The type of pancreatic resection for PDA is dependent upon anatomical location of the dominant mass. Over 70% of PDA presents as a head mass with the remaining masses occurring in the body and tail. While central pancreatectomies have been performed for early tumors in the body, the results of these resections have not proven to be of benefit and concern exists regarding adequate lymph node clearance and an unacceptably high fistula rate.28,29 Thus, the remaining text will focus on either a left- or right-sided pancreatectomy, namely a pancreaticoduodenectomy, or distal pancreatectomy.
The first pancreaticoduodenectomy (PD) was performed by Dr Codavilla in France in 1898, but was popularized later in 1912 by Dr Walter Kausch. Dr Allen Whipple further modified the Kausch approach in 1935 initially as a two-stage surgery that was subsequently modified into a one-stage approach, resulting in the PD known today as the Whipple procedure.30 The modern PD consists of several steps for pancreatic resection, followed by reconstruction. While various modifications exist, the accepted general technique can be divided into the following several steps: (1) Access is gained to the lesser sac through the gastrocolic ligament, and the hepatic flexure of the colon is mobilized inferiorly to allow for access to the duodenum. During this step the infrapancreatic SMV is also identified. (2) A full Kocher maneuver is performed, mobilizing the duodenum medially to the level of the left renal vein confluence into the inferior vena cava. (3) Next a complete portal dissection is performed to identify and ligate the right gastric artery and gastroduodenal artery. Cholecystectomy should be performed if the gallbladder is present and common hepatic duct should be transected proximal to the cystic duct stump. A plane then is developed between the distal divided common bile duct and the hepatic artery to the level of the pancreas. During this step a portal and peripancreatic lymphadenectomy is also performed. (4) The body of the stomach is transected between the third and fourth transverse veins along the lesser curvature and the confluence of the gastroepiploic vessels along the greater curvature. (5) The jejunum is transected 15 cm distal to the Ligament of Treitz and mesentery of the proximal jejunum is divided along the bowel margin to the root of the mesentery where the transected jejunum is passed underneath the root of the mesentery. (6) The pancreas is then divided along the confluence of the SMV and PV, and once parenchymal transection has been completed, the SMV is separated from the uncinate. (7) SMA branches to the uncinate are also divided to completely free the specimen. It is important to skeletonize the SMA during this step in order to increase the rate of a R0 resection. It is this retroperitoneal margin that can be positive in upwards of 20% of cases, even when performed by experienced pancreatic surgeons.31
Reconstruction proceeds in a counter-clockwise fashion beginning with the retrocolic pancreaticojejunostomy, followed by a retrocolic hepaticojejunostomy, and finally an antecolic gastrojejunostomy. Multiple different techniques exist for restoration of pancreatic continuity, but we favor a duct-to-mucosa anastomosis in two layers with an absorbable inner layer with a permanent outer layer. Similar variation exists for hepaticojejunostomy creation, but we favor a single-layered long-term absorbable suture as described in the modified Blumgart–Kelly technique.32 Gastrojejunostomy can be performed in an antecolic fashion in either a hand-sewn or stapled technique.
PYLORUS PRESERVATION VERSUS TRADITIONAL PANCREATICODUODENECTOMY
A larger debate exists regarding pylorus preservation, given the preponderance of complications with delayed gastric emptying (DGE) postoperatively. Watson first reported preservation of intact gastric and duodenal anatomy opting for reconstruction with a duodenojejunostomy in 1944, adapted from Traverso and Longmire's operation.33-35 In addition to normal pyloric function, eliminating a gastroenteric anastomosis leads to avoiding the risk of marginal ulceration. However, despite the anatomic arguments, opponents of this approach site concern for adequate lymph node clearance. Tran et al36 conducted a large multicenter, randomized controlled trial to evaluate whether pylorus preservation affected long-term outcomes compared to traditional PD. A total of 170 patients were enrolled, with 83 undergoing standard PD and 87 undergoing pylorus preservation. Results revealed that there were no statistically significant differences in morbidity, mortality, operative time, blood loss, or hospital stay. Additionally, no differences existed in postoperative DGE (p = 0.80). With regards to lymph node clearance, both groups showed comparable R0 resection rates with standard pancreatectomy achieving 73% versus 82.6% R0 resection in the pylorus preservation group (p = 0.23). Overall survival was similarly unchanged. Tran et al concluded that either approach is acceptable for pancreatic and ampullary carcinomas.
