LIVER TRANSPLANTATION 16:513-519, 2010 ORIGINAL ARTICLE Long-Term Efficacy of Stent Placement for Treating Inferior Vena Cava Stenosis Following Liver Transplantation Jae Myeong Lee, 2 Gi-Young Ko, 1 Kyu-Bo Sung, 1 Dong Il Gwon, 1 Hyun-Ki Yoon, 1 and Sung-Gyu Lee 3 1 University of Ulsan College of Medicine, Asan Medical Center, Department of Radiology and Research Institute of Radiology, Seoul, Korea; 2 Inje University Paik Hospital, Radiology, Seoul, Korea; and 3 University of Ulsan College of Medicine, Asan Medical Center, Division of Hepatobiliary Surgery and Liver Transplantation, Surgery, Seoul, Korea The aims of this study were to evaluate both the efficacy of stent placement for treating inferior vena cava (IVC) stenosis and the patency of hepatic veins (HVs) following IVC stent placement. Fourteen hepatic transplant recipients underwent stent placement to treat IVC stenosis. The median interval between transplantation and stent placement was 32 days. Stents varied from 20-36 mm in diameter and were 60-120 mm long. We retrospectively analyzed the technical and clinical success, changes of hepatic venous flow, and the patency of the IVC stents. Stent placement was successful in all patients. Clinical success was achieved in 12 patients. Four patients underwent HV balloon angioplasty or stent placement through IVC stent meshes either immediately (n ¼ 1) or 12-110 days after (n ¼ 3) IVC stent placement. Nine of the 12 patients were healthy when this manuscript was completed, and the last follow-up computed tomography scan obtained at a median of 65.3 months after IVC stent placement revealed the patency of the stent-placed IVC and HVs. IVC stent placement seems to be an effective treatment with an excellent, long-term patency for treating posttransplant stenosis, although the possibility of hepatic venous outflow abnormalities following IVC stent placement should also be considered. Liver Transpl 16:513-519, 2010. VC 2010 AASLD. Received October 7, 2009; accepted January 2, 2010. Stenosis or occlusion of the inferior vena cava (IVC) is a relatively uncommon complication following liver transplantation (LT). IVC stenosis or occlusion may result from technical factors during the early postoperative period or may present later during the clinical course and reflects intimal hyperplasia and fibrosis at the anastomotic sites or extrinsic compression from the hypertrophied liver graft. 1 3 There is extensive literature concerning the endovascular approach for treating IVC stenosis, including balloon angioplasty and stent placement. The early and intermediate-term results of stent placement for treating IVC stenosis are acceptable; however, the long-term patency of stents in the IVC has not been reported. 1,4 9 In addition, there are only a few reports concerning the changes of hepatic vein (HV) flow following IVC stent placement. 1,8,9 In this article, we discuss the long-term patency of the IVC and the HV in a group of 14 patients who underwent IVC stent placement following LT. PATIENTS AND METHODS Patients The study was approved by our Institutional Review Board for retrospective review of patient charts and radiologic images including Doppler ultrasonography (US), computed tomography (CT), and venograms. From November 1997 through December 2007, 1730 living or cadaveric donor LTs were performed at Abbreviations: CT, computed tomography; HV, hepatic vein; IVC, inferior vena cava; US, ultrasonography. Address reprint requests to Gi-Young Ko, M.D., University of Ulsan College of Medicine, Asan Medical Center, Radiology & Research Institute of Radiology, 388-1 Poongnap-2-dong Songpa-ku, Seoul 138-040, Korea. Telephone: 82-2-3010-4438; FAX: 82-2-476-0090; E-mail: kogy@amc.seoul.kr DOI 10.1002/lt.22021 Published online in Wiley InterScience (www.interscience.wiley.com). VC 2010 American Association for the Study of Liver Diseases.
