With a continuously increasing occurrence rate over the last 20 years (1), pancreatic neuroendocrine tumors (PNETs) now account for 1% of all pancreatic neoplasms (2). This trend may be attributed to increased awareness, diagnostic techniques advancement, or other unidentified environmental factors or genetic factors (3). PNETs may be divided into functional and non-functional (NF-PNET) tumors. In total, 60%–90% of PNETs are clinically non-functional (4, 5). Traditionally considered as less biologically aggressive than pancreatic cancer, PNETs are increasingly recognized for their highly variable pathological potential (6, 7). Many PNETs are indolent with a low metastasis trend and favorable long-term prognosis. In contrast, other high-grade tumors show relentless early metastasis, making their biology more aggressive than ductal adenocarcinoma. Through the generic term “PNET,” a very heterogeneous disease has unfolded before us, which can be defined either as a pancreatic neoplasm or a carcinoma (8).

Surgical resection is the only radical way to treat PNETs. In general, functional PNETs and non-functional tumors >2 cm should be resected according to the recommendations of the National Comprehensive Cancer Network (V2.2021) (9). However, the management of NF-PNETs ≤2 cm remains controversial. Today, many centers (10, 11), as well as the European Neuroendocrine Tumor Society (ENETS) (12) recommend a surveillance strategy for such lesions given their low malignant potential, slow growth, and high incidence of postoperative mortality and morbidity. On the contrary, some studies have reported that surgical resection of NF-PNETs ≤2 is associated with a better survival rate (13). There is increasing recognition that the presence of small, high-grade tumors may result in aggressive behavior (14). In addition, there is increasing evidence that nodules, distant metastases, and recurrence may present in tumor cases that meet the preoperative criteria for benign disease (i.e., intrapancreatic tumors ≤2 cm) (15–18). Therefore, there is no consensus on the optimal management of NF-PNETs ≤2 cm nor regarding the impact of surgery on these patients.

The aim of this population-based study was to determine whether surgical treatment exerted a beneficial effect on overall survival (OS) and cancer-specific survival (CSS) of patients or not. In this study, we enrolled patients with NF-PNETs ≤2 cm with different oncological characteristics, including grade (poorly differentiated, undifferentiated or well–moderately differentiated), lymph node status (lymph node metastasis or not), distant metastasis status (distant metastasis metastasis or not), and regional extension (local invasion or not). We present the following article in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting checklist.

With the widespread use of high-quality, cross-sectional imaging, NF-PNETs ≤2 cm are now increasingly identifiable (20). However, the optimal management strategy for patients with such tumors remains controversial. In this study, it was proved that surgery benefits the CSS of the overall population and G1, M1N0G0, and M0N0G0E1 groups, while patients in the M0N0G0E0 and M0N1G0 groups do not gain such a benefit. In a further step, we used multivariate Cox models to find that surgical treatment is an independent prognostic factor of CSS in the overall population and G1, M0N0G0E1, and M1N0G0 groups. Because of the heterogeneity of oncologic behavior and prognosis of PENTs, we should be prudent in discerning the impact of surgery on the overall population, and we believe it is reasonable and necessary to explore the prognostic impact of surgery in PNETs with different biological behaviors.

To the best of our knowledge, the present study is the first to provide compelling evidence to support different impacts of surgical treatment on patients with NF-PNETs ≤2 cm in different subgroups, including those divided according to M status, N status, regional extension, and grade, at the same time. A striking conclusion was obtained in this study by applying multiple imputation to missing data, which reduced the estimation and improved the validity. Moreover, in order to reduce the selection bias or another bias caused by the limitations of real-world research, IPTW was used to weigh the population in the adjusted analysis, which is potentially a more beneficial approach than using common matching techniques, such as retaining all the samples. After learning from prior proposals about conducting survival analysis, the influence of treatment effects on survival was analyzed in the present study, and CSS and OS were used to mitigate the unmeasured selection bias of the treatment effect.

Characterized by their indolent course and lacking early symptoms, the management of NF-PNETs ≤2 cm is still considered to be controversial. The guidelines or expert consensuses that have been published in different regions are not consistent at present. For example, the ENETS guidelines (12) suggest that observation is a reasonable option for patients with NF-PNETs ≤2 cm. However, other studies have pointed out that, even when the tumors are small (≤2 cm), they showed signs of malignant behavior, such as extra-pancreatic extension, lymph nodal metastasis, distant-organ metastasis, and recurrence, which may lead to disease-related death (21). A consensus statement announced by the Chinese Study Group for Neuroendocrine Tumors (CSNET) (22) suggested that a more aggressive approach be undertaken, except in some selected cases of NF-PNETs <1 cm or patients with incidentally discovered and unacceptable surgical risks. Patients with NF-PNETs ≤2 cm should be treated with tumor resection and careful postoperative surveillance. Meanwhile, it was suggested in the North American Neuroendocrine Tumor Society (NANETS) guidelines (23) that observation is an optimal choice for NF-PNETs ≤1 cm, while the management of NF-PNETs 1–2 cm should be considered on an individual basis according to some factors, including patient age, comorbidities, endoscopic ultrasonography–fine-needle aspiration or endoscopic ultrasonography–biopsy findings (grade, Ki-67), tumor growth status, anatomical location, extent of procedure required for complete resection, patient preferences, and access to long-term follow-up. In a meta-analysis of 714 patients with NF-PNETs <2 cm, it was discovered that PNET excision was linked to better 1-, 3-, and 5-year OS rates (13). The conclusion that surgery improves patient survival was then pushed through the literature as proof of its superiority. Therefore, whether patients with NF-PENTs ≤2 cm ought to be treated with surgery should be further explored by more accurate and specific grouping of this population. Our research has solved some of the differences in this area.

