Acute Necrotizing Pancreatitis: Laboratory, Clinical, and Imaging Findings as Predictors of Patient Outcome

Transcription

1 Gastrointestinal Imaging Original Research Brand et al. Acute Necrotizing Pancreatitis Gastrointestinal Imaging Original Research Michael Brand 1 Andrea Götz 1 Florian Zeman 2 Gundula Behrens 3 Michael Leitzmann 3 Tanja Brünnler 4 Okka W. Hamer 1 Christian Stroszczynski 1 Peter Heiss 1 Brand M, Götz A, Zeman F, et al. Keywords: acute necrotizing pancreatitis, clinical scores, CT, laboratory parameters, prognosis assessment DOI: /AJR Received March 4, 2013; accepted after revision September 2, Department of Radiology, University Medical Center Regensburg, Franz-Josef-Strauß-Allee 11, Regensburg, Germany. Address correspondence to M. Brand (michael.r.brand@web.de). 2 Center for Clinical Studies, University Medical Center Regensburg, Regensburg, Germany. 3 Department of Epidemiology and Preventive Medicine, University Medical Center Regensburg, Regensburg, Germany. 4 Department of Internal Medicine I, University Medical Center Regensburg, Regensburg, Germany. This article is available for credit. AJR 2014; 202: X/14/ American Roentgen Ray Society Acute Necrotizing Pancreatitis: Laboratory, Clinical, and Imaging Findings as Predictors of Patient Outcome OBJECTIVE. In patients with acute necrotizing pancreatitis, we analyzed whether laboratory and clinical findings determined in the early phase of disease and morphologic features on contrast-enhanced CT (CECT) at the beginning of the late phase of disease are helpful in predicting patient outcome and whether CECT findings provide additional information in establishing prognosis compared with the laboratory and clinical findings. MATERIALS AND METHODS. A retrospective analysis of 99 patients with acute necrotizing pancreatitis was performed. Four laboratory variables (albumin, calcium, C-reactive protein, WBC count) and three clinical variables (Acute Physiology, Age, Chronic Health Evaluation [APACHE] II score; Simplified Acute Physiology Score [SAPS] II; persistent organ failure) were assessed. Five morphologic features on CECT including Balthazar grade and CT severity index were reviewed. The endpoints of patient outcome were peripancreatic or pancreatic infection, need for intervention, duration of organ failure, ICU and hospital stays, and death. Based on receiver operating characteristic curve analysis for infection, highand low-risk groups for each prognostic variable were calculated and univariable and multivariable Cox regression analyses were carried out. RESULTS. In our study population of 99 patients (63 men, 36 women; median age, 52 years; age range, years), 25 patients (25%) developed infection, 42 patients (42%) experienced organ failure, and 12 patients (12%) died. Regarding the laboratory and clinical variables, albumin level, APACHE II score, and particularly persistent organ failure were the strongest independent predictors of patient outcome. Regarding the imaging variables, Balthazar grade and a morphologic feature that takes the distribution of intrapancreatic necrosis into account were the strongest independent predictors. In the multivariable analysis of all studied variables, imaging variables were independent and strong predictors of patient outcome and provided additional information in establishing prognosis compared with clinical and laboratory findings. CONCLUSION. In patients with suspected or proven acute necrotizing pancreatitis, performing CECT at the beginning of the late phase of disease is recommended to identify patients at increased risk for adverse outcomes. A cute pancreatitis is a dynamic process with two overlapping phases of disease [1 3]: In the early phase, a severe course is induced by persistent organ failure due to a systemic inflammatory response that is not necessarily related to the infection or the extent of necrosis [1 3]. Thus, in this phase it is recommended that laboratory and clinical parameters be used as diagnostic tools for risk stratification [1, 4, 5]. This early phase usually lasts for the first week but may continue in the second week. The following late phase, lasting weeks to months, is characterized by signs of persistent inflammation and local complica- tions [1]. In this phase, pancreatic and peripancreatic changes progressively mature, further changes in morphology occur slowly, and imaging-based criteria should allow a definitive distinction between the two types of acute pancreatitis [1, 6, 7]: acute interstitial edematous pancreatitis and acute necrotizing pancreatitis. In patients with interstitial edematous pancreatitis, the pancreas is normal or is edematously enlarged with or without mild inflammatory changes of the peripancreatic or retroperitoneal tissues [1]. Acute necrotizing pancreatitis is diagnosed when necrosis of the pancreatic parenchyma or the peripancreatic tissue is identified. AJR:202, June

2 TABLE 1: Baseline Characteristics of All Patients and Comparisons of Patients With and Those Without a Proven Peripancreatic or Pancreatic Infection and of Brand Patients et al. Who Died and Those Who Survived Characteristic All Patients (n = 99) Infection Death Yes (n = 25) No (n = 74) Yes (n = 12) No (n = 87) Age (y) 52 (18 84) 54 (33 84) 50 (18 83) 59 (18 76) 52 (19 84) Sex, male 63 (64) 13 (52) 50 (68) 9 (75) 54 (62) Body mass index (kg/m 2 ) 27.3 ( ) 28.4 ( ) 26.9( ) 35.4 ( ) 26.9 ( ) Transferred from peripheral hospitals 68 (69) 22 (88) 46 (62) 11 (92) 57 (66) Time (d) after onset of symptoms to First hospital admission 0 (0 12) 0 (0 2) 0 (0 12) 0 (0 1) 0 (0 12) Our hospital 2 (0 20) 4 (0 20) 2 (0 18) 3 (0 18) 2 (0 20) CECT 7 (0 21) 9 (4 21) 6 (0 21) 11 (0 19) 7 (0 21) Laboratory variables Albumin (g/l) 32.2 ( ) 27.2 ( ) 34.1 ( ) 27.4 ( ) 33.7 ( ) Calcium (mmol/l) 2.0 ( ) 1.9 ( ) 2.1 ( ) 1.7 ( ) 2.1 ( ) C-reactive protein (mg/l) 195 ( ) 262 (4 410) 171 ( ) 239 (50 368) 193 ( ) WBC count ( 10 9 /L) 14.8 ( ) 16.2 ( ) 14.2 ( ) 16.1 ( ) 14.4 ( ) Clinical variables APACHE II score 13 (0 44) 14 (4 39) 11 (0 44) 24 (14 44) 11 (0 37) SAPS II 23 (6 79) 27 (13 61) 22 (6 79) 42 (24 79) 21 (6 63) Persistent organ failure (first week) 34 (34) 16 (64) 18 (24) 10 (83) 24 (28) SIRS 92 (93) 25 (100) 67 (91) 12 (100) 80 (92) Imaging variables Balthazar grade E (C E) E (D E) E (C E) E E (C E) CT severity index 6 (3 10) 10 (4 10) 6 (3 10) 9 (6 10) 6 (3 10) PAN equal area 2 (0 3) 3 (0 3) 1 (0 3) 3 (1 3) 2 (0 3) PAN anatomical 2 (0 3) 3 (0 3) 1 (0 3) 3 (1 3) 2 (0 3) ANC grade 1 (0 2) 2 (1 2) 1 (0 2) 2 (0 2) 1 (0 2) Outcome parameters Proven peripancreatic or pancreatic infection 25 (25) 4 (33) 21 (24) Time to infection after onset of symptoms (d) 27 (6 76) 17 (7 62) 27 (6 76) Time from CECT to infection (d) 15 (0 65) 13 (0 16) 20 (0 65) Need for intervention 46 (46) 24 (96) 22 (30) 10 (83) 36 (41) Endoscopic intervention 8 (8) a 7 (28) 1 (1) 1 (8) 7 (8) Percutaneous intervention 33 (33) b 20 (80) 13 (18) 3 (25) 30 (34) Surgical intervention 23 (23) c 14 (56) 9 (12) 10 (83) 14 (16) Time (d) to intervention after onset of symptoms 10 (1 60) 14 (4 49) 10 (1 60) 8 (1 29) 10 (4 60) Organ failure 42 (42) 18 (72) 24 (32) 12 (100) 30 (35) Duration of organ failure (d) 20 (1 88) 27 (12 88) 15 (1 64) 23 (7 62) 20 (1 88) ICU admission 77 (78) 24 (96) 53 (72) 12 (100) 65 (75) Length of ICU stay (d) 11 (1 96) 28 (1 96) 7 (1 88) 24 (7 88) 7 (1 96) Length of hospital stay (d) 25 (4 195) 101 (25 195) 19 (4 119) 28 (7 93) 24 (4 195) Death 12 (12) 4 (16) 8 (11) Time (d) to death after onset of symptoms 28 (7 110) 30 (25 38) 16 (7 110) Note Values of characteristics are given as median (range) or, if binary, as numbers (percentages). CECT = contrast-enhanced CT; APACHE II = Acute Physiology, Age, Chronic Health Evaluation II; SAPS II = Simplified Acute Physiology Score II; SIRS = systemic inflammatory response syndrome; PAN = number of parts of the pancreas that exhibited areas of necrosis; ANC = acute necrotic collection. a All eight patients were treated with additional surgical or percutaneous procedures. b Fourteen of 33 patients were treated with percutaneous drainage only. c Six of 23 patients were treated with surgery only AJR:202, June 2014

