Incremental yield of bronchial washing for diagnosing smear-negative pulmonary tuberculosis

OBJECTIVE: To assess the increased diagnostic yield for pulmonary tuberculosis using bronchial washing cultures compared with sputum cultures. METHODS: Study conducted with 61 adults in Lima, Peru, from January 2006 to December 2007. The yield of sputum cultures was compared with the yield of acid-fast bacilli smears and cultures of bronchial washing for diagnosing pulmonary tuberculosis in suspected cases of clinical tuberculosis with negative acid fast bacilli sputum smears. RESULTS: Twenty seven (95%CI 32;58) of the cases were eventually diagnosed with smear-negative pulmonary tuberculosis. Bronchial washing samples detected 23 (95%CI 72;99) of the smear-negative pulmonary tuberculosis cases compared with 15 (95%CI 37;74) for sputum cultures (p = 0.02). The incremental diagnostic yield of acid fast bacilli smear and culture of bronchial washing specimens over sputum culture was 44% (95%CI 25;65). CONCLUSIONS: In function of the epidemiological context and the resources available, bronchoscopy should be deployed as part of a comprehensive work up that optimizes smear-negative pulmonary tuberculosis diagnosis and minimizes risk and costs. DESCRIPTORS: Bronchoalveolar Lavage. Tuberculosis, diagnosis. Diagnostic Techniques and Procedures. Sensitivity and Specifi city. Comunicação Breve DOI: 10.1590/S0034-8910.2013047004548 814 Incremental yield of bronchial washing for TB Soto A et al According to World Health Organization (WHO), smear negative pulmonary tuberculosis (SNPT) is defi ned as a case with at least two negative sputum acid fast bacilli (AFB) smears and either a positive culture or radiographic abnormalities consistent with tuberculosis and failure to respond to an antibiotic trial.5 Microbiological tests on fl exible bronchoscopy samples are considered useful for the diagnosis of SNPT, but their indication is not standardized. The procedure is performed on the premise that the bacillary load is higher in samples from bronchial secretions than in sputum. However, bronchoscopy is not without risk for patients and may be associated with nosocomial transmission of tuberculosis. The reported sensitivity in patients with clinical suspicion of tuberculosis varies from as low as 6.5% to up to 77%.2 However, evaluations of bronchoscopy are scarce, specifi cally for the detection of SNPT in prospective studies. Additionally, culture in liquid media is seldom performed. It is more sensitive than culture in conventional solid media of sputum and recommended by the WHO,4 but whether this is also the case for bronchoscopy samples is less clear. RESUMO OBJETIVO: Avaliar o aumento do rendimento do diagnóstico de tuberculose pulmonar mediante cultura de lavado brônquico em comparação ao rendimento da cultura de escarro. MÉTODOS: Estudo realizado com 61 adultos da cidade de Lima, Peru, a partir de janeiro de 2006 a dezembro de 2007. Foi comparado o rendimento de culturas de escarro com o de esfregaço de bacilos ácido-resistentes e cultura de amostras de lavado brônquico para diagnóstico de tuberculose pulmonar, em casos suspeitos de tuberculose clínica, com esfregaço de escarro de bacilos ácido-resistentes negativo. RESULTADOS: Vinte e sete (IC95% 32;58) dos casos foram diagnosticados com tuberculose pulmonar de baciloscopia negativa. As amostras de lavabo brônquio detectaram 23 (95%; IC95% 72;99) dos casos de tuberculose pulmonar de baciloscopia negativa, comparados com 15 (IC95% 37;74) de cultura de escarro (p = 0.02). O diagnóstico da tuberculose pulmonar mediante o uso do esfregaço de bacilos ácido-resistentes e da cultura de lavado brônquico tiveram um rendimento adicional de 44% (IC95% 25;65) em relação à cultura de escarro. CONCLUSÕES: Em função do contexto epidemiológico e dos recursos disponíveis, a broncoscopia deve ser adotada como parte de trabalho abrangente que otimize o diagnóstico de tuberculose pulmonar de baciloscopia negativa e minimize riscos e custos. DESCRITORES: Lavagem Broncoalveolar. Tuberculose, diagnóstico. Técnicas e Procedimentos Diagnósticos. Sensibilidade e Especifi cidade.

