Diagnostic efficacy of adenosine deaminase levels in cerebrospinal fluid in patients of tubercular meningitis: A comparison with PCR for Mycobacterium Tuberculosis.

BACKGROUND: The rapid diagnosis of Tubercular meningitis (TBM) is fundamental to clinical outcome. The key to diagnosis lies in Cerebrospinal fluid (CSF) analysis and radiological investigations. There are numerous lacunae in the confirmation of diagnosis of TBM from CSF.
PURPOSE: The aim of present study was to compare the efficacy of CSF adenosine deaminase (ADA) level assays and Polymerase chain reaction (PCR) for Mycobacterium tuberculosis (M. tuberculosis) in the diagnosis of TBM.
METHODS: Fifty four adult patients with suspected TBM and 37 controls were included in the study and CSF analyzed for ADA and PCR for M. tuberculosis. The cases were subdivided into definite (5), highly probable (22), probable (22) and possible TBM (5) as per previously validated criteria. The first two were grouped as "most likely" TBM (27) and last two as "unconfirmed" TBM (27).
RESULTS: The mean ADA of the "most likely" TBM was 29±24, "unconfirmed" TBM was 21 ± 15 and controls were 4.8±2.2 U/L. The ADA levels correlated with CSF proteins, absolute lymphocyte count and the staging of the disease. Using a cut off level of >L10 U/L, CSF ADA had a sensitivity of 92.5% and specificity of 97%. PCR for M. tuberculosis was positive in 12 out of 27 "most likely" TBM cases, 5 out of 27 "unconfirmed" TBM cases and 3 out of 37 controls. PCR for M. tuberculosis had a sensitivity of 44.5% and specificity of 92% in the "most likely" TBM cases.
CONCLUSIONS: ADA is a rapid, inexpensive and sensitive test in the diagnosis of TBM. It is more sensitive than AFB smear and culture. PCR is another rapid test in the diagnosis of TBM with a good specificity, even in those patients already on presumptive anti-tuberculous treatment. However, despite the sensitivity and specificity of CSF ADA, it should be corroborated with AFB smear and CSF PCR.

Introduction

TBM is the most dangerous form of extra pulmonary TB and occurs in about 7-12% of all TB patients in developing countries. Despite availability of effective chemotherapy, mortality and morbidity remain unacceptably high. Delay in the diagnosis and institution of proper treatment is directly related to poor outcome and sequelae, which are severe in 20-25% of cases.[1] Early diagnosis by demonstration of AFB in CSF smear is not consistently positive, since there are only a few organisms in the CSF sample. The delay of 6-8 weeks to obtain culture results and 10-14 days with rapid culture is critical in planning treatment.[2] In smear negative cases with high clinical suspicion of TBM, other rapid diagnostic methods need to be relied upon, or treatment has to be started on the basis of presumptive diagnosis.

ADA, a polymorphic enzyme involved in purine metabolism is found to be elevated in the CSF of TBM patients and gradually returns to normal values after 2-6 weeks of specific treatment. The estimation is easy, fast, inexpensive and can be done in ordinary laboratories. [3] PCR is a powerful tool for rapid diagnosis of infectious diseases. Its advantage lies in the fact that the starting DNA is amplified several times, leading to a high sensitivity. The decrease in specificity due to contamination is a cause for concern. Several authors concluded that PCR cannot be used to start or stop ATT in suspected cases of TBM. [4],[5] With improvement in methods to reduce contamination and by increasing the sample volume, both specificity and sensitivity were found to be increased.[6] The present study compares CSF ADA levels and PCR as rapid methods in diagnosis of suspected TBM.

Methods

Informed consent was obtained as per Institute Ethical Committee before study. Fifty four adult patients with suspected TBM and thirty seven controls were included in the study and the CSFwasanalysedforADAand PCRforM. tuberculosis. The cases were evaluated in detail and subdivided into definite, highly probable, probable and possible TBM cases as per in the clinical criteria (Table 1).[7] ATT and steroids were started in all cases and response evaluated. Controls were taken from patients in Neurology ward in whom TB was not suspected and CSF was analysed for diagnostic purposes like Acute Inflammatory Demyelinating Polyradiculoneuropathy, Multiple Sclerosis, Polyneuropathy, Idiopathic Intracranial Hypertension, etc. ADA estimation in CSF was done using the method of Guisti.[8] PCR was done on CSF placed in microcentrifuge tube (in aliquots) and centrifuged for 10 minutes at 10,000 rpm. To the pellet formed after centrifugation (which contains cells) 20jul of 0.1 %Triton-x-100 is added and incubated for 15 minutes. Primers as used by Scarpellini et al[9] were used for amplication of a conserved repetitive insertional sequence IS 6110.

