Validation of Novel Mycobacterium tuberculosis Isoniazid Resistance Mutations Not Detectable by Common Molecular Tests

Genetic mechanisms of co-emergence of INH-resistant Mycobacterium tuberculosis strains during the standard course of antituberculosis therapy

The incidence of isoniazid (INH) resistant Mycobacterium tuberculosis is increasing globally. This study aimed to identify the molecular mechanisms behind the development of INH resistance in M. tuberculosis strains collected from the same patients during the standard course of treatment. Three M. tuberculosis strains were collected from a patient before and during antituberculosis (anti-TB) therapy. The strains were characterized using phenotypic drug susceptibility tests, Mycobacterial Interspersed Repeated Unit-Variable-Number Tandem Repeats (MIRU-VNTR), and whole-genome sequencing (WGS) to identify mutations associated with INH resistance. To validate the role of the novel mutations in INH resistance, the mutated katG genes were electroporated into a KatG-deleted M. tuberculosis strain (GA03). Three-dimensional structures of mutated KatG were modeled to predict their impact on INH binding. The pre-treatment strain was susceptible to INH. However, two INH-resistant strains were isolated from the patient after anti-TB therapy. MIRU-VNTR and WGS revealed that the three strains were clonally identical. A missense mutation (P232L) and a nonsense mutation (Q461Stop) were identified in the katG of the two post-treatment strains, respectively. Transformation experiments showed that katG of the pre-treatment strain restored INH susceptibility in GA03, whereas the mutated katG genes from the post-treatment strains rendered negative catalase activity and INH resistance. The protein model indicated that P232L reduced INH-KatG binding affinity while Q461Stop truncated gene transcription. Our results showed that the two katG mutations, P232L and Q461Stop, accounted for the co-emergence of INH-resistant clones during anti-TB therapy. The inclusion of these mutations in the design of molecular assays could increase the diagnostic performance.IMPORTANCEThe evolution of drug-resistant strains of Mycobacterium tuberculosis within the lung lesions of a patient has a detrimental impact on treatment outcomes. This is particularly concerning for isoniazid (INH), which is the most potent first-line antimycobacterial drug. However, the precise genetic factors responsible for drug resistance in patients have not been fully elucidated, with approximately 15% of INH-resistant strains harboring unknown genetic factors. This raises concerns about the emergence of drug-resistant clones within patients, further contributing to the global epidemic of resistance. In this study, we revealed the presence of two novel katG mutations, which emerged independently due to the stress exerted by antituberculosis (anti-TB) treatment on a parental strain. Importantly, we experimentally demonstrated the functional significance of both mutations in conferring resistance to INH. Overall, this research sheds light on the genetic mechanisms underlying the evolution of INH resistance within patients and provides valuable insights for improving diagnostic performance by targeting specific mutations.

Mutations and insights into the molecular mechanisms of resistance of Mycobacterium tuberculosis to first-line

Genetically antimicrobial resistance in Mycobacterium tuberculosis is currently one of the most important aspects of tuberculosis, considering that there are emerging resistant strains for almost every known drug used for its treatment. There are multiple antimicrobials used for tuberculosis treatment, and the most effective ones are the first-line drugs, which include isoniazid, pyrazinamide, rifampicin, and ethambutol. In this context, understanding the mechanisms of action and resistance of these molecules is essential for proposing new therapies and strategies of treatment. Additionally, understanding how and where mutations arise conferring a resistance profile to the bacteria and their effect on bacterial metabolism is an important requisite to be taken in producing safer and less susceptible drugs to the emergence of resistance. In this review, we summarize the most recent literature regarding novel mutations reported between 2017 and 2022 and the advances in the molecular mechanisms of action and resistance against first-line drugs used in tuberculosis treatment, highlighting recent findings in pyrazinamide resistance involving PanD and, additionally, resistance-conferring mutations for novel drugs such as bedaquiline, pretomanid, delamanid and linezolid.

