Essentials

In different countries incidence figures vary from 40 to 500 new cases of acute pancreatitis per million of the population each year. Data usually derives from hospital record systems and it is important to identify new patients separately from those with recurrent disease. In large population studies from Scotland and Finland the incidence of the disease has risen steadily to the current 400 patients/million per year. A study including data on 84713 patients from 470 Californian hospitals collected from 1992 to 2000 indicated approximately 300 new patients/million per year. A recent epidemiological study utilizing biochemical factors in diagnosis in north-east England, with individual accurate case verification, yielded just over 600 new patients/million per year, probably a more reliable incidence. The disease was most prevalent in those with poorer socioeconomic status, especially associated with alcohol abuse as the cause.

Overall mortality is from 4.0 to 7.5%, highest in those who are over 70 years, obese individuals, and those with comorbidity at the time of onset. In all parts of the world both prospective and retrospective studies record 45 to 50% of deaths as occurring in the initial 2 weeks of illness, with some studies identifying a peak of deaths in the initial 48 h. Death is usually due to multiple organ failure in which respiratory failure predominates.

Aetiology

Aetiological factors and rare associations of acute pancreatitis are listed in Bullet lists 1 and 2.

Major factors

Biliary disease and alcohol abuse together account for over 80% of patients with acute pancreatitis in most prospective studies.

Gallstones

The increased early use of endoscopic retrograde cholangiopancreatography (ERCP) and endoscopic ultrasound (EUS) results in most so-called idiopathic acute pancreatitis being identified as due to small (1–3 mm diameter) stones, which account for nearly all female patients as well as older males. EUS is both less interventional and more sensitive at detecting these smaller stones than ERCP, but each approach is contingent on the skill of the operator.

Bullet list 1 Aetiological factors in acute pancreatitis (according to frequency)

Bullet list 2 Rare associations with acute pancreatitis

Prospective studies in the United Kingdom, without EUS, have revealed that 40 to 65% of patients with pancreatitis have gallstones. In Ohio (United States of America) a similar incidence was reported, and in Argentina over 80% were found to have small gallstones when faecal sieving was employed. Such studies link the passage of small stones migrating from the gallbladder down the common bile duct and into the gut, after usually transient impaction in the ampulla of Vater, with acute pancreatitis. Increased back pressure in the pancreatic duct is likely to be an important element, but other factors are probably also involved.

Eradication of stones nearly always frees the patient from further episodes of acute pancreatitis.

Alcohol

Pancreatitis due to alcohol abuse occurs in over 80% of patients from downtown New York (United States of America), and around 70% in Helsinki (Finland). This association is usually found in young males who drink in excess of 80 g alcohol per day, but anyone having a heavy regular alcohol intake is at risk. Careful investigation of the possibility of small gallstones in these patients is necessary. Some patients have both the most common aetiologies. Alcohol may provoke acute pancreatitis by intracellular lysosomal release modulated through elevations in cytosolic and mitochondrial ionized Ca²⁺ concentration.

Minor factors

Drugs

The drugs most commonly implicated in causing acute pancreatitis are valproic acid, azathioprine, L-asparaginase, and corticosteroids. There is equivocal evidence regarding thiazide and other diuretics. However, unless viral titres have been determined, together with adequate biliary investigations including endoscopic examination of the ampulla of Vater, it is unwise to ascribe acute pancreatitis to a particular drug. Repeat exposure to the same drug again causing acute pancreatitis is the strongest evidence of a direct association.

Viral infection, particularly mumps, Coxsackie B, and viral hepatitis, can cause acute pancreatitis. One clinical feature that may prove useful is prodromal diarrhoea, which is rare in all other types of acute pancreatitis. Of increasing importance in endemic areas are the effects of HIV infection. There are also concerns that some antiretroviral therapy may cause acute pancreatitis.

A focal area of pancreatic necrosis in a primary attack of acute pancreatitis can cause secondary fibrosis with main duct stricture formation, with segmental ‘upstream’ recurrent attacks of pancreatitis as a consequence. Stricture dilatation or occasionally surgical decompression or distal pancreatectomy may be required.

