Retrospective Diagnostic Accuracy Study of Abbott RealTime MTB against Xpert MTB/RIF Ultra and Xpert MTB/RIF for the Diagnosis of Pulmonary Tuberculosis and Susceptibility to Rifampin and Isoniazid Treatment

High-throughput centralized testing for tuberculosis (TB) and drug resistance is important, but comparative data are limited. In this retrospective cross-sectional study, participants were recruited from Johannesburg, South Africa, and Tbilisi, Georgia. The index tests, Abbott RealTime MTB (RT-MTB) and RealTime MTB RIF/INH (RT-MTB RIF/INH), were performed on specimens stored frozen for an extended period of time (beyond manufacturer-validated specifications) and compared to paired Xpert MTB/RIF Ultra (Xpert Ultra) and Xpert MTB/RIF (Xpert) results obtained with fresh specimens. The detection reference standard was the Mycobacterium tuberculosis complex culture, and for resistance detection, it was phenotypic drug susceptibility testing. The median age of 474 participants was 39 (interquartile range [IQR], 31 to 51) years. On decontaminated sputum, Xpert Ultra had a sensitivity of 91%, compared to 77% for RT-MTB, with a difference of +14% (95% confidence interval [CI], +9.2 to +21%; 18/127). On raw sputum, Xpert Ultra exhibited a sensitivity of 89% and Xpert one of 88%, compared to 80% for RT-MTB, exhibiting differences of +10% (95% CI, +3.3 to +18%; 9/93) and +8.6% (95% CI, +2.4 to +17%; 8/93), respectively. Specificity was ≥98% for all tests. All three tests showed high sensitivity and specificity for detection of rifampin resistance. Abbott assays may have lower sensitivity than Xpert and Xpert Ultra for TB detection but similar performance for detection of resistance. The differences in TB detection may be attributable to differences in testing of frozen (Abbott) versus fresh (Xpert) samples. Studies in compliance with manufacturer’s instructions are required to compare performance. IMPORTANCE In 2019, 10 million people fell ill with tuberculosis (TB), of whom 1.4 million died. There are few comparative studies of diagnostic assays, particularly those aiming to be used in high-throughput laboratories. One such assay is the Abbott RealTime MTB (RT-MTB) and RealTime MTB RIF/INH (RT-MTB RIF/INH), which uses the m2000 platform already in use in many settings for HIV load testing and allows the diagnosis of TB and resistance to two first-line drugs, rifampin and isoniazid. Our study compared the RT-MTB and RT-MTB RIF/INH to the WHO-recommended Xpert MTB/RIF Ultra and Xpert MTB/RIF. The study is the largest comparative study to date and was performed independent of the manufacturer. The study results suggest that the Abbott RealTime MTB may have a lower sensitivity, but the study may have placed the Abbott test at a disadvantage by using frozen samples and comparing the results to those for fresh samples for the Xpert.