Collagen-Coated Acrylic Microspheres for Embolotherapy: In Vivo andIn Vitro Characteristics

Colin P. Derdeyn, Virgil B. Graves, M. Shahriar Salamat, and Alan Rappe

PURPOSE: To evaluate the in vivo and in vitro properties of collagen-coated acrylic microspheresand to compare them with polyvinyl alcohol (PVA) particles. METHODS: Samples of 100- to300-mm, 300- to 500-mm, 500- to 700-mm, and 700- to 900-mm collagen-coated acrylic micro-spheres and 200- to 300-mm PVA particles were suspended in solutions of 50% saline and 50%contrast material. The samples were evaluated for quantitative and qualitative microscopic char-acteristics (shape, size, deformability); injectability via standardized microcatheters; degree ofparticulate penetration in the pig rete mirabile; and reaction of tissue to the particles in 48-hour-and 4-week-old specimens. RESULTS: The acrylic microspheres were spherical and deformable.The sample of 100- to 300-mm microspheres (n 5 202) had a mean diameter of 210 mm (standarddeviation, 43 mm). Hub accumulation, particle aggregation, and catheter occlusion were notobserved with the microspheres (all sizes) but were noted with the PVA particles. The 200- to300-mm PVA particles formed aggregates in the proximal rete. The 100- to 300-mm microsphereswere found throughout the rete and beyond. Chronic transmural and perivascular inflammation wasobserved with the microspheres and the PVA particles. CONCLUSIONS: Particle aggregation andcatheter occlusion do not complicate the transcatheter delivery of collagen-coated acrylic micro-spheres as they do with PVA particles. For a given particle and vessel size, acrylic microspherespenetrate to a much greater extent than the PVA particles. Tissue reaction to acrylic microspheresand PVA particles is similar.

Index terms: Interventional materials, particles and microspheres; Animal studies

AJNR Am J Neuroradiol 18:647–653, April 1997

Therapeutic embolization is a well-estab-lished tool for the treatment of many vascularlesions. Polyvinyl alcohol (PVA) is perhaps themost widely used particulate embolic agent. Itsbiocompatibility is well documented, as is itsefficacy as a relatively permanent embolicagent (1–7). However, the irregular PVA parti-cles tend to aggregate and clump, which maylead to catheter occlusion (8, 9). These and

Received June 20, 1996; accepted after revision October 21.Presented at the annual meeting of the American Society of Neurora-

diology, Seattle, Wash, June 1996.Supported in part by Target Therapeutics Inc, for which Dr Graves is a

paid consultant.From the Departments of Radiology (C.P.D., V.B.G., A.R.) and Pathol-

ogy (M.S.S.), University of Wisconsin Hospitals and Clinics, Madison.Address reprint requests to Colin P. Derdeyn, MD, Mallinckrodt Institute

of Radiology, 510 S Kingshighway, St Louis, MO 63119.

AJNR 18:647–653, Apr 1997 0195-6108/97/1804–0647

© American Society of Neuroradiology

64

other properties may also limit the degree ofparticulate penetration into a vascular lesion (7).

New agents continue to be developed to re-solve these difficulties. Embosphere (GuerbetBiomedical; Louvres, France) is the trade namefor a commercially prepared, hydrophilic, non-resorbable, collagen-coated, acrylic micro-sphere currently used in Europe. The purpose ofthis project was to evaluate and characterize theproperties of these new injectable particles andto compare them with polyvinyl alcohol (PVA)particles.

Materials and MethodsCommercial samples of 100- to 300-mm, 300- to 500-

mm, 500- to 700-mm, and 700- to 900-mm Embospherespackaged in sterile saline (Guerbet Biomedical) and 200-to 300-mm PVA particles (Biodyne, El Cajon, Calif) wereobtained. Each vial of Embospheres and PVA particleswas suspended in 40 to 50 mL of a 50% saline and 50%contrast (Omnipaque 300, Nycomed; New York, NY) so-

7

648 DERDEYN AJNR: 18, April 1997

lution, similar to that used in our clinical practice. Subjec-tively dilute solutions were used, although the concentra-tion of particles in solution was not measured. Fourseparate experiments were pursued.

