Surgery for non-functioning P-NETs has been limited to lesions >2 cm in diameter and/or biological-morphological features of localized, aggressive disease (G2 according to WHO 2010 [12], MPD and/or biliary duct dilatation, jaundice, radiological signs of vascular infiltration). The surgical treatment consists of typical and atypical resections, depending on the tumor site [2, 13].

Typical resections Head P-NETs are treated with pancreatoduodenectomy (PD) while lesions of the body/tail with laparoscopic left pancreatectomy (LP) with or without splenectomy. In large series, PD and LP show mortality rates of 3% and 0%, postoperative pancreatic fistula rates of 30% and 19%, perioperative bleeding rates of 20% and 6%, and secondary diabetes rates of 18% and 26%, respectively [14–16]. Lymphadenectomy should always be performed, but the extent of node resection remains controversial [5, 17]. Total pancreatectomy with or without splenectomy should be performed if the whole gland is involved by the neoplasm, for multifocal lesions or if the leftover parenchyma is not enough to guarantee adequate function [13].

Atypical resections These resections have been proposed for small, low- and intermediate-grade P-NETs. No consensus exists on the dimensional cut-off. Although the risk of malignancy cannot be completely ruled out, a 2-cm cut-off should be safe enough. Middle pancreatectomy is usually performed for tumors of the pancreatic body, whereas an enucleation should be considered only when the MPD can be safely preserved. Parenchyma-sparing resections are associated with a lower incidence of endocrine/exocrine pancreatic insufficiency when compared to standard resections. On the other hand, they are associated with a high incidence of pancreatic fistula (50%) [10, 13]. Lymph node sampling is not indicated [5].

For MEN1-associated non-functioning P-NETs <2 cm, surgery is not generally suggested [2]; lesions >2 cm are treated as previously described. Due to the tendency of MEN1-associated P-NETs to be multifocal, an accurate IOUS examination is recommended to guarantee the best indication for a possible total pancreatectomy [13]. In young MEN1 patients, parenchyma-preserving surgery should be considered to reduce the incidence of diabetes.

Von Hippel-Lindau patients with P-NETs, (a) tumor size ≥3 cm, (b) mutation in exon 3, and (c) tumor doubling time ≤500 days, should be referred for surgery [18].

An observational approach can be safely adopted for small (<2 cm) and indolent (G1) [12] lesions without signs of aggressiveness, especially when a major pancreatic resection would be required [2]. In these patients the risk of nodal involvement is around 8–14% [19]. Morphological characterization should include non-functional and functional imaging [20], while the histological one, when possible, should be based on EUS-guided fine needle aspiration biopsy [21, 22].

The decision to undertake a follow-up versus surgery approach should be collectively shared in a multidisciplinary group and should consider morphological and biological data including patient age, comorbidities, performance status, tumor site, biological impact of a possible surgery, and patient wishes. In the case of a conservative approach, follow-up should be performed with abdominal MRI or EUS every 6–12 months [13].

Pancreatic NETs are the ideal entity for laparoscopy because they are often small and with an indolent biological behavior. A laparoscopic approach in both benign and malignant lesions is safe and feasible and associated with a lower complication rate and a shorter hospital stay [8].

Well-differentiated (G1/G2), non-functioning D-NETs ≤1 cm and confined to the submucosa (low-risk D-NETs) can be safely managed with endoscopy [25, 27, 28]. The two most common endoscopic techniques are endoscopic mucosal resection (EMR) and endoscopic submucosal resection (ESD).

It is unclear whether surgery is more appropriate than endoscopy in the case of well-differentiated D-NETs with a diameter ranging from 10 to 20 mm [25, 28]. Since the risk of nodal metastases is modest for tumors limited to the submucosa, some authors proposed to offer local excision (surgical or endoscopic) to this subgroup of patients [29].

Some papers show considerable rates of adverse events (perforation and bleeding rates up to 39% and 18.4%), especially when the lowest margin of the tumor is just before the muscularis propria [30–32], demonstrating that the risks associated with ESD can be even remarkable and close to the classically reported complication rates of duodenopancreatic surgery. When vertical or lateral resection margins are found positive at histology and no further endoscopic approach is feasible, surgery becomes mandatory.

