Search

Objective: To retrospectively investigate the cause of recurring tuberculosis (rcTB) among participants with pulmonary TB recruited from a prospective population-based study conducted between July 2012 and December 2015. Methods: Mycobacterium tuberculosis complex isolates obtained from rcTB cases were characterized by standard mycobacterial genotyping tools, whole-genome sequencing, and phylogenetic analysis carried out to assess strain relatedness. Results: The majority (58.3%, 21/36) of study participants with rcTB episodes had TB recurrence within 12 months post treatment. TB strains with isoniazid (INH) resistance were found in 19.4% (7/36) of participants at the primary episode, of which 29% (2/7) were also rifampicin-resistant. On TB recurrence, an INH-resistant strain was found in a larger proportion of participants, 27.8% (10/36), of which 40% (4/10) were MDR-TB strains. rcTB was attributed to relapse (same strain) in 75.0% (27/36) of participants and 25.0% (9/36) to re-infection. Conclusion: Our findings indicate that previous unresolved infectiondue to inadequate treatment, may be the major cause of rcTB.

Summary: Background: Molecular diagnostics are considered the most promising route to achievement of rapid, universal drug susceptibility testing for Mycobacterium tuberculosis complex (MTBC). We aimed to generate a WHO-endorsed catalogue of mutations to serve as a global standard for interpreting molecular information for drug resistance prediction. Methods: In this systematic analysis, we used a candidate gene approach to identify mutations associated with resistance or consistent with susceptibility for 13 WHO-endorsed antituberculosis drugs. We collected existing worldwide MTBC whole-genome sequencing data and phenotypic data from academic groups and consortia, reference laboratories, public health organisations, and published literature. We categorised phenotypes as follows: methods and critical concentrations currently endorsed by WHO (category 1); critical concentrations previously endorsed by WHO for those methods (category 2); methods or critical concentrations not currently endorsed by WHO (category 3). For each mutation, we used a contingency table of binary phenotypes and presence or absence of the mutation to compute positive predictive value, and we used Fisher's exact tests to generate odds ratios and Benjamini-Hochberg corrected p values. Mutations were graded as associated with resistance if present in at least five isolates, if the odds ratio was more than 1 with a statistically significant corrected p value, and if the lower bound of the 95% CI on the positive predictive value for phenotypic resistance was greater than 25%. A series of expert rules were applied for final confidence grading of each mutation. Findings: We analysed 41 137 MTBC isolates with phenotypic and whole-genome sequencing data from 45 countries. 38 215 MTBC isolates passed quality control steps and were included in the final analysis. 15 667 associations were computed for 13 211 unique mutations linked to one or more drugs. 1149 (7·3%) of 15 667 mutations were classified as associated with phenotypic resistance and 107 (0·7%) were deemed consistent with susceptibility. For rifampicin, isoniazid, ethambutol, fluoroquinolones, and streptomycin, the mutations' pooled sensitivity was more than 80%. Specificity was over 95% for all drugs except ethionamide (91·4%), moxifloxacin (91·6%) and ethambutol (93·3%). Only two resistance mutations were identified for bedaquiline, delamanid, clofazimine, and linezolid as prevalence of phenotypic resistance was low for these drugs. Interpretation: We present the first WHO-endorsed catalogue of molecular targets for MTBC drug susceptibility testing, which is intended to provide a global standard for resistance interpretation. The existence of this catalogue should encourage the implementation of molecular diagnostics by national tuberculosis programmes. Funding: Unitaid, Wellcome Trust, UK Medical Research Council, and Bill and Melinda Gates Foundation.

