CT Findings of Afferent Loop Varices After Bilioenteric Anastomosis in Patients With Malignant Disease

Gastrointestinal Imaging Original Research Lee et al. CT Findings of Afferent Loop Varices Gastrointestinal Imaging Original Research Dong Ho Lee 1 Young Hoon Kim 2 Yoon Jin Lee 2 Kyoung Ho Lee 2 So Yeon Kim 3 Jai Young Cho 4 Yoo-Seok Yoon 4 Ho-Seong Han 4 Lee DH, Kim YH, Lee YJ, et al. Keywords: choledochostomy, diagnosis, gastrointestinal hemorrhage, MDCT, varicose veins DOI:10.2214/AJR.12.9172 Received April 18, 2012; accepted after revision September 14, 2012. 1 Department of Radiology, Seoul National University Hospital, Seoul, Korea. 2 Department of Radiology, Seoul National University Bundang Hospital, Seoul National University College of Medicine, Institute of Radiation Medicine, Seoul National University Medical Research Center, 300 Gumi-dong, Bundang-gu, Seongnam-si, Gyeonggi-do, 463-707, Korea. Address correspondence to Y. H. Kim (yhkrad@gmail.com). 3 Department of Radiology and Research Institute of Radiology, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Korea. 4 Department of Surgery, Seoul National University Bundang Hospital, Seoul National University College of Medicine, Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, Korea. AJR 2013; 200:1261 1268 0361 803X/13/2006 1261 American Roentgen Ray Society CT Findings of Afferent Loop Varices After Bilioenteric Anastomosis in Patients With Malignant Disease OBJECTIVE. The purpose of this study was to determine the prevalence and CT findings of afferent jejunal loop varices after bilioenteric anastomosis in patients with malignancy. MATERIALS AND METHODS. The study included 99 patients who underwent bilioenteric anastomosis for malignancy and follow-up CT examinations. Two abdominal radiologists reviewed CT images for the presence of afferent loop varices. They also assessed for the presence of extrahepatic portal occlusion or stenosis, other collateral vessels, ascites, splenomegaly, and recurrent tumor. Clinical data regarding age, sex, and histories of pancreaticoduodenectomy and surgery on the portal vein, radiation therapy, and postoperative gastrointestinal bleeding were collected. Univariate and multivariate analyses were performed to discern significant factors associated with afferent loop varices. Cumulative incidence was estimated with the Kaplan-Meier method. RESULTS. Afferent loop varices were found in 22 of the 99 patients (22.2%), and the 48-month cumulative incidence of afferent loop varices was 28.0%. Three of the 22 patients (13.6%) eventually experienced gastrointestinal bleeding. Obstruction in the extrahepatic portal system was the only feature that exhibited a significant difference in multivariate analysis (relative risk, 71.4; 95% CI, 7.4 1000). CONCLUSION. Afferent loop varices detected after bilioenteric anastomosis on CT images had a prevalence of 22.2% and can result in gastrointestinal bleeding. The presence of obstruction in the extrahepatic portal system was significantly associated with afferent loop varices. P ostoperative bleeding is one of the most serious postoperative complications. The causes of postoperative bleeding are varied, and treatment mainly depends on the site. Therefore, finding the exact cause and site of bleeding is imperative [1, 2]. Ectopic varices consist of enlarged portosystemic venous collaterals located anywhere other than the gastroesophageal region [3, 4]. The diagnosis of ectopic varices is challenging, as is identification of bleeding from them [5]. Afferent jejunal loop varices can be detected along the hepaticojejunostomy after bilioenteric anastomosis [6], and these varices have been reported as a cause of postoperative gastrointestinal bleeding [7 13]. Accurate diagnosis of afferent jejunal loop varices as a cause of bleeding is essential to avoid unnecessary invasive procedures, such as arteriography and laparotomy. Because the bleeding does not originate from the artery, the diagnosis of bleeding from ectopic varices, such as afferent loop varices, can be easily overlooked, even with angiography. Be- cause CT is routinely used for the evaluation of postoperative