A number of factors are capable of initiating acute pancreatitis although their mechanism of action is not known (Table 1) 386. Alcoholism and biliary tract disease are the processes most commonly associated with pancreatitis: together they account for 80 per cent of acute cases. The number of patients with idiopathic pancreatitis depends on the extent to which rare causes are searched for.

Acute pancreatitis may be the first manifestation of a tumour causing pancreatic duct obstruction: pancreatic carcinoma should be considered in non-alcoholic patients in whom no biliary disease is identified. Infectious agents which have been implicated as a course of pancreatitis include mumps virus, coxsackie virus, Mycoplasma, and parasites. Many drugs are capable of causing abdominal pain and hyperamylasaemia, though only a few have been implicated as clear causes of pancreatitis: these include thiazide diuretics, 6-mercaptopurine, l-asparaginase, azathioprine, oestrogens, frusemide, methyldopa, sulphonamides, tetracycline, pentamidine, DDI, enalapril, and procainamide. Patients with types I and V hypertriglyceridaemia frequently suffer from pancreatitis, usually associated with extremely high levels of serum triglycerides and lactescent serum. However, hypertriglyceridaemia can also be the result rather than the cause of pancreatitis. Postoperative pancreatitis can be fatal and is often due to iatrogenic pancreatic trauma. Pancreatitis following cardiopulmonary bypass is being recognized with increasing frequency; the majority of cases are mild or even subclinical, but evidence of severe pancreatitis is found in 25 per cent of patients who die following cardiac surgery.

The corrosive activity of pancreatic digestive enzymes is the driving force which differentiates acute pancreatitis from other abdominal inflammatory conditions. Although different agents may trigger acute pancreatitis by a variety of different mechanisms, they all ultimately result in intraparenchymal enzyme activation, tissue destruction, and ischaemic necrosis. The ultimate common finding in pancreatitis is the presence of activated proteolytic enzymes in the pancreatic parenchyma and retroperitoneum.

The pancreas is protected from autodigestion by several mechanisms. Pancreatic digestive enzymes are synthesized, transported, and secreted in the form of inactive precursors. From the time of their synthesis on the rough endoplasmic reticulum until their secretion by exocytosis, the potentially harmful digestive enzymes are sequestered from the cytosol in membrane bound organelles that contain potent inhibitors capable of inactivating prematurely activated trypsin. Other intracellular proteases that might be capable of activating trypsin are segregated from the zymogen granules in their respective organelles. Once discharged from the acinar cell, these proenzymes are transported to the gut through the pancreatic duct, which is not normally permeable to proteins of this size. When the zymogens reach the gut, enterokinase, an enzyme present in duodenal mucosa as well as in bile, cleaves trypsinogen to form activated trypsin. Trypsin then converts other inactive precursors to achieve forms in a cascade fashion.

The mechanism underlying the premature enzyme activation seen in acute pancreatitis, including the factors that inappropriately activate enzymes and the endogenous control mechanisms which combat the process, are poorly understood. &agr;&sub1;-Antitrypsin and &agr;&sub2;-macroglobulin, two potent protease inhibitors, are present in the pancreas, but their significance is unknown.

Once pancreatitis has been initiated, vascular events seem to play a major role in its propagation. Microinfarctions of the pancreas due to inflammation-induced microvascular thrombosis occur in most forms of severe acute pancreatitis. The severity of pancreatitis can be limited experimentally by preserving pancreatic blood flow with the use of dextran or isoproterenol. Conversely, factors which impair pancreatic blood flow are capable of converting mild pancreatitis to a severe form of the disease. It is not clear whether hypoxia or hypoperfusion alone can cause acute pancreatitis, but their role in the disease is supported by the small percentage of patients undergoing cardiopulmonary bypass and victims of accidental hypothermia who develop acute pancreatitis.

In mild pancreatitis the inflammatory response is well controlled. There may be oedema, usually confined to the pancreas, but tissue necrosis is uncommon. In severe pancreatitis the response is uncontrolled, leading to more widespread tissue injury and the many systemic manifestations of the disease. An inflammatory exudate rich in proteolytic enzymes, kinins, and vasoactive substances escapes from the pancreas into the lesser sac, retroperitoneum, and peritoneal cavity. It can then be absorbed into the systemic circulation leading to shock, respiratory failure, and renal failure.

The aetiology of shock in the early phase of acute pancreatitis is multifactorial. Sequestration of fluid in the interstitium, or third space, results in intravascular fluid depletion. However, restoration of a euvolaemic status, as indicated by central venous pressure and pulmonary capillary wedge pressure measurements may not restore normal blood pressure and haemodynamics. In addition, cross-circulation experiments have demonstrated the presence of humoral factors capable of causing hypotension in control animals without pancreatitis. Kinins, serotonin, and vasoactive amines have been implicated as mediators of diminished peripheral vascular resistance and increased vascular permeability. Although attempts to measure plasma levels of these compounds have not always correlated with the haemodynamic parameters, they are always present in high concentration in the peritoneal fluid during acute pancreatitis. Removal of this fluid by peritoneal lavage reverses the haemodynamic alterations.

In a study of 34 patients with severe acute pancreatitis, Cobo found that there was a failure of system vascular resistance to increase appropriately in the face of hypovolaemia. The difference between survivors and non-survivors in this series was the inability of the myocardium to compensate appropriately for this loss of peripheral resistance by increasing cardiac output. This was attributed to a myocardial depressant factor, although the circulating myocardial depressant factor which has been demonstrated in the plasma of hypotensive animals with severe pancreatitis has not been proved to exist in man. In most patients with acute pancreatitis the cardiac index is increased. Elevations in pulmonary vascular resistance have been documented, but these do not correlate with pulmonary capillary wedge pressure. When left ventricular stroke work was plotted as a function of wedge pressure (Sarnoff curve), the curve in patients with acute pancreatitis was shifted downward compared to that in normal patients. These data were interpreted to mean that myocardial performance was depressed even though the cardiac index was increased. The relationship between the Sarnoff curve and the elevated pulmonary vascular resistance is not clear, however, and the existence of a myocardial depressant factor in man remains in doubt.

