Patient survival prediction in locally advanced cervical squamous cell carcinoma using MRI-based radiomics: retrospective cohort study

Abstract

Cervical cancer is a major health concern for women, ranking as the fourth most common cancer and a significant cause of cancer-related deaths worldwide. To enhance prognostic predictions for locally advanced cervical squamous cell carcinoma, we conducted a study utilizing radiomics features extracted from pretreatment magnetic resonance images. The goal was to predict patient survival and compare the predictive value of these features with clinical traits and the 2018 International Federation of Obstetrics and Gynecology (FIGO) staging system. In our retrospective cohort study, we included 500 patients with confirmed cervical squamous cell carcinoma ranging from FIGO stages IIB to IVA under the 2018 staging system. All patients underwent pelvic MRI with diffusion-weighted imaging before receiving definitive curative concurrent chemoradiotherapy. The results showed that the combination model, incorporating radiomics scores and clinical traits, demonstrated superior predictive accuracy compared to the widely used 2018 FIGO staging system for both progression-free and overall survival. Age was identified as a significant factor influencing survival outcomes. Additionally, primary tumour invasion stage, tumour maximal diameter, and the location of lymph node metastasis were found to be important predictors of progression-free survival, while primary tumour invasion stage and lymph node metastasis position individually affected overall survival. During the follow-up period, a portion of patients experienced disease-related deaths or tumour progression/recurrence in both sets. The radiomics-score significantly enhanced prediction ability, providing valuable insights for guiding personalized therapy approaches and stratifying patients into low-risk and high-risk categories for progression-free and overall survival. In conclusion, our study demonstrated the potential of radiomics features as a valuable addition to existing clinical tools like the FIGO staging system, offering promising advancements in managing locally advanced cervical squamous cell carcinoma.