Staging and Staging Modalities

Mediastinal lymph node metastases are present in nearly half of all patients with NSCLC. Accurate staging of NSCLC is crucial in determining treatment options because the detection of mediastinal lymph node metastasis preoperatively has therapeutic implications. In the absence of distant metastasis, the documentation of mediastinal metastasis is probably the most common deterrent to cure.^15–26 The TNM staging system used for lung cancer (see Box 31.1) designates ipsilateral peribronchial, intrapulmonary, or ipsilateral hilar lymph nodes as N1 disease and ipsilateral mediastinal and subcarinal lymph node involvement as N2 disease. Although N2 disease is potentially resectable, most patients with N2 disease receive multimodality treatment. Contralateral lymph node involvement of mediastinal or hilar nodes or either ipsilateral or contralateral scalene or supraclavicular lymph nodes is designated N3 disease, which precludes resection (Table 31.1 and Fig. 31.1, and see Box 31.1).^{12–14,26,27}

Table 31.1 Lymph Node Map Definitions

From Mountain CF, Dresler CM: Regional lymph node classification for lung cancer staging. Chest 11:1718–1723, 1997.

Fig. 31.1 Lymph node stations.

(From Mountain CF, Dresler CM: Regional lymph node stations for lung cancer staging. Chest 111:1718–1723, 1997.)

Various techniques are currently available to diagnose and stage lung cancer, including plain radiography, computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), endobronchial ultrasound (EBUS), and EUS. CT scan of the chest is the current standard by which mediastinal lymphadenopathy is detected. Generally, lymph nodes larger than or equal to 1 cm on chest CT scan are considered abnormal. A review of previously published studies reveals an accuracy of CT staging of the mediastinum of 52% to 88%.^28–38 This variation has been attributed to the wide range of correlation of lymph node size to the presence of malignant involvement. Although the general trend is increased risk for metastasis correlating with increasing lymph node size, lymph node size is not an accurate criterion for assessing risk. Problems associated with size as a criterion include the inability to differentiate inflammatory or reactive lymph nodes from malignant involvement. In one study, 37% of mediastinal lymph nodes that ranged in size from 2 to 4 cm were benign,³⁸ and 40% of enlarged nodes in another series were not cancerous.³⁹ Similarly, normal-sized lymph nodes can contain foci of cancer. McKenna and colleagues⁴⁰ found no correlation between the presence of mediastinal nodal metastases and nodal size. Metastases may be found in 21% of normal-sized nodes.⁴¹

MRI may be slightly superior to CT in the detection of mediastinal disease,⁴² and PET has been shown to be superior to CT for staging for the mediastinum.^43,44 PET does not rely on an arbitrary cutoff of size to diagnose malignant nodes but detects the increased glycolytic rate in metabolically active tumors. In a meta-analysis, PET had a sensitivity of 79% and a specificity of 91% compared with CT, which had sensitivity and specificity of 60% and 77%, for the detection of mediastinal disease.⁴³ In another meta-analysis by Toloza and colleagues,⁴⁴ the performance characteristics of CT, PET, and EUS for staging the mediastinum in NSCLC were compared. PET was more accurate than CT or EUS for detecting mediastinal metastases with a sensitivity of 84% and a specificity of 89% for PET compared with CT (sensitivity 57% and specificity 82%) and EUS (sensitivity 78% and specificity 71%). However, PET is limited for small lesions (≤1 cm), has false-negative results in tumors with low metabolic activity, and has false-positive results in benign lesions such as granulomatous disease. Although PET has a relatively high sensitivity, because of the importance and implications of staging, specificity is still too low, and pathologic staging is still generally sought.^45–47

Fritscher-Ravens and associates⁴⁸ performed a prospective comparison of CT, PET, and EUS for the detection of metastatic lymph nodes metastases in patients with lung cancer being considered for operative resection. After bronchoscopic evaluation, CT, PET, and EUS were performed to evaluate potential mediastinal involvement with bronchoscopic biopsy and cytology–proven (n = 25) or radiologically suspected (n = 8) lung cancer before surgery. Surgical histology was used as the “gold standard” and revealed NSCLC in 30 patients, neuroendocrine tumor in 1 patient, and benign disease in 2 patients. With respect to the correct prediction of mediastinal lymph node stage, the sensitivities of CT, PET, and EUS were 57%, 73%, and 94%; specificities were 74%, 83%, and 71%; and accuracies were 67%, 79%, and 82%. Results of PET could be improved when combined with CT (sensitivity 81%, specificity 94%, accuracy 88%). The specificity of EUS (71%) was improved to 100% by fine needle aspiration (FNA) cytology. The authors concluded that no single imaging method alone was conclusive in evaluating potential mediastinal involvement. They also suggested that CT may be necessary to evaluate the pretracheal region and the rest of the thorax and that PET may be valuable to detect distant metastases.

