At a 1992 international consensus conference in Atlanta, expert opinion was used to establish uniform terminology and a clinically based classification system for acute pancreatitis and its complications [19]. According to the Atlanta classification, the term pancreatic necrosis refers to diffuse or focal areas of nonviable pancreatic parenchyma typically associated with peripancreatic fat necrosis (Table 1). The diagnosis requires contrast-enhanced imaging for confirmation and, by definition, must involve approximately 30% or more of pancreatic parenchyma. The necrosis may be focal or diffuse but rarely involves the entire gland. In the Atlanta classification, pancreatic necrosis is specifically distinguished from other complications of severe acute necrotizing pancreatitis, such as pseudocyst, pancreatic abscess (which occurs in close proximity to the pancreas), and hemorrhagic pancreatitis. Consensus regarding the precise usage of these terms allows for clearer documentation of natural history and more accurate comparison of the results of therapy across different centers. However, in practice, these distinctions often fail to capture the unique context of an individual patient of which there is often more ambiguity in the clinical presentation than these discrete terminologies otherwise imply. Although the Atlanta classification currently remains the standard system for terminology in acute pancreatitis, its application in reported series has been less than uniform, and several groups have suggested modifications that consider the presence of local or systemic complications and the number and type of remote organ system failures [20–22].

Acute pancreatitis has many causes, most commonly alcohol abuse and gallstones, each accounting for approximately 35% of cases [23]. Less-common risk factors include hypertriglyceridemia, hypercalcemia, post–endoscopic retrograde cholangiopancreatography pancreatitis, genetic factors, and anatomic anomalies such as pancreatic divisum [23]. Irrespective of the underlying cause, the development of pancreatitis is thought to derive from inappropriate activation of trypsin and other enzymes within the pancreatic acinar cells and the surrounding parenchyma [24]. Normally, the pancreatic acinar cell synthesizes and stores digestive enzymes in precursor forms within zymogen granules along with natural protease inhibitors, such as pancreatic secretory trypsin inhibitor. Excessive pancreatic stimulation or other insults may, in predisposed individuals, overwhelm or disrupt the normal mechanisms that prevent pancreatic autodigestion [24]. In particular, unregulated activation of enzymes, such as trypsin and elastase, may trigger parenchymal proteolytic damage, leading to influx of inflammatory cells and production of soluble proinflammatory mediators, such as interleukin (IL) 1 and tumor necrosis factor α [25–27].

Necrosis in acute pancreatitis is thought to be multifactorial. Microvascular damage may be induced by reactive oxygen species and other noxious agents [28], as well as compromised tissue oxygen delivery mediated by the vasoactive effects of endothelin and the inhibition of bradykinin [28]. Underresuscitation of patients during the early phases of acute pancreatitis is thought to lead to pancreatic hypoperfusion as the consequence of mesenteric vasoconstriction. It has been reported that the development of pancreatic necrosis is significantly more likely in patients who show early evidence of hemoconcentration, indicating inadequate replenishment of pancreatitis-associated fluid losses due to vomiting and third-space losses [29]. Phospholipase A2 has been implicated as a potential prognostic marker and an important mediator of injury [30]. Necrosis of the peripancreatic vasculature may lead to hemorrhage. Significant bleeding can occur if there is erosion into one of the larger vessels, most commonly involved vessels include the splenic and gastroduodenal arteries or the portal vein [31,32]. These patients often present with massive hematemesis or intraperitoneal hemorrhage and may require urgent embolization or surgical intervention with high-attendant mortality of approximately 60% [31,33].

Infection of pancreatic necrosis is thought to develop via several mechanisms. Systemic and intraperitoneal sites of inflammation have been shown experimentally to induce translocation of intestinal microbes that are then postulated to directly extend into or otherwise seed foci of necrosis [34]. Mesenteric hypoperfusion and retroperitoneal inflammation lead to impaired intestinal motility and consequent bacterial overgrowth, as well as mucosal barrier disruption and an overall repression of host immune responses [35]. Secondary seeding of pancreatic necrosis may derive from remote sites of bacterial colonization or infection, such as central intravascular lines or indwelling urinary catheters. The most common bacteria isolated from the pancreas are gram-negative enteric organisms, such as Escherichia coli and Proteus mirabilis [36]. However, gram-positive cocci are frequently isolated as well [37], suggesting nonenteric sources, including contamination during diagnostic needle aspiration, may also be contributory.