Large-scale genomic analysis of Mycobacterium tuberculosis reveals extent of target and compensatory mutations linked to multi-drug resistant tuberculosis

Resistance to isoniazid (INH) and rifampicin (RIF) first-line drugs in Mycobacterium tuberculosis (Mtb), together called multi-drug resistance, threatens tuberculosis control. Resistance mutations in katG (for INH) and rpoB (RIF) genes often come with fitness costs. To overcome these costs, Mtb compensatory mutations have arisen in rpoC/rpoA (RIF) and ahpC (INH) loci. By leveraging the presence of known compensatory mutations, we aimed to detect novel resistance mutations occurring in INH and RIF target genes. Across ~ 32 k Mtb isolates with whole genome sequencing (WGS) data, there were 6262 (35.7%) with INH and 5435 (30.7%) with RIF phenotypic resistance. Known mutations in katG and rpoB explained ~ 99% of resistance. However, 188 (0.6%) isolates had ahpC compensatory mutations with no known resistance mutations in katG, leading to the identification of 31 putative resistance mutations in katG, each observed in at least 3 isolates. These putative katG mutations can co-occur with other INH variants (e.g., katG-Ser315Thr, fabG1 mutations). For RIF, there were no isolates with rpoC/rpoA compensatory mutations and unknown resistance mutations. Overall, using WGS data we identified putative resistance markers for INH that could be used for genotypic drug-resistance profiling. Establishing the complete repertoire of Mtb resistance mutations will assist the clinical management of tuberculosis.

Genome-Wide Study of Drug Resistant Mycobacterium tuberculosis and Its Intra-Host Evolution during Treatment

The emergence of drug resistant Mycobacterium tuberculosis (MTB) strains has become a global public health problem, while, at the same time, there has been development of new antimicrobial agents. The main goals of this study were to determine new variants associated with drug resistance in MTB and to observe which polymorphisms emerge in MTB genomes after anti-tuberculosis treatment. We performed whole-genome sequencing of 152 MTB isolates including 70 isolates as 32 series of pre- and post-treatment MTB. Based on genotypes and phenotypic drug susceptibility, we conducted phylogenetic convergence-based genome-wide association study (GWAS) with streptomycin-, isoniazid-, rifampicin-, ethambutol-, fluoroquinolones-, and aminoglycosides-resistant MTB against susceptible ones. GWAS revealed statistically significant associations of SNPs within Rv2820c, cyp123 and indels in Rv1269c, Rv1907c, Rv1883c, Rv2407, Rv3785 genes with resistant MTB phenotypes. Comparisons of serial isolates showed that treatment induced different patterns of intra-host evolution. We found indels within Rv1435c and ppsA that were not lineage-specific. In addition, Beijing-specific polymorphisms within Rv0036c, Rv0678, Rv3433c, and dop genes were detected in post-treatment isolates. The appearance of Rv3785 frameshift insertion in 2 post-treatment strains compared to pre-treatment was also observed. We propose that the insertion within Rv3785, which was a GWAS hit, might affect cell wall biosynthesis and probably mediates a compensatory mechanism in response to treatment. These results may shed light on the mechanisms of MTB adaptation to chemotherapy and drug resistance formation.

First-line drug resistance profiling of Mycobacterium tuberculosis: a machine learning approach

The persistence and emergence of new multi-drug resistant Mycobacterium tuberculosis (M. tb) strains continues to advance the devastating tuberculosis (TB) epidemic. Robust systems are needed to accurately and rapidly perform drug-resistance profiling, and machine learning (ML) methods combined with genomic sequence data may provide novel insights into drug-resistance mechanisms. Using 372 M. tb isolates, the combined utility of ML and bioinformatics to perform drug-resistance profiling is demonstrated. SNPs, InDels, and dinucleotide frequencies are explored as input features for three ML models, namely Decision Trees, Random Forest, and the eXtreme Gradient Boosted model. Using SNPs and InDels, all three models performed equally well yielding a 99% accuracy, 97% recall, and 99% F1-score. Using dinucleotide frequencies, the XGBoost algorithm was superior with a 97% accuracy, 94% recall and 97% F1-score. This study validates the use of variants and presents dinucleotide features as another effective feature encoding method for ML-based phenotype classification.

