Evolutionary dynamics of drug resistance in MDR-TB: Heterogeneous minor variants and extensive compensatory mutations.

Multidrug-resistant Mycobacterium tuberculosis complex (MTBC) remains a major global health challenge, yet the drug-specific evolutionary pathways through which resistance accumulates are not fully understood. Heterogeneous minor resistant variants (hSNPs) and compensatory mutations are increasingly recognized as important drivers of the emergence, stability, and transmission of drug-resistant strains. We analyzed 143 clinical MTBC isolates using whole-genome sequencing combined with phenotypic drug susceptibility testing to characterize lineage distribution, resistance-associated mutations, hSNPs, compensatory adaptations, and genomic clustering. We performed evolutionary accumulation modelling with HyperTraPS to infer the most probable order of resistance acquisition. Compensatory mutations in rpoA, rpoC, or the non-RRDR region of rpoB were detected in 73.1% of genotypic rifampicin-resistant isolates, with the rpoB p.Ser450Leu mutation showing the highest frequency of compensatory adaptation (92.2%). Drug-specific hSNP patterns were highly distinct. INH-, RIF-, and PZA-associated hSNPs typically occurred as multiple low-frequency variants, whereas FQ- and SM-associated hSNPs predominantly appeared as single-site variants. Evolutionary modelling indicated that resistance to INH and RIF is generally acquired early, followed by SM and EMB resistance. In the inferred resistance-acquisition pathways, INH resistance was frequently followed by EMB resistance, whereas RIF resistance was positively associated with subsequent SM resistance. In contrast, resistance to PZA, EMB, or SM was associated with a reduced probability of subsequent ETO resistance. The overall genomic clustering rate was 20.98%, and ongoing microevolution within transmission clusters was commonly observed, characterized by the emergence of new resistant variants. These findings elucidate the evolutionary dynamics of multidrug-resistant tuberculosis, emphasizing structured resistance accumulation and within-host heterogeneity with implications for surveillance and treatment optimization.