INTRODUCTION — The prognosis of esophageal cancer is strongly associated with its stage, and the choice of initial treatment is highly dependent on clinical stage at diagnosis. As a result, accurate clinical staging is critical for selecting appropriate treatment options. Endoscopic ultrasonography (EUS) has a central role in the initial staging of esophageal carcinoma and may also be useful for detecting disease recurrence.
This topic review will focus on the role of EUS in the care of patients with esophageal cancer. An overview of the diagnosis and staging of esophageal cancer is presented separately. (See "Clinical manifestations, diagnosis, and staging of esophageal cancer".)
PREOPERATIVE STAGING — Initial evaluation of the patient diagnosed with esophageal cancer centers on the assessment of operative risk and tumor stage. Comorbid conditions may preclude a patient with a potentially resectable tumor from undergoing surgery. Preoperative tumor staging is warranted in patients who are considered to be potential surgical candidates because their disease extent will influence treatment planning (eg, it may reveal that a patient is a candidate for endoscopic resection). (See "Surgical management of resectable esophageal and esophagogastric junction cancers" and "Management of superficial esophageal cancer", section on 'Endoscopic resection'.)
Furthermore, initial (neoadjuvant) chemotherapy with or without radiotherapy may be recommended rather than upfront surgery for those with locally advanced (ie, T3 and/or node-positive) tumors. (See "Radiation therapy, chemoradiotherapy, neoadjuvant approaches, and postoperative adjuvant therapy for localized cancers of the esophagus" and "Multimodality approaches to potentially resectable esophagogastric junction and gastric cardia adenocarcinomas".)
Staging usually begins with a contrast-enhanced computed tomography (CT) scan of the neck, chest, and abdomen to evaluate for the presence of metastatic disease. However, with increasing availability, integrated fluorodeoxyglucose positron emission tomography (FDG-PET)/CT fusion scanning may be more accurate for the detection of stage IV disease and can be used as an initial staging assessment. (See "Clinical manifestations, diagnosis, and staging of esophageal cancer", section on 'Evaluation for distant metastases'.)
A more detailed evaluation of locoregional disease extent (T and N stage) should be obtained if distant metastases are not demonstrated. A number of studies have demonstrated that EUS is more accurate than transabdominal ultrasound, CT scan, magnetic resonance imaging (MRI), or PET scanning for locoregional staging of esophageal cancer [1-17], with an overall accuracy of EUS for T and N staging of 90 percent [18].
Overall, study results will be influenced by the number of early versus advanced T-stage patients as EUS is more accurate in the latter population. A strong correlation exists between tumor differentiation, T-stage and lymph node status [19]. A meta-analysis of diagnostic test characteristics made the following estimates for EUS, CT, and FDG-PET, concluding that the three approaches were complementary [16]:
●For EUS, the sensitivity and specificity for detecting celiac lymph node metastases (considered regional lymph nodes in the staging classification, regardless of primary tumor location) were 85 percent (95% CI 72-99 percent) and 96 percent (95% CI 92-100 percent), respectively. Sensitivity and specificity for other regional lymph node metastases were 80 percent (95% CI 75-84 percent) and 70 percent (95% CI 65-75 percent), respectively.
●For CT, sensitivity and specificity for regional lymph node metastases were 50 percent (95% CI 41-60 percent) and 83 percent (95% CI 77-89 percent), respectively. For abdominal lymph node metastases (ie, nonregional, nonceliac abdominal nodes), these values were 42 percent (95% CI 29-54 percent) and 93 percent (95% CI 86-100 percent), respectively. For distant metastases, these values were 52 percent (95% CI 33-71 percent) and 91 percent (95% CI 86-96 percent), respectively.
●For FDG-PET, sensitivity and specificity for regional lymph node metastases were 57 percent (95% CI 43-70 percent) and 85 percent (95% CI 76-95 percent), respectively. For distant metastases, these values were 71 percent (95% CI 62-79 percent) and 93 percent (95% CI 89-97 percent), respectively.
The tumor, node, metastasis staging system of the American Joint Committee on Cancer and the Union for International Cancer Control for esophageal cancer (adenocarcinoma and squamous cell carcinoma) is used universally (2017, eighth edition) (table 1A and table 1B) and has a slight difference in the definitions of T4a disease compared with the prior version [20,21].
In general, the endosonography report should include the endoscopic findings of tumor location, features (eg, circumferential extent, skip areas, presence/absence of Barrett’s), and anatomic landmarks (gastroesophageal junction, diaphragmatic hiatus, squamocolumnar junction), as well as a description incorporating the T-stage including maximal wall thickness, N-stage (as per 2017 guidelines using N0 - N3), and specific features of identified lymph nodes (location, shape/size/border/echogenicity) [22]. Incomplete staging due to tumoral stenosis should be described when present.