complications and gastrointestinal bleeding, it is important to identify the exact incidence of afferent jejunal loop varices and characterize the CT findings. Results of only a few radiology studies on afferent jejunal loop varices have been published. These previous reports [6, 12] involved limited cohorts (fewer than 30 patients), and prevalence, cumulative incidence, and CT findings associated with the development of afferent jejunal loop varices after bilioenteric anastomosis were not assessed. The purposes of our study were to determine the prevalence and CT findings of afferent jejunal loop varices after bilioenteric anastomosis in patients with various malignant diseases involving the pancreaticobiliary system and to identify the determinant factor for development of afferent jejunal loop varices. Materials and Methods Our institutional review board approved this retrospective study, and the requirement for informed consent was waived. AJR:200, June 2013 1261

Lee et al. Patients who underwent bilioentric anastomosis (n = 120) Patients with afferent loop varices (n = 22) Occlusion in extrahepatic portal vein (n = 19) Fig. 1 Flow diagram of study sample. Included patients (n = 99) Excluded patients who did not undergo follow-up CT (n = 21) Patients without afferent loop varices (n = 77) Occlusion in extrahepatic portal vein (n = 3) Patients A computerized search of the medical records at our hospital identified the records of 120 patients who had undergone bilioenteric anastomosis for various malignant diseases involving the pancreaticobiliary system from May 2003 to May 2008. Among these 120 patients, 21 patients were excluded from this study because follow-up CT was not performed. The other 99 patients (59 men, 40 women; age range, 35 87 years; mean age, 64.7 ± 9.4 [SD] years) were included in the study (Fig. 1). All of these patients had complete medical and surgical records and underwent follow-up CT examinations at least 6 months after surgery. In all cases the follow-up CT examination was performed for the detection of any kind of surgery-related complication and of any kind of disease recurrence at 3- or 6-month follow-up intervals. No patient had any symptoms or signs of liver disease or drug or alcohol abuse. No patient had abnormalities suggesting chronic liver disease or portal hypertension during surgery. The underlying diseases in these patients were bile duct cancer (n = 33), pancreatic cancer (n = 34), ampullary cancer (n = 24), invasive intraductal papillary mucinous carcinoma of the pancreas (n = 3), duodenal cancer (n = 3), gallbladder cancer (n = 1), and stomach cancer (n = 1). In total, 57 patients underwent pylorus-preserving pancreaticoduodenectomy, 22 underwent segmental bile duct resection with choledochojejunostomy or hepaticojejunostomy, 17 underwent a Whipple operation, and three underwent pyloruspreserving pancreaticoduodenectomy with liver resection. Among the 99 patients, 12 patients underwent a surgical procedure on the extrahepatic portal system, and the types of and indications for the surgical procedures are as follows: superior mesenteric vein (SMV) resection and anastomosis in six patients for either tumor invasion (four patients) or attachment of tumor to the SMV (two patients); SMV dissection from attached tumor tissue with manipulation in one patient; main portal vein (MPV) resection and anastomosis due to tumor invasion in four patients; and MPV dissection from attached tumor tissue with manipulation in one patient. The mean interval between surgery and the last follow-up CT examination after surgery was 41.6 months (range, 6 95 months). Follow-up CT images obtained more than 24 months after surgery were available for 63 of the 99 patients. MDCT Acquisition All CT examinations were performed with a 16- MDCT (n = 38) or a 64-MDCT (n = 61) scanner (Brilliance, Philips Healthcare). Initially, an unenhanced scan was obtained from the liver dome to the end of the pancreas. Subsequently, 2 ml/ kg of nonionic contrast material (iomeprol, Iomeron 400, Bracco) was injected at a flow rate of 3 ml/s through an 18-gauge catheter in the antecubital vein. A bolus-tracking method