The conventional approach to resect pancreatic neoplasms has been to perform pancreaticoduodenenctomy (PD) for right-sided (head/uncinate) lesions and distal/subtotal pancreatectomy for left-sided (body/tail) lesions. These are radical procedures with high morbidity, and often result in exocrine and endocrine insufficiency due to the large amount of tissue resected.1 In recent years, enucleation has become popular as an alternative strategy for low-grade pancreatic tumours such as pancreatic neuroendocrine tumors and cystic neoplasms.2

Advocates of this cite the preservation of pancreatic parenchyma as advantageous in reducing pancreatic insufficiency in the long term,3 while opponents are concerned about a possible higher rate of post-operative pancreatic fistula (POPF) following enucleation.45 As this procedure is selectively performed for tumours with low malignant potential, it is ideal for the minimally invasive approach, using either laparoscopy or robotic assistance via the Da Vinci® Surgical System.

However, the safety and efficiency of this technique remains uncertain as there are few studies comparing open, laparoscopic and robotic enucleations. In this study, we aimed to evaluate the role of minimally invasive enucleation (MIEn) by comparing perioperative outcomes with the conventional open approach. Propensity score adjustment was performed to mitigate selection bias.

We performed a retrospective review of all patients who underwent enucleation of pancreatic neoplasms in Singapore General Hospital from 1998 to 2018. This study was approved by our institution review board. Patients were classified into two groups based on intended initial approach, open enucleation (OEn) or MIEn. Decision to perform MIEn or OEn was based on an individual surgeon's decision after discussion with and obtaining informed consent from the patient. This was not based on any formalized institution protocol.

Data collected for comparison were the patient baseline demographics, intra-operative variables, post-operative outcomes, pathologic findings and long-term follow-up. Surgical data were collected from a prospective computerized database (OTM 10, IBM, Armonk, New York), whereas clinical data were collected from another prospective computerized clinical database (Sunrise Clinical Manager version 5.8, Eclipsys Corporation, Atlanta, Georgia).

Table 1 summarizes the baseline demographic characteristics of patients in both groups. A total of 20 patients underwent pancreatic enucleation during the study period, of which 7 (35%) underwent MIEn (5 female, 2 male) while 13 (65%) underwent OEn (7 female, 6 male). Of the 7 MIEn patients, 4 were performed via the laparoscopic approach while 3 were via robotic assistance. There were no significant differences in baseline demographics between the 2 groups. The median age of patients was 43 years (range 28–64) in the MIEn group and 56 years (range 38–66) in the OEn group. Median BMI was 22.3 (range 17.2–33.1) in the MIEn group and 28.2 (range 23.9–30.1) in the OEn group. Patients in both groups had similar American Society of Anaesthesiologist (ASA) scores.

There were no significant differences between the two groups in terms of median tumour size or location (proximal versus distal pancreas). With regards to the tumour pathology, there were 2 cystic lesions (1 pancreatic cyst, 1 branch-duct intraductal papillary mucinous neoplasm (IPMN)) while the rest were solid lesions (11 insulinomas, 1 metastatic lesion to pancreas from renal cell carcinoma, 1 retroperitoneal schwannoma involving part of the pancreas and the rest were non-functioning pancreatic neuroendocrine tumours). 1 (14.3%) patient in the MIEn group and 5 (38.5%) patients in the OEn group had concomitant surgical procedures performed in addition to pancreatic enucleation. The patient in the MIEn group had a laparoscopic resection of a retroperitoneal tumour, while the additional procedures in the OEn group were spleen-preserving distal pancreatectomy, wedge resection of colorectal liver metastasis, radical nephrectomy, cholecystectomy and excision of a duodenal nodule. 6 (30%) patients had previous abdominal surgery, of which 3 (42.9%) were in the MIEn group and 3 (23.1%) were in the OEn group.

Table 2 compares the perioperative outcomes between the MIEn and the OEn groups. On propensity-adjusted analysis, the only significant difference was a reduced rate of readmissions within 30 days in the MIEn group (0 readmissions, 0%) versus the OEn group (4 readmissions, 30.8%) (p=0.0464). There were no conversions to open in the MIEn group, and median operation time was similar in both groups. There was no difference in median estimated blood in both groups, and none of the patients in our series required blood transfusions. The overall morbidity rate was 45.0% and the major complication (Clavien-Dindo>2) rate was 15%; this was similar across both groups. Seven (35%) of our patients had a Grade B/C POPF (6 grade B and 1 grade C), and there was no significant difference between the two groups. Of these, 1 patient underwent percutaneous drainage of intra-abdominal collection, 5 patients had their surgical drains left in-situ for 3–4 weeks, and 1 patient had persistent pancreatic duct leak which eventually required a subtotal pancreatectomy and splenectomy. The MIEn group had a shorter LOS of 5 days (range 3–24) as compared to 8.5 days (range 5–42) in the OEn group (p=0.027), but on propensity-matched analysis this was not significant (p=0.3195). There was no post-operative 90-day/in-hospital mortality in all 20 patients.

