Emerg Radiol DOI 10.1007/s10140-012-1092-6
ORIGINAL ARTICLE
MDCT diagnosis of post-traumatic hepatic arterio-portal fistulas Coung Nguyen & Nitima Saksobahavivat & Kathirkamanathan Shanmuganathan & Scott Steenburg & Fred Moeslein & Stuart E. Mirvis & William Chiu Received: 18 October 2012 / Accepted: 12 November 2012 # Am Soc Emergency Radiol 2012
Abstract The purpose of this study is to evaluate the performance of multidetector computed tomography (MDCT) in diagnosing arterioportal fistulas (APF) in high-grade liver injury. A retrospective analysis of catheter-based hepatic angiograms performed for major penetrating and blunt liver injuries identified 11 patients with APFs. Using the trauma registry, two additional demographically matched groups with and without liver injury were formed. A randomized qualitative consensus review of 33 MDCTs was performed by three trauma radiologists for the following MDCT findings of APF: transient hepatic parenchymal attenuation differences (THPAD), early increased attenuation of a peripheral or central portal vein compared with the main portal vein, and the "double-barrel" or "rail tract" signs. THPAD was the most sensitive finding and also had a high specificity for diagnosing APF. Both the early increased attenuation of a peripheral or central portal vein compared with the main portal vein and the double-barrel or rail tract N. Saksobahavivat Department of Radiology, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, 10400, Bangkok, Thailand K. Shanmuganathan (*) Department of Diagnostic Radiology & Nuclear Medicine, University of Maryland Medical Center, 22 S. Greene Street, Baltimore, MD 21201, USA e-mail:
[email protected] C. Nguyen : S. Steenburg : F. Moeslein : S. E. Mirvis Department of Diagnostic Radiology and Nuclear Medicine, University of Maryland, Baltimore, USA W. Chiu Division of Critical Care, Program in Trauma, R. Adams Cowley Shock Trauma Center, University of Maryland Medical Center, 22 S. Greene Street, Baltimore, MD 21201, USA
signs had a100% specificity and a sensitivity of 64% and 36%, respectively. Measurement of differences in attenuation values between the APF and the contralateral central portal vein was most sensitive and specific in diagnosing APF. Traumatic APF of the liver can be optimally diagnosed with arterial phase imaging of solid organ using MDCT. Keywords Trauma . Liver injury . Arterioportal fistula . Mutidetector CT
Background Recent advances in multi-detector computed (MDCT) technology have made the imaging assessment of the multisystem trauma patient from the vertex of the skull to the symphysis pubis technically feasible [1–5]. Faster acquisition times, improved spatial resolution, rapid reconstruction of images, and reduction in the total amount of radiation dose delivered to the patient compared with a typical segmental acquisition protocol have made integration of the whole-body MDCT (WBMDCT) technique the most popular routine admission examination in major trauma centers [4, 6, 7]. Advances in MDCT technology have increased the number of detector arrays which has allowed wider coverage and rapid scanning. This enables scanners with 40-slice or more to perform imaging of the abdominal solid organs during the late arterial contrast phase using the WBMDCT protocol. It is easier to demonstrate post-traumatic vascular injuries including pseudoaneurysms and arterio-venous fistulas (AVF) in the solid organs during the arterial phase compared with the portal venous phase using the WBMDCT protocol [8–10]. Early arterial phase imaging is optimally suited to demonstrate early filling of peripheral portal vein branch before the main and proximal portal branches opacify from the splanchnic inflow demonstrating a hepatic arterio-portal
Emerg Radiol
fistula (APF) [11–13]. While small peripheral liver APFs usually resolve spontaneously, larger central liver APF can result in portal hypertension and high-output cardiomyopathy [14]. In some cases, APF can cause gastrointestinal bleeding, including esophageal varices and hemobilia [15]. Therefore, early diagnosis and treatment are essential. Endovascular embolization has become the treatment of choice for intrahepatic arteriovenous fistulas, while surgery remains the preferred approach for extrahepatic arteriovenous fistulas. No prior study has evaluated the performance of MDCT in diagnosing APF resulting from blunt or penetrating trauma. The purpose of this retrospective study is to investigate the utility of late arterial phase MDCT in the diagnosis of traumatic APF.