was used with a region of interest placed on the proximal abdominal aorta. Acquisition of arterial phase images was Fig. 2 Graph shows cumulative incidence of postoperative afferent jejunal loop varices. Cumulative Incidence of Varices (%) 100 80 60 40 20 0 started 15 seconds after enhancement in the proximal abdominal aorta reached 200 HU. Portal venous phase images were obtained 60 seconds after triggering. The scanning parameters were as follows: detector collimation, 16 1.5 mm or 64 0.625 mm; pitch, 1.11 1.25; gantry rotation time, 0.5 second for the 16-MDCT scanner and 0.42 second for the 64-MDCT scanner; and 120 kvp. The effective tube current varied between 120 and 260 mas with an automatic tube current modulation technique. Two transverse image datasets were reconstructed with different section thicknesses: 4 mm at 3-mm increments (thick sections) and 2 mm at 1-mm increments (16-MDCT scanners) or 0.67 mm at 0.33-mm increments (64-MDCT scanners) (thin sections). The thin sections were used to reconstruct coronal images with a section thickness of 4 mm and an increment of 3 mm. Imaging Analysis Two board-certificated abdominal radiologists with 11 and 6 years of abdominal CT experience independently reviewed the CT images. The radiologists were aware that the patients had undergone bilioenteric anastomosis for various malignant diseases involving the pancreaticobiliary system but were blinded to clinical information, including surgical and pathology records, and any previous imaging findings. All images were reviewed in stack mode at a PACS workstation. The reviewers determined the presence of postoperative afferent jejunal loop varices, which appeared on the CT images as abnormally engorged tortuous veins in the afferent jejunal loop of the bilioenteric anastomosis. They also tried to measure the diameter of afferent jejunal loop varices when they were present. After interpretation of CT images, the two radiologists compared their findings. In cases of disagreement, the CT images were reviewed by a third experienced abdominal radiologist, and the decision regarding imaging findings 28.0% 0 20 40 60 80 100 Follow-Up Time (mo) 1262 AJR:200, June 2013

CT Findings of Afferent Loop Varices Fig. 3 43-year-old woman with afferent jejunal loop varices. A, Axial portal venous phase CT images obtained 3 months after Whipple operation for cancer of pancreatic head show patent portal vein and bilioenteric anastomosis site without abnormalities. B, Axial portal venous phase CT images obtained 48 months after surgery show abnormally engorged and dilated venous structures (black arrows) in jejunal loop of choledochojejunostomy site. Extrahepatic main portal vein is occluded by recurrent tumor mass (white arrow). C, Axial portal venous phase CT images obtained 51 months after surgery shows more prominently engorged and dilated venous structures (black arrows) in jejunal loop of choledochojejunostomy site. Extrahepatic main portal vein is occluded by recurrent tumor mass (white arrows). D, Coronal reformation images show afferent jejunal loop varices (arrows). A B C D was made by consensus of the three abdominal radiologists. The results of consensus were used for statistical analysis. The radiologists evaluated the presence and level of venous occlusion or stenosis in the extrahepatic portal system, including the MPV, SMV, and splenic vein (SV). Occlusion in the extrahepatic portal system was defined as total luminal occlusion in the extrahepatic portal system, whereas stenosis was considered present when more than 50% luminal narrowing compared with the proximal portion was seen in extrahepatic portal venous tributaries. The presence of portosystemic collateral vessels in regions other than the afferent loop of the bilioenteric anastomosis, such as gastroesophageal varices, ascites, splenomegaly (greater than 13 cm in the long axis), and local tumor recurrence, was also evaluated on follow-up CT images. Local tumor recurrence was diagnosed as histologic confirmation in a biopsy specimen or the appearance of either new nodular or infiltrative soft-tissue lesions with an interval increase in size around the