Congenital or developmental anatomical abnormalities can present with pancreatitis (choledochal cyst, duodenal duplication, anomalous pancreaticobiliary junction). Pancreas divisum (nonunion of main and accessory ducts) occurs in 3 to 5% of people and is not usually a primary cause of pancreatitis.

Periampullary adenoma or carcinoma is an important association, best diagnosed by endoscopic biopsy. With the increase in this approach to diagnosis, tumours at or close to the ampulla have been shown to have a greater association (0.4%) with acute pancreatitis than hyperparathyroidism. Effective treatment of the tumour abolishes recurrent attacks. This usually involves surgical resection, but endoscopic laser therapy or endoscopic papillectomy can be effective in older and less fit patients.

Carcinoma of pancreas can occasionally present with clinical acute pancreatitis and other primary tumours metastasizing to the pancreas may present in this way.

This is now recognized to be an uncommon accompaniment of acute pancreatitis. Indeed, many of the reported patients have also had gallstones. The association is calculated at 0.1%. Removal of a parathyroid adenoma usually prevents further acute pancreatitis since persistent hypercalcaemia appears to be the provoking factor.

Patients with type I and V hyperlipoproteinaemia may develop acute pancreatitis. The significance of this association can be difficult to validate where alcohol ingestion compounds the problem. Those patients with primary hyperlipoproteinaemia with chylomicronaemia and hypertriglyceridaemia are prone to attacks of acute pancreatitis in the absence of alcohol ingestion. Dietary restriction of lipids and various lipid-lowering drugs are valuable in therapy.

This is an important association in the elderly, when pancreatitis may be associated with myxoedema coma. In younger patients, alcohol abuse may be linked, particularly if patients fall asleep out of doors or in a cold, unheated house. Management is directed at gradual warming and supportive measures for organ compromise.

This condition is increasingly being studied since the 1996 discovery of genetic mutations of the cationic trypsinogen gene (PRSS1), which shed light on the mechanism of acute pancreatitis. A Europe-wide study (EUROPAC) has tracked multiple families in the United Kingdom, Germany, and France, and similar work in Japan and the United States of America is ongoing. Somewhat confusingly, the numbering of the two most common defects has altered in the intervening years, with R117H changing to R122H and N211 to N291.

An autosomal dominant family history of pancreatic disease is customary. Severe acute inflammatory changes are rare and diagnosis is often delayed. Patients usually have a long history of recurrent abdominal pain from childhood or adolescence. Changes of chronic fibrosis may be present at diagnosis. Typically chronic pancreatitis is evident by the age of 20 to 40 years, and the risk of pancreatic carcinoma over 60 years is significantly increased.

Mutations in the cystic fibrosis transmembrane conductance regulator gene (CTFR) and the pancreatic secretory trypsinogen inhibitor gene (SPINK1) are also linked to pancreatitis.

Hyperamylasaemia may occur after blunt abdominal trauma, usually from a crush injury to the body of the pancreas against the vertebral column. The risk of associated injuries to surrounding organs must be considered, and assessment by CT scan is important. When there is transaction of the main pancreatic, therapeutic options include endoscopic transpapillary stenting and distal pancreatectomy. Late presentation is associated with pseudocyst formation due to leakage from the damaged pancreatic duct.

Surgical or endoscopic procedures involving the ampulla of Vater can induce pancreatitis. The large decrease in diagnostic ERCP (1% risk of acute pancreatitis) as a result of noninvasive imaging modalities of EUS and magnetic resonance cholangio-pancreatography (MRCP) has reduced the overall incidence of post-procedural acute pancreatitis. The risk of acute pancreatitis increases to 3% where a therapeutic endoscopic sphincterotomy has been performed, and in such patients the potential of iatrogenic duodenal perforation should be explored by CT.

This is a rare condition, which is considered part of the IgG4-related autoimmune disease spectrum and associated with other autoimmune diseases (polyarteritis nodosa, systemic lupus erythermatosus, other vasculitides) and inflammatory bowel disease (Crohn’s disease and ulcerative colitis). It may present as abdominal pain with obstructive jaundice more typical of chronic than acute pancreatitis. Other features may include: (1) an increased IgG4/IgG ratio in serum; (2) homogeneous gland enlargement with a well-defined edge on CT; (3) periductal lymphoplasmocytic infiltrate on biopsy, which may also be associated with abnormalities in the extrahepatic biliary tree resembling sclerosing cholangitis. Focal autoimmune pancreatitis may prove difficult to differentiate from carcinoma. A good response to steroids is diagnostic.

