Computed tomography and histopathological findings after embolization with inherently radiopaque 40μm-microspheres, standard 40μm-microspheres and iodized oil in a porcine liver model

Purpose The present study compared standard computed tomography (CT) and histopathological findings after endovascular embolization using a prototype of inherently radiopaque 40μm-microspheres with both standard 40μm-microspheres and iodized oil in a porcine liver model. Materials and methods Twelve pigs were divided into six study groups, of two pigs each. Four pigs were embolized with iodized oil alone and four with radiopaque microspheres; two animals in each group were sacrificed at 2 hours and two at 7 days. Two pigs were embolized with radiopaque microspheres and heparin and sacrificed at 7 days. Two pigs were embolized with standard microspheres and sacrificed at 2 hours. CT was performed before and after segmental embolization and before sacrifice at 7 days. The distribution of embolic agent, inflammatory response and tissue necrosis were assessed histopathologically. Results Radiopaque microspheres and iodized oil were visible on standard CT 2 hours and 7 days after embolization, showing qualitatively comparable arterial and parenchymal enhancement. Quantitatively, the enhancement was more intense for iodized oil. Standard microspheres, delivered without contrast, were not visible by imaging. Radiopaque and standard microspheres similarly occluded subsegmental and interlobular arteries and, to a lesser extent, sinusoids. Iodized oil resulted in the deepest penetration into sinusoids. Necrosis was always observed after embolization with microspheres, but never after embolization with iodized oil. The inflammatory response was mild to moderate for microspheres and moderate to severe for iodized oil. Conclusion Radiopaque 40μm-microspheres are visible on standard CT with qualitatively similar but quantitatively less intense enhancement compared to iodized oil, and with a tendency towards less of an inflammatory reaction than iodized oil. These microspheres also result in tissue necrosis, which was not observed after embolization with iodized oil. Both radiopaque and standard 40μm-microspheres are found within subsegmental and interlobar arteries, as well as in hepatic sinusoids.


Introduction
Minimally invasive transcatheter embolotherapy is an established treatment for benign and malignant hypervascular tumors. In the liver, transarterial embolization is an evidence-based first-line treatment for patients with intermediate stage hepatocellular carcinoma and a second-line treatment for patients with liver-dominant metastases [1,2]. A broad range of different embolic agents are commercially available for liver tumor embolization [2,3]. However, no consensus is available on the specific type of embolic material that is superior for the different tumor entities regarding oncological efficacy, complication rate, and biocompatibility [2,4,5].
An important issue for effective tumor embolization is the embolic agent being capable of reaching the tumor vasculature for complete devascularization and the chemotherapeutic agent being distributed homogeneously throughout the target volume [2,3,6]. Generally, the smaller the diameter of a particulate embolic agent, the deeper its penetration into the tumor and the greater the treatment effect [3,7]. In this context, embolization with the smallest available embolic agents, such as iodized oil (Lipiodol Ultra-Fluide; Guerbet, Roissy, France) and microspheres (e.g., Embozene 40 [CeloNova BioSciences, San Antonio, USA/Boston Scientific, Marlborough, USA]) bear the risk of complications, such as bile duct injury due to occlusion of the peribilliary plexus, non-target embolization due to arterio-venous shunts or occult arterial collateralization, and fulminant tumor necrosis with subsequent embolism of necrotic tissue [2,8,9]. In order to obtain a controllable level of distal arterial occlusion, narrow-size calibrated small microspheres with a defined and uniform diameter of 75μm are increasingly being used [6,7,10]. Embozene 40 and Tandem 40 microspheres (CeloNova BioSciences, San Antonio, USA/Boston Scientific, Marlborough, USA) are currently the smallest commercially available permanent microspheres that are clinically used for transarterial embolization [2,8,11].
The introduction of inherently radiopaque, x-ray-visible microspheres is another trend in liver tumor embolization [12][13][14][15][16]. Currently, the commercially available Embozene microspheres are not inherently visible applying standard x-ray or magnetic resonance imaging (MRI) techniques. In contrast, iodized oil, one of the first embolic agents used for transarterial embolization of liver tumors, features excellent x-ray visibility applying fluoroscopy, radiography, or computed tomography (CT), and can still be regarded as a standard for transarterial tumor embolization in the liver [17,18].
Animal models have been shown to be suitable for preclinical evaluation of the characteristics of embolic agents, such as visibility, vascular distribution, and effects on the treated tissue [7,[10][11][12][13].
The aim of the present study was to compare a prototype of narrow-size calibrated radiopaque 40μm-microspheres with iodized oil and narrow-size calibrated non-radiopaque standard 40μm-microspheres as embolic agents with particularly deep penetration into the liver applying standard CT and histopathology.

