Complications of pancreato-duodenectomy

Overview

Pancreato-duodenectomy (PD) is the procedure of choice for management of resectable periampullary and pancreatic cancers and some patients with chronic pancreatitis. PD is one of the most major GI/ HPB surgical procedures performed involving resection of multiple organs and reconstruction with multiple anastomoses. While mortality of PD has been brought down to less than 5% morbidity still remains high.

Patients undergoing PD are usually elderly with comorbidities – general complications of a major operation e.g. wound, chest, cardiac and venous thrombo-embolism, are common.

The major intra-operative morbidity of PD is bleeding which can be from multiple sites viz. gall bladder bed, choledochal veins, gastro-colic trunk, pancreato-duodenal veins, jejunal veins, uncinate veins and cut surface of pancreas. An aberrant right hepatic artery (from the superior mesenteric artery) can be injured while dissecting the common bile duct.

Pancreatic leak is defined as presence of amylase rich (>3x serum amylase) fluid in drain beyond 3 days. It is graded as A (no clinical impact), B (persists, infected) or C (systemic sepsis, reoperation). Investigation of choice is computed tomography; a localized collection can be drained percutaneously. Major leak requires reexploration at which lavage and drainage should be performed; no attempt should be made to repair or redo the anastomosis. Soft pancreas and undilated duct are the two most important risk factors for pancreatic anastomotic leak; role of octreotide to prevent the leak is debatable. Preoperative biliary drainage in the form of endoscopic stenting may reduce bleeding complications by controlling coagulopathy but increases the risk of infective complications.

Postoperative bleed can be early or delayed and intra-luminal or intra-abdominal. Early intra-abdominal bleed is surgical – either a slipped ligature e.g. of the gastro-duodenal artery or one of the veins or from the pancreato-duodenal bed; reexploration should be done to control it if it is severe. Early intra-luminal bleed is from one of the anastomoses – commonest being the pancreatic. UGIE rules out bleed from the gastro/duodenal anastomosis (which if present, can be controlled endoscopically). Control of pancreatic stump bleed requires reoperation - taking down anterior layer of the anastomosis or a jejunotomy. Delayed intra-abdominal bleed is from a pseudo-aneurysm of an artery, usually gastro-duodenal caused by erosion by an abscess secondary to a leak; treatment of choice is angio-embolization. Delayed intra-luminal bleed is because of stress ulcers caused by systemic sepsis usually secondary to a leak and intra-abdominal sepsis.

Other common complications include delayed gastric emptying and acute pancreatitis.

Reoperations after PD are frequent – commonest cause is bleed, followed by leak; mortality of reexploration is high. Long term complications include anastomotic (PJ/ HJ) stricture, endocrine and exocrine insufficiency and inability to regain weight and poor quality of life.

Key words:
pancreato-duodenectomy − complications

Introduction

Pancreato-duodenectomy (PD), indicated for periampullary and resectable pancreatic head cancers, some patients with chronic pancreatitis and rarely for complex pancreato-duodenal trauma, is one of the most major procedures performed in GI/ HPB surgery – the ‘Cadillac’ of all operations, as it includes multivisceral resection and reconstruction with multiple (pancreatic, biliary and gastro/ duodenal) anastomoses of which the pancreatic anastomosis is the most precarious – the ‘Achille’s heel’ of PD. Hepatectomy is also a major HPB surgical procedure but it is a resection which does not involve any anastomosis (when done for liver cancer) or only one (biliary) anastomosis (when done for cholangiocarcinoma). Liver transplant is probably the only HPB procedure more complex than PD as it also involves multiple anastomoses, including arterial, venous and biliary (but does not have the precarious and dangerous pancreatic anastomosis cf. PD). In GI surgery, restorative proctocolectomy with ileal pouch anal anastomosis performed for ulcerative colitis and familial adenomatous polyposis, with a long suture line of the ileal pouch and a precarious pouch-anal anastomosis is almost akin to PD; fortunately, a proximal diverting loop ileostomy provides protection against the ill effects of a pouch-anal anastomatic leak, a safety valve not available in PD.