Evaluation of TBMDR® and XDRA® for the detection of multidrug resistant and pre-extensively drug resistant tuberculosis

This study evaluated the diagnostic performance of the AccuPower® TB&MDR Real-Time PCR (TBMDR®) and AccuPower® XDR-TB Real-Time PCR Kit-A (XDRA®) to detect multidrug-resistant (MDR-TB) and pre-extensively drug-resistant tuberculosis (pre-XDR-TB) in comparison with phenotypic drug susceptibility testing (DST) using MGIT 960 on 234 clinical Mycobacterium tuberculosis isolates. Discrepant results were confirmed by direct-sequencing. Sensitivity and specificity of TBMDR and XDRA for cultured isolates were 81.2% and 95.8% for isoniazid (INH) resistance, 95.7% and 95.7% for rifampicin (RIF) resistance, 84.1% and 99.1% for fluoroquinolone (FQ) resistance, and 67.4% and 100% for second-line injectables resistance. The sensitivities of each drug were equivalent to other molecular DST methods. High concordance was observed when compared to direct-sequencing. We also found that TBMDR and XDRA assays can detect INH, RIF and FQ resistance in isolates with low level resistance-associated mutations which were missed by phenotypic DST. Our study showed TBMDR and XDRA assays could be the useful tools to detect MDR-TB and pre-XDR-TB.

Correlating genetic mutations with isoniazid phenotypic levels of resistance in Mycobacterium tuberculosis isolates from patients with drug-resistant tuberculosis in a high burden setting

We analysed mutations in katG, inhA and rpoB genes, and isoniazid phenotypic resistance levels in Mycobacterium tuberculosis isolates from drug-resistant TB patients from São Paulo state, Brazil. Isolates resistant to the critical concentration of isoniazid in MGIT (0.1 µg/mL) were screened for mutations in katG 315 codon, inhA promoter region and rpoB RRDR by MTBDRplus assay and subjected to determination of isoniazid resistance levels by MGIT 960. Discordances were resolved by Sanger sequencing. Among the 203 isolates studied, 109 (54%) were isoniazid-monoresistant, 47 (23%) MDR, 29 (14%) polydrug-resistant, 12 (6%) pre-XDR and 6 (3%) XDR. MTBDRplus detected isoniazid mutations in 75% (153/203) of the isolates. Sequencing of the entire katG and inhA genes revealed mutations in 18/50 wild-type isolates by MTBDRplus (10 with novel mutations), resulting in a total of 32/203 (16%) isolates with no mutations detected. 81/83 (98%) isolates with katG 315 mutations alone had intermediate resistance. Of the 66 isolates with inhA C-15T mutation alone, 51 (77%) showed low-level, 14 (21%) intermediate and 1 (2%) high-level resistance. 5/6 (83%) isolates with mutations in both katG and inhA had high-level resistance. Inferred mutations corresponded to 22% (16/73) of all mutations found in rpoB. Mutations detected in katG regions other than codon 315 in this study might be potential new isoniazid resistance markers and could explain phenotypic resistance in some isolates without katG and inhA classic mutations. In our setting, 16% of isoniazid-resistant isolates, some with high-level resistance, presented no mutations either in katG or inhA.

