Original Articles Indian Pediatrics 2000;37:375-382 |
||||||||||||||||||||||||||||||||||||||||||
Polymerase chain reaction in the diagnosis of tuberculosis |
||||||||||||||||||||||||||||||||||||||||||
Sunil Kumar Jatana,
Key words: Mycobacterium tuberculosis, Polymerase chain reaction, Mantoux (Mx) test, Tuberculosis (TB). Tuberculosis is a major cause of morbidity and mortality throughout the world in both industrialized and developing countries inspite of availability of anti tubercular therapy (ATT) for many years. In 1990, an estimated 8 million people developed TB worldwide with 2.6 to 2.9 million deaths. The majority of these cases occurred in Asia and Africa with an increasing number among HIV infected individuals(1). One of the major obstacles of diagnosing this disease in children is the absence of a sensitive, specific and rapid method of diagnosis. Clinical signs and symptoms are less helpful, as more than 50% show no symptoms at all at the time of presentation(2). This hampers the early initiation of therapy or results in over-diagnosis and unnecessary treatment. TB is difficult to diagnose in children because of poor yield on standard diagnostic laboratory tests and the lack of characteristic symptoms in majority of the children in comparison to adults(3). The "gold standard" of diagnosis of TB in adults is isolation of Mycobacterium tuberculosis, which unfor-tunately, may not be applicable to children. Conventional diagnostic tests, including acid-fast stain and mycobacterial culture, are frequently negative in children with TB disease, because lower M. tuberculosis counts and poor tussive force makes adequate specimen collec-tion difficult in this age group. M. tuberculosis culture takes a longer time, which limits its usefulness in the initial diagnosis(4). Even under the best of conditions, the yield from gastric aspirate cultures in children with TB is very poor and acid fast stains of specimens are almost never positive(2). The epoch making discovery of DNA amplification by polymerase chain reaction (PCR) has altered the entire approach to identification of pathogens, which grow with difficulty or take a long time to grow. This has enabled early detection of mycobacterium DNA directly from clinical specimens, especially of CNS origin which are paucibacillary(5). PCR using insertion sequence IS 6110 as the target, has the potential to overcome limitations of conventional methods and be established as a rapid, sensitive and specific technique for detecting DNA of M. tuberculosis in various clinical specimens(6,7). IS 6110 has been said to be specific to M. tuberculosis as most strains of M. tuberculosis have between 8 and 15 copies(8-10). Studies evaluating the application of this technique to the diagnosis of TB in children have been far and few. We therefore evaluated PCR as a tool to diagnose TB (mainly CNS TB) at an early stage in children in Indian conditions.
From July 1997 to January 1999, patients between the age of 3 months to 10 years admitted in Command Hospital, Central Command, Lucknow were selected for the study. The investigations were conducted in the Command Pathology Laboratory, where a PCR laboratory had been set up for diagnosing cases of M. tuberculosis. Eighty clinical specimens (3 gastric aspirate samples taken from a single patient on 3 consecutive days were considered as one sample) were tested, 41 from children suspected to be suffering from TB and 39 from controls. In probable TB cases, 27 CSF samples were tested in CNS cases, 4 pleural fluid specimens in pleural effusion cases and 10 gastric aspirate samples in cases showing pulmonary infiltrates. Out of 39 control samples, 4 were ascitic fluid, 6 were gastric aspirates, 3 pleural fluids and 26 were CSF. Diagnosis of control cases was totally unrelated to TB. Ascitic fluid was taken from 2 cases of viral hepatitis and 2 cases of portal hypertension. Gastric aspirate specimens were collected from patients admitted for routine surgery, i.e., congenital hydrocele and hernia. Pleural fluid was tapped from 1 case of chronic renal failure and 2 cases of nephrotic syndrome. All samples were transudates. CSF samples were taken from 10 cases of congenital hydrocephalus, 7 cases of febrile seizures, and 9 cases of aseptic meningitis. More CNS cases were included for analysis in this study because it is extremely difficult to isolate and culture M. tuberculosis on CSF specimens and exact diagnosis of CNS TB is difficult. The controls were patients whose diagnosis was unrelated to TB. Cases of probable TB were identified if they met the (1 or 2 + 3) following criteria (based on our criteria):
