ORIGINAL ARTICLE

Evaluation of the Usefulness of PCR in the Diagnosis of Mycobacterium tuberculosis in Tissues and Body Fluids in UP-Philippine General Hospital
Rhoda Lynn C. Orallo, MD1, Myrna T. Mendoza, MD1, Mary Ann D. Lansang, MD1, Concepcion F. Ang, RMT2
ABSTRACT
Background: Numerous studies have been done to evaluate the performance of polymerase chain reaction (PCR) in extrapulmonary specimens in the rapid diagnosis of extrapulmonary tuberculosis (TB). However, due to variability in the sensitivity rates in different studies, the role of PCR remains controversial. Objective: To establish the accuracy of in-house PCR in the rapid diagnosis of TB in tissues and body uids. Methods: An IS6110-based primer in-house PCR protocol was used to detect DNA speci c for Mycobacterium tuberculosis in 124 non-respiratory specimens (tissues and body uids) from 121 patients. The results were compared with the results of cultures performed on Ogawa medium and MB Bact liquid medium, cold Kinyoun staining and histopathology as reference standards. Results: Out of 124 tissues and body uids from 121 patients, 116 specimens from 116 patients were evaluated. Thirty-six patients were classi ed as de nite TB cases, 32 patients were classi ed as clinically presumptive TB cases and 48 patients were non-TB cases. The overall sensitivity and speci city rates, and positive and negative likelihood ratios of the in-house PCR were 61% (95% CI 44%, 75%), 75% (95% CI 61%, 85%), 2.44 (95% CI 1.40, 4.26) and 0.52 (95% CI 0.33, 0.81), respectively, when compared with reference standards such as culture, AFB staining and histopathology results. When results were compared with culture, AFB staining and histopathology results together with the clinically presumptive TB cases as reference standard, the sensitivity and speci city rates, and positive and negative likelihood ratios were 54% (95% CI 43%, 66%), 75% (95% CI 0.61, 0.85), 2.18 (95% CI 1.27, 3.72) and 0.61 (95% CI 0.48, 0.83), respectively. Sub-group analysis done evaluating body uids and tissue biopsies separately showed that in-house PCR performed better on body uids than on tissues. Conclusions: Based on the results of our study, the use of in-house PCR show a potential role in con rming extrapulmonary tuberculosis in body uids. The relatively low sensitivity and speci city limit the use of this test in the detection and con rmation of the diagnosis of extrapulmonary tuberculosis. Keywords: In-house polymerase chain reaction (PCR), extrapulmonary tuberculosis, Mycobacterium tuberculosis

Section of Infectious Diseases, Department of Medicine, Philippine General Hospital, 2Medical Research Laboratory, Philippine General Hospital

20

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

Introduction
Tuberculosis continues to be a major global public health issue. The World Health Organization has estimated an annual incidence of 8.8 million cases of tuberculosis worldwide, with 1.8 million deaths. Most of the disease burden is in Africa and Southeast Asia, where annual incidence rates are 350 and 182 cases per 100,000 population, respectively (WHO)1. In the Philippines, tuberculosis remains in the top 6 leading causes of mortality, with a rate of 38.3 per 100,000 populations for the year 20042. Previous studies on TB patients have clearly demonstrated that early treatment is among the key factors affecting outcome. Culture and sensitivity of the Mycobacterium tuberculosis constitute the gold standard for the diagnosis of tuberculosis. However, these are often not helpful in making the diagnosis in paucibacillary forms of tuberculosis. Microscopy is rapid and inexpensive but exhibits low sensitivity (1020%)3,4. Culture, on the other hand, is not very sensitive (<50%), with results becoming available only after 4-8 weeks.3,4 Nucleic acid ampli cation (NAA) tests have emerged with the goal of enabling clinicians to make a rapid and accurate diagnosis. Polymerase chain reaction (PCR) is the best known NAA test5. It ampli es target nucleic acid regions that uniquely identify the Mycobacterium tuberculosis complex. Several nucleic-acid based ampli cation tests are available for the detection of Mycobacterium tuberculosis. An important advantage of NAA tests is the rapidity by which results can be obtained, about 3-6 hours from the receipt of specimen. The accuracy for NAA tests for tuberculosis has been extensively studied since the early 1990s. However, the exact role of these tests remains controversial. According to the US Centers for Disease Control and Prevention (CDC) and the American Thoracic Society (ATS), NAA tests improve diagnostic certainty but do not replace microscopy and culture.6,7 Currently, there are 4 commercial NAA

tests for the detection of M. tuberculosis: the Amplicor MTB test and its automated version, the Cobas Amplicor MTB test (Roche Diagnostics, Branchburg, NJ); the Enhanced Ampli ed M. tuberculosis Direct Test (E-AMTDT) (Gen-Probe, San Diego, CA); and the BDProbe Tec ET test (Becton Dickinson, Cockeysville, MD). The studies on PCR are mostly on the commercially prepared NAA test kits with speci c procedures and controls and few data are available with regards the use of in-house PCR for the diagnosis of TB. Studies evaluating in-house PCR in parallel with the commercially available NAA tests showed comparable sensitivity and speci city results. The study done by Eing8 compared the Roche Amplicor M. tuberculosis assay with an IS6110-based in-house PCR using TB culture as the gold standard. A total of 1,681 specimens from 833 patients, including specimen types other than sputum (pus, gastric aspirate, urine, biopsy tissues, pleural uid, CSF) were included in the study. The overall sensitivity, speci city, positive and negative predictive values of the Cobas Amplicor M. tuberculosis assay was 66.33, 99.71, 94.36, and 97.66%, respectively. The corresponding values for the in-house PCR were 91.08, 99.85, 97.87, and 99.37%, respectively. In another study by Huang,9 two commercial primer kits and detection systems, the Roche Amplicor M. tuberculosis test and the Digene primer probe kit with the SHARP signal system were compared with in-house PCR as well as standard culture techniques. 113 respiratory specimens from 85 patients were included in the study. The sensitivities of the Roche Amplicor, the Digene system, and the in-house PCR were 73.81, 69.05, and 80.95%, and the speci cities were 97.18, 91.55, and 88.73% respectively. However, in a meta-analysis done by Pai,10 the diagnostic accuracy of in-house PCR was highly variable as compared to the commercial nucleic acid ampli cation tests. The objective of the meta-analysis was to establish the diagnostic accuracy of nucleic acid ampli cation tests for tuberculous meningitis. Measures of diagnostic 21

