CT Signs of Hepatofugal Portal Venous Flow in Patients with Cirrhosis

Transcription

1 Thomas J. Bryce Benjamin M. Yeh Aliya Qayyum Preeyacha Pacharn 2 Nathan M. Bass 3 Ying Lu Fergus V. Coakley Received April, 23; accepted after revision June 26, 23. Department of Medicine and Department of Radiology, Abdominal Imaging Section, University of California at San Francisco, Box 628, 55 Parnassus Ave., San Francisco, CA Address correspondence to F. V. Coakley (fergus.coakley@radiology.ucsf.edu). 2 Department of Radiology, Mahidol University, 2 Prannok Rd., Bangkok, 7 Thailand. 3 Department of Medicine, Division of Gastroenterology, University of California at San Francisco, Box 538, 55 Parnassus Ave., San Francisco, CA AJR 23;8: X/3/ American Roentgen Ray Society CT Signs of Hepatofugal Portal Venous Flow in Patients with Cirrhosis OBJECTIVE. We investigated whether CT signs can be used to predict hepatofugal flow in the main portal vein in patients with cirrhosis. MATERIALS AND METHODS. We retrospectively identified 36 patients with cirrhosis, 8 with hepatopetal and 8 with hepatofugal flow in the main portal vein, who underwent contemporaneous abdominal sonography and CT. Two independent observers evaluated the following features on the randomized CT studies: diameter of the portal, splenic, and superior mesenteric veins; spleen size; and the presence of ascites, varices, or arterial phase portal venous enhancement. These data were correlated with the flow direction seen on sonography. RESULTS. A small main portal vein was the only sign significantly (p.5) predictive of hepatofugal flow by univariate and multivariate analyses. Observers and 2 recorded a portal vein diameter of less than cm in eight (44%) and seven (39%) of the 8 patients with hepatofugal flow compared with one (6%) and none of the 8 patients with hepatopetal flow, respectively (p <.2). Receiver operating characteristic analysis using the size of the portal vein to predict flow direction revealed an area under the curve of.83 for observer and.74 for observer 2. CONCLUSION. A diameter of less than cm for the main portal vein is highly specific, although not sensitive, for hepatofugal portal venous flow in patients with cirrhosis. This sign may be useful when sonography is limited, or this sign may prompt sonographic assessment in patients not known to have hepatofugal flow. H epatofugal portal venous flow in patients with cirrhosis indicates advanced portal hypertension [], and cirrhotic patients with hepatofugal flow have greater hepatic dysfunction [2], a higher incidence of hepatic encephalopathy [3], an increased risk of variceal bleeding [2, 4, 5], poorer response of varices to endoscopic ligation [6], and higher mortality [2, 7], compared with cirrhotic patients with hepatopetal flow. The detection of hepatofugal flow in the main portal vein implies the liver is perfused solely by the hepatic artery, which is of therapeutic importance in planning chemoembolization of a hepatocellular carcinoma or placement of a transjugular intrahepatic portosystemic shunt (TIPS). CT arterioportography and portal venous angiography may be ineffective and therefore contraindicated in the presence of hepatofugal portal venous flow [8]. Doppler sonography is the principal technique used to determine the direction of flow in the portal vein [9, ], although MRI [ 3] and conventional angiography [4, 5] can also be used. To our knowledge, the role of CT in determining the direction of portal venous flow has not been systematically examined, although patients with cirrhosis frequently undergo CT, particularly when hepatocellular carcinoma is a concern. The establishment of CT criteria for the detection of hepatofugal flow in the main portal vein could be helpful in the evaluation of such patients. Therefore, we undertook this study to determine whether CT signs can be used to predict hepatofugal portal venous flow in patients with cirrhosis. Materials and Methods Patients This study was a retrospective single-institution study approved by our institutional review board. Informed consent was not required. We searched our radiology information system (IDXrad [software version 9.7.], IDX Systems, Burlington, VT) AJR:8, December

