Long-Term Recurrence Risk After Curative-Intent Resection Of Well-Differentiated Pancreatic Neuroendocrine Tumours: A Systematic Review And Meta-Analysis

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Background: Recurrence after curative-intent resection of well-differentiated pancreatic neuroendocrine tumours (PanNETs) may occur soon after surgery or after many disease-free years. Its timing and the pathological features that best predict it remain uncertain, complicating postoperative surveillance.Methods: We searched PubMed/MEDLINE, Embase, and Web of Science from inception to Aug 8, 2026, with citation and update searches through Aug 11, 2026. Eligible cohorts evaluated recurrence after resection of well-differentiated, baseline M0 PanNETs with no macroscopic residual disease. Overlapping cohorts were resolved for each analysis. Overall recurrence proportions, Kaplan–Meier landmarks, adjusted hazard ratios (HRs), and adjusted odds ratios were synthesised separately using restricted maximum-likelihood random-effects models with modified Hartung–Knapp confidence intervals. The protocol was registered with PROSPERO (CRD420261473228).Findings: Twenty overlap-minimised cohorts included 4843 patients and 860 recurrences. Pooled overall recurrence was 16·5% (95% CI 13·7–19·7; 95% prediction interval 8·3–30·2; I2=80·1%). Among 971 patients with tumours up to 2 cm, pooled recurrence was 5·2% (3·4–7·7; I2=11·1%); an exploratory within-small-tumour synthesis associated higher grade with recurrence (adjusted HR 4·98, 95% CI 1·89–13·08). Recurrence was 5·3% at 1 year, 12·7% at 3 years, 21·2% at 5 years, and 34·5% at 10 years; the 10-year estimate was exploratory and based on three cohorts. Recurrence hazard was higher with nodal metastasis (HR 2·80, 95% CI 1·80–4·35), G2 versus G1 disease (4·93, 1·92–12·68), and perineural invasion (2·44, 1·02–5·80). All three contributing cohorts showed higher recurrence hazard for well-differentiated G3 versus G1, supporting its clinical classification as high risk, although the pooled magnitude was imprecise (HR 10·46, 95% CI 0·90–122·27); R1 versus R0 was not independently supported.Interpretation: Recurrence after disease-free PanNET resection continues to accumulate for at least a decade, and a 2-cm threshold does not identify a recurrence-free group. Patients with nodal metastasis, higher grade, or perineural invasion merit closer surveillance during the first 2–3 years and continued long-term follow-up. These features should also be used to select and stratify participants in adjuvant trials.