Expert Assessment of Infiltration Depth and Treatment Allocation in Early Barrett Cancer

Aims Early neoplasia arising from Barrett esophagus (BE) can be curatively treated by endoscopic resection. The choice of the resection technique – endoscopic mucosal resection (EMR) or submucosal dissection (ESD) – largely depends on the assumed infiltration depth of the lesion and hence on preprocedural judgment by the endoscopist. To clarify the accuracy of the endoscopic assessment and treatment allocation, we performed a study showing endoscopic photographs to BE experts simulating a second opinion procedure.

Methods 202 cases of early BE neoplasia (82% men, mean age 66.9 years) were selected from our endoscopy database 2009-2022, with cancer in the resection specimen and 3-4 adequate endoscopic images demonstrating the lesion and surrounding BE. 110 resections had been done by ESD and 92 by EMR. These images as well as clinical data (age, sex, BE length) and preprocedural biopsy information were shown to 9 BE experts (>100 BE resections) who were blinded to the resection technique and the final histopathological results. Main outcomes were accuracy and interobserver variability to correctly diagnose a) any submucosal infiltration (T1b versus T1a), or b) deeper submucosal involvement (≥T1bsm2 vs. T1bsm1/T1a). Suggested treatment allocation was also recorded.

Results Of the 202 cases, 148 (73.3%) had stage T1m and 54 T1sm (26.7%; of those, n=35 with sm1 and n=19 with≥sm2) on final histology. The accuracy to diagnose sm infiltration (T1sm vs T1m) ranged between 35.2% and 48.9% for positive and between 77.7% and 82.3% for negative predictive values (overall kappa value 0.41). Overall kappa value was 0.27 for differentiation between deeper sm infiltration (≥T1bsm2) versus more superficial infiltration depth (T1bsm/T1a). Although the raters also differed substantially in their recommendation of the resection method, this was however more consistent with their morphologic assessment of lesions. 88.5% of invisible lesions or those slight irregularities/those classified as Paris IIb/G1 or G2 were allocated to EMR, in this subgroup, the rate of T1a cancers was 87.4% and R0 resection was finally achieved in 90.1%.

Conclusions Precise endoscopic assessment of BE cancer infiltration depth based on gross endoscopic lesion morphology largely fails due to poor sensitivity and high interobserver variability. From an oncologic perspective, it could be concluded that only invisible or flat lesions are safely treated by EMR, the remaining lesion by ESDs even if this could result in overtreatment in a variable percentage of cases. Such an approach should be tested in further prospective outcome studies.

Publikationsverlauf

Artikel online veröffentlicht:
15. April 2024