1) To assess particle shape, consistency, and deform-ability, we first qualitatively evaluated a sample of 100- to300-mm Embospheres in solution under a light micro-scope (310 magnification). The range of particle diame-ters was then measured in a random sample by usinga calibrated ruler. The median, maximum, and mini-mum diameters and the standard deviation were calcu-lated.

2) The injectability of the particles through a 3F/1.5Fcatheter (Balt; Montmorency, France) was evaluated byusing a standard method of infusion. All four availablesizes of Embospheres were evaluated as well as the 200-to 300-mm PVA particles for purposes of comparison. Thecatheter was passed through a serpiginous tubing systemto simulate the cerebral vasculature. Suspended particleswere then hand-injected with a 3-mL syringe. To see theinjected particles best, we placed the catheter tip in a clearcollection bowl on top of a dark background with indirectlighting. The degree of resistance on injection was subjec-tively graded on a scale of 0 to 5, where saline injection is0 and occlusion is 5. The catheter was observed for accu-mulation of particles at the hub. In addition, the passage ofsuspended 100- to 300-mm and 700- to 900-mm Embo-spheres through the clear distal portion of Tracker-18 andTracker-10 catheters (Target Therapeutics; Fremont,Calif) was observed under a microscope.

3) The depth of penetration of PVA particles and Em-bospheres was evaluated in the pig rete mirabile. Retiawere unilaterally embolized to occlusion with 200-300-mm PVA particles (n 5 3) or 100- to 300-mm Embo-spheres (n 5 4) in seven juvenile (20 to 25 kg) Chester-White swine. The contralateral rete served as both thehistologic and arteriographic control. The experimentalprotocol was approved by our institution’s Animal CareCommittee.

The pigs were anesthetized before intubation (Telazol[tiletamine/zolazepam] 7 mg/kg, xylazine 2.2 mg/kg, andatropine 400 mg, intramuscularly). General anesthesiaduring the endovascular procedures was maintained withhalothane. A 6F vascular sheath was percutaneouslyplaced into a common femoral artery via standardSeldinger technique. A 6F guiding catheter (Fasguide;Target Therapeutics) was fluoroscopically placed intoeach common carotid artery for bilateral baseline arterio-grams. The ascending pharyngeal artery, which suppliesthe rete and emerges distally as the internal carotid artery,was selectively catheterized over a guidewire with the useof a microcatheter (Tracker 18; Target Therapeutics)placed through the guiding catheter.

After angiographic confirmation of catheter placement,the rete was embolized to occlusion with the 200- to300-mm PVA particles (n 5 3) or the 100- to 300-mmEmbospheres (n 5 4) for the acute experiments. Comple-tion arteriograms were obtained bilaterally after the micro-catheter was removed. The pigs were killed immediately

after angiography was concluded by means of an intrave-nously administered barbiturate overdose.

The skull base and brain were then dissected out andallowed to fix in a 10% formalin solution for 7 to 10 days.The entire rete (both sides and the midline connections)was then dissected from the surrounding bone and softtissues (rete dissection guide provided by John Roberts,Department of Radiological Sciences, University of Cali-fornia at Los Angeles). After further fixation, the retia weresectioned into 2-mm coronal sections from posterior (atthe junction of the ascending pharyngeal artery and therete) to anterior. The average rete was divided into sixcoronal sections (range, six to eight). The sections werethen embedded in paraffin, sliced in thicknesses of 5 mm,and stained with standard hematoxylin-eosin.

The degree of particulate penetration into the rete wasquantified in the sequential coronal sections from the sixacute preparations using the 100- to 300-mm Embo-spheres (four pigs) and the 200- to 300-mm PVA particles(three pigs). Two observers reviewed the slides from se-quential sections for the presence of PVA or Embosphereparticles. The depth of penetration into the rete was re-corded. An average rete arteriole diameter was establishedby measuring a glutaraldehyde-fixed rete embedded inplastic.

4) The tissue response of the pig rete mirabile to em-bolization was assessed microscopically in 48-hour- and4-week-old specimens from four additional juvenile Ches-ter-White swine. Embolization was performed with 700- to900-mm Embospheres (n 5 2) or 200- to 300-mm PVAparticles (n 5 2) in an identical manner as that describedabove for the acute studies. Pigs were then allowed torecover from anesthesia and returned to the veterinaryfacility. No other procedures were performed on thesepigs, either before or after embolization. The chronic Em-bosphere preparations were done with 700- to 900-mmparticles after it became apparent from the acute studiesthat the 100- to 300-mm particles easily passed throughthe rete into the brain.

