- Email: [email protected]
Development of a Large Animal Model of Cirrhosis and Portal Hypertension Using Hepatic Transarterial Embolization: A Study in Swine
LABORATORY INVESTIGATION
Development of a Large Animal Model of Cirrhosis and Portal Hypertension Using Hepatic Transarterial Embolization: A Study in Swine Rony Avritscher, MD, Kenneth C. Wright, PhD, Sanaz Javadi, MD, Rajesh Uthamanthil, DVM, Sanjay Gupta, MD, Mihai Gagea, DVSc, Roland L. Bassett, MS, Ravi Murthy, MD, Michael J. Wallace, MD, and David C. Madoff, MD
ABSTRACT Purpose: To develop a clinically relevant porcine model of liver cirrhosis with portal hypertension by means of hepatic transarterial embolization. Materials and Methods: Institutional animal care and use committee approval was obtained for all experiments. Pigs received transcatheter arterial infusion of a 3:1 mixture of iodized oil and ethanol into the hepatic artery in volumes of 16 mL in group 1 (n ⫽ 4), 28 mL in group 2 (n ⫽ 4), and 40 mL in group 3 (n ⫽ 4) with intent of bilobar distribution. Hepatic venous pressure gradient (HVPG) measurement, liver function tests, and volumetry were performed at baseline, at 2 weeks, and before necropsy. Results: Cirrhosis was successfully induced in three animals that received 16 mL of the embolic mixture and in all four animals that received 28 mL. The animals in the 40-mL group did not recover from the procedure and were euthanized within 48 h. Increases in HVPG after 6 – 8 weeks versus baseline reached statistical significance (P ⬍ .05). Correlation between degree of fibrosis and volume of embolic agent did not reach statistical significance, but there was a trend toward increased fibrosis in the 28-mL group compared with the 16-mL group. Conclusions: Transcatheter hepatic arterial embolization can be used to create a reliable and reproducible porcine model of liver cirrhosis and portal hypertension.
ABBREVIATION HVPG ⫽ hepatic venous pressure gradient
Cirrhosis is the most common nonneoplastic cause of death in patients with hepatobiliary and digestive disease, and is the ninth most frequent cause of death in the United States
From the Department of Diagnostic Radiology, Interventional Radiology Section (R.A., K.C.W., S.J., S.G., R.M., M.J.W., D.C.M.), and Departments of Veterinary Medicine (R.U.), Veterinary Pathology (M.G.), and Biostatistics (R.L.B.), University of Texas M. D. Anderson Cancer Center, Houston, Texas. Received February 23, 2011; final revision received April 19, 2011; accepted April 21, 2011. Address correspondence to D.C.M., Division of Interventional Radiology, New York Presbyterian Hospital/Weill Cornell Medical Center, 525 E. 68th St., P-518, New York, NY 10065; E-mail: dcm9006@med. cornell.edu Research for this study was supported in part by a grant from the John S. Dunn Research Foundation and by the National Institutes of Health through Support Grant CA016672 to M. D. Anderson Cancer Center. None of the authors have identified a conflict of interest. © SIR, 2011 J Vasc Interv Radiol 2011; 22:1329 –1334 DOI: 10.1016/j.jvir.2011.04.016
(1). Portal hypertension and liver cancer are feared complications of cirrhosis, with an associated 5-year mortality rate exceeding 50% (1–3). There are medications to slow or reverse progression of fibrosis, and interventional procedures that can mitigate the effects of portal hypertension. Animal models are needed to test and validate these approaches (4,5). The animal models of liver cirrhosis currently available are created by using a variety of hepatotoxins, such as carbon tetrachloride. The animals most commonly used for such models are mice, rats, and rabbits (6 – 8). Although these small and medium-sized laboratory animals are well suited for pharmacologic studies, large animal models are required to optimally test imaging tools and percutaneous interventions. In addition, such models can be used as large-animal hepatocellular carcinoma models, given that these tumors mostly arise in background cirrhosis. Suitable large-animal models of cirrhosis are currently lacking. The models currently described in the literature require several
1330 䡲 Swine Model of Cirrhosis and Portal Hypertension via Hepatic Embolization
repeated applications of the hepatotoxins and long induction periods (7,9). In addition, none of these models report concomitant development of portal hypertension. Pavcnik et al (10) attempted a percutaneous transhepatic technique to induce portal hypertension in swine in an experiment with the use of polyvinyl alcohol particles, but the portal pressures always returned to baseline levels at 1-week follow-up studies. The swine species is particularly advantageous in the creation of a large-animal model of cirrhosis because of the anatomic and physiologic similarities between the porcine and human liver (11,12). The use of hepatic arterial embolization with ethanol has previously been reported to produce severe hepatocellular damage by direct tissue toxicity, endothelial damage, and “sludging” of erythrocytes. The addition of iodized oil prolongs the contact time between the ethanol and the tissues and promotes a more homogenous distribution of the agents (13). The purpose of the current study was to develop a suitable and reproducible large animal model of liver cirrhosis.
MATERIALS AND METHODS
Avritscher et al 䡲 JVIR
Figure 1. Transcatheter arterial hepatic embolization using ethanol/iodized oil mixture. Close-up of an anteroposterior radiograph of the abdomen in a later phase of the procedure shows embolic mixture in portal vein branches (arrows). Metallic coils are noted in the gastroduodenal artery (arrowheads). The coils are used to prevent nontarget embolization.
Animal Care The institutional animal care and use committee approved the current study. Animals were maintained in facilities approved by the Association for Assessment and Accreditation of Laboratory Animal Care and in accordance with current United States Department of Agriculture, Department of Health and Human Services, and National Institutes of Health regulations and standards.
Hepatic Venous Pressure Measurement A balloon-occlusion catheter was used to measure free and wedged hepatic venous pressures, and the hepatic venous pressure gradient (HVPG) was calculated according to the standards previously described by Pagan et al (14). Measurements were obtained at baseline before embolization, 2 weeks after embolization, and at 6 – 8 weeks before necropsy.
