The Utility of Polymerase Chain Reaction (PCR) in the Diagnosis of Pulmonary Tuberculosis*

The Utility of Polymerase Chain Reaction (PCR) in the Diagnosis of Pulmonary Tuberculosis* jose Manuel Querol, MD, PhD; Maria Amparo Farga, PhD; Damiana Granda, MD, PhD; Concepcion Gimeno, MD, PhD; and juan Garcia-de-Lomas, MD, PhD A fragment of DNA of 123 bp belonging to insertion sequence IS6110, specific of Mycobacterium tuberculosis complex, was amplified by polymerase chain reaction (PCR) of respiratory samples, for the diagnosis of pulmonary tuberculosis. A total of 314 samples (286 sputum and 28 bronchoalveolar lavages) from 242 patients were evaluated by PCR, and the results were compared with the those obtained by acid-fast-stained smears, culture, and clinical diagnosis. Mycobacterium tuberculosis was detected by PCR in 102 of 105 patients with clinical diagnosis of pulmonary tuberculosis. All smear and culture-positive samples were PCR positive. The sensitivity of PCR, culture, and staining was 97%, 88%, and 65%, respectively, and the specificity was 100% in all cases. In ten patients with old residual lesions, but no active disease, M tuberculosis genome was detected by PCR. In our experience, PCR proved to be a useful method for the rapid diagnosis of pulmonary tuberculosis. (CHEST 1995; 107:1631..'35) AFB=acid-fast bacilli; LJ = Lowenstein-Jensen; PCR= polymerase chain reaction; TB=tuberculosis; ZN=Ziehl Neelsen Key words: 156110; Mycobacterium tuberculosis; PCR; polymerase chain reaction; pulmonary tuberculosis Tuberculosis (TB) is presently an important health problem throughout the world. 1 Despite its progressive decrease in developed countries, the situation has changed in recent years due to the AIDS pandemic. 2 Classically, there is a correlation between the presence of acid-fast bacilli (AFB) in clinical samples and the isolation in culture of Mycobacterium tuberculosis. Moreover, the increase in the incidence of atypical mycobacteria in some clinical syndromes-fundamentally in patients with AIDSurges the need to introduce specific methods for rapid diagnosis to avoid unnecessary or improper treatments. At present, microscopic examination is the sole rapid diagnostic method available. The technique is simple and may be performed in any laboratory. A low sensitivity is its only limitation. Culture, through new radiometric systems, 3 and biphase culture 4 may require more than 2 weeks to confirm diagnosis. Be- *From the Department of Microbiology, University Hospital and Medical School, Valencia, and Division of Internal Medicine and Microbiology, Hospital Lluis Alcanyis, Xativa, Spain. Presented in part at the 32nd Interscience Conference on Antimicrobial Agents and Chemotherapy, Anaheim, Calif, October 11-14, 1992. Manuscript received November 8, 1993; revision accepted October 11. Reprint requests: Dr. Querol, Dept Internal Medicine, Hospital LLuis Alcanyis, 46800-Xativa, Spain sides, research in serology 5 and structural components such as tuberculoestearic acid 6 is not widely used in microbiology laboratories, in some cases due to the high costs involved, and in others because of a lack of specificity, sensitivity, or both. The detection of M tuberculosis by enzymatic amplification (PCR) has been found useful in the diagnosis of pulmonary, pleural, and meningeal TB. 7-9 Several PCR procedures have been described to detect M tuberculosis genome, the target sequence being the main difference, 10-19 due to the influence on specificity and sensitivity. Although the insertion sequence IS6110, specific for M tuberculosis complex and repeated several times in the chromosome is the most frequently used target for diagnosis and epidemiologic purposes, 20-22 the investigation of groel gen may also prove important for initial study in areas where atypical mycobacteria are prevalent and in immunosuppressed patients. We have studied the utility of a PCR method based on the detection of IS6110 sequence, using a primer pair described by Eisenach et al 21 to amplify a 123-bp fragment. We have evaluated the results obtained after application of the method to respiratory samples, comparing them with the classic microbiologic methods, in an attempt to establish the utility of the technique in the diagnosis of pulmonary TB. CHEST I 107 I 6 I JUNE, 1995 1631

