Risk of metastasis and survival in patients with T1-2 non-ampullary duodenal neuroendocrine tumors undergoing different treatment strategies Authors: Cheng Peng, Wan-shu Chen, Yao-wen Hu, Tian-ming Xu, Peng-guang Yan, Xiang Xu, Bo-wen Tian, Ji Li, Jing-nan Li Published: 2026 ABSTRACT AbstractBackgroundOptimal management of T1-2 non-ampullary duodenal neuroendocrine tumors (NAD-NETs) remains controversial, particularly for intermediate-sized tumors (11–20 mm), due to limited evidence. This study aimed to quantify real-world metastasis risk and compare long-term survival after endoscopic therapy (ET) or surgical therapy (ST) for T1-2N0M0 NAD-NETs.MethodsPatients were identified from the Surveillance, Epidemiology, and End Results (SEER) database and divided into two cohorts: Cohort A (n = 2101), including T1–2 NAD-NETs for metastasis risk-factor analysis; and Cohort B (n = 1496), including T1–2N0M0 NAD-NETs treated with ET or ST for survival comparison. Multivariable logistic regression was performed to assess factors associated with metastasis risk in Cohort (A) Propensity score matching was performed to balance covariates in Cohort (B) Cancer-specific survival (CSS) and overall survival (OS) were calculated and compared between the ET and ST groups.ResultsOverall metastasis incidence at diagnosis was 9.42% (198/2101). Independent metastasis predictors included tumor size 11–20 mm (OR = 4.30, 95% CI:3.00–6.16) and > 20 mm (OR = 14.65, 95% CI:9.38–22.94), and muscularis propria (MP) invasion (OR = 4.94, 95% CI:2.41–10.74). Of the 1496 patients with T1-2N0M0 NAD-NETs, the long-term prognosis was comparable between ET and ST (after matching, 5-year CSS: 99.0% vs. 97.1%, P = 0.031; 5-year OS: 84.6% vs. 87.4%, P = 0.37). Subgroup analyses confirmed equivalent survival for ET versus ST across those measuring 11–20 mm (CSS 97.4% vs. 98.1%, P = 0.694) and those measuring > 20 mm (CSS 100% both).ConclusionET provides comparable long-term survival versus surgery for T1-2N0M0 NAD-NETs and ET could be a viable strategy for well- to moderately differentiated NAD-NETs of 11–20 mm without MP invasion.Supplementary InformationThe online version contains supplementary material available at 10.1007/s12672-026-05013-0.Keywords: Duodenal neuroendocrine tumors, Metastasis risk, SEER, Survival, Treatment strategies Introduction Neuroendocrine tumors (NETs) are heterogeneous neoplasms arising from neuroendocrine cells, with the gastroenteropancreatic (GEP) system representing the most common site of origin [1–3]. Duodenal neuroendocrine tumors (D-NETs) account for 11% of small intestinal NETs and their incidence has risen dramatically from 0.27 to 1.1 per 100,000 persons between 1983 and 2010 in the United States (US), largely attributed to improved endoscopic detection [4–6]. D-NETs can be classified as ampullary or non-ampullary (NAD-NETs) based on location. While surgical resection is preferred for ampullary D-NETs because of their aggressive behavior and worse overall survival (OS) [7, 8], the optimal management of NAD-NETs remains contentious, partly due to their heterogeneous biological behavior and unpredictable metastatic potential. Endoscopic therapy (ET) has emerged as a viable alternative for NAD-NETs recently, including techniques such as endoscopic submucosal dissection (ESD), endoscopic mucosal resection and endoscopic full-thickness resection (EFTR). However, current guidelines exhibit notable divergence regarding the selection between ET and surgery, particularly for intermediate-sized tumors (11–20 mm). The European Neuroendocrine Tumor Society (ENETS) recommends ET for lesions ≤ 5 mm and surgery for tumors > 10 mm, reserving ET for 5–15 mm NAD-NETs only in specialized centers with rigorous selection [8]. Similarly, Italian experts also emphasize size-based thresholds, cautiously extending ET eligibility to 11–20 mm tumors in specialized settings [9]. In contrast, the National Comprehensive Cancer Network (NCCN) prioritizes ET or local excision for well-differentiated, nonfunctional NAD-NETs confined to the mucosa/submucosa, regardless of size [10]. This lack of consensus generates clinical uncertainty, especially for tumors exceeding 10 mm where