Complications of PD

While the mortality of PD has been brought down from earlier about 25% to less than 5%, many centres reporting more than a hundred consecutive cases without mortality (Trede 1990, Cameron 1993, Penumadu 2015), the morbidity continues to remain high. Cameron (2015) recently reported a personal series of 2,000 PDs performed from 1960s to 2000s – while the 30 day mortality was 1.4%, morbidity (delayed gastric emptying DGE 25%, pancreatic fistula 16% and wound infection 12%) did not decrease over decades. A three decade review of 2,564 PDs (for periampullery adenocarcinomas performed at the Johns Hopkin’s Hospital reported an overall complication rate of 52%, including DGE (16%), pancreatic fistula (9%) and wound complications (12%) (He 2014).

Complications of PD can be

• i. General i.e. those of any major operation performed usually in an elderly patient with comorbidities e.g. chest complications, venous thrombo-embolism, myocardial infarction etc., and wound complications i.e. infection, dehiscence and incisional hernia
• ii. Specific for PD, which can be

• a) Intraoperative

• b) Postoperative

Etiopathology of complications

Most of the intraoperative complications of PD are related to the proximity of several vessels, arteries and veins, to the organs which are resected viz. common bile duct (CBD), distal stomach, duodeonum, proximal jejunum and head of pancreas.

Pancreatic fistula following PD is a mixed fistula i.e. containing pancreatic juice with enteric contents (cf. pure fistula containing pancreatic juice only following distal or central pancreatectomy) and is more damaging to the adjacent tissues. The pathophysiological effects of activated pancreatic juice, rich in autolytic enzymes viz. amylase, lipase, protease and elastase leaking out of the .pancreatic anastomosis can be disastrous. A small anastomotic leak can go on to become a major disruption; it can erode into other adjacent anastomoses i.e. biliary and gastro/ duodenal and into the nearby exposed vessels and vascular stumps, especially of the gastro-duodenal artery (GDA).

Intraoperative complications

The most common and important intraoperative complication of PD is bleeding. Most patients undergoing PD have obstructive jaundice and the associated coagulopathy may further increase the introperative bleeding. Bleeding can occur from several sites during various steps of mobilization and resection.

1. Cholecystectomy – gall bladder in patients with biliary obstruction is enlarged, distended and vascular; significant bleed can occur from the gall bladder bed during mobilisation of the gall bladder from the liver.
2. CBD – in a stented patient, the CBD is thick, inflamed and vascular and the intramural vessels can bleed after division of the CBD.
3. The gastro-colic trunk may get injured during mobilisation of the transverse mesocolon from the anterior surface of the head of the pancreas.
4. Inferior mesenteric vein may get injured when the duodeno-jejunal flexure is being mobilised.
5. Jejunal mesenteric vesels in a thick fat laden mesentery of an obese patient can retract and bleed to result in a large intramesenteric hematoma.
6. Inferior vena cava may get injured during kocherisation of the duodenum, especially in presence of a large pancreatic head mass.
7. Pancreato-duodenal veins, especially the posterior superior pancreato-duodenal vein (PSPDV) which drains directly into the portal vein (PV) and posterior inferior pancreato-duodenal vein (PIPDV) which drains directly into the superior mesenteric vein (SMV) can very easily be torn while dissecting for their ligation and division – this results in a hole in the PV/ SMV causing a major bleed.
8. Similarly, the multiple small veins of the uncinate process drain directly into the SMV and can very easily be torn while dissecting for their ligation and division resulting in a hole in the SMV causing a major bleed.
9. The superior and inferior transverse pancreatic arteries can bleed from both the cut surfaces of the pancreatic stump after division of the neck of the pancreas.
10. Vessels in the mesopancreas i.e. posterior superior pancreato-duodenal artery (PSPDA) and posterior inferior pancreato-duodenal artery (PIPDA) can retract behind the PV/ SMV to the left and be difficult to control.

