INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The polymerase chain reaction (PCR) assay is becoming widely used for the diagnosis of various bacterial infections (1) including pulmonary tuberculosis (TB) and nontuberculous mycobacteriosis (NTMosis) (4). This innovative development enables us to make an early diagnosis, and treatment can be started immediately after diagnosis (4). Sputum, body fluid, biopsy specimens, and serum have been used for PCR (5). However, it is not yet clear what kind of clinical specimen is the most suitable for detecting mycobacterial infection by PCR assay.

Recently, bronchoalveolar lavage (BAL) was recommended for the diagnosis of various lung diseases, including pulmonary infections (8). Mycobacterium tuberculosis has been detected in BAL fluid by PCR analysis, the sensitivity being over 80% (9). However, pathological examination shows few or no acid-fast bacteria in the inflammatory lesions of active mycobacteriosis, and it may be difficult effectively to wash out M. tuberculosis from the lesions by washing alone.

A few years ago, we developed a rapid and simple method for the identification of mycobacterium species (10). In this method, the mycobacterial genes encoding the 65 kD antigen (11) were amplified using the PCR. Analysis of the PCR products subsequently obtained using restriction fragment length polymorphism (RFLP) after digestion with HaeIII enabled us to identify M. tuberculosis and seven species of nontuberculous mycobacteria (NTM) on the basis of the different electrophoretic patterns.

Fiberoptic bronchoscopy is often performed when, although a lesion is radiologically suspected of mycobacterial infection, a sputum smear is negative (14, 15). In this study, we examined the efficacy of washing the bronchus connecting to the lesion with 20 ml of saline, then applying PCR-RFLP to the bronchial washings for the diagnosis of mycobacterial infections in smear-negative outpatients without human immunodeficiency virus (HIV) infection.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ninety-eight patients were examined in this study. They included 24 patients with active TB lesions (active group), 28 with cured TB lesions (old group), and 46 with nontuberculous opacities (control group) in the chest computed tomography (CT) scan taken before examination by bronchoscopy. All were outpatients who visited Shinshu University Hospital between June 1995 and December 1996. None had an HIV infection. All patients were smear-negative for acid-fast mycobacteria. Clinical data for all 98 patients are summarized in Table 1. All patients were followed for at least 2 yr after the bronchoscope examination.

The patients in the active group visited our hospital for further examination of radiological abnormalities. The chest CT showed nodular consolidation with an irregular border measuring 1 to 3 cm in diameter or segmental consolidations, some of which were accompanied by small satellite nodules. Most of the patients had been examined annually by chest radiograph for health screening, and these nodular consolidations had appeared or become enlarged in chest radiographs or chest CT scans within the year preceding the study. These radiological findings suggested active TB. Laboratory examinations showed no abnormality except a slight elevation of C-reactive protein.

The patients of the old group also presented to our hospital because of chest radiographic abnormalities. Fibrous and nodular opacities were predominant in the upper lobes, but they had not altered in shape or size in chest radiographs for several years. These patients had a past history of TB and their condition was clinically diagnosed as cured pulmonary TB. Most of them had previously been treated with antituberculous therapy.

The control group included 18 patients with lung cancer, nine with sarcoidosis, and 19 with interstitial pneumonia. None of the patients in this group had a history of TB. In all patients with lung cancer diagnosis was made histologically, cytologically, or both, after bronchoscopy. Transbronchial lung biopsy specimens, on the other hand, showed noncaseous granulomatous lesions in five patients with sarcoidosis, a finding compatible with sarcoidosis. In the other four patients a diagnosis was made by a combination of the following clinical findings: (1) bilateral hilar lymphadenopathy, elevation of angiotensin-converting enzyme in serum, and leukocyte analysis of BAL fluid, and (2) uveitis compatible with sarcoidosis. The patients with interstitial pneumonia showed interstitial opacities in the chest CT. They included five with idiopathic pulmonary fibrosis, five with collagen vascular disease (two rheumatoid arthritis, two dermatomyositis, and one systemic lupus erythematosis), four with bronchiolitis obliterans organizing pneumonia, three with hypersensitive pneumonia, one with radiation pneumonia, and one with graft versus host disease.