The mechanism underlying the development of renal failure, a feared complication of acute pancreatitis, is also unclear. While hypotension and hypovolaemia may play some role, renal failure may also develop in patients who have had no preceding episode of shock and hypovolaemia.

In animal models of acute pancreatitis, significant reductions in glomerular filtration rate (GFR) occur prior to changes in peripheral circulation, and fluid resuscitation and restoration of normal cardiac output do not necessarily restore glomerular filtration rate to normal. These changes partially reflect an increase in renal vascular resistance. Most patients developing renal failure in association with acute pancreatitis are initially hypovolaemic with impaired renal perfusion and filtration. Progression to renal failure occurs even though systolic blood pressure is restored and is probably secondary to unidentified humoral factors that exert vasoconstrictive effects on the renal arterioles.

There is no foolproof diagnostic test for acute pancreatitis: the diagnosis rests on interpretation of clinical and laboratory information. In some cases the diagnosis is obvious, but the protean manifestations of this disease may make diagnosis difficult. The typical presentation includes upper abdominal pain, nausea, vomiting, and low grade fever. Tenderness is generally limited initially to the upper abdomen, although in severe cases diffuse peritoneal irritation may be present. In patients with severe peritoneal irritation laparotomy can be the safest means to establish a diagnosis and avoid missing a surgically correctable disease. When acute pancreatitis is found unexpectedly at laparotomy, it should not be a source of embarrassment but may provide the opportunity to treat any biliary pathology. Rarely, a patient has no pain, but presents with distension, ileus, fever, and tachycardia. This form of the disease is more often seen in patients with pancreatic infarction following cardiopulmonary bypass or accidental hypothermia.

Although the serum amylase level is the single most useful laboratory test, there is a significant incidence of false positive and false negative results. The serum amylase level in acute pancreatitis generally rises within 2 to 12 h of onset of symptoms and returns to normal over the next 3 to 5 days. Early resolution of hyperamylasaemia may indicate the early resolution of pancreatitis, but pancreatic necrosis may develop despite a normal serum amylase level, perhaps because of the extensive destruction of the gland. There is no correlation between serum amylase level and the aetiology, prognosis, or severity of the disease. The serum amylase is normal in up to 30 per cent of patients with acute pancreatitis, especially in alcoholics with underlying chronic pancreatitis or patients with hypertriglyceridaemia. Many non-pancreatic diseases may be associated with hyperamylasaemia (Table 2) 387.

Hyperamylasaemia following abdominal surgery (most commonly biliary surgery and gastroduodenal surgery) is then due to genuine pancreatitis, presumably the result of operative trauma to the pancreas or trauma to the major or minor papilla. In many cases postoperative hyperamylasaemia is non-pancreatic in origin and reflects poorly understood metabolic changes. Isoamylase analysis can distinguish pancreatic from non-pancreatic sources of amylase, but this is not possible at most institutions.

Persistent unexplained hyperamylasaemia in a patient with normal renal function may be the result of macroamylasaemia, a term describing the presence of circulating amylase complexed with a macromolecule. This complex is too large to be filtered through the glomerulus and thus accumulates in the blood. The diagnosis is made by measuring amylase clearance, which is low. Macroamylasaemia occurs in 0.4 per cent of the general population and in 5.9 per cent of patients with hyperamylasaemia. Perhaps more common than macroamylasaemia is normal distribution hyperamylasaemia – a condition which the homeostatic balance between amylase production and breakdown is set at a high level. There is no predominant isoamylase type. This is the most likely explanation for persistent unexplained hyperamylasaemia in patients who are otherwise well, and it is important that this conditions is identified if unnecessary tests and treatment for non-existent pancreatitis are to be avoided.

The recent development of rapid accurate lipase assays has facilitated measurement of this enzyme as a marker for pancreatitis. Since the pancreas is the only known source of lipase this may represent a more specific marker for pancreatic disease than total serum amylase. The sensitivity and specificity of serum lipase determination may be equivalent to that of isoamylase measurement.

No radiographic technique is a sensitive index of acute pancreatitis, but some techniques can be important in confirming the clinical impression of pancreatitis, especially when the serum amylase level is normal. Radiography can assist in excluding other causes of acute abdominal pain and aid in the early detection of complications. Ultrasound yields abnormal findings in only 30 to 50 per cent of patients with acute pancreatitis. The major limitation of the tests is that the pancreas cannot be adequately seen in a large number of patients because intra-abdominal gas or excess body fat obscures tissue planes. Computerized tomography (CT) offers the advantages of better visualization of retroperitoneal structure and better overall visualization of the pancreas. The examination is more accurate regardless of the presence of abdominal gas, the patient's size, or clinical status. The changes detected by CT reflect pancreatic inflammation and oedema in approximately 70 per cent of proven cases. A normal CT scan or ultrasound does not exclude the diagnosis of pancreatitis; in fact these scans usually are normal in cases of mild pancreatitis. The CT scan is especially useful in diagnosing later complications of acute pancreatitis such as pancreatic necrosis, pseudocysts, abscess, or peripancreatic fluid collections. In severe pancreatitis, changes seen on early CT scans correlate with the subsequent development of pancreatic abscesses and pseudocysts, and early CT scans may therefore have some prognostic value, although similar information can also be discerned from Ranson's criteria (Table 3) 388.

