OC19.01: Prediction of outcome in endometrial cancer improved by combining traditional with sonographic and demographic parameters: IETA4 cohort follow‐up

Résumé fourni par la source

To develop and validate pre- and postoperative mathematical models to predict recurrence/progression of endometrial cancer, and to evaluate if these models outperform the ESMO/ESGO/ESTRO prognostic risk groups. All 1538 women in the prospective IETA4 cohort who underwent preoperative structured ultrasound examination followed by hysterectomy from 2011 to 2015, were followed up until 2019-2020 or to time of death. The ESMO/ESGO/ESTRO pre- and postoperative risk groups (based on myometrial invasion, grade, stage, LVSI) were forced into the respective model. Based on scientific knowledge we also included age, BMI, extrauterine tumour spread on ultrasound, colour score, multiple-multifocal vascular pattern, endometrial-myometrial border, endometrial echogenicity and tumour AP diameter as candidate variables. A multivariable fractional polynomial algorithm was run to fit a Fine-and-Gray competing risk model, with death from other causes as a competing event. Calibration plots were generated, and Harrell's C discrimination index was calculated. Follow-up data was complete in 1429/1538 (92.9%) patients. Recurrence/progression occurred in 228 (15.9%) and death from other causes in 121 (8.5%), 659 (46%) were low-risk. A preoperative model including the ESMO/ESGO/ESTRO preoperative risk group and the retained variables extrauterine spread on ultrasound, non-uniform echogenicity, age ≥65 years, and tumour AP diameter > 2 cm was superior to ESMO/ESGO/ESTRO preoperative risk group alone, Harrell's C 0.761 (95% CI 0.727-0.795) vs. 0.706 (95% CI 0.674-0.739). The postoperative model including the retained variables age > 65 years and tumour AP diameter > 2 cm, improved Harrell's C from 0.758 (0.723-0.792) to 0.785 (0.754-0.816) compared to including only the 2020 postoperative ESMO/ESGO/ESTRO risks group. Adding sonographic and demographic variables to the ESMO/ESGO/ESTRO prognostic groups improved risk prediction and may be valuable to further individualise patient management.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.