Development and internal validation of a KELIM-integrated prognostic nomogram for long-term survival prediction in...

Epithelial ovarian cancer (EOC) remains the most lethal gynecological malignancy, and accurate prediction of long-term survival remains challenging. This study aimed to evaluate the prognostic associations of prespecified clinically relevant factors with overall survival (OS) and to develop and internally validate a KELIM-integrated prognostic nomogram.

A total of 100 patients with EOC treated at the Second Hospital of Hebei Medical University between May 2014 and June 2021 were retrospectively analyzed. Five clinically relevant predictors were prespecified for multivariable modeling: age, histological subtype, FIGO stage, postoperative residual disease, and KELIM. The primary multivariable analysis included 92 patients with complete data for all five predictors. Internal validation was performed using 1,000 bootstrap resamples.

During a median follow-up of 87 months (95% CI, 65-91 months), 60 deaths occurred, and the Kaplan-Meier-estimated median OS was 57 months (95% CI, 44-70 months). Increasing age, advanced FIGO stage, KELIM <1, and gross postoperative residual disease were independently associated with poorer OS, whereas histological subtype was not independently associated with OS. The optimism-corrected C-index was 0.779. The optimism-corrected time-dependent AUCs for predicting 1-, 3-, and 5-year OS were 0.917, 0.865, and 0.799, respectively, compared with 0.699, 0.733, and 0.679 for FIGO stage alone.

Integrating KELIM with established clinical and surgical prognostic factors may enhance individualized survival prediction and risk stratification in patients with EOC. External validation in independent cohorts is required before clinical implementation.