Diagnostic paracentesis has been used to demonstrate the presence of pancreatitis and predict the severity of the disease. Although the amylase concentration in the peritoneal fluid is of no diagnostic value, the physical properties of the fluid are of interest. A severe attack is heralded by the presence of dark peritoneal fluid and mild attacks by straw coloured fluid. The ability to aspirate more than 10 ml of free peritoneal fluid, regardless of colour may also be indicative of a severe attack. However, the Ranson criteria are at least as accurate as peritoneal aspiration and are less invasive, although it takes 48 h to gather the data.

Although many measures have been tried in an effort to modify the natural history of pancreatitis, no known agent arrests or reverses the inflammatory process. Most attacks of acute pancreatitis are self-limited and subside spontaneously. Existing treatment can only be supportive and directed at complications which may develop. Prevention of future attacks can be accomplished when a remediable cause is identified.

The most important requirement in the early treatment of pancreatitis is maintenance of adequate hydration. If the patient becomes hypovolaemic, and the splanchnic circulation is compromised, the pancreas may become ischaemic, with the potential for the development of complicated pancreatitis. In severe pancreatitis, where large fluid shifts occur, measurements of cardiac output and pulmonary capillary wedge pressure with Swan–Ganz catheters may be necessary, especially if cardiac or renal compromise complicate fluid management. Transfusion may be required if significant anaemia develops. When there is haemodynamic instability in spite of adequate fluid replacement, peritoneal lavage should be performed.

Hypoxaemia is common in acute pancreatitis; 45 per cent of patients may have an arterial Po&sub2; level less than 50 mmHg. The four types of pulmonary disease associated with acute pancreatitis are early hypoxia without radiographic abnormality, respiratory insufficiency with non-specific radiological abnormalities (diaphragmatic elevation, atelectasis, pulmonary infiltrates, pleural effusions), pulmonary oedema occurring early in the illness, and late respiratory failure secondary to systemic sepsis. Respiratory failure is most likely to occur in the most severely ill patients and the need for intubation and ventilatory support indicates the likelihood of a fatal outcome.

Early hypoxaemia is usually associated with hypocarbia and is extremely common during the first 48 h of acute pancreatitis, whether clinically mild or severe. The only clues to the presence of hypoxaemia may be subtle tachypnoea and hyperventilation: the chest radiograph is usually normal. Impaired diffusion capacity, decreased compliance, increased airway resistance, and decreased vital capacity have all been demonstrated during this early phase, although the underlying cause or causes have eluded definitive documentation. Substances whose levels are known to be elevated in acute pancreatitis—amylase, lipase, elastase, insulin, fatty acids and triglycerides—increase oxyhaemoglobin affinity in vitro, and the hyperventilation and hypoxia observed without radiographic abnormality or sense of dyspnoea may relate to this altered affinity. Detection of early hypoxia depends less upon clinical observation than upon serial blood gas determinations. Since early hypoxaemia is both common and insidious, supplemental oxygen should be part of the routine treatment of older patients with acute pancreatitis. Although volume overload probably does not play a primary role in the aetiology, it is sensible to restrict intravenous fluids to a volume necessary to maintain adequate peripheral circulation and urine output rather than trying to achieve arbitrary central venous pressure or pulmonary capillary wedge pressure goals.

Between 30 per cent and 60 per cent of patients who survive beyond the first 48 h of illness develop radiographically demonstrable pulmonary complications. Unlike early occult hypoxaemia, these complications correlate directly with the severity of the underlying pancreatitis: they usually occur with continuing pancreatic injury and are associated with increased mortality. Atelectasis, bibasilar infiltrates, and diaphragmatic elevation are non-specific abnormalities common to any disease involving subdiaphragmatic inflammation, and are partly the result of splinting of the abdominal wall and restricted excursion of the diaphragm due to pain, localized peritonitis, and ascites. These phenomena usually resolve as the underlying pancreatitis subsides. Pneumonitis, commonly the result of aspiration of vomitus, may complicate recovery and its prevention remains a reason for using nasogastric suction in patients with acute pancreatitis. Pleural effusions that occur during the acute phase of the illness are sympathetic collections in response to subdiaphragmatic inflammation. They are usually sterile exudates with low amylase content and they are not generally large enough to compromise ventilation significantly and rarely require removal by thorocentesis.

Treatment of these forms of respiratory failure is supportive and directed primarily at the underlying pancreatitis. Humidification of the airways, supplemental tracheal suctioning, and ventilatory exercise to keep alveoli open are helpful. Antibiotics are not necessary unless radiographic evidence of pneumonia appears; choice of any should be based on the results of sputum Gram stain and culture. If the patient is already being treated with antibiotics for the pancreatitis, a resistant organism may have collected in the respiratory tract and a change of antibiotics is often necessary to treat the pulmonary infection effectively.

Non-cardiogenic pulmonary oedema occurs in 10 to 30 per cent of patients with acute pancreatitis, usually 2 to 4 days after the onset of the attack. The early circulatory dysfunction phase of pancreatitis is generally over by the time respiratory failure occurs; in fact, all of the other clinical features of the attack may be subsiding or may have resolved when respiratory failure first begins to evolve. The chest radiograph shows pulmonary vascular congestion, perihilar fluffy infiltrates, and finally full blown pulmonary oedema. The oedema fluid has a high protein content, unlike the situation in cardiogenic pulmonary oedema in which the fluid is a transudate. The clinical picture resembles that described as the adult respiratory distress syndrome but it evolves rapidly, without sepsis playing a part.

