Redefining endometrial cancer phenotypes in the era of molecular classification and sentinel lymph node mapping: results from a multicenter Italian study

Objective: To describe the clinicopathologic phenotypes of endometrial cancer across molecular sub-types and to identify independent predictors of nodal metastasis and sentinel lymph node (SLN) mapping failure. Methods: This multicenter retrospective study included 2592 endometrial cancer patients with defined molecular classification and SLN mapping. Continuous variables are reported as medians. Tumors were classified as POLE-mutated, mismatch repair-deficient, p53-abnormal, or no specific molecular profile. When hormone receptor status was available, no specific molecular profile tumors were further sub-classified according to European Society of Gynaecological Oncology/European Society for Therapeutic Radiology and Oncology/European Society of Pathology 2025 criteria. Results: Molecular sub-types exhibited distinct phenotypes: POLE-mutated tumors occurred in younger patients (median 56 years), with the lowest body mass index (27.0 kg/m2), smallest tumor size (26 mm), and the lowest nodal metastasis rate (8.0%); p53-abnormal tumors affected older patients (67 years), with the largest tumors (35 mm), frequent non-endometrioid histology (78.4%), and the highest nodal metastasis rate (26.1%); mismatch repair-deficient tumors showed intermediate age (63 years), body mass index (28.5), predominantly endometrioid histology (90.4%), substantial lymphovascular space invasion (28.3%), and a 19.4% nodal metastasis rate; no specific molecular profile tumors, the most common group (55.6%), had the highest body mass index (29.0 kg/m2), were low-grade in 84.2% of cases and endometrioid in 92.5%, and exhibited a low metastasis rate (11.0%). Among no specific molecular profile tumors with available hormone receptor status, a high-risk-like sub-group (22%), characterized by hormone receptor negativity and/or high-grade histology, displayed aggressive features macro-metastases (18.2%), SLN mapping failure (24.3%), substantial lymphovascular space invasion (33.2%), deep myometrial invasion (54.3%), and cervical stromal involvement (23.1%) (resembling a "p53-abnormal-like" phenotype despite their molecular profile). At multivariate analysis, molecular sub-type was not an independent predictor of nodal involvement (p=.20) or SLN mapping failure (p=.75). Conclusions: Molecular sub-types in endometrial cancer reflect distinct and reproducible phenotypes. However, classic histopathologic features remain the strongest predictors of nodal spread and SLN detection failure.