Ascaris lumbricoidis within the ampullary area may manifest as acute pancreatitis clinically, and other worms stuck in this area can produce the same effect.

Bartholomew showed elegantly that the Trinidad scorpion bite was associated with the production of a clinical episode of acute pancreatitis soon after the victim was bitten. Patients customarily brought the dead scorpion to show the doctor.

A few patients present with recurrent abdominal pain associated with raised pressures on sphincter manometry. If this is part of a global gut dysmotility spectrum in an individual patient, then symptoms may originate from motility abnormalities in more than one area of the gastrointestinal tract and division of the sphincter of Oddi may only partly resolve such a person’s symptoms. The benefit of intervention is more likely in patients with type 1 Geenan and Hogan sphincter of Oddi dyskinesia (Table 1). Traditional treatment involves endoscopic sphincterotomy, but the risk of post-ERCP pancreatitis in these patients is high (30%). The use of preliminary sphincter paralysis by the injection of botulinum toxin can provide temporary symptom relief at minimal risk, and helpful information on whether more prolonged response may be achieved by sphincterotomy.

Table 1  Geenan and Hogan classification of sphincter of Oddi dysfunction

 

Biliary SOD	Pancreatic SOD
Type 1	Type 1
Biliary type pain	Pancreatic-type pain
Elevated liver enzymes	Amylase/lipase elevation
CBD dilatation	MPD dilatation
Delayed contrast drainage from the CBD > 45 min	Delayed contrast drainage from the MPD >8 min
Type 2	Type 2
Biliary type pain	Pancreatic-type pain
One or two of the above criteria	One or two of the above criteria
Type 3	Type 3
Biliary type pain only	Pancreatic-type pain only

CBD, common bile duct; MPD, main pancreas duct; SOD, sphincter of Oddi dysfunction.

Pathology

All patients with acute pancreatitis have microscopic evidence of necrosis, but macroscopic changes, particularly black discoloration of necrosis of the pancreas, are confined to the most severe cases. It is more frequent for this gross degree of necrosis to occur in the peripancreatic fatty tissue than in the pancreas itself. When it is present in the pancreas there is usually a panlobular necrosis, and it is impossible to delineate where the disease initiated. In a classical paper, Foulis claimed that the most common microscopic abnormality seen in humans, periductal necrosis, is typical of biliary and alcohol causation. Less commonly a perilobular necrosis is found, usually in patients with hypothermia or gross hypotension. In experimental acute pancreatitis the initial lesion is now considered to be intracellular, featuring coalescence of lysozymes and zymogen granules. Acinar cell disruption is found with many of the hyperstimulation models such as caerulein-induced acute pancreatitis. It is now believed that this initial event may be associated with oxidative stress.

Clinical features

Sudden onset of severe upper abdominal pain with vomiting is the most common manner of presentation. The pain may focus in the epigastrium or right or left upper quadrant, with penetration through to the back, and occasionally it encircles the upper abdomen. Patients who have experienced both a myocardial infarct and acute pancreatitis usually describe the latter pain as being much more severe. However, it tends to lessen in severity progressively over the first 72 h, and it is not usually a significant factor beyond this time. There may be upper abdominal tenderness and guarding, but these signs are often less marked than might be suspected from the severity of the pain, and bowel sounds are usually absent in the early stages.

Up to 90% of patients with acute pancreatitis have troublesome vomiting in the first 12 h of illness, and this contributes to hypovolaemia and hypotension. Clinical jaundice and pyrexia are rare on admission, although abnormalities of biochemical liver blood tests occur in 80% of patients with a biliary aetiology. Occasionally pancreatitis occurs in association with cholangitis, which is of importance in that organ failure may be driven by the cholangitis, not pancreatitis, such that urgent ERCP and biliary decompression is required.