Attempts to understand the pathogenesis of this striking syndrome have focused on haemodynamic changes and on substances thought to affect endothelial permeability. Most patients have normal pulmonary capillary wedge and central venous pressures with normal or elevated cardiac output. Some studies have reported elevations of pulmonary vascular resistance, although the relationship of this to the pathogenesis of pulmonary oedema is not clear. The finding that pulmonary oedema fluid is exudative in the presence of relatively normal haemodynamics implies that pulmonary oedema is a consequence of abnormal permeability of alveolar capillaries. The mechanism of injury to the alveolar capillary membrane is unknown. Several possibilities exist: free fatty acids generated from serum triglycerides by lipase are directly toxic to this membrane and increased serum triglycerides are the best predictor of the likely development of pulmonary oedema. Another potential mechanism involves destruction of pulmonary surfactant. Phospholipase A&sub2; released from the pancreas is capable of splitting lecithin into lysolecithin and a fatty acid. There has not, however, been a consistent correlation between elevated serum phospholipase A&sub2; levels and the severity of pancreatitis or its respiratory complications. A third potential mechanism of endothelial injury involves vasoactive substances. Kinins, vasoactive amines, and other poorly characterized substances liberated in acute pancreatitis have been blamed for increased endothelial permeability, interstitial fluid leak, and hypotension. These substances might affect the permeability of the alveolar capillary membrane as well as that of other capillary beds. Finally, activated pancreatic proteolytic enzymes are capable of initiating intravascular coagulation, resulting in microthrombosis and microembolization in the pulmonary vascular system.

The treatment of pulmonary oedema secondary to pancreatitis is also primarily supportive. Diuresis can be beneficial if filling pressures are high but over diuresis should be avoided because of potential renal compromise. If severe pulmonary oedema and hypoxia supervene, endotracheal intubation, positive pressure ventilation, positive end-expiratory pressure are necessary. Ventilatory support may be required for several weeks, but the alveolar membrane injury usually heals within 7 to 10 days. It is important to be vigilant for the development of pneumonia in this setting: sputum should be obtained daily for Gram staining, and the development of purulent sputum with a predominant organism should be treated as evidence of bacterial pneumonia. The use of albumin and diuretics in this condition is controversial, but is likely to be ineffective when the alveolar capillary membrane is leaky. Likewise, there is no strong evidence that treatment with pharmacological doses of steroids or phospholipase A inhibitors is beneficial. If the hypothesis that lung injury is due to circulatory factors, whatever they may be, is correct, peritoneal lavage should be effective treatment for the pulmonary lesion as well as for the circulatory dysfunction which is its primary indication. However, since respiratory failure very rarely causes death in these patients, and since the current techniques of respiratory support are very effective, it is not an indication for either percutaneous or operative lavage.

‘Putting the pancreas at rest’ by fasting the patient has been a traditional part of medical treatment. Recent experimental and clinical studies have shown that pancreatic secretion is already below basal levels in acute pancreatitis and this finding may explain why somatostatin analogues, which inhibit pancreatic secretion, have little effect on the course of acute pancreatitis. Because studies of patients with mild pancreatitis do not show nasogastric suction to have any benefit it has been fashionable to decry it. Nonetheless, some patients suffer from clear recurrent inflammation when nasogastric suction is discontinued, and all patients with pancreatitis are at risk of vomiting and aspirating because of ileus. Therefore, when ileus or significant abdominal distension is present, a nasogastric tube is useful.

Although H&sub2;-receptor blockers might theoretically confer the same benefits as nasogastric suction they have not been shown to have any effect on the course of pancreatitis. H&sub2;-receptor blockers or antacids are still helpful in prophylaxis against haemorrhagic gastritis to which these patients are susceptible.

Patients with severe pancreatitis often cannot be fed for several weeks or months. Once severe pancreatitis has developed total parenteral nutrition should be instituted. Intravenous fat emulsions do not exacerbate pancreatitis in patients with normal triglyceride levels. If triglyceride levels are raised, however, fat emulsion should not be used. Total parenteral nutrition should be continued until the patient appears clinically well. If there is a large inflammatory mass, residual abdominal tenderness, or hyperamylasaemia, total parenteral nutrition should be continued. When pancreatitis is fully resolved, oral feeding can begin cautiously. Since fat, protein, and solid food are strong stimulants of pancreatic secretion begin with a low fat liquid diet, rich in carbohydrates, and to advance the diet slowly. Rapid introduction of solid food may reactivate resolving pancreatitis.

Antibiotics are generally ineffective in preventing the late septic complications of acute pancreatitis, and their use may even promote selection of organisms that are more difficult to treat later on. It seems reasonable to withhold antibiotic therapy in alcoholic patients with mild to moderate pancreatitis. In severe pancreatitis, however, tissue necrosis is likely and there is a high risk of subsequent infection.

In a large German study 40 per cent of patients with severe pancreatitis requiring surgery had bacterial infection of the pancreas or peripancreatic tissue. The use of prophylactic broad spectrum antibiotics in this group of seventy ill patients has not been adequately studied. Bacteria can be recovered from the peripancreatic fluid in fulminant cases early in the course of the disease. Therefore, pre-emptive treatment with broad spectrum antibiotics might be justified in the hope of limiting bacteria invasion in this period of vulnerability. Since no clinical or biochemical parameter can reliably differentiate sterile pancreatic inflammation from early pancreatic infection, CT-directed percutaneous aspiration of inflammatory pancreatic masses should be used to guide therapy (Fig. 1) 1324. Pancreatic masses or fluid collections in patients with inflammatory signs should be sampled since some may be sterile and others infected. During the course of prolonged pancreatitis, repeated aspiration should be performed each time sepsis is suspected. The presence of infection requires either operative or percutaneous drainage as antibiotics alone have never been shown to cure a pancreatic abscess.