514 LEE ET AL. Underlying TABLE 1. Demographic Data of 14 Patients Interval No. Sex/Age Disease Surgery (days) Clinical Manifestations Stenotic Portion 1 M/47 HBV LC LDLT, Rt 42 Massive JP drain (>8 L/day) Suprahepatic 2 M/53 HBV LC OLT 14 Massive JP drain (>3 L/day) Hepatic 3 F/41 Fuminant LDLT, Rt 66 Massive JP drain (>2 L/day) Suprahepatic hepatitis 4 M/51 HBV LC, HCC 2nd LDLT, Rt 797 Leg edema, effusion, ascites Suprahepatic 5 M/44 HBV LC LDLT, Rt 10 Massive JP drain (>4 L/day) Hepatic 6* M/45 HBV LC 2nd OLT 154 Abnormal liver function, ascites Hepatic 7* F/41 HBV LC LDLT, Rt 11 Massive JP drain (>3 L/day) Hepatic 8 M/54 HBV LC 2nd OLT 4 Abnormal liver function, massive Infrahepatic JP drain (>5 L/day) 9 M/36 HBV LC OLT 2 Abnormal liver function, massive Suprahepatic JP drain (>2 L/day) 10 y M/45 Alcoholic LC OLT 0 Central venous pressure (;) Hepatic 11 F/24 Fulminant 2nd OLT 598 Ascites, hepatic congestion on CT Suprahepatic hepatitis 12 M/36 Budd-Chiari LDLT, Rt 57 Leg edema, IVC graft collapse, Diffuse and hepatic congestion on CT 13 F/67 HCV LC, HCC LDLT, Rt 22 Leg edema Hepatic 14 F/27 Primary biliary 2nd OLT 170 Hepatic congestion on CT Suprahepatic cirrhosis *These patients had undergone stent placement in the right hepatic vein and the inferior hepatic vein 3 and 6 days, respectively, prior to IVC stent placement. y This patient underwent stent placement intraoperatively. our institution. Retrospective review of these patients medical information identified 14 (0.8%) patients who had undergone IVC stent placement to treat IVC stenosis. The patient demographics are shown in Table 1. The initial diagnosis of the IVC stenosis was made using a combination of enhanced CT, Doppler US, and the clinical manifestations. One patient did not undergo CT or US prior to IVC stent placement because stent placement was performed during the LT. In this patient, reduction of central venous pressure due to an IVC kink was observed during closure of the abdominal wall. Therefore, IVC stent placement was performed intraoperatively. Two other patients had already undergone stent placement in the right HV and the inferior HV at 3 and 6 days, respectively, prior to IVC stent placement. Whole-liver LT recipients (n ¼ 7) had undergone 2 standard, end-to-end cavocaval anastomoses, and the living donor LT recipients (n ¼ 6) had undergone endto-end anastomosis of the donor s right HV to the recipient s right HV. Four of the latter patients had also undergone end-to-side anastomosis of the donor inferior HV, V5 or V8 HV (>5 mm in diameter) to the IVC. The remaining 1 living donor LT recipient had undergone IVC reconstruction with a 30-mm-diameter polytetrafluoroethylene graft because of underlying Budd-Chiari syndrome. End-to-side anastomoses of the donor right and inferior HVs to the IVC graft were then performed. Stent Placement Interventional procedures were performed under fluoroscopic guidance with conscious sedation (n ¼ 13) or under general anesthesia (n ¼ 1). Conventional cavography and hepatic venography were performed via the right internal jugular vein and/or the right common femoral vein in order to evaluate the status of the IVC and the HVs flow. Prestenting and poststenting pressure gradients across the IVC stenosis were obtained in all except for 1 patient who underwent stent placement during LT. Significant stenosis was considered to exist when the cavography revealed a stenosis greater than 50% of the normal IVC or when a pressure gradient across a stenosis was more than 5 mm Hg. The criteria of primary stent placement was earlypostoperative (<4 weeks) IVC stenosis to avoid anastomotic rupture during balloon angioplasty. In the remaining patients, balloon angioplasty (>14 mm) was usually performed initially, and stent placement was followed if postangioplasty venography showed elastic restenosis of more than 50% or a pressure gradient of more than 5 mm Hg, or if clinical improvement following balloon angioplasty was unsatisfactory. Poststenting balloon angioplasty was performed in cases with residual waist deformity of the placed stent of more than 50%. Stents (20-36 mm in diameter; 60-120 mm in length; TaeWoong Corp., Seoul, Korea; S&G, Seoul, Korea) with approximately a 20% larger diameter than