Materials and methods This retrospective HIPAA-compliant study was approved by the institutional review board. Patient groups The study group consisted of patients diagnosed with hepatic APF during a 13-month period from 04 Dec2009 to 30 June 2011 as shown in Fig. 1. A search of the trauma registry at the R. Adams Cowley Trauma Center and Radiology Information System identified all patients admitted with liver injuries who underwent both MDCT and catheterbased hepatic angiograms (CBHA) during the study period. Review of all 38 CBHAs and MDCT studies and interpretation found 11 patients with APF (Table 1). In the original interpretations, APFs were correctly diagnosed by MDCT in two patients and by hepatic angiography in four patients. Retrospective review of the hepatic angiograms by a faculty angiographer (FM) revealed seven additional patients. Liver injuries were graded based on the American Association for the Surgery of Trauma–Organ Injury Scale [16].
Fig. 1 Acquisition of study patients
Table 1 Demographic and MDCT parameters of the three patient groups
Mechanism of injury Blunt injury Penetrating injury Grade of injury III IV Sex, female/male Age, mean (range) Site of injury, right/left lobe AIS ISS WBMDCT/APMDCT Initial phase in which the liver was imaged arterial/PV
Study group
Liver injury group
No injury liver group
9 2
9 2
9 2
8 3
7 4
Nil
6: 5 27.1 (19–42) years 9:4
6: 5 26.8 (19–43) years 9:3
6: 5 26.3 (18–42) years 0:0
3.7 34.6 8:3 5:6
4 37.5 9:2 5:6
NA NA 9:2 10:1
AIS abdominal injury severity score, WBMDCT whole body multidetector CT, APMDCT abdomen and pelvis multi-detector CT, ISS injury severity score, NA not applicable
Using the trauma registry, two additional groups of patients with and without liver injuries were selected (Table 1). All three groups included the same number of patients matched for age, sex, abdominal injury severity scores (AIS), and mechanism of injury. The study group and the liver injury group were matched for age, sex, grade of liver injury, anatomical site of liver injury, AIS, technique, and scanner used to perform the examination. MDCT technique All MDCT scans were performed on 40- or 64-slice MDCT using intravenous contrast material (100 mL of 350 mg I2/ mL) injected using a biphasic injection technique (50 mL at 6 mL/s and 50 mL at 4 mL/s). A saline chase using 50 mL of normal saline was injected at 4 mL/s immediately following completion of the injection of intravenous contrast material. Imaging was performed using the WBMDCT technique in eight patients, or only the abdomen and pelvis in three. The WBMDCT scanning started 18 s following initiation of injection of intravenous contrast material, and the abdomen and pelvis scans were timed using a bolus pro-technique with the region of interest (ROI) placed on the descending distal thoracic aorta with a threshold to trigger at 120 HU. Scans were performed with the patient lying supine in the cranio-caudal direction from the vertex of the skull for the WBMDCT and from the diaphragms for the abdominal and pelvis MDCT (APMDCT) to the symphysis pubis. Delayed
Emerg Radiol
imaging was performed during the portal venous phase with both techniques from the diaphragms to the iliac crest. The following scan parameters were used for all examinations: tube voltage 120 kV, reference tube current 350 mAs for WBMDCT and 250 mAs for APMDCT, pitch 0.89, slice collimation 0.5×40 or 64, slice thickness 2 mm, reconstruction interval 1 mm, and rotation time of 1.5 s. Attenuation-based dose tube current modulation was used only for WBMDCT technique.
including tumors, inflammatory lesions, chronic liver disease thrombosis, or compression of the portal vein were present. Quantitative measurements of attenuation values of intravascular contrast material was made by a resident radiologist (CN) and an attending radiologist (NS) with 2 years of experience as shown in Fig. 4. ROIs were manually confined to the intravenous contrast material within the splanchnic and portal venous systems of all three groups of patients without including the adjacent liver parenchyma.