Regarding the follow-up strategy after endoscopic resection, the ENETS guidelines recommend UGI endoscopy, cross-sectional imaging and CgA plasma level assays at 6 and 12 months and then biannually [33].

In consideration of their metastatic potential, functioning D-NETs should not be treated with endoscopy unless surgery is deemed at high risk of morbidity and mortality.

Tumor resection of D-NETs guarantees excellent long-term survival results. In fact, D-NETs are a disease with an indolent nature, hence the benefits of any aggressive surgical treatment should be weighed against its risks.

Current data do not allow us to draw clear conclusions on the surgical treatment of D-NETs. It must be remembered that the overall survival of D-NETs is good, with a 5-year survival ranging from 73.9% to 89.3% [34–36], hence any attempt to reach oncological radicality should be performed.

Several authors reported that the larger the tumor the higher the risk of nodal spread [29, 35, 37], but others did not, at least for the proposed dimensional cutoffs of 1, 1.5 or 2 cm [36, 38]. It is likely that the cut-offs provided did not reflect tumor biology, thus they should be considered as one of the parameters to look at during the therapeutic decision-making, but not the only one.

Regarding the type of surgery, Whipple, transduodenal submucosal excision with or without lymphadenectomy, segmental duodenectomy and antrectomy with D1 duodenectomy have all been adopted as feasible surgical approaches to D-NETs [35–37]. However, none of them has been proved to be more effective than the others.

The ENETS guidelines do not recommend a specific treatment for tumors ranging from 1 to 2 cm [25]. Such cases with a limited vertical invasion (until muscularis propria), without any cross-sectional or EUS suspicion of nodal metastases and with a favorable biology, could be treated with transduodenal local excision without regional lymphadenectomy associated with a strict follow-up or with regional lymphadenectomy, indifferently. A Whipple procedure is indicated for D-NETs >2 cm, for smaller tumors with unfavorable biology, for periampullary and for rare functioning D-NETs (somatostatinomas) [25].

As for functioning D-NETs, the duodenum is the most frequent location of gastrinomas (60% to 75%), whether or not they are MEN1-associated [39]. Nodal metastases from duodenal gastrinomas have been reported in 40–70% of cases [28], but they do not affect the prognosis dramatically or, at least, not as much as pancreatic gastrinomas [40]. Surgery of duodenal gastrinomas can be challenging and depends on the size of the tumor and should always be associated with a regional lymphadenectomy [28, 41]. Intraoperative ultrasound and a wide duodenotomy are required to increase the rate of detection of duodenal gastrinomas [41]. Usually, these D-NETs are excised with a rim of normal duodenum. In cases of large and/or multiple duodenal gastrinomas, not removable by enucleation, a Whipple procedure is the treatment of choice. When a MEN1-associated Zollinger-Ellison syndrome is diagnosed, surgery is recommended for tumors >2 cm. For smaller tumors, surgery should be considered in selected cases only. In fact, smaller tumors are invariably multiple, very often they have already metastasized at the time of diagnosis, their prognosis is good and positive results have been described with the use of proton pump inhibitors to control the acid hypersecretion [28]. A postresection clinical and laboratory (with fasting gastric/secretin tests) follow-up is generally enough [41].

After surgical treatment of non-functioning D-NETs, the follow-up should include cross-sectional imaging, CgA plasma level assay and somatostatin receptor scintigraphy, and it should be scheduled at 6 months and then annually for at least 3 years [33]. At least a 5-year follow-up would be reasonable.

Pancreatic NETs show a moderate/good responsiveness to some antiblastic agents. Single agent chemotherapy in P-NETs have a limited activity. Dacarbazine (DTIC) leads to a response rate (RR) of 26% with a median progression free survival (PFS) of 10 months [44], streptozotocin (STZ) ranges from 21 to 36% with a PFS of 16.5–33 months [45], while chlorzotocin leads to partial response in 30% with a PFS of 17 months [46]. Temozolomide was tested in an heterogenous population of 36 thoraco-abdominal neuroendocrine tumors including only 12 P-NETs. Global RR was 8% and PFS did not exceed 7 months [47].