Surgical resection of lesions of the head of the pancreas

INTRODUCTION — Resection of the head of the pancreas is indicated primarily for neoplasms and necessitates concomitant duodenal resection. It is typically accomplished with a pancreaticoduodenectomy, which may also be performed to manage pancreatic or duodenal trauma and chronic pancreatitis.

Pancreaticoduodenectomy is a complex, high-risk surgical procedure. The lowest operative mortality rates and best long-term cancer outcomes have been demonstrated at high-volume centers [1,2]. In experienced hands, the median operative time for the Whipple procedure is 5.5 hours, with a median blood loss of 350 mL and mortality of less than 4 percent [3].

The perioperative management and general techniques for resection of the head of the pancreas are reviewed here. An overview of the surgical management of cancers involving the exocrine pancreas is presented in detail elsewhere. (See "Overview of surgery in the treatment of exocrine pancreatic cancer and prognosis".)

The techniques for resection of the body and tail of the pancreas are discussed elsewhere. (See "Surgical resection of lesions of the body and tail of the pancreas".)

The techniques of resection of sporadic pancreatic neuroendocrine tumors are discussed elsewhere. (See "Surgical resection of sporadic pancreatic neuroendocrine tumors".)

Surgery for chronic pancreatitis is discussed in another dedicated topic. (See "Surgery for chronic pancreatitis".)

INDICATIONS FOR PANCREATIC HEAD RESECTION — The most common indication for resection of the head of the pancreas is the presence of a malignant or premalignant neoplasm of the pancreas or one of the other periampullary structures (bile duct, ampulla, or duodenum) [4-6]. Certain types of neoplasms may be amenable to local excision (enucleation), and some benign conditions may also require pancreatic head resection (eg, trauma, chronic pancreatitis).

Specific indications for pancreatic head resection include:

●Pancreatic adenocarcinoma (see "Overview of surgery in the treatment of exocrine pancreatic cancer and prognosis")

●Ampullary carcinoma (see "Ampullary carcinoma: Epidemiology, clinical manifestations, diagnosis and staging")

●Ampullary adenoma (see "Ampullary adenomas: Management")

●Neuroendocrine tumors (see "Classification, epidemiology, clinical presentation, localization, and staging of pancreatic neuroendocrine neoplasms")

●Cholangiocarcinoma (see "Adjuvant and neoadjuvant therapy for localized cholangiocarcinoma")

●Duodenal neoplasm (see "Treatment of small bowel neoplasms")

●Combined pancreatic/duodenal injury (see "Management of duodenal trauma in adults" and "Management of pancreatic trauma in adults")

●Chronic pancreatitis (see "Chronic pancreatitis: Management" and "Surgery for chronic pancreatitis")

PREOPERATIVE EVALUATION

Medical assessment and optimization — Patients with indications for resecting the head of the pancreas frequently have significant medical comorbidities. Since most pancreatic resections are performed under elective circumstances, there should be adequate time to assess risk factors and optimize the patient's medical status.

Preoperative medical assessment is discussed elsewhere. (See "Preoperative evaluation and management of patients with cancer" and "Evaluation of cardiac risk prior to noncardiac surgery" and "Evaluation of perioperative pulmonary risk" and "Preoperative medical evaluation of the healthy adult patient".)

For patients taking antiplatelet therapy for primary or secondary prevention of cardiovascular disease, or other indications, we generally stop clopidogrel and full aspirin (325 mg) for at least one week before surgery but allow patients to continue to take baby aspirin (81 mg). Although cessation of aspirin prior to elective pancreatic surgery has been done routinely, evidence suggests that this may not be necessary. In a retrospective review of 1017 patients undergoing pancreas resection, 28.4 percent were maintained on aspirin therapy throughout the perioperative period [7]. Among these patients, 699 underwent pancreaticoduodenectomy, 204 in the aspirin group and 484 in the nonaspirin group. Overall, there were no significant differences between the aspirin and nonaspirin groups for intraoperative blood loss, rate of blood transfusion, or other major procedure-related complications. The aspirin group had a slightly increased rate of cardiovascular complications, but the difference was not significant (10.1 versus 7 percent). (See "Perioperative medication management", section on 'Aspirin'.)

Although many of these patients have lost weight, most are still adequately nourished to safely undergo the pancreatic head resection. However, if the serum albumin is <3 g/dL, or if surgery must be delayed for more than two weeks, we prescribe supplemental nutrition [8]. Nutrition can usually be provided enterally (Ensure, or another similar preparation according to patient preference, for five days before the operation), but on occasion parenteral nutrition may be needed. (See "The role of parenteral and enteral/oral nutritional support in patients with cancer", section on 'The perioperative setting' and "The role of parenteral and enteral/oral nutritional support in patients with cancer" and "Overview of perioperative nutrition support", section on 'Indications'.)

Once a decision has been made to proceed with surgery, the patient should be informed of the potential for surgical complications, as well as the possibility that pancreatic resection may not be possible if unrecognized metastatic or locally advanced unresectable disease is identified.

Preoperative biliary drainage — Patients with pancreatic cancer who are jaundiced are at risk for associated coagulopathy, malabsorption, and malnutrition. It was hoped that relief of biliary obstruction preoperatively would correct these defects and decrease postoperative morbidity and mortality rates. However, results from randomized trials have not been consistent. Uncertainty as to the benefit of preoperative biliary drainage has led to differing approaches. Some surgeons routinely decompress jaundiced patients with an endoscopically placed stent prior to surgery. However, most surgeons (including the author) reserve biliary decompression for patients whose serum bilirubin concentration exceeds 20 mg/dL, for those in whom surgery will be delayed for longer than two weeks, and in those with debilitating pruritus or cholangitis.

Although animal studies suggest that biliary drainage prior to pancreatic resection improved surgical outcomes, clinical studies in humans have not consistently shown a benefit. Multiple randomized trials have been conducted in patients with potentially resectable pancreatic cancer and malignant obstructive jaundice, with one suggesting benefit for preoperative biliary drainage [9], five no benefit [10-14], and three potential harm [10,11,15]. An earlier Cochrane review had concluded that stented patients had significantly fewer postoperative complications but similar postsurgical mortality [16]. However, this analysis did not include data from the subsequent largest randomized trial of preoperative biliary drainage versus no drainage in 202 patients with cancer of the head of the pancreas, which showed a nonsignificant trend toward more surgery-related complications with preoperative biliary drainage (47 versus 37 percent, relative risk 0.79, 95% CI 0.57-1.11) [15]. Later meta-analyses identified six trials (mostly patients with malignant obstructive jaundice, but some with benign disease) and included the larger, later trial in patients with pancreatic cancer [17-19]. Again, no significant differences in mortality were seen, but overall serious morbidity was higher in the preoperative drainage group (60 per 100 patients versus 26 per 100 patients).

When stenting is chosen for jaundiced patients who will undergo pancreatic head resection, the type of stent placed should be reviewed with the surgeon prior to placement. We suggest a plastic biliary stent (≥10 French), keeping the proximal end of the stent at or below the level of the cystic duct takeoff. Metallic stents often incite a severe inflammatory reaction and may be incorporated into the bile duct wall, making them difficult to remove at the time of surgery. A metal stent that is placed too high into the common bile duct can complicate the surgery, although this is less of a concern when the latest covered metal stents are used. However, some surgeons have not found significant adverse effects using self-expanding metallic stents [20]. (See "Endoscopic stenting for malignant biliary obstruction" and "Supportive care for locally advanced or metastatic exocrine pancreatic cancer".)

Imaging — All patients should undergo preoperative imaging that includes multiphasic high-resolution pancreatic protocol computed tomography (CT) with 1 to 2 mm cuts through the pancreas during the early arterial phase and venous phase [21-27].

