TIPS ON TECHNIQUESSection Editor: Shahla Masood

Comparison of In-HousePolymerase Chain ReactionMethod With the RocheAmplicor™ Technique forDetection of Mycobacteriumtuberculosis in CytologicalSpecimensRavi Mehrotra, M.D.,

1 Paul Metz, M.D.,2 and Sue Kohlhepp, Ph.D.

2*

Two techniques have been approved by the United States FDA fordiagnosis of tuberculosis in smear positive sputa: LCX M. tuber-culosis, a ligase chain reaction procedure manufactured by AbbottLaboratories, and Amplicor™, a polymerase chain reaction(PCR) procedure manufactured by Roche. However, these com-mercial methods are expensive and beyond the reach of laborato-ries in most developing countries. We compared the Roche Am-plicor™ kit with an in-house PCR using a primer set forMycobacterium tuberculosis/bovis directed at MPB 64 proteingene. It was able to distinguish between M. tuberculosis, M.avium, M. gordonii, M. intracellulari, and M. kansasii. Fifty-sevencytological samples were submitted to the laboratory for molecu-lar diagnosis of M. tuberculosis. Both procedures were run onevery sample submitted and the two methods agreed completely.The custom-made method is less expensive than the commercialtechnique. Diagn. Cytopathol. 2002;26:262–265.© 2002 Wiley-Liss, Inc.

Key Words: Mycobacterium tuberculosis; polymerase chain reac-tion; rapid detection

Infection with Mycobacterium tuberculosis (Mtb) is in-creasing in most parts of the world and early diagnosisremains the cornerstone of management, especially in view

of the rising incidence of multidrug-resistant disease. Cul-ture techniques for diagnosis remain difficult and time-consuming and molecular diagnostic methods are gainingimportance. DNA probe technology requires a growingorganism and is time-consuming. DNA amplification bypolymerase chain reaction (PCR) provides an alternative toDNA probe assays, with a dramatic improvement in turn-around time and sensitivity of detecting specific DNA se-quences.1 However, the commercial assays are too expen-sive for use in developing countries—the place where theyare needed the most. Work reported from the Latin Amer-ican and Caribbean Network on Tuberculosis (RELACTB)showed that six Latin American laboratories using differentin-house PCR assays did not agree on the presence orabsence of Mtb in the same samples. Forty-one percent ofnegative samples were incorrectly identified and 71 to 30%of the positive samples were misidentified, depending onthe concentration of mycobacteria in each sample.2 With theincreasing cost of medical care even in developed countries,the need for reliable and inexpensive diagnostic methodsremains important.

The objective of this study was to define a less expensive,but equally reliable, PCR as compared to an FDA-approvedcommercially available one. We compared an assay using aprimer set for M. tuberculosis/bovis BCG3,4 and polyacryl-amide gel electrophoresis (PAGE) detection, hereaftercalled the study assay with the Roche Amplicor™ assay.

Materials and MethodsFifty-seven cytological samples were consecutive submis-sions to the clinical laboratory service at the Providence

1Molecular and Tumor Immunology Laboratory, Franz Cancer Institute,Earle A. Chiles Research Institute, Providence Portland Medical Center,Portland, Oregon, and Department of Pathology, Moti Lal Nehru MedicalCollege, Allahabad, India

2Infectious Diseases Laboratory, Earle A. Chiles Research Institute,Providence Portland Medical Center, Portland, Oregon

*Correspondence to: Dr. S.J. Kohlhepp, Infectious Diseases Laboratory,Providence Portland Medical Center, 4805 NE Glisan Street, Portland, OR97213. E-mail: skohlhepp@providence.org

Received 27 August 2001; accepted 17 December 2001Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/dc.10089

262 Diagnostic Cytopathology, Vol 26, No 4 © 2002 WILEY-LISS, INC.

Portland Medical Center, Portland, Oregon. They included19 sputum and bronchial lavage samples, 14 CSF samples,8 body fluid samples including paracentesis and pericardialfluid samples, 4 lymph node samples, and 12 samples fromundesignated sites. All were suspected to be tubercular inetiology; however, neither acid fast staining nor culture wasperformed on the samples.