Value of routine whole genome sequencing for Mycobacterium tuberculosis drug resistance detection

Routine whole genome sequencing (WGS) of pathogens is becoming more feasible as sequencing costs decrease and access to benchtop sequencing equipment and bioinformatics pipelines increases. This study examined the added value gained from implementing routine WGS of all Mycobacterium tuberculosis isolates in New South Wales, Australia. Drug resistance markers inferred from WGS data were compared to commercial genotypic drug susceptibility testing (DST) assays and conventional phenotypic DST in all isolates sequenced between 2016 and 2019. Of the 1107 clinical M. tuberculosis isolates sequenced, 29 (2.6%) were multi-drug resistant (MDR); most belonged to Beijing (336; 30.4%) or East-African Indian (332; 30%) lineages. Compared with conventional phenotypic DST, WGS identified an additional 1% of isolates which were likely drug resistant, explained by mutations previously associated with treatment failure and mixed bacterial populations. However, WGS provided a 20% increase in drug resistance detection in comparison with commercial genotypic assays by identifying mutations outside of the classic resistance determining regions in rpoB, inhA, katG, pncA and embB genes. Gains in drug resistance detection were significant (p = 0.0137, paired t-test), but varied substantially for different phylogenetic lineages. In low incidence settings, routine WGS of M. tuberculosis provides better guidance for person-centered management of drug resistant tuberculosis than commercial genotypic assays.

Characterization of Large Deletion Mutants of Mycobacterium tuberculosis Selected for Isoniazid Resistance

Large genomic deletions (LGDs) (6 to 63 kbp) were observed in isoniazid-resistant Mycobacterium tuberculosis mutants derived from four M. tuberculosis strains. These LGDs had no growth defect in vitro but could be defective in intracellular growth and showed various sensitivities toward oxidative stress despite lacking katG The LGD regions comprise 74 genes, mostly of unknown function, that may be important for M. tuberculosis intracellular growth and protection against oxidative stress.

A retrospective cohort study of isoniazid-resistant tuberculosis treatment outcomes and isoniazid resistance-associated mutations in eastern China from 2013 to 2018

OBJECTIVES

The current situation of isoniazid-resistant, rifampicin-susceptible tuberculosis (Hr-TB) and associated genetic factors is not clear in China.

METHODS

A retrospective cohort study was conducted from 2013 to 2018 in Jiangsu Province, China. Phenotypic Hr-TB were identified by drug susceptibility testing on Lowenstein-Jensen medium and using a Mycobacterium Growth Indicator Tube 960 (MGIT 960) system, and mutations in the katG 315 codon and inhA promoter nucleotides -8, -15 and -16 were determined by GenoType MTBDRplus and sequencing. All of the Hr-TB patients enrolled were followed up until June 2019.

RESULTS

A total of 1416 smear-positive sputum samples were collected, of which 57 were excluded due to the presence of nontuberculous mycobacteria. Finally, 63/1359 (4.6%) were determined as Hr-TB. After follow-up, 11 Hr-TB patients (17.5%) showed an unfavourable outcome, of whom 5 (7.9%) relapsed, 4 (6.3%) had treatment failure and 2 (3.2%) died. A total of 52 isolates (82.5%) were detected with either katG 315 or inhA promoter nucleotide -8, -15 or -16 mutations, whereas no canonical mutations were found in 8 isolates (12.7%); 3 isolates failed in mutation detection. TB history was found to be associated with unfavourable outcomes for Hr-TB (odds ratio = 6.13, 95% confidence interval 1.05-35.82; P = 0.04). However, mutations in katG 315 and the inhA promoter region were not found to be associated with Hr-TB unfavourable outcomes (P = 0.15).

CONCLUSION

Unfavourable outcomes for Hr-TB are serious in eastern China, especially for previously treated patients. Meanwhile, current genetic determination of Hr-TB is inadequate.