Staging — Preoperative imaging evaluation to assess resectability [28], the role of endoscopic ultrasound (EUS), and radiographic staging are discussed in detail elsewhere. (See "Clinical manifestations, diagnosis, and staging of exocrine pancreatic cancer", section on 'Imaging studies' and "Endoscopic ultrasound in the staging of exocrine pancreatic cancer".)

Vascular evaluation — Routine preoperative CT helps to identify hepatic vascular anatomy and prepares the surgeon for any potential vascular anomalies that may be present [29]. The most common aberration in hepatic arterial anatomy is the replaced right hepatic artery arising from the superior mesenteric artery (figure 1). Every attempt should be made to preserve the aberrant vessel unless its resection is oncologically indicated.

In addition, preoperative CT can detect hemodynamically significant mesenteric arterial stenosis. If needed, significant mesenteric arterial stenosis can be treated with a stent to minimize the potential for postoperative ischemic complications [30].

In patients with pancreatic cancer for whom vascular resection is thought to be beneficial, preoperative consultation with a vascular surgeon is ideal to assist with the preoperative vascular evaluation, identify appropriate vascular conduit, and allow coordinated operative scheduling. The experienced surgical oncologist may undertake vascular resection and replacement without assistance, but having a vascular surgeon on standby is prudent. (See 'Vein excision and reconstruction' below and "Overview of surgery in the treatment of exocrine pancreatic cancer and prognosis", section on 'Vascular resection'.)

PREPARATION

Bowel preparation — We instruct patients to eat normally prior to surgery with standard cessation of oral intake beginning at midnight prior to the operation as determined by the American Society of Anesthesiology guidelines. Bowel preparation is not necessary prior to pancreatic resection. (See "Overview of colon resection", section on 'Bowel preparation'.)

One small randomized trial did not identify any beneficial effects for preoperative probiotics or selective decontamination [31]. These were not found to have any influence on bacterial translocation, intestinal permeability, or inflammatory mediator expression.

Antibiotics — We routinely give prophylactic antibiotics before pancreaticoduodenectomy. The first dose is given before the incision is made, and the last dose is given on the evening of the day of operation. The choice of antibiotics should be based on local hospital data that reflect characteristic bacterial pathogens (table 1). (See "Antimicrobial prophylaxis for prevention of surgical site infection in adults" and "Antimicrobial prophylaxis for prevention of surgical site infection following gastrointestinal procedures in adults".)

In a North American randomized trial of 778 patients undergoing pancreaticoduodenectomy, prophylactic piperacillin-tazobactam reduced 30-day surgical site infection rate from 33 to 20 percent compared with cefoxitin (one of the current prophylactic antibiotics choices (table 1)) [32]. Patients who received piperacillin-tazobactam also had a lower rate of postoperative sepsis (4.2 versus 7.5 percent), Clostridioides difficile colitis (0.3 versus 3.5 percent), and clinically relevant postoperative pancreatic fistula (13 versus 19 percent). This trial was terminated early after the second scheduled interim analysis, and no intraoperative bile culture was obtained.

Some have speculated that antibiotics based upon the results of bile cultures obtained at the time of preoperative biliary stenting or intraoperatively would decrease the incidence of infectious complications following pancreaticoduodenectomy [31,33-35]. In a meta-analysis of seven studies, targeted antibiotics were associated with a significantly lower rate of overall postoperative surgical site infections (risk ratio 0.55, 95% CI 0.37-0.81), superficial/deep incisional site infections (risk ratio 0.33, 95% CI 0.18-0.59), and organ space site infection (risk ratio 0.54, 95% CI 0.38-0.76) compared with standard antibiotic therapy [36]. Targeted antibiotics (eg, piperacillin-tazobactam or ampicillin-clavulanic acid) typically cover both Enterococcus and Enterobacter species.

For patients found to have bacteriobilia, the duration of antibiotic therapy remains in question. In a retrospective review of 197 patients who underwent pancreaticoduodenectomy, patients in a "control" group received only perioperative prophylactic antibiotics and were compared with a "treatment" group who were continued on antibiotics until intraoperative bile culture results became available [35]. In the "treatment" group, patients with bacteriobilia received 10 days of antibiotic treatment, which was otherwise discontinued in patients without bacteriobilia. Treatment of bacteriobilia was associated with a decreased rate of postoperative wound infections compared with only perioperative antibiotics in the "control" group (3 versus 12 percent). However, fewer patients in the control group had undergone preoperative biliary stenting (48 versus 67 percent), but a higher number of intraperitoneal drains were placed at the time of resection in the control group (85 versus 38 percent).

Thromboprophylaxis — We recommend pharmacologic thromboprophylaxis for patients undergoing pancreaticoduodenectomy. We also place intermittent pneumatic compression devices prior to induction and continue their use until the patient is ambulatory.

Patients undergoing pancreaticoduodenectomy are at moderate-to-high risk for venous thromboembolism due to the nature of the surgery (major open surgery >45 minutes) (table 2). The presence of malignancy increases the risk, which may require an extended period of pharmacologic thromboprophylaxis [37]. Patients with pancreatic adenocarcinoma have an even greater risk for venous thromboembolism than patients with other common adenocarcinomas [38].

The choice and suggested duration of pharmacologic thromboprophylaxis are discussed elsewhere. (See "Prevention of venous thromboembolic disease in adult nonorthopedic surgical patients".)

A small retrospective study of patients undergoing pancreatic surgery for malignancy found that the percentage of patients with venous thromboembolism decreased from 17.6 to 2.6 percent after implementation of a protocol that encouraged preoperative pharmacologic prophylaxis (5000 units of subcutaneous heparin) by including it in the preoperative checklist [39].

SURGICAL PRINCIPLES

Staging laparoscopy — Small hepatic or peritoneal metastases, which are not evident from the preoperative studies, may be found at surgical exploration in 8 to 15 percent of cases. As an example, 117 of 1423 pancreatectomies (8 percent) intended for resection of pancreatic ductal adenocarcinoma were aborted, most often due to occult liver (79 percent) or peritoneal metastasis (16 percent) [40]. For this reason, some surgeons prefer to begin cancer operations with laparoscopy, which permits examination of the liver and peritoneal surfaces and biopsy of any suspicious areas. If a metastatic tumor is found, laparotomy may be avoided. (See "Clinical manifestations, diagnosis, and staging of exocrine pancreatic cancer", section on 'Staging laparoscopy'.)

For patients undergoing pancreatic head resection, a drawback of staging laparoscopy is the additional time and expense required for the laparoscopic instrumentation and the inability to determine whether any vascular invasion has occurred that would preclude an attempt at resection. Thus, we restrict the use of laparoscopy to cases where there is a high likelihood of unresectability that has not been confirmed preoperatively. Examples include some patients with pancreatic head cancer and computed tomography (CT) evidence of liver surface or other intraperitoneal metastases that have not been proven with biopsy, and patients with pancreatic head cancer and ascites, which is probably due to unrecognized peritoneal metastases. In a retrospective study of occult metastatic disease in pancreatic ductal adenocarcinoma, independent predictors include indeterminate lesions on preoperative CT, tumor size >30 mm, abdominal pain, and preoperative CA 19-9 greater than 192 U/mL [40].

If unresectable disease is found at the time of laparoscopy and the patient has unrelieved obstructive jaundice or is at risk for gastric outlet obstruction, we perform a palliative gastric and/or biliary bypass laparoscopically in most cases [41]. Although postoperative endoscopic placement of a metal biliary stent is an acceptable alternative to operative biliary bypass, it does not work for all patients and is associated with its own complications. (See 'Palliative surgery' below and "Supportive care for locally advanced or metastatic exocrine pancreatic cancer", section on 'Jaundice'.)