EFFICACY OF STENT PLACEMENT FOR TREATING STENOSIS 515 TABLE 2. Clinical Outcomes Following Endovascular Treatments Pressure gradients (mmhg) Clinical No. Procedures Stent Size (mm) Prestenting Poststenting Change of Clinical Manifestations Follow-Up Duration (days) Final Outcome 1 PTA/stenting 24 70 13 4 Disappear 4239 Alive 2 Primary stenting 24 120 7 5 Disappear 3194 Alive 3 Primary stenting 28 80 6 2 Disappear 3102 Alive 4 PTA/stenting 22 120 10 1 Disappear 172 Died of sepsis 5 Primary stenting 28 120 6 1 Disappear 2734 Alive 6 PTA/stenting 28 120 10 4 No change 1 Died of multiorgan failure 7 Primary stenting 20 80 6 1 Disappear 2797 Alive 8 Primary stenting 28 100 3 1 No change 0 Died of transplantation failure 9 Primary stenting 28 100 12 1 Disappeared ascites but stationary liver function 51 Died of intracerebral hemorrhage 10 Primary stenting 24 80 N/A N/A Uneventful 2151 Alive 11 PTA/stenting 24 80 12 4 Disappear 1078 Alive 12 PTA/stenting 36 120 6 2 Disappear 786 Alive 13 Primary stenting 26 80 7 2 Disappear 117 Died of hepatic arterial occlusion 14 PTA/stenting 20 x 60 8 1 Disappear 356 Alive that of the patent IVC adjacent to a stenosis were used. Thesestentswerewovenfromasinglethreadof0.20- mm or 0.25-mm nitinol wire in a tubular configuration and without barbs; they were loaded into a 12-French introducer for deployment. Hepatic venograms and pressure gradients between the HVs and the IVC were obtained following IVC stent placement in patients who had abnormal HV outflow waveform on preprocedural US or hepatic congestion on CT. If the pressure gradient was more than 5 mm Hg, balloon angioplasty and/or stent placement was performed through the IVC stent meshes. In general, anticoagulants were not administered either during or after stent placement except for catheter irrigation using heparin-mixed saline (5000 units/500 ml) which was performed during the procedure. care, an additional surgical or interventional manipulation, adverse sequelae, or death. Patency of the HV and IVC stents were evaluated by enhanced CT, Doppler US, and/or venography. US was routinely performed the day after stent placement, weekly until the patient was discharged from our hospital, and then 1, 6, and 12 months after discharge. CT was performed within 1 week after stent placement and then randomly following discharge. Statistical Analysis The paired t test was performed to analyze the difference in the prestenting and poststenting pressure gradients. The analysis was conducted using SPSS software (version 14.0; SPSS, Chicago, IL). Follow-Up The following parameters were documented retrospectively: technical and clinical success; complications; changes of HV flow following IVC stent placement; and patency of the IVC stents. Technical success was defined as successful reduction of the IVC pressure gradient less than 6 mm Hg and with residual stenosis less than 30% of the normal IVC diameter. In one patient with a preprocedural IVC pressure gradient of 3 mm Hg, residual stenosis less than 30% was considered to be a technical success. Clinical success was defined as amelioration of the patients presenting signs or symptoms. Major complications were defined as those necessitating an increased level of RESULTS The median interval between LT and stent placement was 32 days (range, 0-797 days). Cavogram revealed occlusion of the suprahepatic IVC in 1 patient and stenosis of the suprahepatic (n ¼ 5), hepatic (n ¼ 6), and infrahepatic (n ¼ 1) segmental IVC in 12 patients. One patient had diffuse stenosis of the graft IVC caused by a perigraft hematoma which did not resolve despite drainage tube placement. Cavogram also showed collaterals via the vertebral venous plexus (n ¼ 4) and reflux into the HVs (n ¼ 2) relating to IVC stenosis. No patient had definite IVC thrombosis seen on CT and cavography. Primary IVC stent placement was performed on 8 patients, and balloon angioplasty followed by stent placement was performed on 3
