Role of preoperative serum CA-125 and fibrinogen levels in predicting lymph node metastasis, myometrial invasion, and lymphovascular space invasion in patients with endometrial cancer
To evaluate whether preoperative serum levels of cancer antigen 125 (CA-125) and fibrinogen could provide clinical guidance for surgical planning and treatment management in patients with endometrial cancer. A retrospective study was conducted, including 582 patients who underwent surgery for endometrial cancer between October 2020 and December 2024 at a tertiary centre. Preoperative CA-125 and fibrinogen levels, clinical variables, and postoperative pathological findings were collected from electronic medical records. The primary outcome was lymph node metastasis (LNM). The secondary outcomes were myometrial invasion (MI) ≥ 50% and the presence of lymphovascular space invasion (LVSI). Independent predictors of LNM, MI ≥ 50% and LVSI were evaluated using multivariable logistic regression based on TRIPOD-aligned preoperative and postoperative modelling frameworks. Model performance was assessed through area under the receiver operating characteristic curve (ROC-AUC), calibration metrics, and decision curve analysis. Optimal biomarker thresholds were determined using ROC analysis and the Youden index. Sensitivity analyses examined the diagnostic performance of the Mayo low-risk criteria and the additional value of incorporating CA-125 ≥ 29.1 U/ml. MI ≥ 50%, LVSI and LNM were identified in 37.3%, 33.2% and 13.2% of patients, respectively. An elevated CA-125 level was significantly associated with MI ≥ 50%, LVSI and LNM (all P < 0.001), while the fibrinogen level only differed between LNM groups (P = 0.009). In univariable analyses, multiple clinicopathological factors - including age, body mass index (BMI), tumour grade, histological subtype, cervical stromal involvement, lower uterine segment involvement, tumour size, MI ≥ 50%, LVSI, CA-125 and fibrinogen - were associated with LNM. In the preoperative model (Model 1), CA-125 [odds ratio (OR) 1.02; P < 0.001], older age (OR 1.05; P = 0.001) and lower BMI (OR 0.92; P = 0.001) were independent predictors of LNM, yielding an AUC of 0.805. In the postoperative model (Model 2), LVSI was the strongest predictor (OR 20.46; P < 0.001), and the model demonstrated improved predictive performance (AUC 0.889). CA-125 remained independently associated with LNM in both models. CA-125 alone achieved an AUC of 0.740 with an optimal cut-off of 29.1 U/ml. For the secondary outcomes, age, tumour size, grade and LVSI independently predicted MI ≥ 50% (AUC 0.862), whereas BMI, histological subtype, cervical stromal involvement, MI ≥ 50% and tumour size independently predicted LVSI (AUC 0.886). Fibrinogen did not have independent predictive value for MI or LVSI. In a sensitivity analysis, the addition of CA-125 ≥ 29.1 U/ml to the Mayo low-risk criteria improved the detection of LNM in cases initially classified as low risk. Preoperative CA-125 level in patients with endometrial cancer is associated with MI, LVSI and LNM. CA-125 remained an independent predictor of LNM in both the preoperative and postoperative models, whereas fibrinogen did not retain independent significance in multivariable analyses. Incorporating CA-125 into the Mayo low-risk criteria increased sensitivity for detecting LNM. These findings suggest that combining biochemical markers with clinical and pathological variables may enhance surgical planning, and support the personalization of postoperative treatment strategies.