We retrospectively reviewed the diagnostic performance of EBUS-TBNA in consecutive patients with high clinical suspicion of a centrally located primary lung cancer who had undergone EBUS-TBNA at the Samsung Medical Center between May 2009 and June 2011. The centrally located lung lesions were defined as an intrapulmonary nodule or mass located adjacent to the tracheobronchial tree as visualized on chest CT scan. This study was approved by the Institutional Review Board of the Samsung Medical Center, which waived the requirements for informed consent of the individual patients, given the retrospective nature of the study.

EBUS-TBNA was performed by three trained operators using a curvilinear scanning ultrasound bronchoscope (BF UC260F-OL8, Olympus Ltd., Tokyo, Japan) connected to an ultrasound unit (EU-C60 Olympus Ltd., Tokyo, Japan). The procedures were performed under local anesthesia (lignocaine) and moderate sedation (midazolam). For paratracheal lesions, the scope was positioned endotracheally. For peribronchial lesions the scope was positioned in the respective bronchi in order to visualize the lung lesion. TBNA was performed using a 22-gauge needle (NA-2015X-4022, Olympus Ltd., Tokyo, Japan). Two aspirates were performed with 15 passes (moving needle back & forth in the lesion) per aspirate. The core tissue was expelled onto piece of paper for histological examination and the needle was flushed with saline onto glass slides for cytological examination. The aspirate was smeared onto glass slides, air dried, fixed immediately with 95% alcohol, and stained with Hematoxylin and Eosin (HE). Histological cores were fixed with 10% neutral buffered formalin and stained with HE. Rapid on-site cytological examination was not available. A post-procedure chest X-ray was routinely performed to exclude any procedure-related complications.

Cyto-pathological specimens from EBUS-TBNA were categorized as malignant (presence of malignant cells), benign (normal lung tissue without malignant cells), or inadequate (no or scanty cellular component, blood only, mucus, or benign bronchial cells only). Malignant and benign samples diagnosed by EBUS-TBNA were defined as adequate samples. Tumor-positive findings from the EBUS-TBNA samples were not surgically validated. The diagnostic standard for benign disease was pathological or microbiological confirmation of a specific benign disease, surgical confirmation of lesions showing no malignant cells, or spontaneous regression of lung parenchymal lesion after EBUS-TBNA during follow-up of more than 6 months. Malignant disease was defined as confirmation of malignant cells or tissues from EBUS-TBNA, CT-guided transthoracic needle aspiration or surgical lung biopsy. Cases which did not satisfy the criteria of benign or malignant disease were classified as indeterminate cases.

If the pathology from the EBUS-TBNA samples resulted in a diagnosis of malignancy, corroborative evidence of malignancy was looked for in the form of surgical lung biopsy or biopsy results of tissue obtained from some other site suspected of malignant involvement. The results of the concurrent EBUS-TBNA of mediastinal lymph nodes were excluded from analysis.

All data are presented as a number (%) or median (range) unless otherwise stated. Diagnostic performance analyses were performed in relation to diagnosis of malignancy and diagnosis of all disease (benignity and malignancy). Patients who had an indeterminate final diagnosis were considered a false negative result of EBUS-TBNA. The sensitivity, specificity, and diagnostic accuracy for detecting malignancy were calculated using the standard definitions.

Thirty-seven patients underwent EBUS-TBNA for intrapulmonary lesions adjacent to the central airways. There were 25 (68%) males and the median age was 63 years (40-81). The size of pulmonary lesions on CT varied from 8 to 82 mm in short axis with a median of 27.5 mm (22 lesions were less than 3 cm and 15 were larger than 3 cm). Positron emission tomography (PET) scan was done in 34 (92%) patients and increased FDG uptake was seen in the parenchymal mass lesion in all patients. However, median standard uptake value (SUV_max) was only available in 29 (78%) patients as 8 patients had PET scan done outside our institution. Median SUV_max was 12 (3.3-47.3) (Table 1). Fifteen (40.5%) patients had endobronchial abnormality. Twelve (32%) had mucosal or submucosal abnormality such as exophytic tumor or hyperaemia, and 3 (8%) had extrinsic compression. Twenty-two (59%) had no endobronchial abnormality (Table 1).

