Medical research has made dramatic breakthroughs in the development of implantable biomaterial devices. A significant limitation to their extensive use has been thrombotic complications and the need for systemic anticoagulation.1,2 The complex nature of and redundancy within the clotting cascade makes its selective inhibition a formidable task. The manipulation of the biomaterial surface has produced some improvement as evidenced by the less thrombogenic ventricular assist devices and heparin-bonded circuits, but the formation of a passive surface has not yet been achieved.3In vitro studies can be performed to evaluate new drugs or surface modifications, but the correlation of these studies with in vivo results has been poor4 because the blood biomaterial interface is extremely complex. It requires an intricate knowledge of surface characteristics, rheology, protein binding, platelet adhesion, and complement and leukocyte activation, all of which culminate in the activation of the clotting cascade primarily via the intrinsic pathway. A reliable animal model is needed to test those advances that show promise during in vitro testing. The rabbit has hemostatic similarities to humans, is less expensive than a large animal, and would allow multiple studies to be conducted simultaneously.5,6 We therefore initially developed a venovenous (VV) rabbit model of extracorporeal circulation using radiolabeled 111Indium-oxine (111In) platelets. Although it was an excellent small animal model that incorporated several aspects of clinical extracorporeal life support (ECLS), there were several challenges, such as cannulation difficulties and pump effects. In addition, there was no oxygenator or filter-type modeling included with this model. To improve reproducibility, decrease loss of animals due to cannulation problems, and incorporate oxygenator or filter components, we progressed to the development of an arteriovenous (AV) model of ECLS in the rabbit.
Materials and Methods
For the VV studies, two rabbits were used for each experiment. The first rabbit was used as a platelet donor for the preparation of 111In-labeled platelets, and the second was placed on bijugular venovenous extracorporeal circulation (ECC) for 4 hours. The AV studies required one rabbit per experiment that was placed on AV bypass for 4 hours. All 2.5–3.5 kg New Zealand White rabbits were initially anesthetized with intramuscular injections of 5 mg/kg xylazine solution (Anased injectable, Ben Venue Laboratories Inc., Bedford, OH) and 20 mg/kg ketamine hydrochloride (Ketaset III, Fort Dodge Animal Health, Fort Dodge, IA). Maintenance IV fluids of lactated Ringer’s were given to maintain hemodynamic stability (10 ml/kg/h). Anesthesia was maintained with a continuous IV ketamine infusion (6 mg/kg/h). Additionally, IV boluses of pancuronium bromide (0.6–0.8 mg/kg/h) were administered to maintain paralysis. The animals received tracheotomies and were mechanically ventilated using a Sechrist Infant Ventilator Model IV-100 (Sechrist, Anaheim, CA). For hemodynamic monitoring and blood sampling, the animal’s right carotid artery was cannulated using a 16-gauge IV angiocatheter (Jelco, Johnson & Johnson, Cincinnati, OH) and monitored using a Series 7000 Monitor (Marquette Electronics, Milwaukee, WI). For platelet donor animals, the surgical preparation was complete at this point.
For the VV study animals, both internal jugular veins and the left external jugular vein were exposed and prepared for cannulation. For the AV study animals, the left carotid artery and right external jugular vein were exposed to prepare for the AV placement of the custom extracorporeal circuit. Body temperature was maintained at 37°C, a pH of < 7.30 was treated with 1–2-mmol/kg infusions of NaHCO3, and a mean arterial blood pressure below 50 mm Hg was treated with 10-ml/kg boluses of isotonic fluid. All of these interventions were performed with approval from the University Committee on Use and Care of Animals regulations.