Materials and methods
State Animal Care and Ethics Committee approval was obtained (Regional Council Karlsruhe, Germany; permit number: 35-9185.81/G-17/15). All experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals.

Study groups
Twelve animals were equally and randomly divided into six study groups (Table 1) according to type and suspension of embolic agent and survival time: iodized oil, acute setting (IO.a), iodized oil, subacute setting (IO.sa), radiopaque microspheres in suspension with saline, acute setting (RMS.a), radiopaque microspheres in suspension with saline and iodinated contrast material, subacute setting (RMS.sa), radiopaque microspheres in suspension with saline, iodinated contrast material and heparin, subacute setting (RMS-HEP.sa) and standard microspheres in suspension with saline, acute setting (SMS.a). Commercially available iodized oil was used as the pure embolic agent. The commercial product Embozene 40 (CeloNova BioSciences, San Antonio, USA/Boston Scientific, Marlborough, USA) was used as the standard microspheres, consisting of narrow-size calibrated microspheres with a hydrogel core and a Polyzene-F shell with a mean diameter of 40±10 μm [19,20]. There have been previous preclinical and clinical experiences with these microspheres, including long-term follow-up; thus, no special survival experiments (subacute setting) were performed for this embolic agent in order to reduce the number of animals required [8,11,19]. A prototype of Embozene 40 (CeloNova BioSciences, San Antonio, USA/Boston Scientific, Marlborough, USA) with the addition of barium sulfate as the radiopaque material was used as the radiopaque microspheres.
For the acute setting, a defined volume of radiopaque and standard microspheres suspended in saline, without addition of iodinated contrast material, was used to assess whether microspheres are inherently visible applying standard CT. A defined volume of iodized oil was used as a control. We selected this specific embolization technique for two reasons. First, 0.5 mL of embolic agent has been reported to be sufficient for the occlusion of subsegmental arteries in the same species [20,21]. Second, hyperdense areas diagnosed by CT correspond to enhancement induced by the embolic agent itself, and not by iodinated contrast material [20,21]. For the subacute setting, the embolization endpoint was defined as in clinical settings: stasis within the embolized liver segment but persistence of blood flow within the feeding central liver artery determined by angiography 5 minutes after embolization. For the subacute setting, iodinated contrast material was added to the microspheres as is standard in clinical practice [1][2][3]. The addition of iodinated contrast material to the microspheres results in a homogeneous suspension, which is effective for homogeneous delivery of the microspheres into the microcatheter and obtaining optimal rheological properties for flow-directed embolization [3,22]. A suspension consisting of inherently visible microspheres, normal saline, and iodinated contrast material was used in the latest studies on radiopaque microspheres [14,23]. Accordingly, the visibility of the embolic agent may be increased at 2 hours by iodinated contrast material, however during 7 days of follow-up, the iodinated contrast material washes out; thus the visibility at 7 days is expected to be attributed to the radiopaque microspheres exclusively [13,24,25]. The rationale for adding heparin (study group RMS-HEP.sa) was to reduce the formation of coagulation thrombus within the target vessel during embolization, i.e., to obtain arterial occlusion that is mainly caused by microspheres, and not by an additional coagulation thrombus component, with a potential for recanalization over time, and therefore unpredictable embolization results.

