3569 - Pre-Treatment and Pre-Brachytherapy MRI First-Order Radiomic Features By a Commercial Software As Survival Predictors in Radiotherapy for Cervical Cancer

W. Sittiwong¹, P. Dankulchai¹, P. Wongsuwan¹, T. Prasartseree¹, W. Thaweerat¹, P. Tuntapakul¹, and N. Thornsri²; ¹Division of Radiation Oncology, Department of Radiology, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand, ²Research Unit, Faculty of Medicine Siriraj hospital, Mahidol University, Bangkok, Thailand

Purpose/Objective(s): This study aimed to investigate the performance of prediction models from pre-treatment and pre-brachytherapy MR radiomics for clinical factors (CF), radiomic features (RF), and combined clinical factors with radiomic features (CF+RF) to measure treatment outcomes in patients with locally advanced cervical cancer (LACC) after concurrent chemoradiation (CCRT) and 3D image-guided adaptive brachytherapy (3D-IGABT). Materials/

Methods: The study included 100 patients with LACC who underwent definitive CCRT with IMRT/VMAT technique followed by 3D-IGABT. MRI-based contouring included T2WI and DWI images for primary tumor (GTVp) and lymph nodes (GTVn). The contours were imported to commercially available software to extract first-order radiomic features. Radiomic values from pre-treatment (PreRx), pre-brachytherapy (PreBT), differences between PreRx and PreBT (Diff) radiomic and clinical factors were analyzed using univariate and multivariate Cox regression analysis. Predictive models of CF alone, RF alone and CF+RF were created to predict progression-free survival (PFS), local recurrence-free survival (LRFS), distant metastasis-free survival (DMFS), and overall survival (OS). The performance and the stability of the models were compared by Harrel’s C-index and optimism index, respectively. The final models of PFS and DMFS were adopted to create the proposed nomogram.
Results: The median follow-up time was 24.5 months. Most patients were staged IIIC1 and IIIC2 (73%) and the common histology was squamous cell carcinoma (72%). The mean dose of D90 HR-CTV was 89.94 Gy (EQD2₁₀). The 2-year of PFS, LRFS, DMFS, and OS rates were 71, 88.6, 83.1, and 83.5%, respectively. For the clinical outcomes, CF+RF from PreRx and PreBT resulted in the highest Harrell’s C-index compared with the CF or RF alone and from Diff models. The Harrell’s C-indices from the CF+RF model from PreRx and PreBT for PFS, LRFS, DMFS, and OS were 0.739, 0.873, 0.830 and 0.967, respectively. The optimism indices for PFS, LRFS, DMFS, and OS were 0.438, 0.472, 0.292 and 0.430, respectively. The proposed nomogram of PFS and DMFS consisted of factors from clinical factors, PreRX and PreBT radiomic features.
Conclusion: Radiomic features from the first-order statistics added values to clinical factors to predict the outcomes after CCRT. The highest prediction model performance was for the combined clinical and radiomics from PreRx and PreBT. The proposed nomogram could be a potential predictor to personalize treatment after CCRT for patients with a high risk of progression and distant relapse. However, external validation is needed to confirm the robustness of the nomogram.