Patient-Level Reconstruction of Drug Resistance-Associated Molecular Variants from Mutation-Positive Clinical Tuberculosis tNGS Reports

Purpose: Targeted next-generation sequencing (tNGS) covers multiple drug resistance-associated regions, but drug-organized reports can duplicate variants across drugs and disperse a patient’s variants. Through nomenclature standardization and cross-drug deduplication, we reconstructed nonredundant patient-level variant profiles and characterized multigene co-occurrence, molecular variant complexity, and reported frequency patterns. Patients and Methods: This single-center retrospective descriptive study included each patient’s earliest eligible mutation-positive tuberculosis tNGS report from 2024– 2025. The analysis dataset comprised report-derived information and did not include raw sequencing data or systematically paired phenotypic drug susceptibility testing (DST) results. Variants were standardized using the 2023 World Health Organization catalogue, H37Rv coordinates, and Human Genome Variation Society recommendations. Duplicate patient-gene-variant records were collapsed into unique patient-variant events while retaining drug annotations. Eight mutually exclusive, exhaustive combinations were defined from RIF ( rpoB ), INH ( katG and/or inhA ), and FQ ( gyrA and/or gyrB ) axes. Results: Fifty-seven patients contributed 180 unique patient-variant events involving nine genes and 56 standardized variants. rpoB (84.2%) and katG (64.9%) were the leading genes. The leading variants were katG p.Ser315Thr (63.2%), rpsL p.Lys43Arg (29.8%), and rpoB p.Ser450Leu (26.3%), which together accounted for 37.8% of all events; 33 of 56 variants occurred in only one patient. Seven of the eight RIF/INH/FQ combinations were observed. RIF+/INH+/FQ− was most common (31/57, 54.4%), whereas RIF+/INH+/FQ+ had the highest median numbers of genes and variants (both 5.0). Twenty-three events in 16 patients had reported frequencies below 100%, and pncA - and gyrA -associated events accounted for 60.9%. Conclusion: In this mutation-positive cohort, patient-level reconstruction integrated annotations of the same variant across drugs and characterized hotspots and a long tail of variants, multigene co-occurrence within RIF/INH/FQ combinations, and report-derived frequency patterns. This traceable data structure may support laboratory report review and future genotype–phenotype studies.