Endoscopic Ultrasound in the Diagnosis of Pancreatic Neuroendocrine Neoplasms

(adapted from: Tacelli M et al. “Pancreatic Neuroendocrine Neoplasms: Classification and Novel Role of Endoscopic Ultrasound in Diagnosis and Treatment Personalization”. United European Gastroenterol J. 2025,13:34-43)

Endoscopic ultrasound (EUS) is an advanced diagnostic technique that integrates endoscopy with high-frequency ultrasound imaging. This dual-modality approach allows for a detailed evaluation of the anatomical structures adjacent to the upper gastrointestinal tract—ranging from the esophagus to the duodenum—including the pancreas, biliary system, gallbladder, gastrointestinal wall layers, left liver lobe, splenic hilum, aorta, left kidney, and adrenal gland. EUS plays a central role in the evaluation of the pancreatic parenchyma and ductal system, which due to its deep retroperitoneal location, has been a difficult organ to study by conventional transabdominal ultrasound [1,2,3].

A multimodal diagnostic approach is recommended for pancreatic neuroendocrine neoplasms (pNENs) [4], that includes computed tomography (CT), magnetic resonance imaging, somatostatin receptor imaging, and EUS. However, lacking specific indications regarding the sequence of tests and whether all should be performed in every case, CT and EUS are often considered diagnostically equivalent. Multiple studies have shown that EUS outperforms CT, particularly in identifying small pNENs. One study demonstrated that CT failed to detect 68.4% of lesions under 10 mm and 15% of those under 20 mm, compared to EUS [5]. A meta-analysis estimated an incremental diagnostic yield of approximately 26% for EUS over CT in this setting [6].

Morphological Features and Ancillary Techniques

On B-mode EUS, pNENs may present either as solid masses or with features of central cystic degeneration. Well-differentiated, low-grade pNENs usually appear as homogeneous, hypoechoic, oval-shaped lesions with well-defined margins and a clear “space-occupying” pattern. In contrast, poorly differentiated or high-grade tumors tend to display a heterogeneous texture, irregular margins, internal necrosis, and an infiltrative growth pattern [7].

EUS can be enhanced by adjunct imaging techniques:

– EUS elastography allows real-time measurement of tissue stiffness, with a reported sensitivity of 100% and specificity of 88% in differentiating pancreatic adenocarcinoma from pNENs [8]. However, elastography is highly operator-dependent.
– Contrast-enhanced EUS is a non-invasive technique for evaluating tumor vascularity. Well-differentiated pNENs typically display early hyperenhancement with rapid washout, reflecting high arteriolar vascularity [9–11]. Poorly differentiated pNENs exhibit low microvascular density with sparse vascular networks, and tend to appear hypoenhancing in the arterial phase due to reduced perfusion and contrast uptake. Such imaging features may also correlate with higher Ki-67 index and the presence of necrotic or infiltrative changes on standard EUS.

Tissue Acquisition: Biopsy for Diagnosis and Grading

Despite a EUS-FNA sensitivity of 84.5% and a specificity of 99.4% in the diagnosis of pNENs, sampling failure may still occur in specific clinical scenarios. Notably, tumors located in the pancreatic head – due to their anatomical position and more challenging needle access – and those exhibiting significant intratumoral fibrosis (≥30%) have been independently associated with a reduced diagnostic yield [12]. Fibrotic stroma can limit needle penetration and may reduce the proportion of neoplastic cells in the sample, compromising both diagnostic and grading accuracy.
Several studies assessing the concordance between EUS-FNA/FNB grading and final histology of surgical specimens revealed heterogeneous results across different institutions and patient cohorts. A comprehensive meta-analysis [13] from 2022 found an overall concordance rate of 80.3%, and a notable predominance of undergrading (14.7%) compared to overgrading (3.5%, p < .001), suggesting a systematic risk of underestimating tumor aggressiveness in preoperative settings. The single most influential factor associated with grading mismatches was tumor size, with smaller lesions being more prone to sampling error due to limited material and potential intratumoral heterogeneity, which can lead to inaccurate assessment of the Ki-67 proliferation index. Emerging Applications and Future Perspectives

EUS-guided biopsies are increasingly used for molecular characterization and therapeutic stratification. For example, DAXX/ATRX expression and ALT activation, both linked to tumor aggressiveness, can now be evaluated preoperatively via EUS-guided tissue samples [14].
A recent paper has shown the feasibility of transcriptomic analysis from EUS-FNB samples, initially in PDAC [15] and now being investigated in pNENs as well [16].

Finally, the development of artificial intelligence tools capable of predicting pNEN behavior from EUS imaging is underway [17]. These models aim to incorporate morphologic, vascular, and potentially genomic data to support clinical decision-making with greater precision.