Sputum viability microscopy for tuberculosis

Abstract

Introduction. Sputum microscopy is the most frequently used tuberculosis laboratory test, but poorly assesses infectiousness and treatment response. The viability stain fluorescein diacetate (FDA) identifies metabolically active cells. Objective. To optimise sputum microscopy for predicting tuberculosis infectiousness and treatment outcome. Foundations. FDA sputum microscopy rapidly indicated the concentration of Mycobacterium tuberculosis that had not been sterilised by boiling and that was culturable. A cohort of 35 patients had FDA microscopy performed on their pre-treatment sputum. Lower FDA microscopy results predicted higher risk of tuberculosis disease in their 209 household contacts (adjusted hazard ratio=3.9, p=0.02). During the first 9 days of first-line therapy, these 35 patients had FDA microscopy on repeated sputum samples that rapidly differentiated patients with drug-susceptible tuberculosis from those with multi-drug resistant tuberculosis (p<0.001). Refinement. To optimise sputum collection for programmatic use, a network meta-analysis of 23 studies demonstrated that pooled sputum, as used in the foundation studies, had similar performance to more easily collected, instructed on-demand “spot” sputum collection (p=1.0). FDA microscopy was optimised for use in basic clinical laboratories. This replaced potentially biohazardous and expensive steps with more practicable alternatives. Evaluation. These refined protocols were used to collect pre-treatment sputum samples from 978 patients. Higher mycobacterial concentrations assessed by culture, PCR, conventional or FDA microscopy all predicted greater risk of tuberculosis disease in their 4311 household contacts (p<0.0001), with concordance (C)-statistics between 58-62%. Paired samples, pre-treatment and after 14 days of tuberculosis treatment, were collected from 384 patients. An alert FDA microscopy result at 14 days of treatment predicted adverse treatment outcomes (relative risk=3.0, p<0.0001), and performed better than currently used prognostic markers with 42% sensitivity and 85% specificity. Conclusion. After methodological optimisation, FDA microscopy predicted patient infectiousness and treatment outcome with similar or greater reliability than currently used laboratory tests.