3 Interstitial edematous pancreatitis is usually a self-limiting disease [1]. Acute necrotizing pancreatitis often leads to the development of local complications associated with complex treatment, a prolonged hospital stay, and possible progression to lethal multiple organ failure [8, 9]. Necrosis of the pancreatic parenchyma and of the peripancreatic tissue serves as the potential nidus for infection [4], which occurs as a severe complication in 20 70% of cases [10 12]. In cases of infected necrosis, organ failure and mortality rates rise up to 50% and 20 80%, respectively [10, 12, 13]. Contrast-enhanced CT (CECT) is considered the reference standard for imaging of patients with acute pancreatitis [1, 6]. Several studies have shown that the use of CT scoring systems in the early phase is helpful in predicting disease severity in and outcome of patients with acute pancreatitis [8, 14 17]. However, in recent studies, clinical scoring systems obtained on hospital admission showed an equal accuracy [5] and the use of CT scans obtained in the early phase of disease led to no alteration in diagnosis or clinical management [18]. Hence, with regard to prudent resource utilization and the avoidance of unnecessary radiation exposure, the authors of those studies suggested that imaging should be reserved for patients with clinically predicted severe acute pancreatitis, uncertain diagnosis, clinical worsening, or suspicion of life-threatening complications [5, 18]. The Acute Pancreatitis Classification Working Group [1] recommends using CECT mainly in the late phase of the disease in addition to clinical parameters to identify patients at risk for local complications and to guide therapy. Taking these findings into account, this study focuses on patients with acute necrotizing pancreatitis and on CECT studies performed at the beginning of the late phase of the disease. We analyzed whether laboratory and clinical findings determined in the early phase of disease and morphologic features on CECT at the beginning of the late phase of disease are helpful in predicting patient outcome. In addition, we analyzed whether the findings on CECT provide additional information in establishing prognosis compared with the laboratory and clinical findings. Materials and Methods Patients This retrospective study was performed at a university hospital that is a regional tertiary referral center for pancreatitis. The Independent Ethics Committee at the University of Regensburg con- Acute Necrotizing Pancreatitis TABLE 2: Area Under the Curve (AUC), Cutoff, Sensitivity, and Specificity of Variables for Predicting Peripancreatic or Pancreatic Infection Variable AUC (95% CI) Cutoff Sensitivity, % (95% CI) Specificity, % (95% CI) Age 0.59 ( ) > 50 y 68 (47 85) 53 (41 64) Sex 0.58 ( ) Male 48 (28 69) 68 (56 78) Body mass index 0.55 ( ) > 29 kg/m 2 46 (26 67) 72 (60 83) Laboratory variables Albumin 0.71 ( ) 28.7 g/l 70 (47 87) 74 (62 85) Calcium 0.61 ( ) 1.9 mmol/l 55 (32 76) 67 (54 78) C-reactive protein 0.71 ( ) > 251 mg/l 64 (43 82) 76 (64 85) WBC count 0.54 ( ) > /L 58 (37 78) 62 (50 73) Clinical variables APACHE II score 0.65 ( ) > (53 90) 53 (41 65) SAPS II 0.61 ( ) > (63 95) 36 (25 49) Persistent organ failure (first week) 0.70 ( ) Present 64 (43 82) 76 (64 85) Imaging variables Balthazar grade 0.62 ( ) > D 96 (80 100) 27 (17 39) CT severity index 0.84 ( ) > 6 84 (64 96) 74 (63 84) PAN equal area 0.83 ( ) > 2 84 (64 96) 78 (67 87) PAN anatomical 0.82 ( ) > 2 80 (59 93) 78 (67 87) ANC grade 0.67 ( ) > 1 72 (51 88) 58 (46 70) Note APACHE II = Acute Physiology, Age, Chronic Health Evaluation II; SAPS II = Simplified Acute Physiology Score II; PAN = number of parts of the pancreas that exhibited areas of necrosis; ANC = acute necrotic collection. TABLE 3: Cutoff, Frequency, Rate of Infection and Mortality of the High- and Low-Risk Group Based on the Receiver Operating Characteristic Curve Analysis for the Outcome Infection Variable and Cutoff Total No. of Patients With Infection No. (%) of Patients Who Died Age > 50 y a (33) 6 (12) 50 y 47 8 (17) 6 (13) Sex Male a (21) 9 (14) Female (33) 3 (8.3) Body mass index > 29 kg/m 2a (38) 5 (17) 29 kg/m (22) 3 (5.0) Laboratory variables Albumin 28.7 g/l a (50) 7 (22) > 28.7 g/l 53 7 (13) 4 (7.5) Calcium 1.9 mmol/l a (38) 6 (19) > 1.9 mmol/l (19) 2 (3.8) (Table 3 continues on next page) AJR:202, June

