2042 - DECT-Derived Quantitative Parameters and Clinical Features for Predicting Prognosis in Patients with Locally Advanced Esophageal Squamous Cell Carcinoma after Neoadjuvant Chemoradiotherapy
Presenter(s)
Y. Cui; Department of Radiation Oncology, Shandong Cancer Hospital and Institute, Shandong First Medical University and Shandong Academy of Medical Sciences, Jinan, China
Purpose/Objective(s): This study aims to identify dual-energy CT (DECT) derived quantitative imaging biomarkers and clinical features for predicting the overall survival (OS) of locally advanced esophageal squamous cell carcinoma (LA-ESCC) undergoing neoadjuvant chemoradiotherapy (nCRT).
Materials/Methods: We retrospectively recruited consecutive LA-ESCC patients who underwent surgery following nCRT between December 2019 and January 2023. Pretreatment DECT scans were used to extract quantitative parameters from virtual monoenergetic images (VMI) 40kev, 70kev, 100kev; virtual non-contrast images (VNC), effective atomic number maps, electron density maps, and iodine maps. Clinical characteristics were recorded. Arterial enhancement fraction (AEF) and extracellular volume (ECV) were also calculated. Pathological complete response (pCR) was defined as absence of residual tumor cells in both primary tumor and lymph nodes. Overall survival was recorded during follow-up. Continuous variables were Z-score standardized. Univariate Cox proportional hazards analysis was performed to identify candidate prognostic biomarkers. Variables passing univariate screening were evaluated for multicollinearity using variance inflation factor (VIF), and eligible variables were entered into multivariable Cox regression to determine independent prognostic biomarkers. Prognostic discrimination was assessed using time-dependent ROC analysis and concordance index. Kaplan-Meier analysis was performed for risk stratification.
Results: A total of 53 patients with pretreatment DECT were included, with a median OS of 34 months. Univariate analysis identified pCR, AEF, and D-HU as candidate predictors. All variables demonstrated acceptable multicollinearity (VIF < 5). In multivariable analysis (selection threshold p < 0.1), all three variables were retained: pCR (p < 0.001, HR = 0.258), D-HU (p = 0.033, HR = 1.986), and AEF (p = 0.070, HR = 1.603). The Cox-based prediction model achieved an AUC of 0.848 for 1-year OS and 0.776 for 3-year OS. Kaplan-Meier analysis demonstrated significant stratification between high- and low-risk groups (p = 0.0016).
Conclusion: pCR is a key clinical determinant of survival in LA-ESCC. Quantitative DECT-derived parameters, particularly AEF and D-HU, also serve as equally important and complementary prognostic biomarkers.