Page 1
1 First Multimodal Embolization particles visible on X-ray/CT and MRI S. H. Bartling, Dr.1,2 *, J. Budjan 1*, H. Biton 3, S. Haneder, Dr. 1, B. Kraenzlin, Dr.4, H. Michaely, PD Dr.1, S. Margel, Prof. Dr.3, S. Diehl, PD Dr.1, W. Semmler, Prof. Dr. Dr.
2, N. Gretz, Prof. Dr. 4,5, S.O. Schoenberg, Prof. Dr.1,5, M. Sadick, PD Dr.1 * The authors contributed equally
1 Institute of Clinical Radiology and Nuclear Medicine, University Medical Center Mannheim, Heidelberg University, Mannheim, Germany
2 Department of Medical Physics in Radiology, German Cancer Research Center, Heidelberger, Germany
3 Department of Chemistry, Bar-Ilan University, Ramat-Gan, Israel
4 Medical Research Center, University Hospital Mannheim, Heidelberg University, Mannheim, Germany
5 Institute for Medical Technology, Heidelberg University, Mannheim, Germany Communicating author: Dr. Soenke H. Bartling Institute of Clinical Radiology and Nuclear Medicine, University Medical Center Mannheim, Heidelberg University, Mannheim, Germany Theodor-Kutzer-Ufer 1-3 68167 Mannheim, Germany Phone: +49 621 383 2276 Fax: +49 621 383 3817 E-mail: soenke.bartling@umm.de
Page 2
2
The work was supported by a grant of the German Israel Foundation (61/2007).
Page 3
3 First Multimodal Embolization particles visible on X-ray/CT and MRI
Page 4
4
Abstract Objectives: Embolization therapy is gaining importance in the treatment of malignant and even more in benign lesions. Current embolization materials are not visible in imaging modalities, however, it is assumed that directly visible embolization material may provide several advantages ranging from particle shunt and reflux prevention to improved therapy control and follow-up assessment. X-ray visible as well as MRI visible embolization materials have been demonstrated in experiments. Here, we developed the first embolization materials that are visible in more than one imaging modality at once, namely MRI and X-ray/CT, characterized them and tested them in animal models. Materials and Methods: To reduce the chance of adverse reactions and to ease clinical approval materials have been used that are approved and being used on a routine-basis in diagnostic imaging. X-ray visible Iodine was combined with MRI visible Iron (Fe3+O4) in a macroparticle (diameter 40-200 !m). Its core - consisting of copolymerized monomer MAOETIB [2-methacryloyloxyethyl(2,3,5-triiodobenzoate)] - was coated with paramagnetic iron oxide nanoparticles (USPIO, 150 nm). After ex-vivo testing, including signal to noise measurements in CT and MRI (n=5), its ability to embolize tissue was tested in an established tumor embolization model in rabbits (n=6). Digital subtraction angiography (Integris, Philipps), CT (Definition, Siemens Healthcare Section, Erlangen, Germany) and MRI (3 Tesla Magnetom Tim Trio MRI, Siemens Healthcare Section, Forchheim) was performed before, during and after embolization. Imaging signal changes that could be attributed to embolization particles were assessed by visual inspection and rated on an ordinal scale by three radiologists from 1 to 3. Histology for ex vivo comparison was prepared. Results: Particles provided a sufficient image contrast on X-ray angiography, CT (SNR 13 ±2.5) MRI (SNR 35 ±1) in ex-vivo scans. Successful embolization of renal
Page 5
5
tissue was confirmed by dynamic angiography revealing at least partial perfusion stop in all kidneys. Signal changes that were attributed to particles residing within the kidney were found in all cases in all three imaging modalities. Localization distribution of particles corresponded well in all imaging modalities. Dynamic imaging during embolization provided real-time imaging of inflow of embolization particles within angiography, CT and MRI. Histology allowed a direct visualization of the residing particles as well as associated thrombosis in renal arteries. Visual assessment of embolization particle presence likelihood received full rating scores (153/153) after embolization. Conclusions: For the first time, multimodal visible embolization particles were produced, characterized and tested in-vivo in an animal model. Once introduced in clinical radiology, multimodal embolization particles may provide advantages for prevention of carry-over, may allow direct visualization in a multimodal angiography environment at all times and may improve follow-up examinations by the combination of complementary characteristics of CT and MRI. Dose savings could be realized. Finally, these advantages could contribute to further refine and improve embolization therapy. Additionally, new possibilities in basic embolization research may be opened. Keywords: Embolization, CT, MRI, multimodal
