European Journal of Radiology 52 (2004) 67–72
Peripancreatic vascular abnormalities complicating acute pancreatitis: contrast-enhanced helical CT findings Koenraad J. Mortelé a,∗ , Patricia J. Mergo b , Helena M. Taylor b , Walter Wiesner a , Vito Cantisani a , Michael D. Ernst c , Babak N. Kalantari a , Pablo R. Ros a a
Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, USA Department of Radiology, University of Florida College of Medicine, Health Science Center, P.O. Box 100374, Gainesville, FL 32610-0374, USA c Division of Biostatistics, University of Florida College of Medicine, Health Science Center, P.O. Box 100374, Gainesville, FL 32610-0374, USA
b
Received 16 May 2003; received in revised form 6 October 2003; accepted 9 October 2003
Abstract Objective: To determine the prevalence and morphologic helical computed tomography (CT) features of peripancreatic vascular abnormalities in patients with acute pancreatic inflammatory disease in correlation with the severity of the pancreatitis. Materials and methods: One hundred and fifty-nine contrast-enhanced helical CT scans of 100 consecutive patients with acute pancreatitis were retrospectively and independently reviewed by three observers. CT scans were scored using the CT severity index (CTSI): pancreatitis was graded as mild (0–2 points), moderate (3–6 points), and severe (7–10 points). Interobserver agreement for both the CT severity index and the presence of peripancreatic vascular abnormalities was calculated (K-statistic). Correlation between the prevalence of complications and the degree of pancreatitis was estimated using Fisher’s exact test. Results: The severity of pancreatitis was graded as mild (n = 59 scans), moderate (n = 82 scans), and severe (n = 18 scans). Venous abnormalities detected included splenic vein (SV) thrombosis (31 scans, 19 patients), superior mesenteric vein (SMV) thrombosis (20 scans, 14 patients), and portal vein (PV) thrombosis (17 scans, 13 patients). Arterial hemorrhage occurred in five patients (6 scans). In our series, no cases of arterial pseudoaneurysm formation were detected. The interobserver agreement range for scoring the degree of pancreatitis and the overall presence of major vascular abnormalities was 75.5–79.2 and 86.2–98.8%, respectively. The presence of the vascular abnormalities in correlation with the severity of pancreatitis was variable. Conclusion: Vascular abnormalities are relatively common CT findings in association with acute pancreatitis. The CT severity index is insufficiently accurate in predicting some of these complications since no statistically significant correlation between their prevalence and the severity of pancreatitis could be established. © 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: Pancreas; Diseases; Pancreatitis; Vascular complications; Computed tomography (CT); Helical; Pancreas; CT
1. Introduction Acute pancreatitis is a common illness with a relatively unpredictable severity course. In most patients, fortunately, the acute pancreatic inflammatory disease is mild, only presenting with minimal abdominal pain that usually resolves within a few days [1]. However, in 20–30% of cases, the pancreatitis can be severe and associated with potentially life-treathening complications such as extensive pancreatic necrosis, pancreatic abscess formation, multi-organ failure,
∗ Corresponding author. Tel.: +1-617-732-7624; fax: +1-617-732-6317. E-mail address:
[email protected] (K.J. Mortel´e).
and development of peripancreatic vascular complications [1]. Abdominal contrast-enhanced helical computed tomography (CT) is an excellent method for evaluating vascular involvement in patients with acute pancreatitis [2]. Helical CT not only establishes accurately the diagnosis of pancreatitis and determines the extent of the inflammatory process, it also allows optimal visualization of peripancreatic vascular structures and identification of vascular involvement. The latter is achieved by the capability of helical CT to increase speed of acquisition, decrease scanning time, and the ability to perform contrast-enhanced studies following a bolus injection of intravenous contrast material [2,3]. Moreover, CT allows predicting the severity of the pancreatitis at initial and follow-up examinations using the CT severity
0720-048X/$ – see front matter © 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ejrad.2003.10.006
68
K.J. Mortel´e et al. / European Journal of Radiology 52 (2004) 67–72
index (CTSI), as developed by Balthazar et al. [4]. Although the CT appearances of a vast array of peripancreatic vascular complications in patients with acute pancreatitis are well documented in radiologic literature, information is lacking about their prevalence and CT detectability in a relatively large controlled study group [5–10]. Therefore, the purpose of our study was to evaluate the prevalence and morphologic CT features of various vascular abnormalities complicating acute pancreatitis. In addition, we assessed the prevalence of these complications according to the severity of the pancreatitis.
