A proof-of-concept study: targeted next generation sequencing for the detection of drug-resistance associated mutations in Mycobacterium tuberculosis

Early detection of drug-resistance (DR) in Mycobacterium tuberculosis (MTB) is essential for effective treatment guidance. This study evaluated a proof-of-concept workflow combining multiplex PCR with Nanopore sequencing (ONT) for the detection of DR-associated mutations in cultured samples and assessed its potential for clinical samples. This approach could potentially benefit early clinical decision-making and enable more personalized treatment options. A novel multiplex PCR assay targeting 18 resistance-associated genes was evaluated using a total of 126 MTB-positive samples, consisting of 102 culture-derived and 24 clinical sputum samples collected between 2014 and 2024. Culture-derived samples were compared with Illumina-based whole-genome sequencing (WGS) ( n = 24) and the MTBDRplus assay ( n = 78). Clinical sputum specimens were also assessed against Illumina-based WGS ( n = 24). Detection of resistance-associated mutations (RAMs) was investigated across samples. In culture-derived samples, ROC analysis identified an optimal VAF threshold of 61.3% (AUC = 0.994). RAM detection resulted in a sensitivity of 96.4% and specificity of 100% when compared to Illumina-based WGS ( n = 24). Comparison with MTBDR showed complete agreement for isoniazid resistance. Lower performance was observed for rifampicin resistance but false-positive results may represent true RAMs ( n = 33). The broader ONT panel additionally identified RAMs associated with STM and PZA resistance ( n = 45). In clinical sputum, sequencing depth decreased markedly with increasing Ct values and remained variable between targets, limiting RAM detection. The approach showed high concordance for RAM detection in culture-derived samples, while variable target coverage limited performance in sputum. Optimization of individual amplicons, sequencing coverage and variant calling, followed by evaluation in larger clinical datasets, is required before diagnostic validation for culture-independent DR profiling. Not applicable.