Mansour A. Parsi, MD

A 60-year-old woman has had jaundice, dark-colored urine, and light-colored stools for the past several days. She has no history of jaundice or gallstone disease.

Over a century ago, Courvoisier observed that a palpable gallbladder in a patient with obstructive jaundice is often caused by a non-calculus abnormality of the biliary system, such as pancreatic cancer or cholangiocarcinoma, distal to the insertion of the cystic duct.^1–4 He attributed his findings to a higher likelihood of fibrosis of the gallbladder, with stone disease rendering it less distensible.⁴

Although often associated with malignancy, the Courvoisier sign can also be seen in benign processes causing obstruction of the common bile duct.⁵

For decades after its initial description, the Courvoisier sign was used as an important sign for the differential diagnosis of jaundice, but advances in diagnostic imaging have led to a more accurate and earlier diagnosis with less reliance on this sign. In this patient, tissue diagnosis confirmed a clinical suspicion of pancreatic adenocarcinoma.

A 60-year-old woman has had jaundice, dark-colored urine, and light-colored stools for the past several days. She has no history of jaundice or gallstone disease.

Over a century ago, Courvoisier observed that a palpable gallbladder in a patient with obstructive jaundice is often caused by a non-calculus abnormality of the biliary system, such as pancreatic cancer or cholangiocarcinoma, distal to the insertion of the cystic duct.^1–4 He attributed his findings to a higher likelihood of fibrosis of the gallbladder, with stone disease rendering it less distensible.⁴

Although often associated with malignancy, the Courvoisier sign can also be seen in benign processes causing obstruction of the common bile duct.⁵

For decades after its initial description, the Courvoisier sign was used as an important sign for the differential diagnosis of jaundice, but advances in diagnostic imaging have led to a more accurate and earlier diagnosis with less reliance on this sign. In this patient, tissue diagnosis confirmed a clinical suspicion of pancreatic adenocarcinoma.

A 60-year-old woman has had jaundice, dark-colored urine, and light-colored stools for the past several days. She has no history of jaundice or gallstone disease.

Over a century ago, Courvoisier observed that a palpable gallbladder in a patient with obstructive jaundice is often caused by a non-calculus abnormality of the biliary system, such as pancreatic cancer or cholangiocarcinoma, distal to the insertion of the cystic duct.^1–4 He attributed his findings to a higher likelihood of fibrosis of the gallbladder, with stone disease rendering it less distensible.⁴

Although often associated with malignancy, the Courvoisier sign can also be seen in benign processes causing obstruction of the common bile duct.⁵

For decades after its initial description, the Courvoisier sign was used as an important sign for the differential diagnosis of jaundice, but advances in diagnostic imaging have led to a more accurate and earlier diagnosis with less reliance on this sign. In this patient, tissue diagnosis confirmed a clinical suspicion of pancreatic adenocarcinoma.

Several major gastroenterological and surgical societies have issued guidelines on how to manage acute pancreatitis, based on evidence from high-quality randomized trials and nonrandomized studies as well as on expert opinion.^1–3 Information is limited on how well physicians in the United States comply with these guidelines, but compliance is suboptimal in other developed countries, according to several studies,^4–8 and we suspect that many US physicians are not following the guidelines either.

Acute pancreatitis is a frequent inpatient diagnosis that internists, gastroenterologists, and surgeons all confront. The most common causes are gallstones and heavy alcohol intake. Its management is typically straightforward: intravenous fluids, analgesia, and nothing by mouth. However, treatment of severe cases can be quite complex, particularly if multiple organ systems are involved or if there are local complications.

The primary aim of this article is to raise awareness of recognized deviations from established recommendations that may lead to adverse patient outcomes.

MEASURING ENZYME LEVELS DAILY ADDS COST BUT LITTLE BENEFIT

Problem: Serum amylase and lipase levels are often needlessly measured every day.

Measuring the serum amylase and lipase levels is useful in diagnosing acute pancreatitis, which requires two of the following three features¹:

• Characteristic abdominal pain
• Levels of serum amylase or serum lipase, or both, that are three or more times the upper limit of normal
• Findings of acute pancreatitis on computed tomography (CT).

However, the magnitude or duration of the serum enzyme elevation does not correlate with the severity of the attack. Further, we have noticed that physicians at our hospital often order daily serum amylase and lipase levels in patients admitted with acute pancreatitis.

The American College of Gastroenterology (ACG) guidelines¹ state that daily monitoring of amylase and lipase has limited value in managing acute pancreatitis. Rechecking these concentrations may be reasonable if pain fails to resolve or worsens during a prolonged hospitalization, as this may suggest a recurrent attack of acute pancreatitis or a developing pseudocyst. But in most cases of acute pancreatitis, daily serum enzyme measurements add cost but little benefit.

REGULAR ASSESSMENT IS IMPORTANT

Problem: Often, severity assessments are not performed regularly or acted on.

Most cases of acute pancreatitis are mild, with rapid recovery and excellent prognosis. However, 15% to 20% are severe and may result in a prolonged hospitalization, systemic inflammatory response syndrome (SIRS), multiorgan system failure, and death.

In severe acute pancreatitis, as pancreatic enzymes and inflammatory cytokines damage the blood vessels, a vast amount of fluid leaks out into the interstitial (“third”) space. This fluid extravasation leads to decreased effective circulating volume, local pancreatic necrosis, hemodynamic instability, and end-organ failure.

It is important to recognize severe acute pancreatitis early because the patient needs to be transferred to a step-down unit or intensive care unit to receive optimal fluid resuscitation and supportive care for organ dysfunction. After 48 to 72 hours, a prediction of severe acute pancreatitis should also prompt the physician to order CT to detect pancreatic necrosis, and also to initiate nutritional support.

Assessment of severity begins in the emergency room or on admission to the hospital. Older age, obesity, organ failure, and pulmonary infiltrates or pleural effusions are initial indicators of poor prognosis. Signs of SIRS (high or low core body temperature, tachycardia, tachypnea, low or high peripheral white blood cell count) or organ failure (eg, elevated serum creatinine) are present on admission in 21% of patients with acute pancreatitis.⁹

Hemoconcentration is a marker of decreased effective circulating volume in severe acute pancreatitis. A hematocrit higher than 44% at admission or that rises in the first 24 to 48 hours of admission predicts necrosis.^10,11 However, a more robust marker of organ failure may be the blood urea nitrogen level.¹²

Several clinical scoring systems have been studied for assessing severity.

Limitations of the Ranson score are that it can only be completed after 48 hours, all the data points are not always obtained, and it cannot be repeated on a daily basis. Owing to these limitations and its less-than-optimal predictive value, the Ranson score has fallen into disuse.

The APACHE II (Acute Physiology and Chronic Health Evaluation II) score is more versatile. It is based on multiple clinical and laboratory values, and it correlates very well with the risk of death in acute pancreatitis. Death rates are less than 4% when the APACHE II score is less than 8, and 11% to 18% when it is 8 or higher.¹ The trajectory of the APACHE II score in the first 48 hours is also an accurate prognostic indicator.

Previous limitations of the APACHE II score were that it was complicated and timeconsuming to calculate and required arterial blood gas measurements. Easy-to-use online calculators are now available (eg, www.globalrph.com/apacheii.htm), and the venous bicarbonate level and the oxygen saturation can be substituted for the arterial pH and oxygen partial pressure.

BISAP, a new five-point scoring system,¹⁵ was recently prospectively validated.¹² “BISAP” is an acronym for the five markers it is based on, each of which has been shown to predict severe illness in acute pancreatitis:

• Blood urea nitrogen level > 25 mg/dL
• Impaired mental status
• SIRS
• Age > 60 years
• Pleural effusion.

The presence of three or more of these factors correlates with higher risk of death, organ failure, and pancreatic necrosis.¹²

Compared with APACHE II, BISAP has similar accuracy and is easier to calculate. Also, BISAP was specifically developed for acute pancreatitis, whereas APACHE II is a generic score for all critically ill patients.

The Atlanta criteria¹⁶ define severe acute pancreatitis as one or more of the following:

• A Ranson score of 3 or higher during the first 48 hours
• An APACHE II score of 8 or higher at any time
• Failure of one or more organs
• One or more local complications (eg, necrosis, pseudocysts, abscesses).

Recommendation: Assess severity at least daily

A severity assessment should be performed at admission and at least every day thereafter. Clinical guidelines recognize the importance of severity assessment but vary in their specific recommendations.

The ACG advises calculating the APACHE II score within 3 days of admission and measuring the hematocrit at admission, at 12 hours, and at 24 hours. The level of evidence is III, ie, “from published well-designed trials without randomization, single group prepost, cohort, time series, or matched case controlled studies”.¹

The American Gastroenterological Association (AGA) provides a more generalized recommendation, that “clinical judgment” should take into account the presence of risk factors (eg, age, obesity), presence or absence of SIRS, routine laboratory values (eg, hematocrit, serum creatinine), and APACHE II score when assessing severity and making decisions.²

In a German survey, only 32% of gastroenterologists used the APACHE II score for assessing risk in acute pancreatitis, in spite of national guidelines emphasizing its importance.⁷ Also, not all patients with severe acute pancreatitis are transferred to a step-down unit or intensive care unit as recommended. In a British study,⁴ only 8 (17%) of 46 patients with predicted severe acute pancreatitis were transferred, and 8 of the 38 patients who were not transferred died.

FLUID MUST BE AGGRESSIVELY REPLACED AND MONITORED

Problem: Often, not enough fluid is replaced, or fluid status is not adequately monitored.

Fluid must be aggressively replaced to balance the massive third-space fluid losses that occur in the early inflammatory phase of acute pancreatitis. Intravascular volume depletion can develop rapidly and result in tachycardia, hypotension, and renal failure. It may also impair the blood flow to the pancreas and worsen necrosis.

Animal studies show that aggressive fluid replacement supports the pancreatic microcirculation and prevents necrosis.¹⁷ It may also support the intestinal microcirculation and gut barrier, preventing bacterial translocation.

In humans, no controlled trials have been done to test the efficacy of aggressive fluid resuscitation in acute pancreatitis. However, the notion that intravascular fluid loss contributes to poor outcomes is inferred from human studies showing more necrosis and deaths in patients with hemoconcentration. In one study, patients who received inadequate fluid replacement (evidenced by a rise in hematocrit at 24 hours) were more likely to develop necrotizing pancreatitis.¹⁸

Experts have suggested initially infusing 500 to 1,000 mL of fluid per hour in those who are volume-depleted, initially infusing 250 to 350 mL per hour in those who are not volumedepleted, and adjusting the fluid rate every 1 to 4 hours on the basis of clinical variables.¹⁹ The sufficiency of fluid replacement should be carefully monitored by vital signs, urine output, and serum hematocrit.

On the other hand, overly aggressive fluid resuscitation can be detrimental in patients at risk of volume overload or pulmonary edema. Fluid replacement should be tempered in elderly patients and those with cardiac or renal comorbidities, and may require monitoring of central venous pressure.

The ACG and AGA guidelines both recognize the need for early aggressive volume replacement in acute pancreatitis (level of evidence III), but they do not specify the exact amounts and rates. Young and healthy patients should receive a rapid bolus of isotonic saline or Ringer’s lactate solution followed by an infusion at a high initial maintenance rate.

Few studies have been done to assess physicians’ compliance with recommendations for aggressive volume replacement. In an Italian multicenter study, patients with mild or severe acute pancreatitis received an average of only 2.5 L of fluid per day (about 100 mL/hour).²⁰ Gardner et al²¹ recently summarized the available evidence for fluid support in acute pancreatitis.

NUTRITIONAL SUPPORT

Problem: In many severe cases, enteral or parenteral feeding is not started soon enough.

Nutritional support entails enteral or parenteral feeding when an oral diet is contraindicated. Enteral feeding is usually via a nasojejunal tube, which may need to be placed under endoscopic or radiographic guidance. Neither parenteral nor nasojejunal feeding stimulates pancreatic secretion, and both are safe in acute pancreatitis.

Severe acute pancreatitis is an intensely catabolic state characterized by increased energy expenditure, protein breakdown, and substrate utilization. Patients may not be able to resume an oral diet for weeks or even months, particularly if local complications develop. Early nutritional support has been shown to improve outcomes in severe acute pancreatitis.²² Therefore, nutritional support should be started as soon as possible in severe acute pancreatitis based on initial clinical and radiographic indicators of severity, optimally within the first 2 or 3 days.

Enteral nutrition is preferred to parenteral nutrition in pancreatitis: it is less expensive and does not pose a risk of catheter-related infection or thrombosis or hepatic complications. Also, there is experimental evidence that enteral nutrition may preserve the gut barrier, decreasing mucosal permeability and bacterial translocation.

A number of small randomized trials compared enteral and parenteral nutrition in acute pancreatitis, but they yielded mixed results. A meta-analysis of six trials showed a lower rate of infectious complications with enteral than with parenteral nutrition. ²³ However, no significant difference was found in the rates of death or noninfectious complications.

Recommendation: Enteral feeding, when possible

Nutritional support is unnecessary in most cases of mild acute pancreatitis. Pancreatic inflammation typically resolves within a few days, allowing patients to resume eating. Occasionally, patients in whom pain resolves slowly and who fast for more than 5 to 7 days need nutritional support to prevent proteincalorie malnutrition.

The ACG guidelines¹ and most others suggest that, whenever possible, enteral rather than parenteral feeding should be given to those who require nutritional support. The level of evidence is II (“strong evidence from at least one published properly designed randomized controlled trial of appropriate size and in an appropriate clinical setting”).

However, not all physicians recognize the benefit of enteral feeding. In a cohort of German gastroenterologists, only 73% favored enteral over parenteral feeding in acute pancreatitis.⁷

COMPUTED TOMOGRAPHY

Problem: CT is not done in many patients with severe acute pancreatitis, or it is done too soon during the admission.

Dual-phase, contrast-enhanced, pancreatic-protocol CT provides a sensitive structural evaluation of the pancreas and is useful to diagnose necrotizing pancreatitis. Pancreatic necrosis is correlated with a severe clinical course, the development of single or multiorgan dysfunction, and death.

Recommendation: CT in severe cases

Not every patient with acute pancreatitis needs to undergo CT. Most mild cases do not require routine CT, since necrosis and other local complications are infrequent in this group.

Also, CT is often ordered too soon during the hospitalization. Indicators of severity on CT are not usually evident until 2 to 3 days after admission.²⁵ CT should be considered about 3 days after the onset of symptoms rather than immediately upon admission.

On the other hand, CT at the time of admission may be warranted to rule out other life-threatening causes of abdominal pain and hyperamylasemia (eg, bowel obstruction, viscus perforation). CT may also be useful in the late phase of acute pancreatitis (weeks after admission) to diagnose or monitor complications (eg, pseudocysts, abscesses, splenic vein thrombosis, splenic artery pseudoaneurysms). Magnetic resonance imaging with gadolinium contrast is a reasonable alternative to CT for detecting pancreatic necrosis and other local complications.

