CT of the chest and upper abdomen can be utilized to evaluate the primary tumor and identify any enlarged regional lymph nodes or distant metastatic disease. For T staging, CT is not able to consistently determine the depth of primary tumor invasion [15, 16]. For regional lymph node involvement, the overall accuracy of CT is approximately 50 to 70 % [17]. Because size is the criterion used to define malignant nodes on CT, CT is inherently unable to detect metastases in normal-sized nodes and may identify enlarged, reactive nodes as malignant, leading to false positives. It also has poor sensitivity for celiac axis node involvement [18]. For distant metastases, CT has a sensitivity of 40 to 80 % and a specificity of 25 to 70 % [19]. However, it has limited sensitivity for identifying small metastases, especially within the peritoneum [18, 19].

PET scans may also be used to identify a metabolically active primary tumor and visualize metabolically active lymph nodes or metastatic foci. However, PET is not consistently sensitive in identifying the primary tumor [20–22]. Because of its reduced spatial resolution, PET has also limited ability to differentiate depth of tumor invasion. The sensitivity of PET for lymph node disease is only equivalent to CT, with a pooled sensitivity from 43 to 70 % compared to 41–60 % for CT. Because PET can detect disease in normal-sized lymph nodes, and may be able to differentiate between inflammatory and malignant activity, it has better specificity for N staging than CT, with a pooled specificity from 76 to 95 % compared to 77–89 % [18, 20, 21, 23–25]. PET is particularly limited when assessing nodal disease in the area of the primary tumor due to poor spatial resolution. The greatest utility of PET is in detecting M stage , as it is more sensitive for the detection of unsuspected metastatic disease than CT [21, 22, 24, 26–28]. Therefore, patients with no evident distant disease on CT are typically recommended to undergo PET to assess for metastatic foci. PET scans also offer the opportunity to identify synchronous neoplasms. Care must be taken to investigate areas of uptake that may be synchronous neoplasms, rather than metastases, to avoiding upstaging the primary tumor [29].

The combination of PET/CT is likely superior to CT alone in detecting nodal disease. A recent study suggested significantly higher specificity and positive predictive value for detecting lymph nodes with PET/CT (97 % versus 94 % and 65 % versus 44 %, respectively) [30]. Similarly, PET/CT appears more sensitive than PET alone for nodal disease (46 % versus 33 %) [31].

In patients without evident metastatic disease on CT or PET, EUS plays an important role in the staging of esophageal cancer because of its utility in determining the depth of tumor invasion (T status) and regional lymph node metastases (N status).

Utilizing a 7.5- or 12-MHz echoendoscope, EUS offers the ability to assess the relationship of an esophageal mass to the five different layers of the esophageal wall, as shown in Table 25.4 and Fig. 25.2. Regarding terminology, hyperechoic refers to structures appearing bright (light gray to white like fat and bone), hypoechoic is dark gray and darker than surrounding structures like muscle, and anechoic is black like fluid. It is easy to use a radial echoendoscope for luminal cancer staging , as it provides a 360 degree view and requires less torquing to assess the esophageal wall layers and mediastinal structures. However, utilizing a linear echoendoscope for staging permits the endoscopist to perform fine-needle aspiration without changing echoendoscopes during the procedure. A randomized study compared the radial and linear echoendoscopes for esophageal cancer staging and found 88 % agreement for T staging with more lymph nodes detected using the radial echoendoscope (p = 0.009) [32]. Therefore, this may favor the use of the radial echoendoscope, given that the 2010 TNM staging system now incorporates the number of malignant lymph nodes into N staging. Whichever echoendoscope is utilized, staging should be performed in a systematic manner beginning distally in the stomach and moving proximally to visualize the entire length of the mass, and the highest T stage present within the mass should be reported.

Esophageal cancer appears as a hypoechoic lesion disrupting the usual wall layers (Figs. 25.3, 25.4 and 25.5). There is no difference in the EUS appearance of squamous cell carcinoma and adenocarcinoma [33]. The endoscopic wall layer abnormality has been correlated with the tumor depth of invasion [34]. The accuracy of EUS for T staging, compared to surgical pathology, is clearly superior to CT and PET, and was found to be 80–90 % across various studies and meta-analyses [19, 35]. However, the accuracy of EUS varies by T stage. EUS is better for T3 and T4 tumors with 86 % accuracy for T4. Staging accuracy may be less for superficial tumors (T1) because the low frequency used with the echoendoscope impairs visualization of the muscularis mucosa [36].

EUS also offers the ability to assess for regional lymph node disease . By endosonographic appearance, involved nodes may be enlarged, rounded, uniformly hypoechoic, or well-defined being distinctly differentiated from surrounding fat, as compared to benign nodes which are often hyperechoic, elongated, and with less demarcated borders [37–39]. The presence of all four features is 80–100 % predictive of a malignant node; however, only a minority (20–30 %) of all malignant lymph nodes harbor all four features [37–39]. The accuracy of EUS for N staging is 75 % [35]. EUS is superior to CT and PET for N staging; in one study, the accuracy was 81 % compared to 69 and 56 %, respectively [40]. Because the visual characteristics are subjective, diagnostic error is possible with EUS imaging alone. Utilization of EUS-guided fine-needle aspiration (FNA) for lymph node cytology therefore improves sensitivity and specificity [41]. One study showed that, compared to EUS alone, EUS-FNA improved sensitivity from 63 to 93 %, specificity from 81 to 100 %, and accuracy from 70 to 93 % [42]. When local lymph nodes are seen at the level of the primary tumor, FNA is usually not pursued due to the concern that traversing the tumor carries the risk of false-positive results as well as translocating tumor cells into an area without malignant disease [40]. If multiple periesophageal lymph nodes are visualized away from the primary tumor, it is reasonable and practical to perform FNA of one to two representative malignant-appearing nodes. Additional lymph nodes that appear malignant can be counted as such for staging purposes, based solely on their similar malignant appearance.