Rapid Diagnosis of Tuberculosis Meningitis by Detecting Mycobacterium tuberculosis Cell-Free DNA in Cerebrospinal Fluid

Abstract

Tuberculosis meningitis (TBM) is one of the most severe forms of tuberculosis and often results in severe neurologic sequelae, including death.¹ The mortality rate of TBM is about 30%, and the deaths occur mainly in the first 6 months.² Due to the nonspecific clinical manifestations and paucity of Mycobacterium tuberculosis (MTB) microorganisms in the cerebrospinal fluid (CSF), TBM diagnosis is still quite challenging.

In 2009, the international TBM workshop established a consensus on case definition for TBM research studies.³ According to this definition, diagnostic results of TBM can be divided into definite TBM, probable TBM, possible TBM, and non-TBM. Definite TBM is defined when one or more of the following criteria are positive in CSF: acid-fast staining, mycobacterial culture, or commercial nucleic acid amplification test (NAAT). Acid-fast staining is rapid, but its sensitivity is poor (approximately 10%-20%).^4,5 Mycobacterial culture shows a relatively higher sensitivity, varying from 15% to 66% in TBM,^6,7 but often takes more than 2 weeks. The Xpert MTB/RIF assay (Cepheid), a molecular assay endorsed by the World Health Organization (WHO) for tuberculosis (TB) diagnosis, can detect MTB and rifampicin susceptibility rapidly (<2 hours) using the pellet of CSF after centrifugation. The sensitivity of Xpert is 50% to 70% for TBM diagnosis.^4-6,8 To improve the sensitivity of the assay on paucibacillary samples, the Xpert MTB/RIF Ultra (Ultra) has been developed recently.⁹ The sensitivity of Ultra for TBM is greatly improved compared with the Xpert in a prospective cohort study.⁸ Ultra was recommended for TBM diagnosis by the WHO in November 2017.¹⁰ However, the increased sensitivity of Ultra is reported to have sacrificed specificity.¹¹

Cell free-DNA (cf-DNA) was first reported in human plasma by Mandel and Metais,¹² defined as DNA fragments released from cells into body fluids. Later, cf-DNA from fetuses,¹² tumor cells,¹³ and transplanted organs¹⁴ was found in plasma from pregnant women, patients with cancer, and organ transplant patients, respectively. It is also present in various body fluids, such as urine,¹⁵ CSF,¹⁶ pleural effusion,¹⁷ sputum,¹⁸ and ascites¹⁷ under normal physiologic conditions. Recently, cf-DNA from MTB has been reported in blood,^19,20 plasma,²¹ and pleural fluid.²² These studies provide new insight for diagnosing TB by detecting cell-free Mycobacterium tuberculosis DNA (cf-TB) in various body fluids. CSF is an important specimen for TBM diagnosis. However, the diagnostic value of cf-TB in CSF has not been reported. Here we prospectively evaluated the diagnostic value of the cf-TB–based test for TBM and compared the performance of the test with microscopy, mycobacterial culture, and Xpert.

Materials and Methods

Study Population

Patients suspected of having TBM were enrolled prospectively at Beijing Chest Hospital from January 2017 to October 2018. All participants or their surrogates provided a written informed consent. This study was approved by the ethical and institutional review boards for human investigation at Beijing Chest Hospital.

Clinical Classification of Meningitis

All the enrolled patients with TBM were classified according to the criteria by Marais et al,³ which excluded cf-TB detection in CSF. Definite TBM was diagnosed if MTB could be detected in the brain, spinal cord, or CSF using acid-fast staining, mycobacterial culture, or commercial NAAT. Probable TBM was diagnosed if the total diagnostic score was no less than 10 points when cerebral imaging was not available or no less than 12 points when cerebral imaging was available. Possible TBM was diagnosed if the total diagnostic score was 6 to 9 points when cerebral imaging was not available or 6 to 11 points when cerebral imaging was available.

Non-TBM was diagnosed either as a neurologic disorder without any evidence of infection or as another infection with no response to empirical anti-TB treatment.

Collection of Clinical Data

Each patient had a lumbar puncture aseptically after being admitted. For TBM, 7 to 8 mL of CSF was obtained, and the following four tests were done simultaneously: the CSF samples were sent for acid-fast staining (2 mL), mycobacterial culture (2 mL), Xpert (3 mL for pellet), and cf-TB test (3 mL of supernatant) within 2 hours after sample collection. Ziehl-Neelsen acid-fast staining was performed using the pellet of CSF after centrifugation. Mycobacterial culture was performed using the pellet of CSF on the BACTECMGIT 960 TB System (BD) according to the manufacturer’s instruction.

Then, 3 mL of CSF was used for both the Xpert and cf-TB tests. After centrifugation at 3,000g for 10 minutes, the pellet was used for Xpert, while the supernatant was used for the cf-TB test. The Xpert assay was performed on a Cepheid platform according to the manufacturer’s protocol. In brief, 2 mL of lysis reagent was added to the pellet and vortexed for 1 minute. After 10 minutes of settling, the mixture was transferred into the cartridge for the automatic test. cf-DNA was extracted using a QIAamp circulating nucleic acid kit (Qiagen). The IS6110 sequence, specific for MTB,²⁴ was amplified using the MTB fluorescent polymerase chain reaction diagnostic kit (DaAn Gene) according to the manufacturer’s instruction.