Whenever enlarged lymph nodes are seen in the mediastinum on chest CT scan, standard practice is to perform a lymph node biopsy for more accurate staging. The traditional methods for performing a lymph node biopsy are via CT or bronchoscopy or both. Bronchoscopy with FNA is commonly used to evaluate suspicious paratracheal, hilar, and subcarinal lymph nodes seen on CT.^49–52 The role of bronchoscopy in the diagnosis and staging of NSCLC is well established and has a sensitivity of approximately 60%.^53–59 Bronchoscopy is unable to access the aortopulmonary window or the inferior mediastinal nodes, however. CT-guided biopsy of the mediastinum is limited by overlying vascular and bony structures. When the lymph node status is not determined with CT or bronchoscopy or both, mediastinoscopy and in some cases limited thoracotomy are performed to clarify the disease stage.^37,60–62 However, these procedures are more invasive and require general anesthesia and inpatient recovery, increasing the time, cost, and risk of the staging process.⁶³

Endoscopic Ultrasound

The advent of EUS has made it possible to image the GI tract and surrounding extraluminal structures such as the mediastinum with precise resolution. EUS has played an increasingly important role as an accurate and safe method for staging patients with NSCLC.^64–85 With the advent of transesophageal EUS-guided FNA, suspicious posterior mediastinal lymph nodes, including aortopulmonary window, subcarinal, and inferior (below carina) paraesophageal nodes, can be sampled. Tracheal air artifact generally precludes reliable assessment of the anterior mediastinum lesions, pretracheal nodes, and upper paratracheal nodes; however, the advent of EBUS technology seems to be minimizing this limitation. The use of EUS and EBUS technology can enhance the overall accuracy for detecting mediastinal lymph node metastasis.

The results of an early pilot study evaluating the role of EUS in 17 patients with lung cancer found EUS to be very accurate at detecting mediastinal lymphadenopathy with an overall accuracy of 71% versus 41% for CT.⁸³ However, the capability of performing EUS-guided FNA was unavailable during this initial study. Sampling of suspicious lymph is essential except in N1 nodes. In the 1990s, several prospective studies evaluated the accuracy of EUS, EUS-guided FNA, and chest CT scan in detecting and staging mediastinal lymph node metastasis in patients with NSCLC based on correlation with surgical staging.^69,70,75 Gress and colleagues⁷⁵ reported a study consisting of patients with NSCLC and enlarged mediastinal lymph nodes (>1 cm) seen on chest CT scan. EUS-guided FNA was performed for suspicious contralateral posterior mediastinal or subcarinal lymph nodes. EUS criteria used to differentiate benign from malignant lymph nodes resulted in an accuracy of 84% compared with 49% for CT. The sensitivity and specificity of CT was 64% and 35%, which compared with a sensitivity and specificity of 86% and 83% for EUS. The addition of transesophageal EUS–guided FNA improved the overall accuracy of lymph node staging to 96% with a sensitivity of 93% and a specificity of 100%. The combination of CT and EUS did not improve overall accuracy above that for EUS alone for detecting lymph node involvement. However, the addition of CT aided in evaluating the extent of the lung cancer, detecting distant metastasis not seen by EUS and evaluating anterior and pretracheal nodes, which are not imaged by EUS. EUS was best at accurately detecting mediastinal lymph node metastasis in the aortopulmonary window (station 5), subcarinal (station 7), and paraesophageal (station 8) regions (see Fig. 31.1).

These findings are similar to the studies reported by Giovannini and colleagues⁷³ and Silvestri and coworkers,⁷⁴ who reported sensitivities of 81% and 89% and specificities of 100% each. A recent meta-analysis by Micames and colleagus⁶⁷ showed a pooled sensitivity of 83% and a pooled specificity of 97% for EUS-FNA staging of NSCLC.

