Gastrointestinal and Pancreatic Neuroendocrine Tumors: A Propensity Score-Matched Systematic Analysis of Treatment Strategies, Oncological Outcomes, and Prognostic Determinants

Background: Gastroenteropancreatic neuroendocrine tumors (GEP-NETs) constitute a heterogeneous family of neoplasms arising from diffuse neuroendocrine cells throughout the gastrointestinal tract and pancreas. Their incidence has risen markedly over the past three decades—now estimated at 6.98 per 100,000 person-years—driven largely by improved detection through advanced endoscopic and cross-sectional imaging modalities. Despite this epidemiological shift, the evidence base informing therapeutic decision-making remains substantially hampered by confounding factors inherent to observational datasets, chiefly selection bias between patients receiving surgery, somatostatin analogues (SSAs), peptide receptor radionuclide therapy (PRRT), or targeted molecular agents. Propensity score matching (PSM) represents a robust quasi-experimental statistical methodology designed to reduce such confounding and provide more balanced comparisons across treatment groups in non-randomized settings. Objectives: This systematic review and propensity-matched analysis aimed to synthesize evidence from PSM studies examining treatment outcomes in GEP-NETs, to quantify survival differences across therapeutic strategies after covariate balancing, and to identify independent prognostic determinants of overall survival (OS) and progression-free survival (PFS) in this population. Methods: A systematic literature search was conducted across PubMed/MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials (CENTRAL) from inception through December 2024. Studies employing propensity score matching or inverse probability of treatment weighting (IPTW) in adult GEP-NET populations reporting OS or PFS were eligible. The PRISMA 2020 guidelines governed reporting. Risk of bias was assessed using the Newcastle-Ottawa Scale (NOS) for observational studies. Pooled hazard ratios (HR) were estimated through random-effects meta-analysis; heterogeneity was quantified by I² statistics. Results: Forty-seven studies encompassing 89,412 patients were eligible after full-text screening. After propensity score matching, surgical resection was associated with significantly improved OS compared to non-surgical management across localized and oligometastatic disease (pooled HR 0.48, 95% CI 0.41–0.57; I²=54%). For metastatic disease, PRRT with ¹⁷⁷Lu-DOTATATE demonstrated superior PFS compared to SSA monotherapy (pooled HR 0.28, 95% CI 0.19–0.42; I²=38%). Everolimus-based therapy showed benefit over chemotherapy in well-differentiated G2 tumors (HR 0.61, 95% CI 0.49–0.76). Tumor grade, stage at diagnosis, primary site (pancreatic versus intestinal), Ki-67 proliferative index, and functional status emerged as the most consistent prognostic variables in multivariable propensity-adjusted models. For small non-functional PanNETs ≤2 cm, watchful waiting and surgical resection yielded comparable disease-specific survival after PSM (HR 1.09, 95% CI 0.87–1.36), supporting a surveillance-first strategy. Conclusions: Propensity score-matched analyses substantially reduce treatment selection bias in GEP-NET research and confirm the survival advantage of surgical resection in resectable disease and PRRT in progressive metastatic disease. For small, low-grade PanNETs, a carefully monitored observational approach is oncologically safe. Standardisation of PSM methodology and reporting in future GEP-NET trials is urgently needed to enhance cross-study comparability.