Preprocedure Preparation Twelve domestic pigs weighing a mean of 41.4 kg (range, 29.8 – 46.4 kg) were sedated with an intramuscular injection of a solution containing ketamine hydrochloride (15 mg/ kg), acepromazine (0.15 mg/kg), and atropine sulfate (0.04 mg/kg). Anesthesia was then induced with isoflurane (5%) administered by a facemask. When anesthesia was established, an endotracheal tube was inserted and anesthesia was maintained with isoflurane (1.5%–3%) and oxygen (0.8 L/min). The animals were given intramuscular doses of 5 mg/kg of the antibiotic enrofloxacin (Baytril; Bayer Animal Health, Shawnee Mission, Kansas) before each procedure and once daily for 5 days after each procedure. At baseline, 2 weeks, and immediately before necropsy, all animals underwent nonenhanced computed tomography imaging (HiSpeed Advantage; GE Medical Systems, Milwaukee, Wisconsin) in 5-mm slices of the liver. Animals were weighed before embolization and before necropsy. All experimental and control procedures were performed by experienced radiologists familiar with porcine hepatic vascular anatomy.
Transarterial Procedures Access to the right common femoral artery was obtained by using Seldinger technique under sonographic guidance or via surgical cutdown. A single intravenous bolus of heparin (100 U/kg) was administered after vascular access was secure. A 5-F catheter (Sos-2; AngioDynamics, Queensbury, New York) was advanced over the wire into the aorta, and the celiac axis was selected. Celiac and hepatic digital subtraction angiography was accomplished by injecting iodinated radiographic contrast medium (meglumine diatrizoate; Nycomed, Zurich, Switzerland) through the catheter. A 3-F microcatheter (Tracker 325; Boston Scientific, Natick, Massachusetts) was then advanced coaxially into the common hepatic artery. Embolization of the gastroduodenal and right gastric arteries was then performed with metallic fibered coils. The microcatheter was advanced into the proper hepatic artery. Radiographic contrast medium was injected to confirm catheter position before embolization under fluoroscopy (Fig 1). Three groups of pigs were randomly assigned to re-
Volume 22 䡲 Number 9 䡲 September 䡲 2011
ceive 16 mL (n ⫽ 4), 28 mL (n ⫽ 4), or 40 mL (n ⫽ 4) of a 3:1 ratio mixture of iodized oil (Ethiodol; Savage Laboratories, Melville, New York) and absolute ethanol (13). Because rapid infusion of the mixture can lead to vascular spasm of the parent artery and premature termination of the procedure, the mixture was injected with an automated infusion pump under fluoroscopic monitoring. The infusion was carried out at 0.3 mL/s. The embolization endpoint was delivery of the entire volume of embolic agent. After completion of the procedure, the coaxial catheter system was removed, the right common femoral artery was ligated, and the incision was repaired with primary closure.
Liver Biopsies Two weeks after embolization, multiple liver biopsy samples were acquired under real-time sonographic guidance with an 18-gauge core needle biopsy. The biopsy samples were fixed by immersion in 10% neutral buffered formalin. The fixed tissues were embedded in paraffin blocks, stained with hematoxylin and eosin and Masson trichrome stain, and examined microscopically. Liver lesions were staged according to the METAVIR scoring system (15).
Liver Volumetry and Biochemical Analysis Liver volumes were calculated by using syngo InSpace software (Siemens, Erlangen, Germany). The external liver contours were manually segmented with the proprietary software and subsequently summated to obtain the estimated volumes. The middle hepatic vein was used as the dividing landmark to calculate the volumes of the right and left hepatic lobes. Right and left lobe volumes were analyzed individually. The mean baseline, 2-week, and final postprocedural right and left liver volumes were calculated in cubic centimeters and compared among the different treatment groups. In addition, at the same time points, all experimental animals had blood samples collected for aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, and lactate dehydrogenase.
Necropsy and Histopathologic Evaluation After final measurement of hepatic venous pressure, the animals were euthanized with an overdose of Beuthanasia-D (1 mL per 10 lbs; Schering-Plough, Union, New Jersey), and a complete necropsy was performed. At necropsy, the liver was examined grossly and multiple tissue samples were collected from each lobe of the liver and stained in the same manner described earlier. A veterinary pathologist examined the specimens under light microscopy. The overall presence of fibrosis throughout each lobe of the liver was graded on a scale from 0 to 4, with 0 representing no fibrosis, 1 representing fibrosis affecting less than 10% of the volume of the hepatic lobe, 2 representing fibrosis affecting 10%–20% of the lobe, 3 representing fibrosis affecting 21%–50% of the
1331
Table 1. Changes in Liver Volumes 2 and 8 Weeks after Embolization Interval Before embolization 2 Weeks 8 Weeks
Right Lobe (cm3) 452.2 62.8 ⫾ 96.1 15.5 ⫾ 124.1
Left Lobe (cm3) 328.1 74.5 ⫾ 62.1 104.5 ⫾ 68.8
Note.—Values presented as mean ⫾ SD where applicable.
lobe, and 4 representing fibrosis affecting at least 51% of the lobe.
Statistical Analysis Ordinal logistic regression models were used to assess the association between the fibrosis scores and volume of embolic mixture administered, as well as the association between fibrosis scores and HVPGs. The Wilcoxon rank-sum test was used to compare the liver lobes with respect to the maximum degree of fibrosis. Differences in HVPGs between treatment groups were assessed by using a twosample t test for unequal variances. All tests were conducted at the .05 significance level, and P values were calculated based on the Satterthwaite approximation for the degrees of freedom for the t statistic.
RESULTS Transarterial hepatic arterial embolization was technically successful in all 12 animals. The animals that received 16 mL or 28 mL of the embolic mixture (n ⫽ 8) recovered from anesthesia without complications. However, all animals in the final group that received 40 mL of the mixture (n ⫽ 4) never fully recovered from anesthesia, remaining somnolent and unable to ambulate, and consequently had to be euthanized. The preprocedural weight of animals ranged from 29.8 to 46.4 kg (mean, 41.4 kg), and postprocedural weight ranged from 33 to 54.8 kg (mean, 46.1 kg).
Hepatic Venous Pressure Measurements The 2-week and prenecropsy HVPG values were compared versus the baseline measurement for the entire surviving group (n ⫽ 8), as well as for each group separately. The mean HVPG increases at 2 weeks and 6 – 8 weeks after embolization for all animals were 7.1 mm Hg (P ⫽ .09) and 8.5 mm Hg (P ⫽ .03), respectively. However, the HVPG increase in each individual treatment group did not reach statistical significance.