Hemostasis should be checked and secured before reconstruction is started. It is the author’s practice to place the drain and manipulate the naso-jejunal tube out of the divided stomach/ duodenum before reconstruction is started.

The most important intraoperative injury during PD is to an aberrant (accessory or replaced) right hepatic artery which lies on the right postero-lateral aspect of the CBD. In a stented patient, in fact, the aberrant artery may get embedded in the thick wall of the CBD because of severe pericholedochal inflammation and fibrosis. Common hepatic artery may be mistaken for GDA and accidentally divided. A bulldog clamp should initially be applied on the GDA after its mobilisation – pulsations should be confirmed in the proper hepatic artery in the hepato-duodenal ligament before the GDA is finally ligated and divided. Superior mesenteric artery (SMA), lying to the left of the SMV can very easily get injured when dividing the mesopancreas if too much traction is applied on the mobilised specimen – this pulls the SMA to the right so that it comes to lie on the right side of the SMV thus making it liable to injury. The artery first technique of PD prevents such an injury and is recommended for large pancreatic head and uncinate tumors.

Postoperative complications

The most important postoperative complications of PD are related to leaks from the three anastomoses (pancreatic, biliary and gastro/ duodenal - in that order).

Pancreatico-enteric anastomosis (PEA) leak

Also called pancreatic anastomotic failure (PAF), postoperative pancreatic fistula (POPF)

The PEA is very prone to leak - soft gland in periampullary cancers which does not hold sutures well and undilated (normal) pancreatic duct are the two most important risk factors for PEA leak; leak rates are minimum in patients with firm gland and dilated duct e.g. pancreatic head cancer and chronic pancreatitis. Other predictors for leak include old (>70 years) age (He 2013), male gender, poor nutritional status (serum albumin <4 G/ dL), obesity (higher BMI) (Tsai 2010), long duration of operation, high (>1.000 mL) blood loss and hypotension during operation. A peripancreatic seroma, hematoma or chyloma can get infected to form an intraabdominal abscess which then erodes into the intact pancreatic anastomosis resulting in a secondary PEA leak.

International Study Group on Pancreatic Fistula (ISGPF) defined pancreatic fistula as drain fluid amylase more than 3 times upper normal value of serum amylase 3 days after PD (Bassi 2005). PEA leak usually manifests around day 5−7. A high index of suspicion is required to detect the leak early - unstable vitals (tachycardia, tachypnea and hypoxemia) and unsettled abdomen (distened, tender, silent, intolerant to enteral feeds) are early indicators. Elevated drain fluid amylase on day 1 (Israel 2014) and day 3 or 4 are good predictors of leak; leucocytosis and acidosis are usually present. A computed tomogram (with oral and IV contrast) is the investigation of choice - presence of fluid collection around the pancreas is the indirect evidence of a PEA leak.

PEA leak has been classified as

• A Radiologically (CT) demonstratable leak but with no clinical impact
• B Leak which persists (> 3 weeks) or gets infected (fever, leucocytosis)
• C Systemic sepsis, need for reintervention

The key to management is image guided percutaneous catheter drainage (PCD) of the collection, along with antibiotics and enteral/parenteral nutrition. Rarely, if there is no improvement/ deterioration following PCD, reoperation may be required. At reoperation, all collections should be drained, peripancreatic area thoroughly lavaged and good drainage provided with multiple widebore drains; abdominal wound may be left open. A feeding jejunostomy should be added if it was not done at the time of PD. Any attempts at repair/ refashioning of the anastomosis should be avoided as the sutures are very likely to cut through the inflamed friable (pancreas) gland. In case of a major disruption, the entire anastomosis may have to be taken down – stomach/ jejunum is closed and the pancreatic stump is drained with multiple wide bore drains and a fine tube placed into the pancreatic duct, if possible; this will convert the uncontrolled PEA leak into a controlled external pancreatic fistula (EPF) with which the patient can be discharged home. Most EPF following external drainage of PEA leak will close spontaneously, though this may take weeks or, sometimes, even months. Very rarely, intervention will be required for a persistent EPF – this would be either glue obliteration of the fistulous track close to the pancreas or a surgical fistulo-jejunostomy, again as close to the pancreas stump as possible. Completion total pancreatectomy is on option but not frequently used because of the inevitable brittle diabetes which is very difficult to manage.

