Chapter 85. Acute Pancreatitis in the Critically Ill

• Hypoperfusion injury is the most important of many possible causes of pancreatitis in the ICU setting.
• Serial computed tomography is most useful in confirming the diagnosis and in following the inflammatory process.
• The likelihood of multisystem failure is high, and preventive measures must be instituted early.
• Initial treatment is supportive, with aggressive fluid and electrolyte replacement and close monitoring of hemodynamic, pulmonary, and renal status.
• Surgical intervention is indicated for deteriorating patients with a surgically correctable lesion and in patients with complications of the disease, such as major hemorrhage, abscess, or symptomatic pseudocyst.
• Close supervision by an experienced abdominal surgeon in collaboration with an intensivist is important in recognizing the indications for surgical exploration.

Acute pancreatitis can be a frustrating disease to diagnose and treat. Causes are myriad and poorly understood, pathologic variation is great, diagnosis is often difficult, and treatment in most cases is nonspecific and supportive only. Pancreatitis usually presents as a self-limiting disease from which the patient recovers without complication or intervention. In its most severe form (approximately 10% of cases), it is a potentially lethal disease that is complicated by multisystem failure and sepsis, often requires sophisticated intensive care and judiciously timed surgical débridement, and has a mortality rate of up to 30% to 40%.1–4 Most intensivists therefore are quite familiar with the patient who presents with acute necrotizing pancreatitis and who may occupy a bed for many months, often ultimately dying of septic complications5 that are generalized or are related to the secondary infection of pancreatic necrosis.6

The subject of this chapter is the increasingly recognized entity of acute pancreatitis in the patient who is already critically ill from other causes. Discussion will focus on the associated diseases reported to date, the pathogenesis of this unusual form, its recognition, and its management.

Many causes of acute pancreatitis have been recognized, the most common being alcohol and gallstones. Table 85-1 lists the reported causes of acute pancreatitis.

Whatever the initiating cause, the pathogenetic mechanism common to all forms of acute pancreatitis seems to be an intense systemic inflammatory response syndrome (SIRS) caused by the release of activated pancreatic enzymes7,8 and mediated by cytokines, immunocytes, and the complement system. Inflammatory cytokines (such as tumor necrosis factor) cause macrophages to migrate into tissues distant from the pancreas, including lungs and kidneys. Immunocytes attracted by cytokines released from macrophages release more cytokines, free radicals, and nitric oxide9,10; the result is tissue destruction, fluid and electrolyte loss, hypotension, renal and pulmonary complications, late septic complications, and in 20% to 40% of severe cases, multisystem organ failure (MSOF) and death.11

The list of illnesses in the critically ill patient that may precede the development of pancreatitis in the ICU is growing (Table 85-2). Why are these seemingly disparate groups of patients all at risk for this apparently unrelated and possibly lethal condition? The answer is not known and may be multifactorial.

In the case of pancreatitis developing after cardiopulmonary bypass, important factors may be microembolism and venous thrombosis,12,13 a low postoperative flow state, prolonged bypass,14 intraoperative or postoperative bleeding, hypothermia,15 and the use of both narcotics and vasopressors.16 The frequent occurrence of pancreatitis after cardiac transplantation may be related to the effect of heparin on platelet aggregation17 and to the administration of cyclosporine18 and prednisone,19 as well as to the factors suggested for cardiopulmonary bypass. Abdominal aortic aneurysm repair, especially after acute rupture, may produce hypotension, microemboli, and vasoconstriction, resulting in prolonged hypoperfusion of the pancreas.20

Pancreatitis developing after renal transplantation may be related to hypercalcemia,21 specific operative factors, or immunosuppression as for cardiac transplantation. The reasons for severe pancreatitis after liver transplantation may include direct dissection in some cases, blood loss and hypovolemia, or obstruction to pancreatic venous outflow during prolonged venous bypass. One group has suggested that a hepatitis B infectious factor may be involved.22

Review of these possible mechanisms for the development of pancreatitis in the critically ill patient seems to confirm hypoperfusion injury as a major pathogenetic mechanism. In fact, one group has suggested that a patient in shock from any of a variety of causes (oligemic, cardiogenic, or septic) is at risk of developing pancreatitis.23 The actual mechanism of this injury, resulting in the release of activated enzymes and an inflammatory response, is not known.

