# Improved detection of Mycobacterium tuberculosis in lymph node aspirates through GeneXpert MTB/RIF assay in Bangladesh

## TL;DR

This study evaluates the effectiveness of the Xpert MTB/RIF assay for diagnosing tuberculosis in lymph nodes in Bangladesh, finding it useful but recommending additional testing for confirmation.

## Contribution

The study provides new evidence on the diagnostic accuracy of Xpert for TBL in Bangladesh, where prior data was limited.

## Key findings

- Xpert showed 72.9% sensitivity and 100% specificity against the composite reference standard.
- Cytology had higher sensitivity (84.1%) than Xpert but similar specificity.
- Xpert demonstrated moderate agreement with cytology and poor agreement with AFB microscopy.

## Abstract

Despite tuberculous lymphadenitis (TBL) being the most prevalent type of extrapulmonary tuberculosis, there are limitations in laboratory diagnosis of TBL due to high cost, inadequate diagnostic efficacy and feasibility. Xpert MTB/RIF (Xpert) assay is a design-locked, molecular diagnostic technique that detects Mycobacterium tuberculosis (MTB) genome and rifampicin resistance by targeting rpoB gene and mutations within the gene. Currently, there is limited evidence validating Xpert assay usage for diagnosing TBL in Bangladesh. Therefore, in this study, we evaluated diagnostic efficacy of Xpert, considering composite reference standard (CRS), i.e. combination of acid-fast bacilli (AFB) microscopy, culture, Xpert assay and cytology, as gold standard, and compared it to cytology.

AFB microscopy, culture, cytology, and Xpert assay were conducted on fine needle aspirates collected from 523 presumptive TBL patients. Genomic DNA was extracted from bacterial colonies of culture-positive specimens. In order to confirm presence of MTB, PCR and gel electrophoresis were performed on extracted DNA to detect RD9 region of MTB DNA. Sensitivity, specificity, positive and negative predictive values, and Cohen’s kappa coefficient were determined, and McNemar’s test was performed for Xpert and cytology with respect to CRS. Additionally, latent class analysis was performed to estimate sensitivity and specificity of all four diagnostic modalities.

Xpert showed sensitivity and specificity of 72.9% (261/358) and 100% (165/165) respectively against CRS, with 69.8% sensitivity and 97.1% specificity using Bayesian latent class modeling. In contrast, cytology demonstrated sensitivity and specificity of 84.1% (301/358) and 100% (165/165) against CRS, and 81.9% and 99.9% upon latent class analysis, respectively. Furthermore, Xpert showed moderate agreement with cytology (κ = 0.45, p < 0.0001), fair agreement with culture (κ = 0.30, p < 0.0001), and poor agreement with AFB microscopy (κ = 0.09, p < 0.0001).

Our study findings validate routinely using Xpert assay in TBL diagnosis and enable detecting patients with low bacterial load. However, further assessment via cytology is recommended for confirmation in Xpert-negative patients having patent symptoms.

Not applicable.