Lymphadenectomy — The standard regional lymph node basin for tumors located in the head and neck of the pancreas includes those along the common bile duct; common hepatic artery; portal vein; pyloric, posterior, and anterior pancreaticoduodenal arcades; and along the superior mesenteric vein and right lateral wall of the superior mesenteric artery (figure 2). It is important to identify and properly assess as many regional lymph nodes in the specimen as possible. Based on survival data, evaluation of a minimum of 12 lymph nodes is recommended to accurately stage N0 tumors [42]. The total number of positive lymph nodes is now used to classify the N category of the TNM (Tumor, Node, Metastasis) classification according to the 8^th edition of the American Joint Committee on Cancer (AJCC) cancer staging manual. Metastasis in one to three regional lymph nodes is N1 disease, while metastasis in four or more regional lymph nodes is N2 disease (table 3).

Tumor involvement outside of this regional basin is considered distant metastasis. There is no evidence to support benefit from extended lymphadenectomy [43], and we do not perform extended lymph node dissection. As an example, we biopsy any enlarged or suspicious aortocaval lymph nodes and abort the pancreaticoduodenectomy if the biopsy is positive on frozen section. (See "Overview of surgery in the treatment of exocrine pancreatic cancer and prognosis", section on 'Extent of lymphadenectomy'.)

Vein excision and reconstruction — Pancreaticoduodenectomy with portal vein (PV) or superior mesenteric vein (SMV) resection and reconstruction should be considered a standard approach for pancreatic adenocarcinomas that focally involve the PV or SMV, provided that adequate inflow and outflow veins are present, the tumor does not involve the superior mesenteric artery or hepatic artery, and an R0 resection can be accomplished [44]. Nevertheless, many surgeons, including the author, prefer to treat patients with PV or SMV involvement with neoadjuvant systemic chemotherapy for at least six months before the operation is performed, as it increases the R0 resection rate and overall survival [45]. (See "Overview of surgery in the treatment of exocrine pancreatic cancer and prognosis", section on 'Vascular resection'.)

In a single-center study from Norway, 16 percent of 784 patients who underwent pancreaticoduodenectomy required vein excision and reconstruction [46]. Both operative time (422 versus 312 minutes) and blood loss (700 versus 500 mL) were increased with vein resection. Severe postoperative complications were also more prevalent after vein resection, with one in four patients requiring relaparotomy (most commonly for bleeding). This reoperation rate is higher than reported in the literature for this group of patients; a meta-analysis of 17 studies reported an average reoperation rate of 12 percent (range 0 to 48 percent) [47].

The techniques of vein reconstruction vary and should be guided by preoperative evaluation of vascular involvement [48].

Drainage of the pancreatic bed — Drain placement after pancreatic resection remains common [49,50]. The rationale for leaving a drain after a pancreaticoduodenectomy is to provide controlled external drainage of a pancreatic leak. Inadequate drainage of a leak can lead to an intra-abdominal abscess.

However, the routine use of intra-abdominal drains after pancreaticoduodenectomy is controversial [50-52]. While several observational studies [44,53-56] and two randomized trials [57,58] failed to demonstrate any benefit of drain placement, another trial reported that drain placement lowered the morbidity and mortality rates as well as the severity of complications in patients who underwent pancreatic resection [59].

A 2018 Cochrane meta-analysis of four randomized trials suggested that routine abdominal drainage probably reduced mortality slightly at 90 days (moderate-quality evidence), that use of an active drain compared with the use of a passive drain may slightly reduce the length of hospital stay (low-quality evidence), and that early removal may be superior to late removal for people with low risk of postoperative pancreatic fistula [60].

Thus, until better evidence becomes available, we suggest that drains be placed selectively following pancreatic resection in patients who are at a high risk of developing pancreatic fistula (eg, those with a soft and friable pancreas, those with a small [<2 mm] pancreatic duct) and that drains be removed early when there is no evidence of fistula development. (See 'Drain management' below.)

PANCREATICODUODENECTOMY

Conventional versus modified pancreaticoduodenectomy — The standard operation for pancreatic cancer within the head or uncinate process of the pancreas is pancreaticoduodenectomy. Modifications of conventional pancreaticoduodenectomy (ie, Whipple procedure) have been developed in an attempt to improve outcomes or minimize the morbidity associated with the operation. These include the pylorus-preserving pancreaticoduodenectomy and subtotal stomach-preserving pancreaticoduodenectomy.

The available data suggest that, for suitable cases, perioperative morbidity and mortality and long-term survival are not affected by the use of pylorus-preserving techniques [61,62]. The impact of a pylorus-preserving approach on gastrointestinal function remains an open question, but most agree that the incidence of delayed gastric emptying is not increased when compared with the conventional operation. (See 'Delayed gastric emptying' below.)

Given the potential time savings, lower rate of blood loss, and similar oncologic outcomes, we perform a pylorus-preserving pancreaticoduodenectomy on almost all of our patients [61,63]. However, a conventional pancreaticoduodenectomy (Whipple) operation should be performed if the tumor involves the proximal duodenum, pylorus, or gastric antrum.

Conventional pancreaticoduodenectomy (Whipple) — Conventional pancreaticoduodenectomy involves removal of the pancreatic head, duodenum, first 15 cm of the jejunum, common bile duct, and gallbladder, and a partial gastrectomy (figure 3A).

Pancreaticoduodenectomy is accomplished in an orderly fashion, usually in the following sequence.

●The abdomen is assessed for metastatic disease.

●The duodenum and the head of the pancreas are mobilized, and the superior mesenteric vein is isolated.

●The stomach and remainder of duodenum are mobilized and the proximal extent of the resection is determined.  

●The structures of the porta hepatis are skeletonized.

●The gallbladder is removed and the common hepatic duct is transected.

●The proximal jejunum is mobilized and transected.

●The neck of the pancreas is transected and the remaining attachments of the specimen to the superior mesenteric vein, portal vein, and the superior mesenteric artery are divided.

●Gastrointestinal reconstruction is accomplished by providing drainage of the pancreatic stump and common hepatic duct, and restoring proximal and distal gastrointestinal continuity.

Pylorus-preserving pancreaticoduodenectomy — Pylorus-preserving pancreaticoduodenectomy is a somewhat less radical operation that is being increasingly performed worldwide. The procedure preserves the gastric antrum, pylorus, and proximal 3 to 4 cm of the duodenum (figure 4). Pylorus preservation may decrease the incidence of postoperative dumping and bile reflux gastritis that can occur in some patients undergoing partial gastrectomy.

Several small randomized trials have directly compared a pylorus-preserving with a conventional pancreaticoduodenectomy in pancreatic or periampullary cancer, but with conflicting results [64-71]. Data from uncontrolled studies are also conflicting [72-75].

A 2016 updated meta-analysis from the Cochrane database included eight trials that provided quantitative data on 512 patients [61]. Heterogeneity among the trials was noted by the authors with respect to methodology and outcome parameters. The overall impact of a pylorus-preserving procedure on gastrointestinal function remains unclear. The following findings were noted:

●There were no differences between the two procedures in terms of survival, postoperative mortality, and main morbidities (eg, pancreatic fistula, biliary leakage, or postoperative bleeding), except for delayed gastric emptying.

●Delayed gastric emptying was significantly more common after a pylorus-preserving procedure than after conventional surgery (31 versus 24 percent, odds ratio [OR] 3.03, 95% CI 1.05-8.70).

●Operating time (weighted mean difference -45 minutes, 95% CI -75 to -16 minutes), red blood cell transfusion (weighted mean difference -0.47 units, 95% CI -0.86 to -0.07 units), and intraoperative blood loss (weighted mean difference -0.32 L, 95% CI -0.62 to -0.03 L) were all significantly reduced with the pylorus-preserving procedure.

The technique for pylorus-preserving pancreaticoduodenectomy is described in a dedicated topic. (See "Pylorus-preserving pancreaticoduodenectomy".)

Subtotal stomach-preserving pancreaticoduodenectomy — Subtotal stomach-preserving pancreaticoduodenectomy aims to preserve as much stomach as possible while minimizing problems related to delayed gastric emptying that are associated with preserving the pyloric ring in the face of vagal denervation [76]. In this procedure, the duodenum, pylorus, and 1 to 2 cm of stomach are resected with the pancreatic specimen. Although described, this modification has yet to be validated, and it is uncommonly performed. Due to the retained antrum producing large amounts of gastrin, the incidence of gastric ulceration may be higher with stomach-preserving techniques compared with conventional pancreaticoduodenectomy.