516 LEE ET AL. On Preprocedural TABLE 3. Status of Hepatic Venous Outflow on Doppler US or CT HV Status During IVC Number of HVs Crossed by Number of Patent HVs on the Last Follow-Up Image Follow-Up Duration Tx for Abnormal HV Flow During or After IVC No. US Stenting an IVC Stent CT or US (days) Stenting 1 N/A N/A 2 2 3729 Inferior HV PTA 110 days later 2 Normal N/A 5 5 2754 None 3 N/A y Inferior HV stenosis 2 1 3125 No treatment 4 Occluded Failed negotiation 2 2 113 Failed negotiation the right HV 5 Normal N/A 2 2 1780 Inferior HV stenting 12 days later 6* Normal N/A 3 N/A N/A N/A 7* Normal N/A 2 1 2499 No treatment 8 Normal N/A 3 N/A N/A N/A 9 Normal N/A 3 3 1 None 10 N/A Normal 3 3 1959 None 11 Monophasic in Elevated P.G. 2 2 302 Right and left HVs the right HV in the right & stenting during left HVs IVC stenting 12 Monophasic or Elevated P.G. in 2 2 754 Right and inferior biphasic in all HVs the right HV HV stenting 13 days later 13 Normal N/A 2 2 116 None 14 Reversed flow Normal 3 3 229 None in the right HV *These patients had undergone stent placement in the right hepatic vein and the inferior hepatic vein 3 and 6 days, respectively, prior to IVC stent placement. y The stenosis was already diagnosed on a HV venogram obtained 17 days prior to IVC stent placement. patients. In 3 other patients, stent placement was performed 3-14 days after initial balloon angioplasty because of the patients unsatisfactory clinical improvement. Technical success was achieved in all patients, and IVC stents were placed from the suprarenal to the suprahepatic IVC. Poststenting cavogram showed fluent IVC flow with disappearance of vertebral venous plexus and HV reflux in all patients. The pressure gradients across the IVC stenoses before and after stent placement were 8.2 6 3.0 mm Hg (range, 3-13 mm Hg) and 2.2 6 1.5 mm Hg (range, 1-5 mm Hg), respectively (P < 0.001; Table 2). On preprocedural US or venography, abnormal HV flow was seen in 5 patients, 3 of whom also had hepatic congestion in the liver graft as seen on CT (Table 3). Hepatic venography following IVC stent placement was performed in 6 patients, including the 5 with abnormal HV flow seen on preprocedural US and 1 patient who underwent intraoperative IVC stent placement; 3 of the 6 patients showed abnormal HV outflow. One patient (No. 3) revealed a tight stenosis of the inferior HV anastomosis; however, further treatment was not performed because at that time the patient had a patent right HV with normal liver function and without hepatic congestion. In fact, the stenosis was already diagnosed on an HV venogram obtained 17 days prior to IVC stent placement. One patient (No. 11) who had hepatic congestion seen on the preprocedural CT revealed high-pressure gradients between both the right and left HVs and the IVC. Therefore, stents were placed following balloon angioplasty through IVC stent meshes in each HV. Thereafter, the pressure gradients declined from 10 to 4 mm Hg and from 12 to 2 mm Hg, respectively, and follow-up US and CT obtained within 1 week following stent placement revealed patent HV outflow without hepatic congestion. In another patient (No. 4), negotiation of the occluded right HV failed. Major complications occurred in 2 patients (No. 6 and 8). These 2 patients died of multiorgan failure and transplantation failure, respectively, within 2 days after IVC stent placement. Clinical success was achieved in 12 (86%) patients (Table 2). During the mean clinical follow-up period of 57.7 6 48.9 months (range, 1.7-141.3 months; median, 53.8 months), 3 of the 12 patients died of sepsis, intracerebral hemorrhage, or HA occlusion within 6 months after stent placement. In these patients, however, the last follow-up CT or US revealed patent IVC. The remaining 9 patients were healthy when this manuscript was completed, and the last follow-up CT
EFFICACY OF STENT PLACEMENT FOR TREATING STENOSIS 517 Figure 1. Patient number 5. (A) Preprocedural cavogram shows hepatic segmental vena caval stenosis (arrowhead). (B) Cavogram following stent placement shows improvement in the stenosis. (C) Axial CT obtained 58 months following stent placement shows a patent vena cava (arrow) and right HV (curved arrow). (n ¼ 9) and cavogram (n ¼ 1) obtained at a median of 65.3 months (range, 7.6-124.3 months; mean, 63.4 6 41.9 months) after IVC stent placement revealed a patent stent-placed IVC without stent migration (Fig. 1). In all, 30 HV outlets (mean, 2.5 6 1.0; range, 2-5) including the right (n ¼ 12), left and/or middle (n ¼ 7), inferior (n ¼ 8), and V5 or V8 (n ¼ 3) HVs were crossed by IVC stents in the 12 patients with clinical success (Table 3). Except for 1 inferior HV which was not intentionally treated (No. 3), follow-up Doppler US or CT revealed abnormal HV flow or hepatic congestion in 5 (17.2%) HVs of 4 patients. In 2 patients (No. 1 and 5), stenosis of the inferior HV was diagnosed on 12-day and 110-day