Embolization Therapy: Principles and Clinical Applications, 1 Ed.

Bland Embolization

Karen T. Brown

Because the hepatic artery provides most blood flow to hepatocellular carcinoma (HCC), with the portal vein providing trophic blood supply to the liver parenchyma,1 and the paucity of other effective nonsurgical treatments, it is not surprising that transarterial methods of treating unresectable HCC have become a mainstay of therapy over the last four decades. One of the early reports of transcatheter management of HCC by Charnsangavej et al.2 in 1983 described results following two methods of treatment. The first group received continuous hepatic arterial infusion (HAI) of floxuridine, doxorubicin, and mitomycin C in 14 patients requiring a 5-day hospital stay, with two courses of treatment given at 4- to 6-week intervals. The second group of 9 patients (2 of whom crossed over from the HAI group) was treated with hepatic artery embolization (HAE) using Ivalon (Unipoint Industries, High Point, North Carolina). Response rates in the two groups were similar: 71% in HAI group versus 67% in HAE group, with median survival of 12.3 months and 17.4 months, respectively. Later that year, Yamada et al.3 described the use of gelatin sponge cut into 1- to 2-mm pieces and “permeated” with mitomycin C or doxorubicin to treat 120 patients with HCC, infusing this chemotherapy-soaked gelatin sponge into the artery feeding the tumor. The 1-, 2-, and 3-year cumulative survival rates were 44%, 29%, and 15%, respectively. Yamada et al.3 noted that given the knowledge that HCC derives blood supply from “only the hepatic artery,”1 embolization can be expected to result in selective necrosis of the tumor tissue. They also suggested that although the effect of the single dose of mitomycin or doxorubicin may have been enhanced, it is possible that ischemia might be the primary mechanism of tumoricidal effect.3In a 1987 study by Takayasu et al.,4 patients with HCC were divided into three groups for intra-arterial treatment. Group A received intra-arterial Lipiodol (Guerbet USA, Bloomington, Indiana) alone; group B, an emulsion of Lipiodol with doxorubicin; and group C, the same emulsion followed by embolization with gelatin sponge. Group C showed the best therapeutic effect; there was no significant difference in response between groups A and B, with practically no therapeutic effect from Lipiodol alone. Although these results can be interpreted in several ways, the authors pointed out that the results achieved using the anticancer emulsion plus gelatin sponge were superior to those previously described for embolization with gelatin sponge and anticancer agents and posited that the combination of Lipiodol blocking small vessels and gelatin sponge occluding more proximal feeding arteries resulted in an anoxic state “independent of drug sensitivity of the tumor.” They also noted an association between uptake of Lipiodol by the tumor and therapeutic effect. In 1989, Nakamura described a similar effect on survival, comparing 100 patients treated with doxorubicin, iodized oil, and gelatin sponge to 104 historical controls embolized with gelatin sponge and either doxorubicin (96) or mitomycin (8) achieving 1-, 2-, and 3-year survivals of 53%, 33%, and 18% in the iodized oil group and 45%, 16%, and 4% in those treated with only gelatin sponge plus chemotherapy. Taken together, these findings confirm that transcatheter treatment of HCC brought about a radiologic response; however, several questions remained unanswered. What was the primary driver of that therapeutic effect: chemotherapy or ischemia? Does Lipiodol indeed prolong the intratumoral dwell time of chemotherapy? And, finally, does an imaging response to treatment translate into improved survival?