Although still not approaching the “gold standard” of mediastinoscopy and lymph node biopsy, EUS-FNA has greatly improved minimally invasive staging of NSCLC. As noted earlier, where EUS fails is in the anterior mediastinum where tracheal air artifact generally precludes reliable assessment of the anterior mediastinum lesions, pretracheal nodes, and upper paratracheal nodes. In an attempt to view the anterior mediastinum, radial EBUS was initially developed in the 1990s. There was limited use of the technology, however, because it was impossible to perform real-time guided FNA of lesions. In more recent years, convex, linear probe technology has been developed and now allows for real-time EBUS-guided transbronchial needle aspiration (TBNA). A systematic review of EBUS-TBNA showed that it was a safe and effective modality to aid in the staging of NSCLC with a sensitivity ranging from 85% to 100% without reported complications.⁶⁸ Gu and colleagues⁶⁹ performed a meta-analysis of studies aimed at measuring EBUS-TBNA accuracy and found a 93% sensitivity and 100% specificity in detecting mediastinal lymph node metastasis. Given the high sensitivity and specificity of both EUS-FNA and EBUS-FNA, many investigators have been seeking to perform a complete, “medical mediastinoscopy,” with the hope of avoiding the more traditional, invasive staging, including mediastinoscopy, of NSCLC before operative intervention.

The largest study to date was performed by Wallace and colleagues,⁷⁰ who compared the use of TBNA, EUS-FNA, and EBUS-TBNA using pathologic confirmation of malignancy or benign disease as the diagnostic standard. This group found EBUS-FNA to be more sensitive, with a higher negative predictive value than TBNA. More importantly, they found the combination of EBUS-FNA with EUS-FNA to have a sensitivity of 93%, a positive predictive value of 100%, and a negative predictive value of 97%. These results suggest that EBUS-FNA may complement EUS-FNA and possibly could provide near-complete, minimally invasive staging of the mediastinum in patients with NSCLC. These combined modalities may eventually eliminate the need for mediastinoscopy or other invasive surgical procedures for staging purposes.

Endoscopic Ultrasound Technique for Imaging the Mediastinum

After informed consent is obtained, the patient is placed in the left lateral decubitus position. Conscious sedation is administered, and the echoendoscope is passed to the gastroesophageal (GE) junction in a manner similar to the passage of a duodenoscope. Many endosonographers first perform staging with a radial scanning echoendoscope and then switch to a dedicated biopsy echoendoscope or a curved linear array echoendoscope if FNA is to be performed.

When imaging the mediastinum, a thorough understanding of mediastinal anatomy relative to the esophagus is essential to perform a complete examination. Scanning should begin distally below the GE junction while withdrawing the scope proximally because this limits scanning artifacts. It is useful to press the transducer with its balloon partially inflated against the gut wall to minimize air artifacts and help to anchor the transducer. In addition, the suction port is depressed to maintain optimal imaging by minimizing intraluminal air.

As the endosonographer withdraws the radial scanning echoendoscope, the aorta should be maintained in the 5 o’clock to 6 o’clock position in the ultrasound field, which generally allows proper orientation of the paraesophageal structures. The aorta is easily recognized as a circular anechoic structure, approximately 1.5 to 2 cm in diameter, with a relatively bright border resulting from back wall enhancement (a normal artifact seen in vessels). As the echoendoscope is withdrawn into the distal esophagus, the aorta, spine, left lobe of the liver, inferior vena cava, and heart can be seen. The spine is also easily identified in the 7 o’clock position next to the aorta and has irregular echo features with artifacts produced by poor penetration of echoes through bony structures. The left lobe of the liver appears at the 6 o’clock to 12-o’clock position, and often the hepatic veins and inferior vena cava can be seen as they course through the liver. Slightly more proximally, the beating of the heart is appreciated as the left atrium comes into view at the 12 o’clock position. The mitral valve leaflets can be seen as the valve opens from the left atrium into the left ventricle, and the pulmonary veins can be seen entering the left atrium. The left pulmonary artery arches posteriorly to the left of the ascending aorta and tends to be easier to view than the right pulmonary artery, which can be seen just below the carina. The left ventricle, right atrium, and right ventricle lie deep to the left atrium; it can be more difficult to visualize these structures completely. As the aorta moves toward the left, the spine and azygos vein are seen posteriorly. The aortic outflow tract can be appreciated as the endoscope is withdrawn further. The spine continues to be a useful landmark because it consistently appears as a hyperechoic structure located posteriorly throughout the chest. Careful inspection of this area may reveal the thoracic duct adjacent to the aorta and spine.