In patients who have severe acute pancreatitis and compromised renal function (serum creatinine > 1.5 mg/dL), CT can be performed without contrast to assess severity based on a limited Balthazar score (ie, without a necrosis score). Studies in rats suggest that iodinated contrast may decrease pancreatic microcirculation and worsen or precipitate necrosis,26 although published human studies do not support this contention.27,28

Guidelines uniformly recommend CT for patients with severe acute pancreatitis (the ACG guideline gives it a level of evidence of III), but this recommendation is not always followed. A study from Australia showed that CT was done in only 27% to 67% of patients with severe acute pancreatitis.⁵ In a British study, only 8 of 46 patients with clinically predicted severe pancreatitis underwent CT within the first 10 days of admission.⁴

SUSPECTED INFECTED NECROSIS

Problem: Fine-needle aspiration is not done in many cases of suspected infected necrosis.

Approximately one-third of patients with necrotizing pancreatitis develop infected necrosis. The death rate for patients with infected pancreatic necrosis is high—30%, compared with 12% in those with sterile necrosis.¹ Differentiating sterile and infected necrosis is therefore essential.

Clinical signs such as fever are poor predictors of infection. Signs of SIRS can be present in both sterile and infected necrotizing pancreatitis.

Recommendation: Fine-needle aspiration of necrosis

For the reasons given above, the findings of necrosis on CT and persistent SIRS should prompt consideration of fine-needle aspiration with Gram stain and culture to differentiate sterile and infected necrosis (ACG guideline, level of evidence III).¹ If infection is confirmed, surgical debridement should be strongly considered. Other less-invasive approaches such as endoscopic debridement can be used in selected cases.

Fine-needle aspiration of necrosis is too often neglected. In a cohort of German surgeons, only 55% complied with International Association of Pancreatology recommendations to perform biopsy to differentiate sterile from infected necrosis in patients with signs of sepsis.²⁹

BROAD-SPECTRUM ANTIBIOTICS

Problem: Broad-spectrum antibiotics are often used inappropriately in patients with mild acute pancreatitis and in patients with sterile necrotizing pancreatitis who are clinically stable and have no signs of sepsis.

Antibiotics are not indicated in mild acute pancreatitis. A limited course of antibiotics is typically indicated in severe cases with suspected or proven infected necrosis (in conjunction with surgical necrosectomy). However, the use of antibiotics in sterile necrosis has been very controversial.

At least six small, nonblinded, randomized trials have evaluated the benefit of giving antibiotics prophylactically for presumed sterile necrosis. A recent Cochrane analysis of five of these trials (294 patients) suggested that patients who got antibiotics had a lower risk of death (odds ratio 0.37, 95% confidence interval [CI] 0.17–0.83) but no difference in the rates of pancreatic infection or surgery.³⁰ These paradoxical results suggest that antibiotics may prevent death by preventing nonpancreatic infections (eg, pneumonia, line infections) rather than by preventing infection of necrotic pancreatic tissue. The five trials in the meta-analysis are limited by significant methodologic heterogeneity and by lack of double-blinding.

In spite of the overall lower death rate observed in the meta-analysis, the prophylactic use of antibiotics in sterile necrosis remains controversial. One concern is that patients given long prophylactic courses of antibiotics may develop resistant bacterial or fungal infections. However, the Cochrane and other meta-analyses have not shown a higher rate of fungal infections in those given antibiotics.³¹

Recommendation: No routine antibiotics for mild cases

The AGA guidelines recommend against routinely giving antibiotics in mild acute pancreatitis and do not provide strict recommendations for prophylactic antibiotic use in necrotizing acute pancreatitis.² The guidelines state that antibiotics can be used “on demand” based on clinical signs of infection (eg, high fevers, rising leukocytosis, hypotension) or worsening organ failure.

If a purely prophylactic strategy is used, only patients at high risk of developing infection (eg, those with necrosis in more than 30% of the pancreas) should receive antibiotics. Antibiotics with high tissue-penetration should be used, such as imipenem-cilastin (Primaxin IV) or ciprofloxacin (Cipro) plus metronidazole (Flagyl).

Adherence to these guidelines is not optimal. For example, in an Italian multicenter study, 9% of patients with mild acute pancreatitis were treated with antibiotics.¹⁹ Moreover, many patients with proven infected necrosis received antibiotics that do not penetrate the pancreatic tissue very well.

ERCP IN SEVERE BILIARY ACUTE PANCREATITIS

Problem: Endoscopic retrograde cholangiopancreatography (ERCP) often is performed inappropriately in mild biliary acute pancreatitis or is not performed urgently in severe cases.

In most cases of mild biliary pancreatitis, the stones pass spontaneously, as verified by cholangiography done during laparoscopic cholecystectomy. Ongoing ampullary obstruction by impacted biliary stones can perpetuate pancreatic inflammation and delay recovery.

Two early randomized trials showed a benefit from early ERCP (within 72 hours) with sphincterotomy and stone extraction, primarily in those with severe biliary acute pancreatitis or ascending cholangitis,^32,33 but a third trial failed to reveal a benefit.³⁴ A Cochrane metaanalysis of these three trials failed to show a lower death rate with ERCP in mild or severe biliary pancreatitis.³⁵ However, early ERCP did prevent complications in severe biliary pancreatitis (odds ratio 0.27, 95% CI 0.14–0.53).

Later, a fourth randomized trial was restricted to patients with suspected biliary pancreatitis, evidence of biliary obstruction, and no signs of cholangitis³⁶: 103 patients were randomized to undergo either ERCP within 72 hours or conservative management. No difference was observed in rates of death or organ failure or in the CT severity index.

Recommendation: ER CP for suspected retained stones

ERCP has a limited role in patients with biliary pancreatitis, being used to clear retained bile duct stones or to relieve ongoing biliary obstruction.

The decision to perform ERCP before surgery should be based on how strongly one suspects retained stones. ERCP is most appropriate if the suspicion of retained stones and the likelihood of therapeutic intervention are high (eg, if the serum bilirubin and alkaline phosphatase levels are rising and ultrasonography shows a dilated bile duct). If there is moderate suspicion, a safer and less-invasive imaging study such as magnetic resonance cholangiopancreatography (MRCP) or endoscopic ultrasonography can be done to screen for bile duct stones before proceeding to ERCP.

The ACG guidelines suggest urgent ERCP (preferably within 24 hours) for those with severe biliary pancreatitis complicated by organ failure or those with suspicion of cholangitis. The level of evidence is I, ie, “strong evidence from at least one published systematic review of multiple well-designed randomized controlled trials.”¹

Elective ERCP is recommended for those who are poor surgical candidates. ERCP is also recommended for those with rising liver enzyme values or imaging findings suggesting a retained common bile duct stone (including intraoperative cholangiography). Endoscopic ultrasonography or MRCP is recommended for those with slow clinical resolution, who are pregnant, or in whom uncertainty exists regarding the biliary etiology of pancreatitis.

Compliance rates with these and similar guidelines are not adequate. In an audit of adherence to the British Society of Gastroenterology guidelines, early ERCP was performed in only 25% of patients with severe biliary acute pancreatitis.⁶

LAPAROSCOPIC CHOLECYSTECTOMY FOR MILD BILIARY PANCREATITIS

Problem: Laparoscopic cholecystectomy is not done at admission or within 2 weeks in many patients with mild biliary pancreatitis.

If the gallbladder is not removed, biliary pancreatitis may recur in up to 61% of patients within 6 weeks of hospital discharge.³⁷ This is the basis for guideline recommendations for surgery (or a confirmation of a surgery date) prior to hospital discharge.

The International Association of Pancreatology recommends early cholecystectomy (preferably during the same hospitalization) for patients with mild gallstone-associated acute pancreatitis.³⁸ In severe gallstone-associated acute pancreatitis, cholecystectomy should be delayed until there is sufficient resolution of the inflammatory response and clinical recovery. The AGA guidelines advocate cholecystectomy as soon as possible and in no case later than 4 weeks after discharge to prevent relapse. ERCP with biliary sphinc-terotomy may also protect against relapse in those who are not fit to undergo surgery.

Recommendations for definitive management of gallstones (laparoscopic cholecystectomy or ERCP, or both) are not always followed. For example, a British study showed 70% compliance with this recommendation.⁴ A similar compliance audit in Germany revealed that cholecystectomy was performed during the initial hospital stay in only 23% of cases.⁷ In a New Zealand study, a regular compliance audit with feedback to surgeons resulted in an increase in the early cholecystectomy rate from 54% to 80%.⁸

Several major gastroenterological and surgical societies have issued guidelines on how to manage acute pancreatitis, based on evidence from high-quality randomized trials and nonrandomized studies as well as on expert opinion.^1–3 Information is limited on how well physicians in the United States comply with these guidelines, but compliance is suboptimal in other developed countries, according to several studies,^4–8 and we suspect that many US physicians are not following the guidelines either.

Acute pancreatitis is a frequent inpatient diagnosis that internists, gastroenterologists, and surgeons all confront. The most common causes are gallstones and heavy alcohol intake. Its management is typically straightforward: intravenous fluids, analgesia, and nothing by mouth. However, treatment of severe cases can be quite complex, particularly if multiple organ systems are involved or if there are local complications.

The primary aim of this article is to raise awareness of recognized deviations from established recommendations that may lead to adverse patient outcomes.

MEASURING ENZYME LEVELS DAILY ADDS COST BUT LITTLE BENEFIT

Problem: Serum amylase and lipase levels are often needlessly measured every day.

Measuring the serum amylase and lipase levels is useful in diagnosing acute pancreatitis, which requires two of the following three features¹:

• Characteristic abdominal pain
• Levels of serum amylase or serum lipase, or both, that are three or more times the upper limit of normal
• Findings of acute pancreatitis on computed tomography (CT).

However, the magnitude or duration of the serum enzyme elevation does not correlate with the severity of the attack. Further, we have noticed that physicians at our hospital often order daily serum amylase and lipase levels in patients admitted with acute pancreatitis.

The American College of Gastroenterology (ACG) guidelines¹ state that daily monitoring of amylase and lipase has limited value in managing acute pancreatitis. Rechecking these concentrations may be reasonable if pain fails to resolve or worsens during a prolonged hospitalization, as this may suggest a recurrent attack of acute pancreatitis or a developing pseudocyst. But in most cases of acute pancreatitis, daily serum enzyme measurements add cost but little benefit.

REGULAR ASSESSMENT IS IMPORTANT

Problem: Often, severity assessments are not performed regularly or acted on.

Most cases of acute pancreatitis are mild, with rapid recovery and excellent prognosis. However, 15% to 20% are severe and may result in a prolonged hospitalization, systemic inflammatory response syndrome (SIRS), multiorgan system failure, and death.

In severe acute pancreatitis, as pancreatic enzymes and inflammatory cytokines damage the blood vessels, a vast amount of fluid leaks out into the interstitial (“third”) space. This fluid extravasation leads to decreased effective circulating volume, local pancreatic necrosis, hemodynamic instability, and end-organ failure.

It is important to recognize severe acute pancreatitis early because the patient needs to be transferred to a step-down unit or intensive care unit to receive optimal fluid resuscitation and supportive care for organ dysfunction. After 48 to 72 hours, a prediction of severe acute pancreatitis should also prompt the physician to order CT to detect pancreatic necrosis, and also to initiate nutritional support.

Assessment of severity begins in the emergency room or on admission to the hospital. Older age, obesity, organ failure, and pulmonary infiltrates or pleural effusions are initial indicators of poor prognosis. Signs of SIRS (high or low core body temperature, tachycardia, tachypnea, low or high peripheral white blood cell count) or organ failure (eg, elevated serum creatinine) are present on admission in 21% of patients with acute pancreatitis.⁹

Hemoconcentration is a marker of decreased effective circulating volume in severe acute pancreatitis. A hematocrit higher than 44% at admission or that rises in the first 24 to 48 hours of admission predicts necrosis.^10,11 However, a more robust marker of organ failure may be the blood urea nitrogen level.¹²

Several clinical scoring systems have been studied for assessing severity.

Limitations of the Ranson score are that it can only be completed after 48 hours, all the data points are not always obtained, and it cannot be repeated on a daily basis. Owing to these limitations and its less-than-optimal predictive value, the Ranson score has fallen into disuse.

The APACHE II (Acute Physiology and Chronic Health Evaluation II) score is more versatile. It is based on multiple clinical and laboratory values, and it correlates very well with the risk of death in acute pancreatitis. Death rates are less than 4% when the APACHE II score is less than 8, and 11% to 18% when it is 8 or higher.¹ The trajectory of the APACHE II score in the first 48 hours is also an accurate prognostic indicator.

Previous limitations of the APACHE II score were that it was complicated and timeconsuming to calculate and required arterial blood gas measurements. Easy-to-use online calculators are now available (eg, www.globalrph.com/apacheii.htm), and the venous bicarbonate level and the oxygen saturation can be substituted for the arterial pH and oxygen partial pressure.

BISAP, a new five-point scoring system,¹⁵ was recently prospectively validated.¹² “BISAP” is an acronym for the five markers it is based on, each of which has been shown to predict severe illness in acute pancreatitis:

• Blood urea nitrogen level > 25 mg/dL
• Impaired mental status
• SIRS
• Age > 60 years
• Pleural effusion.

The presence of three or more of these factors correlates with higher risk of death, organ failure, and pancreatic necrosis.¹²

Compared with APACHE II, BISAP has similar accuracy and is easier to calculate. Also, BISAP was specifically developed for acute pancreatitis, whereas APACHE II is a generic score for all critically ill patients.

The Atlanta criteria¹⁶ define severe acute pancreatitis as one or more of the following:

• A Ranson score of 3 or higher during the first 48 hours
• An APACHE II score of 8 or higher at any time
• Failure of one or more organs
• One or more local complications (eg, necrosis, pseudocysts, abscesses).

Recommendation: Assess severity at least daily

A severity assessment should be performed at admission and at least every day thereafter. Clinical guidelines recognize the importance of severity assessment but vary in their specific recommendations.

The ACG advises calculating the APACHE II score within 3 days of admission and measuring the hematocrit at admission, at 12 hours, and at 24 hours. The level of evidence is III, ie, “from published well-designed trials without randomization, single group prepost, cohort, time series, or matched case controlled studies”.¹

The American Gastroenterological Association (AGA) provides a more generalized recommendation, that “clinical judgment” should take into account the presence of risk factors (eg, age, obesity), presence or absence of SIRS, routine laboratory values (eg, hematocrit, serum creatinine), and APACHE II score when assessing severity and making decisions.²

In a German survey, only 32% of gastroenterologists used the APACHE II score for assessing risk in acute pancreatitis, in spite of national guidelines emphasizing its importance.⁷ Also, not all patients with severe acute pancreatitis are transferred to a step-down unit or intensive care unit as recommended. In a British study,⁴ only 8 (17%) of 46 patients with predicted severe acute pancreatitis were transferred, and 8 of the 38 patients who were not transferred died.

FLUID MUST BE AGGRESSIVELY REPLACED AND MONITORED

Problem: Often, not enough fluid is replaced, or fluid status is not adequately monitored.

Fluid must be aggressively replaced to balance the massive third-space fluid losses that occur in the early inflammatory phase of acute pancreatitis. Intravascular volume depletion can develop rapidly and result in tachycardia, hypotension, and renal failure. It may also impair the blood flow to the pancreas and worsen necrosis.

Animal studies show that aggressive fluid replacement supports the pancreatic microcirculation and prevents necrosis.¹⁷ It may also support the intestinal microcirculation and gut barrier, preventing bacterial translocation.

