David L. Diehl, MD, FACP, FASGE and Shaffer R.S. Mok, MD, MBS
Advancements in endoscopy techniques and devices have expanded the boundaries of endoscopic resection of gastrointestinal (GI) lesions. Careful endoscopic assessment is vital in deciding on the most appropriate therapy for any given lesion. Endoscopic ultrasound (EUS) is an important tool for the characterization of mucosal and submucosal GI lesions, and for the staging of neoplastic lesions. EUS, in combination with fine-needle aspiration (FNA) or biopsy (FNB), is a valuable tool in the depth evaluation and staging of lesions prior to treatment. This chapter discusses the role of EUS in the evaluation of mucosal and submucosal GI lesions.
EUS evaluation can be performed with radial and curvilinear array echoendoscopes, as well as miniaturized through-the-scope ultrasound probes. The radial echoendoscope has a 360-degree field of view and scans in a plane perpendicular to the long axis of the scope (Figure 5-1). A water-filled balloon over the ultrasound transducer enables optimal acoustic coupling to the mucosa. The curvilinear echoendoscope has a 180-degree scan area along the long axis of the scope (Figure 5-2). With this instrument, more scope torqueing is required for adequate visualization. Real-time guidance of a biopsy needle into a target lesion is the main advantage of the linear echoendoscope.
A through-the-scope ultrasound probe (miniprobe) is available for visualizing small lesions (Figure 5-3). A 20 MHz miniprobe is most commonly used for optimal resolution of the GI tract wall layers. Imaging is performed with the transducer parallel to the mucosal surface of a lesion, typically with instillation of water.
Examinations with the radial and linear echoendoscopes are typically performed at ultrasonic frequencies of 5 to 7.5 MHz to evaluate the relation of the lesion to the wall of the organ, to visualize the presence of nodes, and to evaluate for invasion of wall layers or adjacent organs. The frequency can be increased to 10 MHz to evaluate the wall layers in more detail. In contrast, use of the higher frequency miniprobe allows for a more detailed evaluation of the luminal layers, but depth of penetration is limited.
The water-fill technique is used to improve acoustic coupling and visualization of the target lesion in the GI tract. Instillation of “deaerated” water through the endoscope working channel allows the miniprobe to be underwater during EUS visualization. The water-fill technique carries a risk of tracheal aspiration, so attention and care is required. In addition, some parts of the stomach, such as the antrum, can be hard to visualize underwater because of preferential filling of the fundus.
As an alternative to the water-fill technique, a water-filled condom may be secured around the end of the echoendoscope to improve acoustic coupling (“condom technique”).1,2 Limitations of this technique include difficulty with endoscopic visualization of the lesion as well as difficulty with advancement of the endoscope through the upper esophageal sphincter.
The tubular GI tract has layers composed of mucosa, submucosa, muscularis propria, serosa (in most areas), and adventitia (Figure 5-4). EUS imaging at 5 to 10 MHz, visualizes 5 layers (Figure 5-5): a hyperechoic layer 1 (superficial mucosal layer), a hypoechoic layer 2 (deep mucosal layer), a hyperechoic layer 3 (submucosa), a hypoechoic layer 4 (muscularis propria), and a hyperechoic layer 5 (adventitia). Nine layers may be seen at higher frequencies under optimal imaging conditions.
Endoscopic Ultrasound for T Staging of Esophageal Malignancy
The American Joint Commission on Cancer criteria for T staging of esophageal cancer are outlined as follows: involvement of tumor within the first or second echoendoscopic layer corresponds to T1a disease, third layer T1b, fourth layer T2, fifth layer T3, invasion beyond the fifth layer into the surrounding structures (pericardium, pleura, diaphragm, aorta, trachea, or vertebral body) T4 (Figure 5-6).3
An esophageal lesion restricted to the mucosal layer has less than a 5% risk for lymph node metastasis, and endoscopic mucosal resection (EMR) or endoscopic submucosal dissection (ESD) is indicated in this setting. On the other hand, the risk of nodal metastasis increases to 12% to 27% with submucosal invasion (SMI) and makes the decision for endoscopic resection difficult.4 Furthermore, the rate of nodal metastasis increases to 36% to 46% with invasion of the deep submucosa and muscularis propria.
Several meta-analyses have shown EUS to be useful in the T staging of esophageal cancer, with pooled sensitivities and specificities for T2 lesions of 81% and 96%, for T3 lesions 91% and 84%, and for T4 lesions 92% and 97%.5,6 Sensitivity and specificity for T1 lesions5–7 were 82% and 99%, respectively. For T1a tumors, EUS had 85% sensitivity and 87% specificity, with overstaging of 4.3%. For T1b tumors, sensitivity and specificity was 86%, and understaging was seen in 15%.5 Other studies have found that T1a tumors were correctly staged in only 39% of cases, and T1b in 51%. This limited the accuracy of EUS to 56% to 67%.7,8 The explanation for these differences is unclear.