Consistent with another study (24) that evaluated 709 patients who had stage I disease according to the eighth edition of the American Joint Committee on Cancer criteria, our subgroup analysis of the M0N0G0 group found that, while surgical resection was associated with improved OS, there was no benefit offered to CSS. Compared to this earlier study, our investigation further confirmed that surgical treatment did not improve the OS or CSS of M0N1G0 patients, and this result may be related to the unclear risk-stratification of lymph node metastasis and the inconsistent impact of lymph node metastasis on survival among patients with NF-PENTs ≤2 cm (25–28). Another study using the SEER database analyzed 2,158 patients with a median tumor size of 5 cm and found that lymph node status was not a predictor of overall survival (29). These findings reinforced the observation that close follow-up is a reasonable strategy for M0N0G0 and M0N1G0 patients with NF-PNETs ≤2 cm. In addition, our study conducted a subgroup analysis of the population with local invasion (M0G0N0E1 group), where CSS was significantly improved by surgical treatment. This result may be related to the higher degree of malignancy of NF-PENTs invading the surrounding tissues of the pancreas. In addition, regional extension is related to the tumor growth rate and its anatomical location, which can be monitored by imaging methods like magnetic resonance imaging/computed tomography (30). A retrospective multi-institutional analysis that included 119 cases with confirmed NF-PNETs and resection suggested that vascular/perineural invasion is a significant prognostic factor of recurrence (31).

In order to explore the role of surgical treatment in a population with different tumor grades, a study including 380 patients (32) reported an interesting conclusion that surgical resection provides a survival benefit for patients with NF-PNETs ≤2 cm, even those with favorable well-differentiated and moderately differentiated histologies, and tumor size and margin status were not predictors of survival. Meanwhile, other studies have pointed out that poor pathological grades are signs of malignant behavior and commonly considered to be risk factors affecting the prognosis of NF-PNETs ≤2 cm (15, 33, 34). We support the conclusion that patients with poorly differentiated or undifferentiated tumors could benefit from surgical treatment, regardless of distant metastasis, lymph node metastasis, or regional extension. However, we believe that the conclusion that patients with favorable well-differentiated tumors could benefit from surgical treatment should be considered on a case-by-case basis. As mentioned earlier, we found that surgical treatment could not provide survival benefits to patients with well-differentiated tumors (those in the M0G0N0E0 and M0N1G0 groups) so long as their tumors did not appear to have regional extension or distant metastasis.

In addition, many studies have pointed out that distant metastasis is associated with the poor prognosis of NF-PNETs ≤2 cm (35–37), and we suggested that M1N0G0 patients should undergo surgery to obtain survival benefits unlike patients with pancreatic cancer, even if the grade is well-differentiated. Ye et al. (38) evaluated 758 NF-PNET patients with distant metastasis and reported that the median OS of patients treated with surgery was noticeably higher than that of those who were not treated with surgery (79 vs. 24 months). The median CSS of patients who underwent surgical resection of the primary tumor was 81 months, while that of patients who did not undergo surgical resection of the primary tumor was 26 months. However, this study included 627 NF-PNETs > 2 cm, and the NF-PNET ≤2 cm cohort was not analyzed separately from this group, so the study conclusions may be driven by these larger tumors. Therefore, our research is more convincing in terms of providing accurate findings for patients with NF-PNETs ≤2 cm with distant metastasis.

The current study has several limitations. First, given its retrospective nature, the present study has some inherent selection bias. Although the measured confounders that may influence the treatment choice and final treatment effect of patients were well balanced, any remaining unbalanced and unmeasured confounders would still bring about some bias. Second, the variables used for the survival analysis and multivariate Cox regression analysis were only a subset of the clinical and pathological features. Some important tumor markers, such as the Ki-67 index, and some positive prognostic variables, such as surgical margin status, were not available in the SEER datasets. Third, subgroup analysis was not performed for tumors ≤1 cm or 1–2 cm, and some studies have suggested these values are more rational tumor size cutoffs to identify malignancy (39–41). However, this discrepancy does not completely limit the unique effect of surgical resection of NF-PNETs ≤2 cm that we report here.

In conclusion, we have demonstrated that, first, as long as there is no distant metastasis and the tumor is well–moderately differentiated, whether lymph node metastasis occurs or not, patients may not benefit from surgery. However, if this population shows local invasion, surgery should be performed. Second, surgery should be performed if the grade is poorly undifferentiated, regardless of distant metastasis, lymph node metastasis, or regional extension. At last, patients with distant metastasis, regardless of their tumor grade, can benefit from surgery. These findings could help clinicians make better decisions about whether to choose surgery for patients with NF-PNETs ≤2 cm.