Pharmacokinetic data supporting the use of intra-arterial chemotherapy or chemotherapy plus Lipiodol administered as an emulsion with an embolic agent, as commonly used today, is less than convincing. Studies clearly demonstrating high and prolonged concentration of chemotherapeutic agents within the tumor were performed using mitomycin C, doxorubicin, and aclarubicin dissolved in hydrocarbon solvents and then in Lipiodol5 or using a lipophilic agent,6 methods which are not used clinically. In the animal study by Konno,5 when the chemotherapeutic agent was dissolved in water and then mixed with Lipiodol and administered as an emulsion, concentration of drug in the tumor was high immediately but low at 6 hours, 1 day, and 7 days. In a study of 18 patients by Raoul et al.,7 doxorubicin was given to patients intra-arterially using three different methods: alone as an infusion, emulsified with Lipiodol, or with Lipiodol and gelatin sponge. There was no significant difference in total amount of doxorubicin released into the circulating blood, but patients in whom gelatin sponge was used had less released within the first hour of treatment. Another study evaluated intra-arterial doxorubicin versus doxorubicin with Lipiodol8 and found no difference in the area under the concentration-time curve, or terminal half-life, and no difference in pharmacokinetic profile or systemic toxicity using the same dose schedule compared to administering the doxorubicin intravenously. Even if the pharmacokinetic profile was more encouraging, none of these anticancer agents had ever been demonstrated to have a significant effect on the survival of patients with HCC when administered intravenously, making it difficult to know which agent, or which combination of agents, to use clinically. There are many embolic agents available for embolization. This has led to myriad methods of transcatheter intra-arterial therapy for HCC, limiting the ability to compare results between different groups. Meta-analysis by Simonetti et al.9 in 1997 failed to support the effectiveness of nonsurgical treatments for HCC, but some evidence of “moderate benefit” emerged from trial using tamoxifen and transcatheter arterial embolization (without iodized oil). Combining the heterogeneous population of patients with HCC who present with varied etiology and stage of underlying liver disease, size and number of tumors, and vascular invasion status, with the diverse methods used to treat them, made it even more difficult to compare results of treatment between diverse investigators, and there were no strong randomized trials.

The lingering question of whether reported radiologic response would translate into survival benefit was addressed with two randomized trials reported in 2002, which compared transarterial therapies to supportive care. Lo et al.10 treated a group of 80 patients, most had underlying hepatitis B, with either transarterial chemoembolization (TACE) using an emulsion of cisplatin and Lipiodol (Lipiodol Ultra-Fluide; Guerbet, Bloomington, Indiana) with gelatin sponge particles or best supportive care, using survival as the primary end point. There were 40 patients in each group; the TACE group received a total of 192 courses of embolization, median 4.5 per patient. TACE was associated with a significantly better actuarial survival of 57%, 31%, and 26% at 1, 2, and 3 years compared to 32%, 11%, and 3% in the best supportive care group. Later that same month, Llovet et al.11 published a randomized trial with three arms: chemoembolization with doxorubicin, lipiodol, and gelatin sponge; gelatin sponge alone; and best supportive care. Again, the primary end point was survival; response was a secondary end point. The study design was sequential; patients were assessed every 3 months and the trial was stopped when a significant survival advantage was demonstrated for the chemoembolization group. At the time the study was stopped, the Z value of the sequential triangular test for the embolization group remained within the triangular boundaries, indicating the need to recruit additional patients to achieve a valid conclusion. Survival probabilities at 1 and 2 years were 75% and 50% for embolization, 82% and 63% for chemoembolization, and 63% and 27% for symptomatic treatment, respectively. Although the authors believe that chemoembolization leads to a survival benefit compared to embolization alone, this could not be concluded statistically because the study was stopped before such a benefit could be demonstrated or refuted. Additional patients would have needed to be recruited into the embolization arm to reach a valid conclusion with regard to the impact of embolization versus symptomatic treatment. Of interest, 30 patients treated achieved an objective response by imaging: 16 were in the embolization group, despite the fact that only 35 patients in that group were treated, and the other 14 were among the 40 patients treated with chemoembolization. Although this study provides level 1 statistical evidence that chemoembolization offers a survival benefit compared to best supportive care, it does not allow for any statement regarding the effectiveness of chemoembolization versus embolization alone or embolization versus best supportive care.

That same year, Camma et al.12 published a meta-analysis of randomized chemoembolization trials. There were 18 randomized controlled trials pooled for analysis. The authors concluded that “chemoembolization significantly reduced the overall 2-year mortality rate (odds ratio, 0.54; 95% CI: 0.33, 0.89; P = .015) compared with nonactive treatment . . . overall mortality was significantly lower in patients treated with transarterial embolization (TAE) than in those treated with transarterial chemotherapy (odds ratio, 0.72; 95% CI: 0.53, 0.98, P = .39) and that there is no evidence that transarterial chemoembolization is more effective than TAE (odds ratio 1.007; 95% CI: 0.79, 1.27; P = .95), which suggests that the addition of an anticancer drug did not improve the therapeutic benefit.”12