Two Novel katG Mutations Conferring Isoniazid Resistance in Mycobacterium tuberculosis

Tuberculosis (TB), an infectious disease caused by Mycobacterium tuberculosis, is among the top 10 leading causes of death worldwide. The treatment course for TB is challenging; it requires antibiotic administration for at least 6 months, and bacterial drug resistance makes treatment even more difficult. Understanding the mechanisms of resistance is important for improving treatment. To investigate new mechanisms of isoniazid (INH) resistance, we obtained three INH-resistant (INH-R) M. tuberculosis clinical isolates collected by the Taiwan Centers for Disease Control (TCDC) and sequenced genes known to harbor INH resistance-conferring mutations. Then, the relationship between the mutations and INH resistance of these three INH-R isolates was investigated. Sequencing of the INH-R isolates identified three novel katG mutations resulting in R146P, W341R, and L398P KatG proteins, respectively. To investigate the correlation between the observed INH-R phenotypes of the clinical isolates and these katG mutations, wild-type katG from H37Rv was expressed on a plasmid (pMN437-katG) in the isolates, and their susceptibilities to INH were determined. The plasmid expressing H37Rv katG restored INH susceptibility in the two INH-R isolates encoding the W341R KatG and L398P KatG proteins. In contrast, no phenotypic change was observed in the KatG R146P isolate harboring pMN437-katG. H37Rv isogenic mutant with W341R KatG or L398P KatG was further generated. Both showed resistant to INH. In conclusion, W341R KatG and L398P KatG conferred resistance to INH in M. tuberculosis, whereas R146P KatG did not affect the INH susceptibility of M. tuberculosis.

Identifying isoniazid resistance markers to guide inclusion of high-dose isoniazid in tuberculosis treatment regimens

OBJECTIVES

Effective use of antibiotics is critical to control the global tuberculosis pandemic. High-dose isoniazid (INH) can be effective in the presence of low-level resistance. We performed a systematic literature review to improve our understanding of the differential impact of genomic Mycobacterium tuberculosis (Mtb) variants on the level of INH resistance. The following online databases were searched: PubMed, Web of Science and Embase. Articles reporting on clinical Mtb isolates with linked genotypic and phenotypic data and reporting INH resistance levels were eligible for inclusion.

METHODS

All genomic regions reported in the eligible studies were included in the analysis, including: katG, inhA, ahpC, oxyR-ahpC, furA, fabG1, kasA, rv1592c, iniA, iniB, iniC, rv0340, rv2242 and nat. The level of INH resistance was determined by MIC: low-level resistance was defined as 0.1-0.4 μg/mL on liquid and 0.2-1.0 μg/mL on solid media, high-level resistance as >0.4μg/mL on liquid and >1.0 μg/mL on solid media.

RESULTS

A total of 1212 records were retrieved of which 46 were included. These 46 studies reported 1697 isolates of which 21% (n = 362) were INH susceptible, 17% (n = 287) had low-level, and 62% (n = 1048) high-level INH resistance. Overall, 24% (n = 402) of isolates were reported as wild type and 76% (n = 1295) had ≥1 relevant genetic variant. Among 1295 isolates with ≥1 variant, 78% (n = 1011) had a mutation in the katG gene. Of the 867 isolates with a katG mutation in codon 315, 93% (n = 810) had high-level INH resistance. In contrast, only 50% (n = 72) of the 144 isolates with a katG variant not in the 315-position had high-level resistance. Of the 284 isolates with ≥1 relevant genetic variant and wild type katG gene, 40% (n = 114) had high-level INH resistance.

CONCLUSIONS

Presence of a variant in the katG gene is a good marker of high-level INH resistance only if located in codon 315.

Genome sequencing of Mycobacterium tuberculosis clinical isolates revealed isoniazid resistance mechanisms undetected by conventional molecular methods

