Improved detection of extrapulmonary and paucibacillary pulmonary tuberculosis by Xpert MTB Host Response in a TB low endemic, high resource setting.

Background: Xpert MTB-Host Response (MTB-HR) has reached WHO test targets for pulmonary tuberculosis (PTB) with high bacillary loads. Our aim was to investigate the contribution of MTB-HR as a non-sputum, near point-of-care (POC) method for diagnosis of other prioritized groups, where MTB detection is more complicated, such as extrapulmonary tuberculosis (EPTB) and paucibacillary PTB.

Methods: Individuals with presumed TB disease were prospectively included in Stockholm, Sweden (n=307) and underwent MTB-HR venous and capillary testing in parallel. Clinical characterization was based on symptoms, microbiological results (microscopy, PCR and culture), radiological assessment and a panel of routine biochemical tests. ROC-analysis was performed in PTB and EPTB calculating cut-offs for maximized sensitivity and specificity including WHO targets for screening and diagnostic tests.

Findings: MTB-HR performed equally well in individuals with microbiologically confirmed PTB (AUC 0.84; 95% CI: 0.78-0.90, n=69) and EPTB (AUC 0.82 (95% CI: 0.75-0.90, n=34). Based on Youden index cut-offs the NPV was high both in microscopy- and PCR-negative PTB (-1.27, NPV 94%) and in EPTB (-1.58, NPV 95%) and fulfilled the minimum TPP sensitivity requirement for confirmed EPTB. In individuals without TB (n=204), the majority had pulmonary infectious diseases. There was a close to perfect correlation between venous and capillary samples (r= 0.97, p<0.001).

Interpretation: Capillary Xpert MTB-HR improves detection of sputum PCR-negative, culture-verified PTB and is promising as a rule-out test in EPTB. There was a strong correlation between MTB-HR score and bacterial burden. We suggest a graded MTB-HR score as more clinically relevant than a binary test result.