Biochemical abnormalities

A multitude of biochemical phenomena are found in acute pancreatitis. Various pancreatic enzymes are released that are useful as diagnostic markers. Acinar cell disruption leads to high serum levels of amylase, lipase, trypsin, chymotrypsin, phospholipase, elastase, trypsinogen activation peptide, and phospholipase activation peptide, and these are also elevated in peritoneal and retroperitoneal tissues as well as lymph. The cheapest and most durable of these measurements as a diagnostic marker has been the total activity of serum amylase, and levels over three times the upper limit of normal in blood are now usually taken as diagnostic of acute pancreatitis in an appropriate clinical context. The serum lipase activity is a more specific measure and is almost as cheap to measure; levels of twice the upper limit of normal are significant.

C-reactive protein (CRP) is an acute phase reactant, which 36–48 h after onset of acute pancreatitis can be used for both assessment of severity (>150 mg/litre) and longitudinal monitoring of progress (see below).

Very high concentrations of circulating cytokines occur in the blood at an early stage in the disease, including tumour necrosis factor α (TNFα), platelet activating factor (PAF), and interleukin 6 (IL-6) with maximal levels in those with severe pancreatitis.

Diagnosis

The diagnosis is usually made from the clinical presentation, particularly the rapid onset of upper abdominal pain and vomiting, associated with an elevation of serum amylase and lipase. Where doubt exits, CT will reveal pancreatic swelling, fluid collection, and change in density of the gland. Contrast-enhanced CT is mandatory to identify areas of pancreatic ischaemia and infarction. MRI may ultimately replace contrast-enhanced CT in this area, but currently difficulties in access or with the availability of MRI-compatible equipment in the intensive care unit often limit its applicability in patients with organ failure. Occasionally the diagnosis may be made at laparotomy, when cholecystectomy plus cholangiography is good treatment, otherwise simple washout and closure is all that should be done.

Differential diagnosis

The differential diagnosis is that of an acute abdomen (Bullet list 3). Perforated peptic ulcer may be associated with a raised amylase (although it is unusual for blood amylase to be elevated in <5 h of pain onset), but plain abdominal radiograph or CT usually confirms the diagnosis (most typically by demonstrating free intraperitoneal air). A dissecting aortic aneurysm usually presents with an initial history of chest pain in a known hypertensive; abdominal pain and loss of arterial pulses occur later. High amylase activity in ectopic pregnancy derives from the fallopian tubes, but the clinical presentation should not be mistaken for acute pancreatitis. Mesenteric ischaemia or infarction are important to consider: CT of abdomen and the associated metabolic acidosis help the differentiation.

Elevated amylase activities are frequently present in patients with renal failure, while a lifetime of high amylase occurs in those with macroamylasaemia, when failure to filter the amylase complex molecule results in very low urine levels. Occasionally a presentation of diabetic ketoacidosis exhibits exocrine involvement by elevated amylase/lipase.

Bullet list 3 Differential diagnosis of acute pancreatitis

Grading disease severity

The importance of objective grading of disease severity is that less experienced clinicians can direct the more serious cases to high-dependency or intensive care facilities at an early stage of their illness, or instigate contrast enhanced CT scanning and early ERCP for patients who will derive most benefit. Grading is also useful for trials of different forms of therapy. The original Ranson grading system of 11 prognostic factors was developed for patients with acute pancreatitis due to alcohol abuse, but later an alternative system was introduced by him for those with gallstones. A single system, validated for both the common causes, is the modified Glasgow scoring system of eight prognostic factors (Bullet list 4). Validation came from a multicentre randomized British study by Corfield that assessed the place of peritoneal lavage in the management of severe acute pancreatitis.

Atlanta criteria

In 1992 in Atlanta a group of 40 multidisciplinary experts decided that severe acute pancreatitis was defined by the presence of failure of one or more organs, or the development of a major later complication (infected necrosis, abscess, or pseudocyst).

It is now recognized that neither transient hypoxaemia nor pseudocyst development is a good criterion for severe acute pancreatitis.

This can be very useful in confirming the diagnosis and also in grading severity of disease. It is very helpful in assessing an individual patient by demarcating location and extent of pancreatic damage. The area of nonperfused pancreas corresponds to the extent of necrosis. More than 50% necrosis (especially if the head of the pancreas is involved) is associated with the most severe disease.