Histologic correlates of the endoscopic ultrasound image — Dedicated echoendoscopes operating at frequencies of 7.5 and 12 MHz are able to visualize the esophageal wall as a five-layered structure (image 1 and figure 1):
●First hyperechoic layer: superficial mucosa
●Second hypoechoic layer: deep mucosa
●Third hyperechoic layer: submucosa
●Fourth hypoechoic layer: muscularis propria
●Fifth hyperechoic layer: adventitia
The ability to visualize the five layers permits a detailed understanding of the degree of tumor infiltration into the wall layers, providing an assessment of the tumor stage (T stage) (table 2 and image 2A-D) [23]. However, one limitation of the standard dedicated echoendoscopes is that they operate at relatively low frequency and, as a result, cannot visualize the muscularis mucosa [24,25]. High frequency miniprobes (20 MHz) (picture 1) provide a more detailed visualization, permitting delineation of nine layers in the esophageal wall [24,25]:
●First and second layer: superficial mucosa (hyper and hypoechoic respectively).
●Third layer: lamina propria (hyperechoic).
●Fourth layer: muscularis mucosa (hypoechoic).
●Fifth layer: submucosa (hyperechoic).
●Sixth, seventh, and eighth layer: (hypo, hyper, and hypoechoic respectively) inner circular muscle and outer longitudinal muscle of the muscularis propria with intermuscular connective tissue.
●Ninth layer: adventitia (hyperechoic).
This may have particular importance when evaluating superficial lesions for which nonsurgical therapy (such as endoscopic mucosal resection or photodynamic therapy) is being considered (see below).
EUS for T staging of superficial tumors — Early esophageal cancers are those that are classified as Tis (high-grade dysplasia, which includes all noninvasive neoplastic epithelial that was formerly called carcinoma in situ) or T1 tumors, which are split into T1a and T1b subcategories depending on the depth of invasion (table 1A and table 1B). These are referred to as superficial esophageal cancers. The risk of nodal metastases is higher for T1b than T1a tumors, and within these subcategories, risk may be further stratified according to depth of invasion and the presence of lymphovascular invasion (LVI). (See "Management of superficial esophageal cancer", section on 'Pathologic subclassification and the risk of nodal metastases'.)
The role of EUS for T staging and subsequent management of superficial tumors has been controversial. If the EUS identifies only mucosal (T1a) disease, endoscopic mucosal resection is usually the next step to remove the tumor and precisely define the depth of invasion. The pathology result from the endoscopic mucosal resection (particularly the presence or absence of LVI) can then be used to guide the decision as to whether endoscopic therapy alone is sufficient or if surgery should be recommended. On the other hand, if the EUS identifies esophageal cancer that invades beyond the muscularis mucosa (T1b), or if there is evidence of lymph node involvement, then surgical therapy is frequently recommended. (See "Surgical management of resectable esophageal and esophagogastric junction cancers" and "Radiation therapy, chemoradiotherapy, neoadjuvant approaches, and postoperative adjuvant therapy for localized cancers of the esophagus" and "Management of superficial esophageal cancer".)
However, this is a controversial area and some have suggested that EUS should not be used to determine which patients should go to surgery, and that instead, patients should undergo a careful endoscopic examination and endoscopic mucosal resection without EUS [26]. Studies on the accuracy of EUS for staging superficial esophageal cancer are discussed separately. (See "Management of superficial esophageal cancer", section on 'Accuracy of EUS in the staging evaluation'.)
Although T1 lesions can be identified using echoendoscopes operating at 7.5 and 12 MHz, distinguishing subgroups of the T1 stage (muscularis mucosa and/or submucosal invasion) (figure 2) can be problematic with these instruments because the frequencies are too low to visualize the muscularis mucosa. High-frequency catheters permit determination of the tumor extension into the muscularis mucosa with an accuracy as high as 84 percent when histology is used as the gold standard [25]. (See "Clinical manifestations, diagnosis, and staging of esophageal cancer".)
Despite these advantages, high-frequency catheters have some drawbacks. Their limited depth of penetration into surrounding tissues (approximately 3 cm) precludes adequate assessment of the N stage [27]. As an example, in one report of 54 patients undergoing preoperative EUS, the accuracy of N staging in traversable tumors was significantly worse with the high-frequency linear catheter than with the conventional radial-scanning echoendoscope (48 versus 90 percent) [28].