Animal procedure and embolization technique
German Landrace pigs (body weight 33-37 kg) were used. The survival time was 2 hours (acute setting; n = 6) or 7 days (subacute setting; n = 6). Before the first intervention, the animals were fasted overnight. For the subacute setting, during the first 12 hours after intervention, animals only received water before food and liquid were made available ad libitum. The animals were followed, with daily assessment of activity, pain, dietary intake, and urine and fecal output. Analgesia was administered daily using carprofen (2 mg/kg; Rimadyl, Pfizer Animal Health, New York, USA). The animal housing environment (cage size of 3 m 2 per animal, individual housing) was maintained at a temperature of 18 C˚to 22˚C at 60% to 80% humidity with a 12-hour light-dark cycle. All interventions were performed under deep general anesthesia. Anesthesia was induced with an intravenous injection of ketamin (10 mg/kg; Ketamin, Medistar, Hannover, Germany), midazolam (0.4 mg/kg; Dormicum, Roche, Basel, Switzerland) and azaperone (6 mg/kg; Stresnil, Janssen Animal Health, Beerse, Belgium) and maintained with repetitive intravenous injections of ketamine and midazolam (dosed according to the effect). After cutdown, a 4F introducer sheath was inserted in the femoral artery. After positioning a 4F Cobra catheter in the main hepatic artery, a 2.8F microcatheter (Progreat; Terumo, Tokyo, Japan) was coaxially positioned in a segmental artery. In each animal, one segmental artery (either of the right or left liver lobe) was embolized. Pre-embolization angiograms were performed for treatment planning. Post-embolization angiograms were only performed for the subacute setting in order to confirm the embolization endpoint as described above. Before, 2 hours after or 2 hours and 7 days after embolization, a non-enhanced CT was applied with a standard scanner (Somatom Definition Flash; Siemens Medical Solutions, Forchheim, Germany). CT data were acquired according to clinical practice using an automated dose optimization software (CareDose and CarekV; Siemens Medical Solutions, Forchheim, Germany; dose setting for CTDI vol : 6.0 mGy) and a collimation of 64x0.6 mm. CT images were reconstructed as multi-planar images by applying an iterative reconstruction software (Admire 3/5; Siemens Medical Solutions, Forchheim, Germany; slice thickness: 3/2 mm, reconstruction kernel: I30-30). The animals were sacrificed with an intravenous injection of 20 mL potassium chloride 7.5%. After sacrifice livers were harvested and tissue from embolized and non-embolized lobes preserved in a solution of 4% formalin. In each liver, 10 tissue fractions of the embolized liver segments were paraffin-embedded. The embolized liver segments were located by the macroscopic signs of embolization (e.g. tissue necrosis) and/or by macroscopic landmarks, defined under consideration of the pre-embolization angiograms and the post-embolization CT images.

Histopathology
For the histopathological work-up, 4μm sections were obtained from the paraffin-embedded specimens. Hematoxylin and eosin (HE), Elastica van Gieson (EvG), and Masson Golder (MG) staining was performed. For direct visualization of iodized oil, Sudan III staining was performed, which specifically stains lipids [26].

Study goals
The visibility of the embolic agents on CT was assessed 2 hours (acute and subacute setting) and 7 days after embolization (subacute setting only) by D.F.V. and C.M.S. with 5 and 10 years of experience in radiology, respectively. The enhancement pattern was described qualitatively and compared quantitatively between the different study groups. The vascular distribution pattern of the embolic agents by applying histopathology was described. Inflammation, necrosis and other conspicuities of the embolized liver tissue were assessed.

Results
All animals were treated as planned. The six animals with the subacute setting ate and gained weight as expected. No complications occurred during the embolization procedure, and there were no adverse events or signs of toxicity during follow-up.

Standard CT findings
CT findings are summarized in Table 2. In all animals, compared to the pre-embolization CT, a slight diffuse increase in the density of the entire liver (<15 Hounsfield Units [HU]) was observed 2 hours after embolization, which is most likely attributed to the retention iodinated contrast material after diagnostic angiography. This slight diffuse increase in the density of the entire liver was fully reversible during the 7 days of follow-up. After embolization, the embolic agents, if visible, exhibited two different enhancement patterns: linear hyperdensities, corresponding to the embolic agent occluding subsegmental and interlobular arteries (defined as arterial enhancement) and patchy hyperdensities, corresponding to the embolic agent occluding interlobular arteries and sinusoids (defined as parenchymal enhancement). For IO.a (Fig  1), intense arterial and parenchymal enhancement was observed in both animals, though with different intensity. For IO.sa, 2 hours after embolization, the same type of arterial and parenchymal enhancement was observed as for IO.a, though with a lower intensity. During the 7 day follow-up, the intensity of both arterial and parenchymal enhancement diminished in both animals. For RMS.a (Fig 2), similar to IO.a, both types of enhancement were noted in both animals, though, with a lower intensity. In addition, the intensity of arterial and parenchymal enhancement was different in both animals in this group. Two hours after embolization, the arterial and parenchymal enhancement was more intense for RMS.sa than RMS.a. Furthermore, 2 hours after embolization, there was a higher intensity of arterial enhancement and a comparable intensity of parenchymal enhancement in RMS-HEP.sa compared to RMS.sa (Fig 2). Seven days after embolization, hypodensities were present in the embolized liver parenchyma only for RMS-HEP.sa, corresponding to areas of tissue necrosis. In summary, the type of enhancement was comparable throughout all study groups with iodized oil and radiopaque microspheres, but the arterial and parenchymal enhancement was less intense for radiopaque microspheres. Furthermore, the intensity of both arterial and parenchymal enhancement diminished over the 7 days of follow-up for all subacute animals. For SMS.a, there was neither arterial nor parenchymal enhancement 2 hours after embolization (Fig 3).