Hypoperfusion may play an important role in the conversion of a less severe, self-limiting pancreatitis into the more lethal necrotizing form, even when the more common etiologic factors, such as alcohol and gallstones, are major contributors in the non-critical-care setting.24,25

Many major upper abdominal procedures have been reported to be complicated by pancreatitis, usually as a result of direct dissection or injury of the gland at surgery (see Table 85-2). However, since any patient subjected to a period of hypotension intraoperatively is a candidate for hyperamylasemia, pancreatitis, or both, the cause of pancreatitis in these cases may have been multifactorial rather than purely mechanical, as originally thought.

Whatever the cause of new pancreatitis in an ICU patient, the physician must diagnose the condition early, attempt to assess the severity of the process, institute appropriate aggressive supportive management, and use judgment in deciding on the timing and type of surgical intervention, if necessary. Most patients improve even if the disease is virtually ignored; those few whose pancreas becomes necrotic and infected need aggressive and timely surgery to avoid MSOF and death.

Clinical signs and symptoms are notoriously variable in patients with pancreatitis,20 and they are even more difficult to evaluate in the critically ill. The patient may complain of abdominal pain, nausea, and vomiting; often there are nonspecific features such as elevated temperature, leukocytosis, or in more severe cases, unexplained hemodynamic instability and increased fluid requirements without major clinical findings. The serum amylase concentration,26–28 random urinary amylase concentrations, and the amylase-creatinine clearance ratio may be useful indicators,29,30 but they are not reliable in all patients. The serum lipase concentration31 may be a more reliable and specific indicator than the serum amylase level, but its use has not gained wide acceptance. Serum lipase rises somewhat more slowly than amylase, peaks at 24 hours, and normalizes in 1 to 2 weeks.32 The concentrations of isoamylase,33 serum elastase, phospholipase A2, and trypsin have been suggested as tests but have not been practically useful.34,35

In addition to clinical evaluation and frequently unreliable biochemical tests, morphologic assessment of the retroperitoneum is necessary. For patients presenting initially with suspected pancreatitis, standard abdominal and chest films usually were obtained routinely to rule out other significant catastrophes, such as perforation and mechanical intestinal obstruction. These films also may suggest pancreatitis by demonstrating pleural effusions, localized jejunal or colonic ileus, the outline of a widened duodenal C loop, pancreatic calcification, associated radiopaque gallstones, or obliteration of psoas shadows and free intraperitoneal fluid. In some cases,36 laparoscopy may be useful in making a diagnosis of pancreatitis, although early intervention of any kind is to be avoided unless absolutely necessary. However, the increased availability and efficacy of ultrasound and computed tomographic (CT) scanning have made them the mainstays of routine investigation and monitoring of retroperitoneal inflammation.

Ultrasound

Ultrasonography is perhaps the most available test and can be carried out at the bedside of an acutely ill patient. It remains the most convenient modality for edematous pancreatitis that responds rapidly to conservative therapy, for suspected biliary pancreatitis with a mild clinical course, and for the monitoring of a retroperitoneal collection or phlegmon for resolution or the development of a pseudocyst.37 It also has been suggested that Doppler ultrasound may be valuable in detecting pseudoaneurysms 0.1 cm in diameter. However, paralytic ileus and obesity together severely hamper the ultrasonographer's ability to evaluate the retroperitoneum in the acute stages.

The current generation of high-resolution real-time scanners makes it possible to examine in detail the pancreatic gland, its peripancreatic compartments, and the associated biliary tract, with the patient in the semierect, sitting, and supine positions. These maneuvers, unfortunately, may be difficult to perform in a critically ill patient.

