Predictive value of deep myometrial invasion for lymph node metastasis in low-grade endometrioid carcinoma and its implications for surgical decision-making.

ObjectiveTo evaluate the predictive value of deep myometrial invasion (DMI) for lymph node metastasis (LNM) in low-grade endometrioid carcinoma (LGEC) and its implications for surgical decision-making.MethodsA retrospective cohort study was conducted on 212 patients with pathologically confirmed LGEC (G1/G2) who underwent surgical treatment at Women's Hospital, School of Medicine, Zhejiang University between January 2018 and December 2023. Clinicopathological variables were analyzed using univariate and multivariate logistic regression to identify independent risk factors for LNM. The predictive performance of DMI and other significant variables was assessed by sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and receiver operating characteristic (ROC) curves. A combined predictive model incorporating all independent factors was developed and internally validated using calibration curves and the Hosmer-Lemeshow test. External validation was performed in an independent cohort of 54 patients.ResultsLNM occurred in 9.4% (20/212) of patients. Multivariate analysis identified DMI ≥50% (OR=3.58, 95% CI: 1.23-10.41, P=0.019), cervical stromal invasion (CSI; OR=5.24, 95% CI: 1.47-18.73, P=0.011), and elevated preoperative CA125 >35 U/ml (OR=4.14, 95% CI: 1.42-12.07, P=0.009) as independent predictors. The combined model achieved a sensitivity of 90% and an NPV of 99% for LNM prediction; external validation yielded an NPV of 97.44%, confirming its robustness for identifying low-risk patients. Patients with DMI were more likely to undergo laparotomy than those with superficial invasion (50.0% vs. 18.3%, P<0.001) and more frequently received postoperative adjuvant therapy (52.1% vs. 12.8%, P<0.001).ConclusionDMI, CSI, and elevated CA125 are independent risk factors for LNM in LGEC. The high sensitivity and NPV of the combined predictive model support a selective lymphadenectomy strategy to avoid overtreatment in low-risk patients, while DMI also informs surgical approach selection and adjuvant therapy decisions.