L1CAM Expression in Microcystic, Elongated, and Fragmented (MELF) Glands Predicts Lymph Node Involvement in Endometrial Carcinoma

Simple Summary L1CAM overexpression (>= 10%) and the microcystic, elongated, and fragmented (MELF) pattern of invasion have previously been assessed as prognostic factors in endometrial carcinoma. We aimed to assess the relationship between L1CAM expression, MELF glands, and lymph node involvement in endometrial carcinoma, as all these factors are related to epithelial-to-mesenchymal transition. We evaluated L1CAM expression in 58 cases of uterine-confined, low-grade endometrioid carcinomas. We found that most cases (65.5%) expressed L1CAM in a limited manner to MELF glands. Cases with L1CAM expression in >= 10% of the MELF component showed a significantly higher tendency to lymph node spread (p < 0.001), even when adjusted for lymphovascular space invasion, depth of myometrial invasion and p53/mismatch repair status. On this account, L1CAM expression in the MELF component may stratify the prognosis and management in patients with uterine-confined, low-grade carcinomas. In endometrial carcinoma, both L1CAM overexpression and microcystic, elongated and fragmented (MELF) patterns of invasion have been related to epithelial-to-mesenchymal transition and metastatic spread. We aimed to assess the association between L1CAM expression, the MELF pattern, and lymph node status in endometrial carcinoma. Consecutive cases of endometrial carcinoma with MELF pattern were immunohistochemically assessed for L1CAM. Inclusion criteria were endometrioid-type, low-grade, stage T1, and known lymph node status. Uni- and multivariate logistic regression were used to assess the association of L1CAM expression with lymph node status. Fifty-eight cases were included. Most cases showed deep myometrial invasion (n = 42, 72.4%) and substantial lymphovascular space invasion (n = 34, 58.6%). All cases were p53-wild-type; 17 (29.3%) were mismatch repair-deficient. Twenty cases (34.5%) had positive nodes. No cases showed L1CAM positivity in >= 10% of the whole tumor. MELF glands expressed L1CAM at least focally in 38 cases (65.5%). L1CAM positivity in >= 10% of the MELF component was found in 24 cases (41.4%) and was the only significant predictor of lymph node involvement in both univariate (p < 0.001) and multivariate analysis (p < 0.001). In conclusion, L1CAM might be involved in the development of the MELF pattern. In uterine-confined, low-grade endometrioid carcinomas, L1CAM overexpression in MELF glands may predict lymph node involvement.