Robotic Parenchymal-Sparing Pancreatectomy: A Systematic Review

Simple Summary Non-anatomic pancreatic resections such as enucleation, duodenum-preserving partial pancreatic head resection, central pancreatectomy, and uncinate resection allow for the preservation of more pancreatic parenchyma than standard resections, i.e., Whipple and distal pancreatectomy. These lead to a significantly lesser degree of endocrine and exocrine insufficiency. Robotic approaches are increasingly being adopted for these technically challenging parenchymal-sparing procedures. The aim of our study was to evaluate the use and added value of the robotic approach compared to open approaches and standard anatomic resections. We carried out a systematic review of the available literature surrounding robotic parenchymal-sparing pancreatectomy and found that while postoperative pancreatic fistula remains common, severe complications are exceedingly rare, and rates of endocrine and exocrine insufficiency are negligible after these procedures. Abstract Background: Parenchymal-sparing approaches to pancreatectomy are technically challenging procedures but allow for preserving a normal pancreas and decreasing the rate of postoperative pancreatic insufficiency. The robotic platform is increasingly being used for these procedures. We sought to evaluate robotic parenchymal-sparing pancreatectomy and assess its complication profile and efficacy. Methods: This systematic review consisted of all studies on robotic parenchymal-sparing pancreatectomy (central pancreatectomy, duodenum-preserving partial pancreatic head resection, enucleation, and uncinate resection) published between January 2001 and December 2022 in PubMed and Embase. Results: A total of 23 studies were included in this review (n = 788). Robotic parenchymal-sparing pancreatectomy is being performed worldwide for benign or indolent pancreatic lesions. When compared to the open approach, robotic parenchymal-sparing pancreatectomies led to a longer average operative time, shorter length of stay, and higher estimated intraoperative blood loss. Postoperative pancreatic fistula is common, but severe complications requiring intervention are exceedingly rare. Long-term complications such as endocrine and exocrine insufficiency are nearly nonexistent. Conclusions: Robotic parenchymal-sparing pancreatectomy appears to have a higher risk of postoperative pancreatic fistula but is rarely associated with severe or long-term complications. Careful patient selection is required to maximize benefits and minimize morbidity.


Introduction
The traditional approach to the resection of any pancreatic mass involves pancreaticoduodenectomy in the case of right-sided lesions and a distal pancreatectomy and splenectomy for left-sided lesions. Even in high-volume centers, these operations have a significant rate of complications, such as pancreatic fistula, intra-abdominal infection, hemorrhage, and delayed gastric emptying. More recently, parenchymal-sparing pancreatectomy (PSP) has become an accepted approach in the management of indolent and premalignant pancreatic pathologies such as pancreatic cystic lesions and well-differentiated pancreatic neuroendocrine tumors (PanNET), as minimizing morbidity and disability to continuous variables. Medians and (interquartile) ranges were converted to means and standard deviations using the formulas described by Wan et al. [6].

Outcomes of Interest
Primary outcomes of interest were postoperative exocrine and/or endocrine insufficiency, overall and severe morbidity (severe morbidity was defined as greater than Clavien-Dindo (CD) grade 3), and risk of postoperative pancreatic fistula (POPF). Secondary outcomes of interest were short-term operative outcomes; these include operative time, length of hospital stay, estimated blood loss, number of transfusions, rate of conversion to open surgery, and need for reoperation.

