The management of patients with HGD in BE remains controversial, largely because the natural history of this condition is so poorly defined. For example, Reid et al. reported a 59 % 5-yr cumulative cancer incidence among 76 patients with HGD in BE [1],whereas Schnell et al. found that only 12 (16 %) of their 75 patients with HGD developed adenocarcinoma during a mean follow-up period of 7.3 yr [2]. Four management options have been proposed for patients with HGD in BE: (1) esophagectomy, (2) endoscopic ablative therapies, (3) EMR, and (4) intensive endoscopic surveillance, in which invasive therapy is withheld until biopsy specimens show evidence of invasion [3]. Clearly, there are risks associated with all of these approaches, and all would benefit from accurate staging of the neoplasia.
Estimation of the depth of tumor penetrationThe less than perfect reliability of endoscopic staging with the adjunct of chromo endoscopy and magnification endoscopy for the estimation of cancer depth before EMR can be improved by the use of endoscopic ultrasonography (EUS), in particular highfrequency US probe sonography (HFUPS). HFUPS may distinguish nine layers within the wall of GI organs in contrast to a five-layered structure seen with conventional EUS, thus providing better images that are useful in the evaluation of transmural penetration and in differentiating cancers limited to the mucosa from those with submucosal penetration. The main limitation of HFUPS is its tendency to overstage early lesions. The diagnostic accuracy of HFUPS in assessing depth of tumor invasion in the esophagus, stomach and colon ranges from 67 % to 94 % among the published studies [4 ± 9]. This high variability in HFUPS accuracy may reflect the use of probes with different penetration (15MHz versus 20MHz) and differences in the patient populations studied. Accuracy of HFUPS, in fact, is significantly better for elevated type lesions than for depressed ones [7].When different techniques are used in combination to select lesions suitable for EMR, the overall accuracy is high [10]. Because of the limitations of these staging techniques, however, it has been suggested that when a lesion meets the generally accepted criteria of size, and appearance is encountered, EMR can be performed without prior HFUPS, as long as the lesion can safely be removed in its entirety [11 ± 12]. Submucosal injection used to facilitate EMR can also help to decide whether or not to continue with the procedure [13]. The observation of a bleb formation with elevation of the overlying mucosa indicates the absence of deep submucosal involvement and the feasibility of EMR [30] On the other hand, the dense fibrosis associated with deep submucosal invasion prevents fluid infiltration through the submucosal connective tissue, decreasing bleb formation and elevation of the lesion [14]. This so-called ªnon-lifting sign has been found to have 100 % sensitivity, 99 % specificity, and 83 % positive predictive value for invasive carcinoma in patients with early cancer of the colon [15...