Background: Rapid molecular diagnostics, with their ability to quickly identify genetic mutations associated with drug resistance, have great potential to curb the spread of drug-resistant tuberculosis (DR-TB). Unfortunately, the technical and diagnostic performance of these technologies has yet to be thoroughly evaluated in diverse clinical environments, and the reliability and clinical relevance of the tuberculosis (TB) resistance mutations identified by these assays has not been well characterized.
Methods: These analyses use data from a multisite, observational cohort study conducted by the Global Consortium for Drug Resistant TB Diagnostics (GCDD). Chapter 2 assesses the diagnostic performance of a pyrosequencing assay in this study and applies logistic regression to identify the variables associated with poor assay technical performance. Chapter 3 presents the frequency and distribution of all TB resistance mutations identified between clinical sites. Chapter 4 applies multivariate logistic regression to determine the associations between TB resistance mutations and patient mortality for the GCDD study cohort.
Results: Chapter 2 found pyrosequencing assay performance to be high, overall, though sensitivity fluctuated between sites. Smear negativity, culture negativity, site (Moldova), and sequencing of the rpoB, gyrA, and rrs genes were associated with poor assay technical performance. Chapter 3 found inclusion of the ahpC and eis promoter gene regions to be critical for optimal test sensitivity for isoniazid resistance detection in South Africa and kanamycin resistance detection in Moldova. Chapter 4 found that the detection of a mutation associated with high-level fluoroquinolone or kanamycin resistance was significantly associated with higher odds of patient mortality by 52 weeks.
Conclusions: A DR-TB diagnostic demonstrated excellent specificity in a large, multisite study, though performance was influenced by regional genetics and technical errors. Findings of the high prevalence of low-global frequency TB resistance mutations in specific sites emphasizes that these mutations may be important resistance markers in certain regions. The observation that patients with high-level gyrA and rrs resistance mutations had higher mortality odds provides strong evidence that certain high-level resistance mutations are clinically relevant. Together, these findings underscore the potential for molecular diagnostics to replace conventional growth-based diagnostics as predictors of phenotypic drug resistance and clinical outcome.