Introduction

Pancreatic endocrine tumors (PNET) represent a very heterogeneous class of neoplasms. Their biological behavior is variable and includes a spectrum that extends from relatively indolent to extremely aggressive [1, 2]. Thus, in 2010, the European Neuroendocrine Tumor Society (ENETS) proposed a combination between evaluation of grading, based on mitotic rate and Ki-67 index, and the TNM classification [3]. According to this staging system, neuroendocrine neoplasms can be defined as G1 (mitotic count <2 per 10 HPF; Ki-67 index ≤2 %), G2 (mitotic count 2–20 per 10 HPF; Ki-67 index 3–20 %), and G3 (mitotic count >20 per 10 HPF; Ki-67 index >20 %) [4]. The different histological features of these tumors will affect the prognosis, so that the grade of malignancy has a fundamental role in therapeutic strategy. Enucleation (EN) and middle pancreatectomy (MP) have been proposed as a treatment for G1 and G2 NETs, where the need for proper oncological treatment has to be balanced with the risk of long-term dysfunction of the pancreas, an important aspect for patients with long life-expectancy [5].

Parenchyma-sparing surgery (PSS) can be technically demanding and requires specific surgical experience, so it is performed less frequently and mainly in specialized centers.

The aim of this study is to analyze the outcomes of PSS for PNET in an Italian high-volume center.

Methods

All patients with a histological diagnosis of PNET who underwent surgical resection between January 2010 and January 2016 were included in this study. The procedures were performed by the Pancreatic Surgery Unit of Humanitas Research Hospital.

Diagnosis was achieved by computed tomography (CT), magnetic resonance imaging (MRI), gallium-68 positron emission tomography (68Ga-PET), or endoscopic ultrasound (EUS) with or without fine-needle aspiration (FNA).

Patients were treated with pancreaticoduodenectomy (PD), distal pancreatectomy (DP), total pancreatectomy (TP), MP, or EN according to tumor location, pre-operative suspect of malignancy, and size of the lesion. In particular, for suspected malignancy at pre-operative study, multifocal neoplasms, lesions embedded deep in the parenchyma of the pancreatic head, and\or in contact with Wirsung duct, PD, DP, or TP were performed. In case of small lesions without pre-operative signs of malignancy, PSS was performed. In these cases, the choice between EN and MP was primarily based on the location of the tumor. When neoplasms were located superficially, EN was chosen. For lesions of the neck or proximal body of the pancreas, embedded deep in the parenchyma, or close to the main pancreatic duct, MP was performed. Intraoperative ultrasound was used when necessary to assess tumor proximity to the Wirsung duct and to rule out the presence of other lesions.

Patients were regularly followed-up after resection. Evaluation included clinical examination and imaging studies, such as CT or MRI. It was considered that recurrence had occurred if the imaging studies demonstrated recurrent disease in the pancreatic remnant and\or new lesions suspicious for NET in other organs or in lymph nodes. Follow-up data were collected until March 31, 2016.

Demographic, pre-operative, intraoperative, post-operative, and discharge data were collected in a prospective database. Pre-operative mortality was defined as in-hospital death or within 30 days from surgery. Readmission rate was evaluated until 90 day after discharge. Morbidity was coded according to Clavien–Dindo classification [6]. Pancreatic fistula was defined according to ISGPF [7]. All tumors were classified according to the TNM and the 2010 World Health Organization (WHO) classification as G1 (well differentiated with benign characteristics), G2 (well differentiated with low-grade malignant characteristics), or G3 (poorly differentiated with high-grade malignant characteristics) [8–10].

Statistical analysis

All the analyses were performed using IBM SPSS Statistics Version 20 as statistical software. Student’s t test and Chi-square test were used to analyze continuous and categorical outcomes, respectively. Survival analysis was performed using non-parametric Kaplan–Meier methodology. All outcomes were considered statistically significant when p < 0.05.

Results

Between January 2010 and January 2016, 99 patients underwent surgery for PNETs in our pancreatic surgery unit. Demographic, clinical characteristics, and outcomes are summarized in Table 1.

Table 1 Demographic, clinical characteristics, and outcomes of patients who underwent parenchyma-sparing resections and the traditional resections for pancreatic endocrine tumors
Full size table

PSS was performed in 22 (22.2 %) cases, with 18 (82 %) EN and 4 (18 %) MP. Standard surgery was performed for 77 patients, in particular 25 (32 %) patients underwent PD, 46 (60 %) DP, and 6 (8 %) TP. In 23 (50 %) cases of DP and in 3 (16.7 %) cases of EN, surgical procedure was performed by laparoscopy.

PSS patients’ median age was 57 (27.82) years, and the majority of them were male (n = 13; 59 %), while patients submitted to the conventional surgery had a median age of 59 (22.82) years and 65 % were male (n = 50).

