Gastroenteropancreatic neuroendocrine tumors (GEP-NETs) are biologically heterogeneous neoplasms originating from neuroendocrine cells of the gastrointestinal tract and pancreas (1). Although they are relatively rare in clinical practice, the incidence of GEP-NETs has seen an incremental increase over the last three decades (2–4). Currently, surgical resection is the mainstay treatment for localized GEP-NETs. The tumor size influences decisions regarding the treatment choice for well-differentiated GEP-NETs. According to the current consensus, given the increased risk of nodal metastasis in GEP-NETs > 20 mm, tumor resection with regional lymphadenectomy is recommended as a surgical routine irrespective of the primary subsite (5). However, for GEP-NETs ≤ 20 mm, which are designated as small GEP-NETs, the optimal surgical strategy has yet to be determined. The inconclusive guidelines with regard to the surgical procedures are due to the ambiguity of nodal metastasis risk at different anatomical subsites. A limited number of studies have evaluated on the regional lymph node metastasis (LNM) patterns in small GEP-NETs with most of them assessing a single site of origin (6–12). Moreover, the impact of regional nodal involvement on prognosis in small GEP-NETs has not been conclusively determined. Blakely et al. (11) reported worse survival among patients with LNM in pancreatic NETs (p-NETs) ≤ 20 mm. However, this survival disadvantage was not observed in appendiceal NETs (a-NETs) (13). These discrepancies among studies provide inconclusive evidence for regional lymphadenectomy of small GEP-NETs, particularly in tumors of 10 to 20 mm. Given the heterogeneity and rarity of small well-differentiated GEP-NETs, we used a large national database to investigate LNM patterns in small GEP-NETs and to evaluate the benefits of regional lymphadenectomy, depending on the anatomical subsites and tumor size.

LN involvement, a critical predictor for poor survival, is rarely encountered in well-differentiated GEP-NETs measuring ≤ 20 mm (15, 16). In our ER cohort, a significantly increased cancer-specific mortality rate was also observed after histologically proven LNM (HR, 1.58; 95% CI, 1.13–2.19; p = 0.007). Therefore, it is important to elucidate the actual LNM patterns in small GEP-NETs. We designed this first and largest population-based study with long-term follow-up, to accurately describe the actual distribution of LNM and evaluate the equivocal survival benefits from regional lymphadenectomy in small GEP-NETs.

Well-differentiated NETs from the stomach are incidentally detected during gastroscopy and frequently manifest with small size (≤20 mm) (17). Most of the small well-differentiated tumors belong to type 1 g-NETs (17–19). Based on the European Neuroendocrine Tumor Society (ENETS) guidelines, for this type with superficial and low-grade features, endoscopic resection and surveillance are considered adequate because of the low tumor-related mortality and acceptable LNM rates (less than 2%–9%) (4, 17, 18). Herein, the LNM rate was 4.1% for g-NETs ≤ 10 mm. However, for tumors of 11–20 mm, there was a remarkable increase in the LNM rate, which reached 12.1%, with previous studies reporting a range of 8.6 to 15.3% (6, 20, 21). Consistent with a recent study (22), we found that LNM was slightly associated with unfavorable OS in small g-NETs (HR, 2.48; p = 0.061) and regional lymphadenectomy had survival advantages for g-NETs 11–20 mm (HR, 0.39; 95% CI, 0.16–0.96; p = 0.041). In Japan, lymphadenectomy has been suggested for g-NET 11–20 mm. Given the non-negligible risk of LNM, it seems to be a more appropriate surgical strategy (17, 19).

The small intestine is a common site for GEP-NETs (4, 23). However, si-NETs are unique malignancies with poor prognosis. Despite their small sizes, si-NETs frequently present with LNM and approximately 46%–98% of patients have nodal involvement after regional lymphadenectomy (23, 24). Herein, LN positivity was pathologically detected in 49.1% of si-NETs ≤ 20 mm. Given this high propensity for regional lymphatic spread, mesenteric lymphadenectomy during small bowel resection is recommended by the ENETS guideline (25). Our findings and those of previous studies further substantiated this recommendation, given that si-NETs that underwent lymphadenectomy exhibited improved survival, relative to those receiving LR only, irrespective of tumor size (26).