2. Materials and methods 2.1. Subjects During a 17-month period, helical CT imaging was performed in 100 consecutive patients (47 women and 53 men; age range 12–80 years (mean: 51 years) with a clinical and biochemical diagnosis of acute pancreatic inflammatory disease. The etiology of the pancreatitis included: idiopathic pancreatitis (n = 25), alcoholic pancreatitis (n = 14), biliary pancreatitis (n = 9), iatrogenic pancreatitis (n = 22) (post-ERCP (n = 10); drug-induced (n = 6), post-surgery (n = 6)), acute exacerbation of chronic pancreatitis (n = 15), association with pancreatic mass (n = 6), hypertriglyceridemia/hyperlipidemia (n = 4), traumatic pancreatitis (n = 2), autoimmune pancreatitis (n = 2), and hereditary pancreatitis (n = 1). Fifty-four patients experienced previous episodes of acute pancreatitis. The interval between diagnosis and CT evaluation was less than 1 week in 90% of patients (range: 0–7 days; mean: within 0.6 day), and slightly longer in the remainder 10 patients (range: 8–21 days; mean: 13 days). One hundred and fifty-nine contrast-enhanced helical CT scans were included in the review process: we interpreted the 100 scans performed at admission in each of these 100 patients, extended with all consecutive second and third follow-up scans in 43 and 16 patients, respectively. 2.2. Imaging technique All scans were performed on a helical CT unit (HiLite Advantage; General Electric Medical Systems, Milwaukee, WI). Contrast-enhanced CT (CECT) images were obtained following intravenous administration of 120–150 ml of iodinated contrast material, injected at a rate of 2–4 ml/s using a mechanical power injector. Opacification of the digestive tract was obtained by oral administration of 900 ml of positive contrast agent prior to the examination. Cranio-caudal pancreatic helical scanning was performed during a breath-hold period of 24–32 s, and commenced 30–50 s from the start of the contrast injection (pancreatic parenchymal phase). All scans were obtained using a collimation width of 5 mm, table increment of
7 mm/s (pitch = 1.2), and reconstruction slice width of 5 mm. 2.3. Image analysis One hundred and fifty-nine contrast-enhanced CT scans were retrospectively and independently reviewed on hard copy films by three observers. Initially, we scored the grade of the pancreatitis using the CT severity index developed by Balthazar et al. [4] (Table 1). The severity of pancreatitis was categorized as mild (score: 0–2 points), moderate (score: 3–6 points), or severe (score: 7–10 points). Subsequently, the presence of vascular abnormalities was determinated. Findings specifically recorded, included arterial vascular compromise (pseudoaneurysm formation, arterial hemorrhage) and venous thrombosis. Arterial hemorrhage was defined as focal contrast extravasation within the pancreas or peripancreatic tissue. Venous thrombosis was present to our criteria if an intraluminal hypoattenuating thrombus was present (acute and subacute thrombosis) or the vein was undetectable (chronic thrombosis). 2.4. Statistical analysis Interrater agreement for the severity of pancreatitis and presence of vascular complications was expressed by means of exact-percentage agreement, along with the K-statistic, which is used to estimate the proportion of interrater agreement above that expected by chance. The K-statistic was calculated for each study parameter. A weighted K-statistic of 0.41–0.60 was considered to indicate moderate agreement, a weighted K-statistic of 0.61–0.80 was considered to indicate good agreement, and a weighted K-statistic of 0.81–1.00 was considered to indicate excellent agreement. To correlate the prevalence of the complications and the severity of the pancreatitis, a consensus score for both was obtained. Consensus was determinated if results of three Table 1 CT severity index Prognostic indicators
Characteristics
Points
Pancreatic inflammation
Normal pancreas Focal or diffuse enlargement of the pancreas Intrinsic pancreatic abnormalities with inflammatory changes in the peripancreatic fat Single, ill-defined fluid collection or phlegmon Two or more poorly defined collections or presence of gas in or adjacent to the pancreas
0 1
No necrosis 30% or less 30–50% Greater than 50%
0 2 4 6
Pancreatic necrosis
2
3 4
K.J. Mortel´e et al. / European Journal of Radiology 52 (2004) 67–72
69
or two out of three readers agreed. Subsequent correlation between the prevalence of the complications and the severity of the pancreatitis was obtained by the use of Fisher’s exact test.
