Clinical and bacterial determinants of unfavorable tuberculosis treatment outcomes: an observational study in Georgia.

BACKGROUND: Tuberculosis (TB) remains a major public health concern. Improving TB control programs and treatment success requires a deeper understanding of the factors that determine disease presentation and treatment outcomes. While the importance of patient factors is well established, our understanding of the bacterial determinants of disease presentation and treatment outcomes in TB remains limited. METHODS: In this study, we analyzed the Mycobacterium tuberculosis complex (MTBC) genomes and the associated clinical data from 4529 TB patients in the country of Georgia covering a period of 13 years. We used multivariable modeling together with genome-wide association studies (GWAS) to identify patient and bacterial factors that determine TB disease manifestation and clinical outcomes. RESULTS: Multivariable modelling confirmed the role of demographic and clinical factors in determining treatment outcomes, as well as the efficacy of novel TB treatments containing bedaquiline. In addition, we found that several bacterial factors, including the MTBC lineage, the specific mutations conferring resistance to rifampicin and fluoroquinolones, as well as a high bacterial burden, were associated with unfavorable outcomes. GWAS analyses revealed no bacterial genetic mutations associated with treatment outcomes beyond the known drug resistance-conferring mutations. However, we found that mutations in the bacterial gene sufD were linked to a reduced risk of lung cavities and a lower bacterial burden within patients. By contrast, specific mutations conferring resistance to rifampicin and fitness compensatory mutations were associated with a higher bacterial burden. CONCLUSIONS: Our results show that both patient and bacterial factors determine disease presentation and clinical outcomes in TB. They also support the rationale of optimizing treatment regimens against drug-resistant TB with existing drugs based on the specific genetic features of the pathogen. Finally, our results highlight sufD as a possible therapeutic candidate.