This can be used to grade the severity of many diseases and has been shown to be useful in acute pancreatitis. It takes time for an individual clinician to learn to use the system, but it has the advantage that it can be applied throughout the first week of illness. The higher the score the worse the prognosis, and patients with the most severe acute pancreatitis have scores in excess of 10. A recent large (>1500 patients) clinical study assessing the potential role of a PAF antagonist in the management of higher-risk patients revealed major concerns about an admission entry criterion of APACHE II score of 6 or more to select patients, as the mortality rate in each of three groups of approximately 500 patients was less than 10%. Admission APACHE II scores of 8 or above are now recommended to identify severe disease.

Bullet list 4 Glasgow Prognostic Score—three or more prognostic factors indicates severe acute pancreatitis

Organ failure scoring

Clinical assessment by an expert is probably better than any of the other systems described to identify the most ill patients. Quantification of the clinical assessment by scoring 0 to 4 points for each of several organ systems using the modified Marshall Critical Illness Scoring System (Table 2) has revealed that 44% of patients with an APACHE II score of 6 or more will show an organ failure score of 2 or more at admission (Table 3). The hepatic index in the Marshall score is excluded in acute pancreatitis assessment as the presence of stones in the bile duct is so common. Many patients improve fairly rapidly with intensive supportive intravenous fluids and oxygen: they are regarded having low risk transient organ failure with mortality of less than 1%. By contrast, those who continue with a minimum modified Marshall score of 2 have persistent organ failure and high mortality (35–55%). It is this dynamic aspect of organ failure that is not recognized by the Atlanta criteria.

Other predictors

Many studies indicate that those over 70 years have a higher mortality, with one study finding 18.8% risk of death for this group. Any form of chronic cardiorespiratory or similar impairment is most common in this age group and further increases the risk of death.

Acute pancreatitis carries a significantly higher mortality and morbidity in patients with a body mass index of more than 30 kg/m², mainly because of increased risk of hypoxaemia, but also from other associated factors.

Table 2  Modified Marshall Critical Illness Scoring System (excludes hepatic index)

 

Score	0	1	2	3	4
Systolic BP (mm Hg)	>90	<90 responds to fluid	<90 poor response to fluid	<90 (pH<7.3)	<90 (pH<7.2)
	>400	301–400	201–300	101–200	<101
Glasgow Coma Score	15	13–14	10–12	6–9	<6
Platelet count × 109/litre	>120	81–120	51–80	21–50	<21
Creatinine (µmol/litre)	<134	134–169	170–310	311–439	>439

FiO 2, fraction of inspired oxygen; PO 2, partial pressure of oxygen.

Table 3  Importance of the dynamics of organ failure in acute pancreatitis

 

	Patient nos.	Deaths
Transient organ failure	104 (25.3%)	1 (1.0%)
Persistent organ failure	123 (29.9%)	47 (38.2%)
No organ failure	184 (44.8%)	5 (2.7%)

Outcome derived from combined data from Buter et al. (2002) and Johnson and Abu Hilal (2004). All patients had a minimum APACHE II score of 6 at admission.

CRP

The main value of this acute phase reactant is at around 36 to 48 h of illness, when values between 200 and 600 mg/litre are found in the most severely ill patients.

Preliminary studies measuring this substance hold considerable promise that it may be the most useful single marker of disease severity, but additional clinical studies are needed and its measurement is not routine clinical practice.

The activation of trypsinogen releases trypsin and a small peptide (trypsinogen activation peptide), which passes unchanged in the urine, where its level can be used as a marker of severity. It has now been employed in two clinical studies with the promise of this being a valuable step forward in assessment of severity, but its measurement is not routine clinical practice.

Clinical progress and outcome

Mild acute pancreatitis

Those patients who fail to meet objective criteria of severe acute pancreatitis tend to have a low mortality (maximum 2%) and rarely need to be in hospital beyond 7 to 10 days. Simple therapeutic measures normally suffice through the first 24 to 48 h, after which time monitoring and support can usually be discontinued, but it is best to assume at presentation that a patient may develop severe disease and to provide monitoring within a high dependency environment when possible.

Good prospective studies of analgesia in acute pancreatitis are notable for their absence. Pain is usually treated with morphine or buprenorphine. Care is required if nonsteroidal anti-inflammatories are used because they may exacerbate renal compromise. Epidural analgesia is employed in some German hospitals. Agitation is common, particularly with alcoholic pancreatitis, and intravenous benzodiazepines or haloperidol may also be required.