According to World Health Organization (WHO), smear negative pulmonary tuberculosis (SNPT) is defi ned as a case with at least two negative sputum acid fast bacilli (AFB) smears and either a positive culture or radiographic abnormalities consistent with tuberculosis and failure to respond to an antibiotic trial. 5 Microbiological tests on fl exible bronchoscopy samples are considered useful for the diagnosis of SNPT, but their indication is not standardized. The procedure is performed on the premise that the bacillary load is higher in samples from bronchial secretions than in sputum. However, bronchoscopy is not without risk for patients and may be associated with nosocomial transmission of tuberculosis.
The reported sensitivity in patients with clinical suspicion of tuberculosis varies from as low as 6.5% to up to 77%. 2 However, evaluations of bronchoscopy are scarce, specifi cally for the detection of SNPT in prospective studies. Additionally, culture in liquid media is seldom performed. It is more sensitive than culture in conventional solid media of sputum and recommended by the WHO, 4 but whether this is also the case for bronchoscopy samples is less clear.

INTRODUCTION
Further evaluation of the yield of bronchoscopy is needed in order to generate evidence-based recommendations for the use of the procedure in patients with clinical suspicion of SNPT. This study aimed to prospectively assess the incremental diagnostic yield of acid fast bacilli (AFB) smear and culture of a bronchial washing sample over the yield of sputum sample culture in such patients. We also compared the yield of liquid culture compared to conventional solid culture in bronchial washing samples.

METHODS
Study conducted in adults that attended the Hospital Nacional Hipólito Unanue, a tuberculosis reference hospital in Lima, Peru, from January 2006 to December 2007. We included patients with clinical suspicion of pulmonary tuberculosis (cough for 14 days or more) and two or more negative AFB sputum smears and one additional negative concentrated smear, who did not improve after a two week antibiotic trial. Patients with high clinical suspicion of tuberculosis who were put on anti-tuberculous treatment without further investigation by the treating pneumologist were excluded from the analysis.
All patients underwent bronchoscopy with bronchial washing, without bronchoalveolar lavage. The procedures were executed by trained pneumologists. Patients were pre-medicated with 0.5 mg atropine and then sedated with midazolam. Bronchial washing was performed by instilling 20 to 50 ml of 0.9% isotonic saline in the tracheobronchial tree up to the sub-segmental level. The sample was obtained by aspiration into a trap connected to a suction tube.
The sputum sample taken prior to the procedure was concentrated, decontaminated and cultured in solid Ogawa medium using standard methodology 1 and in mycobacteria growth indicator tube, manually read as previously described. 3 Bronchial washing samples were concentrated, directly examined for AFB and cultured in the same way.

RESULTS
Sixty-one patients were recruited. The median age was 33 (interquartile range 26 to 43 years old) and 52% were males; 21% had a previous history of tuberculosis. Fourteen patients refused HIV testing and one out of the 47 patients tested had a positive test. SNPT was diagnosed in 27/61 (95%CI 32;58) suspected cases. There were no complications from the procedure and bronchial washing samples were able to be obtained from all recruited patients.

DISCUSSION
The incremental yield of AFB smear and culture of bronchial washing specimens over sputum cultures (44.0%) is substantial in suspected cases of SNPT, and the diagnostic yield (85.0%) is high. The latter fi nding is consistent with the literature, but yield is highly dependent on the setting. Nevertheless, a few patients with positive sputum cultures have negative bronchial washing examinations. This may be due to the bacteriostatic effect of the local anesthetic, xylocaine 2 or to delays in processing the bronchial washing samples. Bronchoalveolar lavage could possibly result in an even better diagnostic yield but is not frequently performed in resource constrained settings. A considerable proportion of bronchoscopies could be foregone if the results of sputum cultures became more rapidly available: half of patients with SNPT diagnosed through bronchoscopy eventually had a positive sputum culture. Liquid media cultures give faster results and are more sensitive than culture in standard solid media and should be made more generally available. 4 The diagnostic yield of bronchial washing samples is better than that of sputum samples, but they are complementary and should be included in the work-up in function of the epidemiological context, patient presentation and resources available. Sputum induction, post-bronchoscopic sputum samples and the deployment of molecular techniques could all further