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

accuracy were pooled using a random effects model and 49 studies were included. Their summary estimates in 14 studies with commercial NAA tests were: sensitivity 0.56 (95% CI 0.46, 0.66), speci city 0.98 (0.97, 0.99), positive likelihood ratio 35.1 (19.0, 64.6), and a negative likelihood ratio 0.44 (0.33, 0.60). However, the summary estimates in the 35 studies with inhouse PCR could not be established due to wide variability in test accuracy. Most of the PCR studies done on respiratory specimens show that the sensitivity of PCR ranges from 74% to as high as 98% and speci city of 85% to as high as 100%.11,12,13 However, this method failed to show consistent results in extrapulmonary specimens, with sensitivity ranging from 45% to as high as 95%.14,15,16,17,18 As such, the ampli cation techniques do not yet have an established role in the diagnosis of extrapulmonary tuberculosis. In a study in Turkey by Ozkutuk evaluating the performance of Cobas Amplicor MTB (CAMTB) test for pulmonary and extrapulmonary specimens, results indicate that the CA-MTB is a rapid test for detection of tuberculosis in pulmonary specimens, but does not perform well enough in extrapulmonary specimens. A total of 424 specimens were obtained from the suspected TB patients from January 2003 to August 2004. All specimens (173 pulmonary and 251 extrapulmonary specimens) were processed, stained, cultured and assayed using the CAMTB test for identi cation of Mycobacterium tuberculosis. The CA-MTB test results were compared to culture and acid-fast staining as gold standard. The sensitivity, speci city, positive and negative predictive values of CA-MTB were determined as 73%, 100%, 100%, and 97% for pulmonary specimens, and 45%, 100%, 100% and 96% for extrapulmonary specimens respectively. The sensitivity of the test for acidfast bacilli (AFB) smear positive pulmonary and extrapulmonary specimens was 92% and 75%.
19

AMLPICOR for detecting cases of de nite and probable TBM in patients from whom CSF specimens have been collected less than 10 days into antituberculous treatment was 60.0% and a speci city of 100%. A study was done by Lemaitre21 comparing TB culture with real time PCR and Ampli ed Mycobacterium tuberculosis Direct Test (AMTDII) in 100 respiratory and nonrespiratory (lymph node, tissues, urine, synovial uid, abscess and pleural uid) clinical specimens. The overall sensitivities of the real-time PCR method and AMTDII were similar for respiratory and nonrespiratory specimens. For real-time PCR, sensitivity, speci city, NPV and PPV for nonrespiratory specimens were 80, 100, 92 and 100% respectively. The corresponding values for AMTDII were 70,100, 89 and 100% respectively. In a study done by Hasaneen,22 PCR was applied to detect the Mycobacterium tuberculosis in 33 of the pleural biopsy specimens, and compared it to the results of pleural uid and biopsy cultures performed on either Lowenstein medium or BACTEC 12B liquid medium, Ziehl-Neelsen staining and histopathology in 45 patients with pleural effusion. The overall accuracy of PCR of pleural biopsy was similar to the results of pleural biopsy culture. PCR of pleural biopsy specimens had 90% sensitivity and 100% speci city. The PPV and the NPV for pleural biopsy specimen cultures were 100% and 90.5% vs 100% and 86.7% for pleural biopsy specimen PCRs. This study is limited by its low sample size. These 2 succeeding large studies done in Greece showed a relatively good performance of the commercially available PCR kits in the diagnosis of TB in respiratory and nonrespiratory specimens. An 8-year prospective study (January 1995-January 2003) was done by Michos23 evaluating the diagnostic performance of Amplicor MTB on 2,296 respiratory (sputum and BAL) and nonrespiratory specimens (gastric uid, pleural, pericardial, abscess, urine, bone) from 2,296 patients with suspected tuberculosis. All specimens were examined blindly by direct microscopy, culture and PCR. The sensitivity,

A prospective study was done by Bonington20 evaluating the use of Roche AMPLICOR TB PCR in the diagnosis of tuberculous meningitis. The sensitivity of TB 22