bilioenteric or pancreaticoenteric anastomotic site on follow-up CT images. In the current study, 45 of 99 patients had local tumor recurrence. The diagnosis of local tumor recurrence was made by percutaneous ultrasound-guided biopsy with histologic confirmation in four patients and with clinical and radiologic findings in 41 patients. Evaluation of Medical Record and Operative Findings An author not involved in image interpretation reviewed the medical and surgical records of the 99 patients to determine whether pancreaticoduodenectomy, surgical procedures on the extrahepatic portal system (such as resection and anastomosis), and postoperative radiation therapy (including the operative field) had been performed. The presence of a gastrointestinal bleeding history during follow-up was also identified. If the patient had a history of gastrointestinal bleeding, the possible causes and hospital course were reviewed. Patient survival status was also searched and recorded. Statistical Analysis Patient demographic data, including age and sex, presence of occlusion or stenosis in the extrahepatic portal system, presence of other portosystemic collateral vessels, splenomegaly, ascites, and local tumor recurrence, were compared between the patients with and those without afferent loop varices. The presence of pancreaticoduodenectomy, surgical procedure for portal venous system, and postoperative radiation therapy was also compared. Univariate analysis was performed by Fisher exact test and independent samples Student t test. Variables with p 0.25 in univariate analysis were chosen for AJR:200, June 2013 1263

Lee et al. A E C Fig. 4 56-year-old woman with afferent jejunal loop varices with obstruction in extrahepatic portal system associated with intestinal bleeding treated with portal vein stent. A, Axial portal venous phase CT images show abnormally engorged and dilated venous structures (black arrows) in jejunal loop of choledochojejunostomy site. Esophageal varices (white arrow) are other collateral pathways. B, Coronal reformation images show afferent jejunal loop varices (black arrows) along choledochojejunostomy site. Proximal superior mesenteric vein obstruction (white arrows) is evident, and main portal vein beyond obstruction site is not. C, Direct portograms through segment V portal vein branch show obstruction in superior mesenteric and main portal veins. Abnormally engorged and dilated veins (arrow) consistent with afferent loop varices are evident. D, Technetium-99m labeled RBC bleeding scan shows accumulation of radiotracer (arrow) at right upper quadrant of abdomen, corresponding to bleeding from afferent loop varices of choledochojejunostomy. E, Portogram obtained after portal vein stent insertion shows flow from superior mesenteric vein to main portal vein. Abnormally engorged and dilated veins consistent with afferent loop varices have disappeared. After portal vein stenting, patient recovered without another bleeding episode. D B 1264 AJR:200, June 2013

CT Findings of Afferent Loop Varices Cox proportional hazard regression analysis as multivariate analysis. Relative risk and 95% CIs were obtained. Actuarial analysis of the cumulative incidence of development of postoperative afferent jejunal loop varices was also performed with the Kaplan-Meier method, and the log-rank test was used to compare the cumulative incidences between certain groups as necessary. Interobserver agreement between two readers was evaluated for assessment of afferent loop varices, obstruction or stenosis in the extrahepatic portal system, presence of other collaterals, recurrent tumor, and ascites by use of weighted kappa statistics. We considered a kappa value greater than 0.81 to represent excellent agreement and values of 0.61 0.80 and 0.41 0.60 to represent good and moderate agreement. A value less than 0.40 was considered to represent fair agreement. Data were analyzed with statistical software (Medicalc, MedCalc Software). A value of p < 0.05 was considered statistically significant. Results Prevalence and Cumulative Incidence of Afferent Jejunal Loop Varices Among the 99 patients, 22 patients (22.2%) were found to have