The right lung appears as hyperechoic rings emanating from the 9 o’clock position, whereas the left lung appears at the 2 o’clock position. In the midesophagus, the right and left bronchi are easily demarcated by the hyperechoic rings (echogenic air) seen at the 11 o’clock and 1 o’clock positions. The two bronchi join together to form the trachea normally at 27 to 28 cm from the incisors. The azygos vein can be seen coming into position to the right of the aorta and moves anterior to the spine and toward the right lung. As the endoscope is withdrawn further, the azygos vein can be seen to move forward and extend anteriorly into the superior vena cava. The ascending aorta can be difficult to trace because this structure runs deep to the hilar structures (pulmonary vessels), and because of air within the bronchi and trachea, the ascending aorta is often not fully imaged. In the proximal esophagus, the aortic arch is identified on the left and moves rightward and anteriorly across the screen. In the cervical esophagus, above the level of the aortic arch, the carotid vessels and, occasionally, the thyroid gland can be seen (Figs. 31.2, 31.3, and 31.4).

Fig. 31.2 A, Radial imaging at the distal esophagus showing the liver, inferior vena cava (IVC), and aorta (AO). The spine lies immediately deep to the aorta. B, Imaging slightly higher in the distal esophagus showing the right lower lobe of the lung (RLL), left lung (LL), aorta (AO), spine (SP), and a portion of the left atrium (LA).

Fig. 31.3 Radial endoscopic ultrasound (EUS) from the distal esophagus. The left atrium (LA), right atrium (RA), right ventricle (RV), and left ventricle (LV) can be seen. The leaflets of the mitral valve and the base of the aortic outflow tract (AOFT) are also visible.

Fig. 31.4 Radial endoscopic ultrasound (EUS) from the most proximal aspect of the esophagus. The left common carotid artery (LCA), left internal jugular vein (LIJ), right internal jugular vein (RIV), right common carotid artery (RCA) are seen. In addition, the thyroid gland is seen on either side of the trachea (TR).

Evaluation of the mediastinum using linear EUS requires rotation of the echoendoscope every few centimeters for a thorough evaluation. As in radial EUS, vascular structures provide the major landmarks for orientation, and the home base structure is the descending aorta, which is first located approximately 35 cm from the incisors. The echoendoscope is rotated initially clockwise (right) bringing structures anterior to the esophagus into view and then counterclockwise (left) bringing posterior structures into view. The left atrium is found by rotating the shaft of the scope 180 degrees in the distal esophagus to midesophagus until a large, echolucent structure is seen within which the mitral valve leaflets are located. By tipping the scope upward and with slight withdrawal, the subcarinal lymph node station is located immediately beneath the endoscope at approximately 27 cm between the left atrium and right pulmonary artery. The aortopulmonary window is located by following the descending aorta cephalad to the arch and pushing the endoscope in again about 2 cm. The endoscope is turned 90 degrees clockwise and tipped up slightly until a cross-sectional view of the aortic arch and the more distally located left pulmonary artery are seen. The area between these structures is known as the “AP window.” Another potentially important area for FNA of lymph nodes is the celiac axis. This area is located by finding the abdominal aorta at the level of the GE junction and the takeoff of the celiac artery with the superior mesenteric artery just distal to this (Figs. 31.5 and 31.8).

Fig. 31.5 A, Linear array imaging of the celiac axis is depicted showing the takeoff of the superior mesenteric artery (SMA) and celiac axis (CX) along with the hepatic artery (HA) and splenic artery (SA) branches. B, More typical linear images as seen from the proximal stomach. C, Radial imaging showing the celiac axis.

Fig. 31.6 A, Curved linear array view through the midesophagus of the subcarinal (SC) region. The left atrium (LA) is next to the right pulmonary artery (PA) with the ascending aorta lying deep to these structures. B, Linear array view through the midesophagus of the aortic arch (AA) and the pulmonary artery (PA)—the “AP window.” C, Radial imaging of the area of the AP window.