In spite of the central role of enzymatic digestion in its pathogenesis exogenous antiproteases are ineffective in limiting the progression of human acute pancreatitis.

Hypocalcaemia occurs in severe pancreatitis and a markedly reduced serum calcium is a poor prognostic sign. Only ionized calcium has physiological activity; any assessment of hypocalcaemia should therefore include either measurements of ionized calcium or a correction of the total calcium level on the basis of corresponding serum albumin level. Most cases of hypocalcaemia are due to reduction of serum albumin. Nonetheless hypoalbuminaemia does not account for all instances of hypocalcaemia in acute pancreatitis. Although extensive fat necrosis may be associated with deposition of large quantities of calcium in these areas, administration of exogenous calcium and the usual calcium homeostatic mechanism involving parathormone does not restore normocalcaemia. Studies of parathormone secretion have been contradictory, as have studies implicating calcitonin as a cause of hypocalcaemia. Hypomagnesaemia is common in alcoholic patients, and since this is necessary for both parathormone secretion and mobilization of calcium from bone, hypomagnesaemia may contribute to hypocalcaemia. In general, it is unnecessary to treat hypocalcaemia in these patients as they are rarely symptomatic. However, if neuromuscular irritability develops or the Q–T interval becomes prolonged intravenous calcium should be administered. Hypomagnesaemia and hypokalaemia, if present, must be corrected prior to administration of intravenous calcium.

Rigid control of blood glucose is unnecessary and unwise. Hyperglycaemia tends to be transient and is associated with high glucagon levels, which fall rapidly when the disease subsides. Therefore, insulin should be administered cautiously and blood glucose levels of 200 to 300 mg/dl should be tolerated.

No single operative treatment and probably no operation cures pancreatitis. The role of surgery is reactive and responsive to particular complications as they evolve in a small minority of patients.

Gallstones can be recovered from the stools of most patients with gallstone pancreatitis, emphasizing the importance of gallstones passing into the common bile duct and then obstructing or injuring the pancreatic duct orifice. Over 60 per cent of patients have stones impacted at the ampulla during the first 48 h of gallstone-induced pancreatitis. This observation has led some to advocate early common duct exploration (within 48 h) to remove impacted stones, with the intention of aborting progression from oedematous to necrotizing pancreatitis. In one series, removal of the impacted stone during the first 48 h of the attack by common duct exploration and, if necessary, transduodenal sphincteroplasty seemed to reduce mortality from 16 per cent to 2 per cent.

The concept that early stone removal is beneficial has several flaws. First, the differentiation of gallstone-induced pancreatitis from other forms of the disease is often difficult. Biochemical criteria such as elevated transaminases are helpful, but direct cholangiography is needed to be sure. Secondly, transhepatic cholangiography in patients with gallstone pancreatitis easily demonstrates the presence of common duct stones, but these are rarely impacted. Thirdly, many of the patients studied have undoubtedly had only chemical hyperamylasaemia induced by the obstructing stone, not true pancreatic inflammation. Fourthly, many investigators have found higher complication and mortality rates among patients subjected to early biliary surgery. Ranson operated on 22 patients with gallstone pancreatitis during the first week after the attack with a 23 per cent mortality rate (three-quarters of the patients operated on within 48 h died), whereas there were no deaths among 58 patients treated non-operatively until the pancreatitis subsided, with subsequent cholecystectomy and common duct exploration. Similarly, in a study of 172 patients Kelley found impacted stones in 63 per cent of patients operated upon within 72 h, but had a 12 per cent operative mortality rate in these patients. In contrast, if operation was delayed by 5 to 7 days, only 5 per cent still had impacted stones and the operative mortality was nil. In 15 per cent of his patients the pancreatitis did not subside but progressed and forced earlier operative treatment.

Transduodenal sphincteroplasty in the presence of acute pancreatitis carries an appreciable risk of abscess or duodenal fistula formation. Because 95 per cent of cases of gallstone pancreatitis subside with medical management and without progression to a fulminant form, and 95 per cent of stones pass spontaneously in the first week, surgical intervention to remove the stone within 48 h does not seem justifiable at present. Cholecystectomy and, if still necessary, common duct exploration, may be safely and effectively delayed until pancreatitis subsides, generally during the same hospital admission.

Endoscopic retrograde cholangiopancreatography may be carried out safely by experienced endoscopists in up to 90 per cent of patients with gallstone pancreatitis. Prophylactic antibiotics should be administered and injections into the pancreatic duct should be avoided. The earlier endoscopic retrograde cholangiopancreatography is performed, the more frequently impacted stones are found, an observation confirming the studies of Kelley and Acosta. When an impacted stone is found and endoscopic sphincterotomy with stone removal is successful rapid improvement usually ensues. In patients with an impacted stone and complicating sepsis, a situation that is rare, urgent decompression is likely to be beneficial.

If gallstone pancreatitis is triggered by an initial insult from passage or impaction of a stone, removal of the stone after the insult may or may not alter the subsequent course of events. If early endoscopic sphincterotomy is to have an impact, however, it should be performed within 48 h. The challenge is to identify early in the presentation those patients who are likely to benefit from endoscopic sphincterotomy.

The concept of an operation designed to drain the pancreatic bed and to reduce stimulation of the gland by placement of sump drains in the lesser sac, cholecystostomy tube or T-tube, gastrostomy, tube, and jejunostomy tube was popular in the 1970s. Review of the experience with this approach revealed that only patients who were judged to be dying in shock after 24 to 48 h of maximal supportive care seemed to benefit. The success of the operation in these patients was felt to be due to the removal of toxic ascites—a result which could also have been achieved with placement of percutaneous peritoneal lavage catheters.