Obtaining an adequate acoustic coupling with the esophageal wall can also be problematic [27]. The technique most commonly used to facilitate these examinations consists of attaching a latex condom to the distal end of the endoscope and then instilling water through the biopsy channel of the scope. This provides a fluid bath through which the catheter can image the infiltration of the tumor into the different layers [29]. A balloon sheath (Olympus UM-BS20-26R) placed over the ultrasound catheter (Olympus UM-2R and UM-3R) has been specifically developed for this use [30]. Combined use of the balloon-sheathed catheter with a dual lumen gastroscope provides an efficient method of accurately staging patients with early esophageal carcinoma [31].
EUS for staging locally advanced tumors — The large caliber of dedicated radial echoendoscopes (12.7 mm in diameter) may preclude complete EUS staging in patients with locally advanced esophageal tumors, which may be associated with esophageal stenosis. This was illustrated in a series of 113 patients who underwent esophagectomy in whom the accuracy of EUS for T and N staging was much higher for traversable (81 and 86 percent, respectively) as compared with nontraversable tumors (28 and 72 percent, respectively) [32].
Several options are available to improve accuracy in these settings:
●Dilatation of the lumen to a diameter of 14 to 16 mm with either a Savary dilator or a controlled radial expansion (CRE) balloon may permit traversal of stenotic lesions in many cases [32]. However, a perforation frequency of up to 24 percent has been described with this practice [33,34]. The high perforation rate may in part be due to the blunt tips of the older instruments used in the reports cited above [33,34]. Echoendoscopes that have been developed subsequently appear to reduce the risk of perforation [32,33]. Patients with severely stenotic tumors may be best served by serial, progressive dilation over several days rather than a single dilation when attempting to gain access across the tumor for purposes of staging. In one report, Savary dilation to 14 to 16 mm permitted passage of the echoendoscope in 85 to 100 percent of patients without complications [35]; two additional dilators of incremental diameter were passed once resistance was encountered, and the dilation was limited to <13 mm in one-third of patients [35].
●Ultrasound catheters introduced through the biopsy channel of the endoscope can traverse tight strictures due to their small caliber (3 mm in maximum diameter). Although this may improve the accuracy of T and N staging [36], limited depth of penetration can lead to incomplete assessment of locoregional spread.
In summary, these techniques permit complete examination of stenotic tumors in most patients. The additional information that may be obtained must be balanced by the risk of perforation when dilation is undertaken. A direct comparison of these techniques to determine their comparative accuracy has not been performed. Our practice is to perform endoscopy with a standard gastroscope initially at the time of EUS. If the gastroscope cannot traverse the stenosis, we do not dilate the lesion, as we infrequently are able to pass the echoendoscope even with this intervention. In patients with circumferential stenosis permitting passage of a gastroscope but not the echoendoscope, judicious dilation is undertaken. We are particularly cautious when semi-circumferential involvement is present, since the normal (and hence thinner) esophageal wall may be at increased risk of tearing in this setting, particularly if the proximal esophagus is dilated.
Accuracy of EUS for determining unresectability — The definition of a T4 tumor has been refined in 2017 guidelines. In the staging classification, tumors staged as T4b are considered unresectable, while advanced tumors (T4a) at EUS typically results in chemoradiation (in the absence of distant metastases) followed by restaging and surgical resection if possible [37]. In the guidelines, T4a tumors include tumors involving the pleura, pericardium, diaphragm for both SCC and adenocarcinomas (table 1A-B), and for adenocarcinomas of the distal esophagus and esophagogastric junction (EGJ), they include tumors invading the azygos vein, diaphragm, or peritoneum [38].
Unresectability is also suggested by EUS in patients with tumors arising above the level of the carina, with extension through the esophageal wall into the mediastinum. It may be difficult or impossible to obtain a safe margin of resection between the tumor and the trachea and/or bifurcation in such patients [39].
EUS for preoperative lymph node staging — Endosonographic criteria that are suggestive of malignant involvement of visible lymph nodes include a width greater than 10 mm, round shape, smooth border, and echo-poor pattern (table 3 and image 3) [40,41]. Of these, an echo-poor pattern and width >10 mm appear to be the most specific for malignancy. When all four suspicious features are present in a visualized lymph node, there is an 80 to 100 percent chance of metastatic involvement [40,41]. However, only 25 percent of malignant nodes will have all of these features. These results demonstrate the limitations of EUS criteria for preoperative determination of lymph node staging, especially if the nodes are small.