Liver Volumetry and Laboratory Analysis The 2-week and final liver volumes were compared versus baseline volume for the entire surviving group (n ⫽ 8). Results for the comparison are shown in Table 1. Mean right and left lobe measurements before embolization were
1332 䡲 Swine Model of Cirrhosis and Portal Hypertension via Hepatic Embolization
Avritscher et al 䡲 JVIR
Figure 2. Hepatic arterial embolization with ethanol and iodized oil causes cirrhosis. Representative histologic images show the effects of hepatic arterial embolization in the liver parenchyma. (a) Bridging fibrosis (arrows) and micronodular formation (N) are seen (original magnification ⫻ 40; Masson trichrome stain). (b) Fibroblast proliferation is seen in the portal triads, with extensive periportal collagen deposition (arrows; Masson trichrome stain; original magnification, ⫻ 400). (c) Micronodular formation (N) with biliary hyperplasia is visible (arrows; hematoxylin and eosin stain; original magnification, ⫻ 100). (Available in color online at www.jvir.org.)
Figure 3. Hepatic arterial embolization with ethanol and iodized oil causes cirrhosis. Representative gross pathologic photographs after transcatheter arterial hepatic embolization with ethanol and iodized oil. Photos after liver explantation show various degrees of parenchymal atrophy involving two or more lobes, with compensatory hypertrophy of unaffected lobes. (a) Marked fibrosis and atrophy of the right lateral (RL), right medial (RM), and left medial (LM) lobes are seen, along with compensatory hypertrophy of the left lateral (LL) lobe (arrows). Atrophy of gallbladder (GB) is also noted. (b) Macronodular cirrhosis is present throughout the liver surface (arrows). (Available in color online at www.jvir.org.)
452.3 cm3 and 328.1 cm3. The mean increases in right and left lobe volumes after 8 weeks were 15.5 cm3 and 104.5 cm3, respectively. Results of comparisons between liver enzymes—aspartate and alanine aminotransferases, alkaline phosphatase, and lactate dehydrogenase—and fibrosis were not statistically significant (P ⫽ .87, P ⫽ .31, P ⫽ .11, and P ⫽ .29, respectively).
Pathologic Assessment Light microscopic examination of the histologic samples demonstrated bridging fibrosis and regenerative nodules in three of four pigs treated with 16 mL of embolic mixture and in all four pigs treated with 28 mL of embolic mixture. Moderate to marked periportal ductular reaction with surrounding neutrophilic infiltration was visualized (Fig 2). Multiple hepatic arterial branches in the portal triads demonstrated ulcerated or hyperplastic endothelium with occasional obliteration of the vascular lumen. The remaining pig
from the 16-mL dose group showed only focal areas of mild portal fibrosis and no regenerative nodules. Gross examination of explanted cirrhotic livers revealed discoloration and atrophy of at least two hepatic lobes with compensatory hypertrophy of the nonaffected lobes. The affected surfaces were firm and nodular (Fig 3). Marked to moderate ascites was present in two of the eight animals, whereas the other six exhibited mild amount of ascites fluid. The fibrosis scores for all biopsy samples are shown in Table 2, and the extent of parenchymal involvement in the explanted livers is shown in Table 3. The extent of fibrosis in each individual lobe was correlated with the volume of mixture administered in the two separate groups based on ordinal logistic regression models. The results did not show statistical significance (right medial lobe, P ⫽ .29; right lateral lobe, P ⫽ .14; left medial lobe, P ⫽ .63; and left lateral lobe, P ⫽ .72). Correlation between severities of fibrosis in the entire right versus left lobes was assessed
Volume 22 䡲 Number 9 䡲 September 䡲 2011
1333
Table 2. Histologic Assessment of Portal Fibrosis in Biopsy Samples 2 Weeks after Embolization Subject 16-mL group 1 2 3 4 28-mL group 5 6 7 8
METAVIR Score 2 1 0 0 4 4 4 2
Note.—METAVIR score: 0, no fibrosis; 1, portal fibrosis with no septa; 2, portal fibrosis with few septa; 3, bridging fibrosis with many septa; 4, cirrhosis.
Table 3. Histologic Assessment of Extent of Portal Fibrosis in Explanted Livers 8 Weeks after Embolization Extent of Portal Fibrosis
Subject
Right Lateral
Right Medial
Left Lateral
Left Medial
16-mL group 1
0
1
1
1
0 0 4
3 1 2
4 4 0
4 4 2
3 3 3 4
0 3 3 4
3 3 3 4
0 3 2 4
2 3 4 28-mL group 5 6 7 8
Note.—Extent of portal fibrosis: 0, no fibrosis; 1, minimal fibrosis (findings present in ⬍ 10% of the lobe); 2, mild fibrosis (findings present in 11%–20% of the lobe); 3, moderate fibrosis (findings present in 21%–50% of the lobe); 4, severe fibrosis (findings present in ⬎ 50% of the lobe).
with the Wilcoxon rank-sum test. The results also failed to reach statistical significance (P ⫽ .75), but showed a trend toward greater degree of fibrosis in the 28-mL group.