Major distal pancreatic resection can be accomplished in the face of acute pancreatitis with a mortality of approximately 40 per cent, while pancreaticoduodenectomy or total pancreatectomy carries a mortality rate of 60 per cent or more: the key question is how to select those patients likely to benefit. The surgeon must decide which part of the pancreas to resect, and how much to resect. This decision can be very difficult, because surface changes may not represent the degree of central pancreatic injury and because several days are required for the changes of pancreatic devitalization to become visible. In the first few days there is only massive swelling, with or without haemorrhagic staining. CT scanning following intravenous contrast administration may demonstrate which areas of the pancreas are being perfused and which are not. This has the potential to guide early surgical debridement, but there have been no reports demonstrating improved survival in patients with pre-emptive early resection compared with those treated by later debridement of clearly demarcated necrotic tissue. Since the operative mortality of early pancreatic resection is so high, it is preferable to delay surgery until areas of necrosis are clearly demarcated, or there is proven bacterial infection.

Many of the systemic effects of pancreatitis are believed to be mediated by kinins and other vasoactive amines such as kallikrein and bradykinin which can be found in the dark brown exudate which accompanies fulminant pancreatitis. The effects of these metabolic substances are manifested by capillary leak, low peripheral vascular resistance, and hyperdynamic shock. Peritoneal lavage to remove this toxic ascites frequently leads to a rapid improvement in haemodynamic and respiratory function, whereas haemodialysis does not.

The precise role of peritoneal lavage is controversial. Peritoneal lavage is not a treatment of pancreatitis itself, but only reverses some of the early phase systemic effects which are mediated by circulatory toxins. It is therefore of no benefit in mild to moderate degrees of pancreatitis and it does not alter the progression of pancreatic injury or prevent the intermediate or late phase developments of pancreatic necrosis and abscess. Likewise, there is no benefit in treating the signs and symptoms that develop after the first few days because the cause of late inflammation is more likely to be necrosis, abscess, or pseudocyst.

Attempts have been made to select patients in whom lavage would be helpful. Several carefully performed studies based on Ranson's signs or similar systems, which stratify for severity and risk of dying, have demonstrated no benefit in any objective parameter. However Ranson's signs are only overall predictors of mortality. No study has yet been performed which addresses the use of peritoneal lavage in the treatment of early phase shock or even identifies a subset of patients in early phase shock. These are precisely the patients in whom striking immediate improvement is commonly seen. Many still feel, therefore, that peritoneal lavage is beneficial when there is early evidence of major plasma volume loss, hypotension, pulse below 140 beats/min, or continued clinical deterioration. Lavage should be instituted within 24 h of the onset of illness.

Lavage is performed via a percutaneously placed dialysis catheter and 1 or 2 l of dialysate are used for each lavage. The fluid need not equilibrate, as is the case in peritoneal dialysis, but can be evacuated immediately. The purpose is to wash out the ascitic fluid and its toxins, and the process need be continued only until the systemic effects are reversed. The response to lavage should be quite rapid, within several hours. If not, there may have been an error in diagnosis, or the patient may be suffering from gallstone pancreatitis complicated by cholangitis, or inaccessible loculation of toxic ascites in the lesser sac. The latter may be reached by percutaneous catheter placed under ultrasound or CT guidance. Recent anecdotal reports suggest that lavage of the lesser sac is more effective than general peritoneal lavage.

Irreversible tissue destruction becomes recognized after several days of acute pancreatitis. Phlegmon or swelling, due to oedema and inflammation, is apparent on ultrasound and CT in 30 to 50 per cent of patients and is palpable in 15 to 20 per cent. Intravenous contrast enhanced CT scans are sensitive for the early detection of ischaemic areas which subsequently become necrotic (Fig. 2) 1325. Several serum markers, including C-reactive protein and polymorphonuclear leucocyte-associated elastase, are associated with the development of necrosis and help determine which patients should undergo dynamic intravenous contrast enhanced CT scans. Serial CT scans remain the most accurate means for following the development of liquefaction necrosis. Liquefaction may occur in small well-defined patches or even in large segments such as the distal two-thirds of the gland, with extension of the necrotizing process into the retroperitoneum, perirenal spaces, and mesentery. This process is really regional necrosis, rather than just pancreatic necrosis. The combination of tissue ischaemia and release of activated enzyme into these areas can produce an ongoing necrotic process.

Three factors determine the ultimate outcome of necrosis: the amount, the extent of extrapancreatic necrosis, and, perhaps most important, bacterial contamination. Pancreatic necrosis is found to be infected in 40 per cent of patients. The incidence of infection is maximal during the third week of the disease process.

The appearance of infection, in general, correlates with higher mortality. In reports which differentiate between infected necrosis and pancreatic abscess, patients with infected necrosis require earlier surgery and have a higher mortality rate than those with pancreatic abscess. The combined effects of active pancreatitis and the infected necrosis force early surgical intervention in some of these patients. In contrast, patients whose active phase of pancreatitis is over, leaving them with necrotic tissue which was or would become infected, have a more indolent process which evolves over a longer period of time. When the indications for surgery finally become sufficient, the necrotic tissue has become fully liquefied—that is, an abscess has formed. Infected necrosis and pancreatic abscess are therefore not distinct entities, but are stages of a continuous spectrum of infectious complications.

Infection alone does not create the toxic state. Needle aspirations of pancreatic swellings show unexpectedly high rates of bacterial colonization, often with few or no clinical signs of infection. Furthermore, the haemodynamic consequences of pancreatic necrosis are virtually identical, whether it is infected or sterile. If infected necrosis can be indolent and sterile necrosis can produce severe haemodynamic consequences, the difference between the unstable patient with infected necrosis and the stable patient with a pancreatic abscess (and their correspondingly different mortality rates) must be due to another factor—probably the enzymatic and other biochemical events associated with ongoing acute pancreatitis.

