Optimal Endoluminal Treatment of Barrett's Esophagus: Integrating Novel Strategies into Clinical Practice

Raf Bisschops


Expert Rev Gastroenterol Hepatol. 2010;4(3):319-333. 

In This Article

Indications & Patient Selection for Endoluminal Therapy

Indications for Endoluminal Therapy

The main issues to consider to justify endoluminal therapy in patients with early Barrett's neoplasia are oncological and procedural safety. From this perspective the possible involvement of lymph nodes has to be excluded before treatment. The risk of lymph node metastasis is known from retrospective surgical data and increases significantly when there is deep submucosal involvement.[6] Pretreatment risk assessment of lymph node involvement can be achieved by assessing the endoscopic appearance of the lesions, the use of endoscopic ultrasound (EUS) and computed tomography (CT) of the chest and upper abdomen. In Barrett's-associated high-grade intraepithelial neoplasia (HGIN) or early cancer, ulcerated type III lesions (Paris classification) or large lesions over 2 cm carry an increased risk of lymphatic metastasis. As such, every ER is an advanced staging procedure providing the ultimate T staging of the suspicious lesion. Lesions that are limited to the mucosa (T1a), well or moderately differentiated and without any blood or lymph vessel involvement are eligible for further endoluminal treatment with ablation or follow-up.[7–25] Submucosal invasion (T1b lesion), a poor grade of differentiation or invasion in the blood or lymph vessels in the ER specimen are indications for referral for traditional surgery. Once there is invasion of the submucosa, the risk of lymph node involvement is significantly increased to 15–49% from SM1 to SM3 invasion.[6,26] The risk of lymph node involvement in mucosal tumors is not zero but is outweighed by the surgical risk that ranges between 2 and 20% in mortality and 18 and 48% in morbidity according to the volume load in the centers.[3,25,27–31] Minimally invasive surgical approaches have been reported to have good oncological outcome and a quality of life after surgery that is comparable to the preoperative state.[32] However, these approaches still have an overall mortality of 2.9% and morbidity of 46%. In 5.9% of patients undergoing minimally invasive surgery, a conversion to a classical open procedure is necessary.[33] Nevertheless, during selection of patients for endoluminal treatment, these options need to be discussed, in particular if compliance and adherence to the meticulous follow-up inherent to endoluminal therapy is a problem. The indications for endoluminal therapy are listed in Box 1 & Figure 1.

Figure 1.

Treatment of early neoplasia in Barrett's disease. Example of a treatment algorithm as it is used at UZ Leuven in a multidisciplinary approach with the surgeons. CT: Computed tomography; ER: Endoscopic resection; EUS: Endoscopic ultrasound; G1–G2: Well-to-moderately differentiated; G3: Poorly differentiated; HGIN: High-grade intraepithelial neoplasia; N+: Lymph node metastasis; RFA: Radiofrequency ablation; SRER: Stepwise radical endoscopic resection; T1a: Mucosal lesion; T1b: Submucosal lesion; V+/-: Positive/negative for lymphovascular infiltration.

Pretreatment Work-up in Early Barrett's Cancer

Prior to any curative endoluminal therapy, a proper work-up and staging should be performed. In most series this includes a high-resolution endoscopy, CT of the chest and upper abdomen and an EUS of the esophagus and mediastinum.

The role of advanced endoscopic imaging techniques, such as high-definition or superhigh-resolution endoscopy and virtual chromoendoscopy, is not clear. As a general rule, the endoscope with the highest resolution available should be used to stage and map the entire esophagus prior to endoluminal therapy. In particular, since RFA is now indicated for flat HGIN, it is important to resect small visible lesions prior to any ablation in order to obtain adequate pathological staging. Purely from a physic's point of view, a higher point-to-point resolution enables better identification of smaller lesions. Indirect evidence exists indicating that high-definition endoscopy helps to identify more lesions in Barrett's esophagus. One study addressed the possible use of increased detection of dysplasia by narrow-band imaging (NBI).[34] In 65 patients, high-definition endoscopy with NBI detected more dysplasia in comparison to standard resolution white light endoscopy (57 vs 43%). However, since two imaging improvements were introduced simultaneously (high definition and NBI), it is not clear from this study what the exact contribution of either technique in the detection of dysplasia might be. Curvers et al. assessed the interpretation and appreciation of the quality of vascular and mucosal patterns of images using different imaging modalities (acetic acid or indigo carmin chromoendoscopy, NBI and normal high-definition white-light images).[35] Expert endoscopists were able to distinguish neoplastic from non-neoplastic sites and there was no additional value of the enhancement techniques. This may indicate that the improvement of image quality by high-definition or superhigh-resolution endoscopy may be more important in detecting lesions than advanced enhancement techniques. NBI does, however, help to characterize suspicious lesions. Lesions with irregular mucosal and/or vascular patterns have a very high likelihood of being neoplastic.[36] In particular, a simplified classification of NBI patterns looks promising to assess suspicious areas.[37] Four different types can be distinguished: round pits with regular microvasculature, villous ridge pits, absent pits with regular microvasculature and distorted pits with irregular microvasculature. These findings have a good diagnostic accuracy in predicting the presence of columnar epithelium without intestinal metaplasia (type A), columnar epithelium with intestinal metaplasia (type B and C) and high-grade dysplasia (type D).

Other techniques that help to relocalize lesions or areas of dysplasia are chromoendoscopy, autofluoresence imaging (AFI) and confocal endomicroscopy.

