Development and validation of a nomogram based on three-dimensional quantitative parameters from dual-layer detector spectral computed tomography for predicting treatment response to neoadjuvant chemotherapy in pancreatic ductal adenocarcinoma.

Background: Neoadjuvant chemotherapy (NAC) can improve margin-negative resection rates and control occult systemic disease in pancreatic ductal adenocarcinoma (PDAC). However, not all patients benefit from NAC, and reliable pretreatment response predictors are lacking. This study aimed to develop and validate a nomogram combining three-dimensional (3D) quantitative parameters from dual-layer detector spectral computed tomography (DLCT) and clinical features for predicting treatment response to NAC in PDAC. Methods: This retrospective study enrolled 184 patients with pathologically confirmed PDAC who received 2-3 cycles of NAC (including FOLFIRINOX, gemcitabine-based, or other platinum-containing regimens) between December 2019 and July 2025. After excluding 34 patients due to absence of baseline DLCT, concurrent second primary malignancies, lack of 8-12-week follow-up imaging, or poor image quality, 150 patients were ultimately included. The patients were randomly divided (7:3) into training and validation cohorts. The pretreatment DLCT-based 3D quantitative parameters and clinical features were reviewed. Univariable and multivariable logistic regressions identified independent predictors of NAC response in the training cohort. The predictive performance of a nomogram integrating key predictors was assessed using the area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). Results: The normalized effective atomic number in the portal venous phase (nZeffPVP) [odds ratio (OR), 1.497; 95% confidence interval (CI): 1.173-1.910; P=0.001], Carbohydrate antigen 19-9 (CA19-9) (OR, 0.999; 0.006; 95% CI: 0.998-1.000; P=0.006), and extracellular volume (ECV) (OR, 0.896; 95% CI: 0.837-0.959; P=0.002) of tumors were identified as independent predictors of NAC response. Incorporating all three independent predictors, the nomogram outperformed both the DLCT model and the clinical model, yielding the highest AUC of 0.839 (95% CI: 0.755-0.904) for the training set and 0.849 (95% CI: 0.711-0.938) for the validation set, respectively. When using the optimal cut-off value of the nomogram, sensitivity, specificity, PPV, and NPV for predicting the response group were 69.4%, 87.0%, 73.5%, and 84.5% in the training set, and 93.8%, 69.0%, 62.5%, and 95.2% in the validation set, respectively. Conclusions: The nomogram that integrates nZeffPVP, CA19-9, and ECV offers a promising approach for predicting treatment response to NAC in patients with PDAC. Following large‑scale, multicenter external validation, the nomogram may serve as an adjunctive tool to inform clinical decision-making and risk stratification.