Initial evaluation and research studies in EUS utilized the radial echoendoscope to find lymph nodes, since it provided cross-sectional imaging similar to the computed tomography (CT) scan. However, radial echoendoscopes lacked the capability to provide tissue. Almost invariably, most endosonographers have now moved away from this practice and utilize curvilinear echoendoscopes that allow imaging and fine-needle aspiration (FNA) at the same time.

Endoscopic ultrasound is performed under conscious sedation with the curvilinear array (CLA) echoendoscope. The examination starts by full evaluation of the left lobe of the liver and the left adrenal gland by imaging it from the fundus of the stomach. The descending aorta is identified with the CLA echoendoscope at about 35 cm from the incisor. A continuous and steady push of the CLA echoendoscope to about 45 cm from the incisor, while the aorta is maintained in view, leads to identification of the celiac axis bifurcation. A gentle clockwise maneuver will lead to the identification of a “seagull” shaped organ: the adrenal gland. In patients with metastasis to the adrenal, the gland loses its normal shape and takes the form of a mass. Then the echoendoscope is gradually withdrawn to evaluate the inferior pulmonary ligament nodal region (station 9), the periesophageal areas (station 8), the subcarinal space (station 7), the aortopulmonary window (station 5), and the upper and lower paratracheal area (stations 2 and 4, respectively). To identify the 4L station or even the aortopulmonary (AP) window (station 5), the endoscope is withdrawn from the gastroesophageal junction cephalad to just below the arch while keeping the descending aorta in view. At that particular location, and by torquing 90 degrees clockwise and tipping the up–down dial in the upward direction, the AP window is identified. It is, however, controversial whether the station identified is the AP window station or the 4L station. Once a lymph node is identified, EUS–FNA is performed with a curvilinear echoendoscope (Olympus UC-30P or UCT 140, Olympus, Melville, NY).

All EUS–FNAs are performed using 22-gauge adjustable length Echotip needles (Wilson-Cook Inc., Winston Salem, NC), or the Olympus Ezshot needle. Occasionally, when a core tissue is required, as in the diagnosis of lymphoma, core tissue is obtained utilizing 19-gauge Trucut needle.¹² The procedure carries a very small risk of mediastinitis, bleeding, or stridor in patients with lung disease. Antibiotics are not routinely needed when lymph nodes are being sampled.

EUS, through the esophageal window, is able to assess the relationship of the tumor to the mediastinal vasculature in only a minority of patients owing to its limited penetration in the mediastinum. However, EUS can provide valuable information for vascular invasion when a tumor is seen (Fig. 14-1). One study evaluated 175 patients with lung cancer who underwent EUS.⁴⁵ Ten patients were found by EUS to have stage T4 tumors; 7 were confirmed to be T4 by either surgical exploration (n = 2), CT demonstration of aortic invasion (n = 3), or documentation of malignant pleural effusion (n = 2). Three of the 10 (30%) patients reported as having stage T4 tumors by EUS had T2 disease at surgery and underwent curative resection. Of the remaining
165 patients without evidence of T4 disease at EUS, only one was found to have aortic invasion (T4) at surgery. EUS had a sensitivity of 87.5%, specificity of 98%, positive predictive value of 70%, and negative predictive value of 99% for detecting T4 disease.⁴⁵

EUS–FNA is particularly useful for sampling inferior pulmonary ligament, esophageal, subcarinal, and aortopulmonary window lymph nodes (stations 9, 8, 7, and 5). Nodes anterolateral to the trachea are more difficult to sample reliably.

In a recent comprehensive meta-analysis, results of 18 studies were collated.³² EUS–FNA had a pooled sensitivity of 83% (95% CI, 78%–87%) and a specificity of 97% (95% CI, 96%–98%). For patients with enlarged lymph nodes seen on CT (n = 560), the pooled sensitivity was 90% (95% CI, 84%–94%); the specificity was 97% (95% CT, 95%–98%). For patients without enlarged lymph nodes on CT (n = 175), the pooled sensitivity of EUS–FNA was 58% (95% CI, 39%–75%), and the pooled specificity was 98% (95% CI, 96%–99%). This difference in diagnostic accuracy between studies enrolling patients with adenopathy on CT and those without adenopathy was statistically significant (p = 0.01).

