Minimally invasive pancreatic surgery—where are we going?

Minimally invasive pancreatic surgery (MIPS) was introduced in the mid-1990s by Gagner and co-workers, who reported laparoscopic pancreatoduodenectomy, enucleation, and distal pancreatectomy [1, 2]. Despite these early reports, further implementation of MIPS into clinical practice has been rather slow. Possible reasons for this are the anatomic complexity of the region, limitations of laparoscopic equipment relative to pancreatic surgery, the technically demanding nature of MIPS, and a high risk of postoperative complications.

In the early period of MIPS, minimally invasive distal pancreatectomy (MIDP) and enucleation were mostly considered in selected patients as there was no need for pancreatic anastomoses [3, 4]. The selection criteria for these procedures included tumor type, tumor size, location, involvement of adjacent tissues, and patient characteristics. In contrast, minimally invasive pancreatoduodenectomy (MIPD) was deemed technically challenging and associated with high perioperative risks [5]. Furthermore, its real-life benefits were largely unclear to surgeons. According to the US National Cancer Database (NCDB) review for 2010–2011, patients with lesions in the pancreatic tail were three times more likely to be operated on by MIPS than those with lesions in the pancreatic head [6]. The same study demonstrated that only 13.5% of all patients were subjected to MIPS, while 86.5% were operated on by the open approach.

The first review articles comparing MIPS and its open counterpart found comparable morbidity and mortality, although longer operative time, less blood loss, and shorter hospital stay were typical for the former [3, 5, 7, 8]. At the same time, these studies underscored the low evidence level and selection bias in the minimally invasive arm. Over the past few years, the evidence level on MIPS has somewhat increased as prospective studies, randomized controlled trials (RCTs), and consensus articles have emerged in the literature. In this study, the current evidence on MIPS and its future perspectives are reviewed.

The first MIDP reported by Cuschieri was performed for chronic pancreatitis in 1994 [9]. The proportion of MIDP grew exponentially over the next two decades. In the American population-based study, Tran Cao et al. found that the rate of MIDP has tripled from 2.4% of all distal pancreatectomies in 1998 to 7.3% in 2009 [10]. Data from the nationwide study in the Netherlands demonstrate a significant increase in the proportion of MIDP—from 9 to 47% [11]. In Norway, MIDP currently accounts for 59% of all distal pancreatectomies nationwide [12].

Multiple case–control studies examining outcomes of minimally invasive and open distal pancreatectomy (ODP) have been published to date. Systematic reviews of these studies report reduced intraoperative blood loss, less red blood cell transfusion, and shorter hospital stay for MIDP, but comparable incidence of postoperative pancreatic fistula (POPF), morbidity, and mortality [13, 14]. MIDP was also reported to be associated with relatively high spleen preservation rates compared to its open counterpart, which was explained by the magnified and enhanced visualization facilitating dissection along the splenic vessels and clear identification of small arteries and veins [15, 16]. However, this evidence is based predominantly on retrospective series affected by selection bias. Some studies even observed less postoperative complications following MIDP [7, 17]. At the same time, management of the pancreatic stump during MIDP as one of the most crucial steps of this procedure is as problematic as in open surgery, resulting in similar rates of POPF (19.1 vs. 19.9%, p = 0.73, respectively) [7]. The only RCT published to date suggests less blood loss, longer operative time, lower clinically relevant (grade B/C) delayed gastric emptying rates, shorter time to functional recovery, and better quality of life for MIDP compared with ODP [18]. Interestingly, several prospective studies focusing on costs found MIDP to be as cost effective as ODP [19‐21].

The indications for MIDP have changed significantly over time. Although initially applied mostly for benign and low-grade malignant tumors, MIDP is becoming increasingly used for ductal adenocarcinoma. In France, its proportion in patients with cancer has doubled from 7.3% in 2007 to 14.8% in 2012 [22]. In the Netherlands, the proportion of MIDP for ductal adenocarcinoma has increased from 5 to 51% [11]. Several systematic reviews published on this topic have agreed that typical advantages of the minimally invasive technique also translate into surgery for pancreatic cancer [23, 24], but, most importantly, the long-term oncologic outcomes were comparable between MIDP and ODP [25, 26]. At the same time, MIDP was associated with less lymph node yield, smaller tumor size, and less perineural invasion, which was a result of early stage cancer and a selection bias in the minimally invasive arm.

