Natural Polymorphisms in Mycobacterium tuberculosis Conferring Resistance to Delamanid in Drug-Naive Patients

Mutations in the genes of the F420 signaling pathway of Mycobacterium tuberculosis complex, including dnn, fgd1, fbiA, fbiB, fbiC, and fbiD, can lead to delamanid resistance. We searched for such mutations among 129 M. tuberculosis strains from Asia, South America, and Africa using whole-genome sequencing; 70 (54%) strains had at least one mutation in one of the genes.

I n 2014, the new antituberculosis (anti-TB) drug delamanid (also known as OPC-67683, or Deltyba) was introduced (1). The World Health Organization (WHO) recommends the administration of delamanid if a standard effective drug regimen cannot be prescribed due to drug toxicity or resistance (2,3). Thus, the European Medicines Agency (EMA) conditionally approved delamanid for the treatment of multidrug-resistant (MDR) TB (1,3,4). Of note, 6 years after its market launch, robust and widely accepted breakpoints that define susceptibility and resistance to delamanid still do not exist (5). The few available studies suggest a critical MIC between 0.125 mg/liter and 0.2 mg/liter, and an epidemiological cutoff value (ECOFF) of 0.04 mg/liter (6)(7)(8)(9). This ECOFF is in line with the WHO technical report (10).
This multicenter study has been described in detail elsewhere and is part of the work of the International epidemiology Databases to Evaluate AIDS (IeDEA) (23). We identified putative delamanid resistance-conferring mutations in M. tuberculosis strains from TB patients living with HIV (PLWH) and delamanid-naive, HIV-negative TB patients by whole-genome sequencing (WGS) and MIC determination. We collected information on the demographic and clinical characteristics of patients who were recruited between 2013 and 2016 in Peru, Thailand, Côte d'Ivoire, the Democratic Republic of the Congo (DRC), Kenya, and South Africa (24,25). The Cantonal Ethics Committee in Bern, Switzerland, and local institutional review boards approved the study. Written informed consent was obtained at all locations, except in South Africa, where consent was not required for archived samples.
We included 129 M. tuberculosis isolates, among them 51 isolates (39.5%) from Peru, 13 (10.1%) from Thailand, 49 (38%) from Côte d'Ivoire, 14 (10.9%) from the DRC, and 1 (0.8%) each from Kenya and South Africa. We identified 70 (54.3%) isolates with polymorphisms in at least one of the six F 420 genes compared to the reference genome (Table S1). All patients infected with either of these strains were naive to delamanid. We selected strains fulfilling the following criteria: (i) mutations in a part of the gene encoding regions of catalytic or structural importance predicted by ARIBA and then the PhyResSE pipeline (32,33), (ii) availability of a culture of the strain, and (iii) bacterial growth amenable to microdilution (25). MICs were determined for 10 isolates with mutations in the F 420 genes. Four isolates showed MICs of Ͼ0.015 mg/ liter: specifically, MICs of 0.5 (patient 1), 0.03 (patients 6 and 10), and Ͼ8 (patient 9) mg/liter (Table 1; Fig. S1). The isolate from patient 1 had a polymorphism in fgd1 (Lys270Met) and was susceptible to the six drugs tested (isoniazid, rifampin, ethambutol, pyrazinamide, moxifloxacin, and amikacin). The patient was cured. The isolate from patient 9 had two alterations: a deletion in ddn (Tyr29del) and a nucleotide change in fgd1 (T960C). The strain showed an elevated delamanid MIC and was phenotypically susceptible to six other drugs tested. The patient died. The MIC for the isolates of patients 10 and 6 was above 0.015 but below 0.04 mg/liter (Table 1). This suggests low-level resistance to delamanid (22), which could be due to the combination of various mutations: Ala416Val (fbiC), Trp678Gly (fbiC), Arg64Ser (fgd1), and T960C (fgd1).
In summary, in the subset of 10 isolates with polymorphisms in the six targeted genes, six had no elevated MIC in the microdilution, while four isolates had elevated MICs (Table 1). In line with previous studies, we found that Lys270Met in fgd1 is a natural polymorphism characteristic of M. tuberculosis lineage 4.1.2.1, which may (patients 1 and 6) or may not (patient 7) lead to an increased delamanid MIC (19,34,35). All 16 strains of lineage 4.1.2.1 showed this lineage-specific marker (Table S1). Furthermore, T960C (fgd1) is a synonymous substitution and was found in three other patient isolates which, as expected, did not have a critical MIC. The increase in the delamanid MIC for the isolate of patient 9 was due to the deletion in ddn (7). Our results thus suggest that Tyr29del is a natural polymorphism leading to an increased delamanid MIC. Our study was too small to estimate the prevalence of strains that are naturally resistant to delamanid. In 2020, Lee et al. screened 14,876 M. tuberculosis strains and found 2 strains with Tyr29del, for a prevalence of 0.013% (36). However, in their study, only the ddn gene was screened, and the prevalence of natural resistance could, therefore, be higher.
In conclusion, we confirm that mutations in F 420 genes can confer an elevated delamanid MIC (13,19). Whether our findings also apply to the related drug pretomanid should be investigated in future studies. The occurrence of clinical M. tuberculosis isolates from previously untreated patients for which delamanid MICs are naturally elevated calls for careful drug susceptibility testing (DST) prior to delamanid treatment (5,36). However, access to DST is limited in high-burden countries. This dilemma highlights the conflict between making new drugs available in high-burden countries and avoiding the spread of drug-resistant strains.
Data availability. WGS data from patients' M. tuberculosis strains shown in Table 1 have been submitted to the NCBI (BioProject accession no. PRJNA300846) (Table S1).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.2 MB. SUPPLEMENTAL FILE 2, XLSX file, 0.02 MB.

ACKNOWLEDGMENTS
We thank those at all sites who participated, the patients whose data were used in this study, and Marie Ballif for contributing to the data collection and for critical reading of the manuscript. Calculations were performed on UBELIX (http://www.id.unibe.ch/ hpc), the HPC cluster at the University of Bern.
This research was supported by the Swiss National Foundation (project grants 153442, 310030_166687, 310030_188888, IZRJZ3_164171, IZLSZ3_170834, and CRSII5_ ). This work is solely the responsibility of the authors and does not necessarily represent the official views of any of the institutions mentioned above.
We have no potential conflicts of interest to disclose.