PEA leak is of significance as it may lead to several complications including intraabdominal collection and abscess, delayed intraabdominal bleed (vide infra), delayed gastric emptying and wound complications. PEA leak, collection and abscess may eventually lead to systemic sepsis and multiple organ dysfunction syndrome (MODS).

There are innumerable techniques of PEA, each claiming to be superior to the other in terms of lower risk of leak. Duct-to-mucosa anastomosis (pancreatico-jejunostomy) is usually associated with lower leak rates as compared to gland-to-jejunum dunking/invagination (pancreato-jejunostomy). Pancreatico-gastrostomy (PG) has been found to be better than pancreatico-jejunostomy (PJ) in observational studies but randomised controlled trials do not confirm the same. A metaanalysis of 4 randomized controlled trials and 22 observational clinical studies including 3 752 patients did not reveal adequate evidence of superiority of PG over PJ (He 2013). A recent metaanalysis of 8 studies comparing PG (n=607) with PJ (n=604) showed less leak, fistula and collection following PG; there was, however, no difference in bleed, reoperation, length of stay and mortality (Que 2015). The main disadvantage of PG over PJ is a higher risk of bleeding and the fact that a PG leak results in a complex pancreatic + gastric fistula. An isolated jejunal loop (dual loop reconstruction) has been used for the pancreatic anastomosis to separate it from the other two (biliary and gastro/duodenal) anastomoses. A recent metaanalysis including 802 patients, however, did not show the dual loop reconstruction to be superior to conventional single loop reconstruction (Klaiber 2015). Use of magnification (loupe or microscope) has also been advocated to perform better PEA (Hashimoto 2010). Some surgeons, by choice, ligate/occlude/ block (with fibrin glue or a synthetic polymer) the pancreatic duct thus avoiding a PEA at all but it has not been shown to decrease the risk of PEA leak. In fact, fistula rates are higher but the fistula in such cases is a pure pancreatic fistula which is less dangerous than a mixed pancreatic + enteric fistula following a PEA leak. Transanastomatic stent, either internal or external (transjejunal), has been advocated to reduce the risk of PEA leak but has not been shown to decrease the risk of PEA leak (Zhou 2012).

Somatostatin and its analogus e.g. octreotide, have been used preoperatively to prevent/ decrease the incidence of PEA leak. A recent metaanalysis, however, reported that octreotide had no influence on the incidence of pancreatic fistula; pasireotide is expected to have a potential role in reducing the incidence of PEA leak (Allen 2014). The Author uses octreotide selectively – in patients with a high risk PEA i.e. soft pancreas and undilated duct.

Incidentally, patients who receive neoadjuvent treatment with radiotherapy have lower risk of PEA leak as radiation makes the gland firmer and decreses the exocrine secretion (Ishikawa 1991).

Bleed

Post PD hemorrhage (PPH) is the next important postoperative complication. The International Study Group for Pancreatic Surgery (ISGPS) (Wente 2007) suggested classifying PPH based on

1. Onset – early (within 24 hours ) and delayed (after 24 hours)
2. Site – intraluminal (GI) or extraluminal (intraabdominal)
3. Severity – mild or severe (hemodynamic instability, significant fall in Hb, need for blood transfusion

The concept of sentinel bleed – a minor intermittent bleed followed, a few days later, by a major bleed is debatable. All minor bleeds are not followed by a major bleed and not all major bleeds are preceded by a sentinel bleed.