A combination of targeted molecular methods and phenotypic drug-susceptibility testing is the most widely used approach to detect drug resistance in Mycobacterium tuberculosis isolates. We report the delay in the introduction of an efficient anti-tuberculous drug regimen because of a M. tuberculosis strain displaying a high level of resistance to isoniazid, in the absence of the common mutations associated with isoniazid-resistance, including katG mutations and inhA promoter mutations. Whole-genome sequencing (WGS) identified a large loss-of-function insertion (>1000 pb) at the end of katG in the isolate together with a -57C>T ahpC mutation, a resistance mechanism that would have remained undetected by a conventional molecular targeted approach. A retrospective search using publicly available WGS data of more than 1200 isoniazid-resistant isolates and a similar sized control dataset of isoniazid-susceptible isolates revealed that most (22/31) isoniazid-resistant, KatG loss-of-function mutants had an associated rare ahpC promoter mutation. In contrast, only 7 of 1411 isoniazid-susceptible strains carried a rare ahpC promoter mutation, including shared mutations with the 31 isoniazid-resistant KatG loss-of-function mutants. These results indicate that rare ahpC promoter mutations could be used as a proxy for investigating simultaneous KatG loss-of-function or missense mutations. In addition, WGS in routine diagnosis would improve drug susceptibility testing in M. tuberculosis clinical isolates and is an efficient tool for detecting resistance mechanisms undetected by conventional molecular methods.

Global Distribution and Evolution of Mycobacterium bovis Lineages

Mycobacterium bovis is the main causative agent of zoonotic tuberculosis in humans and frequently devastates livestock and wildlife worldwide. Previous studies suggested the existence of genetic groups of M. bovis strains based on limited DNA markers (a.k.a. clonal complexes), and the evolution and ecology of this pathogen has been only marginally explored at the global level. We have screened over 2,600 publicly available M. bovis genomes and newly sequenced four wildlife M. bovis strains, gathering 1,969 genomes from 23 countries and at least 24 host species, including humans, to complete a phylogenomic analyses. We propose the existence of four distinct global lineages of M. bovis (Lb1, Lb2, Lb3, and Lb4) underlying the current disease distribution. These lineages are not fully represented by clonal complexes and are dispersed based on geographic location rather than host species. Our data divergence analysis agreed with previous studies reporting independent archeological data of ancient M. bovis (South Siberian infected skeletons at ∼2,000 years before present) and indicates that extant M. bovis originated between 715 and 3,556 years BP, with later emergence in the New World and Oceania, likely influenced by trades among countries.

Genetic Diversity and Molecular Epidemiology of Mycobacterium tuberculosis in Roraima State, Brazil

National border areas are special places for the spread of Mycobacterium tuberculosis (MTB). These regions concentrate vulnerable populations and constant population movements. Understanding the dynamics of the transmission of MTB is fundamental to propose control measures and to monitor drug resistance. We conducted a population-based prospective study of tuberculosis (TB) to evaluate molecular characteristics of MTB isolates circulating in Roraima, a state on the border of Venezuela and Guyana. Eighty isolates were genotyped by IS6110-RFLP (restriction fragment length polymorphism), spoligotyping, and 24-locus mycobacterial interspersed repetitive unit-variable number of repeats tandem (MIRU-VNTR). Drug susceptibility tests were performed by using the proportion method and GeneXpert^® MTB/RIF (Cepheid, Sunnyvale, CA). Isolates showing a phenotypic resistance profile were submitted to polymerase chain reaction (PCR) and sequencing. Spoligotyping showed 40 distinct patterns with a high prevalence of Latin-American and Mediterranean (LAM), Haarlem (H), and the "ill-defined" T clades. Mycobacterial interspersed repetitive unit -VNTR and IS6110-RFLP showed clustering rates of 21.3% and 30%, respectively. Drug resistance was detected in 11 (15.1%) isolates, and all were found to have primary resistance; among these, six (8.2%) isolates were streptomycin mono-resistant, four (5.4%) isoniazid mono-resistant, and one (1.3%) multidrug resistant. This is the first study on the molecular epidemiology and drug resistance profile of MTB from Roraima. Herein, we describe high diversity of genetic profiles circulating in this region that may be driven by the introduction of new strain types because of large population flow in this region. In summary, our results showed that analyses of these circulating strains can contribute to a better understanding of TB epidemiology in the northern Brazilian border and be useful to establish public health policies on TB prevention.