In humans, no controlled trials have been done to test the efficacy of aggressive fluid resuscitation in acute pancreatitis. However, the notion that intravascular fluid loss contributes to poor outcomes is inferred from human studies showing more necrosis and deaths in patients with hemoconcentration. In one study, patients who received inadequate fluid replacement (evidenced by a rise in hematocrit at 24 hours) were more likely to develop necrotizing pancreatitis.¹⁸

Experts have suggested initially infusing 500 to 1,000 mL of fluid per hour in those who are volume-depleted, initially infusing 250 to 350 mL per hour in those who are not volumedepleted, and adjusting the fluid rate every 1 to 4 hours on the basis of clinical variables.¹⁹ The sufficiency of fluid replacement should be carefully monitored by vital signs, urine output, and serum hematocrit.

On the other hand, overly aggressive fluid resuscitation can be detrimental in patients at risk of volume overload or pulmonary edema. Fluid replacement should be tempered in elderly patients and those with cardiac or renal comorbidities, and may require monitoring of central venous pressure.

The ACG and AGA guidelines both recognize the need for early aggressive volume replacement in acute pancreatitis (level of evidence III), but they do not specify the exact amounts and rates. Young and healthy patients should receive a rapid bolus of isotonic saline or Ringer’s lactate solution followed by an infusion at a high initial maintenance rate.

Few studies have been done to assess physicians’ compliance with recommendations for aggressive volume replacement. In an Italian multicenter study, patients with mild or severe acute pancreatitis received an average of only 2.5 L of fluid per day (about 100 mL/hour).²⁰ Gardner et al²¹ recently summarized the available evidence for fluid support in acute pancreatitis.

NUTRITIONAL SUPPORT

Problem: In many severe cases, enteral or parenteral feeding is not started soon enough.

Nutritional support entails enteral or parenteral feeding when an oral diet is contraindicated. Enteral feeding is usually via a nasojejunal tube, which may need to be placed under endoscopic or radiographic guidance. Neither parenteral nor nasojejunal feeding stimulates pancreatic secretion, and both are safe in acute pancreatitis.

Severe acute pancreatitis is an intensely catabolic state characterized by increased energy expenditure, protein breakdown, and substrate utilization. Patients may not be able to resume an oral diet for weeks or even months, particularly if local complications develop. Early nutritional support has been shown to improve outcomes in severe acute pancreatitis.²² Therefore, nutritional support should be started as soon as possible in severe acute pancreatitis based on initial clinical and radiographic indicators of severity, optimally within the first 2 or 3 days.

Enteral nutrition is preferred to parenteral nutrition in pancreatitis: it is less expensive and does not pose a risk of catheter-related infection or thrombosis or hepatic complications. Also, there is experimental evidence that enteral nutrition may preserve the gut barrier, decreasing mucosal permeability and bacterial translocation.

A number of small randomized trials compared enteral and parenteral nutrition in acute pancreatitis, but they yielded mixed results. A meta-analysis of six trials showed a lower rate of infectious complications with enteral than with parenteral nutrition. ²³ However, no significant difference was found in the rates of death or noninfectious complications.

Recommendation: Enteral feeding, when possible

Nutritional support is unnecessary in most cases of mild acute pancreatitis. Pancreatic inflammation typically resolves within a few days, allowing patients to resume eating. Occasionally, patients in whom pain resolves slowly and who fast for more than 5 to 7 days need nutritional support to prevent proteincalorie malnutrition.

The ACG guidelines¹ and most others suggest that, whenever possible, enteral rather than parenteral feeding should be given to those who require nutritional support. The level of evidence is II (“strong evidence from at least one published properly designed randomized controlled trial of appropriate size and in an appropriate clinical setting”).

However, not all physicians recognize the benefit of enteral feeding. In a cohort of German gastroenterologists, only 73% favored enteral over parenteral feeding in acute pancreatitis.⁷

COMPUTED TOMOGRAPHY

Problem: CT is not done in many patients with severe acute pancreatitis, or it is done too soon during the admission.

Dual-phase, contrast-enhanced, pancreatic-protocol CT provides a sensitive structural evaluation of the pancreas and is useful to diagnose necrotizing pancreatitis. Pancreatic necrosis is correlated with a severe clinical course, the development of single or multiorgan dysfunction, and death.

Recommendation: CT in severe cases

Not every patient with acute pancreatitis needs to undergo CT. Most mild cases do not require routine CT, since necrosis and other local complications are infrequent in this group.

Also, CT is often ordered too soon during the hospitalization. Indicators of severity on CT are not usually evident until 2 to 3 days after admission.²⁵ CT should be considered about 3 days after the onset of symptoms rather than immediately upon admission.

On the other hand, CT at the time of admission may be warranted to rule out other life-threatening causes of abdominal pain and hyperamylasemia (eg, bowel obstruction, viscus perforation). CT may also be useful in the late phase of acute pancreatitis (weeks after admission) to diagnose or monitor complications (eg, pseudocysts, abscesses, splenic vein thrombosis, splenic artery pseudoaneurysms). Magnetic resonance imaging with gadolinium contrast is a reasonable alternative to CT for detecting pancreatic necrosis and other local complications.

In patients who have severe acute pancreatitis and compromised renal function (serum creatinine > 1.5 mg/dL), CT can be performed without contrast to assess severity based on a limited Balthazar score (ie, without a necrosis score). Studies in rats suggest that iodinated contrast may decrease pancreatic microcirculation and worsen or precipitate necrosis,26 although published human studies do not support this contention.27,28

Guidelines uniformly recommend CT for patients with severe acute pancreatitis (the ACG guideline gives it a level of evidence of III), but this recommendation is not always followed. A study from Australia showed that CT was done in only 27% to 67% of patients with severe acute pancreatitis.⁵ In a British study, only 8 of 46 patients with clinically predicted severe pancreatitis underwent CT within the first 10 days of admission.⁴

SUSPECTED INFECTED NECROSIS

Problem: Fine-needle aspiration is not done in many cases of suspected infected necrosis.

Approximately one-third of patients with necrotizing pancreatitis develop infected necrosis. The death rate for patients with infected pancreatic necrosis is high—30%, compared with 12% in those with sterile necrosis.¹ Differentiating sterile and infected necrosis is therefore essential.

Clinical signs such as fever are poor predictors of infection. Signs of SIRS can be present in both sterile and infected necrotizing pancreatitis.

Recommendation: Fine-needle aspiration of necrosis

For the reasons given above, the findings of necrosis on CT and persistent SIRS should prompt consideration of fine-needle aspiration with Gram stain and culture to differentiate sterile and infected necrosis (ACG guideline, level of evidence III).¹ If infection is confirmed, surgical debridement should be strongly considered. Other less-invasive approaches such as endoscopic debridement can be used in selected cases.

Fine-needle aspiration of necrosis is too often neglected. In a cohort of German surgeons, only 55% complied with International Association of Pancreatology recommendations to perform biopsy to differentiate sterile from infected necrosis in patients with signs of sepsis.²⁹

BROAD-SPECTRUM ANTIBIOTICS

Problem: Broad-spectrum antibiotics are often used inappropriately in patients with mild acute pancreatitis and in patients with sterile necrotizing pancreatitis who are clinically stable and have no signs of sepsis.

Antibiotics are not indicated in mild acute pancreatitis. A limited course of antibiotics is typically indicated in severe cases with suspected or proven infected necrosis (in conjunction with surgical necrosectomy). However, the use of antibiotics in sterile necrosis has been very controversial.

At least six small, nonblinded, randomized trials have evaluated the benefit of giving antibiotics prophylactically for presumed sterile necrosis. A recent Cochrane analysis of five of these trials (294 patients) suggested that patients who got antibiotics had a lower risk of death (odds ratio 0.37, 95% confidence interval [CI] 0.17–0.83) but no difference in the rates of pancreatic infection or surgery.³⁰ These paradoxical results suggest that antibiotics may prevent death by preventing nonpancreatic infections (eg, pneumonia, line infections) rather than by preventing infection of necrotic pancreatic tissue. The five trials in the meta-analysis are limited by significant methodologic heterogeneity and by lack of double-blinding.

In spite of the overall lower death rate observed in the meta-analysis, the prophylactic use of antibiotics in sterile necrosis remains controversial. One concern is that patients given long prophylactic courses of antibiotics may develop resistant bacterial or fungal infections. However, the Cochrane and other meta-analyses have not shown a higher rate of fungal infections in those given antibiotics.³¹

Recommendation: No routine antibiotics for mild cases

The AGA guidelines recommend against routinely giving antibiotics in mild acute pancreatitis and do not provide strict recommendations for prophylactic antibiotic use in necrotizing acute pancreatitis.² The guidelines state that antibiotics can be used “on demand” based on clinical signs of infection (eg, high fevers, rising leukocytosis, hypotension) or worsening organ failure.

If a purely prophylactic strategy is used, only patients at high risk of developing infection (eg, those with necrosis in more than 30% of the pancreas) should receive antibiotics. Antibiotics with high tissue-penetration should be used, such as imipenem-cilastin (Primaxin IV) or ciprofloxacin (Cipro) plus metronidazole (Flagyl).

Adherence to these guidelines is not optimal. For example, in an Italian multicenter study, 9% of patients with mild acute pancreatitis were treated with antibiotics.¹⁹ Moreover, many patients with proven infected necrosis received antibiotics that do not penetrate the pancreatic tissue very well.

ERCP IN SEVERE BILIARY ACUTE PANCREATITIS

Problem: Endoscopic retrograde cholangiopancreatography (ERCP) often is performed inappropriately in mild biliary acute pancreatitis or is not performed urgently in severe cases.

In most cases of mild biliary pancreatitis, the stones pass spontaneously, as verified by cholangiography done during laparoscopic cholecystectomy. Ongoing ampullary obstruction by impacted biliary stones can perpetuate pancreatic inflammation and delay recovery.

Two early randomized trials showed a benefit from early ERCP (within 72 hours) with sphincterotomy and stone extraction, primarily in those with severe biliary acute pancreatitis or ascending cholangitis,^32,33 but a third trial failed to reveal a benefit.³⁴ A Cochrane metaanalysis of these three trials failed to show a lower death rate with ERCP in mild or severe biliary pancreatitis.³⁵ However, early ERCP did prevent complications in severe biliary pancreatitis (odds ratio 0.27, 95% CI 0.14–0.53).

Later, a fourth randomized trial was restricted to patients with suspected biliary pancreatitis, evidence of biliary obstruction, and no signs of cholangitis³⁶: 103 patients were randomized to undergo either ERCP within 72 hours or conservative management. No difference was observed in rates of death or organ failure or in the CT severity index.

Recommendation: ER CP for suspected retained stones

ERCP has a limited role in patients with biliary pancreatitis, being used to clear retained bile duct stones or to relieve ongoing biliary obstruction.

The decision to perform ERCP before surgery should be based on how strongly one suspects retained stones. ERCP is most appropriate if the suspicion of retained stones and the likelihood of therapeutic intervention are high (eg, if the serum bilirubin and alkaline phosphatase levels are rising and ultrasonography shows a dilated bile duct). If there is moderate suspicion, a safer and less-invasive imaging study such as magnetic resonance cholangiopancreatography (MRCP) or endoscopic ultrasonography can be done to screen for bile duct stones before proceeding to ERCP.

The ACG guidelines suggest urgent ERCP (preferably within 24 hours) for those with severe biliary pancreatitis complicated by organ failure or those with suspicion of cholangitis. The level of evidence is I, ie, “strong evidence from at least one published systematic review of multiple well-designed randomized controlled trials.”¹

Elective ERCP is recommended for those who are poor surgical candidates. ERCP is also recommended for those with rising liver enzyme values or imaging findings suggesting a retained common bile duct stone (including intraoperative cholangiography). Endoscopic ultrasonography or MRCP is recommended for those with slow clinical resolution, who are pregnant, or in whom uncertainty exists regarding the biliary etiology of pancreatitis.

Compliance rates with these and similar guidelines are not adequate. In an audit of adherence to the British Society of Gastroenterology guidelines, early ERCP was performed in only 25% of patients with severe biliary acute pancreatitis.⁶

LAPAROSCOPIC CHOLECYSTECTOMY FOR MILD BILIARY PANCREATITIS

Problem: Laparoscopic cholecystectomy is not done at admission or within 2 weeks in many patients with mild biliary pancreatitis.

If the gallbladder is not removed, biliary pancreatitis may recur in up to 61% of patients within 6 weeks of hospital discharge.³⁷ This is the basis for guideline recommendations for surgery (or a confirmation of a surgery date) prior to hospital discharge.

The International Association of Pancreatology recommends early cholecystectomy (preferably during the same hospitalization) for patients with mild gallstone-associated acute pancreatitis.³⁸ In severe gallstone-associated acute pancreatitis, cholecystectomy should be delayed until there is sufficient resolution of the inflammatory response and clinical recovery. The AGA guidelines advocate cholecystectomy as soon as possible and in no case later than 4 weeks after discharge to prevent relapse. ERCP with biliary sphinc-terotomy may also protect against relapse in those who are not fit to undergo surgery.

Recommendations for definitive management of gallstones (laparoscopic cholecystectomy or ERCP, or both) are not always followed. For example, a British study showed 70% compliance with this recommendation.⁴ A similar compliance audit in Germany revealed that cholecystectomy was performed during the initial hospital stay in only 23% of cases.⁷ In a New Zealand study, a regular compliance audit with feedback to surgeons resulted in an increase in the early cholecystectomy rate from 54% to 80%.⁸

Several major gastroenterological and surgical societies have issued guidelines on how to manage acute pancreatitis, based on evidence from high-quality randomized trials and nonrandomized studies as well as on expert opinion.^1–3 Information is limited on how well physicians in the United States comply with these guidelines, but compliance is suboptimal in other developed countries, according to several studies,^4–8 and we suspect that many US physicians are not following the guidelines either.

Acute pancreatitis is a frequent inpatient diagnosis that internists, gastroenterologists, and surgeons all confront. The most common causes are gallstones and heavy alcohol intake. Its management is typically straightforward: intravenous fluids, analgesia, and nothing by mouth. However, treatment of severe cases can be quite complex, particularly if multiple organ systems are involved or if there are local complications.

The primary aim of this article is to raise awareness of recognized deviations from established recommendations that may lead to adverse patient outcomes.

MEASURING ENZYME LEVELS DAILY ADDS COST BUT LITTLE BENEFIT

Problem: Serum amylase and lipase levels are often needlessly measured every day.

Measuring the serum amylase and lipase levels is useful in diagnosing acute pancreatitis, which requires two of the following three features¹:

• Characteristic abdominal pain
• Levels of serum amylase or serum lipase, or both, that are three or more times the upper limit of normal
• Findings of acute pancreatitis on computed tomography (CT).

However, the magnitude or duration of the serum enzyme elevation does not correlate with the severity of the attack. Further, we have noticed that physicians at our hospital often order daily serum amylase and lipase levels in patients admitted with acute pancreatitis.

The American College of Gastroenterology (ACG) guidelines¹ state that daily monitoring of amylase and lipase has limited value in managing acute pancreatitis. Rechecking these concentrations may be reasonable if pain fails to resolve or worsens during a prolonged hospitalization, as this may suggest a recurrent attack of acute pancreatitis or a developing pseudocyst. But in most cases of acute pancreatitis, daily serum enzyme measurements add cost but little benefit.