treatment risks must be carefully weighed. Recent studies challenge these paradigms, reporting comparable 5-year OS between ET and surgery for 11–20 mm T1-2N0M0 NAD-NETs [11, 12], with emerging data even suggest ET feasibility for tumors > 20 mm in highly selected cases [13]. While these findings imply the potential for expanding the indications for ET, they conflict with guideline conservation due to significant limitations: most of the evidence derives predominantly from single centers with small sample sizes, lacks long-term follow-up and may underestimate true metastatic risk due to inadequate stratification [11–19] . To address these critical evidence gaps regarding metastatic risk stratification and long-term therapeutic efficacy in NAD-NETs, we conducted a large-scale, population-based analysis utilizing an extensive national cancer database. Our study aims to quantify the real-world incidence of metastasis in T1-2N0M0 NAD-NETs stratified by tumor size and treatment modality (ET vs. surgery), while concurrently evaluating comparative long-term survival outcomes between these approaches, thereby generating high-level evidence to refine clinical decision-making and resolve existing guideline discrepancies. Methods Study population and data source This is a retrospective, multicenter cohort study using data from the SEER program. The SEER database collects large-scale, real-world cancer data, freely available to the public upon appropriate requests (http://www.seer.cancer.gov). In this study, data obtained from a database with 17 Registries in SEER, which comprised about 26.5% of the US population [20]. We included patients aged > 18 years with histologically confirmed primary duodenal (primary site codes: C17.0, duodenal, non-ampullary) from 2004 to 2022 and with T1-2 stage. NETs histology was confirmed based on the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) (specific histology codes:8150–8157,8240‒8245 and 8249). According to the new TNM classification (American Joint Committee on Cancer, eighth edition), T1 is defined as tumor invading the lamina propria or submucosa and size ≤ 10 mm. T2 is defined as tumor size > 10 mm or muscularis propria (MP) invaded [21, 22]. Patients were excluded if they met any of the following criteria: (1) presence of more than one primary malignant tumor; (2) diagnosis of poorly differentiated or undifferentiated tumors; (3) unknown follow-up time or follow-up duration of less than 1 month; (4) missing crucial variables (tumor size, TNM stage). A two-step cohort design was employed based on the study objectives. Cohort A (n = 2101) was established to investigate the prevalence and risk factors for metastasis, including all T1–2 NAD-NET patients who met the inclusion criteria, regardless of lymph node or distant metastasis status. Subsequently, Cohort B (n = 1496) included patients with confirmed non-metastatic (T1–2N0M0) NAD-NETs who underwent either ET or ST to compare the long-term outcomes of different treatment modalities. The patient selection process is illustrated in Fig. 1. Flow diagram for the selection of non-ampullary duodenal neuroendocrine tumors (NETs). SEER: Surveillance; Epidemiology, and End Results; ET: endoscopic therapy; ST: surgical therapy Variable definitions Demographic data (age, sex, race, median household income (MHI), and residence address), tumor characteristics (tumor size, tumor differentiation, depth of invasion, and metastasis), and survival information during follow-up were extracted. The metastasis in this research included both lymph nodes and distant metastases. The treatment strategies were categorized into ET (codes: 20–27) and ST (codes: 30, 40, 50, and 60) groups according to the surgery codes proposed by the American College of Surgeons Commission on Cancer’s Facility Oncology Registry Data System [23]. Cancer-specific survival (CSS) was defined as death attributed to NAD-NETs from diagnosis. Statistical analysis Continuous variables were expressed as the median and interquartile range (IQR) and compared using the Mann–Whitney U test. Categorical variables were described by frequency (%) and compared by Chi-square test. For metastasis risk analysis in Cohort