89.8 % of patients was studied by CT scan (n = 89) and 69.6 % by EUS (n = 69). MRI was the test of choice in 48 (48.4 %) patients. 68Ga-PET was performed in 47 (47.4 %) cases. The pre-operative histological diagnosis was obtained in 68 individuals (68.6 %). The combination of CT, EUS, and 68Ga-PET was the pre-operative staging of choice in 54.5 and 32.5 % of patients submitted to PSS and the traditional surgery, respectively. In potential candidates for EN, the median distance between tumor and Wirsung was studied using EUS, and was 3.75 (0–15) mm.

Only 4 (13.6 %) tumors in the PSS group and 3 (5 %) of those removed with radical surgery were functioning NETs (insulinoma). Most of the neoplasms were sporadic, while in 3 (3 %) patients, they were associated with Multiple Endocrine Neoplasia Type 1 (MEN1).

Tumors removed with PSS were smaller than those removed with the conventional surgery [median diameter 14 (4–68) mm vs. 23.5 (4–244) mm; p = 0.012].

Histological examination on surgical specimens revealed that most of PSS tumors were well differentiated with benign characteristics (G1 n = 15; 68 %) and there were no poorly differentiated tumors (G3). On the contrary, only 30 (39 %) tumors removed with the traditional surgery were G1 and there were some G3 (n = 9; 11.7 %). Comparing the histological examinations of the two groups, it was confirmed the predominance of low- and intermediate-grade tumors removed using PSS (p = 0.013) (Fig. 1).

Fig. 1
figure 1

Enucleation of pancreatic NET and surgical specimen

Full size image

Nodal sampling was performed in every EN and MP. Only two patients (9 %) showed nodal metastatic disease (N1 according to TNM classification).

The histopathological examination of 4 (5.2 %) patients who underwent the traditional surgery revealed the presence of Mixed Adeno-Neuroendocrine Carcinoma (MANEC). Nine (11.7 %) patients were already metastatic at the time of primary diagnosis.

The median post-operative length of stay was significantly shorter for the patient who received PSS [7 days (5.15) vs. 9 (5.32); p = 0.018].

Eleven (50 %) of these patients developed a complication, but only two (18.2 %) hesitated in a major complication (Clavien–Dindo ≥3). Pancreatic fistula developed in ten (45.5 %) patients but only two (20 %) were type B. There were no type C fistulas in PSS patients. Patients treated with the traditional resections developed complications in 54.5 % of cases (n = 42) with 33.3 % (n = 14) Clavien–Dindo ≥3. Pancreatic fistula occurred in 39 (54.9 %) patients: 14 (56 %) underwent PD and 25 (54.3 %) DP. In this group of patients, there were 56.4 % (n = 39) type A fistulas and 1 (2.5 %) type C. No re-operation was performed in patients submitted to PPS, while four re-operations were needed as a result of complications not related to pancreatic fistula after the traditional surgery.

Readmission rate was 9 % in both groups (n = 7 for the traditional surgery, n = 2 for PSS). Causes of readmission were: hemorrhage related to pancreatic fistula (n = 1), sepsis (n = 1), glycemic decompensation (n = 1), abdominal collection related to pancreatic fistula (n = 5), and pyrexia (n = 2). Post-operative mortality was 0 % for PSS and 3.8 % for the traditional surgery: two patients after PD due to heart attack and hemorrhage, and one after TP due to sepsis.

All patients undergoing PSS are alive and free of recurrence. The two patients with histological evidence of lymph node metastasis (N1) are alive too, with an actual disease-free survival of 25 and 11 months. 7 patients treated with the traditional surgery died during follow-up (overall survival, OS = 90.9 %), and of the remaining patients, 12 have disease relapse (disease-free survival, DFS = 84.4 %).

In major resections, mean OS is 28.7 months, with a mean DFS of 25.1 months, while in PSS, OS and DFS are both 35.1 months. Three (42.8 %) of seven patients died from causes unrelated to disease. Of the remaining four patients, two had MANEC and two were already metastatic at diagnosis.

Discussion

In recent decades, the number of newly diagnosed endocrine tumors has increased progressively, mainly due to improved diagnostic techniques [11–13]. As revealed by the analysis of our cases, the affected population is usually young; in most cases, the diagnosis is incidental and the tumors mean size is small. According to the newest ENETS guidelines [3], PSS is the treatment of choice for functional NETs (insulinoma) in both adults and children [14, 15]. In recent years, PSS have also been proposed for non-functional PNETs with good results [5]. When technically and oncologically feasible, the use of laparoscopic technique or parenchyma preserving surgery is recommended, because this approach can reduce the hospital length of stay, enhance post-operative recovery, and minimize the cosmetic impact of the surgical wound [16–19]. In our series, 50 % of DP and 16.6 % of EN were performed with the laparoscopic technique.