Early intraluminal bleed, manifesting as blood in the nasogastric tube/hematemesis or melena is from one of the anastomoses, most commonly pancreatic. An upper GI endoscopy excludes bleed from the gastro/duodonel anastomosis, which, if present, can be controlled endoscopically using sclerosants and/or energy sources; an expert endoscopist can then pass the scope across the gastro/duodenal anastomosis into the proximal jejunal limb to visualize the biliary and the pancreatic anastomoses also and confirm the source of bleeding. Severe bleeding from the pancreatic stump (usually from the superior and inferior transverse pancreatic arteries running along the superior and inferior borders of the pancreas) usually merits reoperation. In case the PEA was done in a duct-to-mucosa fashion, the anterior layer of the pancreatic anastomosis will have to be opened to control the bleeder on the cut surface of the pancreatic stump; the anastomosis can then be redone. In case of a gland-to-jejunum PEA, the cut surface of the pancreas can be exposed for control of the bleeding by making a jejunotomy along the long axis of the jejunal limb parallel to the PEA. Bleeding from the the cut surface of the pancreatic stump can thus be controlled without disrupting the PEA; jejunotomy is closed after bleeding has been controlled. Early intraluminal bleeed from the pancreatic stump can result in the formation of a clot in the jejunal limb; the clot can then distend the jejunal limb and disrupt the pancreatic anastomosis resulting in a secondary early PEA leak.

Early bleed is more commonly extraluminal (intraabdominal) which manifests as blood in the drain or the wound; it is usually surgical and, if severe, merits reoperation (Correa-Gallego 2012). The usual source is a slipped ligature from a vascular stump e.g. GDA or one of the pancreato-duodenal arteries or the mesopancreas or a diffuse ooze from the pancreatic bed. Major/ named arteries e.g. GDA should preferably be double ligated or ligated and suture transfixed to prevent slippage of the ligature.

Marginal (gastro-jejunostomy) ulcer and gastric mucosal erosions (because of systemic sepsis due to an anastomotic leak) are causes of delayed intraluminal bleed.

Delayed intraabdominal bleed is usually from a pseudoaneurysm of the GDA stump or one of the adjacent exposed arteries e.g. hepatic or splenic, which has been eroded by an abscess formed because of secondary infection of a peripancreatic collection following a PEA leak. Lymph node dissection may result in baring of the adjacent arteries thus making them more prone to erosion. The pseudoaneurysm can be diagnosed on CT angiography (celiac axis and superior mesenteric). Angio-embolization with steel coils or gel foam particles is the treatment of choice; covered vascular stents are better as they avoid inadvertent occlusion of the common/ proper hepatic artery.

A major disruption of the pancreatic anastomosis can cause delayed bleed from the pancreatic stump which can manifest as both intraluminal and intraabdominal bleed; such patients usually need surgical intervention, especially if the bleed is severe.

Inflamed walls of infected collections (secondary to an anastomatic leak) can also cause delayed intraabdominal bleed as also erosion of a vessel by a drain.

Delayed intraabdominal bleed secondary to a PEA leak usually occurs around 7−14 days after PD; several reports have documented patients having delayed bleed even after discharge from the hospital (Correa Gallego 2012).

The ‘dirty double’ combination of pancreatic anastomotic leak and delayed intraabdominal bleed from the GDA stump is associated with high mortality (Balachandran 2004). We have reported the use of a vascularized omental flap which is wrapped around the PJ like a carpet; it separates the GDA stump (posteriorly) from the PJ anastomosis (anteriorly) and protects it from being eroded by a collection secondary to a PJ leak (Kapoor 2006).