DISCUSSION Cirrhosis is the common final pathway of sustained and repetitive injury and healing responses to liver insults and is complicated by the development of portal hypertension and hepatocellular carcinoma (16). An animal model of cirrhosis would be useful in validating the modalities that monitor liver fibrosis and test minimally invasive procedures (17). Our results here demonstrate that a single session of transcatheter hepatic arterial embolization with 16 or 28 mL of a 3:1 mixture of iodized oil and ethanol successfully
induced hepatic cirrhosis and portal hypertension in seven of eight pigs subjected to the procedure. We also demonstrated that volumes equal or greater than 40 mL of the mixture are lethal. The increase in left liver volume might be secondary to preferential arterial flow into the right liver as a result of its larger size, leading to deposition of greater quantities of the agent during embolization. The concept of using an iodized oil and ethanol mixture to produce liver cirrhosis was derived from a previous report on partial hepatic transcatheter embolization in an attempt to cause contralateral hepatic hypertrophy (13). Iodized oil is known to enter the portal vein through the peribiliary plexus after being injected into the hepatic artery (18). Hence, transcatheter hepatic arterial administration of iodized oil produces a dual embolization of the hepatic artery and portal vein. The addition of ethanol is essential, because it enhances the embolic effect by causing endothelial damage and intravascular sludging, thus preventing the mixture from exiting the liver parenchyma. In the absence of ethanol, the iodized oil would pass through the sinusoids without substantial endothelial and hepatocellular injury (18). Previous studies have demonstrated that the 3:1 ratio is optimal, as higher ethanol ratios would lead to excessive arterial spasm, limiting the volume of the mixture that can be infused (19). Direct hepatocellular injury of the iodized oil/ethanol mixture is confirmed in the present study by the presence of necrotic hepatocytes in the periportal spaces. The iodized oil/ethanol mixture also caused severe injury to the biliary epithelium. The transcatheter hepatic arterial approach eliminates the need for multiple treatments, thus reducing the cost of generating cirrhosis models. The technique produces fibrosis within 2 weeks with features persisting for at least 6 – 8 weeks. This represents a much shorter induction period compared with other models described in the literature (6). Carbon tetrachloride is a hepatotoxin most commonly used to create animal models of liver fibrosis and has been used to create liver injury in a wide selection of animals (20,21). Recently, Zhang et al (22) used carbon tetrachloride to produce fibrosis in swine livers. This method of inducing liver cirrhosis is more expensive, labor-intensive, and time-consuming. Our technique decreases the induction time and requires a single initial intervention, minimizing experimental stress to the animals, as well as reducing personnel and equipment utilization. Use of transcatheter embolization for the development of large-animal models for cirrhosis has been previously attempted (10). However, in that study (10), nonspherical polyvinyl alcohol particles were the embolic agent infused. Although the investigators did achieve increases in portal pressures, the results were only transient, lasting approximately 1 week, because of subsequent vascular recanalization. In addition, substantial fibrosis or cirrhosis features were not found on their histopathologic evaluation. None of the animals in our experiment developed diffuse fibrosis or cirrhosis throughout the entire liver. This
1334 䡲 Swine Model of Cirrhosis and Portal Hypertension via Hepatic Embolization
appears to be a result of the asymmetric hepatic arterial anatomy. It is relevant to note that this nonuniform pattern is also observed in humans. There are also potential discrepancies between the biochemical and biologic behavior of this model compared with human disease. Further studies in larger numbers of animals are already under way that are specifically designed to investigate the pathogenesis of this model. In conclusion, we demonstrated that the transcatheter hepatic arterial embolization model using a mixture of ethanol and iodized oil reliably creates fibrosis and cirrhosis in porcine liver after a single session and with short induction times. The ensuing cirrhosis is associated with portal hypertension after 6 – 8 weeks.
REFERENCES 1. Yi NJ, Suh KS, Lee HW, et al. Improved outcome of adult recipients with a high model for end-stage liver disease score and a small-for-size graft. Liver Transpl 2009; 15:496 –503. 2. National Institute of Diabetes and Digestive and Kidney Diseases. Digestive diseases in the United States: epidemiology and impact. NIH Publication 94 –1447. Washington, DC: US Government Printing Office, 1994. 3. Friedman SL, Bansal MB. Reversal of hepatic fibrosis—fact or fantasy? Hepatology 2006; 43(Suppl):S82–S88. 4. Rockey DC. Antifibrotic therapy in chronic liver disease. Clin Gastroenterol Hepatol 2005; 3:95–107. 5. Rockey DC. New therapies in hepatitis C virus and chronic liver disease: antifibrotics. Clin Liver Dis 2006; 10:881–900. 6. Abraldes JG, Pasarin M, Garcia-Pagan JC. Animal models of portal hypertension. World J Gastroenterol 2006; 12:6577– 6584. 7. Sun MA, Wang BE, Annoni G, et al. Two rat models of hepatic fibrosis. A morphologic and molecular comparison. Lab Invest 1990; 63:467– 475.
Avritscher et al 䡲 JVIR
8. Tunon MJ, Alvarez M, Culebras JM, Gonzalez-Gallego J. An overview of animal models for investigating the pathogenesis and therapeutic strategies in acute hepatic failure. World J Gastroenterol 2009; 15:3086 – 3098. 9. Ugazio G, Bosia S, Cornaglia E. Experimental model of cirrhosis in rabbits exposed to carbon tetrachloride by inhalation. Res Commun Mol Pathol Pharmacol 1995; 88:63–77. 10. Pavcnik D, Saxon RR, Kubota Y, et al. Attempted induction of chronic portal venous hypertension with polyvinyl alcohol particles in swine. J Vasc Interv Radiol 1997; 8:123–128. 11. Court FG, Wemyss-Holden SA, Morrison CP, et al. Segmental nature of the porcine liver and its potential as a model for experimental partial hepatectomy. Br J Surg 2003; 90:440 – 444. 12. Swindle M, Smith AC. Comparative anatomy and physiology of the pig. Scand J Lab Anim Sci 1998; 25(Suppl 1):11–21. 13. Madoff DC, Gupta S, Pillsbury EP, et al. Transarterial versus transhepatic portal vein embolization to induce selective hepatic hypertrophy: a comparative study in swine. J Vasc Interv Radiol 2007; 18:79 –93. 14. Bosch J, Abraldes JG, Berzigotti A, Garcia-Pagan JC. The clinical use of HVPG measurements in chronic liver disease. Nat Rev 2009; 6:573–582. 15. Bedossa P, Poynard T. An algorithm for the grading of activity in chronic hepatitis C. The METAVIR Cooperative Study Group. Hepatology 1996; 24:289 –293. 16. Friedman SL, Rockey DC, Bissell DM. Hepatic fibrosis 2006: report of the Third AASLD Single Topic Conference. Hepatology 2007; 45:242– 249. 17. Friedrich-Rust M, Muller C, Winckler A, et al. Assessment of liver fibrosis and steatosis in PBC with FibroScan, MRI, MR-spectroscopy, and serum markers. J Clin Gastroenterol 2010; 44:58 – 65. 18. Kan Z, Wright K, Wallace S. Ethiodized oil emulsions in hepatic microcirculation: in vivo microscopy in animal models. Acad Radiol 1997; 4:275–282. 19. Kan Z, Wallace S. Transcatheter liver lobar ablation: an experimental trial in an animal model. Eur Radiol 1997; 7:1071–1075. 20. Tamayo RP. Is cirrhosis of the liver experimentally produced by CCl4 an adequate model of human cirrhosis? Hepatology 1983; 3:112–120. 21. Slater TF, Cheeseman KH, Ingold KU. Carbon tetrachloride toxicity as a model for studying free-radical mediated liver injury. Philos Trans R Soc Lond B Biol Sci 1985; 311:633– 645. 22. Zhang JJ, Meng XK, Dong C, et al. Development of a new animal model of liver cirrhosis in swine. Eur Surg Res 2009; 42:35–39.