At 48 hours and at 4 weeks, one pig embolized with PVAparticles and one with Embospheres were anesthetizedand catheterized with a diagnostic catheter in a manneridentical to that done for preembolization angiography.Bilateral common carotid artery arteriograms were ob-tained and the pigs were subsequently killed.

The skull base was dissected and allowed to fix beforedissection of the rete, in a manner identical to that used forthe acute preparations. Standard hematoxylin-eosin stainswere used. Tissue reaction to the particles was assessedmicroscopically in the 48-hour and 4-week specimens bya neuropathologist blinded to the age of the specimen.

Results

Particle Evaluation

Embospheres were uniformly spherical inshape and smooth in appearance. They were

AJNR: 18, April 1997 ACRYLIC MICROSPHERES 649

easily deformed by a needle and sometimesindented by adjacent particles (Fig 1). No frag-ments were observed before or after suspensionand manipulation.

A range of sizes was observed for the 100- to300-mm Embosphere particles (Fig 2). Themean particle diameter was 210 mm (standarddeviation, 43 mm) and the minimum and max-

Fig 2. Histogram of Embospheres with diameters of 100 to300 mm.

Fig 1. Embospheres (100 to 300 mm) in saline suspension.The particles are uniformly spherical, with the exception of oneparticle, which appears to be indented by an adjacent sphere(arrowhead). Note the range of particle sizes (original magnifi-cation 310).

imum diameters were 120 mm and 310 mm,respectively (n 5 202).

Particle Injectability

Sizes up to and including the 700- to 900-mmEmbosphere particles suspended well in a so-lution of 50% saline 50% contrast material andinjected easily through the 3F/1.5F microcath-eter (minimum inner diameter, 280 mm). Theease of injection was graded as 1 on a scale of5 (5 for occlusion) for the 100- to 300-mm par-ticles; 2 for the 300- to 500-mm particles; and 3for both the 500- to 700-mm and 700- to900-mm particles. Hub accumulation, particleaggregation, or catheter occlusion were not ob-served. The 200- to 300-mm PVA particles ac-cumulated in the catheter hub after injection of1 mL and occluded the catheter after 4 mL.

Observation of the 100- to 300-mm and the700- to 900-mm Embospheres within theTracker-18 (minimum inner diameter, approx-imately 530 mm) and the Tracker-10 (minimuminner diameter, approximately 360 mm) cathe-ters revealed deformability of the 700- to900-mm particles within the smaller catheter(Fig 3). No adherence to the catheter wall wasobserved.

In Vivo Rete Penetration

A review of the sequential coronal sectionsfrom embolized pig rete mirabile showed muchdeeper penetration of the 100- to 300-mm Em-bospheres than of the 200- to 300-mm PVAparticles (histogram, Fig 4). Many 200- to

Fig 3. Photograph of a 700- to 900-mm Embosphere particlein the distal clear segment of a Tracker-10 catheter (inner diam-eter, approximately 360 mm) shows the deformability of the Em-bospheres (original magnification 34).

650 DERDEYN AJNR: 18, April 1997

300-mm PVA particles were found in aggregatesin the proximal portion of the rete (Fig 5). NoPVA particles were observed in the distal half ofthe rete. The 100- to 300-mm Embosphereswere found in nearly equal density throughoutthe rete and many passed through into the brain(Fig 6). The average rete arteriole diametermeasured 173.3 mm on the glutaraldehyde-fixed specimen.

The arteriographic correlate of this differencein degree of penetration can be seen in Figure 7.

Fig 4. Histogram of depth of particulate penetration. The y-axis plots the percentage of coronal sections containing eitherPVA or Embosphere particles; the x-axis plots the depth of thesection by percent. In the first section, taken at the junction of theascending pharyngeal artery with the rete, all Embosphere prep-arations contained particles (100% of sections). Two of three PVApreparations contained particles on the first section (66% of sec-tions). No PVA particles were observed beyond 60% depth. Em-bospheres were observed in the most distal sections of the rete(100% penetration), at the origin of the internal carotid artery, inthree of four preparations (75% of sections).