If indeed the primary effect of transarterial treatment is from ischemia rather than a local chemotherapeutic effect, it follows that methods that maximize ischemia should be adopted. To that end, calibrated microspheres that are capable of occluding intratumoral vessels should be employed for HAE. In a very elegant study using iron oxide labelled Embosphere Microspheres (Merit Medical Systems, Inc., South Jordan, Utah) published in 2008, Lee and colleagues13 looked at the distribution of 100- to 300-μm and 300- to 500-μm spheres after embolization of VX2 tumor in rabbits. They found that on both magnetic resonance and histology, only the smaller 100- to 300-μm particles penetrated the tumor, whereas the 300- to 500-μm particles were found outside the tumor. Presuming that occlusion of intratumoral vessels will result in the most pronounced ischemia and resultant coagulation necrosis, these finding suggests that 100- to 300-μm particles or smaller should be used for embolization of liver tumors. Indeed, in 2008, Maluccio et al.14 described results obtained in 322 patients thus treated at Memorial Sloan Kettering Cancer Center using small particle embolization alone. Median survival was 21 months, with 1, 2, and 3 years overall survival of 66%, 46%, and 33%, respectively. When patients with extrahepatic disease or portal vein tumor were excluded, the overall survival in the 159 patients with liver-only disease rose to 84%, 66%, and 51% at 1, 2, and 3 years, respectively, with a median survival of 40 months. These results are surprisingly (or perhaps not so surprisingly) similar to the 1- and 2-year overall survival with chemoembolization described by Llovet et al.11 (82% and 63%) and are some of the best results ever reported for transcatheter treatment of HCC.

DEVICE/MATERIAL DESCRIPTION

Small calibrated microspheres without chemotherapy or Lipiodol are used to embolize HCC and other hypervascular tumors. Embosphere Microspheres are preferred by the author because of a demonstrated low incidence of postembolization occlusion of hepatic vessels.15 These particles are made from trisacryl cross-linked acrylic copolymer combined with gelatin and are hydrophilic and compressible. They received 510(k) clearance in the United States in 2000 and are U.S. Food and Drug Administration (FDA) approved for uterine fibroid embolization (UFE). Having been available for almost 15 years, these microspheres are some of the most studied embolic agents currently in use and are available in sizes ranging from 40 to 120 μm to 900 to 1,200 μm. There are other calibrated spherical embolics that can be used for HAE including Bead Block (Biocompatibles UK LTD), available in sizes ranging from 100 to 300 μm to 900 to 1,200 μm and produced from a biocompatible polyvinyl alcohol (PVA) hydrogel, and Embozene Microspheres (CeloNova BioSciences, Inc., San Antonio, Texas), available in sizes ranging from 40 to 900 μm and composed of a hydrogel core with a surface modification of Polyzene-F, a biocompatible polymer. Bead Block received 510(k) approval for neurovascular embolization in 2004 and is now cleared for use in hypervascular tumors. Embozene Microspheres are the most tightly calibrated microspheres currently available. Embozene Microspheres received 510(k) approval from the FDA in December 2008 for use in hypervascularized tumors.

TECHNIQUE

After performing celiac and superior mesenteric angiography to define the direction of portal blood flow, arterial anatomy, and primary blood supply to the tumor(s), a catheter is placed selectively into the vessel or vessels supplying the tumor to be treated. Embolization is performed as selectively as possible often using a coaxial catheter. The catheter is advanced into the vessel(s) to be treated (Fig. 34.1), and the microspheres suspended in contrast and maintained in suspension by admixing through a three-way stopcock (Fig. 34.2) are injected under constant fluoroscopic control until stasis is evident.

Embolization is begun with the smallest size particle; in the case of Embosphere Microspheres, 40- to 120-μm spheres are used except in situations where the risk of systemic embolization is felt to be high, in which case embolization is begun with 100- to 300-μm spheres. Death related to systemic embolization of these small particles to the pulmonary circulation has been described,16 and use of these small particles is discouraged in the following situations: tumors larger than 10 cm, large tumors high in the dome adjacent to the hemidiaphragm, tumors invading the hepatic veins, or with unexpected early hepatic vein drainage demonstrated angiographically.