Even in patients with mild disease it may occasionally be necessary to provide 4 to 5 litres of intravenous fluid in the first 24 h of the illness.

When the acute attack has resolved, it is important to consider prevention of further episodes of pancreatitis. If gallstones have been identified by ultrasound scanning, then laparoscopic or open cholecystectomy should be performed in the same admission, with ERCP sphincterotomy alone considered a satisfactory alternative in patients who are infirm.

Severe acute pancreatitis

The mainstays of treatment are reversing hypoxaemia, restoration of circulating volume, maintaining tissue perfusion, and organ support. Patients who meet objective criteria of severe acute pancreatitis are often tachypnoeic, tachycardic, and hypotensive. They should be managed within a high-dependency environment, with facilities for invasive monitoring. Successful resuscitation and prevention of secondary organ dysfunction is key to improving outcome, especially as 40 to 50% of deaths result from multiple organ failure within the first week of illness rather than the late infective complications.

Body wall staining at the umbilical area (Cullen’s sign) or in the flanks (Grey Turner’s sign) can occur, usually appearing around the fourth day of illness. This results from transudation of blood stained retroperitoneal fluid.

Respiratory failure

Hypoxaemia reflects disease severity in acute pancreatitis, and all systems of grading severity include hypoxaemia. Basal atelectasis and respiratory compromise are very common and must be expected. Early pleural effusions (more commonly on left side) indicate severe disease. The onset of respiratory failure may be insidious, with an initial compensatory tachypnoea easily overlooked, although this can be highlighted by the routine use of Modified Early Warning Score (MEWS) systems that mandate recording of respiratory rate and request for patient review if the MEWS score increases beyond an agreed level. Pulse oximetry is valuable in monitoring, but arterial gas analysis may be needed three or four times in the first 24 h to make sensible decisions about humidified oxygen therapy and possible ventilator support. Most patients can be managed with supplemental humidified oxygen, but assisted ventilation may be required in more severe cases for several weeks.

Circulatory failure

The initial therapy of hypovolaemia is of great importance in maintaining cardiac and renal function. Intravascular fluid deficits in the more severe cases are considerable, often underestimated, and intravascular fluid monitoring is required where tissue perfusion is not easily restored. The combination of fluid lost from vomiting and loss of capillary integrity can be very substantial. The use of catecholamines should only be considered after it is established that adequate fluid restoration has been achieved. In the most severe cases of acute pancreatitis the cardiovascular changes are very similar to those encountered in septic shock, with a high cardiac output and low peripheral vascular resistance. Stress on the heart may cause arrhythmias and ischaemic changes.

Renal failure

After immediate resuscitation, intravenous fluid should be given at sufficient rate to produce 0.5 ml/kg of urine per hour, but oliguric acute tubular necrosis is not uncommon in patients presenting with hypovolaemia and poor tissue perfusion, in which case fluid challenges should not be pursued to the point of inducing pulmonary oedema. The use of mannitol or furosemide is without proven benefit. Renal replacement therapy by haemodialysis or haemofiltration may be required if renal failure becomes established.

Pyrexia

This reflects cell damage and necrosis, as with any condition associated with tissue destruction. High fever within the first 24 h is rare and, if associated with jaundice, is suspicious of ascending cholangitis; consideration should be given to biliary decompression. In the more severe cases of acute pancreatitis the first week is characterized by the development and persistence of a systemic inflammatory response syndrome, one aspect being the development of a swinging pyrexia. This is often inappropriately taken as evidence of infection but is a reflection of the inflammatory cascade and the presence of devitalized tissue in the retroperitoneum.

Specific aspects of management of severe acute pancreatitis

Initial management

Analgesia (with pethidine or buprenorphine) and restoration of circulating volume is the mainstay of early treatment. Although nasogastric decompression has no effect on outcome, it can sometimes alleviate vomiting. In all but the mildest cases, careful monitoring of organ function is required with pulse oximetry, urinary catheterization, and noninvasive blood pressure measurement in addition to biochemical, haematological, and blood gas analysis. Treatment is directed towards support and restoration of lung, renal, and cardiac function.