Our study showed that MIEn was associated with similar rates of overall morbidity, major complication and clinically significant POPF compared to OEn, hence confirming the safety and feasibility of the minimally invasive approach for pancreatic enucleation. To the best of our knowledge, this is only the second study to date to attempt to mitigate selection bias by conducting a propensity score analysis. In our experience, MIEn was also associated with significantly reduced readmission rate compared to OEn after propensity-score adjustment. We had no readmissions in the MIEn group, and amongst the 4 readmissions in the OEn group, 2 were for pain related to the incision site. This reinforces the benefit of reduced surgical trauma from the minimally invasive approach.

Enucleation has been advocated as a method of limited pancreatic resection for a select group of patients with small, low-grade pancreatic neoplasms89 which are situated away from the main pancreatic duct and on the anterior surface of the pancreas. In a meta-analysis of 22 retrospective studies comprising of 1148 patients, pancreatic enucleation was associated with reduced rates of post-operative exocrine and endocrine insufficiency, as well as shorter duration of surgery, lower blood loss and shorter LOS.10 Furthermore, sensitivity analysis of high-volume studies (more than 20 enucleations and more than 4 per year) showed that in specialised centres, pancreatic enucleations can be performed with no increased risk of POPF. Enucleation has also been shown to have excellent short and long-term oncologic outcomes, with no increased risk of tumour recurrence.1112 This may be partly explained by the fact that most tumours resected by enucleation were benign or had low malignant potential to begin with.

The first report of MIEn was a case of laparoscopic enucleation by Gagner et al.13 in 1996. Since then, there have been several comparative case series published in the literature comparing MIEn to OEn, as summarised in Table 3.141516171819 Our institution has been performing minimally invasive pancreatectomies (MIP) for more than a decade,20 and our conversion rates following laparoscopic distal pancreatectomies are comparable to those of high-volume tertiary referral centres.21 We have also successfully performed robotic pancreatectomies, including PD and spleen-saving vessel preserving distal pancreatectomy.2223 Owing to the rarity of cases that are suitable for enucleation, the number of MIEn we have performed to date is relatively small. However, our outcomes are comparable nonetheless. Our vast experience over the years with both open and minimally-invasive pancreatectomies meant that our surgeons had surpassed their learning curves for MIP and were hence very comfortable performing MIEn. There were no conversions in the MIEn group. Median EBL in our MIEn group was lower than in the OEn group, in keeping with findings from other studies,24 although this was not statistically significant likely due to our small number of patients. The magnified view offered by the high-definition cameras in both laparoscopic and robotic surgery, the stable, tremor-free platform of the robot as well as the improved dexterity of the robotic Endowrist, allowed very precise dissection and meticulous hemostasis.

One key concern that surgeons have with performing enucleation is a potentially higher rate of POPF formation.4 In our series, the overall rate of Grade B/C POPF was acceptably low, and was similar in the MIEn group as compared to the OEn group. These findings are concordant with several recent studies which have also reported no difference in the rates of clinically significant POPF between MIEn (both laparoscopic and robotic approach) and OEn.141516171819

Although our operative time was slightly longer than that reported in most series, it is worthwhile to note that more than half of our enucleations were done for tumours in the proximal pancreas, as opposed to most studies where the majority of lesions were left-sided (body/tail). Right-sided dissection is known to be more challenging and therefore associated with longer operating times, as reported by Costi and Sahakyan.252627

In the present study, the median LOS was significantly shorter for the MIEn group before propensity-score adjustment, concordant with the findings of previous studies.141517 Interestingly, there was no significant difference after propensity-score adjustment, similar to the only other propensity-score matched study16 published to date comparing 60 robotic versus 60 open enucleations. This lack of difference in LOS between MIEn and OEn suggests that selection bias may have at least attribute to some of the differences observed in previous studies. The early recovery policy28 adopted at our institution for all pancreatectomy patients and a Type 2 error due to our small sample size may have also contributed to this observation. These findings emphasize the importance of conducting a multi-institution prospective randomized control trial to determine the benefit of MIEn over OEn.

There are several limitations associated with this relatively small single center study. Firstly, due to its retrospective nature, it is subjected to selection bias. Secondly, the small sample size of the study renders it susceptible to both type 1 and type 2 errors. Finally, there was no long-term follow-up to determine functional outcomes, such as rates of endocrine and exocrine insufficiency following enucleation.

In conclusion, this study confirms that MIEn is feasible and safe, and is associated with similar perioperative outcomes and fewer readmissions compared to OEn.