After surgical preparation, baseline measurements were completed including platelet counts using a Coulter Counter (Coulter Electronics, Hialeah, FL), circulating fibrinogen levels using a BBL Fibrosystem fibrometer (Becton Dickinson, Cockeysville, MD), arterial blood gases using an ABL 505 blood gas analyzer (Radiometer Copenhagen, Copenhagen, Denmark) and an OSM3 Hemoximeter (Radiometer Copenhagen, Copenhagen, Denmark), activated clotting times using a Hemochron Blood Coagulation System Model 401 (International Technidyne Corporation, Edison, NJ), and platelet function using an Optical Aggregometer Model 490 (Chrono-Log Corporation, Havertown, PA). The VV studies also included a free plasma 111In sample counted with a gamma counter (Beckman Gamma 5500, Beckman Instruments, Irvine, CA).7
The VV extracorporeal circuits consisted of a 1 m length of 1/4-inch × 1/16-inch surgical grade Tygon poly(vinyl)chloride (PVC) tubing (Norton Corp., Akron, OH), two polycarbonate downsizing connectors (3/16 × 1/4 inches), and two modified 8-Fr PVC umbilical artery cannulae (Argyle Sherwood Medical, St. Louis, MO). No filters or oxygenators were used as part of the extracorporeal circuits. The circuits were primed with 30 ml Lactated Ringers solution with 5 mmol of NaHCO3. Four groups of animals were tested in the development of the VV model. Two of the groups were sham animals, where the entire surgery was performed and the animals cannulated but not placed on extracorporeal circulation (ECC). Of these 2 groups, one was systemically heparinized and the other was not. The other two groups were ECC groups; one group was systemically heparinized and the other group was run without any anticoagulation. The heparinized animals received a 400 units/kg bolus of heparin and were placed on a heparin infusion to maintain activated clotting times (ACT) over 400 seconds (Heparin-Sodium, Elkins-Sinn, Cherry Hill, NJ). All the animals were injected with radiolabeled platelets through the left external jugular vein. Fifteen minutes after platelet injection, repeat baseline measurements were obtained and then hourly sampling occurred for the 4-hour study time or until time of circuit failure. The ECC animals were then cannulated in under 2 minutes and placed on ECC for 4 hours via a Cobe roller pump. (Cobe, Lakewood, CO). The occlusion on the pump was set using a clinical method that allows for minimum flow through the circuit without pump operation.8 The ECC flow rates were monitored using a Transonics ultrasound flowmeter HT 207 with ¼-inch flow probe (Transonic Systems, Ithaca, NY). The animals were run at flow rates of ≥30 ml/kg/min.
After 4 hours of VV ECC or at the time of circuit failure due to thrombus, the circuit was clamped, the time of study end was recorded as a measure of circuit longevity, and the circuit was rinsed with 300 ml of 0.9% sodium chloride (approximately one rabbit’s total blood volume) through the roller pump at the same flow rate as during the study. The circuit was then cut into 5-cm lengths and placed in a gamma counter (Beckman Gamma 5500, Beckman Instruments, Irvine, CA). Each segment was counted over a period of one minute and the sum of these counts was used to determine the amount of injected radioactivity (i.e., platelets) retained on the circuit. This was used as a measurement of platelet adhesion.
Platelet Labeling Procedure.
For 111In-oxine labeling of the donor rabbit platelets, a technique first described by Thakur et al.9 and then modified by Rand et al.7 and Ardlie et al.10 was modified and used. Approximately 90 ml of blood was drawn from each donor rabbit into 30-ml syringes primed with Anticoagulant Citrate Phosphate Dextrose (CPD) Solution USP (Abbott) at a ratio of 1:9 CPD:blood. The blood was then divided into 3-ml aliquots and placed in a centrifuge at 113g for 18 minutes to produce 45–65 ml of platelet rich plasma (PRP). The harvested PRP was divided into 5-ml aliquots; 1 M sodium citrate was added to each aliquot (1:25, NaCitrate:PRP), and after a 5-minute rest period the PRP was spun at 510g for 18 minutes to produce platelet pellets. These pellets were resuspended and combined in a Hepes-based buffer solution.
The resuspended platelets were then incubated with 111In-oxine at a ratio of 2.0 × 106/mm3:10 μCi 111In per milliliter of resuspended platelets. After a 15-minute incubation period, the platelets were centrifuged at 510g for 18 minutes to produce a platelet pellet. The incubation solution was removed and saved, and the pellet was resuspended in a modified Tyrode’s buffer solution to wash the platelets. Once again the platelets were centrifuged to produce a pellet. The wash supernatant was removed and saved and for the final resuspension, the Hepes-based buffer solution was used. The platelets were then drawn into a 3-ml syringe in preparation for injection into the test animal.
Both the incubation and wash supernatants were used to determine the labeling efficiency, which was expected to be around 60–70%. A higher labeling efficiency results in not only cytosol loading of the platelets but also loading of the granules.7 A platelet count and a gamma count of the platelet suspension were measured before injection and used in the calculation of the percentage of injected radioactivity retained by the circuits. Fifteen minutes after injection of the platelets, a free indium level was drawn and if the plasma free indium was >10%, the radioactivity data was not used in the analysis of the study due to concern for autolabeling of plasma proteins and resultant falsely elevated circuit counts.