Another area of controversy concerns how much necrosis can be safely observed. Operative mortality is rare in patients with sterile necrosis operated upon after the first week. However, postoperative complications such as fistula or abscess formation are frequent (13–24 per cent), leading some to question whether or not sterile necrosis should be operated on at all. If necrotic areas are small, sterile collections may resolve. When signs of inflammation are present, percutaneous aspiration should be undertaken to determine if infection is present: when bacteria are found in the aspirate, debridement and drainage are indicated. Larger necrotic areas are problematic because of concern that they will become infected before there is sufficient time for them to reabsorb and heal. The decision to operate on large collections is often made on the basis of clinical signs of inflammation; thus, the actual bacteriological status of the necrotic tissue does not necessarily alter the clinical decision. Sizeable collections which are not shown to be resolving on serial CT scans should be surgically debrided and drained irrespective of their bacteriological status (Fig. 3) 1326. Waiting for signs and proof of sepsis delays diagnosis and intervention, and may lower survival as complications increase.

Peripancreatic fluid collections occur in 10 to 20 per cent of patients with acute pancreatitis. Those persisting beyond the phase of acute inflammation become pancreatic pseudocysts. Pseudocysts occurring as part of the ongoing necrotizing process differ from those which are common in chronic pancreatitis. There is a 10 per cent death rate in patients with acute pseudocysts, due primarily to the severity of the underlying pancreatitis. Up to 15 per cent of patients with acute pseudocysts develop haemorrhage and sepsis while waiting for the cyst wall to become mature. Some acute pseudocysts may resolve spontaneously if the pancreatitis subsides, but those more than 6 cm in diameter which persist for 6 weeks after the onset of pancreatitis and those associated with ongoing pain and inflammation are unlikely to resolve and should be drained.

Drainage by percutaneous aspiration is associated with a very high recurrence rate, and insertion of an indwelling catheter is necessary. Percutaneous drainage is generally not as effective as operative drainage since the cavity contains large amounts of necrotic material which cannot be adequately removed through small percutaneous catheters. Furthermore, if there is a persistent communication with the pancreatic duct in the presence of a proximal obstruction, an external fistula will ensue. The fistula may ultimately require a pancreatic resection for correction, whereas an internal drainage procedure performed in lieu of the initial percutaneous drainage would have been a safer and more definitive approach. There may be a subset of patients with pseudocysts who can be successfully treated with percutaneous catheters, but the criteria for identifying these patients are not yet well established. Pseudocysts may also be internally drained into the stomach or duodenum using endoscopic techniques. The safety of endoscopic cystgastrostomy for acute pseudocysts has yet to be established and for large cysts containing substantial amounts of necrotic material the procedure may be hazardous. Surgical internal drainage remains the standard therapy. Pseudocysts may be drained into the stomach, duodenum, or small bowel (Roux-en-Y limb with minimal morbidity and mortality. The choice of procedure depends on the topography of the pseudocyst and the experience and preference of the surgeon.

Cystgastrostomy is rarely utilized in Europe but is often the fastest and easiest procedure. Except in very large pseudocysts (>15 cm diameter) cystgastrostomy is safe and effective. Giant pseudocysts and those that are not adherent to the stomach should be drained into a Roux-en-Y limb of jejunum. External drainage is necessary for 30 per cent of acute pseudocysts because the cyst wall is too friable to hold sutures. External drainage is frequently required when emergency pseudocyst drainage is performed for sepsis or haemorrhage.

Haemorrhage, a highly lethal complication, is caused by erosion of major blood vessels by elastase and other proteases. The initial lesion is a pseudoaneurysm in which infection is almost always present. If this ruptures, life threatening haemorrhage into a pseudocyst or retroperitoneum occurs. Angiography should be the management step: bleeding can often be fully controlled with angiographic techniques, allowing debridement and vessel ligation to take place under more favourable conditions. When bleeding originates from the main hepatic or splenic artery, angiographic control is difficult. Even if bleeding is not arrested, angiography provides a map that may assist the surgeon in planning the tactics of what is invariably a difficult operation. Although the effects on ultimate survival are difficult to establish without question, reduction in the rate of bleeding or complete cessation of bleeding saves blood, adds time for planning, and may reduce the expenditure of the patient's reserves. Surgery is inevitably required at some point for debridement of the associated regional necrosis and to control sepsis. If the affected area is inadequately drained or debrided, progression of sepsis, recurrence of bleeding, or both will inevitably follow, with fatal results.

Vascular thrombosis can involve the colic branches of the superior mesenteric artery, the splenic artery, or the gastroduodenal artery. Thrombosis manifests as gastrointestinal bleeding from sloughed mucosa, infarction with perforation, fistulae, or late stenosis and stricture. Colonic ischaemia may present with diarrhoea and endoscopic features of pseudomembranous colitis. If bowel infarction occurs it is necessary to exteriorize two stomas rather than risk breakdown of an anastomosis (Fig. 4) 1327. If the duodenum perforates, attempts at local repair are preferable to pancreaticoduodenectomy. Early channelling of any fistulae that develop is beneficial. When well drained, some fistulae close with healing of the underlying process. Gastric fistulae usually do not heal, and the rate of haemorrhage from the fistula tract is high. Biliary fistulae, particularly those developing after choledochotomy can be successfully treated by percutaneous or endoscopic placement of a biliary endoprosthesis.

Gastric outlet obstruction is generally due to duodenal compression and atony or localized ileus from the nearby phlegmon. Gastroenterostomy is occasionally necessary as gastric emptying may be impaired for long periods of time. A gastrostomy tube is helpful as one waits for resolution of the obstruction.