Qualitative and quantitative image analysis
Hepatic arteriography
Three board-certified trauma radiologists (SS, KS, SEM) with 3 to 27 years of experience blinded to the hepatic arteriography results evaluated the soft copy 3-mm thick axial and 2-mm thick multi-planar reformatted sagittal and coronal images of the abdomen and pelvis of all three groups of patients. The studies were presented in a random order to the three radiologists for consensus review for findings of hepatic APF. The radiologists reviewed both the initial and delayed images for the following findings of APF: transient hepatic parenchymal attenuation differences (THPAD) (Fig. 2), early increased attenuation of the peripheral portal vein compared with the central portal veins (Fig. 1, peripheral type), increased attenuation of the peripheral and central portal veins compared with the main portal vein (Fig. 3, central type), local simultaneous enhancement of the hepatic artery, and accompanying adjacent portal vein “double-barrel” (Fig. 2) or “rail track” (Fig. 2) signs. A note was also made of the vascular phase in which the solid organs were scanned during the initial imaging of the abdomen, if findings of hepatic APF were confined to injured liver segment, and if any other underlying causes of APF
Angiograms were performed in Toshiba (Toshiba America Medical systems, Inc., Tustin, CA) or Siemens Artis Zeego (Siemens AG Healthcare, Erlangen Germany) angiography suites. The procedures were performed by board-certified interventional radiologists. Flush aortagrams were performed at the discretion of the radiologist. All patients underwent selective angiography prior to embolization. Embolizations were performed using either gelfoam pledgets or slurry (Pharmacia and Upjohn Company, Kalamazoo, MI), or alternatively if a solitary vessel could be identified, embolization was performed with 0.18 microcoils (Cook Medical, Bloomington, IN).
Fig. 2 A 22-year-old woman pushed from a moving vehicle. Axial MDCT a arterial phase and b portal venous images demonstrate an area of transient hepatic parenchymal attenuation difference (white arrows) in the region of high-grade liver lacerations (black arrows). The double-barrel sign (arrowhead) was seen on multiple other axial images (not shown). d Maximum intensity projection coronal reformatted image demonstrated a peripheral branch of the hepatic artery (white arrow) and portal vein (black arrow) in the peripheral right lobe, the rail track sign. e Right hepatic arteriogram image confirms the arterioportal fistula (black arrow)
Statistical analysis A difference in attenuation values of contrast material within the early enhancing peripheral portal vein branch at the site of injury and attenuation values of contrast material within any other portal or splanchnic vein 20 HU or higher was
Emerg Radiol Fig. 3 A 29-year-old man shot in the right thoracoabdominal region underwent MDCT following damage control surgery. a Axial arterial phase image shows early enhancement of the right peripheral and central portal vein branches (arrows). b Attenuation measurements made of contrast material within the peripheral right portal vein and proximal left portal vein shows a threefold difference in attenuation values. Maximum intensity projection c axial and d coronal reformatted images show both the right hepatic artery (black arrows) and right central and peripheral portal vein (white arrows) due to a arterioportal fistula. e Hepatic arteriogram confirms a central-type arterioportal fistula with opacification of the right main portal vein (arrow) and its branches
considered abnormal and indicative of an APF for this study. The sensitivity, specificity, positive predictive value, and negative predictive value of the various MDCT findings in diagnosing APF were also determined. Fig. 4 Demonstration of anatomical location of measurements of attenuation values of intravascular contrast material
