Pretreatment Staging FDG PET-Derived Radiomic Score Predicts Progression-Free Survival in Locally Advanced Rectal...

Area of Science:

• Oncology
• Radiology
• Medical Imaging

Background:

• Clinical outcomes for locally advanced rectal cancer (LARC) treated with total neoadjuvant therapy (TNT) are inconsistent.
• Existing clinicopathological factors have limited accuracy in predicting prognosis for LARC patients undergoing TNT.

Purpose of the Study:

• To develop a radiomic score (RadScore) using baseline 18F-FDG PET/CT scans.
• To assess the RadScore's independent predictive ability for progression-free survival (PFS) in LARC patients treated with TNT and surgery.

Main Methods:

• Retrospective analysis of 85 LARC patients treated with TNT and surgery.
• Extraction of 140 radiomic features from pretreatment 18F-FDG PET/CT images following IBSI guidelines.
• Development of the RadScore using LASSO-penalized Cox regression for feature selection and internal validation via bootstrap resampling.

Main Results:

• The final RadScore, based on Intensity Histogram Kurtosis and GLCM Difference Entropy, independently predicted PFS (HR = 1.81, p = 0.003).
• The predictive model achieved a C-index of 0.674 (bootstrap-corrected: 0.650).
• Patients in the high-risk group identified by the RadScore had significantly shorter PFS (log-rank p = 0.006).

Conclusions:

• A pretreatment FDG PET-derived RadScore serves as an independent predictor of PFS in LARC patients within a TNT treatment framework.
• This RadScore shows potential as an imaging biomarker for preoperative risk stratification.
• The findings support the use of this RadScore for individualized treatment planning in LARC patients receiving TNT.