Table 1:

Criteria for Diagnosis of TBM [7]

I.	Clinical symptoms and signs
	Mandatory:		Fever and headache >2 weeks
	Optional:		Vomiting, neck stiffness, altered sensorium, seizures or focal Neurologic deficit
II.	Supporting critera
	1.	CSF:	Cells : >20/cmm, lymphocytes : >60% proteins : 100 mg%, sugar: <60% of corresponding blood sugar, negative gram stain, India stain and VDRL where relevant
	2.	CECT/MRI	Showing one or more of : exudates in basal cisterns / Sylvian fissures, gyral enhancement, hydrocephalus, Infarcts, tuberculoma
	3.	Active extraneural TB : as evidenced by appropriate mycobacterial tests, radiology, histopathological examination
	4.	Clinical response to ATT and relief of symptoms
The cases were classified as being:
•Highly probable TBM (met clinical criteria and 3 of the 4 supporting criteria)
•Probable TBM (met clinical criteria and 2 of the 4 supporting criteria)
•Possible TBM (met clinical criteria and 1 out of 4 supporting criteria)
•Confirmed TBM (AFB stain or/and culture of CSF was positive)

35 cycles of PCR were carried out (denaturation at 95°C, annealing at 56°C, and extension at 72°C for 1 minutes each). PCR product was analysed by gel electrophoresis. Genomic DNA of M. tuberculosis was used as a positive control. Statistical analysis was done using Pearsons co-efficient of correlation, chi-square test and student 't' tests for comparison, with p value <0.05 considered as significant.

Results

All the patients had fever for a mean duration of 52.3±73.1 days (median 30). Headache was present in all patients and 49 patients (91%) had vomiting. Forty seven patients (87%) had alteration in sensorium of a mean duration of 6±4 days. Seizures were present in 18 patients (33%) with focal seizures in five. Signs of meningeal irritation were present in 50 patients (93%).

Papilledema was seen in 17 patients (31 %) and abducens nerve palsy was observed in 17 patients (31%) of which it was bilateral in eight. Thirty one patients had ventriculomegaly and 11 of them required ventriculoperitoneal shunt insertion. Eighteen patients had other cranial nerves palsies (oculomotor 17%, facial 13%, vagus 4%, optic 2%, vestibulocochlear 2%, glossopharyngeal 2%). Eight patients had multiple cranial nerve palsies. Five patients had hemiparesis. Chest X-ray abnormalities were seen in 15 patients (28%) (military mottling 8%, cavities 8%, hilar nodes 6%, pleural effusion 6%). CECT head showed basal exudates in 41%, hydrocephalus in 39%, infarcts in 13% and tuberculoma in 13% of all cases studied. No abnormality was detected in cranial imaging of 7 patients (13%). Twelve patients had received ATT prior to admission to hospital and five of them had received the treatment for more than one week. All patients were started on ATT. Five patients developed drug induced hepatitis and were managed accordingly. Four patients expired while in the hospital.

The CSF cell count ranged from 20-1320 cells per cu.mm with a mean of 275.5±357 cells (median 145). The absolute lymphocyte count was 153±178 (median 102). The mean protein level was 297±350 mg% (median 150) and the mean CSF sugar level was 37.3±17.1 mg% (median 38.5). The ratio of CSF glucose to blood sugar was 0.33±0.14 (median 0.32). AFB smear was positive in only one patient and AFB culture was positive in 5 patients (9%). The mean CSF ADA level was 26.1 ± 18.8 μ/L (median 19). The major CSF findings of the TBM groups are shown in Table 2.

Table 2:

CSF findings in the various subgroups of TBM

	All cases of TBM (n=54)	Definite (Culture +ve) (n=5)	Highly probable (n=22)	Probable (n=22)	Possible (n=5)
Total cells (per cumm) Mean±SD (Median)	276±357 (145)	338±401 (300)	290±384 (165)	260±355 (145)	135±177 (60)
Absolute lymphocyte count (per cumm) Mean ± SD (Median)	153±178 (102)	135±147 (100)	198±226 (140)	116±123 (82)	98±123 (45)
Protein (mg%) Mean±SD (Median)	297±350 (150)	163±81 (125)	465±452 (300)	214±241 (92)	89 ±34 (800)
Ratio of CSF sugar to blood sugar Mean±SD (Median)	0.33±0.14 (0.32)	0.32±0.12 (0.28)	0.29±0.12 (0.28)	0.35±0.14 (0.33)	0.48±0.18 (0.48)
CSF ADA levels (U/L) Mean±SD (Median)	26±19 (19)	23.2±8.6 (19)	30.3±24.8 (21)	24.0±15.9 (19)	14.0±4.5 (14)
PCR positivity	17 (32%)	4 (80%)	8 (36%)	5 (23%)	0 (0%)