The custom 50-cm AV extracorporeal circuit was constructed from a 36 cm length of surgical grade PVC Tygon tubing with a 14GA angiocatheter (Jelco, Medex, Lancashire, UK) for the left carotid artery access and a modified 10-Fr thoracic catheter (Argyle Kendall, Tyco, Mansfield, MA) for infusion into the right jugular vein. The Tygon tubing was stepped up from a 15 cm length of ¼-inch inner diameter (ID) tubing on the arterial side, to a 6 cm length of 3/8-inch ID tubing for the creation of a thrombogenicity chamber to mimic an in-line oxygenator or filter. The circuit was then stepped back down to a 15 cm length of ¼-inch ID tubing on the venous side. These circuits were tested with and without folded PVC sheeting (McMaster-Carr, Aurora, OH) within the chamber. Gish Biomedical polycarbonate connectors were used to join the tubing and catheters.
Animals were not systemically heparinized for these AV studies. The flow through the circuit was monitored with an ultrasonic flow probe and flow meter (Transonic HT207, Transonic Systems). The flows were not controlled for these studies and averaged ≥100 ml/kg/h. After 4 hours the circuit was clamped, removed, and rinsed with 60 ml of 0.9% sodium chloride using a catheter-tip syringe. Any remaining blood or clot was noted and recorded. Two-percent glutaraldehyde solution was injected into the circuit for preservation and eventual scanning electron microscopy (SEM).
All study animals before being killed received a bolus of heparin (400 units/kg) so that necropsy results could be properly interpreted. The animal was then euthanized using a 1-ml bolus of Fatal Plus (Vortech Pharmaceuticals, Dearborn, MI). A postmortem examination was performed on all animals and this included examination of the lungs, heart, liver, and spleen. Any thromboembolic phenomena were noted and recorded.
Univariate analysis for continuous variables was performed using Student’s t test and for categorical variables using chi-squared test. Multivariable analysis was performed using a mixed-model analysis of variance to account for both fixed and random variables. The statistical package used was SAS System for Windows V8.
For the VV studies, 48 New Zealand rabbits divided into four groups were studied. These groups included: systemically heparinized sham animals (HepSham, n = 7), nonheparinized sham animals (NoHepSham, n = 8), systemically heparinized ECC animals (HepECC, n = 18), and nonheparinized ECC animals (NoHepECC, n = 15). The sham animals received the surgical preparation for an extracorporeal circuit, but were not actually placed on ECC. The ECC animals were placed on venovenous extracorporeal circulation. Fewer sham animals were studied than ECC animals because, as final methodology was achieved, the studies concentrated on the ECC animal experiments. All four groups had similar baseline measurements for weight, pH, hematocrit, and mean arterial blood pressure, and there were no significant differences among the four groups with respect to hemodynamics and pH control throughout the 4-hour study period.
Platelet consumption among the sham animals remained stable and did not fall below 80% of baseline platelet values during the course of the study, and there were no significant difference between the two groups. Platelet consumption among the ECC animals showed a 40% decrease in platelet count from baseline after 1 hour on ECC (Figure 1). Platelet adherence, measured by percentage of original injected radioactivity remaining on the circuit surface, was similar for both the heparinized and nonheparinized ECC groups at 2.56% and 2.81%, respectively (Figure 2).
Hourly circulating fibrinogen levels for the sham groups remained stable during the course of the experiment, not decreasing by more than 10% of baseline. The heparinized ECC group had circulating fibrinogen levels similar to those of the sham studies, remaining within approximately 15% of baseline. The nonheparinized ECC animals showed significant fibrinogen consumption (p < 0.001) as the levels dropped steadily throughout the experiment to approximately half of baseline measurements (Figure 3). It should be noted that the sample size for the fibrinogen levels is lower than in other measurement parameters as some samples were lost due to premature thawing.
With regard to circuit longevity, all heparinized ECC animals maintained circuit patency and were alive at the end of the 4-hour study period, with no thrombus formation seen in the circuits. All nonheparinized ECC animals maintained circuit patency beyond the first hour of the study, but over the course of the next 3 hours, 47% of the circuits developed gross thrombus formation with study termination at that time (Figure 4). On necropsy, the only group with evidence of thromboembolic phenomena, primarily within the lung and spleen was the nonheparinized ECC group. All other groups had normal necropsy organ findings.
