Preoperative risk stratification in endometrial carcinoma: a pragmatic, complementary nomogram using readily...

Preoperative differentiation between low-risk and non-low-risk endometrial carcinoma (EC) remains suboptimal using conventional assessment (endometrial biopsy combined with MRI); making this distinction is critical for tailoring surgery appropriately. This study aimed to develop a nomogram integrating conventional assessment with readily available variables, as a complementary tool to support risk reclassification and individualized surgical planning in selected clinical scenarios.

A total of 251 patients with primary EC were retrospectively enrolled and divided into training (n = 179) and validation (n = 72) cohorts according to whether the last preoperative transvaginal ultrasound was performed at our institution or at external facilities. A nomogram was constructed using least absolute shrinkage and selection operator regression and multivariable logistic analysis. Model validation included 1,000 bootstrap resamples for optimism correction and Hosmer-Lemeshow testing for calibration, followed by internal validation in the independent validation cohort. Model performance was evaluated using the area under the curve (AUC), calibration slope, Brier score, and decision curve analysis (DCA). The ability of the nomogram to improve risk reclassification beyond conventional assessment was assessed with continuous net reclassification improvement (NRI) and integrated discrimination improvement (IDI).

The nomogram achieved an AUC of 0.902 (95% CI: 0.857-0.948) in the training cohort, with an optimism-corrected AUC of 0.892 after 1,000 bootstrap resamples. In the validation cohort, the AUC was 0.866 (95% CI: 0.773-0.959), calibration was satisfactory (slope, 0.810; Brier score, 0.142; Hosmer-Lemeshow P = 0.156), and the nomogram demonstrated comparable or superior net benefit to conventional assessment across the majority of threshold probabilities, and showed improved risk reclassification (continuous NRI, 1.189, P = 0.019; IDI, 0.126, P = 0.013).

This nomogram, constructed from routinely available variables, demonstrated acceptable calibration in the validation cohort, despite a slight tendency toward risk underestimation. While the incremental AUC over conventional assessment was not statistically significant, the model showed favorable net benefit on DCA and improved risk reclassification, suggesting it may serve as a useful adjunct for risk reclassification in specific clinical settings. Whether this translates into measurable improvements in surgical decision-making and patient outcomes awaits confirmation in prospective multicenter studies.