When a large tumor burden is being treated and there is persistent antegrade flow after 10 mL of the 40- to 120-μm spheres have been injected, embolization is continued with the next largest size, 100- to 300-μm spheres. In situations where there is persistent flow after 10 mL of the 100- to 300-μm spheres have been used, embolization is continued with the next largest size, 300- to 500-μm spheres. There are two reasons to upsize the particles used. In theory, the smallest particles occlude the smallest intratumoral vessels, and it should then be possible to move to the next size to occlude upstream vessels. In addition, it has been demonstrated that it takes a larger volume of smaller particles to occlude a given vascular bed.17 Because the particles are mixed with contrast, embolization of large tumors can require administration of a significant amount of contrast if only small particles are used; moving to larger spheres limits the contrast load and duration of procedure. If more than one vessel supplies the tumor, the catheter is placed sequentially into the next vessels to be embolized and embolization begun again with the smallest 40- to 120-μm spheres. Once again, if there is persistent antegrade flow after 10 mL of the 40- to 120-μm spheres have been injected, embolization is continued with the next largest size, 100- to 300-μm spheres, and so on until all target vessels have been embolized to stasis. Stasis as a hard end point is defined as lack of antegrade flow observed for at least 5 heartbeats18 with lack of opacification of the target vessel on postembolization digital angiography (Fig. 34.3). On occasion, antegrade flow is maintained by development of what have come to be called “embolization sinks,” or spaces within the tumor where contrast pools despite the tumor vessels being completely occluded (Fig. 34.4). In this case, a small amount of PVA may be added to stabilize the end point.

The entire liver is never embolized at once because of the risk of hepatic failure. Patients who have multifocal disease necessitating treatment of both left and right hemilivers to treat the entire tumor burden have only one hemiliver embolized initially. They are brought back for the second embolization to complete the treatment cycle in 3 to 6 weeks depending on their recovery from the first embolization session. Embolization of individual segments on the right and left side of the liver during one embolization session is acceptable as long as the total volume of liver treated is not excessive. For example, one might treat a tumor in the anterior division of the right liver by embolizing segments 5 and 8 along with segment 4 on the left because tumors occurring on the anterior right/segment 4 border often have supply originating from both right and left hepatic arteries.

Noncontrast computed tomography (CT) performed immediately after HAE can provide information regarding the completeness of the embolization that is also prognostic. Contrast is retained within the treated tumor (Fig. 34.5) and both the degree of contrast retention as well as absence of defects in contrast retention have been shown to predict response to treatment.19 This information is also useful for suggesting the presence of other blood supply to the tumor that has not been treated and might be treated either at the same time or planned for the future. This is particularly helpful in the case of supply from extrahepatic vessels because supply from other hepatic arteries can often be demonstrated on the arteriogram obtained following embolization of the branch(es) already treated.

CLINICAL APPLICATIONS

Hepatic arterial embolization can be used to treat any hypervascular liver tumor, including but not limited to HCC and metastases from neuroendocrine tumor. HAE can also be used to treat hypervascular breast, sarcoma, renal cell carcinoma, thyroid, and ocular melanoma metastases. It has never been evaluated in gastrointestinal adenocarcinomas that are typically relatively hypovascular, such as hepatic metastases from colorectal cancer. It can be used in several different clinical settings. First, with curative intent, embolization can be used in an attempt to completely treat hepatic tumor in the case of patients with a single or a few small tumors limited to the liver without extrahepatic disease. Second, palliative embolization is useful for patients with pain related to tumor bulk or hormonal symptoms in the case of patients with metastatic neuroendocrine tumor, for instance. Finally, embolization can be used in an effort to prolong survival in patients who are not candidates for surgical resection, based on extent of disease or underlying cirrhosis/portal hypertension, by preventing progression of hepatic tumor. Embolization is indicated in patients with good performance status (0 or 1) and should not be employed in patients with significant comorbidities or underlying liver disease such as in patients with Child-Pugh class C cirrhosis/portal hypertension whose overall survival is typically dictated by their liver disease and not their tumor. Patients with poor liver function are also at increased risk for postembolization liver failure and other complications.20,21

After HAE, follow-up multiphase CT or magnetic resonance imaging is obtained approximately 4 weeks after complete treatment of the initial tumor burden. The treated tumor should be completely nonenhancing on the arterial phase scans (Fig. 34.6) and low density on the portal venous phase (Fig. 34.7); they are typically smaller as well. This scan serves as the new baseline scan for making future treatment decisions.

