Comparative Accuracy of Tests to Detect Pulmonary Tuberculosis

In a review update, XpertMTB/RIF was found to be sensitive and specific for diagnosing pulmonary tuberculosis (TB) and rifampicin resistance which is consistent with the last review completed in 2014. This update was published in The Cochrane Database of Systematic Reviews and was performed for an upcoming World Health Organization policy review.

The review consisted of randomized trials and cross-sectional and cohort studies using respiratory specimens that evaluated Xpert MTB/RIF, Xpert Ultra, or both against the reference standard, culture for TB and culture-based drug susceptibility testing or MTBDR plus for rifampicin resistance. Databases included in the search were the Cochrane Infectious Diseases Group Specialized Register, MEDLINE, Embase, Science Citation Index, Web of Science, Latin American Caribbean Health Sciences Literature, Scopus, the WHO International Clinical Trials Registry Platform, the International Standard Randomized Controlled Trial Number Registry, and ProQuest, for results up to October 11, 2018, and with no language restrictions.

Search results found that 77 new studies had been completed since the previous review, creating a new total of 95 studies. Of these, 86 studies (42,091 participants) evaluated Xpert MTB/RIF for TB, 57 (8287 participants) for rifampicin resistance in cases of pulmonary TB, and 1 study compared Xpert MTB/RIF and Xpert Ultra on the same participant specimen.

In terms of TB detection, most of the 86 studies had low risk for bias. In 70 studies, with 37,237 unselected participants and high-certainty evidence, Xpert MTB/RIF pooled sensitivity was 85% (95% credible interval [CrI] 82%-88%) and specificity was 98% (95% CrI, 97%-98%). A similar sensitivity and specificity was also found when all studies were included. The studies also showed that for a population of 1000 people where 100 have TB on culture results, 103 would be XpertMTB/RIF-positive and 18 (17%) would be false-positives while 897 would be XpertMTB/RIF-negative and 15 (2%) would be false-negatives. 

Xpert Ultra sensitivity was 88% (95% CI, 85%-91%) vs Xpert MTB/RIF 83% (95% CI, 79%-86%). Xpert Ultra specificity was 96% (95% CI, 94%-97%) vs Xpert MTB/RIF 98% (95% CI, 97%-99%) according to 1 study with 1439 participants and a moderate-certainty evidence. Pooled sensitivity of Xpert MTB/RIF was 98% (95% CI, 97%-98%) in smear-positive and 67% (95% CI, 62%-72%) in smear-negative, culture positive participants, according to data from 45 studies. Based on the data from 14 studies, Xpert MTB/RIF pooled sensitivity was 88% (95% CI, 83%-92%) in HIV-negative and 81% (95% CI, 75%-86%) in participants with HIV. The specificities were similar 98% (95% CI, 97%-99%).

Of note, investigators found that in countries with a high burden of comorbid HIV and TB, Xpert MTB/RIF pooled sensitivity was 83% (95% CrI, 80%-87%), and pooled specificity was 97% (95% CrI, 95%-98%).

For the detection of rifampicin resistance Xpert MTB/RIF pooled sensitivity and specificity were 96% (95% Crl, 94%-97%) and 98% (95% Crl, 98%-99%) based on 48 studies with 8020 participants and high-certainty evidence.

For a population of 1000 people where 100 have rifampicin-resistant TB, 114 would be positive for rifampicin-resistant TB and 18 (16%) would be false-positives while 886 would be negative for rifampicin-resistant TB and 4 (0.4%) would be false-negatives.

According to 1 study of 551 participants and moderate-certainty evidence, Xpert Ultra sensitivity was 95% (95% CI, 90%-98%) vs XpertMTB/RIF 95% (95% CI, 91%-98%) and Xpert Ultra specificity was 98% (95% CI, 97%-99%) vs Xpert MTB/RIF 98% (95% CI, 96%-99%).

Investigators concluded that both tests, “provide accurate results and can allow rapid initiation of treatment for multidrug-resistant tuberculosis” and consistent with previous findings XpertMTB/RIF is sensitive and specific for diagnosing pulmonary TB and rifampicin resistance. It is also more sensitive for TB in smear-positive rather than smear-negative patients as well as HIV-negative vs HIV-positive patients. Xpert Ultra had higher sensitivity and lower specificity for TB and similar sensitivity and specificity for rifampicin resistance compared to XpertMTB/RIF. They also suggested that future studies compare the diagnostic accuracy of Xpert Ultra with other rapid tests, especially in difficult-to-diagnose groups.

Reference

Horne DJ, Kohli M, Zifodya JS, et al. Xpert MTB/RIF and Xpert MTB/RIF Ultra for pulmonary tuberculosis and rifampicin resistance in adults. Cochrane Database Syst Rev. 2019;6:CD009593.