REGULAR ASSESSMENT IS IMPORTANT

Problem: Often, severity assessments are not performed regularly or acted on.

Most cases of acute pancreatitis are mild, with rapid recovery and excellent prognosis. However, 15% to 20% are severe and may result in a prolonged hospitalization, systemic inflammatory response syndrome (SIRS), multiorgan system failure, and death.

In severe acute pancreatitis, as pancreatic enzymes and inflammatory cytokines damage the blood vessels, a vast amount of fluid leaks out into the interstitial (“third”) space. This fluid extravasation leads to decreased effective circulating volume, local pancreatic necrosis, hemodynamic instability, and end-organ failure.

It is important to recognize severe acute pancreatitis early because the patient needs to be transferred to a step-down unit or intensive care unit to receive optimal fluid resuscitation and supportive care for organ dysfunction. After 48 to 72 hours, a prediction of severe acute pancreatitis should also prompt the physician to order CT to detect pancreatic necrosis, and also to initiate nutritional support.

Assessment of severity begins in the emergency room or on admission to the hospital. Older age, obesity, organ failure, and pulmonary infiltrates or pleural effusions are initial indicators of poor prognosis. Signs of SIRS (high or low core body temperature, tachycardia, tachypnea, low or high peripheral white blood cell count) or organ failure (eg, elevated serum creatinine) are present on admission in 21% of patients with acute pancreatitis.⁹

Hemoconcentration is a marker of decreased effective circulating volume in severe acute pancreatitis. A hematocrit higher than 44% at admission or that rises in the first 24 to 48 hours of admission predicts necrosis.^10,11 However, a more robust marker of organ failure may be the blood urea nitrogen level.¹²

Several clinical scoring systems have been studied for assessing severity.

Limitations of the Ranson score are that it can only be completed after 48 hours, all the data points are not always obtained, and it cannot be repeated on a daily basis. Owing to these limitations and its less-than-optimal predictive value, the Ranson score has fallen into disuse.

The APACHE II (Acute Physiology and Chronic Health Evaluation II) score is more versatile. It is based on multiple clinical and laboratory values, and it correlates very well with the risk of death in acute pancreatitis. Death rates are less than 4% when the APACHE II score is less than 8, and 11% to 18% when it is 8 or higher.¹ The trajectory of the APACHE II score in the first 48 hours is also an accurate prognostic indicator.

Previous limitations of the APACHE II score were that it was complicated and timeconsuming to calculate and required arterial blood gas measurements. Easy-to-use online calculators are now available (eg, www.globalrph.com/apacheii.htm), and the venous bicarbonate level and the oxygen saturation can be substituted for the arterial pH and oxygen partial pressure.

BISAP, a new five-point scoring system,¹⁵ was recently prospectively validated.¹² “BISAP” is an acronym for the five markers it is based on, each of which has been shown to predict severe illness in acute pancreatitis:

• Blood urea nitrogen level > 25 mg/dL
• Impaired mental status
• SIRS
• Age > 60 years
• Pleural effusion.

The presence of three or more of these factors correlates with higher risk of death, organ failure, and pancreatic necrosis.¹²

Compared with APACHE II, BISAP has similar accuracy and is easier to calculate. Also, BISAP was specifically developed for acute pancreatitis, whereas APACHE II is a generic score for all critically ill patients.

The Atlanta criteria¹⁶ define severe acute pancreatitis as one or more of the following:

• A Ranson score of 3 or higher during the first 48 hours
• An APACHE II score of 8 or higher at any time
• Failure of one or more organs
• One or more local complications (eg, necrosis, pseudocysts, abscesses).

Recommendation: Assess severity at least daily

A severity assessment should be performed at admission and at least every day thereafter. Clinical guidelines recognize the importance of severity assessment but vary in their specific recommendations.

The ACG advises calculating the APACHE II score within 3 days of admission and measuring the hematocrit at admission, at 12 hours, and at 24 hours. The level of evidence is III, ie, “from published well-designed trials without randomization, single group prepost, cohort, time series, or matched case controlled studies”.¹

The American Gastroenterological Association (AGA) provides a more generalized recommendation, that “clinical judgment” should take into account the presence of risk factors (eg, age, obesity), presence or absence of SIRS, routine laboratory values (eg, hematocrit, serum creatinine), and APACHE II score when assessing severity and making decisions.²

In a German survey, only 32% of gastroenterologists used the APACHE II score for assessing risk in acute pancreatitis, in spite of national guidelines emphasizing its importance.⁷ Also, not all patients with severe acute pancreatitis are transferred to a step-down unit or intensive care unit as recommended. In a British study,⁴ only 8 (17%) of 46 patients with predicted severe acute pancreatitis were transferred, and 8 of the 38 patients who were not transferred died.

FLUID MUST BE AGGRESSIVELY REPLACED AND MONITORED

Problem: Often, not enough fluid is replaced, or fluid status is not adequately monitored.

Fluid must be aggressively replaced to balance the massive third-space fluid losses that occur in the early inflammatory phase of acute pancreatitis. Intravascular volume depletion can develop rapidly and result in tachycardia, hypotension, and renal failure. It may also impair the blood flow to the pancreas and worsen necrosis.

Animal studies show that aggressive fluid replacement supports the pancreatic microcirculation and prevents necrosis.¹⁷ It may also support the intestinal microcirculation and gut barrier, preventing bacterial translocation.

In humans, no controlled trials have been done to test the efficacy of aggressive fluid resuscitation in acute pancreatitis. However, the notion that intravascular fluid loss contributes to poor outcomes is inferred from human studies showing more necrosis and deaths in patients with hemoconcentration. In one study, patients who received inadequate fluid replacement (evidenced by a rise in hematocrit at 24 hours) were more likely to develop necrotizing pancreatitis.¹⁸

Experts have suggested initially infusing 500 to 1,000 mL of fluid per hour in those who are volume-depleted, initially infusing 250 to 350 mL per hour in those who are not volumedepleted, and adjusting the fluid rate every 1 to 4 hours on the basis of clinical variables.¹⁹ The sufficiency of fluid replacement should be carefully monitored by vital signs, urine output, and serum hematocrit.

On the other hand, overly aggressive fluid resuscitation can be detrimental in patients at risk of volume overload or pulmonary edema. Fluid replacement should be tempered in elderly patients and those with cardiac or renal comorbidities, and may require monitoring of central venous pressure.

The ACG and AGA guidelines both recognize the need for early aggressive volume replacement in acute pancreatitis (level of evidence III), but they do not specify the exact amounts and rates. Young and healthy patients should receive a rapid bolus of isotonic saline or Ringer’s lactate solution followed by an infusion at a high initial maintenance rate.

Few studies have been done to assess physicians’ compliance with recommendations for aggressive volume replacement. In an Italian multicenter study, patients with mild or severe acute pancreatitis received an average of only 2.5 L of fluid per day (about 100 mL/hour).²⁰ Gardner et al²¹ recently summarized the available evidence for fluid support in acute pancreatitis.

NUTRITIONAL SUPPORT

Problem: In many severe cases, enteral or parenteral feeding is not started soon enough.

Nutritional support entails enteral or parenteral feeding when an oral diet is contraindicated. Enteral feeding is usually via a nasojejunal tube, which may need to be placed under endoscopic or radiographic guidance. Neither parenteral nor nasojejunal feeding stimulates pancreatic secretion, and both are safe in acute pancreatitis.

Severe acute pancreatitis is an intensely catabolic state characterized by increased energy expenditure, protein breakdown, and substrate utilization. Patients may not be able to resume an oral diet for weeks or even months, particularly if local complications develop. Early nutritional support has been shown to improve outcomes in severe acute pancreatitis.²² Therefore, nutritional support should be started as soon as possible in severe acute pancreatitis based on initial clinical and radiographic indicators of severity, optimally within the first 2 or 3 days.

Enteral nutrition is preferred to parenteral nutrition in pancreatitis: it is less expensive and does not pose a risk of catheter-related infection or thrombosis or hepatic complications. Also, there is experimental evidence that enteral nutrition may preserve the gut barrier, decreasing mucosal permeability and bacterial translocation.

A number of small randomized trials compared enteral and parenteral nutrition in acute pancreatitis, but they yielded mixed results. A meta-analysis of six trials showed a lower rate of infectious complications with enteral than with parenteral nutrition. ²³ However, no significant difference was found in the rates of death or noninfectious complications.

Recommendation: Enteral feeding, when possible

Nutritional support is unnecessary in most cases of mild acute pancreatitis. Pancreatic inflammation typically resolves within a few days, allowing patients to resume eating. Occasionally, patients in whom pain resolves slowly and who fast for more than 5 to 7 days need nutritional support to prevent proteincalorie malnutrition.

The ACG guidelines¹ and most others suggest that, whenever possible, enteral rather than parenteral feeding should be given to those who require nutritional support. The level of evidence is II (“strong evidence from at least one published properly designed randomized controlled trial of appropriate size and in an appropriate clinical setting”).

However, not all physicians recognize the benefit of enteral feeding. In a cohort of German gastroenterologists, only 73% favored enteral over parenteral feeding in acute pancreatitis.⁷

COMPUTED TOMOGRAPHY

Problem: CT is not done in many patients with severe acute pancreatitis, or it is done too soon during the admission.

Dual-phase, contrast-enhanced, pancreatic-protocol CT provides a sensitive structural evaluation of the pancreas and is useful to diagnose necrotizing pancreatitis. Pancreatic necrosis is correlated with a severe clinical course, the development of single or multiorgan dysfunction, and death.

Recommendation: CT in severe cases

Not every patient with acute pancreatitis needs to undergo CT. Most mild cases do not require routine CT, since necrosis and other local complications are infrequent in this group.

Also, CT is often ordered too soon during the hospitalization. Indicators of severity on CT are not usually evident until 2 to 3 days after admission.²⁵ CT should be considered about 3 days after the onset of symptoms rather than immediately upon admission.

On the other hand, CT at the time of admission may be warranted to rule out other life-threatening causes of abdominal pain and hyperamylasemia (eg, bowel obstruction, viscus perforation). CT may also be useful in the late phase of acute pancreatitis (weeks after admission) to diagnose or monitor complications (eg, pseudocysts, abscesses, splenic vein thrombosis, splenic artery pseudoaneurysms). Magnetic resonance imaging with gadolinium contrast is a reasonable alternative to CT for detecting pancreatic necrosis and other local complications.

In patients who have severe acute pancreatitis and compromised renal function (serum creatinine > 1.5 mg/dL), CT can be performed without contrast to assess severity based on a limited Balthazar score (ie, without a necrosis score). Studies in rats suggest that iodinated contrast may decrease pancreatic microcirculation and worsen or precipitate necrosis,26 although published human studies do not support this contention.27,28

Guidelines uniformly recommend CT for patients with severe acute pancreatitis (the ACG guideline gives it a level of evidence of III), but this recommendation is not always followed. A study from Australia showed that CT was done in only 27% to 67% of patients with severe acute pancreatitis.⁵ In a British study, only 8 of 46 patients with clinically predicted severe pancreatitis underwent CT within the first 10 days of admission.⁴

SUSPECTED INFECTED NECROSIS

Problem: Fine-needle aspiration is not done in many cases of suspected infected necrosis.

Approximately one-third of patients with necrotizing pancreatitis develop infected necrosis. The death rate for patients with infected pancreatic necrosis is high—30%, compared with 12% in those with sterile necrosis.¹ Differentiating sterile and infected necrosis is therefore essential.

Clinical signs such as fever are poor predictors of infection. Signs of SIRS can be present in both sterile and infected necrotizing pancreatitis.

Recommendation: Fine-needle aspiration of necrosis

For the reasons given above, the findings of necrosis on CT and persistent SIRS should prompt consideration of fine-needle aspiration with Gram stain and culture to differentiate sterile and infected necrosis (ACG guideline, level of evidence III).¹ If infection is confirmed, surgical debridement should be strongly considered. Other less-invasive approaches such as endoscopic debridement can be used in selected cases.

Fine-needle aspiration of necrosis is too often neglected. In a cohort of German surgeons, only 55% complied with International Association of Pancreatology recommendations to perform biopsy to differentiate sterile from infected necrosis in patients with signs of sepsis.²⁹

BROAD-SPECTRUM ANTIBIOTICS

Problem: Broad-spectrum antibiotics are often used inappropriately in patients with mild acute pancreatitis and in patients with sterile necrotizing pancreatitis who are clinically stable and have no signs of sepsis.

Antibiotics are not indicated in mild acute pancreatitis. A limited course of antibiotics is typically indicated in severe cases with suspected or proven infected necrosis (in conjunction with surgical necrosectomy). However, the use of antibiotics in sterile necrosis has been very controversial.

At least six small, nonblinded, randomized trials have evaluated the benefit of giving antibiotics prophylactically for presumed sterile necrosis. A recent Cochrane analysis of five of these trials (294 patients) suggested that patients who got antibiotics had a lower risk of death (odds ratio 0.37, 95% confidence interval [CI] 0.17–0.83) but no difference in the rates of pancreatic infection or surgery.³⁰ These paradoxical results suggest that antibiotics may prevent death by preventing nonpancreatic infections (eg, pneumonia, line infections) rather than by preventing infection of necrotic pancreatic tissue. The five trials in the meta-analysis are limited by significant methodologic heterogeneity and by lack of double-blinding.

In spite of the overall lower death rate observed in the meta-analysis, the prophylactic use of antibiotics in sterile necrosis remains controversial. One concern is that patients given long prophylactic courses of antibiotics may develop resistant bacterial or fungal infections. However, the Cochrane and other meta-analyses have not shown a higher rate of fungal infections in those given antibiotics.³¹

Recommendation: No routine antibiotics for mild cases

The AGA guidelines recommend against routinely giving antibiotics in mild acute pancreatitis and do not provide strict recommendations for prophylactic antibiotic use in necrotizing acute pancreatitis.² The guidelines state that antibiotics can be used “on demand” based on clinical signs of infection (eg, high fevers, rising leukocytosis, hypotension) or worsening organ failure.

If a purely prophylactic strategy is used, only patients at high risk of developing infection (eg, those with necrosis in more than 30% of the pancreas) should receive antibiotics. Antibiotics with high tissue-penetration should be used, such as imipenem-cilastin (Primaxin IV) or ciprofloxacin (Cipro) plus metronidazole (Flagyl).

Adherence to these guidelines is not optimal. For example, in an Italian multicenter study, 9% of patients with mild acute pancreatitis were treated with antibiotics.¹⁹ Moreover, many patients with proven infected necrosis received antibiotics that do not penetrate the pancreatic tissue very well.

ERCP IN SEVERE BILIARY ACUTE PANCREATITIS

Problem: Endoscopic retrograde cholangiopancreatography (ERCP) often is performed inappropriately in mild biliary acute pancreatitis or is not performed urgently in severe cases.

In most cases of mild biliary pancreatitis, the stones pass spontaneously, as verified by cholangiography done during laparoscopic cholecystectomy. Ongoing ampullary obstruction by impacted biliary stones can perpetuate pancreatic inflammation and delay recovery.