Summary: Background: South Africa has a high burden of rifampicin-resistant tuberculosis (including multidrug-resistant [MDR] tuberculosis), with increasing rifampicin-monoresistant (RMR) tuberculosis over time. Resistance acquisition during first-line tuberculosis treatment could be a key contributor to this burden, and HIV might increase the risk of acquiring rifampicin resistance. We assessed whether HIV during previous treatment was associated with RMR tuberculosis and resistance acquisition among a retrospective cohort of patients with MDR or rifampicin-resistant tuberculosis. Methods: In this retrospective cohort study, we included all patients routinely diagnosed with MDR or rifampicin-resistant tuberculosis in Khayelitsha, Cape Town, South Africa, between Jan 1, 2008, and Dec 31, 2017. Patient-level data were obtained from a prospective database, complemented by data on previous tuberculosis treatment and HIV from a provincial health data exchange. Stored MDR or rifampicin-resistant tuberculosis isolates from patients underwent whole-genome sequencing (WGS). WGS data were used to infer resistance acquisition versus transmission, by identifying genomically unique isolates (single nucleotide polymorphism threshold of five). Logistic regression analyses were used to assess factors associated with RMR tuberculosis and genomic uniqueness. Findings: The cohort included 2041 patients diagnosed with MDR or rifampicin-resistant tuberculosis between Jan 1, 2008, and Dec 31, 2017; of those, 463 (22·7%) with RMR tuberculosis and 1354 (66·3%) with previous tuberculosis treatment. In previously treated patients, HIV positivity during previous tuberculosis treatment versus HIV negativity (adjusted odds ratio [OR] 2·07, 95% CI 1·35–3·18), and three or more previous tuberculosis treatment episodes versus one (1·96, 1·21–3·17) were associated with RMR tuberculosis. WGS data showing MDR or rifampicin-resistant tuberculosis were available for 1169 patients; 360 (30·8%) isolates were identified as unique. In previously treated patients, RMR tuberculosis versus MDR tuberculosis (adjusted OR 4·96, 3·40–7·23), HIV positivity during previous tuberculosis treatment (1·71, 1·03–2·84), and diagnosis in 2013–17 (1·42, 1·02–1·99) versus 2008–12, were associated with uniqueness. In previously treated patients with RMR tuberculosis, HIV positivity during previous treatment (adjusted OR 5·13, 1·61–16·32) was associated with uniqueness as was female sex (2·50 [1·18–5·26]). Interpretation: These data suggest that HIV contributes to rifampicin-resistance acquisition during first-line tuberculosis treatment and that this might be driving increasing RMR tuberculosis over time. Large-scale prospective cohort studies are required to further quantify this risk. Funding: Swiss National Science Foundation, South African National Research Foundation, and Wellcome Trust.

As multi-drug resistant tuberculosis (MDR-TB) continues to spread, investigating the transmission potential of different drug-resistant strains becomes an ever more pressing topic in public health. While phylogenetic and transmission tree inferences provide valuable insight into possible transmission chains, phylodynamic inference combines evolutionary and epidemiological analyses to estimate the parameters of the underlying epidemiological processes, allowing us to describe the overall dynamics of disease spread in the population. In this study, we introduce an approach to Mycobacterium tuberculosis (M. tuberculosis) phylodynamic analysis employing an existing computationally efficient model to quantify the transmission fitness costs of drug resistance with respect to drug-sensitive strains. To determine the accuracy and precision of our approach, we first perform a simulation study, mimicking the simultaneous spread of drug-sensitive and drug-resistant tuberculosis (TB) strains. We analyse the simulated transmission trees using the phylodynamic multi-type birth–death model (MTBD, (Kühnert et al., 2016)) within the BEAST2 framework and show that this model can estimate the parameters of the epidemic well, despite the simplifying assumptions that MTBD makes compared to the complex TB transmission dynamics used for simulation. We then apply the MTBD model to an M. tuberculosis lineage 4 dataset that primarily consists of MDR sequences. Some of the MDR strains additionally exhibit resistance to pyrazinamide — an important first-line anti-tuberculosis drug. Our results support the previously proposed hypothesis that pyrazinamide resistance confers a transmission fitness cost to the bacterium, which we quantify for the given dataset. Importantly, our sensitivity analyses show that the estimates are robust to different prior distributions on the resistance acquisition rate, but are affected by the size of the dataset – i.e. we estimate a higher fitness cost when using fewer sequences for analysis. Overall, we propose that MTBD can be used to quantify the transmission fitness cost for a wide range of pathogens where the strains can be appropriately divided into two or more categories with distinct properties.