One study has retrospectively compared stomach-preserving with standard or pylorus-preserving pancreatoduodenectomy [76]. Although the rate of delayed gastric emptying was marginally less for a stomach-preserving compared with a pylorus-preserving operation, there was no difference from conventional pancreaticoduodenectomy. One trial that compared a stomach-preserving with pylorus-preserving pancreaticoduodenectomy (50 patients in each group) found a reduced incidence of delayed gastric emptying for the stomach-preserving technique (10 versus 6 patients) [77]. There were no differences in the rates of other complications.

Gastrointestinal reconstruction — Following resection, gastrointestinal reconstruction provides a drainage route for the remaining pancreas, the bile duct, and the stomach. A number of methods can be used to accomplish this, but most commonly the free end of jejunum is attached first to the pancreas, next to the bile duct, and last to the stomach or duodenal remnant (figure 3A). Other techniques employ a separate Roux limb of jejunum to drain the stomach (figure 3B), or the Roux limb can be used to drain the pancreas alone, while the proximal jejunum drains the bile duct and stomach.

Pancreatic-enteric anastomosis — Options for managing the pancreatic stump include a pancreaticojejunostomy (PJ) (figure 3A-B) or a pancreaticogastrostomy (PG) (figure 5). The choice is determined by surgeon preference as studies comparing the two techniques reported conflicting results.

Pancreaticojejunostomy versus pancreaticogastrostomy — The optimal method of pancreatic-enteric anastomosis is not clear. Some studies comparing PJ with PG show equivalency [78-83]; others show that fewer postoperative fistulas develop after a PG [78,84,85]. One of the largest trials, however, reported an increased rate of postoperative bleeding events after PG [83].

A meta-analysis of seven randomized trials showed that PG resulted in significantly lower rates of pancreatic (11 versus 19 percent) and biliary fistulas (2 versus 5 percent) compared with PJ [86]. A later Cochrane review of 10 trials, however, showed little or no difference between PG and PJ in rates of overall pancreatic fistula (21.4 percent PG versus 24.3 percent PJ), clinically significant pancreatic fistula (12.8 percent PG versus 19.3 percent PJ), or postoperative mortality (4.8 percent PG versus 3.9 percent PJ) [87].

Based upon available data, some surgeons prefer pancreaticogastrostomy, especially when the pancreas is soft and friable, a condition associated with an increased risk of pancreatic fistula formation. Other surgeons who are experienced with and have had good personal results using pancreaticojejunostomy, however, are not likely to alter their practice [78-82].

Duct-to-mucosa versus invagination — For pancreaticojejunostomy, there are two predominant methods of reconstruction: end-to-side duct-to-mucosa anastomosis (figure 6) or invagination of the pancreatic remnant [88,89]. The invagination method can be performed either end-to-side or end-to-end (figure 7) [89,90]. The choice of the method is again determined by surgeon preference as there is no conclusive evidence to favor one method over the other.

In a 2018 systematic review and meta-analysis of eight trials with a total of 1043 patients undergoing pancreaticojejunostomy after pancreaticoduodenectomy, there was no significant difference between the two anastomotic techniques in terms of overall and clinically relevant postoperative pancreatic fistula (POPF) rate, as well as secondary outcomes. Furthermore, the authors concluded that existing evidence is sufficient to conclude lack of difference and further trials on this subject matter are unlikely to change the outcome [91]. A 2022 Cochrane review that included 11 trials reported similar findings [92].

Adjuncts — Additional techniques are aimed at reducing the incidence of postoperative pancreatic fistula formation. These include pancreatic stenting and application of tissue adhesives. However, study results are conflicted as to whether these adjuncts are helpful or not.

Pancreatic stenting — Placement of a pancreatic stent has the potential to decompress the main pancreatic duct and provide drainage of pancreatic secretions. However, the role of pancreatic stents in preventing pancreatic fistula is unclear. Available studies have important methodologic limitations. Until sufficient evidence is available, pancreatic stenting as a means to prevent postoperative pancreatic fistula following pancreaticoduodenectomy is according to surgeon preference.

At least eight randomized trials have been conducted to investigate the role of pancreatic stenting in pancreaticoduodenectomy. An additional five randomized trials are ongoing, the results of which have not been reported. A 2016 Cochrane review of four trials [93-96] found an overall reduction in the incidence of postoperative pancreatic fistulas for those who were stented compared with those who were not (20.7 versus 28.1 percent), but the difference was not statistically significant (risk ratio 0.67, 95% CI 0.39-1.14) [97]. Another meta-analysis of five trials [93-96,98] showed pancreatic stenting to have decreased the formation of clinically relevant fistulas (14.0 versus 25.8 percent, risk ratio 0.45, 95% CI 0.27-0.76).

Tissue adhesives — Studies on the use of tissue adhesives, particularly fibrin glue, for prevention of postoperative pancreatic fistula have produced conflicting results [99,100]. A 2020 Cochrane systemic review and meta-analysis of four randomized trials found that the effects of fibrin sealants on the prevention of postoperative pancreatic fistula are uncertain in people undergoing pancreaticoduodenectomy [101].

Biliary-enteric anastomosis — The proper hepatic duct is typically drained into the same loop of jejunum that is used to drain the pancreas and the stomach through a hepaticojejunostomy (figure 3A-B). (See "Bile duct resection and reconstruction".)

If the proper hepatic duct is smaller than 10 mm in diameter, a small T-tube may be inserted through the wall of the duct proximal to the anastomosis with the distal end of the T-tube placed through the anastomosis into the jejunal limb. An internal stent through the anastomosis is also satisfactory.

Gastrointestinal anastomosis — The manner in which gastrointestinal continuity is achieved depends upon the type of pancreaticoduodenectomy. In general, the stomach or duodenum (pylorus sparing) can be drained into the same proximal jejunal loop brought up to the right upper quadrant for the pancreatic and biliary anastomosis (figure 3A), or to a separate Roux limb of more distal jejunum (figure 3B). According to two randomized trials, there appear to be no significant differences in the incidence of postoperative pancreatic fistula [102] or delayed gastric emptying [103] whether the same loop or a separate Roux limb is used to drain the stomach/duodenum.

When a single jejunal loop is used, the pancreatic and biliary anastomoses are placed 45 to 60 cm proximal to the gastrojejunostomy. This assures that the gastrojejunostomy is bathed in alkaline secretions, reducing the risk of stomal ulceration. However, it does not prevent reflux of gastric juice and food from the stomach into the biliary tree and pancreas, which can cause cholangitis if the limb is short or obstructed [104-106]. Some surgeons add a Braun enteroenterostomy between the afferent and efferent limb to prevent or alleviate alkaline gastritis (figure 8), but we doubt its usefulness.

The gastrojejunostomy can be placed in front of the colon (antecolic, our preference) or behind the colon (retrocolic). Several small trials generated conflicting results regarding the incidence of delayed gastric emptying relative to the type of reconstruction [107-111]. A 2022 Cochrane systematic review and meta-analysis found no differences between the two techniques with respect to morbidity (including delayed gastric emptying and pancreatic fistula), mortality, length of hospital stay, and quality of life [112].

Open versus minimally invasive — Pancreaticoduodenectomy is a relative latecomer in the minimally invasive surgery revolution. The development of specialized instruments and technology was needed before minimally invasive pancreaticoduodenectomy was even technically feasible. Even with robotic techniques, minimally invasive pancreaticoduodenectomy is a complex operation that is suited for only selected patients in high-volume centers experienced with laparoscopic/robotic techniques [113]. As with open surgery, mortality after minimally invasive pancreaticoduodenectomy was higher in low-volume hospitals than in high-volume hospitals (7.5 versus 3.4 percent), according to a meta-analysis of two registry studies that included close to 20,000 patients [114].

