The World Health Organization (WHO) Global Tuberculosis Report 2020 estimates the number of people living with tuberculosis (TB) at around 10 million in 2019, which makes it the most common cause of death due to a single infectious agent (Chakaya et al., 2021). Tuberculous meningitis (TBM), the most serious form of extrapulmonary TB (Brancusi et al., 2012), is caused by Mycobacterium tuberculosis (M. tuberculosis, MTB) and is associated with significant morbidity and mortality, especially among children and people living with HIV (Ho et al., 2013). Early diagnosis and treatment of TBM is crucial for its prognosis (Garg, 2010). Unfortunately, early diagnosis of TBM is often difficult. Cerebrospinal fluid (CSF) smear, M. tuberculosis culture, and polymerase chain reaction are the gold standards for detecting M. tuberculosis in the CSF (Marais et al., 2010). However, the possibility of identifying acid-fast bacilli in CSF smears is very low, and the culture of M. tuberculosis in CSF is time consuming (Thwaites et al., 2004; Garg, 2019). Although polymerase chain reaction has a higher sensitivity in detecting M. tuberculosis DNA in CSF samples, it also has a higher false-positive rate (Donovan et al., 2020). In response to these challenges, a new technique for the rapid detection of M. tuberculosis has been developed.

In the past decade, the interferon-gamma release assay (IGRA) has become widely used as an immunodiagnostic method for M. tuberculosis infection (Bergot et al., 2018). It detects interferon (IFN)-gamma produced by T cells as a reaction to M. tuberculosis-specific antigens, such as early secretory antigenic target (ESAT)-6 and culture filtrate protein (CFP)-10, which are thought to be present only in M. tuberculosis but not in M. bovis bacille Calmette–Guérin (BCG) vaccine and other mycobacteria (Pai et al., 2006). QuantiFERON-TB Gold In-Tube (QFT-G-IT) (Cellestis, Carnegie, VIC, Australia) and T-SPOT.TB (T-SPOT) (Oxford Immunotec, Abingdon, United Kingdom) are the two most widely used IGRA systems to date, using enzyme-linked immunosorbent assay (ELISA) and enzyme-linked immunospot assay (ELISPOT) to detect IFN-gamma (Pai et al., 2006).

Blood IGRAs are most commonly used. However, an alternative approach of IGRAs using effector T cells from infected TB site specimens may be more likely to detect IFN against TB infection than using peripheral blood mononuclear cells (PBMCs). Recent studies have evaluated the use of CSF IGRAs for the diagnosis of TBM; however, the sample sizes of these included studies were insufficient, and their accuracy was disputable. Therefore, this study aimed to perform a meta-analysis to systematically evaluate and compare the accuracy of blood and CSF IGRAs in the diagnosis of TBM and to review relevant literature.

Since early diagnosis and treatment of TBM save lives, tests must be performed to diagnose TBM quickly and accurately, especially in its early stages (Ho et al., 2013). Unfortunately, the diagnosis of TBM is often ambiguous because of nonspecific clinical manifestations and the low sensitivity of available diagnostic methods (Thwaites et al., 2013). The absolute and most widely used diagnostic tools for TBM are Ziehl-Neelsen staining and culture (Marais et al., 2010), but they are negative in the majority of TBM cases. Our results showed that the sensitivity for CSF culture and/or AFB was 16% in the TBM group. Increasing the volume of CSF (>6 ml) obtained and meticulous microscopy (for at least half an hour) further increases the chance of positive diagnosis (Heemskerk et al., 2018; Bahr et al., 2019). For this reason, we did not restrict diagnostic criteria to microbiological confirmation, which is usually not possible in a routine clinical setting. This may produce bias. The detection of M. tuberculosis DNA in CSF samples using nucleic acid amplification tests (NAATs) are a widely used diagnostic method (Marais et al., 2010). In recent years, some studies on NAATs (Modi et al., 2016 Jyothy et al., 2019; Kwizera et al., 2019; Pormohammad et al., 2019; Siddiqi et al., 2019; Poplin et al., 2020; Slane and Unakal, 2021) reported a potential to rule in or confirm diagnosis (specificity, 80%-100%), but low sensitivity (~40%-96%) precludes the use of these tests to rule out TBM. Up to now, adenosine deaminase (ADA) is still a hot spot in TB diagnosis. Numerous studies have been published regarding the usefulness in TBM. Performance varies according to the assay and cut-off used, the mean sensitivity and specificity of ADA assays were 60%-90% and 80%-90%, respectively (Pormohammad et al., 2017). ADA assays may be useful in identifying TBM, but raised levels may also be seen in other central nervous system diseases such as purulent meningitis (Ekermans et al., 2017) ( Table 3 ).

