Background: Tuberculosis treatment monitoring is hindered by the lack of a rapidly measured biomarker that accurately predicts clinically relevant outcomes. Symptom screening poorly correlates with bacillary burden. Although culture is a direct measure of viable bacillary burden, the long turnaround time makes it clinically irrelevant. Methods: The tuberculosis treatment monitoring potential of stool-based, quantitative PCR was prospectively assessed among 231 microbiologically confirmed tuberculosis participants of all ages from Eswatini, Tanzania, and Mozambique. Stool qPCR results were compared to sputum culture, persistent symptoms, drug resistance, and WHO tuberculosis outcomes. Results: Quantitative bacillary burden measured by stool qPCR strongly correlated with sputum culture at baseline (Spearman correlation rs = 0.79; p < 0.001). Stool was successfully collected at >90% of all timepoints, while sputum collection decreased to <50% at the end of therapy. Participants with isoniazid or rifampin resistance demonstrated decreased bacillary clearance by sputum culture and stool qPCR during the first two weeks of treatment. Participants who remained culture positive at 2 months had a slower decrease in bacillary burden measured by stool qPCR compared to those who were culture negative by 2 months. The odds of a participant being culture positive at two months was associated with i) a lower initial qPCR cycle threshold (Odds ratio (OR) = 0.792; p = 0.004), and ii) a smaller absolute difference between the qPCR cycle threshold measured at two weeks and baseline (OR = 0.72; p = 0.0006). Neither sputum culture, sputum Xpert Ultra, or stool qPCR were associated with resolution of symptoms or in-treatment death. Conclusions: Stool-based tuberculosis treatment monitoring correlates with sputum culture but provides results faster, leverages a more accessible specimen, and identifies patients with tuberculosis who are at risk for drug resistance and persistent 2-month culture positivity. None of the quantitative tests of bacillary burden singularly could predict symptom resolution or death.
New diagnostic tests and alternative samples to diagnose tuberculosis in children are needed to tackle the diagnostic gap of pediatric tuberculosis. The Determine TB LAM Ag test in urine demonstrated sensitivity of 15.9% and specificity of 90.6%. Collecting the urine in adhesive bag was an independent risk factor for positivity (odds ratio, 11.38; P < .01).