Search Results and Study Evaluation
A total of 422 articles were identified (PubMed n = 135, Embase n = 287). After duplicates were removed (n = 46), 376 articles underwent title and abstract screening, and 75 articles were considered for full-text assessment. Only 23 studies were included after full-text assessment for this review.
All case reports and case series were evaluated using the JBI critical appraisal checklist and were found to be of acceptable quality. The median MINORS score for comparative studies was 16. One retrospective study was given a score of 14 [27]. The randomized controlled trial reported by Chen et al. was given a score of 20 [22]. Among the comparative studies, three studies had unique control groups. One study compared robotic enucleation (rEN) to robotic pancreatoduodenectomy/distal pancreatectomy [20]. Another compared a robotic end-to-end pancreatic anastomosis to pancreatojejunostomy (PJ) reconstruction [18]. The last study compared robotic central pancreatectomy (rCP) in elderly patients versus young patients [29]. The remaining comparative studies compare rEN to open EN (oEN) [21,23,25,27] and rCP to open CP (oCP) [19,22,24,26].
For patients who underwent rEN, a mean age of 50.1 years and a mean BMI of 25.2 kg/m 2 were observed. Patients with ASA status <3 comprised 88.2% of the subgroup. Similarly, for patients who underwent rCP, a mean age of 48.6 years and a mean BMI of 23 kg/m 2 were observed and patients with an ASA status <3 comprised 92.5% of the subgroup.
When including only case reports and case series, there were a total of 48 patients with a mean age of 47.6 years and a mean BMI of 19.6 kg/m 2 . Half (n = 24/48, 50%) of the tumors were located in the pancreatic neck/body; 16.7% (n = 8) were in the head, 10.4% (n = 5) in the tail, 2% (n = 1) in the uncinate, and the remainder were in an unknown location. The size of these tumors ranged from 1 to 3.9 cm. The most common pathology among these case reports and case series was PanNET (45.8%, n = 22), followed by SCN (14.6%, n = 7), MCN (12.5%, n = 6), and SPN (18.8%, n = 9). Additionally, there were single cases of IPMN, mass-forming pancreatitis, and incidentally-discovered metastatic renal cell carcinoma.

Surgical Indications and Operative Characteristics
Preoperative evaluation was limited and inconsistently reported in the included studies; however, most studies indicated that a rPSP technique was selected to manage a benign or benign-appearing lesion. Overall, masses in the proximal pancreas comprised 15.5% of lesions and included 10.9% of tumors in the pancreatic head and 4.6% in the uncinate process. Neck/body lesions comprised of 75.4% of all lesions, while tail lesions comprised 7.9%. Tumors included 32.4% PanNETs, 15.4% intraductal papillary mucinous neoplasms (IPMN), 14.4% solid pseudopapillary neoplasms (SPN), and 9.8% mucinous cystic neoplasms (MCN). There were two cases of incidentally-discovered malignant lesions (0.3%) which comprised metastatic hepatocellular carcinoma, and one case of metastatic renal cell carcinoma. The case of metastatic hepatocellular carcinoma was initially diagnosed preoperatively as an SPN or PanNET. The case of metastatic renal cell carcinoma was initially thought to be a benign tumor or a tumor with low-malignant potential on preoperative imaging. The overall median tumor size for rPSP was 2.3 cm. Among the comparative studies that included an open PSP group, rPSP median tumor sizes were generally smaller.
For EN, lesions were predominantly located in the tail (31.1%) and comprised of 9.7% IPMNs, 3% SCNs, 3.7% SPNs, and 92.8% PanNETs. The median tumor size was 1.7 cm. Conversion to oEN occurred in 2.5% patients and was due to technical difficulty or adhesions. One (0.6%) rEN patients underwent relaparotomy. Only one patient required an intraoperative transfusion.
Similarly, for CP, lesions were predominantly located in the neck/body (100%) and comprised of 19.7% IPMNs, 34.1% SCNs, 11.1% MCNs, 18.8% SPNs and 12.3% PanNETs. The median tumor size was 2.6cm. Two (0.6%) patients required conversion to oCP. Eleven (3.2%) patients needed relaparotomy, primarily to manage postoperative bleeding. No patients required intraoperative transfusions. For the rCP subgroup, the distal stump was reconstructed using a pancreato-gastrostomy anastomosis in 8.5% of cases, while a PJ was used in 72.9% of cases. In one series, an end-to-end pancreato-pancreatic anastomosis was used for the reconstruction.
Overall, the mean operative time for rPSP was 169.3 min. The mean operative time for rEN and rCP was 149.2 min and 177.7 min, respectively. Among studies that compared rPSP to open PSP, rPSP had a shorter operative time. Estimated blood loss (EBL) was 108.3 mL in the overall rPSP group and 1149.2 mL and 88.9 mL in the rEN and CP groups respectively. When compared to oEN and oCP, rEN and rCP had significantly lower EBL.