Fig 5. Photomicrograph of arete acutely embolized with 200-to 300-mm PVA particles. Thissection was the first of seven coro-nal sections and was taken at thejunction of the ascending pharyn-geal artery with the rete. A largeartery is occluded in the upper leftcorner of the field. Severalclumped particles of PVA arepresent in the vessel (arrow-heads) with adjacent thrombus.No other particles were identifiedin the field (original magnification34).

Fig 6. Photomicrograph of a rete acutely embolized with 100- to 300-mm Embospheres. This section was the fourth of seven2-mm-thick coronal sections. Several arteries in the field contain the round particles (arrowheads) (original magnification 34).

When occlusion was accomplished with the200- to 300-mm PVA particles, good collateralcrossflow across the anterior rete was noted inall three cases (Fig 7A). After Embosphere oc-clusion, however, no filling across the anteriorrete was observed in three of four cases (Fig7B).

Tissue Reaction

Microscopically, we saw no significant differ-ence in tissue reaction between the PVA parti-cles and the Embospheres in either the 48-houror 4-week specimens. Histologic examinationof the retia harvested at 48 hours showed intra-vascular thrombosis in association with bothPVA and Embosphere particles and no vascularinflammation. Fibrin, platelets, and polymor-phonuclear leukocytes were identified withinthe thrombus. Mild intimal reaction adjacent tothe thrombus was present.

At 4 weeks, evidence of chronic transmuraland perivascular inflammation was observedwith both the Embospheres and the PVA parti-cles (Figs 8 and 9). Transmural inflammationincluded multinucleated giant cells, plasmacells, lymphocytes, and eosinophils in bothpreparations. Many Embospheres remainedpresent at 4 weeks, both intravascularly and inextravascular spaces. No PVA particles wereidentified in the 4-week PVA preparation.

Discussion

A variety of microspheres have been devel-oped for use as embolic agents. They have beenmade from various organic and inorganic ma-terials, including collagen (12), dextran (13),and trisacryl polymer impregnated with gelatin

AJNR: 18, April 1997 ACRYLIC MICROSPHERES 651

Fig 8. Photomicrograph of arete 4 weeks after embolizationwith 700- to 900-mm Embo-spheres. Inflammatory cells arepresent both within the wall of apatent vessel (single small arrow-head) and perivascularly. A parti-cle without an identifiable sur-rounding vessel wall is presentwithin this mass of inflammatorycells (double small arrowheads).Other thrombosed vessels may bepresent with the inflammatorymass. Intravascular thrombosisand perivascular inflammation oflesser magnitude are seen in thelower right corner of the field

(large arrowhead) (original magnification 310).Fig 9. Photomicrograph of a rete 4 weeks after embolization with 200- to 300-mm PVA particles. Transmural inflammatory cells

(arrowhead) are present in the wall of a patent vessel with extensive associated perivascular inflammation. Other normal vessels can beseen within the field (original magnification 310).

Fig 7. Rete anatomy and an-giographic correlate of depth ofparticulate penetration.

A, Subtracted left common ca-rotid artery angiogram (submentalvertex projection) obtained 48hours after complete embolizationof the right rete mirabile with 200-to 300-mm PVA particles. The leftascending pharyngeal artery(large arrowhead) is seen supply-ing the left rete. The fine mesh ofarterioles representing the rete isseen, with flow across midline tosupply both internal carotid arter-ies, which emerge rostrally fromthe distal rete (small arrowheads)(10, 11).

B, Right common carotid ar-tery angiogram obtained aftercomplete embolization of the leftrete with 100- to 300-mm Embo-spheres. Note the absence of col-lateral flow across midline (aster-isk), consistent with more distalpenetration of the rete with Embo-spheres.

(14). Microspheres have been subjectively ob-served to penetrate more deeply into the vas-culature of a lesion than PVA particles and maybe injected with less difficulty (13, 15). How-ever, no objective evaluation of these qualitiesor direct comparison with PVA particles hasbeen performed. And many of these micro-spheres are not widely used or commerciallyavailable. In this study we sought to compareobjectively the in vivo and in vitro behavior of acommercially available collagen-coated acrylicmicrosphere with PVA, a well-established and

frequently used embolic agent whose propertiesand effects are well documented.

Embospheres are uniformly spherical andsmooth in appearance, with a narrow range ofparticle diameters in the sample evaluated. Nosmall fragments were observed. PVA particlesare generally oblong or oval, with irregular,sharp, and angulated projections. After suspen-sion, a few small PVA fragments have beennoted (9).

