INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Tuberculous meningitis (TBM) is the most serious form of extrapulmonary tuberculosis. The disease is frequently fatal if it is not treated early, and in surviving patients neurologic sequelae are common (6, 9-12, 26). In developing countries, TBM occurs in 7 to 12% of persons with active Mycobacterium tuberculosis infection (25). In developed countries, TBM occurs less frequently, but the incidence is increasing in association with the human immunodeficiency virus epidemic (3, 21).

The diagnosis of TBM is difficult due to the low numbers of mycobacteria in the cerebrospinal fluid (CSF) of patients with this disease. In several reported series of TBM in children, acid-fast smears and cultures of the first samples of CSF collected were positive for only 8 to 10% and 29 to 48% of the patients, respectively (5, 17, 19, 27). Tuberculin skin testing is also of limited diagnostic value for this condition (3, 17, 27).

Attention has focused on the development of alternative, rapid, accurate methods for the detection of M. tuberculosis in the CSF. Both indirect and direct assays have been evaluated. Examples of indirect assays include the assessment of the activity of adenosine deaminase, an enzyme that is produced by T lymphocytes and whose levels are elevated in patients whose CSF is infected with M. tuberculosis (sensitivity range, 60 to 100%; specificity range, 84 to 99%) (22, 23), the determination of the ratio of the concentration of bromide between the serum and the CSF after a loading dose of bromide is given (the ratio of the concentration of bromide in the serum to that in the CSF decreases with disruption of the blood-brain barrier, as occurs with TBM) (sensitivity, 80%; specificity, 95%) (29), and the measurement of antibody to tuberculin (purified protein derivative) in the CSF (sensitivity, 24%; specificity, 98%) (28). Potentially useful direct tests which have been evaluated include the analysis for mycobacterial structural components such as tuberculostearic acid (sensitivity, 95%; specificity, 98%) (8) or the detection of mycobacterial antigens (sensitivity, 39 to 100%; specificity, 96 to 99%) (10, 13, 28). However, all of these indirect and direct assays are relatively difficult to perform, lack standardization, and are not commercially available in kit formats.

Several investigators have demonstrated the utility of in-house nucleic acid amplification techniques for the detection of M. tuberculosis complex in CSF (7, 14, 15, 24). One commercial kit which is currently available for the detection of M. tuberculosis complex organisms in respiratory samples is the Gen-Probe Amplified Mycobacterium tuberculosis Direct Test (MTD; Gen-Probe, San Diego, Calif.). The objective of the current study was to determine the ability of the Gen-Probe MTD to detect M. tuberculosis complex organisms in 84 CSF samples.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Subjects. The subjects were children less than 15 years old who presented with signs and symptoms of meningitis from March 1995 through March 1996. All of the children were evaluated and followed throughout their admission at the Dr. Robert Reid Cabral Pediatric Hospital, a 350-bed university teaching hospital located in Santo Domingo, Dominican Republic. There were 358 cases of meningitis during this time period. Twenty-nine (8.1%) were ultimately diagnosed as TBM. Inclusion into the study was based on the ability to obtain an adequate sample for laboratory testing as described below and a complete clinical history, which consisted of symptoms and duration of illness, prior treatment, contact history, Mantoux skin test result, and response to treatment.

Clinical specimens. CSF samples (n = 84) were obtained by lumbar puncture. CSF samples were centrifuged at 2,500 × g for 15 min, and the sediment was evaluated as described below.

Bacterial antigen testing. All samples were screened by latex agglutination testing for Streptococcus pneumoniae, Haemophilus influenzae type b, Neisseria meningitidis groups A, C, Y, and W135, group B streptococcus (Streptococcus agalactiae), and Escherichia coli K1 and N. meningitidis group B according to the manufacturer's instructions (bio-Merieux, St. Louis, Mo.).

Bacterial cultures. All CSF specimens were cultured for general bacteria by inoculating sediment onto a blood and a chocolate agar. These plates were incubated at 35 to 37°C in CO₂ for 2 days and were examined daily for bacterial growth. Bacteria were identified by standard techniques.