Partial common bile duct obstruction due to compression of the intrapancreatic portion of the common bile duct is common. The serum bilirubin may rise to 3 or 4 mg/dl. Operative decompression is almost never necessary unless common duct stones and cholangitis are present. The common bile duct always returns to its normal configuration when the pancreatitis subsides.

Abscesses, which are the most common cause of death in acute pancreatitis, occur in approximately 2.5 per cent of all patients and are related to the severity of the attack. Abscesses are the most likely complication after the second week and arise from secondary infection of necrotic and liquefied pancreas. Abscesses usually contain enteric bacteria, but if broad spectrum antibiotics have been used, Candida species may become a major pathogen. Multiple organisms are cultured from nearly half of all pancreatic abscesses. Patients requiring peritoneal lavage in the early phase of their disease seem to have an increased incidence of abscess formation. However, it is probable that the need for early lavage is simply an indication of basic severity of the pancreatitis, rather than the portal by which infection entered.

The best method for diagnosing pancreatic abscesses is CT scanning. Although it is possible to predict which patients are statistically more likely to form abscesses by the use of early CT scan and a clinical severity score, patients must be evaluated individually. Early CT scan of the pancreas is recommend in patients with moderate or severe pancreatitis. Those with persistent clinical evidence of pancreatic inflammation should undergo CT evaluation every week to enable change in the retroperitoneal anatomy to be monitored. Fever and leucocytosis are not always present with infection and do not reliably differentiate infected from sterile collections. Unless extraintestinal gas is present, the CT scan often cannot distinguish sterile from infected collections, and needle aspiration may be quite helpful. Infections presenting early (in the first 2 to 3 weeks), especially in the patient who has never become free of inflammatory signs, are likely to be associated with ongoing pancreatitis. Abscesses manifesting late (3 weeks or later) in the disease tend to be more bland collections of pus and to behave similarly to other types of intra-abdominal abscesses. Adequate drainage is nonetheless essential; without it the mortality rate is 100 per cent. Antibiotics alone have never cured an abscess and, while limiting the bacteraemia, may select for late fungal superinfection.

Recently there has been guarded enthusiasm for percutaneous drainage of pancreatic abscesses with some authors claiming a success rate of up to 70 per cent. Others have not been as successful. The major shortcoming of percutaneous drainage is the inability to debride thick necrotic material and thus open all the extensions and loculations of what is usually a labyrinthine cavity. Abscesses presenting late in the course of pancreatitis and those developing after initial surgical drainage are most amenable to percutaneous catheter drainage. Most, however, require a combination of surgical and radiological drainage. Although the fluid component of the abscess is often adequately drained percutaneously, surgery is required to remove pieces of necrotic pancreatic debris (Fig. 5) 1328. Patients in whom the abscess can be successfully drained percutaneously often require multiple catheter insertions, multiple catheter manipulations, and long-term catheter drainage.

In recent experience, with the use of aggressive surgical debridement, the mortality rate has been reduced to 5 per cent. Late endocrine and exocrine insufficiency are rare, probably because much of the tissue removed is peripancreatic tissue. Reoperation or subsequent percutaneous radiologically guided drainage is required by 20 per cent of patients. Complications occur frequently after abscess drainage and include fistulae, new abscesses, haemorrhage, renal failure, and wound infection; however most of the patients with these complications do not require further surgery.

A few patients continue to have low grade signs of pancreatic inflammation for many weeks or even several months, without focal collections or areas of necrosis demonstrable by CT scan to target for debridement or drainage. Endoscopic retrograde cholangiopancreatography may identify irreversible injury to the pancreatic duct or underlying anomalies that do not allow the pancreatitis to subside. In other patients, there may be microabscesses or unrecognized duodenal wall injury. Resection of the pertinent area, even if it requires pancreaticoduodenectomy, however radical that may seem, may be the only option left. Distal pancreatectomy is indicated when the pancreatic duct becomes obstructed by the necrotizing process and its healing by scar.

Acosta JM, Pellegrini CA, Skinner DB. Etiology and pathogenesis of acute biliary pancreatitis. Surgery 1980; 88: 118–25.

Beger HG, Bittner R, Block S, Buchler M. Bacterial contamination of pancreatic necrosis: prospective clinical study. Gastroenterology 1986; 91: 433–8.

Cobo JC, Abraham E, Bland RD, Shoemaker WC. Sequential hemodynamic and oxygen transport abnormalities in patients with acute pancreatitis. Surgery 1984; 95: 324–30.

Gerzof SG, et al. Early diagnosis of pancreatic infection by CT-guided aspiration. Gastroenterology 1987; 93: 1315–20.

Ihse I, Evander A, Holmberg JT, Gustafson I. Influence of peritoneal lavage on objective prognostic signs in acute pancreatitis. Ann Surg 1986; 204: 122–7.

Kelley TR. Gallstone pancreatitis: the timing of surgery. Surgery 1980; 88: 345–50.

McMahon MJ, Playforth MJ, Pickford IR. A comparative study of methods for the prediction of severity of attacks of acute pancreatitis. Br J Surg 1980; 67: 22–5.

Ranson JHC. The timing of biliary surgery in acute pancreatitis. Ann Surg 1979; 189: 654–61.

Ranson JHC, et al. Prognostic signs and operative management in acute pancreatitis. Surg Gynecol Obstet 1974; 139: 69–81.

Warshaw AL, Jin, G. Improved survival in 45 patients with pancreatic abscess. Ann Surg 1985; 202: 408–15.

Warshaw AL, Rattner DW. The timing of surgical drainage for pancreatic pseudocysts: clinical and chemical criteria. Ann Surg 1985; 202: 720–4.