afferent jejunal loop varices on follow-up CT images (mean followup period, 15.5 ± 12.4 months; range, 2 48 months). Two reviewers had the same results regarding presence of afferent loop varices in all cases. The overall 12-, 24-, 36-, and 48-month cumulative incidences of afferent jejunal loop varices were 11.5%, 20.5%, 23.7%, and 28.0% (Fig. 2). The varices identified as abnormally engorged tortuous veins were found along the afferent jejunal loop of a choledochojejunostomy (n = 20) or hepaticojejunostomy (n = 2) in all 22 patients (Figs. 3 and 4). The diameter of afferent jejunal loop varices ranged from 6 to 11 mm (mean, 7.82 ± 1.73 mm). Evaluation of CT Findings for Development of Afferent Jejunal Loop Varices Table 1 summarizes the patient demographics and CT findings for the groups with and without afferent jejunal loop varices. In terms of CT findings, good to excellent agreement was achieved between two reviewers (ascites, κ = 0.730; other collaterals, κ = 0.865; obstruction in extrahepatic portal system, κ = 0.923; stenosis in extrahepatic portal system, κ = 0.729; recurrent tumor, κ = 0.798). No significant differences between patients with and without afferent loop varices in age (mean age, 65.4 vs 64.5 years; p = 0.78) or sex (male-to-female ratio, 13:9 vs 46:31; p = 1.0) were found. TABLE 1: Patient Demographics and CT Findings Variable Afferent Loop Varices (n = 22) No Afferent Loop Varices (n = 77) Age (y), mean ± SD 65.3 ± 12.1 64.5 ± 8.6 0.78 Sex Men 13 (59.1) 46 (59.7) 1.0 Women 9 (40.9) 31 (40.3) Extrahepatic portal vein obstruction Present 19 (86.4) 3 (3.9) < 0.001 Absent 3 (13.6) 74 (96.1) Extrahepatic portal vein stenosis Present 2 (9.1) 19 (24.7) 0.15 Absent 20 (90.9) 58 (75.3) Other collaterals b Present 16 (72.7) 7 (9.1) < 0.001 Absent 6 (27.3) 70 (90.9) Splenomegaly Present 1 (4.5) 2 (2.6) 0.53 Absent 21 (95.5) 75 (97.4) Ascites Present 14 (63.6) 9 (11.7) < 0.001 Absent 8 (36.4) 68 (88.3) Recurrent tumor Present 18 (81.8) 27 (35.1) < 0.001 Absent 4 (18.2) 50 (64.9) Method of surgery Pancreatoduodenectomy 18 (81.8) 59 (76.6) 0.77 Bile duct resection 4 (18.2) 18 (23.4) Surgical procedure on portal vein Present 4 (18.2) 8 (10.4) 0.46 Absent 18 (81.8) 69 (89.6) Radiation therapy Present 5 (22.7) 18 (23.4) 1.00 Absent 17 (77.3) 59 (76.6) Note Data in parentheses are percentages. a Fisher exact test and independent samples Student t test. b Portosystemic collateral vessels in other than the jejunal loop of a bilioenteric anastomosis. A total of 22 (22.2%) patients had obstruction in the extrahepatic portal system on follow-up CT images. The portal venous obstruction was found along the MPV, SMV, and SV (n = 11) and in the MPV alone (n = 5), SMV alone (n = 1), MPV and SMV (n = 2), SV alone (n = 1), MPV and SV (n = 1), and SV and SMV (n = 1). The causes of obstruction in the extrahepatic portal system were as follows: encasement by recurrent tumor mass (n = 19) and postoperative adhesion-related fibrotic change (n = 3). Among the 22 patients with occlusion in the extrahepatic portal system, afferent jejunal loop varices were identified in 19 (Figs. 3 and 4). A significant difference was found for the presence of portal venous occlusion between patients with and without afferent loop varices (p < 0.001). Luminal stenosis of the portal venous system was found in 21 of 99 (21.2%) patients. Luminal stenoses were seen in the MPV (n = 10), SMV (n = 8), and SV (n = 3). The causes of stenosis in the extrahepatic portal system were recurrent tumor mass in 13 cases p a AJR:200, June 2013 1265