4 Brand et al. TABLE 3: Cutoff, Frequency, Rate of Infection and Mortality of the High- and Low-Risk Group Based on the Receiver Operating Characteristic Curve Analysis for the Outcome Infection (continued) Variable and Cutoff Total No. of Patients firmed that, according to state law, neither ethics committee approval nor the committee s opinion is necessary for retrospective studies like this one. From a search of the hospital patient database, all adult patients (age 18 years) were identified who were treated at our hospital between January With Infection No. (%) of Patients Who Died C-reactive protein > 251 mg/l a (47) 5 (15) 251 mg/l 61 9 (15) 6 (10) WBC count > /L a (34) 6 (15) /L (19) 5 (9.3) Clinical variables APACHE II score > 11 a (35) 11 (22) (14) 0 SAPS II > 17 a (31) 11 (17) (14) 0 Persistent organ failure (first week) Present a (47) 10 (29) Absent 65 9 (14) 2 (3.1) Imaging variables Balthazar grade > D a (31) 12 (15) D 21 1 (4.8) 0 CT severity index > 6 a (53) 8 (20) (6.8) 4 (6.8) PAN equal area > 2 a (57) 9 (24) (6.5) 3 (4.8) PAN anatomical > 2 a (56) 8 (22) (7.9) 4 (6.3) ANC grade > 1 a (37) 8 (16) (14) 4 (8.0) Note APACHE II = Acute Physiology, Age, Chronic Health Evaluation II; SAPS II = Simplified Acute Physiology Score II; PAN = number of parts of the pancreas that exhibited areas of necrosis; ANC = acute necrotic collection. a High-risk group and December 2010 and had diagnosis code K85, which is defined as acute pancreatitis according to the International Statistical Classification of Diseases and Related Health Problems (10th revision, German Modification). A total of 824 patients were identified (Fig. 1). The medical records of the 824 patients identified were studied to determine whether they had an episode of acute pancreatitis between January 2005 and December 2010 and had undergone CECT of the abdomen within 21 days after the onset of acute pancreatitis. The onset of acute pancreatitis was defined as the time of onset of abdominal pain [1]. Acute pancreatitis was diagnosed if two or more of the following findings were present: characteristic abdominal pain; serum amylase or lipase level three or more times higher than the upper limit of normal (i.e., > 210 and 180 U/L, respectively); and an imaging study (CT, MRI, or transabdominal sonography) showing changes consistent with acute pancreatitis [1, 5, 19]. Three hundred twenty-six patients fulfilled these criteria. Of these 326 patients, 187 were excluded because they presented with acute-onchronic pancreatitis (n = 73); had a malignancy (n = 66); had undergone partial pancreatic resection (n = 21); had undergone abdominal surgery within 2 weeks before the onset of acute pancreatitis (n = 15); or had a severe disease not related to the pancreas, with mild pancreatitis being a secondary diagnosis (n = 12). All available CECT studies and MRI studies of the remaining 139 patients were reviewed by two radiologists to assess in consensus the diagnosis of interstitial edematous pancreatitis versus acute necrotizing pancreatitis according to the revised Atlanta Classification [1]. In two patients, the CECT examination was considered to be of insufficient quality because of low signal-to-noise ratio due to patient obesity or insufficient contrast enhancement of the abdominal parenchymal organs. Three additional patients were excluded because calcifications typical of chronic pancreatitis were identified on CECT, which is suggestive of acute-on-chronic pancreatitis. The final diagnosis of acute necrotizing pancreatitis was made in 99 of 134 patients. These 99 patients represent the study population. The laboratory, clinical, and imaging findings as well as the outcomes of these patients are outlined in this article. In the remaining 35 of 134 patients, the final diagnosis was interstitial edematous pancreatitis. Twelve of these 35 patients (34%) were admitted to the ICU. The median duration of the hospital stay and ICU stay was 9 days (range, 2 27 days) and 2 days (range, 1 12 days), respectively. Two patients experienced organ failure for 9 and 10 days, respectively. None of the 35 patients with interstitial edematous pancreatitis exhibited proven peripancreatic or pancreatic infection, needed an intervention (defined as percutaneous or endoscopic drainage or percutaneous, endoscopic, or surgical necrosectomy), or died within 250 days after the onset of acute pancreatitis AJR:202, June 2014

5 Acute Necrotizing Pancreatitis TABLE 4: Univariable Cox Regression Analysis for Predicting Peripancreatic or Pancreatic Infection, Intervention, End of Organ Failure, Discharge From ICU, Discharge From Hospital, and Mortality Infection a Intervention a End of Organ Failure b Discharge From ICU b Discharge From Hospital b Mortality a HR c 95% CI p HR c 95% CI p HR c 95% CI p HR c 95% CI p HR c 95% CI p HR c 95% CI p Variable Age Sex Body mass index Laboratory variables Albumin < < < Calcium C-reactive protein < WBC count Clinical variables APACHE II score < NP SAPS II NP Persistent organ < < < < failure (first week) Imaging variables Balthazar grade NP < NP CT severity index < < < PAN equal area < < < PAN anatomical < < < ANC grade < Note HR = hazard ratio, APACHE II = Acute Physiology, Age, Chronic Health Evaluation II; NP = statistical analysis was not performed because of the small number of events within the groups; SAPS II = Simplified Acute Physiology Score II; PAN = number of parts of the pancreas that exhibited areas of necrosis; ANC = acute necrotic collection. a The higher the HRs, the higher the risk for infection, intervention, and mortality for the high-risk group of each variable. b The lower the HRs, the lower the chance of ending organ failure and of discharge from ICU and from hospital for the high-risk group of each variable. For instance, the HR of PAN equal area indicates that patients of the high-risk group had an almost sixfold lower chance of discharge from hospital (HR, 0.17) at any time during the observation period compared with the low-risk group and that the mortality risk of the high-risk group was more than fivefold increased (HR, 5.4) compared with patients of the low-risk group. c HR comparing the high- and low-risk groups for each variable with the low-risk group as reference. Data Review The clinical data of the study population and details about surgical and radiologic procedures were collected by reviewing patient charts and surgical and radiologic reports. Laboratory and Clinical Variables The following variables were determined within 36 hours (most within 24 hours) after admission to our hospital: serum levels of albumin, calcium, C-reactive protein, and WBC count; the Acute Physiology, Age, Chronic Health Evaluation (APACHE) II score [20]; and the Simplified Acute Physiology Score (SAPS) II [21]. The presence of persistent organ failure within the first week after the onset of acute pancreatitis was evaluated. Organ failure was defined as a score 2 in one or more of the three (respiratory, renal, and cardiovascular) organ systems of the modified multiple organ dysfunction score [22]. Persistent organ failure was defined as organ failure that exceeded 48 hours [1]. Additionally, the occurrence of systemic inflammatory response syndrome (SIRS) [23] within the first week after admission to our hospital was recorded. Imaging Variables Because the focus of our study is on imaging variables of CECT examinations performed at the beginning of the late phase of disease, only CECT studies performed within 21 days after the onset of symptoms were chosen for evaluation. A CECT study obtained after the first week of the disease should allow a definitive diagnosis of interstitial edematous pancreatitis versus acute necrotizing pancreatitis [1], and CTbased classification systems have the highest diagnostic and predictive accuracy when scanning is performed 6 10 days after the onset of acute pancreatitis [6]. Hence, if one patient underwent two or more CECT examinations within 21 days after the onset of acute pancreatitis, the following procedure was applied to choose the study CECT: Five time intervals between the onset of acute pancreatitis and the performance of CECT were defined with the following priority: 6 10 days (first priority), 4 5 days (second priority), days AJR:202, June