Medical records were reviewed to obtain mechanism of injury, injury severity scores, AIS, length of stay in hospital, hepatic complication, outcome of surgical intervention, and CBHA embolization. A generalized linear model (GLIM)
Emerg Radiol
binomial logistical regression was used to determine if the arterial phase or the portal venous phase was useful to demonstrate the MDCT findings of arterio-portal fistulas. A pvalue 20 HU
RPV–MPV or LPV–MPV>20 HU
2/5 5/5 2/2 5/8
3/5 5/5 3/3 5/7
2/5 (40 %) 8/10 (80 %) 2/4 (50 %) 8/11 (73 %)
2/5 (40 %) 10/10 (100 %) 2/2 (100 %) 10/13 (77 %)
1/6 (17 %) 12/12 (100 %) 1/1 (100 %) 12/17 (71 %)
0/6 (0 %) 12/12 (100 %) 0 (0 %) 12/18 (67 %)
Solid organs in arterial phase
PV portal vein, MPV main portal vein, LPV left portal vein, RPV right portal vein, PPV positive predictive value, NPV negative predictive value
Sensitivity 4/5 (80 %) Specificity 5/5 (100 %) PPV 4/4 (100 %) NPV 5/6 (83 %) Solid organs in P-V phase Sensitivity 1/6 (17 %) Specificity 6/6 (100 %) PPV 1/1 (100 %) NPV 6/11 (55 %)
(40 %) (100 %) (100 %) (63 %)
1/6 (17 %) 6/6 (100 %) 1/1 (100 %) 6/11 (55 %)
(60 %) (100 %) (100 %) (71 %)
1/6 (17 %) 6/6 (100 %) 1/1 (100 %) 6/11 (55 %)
Emerg Radiol
and technique using multiphase helical CT [11, 13, 20, 21]. The majority of posttraumatic APF diagnosed using CT described in the literature occurred from an iatrogenic etiology [11, 12, 20, 22]. No prior study has described the MDCT findings or technique required to demonstrate APFs following blunt and penetrating injuries. The current study only includes patients who sustained blunt force or penetrating trauma. The whole-body CT technique, a popular method for early initial evaluation of multitrauma patients in major trauma centers, was performed in 8 of the 11 patients (73 %) with APF using 40- or 64-slice MDCT scanners which allowed us to obtain images through the liver and spleen in the arterial or early portal venous phase [1–6]. Arterial phase images are ideally suited to demonstrate early filling of the peripheral branches of the portal vein prior to filling of the main, right, and left portal veins from splanchnic inflow. This finding was seen in 80 % (4/5) of the initial scans performed during the arterial phase and in 50 % (3/6) of scans performed during the early portal venous phase, with an overall sensitivity of 64 % (7/11) for the study group. With advancement of MDCT technology, if faster 128- or 256-slice MDCT become routinely available in major trauma centers, it will be feasible to image the solid organs during the arterial phase with the whole-body MDCT technique which is ideally suited to demonstrate the early filling of the portal vein branches. Both the double-barrel and rail track signs are variations of the early visualization of the peripheral portal vein. The double-barrel sign is seen when the artery and adjacent portal vein forming the APF run perpendicular to the plane of the image. The rail track sign is seen if these two vessels run in the plane of the image. These two signs are only seen when the solid organs are imaged during the arterial phase. Both signs have a very high specificity but a low sensitivity. In our study, THPAD was the most sensitive MDCT finding (82 %, 9/11) and also had a high specificity (95 %, 21/22) for diagnosing APF. This finding results from contrast material from the high-pressure hepatic arterial system entering through the organic intrahepatic fistula to the lowpressure portal venous system, causing increased attenuation of the liver parenchyma. The difference in attenuation between the parenchyma supplied by APF and the adjacent normal parenchyma is further enhanced by the dilution of contrast material by the unopacified portal flow [20, 21]. Typically, THPAD appears as a wedge-shaped area with straight margins seen in the periphery of the liver. In this study, all THPADs were seen either within or at the periphery of the injured segment (Fig. 2). When the initial scans imaged the solid organs during the arterial phase versus the early portal venous phase, THPAD was accentuated leading to increased sensitivity and specificity for APF diagnosis significantly (p00.0357). THPAD was also seen in one patient within the liver injury group. Review of the hepatic