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

speci city, and positive predictive value of the assay was 97.2, 100, and 100% for smear positive specimens and 75.3, 97, and 47.5% for smear negative specimens respectively. The lowest sensitivities were observed in pleural uid (60%) and abscess aspirates (64%). Likewise, in another 4-year prospective study (September 1998-December 2001) by Levidiotou24, 7,722 respiratory (sputum, bronchial aspirates, BAL) and 1,451 nonrespiratory (urine, pleural uids, gastric aspirates, blood, pus, CSF, synovial uid, ascitic uid, pericardial uid, biopsy specimens and BMA) specimens were collected from 3,321 patients with signs and/or symptoms suggesting pulmonary or extrapulmonary TB. Cobas Amplicor MTB was performed and results were compared to those of culture. For respiratory specimens, the overall sensitivity, speci city, positive and negative predictive values of CAMTB 84.5, 99.8, 94.1 and 99.4% respectively. For nonrespiratory specimens, the overall sensitivity, speci city, positive and negative predictive values of CA-MTB 82.5, 99.8,94.3 and 99.4%, respectively, when compared with the results of diagnostic culture. In the University of the PhilippinesPhilippine General Hospital (UP-PGH), a study done by Montoya and Berba25 using CSF specimens in the diagnosis of tuberculous meningitis, compared the usefulness of the AMPLICOR MTB PCR and PCR with nested ampli cation. The results showed low sensitivity rates for both AMPLICOR MTB PCR (30.6%) and PCR with nested ampli cation (36.1%). The speci city of both tests was the same at 79.3%. The authors cited some of the possible reasons for the low sensitivity and speci city in both the AMPLICOR PCR and nested PCR, such as history of previous antituberculous treatment or ongoing antituberculous therapy when the specimens were collected; aliquot phenomenon; procedure-related problems during nucleic acid extraction; and denaturation. The UP-PGH is equipped with a standard PCR method using the element of IS6110 for the detection of Mycobacterium tuberculosis.

However, its usefulness in terms of its accuracy and its relevance in the local setting has not been thoroughly evaluated. Because of the need for further evaluation of these tests, especially the in-house PCR in the diagnosis of extrapulmonary tuberculosis, this study was undertaken, with the goal of improving its clinical usefulness and applicability in the local setting. The aim of this study was to evaluate the usefulness of our in-house PCR in the diagnosis of tuberculosis, speci cally in tissues and body uids. Speci cally, it aims to establish the overall accuracy (sensitivity and speci city) and to evaluate the clinical usefulness (likelihood ratios) of PCR in the diagnosis of TB in tissues and body uids.

Methods
Participants
The study took place at the UP-PGH, particularly in the following wards: Medical, Surgical, Neurology, Neurosurgical, Orthopedic and Pediatric Departments. The study included both adult and pediatric patients, both male & female, admitted at any of the wards of the UP PGH from March 1, 2006-February 28, 2007. Tissue (including bone) biopsies, body uids (such as cerebrospinal, pleural, peritoneal, pericardial, synovial uids and urine) and aspirates from hepatic, renal, subcutaneous or intramuscular abscesses were collected prospectively from admitted patients with suspected tuberculosis. The starting point of the study was referrals from the emergency room or from the different wards where a diagnosis of extrapulmonary tuberculosis had been entertained.

Inclusion criteria
Any age group Admitted at the medical, surgical, neurology, neurosurgical, orthopedic or pediatric wards For suspected cases of CNS TB, 2 or more of the following should be present: low-grade fever, 2-3 weeks of protracted headache, 23

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

24

vomiting, confusion, meningismus or focal neurologic signs For suspected cases of tuberculous effusions, the following should be present: fever, 2-3 weeks of cough with pleuritic chest pains AND unilateral pleural effusion by CXR For suspected cases of tuberculous pericarditis, 2 or more of the following should be present: fever, dyspnea, chest pains, tachycardia, distant heart sounds, friction rubs, distended neck veins, hepatomegaly, paradoxical pulse, low-voltage QRS complexes on EKG, cardiomegaly on CXR AND pericardial effusion OR pericardial brosis or thickening on 2D echocardiography For suspected cases of skeletal TB (Potts Disease), any of the following should be present: 2-3 weeks history of back pain/stiffness or weakness, tender spinal prominence or gibbus, or paralysis of the lower extremities AND x-ray ndings of anterior wedging of 2 adjacent vertebral bodies with destruction of intervening disk; and for suspected cases of peripheral osteoarticular TB, any of the following should be present: monoarticular or polyarticular arthritic pain and swelling of 2-3 weeks duration, periarticular cold abscesses and draining sinuses formation AND x-ray ndings of periarticular bony destruction For suspected cases of genitourinary TB, 2 or more of the following should be present: back/ ank pains, dysuria, frequency, epidydimitis/orchitis, scrotal mass, chronic prostatitis, sterile pyuria AND abnormal IVP ndings For suspected cases of abdominal TB (which includes the gastrointestinal tract, peritoneum, lymph nodes, liver spleen and pancreas), 2 or more of the following should be present: fever, weight loss, anorexia, abdominal pain/ abdominal distention, abdominal mass, signs of intestinal obstruction, diarrhea, jaundice, hemorrhage, malabsorption AND ultrasound ndings of ascites or enlarged lymph nodes For suspected cases of tuberculous lymphadenitis, the following should be

present: unilateral, single or multiple painless, rm lymphadenopathies in the cervical or supraclavicular areas with or without sinus tract formation Patients with suspected extrapulmonary tuberculosis on anti-tuberculosis treatment for less than 1 month

Exclusion criteria
Patient with suspected extrapulmonary tuberculosis on 1 month of anti-TB medications

Reference Standard
The diagnosis of TB was con rmed with positive AFB smear and/or positive growth of Mycobacterium tuberculosis and/or tissue diagnosis of chronic granulomatous in ammation with caseation necrosis and Langhans type giant cells and was considered as the reference standard of the study. Each sample (tissue or body uid) was divided into 3 portions: one portion was used for AFB smear, second portion was used for TB culture and the third portion was used for PCR. All of these procedures were done at the Infectious Diseases Section- Medical Research Laboratory (IDS-MRL), UP-PGH. One laboratory technician performed the AFB smear and TB culture while another laboratory technician performed the PCR. A separate set of tissue samples (taken at the same area where specimens for AFB smear/TB culture and TB PCR were biopsied) were submitted to the pathology section of the UP-PGH for histopathologic diagnosis. A secondary analysis was done using AFB smear, TB culture, histopathology together with clinically presumptive TB diagnosis as reference standard. Clinically presumptive TB are cases of TB with a negative AFB smear, no growth on culture, histopathologic report of chronic granulomatous in ammation without caseation necrosis or Langhans type giant cells but clinical picture was consistent with TB, biochemical and radiologic imaging compatible with TB and clinical response to treatment determined 2 months after commencing antituberculous treatment.