Few studies have compared CT, PET, and EUS performance in staging the mediastinum. A study from Germany that included 33 patients showed that EUS–FNA was superior to CT and positron emission tomography (PET) combined.¹⁷ A larger study from the United States enrolled 104 patients with proven or suspected lung cancer who, by PET or CT, had an abnormal lymph node in the posterior mediastinum (station 5, 7, 8, or 9). These patients were then referred for EUS–FNA to sample posterior mediastinal lymph nodes after EUS and PET scanning.¹⁰ The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of EUS–FNA were 92.5%, 100%, 100%, 94%, and 97%, respectively. EUS–FNA was more accurate and had a higher positive predictive value than the PET or CT scanning (p = 0.001) in confirming cancer in the posterior mediastinal lymph nodes.

In addition to identifying nodal involvement, EUS–FNA can reliably sample the left lobe of the liver and the left adrenal gland (Fig. 14-2) for metastasis, resulting in a significant impact on patient management. EUS can identify the left adrenal gland in 97% of cases.^7,9 In a recent multicenter study, EUS–FNA of the left adrenal was performed in 31 patients with thoracic or gastrointestinal (GI) malignancies.¹³ Tissue adequate for interpretation was obtained in all patients; the median number of needle passes was 4.5 (range 1–8). No immediate complications were encountered. EUS-guided FNA confirmed malignant left adrenal involvement in 42% (13 of 31) of the patients. Patients with malignant left adrenal masses were more likely to have known cancer at another site. Patients with benign masses were more likely to have preservation of the normal sonographic appearance of the adrenal gland (“seagull” configuration) compared with those with malignant masses. The accuracy of EUS imaging based on size (≥3 cm) alone was 81%. The authors concluded EUS-guided FNA of the left adrenal gland is a minimally invasive, safe, and highly accurate method that confirms or excludes malignant adrenal involvement in patients with thoracic or GI malignancies. The complication rate of the transgastric EUS-guided FNA approach appears to be much lower than that of the transcutaneous approach, since no organs are traversed except the wall of the stomach.

EUS–FNA can sample mediastinal lymph nodes after chemotherapy and radiation to identify responders. Two studies, although with small numbers, confirm that the concept is feasible. The first study enrolled 19 patients.³ The positive predictive value, negative predictive value, sensitivity, specificity, and diagnostic accuracy of EUS–FNA for mediastinal restaging after induction chemotherapy were 100%, 67%, 75%,
100%, and 83%, respectively. Similar results were shown by Varadarajulu and colleagues.⁴⁴ Fourteen patients were enrolled in this study. Restaging by EUS suggested disease response in 7 patients and residual disease in 6; tissue yield was unsatisfactory in 1 patient. Final diagnosis on the 7 patients in whom EUS suggested N0 disease was established at surgery: EUS was true negative in 6 and false negative in 1. The diagnostic accuracy of EUS–FNA for predicting mediastinal response to preoperative therapy was 86%.⁴⁴ EUS–FNA appears to be an accurate, safe, and minimally invasive diagnostic technique for the restaging of mediastinal lymph nodes in patients with non-small-cell lung cancer.

TBNA alone is limited in that it is a blind technique and yield is related to the size and location of the lesion. In general, it has been underutilized. Reported sensitivity averages 76% (range 14%–100%) and negative predictive value (NPV) averages 71% (range 36%–100%).⁴²

Yasufuku and associates⁵⁴ demonstrated the usefulness and safety of real-time EBUS-guided TBNA in the evaluation of mediastinal and hilar adenopathy and reported their results in 70 patients in 2004. The accuracy of distinguishing benign from malignant lymph nodes was 97.1%. Herth and colleagues²⁰ entered 242 consecutive patients into a prospective study evaluating the feasibility of EBUS for TBNA guidance and demonstrated successful lymph node assessment in 86% of patients. The diagnostic yield was 71%. In a randomized trial of 200 patients, Herth et al.¹⁹ showed that EBUS guidance significantly increased the yield of TBNA in all lymph node stations except in the subcarinal region.

Recent series reporting the results of EBUS–FNA for the mediastinal lymph node staging of lung cancer are detailed in Table 14-1. With the exception of the study by Wallace et al.,⁵¹ the sensitivities and NPVs have been high. Of note is a study by Herth et al.²² evaluating the accuracy of EBUS–TBNA in 100 non-small-cell lung cancer patients with CT showing no enlarged lymph nodes (all nodes ≤1 cm in diameter). Malignancy was detected in 19 patients and missed in two. The sensitivity and NPV of EBUS–TBNA for detecting malignancy in a radiologically normal mediastinum was 92.3% and 96.3%, respectively. In a subset of patients with both a CT- and PET-negative mediastinum reported by Wallace and associates,⁵¹ the NPV was 89%.