The transcatheter delivery of Embosphereswas easier than with PVA particles. Hub accu-

652 DERDEYN AJNR: 18, April 1997

mulation and catheter occlusion were not ob-served with Embospheres, even with particlediameters larger than the minimum inner diam-eter of a flow-directed catheter. Particles aslarge as 700 to 900 mm could be injectedthrough a catheter with a minimum inner diam-eter of 280 mm. Hub accumulation and catheterocclusion occurred in this catheter with 200- to300-mm PVA particles.

Embospheres of similar size to PVA particlesshowed remarkable differences in the degree ofparticulate penetration. Whereas 200- to300-mm PVA particles often formed aggregatesin the distal ascending pharyngeal artery or inthe proximal rete, 100- to 300-mm Embo-spheres completely packed the rete and passedthrough it. These differences may be attributedto the lack of aggregation, the smooth and hy-drophilic surface, and the deformability of theEmbosphere particles. These same propertiesmay also account for the relative ease of injec-tion of the particles and the lack of hub accu-mulation and catheter occlusion observed withEmbospheres.

The degree of penetration of PVA particlesinto a vascular system may be affected bymany factors. The advertised size of a PVA par-ticle is primarily a function of its intermediateaxis, which is responsible for a given particle’sability to pass through a square sieve hole (9).The long axes of these irregular particles areoften much longer than their intermediate axes(9). One might expect that these particleswould align their long axis parallel to the flow ina vessel and become wedged at a point at whichthe vessel diameter is close to the intermediatediameter of the particle. However, Quisling et al(7) have shown that PVA particles in vivo tendto adhere to the walls of larger diameter vesselsrather than become wedged in a distal vessel.Our results support these observations. In addi-tion, PVA particles tend to clump and aggregatewithin the catheter hub. Similar accumulationsmay occur within the vessel at sites of particleadherence. For these reasons, the occlusionachieved with PVA may often occur more prox-imally than expected within a vessel for a givenparticle and given luminal diameter.

Our results provide laboratory evidence tosupport the empirical observations of Beaujeuxand coworkers (15) regarding the use of gela-tin-coated acrylic microspheres in patients withvascular lesions. They noted that larger andlarger microsphere particles were used succes-

sively during an embolization treatment to ob-tain complete occlusion. They postulated thatthis process was necessitated by the fact thatthe physicians’ initial choice of particle size wasbased on experience with PVA particles.

The difference in the degree of particulatepenetration has considerable clinical impor-tance. Greater penetration can be advanta-geous, leading to more effective transarterialembolization of many lesions. Presurgical em-bolization of meningiomas may constitute anideal application for Embospheres, as theremay be less potential for proximal vessel occlu-sion and subsequent need to develop collateraldural supply before surgery. However, lesionswith evidence of arteriovenous shunting willneed to be approached with great caution.Choosing a size of Embosphere solely on thebasis of experience with PVA particles couldresult in significant distal embolization (ie, pul-monary embolization) as a result of shuntingthrough the lesion. The response of the pulmo-nary circulation to Embosphere embolization isnot known. Until more is understood about theoptimal size of Embospheres for a given vesseldiameter and the consequences of pulmonaryembolization, the use of Embospheres for treat-ment of lesions with arteriovenous shunting,such as arteriovenous malformations, cannotbe recommended.

Both the Embospheres and PVA particles in-cited a moderate transmural inflammatory re-sponse. Turjman and coworkers (12) observeda similar pattern of transmural and adventitialinflammation in the pig rete at 3 to 5 weeks afterembolization with collagen microspheres. Thisinflammation may be due in part to effects ofvessel occlusion and/or accompanying throm-bus. Whether the collagen coating itself is re-sponsible for the inflammatory response is notknown. PVA may elicit a mild inflammatory re-action, particularly when the embolized vesselis completely occluded (4, 7, 16). When PVAparticles are not occlusive, they may not elicitany inflammatory response (7). No PVA parti-cles were identified in the 4-week specimen inour study, which may have been due to occlu-sion of the rete at the distal ascending pharyn-geal artery, an area that was not included in thedissected specimen.

In summary, collagen-coated acrylic micro-spheres may offer some advantages over PVAparticles for the embolization of vascular lesionswith no evidence of arteriovenous shunting.