Lee et al. TABLE 2: Associated Findings for the Development of Afferent Jejunal Loop Varices Cumulative Incidence of Varices (%) 100 80 60 40 20 0 Variable p a Relative Risk b Extrahepatic portal vein obstruction < 0.001 71.4 (7.4 1000) Other collaterals c 0.615 0.7 (0.2 2.6) Ascites 0.645 0.7 (0.2 2.9) Local tumor recurrence 0.075 4.0 (0.9 18.5) Extrahepatic portal vein stenosis 0.088 8.7 (0.7 100) a Cox proportional hazard regression analysis. b Data in parentheses are 95% CI of relative risk. c Portosystemic collateral vessels in other than the jejunal loop of bilioenteric anastomosis. 0 20 40 60 80 100 Follow-Up Time (mo) 94.5% With PV obstruction 5.0% Without PV obstruction Fig. 5 Graph shows cumulative incidence of postoperative afferent jejunal loop varices according to patient group, that is, with or without obstruction in extrahepatic portal system. PV = portal vein. and postoperative adhesion-related fibrotic change in eight cases. No significant difference was found between patients with and without afferent loop varices with respect to presence of luminal stenosis of the portal system (p = 0.15). On follow-up CT images, portosystemic collateral vessels in regions other than the afferent jejunal loop of the bilioenteric anastomosis were found in 23 of 99 (23.2%) patients who had obstruction (n = 15) or stenosis (n = 8) in the extrahepatic portal system. These other portosystemic collaterals were found in the distal esophagus (n = 11), gastric fundus (n = 9), sigmoid mesocolon (n = 1), ascending colon (n = 1), and mesentery (n = 1). Among these 23 patients with other portosystemic collaterals, 16 (69.6%) patients had afferent jejunal loop varices (Fig. 4). A significant difference was observed for the presence of other portosystemic collaterals between patients with and without afferent loop varices (p < 0.001). The presence of ascites (14 of 22 [63.4%] vs 9 of 77 [11.7%] patients) and local tumor recurrence (18 of 22 [81.8%] vs 27 of 77 [35.1%]) were also significantly different between patients with and without afferent loop varices (p < 0.001) (Fig. 3). The presence of splenomegaly did not exhibit significant difference between the two groups (1/22 [4.5%] vs 2/77 [2.6%]; p = 0.53). History of surgical procedure on the portal system, such as resection and anastomosis, presence of pancreaticoduodenectomy, and postoperative radiation therapy were not significantly different between patients with and without afferent loop varices (p = 0.46, p = 0.77, p = 1.00). Multivariate analysis with a Cox proportional hazard regression analysis was performed for the presence of occlusion in the extrahepatic portal vein, other portosystemic collaterals, ascites, local tumor recurrence, and stenosis in the extrahepatic portal vein. Occlusion in the extrahepatic portal system was the only statistically significant finding in patients with afferent loop varices (p < 0.001). In patients with extrahepatic portal vein occlusion, the relative risk of developing afferent loop varices was 71.4 times that of patients without portal venous occlusion (Table 2). The cumulative incidence of afferent jejunal loop varices was also obtained in patient groups according to the presence of occlusion in the extrahepatic portal system. The overall 12-, 24-, 36-, and 48-month cumulative incidences of afferent jejunal loop varices in patients without extrahepatic portal venous occlusion were 1.4%, 3.0%, 5.0%, and 5.0% (Fig. 5). In contrast, the overall 12-, 24-, 36-, and 48-month cumulative incidences of afferent jejunal loop varices in patients with extrahepatic portal venous occlusion were 47.8%, 79.1%, 84.3%, and 94.5% (Fig. 5). This difference was also statistically significant (p < 0.001). History of Gastrointestinal Bleeding and Patient Survival Among the 22 patients with afferent jejunal loop varices, three patients (13.6%) eventually had upper gastrointestinal bleeding (Fig. 4). The characteristics of these three patients are summarized in Table 3. During the follow-up period, six of 22 (27.3%) patients with and 18 of 77 (23.4%) patients without afferent loop varices died. This difference was not statistically significant. Discussion In this study, afferent jejunal loop varices developed after bilioenteric anastomosis for malignancy in 22 of 99 (22.2%) patients, and the 48-month cumulative incidence was 28.0%. The prevalence in this study was comparable to that documented in a previous study [6], in which the prevalence was 16.1% (5/31). This study showed that obstruction in the extrahepatic portal system was found more frequently in patients with afferent loop varices than in those without afferent loop varices. Multivariate analysis with Cox proportional hazard regression analysis gave a relative risk of 71.4. Most of the patients