Diagnosing drug-resistance remains an obstacle to the elimination of tuberculosis. Phenotypic drug-susceptibility testing is slow and expensive, and commercial genotypic assays screen only common resistance-determining mutations. We used whole-genome sequencing to characterise common and rare mutations predicting drug resistance, or consistency with susceptibility, for all first-line and second-line drugs for tuberculosis. Between Sept 1, 2010, and Dec 1, 2013, we sequenced a training set of 2099 Mycobacterium tuberculosis genomes. For 23 candidate genes identified from the drug-resistance scientific literature, we algorithmically characterised genetic mutations as not conferring resistance (benign), resistance determinants, or uncharacterised. We then assessed the ability of these characterisations to predict phenotypic drug-susceptibility testing for an independent validation set of 1552 genomes. We sought mutations under similar selection pressure to those characterised as resistance determinants outside candidate genes to account for residual phenotypic resistance. We characterised 120 training-set mutations as resistance determining, and 772 as benign. With these mutations, we could predict 89·2% of the validation-set phenotypes with a mean 92·3% sensitivity (95% CI 90·7-93·7) and 98·4% specificity (98·1-98·7). 10·8% of validation-set phenotypes could not be predicted because uncharacterised mutations were present. With an in-silico comparison, characterised resistance determinants had higher sensitivity than the mutations from three line-probe assays (85·1% vs 81·6%). No additional resistance determinants were identified among mutations under selection pressure in non-candidate genes. A broad catalogue of genetic mutations enable data from whole-genome sequencing to be used clinically to predict drug resistance, drug susceptibility, or to identify drug phenotypes that cannot yet be genetically predicted. This approach could be integrated into routine diagnostic workflows, phasing out phenotypic drug-susceptibility testing while reporting drug resistance early. Wellcome Trust, National Institute of Health Research, Medical Research Council, and the European Union.

The rise of antibiotic-resistant bacteria has led to an urgent need for rapid detection of drug resistance in clinical samples, and improvements in global surveillance. Here we show how de Bruijn graph representation of bacterial diversity can be used to identify species and resistance profiles of clinical isolates. We implement this method for Staphylococcus aureus and Mycobacterium tuberculosis in a software package (‘Mykrobe predictor’) that takes raw sequence data as input, and generates a clinician-friendly report within 3 minutes on a laptop. For S. aureus, the error rates of our method are comparable to gold-standard phenotypic methods, with sensitivity/specificity of 99.1%/99.6% across 12 antibiotics (using an independent validation set, n=470). For M. tuberculosis, our method predicts resistance with sensitivity/specificity of 82.6%/98.5% (independent validation set, n=1,609); sensitivity is lower here, probably because of limited understanding of the underlying genetic mechanisms. We give evidence that minor alleles improve detection of extremely drug-resistant strains, and demonstrate feasibility of the use of emerging single-molecule nanopore sequencing techniques for these purposes.

Drug-resistant tuberculosis threatens recent gains in the treatment of tuberculosis and human immunodeficiency virus (HIV) infection worldwide. A widespread epidemic of extensively drug-resistant (XDR) tuberculosis is occurring in South Africa, where cases have increased substantially since 2002. The factors driving this rapid increase have not been fully elucidated, but such knowledge is needed to guide public health interventions. We conducted a prospective study involving 404 participants in KwaZulu-Natal Province, South Africa, with a diagnosis of XDR tuberculosis between 2011 and 2014. Interviews and medical-record reviews were used to elicit information on the participants’ history of tuberculosis and HIV infection, hospitalizations, and social networks. Mycobacterium tuberculosis isolates underwent insertion sequence (IS)6110 restriction-fragment–length polymorphism analysis, targeted gene sequencing, and whole-genome sequencing. We used clinical and genotypic case definitions to calculate the proportion of cases of XDR tuberculosis that were due to inadequate treatment of multidrug-resistant (MDR) tuberculosis (i.e., acquired resistance) versus those that were due to transmission (i.e., transmitted resistance). We used social-network analysis to identify community and hospital locations of transmission. Of the 404 participants, 311 (77%) had HIV infection; the median CD4+ count was 340 cells per cubic millimeter (interquartile range, 117 to 431). A total of 280 participants (69%) had never received treatment for MDR tuberculosis. Genotypic analysis in 386 participants revealed that 323 (84%) belonged to 1 of 31 clusters. Clusters ranged from 2 to 14 participants, except for 1 large cluster of 212 participants (55%) with a LAM4/KZN strain. Person-to-person or hospital-based epidemiologic links were identified in 123 of 404 participants (30%). The majority of cases of XDR tuberculosis in KwaZulu-Natal, South Africa, an area with a high tuberculosis burden, were probably due to transmission rather than to inadequate treatment of MDR tuberculosis. These data suggest that control of the epidemic of drug-resistant tuberculosis requires an increased focus on interrupting transmission. (Funded by the National Institute of Allergy and Infectious Diseases and others.)

