Endobronchial Ultrasound Transbronchial Needle Aspiration in Thoracic Diseases: Much More than Mediastinal Staging

Background and Objective EBUS-TBNA has revolutionized the diagnostic approach to thoracic diseases from a surgical to minimally invasive procedure. In non small-cell lung cancer (NCSLC) patients, EBUS-TBNA is able to dictate the consecutive therapy both for early and advanced stages, providing pathological diagnosis, mediastinal staging, and even adequate specimens for molecular analysis. This study reports on the ability of EBUS-TBNA to make different diagnoses and dictates the consecutive therapy in a large cohort of patients presenting different thoracic diseases. Methods All procedures performed from January 2012 to September 2016 were reviewed. Five groups of patients were created according to the main indications for the procedure. Group 1: lung cancer staging; Group 2: pathological diagnosis in advanced stage lung cancer; Group 3: lymphadenopathy in previous malignancies; Group 4: pulmonary lesions; Group 5: unknown origin lymphadenopathy. In each group, the diagnostic yield of the procedure was analysed. Non malignant diagnosis at EBUS-TBNA was confirmed by a surgical procedure or clinical and radiological follow-up. Results 1891 patients were included in the analysis. Sensitivity, negative predictive value, and diagnostic accuracy in each group were 90.7%, 79.4%, and 93.1% in Group 1; 98.5%, 50%, and 98.5% in Group 2; 92.4%, 85.1%, and 94.7% in Group 3; 90.9%, 51.0%, and 91.7% in Group 4; and 25%, 83.3%, and 84.2% in Group 5. Overall sensitivity, negative predictive value, and accuracy were 91.7%, 78.5%, and 93.6%, respectively. Conclusions EBUS-TBNA is the best approach for invasive mediastinal investigation, confirming its strategic role and high accuracy in thoracic oncology.


Introduction
Mediastinal adenopathy has always been assessed by radiological imaging such as computed tomography (CT) and positron emission tomography (PET), with high sensitivity but a low diagnostic accuracy for the purposes of correct clinical decision-making [1]. To date, mediastinoscopy has been considered the gold standard for diagnosis and mediastinal staging with high sensitivity and accuracy, but the procedure has been progressively underused due to its high invasiveness, risk of complications, and the need to be performed in experienced centres [2].
In the early 2000's, a minimally invasive convex probe endobronchial ultrasound (EBUS) procedure able to perform real-time transbronchial needle aspiration (TBNA) was described with high accuracy for mediastinal and hilar lymph node staging [3]. Since then, EBUS-TBNA has gradually changed the way mediastinal staging is performed and rapidly improved its value with new indications in lung cancer management, becoming the standard of care [4].
In thoracic oncology, EBUS-TBNA has revolutionized the diagnostic approach to lung cancer and other neoplasms from a surgical to minimally invasive procedure. Especially in non small-cell lung cancer (NCSLC), EBUS-TBNA is able to dictate the consecutive therapy both for early and advanced stages, providing pathological diagnosis, mediastinal staging, and even adequate specimens for molecular analysis [5]. In addition, EBUS-TBNA has been described in di erent clinical scenarios, particularly for the diagnosis and denition of granulomatosis such as sarcoidosis [6] and tuberculosis [7] and for pathological assessment of mediastinal and hilar recurrences from previous malignancies [8,9].
is study reports the largest published experience in the use of EBUS-TBNA in a high-volume thoracic oncology institution. We aimed to describe the utility and diagnostic yield of EBUS-TBNA in di erent clinical scenarios in thoracic diseases, dividing our series into ve di erent groups according to the main indication for the procedure (Group 1: lung cancer staging, Group 2: pathological diagnosis in advanced stage lung cancer, Group 3: lymphadenopathy in previous malignancies, Group 4: pulmonary lesions, and Group 5: unknown origin lymphadenopathy) and reporting our results in terms of sensitivity, negative predictive value, and diagnostic accuracy.