The first PCR procedure was performed with Amplicor™reagents and procedures purchased from Roche DiagnosticSystems (Branchburg, NJ). This procedure targets 16sRNAgenes and detects amplified products with a specific probeand colorimetric assay. All samples were checked for inhi-bition by the addition of previously extracted Mtb DNA asrequired in the Roche protocol.

The study PCR procedure used Roche extraction andprocedures published by Shankar et al.3 to amplify the genefor the MPB 64 protein. Briefly, the extraction procedureinvolved centrifuging a decontaminated sputum or bron-chioalveolar lavage sample through a wash solution. Non-respiratory samples were extracted without decontamina-tion. After centrifugation at 12,500g for 10 min thesupernatant was removed and discarded. Lysis buffer wasadded, the pellet resuspended, and heated at 60°C for 45min. Neutralization buffer was added and the samples takento the amplification step. We purchased primers from LifeTechnologies (Rockville MD). Primer 1 (5� to 3�) was TCCGCT GCC AGT CGT CTT CC. Primer 2 (5� to 3�) wasGTC CTC GCG AGT CTA GGC CA. Reagents for the PCRreaction were purchased from Perkin Elmer (Foster City,CA). Each PCR reaction tube contained 5 �l 10X PCRbuffer, 1 �l each dNTP, 50 nmoles both primers, 0.25 �lTaq (1.25 U), DNAase- and RNAase-free water to 45 �ltotal volume. Five �l of sample extracted by Roche reagentsand protocols was added to the 45 �l master mix for a totalvolume in the PCR reaction tube of 50 �l. PCR was per-formed in a Perkin Elmer™ 2400 thermal cycler using theprotocol: 94°C 5 min, 35 cycles of 94°C 30 sec, 55°C 1 min,72°C 1 min, followed by 72°C 7 min and 4°C hold. Theproduct was detected with ethidium bromide staining ofpolyacrylamide gel. These samples were also checked forinhibition by the addition of previously extracted DNA.Sensitivity and specificity of the second assay was com-pared to that of the Roche assay.

Control organisms M. avium subspp avium ATCC#25291, M. gordonae ATCC #14470, M. intracellularae

ATCC #13950, M. kansasii ATCC #12478, and M. tuber-culosis ATCC #25177 were purchased from either theAmerican Type Culture Collection (Rockville, MD) or Re-mel (Lenexa, KS). All organisms were suspended in sterilesaline at approximately 108 CFU/ml using 0.5 McFarlandstandard for comparison. Two hundred �l of these suspen-sions were extracted with Qiagen procedures and reagentsand amplified with the study PCR procedure.

ResultsAll samples yielded the same PCR result with both assays.Table I details the source of each sample as well as the PCRresults for both assays. One-third of the samples were spu-tum or bronchial lavage samples and one-quarter were ce-rebral spinal fluid samples. The next-largest category wasundesignated samples. Only one sputum sample was inhib-ited and it was inhibited in both assays. There were sevensamples positive for Mtb, four in the undesignated group,two in the sputum/bronchial lavage group, one in the otherbody fluid group, which upon further examination of therecords was a lymph node aspirate. Both assays detectedMtb in the same samples and found the same samples to benegative. There was complete agreement between the twoPCR procedures. No comparison to culture or stain wasavailable since samples were submitted only for molecularanalysis.

Figure 1 shows the electrophoretic pattern of PCR prod-ucts of Mtb, M. avium, and M. gordonii, M. intracellulari,and M. kansasii for the study procedure. Mtb yielded twoamplification products, one at 200 basepairs (bp) and thesecond, larger one at 730 bp. Mycobacterium avium yieldeda single 520 bp single product. Mycobacterium gordonaeyielded a single 395 bp product, M. intracellularae yieldeda single 130 bp product, and M. kansasii yielded two prod-ucts, one at 265 bp and the other at 300 bp. Clinical samplescan also produce multiple amplicons but this has not inter-fered with detection of Mtb amplicons.