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

Procedure for Processing Tissue Specimens

Tissue samples were placed in a sterile vial with NSS or distilled water. The tissue samples were cut (approximately 5mg) into smaller pieces with sterile surgical knives on a sterile petri dish or were ground with sterile pestle and mortar. Tissue samples were then placed on a 50 ml sterile conical tube. The tissue samples were diluted with sterile 0.9% NaCl. Sterile glass beads were added and vortexed to further homogenize the sample. At least 400 to 1000 L of the soup was taken for PCR. The remaining sample was processed for AFB smear and TB culture. About 10 L of the supernatant was placed to a clean glass slide where it was air dried, xed and then stained with Cold Kinyoun method. 0.01 ml of the supernatant was inoculated into to an Ogawa medium and 0.5 ml to an MB Bact liquid medium and incubated at 37oC. Growth on the culture media was observed until 6 weeks.

Procedure for in-house PCR

Samples for PCR have undergone the following steps: 1. Extraction of target DNA One ml of the sample was centrifuged at 13,000 x g for 10 minutes at 4oC. The pellet was resuspended in 1ml 20 mM Tris HCl buffer, vortexed and spinned again for 10 minutes. The supernatant was discarded and the pellet was resuspended in 80 L of 20 mM of Tris HCL. The 10 x digestion buffer was added to the suspension (5% Tween 80, 2mg/ml proteinase K, 0.2 M Tris HCl ph 8.3) and then incubated at 60oC for 1 hour to 18 hours depending on the type of specimen. Samples were boiled for 15 minutes at 100oC to inactivate the proteinase K. 2. PCR (DNA ampli cation) PCR was performed in a 50 L reaction mixtures containing 1X PCR buffer; 1.5mM of MgCl2; 250 M concentrate of each deoxynucleotide triphosphate; 1U Taq polymerase; 10 M each of the INS 1 and INS 2 primers (5 CGT GAG GGC ATC GAG GTC GC 3 5 GCG TAG GCG TCG GTG ACA AA 3 ) and 10 l of the DNA. Ampli cation was carried out following 95oC for 5 minutes (1x), 95oC for 1 min; 65oC for 30 seconds; 72oC for 2 minutes for 35 cycles and a nal extension at 72oC for 10minutes (1x). 3. Analysis of DNA (Gel electrophoresis) Ten L of the of the PCR product was mixed with 2.0 L of 6x loading dye and loaded into 1.5% agarose gel containing 1.0 L of ethidium bromide and was run at 30 minutes at 100 V in 1x TBE buffer. Expected product of 245 bp was then visualized in the UV transilluminator and photographed with Polaroid camera.

Procedure for Processing Body Fluids

All body uids could be directly processed for DNA extraction, AFB smear and TB culture. About 400 to 1000 L of the uid was placed in a microcentrifuge tube for DNA extraction. The remaining pellet was inoculated (0.01ml) to an Ogawa medium and to the MB Bact Liquid medium (0.5ml) and incubated at 37oC. Growth on the culture media was observed until 6 weeks. All specimens for TB culture, if not immediately submitted to the IDS-MRL, were stored in the refrigerator for a maximum of 72 hours except for CSF specimens. CSF specimens for TB culture was immediately sent to the laboratory and planted in the culture medium. All specimens for TB PCR (both tissues and body uids), if not processed immediately were stored in the freezer at -70oC.

Quality Control Procedures

To eliminate false positive or false negative 25

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

results, each run included a positive genomic and specimen control. The negative control consisted of distilled water. To correct the de ciencies mentioned in the study by Montoya and Berba et al, the following were undertaken in the present study: 1. To eliminate false negative results, patients who have been on 4 weeks of anti-TB medications were excluded in the study. According to the studies by Lin26 and Donald,27 the sensitivity of PCR falls considerably if the patient is on more than 4 weeks of anti-TB treatment. 2. Since the extraction of nucleic acid is probably, the most critical step in the PCR that will determine the amount of nucleic acid available, procedure for the DNA extraction has already been optimized in accordance with the internationally accepted standards. 3. To prevent denaturation from happening, proper specimen handling was practiced at all times. Once specimen was collected, it was stored in the freezer promptly or submitted to the IDS-MRL. 4. To avoid the aliquot phenomenon, at least 10cc of body uid was collected and centrifuged for PCR, and 5cc for culture and smear. For tissues, the specimen was macerated with NSS and the soup was taken for PCR, culture and smear.

Results
The study included 124 non-respiratory specimens from 121 patients. Five patients were excluded in the study: 2 patients were already on more than 1 month of antituberculous therapy when the specimens were collected, 2 patients were lost to follow-up and 1 patient whose diagnosis of TB was doubtful but could not be totally ruled out with con dence. There were 116 non-respiratory specimens collected (CSF, pleural uid, synovial uid, pericardial uid, ascitic uid, vitreous uid, bile, urine, abscess aspirates, bone marrow aspirates, tissue biopsies) from 116 patients. There were 63 males (54%) and 53 females (46%). The mean (SD) age was 39 (16.45) years. The patients were categorized into the following 3 groups: Group 1: These were patients with de nite tuberculosis. This group included 36 patients, with 36 clinical specimens. Of these 36 clinical specimens, 20 were smear positive, 25 showed growth of Mycobacterium tuberculosis, and 8 tissue biopsies had histopathologic diagnosis of TB (Table 1). There were 22 PCR positive specimens.
Table 1. AFB staining, culture and histopathologic results of different types of body uids and tissue biopsies, UP-PGH (n =116)
Types of specimen n AFB TB Histopath smear Culture + + + 0 3 2 0 0 0 1 2 4 0 8 20 1 2 2 0 1 1 2 2 5 1 7 25 0 0 0 0 0 0 0 0 0 0 8 8