6 Brand et al. TABLE 5: Multivariable Analysis of Laboratory and Clinical Variables Variable HR a 95% CI p Infection Albumin C-reactive protein Intervention Calcium Persistent organ failure (first week) Chance of discharge from ICU Albumin APACHE II score Persistent organ failure (first week) Chance of discharge from hospital Age Albumin APACHE II score < Persistent organ failure (first week) Mortality Persistent organ failure (first week) Note HR = hazard ratio; APACHE II = Acute Physiology, Age, Chronic Health Evaluation II. a HR comparing the high- and low-risk groups for each variable with the low-risk group as reference. TABLE 6: Multivariable Analysis of Imaging Variables Variable HR a 95% CI p Infection PAN equal area < Intervention PAN anatomical < ANC grade Chance of ending organ failure Balthazar grade PAN equal area Chance of discharge from ICU Balthazar grade PAN equal area Chance of discharge from hospital Balthazar grade < PAN equal area < Mortality PAN equal area Note HR = hazard ratio, PAN = number of parts of the pancreas that exhibited areas of necrosis, ANC = acute necrotic collection. a HR comparing the high- and low-risk groups for each variable with the low-risk group as reference. (third priority), 0 3 days (fourth priority), and days (last priority). The time interval with the highest priority in which a CECT examination was performed was determined, and the corresponding CECT defined as the study CECT was evaluated. If two or more CECT studies were performed in the selected time interval, the CECT study that was performed closest to day 6 was chosen. All examinations were performed on MDCT scanners (Somatom Plus 4 or Somatom Sensation 16, Siemens Healthcare). As stated previously [24], the following standard CT protocol for abdominal imaging was applied: For opacification of the gastrointestinal tract, 2000 ml of positive radiocontrast agent was administered orally and 150 ml nonionic contrast medium (300 mg I/mL iopromide [Ultravist, Bayer Schering Pharma] or iomeprol [Imeron, Bracco Altana Pharma]) was power-injected IV at a rate of 3 ml/s in all patients. Portal venous phase scans of the abdomen were acquired in the craniocaudal direction and were reconstructed in the axial plane with a slice thickness of 5 mm. All study CT examinations were independently reviewed on PACS workstations (Syngo-Imaging, version VB36A, Siemens Healthcare) by two radiologists with 4 and 9 years of experience in body imaging, respectively, who were blinded to patient characteristics and outcome. If morphologic features were determined with identical results from both readers, the finding was defined as a definitive finding. Morphologic features that were discrepantly determined by both readers were reevaluated jointly in a second session and the definitive finding was determined through consensus decision. Features Analyzed for Each Patient Balthazar grade The Balthazar grade [25] was analyzed for each patient. CT severity index The CT severity index [26] was analyzed for each patient. Distribution of pancreatic necrosis The distribution of pancreatic necrosis was analyzed for each patient. Areas of pancreatic parenchyma that exhibited nonenhancement on CECT were considered to represent necrosis [1]. The distribution of pancreatic necrosis was evaluated using two classifications. The first classification used two planes. The pancreas was divided into three parts (head, corpus, and tail) with equal area on the transverse plane (Fig. 2A). The two planes were drawn through the pancreas parallel to the long axis of the body, perpendicular to the pancreatic duct, and in a manner so that the craniocaudal projection of the three parts onto the transverse plane exhibited equal area. In practice, the entire pancreas was reviewed on the transverse reconstructions, and each radiologist drew virtually the two planes through the pancreas to define the three parts of the pancreas. The three parts of the pancreas were analyzed separately for the presence of necrosis (Figs. 3 and 4). The number of parts of the pancreas that exhibited areas of necrosis, which we abbreviate hereafter as PAN, was determined and ranged from 0, meaning 0 parts (no intrapancreatic necrosis), to 3, meaning three parts (necrosis in all three parts of the pancreas). We abbreviate this variable as PAN equal area AJR:202, June 2014

7 Acute Necrotizing Pancreatitis The second classification is based on the anatomic definitions in Grey s Anatomy [27]. The pancreas was divided into three parts (Fig. 2B): the extended head (which includes the head, the neck, and the uncinate process, defined as the parts of the pancreas ventral as well as dorsal and to the right of the superior mesenteric vein and portal vein), the corpus (between the extended head and tail), and the tail (defined as the most lateral 3.5 cm of the pancreas to the left). These three parts were analyzed separately for the presence of necrosis (Figs. 3 and 4), and PAN was determined and ranged from 0, meaning 0 parts (no intrapancreatic necrosis), to 3, meaning three parts (necrosis in all three parts of the pancreas). We abbreviate this variable as PAN anatomical. Presence of acute necrotic collections The presence of acute necrotic collections (ANCs), defined according to the revised Atlanta classification [1], at the following two specific sites was determined: the retrocolic part of the anterior pararenal space and the posterior pararenal space. The retrocolic part of the anterior pararenal space was defined as the space dorsal or dorsolateral of the ascending and descending colon, respectively, limited by the parietal layer of the peritoneum anteriorly, the anterior renal fascia posteriorly, and the lateral conal fascia laterally [28, 29] (Fig. 5). The posterior pararenal space is confined by the posterior renal and lateral conal fascia anteriorly and the transversalis fascia posteriorly. Its medial boundary is formed by the psoas major and quadratus lumborum muscles, whereas laterally it communicates with the properitoneal fat compartment [28] (Fig. 6). The presence of ANCs was graded as follows: grade 0, no ANC at either site (i.e., the retrocolic part of the anterior pararenal space and the posterior pararenal space); grade 1, ANC in the retrocolic part of the anterior pararenal space to the right, left, or both; and grade 2, ANC in the posterior pararenal space to the right, left, or both. We abbreviate this variable as ANC grade. Faint morphologic changes were rated as ANC when the change was considered to be more than fat stranding and was assumed that needle puncture would allow aspiration of at least one fluid drop. Figure 6 illustrates an example of borderline changes that were called a positive finding of an ANC. Definition of Severity of Acute Pancreatitis The severity of acute pancreatitis was determined according to the Atlanta criteria [30] published in 1992 and the revised Atlanta classification [1] published in The presence of severe acute pancreatitis according to both classifications was compared. According to the revised Atlanta classification (2012), severe acute pancreatitis is defined by persistent (> 48 hours) organ failure. TABLE 7: Results of Multivariable Analysis of All Studied Variables Variable HR a 95% CI p Infection Albumin PAN equal area < Intervention Persistent organ failure (first week) PAN anatomical < ANC grade Chance of ending organ failure Balthazar grade PAN equal area Chance of discharge from ICU Albumin APACHE II score Persistent organ failure (first week) < PAN equal area Chance of discharge from hospital APACHE II score Persistent organ failure (first week) Balthazar grade PAN equal area < Mortality PAN equal area Note HR = hazard ratio; PAN = number of parts of the pancreas that exhibited areas of necrosis; ANC = acute necrotic collection; APACHE II = Acute Physiology, Age, Chronic Health Evaluation II. a HR comparing the high- and low-risk groups for each variable with the low-risk group as reference. TABLE 8: Area Under the Curve (AUC), Cutoff, Sensitivity, and Specificity of Imaging Variables for Predicting Peripancreatic or Pancreatic Infection of the Patient Subgroup That Underwent Contrast-Enhanced CT 6 10 Days After the Onset of Symptoms Variable AUC (95% CI) Cutoff Sensitivity, % (95% CI) Specificity, % (95% CI) Balthazar grade 0.63 ( ) > D 100 (72 100) 26 (10 48) CT severity index 0.72 ( ) > 6 73 (39 94) 65 (43 84) PAN equal area 0.76 ( ) > 2 73 (39 94) 74 (52 90) PAN anatomical 0.70 ( ) > 1 91 (59 100) 43 (23 66) ANC grade 0.74 ( ) > (72 100) 48 (27 69) Note PAN = number of parts of the pancreas that exhibited areas of necrosis, ANC = acute necrotic collection. Outcome Parameters Six outcome parameters were examined: proven peripancreatic or pancreatic infection, need for intervention, duration of organ failure, duration of ICU stay, duration of hospital stay, and mortality. The diagnosis of infection was based on examination of the material acquired by an intervention or fine-needle aspiration using Gram stain, culture, or both. Interventions were defined as percutaneous or endoscopic drainage or percutaneous, endoscopic, or surgical necrosectomy. Interventions were performed if infection of peripancreatic or pancreatic necrosis was suspected because of acute clinical deterioration with elevation of body temperature, C-reactive protein, and WBC count, bacteremia, or sepsis; if peripancreatic or pancreatic necrosis seemed to impair organ functions as a result of its massive extent or adverse location; AJR:202, June