Page 6
6
Introduction In interventional radiology embolization therapy plays an important role (1). By means of minimal-invasive, catheter based procedures vessels related to pathologic conditions are occluded (1). Embolization therapy is used as a therapeutic option for bleeding (2-6), tumorous processes (7-14) as well as vascular malformation (15, 16). Embolization therapy is well established within multidisciplinary treatment concepts of malignant tumors. In the treatment of hepatocellular or renal carcinomas embolization therapy is used as an approach for tumor down staging to facilitate operations, bridging to liver transplantation, as well as a therapy option in palliative settings (17). Beside this, the role of embolization in benign, tumorous lesions such as uterine fibroids is rapidly growing and has shown to be superior in comparison to surgery in certain cases (18). Embolization therapy is performed in an angiography X-ray based setup, which allows guidance of catheters as well as control of the embolization process (19). State-of-the-art angiography systems provide also a CT scanning mode for intraprocedural CT imaging (20). A tendency exists to use MRI for interventional radiology, allowing a radiation free treatment which is of special importance in benign condition (21-23). Unfortunately, MRI is currently limited as a stand-alone system for interventional radiology and therefore multimodal setups as well as hybrid systems have been introduced (1, 19, 24, 25). Many different embolization materials have been described (1). For tumor embolization particles consisting of polyvinyl alcohol (PVA) (26) or microspheres loaded with chemotherapeutic agents (27) are considered state of the art in clinical treatment concepts. No embolization particles in clinical use are visible within an imaging modality. X-ray visible embolization materials at an experimental stage have been described (28-31) as well as particles directly visible in MRI (32). Directly visible
Page 7
7
embolization materials may provide advantages over non-visible embolization particles for embolization control and therapy follow-up (28, 32-34). Therefore, the motivation of our research was to improve the concept of directly visible embolization materials by developing the first multi-modal embolization material that is visible in more than one imaging modality, namely X-ray/CT and MRI. Its ability to provide image contrast in angiography, CT and MRI was tested in-vitro and in-vivo, further its ability to embolize tissue was demonstrated in an established rabbit tumor embolization model.
Page 8
8
Materials and Methods Multimodal embolization particles Multimodal embolization particles (Fig. 1) consist of an X-ray visible, Iodine containing core and a MRI visible, ultra small paramagnetic iron oxide (USPIO)-based coating. The core was synthesized by suspension homopolymerization of 2-methacryloyloxyethyl (2,3,5-triiodobenzoate) (MAOETIB) together with a low concentration of glycidyl methacrylate (GMA). This resulted in a long polymer P(MAOETIB-GMA) with iodinated, aromatic side chains (35). This core was coated with Fe3+O4 by nucleation and controlled growth mechanism of magnetic iron oxide nanoparticles on its surface, resulting in magnetic Fe3O4/P(MAOETIB-GMA) particles (Fig. 2). For embolization, particles with diameters ranging from 40 to 200 !m were selected by multiple sieving steps. Adhesion effects due to the electric charge of the particles made it necessary to add 25 g/l rabbit albumin (Sigma Aldrich, Germany) to distilled water to disperse the particles and to prevent the particles from sticking to syringe and catheter walls and allow them to float freely in suspension. In-vitro imaging characterization A Petri dish was filled with a 2% agarose solution (Sigma Aldrich, Germany) to a height of 1.5 cm. After 30 min of cooling, particles were placed on the surface and then covered with a 1 cm layer of agar. The agar phantom was imaged within angiography (Tab 1, No. 1), CT (Tab 1, No. 2) and MRI (Tab 1, No. 4). Embolization animal model A standard, well-established tumor embolization animal model has been used (36, 37) to test the particles in-vivo. Here, a rabbit kidney – representing a tumor - was
Page 9
9