A systematic review identified 27 retrospective studies that included close to 7000 patients undergoing pancreaticoduodenectomy (1306 minimally invasive, 5603 open) [115]. Conversion to an open procedure was required in 0 to 40 percent of cases initially attempted laparoscopically [116]. The minimally invasive approach was associated with longer operative times (mean difference [MD] 71 minutes, 95% CI 27-115) but lower intraoperative blood loss (MD -300 mL, 95% CI -400 to -200 mL). The rate of lymph node retrieval was significantly higher in the minimally invasive group (MD 1.34 nodes, 95% CI 0.14-2.53), and the likelihood of an R0 resection was also higher (OR 1.45, 95% CI 1.18-1.78). Hospital stay was significantly reduced for the minimally invasive group (MD -2.95 days, 95% CI -3.91 to -2 days). Minimally invasive pancreaticoduodenectomy was also associated with a reduction in postoperative hemorrhage (OR 1.60, 95% CI 1.03-2.49) and wound infection (OR 0.44, 95% CI 0.30-0.66), while the rate of overall mortality, reoperations, vascular resection, clinically relevant pancreatic fistula, delayed gastric emptying, and bile leak was similar between the groups.

Three randomized trials have compared laparoscopic pancreaticoduodenectomy with the open procedure:

●The first trial from India, which included 64 patients with periampullary tumors, largely echoed the findings of the above meta-analysis that laparoscopic pancreaticoduodenectomy required longer operative time but shorter hospital stay (13 versus 7 days) [117]. Intraoperative blood loss was also reduced with laparoscopic surgery. Other short-term surgical and oncologic outcomes, including mortality, major morbidity, rates of fistula, delayed gastric emptying, and postoperative hemorrhage, lymph node yield, and rate of R0 resection, were all comparable between patients who underwent open versus laparoscopic pancreaticoduodenectomy.

●The second trial from Spain, which included 66 patients, associated laparoscopic pancreaticoduodenectomy with a shorter hospital stay (13.5 versus 17 days), fewer severe complications (5 versus 11 patients), and a more favorable postoperative course [118]. Lymph node yield and resection margins were comparable between the two groups.

●The third trial from the Netherlands (LEOPARD-2), which included 105 patients, found the opposite [119]. Compared with open surgery, laparoscopic pancreaticoduodenectomy was associated with more major postoperative complications (50 versus 39 percent) and more complication-related mortalities (10 versus 2 percent) but no improvement in the rapidity of functional recovery (10 versus 8 days). Although these differences were not statistically significant, the trends were worrisome enough for the trial to be terminated early. Importantly, the trial was conducted in high-volume centers performing >20 procedures annually, and all cases were performed by experienced surgeons.

The only study comparing oncologic outcomes of laparoscopic versus open pancreaticoduodenectomy performed at a high-volume institution came from the Mayo Clinic [120]. Faster recovery and shorter hospitalization associated with laparoscopic pancreaticoduodenectomy allowed earlier initiation of chemotherapy, but overall rates of chemotherapy and overall survival were unchanged.

At some high-volume centers, robot-assisted pancreaticoduodenectomy has been adopted. The robotic platform allows for 3D vision, 540 degree wrist articulation, and tremor stabilization in addition to the conventional advantages of laparoscopy.

Robotic-assisted pancreaticoduodenectomy has only been studied in nonrandomized comparative studies [121-124]. A meta-analysis of 13 such studies found robotic pancreaticoduodenectomy to be associated with a longer operative time (by 71 minutes) but less blood loss (by 374 mL), shorter length of stay (by 5 days), and fewer wound infections than open surgery [125]. Otherwise, the positive margin rate, number of lymph node harvested, and mortality, morbidity, reoperation, and readmission rates were all comparable between robotic and open procedures.

Minimally invasive pancreatic surgery is discussed elsewhere. (See "Minimally invasive pancreatectomy (MIP)".)

ENUCLEATION PROCEDURES — Some small (ie, ≤1 cm) benign lesions in the head of the pancreas can be enucleated, which avoids the need for pancreaticoduodenectomy. Enucleation should not be used when the lesion is large (usually >2 cm) or when there is a chance that the lesion could be malignant. Larger lesions are associated with greater risk for pancreatic or bile duct injury, which could result in pancreatic or biliary fistula. If the lesion is malignant, enucleation does not remove surrounding soft tissue and lymph nodes, which may lessen the chance for cure and prevents adequate staging.

Enucleation is most commonly performed for pancreatic neuroendocrine tumors, especially insulinomas. A retrospective study of neuroendocrine tumors (20 insulinomas and 6 gastrinomas), which compared tumor enucleation of the pancreatic head with pancreaticoduodenectomy, showed similar overall complication rates (69 versus 80 percent), but the rate of clinically relevant pancreatic fistulas was higher after enucleation (40 versus 14 percent) [126]. The rate of reintervention and readmissions did not differ (49 versus 46 percent). However, the incidences of exocrine and endocrine insufficiency were significantly higher after pancreaticoduodenectomy (55 and 19 percent versus 5 and 7 percent).

Prior to enucleation in patients with tumors close to the ductal system, preoperative placement of a pancreatic duct stent may help the surgeon to avoid ductal injury and to identify a major ductal disruption if it occurs. The technique for enucleation is described elsewhere. (See "Surgical resection of lesions of the body and tail of the pancreas", section on 'Technique'.)

PALLIATIVE SURGERY — If the patient is found to have unresectable disease during the course of staging laparoscopy or during an attempted pancreaticoduodenectomy and downstaging neoadjuvant therapy is not an option (eg, liver or peritoneal metastases, very extensive local disease with complete major vascular occlusion, etc), surgical biliary and gastric bypass may be performed.

●A prophylactic antecolic gastrojejunostomy should also be done in most patients who are likely to survive for more than three months. It does not add to the morbidity of the operation.

●Biliary bypass (cholecysto- or choledochojejunostomy) effectively relieves biliary obstruction and eliminates the need for a subsequent biliary stent. If the gallbladder is used for the decompression, the surgeon should be certain that the cystic duct is open and is not likely to become obstructed by the tumor as it grows.

In a contemporary series of over 600 pancreaticoduodenectomies, 17 percent were aborted [127]. Palliative procedures in this cohort had far fewer complications (overall 17 percent, severe 9 percent) than historical series. Palliative procedures, however, delayed systemic therapy (median 47 versus 21 days), and 29 percent of patients never received additional treatments. Palliative procedures markedly decreased the need for future interventions (12 versus 79 percent). Intraoperative decisions regarding palliative procedures must incorporate the functional status and motivations of the patient; these procedures are increasingly safe but may still affect survival.

Alternatively, in experienced centers, endoscopic relief of biliary obstruction is another less invasive but equally efficacious option. The details of endoscopic treatment for malignant pancreaticobiliary obstruction are discussed elsewhere. (See "Endoscopic stenting for malignant biliary obstruction".)

POSTOPERATIVE CARE AND FOLLOW-UP — Most patients do not require intensive care unit monitoring following a pancreaticoduodenectomy. The majority of patients can be discharged within 7 to 10 days of the operation.

We do not routinely use a nasogastric tube after surgery, which did not reduce major complication rate or length of stay compared with no nasogastric tube in a randomized trial [128], and the patient is encouraged to ambulate early. (See "Inpatient placement and management of nasogastric and nasoenteric tubes in adults", section on 'Prophylactic placement' and "Overview of perioperative nutrition support".)

Negative pressure wound therapy (NPWT) has been used by some in an attempt to reduce wound complications. We use it selectively only in patients at high risk for wound infection. A trial of 123 patients undergoing open pancreaticoduodenectomy found that applying negative pressure wound therapy to the (closed) surgical incision at the time of closure reduced surgical site infection by 69 percent (9.7 versus 31.1 percent) [129]. (See "Principles of abdominal wall closure", section on 'Negative pressure dressings'.)