CSF ADA showed a positive correlation with absolute lymphocyte count and V value was 0.15 using Pearson's coefficient of correlation. However, this was not significant on the students 't' test. CSF ADA showed a stronger positive correlation with CSF protein ('r' value 0.55 using Pearson's formula). This was statistically significant (p <0.01). With increasing severity of disease (as per the Medical Research Council Staging), the CSF ADA showed a increasing trend, but it was not statistically significant. There was no significant correlation with the duration of disease. The ADA level tended to decrease within one week of ATT. The ADA levels decreased further with ATT of more than one week and the difference was statistically significant p<0.02).

Overall, PCR positivity was seen in 17 cases of TBM, it being positive in 4 out of 5 cases of definite TBM (80%), 8 out of 22 cases of highly probable TBM (39%), 5 out of 22 cases of probable TBM (17%) and none of the 5 cases of possible TBM.

Out of the 37 patients in the control group, 10 patients had demyelinating illness, 5 each had polyneuropathy and idiopathic intracranial hypertension, 3 had acute inflammatory demyelinating polyradiculoneuropathy, 2 had subacute sclerosing Panencephalitis and 12 had other neurologic disorders. The major CSF findings of the control groups are shown in Table 3.

Table 3:

Control CSF samples (n=37) showing the CSF findings

	All control (n=37)	Demyelination (n=10)	Idiopathic Intracranial Hypertension (n=5)	Polyneuropathy (n=5)	Acute Inflammatory Demyelinating Polyneuropathy (n= 3)	SSPE (n=2)	Others (n= 12)
Total cells (per cumm) Mean±SD (Median)	4.6±26.3 (br/>(0)	-	-	-	-	-	14±46 (0)
Absolute lymphocyte count(per cumm) Mean ± SD (Median)	3.7±3.5 (0)	-	-	-	-	-	11.5±36.8 (0)
Protein (mg%) Mean±SD (Median)	46.5±29.5 (40)	37±5 (38)	24±8 (25)	74±50 (25)	52±11 (56)	40±14 (40)	52±34 (40)
Ratio of CSF sugar to blood sugar Mean±SD (Median)	0.62±0.14 (0.62)	0.66±0.14 (0.66)	0.60±0.09 (0.63)	0.6±0.14 (0.66)	0.6±0.14 (0.60)	0.67±0.08 (0.67)	0.58±0.18 (0.60)
CSF ADA levels (U/L) Mean±SD (Median)	4.8±2.2 (5)	4.0±1.6 (3)	4.4±1.3 (5.0)	5.0±1.9 (6)	5.3±0.6 (5)	5.0±0 (5.0)	5.4±3.3 (5.0)
PCR (false positivity)	3/37 (8%)	-	-	1/5 (20%)	- -	- -	2-12 (17%)

The comparison between CSF ADA levels of TBM group and controls was statistically significant (p <0.001). The comparison of PCR in the TBM cases with control group was statistically significant (p <0.01). The difference between highly probable TBM and unconfirmed TBM cases was also statistically significant (p <0.05).

The sensitivity and specificity of PCR was 37% and 92% respectively in the diagnosis of TBM. The sensitivity was 44.5% for the most likely TBM cases. The positive and negative predictive values were 85% and 48% respectively.

Discussion

Early treatment of TBM is essential to prevent both morbidity and mortality. Rapid diagnostic tests with good sensitivity and specificity are required to aid the presumptive diagnosis, as AFB staining is not sensitive enough to help the clinician in ruling out the possibility of TBM. The CSF smear was positive for AFB in only one patient in the study group. Since the sensitivity is very low, it cannot be relied upon to pick up cases of TBM. Meticulous examination of smear from large volume CSF samples and serial taps is reported to yield better results.[6] The CSF culture grew AFB in 5 patients (9%) in this study. This is in comparison with the previous studies by Shankar et al(12%)[10], Miorner et al(17%)[11] and Nguyen etal(19%)[12]. Even though this is the gold standard, the delay of 6-8 weeks to obtain a positive result coupled with the low sensitivity makes this test less useful to the clinician in the diagnosis of TBM.