embolized (37). All animal experiments were approved by the responsible local authorities. Six New Zealand White rabbits with an average weight of 3.4 kg (± 0.8 kg) were used. Animal anesthesia was initiated and maintained using a combination of Diazepam (2.5 g/kg s.c.), Xylazinhydrochloride (Rompun®, 10mg/kg i.m.) and Ketaminhydrochloride (Ketanest® 70mg/kg i.m.). After additional local anesthesia using Mepicavaine, the right femoral artery was surgically exposed and a 16G standard venous cannula was inserted. A hemostasis valve (Terumo hemostasis valve II, Terumo, Japan) was connected to the cannula. Heparin (1000 IU) was applied to prevent blood clot forming. Using a 0.018'' microcatheter (MicroFerret-18, Cook Medical, USA) one kidney artery was catheterized, a bend micro guide wire (0.016'', Radiofocus Guidewire GT, Terumo, Japan) was used if necessary. In three cases the right kidney artery was probed, the downward oriented upbranching of the right kidney artery from the aorta facilitated catheterization of this side. In three cases the left kidney was selected, because it was assumed that the left kidney might be less prone to movement artifacts during MR imaging. Correct catheter positioning and normal kidney perfusion was confirmed by injecting 1 ml of iodinated contrast media during angiographic series acquisition (Tab 1, No. 1). Study design and imaging In all animals an angiographic times series (Tab. 1, No. 1), CT (Tab. 1, No. 2, 3) and MRI (Tab. 1, No. 5, 6, 7) was performed before and after embolization of the kidney. To demonstrate real-time visibility of the embolization process, additionally continuous imaging during embolization itself was performed in each modality in two animals (Tab 2). In angiography a time series (Tab. 1, No. 1), in CT a continuously updated slap through the midsection of the kidneys were acquired during application
Page 10
10
of the particles (Tab. 1, No. 3) and in MRI a fast, repetitive, coronal EPI (echo planar imaging) sequence (Tab. 1, No. 8) was used. Histology After imaging, the animals were sacrificed and organs were taken and prepared for histology. Both kidneys were cut into eight 0.5 cm horizontal slices each. In addition, control slices of lung, liver and brain were taken. The samples were fixed in 4% formalin, dehydrated using Ethanol and Xylene and embedded in Paraffin. A microtome (Leica RM 2165, Leica, Germany) was used to obtain 3 !m slices. After Haematoxylin and Eosin staining, the slides were examined using bright field microscopy (DMREHC Microscope, Leica, Germany). Data analysis CT signal to noise ratio (SNR) was calculated using a two region of interest (ROI) approach. Two circular ROIs were placed in one slice, one ROI containing only particles, the other only plain agar. SNR was calculated using following formula:
!
SNR =
(mean value of signal in particle ROI) " (mean value of signal in agar ROI)
(standard deviation of signal in agar ROI)
SNR was calculated in 5 different sites for single particles as well as for particle clusters. For SNR measurements in MRI, summation and difference images were calculated out of two identical consecutively acquired T2 weighted scans (Tab. 1, No. 4). In accordance to (38), corresponding ROIs containing particle signal in both summation and difference images were then taken to calculate the SNR of particles in 5 different representative sites using following formula:
!
SNR =
1
2
"
(mean value of signal in sum image ROI)
(mean value of signal in difference image ROI)
Page 11
11
Imaging analysis of the in-vivo studies was performed by three experienced radiologists (one senior resident, two attendings both specialized in interventional radiology). The representation of kidneys in scans acquired before embolization was compared with those acquired after embolization. A visual analysis of the likelihood of embolization particles presence in all three modalities using a three point scale (1: particles not present, 2: particles probably present, 3: particles definitively present) was performed. Focal changes, being hyperdense in CT, dark/hypointense in MRI and dense in X-ray (in comparison to kidney parenchyma) were attributed to residing embolization particles. Since the modalities do not only image the kidney, the remaining organs within the field-of-view were also screened for signal changes through embolization. Likelihoods of embolization effects being present were assessed for macroscopy (1: no changes, 2: probably color changes, 3: color changes definitively present) as well as histology (thrombus, residing particles, (1: no changes, 2: probably thrombus/particles visible, 3: thrombus/particles definitively present)) by a veterinarian with several years of experience in kidney research. For data analysis sum scores over all raters, imaging modalities were calculated and compared.