Statistical Analysis
The sensitivity and speci city rates and positive and negative likelihood ratios of the assay were calculated using as a reference standard the results of the cold Kinyoun staining examination and/or culture and/or histopathology. Secondary analysis was also done using the cold Kinyoun staining examination and/or culture and/or histopathology and clinically presumptive TB diagnosis as reference standard. 26

Cerebrospinal uid Pleural uid Synovial uid Pericardial uid Ascitic uid Vitreous uid Bile Urine Abscess aspirates Bone marrow aspirates Tissue biopsies Total

28 19 8 2 5 1 3 3 8 9 30 116

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

Group 2: These were patients considered as having clinically presumptive TB. This group included 32 patients, with 32 clinical specimens. The speci c ndings of the 32 patients with clinically presumptive TB are listed in Table 2. Of

the 32 clinical specimens, 15 were PCR positive. 14 PCR positive specimens were body uids (7 CSF, 4 pleural uids, 2 synovial uids, 1 ascitic uid) and 1 PCR positive specimen was a bone tissue.

Table 2. Speci c ndings of patients with clinically presumptive TB (N=32)

Findings N

TB Meningitis (N=15) a. CSF lymphocytic pleocytosis with elevated protein and decreased glucose 13 b. Cranial CT with hydrocephalus 9 + basal enhancement 4 + granuloma 2 + arachnoiditis 1 c. Clinical response to treatment 10 Pleural TB (N=7) a. Exudative pleural effusion with lymphocytic predominance 7 b. Unilateral pleural effusion on CXR 5 + pulmonary lesions 1 c. Clinical response to treatment 7 Abdominal TB (HBT, GI, peritoneal TB) (N=3) a. Ascites with lymphocytic predominance and negative bacterial cultures 1 b. UTZ ndings of dilated intrahepatic ducts with stricture formation, hepatic abscesses 1 and lymphadenopathies 2 c. Colonoscopic ndings of ileocecal mass with histologic tissue diagnosis of chronic granulomatous in ammation 1 d. Clinical response to treatment 3 Skeletal TB (Potts Disease, osteoarticular TB) (N=5) a. Exudative synovial effusion with lymphocytic predominance with negative cultures 2 b. Radiologic ndings of destruction of intervertebral disc and adjacent vertebrae 3 + psoas abscess c. Radiologic ndings of soft tissue swelling and joint space narrowing 2 d. Histologic tissue diagnosis of chronic granulomatous in ammation 2 e. Clinical response to treatment 5 TB Pericarditis (N=1) a. Exudative pericardial effusion with lymphocytic predominance 1 b. Histologic tissue diagnosis of chronic granulomatous in ammation 1 c. Clinical response to treatment 1 TB of the Bone marrow (N=1) a. Pancytopenia 1 b. Normocellular bone marrow with trilineage dyspoiesis 1 c. Evidence of con rmed tuberculosis at other sites/ disseminated TB 1 d. Clinical response to treatment 1

Group 3: These were patients with non-tuberculous infectious diseases and non-infectious diseases. This group included 46 patients, with 46 clinical specimens having a de nite or proven diagnosis of fungal, viral or bacterial infections as well as patients with carcinoma. Of the 46 clinical specimens, 12 were PCR positive, of which 5 were tissue biopsies, 7 were from different body uids. 5 clinical specimens were diagnosed cases of malignancy, 1 case of subarachnoid hemorrhage, 2 infectious colitis cases, and 1 case of ulcerative colitis, 1 case of uremia and 1 case of post-operative bacterial meningitis.
Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

27

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

The overall sensitivity, speci city, positive and negative likelihood ratios of the in-house PCR were 61% (95% CI 44%, 75%), 75% (95% CI 61%, 85%), 2.44 (95% CI 1.40, 4.26) and 0.52 (95% CI 0.33, 0.81), respectively, when compared with reference standards such as culture, AFB staining and/or histopathology results (Table 3). When results were compared with culture, AFB staining and/or histopathology together with the clinically presumptive TB cases as reference standards, the sensitivity, speci city, positive and negative likelihood ratios were 54% (95% CI 43%, 66%), 75% (95% CI 61%, 85%), 2.18 (95% CI 1.27, 3.72) and 0.61 (95% CI 0.48, 0.83), respectively (Table 4). In our study, the 2 most frequent body uids were CSF specimens (25%) and pleural uids (16%). Of the 29 CSF specimens, 1 was a de nite case of TB meningitis, which was PCR positive. Of the 15 cases of clinically presumptive TB, 7 were PCR positive. Of the 8 PCR negative clinically presumptive TB meningitis cases, 7 patients clinically improved with anti-tuberculous treatment while 1 patient died of brain herniation. There were also 2 cases of falsely positive results, 1 was a case of subarachnoid hemorrhage and 1 was a case of post-operative bacterial meningitis. Of the 19 pleural uids, there were 3 de nite cases of pleural TB, 1 was PCR positive. There were 7 cases of clinically presumptive pleural TB, 5 of whom were PCR positive. There were no cases of falsely positive results.