Summary: Background: Drug resistance threatens global tuberculosis control. We aimed to examine mortality in patients with tuberculosis from high-burden countries, according to concordance or discordance of results from drug susceptibility testing done locally and whole-genome sequencing (WGS). Methods: In this multicentre cohort study, we collected pulmonary Mycobacterium tuberculosis isolates and clinical data from individuals with tuberculosis from antiretroviral therapy programmes and tuberculosis clinics in Côte d'Ivoire, Democratic Republic of the Congo, Kenya, Nigeria, Peru, South Africa, and Thailand, stratified by HIV status and drug resistance. Sites tested drug susceptibility using routinely available methods. WGS was done on Illumina HiSeq 2500 in the USA and Switzerland, and TBprofiler was used to analyse the genomes. We included individuals aged 16 years or older with pulmonary tuberculosis (bacteriologically confirmed or clinically diagnosed). We analysed mortality in multivariable logistic regression models adjusted for sex, age, HIV status, history of tuberculosis, and sputum positivity. Findings: Between Sept 1, 2014, and July 4, 2016, of 634 patients included in our previous analysis, we included 582 patients with tuberculosis (median age 33 years [IQR 27–43], 225 [39%] women, and 247 [42%] HIV-positive). Based on WGS, 339 (58%) isolates were pan-susceptible, 35 (6%) monoresistant, 146 (25%) multidrug-resistant, and 24 (4%) pre-extensively drug-resistant (pre-XDR) or XDR. The analysis of mortality was based on 530 patients; 63 (12%) died and 77 (15%) patients received inappropriate treatment. Mortality ranged from 6% (18 of 310) in patients with pan-susceptible tuberculosis to 39% (nine of 23) in patients with pre-XDR or XDR tuberculosis. The adjusted odds ratio for mortality was 4·92 (95% CI 2·47–9·78) among undertreated patients, compared with appropriately treated patients. Interpretation: In seven countries with a high burden of tuberculosis, we observed discrepancies between drug resistance patterns obtained locally and WGS. The underdiagnosis of drug resistance resulted in inappropriate treatment and higher mortality. WGS can provide accurate and detailed drug resistance information required to improve the outcomes of drug-resistant tuberculosis in high-burden settings. Our results support WHO's call for point-of-care tests based on WGS. Funding: National Institutes of Allergy and Infectious Diseases, Swiss National Science Foundation, and Swiss National Center for Mycobacteria.

The fast evolution of pathogenic viruses has allowed for the development of phylodynamic approaches that extract information about the epidemiological characteristics of viral genomes. Thanks to advances in whole genome sequencing, they can be applied to slowly evolving bacterial pathogens like Mycobacterium tuberculosis. In this study, we investigate and compare the epidemiological dynamics underlying two M. tuberculosis outbreaks using phylodynamic methods. Specifically, we (i) test if the outbreak data sets contain enough genetic variation to estimate short-term evolutionary rates and (ii) reconstruct epidemiological parameters such as the effective reproduction number.The first outbreak occurred in the Swiss city of Bern (1987–2012) and was caused by a drug-susceptible strain belonging to the phylogenetic M. tuberculosis Lineage 4. The second outbreak was caused by a multidrug-resistant (MDR) strain of Lineage 2, imported from the Wat Tham Krabok (WTK) refugee camp in Thailand into California.There is little temporal signal in the Bern data set and moderate temporal signal in the WTK data set. Thanks to its high sampling proportion (90%) the Bern outbreak allows robust estimation of epidemiological parameters despite the poor temporal signal. Conversely, there is much uncertainty in the epidemiological estimates concerning the sparsely sampled (9%) WTK outbreak. Our results suggest that both outbreaks peaked around 1990, although they were only recognized as outbreaks in 1993 (Bern) and 2004 (WTK). Furthermore, individuals were infected for a significantly longer period (around 9 years) in the WTK outbreak than in the Bern outbreak (4–5 years).Our work highlights both the limitations and opportunities of phylodynamic analysis of outbreaks involving slowly evolving pathogens: (i) estimation of the evolutionary rate is difficult on outbreak time scales and (ii) a high sampling proportion allows quantification of the age of the outbreak based on the sampling times, and thus allows for robust estimation of epidemiological parameters. Keywords: Phylodynamic analysis, Tuberculosis outbreak, Epidemic dynamics, Transmission dynamics