CA19-9 remains the most widely used serum biomarker for the assessment of pancreatic tumors, particularly pancreatic ductal adenocarcinoma (PDAC). Its levels are influenced by tumor burden, cholestasis, inflammation and secretor status, which limits its diagnostic specificity. Meanwhile, endoscopic ultrasound (EUS) provides unmatched accuracy for characterizing pancreatic masses, allowing detailed assessment of tumor size, echostructure, necrosis, vascular invasion and ductal dilatation. However, the relationship between CA19-9 levels and EUS tumor phenotypes remains poorly defined. Understanding this association may help refine diagnostic performance and improve prognostic stratification. The objective of this study was to evaluate the relationship between CA19-9 and the endosonographic characteristics of both primary and metastatic pancreatic tumors in a real-life cohort.

We conducted a monocentric, retrospective observational study including all patients who underwent EUS for suspicion of a pancreatic mass between 2023 and 2025. Clinical data included age, sex, comorbidities and risk factors. EUS variables included tumor location (head/body/tail), maximal diameter, shape, echostructure (homogeneous/heterogeneous), echogenicity, presence of necrosis, vascular infiltration, Wirsung duct dilatation and biliary dilatation. Serum CA19-9 levels were collected at the time of diagnosis. Histology and metastatic status were retrieved when available.

Among the 81 patients, 43 had a fully analysable CA19-9 value and documented metastatic status. Of these, 18 (42%) had metastatic disease and 25 (58%) had localized tumors. Tumors were predominantly located in the pancreatic head. Tumor size ranged from small focal lesions to large infiltrative masses.

CA19-9 levels showed very high variability across patients (range: 2 to 12,000 U/mL). Median CA19-9 did not differ significantly between metastatic and non-metastatic tumors (p = 0.57), reflecting both biological heterogeneity and the confounding effect of biliary obstruction, which was common in tumors of the pancreatic head.

A positive trend was observed between tumor size and CA19-9, although not statistically significant. Tumors with heterogeneous echostructure, intra-tumoral necrosis or vascular infiltration showed higher median CA19-9 levels compared to homogeneous, non-necrotic or non-infiltrative lesions, although again without statistical significance. Tumors located in the pancreatic head displayed the highest CA19-9 levels, likely due to associated cholestasis.

Non-adenocarcinoma tumors (notably neuroendocrine tumors) exhibited very low CA19-9 values, consistent with the known poor sensitivity of the marker in these subtypes.

Taken together, these results illustrate the complex interplay between tumor biology, local aggressiveness assessed by EUS, and CA19-9 expression. They also highlight the limitations of relying solely on CA19-9 for diagnostic or prognostic evaluation, particularly in the presence of biliary obstruction.

In this cohort, CA19-9 levels did not significantly differ between metastatic and non-metastatic pancreatic tumors, but tended to increase in lesions with more aggressive EUS features, including larger size, heterogeneous echostructure, tumor necrosis and vascular infiltration. While CA19-9 remains an imperfect biomarker, its interpretation in conjunction with detailed EUS morphology may enhance diagnostic accuracy and help refine therapeutic decision-making. A larger, prospective multicentric study is warranted to confirm these findings and develop combined EUS–CA19-9 predictive models.