Methods
is single-centre retrospective study with a prospective follow-up was approved by the Institutional Review Board and the individual consent was obtained.
From January 2012 to September 2016, 1958 EBUS-TBNA procedures were performed at our institution. e indications for EBUS-TBNA, lymph node stations, number of lymph nodes sampled, cytological results, and cancer cell type were obtained for the analysis. To better standardize the series, di erent groups of patients were de ned according to the indication for the procedure.
Group 1 included patients referred for EBUS-TBNA for mediastinal staging in potentially operable lung cancer and patients with mediastinal involvement but no evidence of distant metastasis. Patients with proven or suspected NSCLC were included, and pathological cell type was performed in the same procedure. According to our institutional protocol, suspect lymph nodes were de ned as lymph nodes with a pathological PET scan uptake and/or enlarged lymph nodes with more than 1 cm in the short axis at the CT scan.
Group 2 included patients with metastatic and/or bulky mediastinal disease referred for EBUS-TBNA for pathological diagnosis and molecular mutational analysis for targeted therapy; Group 3 included patients with a previous (thoracic or extrathoracic) malignancy who developed mediastinal and hilar lymphadenopathies suspected for recurrence; Group 4 included patients who underwent EBUS-TBNA for primary tissue sampling in pulmonary lesions (paratracheal or peribronchial); and Group 5 included patients with mediastinal and hilar lymphadenopathy of unknown origin with no history of malignancy. Diagnostic sensitivity, accuracy, and negative predictive value were calculated according to standard de nitions. Sensitivity was calculated for the diagnosis of malignancy.
EBUS-TBNA samples were considered diagnostic when a de nitive diagnosis was obtained. Lymph node samples negative for malignancy underwent surgical con rmation (mediastinoscopy or VATS) or clinical and radiological follow-up (at least 12 months of follow-up). Patients with loss of follow-up were excluded.
EBUS-TBNA samples were considered false negative when surgery changed the nal diagnosis, or there was clinical and radiological progression of the disease during follow-up. EBUS-TBNA samples insu cient or inadequate for diagnosis were considered false negative in the calculation of diagnostic accuracy. Patients without a speci c diagnosis at EBUS-TBNA (benign or malignant) but a denitive non-malignant diagnosis at mediastinoscopy were considered false negative.
During EBUS-TBNA procedures, all lymphadenopathies with increased PET scan pathological uptake were sampled in patients with a non-malignant diagnosis at rapid on-site evaluation (ROSE), and at least one mediastinal and two hilar stations were sampled when ROSE was negative for tumour cells and suspected for lymphadenitis granulomatosis.
Patients with negative EBUS-TBNA samples were referred for con rmatory surgical procedures after a multidisciplinary team discussion with thoracic surgeons, pulmonologists, radiologists, oncologists, and radiotherapists when lymphadenopathies were considered highly suspicious for recurrence based on CT and/or PET scan characteristics.

EBUS-TBNA Technical
Aspects. EBUS-TBNA procedures were performed under local anaesthesia (1% lidocaine), and moderate sedation was provided by an anaesthesiologist with spontaneous ventilation. e same team of interventional pulmonologists using a convex probe (EBUS Convex Probe BF-UC180F; Olympus) and a dedicated ultrasound processor (EU-ME1; Olympus) performed all procedures. All EBUS-TBNA specimens were collected with a 22 gauge dedicated needle (Vizishot NA-201SX-4022; Olympus).
A very small amount of the aspirated material was pushed out by the internal stylet and smeared onto glass slides for immediate on-site evaluation (ROSE). Mirror slides were alcohol-xed for posterior evaluation. Air-dried smears were immediately stained with a modi ed May-Grünwald Giemsa stain (MGG Quick Stain; Bio-Optica, Milan, Italy) and evaluated by the cytopathologist to conrm adequate tumour cells and/or lymph node material. e remaining aspirate and other needle passages (at least three needle passages each station) were xed in a formalin solution for cell block processing and histological evaluation. Alcohol-xed smears were stained with Papanicolaou and haematoxylin-eosin stains.

Results
Out of 1958 patients, 1891 were included in the study. Sixtyseven patients were excluded from statistical analysis due to loss of follow-up with the impossibility to con rm the negativity of lymph node samples. Patients' mean age was 65 years (range: 20-92), 1197 (63.3%) were men. e leading indication for EBUS-TBNA at our institution was lung cancer staging in 728 (38.5%) patients followed by pathological diagnosis and mutational status in advanced stage lung cancer in 401 (21.2%) patients, lymphadenopathy in previous malignancies in 320 (17%) cases, diagnosis of pulmonary lesions in 290 (15.3%), and diagnosis of unknown origin lymphadenopathy in 152 (8%) patients. Patient characteristics and indications for the procedures are reported in Table 1.