Whether antibiotic prophylaxis is effective in the prevention of pancreatic infection and reduction in mortality in the setting of severe acute pancreatitis remains somewhat controversial. Although the experimental and theoretical rationale for antibiotic prophylaxis is compelling [38,39], clinical trials have yielded conflicting results, depending on patient inclusion criteria, choice of antibiotic, and timing of administration. Concerns over the possible selection for resistant bacteria and the potential for the emergence of fungal infection have been raised [40]. The most recent systematic review of 7 studies comprising a total of 404 patients found no difference in either mortality or prevention of secondary pancreatic infection [41]. Meta-analysis has further shown no difference in the need for surgical intervention [42]. Given the potential concerns and lack of evidence to support their use, prophylactic antibiotics should not be used routinely in clinical practice. Probiotic (rather than antibiotic) prophylaxis has been suggested as an alternative approach to the prevention of infection in pancreatic necrosis by maintaining normal gut flora. However, a prospective double-blind multicenter trial failed to show a reduction in the risk of infectious complications and was also associated with an unexpected increase in mortality in the probiotic group [43].

Computed tomography (CT) is particularly useful in establishing the presence of pancreatic necrosis, and CT-based grading systems have been proposed as predictors of clinical outcomes [44]. Early noncontrast CT can predict severity and mortality with reasonable accuracy [45], although intravenous contrast-infused protocols are necessary to establish definitively the presence of necrosis (Fig. 1) [46]. By definition, the CT criteria for pancreatic necrosis are diffuse or focal areas of nonenhancement (density less than approximately 50 Hounsfield units) involving more than 30% or 3 cm of the pancreas. Magnetic resonance imaging (MRI), and in particular the combination of T1- and T2-weighted sequences, can also be used to establish the diagnosis and extent of pancreatic necrosis [47]. The additional information provided by MRI cholangiopancreatography (MRCP) may be useful to evaluate ductal anatomy and the presence of bile duct stones, but otherwise there are few, if any, differences between CT and MRI with respect to diagnostic accuracy, prediction of severity, or other outcomes [48].

Fig. 1 Identification of pancreatic necrosis by CT. Areas of necrosis in the head and body/tail of the pancreas is present (asterisk). These areas are poorly enhanced in comparison with the areas of normal-appearing pancreatic tissue (arrows).

Various scoring systems have been devised to predict severity and estimate mortality in acute pancreatitis, beginning with the original Ranson score [49]. The Computed Tomography Severity Index (CTSI), also known as the Balthazar score, is the most widely used radiological system (Table 2) [50,51]. The presence of necrosis features prominently in the CTSI. Although several studies indicate that the CTSI is more accurate in assessing severity of pancreatitis than physiologic scoring systems, such as the APACHE-II (Acute Physiology and Chronic Health Evaluation II), the traditional Ranson criteria, or biomarkers such as C-reactive protein (CRP) levels [52–54], clinical outcomes may correlate more closely with the Ranson criteria or the APACHE-II scores [55].

Table 2 The CTSI (Balthazar score). Patients are assigned a score from 1 to 10 based on CT findings as well as the extent of pancreatic necrosis

Data from Balthazar E, Robinson DL, Megibow AJ, et al. Acute pancreatitis: value of CT in establishing prognosis. Radiology 1990;174(2):331–6.

CT-guided (or, less commonly, ultrasound-guided) aspiration of necrotic pancreatic tissue was introduced in the 1980s as a means to establish the presence or absence of infection and, for many years, formed a standard part of the treatment algorithm in many centers [37,39,56–62]. This practice was based on the prevailing field bias at the time in favor of early operative debridement of infected necrosis and for continued nonoperative management of otherwise-stable patients with sterile necrosis [13]. Theoretical concerns about percutaneous aspiration include the potential for conversion of sterile necrosis to infected necrosis by instrumentation. One study showed a secondary infection rate of 22%, although with a very small sample size [63]. In general, rates of secondary infection are extremely low and virtually noncontributory to the overall infection rate, and thus, image-guided needle aspiration should still be considered a safe diagnostic tool [64]. However, with increased adoption of nonoperative and minimally invasive alternatives even in cases of infected necrosis, the application of this practice has become somewhat less routine. Various serum markers have been proposed as noninvasive alternatives to the diagnosis of necrosis and the presence of infection. Procalcitonin levels of 1.8 ng/mL or more on 2 consecutive days predicted infected necrosis and differentiated it from sterile necrosis with a sensitivity of 95%, specificity of 88%, and accuracy of 90% in one study [65]. A large multicenter trial showed superiority of procalcitonin to CRP in predicting multisystem organ failure in patients with acute pancreatitis [66]. CRP is readily available in most centers and carries a sensitivity and specificity of more than 75% in predicting necrosis at levels greater than 200 mg/L [67,68]. Various other markers, such as IL-6 and IL-8 [69] and serum amyloid A [70], have also been correlated with disease severity, although these tests are not widely available for routine use.