Fig. 31.7 A, Mediastinal lymph node (LN) as imaged with the linear array endoscopic ultrasound (EUS) system. B, The fine needle aspiration (FNA) needle is seen exiting the scope; the tip of the needle is seen to be in the center of the LN. C, High-power magnification of hematoxylin and eosin stain on cell block reveals metastatic adenocarcinoma. D, High-power magnification of immunoperoxidase stain on cell block. Tumor nuclei (arrows) are positive for TTF1 antibody consistent with non–small cell lung carcinoma.

Fig. 31.8 Linear endoscopic ultrasound (EUS) reveals a hypoechoic mass within the mediastinum. EUS-guided fine needle aspiration (FNA) was performed and revealed metastatic non–small cell lung cancer (cytology favored a large cell neuroendocrine-type lesion).

Mediastinal Lymph Nodes

EUS-guided FNA became available after the development of a specially designed needle catheter system consisting of a 4-cm, 23-gauge needle attached to a 180-cm long, 5-Fr aspiration catheter (Echo-Tip; Wilson-Cook Medical, Inc, Winston-Salem, NC) and the 22-gauge, 10-cm needle (GIP, Medi-Globe, Inc, Tempe, AZ). At the present time, several EUS catheter–FNA needle systems are available with various needle lengths (up to 14 cm) and gauges (19-gauge to 25-gauge) available. We routinely perform EUS-guided FNA on all posterior mediastinal lymph nodes that are suspicious for malignant involvement by clinical suspicion or EUS criteria or both. Many patients have more than one suspicious lymph node or finding. In our patients, we sample only the most suspicious lymph node or finding that would have the greatest impact on the clinical staging (i.e., contralateral or subcarinal).

The EUS-guided FNA biopsy technique was initially developed for use with the linear array instrument and has been described elsewhere.^{95–97,102,103} The unique viewing angle of the linear array transducer allows for observation of the needle as it exits the biopsy channel and enables direction of the needle tip into the target lesion. A similar technique using a radial scanning echoendoscope has been reported; however, serious complications have been described via this technique, and it is not recommended.^94,103

EUS-guided FNA involves the insertion of the FNA catheter device through the accessory channel of the echoendoscope followed by deployment of the needle under EUS guidance into the lymph node to be sampled. The handle mechanism is secured to the accessory port, and if the instrument has an elevator, the elevator should be fully released into the down position to allow easy passage of the needle. The elevator can be used during the biopsy to direct the needle gently into the lesion. Doppler is used to identify surrounding vascular structures. The FNA needle is slowly advanced toward the target lesion. With certain needles, it helps if the stylet is withdrawn a few millimeters (2 to 3 mm), and the needle and the stylet are then directed into the target. When the needle has entered the lesion, the stylet is advanced (to clear the needle) and then removed. The endosonographer or assistant applies suction to the catheter system using a 5-mL or 10-mL Luer-Lok syringe. Suction is followed by “in-and-out” movements of the catheter after firmly locking the needle-catheter system to the appropriate depth so that the needle is not advanced beyond a desired depth. Typically, we make 7 to 10 gradual in-and-out movements within the lesion. Before removing the needle, the negative pressure is released slowly, the needle is removed from the lesion, and subsequently the needle system is unscrewed from the echoendoscope. It has been suggested to perform EUS-guided FNA of lymph nodes without the use of suction because suction may result in a bloody sample that may be more difficult for the cytopathologist to examine.¹⁰⁰

Preliminary cytology findings are obtained immediately during the FNA procedure by a cytopathologist or cytotechnologist present during the study. We recommend having a cytopathologist or cytotechnologist present during the EUS-guided FNA portion of the procedure because it can improve the efficiency of the technique. If a cytopathologist or cytotechnologist is unavailable, two to three passes should be taken by the endosonographer for lymph nodes (or liver metastases) and five to six passes should be taken for masses (similar to pancreatic masses) to ensure adequate cellularity in more than 90% of cases.^100,104 However, this approach is associated with a 10% reduction in definitive cytologic diagnoses, increased time and risk, and potentially the need for additional needles.¹⁰⁴