Routine postoperative management is primarily focused on nutrition, the return of bowel function, and management of closed-suction drain output.

Nutrition support — We prefer to feed patients orally after a pancreaticoduodenectomy, if possible, rather than via an alternative route (eg, jejunostomy). A clear liquid diet can usually be initiated by postoperative day 1 or 2; most patients are able to advance to a regular diet over the next 48 to 72 hours. Early oral intake is a component of almost all pancreaticoduodenectomy enhanced recovery after surgery (ERAS) protocols [130].

A trial randomly assigned 114 patients to oral or enteral feeding after pancreaticoduodenectomy [131]. Compared with enteral feeding, oral feeding did not increase the duration or grade of postoperative pancreatic fistulas and was associated with a shorter hospital stay and reduced costs.

A systematic review found that normal oral intake was established more quickly compared with other enteral nutrition routes (eg, jejunostomy, gastrojejunostomy, or total parenteral nutrition) and was associated with a shorter length of stay [132]. However, patients with perioperative complications may require additional nutritional support. In the same systematic review, approximately 25 percent of patients fed orally required nutritional support, either because oral intake was insufficient or due to complications such as pancreatic fistula.

General principles and practices of perioperative nutritional support are discussed separately. (See "Overview of perioperative nutrition support".)

Drain management — Although whether prophylactic abdominal drainage reduces postoperative complications after pancreatic surgery is controversial, the use of surgical drains is a very common practice [133].

Any closed-suction drain placed during a pancreaticoduodenectomy should be managed according to the volume and character of the output. The normal peripancreatic drainage should be clear or straw colored. Green-tinged fluid suggests the presence of bile. If the fluid is cloudy or turbulent, a pancreatic leak should be suspected and a fluid amylase level should be obtained. A drain fluid amylase level that is more than three times higher than the upper limit of normal for the serum amylase level defines a pancreatic leak.

If there is no evidence of a bile or pancreatic leak, the closed-suction drain placed at the time of surgery is usually removed once the patient is tolerating a regular diet. In one meta-analysis, a higher drain amylase concentration on the first day after major pancreatic resection was highly predictive of developing a pancreatic fistula and thereby requiring long-term drainage [134]. Although there is some evidence that early drain removal may reduce intra‐abdominal infection rate, morbidity, and length of hospital stay for people with a low risk of postoperative pancreatic fistula [133], most surgeons tailor drain care to the individual patient's progress, rather than arbitrary values of drain fluid amylase [135,136].

In our experience, most postoperative pancreatic and biliary fistulas close spontaneously, usually within four to six weeks after discharge [137]. If a pancreatic or bile leak is present and the patient is ready for discharge, the drain should remain in place and can be removed in the outpatient setting once the fistula has resolved.

Any T-tube placed to protect the hepaticojejunostomy can be removed in the office three to four weeks later. The T-tube can usually be removed without the need for a prior cholangiogram, provided that there is no evidence of a bile leak.

Pancreatic enzyme supplementation — Most patients who have undergone pancreaticoduodenectomy will not need pancreatic enzyme supplementation. Although many patients will have several bowel movements a day for some weeks following discharge from the hospital as the gastrointestinal tract "readjusts" to the altered anatomy, bowel frequency almost always decreases, and most patients end up with two to three formed bowel movements daily.

If steatorrhea is evident with oily floating stools, then pancreatic enzyme supplements should be prescribed with each meal (at least 30,000 units of lipase component/meal; one-half dose with snacks) (table 4). Antisecretory agents (histamine receptor antagonist, proton pump inhibitor) should also be given to decrease gastric acid secretion. It is almost never necessary to limit fat intake. If this approach does not provide relief, the patient should be referred to a gastroenterologist for further management. (See "Chronic pancreatitis: Management".)

Prophylactic somatostatin analogues — Somatostatin or its analogues (eg, octreotide, lanreotide, pasireotide, vapreotide), which reduce pancreatic, gastric, and enteric secretions, should theoretically be helpful for reducing postoperative pancreatic fistula, but studies evaluating this issue are conflicting [138-146]. (See "Physiology of somatostatin and its analogues".)

As a result, we do not recommend routine use of prophylactic somatostatin or any of its analogues (eg, octreotide) for patients undergoing resection of the head of the pancreas. Rather, somatostatin or its analogues may be used selectively to control the volume of output in patients who develop high-output pancreatic fistulas postoperatively.

No significant differences in mortality have been identified in several systematic reviews whether or not somatostatin or its analogues are used [139,140,142,147]. Somatostatin and its analogues may have an overall benefit in reducing morbidity following pancreatic operations, but the variable methods by which the agent is administered, outcomes reported, and definitions of complications, and particularly the definition of a pancreatic fistula, make it difficult to compare studies or clearly identify a subgroup for whom prophylactic treatment should be considered, and the optimal means by which it should be administered [148]:

●A 2013 Cochrane systematic review and meta-analysis that included 21 trials evaluating somatostatin and its analogues (predominantly octreotide; two studies lanreotide) following a pancreatic surgery (variety of indications and types of surgery) found a decreased incidence of pancreas-specific complications in those patients who received prophylactic somatostatin analogues postoperatively [147]. The incidence of pancreatic fistula was reduced among those who received somatostatin analogues compared with controls (relative risk [RR] 0.66, 95% CI 0.55-0.79); however, there were no significant differences in the rate of pancreatic fistula among the four trials that clearly distinguished clinically significant fistula. A significant decrease in the incidence of sepsis (RR 0.42, 95% CI 0.21-0.85), risk for any complication, and number of complications was seen in the somatostatin analogue group (RR 0.70, 95% CI 0.61-0.80; RR 0.70, 95% CI 0.60-0.82). No significant differences were identified for rates of reoperation, anastomotic leak, postoperative pancreatitis, renal failure, bleeding, abdominal collections, delayed gastric emptying, or length of hospital stay.

●A 2018 meta-analysis of eight randomized trials did not find that prophylactic octreotide significantly reduced the total number of postoperative pancreatic fistulas, the number of clinically significant fistulas, the length of hospital stay, or postoperative mortality [149].

●By contrast, a single-center trial evaluating pasireotide, which has a longer half-life than octreotide and a different binding profile, found that 14 doses of pasireotide administered during the first seven postoperative days reduced the risk of clinically significant pancreatic fistulas, leak, or abscess compared with placebo (9.2 versus 20.9 percent) [138], without increasing cost [150]. These reductions remained significant in the subgroups of patients who underwent pancreaticoduodenectomy (10 versus 21 percent), distal pancreatectomy (7 versus 23 percent), and who had dilated (2 versus 15 percent) or nondilated pancreatic duct (15 versus 27 percent). Another single-center nonrandomized study also found that the routine use of pasireotide reduced the incidence of clinically relevant pancreatic fistula/abscess (13 versus 21 percent) and biliary leakage (0.6 versus 3.4 percent), compared with historical controls [151]. These results, however, have not been reproduced by others [152]. Therefore, results from more multicenter randomized trials are needed before wider, routine use of pasireotide can be recommended.

Readmission prevention — According to a review of the Surveillance, Epidemiology, and End Results (SEER) database, nearly one-fourth of patients in the United States are readmitted to the hospital following pancreaticoduodenectomy [153]. Preoperative patient comorbidities, as opposed to provider- or hospital-specific variables, appear to be more important for predicting who will require readmission.

We advocate close surgeon-patient communication in the immediate postoperative period to avoid unnecessary readmissions. For those who undergo surgery at a tertiary referral center away from home, participation from a local surgeon or other clinicians familiar with the patient and with whom the operating surgeon can communicate can be very helpful.