The recent worldwide survey among pancreatic surgeons performing MIDP demonstrated that only 12% of surgeons found no contraindications for MIDP [27]. In a similar pan-European survey, 45% of surgeons claimed that they had insufficient training in MIDP for all indications [28]. The recent American nationwide study evaluating patient selection factors concluded that MIDP was more often applied in patients with benign lesions and in case of body mass index (BMI) 30–40 kg/m², whereas ODP was a preferable option for tumors >5 cm and if multivisceral resection was needed [29]. These findings were further confirmed by the data from the NCDB, which found that pancreatic ductal adenocarcinoma was independently associated with opting for the open approach [6]. Thus, questions about the oncologic adequacy and possible limits of MIDP in the setting of cancer remain unanswered. A pan-European RCT (DIPLOMA) on minimally invasive vs. open distal pancreatectomy for cancer is currently running and its results are much anticipated [30].

Since the introduction of MIPD more than 20 years ago, the attitude of surgeons towards this procedure has been controversial. First and foremost, MIPD is a highly complex procedure with a steep learning curve. Data from a high-volume center suggest decreased blood loss and conversion rate after 20 MIPDs, a decrease in pancreatic fistula after 40 procedures, and a decrease in operative time after 80 procedures [38]. At the same time, given the relatively small number of centers performing MIPD and strict patient selection criteria, its advantages over open pancreatoduodenectomy (OPD), as well as its oncologic adequacy, were unclear. According to a recent pan-European survey, only 21% of pancreatic surgeons have experience with MIPD [28]. Interestingly, 65% of responders view laparoscopy in pancreatoduodenectomy as a technical problem. The most frequently mentioned technical concern during MIPD is construction of pancreatic anastomosis [27].

During the first decade of laparoscopic pancreatoduodenectomy in the US, the latter was performed in only 4.4% of patients referred to pancreatoduodenectomy [39]. The analysis of the NCDB from Sharpe and co-workers demonstrated that low-volume centers performing <10 laparoscopic pancreatoduodenectomies annually had a twofold increase in 30-day mortality compared to OPD [40]. Another study focused on NCDB reported decreased 90-day mortality in centers performing >6 MIPD annually [41]. Thus, outcomes of MIPD seem to depend highly on the annual volume of the center.

Systematic reviews and meta-analyses comparing MIPD and OPD report less intraoperative blood loss, a lower transfusion rate, longer operative time, and shorter hospital stay for the former [42‐45]. The oncologic outcomes were equivalent to those in OPD [41, 45]. However, these reports are largely based on retrospective studies affected by selection bias, and thus should be interpreted with caution. Three RCTs on MIPD vs. OPD have been published over the past few years. The first study conducted in a single institution in India demonstrated longer operative time, less blood loss, and shorter hospital stay (7 vs. 13 days, p = 0.001) for laparoscopic pancreatoduodenectomy compared with OPD for periampullary tumors [46]. Postoperative outcomes were comparable. This study was performed in a private hospital with a specific focus on laparoscopic surgery, with few surgeons involved and who had passed the training curve by several hundreds of laparoscopic pancreatoduodenectomies. This trial is therefore to be considered an expert trial. Due to an imbalance in benign to malignant cases and more pancreatic cancers in the open group, this trial will not be able to present any useful information on long-term oncological outcomes.

Another single-center RCT (PADULAP) reported less grade ≥3 postoperative complications, a lower comprehensive complication index, and shorter hospital stay for the laparoscopic approach [47]. In contrast, the Dutch multicenter randomized LEOPARD-2 trial was prematurely terminated by the safety monitoring board due to more complication-related mortality in the laparoscopic arm (10 vs. 2%) [48]. Surgeon experience, learning curve, and annual volume are thought to affect this outcome. Of note, among the patients included, there were no benefits in short-term outcomes or functional recovery in the laparoscopy group (which was the primary aim of the study).

Less frequently utilized minimally invasive procedures on the pancreas are enucleation and central pancreatectomy. Compared with pancreatoduodenectomy or distal pancreatectomy, these are parenchyma-sparing procedures aimed at preventing development of pancreatic exocrine and endocrine insufficiency. Minimally invasive pancreatic enucleation (MIPE) has clear indications applicable also in open surgery: benign and low-grade malignant tumors with a small size ≤4 cm, and a clear distance from the main pancreatic duct.

