Minimally invasive pancreaticoduodenectomy with venous resection: results of a systematic review

Introduction

Since the first description in 1994 by Gagner and Pomp, minimally invasive (MIS) pancreaticoduodenectomy (PD) has been considered among the most complex abdominal procedures^[1]. Even if the feasibility and safety of the minimally invasive approach of MIS PD has been demonstrated in several randomized and observational studies, reluctance still exist to embrace MIS for PD^[2-5]. PDs are in fact a complex procedure entailing (1) extensive dissection around the mesenteric and coeliac vessels; (2) dissection above and below the mesocolon (multi-quadrant procedure); (3) a long and a technically challenging digestive reconstruction; and (4) inherent morbidity and mortality which seems not reduced by the MIS approach.

For these reasons, the MIS approach to PD is still not widely practised compared with other procedures such as colonic and gastric resection. However, increased experience with laparoscopy and robotics in surgery has allowed pioneer centres to test the feasibility and safety of these approaches for more advanced procedures. In fact, from a theoretical point of view, the magnified view provided by the laparoscope and/or the 3D vision achieved by robotics can be of great help during the dissection. This enhanced view, coupled by the superior dexterity of the robotic instruments, can be of great help in complex suturing.

Indeed, complex procedures such as renal or splenic artery aneurysm repair, nephrectomy with caval thrombectomy, and kidney and pancreas transplantation have been described in recent years^[6-11]. As a result of these developments and the increased experience achieved with MIS PDs, small series of MIS pancreatectomies with vascular resection have been reported^[3,5,11-19]. The safety and results of this approach remains to be determined. In this article we systematically reviewed the literature on the topic of MIS PD with vascular resection, evaluating the safety and feasibility as well as the outcomes of this approach.

Methods

Results

The initial search resulted in 58 studies. Only 9 studies (reported from US, n = 4; France, n = 1; Italy, n = 2; Russia, n = 1; China, n = 1) matched the inclusion criteria, however^[5,12-19] [Table 1, Figure 1]. This included 140 patients undergoing PD with vascular resection. Sixty-six PDs were performed robotically (47.1%). There were 82 laterals (58.5%), 18 segmental (12.8%) and 29 not specified venous resections (20.7%). One patient had an associated arterial resection (0.7%). Nineteen conversions to open surgery were reported. At least 11 of these conversions were due to venous adhesions^[5]. The conversion rate ranged from 0% to 55%, blood loss ranged from 200 to 842 mL, and operative time ranged from 397 to 518 min. A graft was used for venous reconstruction in 28 patients (20%). Eight deaths (5.7%) were reported postoperatively. Mortality ranged from 0% to 8% in the series, and reoperation was rarely described and reported in five patients (3.5%).

Figure 1. PRISMA flow chart of study selection

Discussion

The wide application of MIS for almost every procedure in digestive surgery during the last twenty years has not involved pancreatic surgery. Reasons for this reluctance are related to the technical challenge of the dissection and reconstruction phases of PD. The fear of having bleeding difficult to control laparoscopically and the challenges posed by the pancreatoenteric reconstruction in the presence of a soft pancreas could have contributed to this phenomenon.

However, expert centres with a large volume of open pancreatic resections have introduced gradually MIS for PD over the last ten years. The increase in experience achieved with standard PDs performed robotically or laparoscopically has allowed these centres to adopt the MIS approach for extended pancreatectomy with vascular resection. The first series described was in 2011 by Giulianotti et al.^[16], who reported two cases of lateral venous resection during robotic PD. Since then, further studies have reported experience with this approach. This current review found 140 PD with venous resection, of which 50% were performed robotically.

Most of these minimally invasive PDs with venous resection were lateral resections (58.5%), which needed either direct suture-repair or patch-interposition. This is certainly related to less advanced cases operated by MIS and to the challenges posed by segmental resection. Segmental resection is, in fact, needed more often in case of long and circumferential venous involvement; can require extensive mesenteric mobilization in order to achieve a tension-free venous approximation; and can require prolonged vascular clamping which can cause bowel oedema impairing the endoscopic view. The largest series to date of robotic PD with venous resection (50 cases) reported only one case of segmental resection^[12], whereas Croome et al.^[14] reported 9 over 22 cases of laparoscopic segmental venous resection.

Patch-repair was the technique of choice in case of a large defect of the lateral venous wall. Peritoneal, bovine pericardium and polytetrafluoroethylene material were variably used for venous patches. Postoperative thrombosis was rarely reported^[3,15,17]. We found a 5.7% postoperative mortality rate which is in the range of that reported in large registry studies in Europe^[20,21]. The causes of mortality were not different to those in open PD, with no specific complications related to the approach used. Blood loss and operative time seem to be comparable to that reported for open surgery.

In conclusion, despite limited experience, the minimally invasive approach to PD with venous resection seems feasible, with an acceptable rate of mortality and morbidity in the hands of highly experienced pancreatic surgeons. The advantages of this approach over open surgery remain to be determined.