Two early randomized trials showed a benefit from early ERCP (within 72 hours) with sphincterotomy and stone extraction, primarily in those with severe biliary acute pancreatitis or ascending cholangitis,^32,33 but a third trial failed to reveal a benefit.³⁴ A Cochrane metaanalysis of these three trials failed to show a lower death rate with ERCP in mild or severe biliary pancreatitis.³⁵ However, early ERCP did prevent complications in severe biliary pancreatitis (odds ratio 0.27, 95% CI 0.14–0.53).

Later, a fourth randomized trial was restricted to patients with suspected biliary pancreatitis, evidence of biliary obstruction, and no signs of cholangitis³⁶: 103 patients were randomized to undergo either ERCP within 72 hours or conservative management. No difference was observed in rates of death or organ failure or in the CT severity index.

Recommendation: ER CP for suspected retained stones

ERCP has a limited role in patients with biliary pancreatitis, being used to clear retained bile duct stones or to relieve ongoing biliary obstruction.

The decision to perform ERCP before surgery should be based on how strongly one suspects retained stones. ERCP is most appropriate if the suspicion of retained stones and the likelihood of therapeutic intervention are high (eg, if the serum bilirubin and alkaline phosphatase levels are rising and ultrasonography shows a dilated bile duct). If there is moderate suspicion, a safer and less-invasive imaging study such as magnetic resonance cholangiopancreatography (MRCP) or endoscopic ultrasonography can be done to screen for bile duct stones before proceeding to ERCP.

The ACG guidelines suggest urgent ERCP (preferably within 24 hours) for those with severe biliary pancreatitis complicated by organ failure or those with suspicion of cholangitis. The level of evidence is I, ie, “strong evidence from at least one published systematic review of multiple well-designed randomized controlled trials.”¹

Elective ERCP is recommended for those who are poor surgical candidates. ERCP is also recommended for those with rising liver enzyme values or imaging findings suggesting a retained common bile duct stone (including intraoperative cholangiography). Endoscopic ultrasonography or MRCP is recommended for those with slow clinical resolution, who are pregnant, or in whom uncertainty exists regarding the biliary etiology of pancreatitis.

Compliance rates with these and similar guidelines are not adequate. In an audit of adherence to the British Society of Gastroenterology guidelines, early ERCP was performed in only 25% of patients with severe biliary acute pancreatitis.⁶

LAPAROSCOPIC CHOLECYSTECTOMY FOR MILD BILIARY PANCREATITIS

Problem: Laparoscopic cholecystectomy is not done at admission or within 2 weeks in many patients with mild biliary pancreatitis.

If the gallbladder is not removed, biliary pancreatitis may recur in up to 61% of patients within 6 weeks of hospital discharge.³⁷ This is the basis for guideline recommendations for surgery (or a confirmation of a surgery date) prior to hospital discharge.

The International Association of Pancreatology recommends early cholecystectomy (preferably during the same hospitalization) for patients with mild gallstone-associated acute pancreatitis.³⁸ In severe gallstone-associated acute pancreatitis, cholecystectomy should be delayed until there is sufficient resolution of the inflammatory response and clinical recovery. The AGA guidelines advocate cholecystectomy as soon as possible and in no case later than 4 weeks after discharge to prevent relapse. ERCP with biliary sphinc-terotomy may also protect against relapse in those who are not fit to undergo surgery.

Recommendations for definitive management of gallstones (laparoscopic cholecystectomy or ERCP, or both) are not always followed. For example, a British study showed 70% compliance with this recommendation.⁴ A similar compliance audit in Germany revealed that cholecystectomy was performed during the initial hospital stay in only 23% of cases.⁷ In a New Zealand study, a regular compliance audit with feedback to surgeons resulted in an increase in the early cholecystectomy rate from 54% to 80%.⁸

Endoscopic therapy has become an alternative to surgery for some patients with acute recurrent pancreatitis, ie, those whose disease is caused by gallstones or other mechanical processes that can obstruct the outflow from the pancreas.

In this paper, we review the specific situations in which endoscopic therapy might be useful in patients with acute recurrent pancreatitis.

ACUTE PANCREATITIS IS MANAGED DIFFERENTLY IF IT RECURS

Recurrent acute pancreatitis is defined as more than one episode of acute pancreatitis.¹ In clinical practice, it is important to distinguish between the first and recurrent episodes of acute pancreatitis.

Most patients who have one episode of acute pancreatitis never have another one.^2,3 Therefore, for patients having an initial attack, we recommend a limited workup that includes a detailed history, laboratory evaluation, and a noninvasive imaging study such as transcutaneous ultrasonography or computed tomography.

On the other hand, people who have a second attack are at higher risk of more recurrences. Therefore, patients having recurrent attacks need a more extensive workup to determine the underlying cause. We recommend referring them to a gastroenterologist for further evaluation.

WHICH CAUSES CAN BE MANAGED ENDOSCOPICALLY?

In the Western world, 70% to 80% of cases of recurrent pancreatitis are due to either alcohol abuse or gallstone disease.^2,4 The rest are related to:

• Autoimmune disorders
• Cancer, including occult malignancies and premalignant conditions such as intraductal papillary mucinous neoplasm
• Chronic pancreatitis
• Drugs
• Heredity
• Metabolic abnormalities (hypertriglyceridemia, hypercalcemia)
• Sphincter of Oddi dysfunction
• Structural or congenital abnormalities (pancreas divisum)
• Trauma.

• Gallstone disease, including biliary microlithiasis and sludge (in patients with or without a gallbladder)
• Sphincter of Oddi dysfunction
• Pancreas divisum
• Obstruction to flow of pancreatic juice.

Endoscopy is not completely benign

Although endoscopic procedures are less invasive than surgery, they are not completely benign. They can cause anxiety and are associated with risks such as bleeding, perforation, and pancreatitis.⁵ The risks, benefits, and alternatives to these procedures should be discussed with the patient, and informed consent should be obtained before any endoscopic procedure.⁶

STONES (LARGE OR SMALL) OR SLUDGE IN PATIENTS WITH A GALLBLADDER

Gallstones can be large, but small stones (microlithiasis) and sludge are more common and therefore account for more cases of pancreatitis.

Strictly defined, microlithiasis refers to stones smaller than 2 mm in diameter in the biliary tract, whereas sludge is a suspension of biliary crystals, mucin, and cellular debris in the gallbladder or bile ducts.⁷ The terms are often used interchangeably, since the conditions often coexist and their treatment is similar.

Theories differ as to how microlithiasis or sludge can cause recurrent pancreatitis. According to one theory, the debris blocks the common channel, increasing the pancreatic intraductal pressure and leading to pancreatitis.⁸ A second theory is that small stones or biliary crystals passing through the sphincter of Oddi cause inflammation, and that repeated inflammation eventually leads to stenosis or dyskinesia of the sphincter, both of which have been associated with pancreatitis.⁹

Studies suggest that microlithiasis and sludge are common causes of recurrent pancreatitis, accounting for about two-thirds of cases according to estimates by Ros et al¹⁰ and Lee et al.¹¹

Detecting small stones and sludge

The diagnosis of microlithiasis and biliary sludge in patients with a gallbladder is based on imaging studies and bile microscopy.¹²

Transabdominal ultrasonography is the imaging study most often used for diagnosing microlithiasis. The technology and expertise for this test are widely available, and it is relatively inexpensive.

Endoscopic ultrasonography is more sensitive for detecting microlithiasis and can examine the distal common bile duct.

Bile microscopy involves obtaining bile from the second portion of the duodenum (via an endoscope or a duodenal tube) or from the bile ducts (by cannulating the common bile duct and stimulating the gallbladder with cholecystokinin). The bile sample is centrifuged and inspected microscopically under plain light and polarized light (which aids the visualization of biliary crystals). The crystals can be cholesterol monohydrate, calcium bilirubinate, or calcium carbonate.^7,13,14

Removing the gallbladder is the treatment of choice for small stones and sludge

Treatments to prevent recurrent attacks of acute pancreatitis due to microlithiasis and sludge include cholecystectomy, biliary sphincterotomy, and ursodioxycholic acid.^10,11,15

In prospective observational studies by Ros et al¹⁰ and Lee et al,¹¹ about half of the patients with recurrent pancreatitis were treated with cholecystectomy, endoscopic sphincterotomy, or ursodioxycholic acid in a nonrandomized fashion. The choice of therapy was based on the patient’s medical status and the preferences of the patient and the physician. Half the patients received no treatment. In both studies the median follow-up was 4 years. Treated patients had a significantly lower rate of recurrent attacks of pancreatitis during follow-up: less than 20% with therapy compared with more than 60% without therapy. Unfortunately, no published study has compared these three treatments head to head.

Cholecystectomy, however, is the most definitive therapy and is generally considered the treatment of choice.

Biliary sphincterotomy is an endoscopic procedure that involves cutting the sphincter of Oddi to allow the stones and sludge to pass more freely. It is as effective as cholecystectomy in preventing recurrent attacks but does not eliminate the risk of cholecystitis and cholangitis (Figure 1). For this reason, it is usually reserved for patients who cannot tolerate surgery due to comorbidities, those who refuse surgery, or those who are pregnant.¹⁶

Ursodeoxycholic acid is a reasonable alternative in patients who cannot tolerate surgical or endoscopic biliary sphincterotomy.^1,17–20 The dosage is 10 mg/kg/day, which can be in two or three divided doses. The optimal duration of treatment is not known; however, since this drug works slowly, it may need to be taken for 2 years or more. Ursodeoxycholic acid is more effective in patients with cholesterol-based stones and crystals. It is not effective for large stones (> 1 cm in diameter) or calcified stones.

STONES AFTER CHOLECYSTECTOMY

Bile duct stones can be classified as primary or secondary. A primary stone is one that remains where it was formed, whereas a secondary stone is one that has migrated from its site of formation.²¹

Some suggest that bile duct stones that are detected within 2 years of cholecystectomy originated in the gallbladder and were missed when the gallbladder was removed (and therefore are considered secondary stones), and that stones that present more than 2 years after cholecystectomy are de novo (ie, primary) stones.^22,23

In any event, stones have been found in the common bile duct in 4% to 24% of patients up to 15 years after cholecystectomy.^24–26 A fair number of these patients have no symptoms.²⁷ Risk factors for stone recurrence are lithogenic bile (ie, high concentration of cholesterol, low concentration of bile salts), biliary stasis, strictures, dilated bile ducts, and advanced age.^28–30

No role for crystal analysis after cholecystectomy

Biliary crystal analysis does not seem to have diagnostic value in patients with recurrent acute pancreatitis after cholecystectomy,³¹ because removing the gallbladder eliminates the crystals and sludge. Imaging studies are therefore the cornerstone of diagnosis.

Transabdominal ultrasonography is the most commonly used initial imaging test. However, abdominal fat and gas in the duodenum can obscure the distal common bile duct and decrease the sensitivity of this test.³²

Endoscopic ultrasonography involves positioning the transducer in the second part of the duodenum, where it can show the adjacent biliary tree without interference from digestive gas or abdominal fat.

Magnetic resonance cholangiopancreatography (MRCP) and endoscopic ultrasonography are both highly sensitive for detecting common bile duct stones and are recommended if they can be done without delay.

Endoscopic retrograde cholangiopancreatography (ERCP). As a rule, patients who are very likely to have gallstones are best served by proceeding directly to ERCP, a procedure that enables both imaging and treatment. However, ERCP exposes the patient to radiation and the risk of pancreatitis, so in some patients (eg, pregnant women, people who recently had acute pancreatitis), one may want to do ultrasonography first.

ERCP is the treatment of choice after cholecystectomy

The treatment of choice in patients with choledocholithiasis is ERCP with biliary sphincterotomy and stone extraction. Success at clearing the biliary tree of all stones depends on the size, number, and location of the stones, the anatomy of the digestive tract and the bile duct, and the experience of the endoscopist. At specialized centers, the rate of successful clearance with subsequent procedures is close to 100%. Large stones may require fragmentation inside the bile duct to aid their removal.³³

SPHINCTER OF ODDI DYSFUNCTION

The sphincter of Oddi, located where the bile and pancreatic ducts penetrate the wall of the duodenum, actually consists of three sphincters: the common, the biliary, and the pancreatic. Its physiologic role is to regulate the flow of bile and pancreatic juice into the duodenum and to prevent reflux into the ducts from the duodenum.³⁴ Its basal pressure is the main regulating mechanism for pancreatic and biliary secretions into the intestine, and its phasic contractile activity is closely associated with duodenal motility.

Sphincter dysfunction: Stenosis, dyskinesia

The sphincter of Oddi can obstruct the flow of bile and pancreatic juice owing either to stenosis or to dyskinesia.^35,36 Stenosis refers to structural alteration of the sphincter, probably from inflammation and subsequent fibrosis. In contrast, dyskinesia refers to a motor abnormality of the sphincter that makes it hypertonic.

Stenosis or dyskinesia can occur in the biliary sphincter, the pancreatic sphincter, the common sphincter, or any combination of the three. For example, dysfunction of the biliary sphincter can cause abnormalities in liver-associated enzyme levels and biliary-type pain, whereas pancreatic sphincter dysfunction can cause recurrent attacks of pancreatitis and pancreatic-type pain.³⁷ Elevated pancreatic sphincter pressure has been shown to correlate with increased pancreatic ductal pressure, suggesting that the sphincter plays a role in the pathogenesis of acute pancreatitis.^23,38

Sphincter pressure can be measured during ERCP, but ERCP is risky

The gold standard for the diagnosis of sphincter of Oddi dysfunction is manometry,^23,35 ie, direct measurement of sphincter pressure via a thin catheter placed inside the pancreatic or biliary sphincter during ERCP (Figure 1).

However, in patients with suspected sphincter of Oddi dysfunction, ERCP with or without manometry is associated with a high rate of complications, with pancreatitis occurring in up to 25% of cases.^39–41 Therefore, several noninvasive and provocative tests have been designed in an attempt to identify patients with this disorder. Unfortunately, none of them seems to be as sensitive and specific as manometry for diagnosing sphincter of Oddi dysfunction, and so they have not gained widespread use.

Opening the sphincter of Oddi with drugs, endoscopy, or surgery

Drug treatment of sphincter of Oddi dysfunction is based on drugs that relax smooth muscle, such as calcium channel blockers and nitrates. The treatment must be lifelong. Also, it does not improve sphincter stenosis, and only half of patients with sphincter dyskinesia respond to it. For these reasons, drug treatment of sphincter of Oddi dysfunction has not gained widespread acceptance.^36,42

Endoscopic sphincterotomy is the current standard endoscopic therapy for sphincter of Oddi dysfunction. This procedure is performed during ERCP and involves cutting the sphincter with electrocautery.

