Detection of Mycobacterium tuberculosis in Sputum Samples byPolymerase Chain Reaction Using a Simplified Procedure
TANIL KOCAGOZ,l* ENGIN YILMAZ,2 5EREF OZKARA,3 SESIN KOCAGOZ,1 MURAT HAYRAN,4MEENA SACHEDEVA,s AND HENRY F. CHAMBERS5
Departments ofMicrobiology, 1 Medical Biology, 2 and Infectious Diseases,4 School of Medicine,Hacettepe University, and Atatiirk Chest Diseases and Chest Surgery Center,3 Ankara, Turkey,and The Medical Service, San Francisco General Hospital, and The Department of Medicine,
University of California, San Francisco,5 San Francisco, California 94110
Received 7 January 1993/Accepted 3 March 1993
A repetitive sequence of Mycobacterium tuberculosis DNA was amplified by polymerase chain reaction(PCR), from sputum samples, for the diagnosis of pulmonary tuberculosis. The method of heating the samplein a boiling water bath to break down the bacterial cell wall and to release the DNA was compared with thatof enzymatic lysis of bacteria and then phenol-chloroform extraction ofDNA. Heating the sample was the bettermethod with a sensitivity of approximately 10 microorganisms. A total of 78 sputum specimens prepared byheating were examined by PCR, and the results were compared with the results of acid-fast stained smears,cultures, and clinical data. M. tuberculosis was detected by PCR in all smear- and culture-positive andsmear-negative, culture-positive cases. Additionally, PCR was capable of detecting four of nine cases whichwere smear and culture negative but clinically suspected of tuberculosis. DNA amplification by PCR is asensitive and specific method for the diagnosis of tuberculosis, and with this simplified DNA isolation procedureit can be used in routine clinical practice.
Although a presumptive diagnosis of pulmonary tubercu-losis can be made on the basis of patient histories and clinicaland radiological findings, the definitive bacteriological diag-nosis of tuberculosis continues to depend on the microscopicexamination of acid-fast stained sputum smears and thencultural confirmation. Direct microscopy by Ziehl-Neelsenstaining to identify acid-fast bacilli (AFB) is the most rapidmethod, but it lacks sufficient sensitivity and specificity. Onthe other hand it takes 4 to 8 weeks to culture pathogenicmycobacteria because of their slowly growing nature. Re-cently there has been great progress in developing rapid,sensitive, and specific tests for the diagnosis of tuberculosis(3).
Amplification of specific nucleic acid sequences by usingpolymerase chain reaction (PCR) has become a powerfultool for the rapid and specific detection of many infectiousagents (15). Many primers which amplify specifically theDNA of Mycobactenum tuberculosis have been designedand successfully used for identification of this microorgan-ism from culture (5, 10, 13, 16) and also from clinical samples(1, 2, 4, 6-8, 12, 14). It has been stated that the technology istoo complex for application in developing countries, but ithas potential for simplification (3); the necessity of furtherefforts to establish the sensitivity of the test and to developroutine and simple procedures for use in clinical practice isemphasized (2). In previous studies the isolation of M.tuberculosis DNA, by enzymatic lysis and phenol-chloro-form extraction, was usually performed before amplification.In this study we used heat as the only agent to break downthe bacteria and to release DNA for amplification by PCR,which simplified the overall procedure, making the methodpotentially useful for routine clinical practice.
* Corresponding author.
