Preoperative CA125 and metabolic factors associated with high-risk pathology in endometrial cancer

Molecular classification is now central to endometrial cancer risk assessment, but its implementation in routine practice varies across clinical settings. Against this background, the present retrospective study examined whether routinely available preoperative CA125 and metabolic factors are associated with postoperative high-risk pathological features in endometrial cancer. A total of 188 patients with surgically treated, pathologically confirmed endometrial cancer at Cangzhou Central Hospital Affiliated to Hebei Medical University (Cangzhou, China) between January 2024 and January 2026 were included. Clinical characteristics, metabolic indicators, preoperative CA125, postoperative pathological variables and available molecular classification data were collected. The primary outcome was the presence of any high-risk pathological factor, defined as high-grade/aggressive histology, deep myometrial invasion, lymphovascular space invasion (LVSI), cervical stromal involvement or lymph node metastasis. Overall, 98 patients (52.1%) had at least one high-risk pathological factor. Molecular classification was missing or undetermined in 66 patients (35.1%); these results were therefore treated as contextual exploratory information. In the prespecified multivariable model, age [odds ratio (OR)=1.061; 95% CI, 1.019-1.108; P=0.005] and log(CA125) (OR=2.059; 95% CI, 1.375-3.242; P<0.001) were associated with the composite high-risk pathological phenotype. Exploratory analyses of individual endpoints showed the most consistent CA125 associations with LVSI positivity and lymph node metastasis. BMI, diabetes, hypertension, metabolic syndrome and routine lipid indicators did not show reproducible associations with high-risk pathological status. Overall, higher preoperative CA125 was independently associated with the composite adverse pathological phenotype and may provide complementary preoperative risk information pending prospective validation.