A.I and Machine Learning Based Risk Prediction Model to Improve the Clinical Management of Endometrial Cancer.

Recruiting

• Conditions: Endometrium Cancer

• Registration Number: NCT06841653
• Lead Sponsor: Regina Elena Cancer Institute

Brief Summary

Prediction of preoperative endometrial biopsy: the evolution from hyperplasia to cancer, the prognosis and the risk of recurrence. Intelligence methods artificial risk will be used to redefine the current risk classes including our profile immuno-mutational to provide a more precise characterization and closer to the real prognosis of the patient.

Detailed Description

Identify new risk factors for endometrial cancer, using an integrated multi-omics approach linked to a specific immune pattern (called MOMIMIC score) useful for improving oncology and surgery precision. The aim is to evaluate the predictive value of the MOMIMIC score for early identification of progression from precancerous lesions to endometrial carcinoma, prognosis and relapses, to help the clinician in the decision to treatments. Through the identification during hysteroscopy of the most appropriate site for biopsies targeted endometrials, through an artificial intelligence algorithm applied to the video system hysteroscopic which, by comparing the information from the omics approach and the hysteroscopic image combined with radiogenomic information, it could help the gynecologist in the procedure and provide information on the prognosis through the omics-iconographic profile in order to calculate a preoperative predictive score. Furthermore by modulating the surgical radicality, according to the information obtained, there will be a tendency to preserve fertility in young patients with a low-risk profile (since currently the risk factors are not sufficient to discriminate for a non-treatment radical). This will help the surgeon through an artificial intelligence algorithm applied to the system robotic/laparoscopic video, will guide the operator in decision-making procedures regarding the resection margins tumor, metastasis localization, pathological lymph node detection, and imaging driven by biomolecular information.

Study Timeline

• Study Start: 2024-06-20
• Status Verified: 2025-02
• Study First Submitted: 2025-02-18
• Study First Submitted QC: 2025-02-18
• Last Update Submitted: 2025-02-18
• Study First Posted: 2025-02-24
• Last Update Posted: 2025-02-24
• Primary Completion: 2026-06-20
• Study Completion: 2026-06-20

Recruitment & Eligibility

• Status: RECRUITING
• Sex: Female
• Target Recruitment: 40

Inclusion Criteria

• Age > 18 years;
• Histological diagnosis of endometrial hyperplasia, endometrioid adenocarcinoma of the endometrium, healthy endometrium in patients undergoing total hysterectomy for benign extra-endometrial disease;
• Written informed consent (to the study and data processing), for the party's patients only prospective and/or in follow-up) For the retrospective cohort: availability of samples adequately stored at the biobank of the Institute and availability of data relating to follow-up (at least 2 years)

Exclusion Criteria

All exclusion criteria adopted in the surgical protocols will be applied to the study. In particular:

• Comorbidities not controlled with adequate medical therapy;
• Infections of the endometrial cavity (pyometra);
• Synchronous cancer;
• Neoadjuvant treatments;
• Previous radiotherapy treatments of the pelvic region;
• Hormone therapies.

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Primary Outcome Measures

Name	Time	Method
OS (overall survival)	24 months	The study will evaluate the predictive value of the MultiOMics-IMmune-Iconographic model (global mutational profiling, RNA-seq of single cells coupled with the Spatial transcriptomics, proteomic and metabolomic profile) following the data obtained from the identification of new risk factors for endometrial carcinoma, in patients at high or low risk. They will be tested from Random Survival Forest to determine how capable a feature is discriminate between the 4 groups in terms of OS (overall survival). The selected features will be used in combination with the known prognostic clinical and histopathological risk factors described by ESMO-ESGO-ESTRO.
DSF (disease-free survival)	24 months	The study will evaluate the predictive value of the MultiOMics-IMmune-Iconographic model (global mutational profiling, RNA-seq of single cells coupled with the Spatial transcriptomics, proteomic and metabolomic profile) following the data obtained from the identification of new risk factors for endometrial carcinoma, in patients at high or low risk.; They will be tested via Random Survival Forest to determine how capable a feature is discriminate between the 4 groups in terms of impact on progression to cancer, recurrence, DFS (disease-free survival).

Secondary Outcome Measures

Name	Time	Method
Area under the curve (AUC)	24 months	In order to obtain a more robust estimate of accuracy of the MultiOMics-IMmune predictive signature, for validation, we will use two groups of patients composed of a minimum of 200 cases (100 high risk and 100 low risk), at a reduction from 30% confidence interval to 95% when signature performance are kept constant. Considering the area under the curve (AUC).
Accuracy (ACC)	24 months	In order to obtain a more robust estimate of accuracy of the MultiOMics-IMmune predictive signature, for validation, we will use two groups of patients composed of a minimum of 200 cases (100 high risk and 100 low risk), at a reduction from 30% confidence interval to 95% when signature performance are kept constant. Considering Accuracy (ACC).

Trial Locations

Locations (1)

IRCCS National Cancer Institute "Regina Elena"; 🇮🇹 Rome, Italy