In 2005, Lin et al37 reported a smaller cohort of patients based on the same study design. Like their counterparts, they saw no significant differences in operative time, blood loss, hospital stay, or overall mortality, but did report a statistically significant difference in the incidence of DGE with 42% of patients experiencing DGE in the pylorus preservation group versus 0% in the traditional pancreatectomy group (p = 0.5). These results have been recapitulated in other small studies. Most recently, a Cochrane Review published in 2016 evaluated eight randomized controlled trials comparing standard pancreatectomy versus pylorus preservation.38 Wide variety existed in the study size and design, but no statistical differences were able to be elucidating regarding major morbidity, mortality, or operative characteristics. Though standardization of these studies may be required for a definitive result, current data suggest that operative experience and surgeon comfort should determine the operative approach.
Nodal positivity in pancreatic cancer greatly affects the long-term prognosis of patients after PD. Because of this, many have speculated that greater clearance of the nodal basins may confer a survival benefit after surgery. Many trials have examined this hypothesis without definitive data to support the theoretical construct. The first trial of this kind was published by Pedrazzoli et al39 in 1998. They prospectively assessed 81 patients who received empiric extended lymphadenectomy to include standard peripancreatic and subpyloric nodal basins; in addition, nodes along the hepatic hilum, circumferential clearance of the celiac and SMA origins, and para-aortic nodal dissection from the diaphragmatic hiatus to the inferior mesenteric artery (IMA) and lateral to bilateral renal hila were removed. Their data showed no differences in operative time compared to standard lymphadenectomy with a statistically increased node retrieval. Despite this, there was no statistical difference in survival with standard lymphadenectomy having actuarial survival of 12% versus 6% in the extended pancreatectomy group. Secondary analysis did show, however, that in those patients with nodal positivity, extended lymphadenectomy did confer a survival advantage (p < 0.050). In a separate study, Riall et al40 also demonstrated no increase in 5-year survival when extended lymphadenectomy was performed (p = 0.13). In addition, an increase in overall complication rates secondary to higher incidence of DGE and pancreatic fistula was noted. These data have been reproduced in other studies and comprehensive literature reviews by Sergeant et al,41 Fuji,42 Yeo et al,43 and Capussotti et al44 (see Table 141-3).
Comparison of Outcomes of Selected Trials Comparing Standard Pancreaticoduodenectomy to Standard Pancreaticoduodenectomy with Extended Lymphadenectomya
|Favorite Table|Download (.pdf) TABLE 141-3:
Comparison of Outcomes of Selected Trials Comparing Standard Pancreaticoduodenectomy to Standard Pancreaticoduodenectomy with Extended Lymphadenectomya
|Author ||Number of Patients ||Morbidity ||Mortality ||Median Survival ||Median Survival in LN+ Patients ||R0 Resection ||Total Nodes Resected ||LN+ |
|(%) ||(%) ||(Months) ||(Months) ||(%) ||(Mean) ||(%) |
|Pedrazzoli et al39 || || || || || || || || |
| Total ||81 ||48 ||5 ||N/A ||N/A ||75 ||N/A ||59 |
| SP ||40 ||50 ||5 ||11 ||N/A ||72.5 ||13.3 ||60 |
| ExSP ||41 ||46 ||5 ||17 ||N/A ||78 ||19.8b ||58.5 |
|Riall et al40 || || || || || || || || |
| Total ||294 ||36 ||3 ||N/A ||N/A ||87 ||N/A ||80 |
| SP ||146 ||29 ||4 ||20 ||19 ||76 ||17 ||82 |
| ExSP ||148 ||43b ||2 ||22 ||18 ||94b ||28.5b ||77 |
|Capussotti et al44 || || || || || || || || |
| Total ||149 ||38 ||5 ||15 ||N/A ||N/A ||N/A ||59 |
| SP ||112 ||38 ||6.3 ||N/A ||N/A ||N/A ||10.8 ||54 |
| ExSP ||37 ||35 ||5.4 ||N/A ||N/A ||N/A ||32.4b ||68 |
Pawlik et al45 demonstrated in a retrospective analysis of 158 patients that only 0.3% of patients could theoretically expect a R0 resection margin with removal of secondary echelon nodes as performed during an extended lymphadenectomy. For this number to reach statistical significance over 200,000 patients would need to be accrued in each study arm. Because of these results, standard extended lymphadenectomy is not recommended. This is based on the principle that at the time of diagnosis, for most patients, pancreatic cancer is already a systemic disease. It is our practice to remove lymph nodes along the periportal and peripancreatic regions during PD. We do not recommend performing extended lymphadenectomies.