The lymph node stage is further categorized according to the number of involved nodes (no involved nodes (N0), 1 or 2 (N1), 3 to 6 (N2), and 7 or more (N3)) involved lymph nodes, but not the location of the nodes. However, most EUS/CT/PET staging studies have not included this refinement when reporting N staging accuracy.
As noted above, EUS has an accuracy of more than 80 percent for detecting malignant nodes in the cervical paraesophageal, right recurrent laryngeal, left paratracheal, upper and lower paraesophageal, infra-aortic, infracarinal, lower posterior mediastinal, and perigastric regions. Greater numbers of malignant-appearing periesophageal lymph nodes detected by EUS predicted worse survival [42,43]. (See "Clinical manifestations, diagnosis, and staging of esophageal cancer".)
The demonstration of clinically suspicious lymph nodes may support the selection of induction chemoradiotherapy over surgery alone, particularly in a patient with T2 disease [44]. (See "Radiation therapy, chemoradiotherapy, neoadjuvant approaches, and postoperative adjuvant therapy for localized cancers of the esophagus".)
Endoscopic ultrasound-guided fine-needle aspiration biopsy — Endoscopic ultrasound-guided fine-needle aspiration (EUS FNA) may improve the accuracy of N staging by providing cytologic confirmation of metastatic disease from accessible nodes, as long as the primary tumor is not in the pathway of the aspiration needle. EUS FNA cytology results may not be reliable if the needle has to traverse the primary tumor.
Sensitivity, specificity, and accuracy of EUS FNA for locoregional lymph nodes are all over 85 percent when surgical resection specimen or cytology results are considered as the gold standard [45-47]. The accuracy of EUS FNA for esophageal cancer staging appears to be similar to that reported for other gastrointestinal malignancies (table 4) [48-51].
EUS FNA appears to improve the accuracy of nodal staging beyond that achieved by EUS alone. One study from our institution compared lymph node staging using EUS alone in a historical cohort of 33 patients versus 31 patients who underwent EUS FNA of non-peritumoral lymph nodes for preoperative lymph node staging [50]. Compared with EUS alone, EUS FNA was associated with significantly better sensitivity (93 versus 63 percent) and accuracy (93 versus 70 percent). A comparison of specificity was limited in this study, since only one patient underwent a biopsy for which benign pathology was reported. EUS FNA was less costly than CT FNA and surgery. The cost-saving was principally a result of a reduction in unnecessary surgery [52].
These results were supported by a prospective study conducted at our institution that compared the performance characteristics of CT, EUS, and EUS FNA for preoperative lymph node staging of esophageal carcinoma in 125 patients [53]. EUS FNA was more sensitive than CT (83 versus 29 percent) and more accurate than CT (87 versus 51 percent) or EUS (87 versus 74 percent) for nodal staging (table 5). Direct surgical resection was contraindicated in 77 percent of patients due to detection of advanced locoregional/metastatic disease.
Optimal criteria for identifying malignant lymph nodes based upon EUS criteria and for helping to select patients for whom EUS FNA is required continue to evolve. Initial experience suggests that a selective approach based upon modified EUS criteria was cost-saving compared with an approach of routine FNA [54]. Furthermore, the modified EUS criteria (four standard criteria plus EUS-identified celiac lymph nodes, >5 lymph nodes, or EUS T3/4 tumor) were more accurate than standard criteria (hypoechoic, smooth border, round, or width >5 mm) at identifying malignant lymph nodes.
Interobserver variation and EUS learning curve — The accuracy of EUS is operator-dependent [55]. The available evidence suggests that interobserver reliability (different observers staging the same patients) is influenced by experience and tumor stage [18,56-59]. As a general rule, there is good agreement among experienced endosonographers (defined as >50 EUS examinations in patients with esophageal cancer) for both T and N stage, except for T2 tumors in which agreement was poor. Interobserver reliability was lower when inexperienced endosonographers (<20 EUS examinations in patients with esophageal cancer) were tested. Technical factors, including balloon overinflation causing a blurring of the distinction between the esophageal wall layers, tangential imaging, and inadequate use of higher scanning frequencies (12 MHz), may be responsible for staging errors among inexperienced endosonographers [57,59,60].
Among expert endosonographers (>75 esophageal cancer examinations) [61], overstaging of esophageal carcinomas is more frequent than understaging, occurring in 8 to 14 percent of patients [62]. This is particularly evident in T2 lesions and may be attributed to peritumoral inflammation leading to an overestimation of mural penetration [62]. In contrast, understaging occurs in 3 to 15 percent of cases and is frequently associated with T3 tumors with microscopic infiltration of the adventitia that is beyond the resolution capabilities of the available echoendoscopes [8,59].
