Discussion
The incidence of postoperative pancreatic fistula ranges from 2 to more than 20 % after open pancreaticoduodenectomy, and from 1.8 to 20 % after LPD. It is important to achieve a good pancreatic-enteric anastomosis, because a postoperative pancreatic fistula may lead to major complications, prolonged hospital stay, and mortality. Minilaparotomy has been advised to ensure safe anastomosis. Although a hybrid laparoscopic-open technique may reduce operative risk, it also results in loss of the potential advantages of minimally invasive surgery. We developed a new PG technique to enable safe reconstruction in pure LPD.
Although PG has been considered an acceptable method of reconstruction after pancreaticoduodenectomy over the past 50 years, there is still controversy regarding the relative superiority of PG versus PJ in terms of outcomes. Wellner et al. reported that PG was superior to PJ in terms of postoperative pancreatic fistula formation judged according to the ISGPF criteria. Also the recent meta-analysis done on PJ vs PG after PD revealed that PG seems to be superior to PJ in reducing the incidence of pancreatic fistula formation and intra-abdominal fluid collection.
The standard pancreatic-enteric anastomosis performed during LPD is PJ. Only one published case report has described reconstruction with PG in LPD. In that case, the remnant pancreas was invaginated into the stomach and was fixed in place with two continuous purse-string sutures around the incision in the gastric wall using self-retaining monofilament sutures (V-Loc 180 3–0, Covidien). Our technique is relatively simpler to perform. We created a small hole in the posterior wall of the stomach and dilated it bluntly. The remnant pancreas was then pulled into the stomach, and easily positioned so that only a few sutures were required between the pancreatic capsule and gastric mucosa to hold it in place.
In LPD, reconstruction is usually performed by end-to-side PJ with duct-to-mucosa anastomosis. Just as in open surgery, LPD carries an increased risk of pancreatic fistula formation in patients with a small pancreatic duct. This increased risk may be attributed to the technical difficulty of performing the duct-to-mucosa anastomotic portion of the pancreatic-enteric reconstruction. In such patients, magnification laparoscopy can be useful for performing duct-to-mucosa anastomosis, but the restricted range of motion of laparoscopic forceps sometimes makes this anastomosis difficult. Our technique does not require duct-to-mucosa anastomosis, and it can be easily used in patients with a small pancreatic duct.
Our technique may also reduce the risk of intra-abdominal abscess formation due to minor leakage of pancreatic juice from the injured pancreatic capsule, because the sutures between the pancreas and the gastric wall are placed inside the stomach. As damage to the pancreatic capsule outside the stomach can be avoided, this technique may be safe in patients with a soft pancreatic texture.
One patient in our series developed a postoperative pancreatic fistula (ISGPF grade B). In this patient, only two sutures were placed between the pancreatic capsule and the gastric mucosa, which was probably inadequate and may have contributed to fistula formation. To reduce the risk of postoperative pancreatic fistula after PG, we suggest placement of a sufficient number of sutures between the pancreatic capsule and the gastric mucosa.
This patient also received only a short internal plastic stent across the PG site. A meta-analysis of randomized clinical trials found that placement of a stent in the pancreatic duct did not reduce the incidence of postoperative pancreatic fistula. However, subgroup analysis found that use of an external stent significantly reduced the incidence of postoperative pancreatic fistula. Other randomized clinical trials found that external duct stenting after pancreaticoduodenectomy reduced the risk of clinically relevant postoperative pancreatic fistula formation. The majority of the selected patients of these studies used PJ for reconstruction, and subgroup analysis for external duct stenting in PG was not reported. Placement of an external stent across the PG anastomosis is not necessarily an essential part of PG, but could be used adjunct to reduce the risk of pancreatic fistula formation.
One of the disadvantages of our technique is that it may result in delayed gastric emptying, which is one of the most common postoperative complications after pancreatic surgery, occurring in 19–57 % of patients. In patients with PG, gastric peristalsis is disturbed because the posterior wall of the stomach is held in place by the PG anastomosis. Additionally, incision of the anterior wall of the stomach increases the risk of delayed gastric emptying. In our technique, the anterior wall of the stomach is incised, sutured, and attached to the abdominal wall by the gastrostomy, which may cause delayed gastric emptying.
The long-term oncologic and surgical outcomes after use of our procedure should be investigated, and future research should investigate whether LPD provides any significant advantages over other methods of performing pancreaticoduodenectomy. It is difficult to draw any sound conclusions about the safety or limitations of our technique with so little information about patient selection, but we consider our technique a relatively easy method for reconstruction in pure LPD. Our technique may also provide an alternative reconstruction method for use in a hybrid procedure. As reconstruction with PG in LPD is still a new technique, further clinical evaluation to compare outcomes between the use of PG and PJ in LPD is warranted.