Again, controversy exists regarding operative management of pancreatic adenocarcinoma that abuts or invades critical venous structures. Arterial involvement has been shown to be a contraindication to surgical resection, even with adequate reconstruction, as these patients have overall poor prognosis. Different approaches exist regarding appropriate management of venous invasion, as is seen by examining the practices of large volume centers. The MD Anderson group routinely performs neoadjuvant therapy prior to operative resection, whereas some other practitioners advocate neoadjuvant therapy only if there is greater than 180 degree involvement of the SMV. Regardless, once considered a contraindication to any surgical approach, venous involvement no longer negates operative resection.
In 2006, Riediger et al46 examined a 10-year experience of pancreaticoduodenectomies totaling 222 patients. Fifty-three of these patients underwent SMV/PV resection and outcomes were retrospectively analyzed compared to the standard PD patients. The vein resections included wedge resections, and complete resection with interposition graft. Long-term survival was not affected by need for vein resection. Additionally, perioperative outcomes and characteristics were not statistically different. The authors did demonstrate longer operative time in the vein resection group and multivariate analysis revealed that positive margins and undifferentiated tumors were the only factors affecting long-term outcomes. They concluded that vein resection should always be performed if needed to achieve a negative margin.
This study followed a large retrospective analysis of 291 patients from the MD Anderson Cancer Center in which 110 patients underwent vascular resection.47 Despite a higher incidence of positive margins in the vascular resection group (p = 0.02), there was no difference in median survival between the two groups (p = 0.18). More recently Menon et al48 published their data in 2013, which showed similar results. Despite longer operative times when vascular reconstruction was performed (p < 0.01), there were no short- or long-term differences in outcomes. Specifically they demonstrated that median recurrence-free survival was not statistically different between groups (p = 0.37). PV and SMV resections and reconstruction should be considered when venous involvement precludes surgical resection.