The abnormalities detected by ultrasonography are variable. The gland itself may be diffusely enlarged and hypoechoic or have a normal appearance, depending on a number of factors, such as the timing of the study, the presence of underlying chronic disease, and the amount of intrapancreatic fat or hemorrhage. Ductal diameter is similarly variable and may be increased, normal, or decreased. The ultrasonographer is just as interested in the peripancreatic retroperitoneal spaces, looking for evidence of acute fluid collections in the lesser sac, phlegmon, and hemorrhage. It is also possible to detect displacement of the gastrointestinal tract and the presence of intraperitoneal fluid. Fluid collections in the lesser sac usually regress as the clinical picture improves. Those which do not and are surrounded by a dense pseudocapsule legitimately may be termed chronic pseudocysts and often are treated by radiologic or surgical intervention.

Critical in the assessment of the retroperitoneum is the detection of scattered pancreatic necrosis and of the involvement of peripancreatic tissues with fat necrosis, hemorrhage, and ultimately, abscess formation; unfortunately, ultrasonography is limited in its ability to define these subtle changes in tissue density.37

Computed Tomography

Computed tomography has become the “gold standard” for assessing retroperitoneal morphology, especially in complicated cases, and now is increasingly available in every clinical setting. The procedure in the acutely ill patient may be somewhat inconvenient and cumbersome because of endotracheal intubation, multiple lines, and hemodynamic instability, but these problems are not insurmountable. Images of the retroperitoneum are not compromised by obesity or bowel gas, as they may be with ultrasound.

A mildly inflamed pancreas may show slight diffuse enlargement with irregular borders and patchy densities. Peripancreatic fat may increase in density as it undergoes necrosis. The more severe forms of pancreatitis are accompanied by an increase in peripancreatic inflammatory exudate and intrapancreatic fluid collections. Often the pancreas is markedly enlarged, with irregular fluid collections in the retroperitoneum and poor enhancement in scattered areas within the gland. Unlike ultrasonography, computed tomography is an excellent method for identifying high-density fluid collections in the transverse mesocolon, small bowel mesentery, and pararenal and perirenal spaces38; by means of dynamic CT scanning, areas of necrosis in the gland itself may be demonstrated. In sick patients, however, one always must be aware of the potential negative effects of contrast infusion both on the patient and on the pancreatitis.39

The critical value of CT scanning is in the detection and follow-up of significant complications. A pancreatic abscess consists of pancreatic necrosis that becomes infected and liquefied; it appears on CT scans as ill-defined or poorly encapsulated fluid collections of different densities, with air bubbles being recognized in about 20% of cases. Pseudoaneurysms also may be detected by CT scanning and, with enhancement, may be highlighted.

Since infusion during CT scanning may present significant risks for the hypovolemic unstable patient, future applications of magnetic resonance imaging (MRI) must be considered in this situation. Currently, however, CT scanning remains the most useful tool for evaluation of retroperitoneal morphology.

Severity of Disease

The intensivist must identify not only the presence of the new condition but also its severity and the likelihood of progression to necrosis and complications. Staging of the acute disease is useful in selecting certain patients for early aggressive supportive therapy, those who may require immediate referral to a specialized center, and those who may be candidates for entry into clinical trials. Ranson's criteria40 were used extensively in the past, but currently, a number of prognostic scoring systems with clinical, biochemical, and radiologic features are available. The Balthazar score is based primarily on CT findings, focusing on the presence or absence of pancreatic necrosis. It has been suggested that CT scores are more valuable than the Acute Physiologic and Chronic Health Evaluation (APACHE) II score in predicting regional complications associated with pancreatitis.41 Many use the Atlanta classification, which defines severe acute pancreatitis using features such as clinical presentation, the Ranson criteria, the APACHE II criteria, and evidence of organ failure and anatomic changes. These indicators are directed toward assessing the systemic effects of the disease, and in a critically ill patient, a number of them already may be present before the onset of the pancreatitis.