The Role of Endoscopic Ultrasound in Management of Barrett’s Esophagus
Evaluation of Barrett’s esophagus (BE)-associated neoplasia by EUS has been extensively evaluated. Though several studies have suggested a positive role for EUS in the setting of BE, early studies also utilized EMR as a comparative group, demonstrating high concordance.9–13 More recent studies have concluded that EUS has limited value in evaluation of BE14–16 because of the possibility of overstaging and understaging. Furthermore, EUS has been found to confer no additional benefit compared to careful chromoendoscopic mucosal inspection of BE alone.17
Diagnostic Performance of Endoscopic Ultrasound for T Staging of Gastric Cancer
The overall accuracy for EUS in predicting T stage of gastric cancer is 83% (range, 71% to 92%).18 Meta-analyses have demonstrated accuracies of 77% to 83% for T1 lesions, 65% for T2, 85% to 86% for T3, and 66% to 79% for T4 lesions.19–21
The goal for EUS prior to endoscopic resection is to differentiate between mucosal (M) superficial SMI (SM1: depth of less than 500 µm) with a low risk of nodal metastasis from lesions with deep SMI (SM2: depth of 500 µm or more) carrying a higher risk of nodal disease. SM2 and deeper lesions are typically treated surgically. ESD may be an option if the patient is not considered to be a surgical candidate.
The diagnostic accuracy of EUS prior to EMR or ESD has been evaluated in meta-analyses.22 Sensitivity of EUS was 88%, specificity 80%, and accuracy of 93%. When further separating lesions appropriate for ESD (mucosal or SM1) vs not (SM2 and above), the accuracy of EUS was 98% to 90%.23 Hence, endoscopic evaluation in combination with EUS plays an important role in predicting ESD-appropriate lesions.24
Endoscopic Ultrasound Characteristics Affecting T Staging
The accuracy of high-frequency EUS miniprobe imaging diminishes for lesions larger than 20 mm.25 The overall accuracy of EUS decreases from 88% for lesions < 30 mm to 44% for lesions ≥ 30 mm.22 A depression or ulceration in the lesion may lead to a decline in EUS accuracy from 83% to 70% for superficial (< 500 µm) and 46% for deep (≥ 500 µm) ulcers leading to overstaging.24,26–28 The presence of an ulcer on the surface of the lesion limits the ability to detect microinvasion, which is vital to differentiating SM1 from SM2 tumors. Submucosal fibrosis also predicts incomplete resection with ESD.29,30 Nondepressed lesions were found to be understaged in one study and overstaged lesions in another.31,32 Poorly differentiated histology was associated with EUS overstaging in 71% of lesions in one study.31
EUS staging of rectal malignancies has an accuracy of 80% to 95% for T staging33–35 and 70% to 75% for nodal staging.36,37 A miniprobe is preferred for the evaluation of more proximal lesions. Miniprobe EUS has demonstrated accuracy for evaluating and differentiating superficial (T1a from T1b) colorectal lesions suitable for EMR/ESD.38,39
As in the foregut, tumor depth has an impact on the overall risk of cancer recurrence.40 Prognostically, a tumor depth within the submucosa of SM2/3 (> 1000 µm) has a higher risk of recurrence when compared to a lesion with superficial submucosal involvement (< 1000 µm).41 Similar to the upper GI tract, increased tumor thickness (≥ 19 mm) or inflammation and fibrosis29,30 can limit the effectiveness of EUS staging.
Close endoscopic evaluation of lesions using the Paris, Kudo, and other classifications are valuable for predicting SMI42,43 and should be routinely applied in the assessment of any lesion for resection. These classification systems are discussed in greater detail elsewhere in this book.