The mean ADA in TBM group was 26.1 ±18.8 U/L which is comparable to that in adults with TBM observed by Malen et al(26.2 U/L).[3] The CSF ADA levels correlated with absolute lymphocyte counts and CSF proteins (more with the latter), as reported in previous studies.[3],[13] ADA levels tend to decrease with therapy significantly if treatment duration was beyond one week (p<0.02). A cut off value of >.10 U/L gives a good sensitivity and specificity. The overall results are in comparison with the previous studies as shown in Table 4. However, elevated levels of CSF ADA are not specific for TBM. Diseases like pyogenic meningitis, CNS lymphoma, and fungal meningitis were shown to have elevated CSF ADA.[3]

Table 4:

Comparison of Sensitivity and Specificity of CSF ADA in the diagnosis of TBM in the present study with previous studies

Sensitivity (%)	Specificity (%)	Year (Ref)
90	NA	Malan et al, 1984
73	71	Coovadia et al, 1986
99	100	Ribera et al, 1987
50	90	Kaur et al, 1992
44	75	Gambhir et al, 1999
87	86	Pushpa et al, 2000
92.5	97	Present Study

In the most probable TBM group, the sensitivity of PCR was 44.5% and specificity was 92%. The sensitivity of PCR is low when compared with most of the previous studies (Table 5). In the present study, both TBM and control samples were marked as suspected TBM and thus were blinded for processing of PCR. Previous studies tested a smaller number of patients and used a selected patient group[7] or had a higher number of false positives in PCR.[10] Out of the 5 culture positive cases, one was PCR negative. This discrepancy can be explained by the differences in the volumes of sample tested. It would thus be advisable to send a large volume of CSF for PCR analysis. False negative reactions can also be caused by inhibitors of Taq polymerase,[4] when the CSF is contaminated with blood. Contamination of CSF sample with blood may not always be detected by the naked eye. A better DNA extraction procedure that could take care of the inhibitory proteins in the CSF may be helpful in increasing the sensitivity of PCR.

Table 5:

Comparison of Sensitivity and Specificity of PCR in diagnosis of TBM In the present study with the previous studies

Sensitivity (%)	Specificity (%)	Study Year (Ref)
83	100	Kaneko et al, 1990
65	89	Shankar et al, 1991
76	100	Lin et al, 1995
48	100	Kox et al, 1995
54	94	Miorner et al, 1995
82	100	Nguyen et al, 1996
60	100	Bonington et al, 1998
44.5	92	Present Study

Prior treatment with ATT decreased the sensitivity of PCR in this study. Still, there were 29% PCR positive cases, in those treated for a week and one case was positive even after 3 months of ATT. Lin etal concluded that it is still valuable to apply PCR in clinically suspected TBM who have already received a therapeutic trial of ATT even upto 3 weeks.[19] In a study by Kaneko etal, a CSF sample was positive for TB even after one month of ATT.[18] The persistent presence of M. tuberculosis DNA in CSF might indicate that ATT induces release of mycobacterial DNA in CSF. This is especially important in tertiary care centers, where many cases at presentation are already on presumptive ATT which further decreases the chances of obtaining an AFB stain or culture. It can be suggested that PCR is still useful in those who have already received a trial ATT, even upto 3 months.

Some studies have already shown that a nested PCR protocol could also improve the sensitivity of detecting mycobacterial DNA in clinical samples.[22] Conventional method using a one step amplification of DNA is associated with a low sensitivity. Two step nested amplification of DNA enhances the sensitivity by several folds.[22] Even though nested amplification protocol could not be used in our study due to technical reasons, this can be recommended for further studies as it has been shown to increase the sensitivity.

In a recent study from Vietnam, it was shown that the number of copies of insertional sequence IS6110 in 168 strains of M. tuberculosis varied from 0 to 23. Three strains from North Vietnam lacked IS6110 strains and other had 3 to 14 copies of IS6110 while Southern strains had 15-23 copies of IS6110. If the strain does not have IS6110, then PCR cannot give a positive result even when there is bacterial DNA in the CSF. Whether similar variations in insertional sequence exist in our patients require further studies and confirmation in each areas.[23]

In the present study, CSF samples from three controls were positive by PCR. Since these patients had no cells in the CSF, silent tubercular lesions in the meninges are unlikely in the absence of symptoms. These positive results may be attributed to possible cross contamination of samples during initial handling, as has been reported in the literature.[12],[18],[21]

Conclusion

ADA is a rapid, inexpensive and sensitive test in the diagnosis of TBM. It is sensitive than AFB smear and culture and can be suggestive of the diagnosis of TBM. PCR is a rapid test in the diagnosis of TBM with a good specificity. Even in those patients already on presumptive ATT upto 3 months, PCR can be helpful. By doing PCR with a larger volume of uncontaminated CSF, better DNA extraction techniques and a nested PCR protocol, the sensitivity can be increased. CSF ADA level • 10 μ/L is sensitive and can suggest the diagnosis of TBM, especially if the clinical suspicion is high. Other techniques like AFB smear, PCR for M.tuberculosis and rapid culture would be required for confirmation of the diagnosis.