A composite reference standard is needed for bedaquiline antimicrobial susceptibility testing for Mycobacterium tuberculosis complex

Abstract

A composite reference standard minimises false-susceptible AST results for bedaquiline https://bit.ly/3wAVvFm

FIGURE 1

Illustrative plot of different bedaquiline minimum inhibitory concentration (MIC) distributions, where the clinical breakpoint (CB)^S/R corresponds to the epidemiological cut-off (ECOFF) [8]. Because the dosing of bedaquiline is fixed and no regulator has endorsed that higher exposures at particular sites overcome modest MIC increases, there is no “intermediate” or “susceptible, increased exposure” range for bedaquiline, which means that “intermediate” should not be used to refer to mutations either [2, 8]. The relative frequency of the different MIC distributions is not representative (e.g. atpE resistance mutations are rarer than those in mmpR5) and the relative MIC increases were chosen for illustrative purposes given that the different mechanisms have never been tested in the same study using the same method under controlled conditions (e.g. with an on-scale quality control (QC) strain in every batch and with sufficiently low antibiotic concentrations to obtain untruncated MICs for all distributions [3, 13, 15, 17]). Most atpE resistance mutations confer large MIC increases (>16-fold), meaning that these mutations test reliably as resistant [3, 13]. Full loss-of-function (LoF) mmpR5 mutants that cause maximal overexpression of the mmpL5-mmpS5 efflux pump confer smaller relative MIC increases (4- to 8-fold) that would be expected to be even more modest for mmpR5 mutants that retain some repressor activity [13]. Given the inherent technical variability of MIC testing, the reproducibility of the latter borderline mutants would be particularly poor at the CB^S/R. In fact, the overlap between the susceptible (S) MIC distribution of mmpR5 borderline/full LoF mutants is likely exacerbated by the modest collateral hyper-susceptibility conferred by katG mutations (i.e. such mutants have approximately 2-fold lower bedaquiline MICs) [20]. The misclassification of those mutants as susceptible could be minimised by setting an area of technical uncertainty (ATU) that corresponds to the CB^S/R. By contrast, some mutations in resistance genes do not affect the phenotype and the C-11A mmpR5 promoter mutation correlates with a borderline hyper-susceptible phenotype [17]. Such neutral and modest hyper-susceptible mutations are not distinguished in the World Health Organization (WHO) mutation catalogue and would be classified as group 4/5 “not associated with resistance (interim)” [3]. Lastly, LoF mutations in either subunit of the mmpL5-mmpS5 efflux pump should result in a more marked hyper-susceptible phenotype that must be considered for genotypic AST (i.e. group 1/2 mmpR5 mutations cannot confer bedaquiline resistance if genetically linked to a LoF mutation in either subunit, but WHO did not endorse epistasis for mmpS5 as the available dataset lacked clinical mmpS5 mutants [3, 17]). pepQ mutations likely confer similar MIC increases to mmpR5 but appear to be much rarer and are not affected by LoF mutations in mmpL5-mmpS5 [3].