As stated above, those who have severe acute pancreatitis are usually very ill at the time of admission or within 24 h and warrant high-dependency or intensive care therapy. Close routine monitoring for system failures as well as biochemical and haematological abnormalities is required. An algorithm for suggested steps in the management of severe acute pancreatitis is based on the United Kingdom National Guidelines published in 1998 and revised in 2005. Particular interventions are discussed below.

Endoscopic retrograde cholangiopancreatography (ERCP)

The one group of patients with proven benefit from ERCP and biliary decompression are those with pancreatitis associated with cholangitis (jaundice and pyrexia), where the cholangitis, rather than any pancreatic inflammation, is driving organ dysfunction. The role of ERCP is more debatable in other circumstances. There is no role for early ERCP in mild disease. For severe acute pancreatitis there are four randomized trials of early ERCP, three reported in peer-reviewed journals and one in abstract form. The evidence is equivocal, and ERCP is probably difficult to justify in patients who are not jaundiced. No study has addressed the question of whether sphincter decompression in the absence of a common bile duct stone affects disease progression.

Several studies in the 1990s advocated the early initiation of prophylactic antibiotics such as imipenem for 10 to 14 days in patients with predicted severe disease, and most clinicians adopted this policy. However, the quality of these clinical trials was not to the accepted standard for randomized double-blind studies, and two larger higher-quality multicentre studies—both mainly including patients with CT-proven necrosis—have more recently failed to demonstrate benefit in terms of either the primary endpoint of pancreatic infection, or the secondary one of mortality (Table 4. One of these employed ciprofloxacin and metronidazole vs placebo in 114 patients, and meropenem vs placebo was used in 100 patients in the other study. Although it could be argued that neither of these trials was adequately powered, the clinical effort and expense of mounting even larger antibiotic studies of only the most severely ill patients, with poor prospects of a positive outcome, means such a future trial is highly improbable. It seems appropriate to recommend that antibiotic therapy should not be given until and unless a specific indication arises.

Before the 1960s, gut rest was not considered an issue, but intake was often limited because of gastric stasis. The development of modern parenteral support in the late 1960s, along with laboratory studies investigating the stimulation of pancreatic exocrine function, led to the avoidance of enteral intake during the acute phase of disease. However, randomized studies have since shown that early nasojejunal nutritional support is cheaper and associated with fewer side effects than total parenteral nutrition, also that there is no difference in outcome with nasogastric compared to nasojejunal feeding. In clinical practice, therefore, the mode of nutrition support does not appear to affect the disease process, and the choice of delivery relates to tolerance and minimizing morbidity associated with the delivery system.

Small studies indicated potential benefit from adding probiotics to nasoenteral feeds, but surprising and disturbing adverse effects were found when the large Dutch PROPATRIA randomized study of 298 patients was reported in 2008: infective complications occurred in approximately 30% of each group of patients, but the mortality in the probiotic group was 16% vs 6% in the placebo group, and bowel ischaemia or infarction was found in eight fatal cases in the treatment group but none of the placebo group. Probiotics should not be used.

In the 1990s, there was considerable interest in the potential of manipulating the inflammatory response to prevent cytokine cascade modulated multiorgan failure, but the initial promise of the PAF antagonist studies proved unfounded after a multinational study of >1500 patients failed to show benefit. In a similar way, high dose intravenous antiprotease and antioxidant therapy as well as the somatostatin analogue octreotide showed no benefit in randomized double-blind studies. There is no clinically available specific intervention that can be recommended in addition to respiratory, fluid, and circulating volume support in the early phase of disease.

Table 4  Deaths in early antibiotic randomized double-blind studies in severe acute pancreatitis

 

Study	Antibiotic	Placebo
Isenmann et al. (2004)	3/58	4/56
Dellinger et al. (2007)	10/50	9/50
Totals	13/108	13/106

Surgical intervention

The roles of surgery in acute pancreatitis are shown in Bullet list 5. There is no place for early intervention, other than occasionally to establish the diagnosis where diagnostic doubt remains despite clinical, biochemical, and radiological assessment, but surgery does have an important role in the management of complications.

In patients with mild attacks, gallstones (when relevant) should be eradicated, ideally within the same hospital admission. In those with severe disease, the role of surgery is aimed at the control of sepsis, usually several weeks after disease onset. Most sterile collections can be managed conservatively and drainage achieved with low morbidity as a delayed procedure once the acute phase has passed.