Table!-Patients Classified by Clinical Diagnosis, Age, and Sex Group Pulmonary tuberculosis Residual tuberculosis Controls Patients N 105 44 93 Age (X, Limits) 38 (13-86) 66 (34-83) 60 (14-90) MATERIAL AND METHODS Sex, M/ F 67/ 38 34/ 10 74/ 19 A total of 242 patients were classified by clinical diagnosis: pulmonary tuberculosis, 105; residual tuberculosis, 44; and a control group of 93 patients without TB (Table 1). Diagnosis of TB was based on the isolation and identification of M tuberculosis in clinical samples of pulmonary origin in 92 patients, while in another 13 cases the diagnosis was established by clinical, epidemiologic, radiologic, and therapeutic criteria. The more relevant clinical characteristics in this group are outlined in Table 2. The residual TB group included 44 patients, all of whom presented with radiologically manifest, severe residual lesions suggesting old TB. Active disease was excluded in all these patients by the culture of sputum specimens. The control group of 93 patients presented with acute pneumonia (n=47), chronic obstructive lung disease (n=30), and lung neoplasia (n=16). Clinical Samples and Microbiology Studies Three-hundred fourteen samples (286 sputum and 28 bronchoalveolar lavages) were stained by Ziehl-Neelsen (ZN), and culture was performed in Li:iwenstein-Jensen (LJ). The sputum was previously digested and decontaminated by the N-acetyl-Lcysteine and NaOH method. Preparation of Samples for PCR Analysis Three-hundred fifty microliters of the homogenized sample was suspended in a lytic solution (Tris-HCl, ph 8.3, 10 mm; KCl, 50 mm; MgCl2, 1.5 mm; nonidet P-40, 0.45%; polysorbate 20 [Tween-20], 0.45%) and treated with proteinase K (l mg/ ml) and SDS (sodium dodecyl sulfate) for l h at 37 C. Once digested, DNA was extracted by the phenol-chloroform-isoamylalcohol (25:24:1) method after treatment with NaCl 5M and CTAB (hexadecyl-trimethylammonium bromide). 23 DNA was purified by precipitation with isopropyl alcohol and washed with 70% cold ethanol. The pellet was then suspended in distilled water and maintained at -40 C until used. Amplification and Detection of Mycobacterial DNA A pair of previously described primers were used. 21 The oligonucleotides 5'-CCTGCGAGCGTAGGCGTCGG and 5' CTCGTCCAGCGCCGCTTCGG, synthetized with a DNA syn- Table 2-The More Relevant Clinical and Epidemiologic Characteristics of Patients With Pulmonary Tuberculosis Clinical data History of tuberculosis History of recent contact Cough and/ or expectoration Hemoptysis Cavitation in lung radiograph Positive Mantoux (induration >5 mm) 1632 N (%) 12 (11) 16 (15) 80 (77) 34 (33) 64 (60) 85 (80) Table 3-Summary of Results Obtained With the Diagnostic Method Applied Stain Culture PCR..-----"'----, Group + + + Pulmonary 68 37 92 13 102 3 tuberculosis (n=105) Residual tuberculosis 0 44 0 44 10 34 (n=44) Control group 0 93 0 93 0 93 (n=93) thetizer (Mod 320; Applied Biosystems) were used to amplify a 123-bp fragment from the insertion sequence IS6110. The amplification was performed in a final volume of 100 ~ t L containing 50 pmol of each primer; 0.25 mm of each of the deoxynucleotide triphosphates dgtp, datp, dttp, and dctp (Boehringer); 10 mm Tris-HCl, ph 8.0; 50 mm KCl; 2 mm MgCI2, and 2 U Tag-polymerase (Perkin-Elmer/ Cetus) and 25 ~ tof L diluted sample. Cycles were initiated at 94 C for 7 min, followed by 35 cycles at 94 C/ 2 min, 68 C/ 2 min, and 72 C 2 min, increasing each extension step in every cycle by 3 s. Amplification was detected using 3% agarose gel electrophoresis with ethidium bromide. In each experiment, we included a positive control of M tuberculosis DNA extracted from a culture, as well as a negative control, replacing the sample with distilled water. RESULTS Ziehl-Neelsen staining was positive in 68 cases (64.7%) out of 105 patients diagnosed as having pulmonary tuberculosis. Fifty (83%) of 60 patients with cavitary lesions showed positive results by microscopy. In nine patients from this group, we performed fiberoptic bronchoscopy, and the bronchoalveolar lavage obtained proved positive on staining in two cases. In the remaining 37 patients from the same group, 4 samples from each patient were analyzed, all being negative. The culture on LJ was positive in 92 cases (87.