Risk factors associated with large clusters of tuberculosis patients determined by whole-genome sequencing in a high-tuberculosis-burden country.

Whole-genome sequencing (WGS) analysis has great discriminative power for detecting similar molecular fingerprints of suspected tuberculosis (TB) clusters. The proportion of TB cases within clusters and the associated risk factors are important epidemiological parameters guiding appropriate outbreak control strategies in endemic settings. We conducted a hospital-based TB case-cohort study between 2003 and 2011 in the northernmost province of Thailand. We identified TB clusters by Mycobacterium tuberculosis WGS and analysed the risks of TB clustering and the characteristics of large clusters compared with small clusters. Among 1146 TB isolates, we identified 77 clusters with 251 isolates defined by a 5-single-nucleotide variant (SNV) cutoff and 112 clusters with 431 isolates defined by a 12-SNV cutoff. Twelve large clusters with 6 isolates or more in each cluster were identified by a 12-SNV cutoff. Sublineage 2.2.1 (both Ancestral and Modern) strains and imprisonment were independently associated with large clusters. Furthermore, although large clusters of Lineage 2.2.1/Ancestral strains included a high number of prisoners, Lineage 2.2.1/Modern strain clusters were only associated with treatment failures and drug resistance. Heterogeneity among lineage strains was observed with respect to large-cluster characteristics. Patients with an increased TB-transmission tendency should be priority targets for contact investigations and outbreak interventions to prevent ongoing transmission.