Other Endoscopic Ultrasound Findings That May Affect Endoscopic Submucosal Dissection Practice
EUS can be used to detect submucosal fibrosis, which may predict incomplete ESD. Hyperechoic interruption of the third layer > 5 mm may predict SM fibrosis and incomplete resection.29,44 EUS has also been evaluated for determining the presence of intervening blood vessels that could affect the safety and efficiency of ESD. Though one study demonstrated no benefit for EUS to predict endoscopic complication, another showed value.45,46
The detection of malignant lymph nodes on EUS would be an absolute contraindication for endoscopic resection of any lesion. Endosonographically, suspicion for malignancy within a lymph node is raised by defined criteria: size > 1 cm, round appearance, hypoechoic texture, and well-circumscribed borders.47 The risk of positivity increases with each additional criterion48 (Figure 5-7). The sensitivity and specificity of EUS in esophageal or gastric malignancy for predicting nodal disease has been found to range from 69% to 74% and 80% to 84%, respectively.19,20 The sensitivity for predicting nodal disease is increased in nodes ≥ 20 mm.49
Limitations of Endoscopic Ultrasound for Staging Lesions Prior to Endoscopic Submucosal Dissection
The imperfect resolution of EUS is an important potential limitation of this imaging technology.50–52 There are several factors that can decrease the accuracy of EUS imaging: larger lesions, undifferentiated histology, ulceration, and submucosal fibrosis.53,54
Difficulty in obtaining adequate images can occur in up to one-third of cases.44 Accurate EUS evaluation can be particularly problematic in the stomach when compared to other anatomical locations. It is harder to fill the distal stomach with water for acoustic coupling, which in turn may limit the ability to adequately visualize the lesion on EUS. Furthermore, the thinner and vascular submucosal layer in the upper stomach can be challenging for the detection of SMI on EUS.30
Optimal imaging with EUS for lesion characterization and staging can be a time-consuming procedure. In addition, “water-filling” in the upper GI tract can carry a risk of aspiration during imaging because fairly large quantities of water may be necessary to obtain immersion of the mucosa to be examined. Many of the studies of EUS in the staging of early gastric cancer limited their analysis to those cases for which optimal EUS imaging was possible, which may lead to overestimation of the value of EUS.53
EUS has been shown to have a high negative predictive value (158/164, 96%) for deep SMI. Hence, it has been proposed that in cases for which deep SMI is suspected by white-light endoscopy, an ESD approach can still be considered if EUS does not show evidence of invasion.54
Several studies originating from Asia have favored endoscopic inspection and not EUS as part of the evaluation prior to EMR or ESD.19,20 In some cases, this may have resulted from evidence of overstaging of gastric lesions with EUS, leading to unnecessary gastrectomy.55–57 Evolving data from non-Asian centers also demonstrated equivalent accuracies of endoscopic inspection compared to EUS for predicting SMI.58,59 Hence, whether EUS must be routinely performed in addition to endoscopic assessment remains debatable.
The evaluation of subepithelial lesions (SELs) in the GI tract starts with EUS. Understanding the location within the luminal wall, typical lesion features on imaging, along with prognostic findings on EUS are important in dictating the approach to SELs. EUS also has a role in surveillance, as changes in SEL size or appearance can be discerned.
The layer that the lesion arises from can often be determined on EUS, which may further guide the type of endoscopic resection if indicated (eg, ESD, endoscopic full-thickness resection, submucosal tunneling endoscopic resection). Most case series and reports of endoscopic management of SELs have included EUS examination prior to resection.60 This practice is supported by the strong correlation between the appearance on EUS and final pathologic diagnoses for SELs.61 EUS may also serve as a prognostic tool for the technical success of the endoscopic resection, as studies looking at the role of EUS in predicting en bloc resection have demonstrated improved resection results and fewer adverse events when a muscularis propria layer or plane was detected.62–64
EUS-guided needle biopsy is useful in obtaining a tissue diagnosis of SELs. In this way, a tissue diagnosis may be obtained that can be used to favor EUS surveillance without resection, or in other cases, suggest that resection (endoscopic or surgical) should be pursued.
Endoscopic Approach to Diagnosis of Subepithelial Lesions
The overall prevalence of SELs in routine endoscopic examinations is 0.4%. The combined accuracy of diagnostic EUS for predicting the identity of an SEL ranges from 46% to 83%,36,65–69 and increases to 63% to 98%70–72 when adding FNA or FNB.
A careful appraisal of the endosonographic layer from which the lesion is arising is an important first step. Next, analysis of echo pattern, internal features, outer border, vascularity, relation to other organs and surrounding lymphadenopathy (if present) should be performed.
Different lesions are typically found in specific layers (Table 5-1).
Lipomas can be found anywhere within the GI tract, typically are in the third endosonographic layer, and are homogeneous and hyperechoic61,73 (Figures 5-8A and B). They have the classic “pillow sign” during endoscopic “palpation,” and unroofing will often identify yellow lipomatous tissue. These lesions follow a benign course, and identification does not require removal, unless bleeding, obstruction, or inability to differentiate from liposarcoma are present.
GIST, gastrointestinal stromal tumor; NA, not applicable.
Portrayal of each SEL in each echoendoscopic layer with included characteristics seen on EUS.