Comparison between EUS-guided fine-needle aspiration cytology and EUS-guided fine-needle biopsy histology for the evaluation of pancreatic neuroendocrine tumors

Background/objectives: Studies comparing EUS-guided fine-needle aspiration (EUS-FNA) with EUS-guided fine-needle biopsy (EUS-FNB) for the evaluation of pancreatic neuroendocrine tumors (pNETs) are lacking. We aimed at comparing EUS-FNA with EUS-FNB in terms of Ki-67 proliferative index (PI) estimation capability, cellularity of the samples, and reliability of Ki-67 PI/tumor grading compared with surgical specimens. Methods: Patients diagnosed with pNETs on EUS and/or surgical specimens were retrospectively identified. Specimens were re-evaluated to assess Ki-67 PI feasibility, sample cellularity by manual counting, and determination of Ki-67 PI value. Outcomes in the EUS-FNA and EUS-FNB groups were compared. Kendall rank test was used for Ki-67 PI correlation between EUS and surgical specimens. Subgroup analysis including small (≤20 mm), non-functioning pNETs was performed. Results: Three-hundred samples from 292 lesions were evaluated: 69 EUS-FNA cytology and 231 EUS-FNB histology. Ki-67 PI feasibility was similar for EUS-FNA and EUS-FNB (91.3% vs. 95.7%, p = 0.15), while EUS-FNB performed significantly better in the subgroup of 179 small pNETs (88.2% vs. 96.1%, p = 0.04). Rate of poor cellulated (<500 cells) specimens was equal between EUS-FNA and EUS-FNB. A significant correlation for Ki-67 PI values between EUS and 92 correspondent surgical specimens was found in both groups, but it was stronger with EUS-FNB (tau = 0.626, p < 0.0001 vs. tau = 0.452, p = 0.031). Correct grading estimation was comparable between the two groups (p = 0.482). Conclusion: Our study showed stronger correlation for Ki-67 values between EUS-FNB and surgical specimens, and that EUS-FNB outperformed EUS-FNA in the evaluation of small pNETs. EUS-FNB should become standard of care for grading assessment of suspected pNETs.