Association Between the Phenotype and Genotype of Isoniazid Resistance Among Mycobacterium tuberculosis Isolates in Thailand

Purpose

The emergence of isoniazid-resistant tuberculosis (HR-TB) is a global public health problem, causing treatment failure and high mortality rates. This study aimed to determine the minimal inhibitory concentration (MIC) of isoniazid and detect the gene mutation in HR-TB and any association between the level of isoniazid resistance and gene mutation.

Methods

We collected 74 clinical HR-TB isolates from two tertiary-care centers in Thailand. MICs were established using broth macrodilution. A line probe assay (LPA) was used to detect gene mutations that confer resistance to isoniazid, rifampicin, aminoglycosides, and fluoroquinolones.

Results

Sixty-one (82.4%) isolates were monoresistant to isoniazid and 44 (72.1%) were highly resistant to isoniazid. From the clinical isolates, the range of isoniazid MICs was 0.4–16 μg/mL. The katG S315T gene mutation was the prominent mutation in both isoniazid-monoresistant TB (70.5%) and multidrug-resistant TB (72.7%) isolates. The positive predictive value (PPV) of katG was 100% in detecting high levels of isoniazid resistance. The PPV of the inhA mutation was 93.8% in detecting low levels of isoniazid resistance. Five isolates (6.8%) exhibited low-level phenotypic resistance, whereas an LPA failed to detect an isoniazid gene mutation. Our study found one HR-TB isolate with a gyrA fluoroquinolone-resistant gene mutation.

Conclusion

Most HR-TB isolates had high isoniazid-resistance levels associated with the katG gene mutation. High-dose isoniazid should be used with caution in patients with HR-TB. Early detection of drug resistance by genotypic assay can help determine an appropriate regimen.

Utility of Targeted, Amplicon-Based Deep Sequencing To Detect Resistance to First-Line Tuberculosis Drugs in Botswana

Multidrug-resistant tuberculosis (TB) is an alarming threat, and targeted deep sequencing (DS) may be an effective method for rapid identification of drug-resistant profiles, including detection of heteroresistance. We evaluated the sensitivity and specificity of targeted DS versus phenotypic drug susceptibility testing (pDST) among patients starting first-line anti-TB therapy in Botswana. Overall, we found high concordance between DS and pDST. Lower sensitivity of DS, which targets established high-confidence resistance variants, was observed for detecting isoniazid resistance among HIV-infected patients.

Application of machine learning techniques to tuberculosis drug resistance analysis

MOTIVATION

Timely identification of Mycobacterium tuberculosis (MTB) resistance to existing drugs is vital to decrease mortality and prevent the amplification of existing antibiotic resistance. Machine learning methods have been widely applied for timely predicting resistance of MTB given a specific drug and identifying resistance markers. However, they have been not validated on a large cohort of MTB samples from multi-centers across the world in terms of resistance prediction and resistance marker identification. Several machine learning classifiers and linear dimension reduction techniques were developed and compared for a cohort of 13 402 isolates collected from 16 countries across 6 continents and tested 11 drugs.

RESULTS

Compared to conventional molecular diagnostic test, area under curve of the best machine learning classifier increased for all drugs especially by 23.11%, 15.22% and 10.14% for pyrazinamide, ciprofloxacin and ofloxacin, respectively (P < 0.01). Logistic regression and gradient tree boosting found to perform better than other techniques. Moreover, logistic regression/gradient tree boosting with a sparse principal component analysis/non-negative matrix factorization step compared with the classifier alone enhanced the best performance in terms of F1-score by 12.54%, 4.61%, 7.45% and 9.58% for amikacin, moxifloxacin, ofloxacin and capreomycin, respectively, as well increasing area under curve for amikacin and capreomycin. Results provided a comprehensive comparison of various techniques and confirmed the application of machine learning for better prediction of the large diverse tuberculosis data. Furthermore, mutation ranking showed the possibility of finding new resistance/susceptible markers.

