INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary tuberculosis (PTB) remains one of the most important health problems in the world, with an estimated 8 million new cases of tuberculosis (TB) and 1.9 million deaths having occurred in 1997 (1). The World Health Organization (WHO) recommends the detection of acid-fast bacilli (AFB) in respiratory specimens as the initial approach to the diagnosis of PTB (2). However, this method has a low sensitivity and has little value in patients who cannot produce sputum spontaneously (3, 4).

In Brazil, with an estimated annual prevalence of 129,000 cases of TB, approximately 22% of adult human immunodeficiency virus (HIV)-seronegative patients with suspected PTB do not produce sputum spontaneously or have negative AFB smears (5). Thus, the diagnosis of TB in these patients is difficult, and in most cases they are treated empirically on the basis of clinical and chest radiographic findings. However, empiric therapy may result in unnecessary cost and toxicity. Moreover, HIV-seropositive patients who do not produce sputum often undergo expensive and more invasive procedures like fiberoptic bronchoscopy with bronchoalveolar lavage (BAL).

A single report describes the yield of sputum induction (SI) in comparison with BAL for the diagnosis of PTB in HIV-seronegative patients (6). To our knowledge no study has prospectively compared SI with BAL in HIV-seropositive and -seronegative patients who reside in a region with a high prevalence of TB and HIV infection. We therefore conducted a study of this in such a region.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection

The study prospectively evaluated patients 18 yr of age and older who had respiratory symptoms and radiographic findings consistent with PTB, but who were unable to produce sputum spontaneously or had one sample of spontaneously produced sputum that was smear-negative for AFB. All patients who attended and were referred from Rio de Janeiro outpatient clinics in the 3-1 Planning Area (primary health units) or from the Ambulatory Service of the Hospital Universitario Clementino Fraga Filho (HUCFF) (a tertiary health care unit) between April 1, 1996 and March 31, 1998 were screened for the study. Patients were excluded if they had asthma, a pleural effusion, a hematocrit of less than 30%, a Karnofsky score of less than 50, a suspected or known pregnancy, or signs of respiratory or cardiac failure. In addition, patients were excluded if, during the 3 mo prior to enrollment, they had had hemoptysis or had received antituberculosis therapy for more than 1 wk. Informed written consent was obtained from all participants, and the study was approved by the Ethics Committees of the HUCFF and the University of California, San Francisco.

Diagnostic Procedures

All patients included in the study underwent a standardized interview, physical examination, chest radiography, serology for HIV infection by an enzyme-linked immunosorbent assay (ELISA) (with Western blot confirmation), and SI with a 3% hypertonic saline solution generated by a DeVilbiss ultrasonic nebulizer (Ultra Neb 99; Home and Aid Healthcare, Somerset, PA) followed on the same day by fiberoptic bronchoscopy with BAL. SI and fiberoptic bronchoscopy were performed as previously described (7, 8).

All specimens were stained with hematoxylin and eosin, Papanicolaou stain, Ziehl-Neelsen stain, and Grocot's methenamine silver stain, and were cultured in Löwenstein-Jensen and Sabouraud's medium according to standard protocol for the detection of mycobacteria (9). All specimens that were culture positive for mycobacteria were tested with biochemical methods to distinguish Mycobacterium tuberculosis from other, nontuberculous mycobacteria (NTM). All chest radiographs were evaluated by the same pulmonologist (S.L.M.S.). Confirmed PTB was defined as a positive culture for M. tuberculosis in a SI specimen and/or in bronchoalveolar lavage fluid (BALF). Presumptive PTB was defined as unequivocal improvement after 3 mo of antituberculosis treatment as judged by an independent pulmonologist in a blinded review. Inactive PTB was considered to be present if AFB smears and mycobacterial cultures were negative and chest radiographic findings were consistent with healed tuberculosis, and did not change after 6 mo of follow-up. A diagnosis of Pneumocystis carinii (PC) infection was made on the basis of a positive methenamine silver stain, and a diagnosis of unspecified bacterial and/or viral infection was made on the basis of clinical, radiographic, and laboratory findings associated with a favorable outcome after treatment with specific antimicrobial therapy and/or use of supportive measures. Lung cancer was diagnosed by anatomopathologic or cytopathologic studies of material obtained by bronchoscopy. All patients were followed for at least 6 mo after inclusion in the study.