Numerous blood tests have been suggested as indicators of severity (areas of necrosis within the gland), including the levels of cyclic adenosine monophosphate, albumin, methemalbumin,42 ribonuclease,43 C-reactive protein,44,45 fibrinogen,34 and urinary trypsinogen-activated polypeptide.46 Another group has reported that the volume and color of peritoneal aspirate at diagnosis can be useful.47 None of these tests has proved as accurate, however, as clinical assessment of the patient and serial CT scans of the abdomen, especially in the ICU setting.48,49 The development of abdominal pain and tenderness, fever, leukocytosis, hemodynamic instability, and CT changes within 1 to 2 weeks of the presumed onset of the pancreatitis provides the best assessment of the severe complications of the disease.

In a patient who is deteriorating or is persistently septic and whose CT scan demonstrates features typical of pancreatic necrosis as listed earlier, percutaneous aspiration of the retroperitoneum with microscopic examination and cultures can be valuable50–52; this approach is used frequently in our unit to direct further management.

After the patient has been stabilized systemically and the disease has been identified as severe, attention is turned initially to the possible etiologic factors listed in Table 85-1. An ICU patient may develop pancreatitis related to causes other than the described hypoperfusion injury; drugs, gallstones, or other obstructive mechanical factors may be critical. Very frequently, as is the case with gallstone pancreatitis, the disease is short and self-limiting, and in selected patients it can be treated with immediate or delayed endoscopic or surgical intervention as the case demands. The most common scenario is quick resolution of the acute inflammation and a recommendation either for surgery in the same admission if the gallbladder is in situ53 or for endoscopic sphincterotomy and gallstone removal if the gallbladder was removed previously. In the critically ill person, however, this intervention, unless indicated early by persistent jaundice and acute pancreatitis, should be delayed until after resolution of the primary illness. Some groups, however, advocate routine early endoscopic sphincterotomy and claim modest attendant morbidity.54–57

Assuming that the disease in the ICU patient was initiated by a hypoperfusion injury to the gland, however, and that no immediate “corrective” intervention is appropriate, the principles of treatment include careful monitoring, correction of the metabolic and hemodynamic effects of the inflammatory process, respiratory and renal support, nutrition, control of pancreatic enzyme secretion and its sequelae, and prevention and treatment of the local complications of necrotizing pancreatitis. The evaluation and management of such a patient from the outset must be a collaborative effort between the intensivist and an abdominal surgeon experienced in the medical and surgical measures that may become necessary. The decision making with regard to surgery is based to some extent on hard data that can be measured, such as biochemical indicators and radiographs, but it is also based on clinical experience and intuition.58,59

Monitoring

The critically ill patient who develops pancreatitis is likely already undergoing hemodynamic monitoring in the ICU setting, with close evaluation of heart rate, arterial blood pressure, hourly urine output, and central venous pressure. Since patients with severe pancreatitis may well have major fluid derangement, renal failure, sepsis, and especially pulmonary complications, a pulmonary artery catheter may be useful for monitoring, but the impact of this device remains controversial.

Frequent determinations of blood counts, blood urea nitrogen (BUN), blood sugar, electrolytes, creatinine, calcium, magnesium, and arterial blood gas levels are useful in detecting deteriorating pulmonary or renal status, the adequacy of treatment, and the evolution of septic complications. In a patient who is slow to respond, serial ultrasound examinations and CT scans especially are critical to the ongoing evaluation of the retroperitoneum.

Metabolic and Hemodynamic Effects

The need for volume replacement in patients with severe pancreatitis cannot be overemphasized. Such patients, in effect, have a massive retroperitoneal burn, with ongoing third-space losses in the retroperitoneum, peritoneal cavity, and bowel lumen. Requirements may vary from small amounts to 8 to 10 L of isotonic fluid in the first 24-hour period. In the ICU, especially if the patient has a preexisting brittle cardiovascular status, fluid replacement and its effects may have to be followed even more closely; however, a patient who develops pancreatitis after a cardiopulmonary bypass or cardiac transplantation still requires adequate salt and fluid replacement. Patients who develop hemodynamic instability after appropriate massive fluid administration may benefit from support with vasoactive drugs.