angiograms of the patient indicated no sub-selective images were performed. It is difficult to know if this single patient had a small APF that was not diagnosed by angiography or the shunting was due to the grade IV liver injury. On prospective interpretation of the MDCTs and angiography examinations, the majority of APFs was not identified and was only diagnosed on retrospective review of the angiograms. THPAD seen within the injured segment can act as a sensitive maker of potential posttraumatic APF. The radiologist reviewing a CT demonstrating hepatic injury should look for THPAD and if present should search for the more specific APF findings, the rail track and double-barrel signs. Since a small APF may only be diagnosed by performing sub-selective angiograms of the injured hepatic segment where the THPAD is seen, MDCT can be helpful in localizing the anatomical segment of the APF for the interventional radiologist. The most useful quantitative measurement to diagnose an APF was measuring the difference in attenuation values of contrast material seen within the early filling peripheral portal vein branch and the contralateral portal vein. When the solid organs were imaged during the arterial phase, the sensitivity of this measurement increased 80 %, and the difference in attenuation values of contrast material seen within the early filling peripheral portal vein branch was more than 20 HU compared with all attenuation values of contrast material measured portal and splanchnic systems (Table 3). APFs encompass all fistulas between the splanchnic artery (hepatic artery, 65 %; splenic artery, 11 %; superior mesenteric artery, 10 %) and portal vein and are being encountered increasingly following penetrating trauma and post-transhepatic intervention procedures [22–24]. Wellrecognized complications of untreated APF include development of portal hypertension typically presenting with gastrointestinal bleeding, high-output cardiac failure, ascites, and diarrhea [24, 25]. The primary goal of treatment is to prevent development of portal hypertension. In this study, all four patients who were diagnosed prospectively at hepatic angiography were treated by embolization. Follow-up is not available on the outcome of the seven patients who were retrospectively diagnosed when the angiograms were reviewed for this study. Review of the literature indicates optimal therapy is not well-defined and depends on the size of the APF. Small APFs can be carefully followed using Doppler ultrasound, but larger central types of fistulas require interventional radiology, surgery alone, or a combination of both procedures. This study has several limitations. This is a retrospective study and only included a small number of patients. Two different scanners were used to image the study groups. Though careful attention was made in placing the ROI on the vascular contrast material within the
Emerg Radiol
peripheral portal vein branches, the very small diameter of the vessels may have led to some error in measuring attenuation values through partial volume effect of adjacent liver parenchyma. Conflict of interest The authors declare that they have no conflict of interest.
