Use of reverse dot blot hybridization DNA-array and targeted next-generation sequencing for the detection of drug resistance in patients with tuberculosis: a prospective diagnostic accuracy study

Phenotypic drug susceptibility testing (pDST) for Mycobacterium tuberculosis complex (MTC) is slow, and routine molecular assays detect only a subset of resistance-associated mutations. We aimed to assess the diagnostic accuracy of the TB Resistance DNA-Array (GenID GmbH, Germany) for detecting isoniazid (INH) and rifampicin (RIF) resistance, and to evaluate the added value of Deeplex® Myc-TB, a culture free, targeted next-generation sequencing (tNGS) from GenoScreen, France, for identifying drug resistance and resolving discrepancies with pDST. Clinical samples from 234 individuals with bacteriologically confirmed pulmonary tuberculosis (TB) were analyzed: 98 (41.9%) from Ukraine, 25 (10.7%) from Moldova, and 111 (47.4%) from Spain. TB Resistance DNA-Array and tNGS results were compared with pDST and cross-evaluated to identify resistance-conferring mutations. Among 196 sputum specimens tested with the TB Resistance DNA-Array, 98 (50.0%) produced conclusive RIF results and 102 (52.0%) conclusive INH results. Among conclusive outputs, the assay showed high sensitivity and specificity for RIF (94.3% and 98.4%) and INH (95.7% and 94.6%). Deeplex® Myc-TB generated conclusive results for 24/30 (80.0%) sputum samples and 21/23 (91.3%) culture isolates from 53 individuals. tNGS demonstrated high sensitivity for INH and RIF (both > 97.0%) but misclassified 2 of 8 INH-susceptible cases and 2 of 9 RIF-susceptible cases as resistant, suggesting a potential risk of overtreatment with second-line drugs. For linezolid, tNGS showed perfect specificity (100%); however, 1 of 3 cases was falsely categorised as resistant. For fluoroquinolones, sensitivity and specificity were 90.0% and 83.3%, respectively. The TB Resistance DNA-Array performed reliably for detecting RIF and INH resistance in sputum samples with conclusive results. Integration of tNGS can improve interpretative algorithms tailored to clinically relevant and region-specific resistance mutations.