MATERIALS AND METHODSPrimers. The primers used for the amplification were
originally designed by Eisenach et al. from sequences whichare repeated several (10 to 16) times in the chromosome ofM. tuberculosis (6). The sequences of the primers (synthe-sized by MedProbe A.S., Oslo, Norway), which amplify a123-bp fragment of the repetitive sequence (6) were (5' to 3')CCTGCGAGCGTAGGCGTCGG and CTCGTCCAGCGCCGCTITCGG.Comparison ofDNA isolation methods. Two different DNA
isolation methods for preparation of template DNA werecompared. A suspension of a clinical isolate of M. tubercu-losis having -105 microorganisms per ml was prepared in TEbuffer (10 mM Tris [pH 8.0], 1 mM EDTA), by adjusting itsturbidity with McFarland standards. A total 10 ,ul of thissuspension (-103 organisms) was put into two tubes. One ofthese was placed on boiling water for 10 min and then useddirectly for amplification without any further processing.The other tube was centrifuged, and the pellet was resus-pended in TE containing lysozyme (10 mg/ml) and incubatedon ice for 30 min. Proteinase K and sodium dodecyl sulfate(SDS) were then added to final concentrations of 10 mg/mland 2%, respectively, and the incubation was continued at37°C for 3 h. DNA was extracted by phenol-chloroform andprecipitated by ethanol. The pellet was resuspended in 10 ,ulof TE buffer and used in PCR.PCR procedure. A reaction mixture of 100 ,ul containing
the primers (20 pmol each), lx Taq polymerase buffer (25mM Tris HCl [pH 9.5], 50 mM KCl, 10 mM MgCl2, 1 mMdithiothreitol, bovine serum albumin [1 mg/ml], activatedcalf thymus DNA [300 ,ug/ml], 0.2 mM [each] deoxynucleo-side triphosphates [dNTP], Taq polymerase [AmershamInternational; 2.5 U]), and the sample was prepared. TheDNA was denatured for 5 min at 94'C; 35 amplificationcycles were performed with an automated thermal cycler(DNA Thermal Cycler; Perkin-Elmer Cetus). Each cycleconsisted of denaturation at 94°C for 1 min, annealing of
primers at 68°C for 2 min, and primer extension at 72°C for2 min. The presence of the 123-bp amplification product wasanalyzed by electrophoresis of 10 ,ul of the amplified mixtureon a 2% agarose gel. The DNA was stained by ethidiumbromide and photographed on a UV transilluminator.
Determination of sensitivity. To determine the sensitivityof detection of M. tuberculosis, 10-fold serial dilutions of thesame clinical isolate mentioned above were prepared, andDNA was prepared by just boiling the organisms. Tubescontaining 108 to _ 102 organisms per ml were placed onboiling water for 10 min, and 10 [lI from each tube (.106 to-1 organism) was used for amplification.
Patients and clinical specimens. Samples were obtainedfrom 78 patients who were admitted to Ataturk ChestDiseases and Chest Surgery Center, Ankara, Turkey, withor without suspected tuberculosis. The clinical diagnosis oftuberculosis was established by patient histories, clinical andradiological findings, and response to antituberculosis drugtherapy. In Turkey, the prevalence of being seropositive forhuman immunodeficiency virus (HIV) is very low. Accord-ing to the statistics of the Ministry of Health, a total of 129patients seropositive for HIV and 67 cases of AIDS werereported by 29 February 1992 (9). In a survey, 262 patientswith tuberculosis were tested for HIV, and all of them werefound to be negative (18). In another survey done at the samehospital from where the samples were collected for thisstudy, 3 of 357 patients with tuberculosis were found out tobe HIV seropositive (11). Therefore, the patients included inthis study were not likely to be HIV positive.Sputum samples were treated with the standard protocol
of n-acetylcysteine-NaOH and concentrated by centrifuga-tion (1,500 x g). From the sediments smears were made,stained by Ziehl-Neelsen, examined for AFB, and inoculatedon Loewenstein-Jensen medium. The remaining of the sed-iments were stored at -20°C until used for PCR amplifica-tion.
Specimen processing and DNA amplification. The sedi-ments of sputum samples were washed at least three timesby centrifugation (12,000 x g) and resuspension in TEbuffer. This was critical to rid the sample of proteins andsalts produced by decontamination with NaOH. The finalsediment was resuspended in 50 ,ul ofTE buffer, and the tubewas placed in a boiling water bath for 10 minutes. Of this, 10RI was used for PCR. The amplification was performed bythe above-described procedure. M. tuberculosis from cul-ture and human epithelial DNA were used as positive andnegative controls.
M I 2 3 4
bp
1353
603
310
118
FIG. 1. Comparison of the 123-bp amplification products by PCRfrom M. tuberculosis DNA isolated by two different methods fromthe same number (_103) of bacteria. A total of 10 pJ from each tubeafter amplification was run on a 2% agarose gel. Lane M, HaeIII-digested phage 4X174 DNA as molecular size markers; lane 1, DNAprepared by enzymatic lysis followed by phenol-chloroform extrac-tion; lanes 2 and 4, no DNA; lane 3, template DNA prepared byheating the bacteria on boiling water.