AVAILABILITY AND IMPLEMENTATION

The source code can be found at http://www.robots.ox.ac.uk/ davidc/code.php.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Overcoming the pitfalls of automatic interpretation of whole genome sequencing data by online tools for the prediction of pyrazinamide resistance in Mycobacterium tuberculosis

OBJECTIVES

Automated online software tools that analyse whole genome sequencing (WGS) data without the need for bioinformatics expertise can motivate the implementation of WGS-based molecular drug susceptibility testing (DST) in routine diagnostic settings for tuberculosis (TB). Pyrazinamide (PZA) is a key drug for current and future TB treatment regimens; however, it was reported that predictive power for PZA resistance by the available tools is low. Therefore, this low predictive power may make users hesitant to use the tools. This study aimed to elucidate why and to uncover the real performance of the tools when taking into account their variation calling lists (manual inspection), not just their automated reporting system (default setting) that was evaluated by previous studies.

METHODS

WGS data from 191 datasets comprising 108 PZA-resistant and 83 susceptible strains were used to evaluate the potential performance of the available online tools (TB Profiler, TGS-TB, PhyResSE, and CASTB) for predicting phenotypic PZA resistance.

RESULTS

When taking into consideration the variation calling lists, 73 variants in total (47 non-synonymous mutations and 26 indels) in pncA were detected by TGS-TB and PhyResSE, covering all mutations for the 108 PZA-resistant strains. The 73 variants were confirmed by Sanger sequencing. TB Profiler also detected all but three complete loss, two large deletion at the 3'-end, and one relatively large insertion of pncA. On the other hand, many of the 73 variants were lacking in the automated reporting systems except by TGS-TB; of these variants, CASTB detected only 20. By applying the 'non-wild type sequence' approach for predicting PZA resistance, accuracy of the results significantly improved compared with that of the automated results obtained by each tool.

CONCLUSION

Users can obtain more accurate predictions for PZA resistance than previously reported by manually checking the results and applying the 'non-wild type sequence' approach.

Redefining MDR-TB: Comparison of Mycobacterium tuberculosis clinical isolates from Russia and Taiwan

Multidrug-resistant (MDR) and extensively drug-resistant (XDR) tuberculosis are global challenges due to the limited number of effective drugs for treatment. Treatment with less than 4-5 effective drugs might lead to the further emergence of drug resistance and poor clinical outcomes. For better prediction of treatment outcomes, we compared drug-resistance profiles of consecutive clinical MDR Mycobacterium tuberculosis isolates from high- and low-burden settings. This was a retrospective cohort study. We analysed 225 and 229 MDR isolates from Moscow (Russia) and Taiwan, respectively, obtained between 2014 and 2015. Drug susceptibility testing was performed by the Bactec MGIT 960 automated system and the agar proportion method. Detection of resistance-associated mutations in the M. tuberculosis genome was carried out by an array and/or sequencing of selected loci. The principal differences between resistance profiles of MDR isolates in the two countries were the percentages of pre-XDR (40.9% vs. 14.8%) and XDR (34.7% vs. 1.7%) isolates, both of which were significantly higher in Moscow isolates. Forty-eight (33%) of 147 MDR and pre-XDR Russian isolates fall into a group with less than four effective drugs, which accounts for 40% (N = 120) of these isolates. The other 60% in this group were XDR strains (N = 72). Consequently, the average number of effective anti-tuberculosis drugs for MDR-TB treatment was lower for Russian isolates (3 vs. 7). Furthermore, a notable percentage (9%) of isolates resistant to kanamycin harboured mutations in the whiB7 locus, which was not detected by molecular tests targeting common mutations in the rrs and eis loci. We found that 98.2% and 45.9% of MDR isolates from Moscow and Taiwan, respectively, were resistant to streptomycin. Molecular tests for detecting resistance to drugs other than rifampicin, isoniazid, fluoroquinolones, and second-line injectable drugs are needed for individualized therapy. The conventional MDR treatment schemes most probably fail in these cases due to the limited number of effective drugs.