Statistical Analysis

Results were analyzed with the Epi Info 6.0 statistical software package. Sensitivity, specificity, and predictive values of the SI and BAL procedures were calculated. The kappa test was used to evaluate the agreement between SI and BAL.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Between April 1, 1996 and March 31, 1998, 296 patients were eligible for enrollment on the basis of clinical and radiographic findings yielding suspicion of PTB, but with inability to produce sputum spontaneously or ability to produce one sample of sputum that was smear-negative for AFB. Ten patients were excluded because they had asthma (two cases), a pleural effusion (two cases), or hemoptysis in the week previous to enrollment (one case). Five patients were excluded because they had received anti-TB treatment in the previous 3 mo. An additional 35 individuals were not included because we were unable to complete the clinical follow-up at 6 mo (15 cases), the patient's cultures were contaminated (six cases), respiratory specimens were missing (six cases), or the subject was unable to produce sputum despite sputum induction (eight cases). Thus, 251 patients were included for analysis. One hundred forty-three (57%) of the patients were diagnosed as having PTB; a further 36 (14%) had inactive TB, 45 (18%) had an unspecified pulmonary infection, 16 (6%) had lung cancer, five (2%) had silicosis, three (1%) had pulmonary mycosis, two (1%) had Kaposi's sarcoma and one (0.5%) with P. carinii infection. Among patients with PTB, 90 were male, and among individuals with other diseases, 82 were male (p = 0.03), with the mean ages of the two groups being 37 and 43 yr (p < 0.001), respectively. Eighty-nine (62%) of the cases of PTB were 40 yr old or younger (p < 0.005). One hundred (70%) patients with PTB and 52 (35%) patients with other diseases were referred by outpatient units (p < 0.005). In 118 (19 HIV-seropositive) patients, the diagnosis of PTB was based on a positive mycobacterial culture, and in 25 (six HIV-seropositive) patients the diagnosis was presumptive. Forty-four (17%) patients were HIV-seropositive, including 25 (17%) of the patients with PTB. The chest radiographic findings were suggestive of PTB in 101 (71%) patients with confirmed TB and 69 (64%) patients with other diseases (p = 0.21).

The results of AFB smears and mycobacterial cultures of specimens obtained by SI and BAL among HIV-seropositive and -seronegative patients are shown in Table 1. The AFB smear results for specimens obtained by SI and BAL were concordant in 98% (245 of 251) of cases. The M. tuberculosis culture results for SI and BAL specimens were concordant in 224 (89%) cases. The kappa test result for agreement between a diagnosis based on SI versus BAL specimens was 0.93 for AFB smear and 0.75 for culture. Among patients with PTB, SI led to a diagnosis in 94 (66%) and BAL led to a diagnosis in 103 (72%) (p > 0.05).

Among HIV-seronegative patients, the results of AFB smears of specimens obtained by SI and BAL were concordant in 202 (98%) cases (kappa test = 0.92), and among HIV-seropositive patients the results of the two procedures were concordant in 43 (98%) cases (kappa test = 0.92) (Table 1).

The results of mycobacterial cultures of specimens obtained by SI and BAL from HIV-seronegative patients were concordant in 186 (90%) cases (kappa test = 0.78) (Table 1). Agreement between cultures of specimens obtained by SI and BAL in HIV-seropositive patients occurred in 38 (86%) cases (kappa test = 0.69).

Sensitivities, specificities, and predictives values of AFB smear and culture for the diagnosis of PTB in specimens obtained by SI and BAL are shown in Table 2. There were no statistically significant differences between the sensitivities of AFB smears in HIV-seronegative and -seropositive patients whether the specimen was obtained by SI (p = 0.9) or by BAL (p = 0.9). Nor were there any significant differences between the results of culture for M. tuberculosis with SI (p = 0.6) or BAL (p = 0.2). No specimen with a positive AFB smear or positive mycobacterial culture result was identified as showing an NTM or as reflecting a non-TB diagnosis. Seventy-five (52%) patients with a final diagnosis of PTB had a negative AFB smear in one sample of spontaneous sputum before being referred to the pulmonology service of our hospital, and among these cases, 26 (35%) had a positive AFB smear with SI and 30 (40%) had a positive AFB smear with BAL (kappa test = 0.85). Of the 68 subjects with PTB who were unable to produce a sputum specimen spontaneously, 23 (34%) had a positive AFB smear with SI and 25 (37%) had a positive smear with BAL (kappa test = 0.92).