All the clinical specimens of suspected cases were subjected to AFB smear and culture on LJ medium. To avoid false positivity arising from cross contamination, DNA extraction and PCR was done in a separate laboratory away from culture laboratory. A bio-safety laminar flow hood and separate set of pipettes were reserved for PCR(11). Extraction of DNA(12): 200 ml of sample was mixed with 200 ml of lysis buffer (5.3 M Guanidine isothionate, 10 mM Diethyribiol, 1% Tween 20, 0.3 M Sodium acetate, 50 mM Sodium citrate) and the mixture was incubated at 65° C for 10 minutes. 50 ml of glass mixing powder was added to the solution and incubated at room temperature with occasional shaking. Thereafter, it was centrifuged at 1600 g for 1 minute. Supernatant was discarded and matrix re-suspended in wash buffer (50% ethanol, 10 mM Tris HCl, 100 mM NaCl). Bound DNA was eluted by periodic mixing in 100 ml of 10 mM Tris HCl. Primers: The insertion sequence IS 6110 was the target for the PCR. Primers T4 , T5 were selected for amplification of a 123 base pair nucleotide sequence in IS 6110. The sequence was as follows:
The buffers, Taq polymerase enzymes and custom synthesized primers were all obtained from Bangalore Genei Ltd, Bangalore and used for PCR. Amplification of DNA: The reaction buffers dNTPs and Taq polymerase enzymes were all purchased from Bangalore Genei Ltd. PCR reaction was done in 50 ml vol. Reaction master contained 10X reaction buffer (5 ml/sample), dNTPs (200 ml/sample), forward and reverse primers (1 ml/sample) and Taq polymerase enzyme (1 unit/sample). Amplification was carried out on a Cambridge Techne thermo-cycler with a heated lid. The hot start method was employed at 94° C for 2 minutes initially. Thereafter, amplification was carried out for 35 cycles at –90° C for 1 minute (denaturation), 60° C for 1 minute (annealing) and 72° C for 1 minute (extension). Extra extension was carried out at 72° C for 10 minutes. The ampli-fication products were analyzed on 2% agarose gel. The bands were visualized by staining with ethidium bromide under ultraviolet light. Positive and negative controls were run with each batch or sample analyzed. The following precautions were taken to avoid false negative cases in the study(13): (a) Use of guanidine isothiocyanate in the lysis buffer, during pre-treatment of samples for DNA extraction; (b) Testing of several speci-mens; and (c) All specimens were concentrated by centrifugation before proceeding for analysis.
A total of 80 children were evaluated, 41 cases of probable TB and 39 as controls. Out of 41 cases of probable TB, there were 8 cases of frank meningitis, 19 cases of probable CNS TB, based on clinical CSF and CT scan findings. There were 4 cases of pleural effusion and 10 cases probably suffering from pulmo-nary TB based on clinical and radiological criteria and had not improved with treatment with appropriate antibiotics. One of these patients showed cavitation on radiological examination. Nineteen cases (other than frank meningitis), of probable CNS TB were positive for PCR and CSF repeated after 10 days did not show any change. Seventeen out of 19 cases had shown positive tuberculin test while the other two cases were given BCG diagnostic test which showed an accelerated reaction. All the 19 cases were treated with ATT and showed improvement in clinical and CSF findings. All cases of probable TB meningitis showed CSF findings consistent with TB and PCR was positive in all of them. Two of the 8 cases were of disseminated TB showing non-homogenous opacities on radiological examination of the chest. CECT done later showed basal exudates. All these cases were put on ATT and inspite of that 4 of them developed obstructive hydro-cephalus. The CSF samples did not show AFB on staining nor could M. tuberculosis be grown on culture with LJ medium. Considering the paucibacillary nature of CNS TB with poor yield of AFB from CSF, these 18 cases and 9 cases of probable TB meningitis were considered to be having TB, based on clinical, radiological, CSF findings, Mx/BCG testing, positive PCR report and improvement on exhibiting ATT. All 4 cases of pleural effusion were positive with PCR but M. tuberculosis was grown on culture on only one specimen while in the rest of the specimens there was no yield on culture. None of the samples showed AFB on smear. Three out of 4 pleural effusion cases were positive with tuberculin test. In one case BCG diagnostic test showed accelerated reaction. In 10 cases of probable pulmonary TB showing pulmonary infiltrates on radiological examination, only two gastric aspirate samples were positive for PCR and yielded M. tuberculosis on culture while none showed AFB on staining. Out of the two positive cases, one had cavitation in radiological examination of chest. Nine out of 10 pulmonary cases were positive with tuberculin test. One case showed accelerated reaction on BCG diagnostic testing. All cases were given ATT and improved. Thus in gastric samples, as with mycobacterial culture, sensitivity was only 20% (2 out of 10 positive with PCR) in comparison with clinical criteria of pulmonary TB. The details are shown in Table I. In the control samples all CSF and gastric aspirate samples were negative with PCR but two ascitic fluid samples were positive though there was no clinical evidence of TB in these cases. Thus, 2 out of 39 control samples gave a false positive result (7%) (Table II).