%). All samples with a positive result of microscopic examination for AFB were confirmed by culture (Table 3). There was no microscopic examination or culture positivity among patients with residual TB or in the control group. PCR detected M tuberculosis genome in 102 (97%) of 105 patients diagnosed clinically as having tuberculosis. Among nine samples of bronchoalveolar lavage studied, PCR was positive in eight. All culturepositive samples were also PCR positive (Table 4). In ten patients with positive PCR, the culture was negative: four patients came from a school with a TB outbreak and presented with a pulmonary lesion in the x-ray examination; three had noncavitary infiltrates detected while studying contacts of patients with active pulmonary TB; one 14-year-old female patient suffered from erythema nodosum without Clinical Investigations

Table 4-Comparison of Results Obtained by Culture and PCR Culture+ Culture - Group PCR+ PCR- PCR+ PCR- Pulmonary tuberculosis 92 0 10 3 (n =l05) Residual tuberculosis 0 0 10 34 (n =44) identifiable parenchymal lesions in the chest radiograph (her mother had an infiltrate in the superior right lobe); another patient suffered from tuberculous meningitis and had a normal chest radiograph; and finally, a female patient exhibited a right basal infiltrate and hiliar adenopathies in radiographs. In only three patients with a clinical diagnosis of pulmonary TB were the sputum samples obtained negative by both culture and PCR. In one patient, initially studied for a hemoptysis episode, with normal chest radiograph, microscopic and PCR findings, we were able to establish the diagnosis 2 months later by detection of AFB in sputum and positive culture. In another patient, a nurse from our hospital, the diagnosis was of high probability, based on the detection of a cavitary pulmonary nodule in the right upper lobe, and confirmed by CT of the lung, positive tuberculin test, and a favorable response to antituberculous treatment. In the last case, a female patient with moderate fever, hemoptysis was seen together with a left hiliar node that improved after specific treatment. There was no microscopic examination or culture positivity among patients with residual TB or in the control group. We were unable to detect mycobacterial DNA by PCR in any control (Table 3). However, PCR was positive in 10 of 44 patients (22.7%) previously diagnosed as having TB and with an old residual lesion in chest radiographs. Most of these patients consulted for hemoptysis or presented respiratory infection overlying chronic bronchitis. In none of ten patients with a residual lesion and positive PCR could active disease be shown; the patients remained asymptomatic after 1 year, and both periodic microscopic examination and sputum culture were negative. In 29 patients with pulmonary TB and with positive microscopy, culture, and PCR, we obtained 72 sputum samples at intervals-weekly in the first month and monthly afterwards during treatment. After 8 weeks, only 1 of 12 patients continued to have microscopically positive results, while 5 were culturepositive. PCR was positive in ten cases. After 3 months, none of the patients had positive results by conventional microbiology methods, and in only one Table 5-Sensitivity (S) and Specificity (E) of Methods Used for the Diagnosis of Pulmonary Tuberculosis, Taking Clinical or Microbiologic Diagnosis as Reference Clinical Diagnosis Culture Method s (%) E (%) s (%) E (%) Ziehi-Neelsen stain 64.7 100 74 100 Culture in Lowenstein 87.6 100 100 100 PCR 97 100 100 90.5 patient could M tuberculosis genome be detected by PCR in a bronchoalveolar lavage sample obtained for other purposes. DISCUSSION The rapid diagnosis of lung TB continues to be based on the detection of AFB in sputum by ZN staining. The sensitivity of microscopy depends on the clinical presentation, and is greater in cavitary forms with hemoptysis. More than 10,000 bacilli per milliliter of sputum are necessary to secure microscopic positivity. 