Development of a Large Animal Model of Cirrhosis and Portal Hypertension Using Hepatic Transarterial Embolization: A Study in Swine Rony Avritscher, MD, Kenneth C. Wright, PhD, Sanaz Javadi, MD, Rajesh Uthamanthil, DVM, Sanjay Gupta, MD, Mihai Gagea, DVSc, Roland L. Bassett, MS, Ravi Murthy, MD, Michael J. Wallace, MD, and David C. Madoff, MD
ABSTRACT Purpose: To develop a clinically relevant porcine model of liver cirrhosis with portal hypertension by means of hepatic transarterial embolization. Materials and Methods: Institutional animal care and use committee approval was obtained for all experiments. Pigs received transcatheter arterial infusion of a 3:1 mixture of iodized oil and ethanol into the hepatic artery in volumes of 16 mL in group 1 (n ⫽ 4), 28 mL in group 2 (n ⫽ 4), and 40 mL in group 3 (n ⫽ 4) with intent of bilobar distribution. Hepatic venous pressure gradient (HVPG) measurement, liver function tests, and volumetry were performed at baseline, at 2 weeks, and before necropsy. Results: Cirrhosis was successfully induced in three animals that received 16 mL of the embolic mixture and in all four animals that received 28 mL. The animals in the 40-mL group did not recover from the procedure and were euthanized within 48 h. Increases in HVPG after 6 – 8 weeks versus baseline reached statistical significance (P ⬍ .05). Correlation between degree of fibrosis and volume of embolic agent did not reach statistical significance, but there was a trend toward increased fibrosis in the 28-mL group compared with the 16-mL group. Conclusions: Transcatheter hepatic arterial embolization can be used to create a reliable and reproducible porcine model of liver cirrhosis and portal hypertension.
ABBREVIATION HVPG ⫽ hepatic venous pressure gradient
Cirrhosis is the most common nonneoplastic cause of death in patients with hepatobiliary and digestive disease, and is the ninth most frequent cause of death in the United States
From the Department of Diagnostic Radiology, Interventional Radiology Section (R.A., K.C.W., S.J., S.G., R.M., M.J.W., D.C.M.), and Departments of Veterinary Medicine (R.U.), Veterinary Pathology (M.G.), and Biostatistics (R.L.B.), University of Texas M. D. Anderson Cancer Center, Houston, Texas. Received February 23, 2011; final revision received April 19, 2011; accepted April 21, 2011. Address correspondence to D.C.M., Division of Interventional Radiology, New York Presbyterian Hospital/Weill Cornell Medical Center, 525 E. 68th St., P-518, New York, NY 10065; E-mail: dcm9006@med. cornell.edu Research for this study was supported in part by a grant from the John S. Dunn Research Foundation and by the National Institutes of Health through Support Grant CA016672 to M. D. Anderson Cancer Center. None of the authors have identified a conflict of interest. © SIR, 2011 J Vasc Interv Radiol 2011; 22:1329 –1334 DOI: 10.1016/j.jvir.2011.04.016
(1). Portal hypertension and liver cancer are feared complications of cirrhosis, with an associated 5-year mortality rate exceeding 50% (1–3). There are medications to slow or reverse progression of fibrosis, and interventional procedures that can mitigate the effects of portal hypertension. Animal models are needed to test and validate these approaches (4,5). The animal models of liver cirrhosis currently available are created by using a variety of hepatotoxins, such as carbon tetrachloride. The animals most commonly used for such models are mice, rats, and rabbits (6 – 8). Although these small and medium-sized laboratory animals are well suited for pharmacologic studies, large animal models are required to optimally test imaging tools and percutaneous interventions. In addition, such models can be used as large-animal hepatocellular carcinoma models, given that these tumors mostly arise in background cirrhosis. Suitable large-animal models of cirrhosis are currently lacking. The models currently described in the literature require several
1330 䡲 Swine Model of Cirrhosis and Portal Hypertension via Hepatic Embolization
repeated applications of the hepatotoxins and long induction periods (7,9). In addition, none of these models report concomitant development of portal hypertension. Pavcnik et al (10) attempted a percutaneous transhepatic technique to induce portal hypertension in swine in an experiment with the use of polyvinyl alcohol particles, but the portal pressures always returned to baseline levels at 1-week follow-up studies. The swine species is particularly advantageous in the creation of a large-animal model of cirrhosis because of the anatomic and physiologic similarities between the porcine and human liver (11,12). The use of hepatic arterial embolization with ethanol has previously been reported to produce severe hepatocellular damage by direct tissue toxicity, endothelial damage, and “sludging” of erythrocytes. The addition of iodized oil prolongs the contact time between the ethanol and the tissues and promotes a more homogenous distribution of the agents (13). The purpose of the current study was to develop a suitable and reproducible large animal model of liver cirrhosis.
MATERIALS AND METHODS
Avritscher et al 䡲 JVIR
Figure 1. Transcatheter arterial hepatic embolization using ethanol/iodized oil mixture. Close-up of an anteroposterior radiograph of the abdomen in a later phase of the procedure shows embolic mixture in portal vein branches (arrows). Metallic coils are noted in the gastroduodenal artery (arrowheads). The coils are used to prevent nontarget embolization.
Animal Care The institutional animal care and use committee approved the current study. Animals were maintained in facilities approved by the Association for Assessment and Accreditation of Laboratory Animal Care and in accordance with current United States Department of Agriculture, Department of Health and Human Services, and National Institutes of Health regulations and standards.
Hepatic Venous Pressure Measurement A balloon-occlusion catheter was used to measure free and wedged hepatic venous pressures, and the hepatic venous pressure gradient (HVPG) was calculated according to the standards previously described by Pagan et al (14). Measurements were obtained at baseline before embolization, 2 weeks after embolization, and at 6 – 8 weeks before necropsy.