2 Bryce et al. and medical records for the period of October 998 to March 22 to identify patients meeting the following sequential criteria: abdominal sonography report containing the term cirrhosis ; contemporaneous (within 5 months) abdominal CT; direction of flow in the main portal vein described in the sonography report; absence of TIPS or thrombosis of the main portal vein; and histologic diagnosis of cirrhosis (n = ) or a clear clinical diagnosis of cirrhosis documented in the medical record (n = 25). We identified 8 patients with hepatofugal flow in the main portal vein who met these criteria. We then randomly selected an additional 8 patients with hepatopetal flow who also met these criteria to form the final study population of 36 patients. None of the patients was assigned a Child-Pugh class because not all required data were available retrospectively. The mean age of patients in the hepatopetal group was 59 years (range, 44 8 years) versus 5 years (range, 7 73 years) in the hepatofugal group. The hepatopetal group comprised men and eight women, and the hepatofugal group comprised seven men and women. In the hepatopetal group, the diagnosis of cirrhosis was established by histology in six patients and on clinical grounds in 2; in the hepatofugal group, cirrhosis was established by histology in five patients and on clinical grounds in 3. Of those patients without histologic confirmation, cirrhosis was diagnosed by a gastroenterologist in all but two, both of whom had hepatofugal portal venous flow. In these two patients, cirrhosis was diagnosed on the basis of clinical evidence including a history of severe alcohol abuse; gross findings of cirrhosis on imaging studies; and large varices, ascites, hypoalbuminemia, and coagulopathy. The causes for cirrhosis in the group with hepatopetal flow and in the group with hepatofugal flow, respectively, included the following: chronic viral hepatitis, and four patients; viral hepatitis and ethanol abuse, two patients and one patient; ethanol abuse alone, and six patients; autoimmune hepatitis, one and three patients; nonalcoholic steatotic hepatitis, one patient and patients; primary sclerosing cholangitis, one and three patients; and cystic fibrosis, and one patient. Cirrhosis was cryptogenic in two patients with hepatopetal flow. The median time between sonography and CT was 23 days (range, 48 days) in the hepatopetal group and 3 days (range, 97 days) in the hepatofugal group. If sonography had been performed more than 48 hr before or after the CT examination (hepatopetal group, n = 4; hepatofugal group, n = 9), we identified a second Doppler sonographic study of flow direction in the main portal vein for each patient so that sonography occurred before and after the CT examination (hepatopetal group, n = 9; hepatofugal group, n = 5). Alternatively, if CT had not been performed between the sonographic examinations, we identified the two sonographic examinations closest in time to the CT examination (hepatopetal group, n = 2; hepatofugal group, n = 2). In these patients, the time lag between CT and sonography for the two patients with hepatopetal flow was 3 and 3 days, and for the two patients with hepatofugal flow, the lag was 37 and 62 days. Additional studies were not available for the remaining three hepatopetal and two hepatofugal patients. All additional sonographic studies evaluated confirmed the flow direction seen on the initial study. Imaging Technique All CT examinations were performed on multidetector scanners (LightSpeed or HiSpeed, General Electric Medical Systems, Milwaukee, WI). Thirty-four of the 36 patients received 5 ml of IV iohexol (Omnipaque 35, Nycomed Amersham, Princeton, NJ), and images were acquired in the portal venous phase of enhancement (7-sec scan delay with 5-mm slice collimation). Images were also acquired in the arterial phase of enhancement (45-sec scan delay with 2.5-mm slice collimation) in 2 of these patients. Two patients (one with hepatopetal and one with hepatofugal flow) did not receive IV contrast material. All patients received oral diatrizoate meglumine (Hypaque, Nycomed Amersham). All images were contiguous. Color Doppler sonography of the direction of flow in the main portal vein was performed using a scanner (Sequoia 52, Acuson Solutions, Mountain View, CA) with a.75- to 4-MHz sector transducer (4V, Acuson Solutions) or a 2.5- to 4-MHz sector transducer (4V2, Acuson Solutions). CT Interpretation Two radiologists independently reviewed the randomized CT images of all 36 patients on a PACS (picture archiving and communication system) workstation (Impax DS 3 [release 4.], Agfa, Mortsel, Belgium). Observers were unaware of clinical and sonographic findings. Both observers recorded the following CT signs: short-axis diameter of the main portal vein, which was measured midway between the splenoportal confluence and the portal vein bifurcation in the porta hepatis; maximum short-axis diameter of the superior mesenteric vein, measured on the first image that was clearly inferior to the splenoportal confluence; diameter of the splenic vein, measured adjacent to the midportion of the pancreatic tail; presence or absence of ascites; presence or absence of varices; and spleen size, recorded as the maximum axial diameter. In addition, for multiphase CT examinations (n = 2), the presence or absence of early (i.e., arterial phase) enhancement of the main portal vein was recorded. Data Analysis Statistical analysis was performed using statistical analysis software (SAS version 8., SAS, Cary, NC). Continuous data (vessel