Clinical and oncological outcomes of gastroduodenal neuroendocrine tumours after endoscopic resection in the Western setting - a target trial emulation.

Abstract

Background and aims: Gastroduodenal neuroendocrine tumours (GD-NETs) are increasingly diagnosed. We aimed to (i) evaluate the long-term outcomes following endoscopic resection (ER) of GD-NETs, and (ii) compare the effectiveness of different ER techniques.

Methods: We conducted a multinational retrospective study (24 centres, 10 Western countries) of consecutive patients undergoing ER of GD-NETs. The main evaluated outcomes were the frequency of curative resection and residual disease. Additional evaluated outcomes included R0 resection rate and local recurrence. We performed a target-trial emulation to compare endoscopic submucosal dissection (ESD), conventional endoscopic mucosal resection (C-EMR), and modified-EMR for these outcomes, while adjusting for potential cofounders.

Results: We analysed 444 GD-NETs (304 gastric, 140 duodenal) from 434 patients (median follow-up of 36 months). The curative resection rate was 51.6% for gastric NETs and 33.6% for duodenal NETs. The residual disease rate was 19.0% for gastric NETs and 10.8% for duodenal NETs. GD-NET-related mortality was 0.2%. Target-trial emulation demonstrated that ESD, compared with C-EMR, achieved a significantly higher R0 resection (60.9% vs. 45.6%; risk difference=15.3 percent points [pp]) and lower risk of local recurrence at 4 (risk difference=12.0 pp) and 5 years (risk difference=14.3 pp). No significant differences were observed between the three ER approaches for other outcomes.

Conclusion: GD-NETs were associated with frequent non-curative ER resections, but low recurrence and tumour-related mortality. ESD achieved a higher adjusted R0 resection rate and lower local recurrence compared with C-EMR. These results support the refinement of not only the curative resection criteria but also patient and ER technique selection.