Chromoendoscopy with methylene blue helps to relocalize dysplastic areas since they are reported to remain unstained.[38–41] However, this was not confirmed by other authors who even found an increased uptake of methylene blue in dysplastic areas.[42] Acetic acid helps to highlight the pit pattern of Barrett's mucosa. A similar classification as that mentioned for NBI can be used to differentiate between columnar-lined epithelium with or without specialized intestinal metaplasia and intraepithelial neoplasia. In one study, acetic acid staining upgraded the worst histological stage in 24% of the patients in comparison to the standard Seattle protocol.[43] AFI, and more particularly trimodal imaging in combination with NBI, looks very promising as an endoscopic tool to identify neoplastic Barrett's mucosa.[44–46] In one study, the use of AFI as a red flag technique increased the detection of neoplastic lesions from 53 to 90% (11 additional patients). Both high-resolution endoscopy and AFI had a high false-positive rate for detection of lesions, which could be decreased by additional inspection with magnification NBI, but at the expense of missing three patients with a neoplastic lesion.[46]

Confocal endomicroscopy is a new endoscopic imaging technique that enables one to obtain images with a subcellular resolution. Confocal endomicroscopy during ongoing endoscopy can differentiate between gastric mucosa, specialized intestinal metaplasia and neoplastic changes with a high diagnostic accuracy[47–49] and can help to localize the site for ER.[49,50]

The role of EUS for T and N staging has been well studied and is currently clear. Conventional EUS with a 7.5–12-MHz probe does not usually allow visualization of the muscularis mucosae and is, therefore, not suitable for differentiating between T1a and T1b disease. The introduction of high-frequency EUS probes (12–30 MHz) permits viewing of four mucosal layers (two layers of epithelium, lamina propria and muscularis mucosae) and theoretically should allow for a better differentiation between mucosal and submucosal lesions and help in patient selection. However, several recent studies have addressed this hypothesis and have shown that the accuracy of T-staging for early lesions is relatively adequate, but insufficient in comparison to staging ER. In the most recent studies, sensitivity and specificity for diagnosing T1a disease was 62–100% and 65–94%, respectively.[51–65] In a recent study,[53] the appearance of early lesions on high-resolution endoscopy was prospectively compared with high-frequency EUS in 100 patients (81 adenocarcinoma) with early esophageal neoplasia. The accuracy for staging early esophageal neoplasia was 83.4 and 79.6% for high-resolution endoscopy and high-frequency EUS, respectively. The sensitivity for mucosal tumors was also high for both techniques (94.1 and 91.2%, respectively) but only 56 and 48%, respectively, for submucosal lesions. This study showed that tumors located close to the gastroesophageal junction and tumors with a superficial infiltration of the submucosal layer were more likely to be incorrectly staged. Nonetheless, this study showed that high-resolution endoscopy in the hands of endoscopists with extensive experience in ER is just as effective as EUS in assessing T staging in early esophageal lesions. A recent study indeed showed that the macroscopic Paris classification of lesions helps to predict final T staging after ER, and helps to decide whether endoluminal therapy can be performed. Pech et al. showed that type IIb lesions contain neoplastic features that are confined to the mucosal layer in 96% of cases and are more likely to be HGIN.[59] In type IIa and IIc lesions a favorable stage was less frequently observed. In the experienced hands of the Wiesbaden group, the interobserver agreement for the classification of lesions was good (κ-value 0.86). It is, however, not clear how well this applies to the less experienced endoscopist. In general, however, macroscopic appearance of the lesions should be the main determinant for deciding on a staging ER. From the studies previously mentioned, it should be concluded that there is a limited, if any, role for EUS in T staging of early superficial lesions in Barrett's disease.

The goal of EUS in early Barrett's neoplasia is mainly limited to exclude pathological lymph node involvement. Lesions containing only HGIN are not at risk for metastatic disease. However, particularly in a patient with a long Barrett's segment, the esophagus is often harboring a more advanced lesion that is only diagnosed after ER or surgery.[66] Since the endoscopic T staging is not fully accurate and the risk for lymph node metastasis is not entirely zero, EUS is the method of choice to perform N staging prior to endoluminal therapy. Lymph node characteristics that need be assessed include diameter, echo texture (homogeneous/heterogeneous), shape (round/elliptical or triangular), demarcation (sharp/fussy) and echogenicity (hypo/hyperechoic). Lymph nodes measuring more than 10 mm in diameter with a hypoechoic, homogeneous appearance, a round shape and sharp demarcation are very likely to be malignant.[63] However, all four features are only present in 25% of malignant lymph nodes. In general the accuracy of EUS for lymph node metastasis is 64–90%.[56,60,64] A recent study in 214 patients with esophageal cancer showed that EUS was correct in only 64.5% of patients to predict final N staging. Sensitivity of EUS was very good (93.8%) but specificity was very poor (20%), which may have a major impact on management and, in the case of early cancer, patients may be falsely referred for surgery.[64] Fine needle aspiration significantly helps to improve lymph node staging and has a sensitivity and specificity of approximately 90%.[67]

In the case of early neoplastic neoplasia one can wonder whether there is a role for EUS at all. If a lesion looks amendable by endoluminal therapy, a staging ER can be performed and if the final pathological staging shows no submucosal infiltration and no signs of lymphovascular invasion, the risk of nodal disease is very low. Regardless, a lymph node looking suspicious on EUS should be confirmed as malignant using fine needle aspiration to be of any therapeutic consequence.

The role of the CT scan or 18F-fluoro-2-deoxy-D-glucose PET appears to be very limited in staging lymph nodes in early esophageal cancer.[60,62] First, only the size of the lymph nodes are taken into account on a CT scan, whereas more features are assessed during EUS. Second, a recent study compared the diagnostic accuracy of PET–CT with EUS.[60] In this study, sensitivity and specificity for N staging by EUS were 80 and 70%, respectively, whereas for CT they were 50 and 83% and for PET 57 and 85%, respectively.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: