High fluoroquinolone resistance proportions among multidrug-resistant tuberculosis driven by dominant L2 Mycobacterium tuberculosis clones in the Mumbai Metropolitan Region

Abstract

Background: Multidrug-resistant (MDR) Mycobacterium tuberculosis complex (MTBC) strains are a serious health problem in India, also contributing to one-fourth of the global MDR tuberculosis (TB) burden. About 36% of the MDR MTBC strains are reported fluoroquinolone (FQ) resistant leading to high pre-extensively drug-resistant (pre-XDR) and XDR-TB (further resistance against bedaquiline and/or linezolid) rates. Still, factors driving the MDR/pre-XDR epidemic in India are not well defined.

Methods: In a retrospective study, we analyzed 1852 consecutive MTBC strains obtained from patients from a tertiary care hospital laboratory in Mumbai by whole genome sequencing (WGS). Univariate and multivariate statistics was used to investigate factors associated with pre-XDR. Core genome multi locus sequence typing, time scaled haplotypic density (THD) method and homoplasy analysis were used to analyze epidemiological success, and positive selection in different strain groups, respectively.

Results: In total, 1016 MTBC strains were MDR, out of which 703 (69.2%) were pre-XDR and 45 (4.4%) were XDR. Cluster rates were high among MDR (57.8%) and pre-XDR/XDR (79%) strains with three dominant L2 (Beijing) strain clusters (Cl 1-3) representing half of the pre-XDR and 40% of the XDR-TB cases. L2 strains were associated with pre-XDR/XDR-TB (P < 0.001) and, particularly Cl 1-3 strains, had high first-line and FQ resistance rates (81.6-90.6%). Epidemic success analysis using THD showed that L2 strains outperformed L1, L3, and L4 strains in short- and long-term time scales. More importantly, L2 MDR and MDR + strains had higher THD success indices than their not-MDR counterparts. Overall, compensatory mutation rates were highest in L2 strains and positive selection was detected in genes of L2 strains associated with drug tolerance (prpB and ppsA) and virulence (Rv2828c). Compensatory mutations in L2 strains were associated with a threefold increase of THD indices, suggesting improved transmissibility.

Conclusions: Our data indicate a drastic increase of FQ resistance, as well as emerging bedaquiline resistance which endangers the success of newly endorsed MDR-TB treatment regimens. Rapid changes in treatment and control strategies are required to contain transmission of highly successful pre-XDR L2 strains in the Mumbai Metropolitan region but presumably also India-wide.

Keywords: Fluoroquinolone resistance; India; Pre-XDR/XDR-TB; Multidrug-resistant TB; Pre-XDR/XDR-TB transmission; Resistant TB; Transmission success; Tuberculosis.

Fig. 1

Study flowchart. In- and exclusion criteria for strains are reported in the two rhombuses. Final dataset consists of 1852 strains

Fig. 2

Genotype and resistance category distribution of strains across 1852 clinical isolates (Total) and within the three major clusters. A Distribution of resistance category across the 1852 isolates; resistance categories are RR (rifampicin resistant), nonMDR (resistant, but not multi drug resistant [MDR]), MDR, pre-XDR (pre-extensively drug resistant), and XDR (extensively drug resistant); of the total strain population, about 37% are susceptible (S) to all drugs, 38% of strains are pre-XDR whereas 2.4% are XDR in the cohort. The resistant category distribution of clusters 1, 2, and 3 differs, with all strains being at least MDR). B Proportion of strains with known resistance mutations per lineage. Each bar represents a specific antibiotic or compensatory effect. L2 strains, especially the ones from clusters 1–3, have the highest proportion of resistance mutations and also compensatory effects. Abbreviations: INH, isoniazid; RMP, rifampicin; Comp., compensatory mutation; EMB, ethambutol; PZA, pyrazinamide; MFX, moxifloxacin; Inj., injectables; PTH, prothionamide; PAS, para-aminosalycilic acid; CS, cycloserine; BDQ, bedaquiline; CFZ, clofazimine; LZD, linezolid; L1, Lineage 1; L2, Lineage 2; L3, Lineage 3; L4, Lineage 4

Fig. 3

Maximum likelihood phylogeny based on the concatenated SNP sequence of 258 MTBC strains from allele-based cluster 1. The concatenated SNP sequence consists of 356 parsimony-informative, 961 singleton sites, and 1106 constant sites; mutations related to respective drugs and resistance status are color coded and expressed as annotation rings on the tree. SNP-based clusters with maximum distance of 12 (d12) is plotted on the outer ring. Abbreviations: INH, isoniazid; EMB, ethambutol; PZA, pyrazinamide; FQ, fluoroquinolones

Fig. 4

Geographical occurrence of patients with maximum SNPs distance of 12 for cluster 1 across Mumbai Metropolitan Region. The geographical distribution underlines the widespread of all cluster 1 SNP subgroups across the city and its neighboring areas. Boundaries of the map for the neighboring regions are not available online

Fig. 5

MTBC lineage 2 exhibits greater epidemic success than Lineages 1, 3, and 4. Shown are boxplots and distribution of THD success indices in MTBC Lineages 1–4, with Lineage 2 split into clusters 1–3 and other strains. THD success indices were larger in Lineage 2 compared with other lineages, using both a long-term (A) and a short-term (B) analysis time scale. Within Lineage 2, clusters 1–3 had similar long-term success as other strains over a long-term time scale (A) but had superior success in the short-term (B), suggesting that strains from clusters 1–3 became successful only recently. P-values obtained from 2-sided Mann–Whitney U test

Fig. 6

Compensatory mutations correlate with increased epidemic success of MDR MTBC strains. Shown are boxplots and distribution of THD success indices with a 20-year time scale in MTBC lineages 2–4 with resistance profile categorized as not-MDR, MDR, and MDR + (pre-XDR/XDR). Within each lineage and resistance profile, strains without compensatory mutations (green) were compared with those with at least 1 compensatory mutation (orange). Compensatory mutation(s) correlated with increased success indices in MDR lineage 2, MDR + lineages 2 and 3, but not in lineage 4 strains. In nonMDR lineage 2 strains, compensatory mutation(s) correlated with lesser success. P-values obtained from 2-sided Mann–Whitney U test. Lineage 1 was excluded and clusters 1–3 were pooled with other lineage 2 isolates because all lineage 1 and clusters 1–3 strains had at least 1 compensatory mutation