Diagnostic Utility of Mycobacterium Tuberculosis Cell-free DNA in Hong Kong

Tuberculosis (TB) remains a key infectious disease burden globally. According to the Global TB Report by the World Health Organization (WHO) in 2023, Over seven million people were infected with TB in 2022, and 1.3 million died from the disease. China is among the top three countries with the highest TB burden, while Hong Kong has an intermediate burden of TB, with an average of 4,000 new cases per year.

TB infection can cause disease of various manifestations outside the lung parenchyma in about 25% of patients (extrapulmonary TB, EPTB), and is associated with worse treatment outcomes than pulmonary TB. Tuberculous pleuritis (TBP) is its most prevalent form in Hong Kong and China (41% to 50% of all EPTB). Although TBP can present with acute to subacute occurrence of unilateral pleural effusion, breathlessness and fever, its presentation is frequently non-specific. Due to its clinical and epidemiological importance, accurately discriminating TBP from other types of pleural effusion is crucial to shorten the diagnostic process and optimise the outcomes for these patients.

Laboratory challenges for diagnosing TBP Diagnosing TBP is challenging as the disease is caused by a small number of bacteria (paucibacillary nature) in the pleural space, which limits the diagnostic performance of various conventional diagnostic modalities. These limitations include low sensitivity for pleural fluid microbiological tests (acid-fast bacilli [AFB] stain, Mycobacterium tuberculosis [MTB] culture and MTB polymerase chain reaction [PCR]) and lack of specificity for pleural fluid biomarkers (adenosine deaminase [ADA]) especially in low TB prevalence regions. Although the yields of MTB culture and PCR on pleural tissue are higher, pleural biopsy (by bedside procedure or pleuroscopy) carries invasiveness and requires an additional procedure. Currently, no ideal test can achieve multiple diagnostic goals simultaneously, including high detection sensitivity, high specificity to exclude other diseases, short turnaround time, low invasiveness and detection of drug resistance. The diagnosis of TBP is practically based on a composite of clinical, radiological and laboratory endpoints. The diagnosis may occasionally be established clinically without any microbiological or histological proof (probable TBP).

Consequences of delayed TBP diagnosis The key clinical drawbacks of difficulty in diagnosing TBP include unnecessary invasive pleural interventions and delayed initiation of curative treatment. Our internal audit recently revealed that the initiation of treatment was delayed while waiting for confirmatory test results, and patients required a median of one additional procedure to establish the diagnosis, which included pleural biopsy with a more invasive nature. These patients had long hospital stays for workups and trials of multiple antibiotics for unresolved fever. Some patients may instead receive empirical anti-TB treatment based on compatible clinical presentations without confirmatory diagnostic investigation results. However, this is not without risk (e.g. hepatotoxicity), particularly in the elderly, who are more susceptible to TBP. A better diagnostic tool for TBP would mitigate these clinical challenges.

Liquid biopsy in pleural effusion: latest development from our group Liquid biopsy by detecting the cell-free DNA (cfDNA) in body fluids is a potential solution. When TB bacteria die or break down, the DNA fragments shed into the pleural space, forming Mycobacterium tuberculosis cfDNA (MTB cfDNA). Detecting the MTB cfDNA in the pleural fluid can directly prove the presence of TB bacilli or their subunits in the pleural space, which is an appealing option. Previous studies had employed MTB cfDNA to diagnose TBP, but the performance was variable. Che et al first measured MTB cfDNA by detecting the IS6110 insertion sequence. The sensitivity of diagnosing TBP was 75.0% (95% confidence intervals [CI] 63.7-86.3%) with specificity at 100.0% (95% CI 100.0-100.0%), and pleural fluid MTB culture-negative cases was included in the TBP group. The diagnostic sensitivity of MTB cfDNA was better than pleural fluid ADA, MTB culture and MTB PCR by Xpert MTB/RIF. Yang et al employed the same laboratory technique and found a high diagnostic sensitivity at 96.6% (95% CI 92.0-100.0%) in TBP cases defined by pleural fluid MTB or PCR positive status. However, the diagnostic sensitivity dropped to 79.5% (95% CI 72.4-86.7%) if pleural fluid MTB culture or PCR-negative cases were included in the TBP group. Similar findings were also reported by Sharma et al. All these study results reflected the heterogeneous diagnostic performance of MTB cfDNA. A negative pleural fluid MTB culture or PCR status, signifying a low bacterial load, can impair the clinical utility of MTB cfDNA.

Our group has recently developed a new laboratory assay measuring the MTB cfDNA levels, involving targeted sequencing through a capture-probe system and novel bioinformatics algorithm. The diagnostic performance remained robust even though pleural fluid MTB culture-negative cases were included, which can potentially obviate another invasive diagnostic procedure.

Need for a validation cohort before introducing to clinical practice Since the development cohort only contains a limited number of pleural fluid samples with definitive diagnoses, the true diagnostic accuracy of the novel MTB cfDNA assay can only be determined in a prospective, non-selective cohort reflecting the real-world situation, which will give a realistic diagnostic performance. This will confirm the clinical utility of MTB cfDNA assay before being introduced in clinical practice. We hypothesise that the novel MTB cfDNA assay remains robust in a prospective cohort and outperforms previously reported methods in discriminating pleural effusions due to TBP from non-TBP.