A, a stepwise logistic regression model of metastasis factors was established, and the results were expressed as odds ratio (OR) and 95% confidence interval (CI). For survival analysis in Cohort B, patients were divided into two groups in terms of treatment modalities: ET and ST. To mitigate selection bias, propensity score matching (PSM) estimated via a multivariable logistic regression model was performed to balance potential confounding variables between the two treatment groups. Matching was conducted using optimal matching with a 1:1 ratio and no replacement. Kaplan-Meier survival curves and log-rank tests were used to compare CSS and OS between the ET and ST groups, with 5-year survival rates also reported. Furthermore, in Cohort B, we conducted a subgroup analysis based on tumor size, invasion depth and tumor differentiation. Additionally, in a separate analysis using Cohort A, Cox regression analysis was used to calculate the hazard ratio (HR) for the effect of metastasis on survival. Data was acquired using SEER*Stat (9.0.40) and all statistical analyses were performed by R software 4.5.0 (https://www.r-project.org/). The two-sided P-value less than 0.05 was considered statistically significant in all calculations. Results Demographics and prevalence of metastasis A total of 2101 patients with stage T1-2 NAD-NETs were included. Overall, 198 patients (9.42%) presented with metastasis at diagnosis. Specifically, 151 patients (7.19%) had lymph node metastasis only, 27 patients (1.29%) had distant metastasis only, and 20 patients (0.95%) had both lymph node and distant metastases. The two groups had significant age differences (P < 0.001), tumor size (P < 0.001), tumor differentiation (P < 0.001), and invasion depth (P < 0.001). Table 1 shows the clinical features of metastasis and non-metastasis groups. Table 1Demographic Characteristics of patients stratified by metastasis with T1-2 non-ampullary duodenal neuroendocrine tumorsCharacteristicsOverall (n = 2101)No metastasis(n = 1903)Metastasis (n = 198)P-valueAge (median [IQR])63.00 [54.00, 71.00]64.00 [55.00, 72.00]59.00 [50.25, 68.75]< 0.001Age (%)< 0.001 <60y815 (38.8)712 (37.4)103 (52.0) >=60y1286 (61.2)1191 (62.6)95 (48.0)Sex (%)0.016 Male1031 (49.1)950 (49.9)81 (40.9) Female1070 (50.9)953 (50.1)117 (59.1)Race (%)0.613 White1390 (66.2)1253 (65.8)137 (69.2) Black513 (24.4)468 (24.6)45 (22.7) Others198 (9.4)182 (9.6)16 (8.1)Differentiation (%)< 0.001 Well1500 (71.4)1355 (71.2)145 (73.2) Moderately164 (7.8)128 (6.7)36 (18.2) Unknown437 (20.8)420 (22.1)17 (8.6)Invasion depth (%)< 0.001 Intramucosal306 (14.6)294 (15.4)12 ( 6.1) SMI363 (17.3)340 (17.9)23 (11.6) MPI156 (7.4)117 (6.1)39 (19.7) Unknown1276 (60.7)1152 (60.5)124 (62.6)T stage (%)< 0.001 T11449 (69.0)1396 (73.4)53 (26.8) T2652 (31.0)507 (26.6)145 (73.2)Size (%)< 0.001 0–10 mm1540 (73.3)1472 (77.4)68 (34.3) 11–20 mm429 (20.4)353 (18.5)76 (38.4) > 20 mm132 (6.3)78 (4.1)54 (27.3)MHI (%)0.942 <$65,000567 (27.0)514 (27.0)53 (26.8)> =$65,0001534 (73.0)1389 (73.0)145 (73.2)Residence address (%)0.125 Nonmetropolitan267 (12.7)235 (12.3)32 (16.2) Metropolitan1834 (87.3)1668 (87.7)166 (83.8)IQR: Interquartile range; Others: American Indian, Alaska Native, Asian/Pacific Islander; SMI: Submucosal involvement; MPI: Muscularis propria invaded; MHI: Median household income Demographic Characteristics of patients stratified by metastasis with T1-2 non-ampullary duodenal neuroendocrine tumors IQR: Interquartile range; Others: American Indian, Alaska Native, Asian/Pacific Islander; SMI: Submucosal involvement; MPI: Muscularis propria invaded; MHI: Median household income In addition, subgroup analysis indicated that the metastasis rate of the tumor > 20 mm (40.9%, 54/132) significantly increased, compared with tumor of 0–10 mm (4.4%, 68/1540) and 11–20 mm (17.7%, 76/429), respectively. Among 1800 patients with available data on depth of invasion, metastasis was identified in 25.0% (39/156) of tumor with MP infiltration and 4.5% (74/1644) of tumors limited to mucosa or submucosa (Supplementary Table bS1). Survival and logistic regression of metastasis Survival of patients with T1-2 NAD-NETs in the non-metastatic group was significantly better than