Delayed gastric emptying

Delayed gastric emptying (DGE) is the commonest complication of PD. It is more common following pylorus preserving PD (PPPD) than after classical (Whipple’s) PD and is more common in diabetics. A formal lymph node discsection is also more frequently complicated by DGE. A review of pylorus resecting subtotal stomach preserving PD has shown reduced incidence of DGE (Hanna 2015). Antecolic route of GJ may be associated with lower incidence of DGE than retrocolic route (Eshuis 2012). Side to side gastric greater curve to jejunal anastomosis has been shown to be associated with lower incidence of DGE than end to side gastric stump to jejunal anastomosis (Nakamura 2015). DGE can occur in the absence of an anastomotic leak or intrabdominal sepsis but the commonest cause of DGE remains a retrogastric collection secondary to a PEA leak. DGE manifests as hight nasogastric aspirate, delayed (>3 days) removal of naso-gastric tube, need for reinsertion of naso-gastric tube and intolerance of oral feeds after day 7 (Wente 2007). Gastric emptying scintigraphy is a useful investigation to diagnose and grade the severity of DGE (van Samkar 2013). Management includes prokinetics e.g. metoclopramide and erythromycin, and enteral (nasojejunal or jejunostomy) or parenteral nutrition. Most cases settle over days or weeks; very rarely, if the DGE persists for more than a few weeks, surgical intervention is required which is in form of conversion to Roux-en-Y gastro-jejunostomy.

HJ leak

Hepatico-jejunostomy (HJ) leak manifests as bile in the drain; amylase levels in the drain fluid are high as the pancreas is also draining into the same jejunal limb. PCD is required to drain any collection; a minor HJ leak may stop on its own. If the leak is large or persists, percutaneous transhepatic biliary drainage (PTBD) may be required. Most HJ leaks will respond to PTBD; reoperation is very rarely required.

GJ/ DJ leak

Gastro-jejunostomy (GJ)/duodeno-jejunostomy (DJ) leak presents with enteric contents/ oral feeds in the drain. Oral administration of a blue dye confirms the diagnosis at bed side but CT should be done to look for any collection. Most patients with GJ/DJ leak will require surgical intervention to drain the collection and repair the anastomosis; feeding jejunostomy should be added if it was not performed earlier.

Acute pancreatitis

Post PD acute pancreatitis is an uncommon complication. Clinical picture is similar to that of PEA leak (which is, however, not seen on CT) and management is largely conservative and supportive.

Chylous ascites

Extensive nodal dissection around the SMA origin from the aorta and vascular (PV) resection are associated with the higher risk of chylous ascites which manifests as high drain output which is clear serous to begin with but turns milky after introduction of polymeric enteral feeds. It is diagnosed by high triglyceride levels in the drain fluid. Management includes enteral feeding with medium chain triglycerides (MCT) or total parenteral nutrition (Assumpcao 2008).

Diarrhea

Post PD diarrhea is more common after radical extended retroperitoneal lymphadenectomy.

R2 resection

The commonest margin which can turn out to be positive on histopathology after PD is the uncinate (variously called posterior, retroperitoneal or SMA) margin in a pancreatic head cancer. The other margin which can be positive is the pancreatic neck margin. In a lower end cholangiocarcinoma, the CBD margin may be positive.

In patients with vascular (venous or arterial) involvement, which was not obvious on CT but is detected for the first time at operation, a palliative (R2) resection may be the only option as such an involvement is usually found at a late stage of operation when some irreversible steps have been taken. While venous (SMV/ PV) resection and reconstruction is an option, arterial resection is not recommended. Artery (SMA) first approach is advisable in large head and uncinate tumors – this detects the arterial involvement at an earlier stage and thus avoids an R2 resection.

Readmission

Median hospital stay after PD is about 14 days; readmissions after discharge, however, are very frequent. In a SEER – Medicare database review, readmission to hospital within 30 days after discharge was required in 292 (21%) out of 1 370 patients who underwent PD; reasons included dehyderation, gastric outlet obstruction, cholangitis, venous thrombo-embolism and pneumonia (Hyder 2013). Sadot (2014) reported 30 day and 90 day readmission rates of 23% and 29% after 490 pancreatic resections; the commonest cause of readmission was infectious complications.

Long term complications

Anastomotic stricture

PJ stricture may lead to obstructive chronic pancreatitis; if symptomatic, lateral pancreatico-jejunostomy may be required to drain the dilated pancreatic duct. HJ stricture causing cholangitis may occur in a patient who had an HJ leak in the postoperative period; treatment of choice is re-HJ as a stump of the common hepatic duct is invariably available for the anastomosis.