Group 1: Lung Cancer Staging.
Out of 728 (60.2%) patients, 438 underwent EBUS-TBNA for mediastinal staging and the diagnosis of the cell-type tumour in the same procedure, and 244 (33.5%) had a previous pathological diagnosis of NSCLC, and EBUS-TBNA was performed for complete mediastinal staging, and in 46 (6.3%) patients, EBUS-TBNA was performed for restaging the mediastinum after induction chemotherapy for stage IIIA (pN2) NSCLC. In this group, EBUS-TBNA was diagnostic for malignancy in 485 (66.6%) patients, and a mean of 1, 5 lymph nodes per patient was biopsied. In 187 (25.7%) cases, EBUS-TBNA revealed normal or reactive lymph node samples, and in six (0.8%), EBUS-TBNA revealed a granulomatous reaction in ve cases of sarcoid-like reaction compatible with sarcoidosis and one case of necrotizing granulomatosis with a positive culture for Mycobacterium tuberculosis.

Group 2: Pathological Diagnosis and Mutational Status in Advanced Stage Lung Cancer.
is group comprised 401 patients. In 344 (85.8%) patients, EBUS-TBNA was performed

Discussion
Many literature studies have reported on EBUS-TBNA experience for mediastinal staging in lung cancer patients and some have described the utility of EBUS-TBNA in other pathologies with mediastinal lymphadenopathies such as sarcoidosis, tuberculosis, and lymphoma [10,11].
During the last decade, EBUS-TBNA has become an essential diagnostic procedure in thoracic disease, revolutionizing the approach to diagnosis and treatment and guiding the best treatment option in a large number of patients. To date, the procedure has been widely used to perform mediastinal staging and has been included in almost all guidelines as the preferred rst approach for invasive mediastinal staging in lung cancer patients [4,12].
A recent meta-analysis showed a sensitivity of EBUS-TBNA for mediastinal staging ranging from 81 to 95% [13]. Likewise, an Italian multicentre trial published in 2017 analysed 485 patients showing a sensitivity of 90% with a diagnostic accuracy of 93% [14]. In agreement with these publications, our study showed an EBUS-TBNA sensitivity for malignant disease of 90.7% and a diagnostic accuracy of 93.1% in 728 patients in our Group 1 patients (lung cancer staging). Surprisingly, the nal EBUS-TBNA diagnosis was lymphoma in ve patients with a single pulmonary lesion and mediastinal lymphadenopathies highly suspected for lung cancer. EBUS-TBNA samples were able to subtype the lymphoma in three of the ve (60%) cases achieving a de nitive diagnosis of Hodgkin's lymphoma, classic variation, showing the feasibility of lymphoma diagnosis and subclassi cation by EBUS-TBNA samples even if the need for further histological evaluation remains essential in many cases [15,16]. In the subgroup of patients who underwent EBUS-TBNA for restaging the mediastinum after neoadjuvant chemotherapy, there were a relative higher incidence of inadequate samples due to the presence of brosis in the mediastinum and the di culty of the procedure specially after chemoradiotherapy.
Considering our Group 2 (pathological diagnosis and mutational status in advanced stage lung cancer), EBUS-TBNA sensitivity and diagnostic accuracy were both 98.5%. EBUS-TBNA also provided adequate material for molecular analysis whenever requested in advanced stage adenocarcinomas in 98% of cases. is datum con rms our previous published experience of molecular analysis with EBUS-TBNA specimens where we showed a sensitivity of 96.9% [17]. Our previous study also demonstrated that molecular analysis obtained from EBUS-TBNA specimens was equivalent to that obtained from surgical specimens, with no di erences in terms of sensitivity and diagnostic accuracy [17].
Due to the low invasive and the feasibility of repeated procedures, EBUS-TBNA is the ideal procedure to establish a cell-type diagnosis and to provide adequate specimens for molecular assay in advanced lung cancer patients. e procedure's high accuracy and low risk of complications make it suitable for low performance status patients who could still bene t from targeted therapy [18].
In Group 3 (lymphadenopathy in previous malignancy), EBUS-TBNA showed a sensitivity and diagnostic accuracy of 92.4% and 94.7%, respectively, in agreement with a previous similar study published by Navani et al. and also with our previous reported experience in this subset of patients [8,9]. In this group of patients, EBUS-TBNA was a crucial diagnostic approach. e results of this study showed that 35.3% of patients with lymphadenopathy suspected for a recurrence did not really have a recurrence, and EBUS-TBNA revealed a di erent diagnosis. From all nonmalignant diagnoses, lymphadenitis granulomatosis was found in 36% of patients. EBUS-TBNA specimens provided adequate diagnosis avoiding unnecessary treatments in cases of nonmalignant disease and guiding therapy in cases of recurrence.
In the diagnosis of Group 4 patients (pulmonary lesions), EBUS-TBNA demonstrated a sensitivity and accuracy of 90.9% and 91.7%, respectively, in line with Nakajima et al.'s published data (sensitivity of 94.1% and diagnostic accuracy of 94.3% in the diagnosis of peribronchial and peritracheal lesions) [19]. Clinical presentations vary widely in this group, and EBUS-TBNA was able to establish a correct diagnosis in the vast majority of cases also in nonmalignant intrapulmonary lesions. Some interesting ndings in this group included the diagnosis of a benign leiomyoma, a mycetoma, and ectopic thymus tissue.
In the investigation of Group 5 (unknown origin lymphadenopathy), EBUS-TBNA presented a diagnostic accuracy of 84.2%. Due to the very low prevalence of malignancy in this group of patients, sensitivity was not surprisingly very low (25%). e vast majority of patients in this group had an in ammatory disease with the most frequent diagnosis of granulomatous lymphadenitis compatible with sarcoidosis.
Our results refer to the largest published series in EBUS-TBNA, in a high-volume thoracic oncology institution, demonstrating the utility and high accuracy of EBUS-TBNA in di erent clinical scenarios in daily clinical practice.
Some additional ndings have also emerged from this study. First, our study showed a wide use of EBUS-TBNA in elderly patients (732 patients over 70 years old and 120 over 80 years old), with a very low rate (0.74%) of postprocedure minor complications and no major events. Other interesting ndings included the possibility to repeat biopsies easily and safely, obtaining tumour specimens for genetic assessment in stage IV NSCLC patients and in other tumour recurrences without the need for further more invasive procedures. ese aspects highlight the importance of EBUS-TBNA as a diagnostic approach in di erent thoracic oncology scenarios, underpinning modern oncological strategy also for fragile patients and thereby avoiding more invasive and demanding procedures.
An essential aspect to achieve optimal results is the specimen handling in EBUS-TBNA [5]. e use of ROSE and synergy with the pathologist is crucial during the procedure both to obtain the best specimen collection and to achieve all pathological, immunohistochemistry, and molecular analyses. Despite some di erent previous points of view regarding ROSE [20][21][22], a recent metaexpert panel review showed that ROSE is necessary to reach all molecular analyses and to prevent invasive surgical procedures after EBUS-TBNA [23].
Our study has some limitations. e rst is its retrospective nature, which probably limits and in uences all of Table 4: Diagnostic performance of EBUS in the di erent groups.

