Endoscopic resection has become standard therapy for selected patients with EGC . Ruling out lymph node metastasis (and the risk thereof) is a critical step prior to attempting EMR or ESD . Lymph node metastasis is rare in small carcinomas with the intestinal phenotype that are confined to the mucosa, while the risk increases with SM invasion . Therefore, endoscopic assessment prior to treatment is crucial for deciding whether the tumor is suitable for EMR . However, the preoperative diagnostic accuracy is not high, and the number of cases with SM invasion detected in the EMR or ESD specimens is increasing .
In an attempt to identify histologic features observed in pinch biopsies associated with SM invasion in EMR specimens, we analyzed the cases and found that poorly differentiated component, IEND, cribriforming, and papillary architecture were significantly associated with SM invasion in both univariate and multivariate analyses. However, the degree of differentiation did not always correlate with SM invasion; this is likely a selection bias because most cases that are commonly considered for EMR are the differentiated type. Actually, in the validation cohort, histologic grade was significantly associated with SM invasion. IEND is a distinct pattern of necrosis composed of amorphous eosinophilic material admixed with necrotic epithelial fragments within the lumen of a dilated atypical gland,  and it is predominantly found in moderately differentiated gastric adenocarcinomas. In this study, IEND was significantly more frequent in SMiGCs compared to IMCs. Distinct architectural patterns, especially papillary architecture and cribriforming, were also significantly more common in pre-treatment biopsies of cases with SM invasion than in those limited to the mucosa. These two architectural patterns have the same significance in colorectal adenocarcinomas [7, 20, 21].
In our study, poorly differentiated component was the most significant pathologic factor significantly associated with SM invasion in the validation cohort. This result is consistent with the observations by Kuroda et al., who reported that poorly differentiated component was closely associated with SM invasion in the diffuse type of EGCs and lymph node metastasis . In SMiGCs, poorly differentiated component was reported to be significantly associated with higher risks of lymph node metastasis [19, 29]. Zheng et al. also reported that mixed-type gastric carcinomas are larger, more deeply invasive into the wall, and associated with a higher frequency of lymph node metastasis compared to pure intestinal or diffuse-type carcinomas .
The present study had several limitations. Pretreatment gastric biopsy may not fully indicate neoplastic progression in a patient with EGC . An endoscopist may miss the invasive component of a tumor that harbors “SM invasion-associated” histologic features. Moreover, although specificity was high, the sensitivity was very low. Furthermore, endoscopic biopsy specimens are small, and thus significant histologic features may be missed. Despite these limitations, multiple biopsies, expertise in endoscopy, and higher resolution endoscopy would be helpful to get higher sensitivity.
In conclusion, poorly differentiated component is the sole significant histopathologic predictor of SM invasion in gastric pre-treatment biopsies and should be evaluated and described in pathology reports.