Project no.: HIV-NAT 188/IeDEA TB genomics
This study will investigate the macroevolution of drug-resistant M. tuberculosis.
Brief Summary: The gold standard for drug susceptibility testing (DST) in Mycobacterium tuberculosis involves time-consuming assessments of bacterial growth at defined drug concentrations. Whole genome sequencing allows rapid detection of drug-resistant M. tuberculosis isolates. However, high-quality data linking quantitative phenotypic DST and genomic data has thus far been lacking.
We determined drug resistance profiles of 177 genetically diverse clinical M. tuberculosis isolates from Democratic Republic of the Congo, Ivory Coast, Peru, Thailand, South Africa and Switzerland by semi-quantitative and quantitative phenotypic DST for 11 antituberculous drugs using the BD BACTEC MGIT 960 system and 7H11 agar dilution. We compared phenotypic drug susceptibility results with predicted drug resistance profiles inferred by whole genome sequencing.
Both phenotypic DST methods identically classified the strains into resistant/susceptible in 73-99% of the cases, depending on the drug. Increased minimal inhibitory concentrations (MICs) could be explained by mutations identified by whole genome sequencing. The overall sensitivity and specificity of predicting drug resistance based on whole genome sequences were 87.5% and 82.1%, respectively. Wild type and mutant MIC distributions overlapped partially due to mutational combinations, but this had no clinical implications.
Whole genome sequencing has high predictive power to infer resistance profiles without the need for time-consuming phenotypic methods.
Results: We included 738 TB patients; median age was 34 years (interquartile range [IQR]: 27-43); 269 (36%) female; 304 (41%) HIV-positive (median CD4 count at TB diagnosis: 192 cells/µl; IQR: 80-369) and 175 (57%) were on ART at TB treatment start. The NRC identified 460 (62%) pansusceptible TB cases, 43 (6%) mono-resistance, 205 (28%) multidrug resistance (MDR), and 30 (4%) pre-extensive drug resistance (pre-XDR) or XDR cases. Local clinics diagnosed drug resistances using Xpert MTB/RIF (276 cases, 37%), culture (259, 35%), line-probe assay (LPA, 15, 2%), a combination of the above (99, 13%), and unknown method (89, 12%). In 590 cases (80%), NRC confirmed drug resistance profiles obtained at clinics. Among the 148 (20%) discrepant profiles, 56 (38%) were initially tested with Xpert MTB/RIF, 60 (40%) with culture, 4 (3%) with LPA, and 28 (19%) with combined tests. Among cases with discrepant resistance profiles, 10 (7%) were under-treated, 21 (14%) were over-treated, and the remaining 117 (79%) cases were appropriately treated. Mortality was loweramong patients whose resistance profiles were concordant or discrepant and leading to over-treatment, compared to patients with discrepant results leading to under-treatment: 39 (7%), 4 (6%), and 15 (19%) deaths, respectively (p<0.001). There was no evidence for an association between HIV status and accuracy of drug resistance profiles.
634 tuberculosis patients were included; median age was 33.2 years, 239 (37.7%) were female, 272 (42.9%) HIV-positive and 69 (10.9%) patients died. Based on the reference laboratory DST, 394 (62.2%) strains were pan-susceptible, 45 (7.1%) mono-resistant, 163 (25.7%) multidrug-resistant (MDR-TB), and 30 (4.7%) had pre-extensive or extensive drug resistance (pre-XDR/XDR-TB). Results of reference and local laboratories were discordant in 121 (19.1%) cases, corresponding to a sensitivity of 84.3% and a specificity of 90.8%. In patients with drug-resistant tuberculosis, discordant results were associated with increased mortality (risk ratio 1.81; 95% CI 1.07-3.07). In logistic regression, compared to adequately treated patients with pan-susceptible strains, the adjusted odds ratio for death was 4.23 (95% CI 2.16-8.29) for adequately treated patients with drug-resistant strains and 21.54 (95% CI 3.36-138.1) for inadequately treated patients with drug-resistant strains. HIV status was not associated with mortality.