AJNR: 18, April 1997 ACRYLIC MICROSPHERES 653

Embospheres are uniformly spherical andsmooth with a narrow range of particle diame-ters for a given size. Their transcatheter deliveryis easier and less complicated by particulateaggregation and catheter occlusion than arePVA particles. Embospheres can be injectedthrough small flow-directed catheters. Thechronic tissue reaction observed with Embo-spheres is similar to that seen with PVA. For agiven particle and vessel size, Embospherespenetrate to a much greater extent than PVA par-ticles. The major disadvantage of Embosphereslies in their potential to pass through a lesion tothe pulmonary circulation. Until more is knownabout their optimal size and their effect on thepulmonary circulation, great caution should beexercised in using Embospheres to treat lesionswith evidence of arteriovenous shunting.

References1. Tadavarthy SM, Moller JH, Amplatz K. Polyvinyl alcohol (Ivalon):

a new embolic material. AJR Am J Roentgenol 1975;125:609–616

2. Tadavarthy SM, Coleman CC, Hunter D, Casteneda-Zuniga WR,Amplatz K. Polyvinyl alcohol (Ivalon) as an embolic agent. SeminIntervent Radiol 1984;1:101–109

3. Lanman TH, Martin NA, Vinters HV. The pathology of encephalicarteriovenous malformations treated by prior embolotherapy.Neuroradiology 1988;30:1–10

4. White RI, Strandberg JV, Gross GS, Barth KH. Therapeutic embo-lization with long-term occluding agents and their effects on em-bolized tissues. Radiology 1977;125:677–687

5. Chuang VP, Tsai CC, Wright K, Wallace S, Charnsanavej C.

Experimental canine hepatic artery embolization with polyvinylalcohol foam particles. Radiology 1982;145:21–25

6. Scialfa G, Scotti G. Superselective injection of polyvinyl alcoholmicroemboli for the treatment of cerebral arteriovenous malfor-mations. AJNR Am J Neuroradiol 1985;6:957–960

7. Quisling RG, Mickle JP, Ballinger WB, Carver CC, Kaplan B.Histopathologic analysis of intraarterial polyvinyl alcohol micro-emboli on rat cerebral cortex. AJNR Am J Neuroradiol 1984;5:101–104

8. Jack CR, Forbes G, Dewanjee MK, Brown ML, Earnest F. Polyvinylalcohol sponge for embolotherapy: particle size and morphology.AJNR Am J Neuroradiol 1985;6:595–597

9. Derdeyn CP, Moran CJ, Cross DT III, Dietrich HH, Dacey RG Jr.Polyvinyl alcohol particle size and suspension characteristics.AJNR Am J Neuroradiol 1995;16:1335–1343

10. Daniel PM, Dawes JDK, Prichard MML. Studies of the carotid reteand its associated arteries. Philos Trans R Soc Lond [Biol] 1953;237B:173–208

11. Gillian LA. Blood supply to brains of ungulates with and without arete mirabile caroticum. J Comp Neurol 1974;153:275–290

12. Turjman F, Massoud TF, Vinters HV, et al. Collagen microbeads:experimental evaluation of an embolic agent in the rete mirabile ofthe swine. AJNR Am J Neuroradiol 1995;16:1031–36

13. Dion JE, Rankin RN, Vinuela F, Fox AJ, Wallace AC, Mervart M.Dextran microsphere embolization: experimental and clinical ex-perience with radiologic-pathologic correlation. Radiology 1986;160:717–721

14. Laurent A, Beaujeux R, Wassef M, Rufennacht D, Boschetti E,Merland J-J. Trisacryl gelatin microspheres for therapeutic em-bolization, I: development and in vitro evaluation. AJNR Am JNeuroradiol 1996;17:533–540

15. Beaujeux R, Laurent A, Wassef M, Casasco A, Gobin Y-P, AymardA, Rufennacht D, Merland J-J. Trisacryl gelatin microspheres fortherapeutic embolization, II: clinical evaluation in tumors and ar-teriovenous malformations. AJNR Am J Neuroradiol 1996;17:541–548

16. Casteneda-Zuniga WR, Sanchez R, Amplatz K. Experimental ob-servations on short- and long-term effects of arterial occlusionwith Ivalon. Radiology 1978;126:783–785