SI and BAL procedures were well tolerated by most patients. Adverse effects were reported by eight (3%) patients undergoing SI and by 10 (4%) patients undergoing bronchoscopy with BAL. These adverse effects were a persistent cough (three cases with SI and five cases with BAL), headache (two cases with SI and two cases with BAL), dizziness (three cases with SI), and dyspnea (three cases with BAL).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study demonstrated that SI is a safe and effective means of obtaining specimens for AFB smear and culture in persons suspected of having PTB but who cannot produce sputum spontaneously or who are smear-negative for AFB. These scenarios are common in the workup of suspected TB. Between 1995 and 1997 in the city of Rio de Janeiro, for example, 27% of patients with PTB were unable to produce sputum and 21% were smear-negative for AFB (10). In our study, nebulization with 3% hypertonic saline solution induced sputum production in 97% of such patients.

The sensitivities of the AFB smear and culture for M. tuberculosis were similar for patients who underwent both SI and BAL. These results are similar to those obtained by Anderson and colleagues in 101 HIV-seronegative patients who underwent both procedures (6). In our study, the agreement between the results of AFB smear and culture for M. tuberculosis with SI and BAL specimens was excellent in both HIV-seropositive and -seronegative patients. The differences between the sensitivities of AFB smear and culture for M. tuberculosis in HIV-seropositive patients and in HIV-seronegative patients were not statistically significant.

The positive predictive value (PPV) of an AFB smear for the diagnosis of PTB was 100% among both HIV-seronegative and -seropositive patients undergoing SI or BAL, in contrast to findings of other investigators, who reported a PPV of 60 to 70% for AFB smear in specimens obtained by SI and BAL (6, 11). These differences reflect the different prevalence of isolation of NTM in respiratory specimens. The lack of isolation of NTM from our patients is difficult to explain, since previous studies have shown that Mycobacterium avium-complex infection is common in AIDS patients in other parts of Brazil (12). It is possible that our laboratory had methodologic problems in culturing M. avium from sputum or BAL. If this is the case, the specificity of an AFB smear of specimens obtained by SI or BAL would be significantly reduced.

The sensitivities of AFB smears of specimens obtained by SI and BAL from HIV-seronegative patients were 34% and 38%, respectively. These values are greater than those reported by Anderson and colleagues (19% and 12%, respectively) (6). The frequencies of positive AFB smears and cultures for M. tuberculosis with specimens obtained by SI from patients who were HIV-seropositive were also higher in our series than in a study in Malawi (13). These differences may be related to differing patient populations, different procedural techniques, and different levels of laboratory specialization. For example, in our study, most of the patients were referred from outpatient clinics (primary health care units) whose laboratories are not specialized in the diagnosis of TB, and some of the results for these patients may therefore have been falsely negative. Supporting this supposition is that among 75 PTB patients who had a negative AFB smear from one spontaneously produced sputum specimen, 26 had a positive smear with SI and 30 had a positive smear with BAL. It is important to note that SI did not yield specimens for eight patients. Because these patients were excluded from the study, we are unable to evaluate the effect of their inclusion on our results. However, if these patients had undergone bronchoscopy with BAL, it is possible that the yield of SI would have decreased relative to that of BAL.

The yield of culture for M. tuberculosis among HIV-seronegative and -seropositive patients was almost twice as high as the yield for AFB smears, confirming the importance of culture for the diagnosis of PTB in our patients.

Only eight patients who underwent SI experienced side effects, a rate similar to the rate of side effects experienced with bronchoscopy with BAL. In a similar study, Anderson and coworkers reported no complications with SI and only two minor complications with bronchoscopy (6).

We conclude that SI is a safe procedure with a high diagnostic yield and a high agreement with results of fiberoptic bronchoscopy for the diagnosis of PTB in HIV-seronegative and -seropositive patients. In areas in which fiberoptic bronchoscopy is not readily available, and as part of the workup of suspected TB prior to bronchoscopy, SI offers an alternative or additional approach to the diagnosis of sputum-smear-negative TB, and would enhance sensitivity for the diagnosis of TB in resource-poor areas.