24 The success of microscopy is highly variable (22 to 96%), 25-27 though most authors rate it at around 60%. 28-31 In our study, microscopic sensitivity was 64.7% and specificity was 100%. The high specificity of staining is due to the low prevalence of nontuberculous mycobacteria in our area, as they cannot be discerned from M tuberculosis on the smear. Culture sensitivity using LJ medium compared with clinical diagnosis was 87.6% (Table 5). The PCR assay used in this study is based on the method previously described. 21 It involves the amplification of a fragment of 123 bp in the IS6110, specific for M tuberculosis complex. We used this sequence in view of its high specificity and also because it provides a higher sensitivity due to its repetition along the chromosome (6 to 20 times in M tuberculosis). PCR was positive in 97% of patients with the clinical diagnosis of pulmonary TB. Only three patients were PCR and culture negative (Table 4). In none of the samples from these three patients were reaction inhibitors detected, and we consider them false-negatives as judged by clinical diagnosis. All culture-positive cases were also PCR positive. This means that with respect to culture, PCR sensitivity was 100%. PCR specificity in relation to the initial clinical diagnosis of TB was 100% (Table 5). However, when considering culture as the reference method, specificity was 90.5%, but decreased to 87% when including all culture-negative patients, including both the control group and residual TB group. The absence of a gold standard for the diagnosis of TB makes the evaluation of any new method diffi- CHEST I 107 I 6 I JUNE, 1995 1633

cult, for culture is not always positive in patients with clinical evidence of TB-eg, in primoinfection and noncavitary forms of the disease, where anti-tb treatment is often initiated on an empirical basis. Polymerase chain reaction was positive in ten nontreated patients with negative culture. This lack of agreement between the two methods may be due to their differences in sensitivity. Culture sensitivity in LJ medium is greater than 100 mycobacteria per milliliter, 24 while PCR could detect 1 to 10 mycobacterias.21 Besides, culture only permits the development of viable bacteria, while PCR can be positive in the presence of bacteria nonviable due to treatment. In 72 samples obtained during treatment, microscopy was positive in 24, culture in 46, and PCR in 62. The greater sensitivity of PCR over culture makes it useful in the diagnosis of pulmonary TB with a low rate of bacteria elimination, 32 as in the primary infections and infiltrates without cavitation seen in some of our patients. Although we could not quantify the number of mycobacteria in samples, there were differences in the intensity of the amplification obtained related to the presence or absence of mycobacteria in smear-negative specimens. In ten patients with old residual lesions in chest radiographs in whom we could not detect mycobacteria by culture, genome was detected by PCR. Clinical evolution, without antituberculous treatment, and followed for 18 months demonstrated no change in the clinical situation. In justification of this, it is well known from epidemiologic data that the previous existence of residual lesions in chest radiographs constitutes a risk factor for developing active TB (reactivation), estimated at between 2 and 13.6/ 1,000 individuals yearly. 