Preprocedure Preparation Twelve domestic pigs weighing a mean of 41.4 kg (range, 29.8 – 46.4 kg) were sedated with an intramuscular injection of a solution containing ketamine hydrochloride (15 mg/ kg), acepromazine (0.15 mg/kg), and atropine sulfate (0.04 mg/kg). Anesthesia was then induced with isoflurane (5%) administered by a facemask. When anesthesia was established, an endotracheal tube was inserted and anesthesia was maintained with isoflurane (1.5%–3%) and oxygen (0.8 L/min). The animals were given intramuscular doses of 5 mg/kg of the antibiotic enrofloxacin (Baytril; Bayer Animal Health, Shawnee Mission, Kansas) before each procedure and once daily for 5 days after each procedure. At baseline, 2 weeks, and immediately before necropsy, all animals underwent nonenhanced computed tomography imaging (HiSpeed Advantage; GE Medical Systems, Milwaukee, Wisconsin) in 5-mm slices of the liver. Animals were weighed before embolization and before necropsy. All experimental and control procedures were performed by experienced radiologists familiar with porcine hepatic vascular anatomy.
Transarterial Procedures Access to the right common femoral artery was obtained by using Seldinger technique under sonographic guidance or via surgical cutdown. A single intravenous bolus of heparin (100 U/kg) was administered after vascular access was secure. A 5-F catheter (Sos-2; AngioDynamics, Queensbury, New York) was advanced over the wire into the aorta, and the celiac axis was selected. Celiac and hepatic digital subtraction angiography was accomplished by injecting iodinated radiographic contrast medium (meglumine diatrizoate; Nycomed, Zurich, Switzerland) through the catheter. A 3-F microcatheter (Tracker 325; Boston Scientific, Natick, Massachusetts) was then advanced coaxially into the common hepatic artery. Embolization of the gastroduodenal and right gastric arteries was then performed with metallic fibered coils. The microcatheter was advanced into the proper hepatic artery. Radiographic contrast medium was injected to confirm catheter position before embolization under fluoroscopy (Fig 1). Three groups of pigs were randomly assigned to re-
Volume 22 䡲 Number 9 䡲 September 䡲 2011
ceive 16 mL (n ⫽ 4), 28 mL (n ⫽ 4), or 40 mL (n ⫽ 4) of a 3:1 ratio mixture of iodized oil (Ethiodol; Savage Laboratories, Melville, New York) and absolute ethanol (13). Because rapid infusion of the mixture can lead to vascular spasm of the parent artery and premature termination of the procedure, the mixture was injected with an automated infusion pump under fluoroscopic monitoring. The infusion was carried out at 0.3 mL/s. The embolization endpoint was delivery of the entire volume of embolic agent. After completion of the procedure, the coaxial catheter system was removed, the right common femoral artery was ligated, and the incision was repaired with primary closure.
Liver Biopsies Two weeks after embolization, multiple liver biopsy samples were acquired under real-time sonographic guidance with an 18-gauge core needle biopsy. The biopsy samples were fixed by immersion in 10% neutral buffered formalin. The fixed tissues were embedded in paraffin blocks, stained with hematoxylin and eosin and Masson trichrome stain, and examined microscopically. Liver lesions were staged according to the METAVIR scoring system (15).
Liver Volumetry and Biochemical Analysis Liver volumes were calculated by using syngo InSpace software (Siemens, Erlangen, Germany). The external liver contours were manually segmented with the proprietary software and subsequently summated to obtain the estimated volumes. The middle hepatic vein was used as the dividing landmark to calculate the volumes of the right and left hepatic lobes. Right and left lobe volumes were analyzed individually. The mean baseline, 2-week, and final postprocedural right and left liver volumes were calculated in cubic centimeters and compared among the different treatment groups. In addition, at the same time points, all experimental animals had blood samples collected for aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, and lactate dehydrogenase.
Necropsy and Histopathologic Evaluation After final measurement of hepatic venous pressure, the animals were euthanized with an overdose of Beuthanasia-D (1 mL per 10 lbs; Schering-Plough, Union, New Jersey), and a complete necropsy was performed. At necropsy, the liver was examined grossly and multiple tissue samples were collected from each lobe of the liver and stained in the same manner described earlier. A veterinary pathologist examined the specimens under light microscopy. The overall presence of fibrosis throughout each lobe of the liver was graded on a scale from 0 to 4, with 0 representing no fibrosis, 1 representing fibrosis affecting less than 10% of the volume of the hepatic lobe, 2 representing fibrosis affecting 10%–20% of the lobe, 3 representing fibrosis affecting 21%–50% of the
1331
Table 1. Changes in Liver Volumes 2 and 8 Weeks after Embolization Interval Before embolization 2 Weeks 8 Weeks
Right Lobe (cm3) 452.2 62.8 ⫾ 96.1 15.5 ⫾ 124.1
Left Lobe (cm3) 328.1 74.5 ⫾ 62.1 104.5 ⫾ 68.8
Note.—Values presented as mean ⫾ SD where applicable.
lobe, and 4 representing fibrosis affecting at least 51% of the lobe.
Statistical Analysis Ordinal logistic regression models were used to assess the association between the fibrosis scores and volume of embolic mixture administered, as well as the association between fibrosis scores and HVPGs. The Wilcoxon rank-sum test was used to compare the liver lobes with respect to the maximum degree of fibrosis. Differences in HVPGs between treatment groups were assessed by using a twosample t test for unequal variances. All tests were conducted at the .05 significance level, and P values were calculated based on the Satterthwaite approximation for the degrees of freedom for the t statistic.
RESULTS Transarterial hepatic arterial embolization was technically successful in all 12 animals. The animals that received 16 mL or 28 mL of the embolic mixture (n ⫽ 8) recovered from anesthesia without complications. However, all animals in the final group that received 40 mL of the mixture (n ⫽ 4) never fully recovered from anesthesia, remaining somnolent and unable to ambulate, and consequently had to be euthanized. The preprocedural weight of animals ranged from 29.8 to 46.4 kg (mean, 41.4 kg), and postprocedural weight ranged from 33 to 54.8 kg (mean, 46.1 kg).
Hepatic Venous Pressure Measurements The 2-week and prenecropsy HVPG values were compared versus the baseline measurement for the entire surviving group (n ⫽ 8), as well as for each group separately. The mean HVPG increases at 2 weeks and 6 – 8 weeks after embolization for all animals were 7.1 mm Hg (P ⫽ .09) and 8.5 mm Hg (P ⫽ .03), respectively. However, the HVPG increase in each individual treatment group did not reach statistical significance.