from patients with diagnosis of tuberculosis and 21 werefrom patients with nontuberculous pulmonary disease, onthe basis of clinical and radiological findings. All of the 26samples which were smear and culture positive were alsopositive with PCR; 3 samples which were negative on thesmear for AFB but culture positive were PCR positive.There were 9 patients who were smear and culture negativebut for whom tuberculosis was strongly suspected clinically.Four of these were PCR positive. There were 19 patientswith pulmonary tuberculosis who were receiving antituber-culosis therapy. Of these, 12 were smear and/or culturepositive before therapy but were both smear and culturenegative at the time the sample was taken for PCR. Theduration of therapy ranged between 2 weeks and 1 year. Twoof these were positive with PCR. One of these cases hadtaken therapy for 2 weeks and the other had therapy for 3weeks. PCR-negative patients had taken at least 4 weeks oftherapy. Seven patients with low clinical suspicion of tuber-culosis but for whom smears and cultures were repeatedlynegative were negative by PCR. Finally, of the 21 patientswith nontuberculous pulmonary disease with sputum smearand culture negative for AFB, 1 was positive with PCR.
RESULTS
The amplification of the 123-bp fragment of M. tuberculo-sis DNA by PCR, after preparation of the template DNA bytwo different methods, is shown in Fig. 1. A stronger bandwas produced by PCR amplification when bacterial DNAwas obtained by boiling compared with lysis by lysozymeand SDS and then proteinase K digestion and phenol-chloroform extraction. The experiment was repeated twicewith the same results (data not shown). The sensitivity ofPCR for the detection of M. tuberculosis with the boilingmethod was determined by making serial 10-fold dilutions ofthe bacterial suspension (Fig. 2). Less than 10 microorgan-isms can be detected.A total of 78 sputum samples were tested by PCR with the
boiling method. (some are shown on Fig. 3.) The comparisonof amplification results with smears, cultures, and clinicaldata is summarized in Table 1. A total of 57 of these were
A. N 1 ~2 3 4 5 6 7 8 9
'3_ _ _ _ _ _
1W3
FIG. 2. Detection ofM. tuberculosis by PCR with template DNAprepared by heating the bacteria on boiling water. Amplificationproducts from serial 10-fold dilutions of M. tuberculosis. A total of10 ,ul from each tube after amplification was run on a 2% agarose gel.Lane M, HaeIII-digested phage qX174 DNA as molecular sizemarkers; lanes 1 to 7, DNA from reaction tubes containing -106,_105, 1 _103, _ 102, -10, and -1 bacterium, respectively; lane8, human epithelial DNA; lane 9, no DNA.
DETECTION OF M. TUBERCULOSIS IN SPUTUM SAMPLES BY PCR 1437
M 1 2 3 4 5 6 7 8 9bp
1353603310118
FIG. 3. Amplification products (123 bp) by PCR from M. tuber-culosis in sputum samples. Lane M, HaeIII-digested phage +X174DNA as molecular size markers; lanes 1, 2, 4, and 6 are positive;lanes 3, 5, and 7 are negative; lane 8, M. tuberculosis from culture;lane 9, human epithelial DNA.
Unfortunately there is no real "gold standard" to compareand to determine the sensitivity and specificity of eachmethod. Table 2 shows the sensitivity, specificity, andpredictive valves of PCR compared with AFB smear andculture, assuming clinical diagnosis as the gold standard.
DISCUSSIONDNA amplification by PCR is a rapid and sensitive method
for detection of M. tuberculosis in sputum samples. Prepar-ing samples by boiling simplifies the method so that it can beused in routine clinical practice. The repetitive nature of thetarget sequence amplified by PCR probably contributes tothe high sensitivity. Fewer than 10 microorganisms could bedetected. Precise estimation of the sensitivity is difficult.Since the number of organisms used in PCR to determine thesensitivity of the assay was adjusted depending on theturbidity of the suspension, the actual number of organismsper sample may have been higher, because of possibleclumping of organisms, and thus the sensitivity may havebeen overestimated. All culture-positive samples, includingthree that were smear negative, were positive with PCR. Ofthe nine samples which were both smear and culture nega-
TABLE 1. Correlation of PCR findings with smear and cultureresults and clinical data
Patient status No. of No. positive
patients by PCR
Smear and culture positive 26 26Smear negative, culture positive 3 3Smear and culture negative, with high 9 4
clinical suspicion; not on treatment attime of studya
Smear and culture positive previously, 12 2became negative by therapy at thetime sample was taken for PCR
On treatment for tuberculosis, repeat- 7 0edly smear and culture negative, withlow clinical suspicion
Smear and culture negative; nontuber- 21 1culous pulmonary disease
a Of the four patients who were smear and culture negative, PCR positive,and with high clinical suspicion, two had pleural fluid with adjacent paren-chymal lesion and were diagnosed to have tuberculosis by pleural biopsy; theother two had moderate fibronodular infiltration without cavitation. All ofthese patients had symptoms for more than 4 weeks and responded toantituberculosis therapy, showing clinical and radiological improvement.