Endoscopic pancreatic sphincterotomy prevents recurrent attacks of pancreatitis in patients with pancreatic sphincter dysfunction in more than 60% of cases.^23,43–46 A potential complication is pancreatitis, which occurs more often in patients with pancreatic sphincter dyskinesia. Placing a stent in the pancreatic duct after pancreatic sphincterotomy reduces the risk of pancreatitis after ERCP.^37,47,48

Surgery. Pancreatic sphincterotomy can also be done surgically, most commonly via transduodenal pancreatic sphincteroplasty. Surgical sphincteroplasty is as effective as endoscopic sphincterotomy for preventing recurrent attacks of pancreatitis in patients with pancreatic sphincter dysfunction.⁴⁹ However, endoscopic therapy is much less invasive and remains the preferred treatment for sphincter of Oddi dysfunction in most centers with experience in this technique.⁵⁰

PANCREAS DIVISUM

Pancreas divisum is the most common congenital anomaly of the pancreatic duct. Autopsy studies show it occurs in 5% to 10% of the population.^51–53

At approximately the 5th week of gestation, there are two pancreatic buds: a ventral and a dorsal bud. The ventral bud eventually gives rise to part of the pancreatic head and uncinate process of the pancreas in the adult. The dorsal bud eventually gives rise to the rest of the pancreatic head, the pancreatic body, and the pancreatic tail. At 6 to 7 weeks of gestation, the ventral bud rotates clockwise and lies posterior to the dorsal bud. At this stage, both the dorsal and ventral pancreata have their own ducts, which do not communicate with each other. Normally, the ventral and dorsal pancreas and their ducts fuse together at 8 weeks of gestation; in people with pancreas divisum, this ductal fusion does not occur.⁵¹

The pancreas secretes 1.5 L of fluid per day. Normally, 90% to 95% of this volume drains through the major papilla. In people with pancreas divisum, 90% to 95% of the fluid drains through the minor papilla.

People with pancreas divisum are a heterogeneous group. Most have no symptoms, and their ductal anatomy is diagnosed only incidentally. However, a subgroup is prone to develop acute pancreatitis. The cause is thought to be the small diameter of the minor papilla, which poses a relative obstruction to the flow of pancreatic juice.⁵⁴ Direct support for this theory comes from a study in which investigators measured pancreatic ductal pressures in eight people with normal anatomy and six people with pancreas divisum. The pressure in the main pancreatic duct in those with pancreas divisum was significantly higher than in those with normal anatomy.⁵⁵ Additional evidence in favor of this theory is the effectiveness of treatment, which involves widening the minor papillary opening (minor papillary sphincterotomy).

Diagnosis of pancreas divisum

The diagnosis of pancreas divisum is based on imaging studies, and ERCP remains the gold standard for patients with equivocal results on noninvasive imaging. However, MRCP, especially secretin-enhanced MRCP, is as accurate as ERCP. In most cases, MRCP has replaced ERCP for the diagnosis of this condition, although a recent study suggests that MRCP is inferior to ERCP in the diagnosis of pancreas divisum.⁵⁶ We recommend secretin-enhanced MRCP for this purpose.

Computed tomography and endoscopic ultrasonography can also diagnose pancreas divisum, but their diagnostic accuracy is lower than that of ERCP and MRCP.

Minor papillary sphincterotomy

Treating recurrent pancreatitis due to pancreas divisum involves relieving the relative obstruction of the minor papilla by minor papillary sphincterotomy. This can be done surgically or endoscopically (Figure 1).

Surgery. No randomized, controlled study has yet assessed the efficacy of surgical sphincteroplasty for recurrent pancreatitis in patients with pancreas divisum. However, retrospective studies and one prospective study have been published.^57,58

In the retrospective study with the largest number of patients, Warshaw et al⁵⁷ reported their experience in 49 patients who had recurrent pancreatitis due to pancreas divisum. After surgical sphincteroplasty, the patients were followed for a mean of 53 months; 40 (82%) of the 49 patients had no further episodes of acute pancreatitis during this time.

Bradley and Stephan⁵⁸ studied 37 patients with pancreas divisum and recurrent pancreatitis.⁵⁸ After surgical sphincteroplasty, the patients were followed for a mean of 60 months; 31 of the 37 patients had no further attacks, a success rate of 84%.

Endoscopic therapy. As with surgical therapy trials, most trials of endoscopic therapy of recurrent pancreatitis in patients with pancreas divisum are small case series. In a retrospective study with one of the largest number of patients, Heyries et al⁵⁹ reported their experience with 24 patients with pancreas divisum and recurrent pancreatitis. After undergoing endoscopic minor papillary sphincterotomy, all patients were followed for a mean of 39 months, during which 22 (92%) did not have further episodes of acute pancreatitis.

In the only randomized controlled trial available, 19 patients with recurrent pancreatitis and pancreas divisum underwent either no treatment or endoscopic minor papillary sphincterotomy.⁶⁰ In the treatment group, 9 of 10 patients had no further episodes of acute pancreatitis during the 3 years of follow-up, while 6 of 9 patients who were randomized to no treatment had at least one episode.⁶⁰

Although surgical and endoscopic minor papillary sphincterotomy are equally effective, endoscopic therapy is preferred since it is less invasive, is associated with less morbidity, and costs less. It is also more convenient for patients, since it is an outpatient procedure. Surgical treatment is usually reserved for those in whom endoscopic treatment has failed or is not technically possible.

OTHER PROCESSES OBSTRUCTING THE FLOW OF PANCREATIC JUICE

Any process preventing free flow of pancreatic juice can lead to acute pancreatitis. The cause of the blockage can be around the ampulla, in the ampulla, or in the duct.⁶¹

Periampullary lesions, tumors, or polyps can press on the ampulla and cause complete or relative obstruction of the pancreatic duct with a subsequent increase in intraductal pressure and, thus, acute pancreatitis.⁶² Tumors or polyps of the ampulla, such as ampullary adenoma or carcinoma, can cause pancreatitis by directly obstructing the pancreatic duct where it opens into the duodenum.^63–66 Intraductal processes such as ductal adenocarcinoma, intraductal papillary mucinous tumor, pancreatic duct stone, and intraductal stricture due to cancer, chronic pancreatitis, or trauma can also cause pancreatitis by preventing free flow of pancreatic juice.^67–71

Although it is well known that sequelae of severe chronic pancreatitis such as ductal strictures or intraductal stones can lead to recurrent attacks of acute pancreatitis by preventing the free flow of pancreatic juice, a relationship also seems to exist between early chronic pancreatitis and recurrent acute pancreatitis.⁷² Several studies have shown that up to 50% of patients with idiopathic recurrent pancreatitis have evidence of chronic pancreatitis.^72–74 However, it is still unclear whether early chronic pancreatitis is the underlying cause of the recurrent attacks of acute pancreatitis or whether recurrent attacks of acute pancreatitis might have led to the development of chronic pancreatitis.

Diagnosis

Ampullary and periampullary neoplasms can be diagnosed endoscopically. Intraductal lesions such as strictures can be diagnosed by MRCP, especially secretin-enhanced MRCP, or by ERCP. ERCP has the additional advantage of being able to deliver treatment, ie, balloon dilation and stenting. In the case of ductal strictures, upsizing of the stents or placement of multiple stents during subsequent procedures is usually needed. Pancreatic ductal calcifications associated with chronic pancreatitis are usually radiopaque and are easily visible on plain films or computed tomography of the abdomen. Parenchymal and ductal changes of chronic pancreatitis can be diagnosed by endoscopic ultrasonography.

Treatment

The treatment is to relieve the obstruction and re-establish the free flow of pancreatic juice.

Periampullary tumors or polyps can be resected surgically or, if they involve only the mucosa, by endoscopic mucosal resection. Ampullary adenomas can be resected endoscopically. Ampullary carcinomas usually require surgical resection.

Small, nonobstructive stones in the pancreatic duct can be removed during ERCP.⁷⁵ Larger stones may need to be fragmented by extracorporeal shock wave lithotripsy to facilitate removal by ERCP.⁷⁵

Intraductal strictures should raise the suspicion of pancreatic adenocarcinoma, especially in older patients.⁶¹ In these cases, relief of the obstruction by placement of a pancreatic stent can prevent further attacks of pancreatitis until a diagnosis can be established and a more definitive treatment can be offered.

Endoscopic therapy has become an alternative to surgery for some patients with acute recurrent pancreatitis, ie, those whose disease is caused by gallstones or other mechanical processes that can obstruct the outflow from the pancreas.

In this paper, we review the specific situations in which endoscopic therapy might be useful in patients with acute recurrent pancreatitis.

ACUTE PANCREATITIS IS MANAGED DIFFERENTLY IF IT RECURS

Recurrent acute pancreatitis is defined as more than one episode of acute pancreatitis.¹ In clinical practice, it is important to distinguish between the first and recurrent episodes of acute pancreatitis.

Most patients who have one episode of acute pancreatitis never have another one.^2,3 Therefore, for patients having an initial attack, we recommend a limited workup that includes a detailed history, laboratory evaluation, and a noninvasive imaging study such as transcutaneous ultrasonography or computed tomography.

On the other hand, people who have a second attack are at higher risk of more recurrences. Therefore, patients having recurrent attacks need a more extensive workup to determine the underlying cause. We recommend referring them to a gastroenterologist for further evaluation.

WHICH CAUSES CAN BE MANAGED ENDOSCOPICALLY?

In the Western world, 70% to 80% of cases of recurrent pancreatitis are due to either alcohol abuse or gallstone disease.^2,4 The rest are related to:

• Autoimmune disorders
• Cancer, including occult malignancies and premalignant conditions such as intraductal papillary mucinous neoplasm
• Chronic pancreatitis
• Drugs
• Heredity
• Metabolic abnormalities (hypertriglyceridemia, hypercalcemia)
• Sphincter of Oddi dysfunction
• Structural or congenital abnormalities (pancreas divisum)
• Trauma.

• Gallstone disease, including biliary microlithiasis and sludge (in patients with or without a gallbladder)
• Sphincter of Oddi dysfunction
• Pancreas divisum
• Obstruction to flow of pancreatic juice.

Endoscopy is not completely benign

Although endoscopic procedures are less invasive than surgery, they are not completely benign. They can cause anxiety and are associated with risks such as bleeding, perforation, and pancreatitis.⁵ The risks, benefits, and alternatives to these procedures should be discussed with the patient, and informed consent should be obtained before any endoscopic procedure.⁶

STONES (LARGE OR SMALL) OR SLUDGE IN PATIENTS WITH A GALLBLADDER

Gallstones can be large, but small stones (microlithiasis) and sludge are more common and therefore account for more cases of pancreatitis.

Strictly defined, microlithiasis refers to stones smaller than 2 mm in diameter in the biliary tract, whereas sludge is a suspension of biliary crystals, mucin, and cellular debris in the gallbladder or bile ducts.⁷ The terms are often used interchangeably, since the conditions often coexist and their treatment is similar.

Theories differ as to how microlithiasis or sludge can cause recurrent pancreatitis. According to one theory, the debris blocks the common channel, increasing the pancreatic intraductal pressure and leading to pancreatitis.⁸ A second theory is that small stones or biliary crystals passing through the sphincter of Oddi cause inflammation, and that repeated inflammation eventually leads to stenosis or dyskinesia of the sphincter, both of which have been associated with pancreatitis.⁹

Studies suggest that microlithiasis and sludge are common causes of recurrent pancreatitis, accounting for about two-thirds of cases according to estimates by Ros et al¹⁰ and Lee et al.¹¹

Detecting small stones and sludge

The diagnosis of microlithiasis and biliary sludge in patients with a gallbladder is based on imaging studies and bile microscopy.¹²

Transabdominal ultrasonography is the imaging study most often used for diagnosing microlithiasis. The technology and expertise for this test are widely available, and it is relatively inexpensive.

Endoscopic ultrasonography is more sensitive for detecting microlithiasis and can examine the distal common bile duct.

Bile microscopy involves obtaining bile from the second portion of the duodenum (via an endoscope or a duodenal tube) or from the bile ducts (by cannulating the common bile duct and stimulating the gallbladder with cholecystokinin). The bile sample is centrifuged and inspected microscopically under plain light and polarized light (which aids the visualization of biliary crystals). The crystals can be cholesterol monohydrate, calcium bilirubinate, or calcium carbonate.^7,13,14

Removing the gallbladder is the treatment of choice for small stones and sludge

Treatments to prevent recurrent attacks of acute pancreatitis due to microlithiasis and sludge include cholecystectomy, biliary sphincterotomy, and ursodioxycholic acid.^10,11,15

In prospective observational studies by Ros et al¹⁰ and Lee et al,¹¹ about half of the patients with recurrent pancreatitis were treated with cholecystectomy, endoscopic sphincterotomy, or ursodioxycholic acid in a nonrandomized fashion. The choice of therapy was based on the patient’s medical status and the preferences of the patient and the physician. Half the patients received no treatment. In both studies the median follow-up was 4 years. Treated patients had a significantly lower rate of recurrent attacks of pancreatitis during follow-up: less than 20% with therapy compared with more than 60% without therapy. Unfortunately, no published study has compared these three treatments head to head.

Cholecystectomy, however, is the most definitive therapy and is generally considered the treatment of choice.

Biliary sphincterotomy is an endoscopic procedure that involves cutting the sphincter of Oddi to allow the stones and sludge to pass more freely. It is as effective as cholecystectomy in preventing recurrent attacks but does not eliminate the risk of cholecystitis and cholangitis (Figure 1). For this reason, it is usually reserved for patients who cannot tolerate surgery due to comorbidities, those who refuse surgery, or those who are pregnant.¹⁶

Ursodeoxycholic acid is a reasonable alternative in patients who cannot tolerate surgical or endoscopic biliary sphincterotomy.^1,17–20 The dosage is 10 mg/kg/day, which can be in two or three divided doses. The optimal duration of treatment is not known; however, since this drug works slowly, it may need to be taken for 2 years or more. Ursodeoxycholic acid is more effective in patients with cholesterol-based stones and crystals. It is not effective for large stones (> 1 cm in diameter) or calcified stones.

STONES AFTER CHOLECYSTECTOMY

Bile duct stones can be classified as primary or secondary. A primary stone is one that remains where it was formed, whereas a secondary stone is one that has migrated from its site of formation.²¹

Some suggest that bile duct stones that are detected within 2 years of cholecystectomy originated in the gallbladder and were missed when the gallbladder was removed (and therefore are considered secondary stones), and that stones that present more than 2 years after cholecystectomy are de novo (ie, primary) stones.^22,23

In any event, stones have been found in the common bile duct in 4% to 24% of patients up to 15 years after cholecystectomy.^24–26 A fair number of these patients have no symptoms.²⁷ Risk factors for stone recurrence are lithogenic bile (ie, high concentration of cholesterol, low concentration of bile salts), biliary stasis, strictures, dilated bile ducts, and advanced age.^28–30

No role for crystal analysis after cholecystectomy

Biliary crystal analysis does not seem to have diagnostic value in patients with recurrent acute pancreatitis after cholecystectomy,³¹ because removing the gallbladder eliminates the crystals and sludge. Imaging studies are therefore the cornerstone of diagnosis.

Transabdominal ultrasonography is the most commonly used initial imaging test. However, abdominal fat and gas in the duodenum can obscure the distal common bile duct and decrease the sensitivity of this test.³²

Endoscopic ultrasonography involves positioning the transducer in the second part of the duodenum, where it can show the adjacent biliary tree without interference from digestive gas or abdominal fat.

Magnetic resonance cholangiopancreatography (MRCP) and endoscopic ultrasonography are both highly sensitive for detecting common bile duct stones and are recommended if they can be done without delay.

Endoscopic retrograde cholangiopancreatography (ERCP). As a rule, patients who are very likely to have gallstones are best served by proceeding directly to ERCP, a procedure that enables both imaging and treatment. However, ERCP exposes the patient to radiation and the risk of pancreatitis, so in some patients (eg, pregnant women, people who recently had acute pancreatitis), one may want to do ultrasonography first.