PERIOPERATIVE MORBIDITY AND MORTALITY — In experienced high-volume centers, pancreaticoduodenectomy can be performed with a low postoperative mortality of <4 percent and morbidity of 40 to 50 percent. The five-year survival for completely resected patients is 20 to 30 percent. (See "Overview of surgery in the treatment of exocrine pancreatic cancer and prognosis", section on 'Outcomes of pancreaticoduodenectomy'.)

Mortality — Although previously associated with high morbidity and mortality rates, modern series show that in experienced hands, the conventional Whipple procedure is associated with a perioperative mortality rate of less than 4 percent, and 20 to 30 percent five-year survival in completely resected patients [154-160]. This relatively low perioperative mortality rate represents a decline from over 15 percent in the 1970s, making the pancreaticoduodenectomy a more attractive option.

One of the most important reasons for this is the greater experience of a limited number of surgeons who perform the procedure regularly in high-volume institutions [161-164]. The relationship between surgeon and hospital volume and postoperative mortality and survival after pancreatic cancer surgery has been evaluated by several meta-analyses [165,166]. In one meta-analysis that included 14 studies, there was a significant association between hospital volume and postoperative mortality (odds ratio [OR] 0.32, 95% CI 0.16-0.64), as well as overall survival (hazard ratio for death 0.79, 95% CI 0.70-0.89) [165]. However, morbidity rates remain high (40 to 50 percent).

In a French study of 3195 patients undergoing pancreatic resection at high-volume centers, 4.4 percent died within 90 days of surgery [167]. The cause of death was determined in 90 percent of deaths, of which 30 percent were preventable. Young age and arterial resection (especially unplanned) were often associated with avoidable mortality.

Previous observational studies have shown that mortality after pancreatic resection is related to failure to rescue rather than to complications [168]. As such, early recognition and minimally invasive management of postoperative complications may reduce mortality rates [169].

Delayed gastric emptying — Vomiting after oral intake is resumed can be a manifestation of delayed gastric emptying. The mean incidence of delayed gastric emptying is 17 percent, although the range varies widely among trials [170]. Possible risk factors for delayed gastric emptying include prior abdominal surgery, history of cholangitis, and diabetes mellitus [171-173]. It is important to not overlook the possibility of pancreatic fistula with an intra-abdominal fluid collection as a cause of delayed gastric emptying. Thus, an abdominal computed tomography (CT) scan should be done if vomiting occurs.

The incidence of delayed gastric emptying does not appear to differ following conventional versus pylorus-preserving pancreaticoduodenectomy [62] or single loop versus Roux-en-Y gastrointestinal reconstruction. (See 'Pylorus-preserving pancreaticoduodenectomy' above and 'Gastrointestinal anastomosis' above.)

Pancreatic fistula — Postoperative pancreatic fistula (POPF) is diagnosed when the amylase content of any measurable volume of fluid from the pancreatic drain on or after postoperative day 3 is greater than three times the upper limit of the normal serum amylase content, based on standards from the International Study Group on Pancreatic Fistulas (ISGPF) [174,175]. Clinically relevant postoperative pancreatic fistulas (grade B or C [176]) occur in approximately 5 to 10 percent of patients, although in some series the incidence is as high as 22 percent [177,178]. Pancreatic fistulas can lead to sepsis and hemorrhage if they are not adequately drained externally. These complications are associated with mortality of 20 to 40 percent, prolonged hospitalization, and increased hospital expenses [177].

The risk of developing POPF is increased with various factors, such as high body mass index (BMI), preoperative comorbidities such as jaundice, as well as a soft pancreas and a narrow pancreatic duct. Other factors that have been associated with an increased risk of POPF include drain amylase >4000 U/L on postoperative day 1, increased intraoperative blood loss, and prolonged operative time [177,179-181]. In one study, a risk score calculated using only BMI and pancreatic duct width on preoperative CT imaging, accurately predicted the risk of POPF [182].

In a study of 2752 patients under pancreatic resection, 256 patients (9.3 percent) developed a clinically relevant POPF (ISGPF grade B or C) [183]. Detection of microorganisms in the POPF was associated with a higher rate of post-pancreatectomy hemorrhage (42 versus 22 percent), sepsis (38 versus 7 percent), wound infection (30 versus 7 percent), reoperation (48 versus 11 percent), overall 90 day mortality (20 versus 4 percent), and length of hospital stay (median 42 versus 26 days). An infected POPF was identified as an independent risk factor for sepsis, wound infection, and reoperation by multivariate analysis. However, it is unclear how this increased risk for complications can be best mitigated (eg, by earlier antibiotic therapy, or by procedural or surgical intervention).

In the past decade, neoadjuvant therapy has been used more often for patients with borderline and locally advanced pancreatic cancer. In such patients, the risk factors, incidences, and implications of POPF after pancreaticoduodenectomy may be different. In a single-center retrospective study of 753 patients, the rate of clinically relevant POPF (CR-POPF) was 3.6-fold lower in patients receiving neoadjuvant therapy versus upfront resection (3.8 versus 13.8 percent) [184]. However, overall survival was worse for patients who develop CR-POPF after neoadjuvant therapy (17 months with CR-POPF versus 34 months without CR-POPF). In multivariate analysis, CR-POPF following neoadjuvant therapy was not associated with the usual risk factors except one, soft pancreatic texture. The authors of this study suggested that neoadjuvant therapy altered the normal pancreatic parenchyma as much as the diseased portion to make it more fibrotic and resistant to POPF but that a persistently soft parenchyma following neoadjuvant therapy portends poor response to neoadjuvant therapy, hence the worse survival.

Prevention — Prevention of postoperative pancreatic fistula using various techniques for handling the pancreatic stump or anastomosis, prophylactic somatostatin and its analogues, and prophylactic pancreatic duct stenting has been attempted with varying success. In a 2022 meta-analysis of 12 interventions, external pancreatic duct drainage was the only intervention associated with reduced rates of both CR-POPF (OR 0.40, 95% CI 0.20-0.80) and all POPF (OR 0.42, 95% CI 0.25-0.70) [185]. Ulinastatin was associated with reduced rates of CR-POPF (OR 0.24, 95% CI 0.06-0.93). Invagination (versus duct-to-mucosa) pancreatojejunostomy was associated with reduced rates of all-POPF (OR 0.60, 95% CI 0.40-0.90).

Specific methods are discussed above for resection of the pancreatic head and discussed separately for resection of the tail of the pancreas. (See 'Pancreaticoduodenectomy' above and 'Adjuncts' above and 'Prophylactic somatostatin analogues' above and "Surgical resection of lesions of the body and tail of the pancreas", section on 'Central pancreatectomy' and "Surgical resection of lesions of the body and tail of the pancreas", section on 'Postoperative care and follow-up'.)

Bile leak — Bile leaks from the choledochal-jejunal anastomosis occur in 1 to 3 percent of cases and are heralded by the appearance of bile in the drainage fluid [186]. If this occurs, the drain should be left in place until the leak stops. If it is still present when the patient is ready for discharge, the patient can go home with the drain in place and have the drain removed in the office after the leak stops. (See 'Drain management' above.)

New-onset diabetes — Pancreaticoduodenectomy may lead to new-onset diabetes mellitus. The incidence is best estimated by a systematic review and meta-analysis of 22 studies [187]. The mean weighted overall proportion of new-onset diabetes mellitus after pancreaticoduodenectomy was 16 percent (95% CI 12 to 20 percent); 6 percent developed insulin-dependent new-onset diabetes (95% CI 4 to 10 percent). The risk is the same regardless of whether the indication for the surgery is benign or malignant.

In another study of 50 patients with a median follow-up of 32 months, the incidence of new-onset impaired glucose tolerance/diabetes mellitus was 22 percent after pancreaticoduodenectomy, and a percentage of pancreatic remnant volume <48.8 percent was an independent factor associated with new-onset diabetes [188].

In another study of 403 patients followed for at least 10 years, approximately 16.6 percent developed diabetes following pancreaticoduodenectomy [189]. A preoperative HbA1c >5.4 percent predicted new-onset diabetes with a relative risk of 2.944.