To address this, newer tests have emerged for the early diagnosis of TBM, such as IGRAs. The design and development of IGRAs were based on the fact that T lymphocytes release IFN-gamma in response to the stimulation of M. tuberculosis-specific antigens such as ESAT-6 and CFP-10 (Andersen et al., 2000; Pai et al., 2007; Stevens et al., 2019). Moreover, blood-based tests have been widely evaluated and are considered promising tools for the rapid detection of M. tuberculosis (Westermann and Pabst, 1992; Meier et al., 2005; Kawamura et al., 2012; Rangaka et al., 2012). In addition, the accuracy of CSF IGRAs has also been evaluated for the diagnosis of TBM (Kim et al., 2008; Thomas et al., 2008; Kim et al., 2010; Patel et al., 2010; Park et al., 2012; Zhang et al., 2013; Caliman-Sturdza et al., 2015; Qin et al., 2015; Lu et al., 2016; Pan et al., 2017). However, no studies comparing blood and CSF IGRAs have been performed yet. Therefore, we performed this meta-analysis to systematically evaluate and compare the diagnostic accuracy of blood and CSF IGRAs for diagnosing TBM. The obtained overall sensitivity, specificity, PLR, NLR, DOR, and the SROC AUC in blood samples were 74% (95% CI: 0.65-0.82), 78% (95% CI: 0.68-0.86), 3.38 (95% CI: 2.26-5.06), 0.33 (95% CI: 0.23-0.46), 10.25 (95% CI: 5.46-19.25), and 0.83 (95% CI: 0.79-0.86), respectively, and for CSF, these values were 79% (95% CI: 0.71-0.85), 95% (95% CI: 0.88-0.98), 16.30 (95% CI: 6.50-40.83), 0.22 (95% CI: 0.16-0.31), 57.93 (95% CI: 22.56-148.78), and 0.91 (95% CI: 0.88-0.93), respectively. The post-test probability of blood samples was 46%/8% and 80%/5% in CSF samples. These results suggest that the IGRAs carried out in CSF have better diagnostic sensitivity and specificity than IGRAs in blood for the diagnosis of TBM.

However, even if the sensitivity, specificity, and post-test probability were higher for CSF IGRAs than for blood IGRAs in this analysis, the diagnostic sensitivity of CSF IGRAs alone was not high enough to support the diagnosis of TBM. Nevertheless, CSF IGRAs have some advantages over blood IGRAs. When peripheral blood IGRAs are used alone, it becomes difficult to distinguish active tuberculosis (ATB) from latent tuberculosis infection in a clinical setting (Pai et al., 2004; Gao et al., 2017). In an ATB stimulation, antigen-specific T lymphocytes are recruited to the sites of infection and proliferate rapidly (Schwander et al., 1998; Hirsch et al., 2001). Because of this, CSF IGRA is better at distinguishing ATB and has a higher ability to predict the location of the M. tuberculosis infection, especially in TBM cases.

The current analysis had some limitations. First, among the ten included studies that utilized CSF IGRA, the sample size was fairly small. Second, the cut-off and incubation cell numbers across the studies were inconsistent. This is an important consideration because the cut-off values greatly influence the sensitivity and specificity. Finally, in some of the included studies, the diagnosis of TBM patients was not confirmed by microbiological methods (smear or culture).

The results of this meta-analysis showed that CSF IGRAs had higher sensitivity and specificity in the diagnosis of TBM than peripheral blood IGRAs. However, carefully designed higher-quality independent studies are required to reliably compare the diagnostic accuracies of blood and CSF IGRAs.

AW, E-LL, and FH designed the study. The manuscript was written by AW, E-LL, and FH with the final approval. S-ML and Y-LZ did the data searches and study selection. W-FC and D-YY did the data synthesis and created the tables and figures. All authors contributed to the article and approved the submitted version.

This study was supported by Jiangxi Provincial Health Commission Science and Technology Foundation grants 20203025.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.