Benign lymphadenopathy can be seen in inflammatory conditions such as sarcoidosis and in infections such as tuberculosis and histoplasmosis. EUS-FNA is a safe and minimally invasive method of obtaining tissue and has a robust yield when the index of suspicion is high [11–14]. A recent randomized study of 301 patients suspected of having sarcoidosis compared conventional bronchoscopy with transbronchial and endobronchial mucosal biopsies to EUS or EBUS with FNA [15]. In the EUS/EBUS arm, 22-gauge needles were used, and when a cytologist was not present, 4 passes were performed into the lymph node. Diagnostic yield for detecting granulomas was significantly higher in the ultrasound arm (80 vs. 53 %, p < 0.001) with EUS performing better than EBUS (88 vs. 66 %, p < 0.01). A retrospective study evaluated 124 patients with mediastinal lymphadenopathy of unclear etiology who underwent EUS-FNA using predominantly a 22-gauge needle with 4 passes into the nodes with onsite cytopathology [14]. EUS-FNA was highly accurate for diagnosing sarcoidosis with a sensitivity and specificity of 89 and 96 %, respectively. Most of the lymph nodes were located in the subcarinal or the AP window. In another retrospective study of 49 patients who had mediastinal masses without known lung cancer, the overall diagnostic yield of EUS-FNA was 94 % [16]. About half these patients had benign diseases which included benign lymph nodes, histoplasmosis, sarcoidosis, and duplication cyst, and FNA was performed with a 22-gauge needle and ROSE. EUS-FNA also accurately distinguishes between mediastinal lymphadenopathy resulting from sarcoidosis and tuberculosis [17]. In this study, patients had undergone an extensive negative evaluation including bronchoscopy and bronchoalveolar lavage before EUS-FNA. A 22-gauge needle was used with 2 passes for cytology and 1 for microbiology. EUS-FNA provided a definite diagnosis in 89 % of the cases with 86 % sensitivity and 100 % specificity for tuberculosis and 100 % sensitivity and 93 % specificity for sarcoidosis.

EUS-FNA is an excellent tool for detecting malignant disease in the mediastinum. Bronchoscopy with or without EBUS is often the first diagnostic test in lung cancer. EUS-FNA is an excellent adjunct tool when a diagnosis has not been made by bronchoscopy, particularly when inferior or posterior lymphadenopathy is present. EUS detected lung cancer in 25/35 patients suspected of having lung cancer despite a negative bronchoscopy and missed the diagnosis in only 1/35 [18]. The other 9 patients had benign disease to yield an overall diagnostic sensitivity and specificity for EUS-FNA of 96 and 100 %. EUS-FNA was performed of lymph nodes in the high paratracheal, aortopulmonary, subcarinal, paraesophageal, and hilar regions. The ability of EUS to sample subcentimeter nodes makes it superior to CT and PET [19].

In a retrospective study on patients with non–lung cancer malignant mediastinal disease, EUS-FNA detected colon cancer, breast cancer, laryngeal cancer, renal cell cancer, lung cancer, and metastatic disease from an unknown site in 22/49 patients. The accuracy of EUS-FNA for malignant and benign diagnoses was over 90 % [16].

A recent retrospective study evaluated the value of EUS and EUS-FNA in diagnosing mediastinal lymphadenopathy as benign, sarcoidosis, lymphoma, or metastatic disease in patients not suspected of having lung cancer [20]. While larger node size was associated with malignancy, a large overlap of size between benign and malignant nodes made this unreliable as the sole criterion for diagnosis. Similarly, the combination of echo features suggestive of malignancy (round, well defined, homogeneous, size > 1 cm) could not reliably differentiate benign from malignant lymph nodes as approximately 20 % of sarcoidosis, 40 % of lymphoma, and 20 % of metastases exhibited these features. Therefore, for mediastinal lymph nodes diagnosed as benign by imaging, further evaluation should occur to rule out malignancy.

EUS can define the primary tumor and its relationship to surrounding structures, particularly invasion into vasculature, such as the left atrium and other mediastinal structures (e.g., aorta, azygos vein) to establish T4 disease with 87 % sensitivity and 98 % specificity [24]. Documentation of T4 disease would preclude surgery.

Mediastinal lymph node evaluation is the primary role of EUS and EUS-FNA in lung cancer staging (Fig. 29.10). According to the American Joint Commission on Cancer (AJCC), mediastinal lymph nodes are stage N2 when they are ipsilateral (same side) or midline (subcarina is always considered midline) to the tumor. Contralateral lymph nodes are considered N3 and carry a worse prognosis than N2. Thus, contralateral lymph nodes should be sampled first. If N3 lymph nodes are confirmed by ROSE to be positive, then no further sampling is needed. In most cases, we recommend sampling all suspicious (> 1 cm, round, hypoechoic, sharp borders) N3 lymph nodes, then, if necessary, all suspicious N2 lymph nodes. In the absence of a suspicious lymph node, it may still be valuable to sample visible lymph nodes in the common stations.

Patients with abnormal mediastinal lymph nodes by conventional cross-sectional imaging (CT scan) or PET scan should undergo lymph node sampling [25]. The sensitivity and specificity of diagnosing metastatic disease by imaging alone are inadequate. EUS-FNA is effective at detecting metastatic disease in lymph nodes with an accuracy of 83–97 % and a sensitivity of 84–92 % [11, 21, 23, 26–31]. A prospective study demonstrated sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of 92, 100, 100, 94, and 97 %, respectively, in a series of 104 patients with suspected lung cancer who underwent mediastinal lymph node EUS-FNA [27]. As previously described, EUS-FNA and EBUS-TBNA can be used together for incremental benefit. EBUS provides access to anterior and superior mediastinal lymph nodes, while EUS allows access to posterior and inferior mediastinal lymph nodes . When feasible, combined EUS and EBUS in the same setting are most efficient for patient care; however, there are many practical challenges, especially if performed by two separate (GI and pulmonary) physicians. A practical approach is to first do the procedure most likely to confirm nodal metastases (EBUS for anterior lymph nodes or subcarina; EUS for AP window, posterior, and lower paraesophageal), followed by the other procedure if the first is negative.

Endosonography may be uniquely useful in identifying and sampling lymph nodes based on EUS criteria in the absence of lymphadenopathy on cross-sectional imaging [32, 33]. In two studies, EUS was able to detect metastatic disease in 25 % of patients with no nodal enlargement by CT scan and also detected advanced local disease in 12 % and thus prevented unnecessary surgery. The lesions detected included invasion of mediastinal structures, contralateral and distant lymph node involvement, esophageal involvement, and adrenal metastases [34, 35]. Thus, EUS (or EBUS) should be considered in most patients with suspected lung cancer. An exception is a patient with a peripheral T1 tumor and negative CT/PET of the mediastinum, in whom mediastinal metastases are very rare.

A few small studies have demonstrated a benefit in analyzing the cytology samples for genes that may detect micrometastases in up to 19 % of cytologically negative lymph nodes [36–38].

While distant metastatic disease is generally detected by cross-sectional imaging, EUS provides a unique opportunity to evaluate and sample abdominal metastatic disease such as celiac lymph nodes and liver and adrenal metastasis at the same time as mediastinal EUS [9, 27, 34]. Endosonography identified celiac lymphadenopathy in 11 % of patients undergoing lung cancer staging and was superior to CT for detecting distant metastases (97 vs. 89 %, p = 0.02) [39]. This was mainly due to the superior diagnostic yield of EUS-FNA over CT for detecting malignant celiac nodes (100 vs. 50 %, p < 0.05). The presence of malignant celiac lymph nodes portends a poor prognosis.