Histological examinations of patients who have undergone the traditional resections are rather heterogeneous, compared with those of the patients selected for the PSS. In fact, no PSS patient was found to be affected from poorly differentiated tumor with malignant behavior (G3), while 68 % had a diagnosis of tumor with low risk (G1). This result confirms the correct indication for a less radical treatment. However, comparing the histological examinations of patients undergoing radical surgery with those undergoing PSS, there are no substantial differences between the tumors with similar behavior. In fact, 80.3 % (n = 45) of the tumors with mitotic index ≤2 (G1) removed with the conventional surgery showed a mitotic index lower than 1 mitosis\10 HPF. These data are comparable with that found in histological examinations of patients undergoing PSS (72.7 %, n = 16). This indicates that the choice of surgical technique for tumors with pre-operative low risk is often dictated by technical reasons rather than oncological ones. As a consequence, accuracy of pre-operative staging plays a fundamental role. EUS is our test of choice to study tumor relations with Wirsung duct and was performed in all patients who were candidates for PSS. In selected cases (3), a pancreatic stent was placed during the procedure to protect the Wirsung duct.

Nevertheless, we found two (9.1 %) cases with nodal positivity after PSS, one after EN, and one after MP. Even if these two patients are still alive and free of recurrence, this result raises concerns about oncological safety of PSS. In fact, lymphadenectomy in EN and MP consists in a sampling of a few numbers of regional lymph nodes. In our EN, the metastatic lymph node was the only one collected, and in the MP, it was 1 of 3. Pre-operative staging was performed in both the cases with CT, MRI, EUS with FNA, and 68Ga-PET, and no suspicious signs of lymph node metastasis were found.

Analyzing the histological behavior of these two NET, we found that in both the cases, primary tumor had Ki-67 <2 at pre-operative cytology and specimen analysis confirmed that these two tumors were G1. Retrospectively, we can affirm that in these two cases, the oncological treatment is questionable. On the other hand, the lymph node sampling in PSS cannot guarantee that all PNETs classified as N0 are actually free of lymph node metastasis.

This theme is deeply debated, in fact, in literature, the relevance of lymph nodal metastasis remains controversial. Some studies showed no association between nodal positivity and patients’ survival, while others showed a decrease in DFS and OS [20–22]. The minimum number of lymph nodes required in PSS is not standardized yet, and this is an important bias in the correct surgical staging of PNET. In fact, some authors reported an estimated number of lymph nodes retrieved going from 0 to 5 (median 0) [23]. Certainly, a potentially efficient way to reduce this problem is to improve the lymph node sampling during PSS to guarantee the best oncological result within this surgical technique. In light of these considerations, the benefits of parenchyma-sparing surgery are inevitably opposed by the flaw in nodal sampling. Therefore, a meticulous pre-operative and intraoperative analysis of each case becomes mandatory, also resorting to extemporaneous intraoperative histological examination [24, 25].

Another critical point of this sparing surgery is the incidence of post-operative complications that authors usually report as higher than the traditional procedures. Even if the literature can be frequently found an high incidence of post-operative complications after PSS, in particularly post-operative pancreatic fistulas, in reality, the most recent meta-analysis and reviews reported similar percentages of overall complications and fistulas for PSS and the traditional surgery in high-volume centers [26, 27], and our data confirmed these records. In our study, overall morbidity in PSS group was 50 % and no statistically significant differences were found in the rate of post-operative complications compared with patients submitted to the conventional surgery. Instead, there is a relevant difference in the severity of the complications. In fact, 81.8 % of patients undergoing PSS did not develop serious complications, while only 18.2 % was classified as Clavien–Dindo ≥3.

There are no significant differences in the incidence of pancreatic fistula. No grade C fistula occurred, regardless of the surgical technique and the type of procedure performed. Even if the incidence of pancreatic fistula after PSS is similar to that seen in the traditional surgery, almost all the fistulas that developed after PSS (80 %) had a low clinical impact (grade A). This suggests that, even if the PSS is not devoid of morbidity, it is a safe technique that does not expose the patient to greater risk of developing serious post-operative complications compared with the traditional resections and allows the preservation of a better long-term functionality of the gland.

As further confirmation of the safety of PSS, we calculated the same readmission rate of 9 % for patients undergoing PSS and for those undergoing the conventional surgery, whereas no patients operated with sparing surgery underwent re-operation nor during hospitalization nor after readmission. The readmission rate is in line with data reported by authors, ranging from 12 to 15 % depending on center experience [28–30].

All patients submitted to PSS are at the moment alive and disease-free. Our post-operative follow-up is too short to draw conclusions about oncological safety of PSS; however, data form literature confirm that in selected cases, when an accurate pre-operative staging and meticulous intraoperative evaluation have been performed, parenchyma-sparing pancreatectomy is a valuable technique that allows OS and DFS similar to those obtained after the traditional surgery [28, 31–33].

In conclusion, when possible, G1 and G2 tumors can be treated with parenchyma-sparing surgery. In fact, PSS is not burdened by a higher rate of complications compared with the traditional surgery. Even if the risk of failure of oncological radicality exists, this does not appear to be relevant for the prognosis of the patients, provided that these are carefully selected. In order for this technique to be the optimal surgical option, a cautious choice of indication is mandatory, through an accurate pre-operative and intraoperative staging. Moreover, this kind of surgery has to be conducted in experienced centers, because a meticulous surgical technique is required. And not least, an extensive nodal sampling and a long follow-up are required for the best oncologic outcome.