those in the metastatic group, with a 5-year CSS of 97.3% and 92.5%, respectively (P < 0.001) (Fig. 2b). Kaplan–Meier curves for T1-2 non-ampullary duodenal neuroendocrine tumors (NAD-NETs) patients stratified by metastasis in matched cohort. a 5-year overall survival: 84.4% vs. 60.1%, P = 0.261. b 5-year cancer-specific survival: 97.3% vs. 92.5%, P < 0.001. c Multivariate logistic regression for the metastasis of T1-2 NAD-NETs. Complex model: The P-value of all variables was calculated based on univariate logistic regression separately. Simple model: According to the complex model and the stepwise regression method (Akaike information criterion (AIC) minimization principle). Unadjusted HR: Unadjusted hazard ratio; 95% CI: 95% confidence interval In the logistic regression analysis, tumor moderate differentiation (OR = 1.98; 95% CI 1.24–3.12; P = 0.003), MP invasion (OR = 4.94; 95% CI 2.41–10.74; P < 0.001), tumor size (11–20 mm) (OR = 4.30; 95% CI 3.00–6.16; P < 0.001) and > 20 mm (OR = 14.65; 95% CI 9.38–22.94; P < 0.001) were independent risk factors for metastasis (Fig. 2c). Demographics according to treatment modalities Ultimately, 1496 T1-2N0M0 NAD-NETs patients were enrolled for to compare the survival undergoing different treatment strategies; 1122 (75%) received ET, and 374 (25%) received ST. Overall, ET was used more frequently in recent years (69% in 2004–2010, 70.4% in 2011–2016, and 81.1% in 2017–2022). Among these, the use of ET in tumors measuring 0–10 mm rose from 70.8% in 2004–2010 to 75.1% in 2011–2016, and further to 87.7% in 2017–2022. However, the proportion of 11–20 mm receiving ET remained relatively stable across all time periods (65.7% in 2004–2010, 52.4% in 2011–2016, and 57.3% in 2017–2022). Tumor size (P < 0.001), and depth of invasion (P < 0.001) were statistically significant between the ET and ST groups. PSM was used to balance these covariables between the two therapeutic groups. After matching, there were 374 patients in each group; all covariates of the two groups were balanced. The clinical features of the two groups before and after matching are provided in Table 2. Table 2Comparison of clinical features of patients with T1-2N0M0 non-ampullary duodenal neuroendocrine tumors undergoing endoscopic therapy (ET) and surgical therapy (ST) in the unmatched and matched cohortCharacteristicsBefore matchingP-valueAfter matchingP-valueOverall(n = 1496)Overall(n = 748)ET (n = 1122)ST (n = 374)ET (n = 374)ST (n = 374)Age (median [IQR])63.00 [54.00, 71.00]63.00 [55.00, 70.00]0.57662.50 [53.00, 71.00]63.00 [55.00, 70.00]0.918Sex (%)0.3250.660 Male553 (49.3)196 (52.4)203 (54.3)196 (52.4) Female569 (50.7)178 (47.6)171 (45.7)178 (47.6)Year of diagnosis (%)< 0.0010.400 2004–2010205 (18.3)92 (24.6)107 (28.6)92 (24.6) 2011–2016364 (32.4)153 (40.9)151 (40.4)153 (40.9) 2017–2022553 (49.3)129 (34.5)116 (31.0)129 (34.5)Race (%)0.0620.751 White730 (65.1)264 (70.6)267 (71.4)264 (70.6) Black273 (24.3)84 (22.5)86 (23.0)84 (22.5) Others119 (10.6)26 ( 7.0)21 ( 5.6)26 ( 7.0)Differentiation (%)0.1840.390 Well818 (72.9)258 (69.0)245 (65.5)258 (69.0) Moderate71 ( 6.3)33 ( 8.8)30 ( 8.0)33 ( 8.8) Unknown233 (20.8)83 (22.2)99 (26.5)83 (22.2)Invasion depth (%)< 0.0010.490 Intramucosal177 (15.8)56 (15.0)59 (15.8)56 (15.0) SMI189 (16.8)103 (27.5)116 (31.0)103 (27.5) MPI55 ( 4.9)46 (12.3)35 ( 9.4)46 (12.3) Unknown701 (62.5)169 (45.2)164 (43.9)169 (45.2)T stage (%)< 0.0010.374 T1893 (79.6)211 (56.4)224 (59.9)211 (56.4) T2229 (20.4)163 (43.6)150 (40.1)163 (43.6)Size (%)< 0.0010.591 0–10 mm934 (83.2)232 (62.0)245 (65.5)232 (62.0) 11–20 mm159 (14.2)116 (31.0)107 (28.6)116 (31.0) > 20 mm29 ( 2.6)26 ( 7.0)22 ( 5.9)26 ( 7.0)MHI (%)0.2051.000 <$65,000276 (24.6)105 (28.1)106 (28.3)105 (28.1) > =$65,000846 (75.4)269 (71.9)268 (71.7)269 (71.9)Residence address (%)0.5481.000 Nonmetropolitan129 (11.5)48 (12.8)48 (12.8)48 (12.8) Metropolitan993 (88.5)326 (87.2)326 (87.2)326 (87.2)ET: Endoscopic therapy; ST: Surgical therapy; IQR: Interquartile range; Others: American Indian, Alaska Native, Asian/Pacific Islander; SMI: Submucosal involvement; MPI: Muscularis propria invaded; MHI: Median household income Comparison of clinical features of patients with T1-2N0M0 