Because hemorrhage can occur as a complication of the retroperitoneal extravasation of proteolytic enzymes, administration of blood may be necessary; for those who develop hypoalbuminemia as a result of significant protein loss, plasma or albumin or both may be indicated.

It should be stressed that the reason why one patient progresses to pancreatic necrosis and/or the development of SIRS and another has a benign, self-limiting course remains a mystery; however, in the usual case of pancreatitis, and especially in a critically ill patient in the ICU, this deterioration may well be exacerbated by hypotension and hypoperfusion related to inadequate attention to volume replacement.

Metabolic abnormalities should be recognized and treated. For example, early fluid management and stabilization usually are sufficient to correct metabolic acidosis. Whether hypocalcemia—even ionized hypocalcemia—should be corrected is controversial, given the central role of elevated intracellular calcium in the pathogenesis of pancreatitis.60

Respiratory and Renal Support

Many mechanisms are implicated in the development of typical adult respiratory distress syndrome (ARDS) in this setting. After major heart or transplantation surgery, the patient's pulmonary function already may be compromised by anesthetics, opiates, and pain related to the procedure. In addition, elevation of the diaphragm, pleural effusion, abdominal pain leading to further splinting and atelectasis, and overaggressive fluid replacement may contribute to the respiratory failure. The most important factors may occur in the lung itself, with increased capillary permeability related to serotonin release, breakdown of surfactant by the enzyme phospholipase A released by the necrosing pancreas, or activation of the complement cascade with elaboration of vasoactive prostaglandins and leukotrienes. Whatever the mechanisms, aggressive ventilatory support is required frequently, and this possible need should be anticipated. If the patient was ill previously and develops respiratory insufficiency along with acute pancreatitis or requires multiple operations for complications of the latter, ventilatory support may be prolonged for many weeks. Tracheostomy may become necessary for adequate tracheobronchial toilet or occasionally for protracted ventilation.

No specific renal insult has been recognized in the patient with acute pancreatitis, apart from the secondary effects of volume loss and hypotension. However, as reviewed previously, many patients with ICU-related pancreatitis may have a hypoperfusion injury as a cause; therefore, renal dysfunction may be anticipated in many. Aggressive support with fluids, judicious use of diuretics when fluid has been replaced adequately, and hemodialysis in selected cases of acute tubular necrosis are all part of the therapeutic armamentarium.

One group examining the incidence of pancreatitis in patients undergoing abdominal aortic aneurysm repair found that there was a high correlation of pancreatic injury with ischemic injury to the kidney23 (acute tubular necrosis), suggesting that the two conditions have similar etiologies and indicating the need for close monitoring of renal function.

Nutrition

Nutritional support for the seriously ill patient is critical, even though no specific benefit to the patient with an inflamed pancreas has been proven. Most of the conditions listed in Table 85-2 occur in patients at risk for developing pancreatitis; nevertheless, these patients enter the ICU with an expectation of returning to normal feeding within days. The recognition of pancreatitis, however, should carry with it the realization that a prolonged stay is possible. All too often the intensivist withholds total parenteral or enteral nutrition (TPN or EN) until the course of the disease is declared, only to find that after 2 weeks the patient is not improving significantly and now has been starved unnecessarily.

The patient who develops pancreatitis in this setting should be regarded with suspicion, and nutrition should be offered early with an objective of “pancreatic rest.” Long-term TPN in a sick, possibly septic patient carries its own frequent complications, among which are line sepsis and opportunistic infections; enteral feeding through a nasojejunal or percutaneous jejunal tube is a valuable adjunct and should, whenever possible, be the feeding method of choice if severe paralytic ileus does not preclude its use.61–63 Infusion of nutrients into the proximal jejunum bypasses the stimulatory effect of feeding on pancreatic secretion, has fewer complications, is much less costly than TPN, helps maintain mucosal function, and limits absorption of endotoxins and cytokines from the gut.64–67 The feeding tubes also can be placed via laparoscopy or combined endoscopic and radiologic means.68

Control of Pancreatic Enzyme Secretion

Many therapeutic steps have been taken to minimize pancreatic enzyme secretion or limit its effects. Studies to examine the possible benefits in acute pancreatitis of nasogastric suction,69 anticholinergic agents,70 H2-receptor blockers,71 somatostatin,72,73calcitonin,74 and glucagon75 have not supported the routine use of these agents.