Definition of suspected TBM. Patients with suspected TBM were those with fever and stiff neck for longer than 2 weeks, with analysis of CSF showing pleocytosis (>10 leukocytes/ml), an elevated protein concentration (>40 mg/dl), and an amount of glucose in CSF less than 60% of the amount of glucose in serum, and with negative bacterial antigen tests and cultures. In addition, the patients had to be positive for at least two of the following supporting criteria: (i) close contact (i.e., at least 2 days per week in the same domicile) with a patient with a known case of active tuberculosis, (ii) positive Mantoux skin test (i.e., induration, >5 mm), (iii) microbiologic or radiographic evidence of active tuberculosis at an extraneural site, (iv) clinical response to antituberculous drugs, or (v) cranial computed axial tomography showing central nervous system densities or basilar exudates. A clinical response to antituberculous drugs was defined as clinical improvement on a quadruple-drug regimen of isoniazid, rifampin, streptomycin, and pyrazinamide at standard doses. This improvement could occur following a lack of clinical improvement to standard dual-antibacterial therapy of chloramphenicol and ampicillin or when this standard antibacterial therapy was not used.

Mycobacteria cultures. For patients with clinically suspected TBM, CSF sediment was resuspended by adding approximately 3 ml of prepared phosphate buffer at pH 6.8, and then an aliquot (500 µl) was inoculated into a single Lowenstein-Jensen slant tube and the tube was incubated at 35 to 36°C for 12 weeks. Any growth on this medium was confirmed as M. tuberculosis by Ziehl-Neelsen staining and conventional biochemical testing (18). An additional amount of CSF sediment was heat fixed onto a glass slide and was stained by the Ziehl-Neelsen method.

MTD. Portions of all CSF sediments were transported overnight on dry ice to the Mayo Clinic Microbiology Laboratory located in Rochester, Minn. Upon receipt, the samples were placed in a freezer at 70°C until they were further analyzed. CSF samples were processed and analyzed according to the manufacturer's instructions for sputum, but with the following modifications adapted from Pfyffer and colleagues (20): (i) whenever possible, the amount of sample was increased 10-fold (500 µl was used instead of 50 µl; if there was an insufficient amount of sample, the total volume was increased to 500 µl with sterile water); (ii) the sample was pretreated with a denaturation agent (3.26% sodium dodecyl [lauryl] sulfate [SDS], 1.0% NaOH), (iii) and the nucleic acid amplification time was increased from 2 to 3 h. Cutoff values for a positive result were 30,000 relative light units (RLUs), as recommended by the manufacturer. Lower cutoff values for a positive result were also assessed.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

None of the CSF samples had positive smears by Ziehl-Neelsen staining. M. tuberculosis was isolated from five specimens. M. tuberculosis was not recovered from 19 specimens, but these were from patients who fulfilled the criteria for probable TBM. Table 1 presents the specific supporting criteria for each of the 19 patients with probable TBM listed according to RLU value. Three of the 19 patients met two of the supporting criteria, 12 of the 19 patients met three of the supporting criteria, and 4 of the 19 patients met four of the supporting criteria. All CSF specimens from patients with probable TBM had elevated leukocyte counts (mean, 435 × 10⁹ ± 326 × 10⁹/liter), elevated protein concentration (mean, 206 ± 100 g/dl), and a CSF glucose concentration less than 60% of the blood glucose concentration (mean, 35 ± 11 mg/dl). Eighty percent of the patients with probable TBM had a lymphocytic predominance in their CSF.

In no patient was an M. tuberculosis complex organism detected by the Gen-Probe MTD when criteria for probable TBM were not fulfilled or when other bacteria were isolated or other infectious agents were felt to be responsible for meningitis. Therefore, the MTD was 100% specific (Tables 2 and 3). For cases in which M. tuberculosis was isolated from the CSF or patients fulfilled the criteria for probable TBM, the sensitivity of the MTD was 33%. Lowering of the cutoff value for a positive result to approximately one-third of the recommended value improved the sensitivity of the MTD. Not shown in Table 3 are additional calculations for sensitivity and specificity when the cutoff values were decreased further. These calculations demonstrated that the sensitivities of the MTD remained essentially unchanged, but the specificities were appreciably worse.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Cultures of spinal fluid for M. tuberculosis are frequently negative for patients with TBM. Therefore, in the current study, the results of the Gen-Probe MTD were compared to those of a "gold standard" arbitrarily defined as either a positive CSF culture result for M. tuberculosis or clinical or laboratory criteria supporting a diagnosis of probable TBM. These criteria have been reported elsewhere and include contact history (exposure to persons with active tuberculosis), skin testing, the presence of extraneural tuberculosis, response to antituberculosis treatment, characteristic cranial computed tomography findings, and the results of CSF cellular and chemistry analyses (1, 4, 5, 17). In the current study, all patients that we felt had probable TBM had negative results for bacterial antigen tests and negative cultures for general pathogenic bacteria but had elevated leukocyte counts, elevated protein concentrations, and decreased glucose concentrations in the spinal fluid and were positive for at least two of the clinical criteria given above.

