Most of PNET are incidentally discovered during imaging follow-up performed for other reasons, also due to the widespread use of high-resolution images [27]. The incidence of PNENs ≤2 cm increased by 710.4% (annual percentage change 12.8%) over the last 22 years [28]. As a consequence many of these tumors have small dimension and most of patients are totally asymptomatic. The incidental diagnosis represents a favorable predictor of overall survival for patients with PNENs. Crippa et al. published a series of 355 patients of which 124 (35%) incidentally diagnosed, and they showed that the 5-year progression-free survival (PFS) was 83% and 32% for incidental and symptomatic NF-PNENs, respectively [27]. Cheema et al. recently demonstrated that progression-free survival rates are significantly higher in patients with incidentally detected lesions [29]. The presence of symptoms occurs usually in patients with larger lesions or in the presence of advanced disease. Patients often present late in their course with symptoms of mass effect or with symptoms related to metastases including abdominal pain, weight loss, and jaundice [30, 31]. The small dimension of the tumor, in association with the incidental finding, is associated with a negligible risk of recurrence after radical surgery [32, 33]. Bettini et al. showed that only 6% of NF-PNENs ≤2 cm have an aggressive behavior after radical surgery when incidentally discovered [32]. The choice to treat these lesions with surgery or conservatively should be balanced with the risks and benefits of the surgery itself. Pancreatic resections are associated with a high risk of morbidity and mortality rate remains around 2% also in high-volume centers [34]. Recently, Ateema et al. published a comparison between resections for PNENs and resections for other lesions in terms of postoperative pancreatic fistula (POPF) rate. They showed that grade B and C POPF [35] have higher incidence after resection for PNENs compared with resection for other lesions (22.7% vs 17.2), and if compared only with PDAC, this difference is highly significant (22.7% vs 9%) [36]. In this setting, considering the less aggressive behavior of asymptomatic NF-PNENs ≤2 cm, conservative management has been proposed for these tumors. Recently, several experiences have demonstrated the safety of an active surveillance for small, asymptomatic PNETs [37, 38]. No disease progression among observed NF-PNEN ≤2 cm was reported in any of the published series, and surgery during follow-up was indicated for 4–20% of patients [37]. However surgery is still the treatment of choice for small NF-PNETs in selected cases. Main reasons for surgery are a rapid tumor size increasing, a dilatation of the MPD, a suspicious lymph node metastases, the presence of a Ki67 >2%, and an excessive anxiety experienced by the patient during follow-up. In these cases, surgery should be tailored on the patients’ characteristics and on the localization of the tumors. Conservative, parenchyma-preserving procedures should be avoided in the presence of suspicious signs for malignancy (i.e., lymph node involvement at imaging or dilatation of the MPD). In the remaining cases, middle pancreatectomy is indicated for small tumors of the pancreatic body, whereas an enucleation should be considered only if the main pancreatic duct can be preserved. The main advantage for atypical resections is the possibility to reduce as much as possible the risk of endocrine and/or exocrine insufficiency compared to standard resections although they are associated with a high rate of pancreatic fistulas [39, 40].

In both enucleation and middle pancreatectomy, a lymphadenectomy is not routinely performed, but a nodal sampling should be always recommended for final histological assessment. Atypical resection should be considered only for small lesions with benign or uncertain behavior. Surgical resections can be performed either open or laparoscopically. Both laparoscopic enucleation and distal pancreatectomy are safe and feasible in patients with PNETs [41].

Lesions larger than 2 cm show different behavior: they are associated with a higher risk of malignancy and a poorer disease-free survival. The risk of nodal metastases is also higher and is up to 50% in case of lesions >4 cm [32]. In most series, NF-PNETs >2 cm are usually diagnosed at an advanced stage with 60–85% presenting with synchronous liver metastases [42, 43]. Considering the risk of malignancy all these lesions should be treated with a standard resection and adequate lymphadenectomy. In case of lesions localized in the head of the pancreas, PD is the treatment of choice, whereas in lesions of the body and tail of the pancreas, left pancreatectomy (LP) and splenectomy are indicated.

Neuroendocrine carcinomas (NECs) are defined as lesions with a mitotic count of >20 mitoses per 10 high-power fields and/or a Ki67 index of >20% (WHO 2010 G3) [44]. Pancreatic NECs are rare (only 2–3% of all PNENs) and are associated with a poor prognosis [45]. Most of patients with NECs are diagnosed at advanced stage with a high rate of nodal metastasis, higher than 90% in some series [46]. Median survival ranges from 11 to 21 months [47]. Curative (R0/R1) resection of pancreatic NEC is associated with improved survival, and in localized and resectable disease, surgical exploration for potentially curative resection should be considered. Resection of primary pancreatic neuroendocrine neoplasms in the presence of not resectable hepatic metastases is still controversial, but considering their poor outcomes, palliative resection of the primary pancreatic NEC in the setting of not resectable liver metastases cannot be recommended [48, 49].