Similarly, the effects of the proteolytic enzyme inhibitor aprotinin (trasylol),76,77 the antitrypsin agent ε-aminocaproic acid (EACA),78 and a phospholipase A inhibitor79 have not been convincingly beneficial. Nasogastric suction is still used routinely, especially in the obtunded patient, to decompress paralytic ileus, control vomiting, perhaps decrease stimulation of the pancreas, and prevent aspiration.

The numerous modalities used and tested in the treatment of the severe form of this disease attest to the failure of most of them to alter the course of the inflammatory process. The mainstay of initial treatment is resuscitation with fluid and electrolytes combined with respiratory and nutritional support, as well as attention to the primary disease state in the case of a critically ill patient. Patients who die from necrotizing pancreatitis, however, do not usually die in the early resuscitation phase, or even in the first 2 weeks, owing to excellent intensive care.

The next step is to evaluate and follow the pathologic process in the retroperitoneum while all supportive measures are continued.

At the outset, the clinician must be aware of the pathologic conditions that occur secondarily as a result of acute pancreatitis. A relatively uncommon but potentially lethal complication is that of intraabdominal hemorrhage requiring immediate intervention.80 The local complications of necrotizing pancreatitis are pancreatic necrosis, pancreatic abscess, and pseudocyst formation81; the process also may have more remote effects, such as formation of multiple intraabdominal collections, colonic perforation,82 and even gallbladder necrosis83 and abdominal compartment syndrome.84 Aggressive early hemodynamic stabilization is probably the only rational way to prevent further hypoperfusion injury to the pancreas and the development of these complications. Institution of peritoneal dialysis via the percutaneous or minilaparotomy route accelerates early improvement in some patients but does not appear to offer any benefit in decreasing the likelihood of complications and death.85 Patients in whom retroperitoneal necrosis (fat or pancreatic) develops are at high risk for morbidity and death.6 Those in whom this complication does not develop usually will survive with appropriate supportive measures only.

Pancreatic and Peripancreatic Necrosis

Pancreatic necrosis consists of patchy devitalization of the pancreatic gland; it develops in a few days to weeks after the onset of the inflammatory process. The patient usually has abdominal pain, persistent fever (possibly low grade), and leukocytosis. CT scan with vascular enhancement may demonstrate local or diffuse areas of nonenhancement. Needle aspirates obtained at this stage will show no growth on culture if necrosis is present without infection.

Treatment of this entity is controversial. It is difficult to predict how many cases of pancreatic necrosis (without infection) will resolve spontaneously with simple supportive management and how many will develop sepsis and clearly require débridement. Undoubtedly, the condition does resolve in some patients, who clearly would not benefit from laparotomy. Every pancreatic surgeon has had the experience of extensively débriding sterile retroperitoneal necrosis, subjecting the patient to perhaps many future débridements (which then, of course, involve infected tissue), and wondering whether the original noninfected condition might have resolved spontaneously. In addition, many “sequestrectomies” performed for pancreatic necrosis likely are excisions only of necrotic fat and not of necrotic gland. No comparison of these two pathologic entities in terms of morbidity and mortality has been reported. A retroperitoneum with a well-perfused pancreas surrounded by necrotic fat may carry a different outlook than if the gland itself is compromised. Therefore, removal of uninfected necrotic fat only may be a fruitless and conceivably dangerous exercise.