Early Postoperative Outcomes
The overall LOS was 16.9 days, with a LOS of 14.5 days for rEN and 17.6 days for rCP, respectively. Among studies that compared rPSP to oPSP, LOS was also shorter for the rPSP group. Major complications occurred in 5.4% of cases in the rPSP group. Specifically, clinically significant POPF occurred in 29.8% of cases. Among all included patients, zero events of postoperative mortality occurred.

Long-Term Postoperative Outcomes (Endo/Exo)
The median follow-up for rPSP patients was 24.1 months. Disease recurrence was inconsistently reported and was therefore not included in this review. During the follow-up period, exocrine deficiency occurred in 0.54% of cases in the rPSP group, 0% in the rEN subgroup, and 0.7% in the rCP subgroup. Similarly, endocrine deficiency occurred in 2.9%, 0%, and 5% in the rPSP, rEN, and rCP groups, respectively. The outcomes of case reports/series and comparative studies are summarized in Tables 3 and 4, respectively. The outcomes stratified by procedure and approach are listed in Table 5.

Discussion
In this systematic review, we find that rPSP is commonly performed in the treatment of benign or indolent pancreatic lesions such as SPN and PanNET. This is the first review to focus on robotic parenchymal sparing procedures. We find that the robotic approach led to no deaths and was rarely linked to severe complications such as reoperation for bleeding. However, postoperative pancreatic fistula was common after rPSP and occurred in over 30% of all cases, but all were grade A or B and did not require reoperation. When compared to the open approach for similar indications, rPSP led to a longer average operative time, shorter length of stay, and higher estimated intraoperative blood loss. Postoperative endocrine and exocrine insufficiency were nearly nonexistent after rPSP. As more cystic and premalignant lesions of the pancreas are incidentally discovered with modern imaging, it is important to consider short-term disability and long-term function after pancreatic surgery. The robotic approach may allow patients to better realize the long-term benefits of parenchymal preservation with a nearly negligible risk of diabetes and minimal effect on the long-term quality of life. This is particularly important when considering rPSP for patients with SPN and PanNET, which are younger patients with a normal life expectancy [31,32]. However, short-term complications such as clinically significant POPF are common, and patients should be well-counseled about the risks and benefits of rPSP.
J.M.T. Finney first described a central pancreatectomy at Johns Hopkins as early as 1910, and since then, many other non-anatomic approaches have been developed [33]. These parenchymal-sparing operations-enucleation, uncinate resection, DPPPHR, and central pancreatectomy-all have a higher risk of early complications such as POPF when compared to standard anatomical resections (pancreatoduodenectomy or distal pancreatectomy). POPF occurs in up to 41% of patients undergoing enucleation, with half of those being significant enough to require intervention [34]. This is in comparison to an approximately 10-20% rate of clinically relevant POPF in patients undergoing pancreatoduodenectomy [35]. However, central pancreatectomy also leads to a lower rate of exocrine and endocrine insufficiency. The occurrence of postoperative endocrine insufficiency, intraoperative blood loss, and length of stay were all also lower with enucleation in a meta-analysis [36]. Similarly, in a randomized study by Büchler et al., patients undergoing DPPPHR had less pain, better glucose tolerance, and more rapid weight gain after surgery than those undergoing pancreatoduodenectomy [37]. Altogether, we find that POPF remains common after rPSP, although severe complications, endocrine/exocrine insufficiency, and reintervention are exceedingly rare. Parenchymal-sparing approaches prioritize long-term outcomes at the cost of temporary short-term morbidity, and patients should be acutely aware of these risks before committing to these operations.
When comparing rPSP to open PSP, we found no significant differences in overall morbidity using a robotic approach. The largest meta-analysis of 1004 patients undergoing minimally invasive central pancreatectomy showed no difference in the rate of severe complications, POPF, or new endocrine insufficiency with a minimally invasive approach [38].