3. Results 3.1. CT severity index (CTSI) According to the CTSI, the severity of the acute pancreatitis in our series (159 scans) was graded as mild (n = 59), moderate (n = 82), and severe (n = 18), respectively. Readers agreed on these observations 75.5–79.2% of time. The K-statistic for each pair of readers ranged from 0.59 to 0.65, indicating moderate to good agreement.
Fig. 1. Splenic vein thrombosis. Axial contrast-enhanced CT image in a patient with mild pancreatitis shows intraluminal hypo-attenuating thrombus in the splenic vein (black arrow).
Table 2 Interrater agreement (presence of peripancreatic vascular complications)
3.2. Vascular involvement Significant venous vascular abnormalities observed included thrombosis of the splenic vein (SV), superior mesenteric vein (SMV), and portal vein (PV). Thrombosis of the SV, SMV, and PV was judged to be present in 31, 20, and 17 scans, respectively (19, 14, and 13 patients) (Figs. 1–3). Interobserver agreement for the presence of major venous thrombosis ranged from 86.2 to 95.6% with K values between 0.45 and 0.82. According to the previously defined subgroups of CT severity indexes, the prevalence of thrombosis of the SV, SMV, and PV was 1.7, 22.0, 66.7; 0, 15.9, 38.9; and 1.7, 14.6, 22.2%, respectively (Table 2). Arterial hemorrhage was detected in 6 scans (five patients) (Fig. 4). The interobserver agreement for the presence of
Portal vein (PV) thrombosis Superior mesenteric vein (SMV) thrombosis Splenic vein (SV) thrombosis Arterial hemorrhage Pseudoaneurysm
Agreement (%)
Kappa
86.2–92.5 93.7–95.6
0.46–0.58 0.73–0.80
88.7–94.3 96.9–98.7 98.1–99.4
0.62–0.82 0.60–0.79 NA
arterial hemorrhage ranged from 96.7 to 98.8%, and K values between 0.60 and 0.79. Arterial hemorrhage, according to the different subgroups of pancreatitis, was present in 0, 6.1, and 5.6% of patients, respectively. In our large series of 100 patients, no cases of arterial pseudoaneurysm formation were detected.
Fig. 2. Superior mesenteric vein thrombosis. Axial CECT image in a patient with moderate pancreatitis shows non-enhancing intraluminal thrombus in the superior mesenteric vein (arrow). Note the enhancement of the venous wall due to opacification of the vasa vasorum (arrowhead).
70
K.J. Mortel´e et al. / European Journal of Radiology 52 (2004) 67–72
Fig. 3. Portal vein thrombosis. Axial CECT image in a patient with moderate pancreatitis shows complete thrombosis of the portal vein (arrowheads) and hypertrophic hepatic artery (arrow).
Fig. 4. Arterial hemorrhage. Axial CECT image in a patient with moderate pancreatitis shows necrosis of pancreatic body, intraluminal thrombus within the portal vein (arrow), and acute hematoma within the right abdominal wall. At the level of the pancreatic body, a small area of contrast extravasation (arrowheads), compatible with intrapancreatic hemorrhage, is demonstrated.
