INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Methods for the diagnosis of tuberculosis have improved in recent years, and several molecular techniques for its diagnosis have been introduced for clinical use. However, the stained smear of sputum, sputum culture, and chest radiography remain the principal tools used by most clinicians for the diagnosis of pulmonary tuberculosis (1). Some of these traditional methods require weeks before results are produced, and sensitivity is low when sputum samples contain only a small number of organisms.

Several methods for DNA and RNA amplification have been shown to be sensitive and specific for rapid detection of Mycobacterium tuberculosis in sputum and other body fluid samples (2). These methods provide several advantages over culture, including confirmation of the presence of M. tuberculosis within 1 to 3 d as compared with 2 to 6 wk with culture techniques. The use of DNA amplification for detection of M. tuberculosis in formalin-fixed, paraffin-embedded tissue samples would be useful for patients in whom diagnosis depends on tissue examination rather than detection of M. tuberculosis in body secretions. Unfortunately, there are frequent occasions when tissue obtained by biopsy is not sent for culture because the diagnosis of tuberculosis was not a clinical consideration before the report of findings on microscopic examination of the tissue. Polymerase chain reaction (PCR) amplification of M. tuberculosis DNA can provide much needed help in these circumstances because this methodology can detect M. tuberculosis in tissue samples even though the tissues have been preserved in formalin or other substances that preclude the possibility of culture.

Higuchi and colleagues achieved the first retrieval of phylogenetically informative DNA sequences from tissue obtained from a museum specimen of an extinct quagga, a member of the genus Equus (horse) (14). Archaeologic remains generally do not yield amplification products above 150 bp in size, whereas better preserved specimens may allow the amplification of sequences up to 500 bp long (15). PCR amplification has been used to establish the presence of organisms of the M. tuberculosis complex in a naturally mummified woman who died approximately 900 yr ago (16). In situ hybridization after amplification by PCR of formalin-fixed tissue has been shown to be sensitive for the detection of human papillomavirus DNA (17) and Mycobacterium leprae DNA (18). The type of fixative, duration of fixation, and molecular weight of the expected PCR product have important effects on PCR amplification (19, 20). Unbuffered and neutral-buffered formalin preparations may inhibit the efficiency of PCR amplification and diminish the sensitivity of the assay (19).

A PCR assay done on paraffin-embedded lung tissue of mice experimentally infected with the H37Rv strain of M. tuberculosis detects as few as nine organisms in a 5-µm section of tissue (23). Fixation in 10% neutral-buffered formalin for up to 7 d has negligible effect on the sensitivity of the assay. Previous studies have successfully used PCR to detect M. tuberculosis DNA in formalin-fixed tissues, but only a single study reported the sensitivity and specificity of the technique (24). To expand the reported experience with the detection of M. tuberculosis in formalin-fixed tissue samples, we report our experience in the diagnosis of tuberculosis by PCR amplification of DNA in a variety of fixed tissue specimens. In addition, we compared our PCR results with the clinical and microbiologic findings for whether or not the patient on whose specimen PCR was done was diagnosed as having tuberculosis.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients and Tissue Specimens

Sixty formalin-fixed, paraffin-embedded tissues submitted from various clinical laboratories in the United States, Canada, Europe, and Australia, which had been reviewed microscopically and read as possibly reflecting tuberculosis, were sent to our laboratory to be tested for the presence of M. tuberculosis DNA, using PCR for amplification. For all of the 60 subjects from whom the tissues came, a clinical diagnosis for the presence or absence of tuberculosis was made by the clinician in charge without the use of PCR results, using clinical, laboratory and epidemiologic data together with the clinical response observed after antituberculosis or other therapy was given.

Culture for mycobacteria and histologic studies were performed by the laboratories submitting the tissue specimens, using standard laboratory procedures. The tissues studied included lung (n = 18); brain and spinal cord (n = 7); skin and subcutaneous tissue (n = 8); lymph node (n = 9); pleura, liver, and gastrointestinal tissues (n = 4 each); spleen, bone, and bone marrow (n = 2 each); and synovium and aorta (n = 1 each).