Owing to the heterogeneous nature of GEP-NETs, the malignant potential of a-NETs and r-NETs differs from those derived from the small intestines (4). A-NETs and r-NETs ≤ 10 mm are less likely to present with nodal involvement. In previous studies, only one patient (3.8%) with a-NETs ≤ 10 mm harbored occult LNM, and 9.2% of G1/2 r-NETs ≤ 10 mm, which were 4.0% and 6.8% herein (7, 27). Thus, sample appendectomy or endoscopic resection are the mainstay treatment for a-NETs and r-NETs ≤ 10 mm (7, 28). Our data revealed excellent long-term survival of a-NETs and r-NETs ≤ 10 mm underwent LR, with 10-year OS rates of 97.4% and 92.4%, respectively. However, for tumors 11–20 mm, the optimal surgical strategy remains elusive. Some investigators suggested that, because of the indolent nature, appendectomy and complete endoscopic resection are sufficient for a-NETs and r-NETs of 11–20 mm (5, 8, 28), whereas other investigators are prudent because of the elevated risk of LNM in these patients, which were also observed in our study (a-NETs, 29.8%; r-NETs, 40.3%) (9, 27, 29). A tumor size of 11–20 mm is identified as a predictor for LNM in a-NETs (OR, 5.5; p < 0.001) and r-NETs (OR, 4.24; p < 0.01) (9). These findings do not demonstrate that additional lymphadenectomy is worthwhile for tumors of 11–20 mm. Herein, radical resection did not appear to improve OS, as has been previously reported (8, 10). This discrepancy might be attributed to the increased morbidity and decreased quality of life after radical resection. Therefore, additional lymphadenectomy should be cautiously recommended for a-NETs and r-NETs of 11–20 mm. The treatment algorithm based solely on tumor size is insufficient and ambiguous. We believe that lymphadenectomy is warranted for tumors of 11–20 mm with predictors of aggressive tumor behaviors, such as grade 2-3, muscularis, or lymphovascular invasion (7, 9, 30).

C-NETs are always large tumors, and small lesions are rarely encountered during colonoscopy. They tend to behave aggressively and have a high metastatic potential for LN (4). The National Comprehensive Cancer Network (NCCN) guidelines recommend bowel resection with regional lymphadenectomy, irrespective of tumor size; however, the standard surgical strategy for small c-NETs has not been clearly elucidated (31). We confirmed a high risk of LNM in small c-NETs. LNM occurred in 24.9% of tumors ≤ 10 mm and 76.1% of tumors 11–20 mm. Comparable findings reported by Natour et al. (32) were 26.9% in c-NETs ≤ 10 mm and 65.1% in 11–20 mm after LN sampling. Bowel resection with lymphadenectomy remained the mainstay treatment in our 11–20 mm cohort, and was shown to improve OS (HR, 0.07; 95% CI, 0.01–0.48; p = 0.007). However, for c-NETs ≤ 10 mm, this survival benefit could not be observed after ER. In an analysis of nationwide registers in Japan, Konishi et al. (33) considered that c-NETs ≤ 10 mm without lymphatic invasion could be curatively treated by LR. The optimal treatment strategy for tumor ≤ 10 mm warrants further investigation.