EBUS
Overall Group 1  Group 2  Group 3  Group 4  Group 5  Total  1891  728  401  320  290  152  True negative  441  193  6  97  25  120  False negative  121  50  6  17  24  24  True positive  1329  485  389  206  Canadian Respiratory Journal 5 the variables included in the analysis. Another bias is that not all patients underwent a surgical procedure (mediastinoscopy or VATS) to con rm negative EBUS-TBNA samples but all patients had a clinical and radiological follow-up longer than 15 months. In our experience, EBUS-TBNA is not just a technique, but a philosophy for the investigation of thoracic disease. EBUS-TBNA feasibility and optimal results largely depend on the cytopathology and pulmonologists' experience but also the way the procedure is performed. Adequate sedation and ROSE are crucial and should always be used. In experienced hands, EBUS-TBNA o ers everything chest physicians need for the best medical practice without further invasive procedures. Surgical procedures are still required in highly suspicious NSCLC cases when EBUS-TBNA is negative for malignancy and in some cases for lymphoma subtyping.
In conclusion, EBUS-TBNA represents the best diagnostic approach in many di erent clinical scenarios in high-volume thoracic oncology centres and dictates the consecutive treatment option in the vast majority of patients. EBUS-TBNA is useful and accurate in a high percentage of cases underpinning modern oncological therapy, guiding the best treatment options in lung cancer and avoiding useless and invasive procedures in case of benign disease.
Based on our previous expertise, we are developing new applications for EBUS-TBNA such as a new protocol for EBUS-TBNA primary lung cancer cell culture and micro-RNA pro le in stage IIIA (pN2) NSCLC patients, testing the feasibility of a microRNA pro le obtained from EBUS-TBNA [24] and the correlation with the prediction of chemotherapy response. In addition, new molecular analyses are being developed, and the correlation between PDL-1 expression and tumour histology is under evaluation.