33 These cases of reactivation reflect the importance of a latent population of bacilli in residual lesions, probably detected in our study with PCR. This finding requires more extensive and prolonged investigation and if confirmed, would require the study of chemoprophylactic regimens other than those presently accepted in these patients. In addition to the greater sensitivity of PCR, another possible explanation for this discrepancy between PCR and culture are the false-positives attributable to sample contamination with DNA from other simultaneously processed and highly bacillus-laden clinical samples (crossover contamination between specimens) or with products from previous amplifications of the same target sequence. This form of contamination can be avoided with relative ease by using careful techniques and observing stringent quality control practices. In our study, both sample processing until obtaining DNA and the amplification procedure were performed in independent laboratories, and frequent negative controls were used in each experiment; as a result, the culture negative- 1634 PCR positive cases were unlikely to be false-positives. In 80% of patients with TB and in 30% of the controls, PPD testing was positive. As could be expected, we found no relationship between PCR and skin testing. The results obtained in our study are similar to those reported by other authors, though in many cases comparisons are difficult when clinical criteria for diagnosis are used. Eisenach et ap 4 studied 162 sputum samples, and found a correlation between culture and PCR in 100% of nontreated patients. Brisson-Noel et al, 35 using amplification of groel gen and the insertion sequence 156110, also found a correlation between PCR and clinical or bacteriologic data in 97.4% of cases. Other studies have been made using genomic sequences different to groel gen and insertion sequence 156110, with similar sensitivity results.l 1-19 However, we consider that to diagnose TB it is preferable to use amplification methods to detect the sequence 156110, while groel gen should be reserved for diagnosis in areas of high prevalence of atypical mycobacteria and in the initial study of immunodepressed patients secondary to human immunodeficiency virus infection. This is because we should exclude the presence of other nontuberculous mycobacteria whose genes are not detected by the amplification of 156110, which is exclusive for M tuberculosis complex, while the groel gen is common to all mycobacteria. To conclude, PCR is a rapid, sensitive, and specific method for the diagnosis of pulmonary TB. In some clinical situations, mycobacterial gene detection by PCR may be of uncertain significance. Further study is needed in such cases. However, due to its simplicity and rapid results in pulmonary TB, the introduction of PCR in microbiology laboratories will help decrease the number of cultures made to isolate mycobacteria. Culture would thus be limited to PCRpositive samples for sensitivity testing and to clinical situations in which the utility of PCR is uncertain. REFERENCES 1 Comstock GW. Epidemiology of tuberculosis. Am Rev Respir Dis 1982; 125:8-15 2 Narain JP, Raviglione MC, Kochi A. HIV-associated tuberculosis in developing countries: epidemiology and strategies for prevention. Tubercle 1992; 73:311-21 3 Morgan MA, Horstmeier CD, Deyoung DR, et al. Comparison of a radiometric method (BACTEC) and conventional media for recovery of mycobacteria from smear-negative specimens. J Clin Microbiol 1983; 18:384-88 4 Isenberg HD, Damato RF, Heifets L, et al. Collaborative feasibility study of a biphasic system (Roche Septi-Check AFB) for rapid detection and isolation of mycobacteria. J Clin Microbial 1991; 29:1719-22 5 Daniel TM. Antibody and antigen detection for the immunodiagnosis of tuberculosis: what not? what more is needed? where do we stand today? J Infect Dis 1988; 158:678-80 Clinical Investigations

6 French GL, Chan CY, Cheung SW, et al. Diagnosis of pulmonary tuberculosis by detection of tuberculostearic acid in sputum by using gas chromatography-mass spectrometry with selected ion monitoring. J Infect Dis 1987; 156:356-62 7 Pao CC, Yen B, You J, et al. Detection and identification of Mycobacterium tuberculosis by DNA amplification. J Clin Microbiol1990; 28:1877-80 8 De Witt D, Maartens G, Steyn L. A comparative study of the polymerase chain reaction and conventional procedures for the diagnosis of tuberculous pleural effusions. Tubercle 1992; 73:262-67 9 Shankar P, Manjunath N, Mohan KK, et al. Rapid diagnosis of tuberculous meningitis by polymerase chain reaction. Lancet 1991; 337:5-7 10 Hance AJ, Grandchamp B, Levy-Frebault V, et al. Detection and identification of mycobacteria by amplification of mycobacterial