ERCP is the treatment of choice after cholecystectomy

The treatment of choice in patients with choledocholithiasis is ERCP with biliary sphincterotomy and stone extraction. Success at clearing the biliary tree of all stones depends on the size, number, and location of the stones, the anatomy of the digestive tract and the bile duct, and the experience of the endoscopist. At specialized centers, the rate of successful clearance with subsequent procedures is close to 100%. Large stones may require fragmentation inside the bile duct to aid their removal.³³

SPHINCTER OF ODDI DYSFUNCTION

The sphincter of Oddi, located where the bile and pancreatic ducts penetrate the wall of the duodenum, actually consists of three sphincters: the common, the biliary, and the pancreatic. Its physiologic role is to regulate the flow of bile and pancreatic juice into the duodenum and to prevent reflux into the ducts from the duodenum.³⁴ Its basal pressure is the main regulating mechanism for pancreatic and biliary secretions into the intestine, and its phasic contractile activity is closely associated with duodenal motility.

Sphincter dysfunction: Stenosis, dyskinesia

The sphincter of Oddi can obstruct the flow of bile and pancreatic juice owing either to stenosis or to dyskinesia.^35,36 Stenosis refers to structural alteration of the sphincter, probably from inflammation and subsequent fibrosis. In contrast, dyskinesia refers to a motor abnormality of the sphincter that makes it hypertonic.

Stenosis or dyskinesia can occur in the biliary sphincter, the pancreatic sphincter, the common sphincter, or any combination of the three. For example, dysfunction of the biliary sphincter can cause abnormalities in liver-associated enzyme levels and biliary-type pain, whereas pancreatic sphincter dysfunction can cause recurrent attacks of pancreatitis and pancreatic-type pain.³⁷ Elevated pancreatic sphincter pressure has been shown to correlate with increased pancreatic ductal pressure, suggesting that the sphincter plays a role in the pathogenesis of acute pancreatitis.^23,38

Sphincter pressure can be measured during ERCP, but ERCP is risky

The gold standard for the diagnosis of sphincter of Oddi dysfunction is manometry,^23,35 ie, direct measurement of sphincter pressure via a thin catheter placed inside the pancreatic or biliary sphincter during ERCP (Figure 1).

However, in patients with suspected sphincter of Oddi dysfunction, ERCP with or without manometry is associated with a high rate of complications, with pancreatitis occurring in up to 25% of cases.^39–41 Therefore, several noninvasive and provocative tests have been designed in an attempt to identify patients with this disorder. Unfortunately, none of them seems to be as sensitive and specific as manometry for diagnosing sphincter of Oddi dysfunction, and so they have not gained widespread use.

Opening the sphincter of Oddi with drugs, endoscopy, or surgery

Drug treatment of sphincter of Oddi dysfunction is based on drugs that relax smooth muscle, such as calcium channel blockers and nitrates. The treatment must be lifelong. Also, it does not improve sphincter stenosis, and only half of patients with sphincter dyskinesia respond to it. For these reasons, drug treatment of sphincter of Oddi dysfunction has not gained widespread acceptance.^36,42

Endoscopic sphincterotomy is the current standard endoscopic therapy for sphincter of Oddi dysfunction. This procedure is performed during ERCP and involves cutting the sphincter with electrocautery.

Endoscopic pancreatic sphincterotomy prevents recurrent attacks of pancreatitis in patients with pancreatic sphincter dysfunction in more than 60% of cases.^23,43–46 A potential complication is pancreatitis, which occurs more often in patients with pancreatic sphincter dyskinesia. Placing a stent in the pancreatic duct after pancreatic sphincterotomy reduces the risk of pancreatitis after ERCP.^37,47,48

Surgery. Pancreatic sphincterotomy can also be done surgically, most commonly via transduodenal pancreatic sphincteroplasty. Surgical sphincteroplasty is as effective as endoscopic sphincterotomy for preventing recurrent attacks of pancreatitis in patients with pancreatic sphincter dysfunction.⁴⁹ However, endoscopic therapy is much less invasive and remains the preferred treatment for sphincter of Oddi dysfunction in most centers with experience in this technique.⁵⁰

PANCREAS DIVISUM

Pancreas divisum is the most common congenital anomaly of the pancreatic duct. Autopsy studies show it occurs in 5% to 10% of the population.^51–53

At approximately the 5th week of gestation, there are two pancreatic buds: a ventral and a dorsal bud. The ventral bud eventually gives rise to part of the pancreatic head and uncinate process of the pancreas in the adult. The dorsal bud eventually gives rise to the rest of the pancreatic head, the pancreatic body, and the pancreatic tail. At 6 to 7 weeks of gestation, the ventral bud rotates clockwise and lies posterior to the dorsal bud. At this stage, both the dorsal and ventral pancreata have their own ducts, which do not communicate with each other. Normally, the ventral and dorsal pancreas and their ducts fuse together at 8 weeks of gestation; in people with pancreas divisum, this ductal fusion does not occur.⁵¹

The pancreas secretes 1.5 L of fluid per day. Normally, 90% to 95% of this volume drains through the major papilla. In people with pancreas divisum, 90% to 95% of the fluid drains through the minor papilla.

People with pancreas divisum are a heterogeneous group. Most have no symptoms, and their ductal anatomy is diagnosed only incidentally. However, a subgroup is prone to develop acute pancreatitis. The cause is thought to be the small diameter of the minor papilla, which poses a relative obstruction to the flow of pancreatic juice.⁵⁴ Direct support for this theory comes from a study in which investigators measured pancreatic ductal pressures in eight people with normal anatomy and six people with pancreas divisum. The pressure in the main pancreatic duct in those with pancreas divisum was significantly higher than in those with normal anatomy.⁵⁵ Additional evidence in favor of this theory is the effectiveness of treatment, which involves widening the minor papillary opening (minor papillary sphincterotomy).

Diagnosis of pancreas divisum

The diagnosis of pancreas divisum is based on imaging studies, and ERCP remains the gold standard for patients with equivocal results on noninvasive imaging. However, MRCP, especially secretin-enhanced MRCP, is as accurate as ERCP. In most cases, MRCP has replaced ERCP for the diagnosis of this condition, although a recent study suggests that MRCP is inferior to ERCP in the diagnosis of pancreas divisum.⁵⁶ We recommend secretin-enhanced MRCP for this purpose.

Computed tomography and endoscopic ultrasonography can also diagnose pancreas divisum, but their diagnostic accuracy is lower than that of ERCP and MRCP.

Minor papillary sphincterotomy

Treating recurrent pancreatitis due to pancreas divisum involves relieving the relative obstruction of the minor papilla by minor papillary sphincterotomy. This can be done surgically or endoscopically (Figure 1).

Surgery. No randomized, controlled study has yet assessed the efficacy of surgical sphincteroplasty for recurrent pancreatitis in patients with pancreas divisum. However, retrospective studies and one prospective study have been published.^57,58

In the retrospective study with the largest number of patients, Warshaw et al⁵⁷ reported their experience in 49 patients who had recurrent pancreatitis due to pancreas divisum. After surgical sphincteroplasty, the patients were followed for a mean of 53 months; 40 (82%) of the 49 patients had no further episodes of acute pancreatitis during this time.

Bradley and Stephan⁵⁸ studied 37 patients with pancreas divisum and recurrent pancreatitis.⁵⁸ After surgical sphincteroplasty, the patients were followed for a mean of 60 months; 31 of the 37 patients had no further attacks, a success rate of 84%.

Endoscopic therapy. As with surgical therapy trials, most trials of endoscopic therapy of recurrent pancreatitis in patients with pancreas divisum are small case series. In a retrospective study with one of the largest number of patients, Heyries et al⁵⁹ reported their experience with 24 patients with pancreas divisum and recurrent pancreatitis. After undergoing endoscopic minor papillary sphincterotomy, all patients were followed for a mean of 39 months, during which 22 (92%) did not have further episodes of acute pancreatitis.

In the only randomized controlled trial available, 19 patients with recurrent pancreatitis and pancreas divisum underwent either no treatment or endoscopic minor papillary sphincterotomy.⁶⁰ In the treatment group, 9 of 10 patients had no further episodes of acute pancreatitis during the 3 years of follow-up, while 6 of 9 patients who were randomized to no treatment had at least one episode.⁶⁰

Although surgical and endoscopic minor papillary sphincterotomy are equally effective, endoscopic therapy is preferred since it is less invasive, is associated with less morbidity, and costs less. It is also more convenient for patients, since it is an outpatient procedure. Surgical treatment is usually reserved for those in whom endoscopic treatment has failed or is not technically possible.

OTHER PROCESSES OBSTRUCTING THE FLOW OF PANCREATIC JUICE

Any process preventing free flow of pancreatic juice can lead to acute pancreatitis. The cause of the blockage can be around the ampulla, in the ampulla, or in the duct.⁶¹

Periampullary lesions, tumors, or polyps can press on the ampulla and cause complete or relative obstruction of the pancreatic duct with a subsequent increase in intraductal pressure and, thus, acute pancreatitis.⁶² Tumors or polyps of the ampulla, such as ampullary adenoma or carcinoma, can cause pancreatitis by directly obstructing the pancreatic duct where it opens into the duodenum.^63–66 Intraductal processes such as ductal adenocarcinoma, intraductal papillary mucinous tumor, pancreatic duct stone, and intraductal stricture due to cancer, chronic pancreatitis, or trauma can also cause pancreatitis by preventing free flow of pancreatic juice.^67–71

Although it is well known that sequelae of severe chronic pancreatitis such as ductal strictures or intraductal stones can lead to recurrent attacks of acute pancreatitis by preventing the free flow of pancreatic juice, a relationship also seems to exist between early chronic pancreatitis and recurrent acute pancreatitis.⁷² Several studies have shown that up to 50% of patients with idiopathic recurrent pancreatitis have evidence of chronic pancreatitis.^72–74 However, it is still unclear whether early chronic pancreatitis is the underlying cause of the recurrent attacks of acute pancreatitis or whether recurrent attacks of acute pancreatitis might have led to the development of chronic pancreatitis.

Diagnosis

Ampullary and periampullary neoplasms can be diagnosed endoscopically. Intraductal lesions such as strictures can be diagnosed by MRCP, especially secretin-enhanced MRCP, or by ERCP. ERCP has the additional advantage of being able to deliver treatment, ie, balloon dilation and stenting. In the case of ductal strictures, upsizing of the stents or placement of multiple stents during subsequent procedures is usually needed. Pancreatic ductal calcifications associated with chronic pancreatitis are usually radiopaque and are easily visible on plain films or computed tomography of the abdomen. Parenchymal and ductal changes of chronic pancreatitis can be diagnosed by endoscopic ultrasonography.

Treatment

The treatment is to relieve the obstruction and re-establish the free flow of pancreatic juice.

Periampullary tumors or polyps can be resected surgically or, if they involve only the mucosa, by endoscopic mucosal resection. Ampullary adenomas can be resected endoscopically. Ampullary carcinomas usually require surgical resection.

Small, nonobstructive stones in the pancreatic duct can be removed during ERCP.⁷⁵ Larger stones may need to be fragmented by extracorporeal shock wave lithotripsy to facilitate removal by ERCP.⁷⁵

Intraductal strictures should raise the suspicion of pancreatic adenocarcinoma, especially in older patients.⁶¹ In these cases, relief of the obstruction by placement of a pancreatic stent can prevent further attacks of pancreatitis until a diagnosis can be established and a more definitive treatment can be offered.

Endoscopic therapy has become an alternative to surgery for some patients with acute recurrent pancreatitis, ie, those whose disease is caused by gallstones or other mechanical processes that can obstruct the outflow from the pancreas.

In this paper, we review the specific situations in which endoscopic therapy might be useful in patients with acute recurrent pancreatitis.

ACUTE PANCREATITIS IS MANAGED DIFFERENTLY IF IT RECURS

Recurrent acute pancreatitis is defined as more than one episode of acute pancreatitis.¹ In clinical practice, it is important to distinguish between the first and recurrent episodes of acute pancreatitis.

Most patients who have one episode of acute pancreatitis never have another one.^2,3 Therefore, for patients having an initial attack, we recommend a limited workup that includes a detailed history, laboratory evaluation, and a noninvasive imaging study such as transcutaneous ultrasonography or computed tomography.

On the other hand, people who have a second attack are at higher risk of more recurrences. Therefore, patients having recurrent attacks need a more extensive workup to determine the underlying cause. We recommend referring them to a gastroenterologist for further evaluation.

WHICH CAUSES CAN BE MANAGED ENDOSCOPICALLY?

In the Western world, 70% to 80% of cases of recurrent pancreatitis are due to either alcohol abuse or gallstone disease.^2,4 The rest are related to:

• Autoimmune disorders
• Cancer, including occult malignancies and premalignant conditions such as intraductal papillary mucinous neoplasm
• Chronic pancreatitis
• Drugs
• Heredity
• Metabolic abnormalities (hypertriglyceridemia, hypercalcemia)
• Sphincter of Oddi dysfunction
• Structural or congenital abnormalities (pancreas divisum)
• Trauma.

• Gallstone disease, including biliary microlithiasis and sludge (in patients with or without a gallbladder)
• Sphincter of Oddi dysfunction
• Pancreas divisum
• Obstruction to flow of pancreatic juice.

Endoscopy is not completely benign

Although endoscopic procedures are less invasive than surgery, they are not completely benign. They can cause anxiety and are associated with risks such as bleeding, perforation, and pancreatitis.⁵ The risks, benefits, and alternatives to these procedures should be discussed with the patient, and informed consent should be obtained before any endoscopic procedure.⁶

STONES (LARGE OR SMALL) OR SLUDGE IN PATIENTS WITH A GALLBLADDER

Gallstones can be large, but small stones (microlithiasis) and sludge are more common and therefore account for more cases of pancreatitis.

Strictly defined, microlithiasis refers to stones smaller than 2 mm in diameter in the biliary tract, whereas sludge is a suspension of biliary crystals, mucin, and cellular debris in the gallbladder or bile ducts.⁷ The terms are often used interchangeably, since the conditions often coexist and their treatment is similar.

Theories differ as to how microlithiasis or sludge can cause recurrent pancreatitis. According to one theory, the debris blocks the common channel, increasing the pancreatic intraductal pressure and leading to pancreatitis.⁸ A second theory is that small stones or biliary crystals passing through the sphincter of Oddi cause inflammation, and that repeated inflammation eventually leads to stenosis or dyskinesia of the sphincter, both of which have been associated with pancreatitis.⁹

Studies suggest that microlithiasis and sludge are common causes of recurrent pancreatitis, accounting for about two-thirds of cases according to estimates by Ros et al¹⁰ and Lee et al.¹¹

Detecting small stones and sludge

The diagnosis of microlithiasis and biliary sludge in patients with a gallbladder is based on imaging studies and bile microscopy.¹²

Transabdominal ultrasonography is the imaging study most often used for diagnosing microlithiasis. The technology and expertise for this test are widely available, and it is relatively inexpensive.

Endoscopic ultrasonography is more sensitive for detecting microlithiasis and can examine the distal common bile duct.