Locations of primary tumors sampled by EBUS-TBNA and puncture sites are shown in Table 2. Right upper lobe lesions were targeted predominantly through the trachea, right main stem bronchus, or bronchus intermedius. Left upper lobe masses were sampled through the left upper lobar bronchus at its orifice, the distal left main bronchus and the left side of the trachea. Both lower lobar masses were targeted through the distal main bronchus, bronchus intermedius or the basal trunk. Two patients experienced complications. One developed pneumothorax requiring chest tube insertion and another patient developed moderate bleeding which resolved spontaneously. Both patients recovered from their complications uneventfully. A representative case of EBUS-TBNA for lung parenchymal lesion is described in Fig. 1.

Twenty-two out of 37 (59.4%) patients had undergone mucosal and transbronchial biopsies by conventional bronchoscopy. Out of these, 19 patients had it done at a median (range) interval of 4.5 (1-7) days prior to EBUS-TBNA and 3 patients had it done in the same sitting as EBUS-TBNA. The results of conventional bronchoscopic biopsy were non-diagnostic in 17 patients (15 patients in the prior to EBUS-TBNA group and in 2 patients in the same sitting group) with an overall non-diagnostic rate of 17/22 (77.3%). Out of 5 cases that had positive bronchoscopic biopsy, one had a slit like narrowing of the distal left main bronchus with a submucosal lesion, one had a submucosal lesion in the left upper lobe, two had mucosal abnormality in the right upper lobe and one had a left lower lobe mucosal lesion. EBUS-TBNA was performed on these patients, despite positive yield, to ascertain the cell type when the cell type was not clearly discernible (doubtful) based on conventional bronchoscopy, or to obtain additional tissue for mutation analysis (for patients who had bronchoscopy done outside our institute). Seventeen cases were judged to be non-accessible by bronchoscopic biopsies based on chest CT scans and bronchoscopic findings. Sputum was evaluated in 21 patients and was positive in one (2.7%). TTNA was done in 7 patients and was positive in one (2.7%) of these patients (Table 3).

Out of the 32 cases of malignancy detected by EBUS-TBNA, corroborative evidence of malignancy was found in the cerebrospinal fluid in 1 case, brain biopsy in 1 case, lobectomy in 1 case, tonsillectomy in 1 case, colon polypectomy in 1 case, tonsil and duodenal biopsy along with bone marrow biopsy in 1 case, pleural fluid in 1 case, thyroid fine-needle aspiration in 1 case and supraclavicular lymph node aspiration in 4 cases (Table 4).

Pathologic diagnosis was achieved in 32 patients (86.4%) by EBUS-TBNA (Table 5, Fig. 2). The diagnoses established were lung cancer in 30 patients [7 small cell lung cancer (SCLC), 23 non-small cell lung cancer (NSCLC)], lymphoma in one and malignant fibrous histiocytoma in one (Table 5, Fig. 2). EBUS-TBNA was non-diagnostic in 5 patients. Out of these, one patient was diagnosed with Epstein-Barr virus (EBV) related lymphoproliferative disorder by lobectomy, one was diagnosed with tuberculosis based on positive culture of bronchoalveolar lavage fluid, one was diagnosed with benign disease due to interval radiological resolution, one was diagnosed with non-small cell lung carcinoma on bronchoscopic biopsy and one did not undergo any confirmatory test due to refusal. Accordingly, diagnoses of malignant disease, benign disease and indeterminate cases were confirmed in 34, 2, and 1 patient, respectively. Among 23 patients with a final diagnosis of NSCLC, 16 had advanced disease (14 with stage IV disease and 2 with stage IIIB disease), five had stage IIIA disease, one had stage IIB disease and one had stage IB disease. Among patients with a final diagnosis of SCLC, five had extensive disease and two had limited disease. Overall, three (8%) patients underwent surgery with curative intent. The diagnosis in these patients was EBV related lymphoproliferative disease (n=1), stage IB NSCLC (n=1) and malignant fibrous histiocytoma (n=1).

The sensitivity (95% confidence interval), specificity (95% confidence interval) and diagnostic accuracy of EBUS-TBNA for the diagnosis of malignancy was 91.4% (89.2-100), 100% (15.0-85.0) and 91.9%, respectively. The diagnostic sensitivity (95% confidence interval) of EBUS-TBNA in detecting both malignancy and benignity (all cases) was 86.5% (71.2-95.4) (Table 6).