Genomic signatures of pre-resistance in Mycobacterium tuberculosis

Abstract

Recent advances in bacterial whole-genome sequencing have resulted in a comprehensive catalog of antibiotic resistance genomic signatures in Mycobacterium tuberculosis. With a view to pre-empt the emergence of resistance, we hypothesized that pre-existing polymorphisms in susceptible genotypes (pre-resistance mutations) could increase the risk of becoming resistant in the future. We sequenced whole genomes from 3135 isolates sampled over a 17-year period. After reconstructing ancestral genomes on time-calibrated phylogenetic trees, we developed and applied a genome-wide survival analysis to determine the hazard of resistance acquisition. We demonstrate that M. tuberculosis lineage 2 has a higher risk of acquiring resistance than lineage 4, and estimate a higher hazard of rifampicin resistance evolution following isoniazid mono-resistance. Furthermore, we describe loci and genomic polymorphisms associated with a higher risk of resistance acquisition. Identifying markers of future antibiotic resistance could enable targeted therapy to prevent resistance emergence in M. tuberculosis and other pathogens.

Fig. 1. Phylogenetic analysis of 3134 Mycobacterium tuberculosis isolates from Lima, Peru.

Colors represent different lineages and sublineages. a Maximum likelihood phylogeny. Scale in number of substitutions per genome. b Violin plots showing the posterior density distribution of the inferred substitution rate in substitutions per genome per year derived by sampling from 10⁷ MCMC iterations. The substitution rate was estimated separately for lineage 4 (blue) and lineage 2 (red). Box plots inside the violin indicate the median value of the distribution (black horizontal line) and the interquartile range. Whiskers denote 1.5x the interquartile range, while outliers are plotted as individual points. c Time-calibrated phylogeny of lineage 4. d Time-calibrated phylogeny of lineage 2.

Fig. 2. Inferred posterior density distribution of the earliest occurrence of resistance to first line antituberculous drugs.

Posterior density distribution inferred using a time-calibrated phylogeny for both lineage 4 and lineage 2. Arrows represent the approximate time of antibiotic introduction.

Fig. 3. Dynamics of non-synonymous mutations in rpoC.

a, b Cumulative number of non-synonymous mutations in rpoC over time. The x-axis represents the years since the inferred time of rifampcin resistance (time 0). Dark blue line shows the cumulative number of mutations for the ML tree, while the 95% confidence interval (shaded area) is inferred by repeating the analysis in 100 bootstrap phylogenies. The analysis was performed separately for a lineage 2 and b lineage 4.

Fig. 4. Phylogeographic analysis of Mycobacterium tuberculosis introductions to Peru.

a, b Inferred introductions of Mycobacterium tuberculosis in Peru. The top part shows a time-calibrated phylogeny, with inferred introductions to Peru highlighted in the nodes with colors representing the country from which the clade was introduced. Peruvian clades are shown in blue. The bottom part shows the estimated year of introduction. The analysis was done separately for a lineage 2 and b lineage 4. For visual representation purposes, only the year of introduction of clades with more than 10 tips are shown for lineage 4.

Fig. 5. Hazard ratio and Kaplan–Meier curve for different sublineages of Mycobacterium tuberculosis.

a–c Top: Hazard ratio (HR). Points and error bars represent the HR estimate and the 95% CI, respectively. The p-value for the HR was calculated using the likelihood ratio test. Bottom: Kaplan–Meier curve and numbers at risk. Y-axis represents the probability of remaining susceptible to any antibiotic, while the X-axis shows the time in years or the distance in branch length. Shaded areas show the 95% CI. Kaplan–Meier curves were compared and p-values were derived using the log-rank test. a Depicts the HR of lineage 2 compared to lineage 4 in the Peruvian dataset (HR 3.36, 95% CI 2.10–5.38, Likelihood ratio test p-value = 4.25 × 10⁻⁷) and the different Kaplan–Meier curve for lineage 2 and lineage 4 (Log-rank test p-value = 1.2 × 10⁻⁹). b Same metrics for the Samara dataset (HR 4.82, 95% CI 3.74–6.21, Likelihood ratio test p-value = 6.8 × 10⁻³⁴; Kaplan–Meier curve Log-rank test p-value = 1 × 10⁻³⁹). c Shows HR between lineage 2 and the different sublineages of lineage 4 found in the Peruvian dataset (LAM9, LAM3, LAM11, Haarlem, X type and T type), using LAM3 as a reference (lineage 2 HR 3.32, 95% CI 1.84–6.28, Likelihood ratio test p-value = 1.9 × 10⁻⁴, all other p-values > 0.2; Kaplan–Meier curve Log-rank test p-value = 6.9 × 10⁻⁸). Statistical significance of the hazard ratio differences presented next to the CI bars (*p < 0.05; **p < 0.01; ***p < 0.001).

Fig. 6. Hazard ratio and Kaplan–Meier curve for rifampicin acquisition.

a, b Top: Hazard ratio (HR). Points and error bars represent the HR estimate and the 95% CI, respectively. The p-value for the HR was calculated using the likelihood ratio test. Bottom: Kaplan–Meier curve and numbers at risk. Y-axis represents the probability of remaining susceptible to rifampicin, while the X-axis shows the time in years or the distance in branch length. Shaded areas show the 95% confidence interval. P-values for the Kaplan–Meier curves were calculated using the log-rank test. a Depicts the risk of acquiring rifampicin resistance from an already isoniazid mono-resistant background compared to a drug susceptible one (HR 15.12, 95% CI 10.54–21.69, Likelihood ratio test p-value = 1.3 × 10⁻⁴⁰) and the Kaplan–Meier curves for the different backgrounds (Log-rank test p-value = 2.7 × 10⁻⁶³). b Same metrics for the Samara dataset (HR 37.28, 95% CI 18.81–73.88, Likelihood ratio test p-value = 3.4 × 10⁻²⁵; Kaplan–Meier curve p-value = 4.6 × 10⁻⁶³). Statistical significance of the hazard ratio differences presented next to the CI bars (*p < 0.05; **p < 0.01; ***p < 0.001).

Fig. 7. Genome-Wide association study (GWAS) results.

a Manhattan plot for GWAS on increased risk of drug resistance acquisition in lineage 4. The red line represents the Bonferroni corrected p-value threshold of 3.37×10⁻⁵. Labels show the genes where the significant hits are located. Colors indicate the hazard ratio, with a scale of blue representing hazards ratio lower than 1 and a scale of reds for hazard ratios higher than 1. b Kaplan–Meier curve and numbers at risk of a 9 bp deletion in the gene lppP comparing the probability of remaining susceptible between those nodes without the deletion (blue) and those with it (red). Shaded areas represent the 95% CI. The p-value for the Kaplan–Meier curves was calculated using the log-rank test.