Endocrine and exocrine functions are still preserved one year after a minimally invasive central pancreatectomy [39]. Similarly, minimally invasive enucleations were associated with a significantly decreased overall complication and POPF rate compared to similar open approaches [40]. Large studies of rPSP are still rare, but a meta-analysis of 13 studies of robotic central pancreatectomy similarly showed an overall complication rate of 58%, POPF in 41%, and a negligible rate of endocrine insufficiency [41]. A robotic approach may also lead to better splenic vessel preservation and fewer conversions to open surgery when compared to laparoscopic enucleations [40]. The benefits seen with rPSP over laparoscopic and open approaches may be due to enhanced visualization and dexterity afforded by the robotic platform, allowing for more careful dissection between small tumors and the pancreatic duct. These advantages seem common to all rPSP procedures, although the learning curve for robotic pancreatic resections remains steep.
Although the learning curve for rPSP approaches has yet to be defined, there is evidence that proficiency for robotic pancreaticoduodenectomy occurs after a minimum of 40 cases [42]. Among the cohort studies and randomized trials included in our review, we find that results are mixed with regard to improvement with rPSP over time. The most recent study on robotic enucleation with enough patients to make meaningful comparisons is from Caruso et al. in 2022, who found no severe complications or grade B/C pancreatic fistulas in their cohort of 40 patients, whereas prior studies had several patients with such complications [21]. On the other hand, Shi et al. published the largest and most recent series of central pancreatectomy in 2020, and they found a higher rate of grade B/C POPF (34.5% vs. 21.7%) in their patients undergoing robotic vs. open central pancreatectomy [26]. Although parenchymal-sparing approaches are technically challenging, authors have pointed to the improved visualization and dexterity with the robot and the use of pancreaticogastrostomy to decrease the need for a Roux-en-y reconstruction after central pancreatectomy [13] as reasons why rPSP approaches may actually be relatively easy to adopt. This review has several limitations. Firstly, the procedures included together are technically very different, with disparate risk profiles. However, they do share POPF as a common and oft-feared complication, and this does not seem to be significantly elevated in patients undergoing rPSP. Of the studies included, there was only one randomized trial [22]; there were many case series and case reports among the remainder. As such, most studies were likely subject to a significant selection bias wherein we would expect healthier patients with more accessible tumors and favorable anatomy to undergo robotic surgery. In the studies included in this review, patients were mostly similar with regard to lesion size, BMI, age, and baseline ASA, but certain studies had significantly younger patients [24], smaller tumors [25], and healthier patients in their robotic cohort [13]. Furthermore, patients undergoing rPSP have been highly selected and include very few cases of malignancy, i.e., pancreatic ductal adenocarcinoma, and so these results may not be applicable to this subset of patients for which the majority of pancreatectomies are performed. However, these promising results for rPSP demonstrate that good results are achievable for selected patients in the hands of highly-trained individuals. These results may one day be more generalizable as these techniques are more widely adopted in practice and outcomes improve.

Conclusions
Ultimately, rPSP appears to be a safe approach for benign or premalignant pancreatic pathology. These findings show that rPSP may be effective for the management of nonmalignant cystic pancreatic lesions, preserving pancreatic function and maximizing longterm benefits. Further research with larger sample sizes from high-volume centers of excellence in hepatopancreatobiliary surgery is needed before these techniques are widely adopted, as implementation has been thus far limited to a handful of centers with very carefully selected patients and demonstrating that these procedures can be performed safely at a larger scale is essential.

Data Availability Statement:
Publicly available studies were used in this systematic review. These can be found in PubMed (Medline) and Embase.

Conflicts of Interest:
The authors declare no conflict of interest.