Two samples obtained by bronchoscopy from each patient were examined by culture and PCR-RFLP. The brushes used to brush the lesions were washed in 5 ml of sterile saline (brushing sample). Aliquots of 3 to 10 ml of saline were aspirated through the forceps channel after 20 ml sterile saline had been injected into the bronchus by means of which the lesions were biopsied or brushed (washing sample).

DNA from each sample was prepared as described elsewhere (10). Briefly, 1 N sodium hydroxide was added to precipitate the samples. After centrifugation, the pellet was resuspended in 0.5 ml of lysis buffer, and ribonuclease (RNase) was added to give a final concentration of 0.1 mg/ml. Incubation then proceeded for 30 min at 37° C. Proteinase K was added to this resuspension to give a final concentration of 0.2 mg/ ml, and incubation then proceeded overnight at 37° C. The DNA was extracted using a phenol/chloroform extraction method, followed by precipitation in ethanol and resuspension in sterilized distilled water.

The DNA amplification and analysis methods have also been described previously (10). DNA amplifications were performed using primers of TB1 and TB2 to recognize sequences in the 65 kD antigen gene, which is common in mycobacterial species (11, 13). Another primer was also used as an anti-sense primer for the second round PCR (Table 2). A reaction mixture of 25 µl volume containing Low/ Med buffer (50 mM Tris, pH 8.3, 1.5 mM MgCl₂, and 0.25 mg/ml bovine serum albumin; Idaho Technology Inc., Idaho Falls, ID), 0.2 mM of each deoxynucleotide triphosphate, 0.6 U AmpliTaq DNA polymerase (Perkin-Elmer, Norwalk, CT), and 25 pmol of each primer. One microliter of the first-round PCR products was added to the second reaction as the template DNA. The first- and second-round PCR were carried out under the following conditions for 40 cycles and 20 cycles of amplification, respectively: denaturation at 94° C for 15 s, annealing at 60° C for 15 s, and primer extension at 72° C for 30 s. The cycles were preceded by a 1-min preheating at 94° C, and were followed by a 2-min extension at 72° C. In both the first- and second-round reactions, negative and positive controls were included in every experiment. For the negative and positive reactions, we added sterilized distilled water and genomic DNA of M. tuberculosis instead of sample DNA, respectively. As an internal control, an 81-bp fragment of the Ki-ras gene was amplified by PCR using primers KR-1 and KR-17 (Table 2). Preparations of PCR mixture without template DNA were mixed on a desktop-type clean-bench to avoid false positives caused by contamination.

The PCR products were then analyzed by electrophoresis on a 3% agarose gel stained with ethidium bromide, and examined for the presence of a 368-bp fragment under ultraviolet irradiation. Then, positive products were digested with restriction enzyme HaeIII in order to identify M. tuberculosis and NTM on the basis of the RFLP patterns (Figure 1). The M. tuberculosis complex included M. tuberculosis, M. bovis, M. africanum, and M. microti.

View larger version (42K): [in this window] [in a new window]   Figure 1.   HaeIII digestion patterns for the second-round PCR products (368 bp) 1: M. tuberculosis, 2: M. marinum, 3: M. scrofulaceum, 4: M. avium, 5: M. intracellulare, 6: M. fortuitum, 7: M. chelonei, 8: M. gastri, M: size marker.

A mixture of the brushing and washing samples from each patient was also examined by culture for 3 mo using 1% Ogawa's medium.