POTENTIAL COMPLICATIONS

Most patients develop postembolization syndrome (PES) consisting of some combination of pain, nausea/vomiting, and fever. PES symptoms usually abate over 24 to 48 hours. Elevation of liver enzymes is seen beginning the day after embolization but peaking on the second or third day posttreatment. The liver enzymes usually return to baseline with 1 month. Complications related to angiography such as vessel dissection and groin hematoma can also occur. Liver abscess is a rare but serious complication of HAE that is seen most commonly in patients with previous bilioenteric anastomosis (BEA) or any reason for compromise of the sphincter of Oddi, including preexisting biliary drainage catheter, biliary stent, or previous sphincterotomy. In a review of over 2,000 embolization procedures in 971 patients, Mezhir22found that 33% (11/34) of patients with BEA developed a postprocedure liver abscess compared to 1 of 916 (0.1%) patients with an intact sphincter. Liver abscess can occur in this setting despite a prolonged course of antibiotics in the periprocedural period. Postembolization liver failure or death are exceedingly rare, occurring in less than 1% of patients with proper patient selection.21 Embolization of small particles into the pulmonary circulation via shunting to systemic veins, one of the most dreaded complications of this procedure given its often fatal outcome, can occur23 and is best avoided by upsizing microsphere size in tumors larger than 10 cm, large tumors high in the dome adjacent to the hemidiaphragm, tumors invading the hepatic veins, or with unexpected early hepatic vein drainage demonstrated angiographically as outlined previously.

TIPS AND TRICKS

Objective or Problem

Tip or Trick

How to avoid surprises during embolization procedure?

Carefully review recent multiphase CT/MRI prior to procedure: identify target, know the arterial anatomy, and anticipate nonhepatic arterial supply.

Completely evaluate arterial anatomy angiographically, even with excellent pretreatment multiphase imaging.

Always begin with celiac and superior mesenteric artery angiography.

Do not underestimate underlying liver disease.

Get good portal phase angiographic imaging for direction of portal flow; cannot get that from CT.

Difficult celiac?

Start with Cobra (Angiodynamics, Latham, New York), then use Simmons 2 catheter, which will usually reach into common hepatic artery so at least seated in “parent” vessel.

Where to reform Simmons?

Avoid cerebrovascular accident territory; use 65 cm Simmons 2 catheter and reform over aortic bifurcation.

How to save contrast?

Use road mapping or blend mask.

How not to mix up syringes used for embolization with those used for control angiography?

Use color-coded syringes for embolization material, flush, and contrast during embolization (Fig. 34.8).

How to maximize intratumoral embolization?

Begin with smallest microspheres and move to next largest size after using 10 mL in any vessel.

Avoid dissection of or bubbles in vessel to be embolized.

Always advance catheters over guidewires and make sure you have blood return or backbleeding from catheter before you inject.

Persistent antegrade flow after tumor vessels occluded?

Stabilize embolization end point (get to “stasis”) using 100-μm PVA.

Continued flow into embolization “sink”?

Get to stasis with 100-μm PVA.

How to mix PVA?

Add 10-mL full-strength contrast to vial (1 mL) of PVA; shake/mix. May need to add more contrast for small inner diameter coaxial catheters to avoid catheter occlusion.

Tumor in “watershed area” like 8/4 border?

Even if tumor appears to be totally supplied by one vessel, always study the “other” vessel before quitting.

Tumor located adjacent to hemidiaphragm or high in dome of liver?

Always study phrenic artery selectively.

Do you see prominent intercostal vessels on preembolization arterial phase CT?

Study those intercostals.

How do you find phrenic artery?

Find on preprocedure arterial phase imaging. Phrenic can arise from left of aorta or be reconstituted from renal or adrenal arteries.

REFERENCES

 1. Breedis C, Young G. The blood supply of neoplasms in the liver. Am J Pathol. 1954;30(5):969–977.

 2. Charnsangavej C, Chuang VP, Wallace S, et al. Work in progress: transcatheter management of primary carcinoma of the liver. Radiology. 1983;147(1):51–55.

 3. Yamada R, Sato M, Kawabata M, et al. Hepatic artery embolization in 120 patients with unresectable hepatoma. Radiology. 1983;148(2):397–401.

 4. Takayasu K, Shima Y, Muramatsu Y, et al. Hepatocellular carcinoma: treatment with intraarterial iodized oil with and without chemotherapeutic agents. Radiology. 1987;163(2):345–351.

