Clinical Scorecard: Predicting Surgical Intervention in Ulcerative Colitis Resistant to Pharmacological Treatment Using Radiomics
At a Glance
Category
Detail
Condition
Ulcerative colitis (UC), an inflammatory bowel disease causing autoimmune inflammation of the colon
Key Mechanisms
Chronic and recurrent inflammation with relapse and remission; radiomics extracts high-dimensional imaging features from CT to predict treatment outcomes
Target Population
Patients hospitalized for relapse of UC with moderate to severe disease requiring CT imaging
Care Setting
Hospital inpatient setting during relapse requiring assessment for medical versus surgical treatment
Key Highlights
8–24% of UC patients still require surgery despite advances in drug therapy
Radiomics applied to admission CT images can predict the need for surgery versus medical treatment
Machine learning with radiomics features combined with clinical data improves prediction accuracy
Guideline-Based Recommendations
Diagnosis
Endoscopic diagnosis by IBD specialists to confirm UC
Use of noncontrast pelvic CT imaging at admission for relapse to obtain radiomics data
Management
Initial induction of remission with drug therapy in nonemergent relapse
Consider surgery if refractory to drug therapy, based on clinical, laboratory, and radiomics data
Surgical intervention includes total or subtotal colorectal resection
Monitoring & Follow-up
Clinical assessment of relapse severity using Lichtiger score and extent of inflammation
Monitoring response to drug therapy during hospitalization
Use of radiomics score and nomogram integrating clinical factors to predict surgical need
Risks
Emergency surgery indicated for perforation, life-threatening bleeding, or toxic megacolon
Surgery carries risks inherent to colorectal resection and should be individualized
Patient & Prescribing Data
Patients hospitalized for UC relapse with moderate to severe disease
Medication regimens include calcineurin inhibitors, anti-TNF agents, vedolizumab, and ustekinumab; choice and dosing individualized by IBD specialists
Clinical Best Practices
Perform CT imaging within 3 days prior to or 1 day after admission for relapse
Manual segmentation of rectal wall on axial CT images for radiomics feature extraction
Use LASSO regression with cross-validation for radiomics feature selection
Integrate radiomics score with clinical factors in a nomogram to guide surgical decision-making
Consider patient preference and clinical context when deciding on surgery