References 1. Loupatatzis C, Schindera S, Gralla J et al (2008) Whole-body computed tomography for multiple traumas using a triphasic injection protocol. Eur Radiol 18(6):1206–1214 2. Huber-Wagner S, Lefering R, Qvick LM, Körner M, et al. (2009) Effect of whole-body CT during trauma resuscitation on survival: a retrospective, multicentre study. Lancet 25;373(9673):1455–61 3. Leidner B, Beckman MO (2001) Standardizing whole-body computed tomography as a screening tool in blunt trauma. Emerg Radiol 8:20–28 4. Fanucci E, Fiaschetti V, Rotili A, Floris R, Simonetti G (2007) Whole body 16-row multislice CT in emergency room: effects of different protocols on scanning time, image quality and radiation exposure. Emerg Radiol 13(5):251–257 5. Karlo C, Gnannt R, Frauenfelder T et al (2011) Whole-body CT in polytrauma patients: effect of arm positioning on thoracic and abdominal image quality. Emerg Radiol 18(4):285–293 6. Ptak T, Rhea JT, Novelline RA (2003) Radiation dose is reduced with a single-pass whole-body multi-detector row CT trauma protocol compared with a conventional segmented method: initial experience. Radiology 229(3):902–905 7. Stengel D, Frank M, Matthes G et al (2009) Primary pan-computed tomography for blunt multiple trauma: can the whole be better than its parts? Injury 40(Suppl 4):S36–S46 8. Atluri S, Richard HM 3rd, Shanmuganathan K (2011) Optimizing multidetector CT for visualization of splenic vascular injury. Validation by splenic arteriography in blunt abdominal trauma patients. Emerg Radiol 18:307–312 9. Boscak A, Shanmuganathan K (2012) Splenic trauma: what is new? Radiol Clin North Am 50:105–122 10. Marmery H, Shanmuganathan K, Mirvis SE et al (008) Correlation of multidetector CT findings with splenic arteriography and surgery: prospective study in 392 patients. J Am Coll Surg 206(4):685–693
11. Quiroga S, Sebastià MC, Moreiras M, Pallisa E, Rius JM, AlvarezCastells A (1999) Intrahepatic arterioportal shunt: helical CT findings. Eur Radiol 9:1126–1130 12. Ahn JH, Yu JS, Hwang SH, Chung JJ, Kim JH, Kim KW (2010) Nontumorous arterioportal shunts in the liver: CT and MRI findings considering mechanisms and fate. Eur Radiol 20:385–394 13. Torabi M, Hosseinzadeh K, Federle MP (2008) CT of nonneoplastic hepatic vascular and perfusion disorders. Radiographics 28 (7):1967–1982 14. Guzman EA, McCahill LE, Rogers FB (2006) Arterioportal fistulas: introduction of a novel classification with therapeutic implications. J Gastrointest Surg 10:543–550 15. Tasar M, Gulec B, Bozlar U, Saglam M, Ugurel MS, Ucoz T (2005) Intrahepatic arterioportal fistula and its treatment with detachable balloon and transcatheter embolization with coils and microspheres. Clinical Imaging 29:325–330 16. Moore EE, Cogbill TH, Jurkovich GJ, Shackford SR, Malangoni MA, Champion HR (1995) Organ injury scaling: spleen and liver (1994 revision). J Trauma 38:323–324 17. Ibn Majdoub Hassani K, Mohsine R, Belkouchi A, Bensaid Y (2010) Post-traumatic arteriovenous fistula of the hepatic pedicle. J Vasc Surg 147:e333-e336. Epub 2010 Oct 6 18. Lumsden AB, Allen RC, Sreeram S, Atta H, Salam A (1993) Hepatic arterioportal fistula. Am J Surg 59(11):722–726 19. Wachsberg RH, Bahramipour P, Sofocleous CT, Barone A (2002) Hepatofugal flow in the portal venous system: pathophysiology, imaging findings, and diagnostic pitfalls. Radiographics 22(1):123–140 20. Choi D, Choo SW, Lim JH, Lee SJ, Do YS, Choo IW (2001) Opacification of the intrahepatic portal veins during CT hepatic arteriography. J Comput Assist Tomogr 25:218–224 21. Quiroga S, Sebastià C, Pallisa E, Castellà E, Pérez-Lafuente M, Alvarez-Castells A (2001) Improved diagnosis of hepatic perfusion disorders: value of hepatic arterial phase imaging during helical CT. Radiographics 21:65–81 22. Lee SJ, Lim JH, Lee WJ, Lim HK, Choo SW, Choo IW (1997) Transient subsegmental hepatic parenchymal enhancement on dynamic CT: a sign of postbiopsy arterioportal shunt. J Comput Assist Tomogr 21:355–360 23. Lumsden AB, Allen RC, Sreeram S, Atta H, Salam A (1993) Hepatic arterioportal fistula. Am Surg 59:722–726 24. Guzman EA, McCahill LE, Rogers FB (2006) Arterioportal fistulas: introduction of a novel classification with therapeutic implications. J Gasrointest Surg 10(4):543–550 25. Van Way CW, Crane JM, Riddell DH, Foster JH (1971) Arteriovenous fistula in the portal circulation. Surgery 70:876–890
