Preoperative grading of pancreatic neuroendocrine tumors (pNETs) plays a pivotal role in guiding surgical and oncological management. Endoscopic ultrasound-guided tissue acquisition (EUS-TA) enables cytological and histological assessment through fine-needle aspiration (FNA) and fine-needle biopsy (FNB), both currently recommended for tissue sampling. However, their comparative performance in achieving accurate grading consistent with surgical specimens remains unclear.

This study aimed to evaluate the performance of EUS-guided tissue acquisition (EUS-TA) for the preoperative characterization and grading of pancreatic neuroendocrine tumors (pNETs), comparing the relative performance and safety of FNA and FNB to the postoperative histological gold standard.

We retrospectively analyzed 63 patients with histologically confirmed pNETs who underwent preoperative EUS-guided tissue acquisition (FNA or FNB) and subsequent surgical resection at the S. Orsola-Malpighi University Hospital in Bologna (2010–2024). Demographic, clinical, radiological, and procedural data were collected, including tumor size, site, morphology, sampling technique and adverse events. Cytological (FNA) and histological (FNB) samples were processed and graded according to WHO 2022 criteria. Diagnostic yield, accuracy, and grading adequacy were assessed. Multivariable logistic regression identified predictors of grading adequacy (sex, age, site, morphology, size). Concordance between preoperative and surgical grading was measured as percent agreement, Cohen’s kappa, and Gwet’s AC. Statistical significance was set at p<0.05.

Among 63 resected pNETs (19 FNA, 44 FNB), overall diagnostic yield was 95.2% (60/63) and diagnostic accuracy 98.3% (59/60). Grading was achievable in 84.1% of cases, with 93.2% for FNB and 63.2% for FNA (OR 7.97; 95% CI 1.78–35.65; p=0.006). Lesion size ≥20 mm independently predicted grading adequacy (OR 8.36, 95% CI 1.10–63.6; p=0.040). Overall concordance between preoperative and surgical grading was 69.8% (κ = 0.44; Gwet’s AC = 0.59), similar for FNA (75.0%) and FNB (68.3%). Misgrading occurred in 30.2% (undergrading 24.5%, overgrading 5.7%). In lesions <20 mm, concordance was 75.0%. Safety was favorable: minor bleeding 15.9%, mild pancreatitis 1.6%, no severe events.

Our findings confirm the high diagnostic accuracy of EUS-guided sampling in pNETs and highlight the significant advantage of FNB in obtaining histological material suitable for grading compared to FNA. Larger lesions (>20 mm) favored grading adequacy, while overall concordance with surgical grading was moderate (69.8%). FNB therefore appears to be the most reliable method for preoperative grading of pNETs.