RESTAGING AFTER NEOADJUVANT THERAPY — One of the most controversial areas in oncology has been the optimal treatment of potentially resectable esophageal cancer, in particular, the benefit of multimodality therapy (ie, preoperative chemoradiotherapy or perioperative chemotherapy alone over resection alone). However, clinical practice is evolving toward initial (neoadjuvant) therapy rather than resection for tumors that are T3 and/or node-positive at presentation. (See "Radiation therapy, chemoradiotherapy, neoadjuvant approaches, and postoperative adjuvant therapy for localized cancers of the esophagus" and "Multimodality approaches to potentially resectable esophagogastric junction and gastric cardia adenocarcinomas".)
For patients who undergo preoperative chemoradiotherapy or chemotherapy alone, it is frequently desirable to evaluate the response to treatment prior to recommending tumor resection. Unfortunately, the accuracy of endoscopic ultrasound (EUS) is limited in this setting. As an example, in a meta-analysis involving 16 studies with 724 patients, the pooled sensitivity and specificity of EUS in staging esophageal cancer after neoadjuvant chemotherapy varied with tumor stage (table 2) [63]. The sensitivity was relatively high in T3 disease (81 percent) while specificity was high in other T stages (T1, T2, and T4). Results did not appreciably change between the periods of 1991 to 2000 and 2001 to 2013 despite introduction of new technology. A possible explanation for the discrepancy between endoscopic and subsequent pathologic staging is that EUS may not be able to differentiate between post-treatment inflammation/fibrosis and residual tumor.
DETECTION OF LOCOREGIONAL RECURRENCE — Patients who present with symptoms or signs worrisome for locoregional recurrence may have an initially negative endoscopic and radiographic evaluation. In this setting, endoscopic ultrasound (EUS) is extremely accurate (sensitivity >92 percent and specificity >96 percent) for detecting locoregional relapse, and it should be considered in the work-up of such patients [64-66].
The benefit of EUS for surveillance of resected patients has also been evaluated. In one series of 45 patients who had undergone resection for localized esophageal carcinoma, EUS examination was performed every six months for a period of two years [65]. The positive predictive value of an abnormal EUS for tumor recurrence was 92 percent, and two-thirds of patients who had tumor relapse documented by EUS were asymptomatic at the time of diagnosis. Although EUS may be an important tool for postoperative surveillance, it is unclear if early detection of tumor recurrence improves survival. As a result, EUS is not included as a routine component of the posttreatment surveillance strategy. (See "Radiation therapy, chemoradiotherapy, neoadjuvant approaches, and postoperative adjuvant therapy for localized cancers of the esophagus", section on 'Surveillance strategy' and "Multimodality approaches to potentially resectable esophagogastric junction and gastric cardia adenocarcinomas", section on 'Post-treatment surveillance'.)
On the other hand, for patients with signs or symptoms suspicious for recurrence, EUS with FNA should be performed to establish a diagnosis.
SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Esophageal cancer".)
SUMMARY AND RECOMMENDATIONS
●The prognosis of esophageal cancer is strongly associated with its stage, and the choice of initial treatment is highly dependent on clinical stage at diagnosis. As a result, accurate clinical staging is critical for selecting appropriate treatment options. (See 'Introduction' above.)
●Endoscopic ultrasonography (EUS) is the most accurate method for locoregional staging of esophageal cancer. EUS should be performed to assess T and N stage in patients being considered for surgery once distant metastases have been excluded by computed tomography and/or integrated fluorodeoxyglucose positron emission tomography (FDG/PET)/CT scan. The addition of EUS-guided fine-needle aspiration has further improved lymph node staging accuracy and should be performed routinely when confirmation of metastatic lymphadenopathy will alter the approach to treatment. (See 'Preoperative staging' above.)
●Limitations caused by locally advanced, stenotic tumors that preclude echoendoscope passage and, therefore, complete staging, are being overcome by smaller caliber instruments. Reports comparing EUS with PET support the superior performance of EUS in locoregional staging. Although EUS has a limited role in restaging patients post chemotherapy and/or radiation therapy, it is the most sensitive technique for detecting locoregional tumor recurrence. (See 'Restaging after neoadjuvant therapy' above and 'Detection of locoregional recurrence' above.)
●The clinical manifestations, diagnosis, and staging of esophageal cancer are presented in more detail separately. (See "Clinical manifestations, diagnosis, and staging of esophageal cancer".)
ACKNOWLEDGMENT — The author and UpToDate thank Dr. Enrique Vazquez-Sequeiros, who contributed to earlier versions of this topic review.