8 Brand et al. and if infection of peripancreatic or pancreatic necrosis was proven by fine-needle aspiration. The first-line intervention was percutaneous drainage therapy. If percutaneous drainage placement was not possible because of anatomic conditions, endoscopic drainage therapy was attempted. Active percutaneous and, where not feasible, endoscopic necrosectomy was indicated if sepsis and solid tissue within the necrosis persisted despite maximum medical support and drainage therapies or if abdominal compartment syndrome was diagnosed. Surgical necrosectomy was considered in cases of clinical deterioration despite percutaneous or endoscopic interventions or when percutaneous or endoscopic interventions were not feasible. Organ failure was defined as mentioned earlier. The durations of organ failure, ICU stay, and hospital stay were calculated including the time periods spent at the transferring hospitals. If a patient died during hospitalization, these three time lines were censored accordingly. Survival time was defined as the time interval between the onset of acute pancreatitis and patient death and was censored at day 250. Observations until day 250 after the onset of acute pancreatitis were based on medical reports or telephone follow-up. TABLE 9: Cutoff, Frequency, and Rate of Infection of the High- and Low-Risk Groups of Imaging Variables of the Patient Subgroup That Underwent Contrast-Enhanced CT 6 10 Days After the Onset of Symptoms Variable and Cutoff Total No. of Patients No. (%) of Patients With Infection Balthazar grade > D a (39) D 6 0 CT severity index > 6 a 16 8 (50) (17) PAN equal area > 2 a 14 8 (57) (15) PAN anatomical > 1 a (43) (9) ANC grade > 1 a (48) Note PAN = number of parts of the pancreas that exhibited areas of necrosis, ANC = acute necrotic collection. a High-risk group. Statistical Analysis Descriptive statistics were used for patient characteristics, including individual and clinical data; predictor variables; and outcome parameters. Values are given as a total number with the percentage or as a median with the range. The laboratory, clinical, and imaging variables as well as age, sex, and body mass index were tested for their capability to predict patient outcome. In patients with acute necrotizing pancreatitis, a peripancreatic or pancreatic infection is a common starting point for a complicated and protracted clinical course. Hence, for each variable, the area under the curve (AUC) was calculated from receiver operating characteristic (ROC) curves for predicting peripancreatic or pancreatic infection. AUCs greater than 0.8 and AUCs ranging from 0.6 to 0.8 were considered to indicate good and moderate accuracy, respectively [31]. The AUCs for each variable were compared using a method proposed by DeLong et al. [32]. For each variable, the threshold with the highest Youden index for predicting infection was calculated. This cutoff value was used to stratify the patient population into a high-risk group and a low-risk group. To compare the outcomes of the high- and lowrisk groups, a univariable Cox regression analysis for each dichotomized variable was performed. Variables with a p value < 0.05 were introduced to multivariable backward stepwise Cox regression analyses using the likelihood ratio test. The p value for variable selection and exclusion in the models was For all covariates, the absence of multicollinearity and the assumption of proportional hazards were tested with correlations and log minus log plots, respectively. The goodness-of-fit of these multivariable models was tested with an approximation to the test of Hosmer and Lemeshow, described in [33] and as suggested by Bradburn et al. [34], using ranked prognostic indexes. An analysis of the imaging variables of the patient subgroup who underwent CECT within 6 10 days after the onset of symptoms was added. Statistical analyses were conducted using statistics software (SPSS Statistics, version 19.0, SPSS; MedCalc, version , MedCalc; and R, version , R Foundation for Statistical Computing). A two-sided p value < 0.05 was considered statistically significant. Results The baseline characteristics of all patients, patients with and those without proven peripancreatic or pancreatic infection, and patients who died and those who survived are shown in Table 1. The cause of pancreatitis was alcoholic (n = 33), biliary (n = 31), postendoscopic retrograde cholangiopancreatography (n = 7), hyperlipidemia (n = 6), drug-related (n = 4), hyperglycemia (n = 2), posttraumatic (n = 1), and idiopathic or unknown (n = 15). The values for albumin and calcium were available in 85 patients (86%). C-reactive protein, WBC count, APACHE II score, and SAPS II were assessable in 95 of 99 patients (96%). Within the first week after the onset of acute pancreatitis, 34 patients (34%) experienced persistent (> 48 hours) organ failure. During the entire hospital stay, 41 patients (41%) presented with persistent organ failure and one patient (1%) with transient ( 48 hours) organ failure. Seventy-seven of 99 patients (78%) presented with pancreatic parenchymal necrosis with or without extrapancreatic necrosis. Twenty-two patients (22%) presented with only extrapancreatic necrosis. Twelve patients (12%) died; the cause of death was multiple organ failure in all cases. Most patients (92/99, 93%) presented with SIRS. According to the definitions of the Atlanta criteria 1992 [30] and the revised Atlanta classification 2012 [1], 99 and 41 patients experienced severe acute pancreatitis, respectively. Because a SIRS and severe acute pancreatitis according to the Atlanta criteria 1992 were present in most patients (> 90%), these variables were not applicable for risk stratification. Table 2 presents the AUC, the cutoff value that was used to allocate patients to the high- and low-risk groups, and the sensitivity and specificity of all variables for predicting 1222 AJR:202, June 2014

9 Acute Necrotizing Pancreatitis TABLE 10: Univariable Cox Regression Analysis of Imaging Variables for Predicting Peripancreatic or Pancreatic Infection, Intervention, End of Organ Failure, Discharge From ICU, and Discharge From Hospital of the Patient Subgroup That Underwent Contrast-Enhanced CT 6 10 Days After the Onset of Symptoms Infection a Intervention a End of Organ Failure b Discharge From ICU b Discharge From Hospital b HR c 95% CI p HR c 95% CI p HR c 95% CI p HR c 95% CI p HR c 95% CI p Variable Balthazar grade NP NP NP NP NP NP NP NP NP < CT severity index PAN equal area PAN anatomical NP NP NP ANC grade NP NP NP Note HR = hazard ratio, NP = statistical analysis was not performed because of the small number of events within the groups, PAN = number of parts of the pancreas that exhibited areas of necrosis, ANC = acute necrotic collection. a The higher the HRs, the higher the risk for infection and intervention for the high-risk group of each variable. b The lower the HRs, the lower the chance of ending organ failure and of discharge from the ICU and from the hospital for the high-risk group of each variable. c HR comparing the high- and low-risk groups for each variable with the low-risk group as reference. peripancreatic or pancreatic infection. Three variables the CT severity index, PAN equal area, and PAN anatomical showed good accuracy (AUC > 0.8) in predicting infection. Albumin level, C-reactive protein level, APACHE II score, persistent organ failure, Balthazar grade, and ANC grade showed moderate accuracy (AUC = ) and their confidence intervals did not include the value of 0.5, which would indicate random accuracy. The ROC curves of the imaging variables, the clinical score with the highest AUC (APACHE II score), persistent organ failure, and the laboratory variable with the highest AUC (albumin) are presented in Figure 7. Table 3 presents the frequencies and rates of infection and mortality of the high- and low-risk groups. The highest rates of infection and mortality were present in the high-risk group of PAN equal area and persistent organ failure, respectively. In the univariable analyses, several laboratory, clinical, and imaging variables were significant predictors of patient outcome (Table 4). For three imaging findings (CT severity index, PAN equal area, PAN anatomical), the hazard ratio for a peripancreatic or pancreatic infection was 14, indicating that patients of the high-risk group had a considerably increased risk of developing infection compared with those in the low-risk group. For persistent organ failure, the hazard ratio for mortality was 11, indicating that patients of the high-risk group had a considerably increased risk of death. PAN equal area was the only variable that predicted a statistically significantly increased risk of the high-risk group for all six outcome parameters. PAN anatomical and persistent organ failure predicted a statistically significantly increased risk of the high-risk group for all outcome parameters except chance of ending organ failure. Of note, the number of events in the low-risk group of Balthazar grade, APACHE II score, and SAPS II was 1 for the outcomes infection and mortality (Table 3); accordingly, in these cases univariable regression analysis was not performed because of the small number of events. For some variables, the outcomes of the high- and lowrisk groups are illustrated in Kaplan-Meier plots (Figs. 8 11). In the multivariable analyses, there was no evidence for multicollinearity or a violation of the proportional hazards assumption. For all outcome parameters, the covariates formed sufficient models with no evidence of a poor fit according to the ranked prognostic indexes. The results of the multivariable analyses are presented in Tables 5 7. Regarding the laboratory and clinical variables, determined in the early phase of disease, albumin, APACHE II score, and particularly persistent organ failure were the strongest independent predictors of patient outcome (Table 5). Regarding the imaging variables, Balthazar grade and PAN equal area were the strongest independent predictors (Table 6). In the multivariable analysis of all studied variables, the prognostic models for the outcomes end of organ failure and mortality included only imaging variables as independent predictors and the remaining variables provided no additional gain of information (Table 7). The prognostic models for the outcomes infection, need for intervention, and discharge from hospital included imaging variables as well as albumin, persistent organ failure, and APACHE II score, and the imaging variables were stronger predictors according to the hazard ratios. The prognostic model for the outcome discharge from ICU included albumin, APACHE II score, persistent organ failure, and PAN equal area, and persistent organ failure was the strongest predictor according to the hazard ratio. PAN equal area contributed as a covariate to five of the six prognostic models and it was the only independent predictor for mortality. For 34 patients, the study CECT was performed within 6 10 days after the onset of symptoms. The results of the analysis of this patient subgroup are shown in Tables Only one patient in this subgroup died; hence, mortality was not included in the analysis. The findings of the subgroup analysis showed, by trend, results similar to the results of the entire study population, but with larger CIs and lower statistical significance in the univariable analysis. Thus, a multivariable analysis was not performed. Discussion Risk stratification, therapeutic guidance, and reliable assessment of prognosis in patients with acute pancreatitis continue to be AJR:202, June