For the AV model, multiple circuit designs were developed and changes to the design were based upon platelet consumption before finalizing the current model. The previous VV model resulted in a 40% platelet consumption after 1 hour consistent with what is seen clinically, but did not fully imitate clinical ECLS due to the absence of an oxygenator-type device.
A total of 49 New Zealand White rabbits were used in the development of this animal model. Five groups were studied: a standard AV with no chamber (n = 4), a 1st Generation Chamber with variations to a fiber mesh configuration within the chamber (n = 13), a 2nd Generation with variations to a solid PVC sheeting configuration within the chamber (n = 18), a 2nd Generation empty chamber, the final model (n = 6), and a sham group that received the surgical preparation and catheters for the experiment but were not placed on ECC (n = 8).
All animal groups had similar baseline measurements for weight, pH, hematocrit, and hemodynamics. There were no significant differences among the groups with respect to hemodynamic variables and pH throughout the 4-hour study period.
Arteriovenous Circuit Design Results
The AV circuit began as a simple 50 cm shunt of 3/16-inch ID Tygon with a 14-GA angiocatheter for the arterial drainage and a 10-Fr thoracic catheter for the venous reinfusion. This standard AV model with no chamber did not create any gross thrombus in the circuit and did not achieve a platelet consumption level of > 40% of baseline, which was desired for this type of thrombogenicity model. In response to these results, the 1st Generation circuit was developed and included a chamber to increase thrombogenicity and allow the option of adding material within it to increase the surface area. This chamber design was a simple way of mimicking either a filter or an oxygenator. The 1st Generation version used Tygon tubing that stepped up from 3/16-inch ID tubing at the arterial access to ¼-inch ID before the chamber and then to 3/8-inch ID tubing for the chamber itself. This was then stepped down in reverse order to complete the circuit at the venous return cannula (Figure 5).
This design performed reasonably well with a 40% drop in platelet count after 4 hours and with the addition of fiber mesh material within the chamber a drop to 50% of baseline. The difficulty identified with the fiber mesh material was that it would be difficult to evenly coat the mesh with nonthrombogenic polymer coating and maintain the porosity of the material. This particular design also required gluing at the joints and many step-ups (via tubing size changes) allowing for the multiple areas of blood stagnation thereby not truly mimicking an inline filter or oxygenator. The reproducibility was difficult with this design.
The 2nd Generation model eliminated the step-ups and used connectors instead of glue to join the tubing segments (Figure 6). The fiber mesh material within the chamber was replaced with nonporous PVC sheeting to mimic a solid silicone oxygenator type configuration and allow for easier polymer coating application. The PVC sheeting was tested when folded both vertically (short-wise) and horizontally (lengthwise) (Figure 6). Circuit flow was more difficult to establish when the vertical-fold PVC sheeting was used. A 50% drop in platelet count was achieved using either PVC sheeting method. The 2nd Generation model was also tested without any material within the test chamber and, surprisingly, this proved to be the most thrombogenic model. This final iteration was chosen for our model. The final circuit design was a 0.5 m length of ¼-inch ID Tygon before and after a 3/8-inch ID Tygon clotting chamber.
Final AV Circuit Design Results
The 2nd Generation empty chamber (final model) resulted in an almost 70% reduction in platelet count as shown in Figure 7. Sham surgeries that involved no flow with and without heparin had < 25% platelet consumption. This drop in platelets is thought to be due to the surgery itself and the fact that the animals are on a ventilator, received IV fluids, and are anesthetized. Hematocrits did not change significantly during a 4-hour time period and so this is not a dilutional decrease. The platelet function was measured using platelet aggregometry, the results are shown in Figure 8. The absence of shear forces on the platelets of the sham animals can explain the aggregometry results, although the rationale behind the increased function after 2 hours is not fully understood.11 After 4 hours, the 2nd Generation empty chamber group showed a reduction in platelet function to approximately 10% of the baseline. The sham group initially demonstrated an increase in platelet function after the second hour of the study and then dropped to 50% of their baseline function. The SEM pictures further supported the thrombogenicity of this model by showing platelets in various stages of activation with fibrin cross-linking (Figure 9).