Tissue Processing for PCR

From each paraffin-embedded tissue block, 25-µm-thick sections were cut with a microtome blade. To prevent carryover tissue contamination of the samples, the microtome blade was cleaned with octane and 100% ethanol after sectioning each sample. The paraffin from each tissue section was extracted using octane and 100% ethanol (23, 35). The tissue was digested at 55° C for 12 to 14 h in a solution (500 µl) containing proteinase K, Tris-HCl, ethylenediamine tetraacetic acid (EDTA), and Tween 20. The proteinase K was subsequently inactivated by heating the reaction mixture to 95° C for 10 min. Mycobacterial cell lysis was achieved by the freeze-thaw method, and the liberated DNA was extracted with a Gene Clean kit (BIO 101, La Jolla, CA) according to the manufacturer's instructions (23). The purified DNA was suspended in 30 µl of sterile deionized water.

DNA Amplification

Using disposable, positive-displacement pipettes, 5 µl of the DNA extract from each sample was added to 50 µl of a PCR reaction mixture containing 10 mM Tris-HCl; 50 mM KCl; 3.75 mM MgCl₂; 0.18 mM each of deoxyadenosine triphosphate (dATP), deoxycytosine triphosphate (dCTP), deoxyguanosine triphosphate (dGTP), and deoxyuridine triphosphate (dUTP); 0.001% gelatin, 0.68 mg/ml acetylated bovine serum albumin (BSA); 1,000 copies of a previously described internal control DNA sequence (2); 1 U of Taq DNA polymerase; and 0.18 µm of primers T4 and T5. The sequence of each oligonucleotide has been described (36). To reduce the risk of contamination from previous PCR products uracil-N-glycosylase (UNG) was added to the complete mixtures containing target DNA, and the mixtures were heated for 10 min at 50° C. This step would inactivate any uracil-containing DNA that might have been present as a result of contamination of the test samples from previous amplification reactions (37). To reduce the effects of low-stringency primer extension, a "hot-start" was used, at 94° C for 10 min, and the reaction was then amplified for 35 cycles using a thermal cycler (Gene Amp PCR system 9600; Perkin-Elmer Cetus, Norwalk, CT). Each cycle consisted of a denaturation step at 94° C for 45 s, annealing at 68° C for 45 s, and extension at 72° C for 2 min.

DNA Detection

M. tuberculosis-specific PCR products were detected in a liquid hybridization assay, using a ³²P-labeled LK229 probe (2, 23, 36). This probe hybridizes with sequences internal to the primers for the 123-bp M. tuberculosis amplimer to generate a hybrid molecule that is identified autoradiographically following electrophoresis through 12% nondenaturating polyacrylamide. The 600-bp PCR product generated by amplification of the internal control sequence has no homology to LK229, and is detected by ethidium bromide staining of gels, with ultraviolet transillumination at 302 nm. The LK229 probe sequence has been previously described (23).

Control Procedures

All testing was done as recommended in the guidelines for molecular diagnostic methods, with unidirectional workflow and physical separation of reagent preparation, amplification, and product detection procedures (38). Controls included positive and negative formalin-fixed, paraffin-embedded tissues processed in the same manner as the test samples. Positive tissue was obtained from mice experimentally infected with M. tuberculosis. As a control for cell lysis, a broth culture containing 1,000 colony forming units (cfu) of the H37Rv strain of M. tuberculosis was processed in a manner identical to the test samples. To monitor for inhibition of amplification, internal control DNA was added to each PCR reaction mixture. The internal control DNA was amplified with use of the same set of IS6110 primers as required for the test specimens, but was easily separated from the M. tuberculosis DNA amplified products because of its larger size (600 bp) (2). Additional negative controls included water and digestion buffer, as well as reagent mixtures containing all of the test components with and without UNG.