Twelve flexible fiberoptic bronchoscopes (Olympus 1T200, P200, P240 etc.; Olympus, Tokyo, Japan) were available in our hospital, and a given bronchoscope was used for one patient only on a given day. The bronchoscopes were cleaned and disinfected immediately after each use. First, the outside of the bronchoscope and the suction channel were cleaned with tap water. Then, the bronchoscopes were disinfected in an automated washing machine (EW20 or OER; Olympus) as follows. They were initially cleaned with neutral detergent or by sonication for 15 min, disinfected by immersion in a 3.09% glutaraldehyde solution (Steriscope; Maruishi Chem. Co., Osaka, Japan) for 45 min, rinsed in tap water for 15 min, and finally flushed with 70% alcohol. They were then disinfected in ethylene oxide gas at 37° C, 600 millibar for 5.5 h, and aerated for 12 h. The tap water and the sterile water flushed through the bronchoscopes after disinfection and before use were also examined by culture and PCR-RFLP for mycobacteria in several cases. All such materials were confirmed negative for all mycobacterial species examined.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We diagnosed TB or NTMosis by positive culture or radiological improvement after antituberculous therapy, and TB-suspicious or NTMosis-suspicious cases by negative culture, positive PCR-RFLP, and absence of radiological alteration after the therapy.

The PCR and culture findings from the patients in the active group are summarized in Table 3. Among the 24 patients, M. tuberculosis was cultured from the mixture of brushing and washing samples in 11 patients, and NTM were cultured in five. No patient showed two different species by culture. The nodular consolidation or segmental consolidation was reduced in size after antituberculous therapy in six of the eight culture-negative patients; however, the consolidation did not alter in the remaining two patients. We diagnosed TB in 17 patients, NTMosis in five, and characterized two patients as TB-suspicious. PCR-RFLP of the brushing sample was positive in 14 of the 19 TB or TB-suspicious patients and in four of the five with NTMosis. On the other hand, PCR-RFLP of the washing sample was positive in 12 of the 17 TB or TB-suspicious patients and in all five with NTMosis. PCR-RFLP of the combined brushing and washing samples was positive in all TB or TB-suspicious patients except one, in whom only two colonies of M. tuberculosis were cultured. No patient proved positive for two different species of mycobacterium by PCR-RFLP. The species of mycobacterium identified by PCR-RFLP completely matched the species determined by culture. The accuracy of the PCR-RFLP in diagnosing mycobacterial infection was 88% in the active group when the two TB-suspicious cases were not considered as showing true infection.

The PCR and culture findings in the patients in the old group are summarized in Tables 4 and 5. M. tuberculosis was not cultured at all, but M. avium complex was cultured in one patient. Abnormal opacities did not alter in any patients with or without antituberculous therapy. Consequently, we diagnosed NTMosis in one patient of the old group. PCR-RFLP of the combined brushing and washing samples was positive for M. tuberculosis complex in four patients and for five different species of NTM (M. marium, M. gastri, M. fortuitum, M. intracellulare, and one mycobacterium not identified in this study) in five patients (one species per patient). We classified these subgroups of four and five patients as TB-suspicious and NTMosis-suspicious, respectively.

The PCR and culture findings in the patients in the control group are also summarized in Tables 4 and 5. No M. tuberculosis or NTM was cultured in this group. PCR-RFLP was positive for M. tuberculosis complex in two of the 18 patients with lung cancer and in one of the nine with sarcoidosis. PCR-RFLP was positive for M. gastri in one of the 29 patients with interstitial pneumonia. Three patients and one patient were classified as TB-suspicious and NTMosis-suspicious, respectively. Lobectomy was performed in the two PCR-positive patients with lung cancer, but a lesion suggesting tuberculosis was not observed in the histological preparations. One patient with sarcoidosis and one with collagen vascular disease were followed without antituberculous therapy, and interstitial opacities did not progress.

Among the 98 patients, 17 TB, 6 NTMosis, 9 TB-suspicious, and 6 NTMosis-suspicious cases were evaluated by a combination of PCR-RFLP, culture, treatment, and alteration of radiographic findings. When the TB-suspicious and NTMosis-suspicious patients are not included as showing true infection, the values obtained for the sensitivity of the PCR-RFLP with respect to the brushing samples, the washing samples, and the combined samples were 70, 76, and 91%, respectively, and the values for the specificity of the PCR-RFLP were 85, 89, and 80%, respectively (for the brushing samples, the washing samples, and the combined samples).