LAPAROSCOPIC VERSUS OPEN PANCREATICODUODENECTOMY
With minimally invasive approaches to hepatobiliary surgery becoming more apparent, the question has been raised as to the effectiveness of laparoscopically or robotic-assisted PD. First reported in the literature over 20 years ago, these minimally invasive techniques have gained popularity based on the results of multiple single institutional studies. Palanivelu et al49 published their experience of 42 patients undergoing laparoscopic PD and determined that this approach was safe and within adherence to oncologic principles of surgical resection, with adequate lymph node retrieval, and acceptable rates of postoperative complications as compared to open PD. In a 2-year period, Kendrick and Cusati50 published their single-institutional experience of over 60 patients during the procedural learning curve, showing that laparoscopic PD is not only feasible but also safe and effective. Zeh et al51 showed similar results in their initial robotic-assisted PD experience, demonstrating a margin negative rate of 89% in 50 patients. More recently, Asbun and Stauffer52 showed multiple benefits of laparoscopic pancreatectomy compared to open approaches, with statistically significant differences in intraoperative blood loss (p < 0.001), transfusion need (p < 0.001), length of hospital stay (p < 0.001), and length of ICU stay (p < 0.001). As the experience grows with minimally invasive approaches, the reported benefits continue to be demonstrated. In a large meta-analysis and retrospective review, Correa-Gallego et al53 reviewed 527 published studies of minimally invasive approaches to PD. They revealed a decreased operative blood loss (p < 0.001), increased lymph node retrieval by 3 (p = 0.03), and a decreased likelihood of an R1 resection (p = 0.007) favoring minimal invasive PD. Despite statistically increased operative time (p = 0.003), no differences were seen in perioperative complications, with no statistical difference seen in clinically significant pancreatic fistula rates (p = 0.9). Similar to prior studies, they demonstrated that open approaches had increased hospital stays of 3.7 days (p = 0.02). It should be noted, however, that these series reflect the results of highly skilled surgeons and highly selected patients. It is not clear if these results can be translated to common practice. Though these studies show high variability in study design and demand a high level of operative experience, the preliminary data suggest a need for further study. To date, no prospective trials are available in the literature.
Given the high incidence of pancreatic fistula formation, both after PD and after distal pancreatectomy, much effort has been made to identify factors associated with higher incidence of leaks. Accepted risk factors include prolonged operative time, soft pancreatic gland texture, and a pancreatic duct diameter of <3 mm. Anastomotic or transection techniques have not been proven as associated risk factors. Much attention has been given to the use of prophylactic intraoperative drains, given the high incidence of leak. In a randomized trial, 179 patients were enrolled to receive either prophylactic drain placement or no drain at the time of their pancreatic resection.54 While 12.5% of patients were noted to have high amylase levels in the prophylactically placed drains, this prophylactic placement failed to result in statistically significant differences in number or type of postoperative outcomes, including serious complications. A recent systematic review55 suggests higher absolute complication rates approaching 9% in patients receiving empiric surgical drains; however, when corrected for multiple testing and heterogeneity, the cumulative risk difference in major complications was not statistically different (p = 0.214). While higher complication rates are reported in the literature in groups receiving empiric drain placement, many of these studies are biased in that these are retrospective studies, and most patients who received empiric drains demonstrated many of the known risk factors for fistula formation. Given this, current consensus suggests selective use of intraoperative drains, with an emphasis on early drain removal when possible.
USE OF OCTREOTIDE TO PREVENT PANCREATIC FISTULA
While the relatively high incidence of fistula formation after pancreatic resection has been well established, the impact this has on long-term outcomes is not clear. The International Study Group on Pancreas Surgery defines a pancreatic fistula as representing a failure of healing/sealing of a pancreatic-enteric anastomosis or a parenchymal leak not directly related to an anastomosis.56 This is objectively defined as any measurable volume of fluid with an amylase greater than three times that of the serum amylase when measured on postoperative day 3. They are graded as A, B, or C, depending on severity. Grade A fistulas are those that are asymptomatic and represent a chemical leak. Grade B fistulas are those requiring additional intervention, including, but not limited to, additional drain placement, antibiotic therapy, hospital readmission, or necessitating repeat imaging. Grade C fistulas are those that manifest with any end-organ dysfunction, systemic illness, sepsis, or shock. The resource utilization, economic ramifications, as well as long-term outcomes related to each grade of fistula are yet to be established.
