A Radiomic MRI Predictive Model for Response to Concomitant Chemoradiotherapy in Locally Advanced Cervical Cancer

Cervical cancer is an invasive cancer that develops from the squamous epithelium of the cervix. Worldwide, cervical cancer is the fourth most common cancer in women, with approximately 604,000 new cases in 2020.Treatment for locally advanced cervical cancer (FIGO stage IB3 to IVA) is based on a combination of radiotherapy and chemotherapy (cisplatin 40 mg/m2 x5 or 6 or carboplatin area under the curve 2 if cisplatin is contraindicated). Responses to concomitant chemoradiotherapy remain highly heterogeneous from one woman to another, and predicting the response to these treatments would allow early consideration of alternative therapies for patients identified as less responsive to standard treatments. Indeed, 5-year recurrence-free survival is approximately 79% for stages IB and IIA and 59% for stages III and IVA, with approximately 36% of local failures despite chemoradiotherapy.

The radiomic MRI approach in locally advanced cervical cancers has shown in a few studies to be prognostic for locoregional recurrence or survival. However, these models still need to be explored and validated before they can be implemented in routine clinical practice.

The EPICOL cohort is a clinical-biological cohort of 136 patients treated with chemoradiotherapy for locally advanced cervical cancer at the Montpellier Cancer Institute or Nîmes University Hospital.

The aim is to develop a predictive model of response to chemoradiotherapy based on radiomic data from pelvic MRIs before and after treatment from the EPICOL cohort.

Inclusion criteria

• Patients treated with exclusive radio-chemotherapy for locally advanced cervical cancer (stage Ib-IVb according to the FIGO classification).
• Patients with a minimum of 2 years of post-treatment follow-up.
• Patients for whom the initial biopsy specimen (prior to treatment) is available.
• Patients who have not expressed their opposition to participating in the study.
• Patients who are affiliated with or beneficiaries of a health insurance plan.

Exclusion criteria

• Patients under judicial protection, guardianship, or curatorship