Fig 2. Radiopaque microspheres-standard CT findings. A RMS.a. Non-enhanced CT 2 hours after embolization.
Marked linear hyperdensity (blue asterisk) corresponding to an occluded subsegmental artery (arterial enhancement). The type of arterial enhancement was comparable to iodized oil. B RMS.sa. Non-enhanced CT 2 hours after embolization. Linear (blue asterisk) and patchy hyperdensities (red asterisk) corresponding to occluded subsegmental and interlobular arteries and sinusoids (arterial and parenchymal enhancement). Note the more intense arterial and parenchymal enhancement compared to A. The type of arterial and parenchymal enhancement was comparable with iodized oil. C RMS-HEP.sa. Non-enhanced CT 2 hours after embolization. Note the more intense arterial (blue asterisk) and parenchymal enhancement (red asterisk) than A and B. D RMS-HEP.sa. Non-enhanced CT 7 days after embolization. Note the weaker arterial (blue asterisk) and parenchymal (red asterisk) enhancement compared to C. Note the hypodense areas in the embolized liver parenchyma (blue arrowhead) representing tissue necrosis.

Histopathological findings
Histopathological findings are summarized in Table 3 and Figs 4-6. For RMS.a and SMS.a, but not for IO.a, dilated sinusoids with blood retention were observed. Iodized oil was detected within sinusoids as oval cavities with 10-30 μm in size, representing iodized oil preparation artefacts (HE, MG and EvG staining), and within subsegmental and interlobular arteries as black dots approximately 30-50 μm in size (Sudan III staining). Radiopaque and standard microspheres were detected as 30 to 50 μm round globuli within subsegmental and interlobular arteries and sinusoids. Neither the morphology nor arterial distribution pattern was qualitatively different between the two types of microspheres or between RMS.sa and RMS-HEP.sa. The arterial penetration depth was higher for iodized oil than microspheres, and the arterial distribution pattern was comparable between the subacute and acute setting for iodized oil and microspheres, respectively. An inflammatory response was observed for all types of embolic agents after 7 days. For IO.sa, there was mild to moderate leukocyte infiltration of the lobules and a moderate to severe leukocyte infiltration of the periportal fields, indicating moderate to severe parenchymal and periportal inflammation. Leukocyte infiltration was also observed for RMS.sa and RMS-HEP.sa, though with lower severity, corresponding to mild to moderate inflammation.  were observed with a hemorrhagic rim and necrosis of the periportal fields. No necrosis was detected for IO.a, IO.sa, RMS.a or SMS.a.