Various investigations have been held to assess the efficacy of somatostatin analogues, namely octreotide, in reducing the rate of postoperative pancreatic fistula formation. There have been several trials investigating the use of octreotide in order to reduce the risk of pancreatic leaks and fistulas. Studies from Europe have demonstrated mixed results, and those that have shown a decrease in leak rates have been in the setting of surgical treatment of chronic pancreatitis or for those undergoing distal pancreatectomies.57-59 In a study from MD Anderson Cancer Center, Lowy et al60 examined the use of octreotide with a prospective, randomized trial of 180 patients undergoing PD specifically for malignancy (58% PDA) in which they were randomized to octreotide treatment for the first five postoperative days or no additional treatment. They found no statistically significant difference in either the type or number of complications between the two treatment arms. Yeo et al61 did a similar study in which PD was the only surgical procedure to treat periampullary disease (39% for PDA) and even attempted to control for pancreatic texture (soft, intermediate, or hard). There was no difference in complication rate or pancreatic fistula with the use of octreotide, yet there was an increase in pancreatic fistula rate with soft-textured pancreas versus hard-textured irrespective of octreotide use (25% vs. 0%, respectively; p < 0.001). No data exist suggesting higher complication rates, and many still advocate its use in select high-risk patients, accepting the lack-of-quality comparative effectiveness data regarding cost.
While most pancreatic cancers reside in the pancreatic head at the time of diagnosis, over 25% are to the left of the SMV-portal confluence. For these, operative strategies necessitate a different approach than PD. First described in 1913 by Mayo, distal pancreatectomy requires en bloc mobilization of the pancreas and spleen followed by subsequent pancreatosplenectomy. Similar to PD, the operative goal for resection of tumors of the body and tail of the pancreas consist of tumor resection with adequate margin and adequate clearance of the appropriate lymph node basin. The N1 stations of the body and tail of the pancreas have been described by O'Morchoe.62 The concept surmises that the lymphatic drainage of the body and tail travels along small lymphatic channels on the superior and inferior borders of the pancreas, and are separated based on a four-quadrant system that drains to two large nodal basins. The first is a set of nodes that form a ring around the pancreas, with lymphatic drainage from the right half of the pancreatic body draining to the gastroduodenal and infrapancreatic nodes. Lymphatic drainage from the left half of the pancreatic body travels to the splenic hilum and nodes along the gastrosplenic omentum. The second basin lies anterior to the aorta in proximity to the celiac and SMA axes. While primarily N2 nodes, there are direct lymphatic communications from the pancreas indicating that some nodes are N1 stations.
First described in 2003, Strasberg et al63 adapted a radical anterograde modular pancreatosplenectomy (RAMPS) approach to adequately excise these aforementioned drainage basins. The technique is summarized in the following steps: (1) Step 1 involves accessing the lesser sac, identifying the infrapancreatic SMV, portal dissection, and lymph node clearance of the PV and hepatic artery. The pancreas is transected along the SMV and PV confluence. (2) Step 2 involves an adequate lymphadenectomy around the celiac and left gastric vessels. (3) Step 3 involves dividing the splenic vein at its confluence with the SMV and skeletonization of the SMA. (4) Step 4 ensures an adequate posterior plane often necessitating the dissection of the left renal vein, Gerota's fascia, and adrenal gland.
Using this technique, Strasberg et al64 demonstrated a 91% negative tangential margin rate, with survival comparable to that after PD. Further data collection by Mitchem et al65 suggested a comparable but slightly lower R0 resection rate of 81% with actuarial 5-year survival of 30.4%. Strasberg and Field66 have continued to modify this technique with recent data regarding celiac axis resection now becoming available. Though there is high variability in the results, the proposed benefits of this have not yet been proven with mature data.
Regardless of the operative approach, pancreatic fistula rates remain high after distal pancreatectomy, with rates as high as 64% in a large systematic review. While surgeon experience and patient factors are regarded as relevant risk factors for developing a pancreatic fistula, the optimal surgical technique is unclear. Several techniques to reduce the risk of leak are suggested in the literature, including hand-sewn and stapled stumps, application of mesh, omental, or enteric serosal patches, and pancreaticoeneteric anastomoses. While the literature reports stapled techniques and hand-sewn techniques more frequently, their superiority has not been established. In a large multicenter randomized trial published in The Lancet,67 the authors examined various different techniques for pancreatic stump management. Patients were randomized to either scalpel transection with hand-sewn closure or stapled transection. Details included complete pancreatic transection with a scalpel, followed by either a single stitched or continuous running suture closure of the stump using an absorbable suture in the hand-sewn group. Stapled transections were performed using a linear stapling device, with variations on staple height based on surgeon preference (1.0 to 2.5 mm). Institutions were free to use somatostatin analogues postoperatively based on their own prestudy protocol. Additionally the decision to perform splenectomy was left to the surgeon. The authors demonstrated that no statistical difference existed in postoperative fistula formation rates depending on technique, even when subset analysis was used to assess fistula grade. Results of a post hoc binary multiple logistical regression models of over 15 different factors revealed that only length of hospital stay was associated with fistula formation. Given this, no definitive consensus exists regarding optimal management of the remnant pancreatic stump after distal pancreatectomy.