Liver Volumetry and Laboratory Analysis The 2-week and final liver volumes were compared versus baseline volume for the entire surviving group (n ⫽ 8). Results for the comparison are shown in Table 1. Mean right and left lobe measurements before embolization were
1332 䡲 Swine Model of Cirrhosis and Portal Hypertension via Hepatic Embolization
Avritscher et al 䡲 JVIR
Figure 2. Hepatic arterial embolization with ethanol and iodized oil causes cirrhosis. Representative histologic images show the effects of hepatic arterial embolization in the liver parenchyma. (a) Bridging fibrosis (arrows) and micronodular formation (N) are seen (original magnification ⫻ 40; Masson trichrome stain). (b) Fibroblast proliferation is seen in the portal triads, with extensive periportal collagen deposition (arrows; Masson trichrome stain; original magnification, ⫻ 400). (c) Micronodular formation (N) with biliary hyperplasia is visible (arrows; hematoxylin and eosin stain; original magnification, ⫻ 100). (Available in color online at www.jvir.org.)
Figure 3. Hepatic arterial embolization with ethanol and iodized oil causes cirrhosis. Representative gross pathologic photographs after transcatheter arterial hepatic embolization with ethanol and iodized oil. Photos after liver explantation show various degrees of parenchymal atrophy involving two or more lobes, with compensatory hypertrophy of unaffected lobes. (a) Marked fibrosis and atrophy of the right lateral (RL), right medial (RM), and left medial (LM) lobes are seen, along with compensatory hypertrophy of the left lateral (LL) lobe (arrows). Atrophy of gallbladder (GB) is also noted. (b) Macronodular cirrhosis is present throughout the liver surface (arrows). (Available in color online at www.jvir.org.)
452.3 cm3 and 328.1 cm3. The mean increases in right and left lobe volumes after 8 weeks were 15.5 cm3 and 104.5 cm3, respectively. Results of comparisons between liver enzymes—aspartate and alanine aminotransferases, alkaline phosphatase, and lactate dehydrogenase—and fibrosis were not statistically significant (P ⫽ .87, P ⫽ .31, P ⫽ .11, and P ⫽ .29, respectively).
Pathologic Assessment Light microscopic examination of the histologic samples demonstrated bridging fibrosis and regenerative nodules in three of four pigs treated with 16 mL of embolic mixture and in all four pigs treated with 28 mL of embolic mixture. Moderate to marked periportal ductular reaction with surrounding neutrophilic infiltration was visualized (Fig 2). Multiple hepatic arterial branches in the portal triads demonstrated ulcerated or hyperplastic endothelium with occasional obliteration of the vascular lumen. The remaining pig
from the 16-mL dose group showed only focal areas of mild portal fibrosis and no regenerative nodules. Gross examination of explanted cirrhotic livers revealed discoloration and atrophy of at least two hepatic lobes with compensatory hypertrophy of the nonaffected lobes. The affected surfaces were firm and nodular (Fig 3). Marked to moderate ascites was present in two of the eight animals, whereas the other six exhibited mild amount of ascites fluid. The fibrosis scores for all biopsy samples are shown in Table 2, and the extent of parenchymal involvement in the explanted livers is shown in Table 3. The extent of fibrosis in each individual lobe was correlated with the volume of mixture administered in the two separate groups based on ordinal logistic regression models. The results did not show statistical significance (right medial lobe, P ⫽ .29; right lateral lobe, P ⫽ .14; left medial lobe, P ⫽ .63; and left lateral lobe, P ⫽ .72). Correlation between severities of fibrosis in the entire right versus left lobes was assessed
Volume 22 䡲 Number 9 䡲 September 䡲 2011
1333
Table 2. Histologic Assessment of Portal Fibrosis in Biopsy Samples 2 Weeks after Embolization Subject 16-mL group 1 2 3 4 28-mL group 5 6 7 8
METAVIR Score 2 1 0 0 4 4 4 2
Note.—METAVIR score: 0, no fibrosis; 1, portal fibrosis with no septa; 2, portal fibrosis with few septa; 3, bridging fibrosis with many septa; 4, cirrhosis.
Table 3. Histologic Assessment of Extent of Portal Fibrosis in Explanted Livers 8 Weeks after Embolization Extent of Portal Fibrosis
Subject
Right Lateral
Right Medial
Left Lateral
Left Medial
16-mL group 1
0
1
1
1
0 0 4
3 1 2
4 4 0
4 4 2
3 3 3 4
0 3 3 4
3 3 3 4
0 3 2 4
2 3 4 28-mL group 5 6 7 8
Note.—Extent of portal fibrosis: 0, no fibrosis; 1, minimal fibrosis (findings present in ⬍ 10% of the lobe); 2, mild fibrosis (findings present in 11%–20% of the lobe); 3, moderate fibrosis (findings present in 21%–50% of the lobe); 4, severe fibrosis (findings present in ⬎ 50% of the lobe).
with the Wilcoxon rank-sum test. The results also failed to reach statistical significance (P ⫽ .75), but showed a trend toward greater degree of fibrosis in the 28-mL group.