diameters and spleen size) were examined by univariate analysis using the two-tailed Cochran t test. Noncontinuous data (all other categories) were examined using Fisher s exact test. The univariate analyses were performed separately for the data for each observer. All p values of.5 or less were considered significant. Interobserver agreement for categorical data was measured with kappa statistics [6]. Interobserver agreement for continuous data was assessed using Bland-Altman regression [7] with Bradley-Blackwood p values [8]. Multivariate analysis was performed using logistic regression with a generalized linear model to account for the presence of two observers. Parameters were selected in a stepwise fashion with a type 3 significance level of.5 or less required for factors to remain in the model. Receiver operating characteristic (ROC) analysis and area under the ROC curve (A z ) calculations were performed to evaluate predictive models. Results The CT signs recorded by each observer showed high interobserver consistency for all measurements other than the presence of varices and the size of the spleen. A small main portal vein correlated strongly (p <.) with hepatofugal flow by univariate and multivariate analyses (Fig. ). No other signs achieved significance by either method. ROC analysis using the diameter of the main portal vein to predict hepatofugal flow revealed an A z of.83 for observer and.74 for observer 2 (Fig. 2). Observer found that the main portal vein measured less than cm in eight (44%) of 8 patients with hepatofugal flow, but in only one (6%) of 8 patients with hepatopetal flow (p <.2). Thus, the sensitivity of this sign for predicting hepatofugal flow was 44% and the specificity was 94%. Observer 2 found that the main portal vein measured less than cm in seven (39%) of 8 patients with hepatofugal flow and of 8 patients with hepatopetal flow (p <.), for a sensitivity of 39% and a specificity of %. The -cm measurement was the largest (most sensitive) threshold that could be used to predict hepatofugal flow without significantly reducing specificity, given that both observers measured the main portal vein at or slightly above cm in several patients with hepatopetal flow (Fig. ). A lower threshold would reduce sensitivity without significantly increasing specificity, and a higher threshold would greatly reduce specificity with only mildly increasing sensitivity (Fig. 2). A representative case illustrates the finding of a small portal vein in a patient with hepatofugal flow (Fig. 3). Arterial phase enhancement of the portal vein was seen in only one patient; in that patient, flow in the portal vein was hepatopetal on sonography day before CT (Fig. 4). 63 AJR:8, December 23

3 CT Signs of Hepatofugal Portal Venous Flow.8 2 True-Positive Fraction Diameter of Main Portal Vein (cm) Hepatopetal Hepatofugal Observer Hepatopetal Hepatofugal Observer False-Positive Fraction Fig.. Diagram shows diameters of main portal vein in patients with hepatopetal flow versus hepatofugal flow. Note that main portal vein diameter of less than cm is highly specific for hepatofugal flow. Fig. 2. Diagram shows receiver operating characteristic (ROC) values achieved using small main portal vein diameter to predict hepatofugal flow in main portal vein in patients with cirrhosis. Operating points achieved by predicting hepatofugal flow when main portal vein diameter was less than cm are shown for observer ( ) and observer 2 ( ). Note that at appropriate operating points, this sign is highly specific for hepatofugal flow. Area under ROC curve (Az) was.83 for observer and.74 for observer 2. A B Fig year-old woman with cirrhosis due to alcohol abuse. A, Axial CT scan obtained during portal venous phase shows small (8 mm in diameter) main portal vein (arrow); this finding strongly correlates with hepatofugal flow. B, Sonogram shows hepatofugal flow in main portal vein. AJR:8, December 23 63

4 Bryce et al. Discussion Our finding that a relatively small main portal vein in patients with cirrhosis indicates hepatofugal flow is consistent with the physiology of hepatofugal flow in the cirrhotic liver. The average size of the main portal vein in healthy adults has been found to be. cm [9]; a main portal vein of less than cm can be considered relatively small in a cirrhotic patient because portal hypertension tends to increase portal vein size [2]. Normally, the portal vein receives flow from the superior mesenteric vein and the splenic vein. In patients with cirrhosis and hepatofugal flow in the main portal vein, the portal vein is supplied only by the hepatic artery, which also supplies the hepatic veins. This decrease in flow volume could explain the decreased diameter of the portal vein. (A small main portal vein in cirrhotic patients has been associated with spontaneous splenorenal shunts [2], a phenomenon that would also be expected to reduce hepatic