non-ampullary duodenal neuroendocrine tumors undergoing endoscopic therapy (ET) and surgical therapy (ST) in the unmatched and matched cohort ET: Endoscopic therapy; ST: Surgical therapy; IQR: Interquartile range; Others: American Indian, Alaska Native, Asian/Pacific Islander; SMI: Submucosal involvement; MPI: Muscularis propria invaded; MHI: Median household income Comparison of outcomes with different treatments The KM survival curve and log-rank test were performed to compare the 5-year OS and CSS of the ET and ST groups before and after matching (Fig. 3a–d). Before matching, the 5-year CSS (98.5% vs. 97.1%, P = 0.017) in the ET group was significantly higher than that in the ST group, while there was little difference between the two groups in the 5-year OS (86.3% vs. 87.4%, P = 0.62). After matching, the 5-year OS (84.6% vs. 87.4%, P = 0.37) was similar between the two groups whereas the CSS in ET group was still higher than that in ST group (99.0% vs. 97.1%, P = 0.031). Subsequently, to further investigate the effect of tumor size, differentiation and invasion depth on survival outcomes of early NAD-NETs, we performed a subgroup analysis targeting these three factors (Fig. 4). Kaplan–Meier curve for cancer-specific survival (CSS) and overall survival (OS) for T1-2N0M0 non-ampullary duodenal neuroendocrine tumors (NAD-NETs) patients stratified by treatment. a OS of the unmatched cohort (5-year OS: 86.3% vs. 87.4%, P = 0.62). b CSS of the unmatched cohort (5-year CSS: 98.5% vs. 97.1%, P = 0.017). c OS of the matched cohort (5-year OS: 84.6% vs. 87.4%, P = 0.37). d CSS of the matched cohort (5-year CSS: 99.0% vs. 97.1%, P = 0.031). ET: endoscopic therapy; ST: surgical therapy; CSS: Cancer-specific survival; OS: Overall survival Forest plot of subgroup analysis for T1-2N0M0 non-ampullary duodenal neuroendocrine tumors (NAD-NETs) stratified by treatment. ET: endoscopic therapy; ST: surgical therapy; CSS: Cancer-specific survival; OS: Overall survival In patients with tumors of 11–20 mm, similar 5-year CSS (97.4% vs. 98.1%, P = 0.694) was observed between the two groups. In the matched cohort, there were 48 patients with tumors > 20 mm, the 5-year CSS of both groups was 100%, the 5-year OS of the ET group was 94.4%, and the 5-year OS of the ST group was 85.1%. In patients with well differentiated tumor, we found no apparent difference in 5-year CSS between the two groups. Similarly, no statically significant difference was found between ET and ST in the moderately differentiated tumors (5-year CSS: 93.1% vs. 96.0%, P = 0.6; 5-year OS: 86.1% vs. 86.8%, P = 0.42). As for invasion depth, whether the tumor was confined to mucosal layer or submucosal layer, there was no difference in 5-year OS and CSS between ET and ST groups (Supplementary Fig. S1). In 81 patients with tumor invasion of MP, similar prognosis was found between the two groups (5-year CSS: 100% vs. 92.8%, P = 0.177; 5-year OS: 97.1% vs. 78.8%, P = 0.173). Subgroup analyses stratified by year of diagnosis demonstrated no significant differences in 5-year CSS or OS between the ET and ST groups across any of the three time periods (2004–2010, 2011–2016, and 2017–2022). Discussion NAD-NETs present therapeutic challenges due to limited evidence available from studies involving small cohorts. This large-scale SEER-based analysis provides valuable insights into the management of early-stage NAD-NETs, offering evidence to refine metastasis risk stratification and treatment selection. Our findings suggest that tumor biology and technical resectability, rather than size thresholds alone, should inform therapeutic decisions, and ET demonstrated satisfactory safety and efficacy in selected T1-2N0M0 NAD-NETs. Patients who developed metastasis accounted for 9.42% of T1-2 NAD-NETs and exhibited significantly reduced 5-year CSS to those without metastasis (92.5% vs. 97.3%, P < 0.001), emphasizing both the prognostic significance of metastatic progression and the critical need for metastasis risk stratification. Multivariate analysis identified MP invasion (OR = 4.94) and tumor size > 10 mm (OR = 4.30 for 11–20 mm; OR = 14.65 for > 20 mm) as dominant risk factors, consistent with ENETS 2023 guidelines