Nutritional status

PD results in major changes/ disturbances in the upper GI anatomy and physiology causing nutritional disturbances; most patients do not regain their pre-disease weight even after disease free survival of many years.

Endocrine and exocrine insufficiency

Exocrine insufficiency is more common following PG (than PJ) because of inactivation of the alkaline pancreatic juice by the gastric acid. Endocrine dysfunction (glucose intolerance) develops/ worsens in many patients after PD.

Quality of life (QoL)

About half of the patients undergoing PD may not achieve preoperative QoL even after 6 months (Park 2013).

Management

The investigation of choice for detecting/ruling out most of the common/ major post PD complications is CT with angiography (Raman 2013).

The approach towards the management of post PD complications has shifted from surgical to nonsurgical (percutaneous radiological and therapeutic endoscopic) intervention; severe early intraadbominal bleed being the only complication requiring direct surgical intervention (Tol 2014). A significant number of patients undergoing PD, however, still require reoperation for various reasons (Reddy 2012).

Mortality

The acceptable mortality of PD today is <5%. For a major procedure such as PD, not only 30 day but in-hospital and 90 day mortality should be documented and mentioned as some patients, especially those who develop postoperative complications, have a long (>30 days) hospital stay and can die even after 30 days.

Few centers have reported >100 viz. 118 (Trede 1990), 126 (Penumadu 2015) and 145 (Cameron 1993) consecutive PDs without mortality but none of these centers further reported 200 consecutive PDs without mortality thus indicating that one or more patients died before the center reached the 200 mark without mortality. Volume (number of PDs performed per year) is one of the major factors determining the mortality of PD (Birkmeyer 2002). The commonest cause of death is PEA leak resulting in sepsis (abdominal and systemic) and secondary bleed (Balachandran 2004); other causes include cardiac and pulmonary complications. Risk scores have been described to predict the mortality of PD; one such score includes age, male sex, serum albumin, tumor size and Charlson comorbidity index (Venkat 2011).

Effect on survival

Postoperative complications not only prolong the length of stay and increase the cost of treatment, they also delay time to adjuvant therapy and reduce the likelihood of multimodality adjuvant therapy thus reducing survival (Wu 2014). Post PD complications may suppress cell mediated immunity thus increasing the risk of recurrence and reducing the survival (van der Gaag 2014).

Prevention

Most patients undergoing PD have jaundice. Preoperative biliary drainage, in the form of endoscopic stenting, to bring the serum bilirubin down with an aim to reduce the risk of postoperative complications is a debatable and controvertial issue. Biliary stenting is a double edged sword − while it can reduce bleeding complications (by correcting the associated coagulopathy) it inceases infective complications. In a randomised controlled trial (RCT), in fact, routine preoperative biliary drainage increased the rate of post PD complications (van der Gaag 2010). In spite of evidence against its routine use, however, more than 50% of patients who had PD underwent preoperative biliary drainage in the form of endoscopic stenting (Jinkins 2013). Uncontrolled cholangitis is probably the only definitive indication of preopertaive biliary drainage before PD.

A prospective RCT showed that use of a restrictive fluid regimen with 3% hypertonic saline (vs. Ringer’s lactate) resulted in a significant reduction of postoperative complications (Lavu 2014).

Some surgeons abandoned the use of drains during PD but a prospective RCT showed increased (12% vs. 3%) mortality after PD without drain and the trial had to be stopped prematurely (van Buren 2014).

Conclusion

PD continues to be associated with high morbidity; the dual combination of leak and bleed makes the situation bl(eed)(l)eak!

Conflict of Interests

The authors declare that they have not conflict of interest in connection with the emergence of and that the article was not published in any other journal.

Vinay Kumar Kapoor FRCS

Professor of Surgical Gastroenterology

Sanjay Gandhi Post-graduate Institute of Medical Sciences (SGPGIMS)

Lucknow 226014 UP India

e-mail: vkkapoor.india@gmail.com