Tissues were determined to be PCR-positive when the 123-bp M. tuberculosis DNA fragment was present on the gel, and were declared PCR-negative when this fragment was absent but when the 600-bp fragment of the internal control DNA was present. If the internal control fragment and the 123-bp M. tuberculosis DNA fragment were both absent, the reaction was reported as indeterminate and the test was repeated.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Results obtained from all the tissues studied are shown in Table 1. Of the 60 specimens studied, 57 showed granulomatous inflammation. Fifteen tissue specimens showing granulomatous changes also showed acid-fast bacilli (AFB) on microscopy; nine of these 15 specimens were culture-positive for mycobacteria (seven grew M. tuberculosis, one grew Mycobacterium avium, and one grew Mycobacterium celatum). PCR for M. tuberculosis was positive for all seven tissue specimens that grew M. tuberculosis and was negative for the two specimens that grew nontuberculous mycobacteria. Three tissue specimens that were culture-negative were also PCR-negative for M. tuberculosis, and three specimens were not cultured, of which two were PCR-positive and one was PCR-negative for M. tuberculosis. Forty-two tissue specimens showing granulomatous inflammation were negative for AFB with staining; of these 42 specimens, three were both culture-positive and PCR-positive for M. tuberculosis; one was culture-positive for M. avium and PCR-negative for M. tuberculosis; 34 were culture-negative for mycobacteria, of which 31 were PCR-negative and three were PCR-positive for M. tuberculosis; and four were not cultured, all of which were also PCR-negative for M. tuberculosis. Only three tissue specimens did not show granulomatous inflammation, all of which were culture-negative for mycobacteria and PCR negative for M. tuberculosis.

Correlation of PCR Results with Culture for M. tuberculosis

PCR results for M. tuberculosis DNA obtained on all 53 tissue samples cultured for M. tuberculosis are shown in Table 2. M. tuberculosis was isolated from 10 specimens, and all 10 of these were also PCR-positive for M. tuberculosis DNA. Forty-three specimens were negative for M. tuberculosis by culture, and three of these were positive for M. tuberculosis DNA by PCR. Two of the latter three specimens were obtained from lung tissue of patients who had been sputum culture-positive for M. tuberculosis 2 yr earlier; both patients were nonadherent with their chemotherapy regimens because of side effects. The third patient had chronic monoarticular arthritis, and a synovial biopsy showed noncaseating granulomatous inflammation; no AFB were seen on microscopy, and culture of the patient's synovial tissue was negative. With antituberculosis chemotherapy this third patient showed a good response. With a positive culture for M. tuberculosis used as a "gold standard," PCR for M. tuberculosis DNA in the 53 tissue specimens described here was 100% sensitive and 93% specific, and had a positive predictive value of 76.9% and a negative predictive value of 100%.

Correlation of PCR Results with Clinical Diagnosis

For 58 patients, it was possible to make a clinical diagnosis regarding the presence or absence of tuberculosis from clinical, laboratory, and epidemiologic data without consideration of PCR results. The results of PCR amplification for M. tuberculosis DNA in tissue specimens from these 58 patients is compared with the final clinical diagnosis in Table 3. Nineteen patients had a clinical diagnosis of tuberculosis; the AFB stain of the tissues was positive in eight of these 19 cases and culture for M. tuberculosis was positive in 10. Culture was not done in one patient whose pleural fluid was positive for AFB on staining; this patient showed a good response to antituberculosis therapy. Two patients had been sputum culture-positive for M. tuberculosis 2 yr earlier. Both of these patients were nonadherent with antituberculosis therapy because of side effects; their lung tissues were AFB-negative on smear and culture. One patient had chronic monoarticular arthritis; in this case synovial-tissue biopsy showed noncaseating granuloma, an AFB stain and culture were negative, and there was a good response to antituberculosis therapy.

The tissue specimens from five patients were negative for AFB on staining, culture, and PCR; one of these five patients had a paraesophageal lymph node that showed caseating granulomas; a second had a duodenal mass with caseating granulomas; a third had tissue from a liver biopsy that showed granulomatous inflammation; a fourth had paratracheal lymphadenopathy with granulomas and caseous necrosis; and the fifth had weight loss and small-bowel obstruction. In this last case, biopsy of tissue at the obstruction site showed granulomatous inflammation. All five of these patients improved with antituberculous chemotherapy.

Thus, PCR of tissue for M. tuberculosis DNA was positive in 14 of the 19 patients who had a clinical diagnosis of tuberculosis.

PCR for M. tuberculosis was negative for all 6 patients with nontuberculous mycobacterial infection. The AFB stain of tissue specimens was positive in five of these patients, of whom three were culture-positive for mycobacteria other than M. tuberculosis (two for M. avium and one M. celatum). Three of the six patients with nontuberculous mycobacterial infections were culture-negative for mycobacteria (two of these had M. leprae infection documented by a positive PCR, and one had a skin ulcer that healed after treatment with erythromycin).