TABLE 2. Sensitivity, specificity, and predictive values ofvarious methodse
Method Sensitivity Specificity Predictive value (%)(M) ( ) Positive Negative
a Clinical diagnosis with high suspicion is considered the gold standard. Thetotal cases of tuberculosis was 38 (26 were smear and culture positive; 3 weresmear negative, culture positive; and 9 were smear and culture negative butwith clinical diagnosis of tuberculosis with high suspicion). True negativesnumbered 28 (21 patients with nontuberculous pulmonary disease and 7 whowere always smear, culture, and PCR negative, with low clinical suspicion).
tive, obtained from patients clinically suspected of tubercu-losis, four were positive with PCR, suggesting that the assayis probably more sensitive than the culture by detectingnonviable and/or fewer viable organisms.The one false-positive result from the sputum sample of a
nontuberculous pulmonary disease patient may be due tocross-contamination with target sequences because of anerror in handling the samples. (Although the PCR procedurewas performed in a laboratory located in another building,the processing of the sputum samples for smear, culture, andPCR was performed in the busy laboratory of the chestdiseases hospital, where about 50 AFB-positive smears areidentified each day.) The clinical diagnosis was accepted asthe gold standard of this research, but since there was noway to confirm this in a smear- and culture-negative case,the possibility of false diagnosis must also be considered.The preparation of mycobacterial DNA by just boiling
rather than enzymatic lysis followed by phenol-chloroformextraction was better for amplification with PCR. This mightbe due to the loss of some of the DNA during purificationsteps in the latter method. Thus, the procedure is simplifiedwith the elimination of DNA purification, and the timerequired and the cost of the assay were reduced by half.Sritharan and Barker compared eight different procedures,including different enzymatic and detergent lysis methods, torelease PCR-amplifiable DNA from M. tuberculosis bacilli(17). Of methods which worked, they selected a similarboiling method, because of its simplicity, to examine itsefficacy for detecting M. tuberculosis in clinical samples. Asin our method, they used Tris-EDTA buffer to suspend andboil the samples, but they also included 1% Triton X-100,and the duration of boiling was 30 instead of 10 min. Theydetermined that with this method, fewer than 10 microor-ganisms can be detected by PCR.
All the smear-positive samples were positive with PCR.When the incidence of seropositivity for HIV is very low andMycobacterium avium subsp. intracellulare complex (MAC)is hence a rare possibility, as it was in the patients includedin this study, PCR positivity only confirms that the AFBseen on the smear is M. tuberculosis. Since detection ofAFB by smear is an inexpensive and rapid method from thepoint of beginning antituberculosis therapy, PCR positivityis not very beneficial for the clinician in these cases. How-ever, in HIV-positive patients it is very important to differ-entiate M. tuberculosis from MAC in choosing the righttherapeutic regimen. In these cases the detection of M.tuberculosis by PCR might be very useful, even when thesmear is positive, especially if it can be used together with aPCR procedure that can detect MAC. Since there was nosmear-positive and PCR-negative patient identified in this
study, the utility of this method in HIV-positive casesremains to be determined.The detection of M. tuberculosis with PCR in smear-
negative and culture-positive patients provides the bacterio-logical data 4 to 8 weeks earlier. For smear- and culture-negative specimens, PCR is the only currently availablemethod that can provide bacteriological diagnosis. Sincedetection of AFB by smear is an inexpensive and rapidmethod, for HIV-negative patients clinically suspected oftuberculosis, it might be rational to use PCR for smear-negative specimens.
ACKNOWLEDGMENT
This study was partly financially supported by the Turkish Na-tional Planning Organization.
REFERENCES1. Brisson-Noel, A., C. Aznar, C. Chureau, S. Nguyen, C. Pierre,
M. Bartoli, R. Bonete, G. Pialoux, B. Gicquel, and G. Garrigue.1991. Diagnosis of tuberculosis by DNA amplification in clinicalpractice evaluation. Lancet 338:364-366.