DNA. Mol Microbiol1989; 3:843-49 11 Brisson-Noel A, Lecossier D, Nassif X, et al. Rapid diagnosis of tuberculosis by amplification of mycobacterial DNA in clinical samples. Lancet 1989; 2:1069-71 12 Shankar P, Manjunath N, Lakshmi R, et al. Identification of Mycobacterium tuberculosis by polymerase chain reaction. Lancet 1990; 335:423 13 Manjunath N, Shankar P, Rajan L, et al. Evaluation of a polymerase chain reaction for the diagnosis of tuberculosis. Tubercle 1991; 72:21-7 14 Patel RJ, Fries JW, Piessens WF, eta!. Sequence analysis and amplification by polymerase chain reaction of a cloned D. A fragment for identification of Mycobacterium tuberculosis. J Clin Microbiol1990; 28:513-18 15 Sji:ibring U, Mecklenburg M, Andersen AB, et a!. Polymerase chain reaction for detection of Mycobacterium tuberculosis. J Clin Microbial 1990; 28:2200-04 16 Soini H, Skurnik M, Liipo K, eta!. Detection and identification of mycobacteria by amplification of a segment of the gene coding for the 32-kilodalton protein. J Clin Microbial 1992; 30:2025-28 17 Cousins DV, Wilton SD, Francis BR, et al. Use of polymerase chain reaction for rapid diagnosis of tuberculosis. J Clin Microbial 1992; 30:255-58 18 Del Portillo P, Murillo LA, Elkin M. Amplification of a species-specific DNA fragment of Mycobacterium tuberculosis and its possible use in diagnosis. J Clin Microbial 1991; 29:2163-68 19 Altamirano M, Kelly MT, Wong A, eta!. Characterization of a DNA probe for detection of Mycobacterium tuberculosis complex in clinical samples by polymerase chain reaction. J Clin Microbiol1992; 30:2173-76 20 Thierry D, Brisson-Noel A, Levy-Frebault V, et al. Characterization of a Mycobacterium tuberculosis insertion sequence, 156110, and its application in diagnosis. J Clin Microbiol1990; 28:2668-73 21 Eisenach KD, Cave MD, Bates JH, eta!. Polymerase chain reaction amplification of a repetitive DNA sequence specific for Mycobacterium tuberculosis. J Infect Dis 1990; 161:977-81 22 De Witt D, Steyb L, Shomaker S, et a!. Direct detection of Mycobacterium tuberculosis in clinical specimens. J Clin Microbiol1990; 28:2437-41 23 Ausubel FM, Brent R, Kingston RE, eta!. Current protocols in molecular biology. New York: Greene Publishing Associates and Wiley-Interscience, John Wiley and Sons, 1988;241-42 24 Kim TC, Blackman RS, Heatwole KM, et al. Acid-fast bacilli in sputum smears of patients with tuberculosis. Am Rev Respir Dis 1984; 129:264-68 25 Boyd JC, Marr JJ. Decreasing reliability of acid-fast smear techniques for detection of tuberculosis. Ann Intern Med 1975; 82:489-92 26 Narain R, Subba Rao MS, Chandressekhar P, et al. Microscopy positive and microscopy negative cases of pulmonary tuberculosis. Am Rev Respir Dis 1971; 103:761-73 27 Blair EB, Brown GL, Tull AH. Computer files and analysis of laboratory data from tuberculosis patients: II. Analysis of 6 years data on sputum specimens. Am Rev Respir Dis 1976; 113:427-32 28 Murray PR, Elmore C, Krogstad DJ. The acid-fast stain: a specific and predictive test for mycobacterial disease. Ann Intern Med 1980; 92:512-13 29 Lipsky BA, Gates J, Tenover FC, et al. Factors affecting the clinical value of microscopy for acid-fast bacilli. Rev Infect Dis 1984; 6:214-22 30 Levy H, Feldman C, Sacho H, eta!. A reevaluation of sputum microscopy and culture in the diagnosis of pulmonary tuberculosis. Chest 1989; 95:1193-97 31 Gordin F, Slutkin G. The validity of acid-fast smears in the diagnosis of pulmonary tuberculosis. Arch Pathol Lab Med 1990; 114:1025-27 32 Pierre C, Olivier C, Lecossier D, et al. Diagnosis of primary tuberculosis in children by amplification and detection of mycobacterial DNA. Am Rev Respir Dis 1993; 128:336-42 33 American Thoracic Society / Centers for Disease Control. Control of tuberculosis. Am Rev Respir Dis 1983; 128:336-42 34 Eisenach KD, Sifford MD, Cave MD, eta!. Detection of Mycobacterium tuberculosis in sputum samples using a polymerase chain reaction. Am Rev Respir Dis 1991; 144:1160-63 35 Brisson-Noel A, Aznar C, Chuerau C, et al. Diagnosis of tuberculosis by DNA amplification in clinical practice evaluation. Lancet 1991; 338:364-66 CHEST 1107 161 JUNE, 1995 1635