portal venous perfusion.) Although specific for hepatofugal flow, this sign was of only moderate sensitivity in our study group. Possible explanations for the limited sensitivity may include a tendency for the vein to remain patulous after enlarging in earlier stages of portal hypertension and the possibility that portal hypertension may provide a continuing impetus for venous enlargement, despite a reduction or reversal in flow. Arterial phase enhancement of the portal vein has been reported as a sign of hepatofugal flow [, 22]. In our study, this sign was present in only one of 2 patients undergoing CT with arterial phase contrast enhancement, and Doppler sonography of this patient day before CT showed hepatopetal flow. The finding of arterial phase enhancement of the portal vein on CT in a patient with hepatopetal portal venous flow on sonography is puzzling but might be explained by transient changes in flow direction in the portal venous system. Such flow changes can occur spontaneously in cirrhotic patients [23], in the postprandial state [24, 25], and during hepatic arteriography [26]. Disturbance of baseline physiology during abdominal CT, such as due to breath-hold technique or bolus administration of IV contrast material, may alter portal venous hemodynamics. For example, contrast administration might elevate right heart and hepatic vein pressures, and the subsequent increased shunting of hepatic arterial blood into the portal venous system may result in temporary hepatofugal flow. Our study has several limitations. The sample size was small and may have restricted the power of the study to detect differences between the hepatopetal and hepatofugal groups. For example, the prevalence of ascites and varices was higher in patients with hepatofugal flow, but these differences did not achieve statistical significance. These factors might exhibit an association with hepatofugal flow in a larger study. CT and sonography were not immediately contemporaneous. The lag between the examinations may have lowered the sensitivity of the study for additional predictors of hepatofugal flow, although the possible effects of the time difference would be expected to apply to both hepatopetal and hepatofugal groups of patients. The lag between studies would not be expected to generate spurious associations and should not detract from the finding that a small portal vein size is associated with hepatofugal portal venous flow. Analysis of additional sonography examinations of the study population provided evidence that the direction of portal venous flow in our subjects was stable. Therefore, it is reasonable to expect that the direction of flow at the time of CT would be accurately predicted by the contemporaneous sonographic examination that was evaluated. A B Fig year-old man with cirrhosis due to chronic hepatitis C infection. A, CT scan obtained during arterial phase shows contrast enhancement of portal vein. This sign is thought to indicate hepatofugal portal venous flow. B, Doppler sonogram obtained day earlier than A, however, reveals hepatopetal flow. 632 AJR:8, December 23

5 CT Signs of Hepatofugal Portal Venous Flow High interobserver consistency was not seen for the measurement of spleen size and assessment for the presence or absence of varices. The discrepancy in identification of varices between observers could be explained by a greater sensitivity of observer 2 for varices in borderline cases, given that all cases identified as positive by observer were also called positive by observer 2, but not vice versa. The interobserver variation regarding spleen size may have resulted from the complex and partly subjective methodology used, which involved identifying the maximal single dimension of the spleen on any axial image. This study did not seek to measure the prevalence of hepatofugal flow in the entire population of cirrhotic patients, although this information would be required to determine predictive values. This is important given that our sign, although highly specific, is of limited sensitivity. The prevalence of hepatofugal flow in unselected cirrhotic patients has been previously estimated at % [2, 27], rising to 9% or more in patients with advanced cirrhosis [2, 28]. Finally, the diagnosis of cirrhosis was confirmed histologically in only about one third of the study subjects. However, most patients with a diagnosis of cirrhosis do not undergo histologic confirmation. Restricting the analysis to only those patients with a histologic diagnosis would have selected for substantially more diseased or symptomatic patients, possibly harming the generalizability of our results. Given that all patients in this study had significant risk factors for cirrhosis, had imaging features consistent with cirrhosis, and had either histologic confirmation of cirrhosis or had been assigned the diagnosis of cirrhosis at a tertiary care center (by