with afferent loop varices (86.4%, 19/22) also had obstruction in the extrahepatic portal system, and the 48-month cumulative incidence of afferent loop varices in patients with extrahepatic portal venous occlusion was 94.5%. These results showed the importance of extrahepatic portal venous occlusion in development of afferent loop varices. Our results also suggest that afferent loop varices can form as a collateral route of portal flow in patients with portal vein obstruction along the bilioenteric anastomotic site, where collaterals easily develop [6]. When hepatopetal flow from the mesentery is interrupted as a result of obstruction of the extrahepatic portal system, communicating veins at the bilioenteric anastomotic site become dilated and build collateral pathways to maintain the 1266 AJR:200, June 2013

CT Findings of Afferent Loop Varices TABLE 3: Characteristics of Patients With Bleeding From Afferent Loop Varices Diagnostic Method for Bleeding Treatment Result Local Tumor Recurrence Presenting Symptom Location of Bleeding Indication for Surgery Patient No. Sex Age (y) Type of Surgery Yes Recurrent melena Choledochojejunostomy site Endoscopy a Conservative Improvement 1 F 56 PPPD Ampullary cancer Yes Recurrent melena Hepaticojejunostomy site RBC scan Conservative Death b 2 M 31 Whipple operation Common bile duct cancer No Recurrent melena Choledochojejunostomy site RBC scan Portal vein stent Improved Common bile duct cancer 3 F 56 Pylorus-preserving pancreaticoduodenectomy a Upper gastrointestinal endoscopy revealed hyperemic nodular lesion with blood oozing at afferent jejunal loop. b Uncontrolled recurrent bleeding was considered to be cause of death. hepatopetal flow [6]. In addition, stenosis in the extrahepatic portal system was found in the three patients with afferent loop varices who did not have obstruction in the extrahepatic portal system. In these three patients, stenosis in the extrahepatic portal system was a possible cause of afferent loop varices. This study showed a significant difference in local tumor recurrence between patients with and those without afferent loop varices. In 19 of 22 patients with extrahepatic portal vein obstruction, a recurrent tumor was found around the operation site. Although the difference was not statistically significant in multivariate analysis, the finding implicated local tumor recurrence as a cause of obstruction of the extrahepatic portal system. Therefore, when the local recurrent tumor is found on follow-up CT images after bilioenteric anastomosis in patients with malignancy, the portal system and bilioenteric anastomotic site should be carefully evaluated for the presence of obstruction and development of afferent jejunal loop varices. Three of 22 patients (13.6%) in this study had recurrent bleeding from afferent loop varices, and one of these three patients died of uncontrolled recurrent bleeding. The diagnosis of bleeding from afferent jejunal loop varices can be easily overlooked clinically because esophageal varices are frequently regarded as a cause of bleeding in this clinical situation [10]. Because afferent loop varices can cause lethal recurrent or massive bleeding, as in one of our cases, their presence on follow-up CT images should raise concern. Once afferent loop varices are noted on CT images of patients with gastrointestinal bleeding, these should be regarded as a potential cause of gastrointestinal bleeding. This suspicion can be helpful in avoiding unnecessary invasive procedures, including celiac and superior mesenteric arteriography, and choosing adequate therapeutic options. In the care of patients with afferent loop variceal bleeding, medical and surgical options such as transfusion, administration of β-blockers, revision of bilioenteric anastomosis, and surgical or interventional formation of a portosystemic shunt, can be considered [3, 4]. Placement of a portal vein stent with reconstitution of portal flow is also an option for treatment of bleeding from afferent loop varices, especially when obstruction of the extrahepatic portal system is associated with afferent loop varices [10]. In this study, one patient with