In a sub-group analysis of 69 body uids (CSF, pleural uid, synovial uid, pericardial uid, ascitic uid, vitreous uid, bile, urine), the sensitivity, speci city, positive and negative likelihood ratios of the in-house PCR were 64%(95% CI 39%, 84%), 76%(95% CI 59%, 89%), 2.75(95% CI 1.30, 5.87) and 0.47(95% CI 0.22, 0.97), respectively, when compared with AFB staining and culture results (Table 3). When results were compared with AFB staining and culture together with the clinically presumptive TB cases as reference standard, the sensitivity, speci city, positive and negative likelihood ratios were 55%(95% CI 40%, 69%), 76%(95% CI 60%, 88%), 2.40(95% CI 1.19, 4.86) and 0.57(95% CI 0.38, 0.85), respectively (Table 4). In a sub-group analysis of 30 tissue biopsies (mucosal, pancreatic, hepatic, splenic, brain, bone, pleural, pericardial, skin), the sensitivity, speci city, positive and negative likelihood ratios of the in-house PCR was 56%(95% CI 0.33, 0.77), 44%(95% CI 0.19, 0.73),1.01(95% CI 0.49, 2.10) and 0.98(95% CI 0.39, 2.46), respectively, when compared with AFB staining, culture and/ or histopathology results (Table 3). When results were compared with culture, AFB staining and/ or histopathology together with the clinically presumptive TB cases as reference standards, the sensitivity, speci city, positive and negative likelihood ratios were 50%(95% CI 0.30, 0.70), 44%(95% CI 0.19, 0.73), 0.90(95% CI 0.44,1.85) and 1.12(95% CI 0.48, 2.61), respectively (Table 4).

Table 3. Overall performance of in-house PCR in the diagnosis of TB

Patient Classi cation De nite TB All specimens Body uids (except BMA and abscess) Tissue biopsies 36 14 20 7 24 10 8 0 22 9 14 5 61%(45%, 75%)/ 75%(61%, 85%) 64%(39%, 84%)/ 76%(59%, 89%) 56%(33%, 77%)/ 44%(19%, 73%) 2.44(1.40, 4.26)/ 0.60(0.33, 0.81) 2.75(1.30, 5.87)/ 0.47(0.22, 0.97) 1.01(0.49, 2.10)/ 0.98(0.39, 2.46) Number of AFB TB Histopath specimens smear + culture + + PCR + PCR Sensitivity/Speci city (95% CI) Positive/ Negative LR (95% CI)

18

28

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

Table 4. Performance of in-house PCR in patients with de nite TB and clinically presumptive TB
Patient Classi cation De nite and Clinically Presumptive TB All specimens Body uids (except BMA and abscess) Tissue Biopsies Number of specimens PCR + PCR Sensitivity/Speci city (95% CI) Positive/ Negative LR (95% CI)

71 49 22

36 23 10

32 18 10

54%(43%, 66%)/ 75%(61%, 85%) 55%(40%, 69%)/ 76%(60%, 88%) 50%(30%, 70%)/ 44%(0.19, 0.73)

2.18(1.27, 3.72)/ 0.61(0.45, 0.83) 2.40(1.19, 4.86)/ 0.57(0.38, 0.85) 0.90(0.44, 1.85)/ 1.12(0.48, 2.61)

The sensitivity and speci city rates were generally lower when AFB smear, culture and/or histopathology together with the clinically presumptive TB cases were used as reference standard than when AFB staining, culture and/or histopathology alone was used as reference standard (Table 4). A separate analysis was done for BMA and abscess aspirates and the results are as follows. There were 10 bone marrow aspirates, 1 was culture positive but PCR negative and 1 BMA was clinically presumptive TB case which was also PCR negative. The remaining 8 BMA specimens were non-TB infections and were all PCR negative. Out of 7 abscess aspirates, 5 were culture positive, of which 3 were PCR positive. There were no false positive results for abscess aspirates.

Discussion
The diagnosis of extrapulmonary tuberculosis is made dif cult by the complexity and risks involved in the collection of specimen for examination. It frequently requires invasive or semi-invasive approaches, and often low yield of direct examination and culture of material obtained due to the paucibacillary nature of the disease. This leaves clinicians with a long delay without an af rmative answer. But with the development of newer techniques in molecular biology, this delay in the accurate diagnosis of the disease is minimized. Our study showed a seemingly low sensitivity rate of our in-house PCR which was within the range of sensitivities, between 45% and 95%, as noted in the previous studies. This low sensitivity rate in our study could be attributed to the procedure in sample preparation and DNA extraction. Optimal sample preparation is needed to reduce inhibitors that interfere with

detection. Monitoring and accurate evaluation of PCR inhibitors is possible only if an internal control is included during DNA ampli cation.28 Our in-house PCR uses the proteinase K as pretreatment in sample preparation for the extraction of target DNA but an internal control to detect inhibitors is absent. A study done by Honore-Bouakline29 compared 2 sample preparation methods, pretreatment with proteinase K (PK-Roche) and complete DNA puri cation using cetyltrimethylammonium bromide (CTABRoche) on 144 extrapulmonary specimens. The results showed that CTAB-Roche produced the best sensitivity (53.8%) compared to culture (43.3%), PK-Roche (16%), and smear (13%). High rates of inhibition were obtained particularly in cutaneous biopsy specimens, BMA and abscesses. However, extraction procedures employing multiple steps to purify DNA, as the CTAB-Roche method, appeared to have lower rates of inhibition. This result shows that 29

