complement these findings, we analyzed lineage 1 mutants that had been selected in vitro using clofazimine by Snobre et al. (9). During the first round of selection, an Rv0678 frameshift was selected that conferred elevated bedaquiline and clofazimine MICs, whereas the delamanid and pretomanid MICs did not increase (Table 2) (2). Additional exposure to clofazimine selected fbiC Arg536Leu that correlated with cross-resistance to both delamanid and pretomanid using WHO critical concentrations (8, 10). Moreover, the acquisition of the fbiC mutation correlated with a clofazimine MIC increase by approximately a factor of two, which was in line with the magnitude reported by Waller et al. in an Rv0678 wild-type background, whereas the bedaquiline MIC was not affected (Table 2).

Our in vitro results demonstrate that clofazimine can select for cross-resistance to delamanid and pretomanid in three different lineages (i.e., 1, 2, and 4; [Tables 1 and 2](#T1 T2)), but only after the acquisition of Rv0678 mutations. This raises the possibility that the link between clofazimine and nitroimidazoles in a wild-type Rv0678 background observed by Waller et al. might be due to their use of an auxotroph (4). In fact, the twofold MIC increase reported by Waller et al. contrasts with the hypersusceptibility observed by Gurumurthy et al. for a fbiC knockout mutant of M. bovis Bacillus Calmette-Guérin (4, 14, 15). Data from clinical strains are needed to investigate this contradiction. For example, clofazimine MIC testing of Rv0678 wild-type clinical strains with nitroimidazole cross-resistance caused by fbiA, fbiB, fbiC, fbiD, or fgd1 mutations could be carried out (e.g., using some of the mutants known to have arisen prior to the clinical use of either nitroimidazole [16, 17]). Moreover, relevant insights might be gained by studying serial cultures from patients treated with clofazimine as part of regimens without nitroimidazoles, such as historical regimens prior to the approval of nitroimidazoles or the recently WHO-recommended 9 month all-oral regimen of bedaquiline, linezolid, levofloxacin, clofazimine, and pyrazinamide (1).

In light of these uncertainties, we support the current WHO classification of fbiA, fbiB, fbiC, fbiD, and fgd1 as tier 2 genes for clofazimine that require ongoing monitoring (i.e., not tier 1 resistance mechanisms [2]). In fact, even if data from clinical strains confirmed that the five nitroimidazole resistance genes confer an approximately twofold MIC increase to clofazimine, we believe that patients with rifampicin-resistant tuberculosis for whom other regimens are not an option should not be deprived of clofazimine. This should remain the case until strong clinical and/or pharmacokinetic and pharmacodynamic evidence becomes available that these mechanisms actually lead to worse treatment outcomes (e.g., because clofazimine takes several weeks to achieve steady state using the current dosing that was not optimized for treating tuberculosis [18, 19]). WHO used a similar rationale in its first edition of the mutation catalog (20, 21). Specifically, WHO decided not to classify the eis G−10A, C−12T, and G−37GT promoter mutations as amikacin resistance mutations as they only confer approximately twofold MIC increases, whereas it considered the fourfold MIC increase conferred by eis C−14T and rrs C1402T to be likely clinically significant for the current amikacin dose of 15–20 mg/kg/day (i.e., these mutations had not been considered amikacin resistance mechanisms prior to that [1, 2, 20, 22]). Similarly, Pandey et al. also did not consider ndh to be a resistance mechanism for delamanid, ethionamide, isoniazid, or pretomanid (23).

In summary, the findings from this and related studies underline the value of conducting selection experiments using varied conditions, as these may yield unexpected results. The selection of nitroimidazole resistance by clofazimine needs to be further investigated and monitored in the clinical setting. Based on the available data, we propose that, in the rare event that nitroimidazole resistance mutations in fbiA, fbiB, fbiC, fbiD, or fgd1 are found in a pepQ and Rv0678 wild-type background, this nitroimidazole resistance should not be an exclusion criterion for the use of clofazimine (16, 24). Ddn-mediated nitroimidazole resistance has been described to be frequent in at least one setting but does not impact the clofazimine MIC according to Waller et al. (4, 25). Therefore, the detection of Rv0678 resistance mutations will likely be the most frequent genotypic reason for not using clofazimine in practice.