Abstract
BACKGROUND: Given the poor prognosis of patients with pancreatic ductal adenocarcinoma (PDAC), the accurate stratification of patients at high risk for early recurrence (ER) is an urgent need. Conventional predictors such as carbohydrate antigen 19-9 (CA19-9) and tumor diameter have suboptimal efficacy. Quantitative parameters derived from dual-energy computed tomography (DECT) and the (18)F-fluorodeoxyglucose positron emission tomography-computed tomography ((18)F-FDG-PET/CT) serve as validated imaging biomarkers for aggressive tumor biology. This study aimed to develop an integrative nomogram that combines these imaging markers with clinicopathological factors to preoperatively predict ER in resectable PDAC. METHODS: In this single-center retrospective study, we analyzed 80 patients diagnosed with pathologically confirmed PDAC from November 2021 to July 2023. ER was defined as disease relapse within 12 months postoperatively, and patients were categorized into ER and non-early recurrence (non-ER) groups. Clinicopathological variables, including tumor markers, pathological T stage (pTs), pathological N stage (pNs), tumor location, maximum tumor diameter, perineural invasion (PNI), and lymphovascular invasion (LVI), were collected. The following preoperative DECT parameters were obtained: dual-energy index (DEI), effective atomic number (Zeff), electron density (Rho), fat fraction, iodine concentration (IC), normalized iodine concentration (NIC), and vascular involvement. The maximum standardized uptake value (SUV(max)) values were extracted from the PET/CT images. Univariate and multivariate logistic regression analyses were employed to identify independent clinicopathologic and imaging predictors of early postoperative recurrence, and a nomogram was subsequently constructed. The discrimination, calibration, and clinical utility of the nomogram were evaluated via receiver operating characteristic (ROC) curves, calibration curves, and a decision curve, respectively. RESULTS: Comparative analysis revealed significant differences between the non-ER and ER groups in terms of the maximum tumor diameter, serum CA19-9 level, pNs, LVI, portal-venous-phase (PV-NIC) value, number of veins involved, and SUV(max) (all P values <0.05). Multivariate logistic regression analysis revealed lymph node metastasis [odds ratio (OR) =19.610; 95% confidence interval (CI): 1.211-340.406; P=0.032], a low PV-NIC value (OR =0.769; 95% CI: 0.617-0.945; P=0.028), a greater number of invaded vessels (OR =8.660; 95% CI: 1.083-110.245; P=0.043), and an elevated SUV(max) (OR =1.739; 95% CI: 1.091-4.142; P=0.027) as independent predictors of ER in patients with PDAC. The comprehensive model achieved an area under the curve of 0.979, along with robust calibration (calibration slope =0.91). CONCLUSIONS: The nomogram model based on DECT parameters, the PET/CT SUV(max), and clinicopathological parameters effectively predicted early postoperative recurrence in patients with PDAC.