and recent studies [8, 11]. These findings reinforce the biological rationale underlying guideline recommendations for surgical resection in deeply invasive tumors. However, the findings of this study indicate that survival outcomes were similar across treatments for non-metastatic tumors at diagnosis, even those with size or depth risk factors. This suggests that localized disease may be suitable to less invasive approaches when comprehensively staged [24]. The central findings showed comparable long-term survival between ET and ST in T1-2N0M0 NAD-NETs (5-year CSS 99.0% vs. 97.1%, P = 0.031; OS 84.6% vs. 87.4%, P = 0.37), consistent with growing evidence supporting ET as a viable first-line strategy. Notably, in tumors ≤ 10 mm, ET was associated with significantly higher 5-year CSS compared with ST (99.5% vs. 96.3%, P = 0.003). This difference may partially explain the overall CSS advantage observed in the cohort and suggests that endoscopic management may be particularly beneficial in relatively small, localized lesions. From 2004 to 2022, ET was more widely adopted overall, whereas the proportion of tumors sized 11–20 mm undergoing ET did not show a significant increase. Current ENETS guidelines have not yet provided definitive resection method recommendation for NAD-NETs sized 11–20 mm. In our matched cohort, ET achieved 5-year CSS rates comparable to ST (97.4% vs. 98.1%, P = 0.694) for 11–20 mm tumors (n = 220), challenging the conventional mandates for universal surgical resection beyond the 10 mm threshold. These results resonate with several studies from other institutions. Hers et al. reported no survival difference in a cohort including 38 patients with NAD-NETs sized 11–20 m treated endoscopically versus surgically, as well as Wang et al. (n = 64, 32 ET and 35 ST) and Mirzaie et al. (n = 86, 56 ET vs. 30 ST) similarly demonstrated equivalent disease-specific survival [11, 12, 15]. We refine this perspective based on a larger sample size and subgroup analysis to contextualize outcomes relative to key clinicopathological variables. Remarkably, in the > 20 mm cohort (n = 47), ET resulted in 100% 5-year CSS and 94.4% OS, comparable to or exceeding those of ST. Similar results were found in a single center endoscopic study, where endoscopic R0 resection of larger NAD-NETs (median 12 mm) yielded 87.2% R0 rates and only one recurrence during 62-month follow-up [14]. Moreover, it was found that tumor progression was rare following an R1 resection in some cohorts, suggesting R1 resection may have limited impact on prognosis in selected cases. However, individualized management remains necessary given procedure-related risks [25, 26]. The real-world data suggest that ET may achieve efficacy for carefully selected tumors exceeding traditional size (15 mm) [8], but technical complexity and perforation risks should also be considered [27, 28]. Therefore, ET may be applicable only in highly selected cases with rigorous staging in expert centers and further research with a larger sample size is required. For tumors limited to mucosal or submucosal layers, similar to previous studies, the survival rate of ET was comparable to that of surgery. While MP invasion quadrupled metastasis risk (OR = 4.94), it did not universally preclude successful ET. Paradoxically, MP-invaded tumors managed by ET provided slightly higher 5-year OS versus ST (97.1% vs. 78.8%), although without statistical significance. This may reflect the inherent selection bias favoring superficially invasive MP tumors for ET or surgical morbidity and long-term complications carried by pancreatoduodenectomy. This underscores the emphasis of Italian association for neuroendocrine tumors (Itanet) on endoscopic ultrasound (EUS) staging and multidisciplinary evaluation [9]. Advanced techniques like EFTR/ESD enable R0 resection in recent studies, aligning with our observed efficacy [9, 14, 25, 28, 29]. However, lymphatic metastasis risk remains a concern, necessitating EUS lymph node assessment before ET. Future studies must document resection quality (R0 rates) and local recurrence. These findings support revising current size-based restrictions for endoscopic therapy. For 11–20 mm NAD-NETs, ET