MINIMALLY INVASIVE DISTAL PANCREATECTOMY
Similar to PD, distal pancreatectomy has been approached through minimally invasive techniques. Given that there is no need for a pancreatic anastomosis, far more advancements have been made in this area for distal pancreatectomies. Cuschieri68 first reported successful laparoscopic resection of the distal pancreas in 1994, followed shortly thereafter by Gagner et al's69 report of eight cases of laparoscopic distal pancreatectomy with splenic preservation. Since that time over 400 papers have been published in the English literature showing results of minimally invasive approaches to distal pancreatic resection. These include laparoscopic, laparoscopically and robotic-assisted distal pancreatectomies. Few large-scale, multicenter trials exist to date. In critical examination of the literature, no generalized accepted operative approach exists. Most surgeons have patients positioned either supine or with the left side elevated. Port placement is done under direct visualization and should include a 12-mm trocar to accommodate an endoscopic stapling device. Typically the surgery commences by wide entrance of the lesser sac, with division of the short gastric vessels. This is usually facilitated through the aid of an ultrasonic dissector. In the setting of malignancy, all efforts should be taken to ensure an adequate lymphadenectomy, consistent with that of the RAMPS technique. This can be initiated by dividing adhesions between the posterior stomach wall and pancreas, followed by defining the infrapancreatic border. While some advocate dividing the pancreas at the SMV/portal junction, others have divided the pancreas farther leftward using intraoperative ultrasound guidance to ensure adequate margins. An umbilical tape or vessel looped can be passed circumferentially to allow for traction to be placed on the pancreas to assist in the posterior dissection. Similarly, a hand port can be placed to allow for the surgeon to manually provide countertraction. Most advocate dividing the pancreas en bloc with the splenic vein, and harvesting the splenic artery separately. Management of the pancreatic transection line is highly variable. With this general approach, laparoscopic distal pancreatectomy can be performed safely. Venkat et al70 demonstrated lower operative blood loss and shorter hospital stays (p < 0.001) when distal pancreatectomy was performed for all causes. Additionally, they showed that overall complications were significantly lower in the laparoscopic group (33.9% vs. 44.2%; odds ratio [OR] = 0.73, 95% confidence interval [CI] 0.57 to 0.95), as was surgical site infection (2.9% vs. 8.1%; OR = 0.45, 95% CI 0.24 to 0.82).
Despite the general acceptance of the effectiveness of minimally invasive approaches for distal pancreatic resection, little additional literature exists regarding its utility specifically in the setting of adenocarcinoma. Magge et al71 showed that after adjusting for comorbidity and year of surgery, there were no significant differences in overall survival between open distal pancreatectomy and laparoscopic approaches. They again demonstrated lower operative blood loss and hospital stay favoring minimally invasive approaches, even when patients who necessitated conversion to open distal pancreatectomy were analyzed with intention to treat. Kooby et al72 recapitulated results from other distal pancreatectomy trials, showing that overall complications were significantly lower in the laparoscopic group (33.9% vs. 44.2%; odds ratio [OR] = 0.73, 95% confidence interval [CI] 0.57 to 0.95), as was surgical site infection (2.9% vs. 8.1%; OR = 0.45, 95% CI 0.24 to 0.82), when performed specifically for adenocarcinoma.