2. Brisson-Noel, A., B. Gicquel, D. Lecossier, V. Levy-Frebault, X.Nassif, and A. J. Hance. 1989. Rapid diagnosis of tuberculosis byamplification of mycobacterial DNA in clinical samples. Lancet335:1069-1071.
3. Daniel, T. M. 1990. The rapid diagnosis of tuberculosis. Aselective review. J. Lab. Clin. Med. 116:277-282.
4. Del Portillo, P., L. A. Murillo, and M. E. Patarroyo. 1991.Amplification of a species-specific DNA fragment of Mycobac-terium tuberculosis and its possible use in diagnosis. J. Clin.Microbiol. 29:2163-2168.
5. Eisenach, K. D., M. D. Cave, J. H. Bates, and J. T. Crawford.1990. Polymerase chain reaction amplification of a repetitiveDNA sequence specific for Mycobacterium tuberculosis. J.Infect. Dis. 161:977-981.
6. Eisenach, K. D., M. D. Sifford, M. D. Cave, J. H. Bates, andJ. T. Crawford. 1991. Detection of Mycobacterium tuberculosisin sputum samples using a polymerase chain reaction. Am. Rev.Respir. Dis. 144:1160-1163.
7. Hermans, W. M., A. R. J. Schuitema, D. Van Soolingen,
C. P. H. J. Verstynen, E. M. Bik, J. E. R. Thole, A. H. J. Kolk,and J. D. A. Van Embden. 1990. Specific detection ofMycobac-tenum tuberculosis complex strains by polymerase chain reac-tion. J. Clin. Microbiol. 28:1204-1213.
8. Manjunath, N., P. Shankar, L. Rajan, A. Bhargava, S. Saluja,and Shriniwas. 1991. Evaluation of a polymerase chain reactionfor the diagnosis of tuberculosis. Tubercle 72:21-27.
9. Ministry of Health, Turkey. 1992. HIV positive cases in Turkey.AIDS Bull. 2:1-2.
10. Musial, C. E., L. S. Tice, L. Stockman, and G. D. Roberts. 1988.Identification of mycobacteria from culture by using the Gen-Probe rapid diagnostic system for Mycobacterium avium com-plex and Mycobacterium tuberculosis complex. J. Clin. Micro-biol. 26:2120-2123.
11. Onde, G., U. Onde, T. Sipit, S. Ozkara, F. Duru, 0. OrucZ, andN. Acar. 1992. A study of anti-HIV antibodies in patients withtuberculosis, p. 168. Abstr. 20th Annu. Meet. Turkish Respir.Res. Assoc. 1992.
12. Pao, C. C., B. Yen, J. B. You, J. S. Maa, E. H. Fiss, and C. H.Chang. 1990. Detection and identification of Mycobactenumtuberculosis by DNA amplification. J. Clin. Microbiol. 28:1877-1880.
13. Patel, R. J., J. W. U. Fries, W. F. Piessens, and D. F. Wirth.1990. Sequence analysis and amplification by polymerase chainreaction of a cloned DNA fragment for identification of Myco-bacterium tuberculosis. J. Clin. Microbiol. 28:513-518.
14. Pierre, C., D. Lecossier, Y. Boussougant, D. Bocart, V. Joly, P.Yeni, and A. J. Hance. 1991. Use of a reamplification protocolimproves sensitivity of detection of Mycobacterium tuberculo-sis in clinical samples by amplification of DNA. J. Clin. Micro-biol. 29:712-717.
15. Schochetman, G., C. Y. Ou, and W. K. Jones. 1988. Polymerasechain reaction. J. Infect. Dis. 158:1154-1157.
16. Sjobring, U., M. Mecklenburg, A. B. Andersen, and H. Miorner.1990. Polymerase chain reaction for detection of Mycobacte-rium tuberculosis. J. Clin. Microbiol. 28:2200-2204.
17. Sritharan, V., and R. H. Barker, Jr. 1991. A simple method fordiagnosing M. tuberculosis infection in clinical samples usingPCR. Mol. Cell. Probes 5:385-395.
18. Yener, 0. S., E. Ekinci, and R. Aydilek. 1987. Prevalence ofanti-HIV antibodies in patients with pulmonary tuberculosis.Tubercul. Thorax 35:293-298.
