3758 - Prospective Validation of a CT-Based Radiomics Nomogram for the Early Prediction of Malnutrition in Stage IB-IIA Cervical Cancer Patients Receiving Postoperative (Chemo)radiotherapy

Purpose/Objective(s):

Malnutrition, frequently manifested as skeletal muscle loss, is prevalent among cancer patients and is associated with reduced treatment tolerance and unfavorable clinical outcomes. In cervical cancer (CC) patients undergoing postoperative radiotherapy or chemoradiotherapy (PORT/CRT), early identification of nutritional risk using validated tools such as the Patient-Generated Subjective Global Assessment (PG-SGA) is essential to enable timely nutritional interventions. CT-based radiomics has emerged as a non-invasive approach for developing imaging-derived predictors to support clinical decision-making. In a previous study, we constructed and internally validated a radiomics-based nomogram by extracting quantitative features from the psoas muscle at the third lumbar vertebra (L3) level from planning CT scans, in combination with relevant clinical variables. The model demonstrated promising predictive capability for malnutrition defined by PG-SGA scores in patients with FIGO (2014) stage IB–IIA CC treated with PORT/CRT (DOI: 10.3389/fnut.2023.1113588). This prospective study aims to externally validate the predictive accuracy and clinical applicability of this CT-based radiomics nomogram in an independent patient cohort.

Materials/Methods:

Trial design and statistical methods:

This is a prospective, single-center validation study enrolling patients with FIGO (2014) stage IB–IIA CC scheduled to receive PORT/CRT following radical surgery. The key innovative aspect is the prospective external validation of a previously developed CT-based radiomics nomogram. Radiomic features will be extracted from routinely acquired non-contrast planning CT scans, avoiding additional radiation exposure and financial burden. Selected radiomic features from the L3 psoas muscle, together with clinical variables identified in the prior study, will be incorporated into the nomogram to predict malnutrition, with PG-SGA scores serving as the reference standard. Model performance will be evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and calibration. Based on prior internal validation, an anticipated AUC of approximately 0.80 and an outcome prevalence of 39.0% were assumed. The sample size was calculated to ensure a 95% confidence interval half-width of no more than 0.05 for the AUC estimate.

Planned treatment or intervention: No experimental interventions are administered. All patients receive standard PORT/CRT and nutritional management per institutional protocols.

Major eligibility criteria: Eligible patients include those with FIGO (2014) stage IB–IIA CC who have undergone radical surgery and are scheduled for PORT/CRT, with analyzable planning CT images at the L3 level and completed PG-SGA assessment.

Current enrollment status: The study is currently recruiting.

Clinical trial registry number: NCT05709769.

Results: Not available (Trials in Progress).

Conclusion: Not applicable.