In the last 4 decades, incidences of small p-NETs (≤20 mm) have shown a threefold increase, which could be accounted for by the widespread use of morphological and functional imaging (12). However, due to the uncertain malignant potential for small p-NETs, treatment algorithms remain controversial according to the current guidelines (31, 34). For asymptomatic nfp-NETs ≤ 20 mm, ENETS recommends an active surveillance policy considering the comparable safety of this policy and non-negligible complications of aggressive surgery (34). A recent meta-analysis reported the presence of LNM in 11.2% of small nfp-NETs, which was associated with worse prognosis. Thus, the surveillance policy was challenged and additional lymphadenectomy was suggested by these investigators (35). We found that 7.8% of nfp-NETs ≤ 20 mm patients had occult LNM, in tandem with previous findings from a German study (36). Notably, patients with nfp-NETs ≤ 20 mm who received local enucleation exhibited excellent prognosis after 10 years’ follow-up herein (10-year OS = 100%), compared to 56.4% in the observation group (p = 0.117). Due to the potential selection bias, this reduced OS in the observation arm should be interpreted with caution. The impact of formal lymphadenectomy on survival was not found, which could be attributed to inadequate LN harvest and increased perioperative mortality (12). For fp-NETs ≤ 20 mm, according to the NCCN guidelines, enucleation with peripancreatic LN dissection is recommended, except for insulinomas with a benign course (31). Pancreatic gastrinomas, VIPomas, and glucagonomas harbor malignant potential and are associated with LNM (35, 37–39). Herein, LNM was found in 15.4% of all fp-NETs ≤ 20 mm. The possible explanation for the different LNM patterns between small fp-NETs and nfp-NETs may be the tumor heterogeneity and metastatic potential are enhanced in patients with fp-NETs. However, the underlying mechanisms remain unclear. In tandem with previous findings (40), LNM exhibited a negative effect on survival (HR, 9.84; p = 0.005). Peripancreatic LN removal should be considered as it improved survival (HR, 0.08; 95% CI, 0.02–0.36; p = 0.001). Therapeutic decisions should be made by a multidisciplinary team, especially when surveillance policies or extended surgeries are discussed, to carefully weigh the benefits against the risk for each small p-NET (41).

Our results must be considered in light of several limitations. First, information regarding factors that might predispose one to nodal metastasis and recurrence, such as Ki-67 index, mitotic rate, vertical margin, lymphatic and blood vascular invasion, and genetic backgrounds, is unavailable in the SEER database. Therefore, we were unable to evaluate the association between these pathological parameters and LNM. Another intrinsic limitation is non-negligible variability in the LN yield for each GEP-NET, reflecting real-world clinical settings. Currently, there are no guidelines to make an explicit statement of the optimal number of examined LNs, and most surgeons implement lymphadenectomy based on clinical experience. Our study focuses on patients who underwent ER with the hope of reflecting the actual patterns of regional LNM in small GEP-NETs. In this study, a proportion of patients without examined LNs were regarded as those without LNM, which may have been caused by inadequate LN harvest, resulting in an increased false LN negative rate.

Among localized small GEP-NETs that underwent ER, 2,380 patients (36.0%) were identified as LNM. GEP-NETs with younger age, derived from small intestine or colon, with a tumor size of 11 to 20 mm, and with deeper tumor invasion were more likely to present with LNM. Given the increased prevalence of regional nodal metastasis, tumor resection with lymphadenectomy seemed to be the optimal surgical strategy for si-NETs and fp-NETs ≤ 20 mm, as well as g-NETs and c-NETs of 11–20 mm. Survival benefits of additional lymphadenectomy were not found in a-NETs, r-NETs, and nfp-NETs with a small size.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories can be found in the article/ Supplementary Material . The accession number of SEER 18 registries database is 11713-Nov2019.

YZ, HJ, and YC contributed to the conception and design of the study. YC, ZL, and LJ collected data and organized the database. YC, DM, and ZZ performed the statistical analysis. YZ, HJ, YC, ZL, and LJ interpreted the results. YC and YZ wrote the first draft of the manuscript. YC, YZ, and HJ wrote sections of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

This study is supported in part by the Medical and Health Research Project of Zhejiang Province (grant 2020KY468) and the Natural Science Foundation of Zhejiang Province (grant LY20H160001).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.