 5. Konno T. Targeting cancer chemotherapeutic agents by use of lipiodol contrast medium. Cancer. 1990;66(9):1897–1903.

 6. Egawa H, Maki A, Mori K, et al. Effects of intra-arterial chemotherapy with a new lipophilic anticancer agent, estradiol-chlorambucil (KM2210), dissolved in lipiodol on experimental liver tumor in rats. J Surg Oncol. 1990;44(2):109–114.

 7. Raoul JL, Heresbach D, Bretagne JF, et al. Chemoembolization of hepatocellular carcinomas. A study of the biodistribution and pharmacokinetics of doxorubicin. Cancer. 1992;70(3):585–590.

 8. Johnson PJ, Kalayci C, Dobbs N, et al. Pharmacokinetics and toxicity of intraarterial adriamycin for hepatocellular carcinoma: effect of coadministration of lipiodol. J Hepatol. 1991;13(1):120–127.

 9. Simonetti RG, Liberati A, Angiolini C, et al. Treatment of hepatocellular carcinoma: a systematic review of randomized controlled trials. Ann Oncol. 1997;8(2):117–136.

10. Lo CM, Ngan H, Tso WK, et al. Randomized controlled trial of transarterial lipiodol chemoembolization for unresectable hepatocellular carcinoma. Hepatology. 2002;35(5):1164–1171.

11. Llovet JM, Real MI, Montana X, et al. Arterial embolisation or chemoembolisation versus symptomatic treatment in patients with unresectable hepatocellular carcinoma: a randomised controlled trial. Lancet. 2002;359(9319):1734–1739.

12. Camma C, Schepis F, Orlando A, et al. Transarterial chemoembolization for unresectable hepatocellular carcinoma: meta-analysis of randomized controlled trials. Radiology. 2002;224(1):47–54.

13. Lee KH, Liapi E, Vossen JA, et al. Distribution of iron oxide-containing Embosphere particles after transcatheter arterial embolization in an animal model of liver cancer: evaluation with MR imaging and implication for therapy. J Vasc Interv Radiol. 2008;19(10):1490–1496.

14. Maluccio MA, Covey AM, Porat LB, et al. Transcatheter arterial embolization with only particles for the treatment of unresectable hepatocellular carcinoma. J Vasc Interv Radiol. 2008;19(6):862–869.

15. Erinjeri JP, Salhab HM, Covey AM, et al. Arterial patency after repeated hepatic artery bland particle embolization. J Vasc Interv Radiol. 2010;21(4):522–526.

16. Brown KT. Fatal pulmonary complications after arterial embolization with 40-120- micro m tris-acryl gelatin microspheres. J Vasc Interv Radiol. 2004;15(2, pt 1):197–200.

17. Stampfl S, Bellemann N, Stampfl U, et al. Arterial distribution characteristics of Embozene particles and comparison with other spherical embolic agents in the porcine acute embolization model. J Vasc Interv Radiol. 2009;20(12):1597–1607.

18. Chrisman HB, Minocha J, Ryu RK, et al. Uterine artery embolization: a treatment option for symptomatic fibroids in postmenopausal women. J Vasc Interv Radiol. 2007;18(3):451–454.

19. Wang X, Erinjeri JP, Jia X, et al. Pattern of retained contrast on immediate postprocedure computed tomography (CT) after particle embolization of liver tumors predicts subsequent treatment response. Cardiovasc Intervent Radiol. 2013;36(4):1030–1038.

20. Chung JW, Park JH, Han JK, et al. Hepatic tumors: predisposing factors for complications of transcatheter oily chemoembolization. Radiology. 1996;198(1):33–40.

21. Raoul JL, Sangro B, Forner A, et al. Evolving strategies for the management of intermediate-stage hepatocellular carcinoma: available evidence and expert opinion on the use of transarterial chemoembolization. Cancer Treat Rev. 2011;37(3):212–220.

22. Mezhir JJ. Pyogenic abscess after hepatic artery embolization: a rare but potentially lethal complication. J Vasc Interv Radiol. 2011;22(2):177–182.

23. Brown KT. Re: Fatal pulmonary complications after arterial embolization with 40-120-microm tris-acryl gelatin microspheres. J Vasc Interv Radiol. 2004;15(8):887–888.