The indication for surgical intervention is the development of infection within an acute fluid collection, which is normally heralded by deterioration in biochemical and haematological markers, and deterioration in organ function. Traditional management demanded complete debridement of any devitalized tissue at the earliest opportunity, but it is now clear that maintaining adequate drainage is more important than eradication of necrosis, and options for intervention include open surgical drainage, percutaneous, or endoscopic minimally invasive approaches, with choice of procedure determined by local expertise and preference.

In open surgical debridement, the necrotic tissue is removed by a combination of gentle finger and forceps dissection, after which there are two options: (1) to establish a postoperative lavage system, which may necessitate up to three inflow and three outflow drains because of the tendency for retroperitoneal extension of the infected necrosis down the paracolic gutters and upwards towards the diaphragm; or (2)—if venous ooze of blood is a particular problem—packing of the abdominal cavity with large cotton packs wrapped in nonadhesive paraffin gauze, together with limited closure or nonclosure of the abdominal wall. Such patients are invariably in intensive care and receiving mechanical ventilation. The packs should be changed at intervals of 48 to 72 h, with removal of any extension of infection or necrosis. Abdominal wall closure and postoperative lavage may be established after the second or subsequent operations.

Bullet list 5 Role of surgery in acute pancreatitis

Minimally invasive surgical approaches have been developed to achieve adequate drainage of sepsis whilst minimizing the surgical insult. The first of these was developed in Glasgow, utilizing either a left flank retroperitoneal approach (80%) or right anterior drainage (20%), and is now our standard surgical approach. Other variations, including endoscopic drainage methods, have been described, but our own experience of these has highlighted the difficulties in maintaining drainage in collections with a large solid component, and we find this method more useful in fluid-predominant collections being drained several months after an acute attack. However, to emphasize the point again, in practical terms achieving continued sepsis control is probably more important than the method used.

Specific complications of acute pancreatitis

Acute pancreatic pseudocyst

This is probably a misnomer in that acute fluid collections requiring intervention rarely contain little or no debris (as required by the Atlanta definitions). Unfortunately, necrosis is poorly demonstrated by CT (MRI is much more sensitive), and an acute fluid collection on CT may appear homogeneous and invite radiological intervention. This leads to inadequate drainage and persistent infection. We prefer the terms ‘fluid-predominant’ or ‘solid-predominant’ acute fluid collection or pseudocyst, as this defines subsequent treatment.

Pancreatic pseudocyst most commonly occurs in the lesser sac and often represents a closed pancreatic fistula, as a breach in the main or major pancreatic duct can frequently be demonstrated at ERCP. Transpapillary duct stenting and ERCP should be avoided if possible, as this leads to the introduction of infection in more than 10% of patients. Percutaneous aspiration alone invariably results in recollection of the fluid quite rapidly, while infection is potentially associated with long-term percutaneous drainage. In the absence of sepsis, fluid-predominant acute fluid collections may be adequately drained as a one-step procedure by EUS-guided endoscopic transmural procedures, although repeated tract dilatation is often required. By contrast, solid-predominant acute fluid collections, especially in younger, fit patients, may be best dealt with by internal surgical drainage to the stomach or to a defunctioned Roux loop of jejunum. This procedure can be done either laparoscopically or at open surgery, depending on local expertise, and can also be combined with a cholecystectomy.

Pancreatic ascites

This condition rarely occurs in association with acute pancreatitis, but when it does it is due to spontaneous decompression of a pancreatic pseudocyst, with escape of pancreatic juice into the peritoneal cavity. Amylase-rich fistula fluid is common after surgical drainage and can often be successfully managed by endoscopic transpapillary drainage.

Rare complications

These include splenic vein thrombosis and subcutaneous fat necrosis. Increased availability and use of high-quality imaging techniques mean that splenic and segmental portal vein thrombosis are increasingly detected. Painful subcutaneous bright red nodules (mimicking erythema nodosum) may occasionally occur on upper and lower limbs when spontaneous decompression of a pseudocyst takes place: skin biopsy makes the diagnosis. Cranial areas of fat necrosis have also been noted at autopsy.