10 Brand et al. a challenge, with gradually changing strategies. Various studies have shown that early CT scans used for patient populations that covered the entire range of disease severity from mild to severe pancreatitis are helpful in handling this task [8, 14 17]. Because of the findings of recent studies, a more circumspect use of CT is recommended [5, 18, 35]. The Acute Pancreatitis Classification Working Group [1] advises using CECT primarily in severe cases and in the late phase of disease. Taking these aspects into account, we analyzed severely ill patients with acute necrotizing pancreatitis and focused on the prognostic significance of CECT examinations performed basically at the beginning of the late phase of disease. All CECT images were independently read by two radiologists. Image analysis included two common radiologic scores (Balthazar grade [25] and CT severity index [26]) and three additional variables (PAN equal area, PAN anatomical, ANC grade) that represent modifications of morphologic features that were helpful in predicting the prognosis of patients with acute pancreatitis in previous studies [24, 36]. In addition to the morphologic features on CECT, we analyzed a variety of prognostic laboratory and clinical variables [21, 37 40] to evaluate whether laboratory and clinical findings are helpful in predicting patient outcome in our study population; to compare laboratory, clinical, and imaging variables [5, 41]; and to analyze the additional information when combined [42, 43]. The temporal relation of the onset of acute pancreatitis to all prognostic variables was evaluated, which is a mandatory precondition for the dependable use of predictors and for comparison of different studies [4]. Following the comments of several authors [4, 41, 44], we did not analyze variables to predict a further predictor (e.g., severity of acute pancreatitis) but to predict six direct outcome parameters including the occurrence of a proven peripancreatic or pancreatic infection and mortality. The evaluation of these outcomes included time periods and events in peripheral clinics and an observation until day 250 after the onset of acute pancreatitis based on medical reports or telephone follow-up. Peripancreatic or pancreatic necrosis is a common starting point for a complicated and protracted clinical course [10, 12, 13] and is associated with increased morbidity and mortality [45]. Hence, a proven peripancreatic or pancreatic infection was chosen as an outcome parameter for ROC curve analysis. The CT severity index, PAN equal area, and PAN anatomical all of which consider intrapancreatic necrosis predicted peripancreatic or pancreatic infection most accurately, with AUCs of more than Persistent organ failure within the first week after the onset of acute pancreatitis and albumin level yielded the highest AUCs among the clinical and laboratory variables, respectively. Based on the ROC curve analysis for infection, the patient population was divided into a high-risk group and a low-risk group for each variable. Based on these groups, uni- and multivariable testing was carried out. Regarding the laboratory and clinical variables determined in the early phase of disease, albumin level, APACHE II score, and particularly persistent organ failure were the strongest independent predictors of patient outcome. Regarding the imaging variables, Balthazar grade and PAN equal area were the strongest independent predictors. In the multivariable analysis of all studied variables, imaging variables were independent and strong predictors of patient outcome and provided additional information in establishing prognosis compared with clinical and laboratory findings. The analysis of the imaging variables of the patient subgroup who underwent CECT 6 10 days after the onset of symptoms was limited by the number of patients (n = 34) but showed, by trend, results similar to the results of the entire study population. These results underline the recommendations of the Acute Pancreatitis Classification Working Group [1] to use persistent organ failure in the early phase and imaging variables in the late phase of disease for risk stratification. The imaging variables PAN equal area, PAN anatomical, and ANC grade which in our experience are relatively easy to apply proved to be comparable or superior predictors of patient outcome compared with the common imaging variables Balthazar grade and CT severity index. Hence, their use in daily practice might be favorable. Despite these notable results, the current study contains several limitations. First, 69% of the patients were transferred from peripheral hospitals, reflecting the fact that the University Medical Center Regensburg is the only tertiary referral center for a large geographic region. Hence, our study population was largely representative of patients with a severe clinical course of acute necrotizing pancreatitis and may not have included patients with milder forms of the disease. Nevertheless, our study population represents the characteristics of patients of a tertiary referral center. The second limitation is that the time interval between the onset of acute pancreatitis and performance of CECT varied, because CECT is not performed after a defined time period after the onset of acute pancreatitis at our hospital; rather, CECT is performed when the clinical condition or suspected complications justify performing CECT. In our study, a strict and, in our opinion, best possible procedure was applied to select and analyze the specific CECT examination that was performed most proximal to the beginning of the late phase of disease. Despite this potential limitation, the CECT-based findings showed remarkable performance in predicting patient outcome. Of note, in several other studies, including the original surveys of the Balthazar grade [25] and CT severity index [26], only the time interval between hospital admission and CT examinations was regarded. However, the onset of the first symptoms occurs usually earlier on the axis of time [4] and varies from patient to patient. In the current study, the onset of symptoms was analyzed and was considered for selection of the study CT examinations. The third limitation that is, the albumin and calcium values were available in only 86% of patients is due to the retrospective study design. C-reactive protein, WBC count, APACHE II score, and SAPS II were assessable in 96% of the patients. The fourth limitation is that the approach for performing interventions in patients with acute necrotizing pancreatitis varies among medical centers. This aspect needs to be considered with regard to the outcome parameter need for intervention. The fifth limitation is that the cutoffs used to assign the patients to the high- and lowrisk groups were based on the ROC curve analysis for a proven peripancreatic or pancreatic infection. We considered computing separate cutoffs for each outcome parameter. However, in our opinion, an approach using multiple cutoffs for each variable would pose difficulties in data interpretation and conclusions for clinical practice. Most cutoffs determined in the current study are in broad congruence with established cutoffs or observations for acute pancreatitis: In the Ranson and Imrie scores, the cutoff for age is 55 years. According to the World 1224 AJR:202, June 2014