a gastroenterologist in all but two cases), we believe that the probability of noncirrhotic patients existing within the study population is low. In conclusion, our data suggest that a main portal vein diameter of less than cm is highly specific for hepatofugal portal venous flow in cirrhotic patients, although the sensitivity of this sign is limited. If validated in prospective studies, this sign may be useful when sonography is limited or may prompt sonographic assessment in patients not known to have hepatofugal flow. References. Wachsberg RH, Bahramipour P, Sofocleous CT, Barone A. Hepatofugal flow in the portal venous system: pathophysiology, imaging findings, and diagnostic pitfalls. RadioGraphics 22;22: Tarantino L, Giorgio A, de Stefano G, et al. Reverse flow in intrahepatic portal vessels and liver function impairment in cirrhosis. Eur J Ultrasound 997;6: Smith-Laing G, Camilo ME, Dick R, Sherlock S. Percutaneous transhepatic portography in the assessment of portal hypertension: clinical correlations and comparison of radiographic techniques. Gastroenterology 98;78: Wachsberg RH, Simmons MZ. Coronary vein diameter and flow direction in patients with portal hypertension: evaluation with duplex sonography and correlation with variceal bleeding. AJR 994;62: Komatsuda T, Ishida H, Konno K, et al. Color Doppler findings of gastrointestinal varices. Abdom Imaging 998;23: Matsumoto A, Hamamoto N, Ohnishi A, et al. Left gastric vein hemodynamics and variceal recurrence in patients undergoing prophylactic endoscopic ligation of high-risk esophageal varices. Gastrointest Endosc 999;5: Finucci G, Bellon S, Merkel C, et al. Evaluation of splanchnic angiography as a prognostic index of survival in patients with cirrhosis. Scand J Gastroenterol 99;26: Oliver JH 3rd, Baron RL, Dodd GD 3rd, Peterson MS, Carr BI. Does advanced cirrhosis with portosystemic shunting affect the value of CT arterial portography in the evaluation of the liver? AJR 995;64: Ralls PW. Color Doppler sonography of the hepatic artery and portal venous system. AJR 99;55: Johansen K, Paun M. Duplex ultrasonography of the portal vein. Surg Clin North Am 99;7:8 9. Burkart DJ, Johnson CD, Morton MJ, Ehman RL. Phase-contrast cine MR angiography in chronic liver disease. Radiology 993;87: Imazu H, Matsui T, Noguchi R, et al. Magnetic resonance angiography for monitoring prophylactic endoscopic treatment of high risk esophageal varices. Endoscopy 2;32: Finn JP, Kane RA, Edelman RR, et al. Imaging of the portal venous system in patients with cirrhosis: MR angiography vs duplex Doppler sonography. AJR 993;6: Herlinger H. Arterioportography. Clin Radiol 978;29: Nelson RC, Lovett KE, Chezmar JL, et al. Comparison of pulsed Doppler sonography and angiography in patients with portal hypertension. AJR 987;49: Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 977;33: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 986;: Bradley EL, Blackwood LG. Comparing paired data: a simultaneous test of means and variances. Am Stat 989;43: Niederau C, Sonnenberg A, Muller JE, Erckenbrecht JF, Scholten T, Fritsch WP. Sonographic measurements of the normal liver, spleen, pancreas, and portal vein. Radiology 983;49: Subramanyam BR, Balthar EJ, Raghavendra BN, Lefleur RS. Sonographic evaluation of patients with portal hypertension. Am J Gastroenterol 983;78: Brancatelli G, Federle MP, Pealer K, Geller DA. Portal venous thrombosis or sclerosis in liver transplantation candidates: preoperative CT findings and correlation with surgical procedure. Radiology 2;22: Park CM, Cha SH, Kim DH, et al. Hepatic arterioportal shunts not directly related to hepatocellular carcinoma: findings on CT during hepatic arteriography, CT arterial portography and dual phase spiral CT. Clin Radiol 2;55: Hoevels J, Lunderquist A, Tylen U. Spontaneous intermittent reversal of blood flow in intrahepatic portal vein branches in cirrhosis of the liver. Cardiovasc Radiol 979;2: de Vries PJ, de Hooge P, Hoekstra JB, van Hattum J. Postprandial reversal of the portal venous flow in a patient with liver cirrhosis. Neth J Med 995;47: Tochio H, Kudo M, Nishiuma S, Okabe Y. Intrahepatic spontaneous retrograde portal flow in patients with cirrhosis of the liver: reversal by food intake. AJR 2;77: Ohnishi K, Saito M, Sato S, et al. Direction of splenic venous flow assessed by pulsed Doppler flowmetry in patients with a large splenorenal shunt: relation to spontaneous hepatic encephalopathy. Gastroenterology 985;89: Gaiani S, Bolondi L, Li Bassi S, Zironi G, Siringo S, Barbara L. Prevalence of spontaneous hepatofugal portal flow in liver cirrhosis: clinical and endoscopic correlation in 228 patients. Gastroenterology 99;: von Herbay A, Frieling T, Haussinger D. Color Doppler sonographic evaluation of spontaneous portosystemic shunts and inversion of portal venous flow in patients with cirrhosis. J Clin Ultrasound 2;28: AJR:8, December