Radiomics-Based Prognostication in Primary Sclerosing Cholangitis: A Proof-of-Concept Study

Area of Science:

• Radiology
• Medical Imaging
• Quantitative Biomarkers

Background:

• Primary Sclerosing Cholangitis (PSC) risk assessment using MRI is limited by semi-quantitative methods and interpretation variability.
• Radiomics offers a potential quantitative alternative for objective risk stratification in PSC.
• This study focuses on developing and validating MRI-derived radiomic features for high-risk PSC patient identification.

Purpose of the Study:

• To explore and validate MRI-derived radiomic features for identifying high-risk patients with primary sclerosing cholangitis (PSC).
• To assess the prognostic value of radiomic features for predicting clinical outcomes in PSC.
• To establish a quantitative, standardized, and cost-free biomarker for PSC risk stratification.

Main Methods:

• Prospective recruitment of PSC patients undergoing gadoxetate disodium-enhanced MRI.
• Extraction of whole liver parenchyma radiomic features using PyRadiomics, adhering to IBSI standards.
• Risk categorization based on Mayo Risk Score (MRS) and liver stiffness measurement (LSM), with validation in an independent cohort.
• Survival analysis using Cox regression to evaluate prognostic value for clinical events.

Main Results:

• Five radiomic features were initially associated with high risk in the training cohort.
• GLRLM-Run Entropy in fat-saturation T2 weighted imaging (FS-T2W) was validated as a significant predictor for both MRS and LSM.
• This feature demonstrated excellent predictive performance for clinical outcomes (C-index = 0.857) and prognostic value (HR = 1.480).

Conclusions:

• GLRLM-Run Entropy in FS-T2W MRI is a novel, quantitative radiomics biomarker for PSC risk stratification.
• This feature is standardized, easy to compute, and cost-free, presenting a significant advancement in PSC radiology.
• It holds potential as a key innovation for improving radiology-based biomarkers in PSC management.