DiscussionSeveral methods have been published for quick and accuratediagnosis of tuberculous lymphadenitis. Totsch et al.5 usedthree different in-house PCR methods with paraffin-embed-ded lymph node tissue and reported 100% sensitivity bycombining both methods on five serial sections. Tevere etal.6 presented an in-house PCR method with pan-Mycobac-

Table I. Mycobacterium tuberculosis PCR Results and Source of All Samples

Specimen

Roche procedure Study procedure

Positive Negative Inhibited Positive Negative Inhibited

Sputum 2 16 1 2 16 1CSF 14 14Body fluids 1 7 1 7Lymph nodes 4 4Undesignated 4 8 4 8

COMPARISON OF IN-HOUSE PCR METHOD

Diagnostic Cytopathology, Vol 26, No 4 263

terium primer and Mtb-specific probe and reported 91.9%sensitivity and 99.8% specificity using 662 respiratory tractsecretions. Kunakorn et al.7 reported on errors of in-housePCR assays when used to diagnose Mtb infections and someof the necessary precautions to reduce those errors. RochePCR has been evaluated for the diagnosis of mycobacteriain 17 fine-needle aspirates (FNAs) from cervical lymphnodes8 and showed 76% sensitivity and 100% specificity.Singh et al.9 compared in-house PCR with conventionaltechniques for the detection of Mtb DNA in granulomatouslymphadenopathy. Manitchotpisit et al.10 using an in-houseassay reported combined specificity of FNA and PCR incervical tuberculosis of 84% and sensitivity of 100%. In arecent study, Goel et al.11 compared PCR with conventionaltechniques (FNAC, acid fast staining, culture, and lymphnode biopsy) for diagnosis and found excellent correlation.However, they suggested that in developing countries PCRshould be reserved for cases in which there is a strongclinical suspicion, with equivocal results in the standardtests.

In our hands the study assay performed as well as thecommercial method, which has a published sensitivity of95% and specificity of 100% for AFB-positive sputa. Thestudy method could be completed within hours and is lesslabor-intensive than the commercial assay. The inclusion ofan inhibitor control was valuable in preventing false-nega-tive results from being reported. The study procedure hasthe advantage of being able to differentiate between M.tuberculosis, M. avium M. gordonae, M. intracellulari, andM. kansasii by PAGE. Clinical data, AFB staining, andhistological picture may suggest infection by Mtb but the

disease may actually be caused by an atypical mycobacte-rium. This procedure offers a rapid PCR-based system forthe direct detection of Mtb—thus assisting patient manage-ment by providing prompt supportive evidence of Mtbinfection, as opposed to that caused by other mycobacteria.

PCR has a remarkable sensitivity but it is expensive—especially if one has to rely on commercially prepared kits.We did a less than exhaustive estimate of the costs thatwould be incurred initially and subsequently to implementboth the Roche and the study assay in a laboratory with nohistory of PCR. The thermal cycler, precision pipettors, andplugged tips are common to both assays and their costs arenot included. To begin to perform the Roche procedure oneneeds $1,800 for the kits, $5,500 for a plate reader (Bio-Rad, Hercules, CA) and either $4,875 (Fisher, Orangeburg,NY) or $400 (Fisher) for either an automatic plate washer ora hand-held one. Cost will be approximately $12,000 to$7,700, depending on whether one uses an automatic or ahand-held plate washer. To implement the study assay onewould have to spend $225 for the PCR core reagent kit(Perkin-Elmer), $30 for the primers (Life Technologies,Bethesda, MD), $75 for gel reagents (Bio-Rad), $760 for thepower supply and tank, plates, casting stand, etc., to run thegel, and $2,750 for a gel documentation system from Bio-Rad or $2,200 for the UV light box and associated Polaroidcamera from Fisher. Until we have another extractionmethod we will use the Roche sputum preparation kit cost-ing $188 for the study assay. Cost will be approximately$3,840 to $3,290, depending on the documentation systemused. The continuing cost for the Roche assay would be$1,800/96 tests or $18.75 per test. The continuing cost of thestudy assay would be $255/150 tests for PCR reagents or$1.67 per test and $70/100 assays for gel reagents or $0.70per test—a total of $2.37 per test. The sensitivity andspecificity of the study assay is equal to the Roche Amplicorassay on this limited number (57) of samples and is a lessexpensive procedure both to set up and run.