3.3. Correlation of prevalence of complications and severity of pancreatitis Statistical calculations (Fisher’s exact test) showed a significant correlation between the presence of splenic vein
thrombosis and superior mesenteric vein thrombosis, and the severity of the pancreatitis (P < 0.05). No statistically significant correlation could be calculated for the presence of portal vein thrombosis and arterial hemorrhage (P > 0.05) (Table 3).
K.J. Mortel´e et al. / European Journal of Radiology 52 (2004) 67–72 Table 3 Correlation of prevalence of complications and severity of pancreatitis
PV thrombosis SMV thrombosis SV thrombosis Arterial hemorrhage
Moderate (3–6 points) (%)
Severe (7–10 points) (%)
P-value
14.6 15.9 22.0 6.1
22.2 38.9 66.7 5.6
0.4796 0.0463 0.0004 1
4. Discussion The role and accuracy of helical CT in the diagnosis and staging of acute pancreatitis, as well as its contribution to prognostic significance, has been widely reported in literature [2,3,11–14]. Balthazar and co-workers focused on both the presence and degree of peripancreatic fluid collections and pancreatic necrosis, as seen on contrast-enhanced CT, to develop a CT severity index [4,11]. In this attempt to enhance the value of CT as a predictive indicator for the severity of the pancreatitis, the CTSI was defined to provide an easily attained and widely accessible grading system [11]. Although excellent correlation between the occurrence of most morbidity and mortality in patients with acute pancreatitis and the CTSI score has been established in the past, it is uncertain if the development of some major extra-pancreatic complications are necessary related to the severity of the acute pancreatic inflammatory disease [4]. Among these, the occurrence of peripancreatic vascular complications has not been previously studied with CT for this purpose. However, Dörffel et al. observed using color-Doppler ultrasound that vascular complications were significantly more frequent in alcohol-induced than in gallstone-induced pancreatitis. The prevalence of vascular complications in acute pancreatitis was found in up to 57% of the patients with acute necrotizing pancreatitis [15]. Peripancreatic venous thrombosis is a well-recognized complication of acute pancreatitis and thought to be related to stasis, spasm, and mass effect from surrounding inflamed pancreas as well as direct intimal damage of the venous wall by the liberated enzymes [5–7]. The CT features of acute or subacute venous thrombosis consist of the presence of an enlarged vein with a non-enhancing low-attenuation center and ring enhancement due to opacification of the vasa vasorum [5–7]. In more chronic thrombosis, the vein may be normal sized or small, with sometimes non-visualization, and opacification of collateral veins [5–7]. Thrombosis of the major venous tributaries is known to be associated with potential emergent complications and obviates timely and proper management. In cases of splenic vein thrombosis without portal vein thrombosis, for example, direct drainage of the splenic blood will take place through the short gastric veins and left gastric vein into the portal vein [6]. Due to this increased flow within the gastric wall veins, vein dilatation and eventually gastric varices result [6]. This
71
phenomenon, also called segmental left-sided portal hypertension, may evoke severe gastrointestinal hemorrhage from varices, and urgent splenectomy may be considered. Similarly, in patients with pancreatitis-induced acute portal vein thrombosis with or without portal hypertension, intravascular thrombolysis or surgical porto-systemic shunting has to be envisioned [6]. Major venous thrombosis was relatively common in our series and involved, in decreasing order of frequency, thrombosis of the splenic (19% of patients), superior mesenteric (14%), and portal vein (13%). Although the presence of splenic and superior mesenteric vein thrombosis was significantly correlated with the severity of the pancreatitis, thrombosis of the portal vein was not. This finding is important since it implicates that clinicians and radiologists have to be aware of this complication even in patients with “moderate” pancreatitis, and therapeutic strategies have to modified individually. Identical considerations have to be made regarding