Application of radiomics as a risk stratification tool in node positive penile cancer.

• DOI: 10.1200/jco.2025.43.5_suppl.3
• Publication Date: 2025-02-18T14:30:51Z
• Authors: Adnan Nazir Fazili; Brian Morse; Abraham Ahmed; Jongphil Kim; Daniel Jeong; Philippe E. Spiess; Jad Chahoud
• Keywords: Risk Stratification

3 Background: The risk level of lymph node metastasis in penile cancer is determined using the Graafland criteria which determines how patients are treated. These imaging-based have critical implications on patient care yet not been extensively validated. purpose this study was to validate and assess use radiomics analysis CT scans standardize stratification. Methods: Thirty-eight with squamous cell prior regional lymphadenectomy were included retrospective cohort. Patients chemotherapy for PSCC or any inguinal pelvic radiotherapy excluded. All performed IV contrast axial slice thickness 3mm. analyzed by two experienced oncologic radiologists criteria. largest each segmented features extracted compared pathologic evidence high-risk disease. 308 radiomic identified. Features poor predictive value (i.e. AUC <0.5) Pearson’s correlation coefficient used find clusters (absolute ≥0.95), most feature cluster selected.199 best selected analysis. Two logistic regressions stepwise selection LASSO approach. Results: In 38 subjects, 16 identified as harboring surgical pathology. Applying had an inter-reader concordance 89.5% (34/38), accuracy 86.8% (95%CI: 71.9% - 95.6%); AUC: 0.869 (95% CI: 0.758– 0.981); Sensitivity 87.5%; Specificity 86.4% predicting yielded a model including F54 (compactness) F279 (3D wavelet_P1_L2_C1) multi-order texture-based feature. regression (F54 F279) (AUC = 0.921 ± 0.096) after five-fold cross validation. Conclusions: accurately predicted presence nodal disease pre-operative when assessed expert radiologists. Radiomic shows promise further refining standardizing detection advantage ease training and/or clinical application.