ConclusionsWe have shown that an assay based on published methods3

using PAGE and ethidium bromide detection is less expen-sive and equal to the commercially available and FDA-approved Roche Amplicor™ procedure for M. tuberculosisdetection. Commercial assays are specific for M. tubercu-losis. If the organism is a mycobacterium other than tuber-culosis it will be negative in the commercial assays. Wehave found that this set of primers distinguishes between M.tuberculosis, M. avium, M. gordonae, and M. intracellulariand M. kansasii by PAGE. We suggest that this proceduremay have special relevance for those laboratories for whomthe commercial kit is either not available or is prohibitivelyexpensive.

Fig. 1. Electrophoretic pattern of study procedure PCR products fromMycobacterium tuberculosis, M. avium, M. gordonae, M. intracellularae,and M. kansasii. Ladder is 50, 150, 300, 500, 750, 1,000, 1,500, and 2,000bp. PCR products are electrophoresed at 200 volts for 45 min on 30%polyacrylamide gel.

MEHROTRA ET AL.

264 Diagnostic Cytopathology, Vol 26, No 4

References1. Thompson CH, Rose BR, Cossart YE. Detection of HPV DNA in

archival specimens of cervical cancer using in situ hybridisation andthe polymerase chain reaction. J Med Virol 1992;36:54–59.

2. Suffys P, Palimino JC, Cardoso Leao S, et al. Evaluation of thepolymerase chain reaction for the detection of Mycobacterium tuber-culosis. Int J Tuberc Lung Dis 2000;4:179–183.

3. Shankar P, Manjunath N, Lakshmi R, Aditi B, Seth P, Shriniwas.Identification of Mycobacterium tuberculosis by polymerase chainreaction. Lancet 1990;335:423–424.

4. Narita M, Shibata M, Togashi T, Kobayashi H. Polymerase chainreaction for detection of Mycobacterium tuberculosis. Acta Paediatr1992;81:141–144.

5. Totsch M, Bocker W, Brommelkamp E, et al. Diagnostic value ofdifferent PCR assays for the detection of mycobacterial DNA ingranulomatous lymphadenopathy. J Pathol 1996;178:221–226.

6. Tevere VJ, Hewett P, Dare A, et al. Detection of Mycobacteriumtuberculosis by PCR amplification with pan-Mycobacterium primers

and hybridization to a M. tuberculosis-specific probe. J Clin Microbiol1996;34:918–923.

7. Kunakorn M, Kanchana R, Pracharktam R, Sattaudom C. Overcomingthe errors of in-house PCR used in the clinical laboratory for thediagnosis of extrapulmonary tuberculosis. Southeast Asian J Trop MedPublic Health 1999;30:84–90.

8. Baek CH, Kim S, Ko YH, Chu KC. Polymerase chain reaction detec-tion of Mycobacterium tuberculosis from fine-needle aspirate for thediagnosis of cervical tuberculous lymphadenitis. Laryngoscope 2000;110:30–34.

9. Singh KK, Muralidhar M, Kumar A, et al. Comparison of in housepolymerase chain reaction with conventional techniques for the detec-tion of Mycobacterium tuberculosis DNA in granulomatous lymphad-enopathy. J Clin Pathol 2000;53:355–361.

10. Manitchotpisit B, Kunachak S, Kulapraditharom B, Sura T. Combineduse of fine needle aspiration cytology and polymerase chain reaction in thediagnosis of cervical tuberculosis. J Med Assoc Thai 1999;82:363–368.

11. Goel MM, Ranjan V, Dhole TN, et al. Polymerase chain reaction vs.conventional diagnosis in fine needle aspirates of tuberculous lymphnodes. Acta Cytol 2001;45:333–340.

COMPARISON OF IN-HOUSE PCR METHOD

Diagnostic Cytopathology, Vol 26, No 4 265