An 81-bp fragment of the Ki-ras gene was uniformly amplified from all of the samples. M. tuberculosis and NTM were not detected by either culture or PCR-RFLP in the tap water or the sterile water flushed through the bronchoscopes after disinfection and before use.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have demonstrated that washing the bronchus connecting to the lesions with a small amount of saline (20 ml) after biopsy and brushing with a fiberoptic bronchoscope provides material suitable for the diagnosis of mycobacterial infection. A mixture of the brushing and washing samples provides good material for PCR-RFLP and culture, enabling detection of several species of mycobacteria, including M. tuberculosis.

The sensitivity of the PCR-RFLP with respect to the combined brushing and washing samples was 91% (when the TB-suspicious and NTMosis-suspicious cases were not considered as showing true infection). The sensitivity of PCR for M. tuberculosis has been reported to be 80.2% when BAL is used (9), and 73% when sputum is used (5). In the present study, those patients who were PCR-positive, culture-negative, and showed no response to the therapy were classified as TB-suspicious or NTMosis-suspicious. It was impossible to determine whether the pulmonary lesions were active or cured in these groups because PCR cannot distinguish living mycobacteria from dead ones. Indeed, PCR is often positive when culture is negative (16).

Scratching the lesions by brushing or biopsy may aid the washing out of mycobacteria. This is useful because few mycobacteria can often be detected by pathological examination in granulomatous tissues. In fact, the organisms are usually located within the central, most necrotic zone of the granuloma (17), and identification of the organisms sometimes requires a long and careful search, because few or no mycobacteria are present in tissue specimens (18).

The accuracy of the PCR-RFLP used in this study was 88% in the active group. It would be advisable to perform bronchoscopy and examine the brushing and washing samples by PCR-RFLP when a patient is smear-negative, but radiologically suspected as having active tuberculosis. M. tuberculosis complex, on the other hand, was positive by PCR-RFLP in four patients among the 28 in the old group, but not cultured. We need to follow PCR-RFLP-positive and culture-negative patients carefully to determine whether M. tuberculosis persists in the fibrous lesions.

It is difficult to explain the dissociation between negative PCR and positive culture findings. Only two colonies of M. tuberculosis were cultured from combined brushing and washing samples obtained from one patient with TB in the active group. The same phenomenon occurred in a patient with M. avium complex in the old group. It may be that the presence of only a few mycobacteria is beneath the level of detection when PCR-RFLP is used.

PCR-RFLP was positive for M. tuberculosis complex in two patients with lung cancer and in one with sarcoidosis. The DNA of M. tuberculosis is sometimes observed in sarcoid lesions (19). The two patients with lung cancer had no history of TB and no radiographic findings suggesting TB. Contamination is considered a possible explanation, because tuberculous lesions were not observed in the resected lobes during the therapy for lung cancer.

NTM from the environment could conceivably contaminate the brushing and washing samples obtained from the bronchoscope. In this study, M. gastri was detected using PCR-RFLP in two patients, but was not cultured. This species is sometimes isolated from sputum and is not associated with disease in humans (20). There are various NTM in the environment, including city water (21), and they could contaminate samples obtained after bronchoscopy. However, the DNA of NTM was not found by PCR-RFLP, and NTM were not cultured from the tap water used in our hospital.

Our PCR-RFLP method is a very useful way of distinguishing M. tuberculosis from the other mycobacterial species. In fact, eight kinds of mycobacterial species could be clearly identified from one PCR-RFLP sequence. M. avium or M. intracellulare were detected in five of the 24 patients in the active group (who were clinically suspected of having TB).

In conclusion, a sequential combination of brushing and washing, collecting rinsing samples, and finally analyzing by the PCR-RFLP method is useful for the rapid diagnosis of mycobacterial infection.