The PCR test of M. tuberculosis was negative with the tissue specimens obtained from all 33 patients who had a diagnosis of a disease other than a mycobacterial infection. All of these specimens were negative for AFB on staining, and 29 were culture-negative for mycobacteria. Four of the specimens were not cultured.

In the cases of two patients the final clinical diagnosis remained uncertain because there are insufficient data for ascertaining the presence or absence of tuberculosis. One was an elderly diabetic patient who died after a leg amputation. Lung tissue obtained at autopsy of this patient showed numerous granulomas, with AFB present on stained tissue; culture was not done, but PCR gave a positive result for M. tuberculosis. The second patient had a gastric resection for carcinoma of the stomach, and liver tissue obtained at surgery showed granuloma with AFB present; the PCR for M. tuberculosis was negative, and culture was not done. This patient was given antituberculous chemotherapy, but no follow-up information has been available.

The results of PCR for M. tuberculosis DNA are compared with the final clinical diagnosis in Table 4. In this set of 58 patients, five were AFB-stain- and culture-negative. All five of these patients responded favorably to antituberculosis therapy and were given a clinical diagnosis of tuberculosis. Using the clinical diagnosis as a "gold standard," PCR for M. tuberculosis DNA in these 58 patients' tissues was 73.6% sensitive and 100% specific, having a positive predictive value of 100% and a negative predictive value of 88.6%.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We and others have shown PCR amplification to be sensitive and specific for the rapid detection of M. tuberculosis DNA in sputum samples (2). The appropriate use of these new tests, and guidelines to assist clinicians in using the tests, have been set out in previous publications (39, 40). We have done this assay on formalin-fixed, paraffin-embedded tissue from mice experimentally infected with the H37Rv strain of M. tuberculosis and demonstrated that we can detect as few as nine organisms in a 5-µm section of tissue (23). Others have reported experience with PCR amplification with human tissue. Diaz and coworkers tested 64 tissue samples and found that PCR amplification was 100% sensitive and 100% specific in culture-positive specimens, with 11 of their tissues being culture-positive (24). Popper and colleagues used a method that amplified M. tuberculosis DNA within th 65-kD antigen, and tested seven tissues from patients with culture-proven tuberculosis; all seven were PCR-positive (27). Ghossein and coworkers, amplifying the same target, obtained similar results for 12 tissues (28). There are additional reports of small numbers of patients supporting the usefulness of PCR amplification in cutaneous tuberculosis (30, 34). In our study of 60 tissue specimens, PCR amplification revealed M. tuberculosis DNA in 14 of 19 patients who had a final clinical diagnosis of tuberculosis. Using the clinical diagnosis as the "gold standard," we found no false-positive PCR results for M. tuberculosis DNA among the patients diagnosed as having tuberculosis.

The PCR assay for M. tuberculosis that we have described is very specific, and has been shown to have 100% positive predictive value. The sensitivity and specificity of the assay are 100% when both AFB stain and culture are positive for M. tuberculosis.

Thus, there are a number of previously published reports of studies with humans, our present data, and one report in mice that indicate that PCR amplification in formalin-fixed tissue can detect M. tuberculosis DNA when only a few genomes are present. The need for rapid and sensitive detection of M. tuberculosis DNA in tissue specimens, whether fixed in formalin or fresh, exists in all major hospitals and clinics that evaluate and treat patients with mycobacterial infections. The PCR test for M. tuberculosis DNA with the repeating IS6110 sequence as a target for amplification can provide valuable information for the treating physician. Special attention must be paid to the capacity of M. tuberculosis DNA to remain in tissue specimens for many years even though no viable organisms are present. This point has been confirmed in the study of sputum from patients who have fully recovered after chemotherapy for pulmonary tuberculosis, and such persistence of M. tuberculosis DNA has been observed in tissue from museum specimens (2, 14, 16). It is highly probable, even though the question has not been studied, that healed tuberculous granulomas that are culture- and AFB stain-negative for M. tuberculosis will sometimes be positive for its DNA. Thus, when M. tuberculosis DNA is found in tissue specimens, this information must be taken together with a complete analysis of all other laboratory and clinical data before a final diagnosis of tuberculosis is made. When properly used, the PCR test for M. tuberculosis DNA will increase diagnostic accuracy in cases of perplexing infections that might be tuberculosis, and this should result in improved and timely care for patients.