11 Acute Necrotizing Pancreatitis Health Organization s definition of obesity, a cutoff for BMI of 30 kg/m 2 was used in studies of patients with acute pancreatitis [46 48]. The Imrie score [38] uses a cutoff for serum albumin of 32 g/l. In the Ranson score [39], cutoffs for serum calcium and WBC count are 2 mmol/l and /L respectively. Studies by Rau et al. [49] and Wilson et al. [50] found ideal cutoffs for C-reactive protein at 230 and 210 mg/l, respectively. The universally accepted cutoff value of the APACHE II score for predicting clinically severe disease and mortality is 8 [5]. Prospective studies on acute pancreatitis by Brisinda et al. [51], Wilson et al. [52], and Bollen et al. [5] used APACHE II scores 10 to identify patients at risk. No established cutoff value for SAPS II was found in the literature. In the original study by Balthazar et al. [25], in patients with a Balthazar grade > D, the rates of infected fluid collections and mortality were 61% and 17%, respectively, whereas in patients with Balthazar grade D the rates of infected fluid collections and mortality were 7% and 2%, respectively. In the study resulting in the CT severity index by Balthazar et al. [26], a CT severity index > 6 was the criterion to assign patients to the high-risk group. The cutoffs of the present study and of the aforementioned studies are, although determined in different study populations and by different means, approximately congruent, indicating that the cutoffs are of clinical significance. The sixth limitation is the relatively small numbers of events in particular, for the outcome parameter mortality. However, this survey was the largest to date to compare the use of laboratory, clinical, and imaging findings as predictors of outcome in patients with acute necrotizing pancreatitis. In conclusion, the results of our study underline the recommendation of the Acute Pancreatitis Classification Working Group [1] and provide evidence to our knowledge for the first time that morphologic features on CECT are independent and strong predictors of outcome in patients with acute necrotizing pancreatitis. In particular, imaging variables were powerful predictors of infection, and patients identified to be at increased risk for infection might profit from aggressive monitoring and treatment. Hence, in patients with suspected or proven acute necrotizing pancreatitis, performing CECT at the beginning of the late phase of disease is recommended to identify patients at increased risk for an adverse outcome. Acknowledgment We thank Dietlinde Ulsperger for editing the figures. References 1. Banks PA, Bollen TL, Dervenis C, et al.; Acute Pancreatitis Classification Working Group. Classification of acute pancreatitis: 2012 revision of the Atlanta classification and definitions by international consensus. Gut 2012; 62: Lankisch PG, Pflichthofer D, Lehnick D. No strict correlation between necrosis and organ failure in acute pancreatitis. Pancreas 2000; 20: Mofidi R, Duff MD, Wigmore SJ, Madhavan KK, Garden OJ, Parks RW. Association between early systemic inflammatory response, severity of multiorgan dysfunction and death in acute pancreatitis. Br J Surg 2006; 93: Delrue LJ, Waele JJ, Duyck PO. Acute pancreatitis: radiologic scores in predicting severity and outcome. Abdom Imaging 2010; 35: Bollen TL, Singh VK, Maurer R, et al. A comparative evaluation of radiologic and clinical scoring systems in the early prediction of severity in acute pancreatitis. Am J Gastroenterol 2011; 107: Mofidi R, Patil PV, Suttie SA, Parks RW. Risk assessment in acute pancreatitis. Br J Surg 2009; 96: Bollen TL. Imaging of acute pancreatitis: update of the revised Atlanta classification. Radiol Clin North Am 2012; 50: Casas JD, Díaz R, Valderas G, Mariscal A, Cuadras P. Prognostic value of CT in the early assessment of patients with acute pancreatitis. AJR 2004; 182: Simchuk EJ, Traverso LW, Nukui Y, Kozarek RA. Computed tomography severity index is a predictor of outcomes for severe pancreatitis. Am J Surg 2000; 179: Beger HG, Bittner R, Block S, Büchler M. Bacterial contamination of pancreatic necrosis: a prospective clinical study. Gastroenterology 1986; 91: Papachristou GI, Whitcomb DC. Predictors of severity and necrosis in acute pancreatitis. Gastroenterol Clin North Am 2004; 33: Mortele KJ, Wiesner W, Intriere L, et al. A modified CT severity index for evaluating acute pancreatitis: improved correlation with patient outcome. AJR 2004; 183: Trout AT, Elsayes KM, Ellis JH, Francis IR. Imaging of acute pancreatitis: prognostic value of computed tomographic findings. J Comput Assist Tomogr 2010; 34: Vriens PW, van de Linde P, Slotema ET, Warmerdam PE, Breslau PJ. Computed tomography severity index is an early prognostic tool for acute pancreatitis. J Am Coll Surg 2005; 201: Kemppainen E, Sainio V, Haapiainen R, Kivisaari L, Kivilaakso E, Puolakkainen P. Early localization of necrosis by contrast-enhanced computed tomography can predict outcome in severe acute pancreatitis. Br J Surg 1996; 83: De Waele JJ, Delrue L, Hoste EA, De Vos M, Duyck P, Colardyn FA. Extrapancreatic inflammation on abdominal computed tomography as an early predictor of disease severity in acute pancreatitis: evaluation of a new scoring system. Pancreas 2007; 34: Leung T, Lee C, Lin S, et al. Balthazar computed tomography severity index is superior to Ranson criteria and APACHE II scoring system in predicting acute pancreatitis outcome. World J Gastroenterol 2005; 11: Spanier B, Nio Y, van der Hulst R, Tuynman H, Dijkgraaf M, Bruno M. Practice and yield of early CT scan in acute pancreatitis: a Dutch observational multicenter study. Pancreatology 2010; 10: Banks PA, Freeman ML. Practice guidelines in acute pancreatitis. Am J Gastroenterol 2006; 101: Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985; 13: Le Gall JR, Lemeshow S, Saulnier F. A new simplified acute physiology score (SAPS II) based on a European/North American multicenter study. JAMA 1993; 270: Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL, Sibbald WJ. Multiple organ dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med 1995; 23: Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis: the ACCP/ SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest 1992; 101: Heiss P, Bruennler T, Salzberger B, et al. Severe acute pancreatitis requiring drainage therapy: findings on computed tomography as predictor of patient outcome. Pancreatology 2010; 10: Balthazar EJ, Ranson JH, Naidich DP, Megibow AJ, Caccavale R, Cooper MM. Acute pancreatitis: prognostic value of CT. Radiology 1985; 156: Balthazar EJ, Robinson DL, Megibow AJ, Ranson JH. Acute pancreatitis: value of CT in establishing prognosis. Radiology 1990; 174: Borley NR, Standring S, Gray H. Gray s anatomy: the anatomical basis of clinical practice, 40th ed. Edinburgh, Scotland: Churchill Livingstone, Raptopoulos V, Kleinman PK, Marks S, Snyder M, Silverman PM. Renal fascial pathway: posterior extension of pancreatic effusions within the anterior pararenal space. Radiology 1986; 158: Meyers MA, Charnsangavej C, Oliphant M. Meyers dynamic radiology of the abdomen: normal AJR:202, June