Tuberculosis even today, remains a major health problem among children in the world especially in India. Authentic information about the extent and magnitude of tuberculosis is scant. The problem is further compounded by the fact, that conventional epidemiological tools used for assessment, namely prevalence and incidence of infection, prevalence of suspected and drug resistant cases and mortality, are difficult to apply in this category because of the absence of an objective method or a ‘gold standard’ for diagnosis of tuberculosis(14). Bacterial proof of the infection and positive smears are rarely obtained. Amplification of M. tuberculosis specific DNA sequences in clinical samples is the most sensitive and rapid method of detection available(15,16). PCR has been shown to be a sensitive (88% to 100%)(17,18) and specific (>90%)(17,19) diagnostic test for pulmonary TB in adults. An Indian study using IS6110 as primers for DNA amplification in adults has shown sensitivity for CSF when clinical picture is taken into consideration, as 93% and specificity as 81.5%(20). However, the usefulness of this approach in children has not been well-evaluated(15). If PCR proved to be as sensitive and specific for children also, it would be a great improvement over current diagnostic methods for TB in children(4). We detected mycobacterial DNA after amplification by PCR in clinical samples from 41 children with suspected clinical disease as per clinical criteria mentioned earlier. We used 39 clinical specimens as control from children not suffering from TB. In CSF samples, all children diagnosed to be suffering from CNS TB (diagnosis based on clinical, investigative parameters and response to treatment), were positive with PCR although none of the samples yielded M. tuberculosis on culture. This was considered as 100% positive, since M. tuberculosis from CSF samples can be isolated very rarely(15,21). Various authors have shown a sensitivity ranging from 40% to 100%. Delacourt et al. found 50% sensitivity in CSF samples(15), Smith et al. 40%(4) while Monno et al. showed 100% sensitivity(22) with PCR. Culture results from CSF samples in these studies were very poor. Our results were similar, indicating that PCR is an extremely sensitive test for the diagnosis of CNS TB. PCR will prove to be a valuable diagnostic test for detecting M. tuberculosis in CSF from patients with CNS TB(4,21-23). Only 2 out of 10 gastric aspirate samples analyzed were positive with PCR and M. tuberculosis was grown on culture though the smears were negative signifying sensitivity of 20%. Compared to clinical evaluation, 8 samples gave a false negative result. Various studies have found yield of M. tuberculosis from gastric samples in children ranging from 12% to 41%. Most of the American and European studies(4,15,24) have shown gastric sample positivity between 24% to 41% while Indian studies revealed results between 12% and 20%(14,25) signifying a low yield of M. tuberculosis on culture from gastric samples. However, false negative results could have resulted from (13): (a) The presence of inhibitors not detected by the control amplifica-tion; (b) Non-homogenous distribution of bac-teria in the specimen so that the fraction tested does not contain mycobacteria; and (c) low number of bacilli in the specimen. Also, a number of studies in India have shown that, nucleotide sequence in IS6110 is not always present within strains of M. tuberculosis causing disease in India. Primers with different nucleo-tide sequence are now being used in India by a few laboratories to ensure that all cases are detected by PCR amplification(26). In our study, 2 false positive results were obtained from control samples. False positive results are known to be a problem with PCR technique, both, in research and hospital laboratories(14,27). There are two recognized explanations for false positive PCR results. The most common is the carry over of amplicons from previous reactions (even when the laboratory activities are well separated from each other). Another source is cross contamina-tion with Mycobacterium tuberculosis DNA isolated from positive clinical samples during the processing procedure. The frequency with which these problems occur is difficult to document(4). In our study, considering all the samples evaluated, sensitivity was 100% in CSF and pleural fluid specimens and only 20% in gastric aspirate samples. Specificity of PCR in our study was 94% (2 false positive out of 39) but owing to small number of control samples this figure may be misleading. However, Delacourt et al. have shown a specificity of 100%(15), while Smith et al. have shown 80% specifi-city(4) in their studies. The result s of various studies show that PCR can detect very low levels of AFB in the clinical specimens and specificity ranges from 80-100%(4,15,28). Results are available in 3 days as compared to 3 weeks with culture. However, its strength can be its greatest weak-ness as even the smallest amount of contaminat-ing DNA can be amplified resulting in mis-leading results. Caution is essential while using any sensitive technique. However, by taking certain precautions 100% specificity with PCR is possible. Collecting samples in two volumes, one of which is separately stored for later use if necessary and processing a small number of samples at one time can do this(21). A negative PCR result seems most useful in excluding active TB as a diagnostic consideration in selected patients(17). Our study had certain limitations. Firstly, the control specimens were small in number, hence the specificity results can be misleading. Secondly, more number of patients of pulmo-nary and extra-pulmonary TB need to be evaluated before usefulness of this technique can be conclusively proved. To conclude, more work is needed to evaluate the utility of PCR for diagnosing TB in children in Indian conditions though our preliminary study holds promise. Until advances in PCR technique improve specificity, PCR alone is insufficient as a single diagnostic test for TB in children. In culture negative children, epidemiological, clinical, radiological, tuber-culin testing, and response to treatment, constitute the criteria of diagnosis of TB. Acknowledgements The authors are thankful to Maj. Gen. (Dr.) J.K. Bhalla, Commandant Command Hospital, Lucknow and Col. (Dr.) Y.K. Goorha, previous Commanding Officer and Senior Advisor (Pathology), Command Pathology Laboratory, Lucknow, for their able guidance and en-couragement in completing this study. Contributors: SKJ designed the study selected the patients, collected data, interpreted the results, and drafted the paper; he will act as the guarantor. MNGN participated in designing the study, interpreting the results and helped in drafting the paper. KKL carried out PCR test in the laboratory and participated in data collection. NPS coordinated and supervised the study.
Funding: None.
|