Bile microscopy involves obtaining bile from the second portion of the duodenum (via an endoscope or a duodenal tube) or from the bile ducts (by cannulating the common bile duct and stimulating the gallbladder with cholecystokinin). The bile sample is centrifuged and inspected microscopically under plain light and polarized light (which aids the visualization of biliary crystals). The crystals can be cholesterol monohydrate, calcium bilirubinate, or calcium carbonate.^7,13,14

Removing the gallbladder is the treatment of choice for small stones and sludge

Treatments to prevent recurrent attacks of acute pancreatitis due to microlithiasis and sludge include cholecystectomy, biliary sphincterotomy, and ursodioxycholic acid.^10,11,15

In prospective observational studies by Ros et al¹⁰ and Lee et al,¹¹ about half of the patients with recurrent pancreatitis were treated with cholecystectomy, endoscopic sphincterotomy, or ursodioxycholic acid in a nonrandomized fashion. The choice of therapy was based on the patient’s medical status and the preferences of the patient and the physician. Half the patients received no treatment. In both studies the median follow-up was 4 years. Treated patients had a significantly lower rate of recurrent attacks of pancreatitis during follow-up: less than 20% with therapy compared with more than 60% without therapy. Unfortunately, no published study has compared these three treatments head to head.

Cholecystectomy, however, is the most definitive therapy and is generally considered the treatment of choice.

Biliary sphincterotomy is an endoscopic procedure that involves cutting the sphincter of Oddi to allow the stones and sludge to pass more freely. It is as effective as cholecystectomy in preventing recurrent attacks but does not eliminate the risk of cholecystitis and cholangitis (Figure 1). For this reason, it is usually reserved for patients who cannot tolerate surgery due to comorbidities, those who refuse surgery, or those who are pregnant.¹⁶

Ursodeoxycholic acid is a reasonable alternative in patients who cannot tolerate surgical or endoscopic biliary sphincterotomy.^1,17–20 The dosage is 10 mg/kg/day, which can be in two or three divided doses. The optimal duration of treatment is not known; however, since this drug works slowly, it may need to be taken for 2 years or more. Ursodeoxycholic acid is more effective in patients with cholesterol-based stones and crystals. It is not effective for large stones (> 1 cm in diameter) or calcified stones.

STONES AFTER CHOLECYSTECTOMY

Bile duct stones can be classified as primary or secondary. A primary stone is one that remains where it was formed, whereas a secondary stone is one that has migrated from its site of formation.²¹

Some suggest that bile duct stones that are detected within 2 years of cholecystectomy originated in the gallbladder and were missed when the gallbladder was removed (and therefore are considered secondary stones), and that stones that present more than 2 years after cholecystectomy are de novo (ie, primary) stones.^22,23

In any event, stones have been found in the common bile duct in 4% to 24% of patients up to 15 years after cholecystectomy.^24–26 A fair number of these patients have no symptoms.²⁷ Risk factors for stone recurrence are lithogenic bile (ie, high concentration of cholesterol, low concentration of bile salts), biliary stasis, strictures, dilated bile ducts, and advanced age.^28–30

No role for crystal analysis after cholecystectomy

Biliary crystal analysis does not seem to have diagnostic value in patients with recurrent acute pancreatitis after cholecystectomy,³¹ because removing the gallbladder eliminates the crystals and sludge. Imaging studies are therefore the cornerstone of diagnosis.

Transabdominal ultrasonography is the most commonly used initial imaging test. However, abdominal fat and gas in the duodenum can obscure the distal common bile duct and decrease the sensitivity of this test.³²

Endoscopic ultrasonography involves positioning the transducer in the second part of the duodenum, where it can show the adjacent biliary tree without interference from digestive gas or abdominal fat.

Magnetic resonance cholangiopancreatography (MRCP) and endoscopic ultrasonography are both highly sensitive for detecting common bile duct stones and are recommended if they can be done without delay.

Endoscopic retrograde cholangiopancreatography (ERCP). As a rule, patients who are very likely to have gallstones are best served by proceeding directly to ERCP, a procedure that enables both imaging and treatment. However, ERCP exposes the patient to radiation and the risk of pancreatitis, so in some patients (eg, pregnant women, people who recently had acute pancreatitis), one may want to do ultrasonography first.

ERCP is the treatment of choice after cholecystectomy

The treatment of choice in patients with choledocholithiasis is ERCP with biliary sphincterotomy and stone extraction. Success at clearing the biliary tree of all stones depends on the size, number, and location of the stones, the anatomy of the digestive tract and the bile duct, and the experience of the endoscopist. At specialized centers, the rate of successful clearance with subsequent procedures is close to 100%. Large stones may require fragmentation inside the bile duct to aid their removal.³³

SPHINCTER OF ODDI DYSFUNCTION

The sphincter of Oddi, located where the bile and pancreatic ducts penetrate the wall of the duodenum, actually consists of three sphincters: the common, the biliary, and the pancreatic. Its physiologic role is to regulate the flow of bile and pancreatic juice into the duodenum and to prevent reflux into the ducts from the duodenum.³⁴ Its basal pressure is the main regulating mechanism for pancreatic and biliary secretions into the intestine, and its phasic contractile activity is closely associated with duodenal motility.

Sphincter dysfunction: Stenosis, dyskinesia

The sphincter of Oddi can obstruct the flow of bile and pancreatic juice owing either to stenosis or to dyskinesia.^35,36 Stenosis refers to structural alteration of the sphincter, probably from inflammation and subsequent fibrosis. In contrast, dyskinesia refers to a motor abnormality of the sphincter that makes it hypertonic.

Stenosis or dyskinesia can occur in the biliary sphincter, the pancreatic sphincter, the common sphincter, or any combination of the three. For example, dysfunction of the biliary sphincter can cause abnormalities in liver-associated enzyme levels and biliary-type pain, whereas pancreatic sphincter dysfunction can cause recurrent attacks of pancreatitis and pancreatic-type pain.³⁷ Elevated pancreatic sphincter pressure has been shown to correlate with increased pancreatic ductal pressure, suggesting that the sphincter plays a role in the pathogenesis of acute pancreatitis.^23,38

Sphincter pressure can be measured during ERCP, but ERCP is risky

The gold standard for the diagnosis of sphincter of Oddi dysfunction is manometry,^23,35 ie, direct measurement of sphincter pressure via a thin catheter placed inside the pancreatic or biliary sphincter during ERCP (Figure 1).

However, in patients with suspected sphincter of Oddi dysfunction, ERCP with or without manometry is associated with a high rate of complications, with pancreatitis occurring in up to 25% of cases.^39–41 Therefore, several noninvasive and provocative tests have been designed in an attempt to identify patients with this disorder. Unfortunately, none of them seems to be as sensitive and specific as manometry for diagnosing sphincter of Oddi dysfunction, and so they have not gained widespread use.

Opening the sphincter of Oddi with drugs, endoscopy, or surgery

Drug treatment of sphincter of Oddi dysfunction is based on drugs that relax smooth muscle, such as calcium channel blockers and nitrates. The treatment must be lifelong. Also, it does not improve sphincter stenosis, and only half of patients with sphincter dyskinesia respond to it. For these reasons, drug treatment of sphincter of Oddi dysfunction has not gained widespread acceptance.^36,42

Endoscopic sphincterotomy is the current standard endoscopic therapy for sphincter of Oddi dysfunction. This procedure is performed during ERCP and involves cutting the sphincter with electrocautery.

Endoscopic pancreatic sphincterotomy prevents recurrent attacks of pancreatitis in patients with pancreatic sphincter dysfunction in more than 60% of cases.^23,43–46 A potential complication is pancreatitis, which occurs more often in patients with pancreatic sphincter dyskinesia. Placing a stent in the pancreatic duct after pancreatic sphincterotomy reduces the risk of pancreatitis after ERCP.^37,47,48

Surgery. Pancreatic sphincterotomy can also be done surgically, most commonly via transduodenal pancreatic sphincteroplasty. Surgical sphincteroplasty is as effective as endoscopic sphincterotomy for preventing recurrent attacks of pancreatitis in patients with pancreatic sphincter dysfunction.⁴⁹ However, endoscopic therapy is much less invasive and remains the preferred treatment for sphincter of Oddi dysfunction in most centers with experience in this technique.⁵⁰

PANCREAS DIVISUM

Pancreas divisum is the most common congenital anomaly of the pancreatic duct. Autopsy studies show it occurs in 5% to 10% of the population.^51–53

At approximately the 5th week of gestation, there are two pancreatic buds: a ventral and a dorsal bud. The ventral bud eventually gives rise to part of the pancreatic head and uncinate process of the pancreas in the adult. The dorsal bud eventually gives rise to the rest of the pancreatic head, the pancreatic body, and the pancreatic tail. At 6 to 7 weeks of gestation, the ventral bud rotates clockwise and lies posterior to the dorsal bud. At this stage, both the dorsal and ventral pancreata have their own ducts, which do not communicate with each other. Normally, the ventral and dorsal pancreas and their ducts fuse together at 8 weeks of gestation; in people with pancreas divisum, this ductal fusion does not occur.⁵¹

The pancreas secretes 1.5 L of fluid per day. Normally, 90% to 95% of this volume drains through the major papilla. In people with pancreas divisum, 90% to 95% of the fluid drains through the minor papilla.

People with pancreas divisum are a heterogeneous group. Most have no symptoms, and their ductal anatomy is diagnosed only incidentally. However, a subgroup is prone to develop acute pancreatitis. The cause is thought to be the small diameter of the minor papilla, which poses a relative obstruction to the flow of pancreatic juice.⁵⁴ Direct support for this theory comes from a study in which investigators measured pancreatic ductal pressures in eight people with normal anatomy and six people with pancreas divisum. The pressure in the main pancreatic duct in those with pancreas divisum was significantly higher than in those with normal anatomy.⁵⁵ Additional evidence in favor of this theory is the effectiveness of treatment, which involves widening the minor papillary opening (minor papillary sphincterotomy).

Diagnosis of pancreas divisum

The diagnosis of pancreas divisum is based on imaging studies, and ERCP remains the gold standard for patients with equivocal results on noninvasive imaging. However, MRCP, especially secretin-enhanced MRCP, is as accurate as ERCP. In most cases, MRCP has replaced ERCP for the diagnosis of this condition, although a recent study suggests that MRCP is inferior to ERCP in the diagnosis of pancreas divisum.⁵⁶ We recommend secretin-enhanced MRCP for this purpose.

Computed tomography and endoscopic ultrasonography can also diagnose pancreas divisum, but their diagnostic accuracy is lower than that of ERCP and MRCP.

Minor papillary sphincterotomy

Treating recurrent pancreatitis due to pancreas divisum involves relieving the relative obstruction of the minor papilla by minor papillary sphincterotomy. This can be done surgically or endoscopically (Figure 1).

Surgery. No randomized, controlled study has yet assessed the efficacy of surgical sphincteroplasty for recurrent pancreatitis in patients with pancreas divisum. However, retrospective studies and one prospective study have been published.^57,58

In the retrospective study with the largest number of patients, Warshaw et al⁵⁷ reported their experience in 49 patients who had recurrent pancreatitis due to pancreas divisum. After surgical sphincteroplasty, the patients were followed for a mean of 53 months; 40 (82%) of the 49 patients had no further episodes of acute pancreatitis during this time.

Bradley and Stephan⁵⁸ studied 37 patients with pancreas divisum and recurrent pancreatitis.⁵⁸ After surgical sphincteroplasty, the patients were followed for a mean of 60 months; 31 of the 37 patients had no further attacks, a success rate of 84%.

Endoscopic therapy. As with surgical therapy trials, most trials of endoscopic therapy of recurrent pancreatitis in patients with pancreas divisum are small case series. In a retrospective study with one of the largest number of patients, Heyries et al⁵⁹ reported their experience with 24 patients with pancreas divisum and recurrent pancreatitis. After undergoing endoscopic minor papillary sphincterotomy, all patients were followed for a mean of 39 months, during which 22 (92%) did not have further episodes of acute pancreatitis.

In the only randomized controlled trial available, 19 patients with recurrent pancreatitis and pancreas divisum underwent either no treatment or endoscopic minor papillary sphincterotomy.⁶⁰ In the treatment group, 9 of 10 patients had no further episodes of acute pancreatitis during the 3 years of follow-up, while 6 of 9 patients who were randomized to no treatment had at least one episode.⁶⁰

Although surgical and endoscopic minor papillary sphincterotomy are equally effective, endoscopic therapy is preferred since it is less invasive, is associated with less morbidity, and costs less. It is also more convenient for patients, since it is an outpatient procedure. Surgical treatment is usually reserved for those in whom endoscopic treatment has failed or is not technically possible.

OTHER PROCESSES OBSTRUCTING THE FLOW OF PANCREATIC JUICE

Any process preventing free flow of pancreatic juice can lead to acute pancreatitis. The cause of the blockage can be around the ampulla, in the ampulla, or in the duct.⁶¹

Periampullary lesions, tumors, or polyps can press on the ampulla and cause complete or relative obstruction of the pancreatic duct with a subsequent increase in intraductal pressure and, thus, acute pancreatitis.⁶² Tumors or polyps of the ampulla, such as ampullary adenoma or carcinoma, can cause pancreatitis by directly obstructing the pancreatic duct where it opens into the duodenum.^63–66 Intraductal processes such as ductal adenocarcinoma, intraductal papillary mucinous tumor, pancreatic duct stone, and intraductal stricture due to cancer, chronic pancreatitis, or trauma can also cause pancreatitis by preventing free flow of pancreatic juice.^67–71

Although it is well known that sequelae of severe chronic pancreatitis such as ductal strictures or intraductal stones can lead to recurrent attacks of acute pancreatitis by preventing the free flow of pancreatic juice, a relationship also seems to exist between early chronic pancreatitis and recurrent acute pancreatitis.⁷² Several studies have shown that up to 50% of patients with idiopathic recurrent pancreatitis have evidence of chronic pancreatitis.^72–74 However, it is still unclear whether early chronic pancreatitis is the underlying cause of the recurrent attacks of acute pancreatitis or whether recurrent attacks of acute pancreatitis might have led to the development of chronic pancreatitis.

Diagnosis

Ampullary and periampullary neoplasms can be diagnosed endoscopically. Intraductal lesions such as strictures can be diagnosed by MRCP, especially secretin-enhanced MRCP, or by ERCP. ERCP has the additional advantage of being able to deliver treatment, ie, balloon dilation and stenting. In the case of ductal strictures, upsizing of the stents or placement of multiple stents during subsequent procedures is usually needed. Pancreatic ductal calcifications associated with chronic pancreatitis are usually radiopaque and are easily visible on plain films or computed tomography of the abdomen. Parenchymal and ductal changes of chronic pancreatitis can be diagnosed by endoscopic ultrasonography.

Treatment

The treatment is to relieve the obstruction and re-establish the free flow of pancreatic juice.

Periampullary tumors or polyps can be resected surgically or, if they involve only the mucosa, by endoscopic mucosal resection. Ampullary adenomas can be resected endoscopically. Ampullary carcinomas usually require surgical resection.

Small, nonobstructive stones in the pancreatic duct can be removed during ERCP.⁷⁵ Larger stones may need to be fragmented by extracorporeal shock wave lithotripsy to facilitate removal by ERCP.⁷⁵

Intraductal strictures should raise the suspicion of pancreatic adenocarcinoma, especially in older patients.⁶¹ In these cases, relief of the obstruction by placement of a pancreatic stent can prevent further attacks of pancreatitis until a diagnosis can be established and a more definitive treatment can be offered.