12 Brand et al. and pathologic anatomy, 6th ed. New York, NY: Springer-Verlag, Bradley EL 3rd. A clinically based classification system for acute pancreatitis: summary of the International Symposium on Acute Pancreatitis, Atlanta, Ga, September 11 through 13, Arch Surg 1993; 128: Göçmen E, Klc YA, Yoldaş O, et al. Comparison and validation of scoring systems in a cohort of patients treated for biliary acute pancreatitis. Pancreas 2007; 34: DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: Muche R. Applied survival analysis: regression modeling of time to event data. (book review) Int J Epidemiol 2001; 30: Bradburn MJ, Clark TG, Love SB, Altman DG. Survival analysis. Part III. Multivariate data analysis: choosing a model and assessing its adequacy and fit. Br J Cancer 2003; 89: Spanier BWM, Bruno MJ. Use of early CT scanning in patients with acute pancreatitis. Radiology 2011; 260: Ishikawa K, Idoguchi K, Tanaka H, et al. Classification of acute pancreatitis based on retroperitoneal extension: application of the concept of interfascial planes. Eur J Radiol 2006; 60: Neoptolemos JP, Kemppainen EA, Mayer JM, et al. Early prediction of severity in acute pancreatitis by urinary trypsinogen activation peptide: a multicentre study. Lancet 2000; 355: Imrie CW. Prognostic indicators in acute pancreatitis. Can J Gastroenterol 2003; 17: Ranson JH, Rifkind KM, Roses DF, Fink SD, Eng K, Spencer FC. Prognostic signs and the role of operative management in acute pancreatitis. Surg Gynecol Obstet 1974; 139: Johnson CD. Persistent organ failure during the first week as a marker of fatal outcome in acute pancreatitis. Gut 2004; 53: Gravante G, Garcea G, Ong S, et al. Prediction of mortality in acute pancreatitis: a systematic review of the published evidence. Pancreatology 2009; 9: Balthazar EJ. Acute pancreatitis: assessment of severity with clinical and CT evaluation. Radiology 2002; 223: Robert JH, Frossard JL, Mermillod B, et al. Early prediction of acute pancreatitis: prospective study comparing computed tomography scans, Ranson, Glasgow, Acute Physiology and Chronic Health Evaluation II scores, and various serum markers. World J Surg 2002; 26: De Waele JJ, Blot SI. Severity prediction in acute pancreatitis using the CT severity index: a selffulfilling prophecy. JOP 2006; 7: Petrov MS, Shanbhag S, Chakraborty M, et al. Organ failure and infection of pancreatic necrosis as determinants of mortality in patients with acute pancreatitis. Gastroenterology 2010; 139: Patients were 18 years and had been assigned a K85 diagnosis code between January 2005 and December Patients had an episode of acute pancreatitis and underwent CECT within 21 days of the onset of symptoms All CECT and MRI studies for 139 patients were reviewed 187 Patients were excluded: 73 Patients had acute-on-chronic pancreatitis 66 Patients had a malignancy 21 Patients had undergone pancreatic resection 15 Patients had undergone abdominal surgery 14 days before symptom onset 12 Patients had acute pancreatitis that was secondary to a severe illness 40 Patients were excluded: 2 Patients had CECT examinations that were of insufficient quality 3 Patients had pancreatic calcifications (indicating acute-on-chronic pancreatitis) 35 Patients had interstitial edematous pancreatitis 99 Patients with acute necrotizing pancreatitis were included in the study 46. Johnson CD, Toh SKC, Campbell MJ. Combination of APACHE-II score and an obesity score (APACHE-O) for the prediction of severe acute pancreatitis. Pancreatology 2004; 4: Papachristou GI, Papachristou DJ, Avula H, Slivka A, Whitcomb DC. Obesity increases the severity of acute pancreatitis: performance of APACHE-O score and correlation with the inflammatory response. Pancreatology 2006; 6: Martínez J, Johnson CD, Sánchez-Payá J, de Madaria E, Robles-Díaz G, Pérez-Mateo M. Obesity is a definitive risk factor of severity and mortality in acute pancreatitis: an updated meta-analysis. Pancreatology 2006; 6: Rau B, Steinbach G, Baumgart K, Gansauge F, Grünert A, Beger HG. Serum amyloid A versus C-reactive protein in acute pancreatitis: clinical value of an alternative acute-phase reactant. Crit Care Med 2000; 28: Wilson C, Heads A, Shenkin A, Imrie CW. C-reactive protein, antiproteases and complement factors as objective markers of severity in acute pancreatitis. Br J Surg 1989; 76: Brisinda G, Maria G, Ferrante A, Civello IM. Evaluation of prognostic factors in patients with acute pancreatitis. Hepatogastroenterology 1999; 46: Wilson C, Heath DI, Imrie CW. Prediction of outcome in acute pancreatitis: a comparative study of APACHE II, clinical assessment and multiple factor scoring systems. Br J Surg 1990; 77: Fig. 1 Flowchart outlines inclusion of patients. K85 diagnosis code is defined as acute pancreatitis according to International Statistical Classification of Diseases and Related Health Problems (10th revision, German Modification). CECT = contrast-enhanced CT AJR:202, June 2014

13 Acute Necrotizing Pancreatitis Head SMV Corpus Tail A Fig. 2 Partitioning of pancreas based on two classifications used to analyze distribution of necrosis. Drawings show pancreas in craniocaudal view. PAN = number of parts of pancreas that exhibited areas of necrosis, SMV = superior mesenteric vein. A, Drawing shows partitioning of pancreas based on variable abbreviated as PAN equal area. Pancreas was divided into head, corpus, and tail according to PAN equal area by two planes that run parallel to long axis of body, perpendicular to pancreatic duct, and in manner so that craniocaudal projection of three parts onto transverse plane exhibited equal area. B, Drawing shows partitioning of pancreas based on variable abbreviated as PAN anatomical. Pancreas was divided into extended head (parts of pancreas ventral as well as dorsal and to right of SMV and portal vein), corpus (between extended head and tail), and tail (defined as most lateral 3.5 cm of pancreas to left). A Head SMV Corpus Fig year-old man with acute necrotizing pancreatitis who was discharged from hospital 22 days after admission. Transverse contrast-enhanced CT depicts two areas of pancreas with nonenhancement representing areas of necrosis. PAN = number of parts of pancreas that exhibited areas of necrosis. A, Based on classification variable abbreviated as PAN equal area, both areas of necrosis (arrows) are located within tail (T) of pancreas. Black line represents border that is considered to separate corpus (C) and tail of pancreas. No necrosis of corpus is identified. v = lineal vein. B, Based on classification variable abbreviated as PAN anatomical, there are areas of necrosis (arrows) within corpus (C) and tail (T), which differs from findings based on PAN equal area variable shown in A. Black line represents border that is considered to separate corpus and tail. v = lineal vein. B Tail B AJR:202, June

14 Brand et al. A C Fig year-old woman with acute necrotizing pancreatitis who was discharged from hospital 75 days after admission. PAN = number of parts of pancreas that exhibited areas of necrosis. A C, Transverse contrast-enhanced CT scans at level of pancreatic head (2 cm below confluence of portal and lineal veins) (A and B) and at level of corpus (C) depict areas of necrosis (arrows) of pancreatic head (H, A and B) and corpus (C, A and C). White lines represent borders that are considered to separate head and corpus of pancreas based on variable abbreviated as PAN equal area (A) and variable abbreviated as PAN anatomical (B and C). In this case, areas of necrosis were congruently identified within head and corpus of pancreas using both classifications. Compared with PAN equal area border that separates head and corpus (A), border that separates head and corpus according to PAN anatomical is placed more to left of pancreas (B and C). v = contrast-enhancing vital pancreatic tissue of head and corpus, ANC = acute necrotic collection, asterisks in A and B = superior mesenteric vein, pv in C = portal vein, lv in C = lineal vein. Fig year-old man with acute necrotizing pancreatitis who was discharged from hospital 22 days after admission. Transverse contrast-enhanced CT scan shows acute necrotic collection (ANC) at anterior pararenal space to right (arrows). ANC extends laterally to white line that runs anteroposteriorly through medial border of adjacent colon ascendens (C). Hence, ANC at retrocolic part (RCP) of anterior pararenal space is present (ANC grade 1). Fig year-old man with acute necrotizing pancreatitis and proven infection who was discharged from hospital 59 days after admission. Transverse contrast-enhanced CT scan displays faint changes of density within fatty tissue of posterior pararenal space to left (straight arrows). It was assumed that needle puncture for example, of spot marked with asterisk would allow aspirating at least one fluid drop. Hence, acute necrotic collections (ANCs) at posterior pararenal space are present (ANC grade 2). In addition, ANC is present at retrocolic part of left anterior pararenal space (curved arrows). C = colon descendens. B 1228 AJR:202, June 2014