Statistical Analysis

SPSS version 20.0 software (SPSS) was used for all statistical analyses. Normally distributed data are described as mean ± SD. Nonnormally distributed data are described as median (quartile). McNemar’s test was used to compare the four diagnostic tests. Statistical significance was defined as P < .05.

Results

Characteristics of Patients

In this study, 46 patients with TBM (25 male/21 female, age 38.3 ± 19.4 years) and 22 patients without TBM (15 male/7 female, age 41.1 ± 17.2 years) were prospectively recruited, and they were all negative for human immunodeficiency virus. Patients with TBM included 14 definite cases of TBM, 25 probable cases of TBM, and seven possible cases of TBM. Patients who did not have TBM included 14 with viral meningitis, three with bacterial meningitis, two with intracranial hemorrhage, two with intracranial metastasis of a malignant tumor, and one with cerebral infarction. Detailed clinical information and CSF features of the patients are shown in Table 1 and Table 2.

Sensitivity and Specificity of the cf-TB Test

The copy number of IS6110 in CSF samples ranged from 15.4 to 5.53 × 10⁴ copies/mL, and the positive result was reported when the Ct value was 25 or more according to the instruction. The sensitivity of the cf-TB test was 56.5% (26/46) in the total TBM samples, while the sensitivities in subgroups of definite TBM, probable TBM, and possible TBM were 100% (14/14), 40.0% (10/25), and 28.6% (2/7), respectively Table 3. The sensitivity of definite TBM was significantly higher than that of probable TBM (P < .001) and possible TBM (P < .001). No positive cf-TB test result was found in the non-TBM group. Positive and negative predictive values of the cf-TB test were 100% and 52.4%, respectively (Table 3).

Comparison of the cf-TB Test With Other Diagnostic Methods

To further evaluate the diagnostic value of the cf-TB test for TBM, we also performed smear microscopy, mycobacterial culture, and Xpert with all these CSF samples. Sensitivities and specificities of these assays are listed in Table 3. Based on the clinical classifications (all patients with definite, probable, and possible TBM were considered as having TBM), the sensitivity of the cf-TB test (56.5%, 26/46) was significantly higher than that of microscopy (2.2%, 1/46; P < .001), mycobacterial culture (13.0%, 6/46; P < .001), and Xpert (23.9%, 11/46; P < .001). Specificities of these four methods were all 100%.

The sensitivities for different TBM groups using the four methods are shown in Table 4. In the definite TBM group, the sensitivity of the cf-TB test (100%, 14/14) was significantly higher than that of microscopy (7.1%, 1/14; P < .001) and mycobacterial culture (42.9%, 6/14; P = .013) but not statistically different from Xpert (78.6%, 11/14; P = .248). However, the cf-TB test gave positive results in 40.0% (10/25) of the probable TBM group and in 28.6% (2/7) of the possible TBM group, while other tests were all negative.

All these results above indicated that the cf-TB test was more accurate in diagnosing TBM than other conventional methods.

Discussion

TBM is a serious life-threatening disease worldwide, and its diagnosis is difficult due to nonspecific clinical presentation and the lack of etiologic evidence. The sensitivity of the cf-TB test in diagnosing TBM was significantly higher than that of microscopy, mycobacterial culture, and Xpert. The cf-TB test showed great potential to be an effective diagnostic tool for TBM.

The sensitivities of the cf-TB test and Xpert were higher than those of microscopy and culture. Also, the sensitivity of the cf-TB test was 2.36-fold higher than that of Xpert in this study. The possible reason may be that Xpert detects the rpoB gene, whose copy number is only one per genome, while the copy number of IS6110, the target gene of the cf-TB test, ranges from 10 to 12 in various MTB strains.²⁴ Our results showed that the cf-TB test was not markedly superior to Xpert in the definite TBM group but showed great advantages in diagnosing probable TBM. We have noticed that the sensitivity of Xpert in diagnosing TBM in this study was lower than that reported in other studies.^9,25 We think this might be due to the high proportion of probable and possible cases of TBM (69.6%; 32/46) enrolled in this study.

This study has two major limitations. First, the sample size was relatively small. TBM accounts for only 1% of all tuberculous cases,²⁶ and more studies are needed to evaluate this new method in a large cohort. Second, the cf-TB test was compared with the Xpert but not Ultra, because Ultra was not available when designing this study. Further investigations are needed to compare the performance of Ultra and the cf-TB test. In conclusion, cell-free DNA from MTB is detectable in CSF, and the cf-TB test is a rapid and accurate method for TBM diagnosis.

This work was supported by grants from Beijing Municipal Science and Technology Project (grant Z181100001918027), Beijing Municipal Administration of Hospitals’ Ascent Plan (grants DFL20151501 and DFL20181601), and Beijing Tongzhou Municipal Science & Technology Commission (grant KJ2019CX006).

Acknowledgment

We thank Xiangdong Xu, MD, from the Department of Pathology, UCSD, for his critical review of the manuscript.

References

Author notes