Complementary Roles of Endoscopic Ultrasound and CT/MRI in Pancreatic Lesions with Histopathological or Follow-Up Confirmation: A Retrospective Clinical and Imaging Analysis
Year: 2026
Type: article
Abstract: Background: Differentiating benign from malignant pancreatic lesions requires integrating clinical, biochemical, and imaging data. We identified factors associated with malignancy and evaluated endoscopic ultrasound (EUS) alongside CT/MRI in pancreatic lesions with a reference diagnosis established by histopathology or clinical–radiological follow-up. Methods: In this single-center retrospective study (Sibiu, Romania; 2022–2025), all patients underwent EUS-guided FNA and CT and/or MRI. The reference standard was histopathology in 193 patients (82.1%) and clinical–radiological follow-up in 42 benign lesions (17.9%). Multivariable logistic regression identified factors associated with malignancy. EUS–CT/MRI agreement was assessed by concordance and Cohen’s kappa; diagnostic performance by sensitivity, specificity, predictive values, and Youden index. Results: Of 235 patients, 47 (20.0%) had malignant lesions. Age, ln-transformed CA 19-9, pancreatic duct dilation (OR 18.97; 95% CI 5.95–60.50), and lymphadenopathy were independently associated with malignancy (apparent AUC 0.956; optimism-corrected AUC 0.949 after internal bootstrap validation; no external validation). EUS visualized 46 lesions not detected on CT/MRI. Agreement was 95.8% for size category (κ = 0.905) and 87.8% for morphology (κ = 0.510). Sensitivity was 80.9% for both modalities; specificity 59.0% (EUS) and 51.6% (CT/MRI); Youden index 0.399 and 0.324. All 38 EUS-solid lesions were malignant (all adenocarcinoma) in this selected referral cohort. Conclusions: EUS detected additional lesions, but both modalities showed limited overall discrimination with comparable sensitivity. Findings support complementary use of EUS and CT/MRI within an integrated clinical assessment; the regression model requires validation. Predictive values apply to this selected referral cohort and may not be transferable to unselected populations.
Source: Life
DOI: 10.3390/life16111769
Language: English
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $3,137
OpenAlex ID: W7223168641