Pyrazinamide and fluoroquinolones are essential antituberculosis drugs in new rifampicin-sparing regimens. However, little information about the extent of resistance to these drugs at the population level is available. In a molecular epidemiology analysis, we used population-based surveys from Azerbaijan, Bangladesh, Belarus, Pakistan, and South Africa to investigate resistance to pyrazinamide and fluoroquinolones among patients with tuberculosis. Resistance to pyrazinamide was assessed by gene sequencing with the detection of resistance-conferring mutations in the pncA gene, and susceptibility testing to fluoroquinolones was conducted using the MGIT system. Pyrazinamide resistance was assessed in 4972 patients. Levels of resistance varied substantially in the surveyed settings (3·0-42·1%). In all settings, pyrazinamide resistance was significantly associated with rifampicin resistance. Among 5015 patients who underwent susceptibility testing to fluoroquinolones, proportions of resistance ranged from 1·0-16·6% for ofloxacin, to 0·5-12·4% for levofloxacin, and 0·9-14·6% for moxifloxacin when tested at 0·5 μg/mL. High levels of ofloxacin resistance were detected in Pakistan. Resistance to moxifloxacin and gatifloxacin when tested at 2 μg/mL was low in all countries. Although pyrazinamide resistance was significantly associated with rifampicin resistance, this drug may still be effective in 19-63% of patients with rifampicin-resistant tuberculosis. Even though the high level of resistance to ofloxacin found in Pakistan is worrisome because it might be the expression of extensive and unregulated use of fluoroquinolones in some parts of Asia, the negligible levels of resistance to fourth-generation fluoroquinolones documented in all survey sites is an encouraging finding. Rational use of this class of antibiotics should therefore be ensured to preserve its effectiveness. Bill & Melinda Gates Foundation, United States Agency for International Development, Global Alliance for Tuberculosis Drug Development.

In many countries, regular monitoring of the emergence of resistance to anti-tuberculosis drugs is hampered by the limitations of phenotypic testing for drug susceptibility. We therefore evaluated the use of genetic sequencing for surveillance of drug resistance in tuberculosis. Population-level surveys were done in hospitals and clinics in seven countries (Azerbaijan, Bangladesh, Belarus, Pakistan, Philippines, South Africa, and Ukraine) to evaluate the use of genetic sequencing to estimate the resistance of Mycobacterium tuberculosis isolates to rifampicin, isoniazid, ofloxacin, moxifloxacin, pyrazinamide, kanamycin, amikacin, and capreomycin. For each drug, we assessed the accuracy of genetic sequencing by a comparison of the adjusted prevalence of resistance, measured by genetic sequencing, with the true prevalence of resistance, determined by phenotypic testing. Isolates were taken from 7094 patients with tuberculosis who were enrolled in the study between November, 2009, and May, 2014. In all tuberculosis cases, the overall pooled sensitivity values for predicting resistance by genetic sequencing were 91% (95% CI 87-94) for rpoB (rifampicin resistance), 86% (74-93) for katG, inhA, and fabG promoter combined (isoniazid resistance), 54% (39-68) for pncA (pyrazinamide resistance), 85% (77-91) for gyrA and gyrB combined (ofloxacin resistance), and 88% (81-92) for gyrA and gyrB combined (moxifloxacin resistance). For nearly all drugs and in most settings, there was a large overlap in the estimated prevalence of drug resistance by genetic sequencing and the estimated prevalence by phenotypic testing. Genetic sequencing can be a valuable tool for surveillance of drug resistance, providing new opportunities to monitor drug resistance in tuberculosis in resource-poor countries. Before its widespread adoption for surveillance purposes, there is a need to standardise DNA extraction methods, recording and reporting nomenclature, and data interpretation. Bill & Melinda Gates Foundation, United States Agency for International Development, Global Alliance for Tuberculosis Drug Development.

Bedaquiline (BDQ) is a novel agent approved for use in combination treatment of multi-drug resistant tuberculosis (MDR-TB). We sought to determine BDQ epidemiological cut-off values (ECVs), define and assess interpretive criteria against putative resistance associated variants (RAVs), microbiological outcomes and cross resistance with clofazimine (CFZ). A retrospective cohort study was conducted. Minimal inhibitory concentrations (MIC) to BDQ were determined using 7H9 broth microdilution (BMD) and MGIT960. RAVs were genetically characterised using whole genome sequencing. BDQ ECVs were determined using ECOFFinder and compared with 6-month culture conversion status and CFZ MICs. A total of 391 isolates were analysed. Susceptible and intermediate categories were determined to have MICs of ≤0.125μg/ml and 0.25μg/ml using BMD and ≤1μg/ml and 2μg/ml using MGIT960 respectively. Microbiological failures occurred among BDQ exposed patients with a non-susceptible BDQ MIC, an Rv0678 mutation and ≤2 active drug classes. The Rv0678 RAVs were not the dominant mechanism of CFZ resistance and cross resistance was limited to isolates with an Rv0678 mutation. Criteria for BDQ susceptibility are defined and will facilitate improved early detection of resistance. Cross- resistance between BDQ and CFZ is an emerging concern but in this study was primarily among those with an Rv0678 mutation.