A small number of case–control studies have been published on minimally invasive vs. open pancreatic enucleation. These report shorter operative time, less blood loss, and reduced hospital stay for the former [56‐59]. A systematic review from Guerra et al. confirmed these findings, while the rates of clinically relevant POPF and major complications were similar [60]. Although initially considered for left-sided pancreatic lesions only, MIPE has also been successfully applied for tumors in the pancreatic head [59, 61]. Interestingly, both the laparoscopic and the robotic approach have been equally successful in pancreatic enucleation, as reported in the aforementioned studies. At the same time, strict patient selection criteria and a lack of randomized trials on this topic limit current knowledge on the role of MIPE.

Minimally invasive central pancreatectomy (MICP) is normally applied for benign or low-grade malignant lesions like small neuroendocrine neoplasms located in the pancreatic body and neck, which cannot be removed by MIPE due to tumor proximity to the main pancreatic duct. At the same time, this procedure bears risks and challenges, such as technically demanding pancreatic anastomosis and two potential sources of POPF. Only small case series had been published on MICP until recently [62, 63]. One of the first comparative studies between minimally invasive and open central pancreatectomy conducted by Song et al. demonstrated longer operative time but shorter hospital stay for MICP [64]. Another similar report from Asia suggests less blood loss as well as shorter first flatus and diet start times following laparoscopic central pancreatectomy [65]. Quality of life parameters were higher for laparoscopy compared to the open approach yet without statistical significance. A systematic review and meta-analysis by Xiao and co-workers demonstrated similar POPF rates, morbidity, and incidence of postoperative exocrine and endocrine insufficiency between minimally invasive and open central pancreatectomy [66]. However, the reoperation rate was higher for MICP. Interestingly, the use of pancreaticojejunostomy and pancreaticogastrostomy was almost equal during MICP, while POPF rates between them were similar. Finally, an RCT on robot-assisted laparoscopic vs. open central pancreatectomy published by Chen et al. suggests shorter operative time, less blood loss, lower incidence of clinically relevant POPF, and faster postoperative recovery after laparoscopy [67]. The long-term data on MICP from this trial are awaited.

More than two decades have passed since the introduction of MIPS and five RCTs have been published to date (Table 1). Yet implementation of MIPS has advanced differently for pancreatoduodenectomy and distal pancreatectomy. If the progress of MIDP over the past two decades and its perspectives are obvious, the future of MIPD is unclear. MIDP is currently gaining widespread acceptance among pancreatic surgeons worldwide and will likely soon replace its open counterpart. The main question yet to be answered regarding MIDP is whether it is as oncologically effective and safe as the open approach. Hopefully, the ongoing DIPLOMA trial on minimally invasive vs. open distal pancreatectomy for ductal adenocarcinoma will give all answers and allow for the extension of boundaries for MIDP and its full recognition. Furthermore, the learning curve and implementation of MIDP into clinical practice is rather straightforward and technically feasible. Thus, it has a good chance of becoming the “gold standard” in the near future.

The situation is rather different with MIPD. The rationale for its use remains controversial. Although some studies report typical advantages of the minimally invasive approach also for MIPD, high mortality observed in a number of population-based studies and in the recent Dutch RCT warrants further investigation. This problem was shown to be associated with low-volume centers performing MIPD. Hence, one possible solution could be centralizing the health care system and concentrating MIPD in high-volume centers with a sufficient level of expertise, where the benefits of MIPD can be perceived. All in all, more evidence is needed to clarify the utility of MIPD.

MIPE and MICP are limited to a small number of selected patients. However, these procedures merit further scrutiny, as they bear significant advantages over formal resections in terms of preserving pancreatic gland function and improving patients’ quality of life. Well-designed multicenter RCTs are needed to proceed with the assessment of MIPE and MICP.

The first consensus expert meeting on MIDP and MIPD is awaited in March 2019, presumably leading to evidence-based recommendations regarding the routine implementation of minimally invasive surgery for pancreatic disease.

In conclusion, although the outcomes both for minimally invasive and open pancreatic surgery are dependent on the technical skills of the single surgeon, the optimal clinical environment and a dedicated pancreatic team plays a greater role in managing patient selection, safe surgery, and overcoming postoperative complications [68]. To facilitate further development of MIPS, it has to integrate effectively with the outcome-improving effect of a dedicated pancreatic team.