Discussion
In this experimental study, narrow-size calibrated, very small, inherently radiopaque microspheres were visible 2 hours and 7 days after embolization on standard CT images. In contrast, standard microspheres were not visible. Compared to iodized oil, the enhancement of radiopaque microspheres was qualitatively comparable, but with lower intensity. However, the vascular distribution on histopathology was comparable between radiopaque and standard microspheres, as the sinusoidal penetration depth was not as deep as with iodized oil. The degree of inflammatory response was mild to moderate for radiopaque microspheres and moderate to severe for iodized oil. Different suspensions of the radiopaque microspheres resulted in different degrees of necrosis. Radiopacity of embolic agents is potentially beneficial [6,27,28]. The intratumoral vascular distribution of the embolic agent determined by peri-and post-interventional imaging can be used for objective treatment control and to guide subsequent procedures, such as repeated embolization or thermal ablation procedures [6]. Furthermore, as already shown for iodized oil, non-enhancing areas may be an imaging biomarker of viable tumor or ineffective treatments [27]. Suk et al. demonstrated that contrast saturation in the tumor periphery can have a high positive predictive value for tumor response [28]. In the clinical setting, after embolization with standard microspheres, such imaging features can be achieved only with a low degree of reproducibility since the iodinated contrast agent is rapidly and unpredictably washed out [13,24,25]. Moreover, the transient intratumoral contrast agent saturation may not represent the exact location of the microspheres but mainly stasis of the iodinated contrast agent column [12,13,24,29]. Using inherently radiopaque microspheres, the true intra-and peritumoral microsphere distribution could be assessed objectively, and treatment endpoints could be optimized and standardized (e.g. by using microsphere quantification techniques for the embolized target volume) [12, 13,30]. Another important issue is the sensitivity of the imaging In 2016, the first inherently radiopaque microspheres were approved by the US Food and Drug Administration (LC Bead LUMI [BTG, London, Great Britain]) [23]. LC Bead LUMI is a product of the modification of commercially available and established non-inherently radiopaque microspheres (LC Bead [BTG, London, Great Britain]) by incorporating iodized oil into the microspheres [13,15]. Currently, the smallest commercially available inherently radiopaque microspheres are the above-mentioned LC Bead LUMI (version M1), which have a diameter of 75-150 μm. Even though these microspheres are considered to be small, theoretically they can be too large for deep penetration into the liver and thus ineffective for homogeneous occlusion of both the smallest tumor-feeding arteries and the vascular network of the tumor itself [7,32,33]. Therefore, inherently radiopaque microspheres with a smaller diameter and narrow-size calibration may be advantageous for effective tumor embolization [7,34].
In this study, we demonstrated that microspheres with a mean diameter of 40 μm and a size range of 30-50 μm (95% confidence interval) occlude subsegmental arteries but can also reach interlobular arteries, and even sinusoids. Iodized oil droplets with a diameter of 10-50 μm exhibit deeper penetration into the liver with occlusion of mainly interlobular Note the blood retention within the sinusoids (white arrowheads), which is a potential sign of embolization-related reactive hyperemia. There were no qualitative differences regarding the histopathological findings between the standard and the radiopaque microspheres for the acute setting.
https://doi.org/10.1371/journal.pone.0198911.g006 arteries and sinusoids. The higher intensity of arterial and parenchymal enhancement provided by iodized oil may be attributed to a greater amount and/or higher packing density of the radiopaque component, especially within interlobular arteries and sinusoids. For radiopaque microspheres, the suspension with the addition of iodinated contrast material (study group RMS.sa) exhibited a higher intensity of enhancement 2 hours after embolization, which can be explained by the iodinated contrast material itself and the temporary diffusion of iodinated contrast material into the microsphere core. On the other hand, both arterial and parenchymal enhancement diminishing during the 7 days of follow-up is most likely explained by a combination of the iodinated contrast material being washed out and microsphere redistributions.
The manufacturer of Embozene 40 warns not to include heparinized saline in the embolic mixture as this could lead to microsphere agglomeration. Microsphere agglomeration is unwanted as it may impede microsphere delivery through the catheter (e.g. occlusion) or result in uncontrolled embolization (e.g. non-target embolization or inhomogeneous microsphere distribution in the target volume). In this study, we did not observe any technical failures during embolization or adverse events after embolization for all study groups. However, we observed a higher degree of necrosis after microsphere embolization when heparin was added to the embolic agent (study group RMS-HEP.sa). In this context, it is important to mention the role of heparin in the reduction of coagulation thrombus as one mechanism of temporary arterial occlusion during embolization with microspheres. The higher degree of necrosis can be explained by more effective embolization by heparin-induced reduction of coagulation thrombus formation during embolization, increasing the packing density of the microspheres and reducing the potential for recanalization during follow-up. Embolization-induced necrosis was also observed by Sharma et al. after transarterial embolization without the addition of heparin in a pig liver model using LC BEAD LUMI microspheres 75-150 μm in size [14].
With respect to embolization-induced inflammation, a mild to moderate inflammatory response was observed for radiopaque microspheres, whereas a moderate to severe inflammatory response was observed for iodized oil. This finding has to be kept in mind because any inflammatory response can trigger an entire cytokine cascade, including neoangiogenesis, which may have implications for both tumor response and growth [35]. The mild to moderate inflammatory response for the radiopaque microspheres can be interpreted as acceptable biocompatibility [33,36,37].
This study has limitations. First, the number of animals was small and the specific embolization technique was designed for a proof of principle study. Therefore, a descriptive and comparative statistical analysis (e.g. measurement of HU) was consciously not performed. Second, there was no subacute setting (7 days survival) for the standard microsphere group. Third, 7 days was a short survival time for the subacute setting. Longer follow-up could provide more information on long-term visibility, tissue necrosis, and biocompatibility. Fourth, healthy pig livers were used in this study. The findings could be different for human liver tumors in regards to tumor biology, tumor microenvironment, and the complex tumor vascularity.

Conclusion
Radiopaque 40μm-microspheres are visible on standard CT with qualitatively similar but quantitatively less intense enhancement compared to iodized oil. These microspheres also demonstrate less of an inflammatory reaction than iodized oil, but result in tissue necrosis, which was not observed after embolization with iodized oil. Both radiopaque and standard 40μm-microspheres are found within subsegmental and interlobar arteries, as well as in hepatic sinusoids.