Serial image analysis of Mycobacterium tuberculosis colony growth reveals a persistent subpopulation in sputum during treatment of pulmonary TB

Abstract

Faster elimination of drug tolerant ‘persister’ bacteria may shorten treatment of tuberculosis (TB) but no method exists to quantify persisters in clinical samples. We used automated image analysis to assess whether studying growth characteristics of individual Mycobacterium tuberculosis colonies from sputum on solid media during early TB treatment facilitates ‘persister’ phenotyping. As Time to Detection (TTD) in liquid culture inversely correlates with total bacterial load we also evaluated the relationship between individual colony growth parameters and TTD. Sputum from TB patients in Malawi was prepared for solid and liquid culture after 0, 2 and 4 weeks of treatment. Serial photography of agar plates was used to measure time to appearance (lag time) and radial growth rate for each colony. Mixed-effects modelling was used to analyse changing growth characteristics from serial samples. 20 patients had colony measurements recorded at ≥1 time-point. Overall lag time increased by 6.5 days between baseline and two weeks (p = 0.0001). Total colony count/ml showed typical biphasic elimination, but long lag time colonies (>20days) had slower, monophasic decline. TTD was associated with minimum lag time (time to appearance of first colony1). Slower elimination of long lag time colonies suggests that these may represent a persister subpopulation of bacilli.