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

it is important to prepare a good puri ed DNA sample and eliminate inhibitors in order to make a precise PCR diagnosis, even if it requires a more tedious method. Inhibitors such as blood, heparin, phenol and other substances might be the most possible reasons for such low sensitivity results.23,28,29 As in our study, most false-negative results were seen in tissue biopsies, abscess aspirates and BMA. These specimens most likely contained blood and other substances that inhibit ampli cation. This reaf rms the need for an internal control to detect the presence of such inhibitors and eliminate false-negative results with PCR. A positive genomic and specimen control, as well as a negative control consisting of distilled water which is used by our in-house PCR for quality control measures might be inadequate to recognize these inhibitors. Prior studies have shown that PCRs sensitivity is related to the number of organisms present in the sample.30 However, in our study, there is only a slightly higher PCR positive results observed in smear-positive specimens (11 PCR positives out of 20 smear-positive specimens) with a strikingly higher PCR positive results in smear-negative specimens, with de nite TB (10 PCR positive results out of 16 smearnegative specimens) and in effect, presented the greatest potential bene t for these patients. The apparently lower sensitivity rates for smearpositive specimens could still be attributed to the presence of inhibitors since most of the smearpositive, PCR negative specimens were mostly from tissues and abscess aspirates. This may con rm the potential use of our in-house PCR in body uids, since CSF, pleural, pericardial, synovial, urine and bile uids are relatively sterile specimens and may contain less inhibitor. In the present study, the sensitivity of the assay was higher when AFB smear, culture and histopathologic diagnosis were used as reference standard, as compared when clinically presumptive TB diagnosis was added to the reference standard. Our hypothesis was that the best potential advantage of our in-house 30

PCR should have been seen in patients with clinically presumptive TB owing to the very few tubercle bacilli in extrapulmonary TB. However, although clinical response to anti-tuberculous therapy and high prevalence of TB in our country would strongly suggest the disease, some of our cases may have been misclassi ed as clinically presumptive TB, resulting in the underestimation of the performance of the inhouse PCR. This would imply that it may not be possible to evaluate precisely the accuracy of this assay without a gold standard such as AFB smear, culture and histopathologic diagnosis. While the speci city rate of PCR in the previously mentioned studies ranges from 88 to 100%, a signi cant nding in our present study is its low speci city which challenges our knowledge of the PCR results. For this assay to be clinically useful in our local setting, it is critically important that false-positive results are eliminated and only de nite cases of TB are detected given the cost and adverse drug events related to administration of anti-tuberculous therapy. Most of the falsepositive specimens were from patients with malignancy. Nonetheless, we cannot question the fact that patients who are immunocompromised are at risk to develop TB31, and misclassifying these patients as no TB may be a possibility. However, our diagnosis for no TB patients in our study was based on our reference standard and therefore, the likelihood of misclassifying our patients is not possible. Other false-positive results were cases of non-tuberculous infections, subarachnoid hemorrhage and uremia and were treated as such with noted clinical improvement. Another reason for low speci city could be crosscontamination during the procedure, which is a common problem in laboratories using in-house protocols. Efforts to reduce contamination in our study included the observation of a separate room policy, all pipette tips were used only once and all pipettes were dedicated to a single specimen. Based on the ndings presented here, our in-house PCR may have a potential role in the rapid diagnosis of extrapulmonary TB in body uids, showing a likelihood ratio for a positive test

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

of 2.18 (twice as likely) and a negative likelihood ratio of 0.61 (not very useful negative test, i.e., poor sensitivity). As to its role in tissue biopsies, BMA and abscess aspirates, further studies are recommended to adequately eliminate inhibitors that prevent ampli cation. Further studies should also be done to reduce the nonspeci c ampli cation by the use of chemicals such as psolaren and UV irradiation or use of enzymes such as uracyl-DNA glycosylase that prevent cross-contamination with amplicons. A positive AFB smear with a positive PCR would con rm the diagnosis of TB; however a negative smear with a positive PCR would necessitate clinical correlation. The relatively low sensitivity and speci city limit the use of this test in the detection and con rmation of the diagnosis of extrapulmonary tuberculosis.

Ms. Richell Mojica-Ador for all her help in the completion of this study.

References
1. World Health Organization. Global tuberculosis control. WHO Report 2001. Geneva: World Health Organization, 2001:18-19. 2. Philippine Health Statistics. Department of Health 2004. 3. Thwaites GE, Chau TT, Stepniewka K, et al. Diagnosis of adult tuberculous meningitis by use of clinical and laboratory features. Lancet 2002; 360: 1287-92. 4. Thwaites G, Chau TT, Mai NT, Drobniewski F, McAdam K, Farrar J. Tuberculous meningitis. J Neurol Neurosurg Psychiatry 2000; 68: 289-99. 5. American Thoracic Society. American Thoracic Society Workshop. Rapid diagnostic tests for tuberculosis. What is the appropriate use? Am J Respir Crit Care Med 1997; 155: 1804-14. 6. Centers for Disease Control and Prevention. Update: nucleic acid ampli cation tests for tuberculosis. Morbidity Mortality Weekly Report 2000; 49: 593-94. 7. American Thoracic Society. Diagnostic standards and classi cation of tuberculosis in adults and children. Am J Respir Crit Care Med 2000; 161: 1376-1395. 8. Eing B, Becker A, Sohns A, Ringelmann R. Comparison of Roche Cobas Amplicor Mycobacterium tuberculosis Assay with Inhouse PCR and Culture for Detection of M. tuberculosis. J Clin Microbiol 1998; 36:20232029. 9. Huang T, Liu Y, Lin H, Huang W, Cheng D. Comparison of Roche Amplicor Mycobacterium tuberculosis Assay and Digene SHARP Signal System with Inhouse PCR and Culture for Detection of M. tuberculosis in Respiratory specimens. J Clin Microbiol 1996; 34:3092-3096. 10. Pai M, Flores L, Pai N, Hubbard A, et al. Diagnostic accuracy of nucleic acid ampli cation tests for tuberculous meningitis: a systematic review and meta-analysis. The LANCET 2003; 3:633-643. 11. Cheng VC, Yam WC, Hung IF, et al. Clinical evaluation of the polymerase chain reaction 31

Acknowledgments
Completion of this research would not have been possible without the support of the following: Dr. Benjamin Limson Memorial Research Grant awarded to the principal author during the 2006 Annual Convention of the Philippine Society for Microbiology and Infectious Diseases; Foundation for the Control of Infectious Diseases, Inc.; and Philippine College of Physicians. The authors would like to thank the following: Dr. Sang Nae Cho for his expert technical advise. The fellows of the Section of Infectious Diseases, and to the residents of the Departments of Medicine, Surgery, Orthopedics, Pediatrics and Neurosciences most especially Dr. Jennifer De Guzman, for giving us a hand in gathering clinical data. The staff of the Infectious Disease SectionTB Research Laboratory, Medical Research Laboratory, UP-PGH - Wilma Bulatao and Marc Agnew Cajucom for rendering their technical expertise in the laboratory aspect of the study.