follow-up US, respectively, and the stenosis disappeared following balloon angioplasty and stent placement, respectively. In 1 patient (No. 12), 8-day follow-up US and CT revealed stenoses of the right and the inferior HV outlets with hepatic congestion, and the stenoses with hepatic congestion disappeared following stent placement in each HV through IVC stent meshes (Fig. 2). In 1 patient (No. 7), the inferior HV in which a 6-mm-diameter stent was placed prior to IVC stent placement showed thrombosis on the 30-month follow-up CT. However, further management was not performed because of the patent right HV without any specific clinical manifestations relating to the inferior HV occlusion. DISCUSSION Stenosis of the IVC following LT is uncommon and occurs in less than 3% of LT recipients. 1,9 Balloon angioplasty is primarily used to treat IVC stenosis following LT, although the patency does not last because of the inherent IVC elasticity and elastic recoil of the fibrotic tissue in chronic stenosis. 7,9,10 Weeks et al. 1 reported that initial balloon angioplasty failed in 13 of 16 patients who then required stent placement in their literature review of 43 patients who had IVC abnormalities following LT. In comparison, several literature reports have indicated that stent placement is very effective for treating IVC stenosis and has durable patency as well. 1,4,6 9 Weeks et al. 1 also reported 100% objective patency in 6 patients with an average of 491 days of follow-up, and Borsa et al. 9 reported a primary patency of 83% in 6 patients, with a mean follow-up period of 11 months. We also initially performed balloon angioplasty on 6 patients; however, all of them eventually had to undergo stent placement owing to elastic recoil. Nine of the 14 study patients who had survived until the writing of this manuscript revealed patent IVC without recurrence with a median imaging follow-up of 65.3 months. Therefore, we agree that stent placement is an effective and durable modality for treating IVC stenosis. As for IVC stents, Weeks et al. 1 maintained that the stent required certain properties, such as superior radial strength, barbs to ensure the stability of the stent after deployment, a large stent diameter for caval deployment, and wide interstices to minimize interference with the HV outflow. Although the stents we used in this study had no barbs, stent migration did not occur in any study patient during the follow-up period. This was probably due not only to the sufficient expansile power of the stent but also to the sufficiently large diameters and long lengths of the stent. In the current study, we intentionally placed IVC stents from the suprarenal to the suprahepatic IVC with approximately 20% larger diameters than that of the patent IVC in order to prevent stent migration. In addition, because the stents we used had relatively wide stent meshes, HV balloon angioplasty or stent placement through the IVC stent meshes was possible
518 LEE ET AL. Figure 2. Patient number 12. (A) Preprocedural multiplanar oblique coronal reconstruction CT shows diffuse stenoses of the vena cava graft (arrowheads) caused by perigraft hematoma (curved arrow). Note the grossly patent right HV (arrow). (B) Cavogram reveals the stenoses of the vena cava graft (arrowheads). (C) Cavogram following vena cava stent placement shows much improved stenoses. (D) Sixmonth follow-up multiplanar oblique coronal reconstruction CT shows a patent vena cava as well as the stent-placed HVs. without technical difficulty in 4 patients. Therefore, we assume that the IVC stents we used possess the properties that Weeks et al. 1 recommended. Although IVC stent placement shows encouraging results, there are several potential limitations. First, IVC stent placement may interfere with HV flow because HV outlets are crossed by IVC stent meshes. In addition, there may be a risk of kink or distortion of an HV caused by an overexpanded IVC stent. Weeks et al. 1 reported partial HV outflow obstruction in 1 of 9 patients who underwent Gianturco stent placement to treat IVC stenosis or torsion following LT. They successfully treated the obstruction with HV stent placement through the Gianturco stent interstices 6 days after IVC stent placement. In our study, one patient (No. 11) underwent HV stent placement immediately after IVC stent placement because of high pressure