bleeding from afferent loop varices and extrahepatic portal system obstruction was successfully treated with portal vein stent placement. This study had limitations. First, the design was retrospective; therefore, some degree of selection bias could not be avoided. Second, all patients had undergone bilioenteric anastomosis for malignant tumors. Therefore, application of the study results may be limited to cases of malignant disease. Third, angiography of the portal system was not performed for all patients. Therefore, the flow direction and exact extent of obstruction in the portal system and collateralization were not assessed. However, routine use of angiography cannot be justified unless an interventional procedure has been requested. Fourth, afferent jejunal loop varices were detected and diagnosed only on the basis of findings seen at CT. Thus patients with minor varices might have been excluded from this study. Fifth, the follow-up range in this study (6 95 months) was large, and this might have caused bias in estimating cumulative incidences. Conclusion Afferent loop varices after bilioenteric anastomosis for malignancy had a prevalence and 48-month cumulative incidence of 22.2% and 28.0% and can result in gastrointestinal bleeding. The presence of obstruction in the extrahepatic portal system was significantly associated with afferent loop varices. Thus, careful evaluation for the presence of afferent loop varices should be conducted in the care of patients with extrahepatic portal venous occlusion after bilioenteric anastomosis. References 1. Puppala S, Patel J, McPherson S, Nicholson A, Kessel D. Hemorrhagic complications after Whipple surgery: imaging and radiologic intervention. AJR 2011; 196:192 197 2. Jagad RB, Koshariya M, Kawamoto J, Chude GS, Neeraj RV, Lygidakis NJ. Postoperative hemorrhage after major pancreatobiliary surgery: an update. Hepatogastroenterology 2008; 55:729 737 3. Helmy A, Al Kahtani K, Al Fadda M. Updates in the pathogenesis, diagnosis and management of ectopic varices. Hepatol Int 2008; 2:322 334 4. Norton ID, Andrews JC, Kamath PS. Management of ectopic varices. Hepatology 1998; 28:1154 1158 5. Seeger M, Gunther R, Hinrichsen H, et al. Chronic portal vein thrombosis: transcapsular hepatic collateral vessels and communicating ectopic varices. Radiology 2010; 257:568 578 6. Hashimoto M, Heianna J, Yasuda K, et al. Portal AJR:200, June 2013 1267

Lee et al. flow into the liver through veins at the site of biliary-enteric anastomosis. Eur Radiol 2005; 15:1421 1425 7. Taniguchi H, Moriguchi M, Amaike H, Fuji N, Murayama Y, Kosuga T. Hemorrhage from varices in hepaticojejunostomy in the fifth and tenth year after surgery for hepatic hilar bile duct cancer: a case report. Cases J 2008; 1:59 8. Ota S, Suzuki S, Mitsuoka H, et al. Effect of a portal venous stent for gastrointestinal hemorrhage from jejunal varices caused by portal hypertension after pancreatoduodenectomy. J Hepatobiliary Pancreat Surg 2005; 12:88 92 9. Tang SJ, Jutabha R, Jensen DM. Push enteroscopy for recurrent gastrointestinal hemorrhage due to jejunal anastomotic varices: a case report and review of the literature. Endoscopy 2002; 34:735 737 10. Stafford Johnson DB, Narasimhan D. Case report: successful treatment of bleeding jejunal varices using mesoportal recanalization and stent placement report of a case and review of the literature. Clin Radiol 1997; 52:562 565 11. Moncure AC, Waltman AC, Vandersalm TJ, Linton RR, Levine FH, Abbott WM. Gastrointestinal hemorrhage from adhesion-related mesenteric varices. Ann Surg 1976; 183:24 29 12. Paquet KJ, Lazar A, Bickhart J. Massive and recurrent gastrointestinal hemorrhage due to jejunal varices in an afferent loop: diagnosis and management. Hepatogastroenterology 1994; 41: 276 277 13. Smith AM, Walsh RM, Henderson JM. Novel bile duct repair for bleeding biliary anastomotic varices: case report and literature review. J Gastrointest Surg 2005; 9:832 836 FOR YOUR INFORMATION The comprehensive book based on the ARRS 2013 annual meeting categorical course on Body MRI is now available! For more information or to purchase a copy, see www.arrs.org. 1268 AJR:200, June 2013