DISCUSSION Cirrhosis is the common final pathway of sustained and repetitive injury and healing responses to liver insults and is complicated by the development of portal hypertension and hepatocellular carcinoma (16). An animal model of cirrhosis would be useful in validating the modalities that monitor liver fibrosis and test minimally invasive procedures (17). Our results here demonstrate that a single session of transcatheter hepatic arterial embolization with 16 or 28 mL of a 3:1 mixture of iodized oil and ethanol successfully
induced hepatic cirrhosis and portal hypertension in seven of eight pigs subjected to the procedure. We also demonstrated that volumes equal or greater than 40 mL of the mixture are lethal. The increase in left liver volume might be secondary to preferential arterial flow into the right liver as a result of its larger size, leading to deposition of greater quantities of the agent during embolization. The concept of using an iodized oil and ethanol mixture to produce liver cirrhosis was derived from a previous report on partial hepatic transcatheter embolization in an attempt to cause contralateral hepatic hypertrophy (13). Iodized oil is known to enter the portal vein through the peribiliary plexus after being injected into the hepatic artery (18). Hence, transcatheter hepatic arterial administration of iodized oil produces a dual embolization of the hepatic artery and portal vein. The addition of ethanol is essential, because it enhances the embolic effect by causing endothelial damage and intravascular sludging, thus preventing the mixture from exiting the liver parenchyma. In the absence of ethanol, the iodized oil would pass through the sinusoids without substantial endothelial and hepatocellular injury (18). Previous studies have demonstrated that the 3:1 ratio is optimal, as higher ethanol ratios would lead to excessive arterial spasm, limiting the volume of the mixture that can be infused (19). Direct hepatocellular injury of the iodized oil/ethanol mixture is confirmed in the present study by the presence of necrotic hepatocytes in the periportal spaces. The iodized oil/ethanol mixture also caused severe injury to the biliary epithelium. The transcatheter hepatic arterial approach eliminates the need for multiple treatments, thus reducing the cost of generating cirrhosis models. The technique produces fibrosis within 2 weeks with features persisting for at least 6 – 8 weeks. This represents a much shorter induction period compared with other models described in the literature (6). Carbon tetrachloride is a hepatotoxin most commonly used to create animal models of liver fibrosis and has been used to create liver injury in a wide selection of animals (20,21). Recently, Zhang et al (22) used carbon tetrachloride to produce fibrosis in swine livers. This method of inducing liver cirrhosis is more expensive, labor-intensive, and time-consuming. Our technique decreases the induction time and requires a single initial intervention, minimizing experimental stress to the animals, as well as reducing personnel and equipment utilization. Use of transcatheter embolization for the development of large-animal models for cirrhosis has been previously attempted (10). However, in that study (10), nonspherical polyvinyl alcohol particles were the embolic agent infused. Although the investigators did achieve increases in portal pressures, the results were only transient, lasting approximately 1 week, because of subsequent vascular recanalization. In addition, substantial fibrosis or cirrhosis features were not found on their histopathologic evaluation. None of the animals in our experiment developed diffuse fibrosis or cirrhosis throughout the entire liver. This
1334 䡲 Swine Model of Cirrhosis and Portal Hypertension via Hepatic Embolization
appears to be a result of the asymmetric hepatic arterial anatomy. It is relevant to note that this nonuniform pattern is also observed in humans. There are also potential discrepancies between the biochemical and biologic behavior of this model compared with human disease. Further studies in larger numbers of animals are already under way that are specifically designed to investigate the pathogenesis of this model. In conclusion, we demonstrated that the transcatheter hepatic arterial embolization model using a mixture of ethanol and iodized oil reliably creates fibrosis and cirrhosis in porcine liver after a single session and with short induction times. The ensuing cirrhosis is associated with portal hypertension after 6 – 8 weeks.
REFERENCES 1. Yi NJ, Suh KS, Lee HW, et al. Improved outcome of adult recipients with a high model for end-stage liver disease score and a small-for-size graft. Liver Transpl 2009; 15:496 –503. 2. National Institute of Diabetes and Digestive and Kidney Diseases. Digestive diseases in the United States: epidemiology and impact. NIH Publication 94 –1447. Washington, DC: US Government Printing Office, 1994. 3. Friedman SL, Bansal MB. Reversal of hepatic fibrosis—fact or fantasy? Hepatology 2006; 43(Suppl):S82–S88. 4. Rockey DC. Antifibrotic therapy in chronic liver disease. Clin Gastroenterol Hepatol 2005; 3:95–107. 5. Rockey DC. New therapies in hepatitis C virus and chronic liver disease: antifibrotics. Clin Liver Dis 2006; 10:881–900. 6. Abraldes JG, Pasarin M, Garcia-Pagan JC. Animal models of portal hypertension. World J Gastroenterol 2006; 12:6577– 6584. 7. Sun MA, Wang BE, Annoni G, et al. Two rat models of hepatic fibrosis. A morphologic and molecular comparison. Lab Invest 1990; 63:467– 475.
Avritscher et al 䡲 JVIR
8. Tunon MJ, Alvarez M, Culebras JM, Gonzalez-Gallego J. An overview of animal models for investigating the pathogenesis and therapeutic strategies in acute hepatic failure. World J Gastroenterol 2009; 15:3086 – 3098. 9. Ugazio G, Bosia S, Cornaglia E. Experimental model of cirrhosis in rabbits exposed to carbon tetrachloride by inhalation. Res Commun Mol Pathol Pharmacol 1995; 88:63–77. 10. Pavcnik D, Saxon RR, Kubota Y, et al. Attempted induction of chronic portal venous hypertension with polyvinyl alcohol particles in swine. J Vasc Interv Radiol 1997; 8:123–128. 11. Court FG, Wemyss-Holden SA, Morrison CP, et al. Segmental nature of the porcine liver and its potential as a model for experimental partial hepatectomy. Br J Surg 2003; 90:440 – 444. 12. Swindle M, Smith AC. Comparative anatomy and physiology of the pig. Scand J Lab Anim Sci 1998; 25(Suppl 1):11–21. 13. Madoff DC, Gupta S, Pillsbury EP, et al. Transarterial versus transhepatic portal vein embolization to induce selective hepatic hypertrophy: a comparative study in swine. J Vasc Interv Radiol 2007; 18:79 –93. 14. Bosch J, Abraldes JG, Berzigotti A, Garcia-Pagan JC. The clinical use of HVPG measurements in chronic liver disease. Nat Rev 2009; 6:573–582. 15. Bedossa P, Poynard T. An algorithm for the grading of activity in chronic hepatitis C. The METAVIR Cooperative Study Group. Hepatology 1996; 24:289 –293. 16. Friedman SL, Rockey DC, Bissell DM. Hepatic fibrosis 2006: report of the Third AASLD Single Topic Conference. Hepatology 2007; 45:242– 249. 17. Friedrich-Rust M, Muller C, Winckler A, et al. Assessment of liver fibrosis and steatosis in PBC with FibroScan, MRI, MR-spectroscopy, and serum markers. J Clin Gastroenterol 2010; 44:58 – 65. 18. Kan Z, Wright K, Wallace S. Ethiodized oil emulsions in hepatic microcirculation: in vivo microscopy in animal models. Acad Radiol 1997; 4:275–282. 19. Kan Z, Wallace S. Transcatheter liver lobar ablation: an experimental trial in an animal model. Eur Radiol 1997; 7:1071–1075. 20. Tamayo RP. Is cirrhosis of the liver experimentally produced by CCl4 an adequate model of human cirrhosis? Hepatology 1983; 3:112–120. 21. Slater TF, Cheeseman KH, Ingold KU. Carbon tetrachloride toxicity as a model for studying free-radical mediated liver injury. Philos Trans R Soc Lond B Biol Sci 1985; 311:633– 645. 22. Zhang JJ, Meng XK, Dong C, et al. Development of a new animal model of liver cirrhosis in swine. Eur Surg Res 2009; 42:35–39.