gradients. Although the exact etiology of the high-pressure gradients in this patient was unclear because we did not measure the gradients prior to IVC stent placement, we assume that it might have been caused by an HV kink caused by the expanded IVC stent. In fact, this patient s preprocedural CT and cavogram during IVC stent placement revealed the widely patent HVs but also a stenosis of the suprahepatic IVC. In addition, 3 other patients also underwent HV balloon angioplasty or stent placement several days after IVC stent placement. The connection between IVC stent placement and an HV flow abnormality was unclear in 2 of these 3 patients. However, in the remaining patient (No. 12), we assumed that the IVC stent itself interfered with the HV outflow. In this patient, a right HV venogram with a pressure gradient before and after IVC stent placement had been obtained, and the pressure gradient had risen from 1 mm Hg before IVC stent placement to 5 mm Hg following IVC stent placement. However, further
EFFICACY OF STENT PLACEMENT FOR TREATING STENOSIS 519 management was not performed at that time because of the grossly patent HV outflow and relatively less significant pressure gradient. One-day follow-up US revealed a biphasic waveform of the right HV, but the waveform changed to monophasic with newly developed hepatic congestion on the 8- day follow-up US and CT. Therefore, we recommend careful evaluation of HV outflow following IVC stent placement, because there is a risk of HV outflow disturbance caused by an IVC stent. Second, IVC stents may interfere with retransplantation, particularly if a stent protrudes into the right atrium. However, Weeks et al. 1 reported 2 patients who underwent retransplantation without technical difficulty following IVC stent placement. Although none of our patients underwent retransplantation following IVC stent placement, careful deployment of an IVC stent so that it does not extend into the right atrium, may be important. In summary, IVC stent placement seems to be an effective treatment modality with an excellent longterm patency for treating posttransplant IVC stenosis. However, the possibility of an HV outflow abnormality following IVC stent placement should also be considered. ACKNOWLEDGMENT We thank Bonnie Hami, M.A. (USA), for editorial assistance in preparing this manuscript. REFERENCES 1. Weeks SM, Gerber DA, Jaques PF, Sandhu J, Johnson MW, Fair JH, et al. Primary Gianturco stent placement for inferior vena cava abnormalities following liver transplantation. J Vasc Interv Radiol 2000;11:177-187. 2. Ko GY, Sung KB, Yoon HK, Kim HJ, Song HY, Seo TS, et al. Endovascular treatment of hepatic venous outflow obstruction after living donor liver transplantation. J Vasc Interv Radiol 2002;13:591-599. 3. Darcy MD. Management of venous outflow complications after liver transplantation. Tech Vasc Interv Radiol 2007; 10:240-245. 4. Simo G, Echenagusia A, Camunez F, Quevedo P, Calleja IJ, Ferreiroa JP, et al. Stenosis of the inferior vena cava after liver transplantation: treatment with Gianturco expandable metallic stents. Cardiovasc Intervent Radiol 1995;18:212-216. 5. Zulke C, Berger H, Anthuber M, Jauch KW. Detection of suprahepatic caval stenosis following liver transplantation and treatment via balloon-expandable intravascular stent. Transpl Int 1995;8:330-332. 6. Althaus SJ, Perkins JD, Soltes G, Glickerman D. Use of a Wallstent in successful treatment of IVC obstruction following liver transplantation. Transplantation 1996;61: 669-72. 7. Pfammatter T, Williams DM, Lane KL, Campbell DA Jr, Cho KJ. Suprahepatic caval anastomotic stenosis complicating orthotopic liver transplantation: treatment with percutaneous transluminal angioplasty, Wallstent placement, or both. AJR Am J Roentgenol 1997;168:477-480. 8. Mizuno S, Yokoi H, Yamagiwa K, Tabata M, Isaji S, Yamakado K, et al. Outflow block secondary to stenosis of the inferior vena cava following living-donor liver transplantation? Clin Transplant 2005;19:215-219. 9. Borsa JJ, Daly CP, Fontaine AB, Patel NH, Althaus SJ, Hoffer EK, et al. Treatment of IVC anastomotic stenosis with the Wallstent endoprosthesis after OLT. J Vasc Interv Radiol 1999;10:17-22. 10. Raby N, Karani J, Thomas S, O Grady J, Williams R. Stenoses of vascular anastomoses after hepatic transplantation: treatment with balloon angioplasty. AJR Am J Roentgenol 1991;157:167-171.