Differentially culturable tubercle bacteria as a measure of tuberculosis treatment response

INTRODUCTION: Routine efficacy assessments of new tuberculosis (TB) treatments include quantitative solid culture or routine liquid culture, which likely miss quantification of drug tolerant bacteria. To improve these assessments, comparative analyses using additional measures such as quantification of differentially culturable tubercle bacteria (DCTB) are required. Essential for enabling this is a comparative measure of TB treatment responses using routine solid and liquid culture with liquid limiting dilutions (LLDs) that detect DCTB in sputum. METHODS: We recruited treatment-naïve TB patients, with and without HIV-infection, and serially quantified their sputum for DCTB over the course of treatment. RESULTS: Serial sputum sampling in 73 individuals during their first 14 days of treatment demonstrated that clearance of DCTB was slower compared to routine solid culture. Treatment response appeared to be characterized by four patterns: (1) Classic bi-phasic bacterial clearance; (2) early non-responders with slower clearance; (3) paradoxical worsening with an increase in bacterial count upon treatment initiation; and (4) non-responders with no change in bacterial load. During treatment, LLDs displayed greater bacterial yield when compared with quantitative solid culture. Upon treatment completion, 74% [46/62] of specimens displayed residual DCTB and within this group, two recurrences were diagnosed. Residual DCTB upon treatment completion was associated with a higher proportion of MGIT culture, GeneXpert, and smear positivity at two months post treatment. No recurrences occurred in the group without residual DCTB. DISCUSSION: These data indicate that DCTB assays detect distinct subpopulations of organisms in sputum that are missed by routine solid and liquid culture, and offer important alternatives for efficacy assessments of new TB treatments. The residual DCTB observed upon treatment completion suggests that TB treatment does not always eliminate all bacterial populations, a finding that should be investigated in larger cohorts.