Improving the Sensitivity of the Xpert MTB/RIF Assay on Sputum Pellets by Decreasing the Amount of Added Sample Reagent: a Laboratory and Clinical Evaluation

The Xpert MTB/RIF (Xpert) assay permits rapid near-patient detection ofMycobacterium tuberculosis in sputum; however, the test sensitivity remains suboptimal in paucibacillary specimens that are negative for acid-fast bacilli using smear microscopy. Xpert testing includes dilution with sample reagent, and when processed sputum pellets are tested, the recommended sample reagent/pellet ratio is 3:1. We evaluated whether a decreased sample reagent/pellet ratio of 2:1 increased Xpert sensitivity com-pared to the recommended 3:1. The limit of detection was determined by inoculating serial dilutions ofM. tuberculosis into spu-tum samples, preparing sputum pellets, and testing each pellet by Xpert at both sample reagent ratios. Processed sputum pellets obtained fromM. tuberculosis culture-positive clinical specimens were also tested by Xpert at both ratios. Among spiked sputum pellets, the limit of detection was 1,478 CFU/ml (95 % confidence interval [CI], 1,211 to 1,943) at a 3:1 ratio and decreased to 832 CFU/ml (95 % CI, 671 to 1,134) at 2:1. The proportion of specimens in whichM. tuberculosiswas detected was greater at 2:1 than at 3:1 for almost all numbers of CFU/ml; this difference was most prominent at lower numbers of CFU/ml. Among 134 concen-trated sputum pellets from the clinical study, the sensitivity of Xpert at 2:1 was greater than at 3:1 overall (80 % versus 72%; P 0.03) and for smear-negative specimens (67 % versus 58%; P 0.12). For Xpert testing of sputum pellets, using a lower sampl