may be considered first-line when EUS confirms confinement to mucosa/submucosa. For MP-invaded tumors, ET (preferably EFTR) remains viable when EUS excludes nodal involvement and R0 is anticipated. This approach mirrors Itanet’s cautious extension NCCN’s biology-driven strategy while incorporating stratification by depth and technical feasibility [9, 10]. However, the analysis of results is constrained by several limitations. As a retrospective study, the sample size in certain subgroups (e.g., Tumors > 20 mm) is still few due to the low incidence of NAD-NETs, while to our knowledge, this is the largest study comparing the prognosis of different treatment approaches for T1-2NAD-NETs. In addition, the absence of Ki-67 and mitotic index in the SEER database, which are important for the classification of NETs, results in an omission of these factors from the tumor grading [30]. And details such as endoscopic techniques (e.g., EMR vs. ESD), margin status, and recurrence patterns were not reported in our study due to database limitations, so future studies are needed to explain this issue in the future. To control the influence of biases, we performed PSM matching, though residual selection bias may persist (e.g., ET patients may have more favorable tumor locations). Thus, prospective studies recording the quality of endoscopic resection, adjuvant therapies, and patient-reported outcomes are needed. In addition, as an extensive cancer database in the United States, the SEER database contains accurate and comprehensive patient information, a wide geographical range, and a long follow-up time, which are the advantages of our study. Conclusion In conclusion, this study demonstrates that endoscopic therapy provides comparable long-term survival compared to surgery for T1-2N0M0 NAD-NET and ET appears to be adequate for well- to moderately differentiated 11–20 mm NAD-NETs without MP invasion. Additionally, with multidisciplinary evaluation by integrating tumor differentiation, technical resectability assessed by EUS, and center expertise, ET may be cautiously considered after assessing individual risk and ruling out metastasis for other T1-2N0M0 NAD-NETs. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Acknowledgements The authors thank the Surveillance, Epidemiology, and End Results database for its contribution to this research. Author contributions C.P. and W.C. conceived the study, developed the methodology, operated the software, performed formal analysis, curated the data, wrote the original draft, reviewed and edited the article. T.X. and P.Y. conceived the study, reviewed and edited the manuscript. X.X. and Y.H. performed formal analysis, reviewed, edited, and polished the language of the manuscript. B.T. reviewed and edited the manuscript. J.L. and J.L. conceived the study, developed the methodology, reviewed and edited the manuscript, provided language polishing, supervised the project, and were responsible for funding acquisition. All authors take full responsibility for this work and have reviewed and approved the final manuscript. Funding This study was supported by the National Key R&D Program of China [No. 2025YFC3408700, No. 2025YFC3408701]; National High Level Hospital Clinical Research Funding [No. 2022- PUMCH-C-055] and National Natural Science Foundation of China [Nos. 82400638]. Data availability The dataset(s) supporting the conclusions of this article is(are) available in the Surveillance, Epidemiology, and End Results (SEER) program, [http://www.seer.cancer.gov]. Declarations Ethics approval and consent to participate Data used in this study were acquired from SEER, a publicly available database. Thus the approval of the Ethics Committee of Peking Union Medical College Hospital was not required for this study. Not applicable, as this manuscript does not contain any individual person’s identifiable data or images. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Cheng Peng and Wan-shu Chen have contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The dataset(s) supporting the conclusions of this article is(are) available in the Surveillance, Epidemiology, and End Results (SEER) program, [http://www.seer.cancer.gov].