MEN1 is characterized by the combined occurrence of primary hyperparathyroidism, duodeno-pancreatic neuroendocrine neoplasms, and tumors of the anterior pituitary gland. These tumors include nonfunctioning PNETs (NF-PNETs) (30–80%), gastrinomas (50%), and insulinomas (20%). The presence of malignant NF-PNEN represents one of the leading causes of disease-specific death in these patients [50]. Nowadays, the recommendation for surgical resection has been based on tumor size, because of a higher rate of metastases in patients with larger tumors [51]. Tumor size is significantly correlated with the presence of metastases and tumor size >30 mm is associated with a reduction in survival time. Patients with tumors between 11 and 30 mm have a risk of death slightly, but not significantly, higher than patients with tumors <10 mm [15]. Surgery represents the treatment of choice for NF-PNENs >2 cm associated with MEN1 [52]. Many controversies remain in the management of tumors ≤2 cm. No survival benefit is demonstrated for patients with NF-PNENs ≤2 cm who receive surgery compared with patients who had conservative management [53] [54]. PNENs in the context of MEN1 are often multifocal. In this setting, careful microdissection of the pancreas demonstrates multiple, small microadenomas. While only a minority of the microadenomas acquire the potential to grow unrestrictedly, larger lesions may be genetically unstable, develop secondary mutations, and will grow into clinically relevant lesions. Prophylactic surgery aims to remove these lesions before malignancy develops. However, while recent data show that early diagnosis and surgery improve survival [55], others suggest a more conservative approach, as their data indicate that only tumors >2 cm are associated with an increased risk of malignancy [56]. Pancreatic surgery is graved by a high rate of morbidity as well as a high rate of mortality particularly in non-referral centers. In these patients, the texture of the pancreatic remnant is usually soft. As a consequence there is an increased risk of failure for the pancreatic anastomoses or suturing of the resection margins. Thus, pancreatic surgery in patients with MEN1 syndrome is challenging and associated with high risk of postoperative complications.

Liver metastases (LM) develop commonly (up to 95% in some case series) in the natural history of NENs [57]. Hepatic involvement differs between primaries. Gastric, appendiceal, and rectal tumors rarely cause metastases, while pancreatic and small bowel have 28–78% and 67–91% incidence rates, respectively [58]. LM represent a strong prognostic factor, along with histologic differentiation and proliferative activity. Among patients with LM, the origin of the primary tumor influences prognosis. PNENs have the worst prognosis within all metastatic NENs, with 5-year survival rates of 40–60% [59]. Patients with gastrinoma and no metastatic disease have 95% survival at 20 years, while only a 10-year survival of 15% is reported when diffuse metastatic liver disease is present [60]. Most of the published series on the management of metastatic NENs are heterogeneously made from different primaries including gastro-entero-pancreatic and pulmonary NENs. Although most of these tumors share similar characteristics, long-term outcomes can change considerably among different forms. A recent study focused selectively on PNENs with LM and showed improved overall and progression-free survival in patients who underwent radical surgery comparing to those who underwent palliative surgery and those who had nonoperative treatments [61].

The decision to offer liver surgery is based on a multifactorial assessment taking into account tumor grading (only G1-G2 should undergo liver surgery; G3 cancers have overtly high recurrence rates and disseminated disease), the presence of extrahepatic disease, volume of liver remnant, and the presence of symptoms [62]. This latter is often developed with the onset of hepatic metastases, as the result of tumor-secreted hormones reaching the systemic circulation. Only 20–30% of patients with metastatic disease are suitable for radical intent at presentation [63]. Debulking surgery (R2) in incompletely resectable metastatic disease is not universally accepted, but particularly in symptomatic patients, it may improve the quality of life when medical treatment failed [62]. For surgery with curative intent, ENETS have proposed the following criteria: (i) resectable G1-G2 liver disease with acceptable morbidity and less than 5% mortality, (ii) absence of right heart insufficiency, (iii) absence of unresectable lymph node and extra-abdominal metastases, and (iv) absence of diffuse or unresectable peritoneal carcinomatosis [58]. The type of surgical resection depends on patient and liver conditions (i.e., performance status, number and location of hepatic deposits, complexity of the resection, and predicted future liver remnant). NEN LMs have been classified morphologically as type I (single metastasis), type II (isolated metastatic bulk accompanied by smaller deposits), or type III (disseminated metastatic spread) [64]. For otherwise unresectable disease, two-step approaches have been proposed including portal vein embolization and two-stage hepatectomies including associating liver partition to portal vein ligation (ALPPS) [65, 66]. The overall survival after hepatic resection is 46–86% at 5 years and 35–79% at 10 years [58]. In comparison, patients who do not receive resection of liver metastases show a survival rate of only 30–36% [61, 67]. Selection biases due to better performance status or less advanced disease are likely to have influenced such differences in favor of the outcomes associated with surgery. Nevertheless, resection shows low mortality rate (0–5%) and acceptable morbidity (30%). Interestingly, resections with microscopically invaded margins (R1) do not seem to affect the overall survival [68]. Analysis of histopathology specimens revealed that often the disease burden in the liver is underestimated, with almost half the number of LM from neuroendocrine tumors undetectable on preoperative imaging [69]. These results suggest that NEN LM are frequently more extensive than identified, even intraoperatively, and that a real curative R0/R1 resection is difficult to achieve. As a consequence, a high rate of recurrence is reported, after a median time of 16–20 months and the majority of patients experience recurrence at 5 years. Robust studies comparing nonsurgical against surgical treatments are lacking and those available are subject to relevant selection biases [70]. In the last 20 years, nonsurgical novel techniques have become available and gained popularity mainly as complementary treatment options. The lack of randomized data makes the comparison with a surgical approach in terms of survival benefit and symptomatic relief difficult.