It has been suggested that infection develops in only 40% of patients with pancreatic and peripancreatic necrosis.86 Because of this and the aforementioned fear that surgery in the sterile situation actually may cause more harm than good, we tend to avoid major débridement unless sepsis is confirmed. Therefore, if the patient remains stable hemodynamically, has only low-grade fever and leukocytosis, has CT findings that do not suggest extensive glandular necrosis and infection, and has a sterile percutaneous retroperitoneal sample, we continue with supportive management without intervention. However, the patient ultimately may require intervention even in the absence of gross septic parameters simply because of a prolonged “smoldering” course showing no clinical improvement (Fig. 85-1).

Antibiotic Therapy

The role of antibiotics in the patient with pancreatic necrosis or sepsis is similarly controversial.87–91 Antibiotics are required in patients who have pancreatic necrosis (with or without sepsis), but their purpose is to provide systemic protection rather than local treatment or prophylaxis. Patients with sterile necrosis usually receive broad-spectrum antibiotics for systemic protection while the intensivist monitors the retroperitoneum, although this treatment undoubtedly brings with it a risk of opportunistic infection, such as disseminated candidiasis in the acutely ill host. In recent years, the case for routine antibiotic usage in such patients has strengthened. Certain antibiotics such as imipenem actually penetrate the pancreatic parenchyma, achieving good tissue levels during acute pancreatitis. In several studies, significant reduction in the incidence of septic complications followed the routine use of broad-spectrum antibiotics, but the effect on mortality has been less predictable. One study92 showed that routine use of imipenem-cilastatin altered the bacteriology in severe acute pancreatitis from predominately gram-negative coliforms to predominately gram-positive organisms without altering the incidence of resistance or fungal superinfection.

Prophylactic antibiotics for the purpose of selective decontamination of the gastrointestinal tract may prevent translocation of bacteria to the inflamed pancreatic bed, presumably from the transverse colon.93–96 Patients who develop infected necrosis must be treated actively to prevent or manage bacteremia and frequently require surgical débridement and drainage; however, even that approach is being challenged by some anecdotal evidence that antibiotics alone or in conjunction with percutaneous drainage without surgery may be sufficient to treat some patients with aspiration-proven retroperitoneal sepsis.97–101 This strategy still may be difficult to accept for those intensivists who have treated patients who developed fatal opportunistic infections in the past after long-term antibiotic therapy.

Infected Necrosis and Pancreatic Abscess

Some patients with pancreatic and peripancreatic necrosis develop spontaneous infection of the necrotic tissue presumably because of direct contamination by bacteria from the transverse colon. Although some authors do not distinguish between infected necrosis and pancreatic abscess, others differentiate them on the basis of suitability for percutaneous drainage: Infected pancreatic necrosis usually requires extensive surgical débridement and possibly pancreatic resection because of solid necrotic areas, whereas a pancreatic abscess may require just percutaneous drainage if liquefaction has taken place, leaving little solid necrotic tissue.

Notwithstanding the more conservative approach described earlier in selected patients, established sepsis in the retroperitoneum generally requires intervention to avoid the sequelae of systemic infection. This fact has significant implications in the patient who is critically ill from a primary illness. The diagnosis of retroperitoneal sepsis, whether infected pancreatic necrosis or pancreatic abscess, is made by recognition of clinical deterioration, which may be more precipitous than with pancreatic necrosis and may be accompanied by more pronounced fever and leukocytosis, typical CT findings, and a positive percutaneous aspirate.59,102 When the patient is maintained on broad-spectrum antibiotics (with perhaps decreased immunosuppression in the transplant patient), preparations are made for surgical débridement. Although some cases may be amenable to percutaneous drainage only, as stated previously, in our experience, most require laparotomy.

Principles of Surgical DéBridement

The conventional indications for surgery in acute pancreatitis include failure of early medical management, unresolving gallstone pancreatitis, and pancreatic sepsis. In the critical care setting, aggressive supportive care has virtually eliminated the need for early operation. As stated earlier, a persistent gallstone pancreatitis may be treated effectively by early endoscopic sphincterotomy. Therefore, the main indication for surgery in the critically ill patient with pancreatitis is the late development of necrosis or sepsis.58,59

Recommended methods of surgical débridement vary considerably, and their description is beyond the scope of this chapter. The principles of the different approaches are similar, however; the critical steps are the recognition of the patient in whom débridement is appropriate and quick execution of the procedure, with excellent anesthesia and postoperative intensive care.