Page 12
12
Results In vitro imaging characterization Within all three imaging modalities, the particles provided a clear contrast to the surrounding agar (Fig. 3). On CT and MRI, signal from particles was only found in slices representing the agar layer the particles were embedded in. Spatial distribution of signal changes matched in all three imaging modalities as seen in Figure 3. In CT, single particles showed maximal CT values of 206 ± 30 HU, the density within clusters of particles was 1340 ± 136 HU. The SNR of a single particle was 13 ± 2.5. SNR of particle clusters in CT was 105 ± 11.8. SNR in MR was 35 ± 10. Embolization In all cases renal arteries could be successfully catheterized as confirmed by contrast media injection (Fig. 4 a). The embolization procedure could be visually observed by the performing radiologist without adding radiopaque agents. Embolization particle injection was successful and at first without relevant resistance, yet resistance and manual injection pressure increased during the application process. The embolization was confirmed by patchy (n=2) to complete perfusion defects (n=4) (Fig. 4 b) in the post-embolization kidneys. Image comparison before and after embolization Imaging before, after and during embolization of six rabbits was mostly successful as described in Tab. 2. In only one case X-ray angiography imaging after embolization could not be accomplished due to death of the animal after embolization due to anesthesia complications. Image comparison between scans that were performed before and after injection of embolization revealed differences that were attributed to the embolization particles:
Page 13
13
In all six cases, similar results were found: Within CT there were hyperdense, focal density changes present that were not visible in the kidney parenchyma before embolization (Fig. 5 a, b). Within MRI patchy, dark/hypointense, confluent areas in T2* weighting were visible that were not visible within kidneys before embolization (Fig. 5 c, d). Additional MR sequences also demonstrated signal changes: small focal signal drops in the kidney parenchyma in T2 weighted images (Fig 6 a, b) as well as bigger, round hypointense parenchymal areas in the EPI sequence (Fig 6 c, d). X-ray angiography showed focal, small hyperdense areas that were not visible in kidneys before embolization (Fig 5 e, f). The distribution of the particles was random within kidney parenchyma and varied between animals, in some cases almost all parts of the kidney showed signal changes (n=4), whereof in other cases only parts of the kidney parenchyma showed signal changes (n=2). The distribution of signal changes in all modalities corresponded well in all kidneys (Fig 7). Here, the embolization particles were mainly localized within the middle level and upper pole of the organ. The position of particles visibly correlated well in all three modalities. For all raters maximum scores were reached in all cases in all imaging modalities with regard to particles being visible (resulting in an overall sum score of 153/153 for particle visibility as compared to 0/162 before embolization). In only one case (rabbit number 5) a signal change was found that was not within kidney, here a hyperdensity on CT and corresponding signal void in MRI was detected within the psoas muscle on the same side as the embolized kidney, which was interpreted as a misplaced embolization particle, probably due to reflux beside the catheter. Beside this, no relevant signal changes from embolization or organ alterations outside the kidneys have been found. Dynamic imaging during embolization
Page 14
14
Imaging during embolization was possible in all cases, respectively twice using MRI, CT and X-ray angiography. The wash in of particles was monitored by repetitive, coronal echoplanar imaging. Here, continuously more and more dark areas caused by the susceptibility of the particles became visible. Similarly, continuously acquired CT scans showed an increasing amount of more and more hyperdense points in kidney medulla. Within angiography the particles became consecutively visible within the renal parenchyma. Macroscopy and histology after embolization Embolized kidneys showed an inhomogeneous surface whereof some areas were darker and some brighter. The transition was sharply delineated (Fig. 8 a). Histology revealed particle inside arteries at various sites in all embolized kidneys and particles were found within interlobar, arcuate and up to the interlobular arteries. Additionally thrombi consisting of erythrocytes and fibrin were found in all vessel regions including medullary vessels (Fig. 8 b,c). No particles were found in control kidneys as well as in lung, liver and brain. The sum scores for likelihood of embolization effect was 18/18 for macroscopy and 18/18 for microscopy.
Page 15
15
Discussion: For the first time embolization particles that are visible in more than one imaging modality have been tested in-vitro an in-vivo. The results show that embolization particles can be created that demonstrate sufficient contrast in CT, MRI and angiography so that visibility during and after application can be assured. The benefits of directly visible embolization particles are currently being discussed, while multimodality visible particles have not been discussed yet. Advantages of directly visible embolization particles are proposed in several publications (28, 32-34), however, it has to the best of our knowledge never been shown in a clinical setting and remains therefore an assumption. In the following direct visibility as well as multimodal visibility of embolization particles are discussed in the context of current embolization practice as well as potential future developments of embolization therapy. Directly detectable embolization particles are clearly advantageous in order to detect displaced particles. Without intrinsic imaging potential miscarried particles can only be found if unwanted embolization effects are present and large enough to be detectable by imaging modalities. The combination of CT and MRI visible particles would be clearly favorable for detection of carry-over. The high sensitivity of MRI for detecting changes in susceptibility caused by iron would allow the detection of even small amounts of displaced particles. On the other hand CT would provide complimentary value when particles have to be detected within the lung - a domain where CT is still superior to MRI. Obviously detecting displaced particles after intervention is of less interest than actual real-time prevention of particle carry-over. Carry-over mostly happens either because of particles shunting through tumor vessels or refluxing along the catheter track when tumor vessel flow ceases during the embolization process. Currently,
Page 16
Page 17
Page 18
Page 19
Page 20
Page 21
Page 22
Page 23
Page 24
Page 25
Page 26
Page 27
Page 28
Page 29
Page 30
Page 31
Page 32
Page 33
Page 34
Page 35