Pancreaticojejunal anastomotic stricture — Pancreaticojejunal anastomotic (PJA) stricture is a rare late complication that occurs at a median of 34 months after pancreaticoduodenectomy but only in 1.4 to 11.4 percent of patients, according to a systematic review [190]. Most present with symptoms of postprandial abdominal pain, recurrent acute pancreatitis, and impaired pancreatic function. Secretin-enhanced magnetic resonance cholangiopancreatography (MRCP) is the best modality to confirm the diagnosis. Only symptomatic patients require treatment, which consists of endoscopic dilation (with ultrasound-assisted "rendezvous" techniques), surgical repair, or total pancreatectomy with islet auto-transplantation (last resort).

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Pancreatic cancer".)

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5^th to 6^th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10^th to 12^th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Beyond the Basics topics (see "Patient education: Pancreatic cancer (Beyond the Basics)" and "Patient education: Chronic pancreatitis (Beyond the Basics)")

SUMMARY AND RECOMMENDATIONS

●Indications for pancreatic head resection – The most common indication for pancreaticoduodenectomy is the presence of a malignant or premalignant neoplasm in the head of the pancreas or another periampullary structure (bile duct, ampulla, or duodenum). Other potential indications include pancreatic or duodenal trauma and chronic pancreatitis. (See 'Indications for pancreatic head resection' above.)

●Preoperative biliary stenting – For patients presenting with obstructive jaundice, we suggest against routine preoperative biliary decompression (Grade 2B). However, we perform preoperative biliary stenting in these following clinical scenarios (see 'Preoperative biliary drainage' above):

•The serum bilirubin concentration exceeds 20 mg/dL.

•Surgery will be delayed for longer than two weeks.

•There is intractable cholangitis or pruritus.

When preoperative biliary stenting is elected, we prefer to use a plastic biliary stent (≥10 French), keeping the proximal end of the stent at or below the level of the cystic duct takeoff.

●Preoperative imaging – Preoperative imaging with high-resolution computed tomographic (CT) scan with 1 to 2 mm cuts through the pancreas during the early arterial and venous phase (pancreatic protocol) provides the best assessment of resectability and the peripancreatic vasculature (eg, anatomic anomaly, mesenteric stenosis, need for portal or superior mesenteric vein excision). (See 'Imaging' above.)

●Staging laparoscopy – For patients with a high likelihood of unresectability that has not been confirmed preoperatively, staging laparoscopy permits examination of the liver and peritoneal surfaces as well as biopsy of any suspicious areas. If metastatic tumor is found, laparotomy may be avoided. If the patient is found to have unresectable disease during the course of staging laparoscopy or an attempted pancreaticoduodenectomy and downstaging neoadjuvant therapy is not an option (eg, liver or peritoneal metastases, very extensive local disease with complete major vascular occlusion, etc), surgical biliary and gastric bypass may be performed, or alternatively, the patient may undergo endoscopic biliary stenting. (See 'Staging laparoscopy' above and 'Palliative surgery' above.)

●Vascular resection – Pancreaticoduodenectomy with portal or superior mesenteric vein resection and reconstruction is considered a standard approach for pancreatic adenocarcinomas that focally involve one of the veins, provided that adequate inflow and outflow veins are present, the tumor does not involve the superior mesenteric artery or hepatic artery, and an R0 resection can be accomplished. (See 'Vascular evaluation' above and 'Vein excision and reconstruction' above.)

●Pancreaticoduodenectomy – For most patients, we suggest a pylorus-preserving pancreaticoduodenectomy rather than the conventional pancreaticoduodenectomy (ie, Whipple procedure) (Grade 2C). The advantages of a pylorus-preserving pancreaticoduodenectomy include the potential time savings and lower rate of blood loss, with similar oncologic outcomes. However, a conventional pancreaticoduodenectomy (Whipple) operation should be performed if the tumor involves the proximal duodenum, pylorus, or gastric antrum. (See 'Conventional versus modified pancreaticoduodenectomy' above.)

●Gastrointestinal reconstruction – Following resection, gastrointestinal reconstruction provides a drainage route for the remaining pancreas, the bile duct, and the stomach. There are multiple accepted methods of gastrointestinal reconstruction; the choice is by surgeon preference as one method has not been shown to be consistently superior to others (see 'Gastrointestinal reconstruction' above):

•The pancreatic-enteric anastomosis can be either a pancreaticojejunostomy (figure 3A-B) or a pancreaticogastrostomy (figure 5). (See 'Pancreaticojejunostomy versus pancreaticogastrostomy' above.)

•A pancreaticojejunostomy can be constructed by an end-to-side duct-to-mucosa anastomosis (figure 6) or invagination of the pancreatic remnant. The invagination method can be performed either end-to-side or end-to-end (figure 7). (See 'Duct-to-mucosa versus invagination' above.)

•The biliary-enteric anastomosis is performed as a hepaticojejunostomy to the same loop of jejunum used to drain the pancreas. (See 'Biliary-enteric anastomosis' above.)

•The stomach or duodenum (pylorus sparing) can be drained into the same proximal jejunal loop brought up to the right upper quadrant for the pancreatic and biliary anastomosis (figure 3A) or to a separate Roux limb of more distal jejunum (figure 3B). The reconstruction can be placed anterior (antecolic) or posterior (retrocolic) to the transverse colon. (See 'Gastrointestinal anastomosis' above.)

●Minimally invasive approaches – In some studies, minimally invasive (laparoscopic or robotic) pancreaticoduodenectomy can be accomplished with less blood loss and shorter length of hospital stay, albeit longer operative time, compared with open surgery. However, it is highly complex and therefore should only be performed in selected patients in high-volume centers. (See 'Open versus minimally invasive' above.)

●Enucleation – Some small (ie, ≤1 cm) benign lesions in the head of the pancreas can be enucleated. Compared with pancreaticoduodenectomy, enucleation is more likely to cause a pancreatic fistula rate but less likely to lead to postoperative pancreatic insufficiency. Enucleation should not be used when the lesion is large (usually >2 cm) or when there is a chance that the lesion may be malignant. Enucleation is most commonly performed for pancreatic neuroendocrine tumors, especially insulinomas. (See 'Enucleation procedures' above.)

●Drain management – Following pancreatic resection, we suggest placing drains selectively in patients who are at a high risk of developing pancreatic fistula (eg, those with a soft and friable pancreas, those with a small [<2 mm] pancreatic duct), rather than routinely (Grade 2B). Any drain placed at the time of surgery is usually removed once the patient is tolerating a regular diet, provided there is no evidence of bile or pancreatic leak. If a leak is present, the patient may be discharged home with the drain in place. Most postoperative pancreatic and biliary fistulas close spontaneously, usually within four to six weeks after discharge, after which the drains can be removed in the office. (See 'Drainage of the pancreatic bed' above and 'Drain management' above.)

●Somatostatin analogues – We suggest against routine use of prophylactic somatostatin or any of its analogues (eg, octreotide) for patients undergoing resection of the head of the pancreas (Grade 2C). We limit the use of these medications to control the volume of output in patients who develop high-output pancreatic fistulas postoperatively. (See 'Prophylactic somatostatin analogues' above.)

●Morbidity and mortality – At experienced, high-volume centers, pancreaticoduodenectomy can be performed with a low postoperative mortality of <4 percent and morbidity of 40 to 50 percent. Common complications include delayed gastric emptying (17 percent), pancreatic fistula (5 to 10 percent clinically relevant), and bile leak (1 to 3 percent). The lowest operative mortality rates and best long-term cancer outcomes have been demonstrated at high-volume centers. (See 'Introduction' above and 'Perioperative morbidity and mortality' above.)

ACKNOWLEDGMENTS — The UpToDate editorial staff acknowledges Mark Girgis, MD, and Jonathan King, MD, who contributed to an earlier version of this topic review.