The results of our study demonstrate that the Gen-Probe MTD is a useful rapid test for the diagnosis of TBM. Four of the five CSF samples that grew M. tuberculosis were positive by the Gen-Probe MTD. An additional four samples were positive by the Gen-Probe MTD; all of these patients were felt to have probable TBM. It is possible that M. tuberculosis would have been isolated by culture from these four samples had liquid medium been used, although this was not available at the site where the samples were collected in the Dominican Republic. Due to the cost, space, and technology required, the use of liquid medium-based culture systems was not possible. The sensitivity of this rapid molecular test was greater than that of acid-fast smear examination; none of the CSF samples that were evaluated had positive acid-fast smear results.

The sensitivity of the Gen-Probe MTD was significantly increased (33 to 83%) and the specificity remained unchanged (100%) when the cutoff value for a positive result was decreased from 30,000 RLUs (used for respiratory specimens) to 11,000 RLUs. Further decreases in the cutoff values resulted in essentially no change in sensitivity, but the specificities were appreciably worse.

The specimen processing that we used for the Gen-Probe MTD for CSF samples was modified from that recommended for sputum samples. These modifications were according to a recent study by Pfyffer and colleagues (20). Those investigators showed that the pretreatment of CSF with the denaturing agent SDS-1% NaOH and an increase in the amplification time from 2 to 3 h improved the sensitivity of the Gen-Probe method. SDS denatures protein and enzymes which likely inhibit the nucleic acid amplification reactions. Also in accordance with the recommendations of Pfyffer and colleagues (20), whenever possible we processed 10 times the volume of CSF suggested by the manufacturer (500 versus 50 µl). For 4 of the 19 patients with probable TBM, 50 to 250 µl of CSF was processed; for 2 of the 5 patients culture positive for TBM, 250 µl was processed. Five hundred microliters was processed for each of the 13 remaining patients with probable TBM. These aliquots of CSF were treated with SDS-NaOH and centrifuged, the sediment was resuspended, and the usual test protocol was then followed. Pfyffer and colleagues (20) demonstrated that these modifications enhanced the sensitivity of the MTD for the detection of M. tuberculosis complex organisms in CSF spiked with the organism. Positive MTD results could be obtained for CSF spiked with as few as about five mycobacterial cells per ml (20).

The results that we obtained by the Gen-Probe MTD were in agreement with those of a smaller clinical evaluation published recently by Pfyffer and colleagues (20). They demonstrated that the Gen-Probe MTD was more sensitive than acid-fast smears for the detection of M. tuberculosis in six CSF specimens which were obtained from patients with clinical symptoms and chemical parameters compatible with TBM. One of these six CSF samples was acid-fast smear positive; five grew M. tuberculosis. Pfyffer and colleagues (20) evaluated fresh CSF samples by the Gen-Probe MTD method. It is possible that in our study the sensitivity of the Gen-Probe MTD method would have been enhanced if fresh specimens had been provided and specimens of 500 µl had consistently been evaluated. Freezing and thawing may theoretically affect RNA stability, which may ultimately result in decreased RLU values. Because of a lack of resources on site in the Dominican Republic, CSF samples were frozen, transported to the Mayo Clinic laboratory, and then evaluated.

Commercially prepared, standardized rapid test methods whose performance is equal to or exceeds the performance of acid-fast smears for diagnosing TBM are not available. The results of this study and the study of Pfyffer and colleagues (20) demonstrate that for the diagnosis of TBM, the Gen-Probe MTD is superior to acid-fast staining. Savings in reagent costs and personnel time could be realized if the Gen-Probe MTD were used in place of acid-fast staining for the rapid diagnosis of this disease. Like acid-fast staining, the Gen-Probe should be used only for those patients in whom TBM is suspected (2, 16). The Gen-Probe MTD is approved by the U.S. Food and Drug Administration and is commercially available for the testing of respiratory specimens. Modification of this test method for CSF specimens to include changes in sample preparation and interpretive changes in cutoff values should facilitate the rapid diagnosis of TBM, a frequently fatal illness.