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008

Orallo RL C, et al

Usefulness of PCR in the Diagnosis of Mycobacterium Tuberculosis

for the rapid diagnosis of tuberculosis. J Clin Pathol 2004; 57:281-285. 12. Ashok Rattan. PCR for the diagnosis of tuberculosis: Where are we now? Indian Journal of Tuberculosis 2000; 47:79-82. 13. Bergmann JS, Woods GL. Clinical evaluation of the Roche AMPLICOR PCR Mycobacterium tuberculosis test for detection of M. tuberculosis in respiratory specimens. Journal of Clinical Microbiology 1996; 34:1083-1085. 14. Shah S, Miller A, Mastellone A. Rapid Diagnosis of Tuberculosis in various biopsy and body uid specimens by the Amplicor Mycobacterium tuberculosis PCR test. CHEST 1998; 113:1190-1194. 15. Scarparo C, Piccolo P, Rigon A. Comparison of Enhanced MTB Ampli ed Direct Test with Cobas Amplicor Mycobacterium tuberculosis Assay for Direct Detection of M. tuberculosis complex in respiratory and extrapulmonary specimens. J Clin Microbiol 2000; 38:15591662. 16. DAmato R, Hochstein L, Colaninno P. Application of Roche Amplicor Mycobacterium tuberculosis PCR test to specimens other than respiratory secretions. Diagn Microbiol Infect Dis 1996; 24:15-17. 17. Piersimoni C, Scarparo C, Piccolo P, Rigon A. Performance Assessment of Two Commercial Ampli cation Assay for Direct Detection of M. tuberculosis complex for respiratory and extrapulmonary specimens. J Clin Microbiol 2002; 40:4138-4142. 18. Yee Y, Gough A, Kumarasinghe G. The pattern and utilization of accuracy of a commercial NAA test for the rapid diagnosis of M. tuberculosis in routine clinical practice. Singapore Med J 2002; 43:415-420. 19. Ozkutuk A, Kirdar S, Ozden S. Evaluation of Cobas Amplicor M. tuberculosis test to detect M. tuberculosis in pulmonary and extrapulmonqary specimens. New Microbiol 2006; 29:269-273. 20. Bonington A, Strang JG, Klapper P, et al. Use of Roche AMPLICOR Mycobacterium tuberculosis PCR in the early diagnosis of tuberculous meningitis. Journal of Clinical Microbiology 1998; 36:1251-1254. 21. Lemaitre N, Armand S, et al. Comparison of the Real-Time PCR Method and the GenProbe Ampli ed Mycobacterium tuberculosis 32

Direct Test for Detection of Mycobacterium tuberculosis in Pulmonary and Nonpulmonary Specimens. Journal of Clinical Microbiology 2004;42(9):4307-4309. 22. Hasaneen N, Zaki M, et al. Polymerase chain reaction of pleural biopsy is a rapid and sensitive method for the diagnosis of tuberculous pleural effusion. CHEST 2003;124:2105-2111. 23. Michos A, Daikos G, et al. Detection of M. tuberculosis DNA in respiratory and nonrespiratory specimens by Amplicor MTB PCR. Diagn Microbiol and Infect Dis 2006; 54:121-126. 24. Levidiotou S, Vrioni G, et al. Four-Year Experience of Use of the Cobas Amplicor System for Rapid Detection of M. tuberculosis complex in respiratory and nonrespiratory specimens in Greece. Eur J Clin Microbiol Infect Dis 2003; 22:349-356. 25. Montoya J, Berba R, Ang C, et al. Evaluation of the usefulness of AMPLICOR PCR and PCR with nested ampli cation in the diagnosis of tuberculous meningitis. Phil J Microbiol Infec Dis 2001; 30:44-50. 26. Lin JJ, Harn HJ, Hsu YD, et al. Rapid diagnosis of tuberculous meningitis by polymerase chain reaction assay of cerebrospinal uid. J Neurol 1995; 242:147-152. 27. Donald PR, Victor TC, Jordan AM, et al. Polymerase chain reaction in the diagnosis of tuberculous meningitis. Scand J Infect Dis 1993; 25(5):613-617. 28. Rosenstraus M, Zhuang W, et al. An internal control for routine diagnostic PCR : design, properties, and effect on clinical performance. J Clin Microbiol 1998; 36:191-197. 29. Honore-Bouakline S, Vincensini J, et al. Rapid diagnosis of extrapulmonary tuberculosis by PCR : Impact of sample preparation and DNA extraction. J Clin Microbiol 2003; 41:2323-2329. 30. American Thoracic Society Workshop. Rapid diagnostic tests for tuberculosis. What is the appropriate use? Am J Respir Crit Care Med 1997; 155:1804-1814. 31. Honore-Bouakline S, Vincensini J, et al. Diagnostic value of a nested PCR assay on peripheral blood mononuclear cells from patients with pulmonary and extrapulmonary tuberculosis. Int J Tuberc Lung Dis 2001; 5:754-762.

Philippine Journal of Microbiology and Infectious Diseases Vol. 37 No. 1, January-June 2008