Historically, the mortality of severe acute necrotizing pancreatitis was so high that surgical dogma mandated operative intervention by debridement, drainage, and open packing. By the 1980s, an association between the presence of infection and overall mortality was recognized, and a more selective approach began to be adopted. While in Europe, early surgical debridement was endorsed for all cases of sterile necrosis, in American centers, the practice of early CT-guided aspiration of areas of necrosis became the standard practice, with operative intervention reserved for cases of documented infection or rapid clinical deterioration [8,28,71]. The guidelines from the International Association of Pancreatology (IAP, 2002) [72], based mostly on level II-2 evidence, endorsed the use of fine-needle aspiration biopsy (FNAB) and recommended nonoperative management for FNAB-negative necrosis except in the presence of organ failure that does not improve with nonoperative therapy. However, current practice continues to evolve. Several groups began to report good results with initial nonoperative management of infected necrosis in selected clinically stable patients [14,15,73–75]. Most practitioners agree that most patients with sterile necrosis resolve without surgical intervention [6,76,77]. False-negative culture results may occur in as many as 20% of patients, and infection may be present with minimal signs. In most centers, FNAB is no longer routine. Even staunch advocates for early surgical intervention in patients with suspected infected necrosis are now more willing to accept a clinical picture consistent with infection as sufficient indication, such as persistent fevers, tachycardia, adynamic ileus, and continued or progressive multisystem organ failure. Surgical intervention continues to be indicated in patients with progressive multiorgan dysfunction, clinical deterioration, and failure to respond to supportive therapy [75].

Several factors beyond the presence of infection and the clinical stability of the patient may affect the decision for operative intervention in acute necrotizing pancreatitis. Anatomic considerations, including the degree to which the necrosis extends into surrounding spaces within the retroperitoneum; the presence of vascular complications, such as mesenteric or splenoportal venous thrombosis; obstruction of the gastrointestinal and biliary tracts; development of pancreatic ascites; and other complications, such as ischemia of the transverse colon, may influence decision making.

A critical question is the timing of surgical intervention, and clinical practice has evolved considerably in this regard over the past few years. In the past, when the major option was open transperitoneal necrosectomy, surgical intervention was generally undertaken early in the course of the illness, typically within a few days of onset [78] because it was assumed that early operation, particularly in cases of infection, improves the otherwise-dismal prognosis. Several earlier studies suggested that early intervention within the first 48 to 72 hours is associated with particularly high mortality, and the IAP guidelines of 2002 recommended against surgical intervention within the first 14 days of onset of the illness unless there were specific overriding indications [72]. Subsequent studies have provided evidence that mortality remains high even within the first 14 to 21 days of presentation but can be reduced to less than about 8% when operation is delayed beyond 28 to 30 days [5,79]. One prospective randomized trial examining early versus late necrosectomy was stopped because of high mortality in the early operative group [80]. One explanation for the apparent benefit of delayed surgery is that, over time, the definition and demarcation between normal and necrotic tissues improve to the point to which dissection can be less extensive, the risk of hemorrhage and injury to surrounding organs can be reduced, and unnecessary removal of otherwise-viable pancreatic tissues can be limited [75]. Although the current practice is to delay surgical intervention wherever possible, it is acknowledged that the prolonged use of antibiotics, so typical of this approach, is associated with the emergence of resistant strains and fungal superinfection [75].

The introduction of several interventional alternatives to traditional open necrosectomy has contributed to this evolution in practice. Videoscopic procedures (transperitoneal laparoscopic or retroperitoneal intracavitary approaches), image-guided catheter placement, endoluminal therapy, and various hybrid procedures are significantly less invasive than traditional necrosectomy. Some of these approaches have been advocated as temporizing maneuvers rather than definitive treatment, with the goal of controlling systemic infection and allowing sequestration of necrosis and inflammation such that formal operation can be delayed by the safer 4-week window. However, indications and timing for these alternatives remain to be established. Comparisons among these approaches and with open operation are undermined by small patient numbers, lack of uniform criteria for the reporting of results, variable learning curves, and wide differences in local expertise among various centers. The transferability of the results obtained by highly experienced experts in high-volume settings to less-experienced practitioners in other geographies is questionable. Current practice is thus a moving target, continually shaped by technological advances and accelerating product cycles.