the presence of peripancreatic arterial hemorrhage in acute pancreatitis. Arterial hemorrhage in patients with pancreatitis results of autodigestion of arterial walls by pancreatic enzymes, especially elastase, and is considered the most severe manifestation of the spectrum of vascular complications [5]. Acute hemorrhage may develop from direct vascular erosion, or rupture of a pseudoaneurysm [8,10]. The surgical and radiological literature has stressed the importance of its early detection, since severe peripancreatic hemorrhage can have catastrophic consequences, with an estimated mortality of 37% [8]. Although peripancreatic arterial hemorrhage was relatively uncommon (5%) in our series, there were also no significant data that showed a correlation between the severity of the pancreatitis and the occurrence of this complication. In our study, including 100 patients, no cases of radiographically identifiable pseudoaneurysm formation were seen. This rarity of prevalence is in contrary with previous reports which estimated its frequency to be as high as 10% [8]. Except of potential population sampling errors—especially the low incidence (18%) of severe pancreatitis in our series, or other unrecognized factors, the reason for the discrepancy is unclear. Although speculative, it has to be considered that probably not all of the aneurysms can be identified on CT, even in retrospect, due to the coexistence of hemorrhage, vascular obliteration and distortion. Furthermore, pseudoaneurysm formation typically manifests itself as a late complication of acute pancreatitis and does usually not develop within the first weeks or months after diagnosis [8,14]. In conclusion, our data suggest that major venous thrombosis is a relatively common helical CT finding in association with acute pancreatitis. Arterial hemorrhage is infrequently detected and pseudoaneurysm formation is exceedingly rare. The CT severity index is insufficiently accurate in predicting some of these major complications since no significant correlation between their prevalence and the severity of pancreatitis could be established.
72
K.J. Mortel´e et al. / European Journal of Radiology 52 (2004) 67–72
References [1] Steer ML. Pathophysiology and pathogenesis of acute pancreatitis. In: Bradley III EL (Ed.), Acute pancreatitis: diagnosis and therapy. New York: Raven Press; 1994. [2] Urban BA, Curry CA, Fishman EK. Complications of acute pancreatitis: helical CT evaluation. Emerg Radiol 1999;6:113–20. [3] Bonaldi VM, Bret P, Atri G, Garcia P, Reinhold C. A comparison of two injection protocols using helical and dynamic acquisitions in CT examinations of the pancreas. AJR 1996;167:49–55. [4] Balthazar EJ, Robinson DL, Megibow AJ, Ranson JH. Acute pancreatitis; value of CT in establishing prognosis. Radiology 1990;174:331–6. [5] Vujic I. Vascular complications of pancreatitis. Radiol Clin North Am 1989;27:81–91. [6] Belli M, Jennings CM, Nakielny RA. Splenic and portal venous thrombosis: a vascular complication of pancreatic disease demonstrated on computed tomography. Clin Radiol 1990;41:13–6. [7] Crowe PM, Sagar G. Reversible superior mesenteric vein thrombosis in acute pancreatitis. The CT appearances. Clin Radiol 1995;50:628– 33.
[8] Burke JW, Erickson SJ, Kellum CD, Tegtmeyer CJ, Williamson BRJ, Hansen MF. Pseudoaneurysms complicating pancreatitis: detection by CT. Radiology 1986;161:447–50. [9] Mauro MA, Schiebler ML, Parker LA, Jaques PF. The spleen and its vasculature in pancreatitis: CT findings. Am Surg 1993;3:155– 9. [10] Lee MJ, Saini S, Heller SC, Warschaw AL, Mueller PR. Pancreatitis with pseudoaneurysm formation: a pitfall for the interventional radiologist. AJR 1991;156:97–8. [11] Balthazar EJ. Acute pancreatitis: assessment of severity with clinical and CT evaluation. Radiology 2002;223:603–13. [12] Robinson PJA, Sheridan MB. Pancreatitis: computed tomography and magnetic resonance imaging. Eur Radiol 2000;10:401– 8. [13] Chalmers AG. The role of imaging in acute pancreatitis. Eur J Gastroenterol Hepatol 1997;9:106–16. [14] Merkle EM, Görich J. Imaging of acute pancreatitis. Eur Radiol 2002;12:1979–92. [15] Dörffel T, Wruck T, Rukert RI, Romaniuk P, Dorffel O, Wermke W. Vascular complication in acute pancreatitis assessed by color duplex ultrasonography. Pancreas 2002;21:126–33.