The timing of surgery for pancreatic sepsis is critical,103 in that early débridement (done usually within 3 weeks of the onset of clinical pancreatitis) may encounter poor demarcation between necrotic and healthy tissue, leading to oozing from partially devitalized tissue after necrosectomy. If a wait can be tolerated by the patient, a late laparotomy (at 3 to 6 weeks) is likely to involve the removal of well-demarcated necrosis (often in a single procedure rather than multiple operations).

Surgery should consist of wide débridement of all devitalized peripancreatic fat and pancreatic tissue, followed by adequate dependent drainage, which should be generous enough to allow drainage of thick, purulent, often particulate material.104,105 Since multiple operations for repeated abdominal toilet sometimes are required, some surgeons have advocated open packing of the abdomen in such cases.102,106

Our own preference is for radical excision of all necrotic and infected material followed by generous dependent sump drainage and closure of the abdomen, although some practitioners advocate closed lavage postoperatively.107 Intubation of the gastrointestinal tract (cholecystostomy, gastrostomy, and jejunostomy were advocated enthusiastically in the past108) is to be avoided in this septic setting. Temporary packing of the retroperitoneum and scheduled reoperation occasionally are necessary to control venous oozing from the bed of the necrotic tissue.

The extent of pancreatic resection varies depending on the degree of necrosis of the gland itself. Although some surgeons have recommended routine performance of major pancreatic resection,109,110 we débride the gland itself only if necessary; only necrotic portions are removed, and major anatomic dissections are avoided, if possible, in these often unstable patients.

The patient who suffers infected pancreatic necrosis and who requires intensive care, often multiple operations, and an extended hospital stay clearly will have a protracted convalescence. However, a recent quality-of-life study111 found that the survivors of this morbid disease could look forward to good quality of life.

Pancreatic Pseudocyst

With more frequent use of ultrasound and CT scanning in the early stages of pancreatitis, recognition of fluid collections in and around the lesser sac has become the rule rather than the exception. Most collections probably consist of inflammatory exudate, and in fact, most resolve as the pancreatitis settles (Fig. 85-2). The persistence of a lesser sac fluid collection (or of fluid in a variety of areas, such as the peritoneal cavity, the retroperitoneum, or even the mediastinum) implies a communication with a ruptured pancreatic duct system. A collection that develops thick, fibrous walls consisting of the surrounding stomach, colon, and mesentery properly can be termed a chronic pseudocyst. Diagnosis of such a condition does not mandate immediate treatment unless the pseudocyst becomes symptomatic, grows quickly, becomes infected, bleeds, or ruptures into the peritoneal cavity or adjacent bowel. In an already sick patient, a quickly enlarging or infected pseudocyst is best treated with percutaneous decompression or endoscopic pseudocyst drainage, whereas immediate surgery is required for bleeding or free rupture. Occasionally, transpapillary stent placement is useful if a direct communication exists between the main pancreatic duct and the pseudocyst.112 The presence of a large collection does not of itself necessitate percutaneous decompression; in fact, one group found that collections that were drained early had a significantly increased chance of later developing septic complications requiring surgery and a longer hospital stay.113 It also has been suggested that percutaneous treatment is effective for peripheral collections, whereas more centrally placed collections need operative decompression.114 Spontaneous rupture into the intestinal tract, a rare occurrence, can be treated expectantly in this setting if the patient is otherwise stable. The uncomplicated pseudocyst that is stable in size and is at least 6 weeks old (to allow the walls to “mature” enough to accept sutures) is conventionally drained internally at laparotomy; in the intensive care situation, percutaneous drainage can be done safely to avoid the dreaded complications, but it must be used with full knowledge that recurrence is frequent.