Circulating fibrinogen levels (Figure 10) were also measured hourly. The sham group maintained fibrinogen levels over 80% for the 4 hours of the study. The 2nd Generation empty chamber group showed some consumption of fibrinogen to levels of approximately 70% of baseline at the end of the study period. The 2nd Generation empty chamber had ECC flow rates during the experiment that increased steadily until the end of 4 hours. The steady increase in flow is commonly observed clinically in an AV shunt situation.12 Blood flow through the AV circuit increases and so blood flow to the organs is decreased. To accommodate this lack of blood to the organs the cardiac output is continually increasing in order to perfuse the organs. The ECC flow rate ranged from 120 ± 13 cc/min at the start of the study to 210 ± 12 cc/min by the end. A postmortem exam of the heart, liver, lungs, and spleen was performed on all animals. The sham group, as expected, showed no evidence of thromboembolism to any of the organs inspected. All of the study animals, while surviving the full 4 hours on ECC, exhibited multiple emboli within both lungs.
Most of the testing performed to evaluate the usefulness of new nonthrombogenic surfaces or medications such as antiplatelet agents is performed in vitro using a human blood system.13,14 Because many commercial markers are available, qualitative and quantitative data that are specific for activation of platelets and the coagulation cascade can be generated for analysis. Unfortunately, this does not adequately reflect what is going on within a biologic system during ECC because other processes such as leukocyte and complement activation are intricately involved with the activation of platelets and the coagulation cascade.15 There is a need for the development of a small animal model with a thrombogenic profile similar to humans that can be efficiently used to generate data and censor out less promising devices or medications before large animal or human studies. These models have shown themselves to be an effective way to look at blood-biomaterial reactions in vivo based on platelet consumption, platelet adherence, circulating fibrinogen levels, and circuit patency. They both mimic the physiologic changes during clinical ECLS.
With regard to animal choice, rabbits are small, easily instrumented, and have platelets that are physiologically similar to human platelets.5,16,17 In addition, they are even more sensitive to activation by the various stimuli released during ECC, making them ideal for the testing of nonthrombogenic polymers or alternative anticoagulants. The VV model mimicked clinical VV ECC well. The difficulty with it, however, is that it requires technical and surgical expertise in the placement of the venous drainage cannulae such that cavitation of the blood occurs if cannula placement is not perfect.18 As a result, there may be a loss of study animals before ECC initiation, depending upon the instrumenting personnel. In addition, this particular model did not include any oxygenator or filter-type modeling that, clinically, accounts for over 50% of the artificial surface area.17
The AV model has proven to be the most simple of the two models we have designed. It is an easily reproducible screening method to test nonthrombogenic polymers applied to extracorporeal circuits. It is much less labor-intensive than the VV model and requires less surgical expertise. It eliminates the need for a pump, although one could be added, and with the ease of animal management, multiple studies can be performed at one time.
The AV model demonstrated a 70% consumption in platelets and a 90% decrease in their function; therefore, any test polymer that can significantly reduce this consumption and activation would be worth testing in a large animal model for possible clinical consideration. Material can be added to the test chamber to mimic a clinical filter or oxygenator but is not necessary to induce thrombogenicity. The thrombogenicity of the empty chamber may be attributed to areas of stasis within the chamber as well as increased turbulent flows without the horizontally folded PVC sheet to guide blood in a laminar direction. All components of the circuit can be easily coated with nonthrombogenic polymers, and the incorporation of 111In-oxine radiolabeled platelets to provide additional information about platelet adhesion is easily done. The SEM pictures of the circuit surface showed platelets in various stages of activation with fibrin cross-linking, which further illustrates the thrombogenicity of the model.
Overall, either of these models is a reliable and reproducible method by which to evaluate nonthrombogenic surfaces and new compounds. Each provides both qualitative and quantitative measures of outcome. Being a small animal study, they allow for quick, efficient, and reliable information in the screening of new nonthrombogenic biomaterials and/or anticoagulants so that advancement to more intricate and thorough large animal and human studies can occur.
In our experience with small animal models, the AV model is by far the most robust and user-friendly we have used. Modifications to the model can be easily made without complete model redesign, and the realization of the ultimate goal, ECC without anticoagulation, is close at hand with this model.
1. van Oeveren W, Wildevuur CRH: Blood compatibility of cardiopulmonary bypass circuits. Perfusion
2: 237–44, 1987.
2. Gallimore MJ, Heller W, Fuhrer G, et al
: Contact activation, heparins and cardiopulmonary bypass. Thromb Haemost
68: 91–2, 1992.
3. Draaisma AM, Hazekamp MG: Coated versus noncoated circuits in pediatric cardiopulmonary bypass. ASAIO J
51: 663–664, 2005.
4. Espadas-Torres C, Oklejas V, Mowery K, et al
: Thromboresistant chemical sensors using combined nitric oxide release/ion sensing polymeric films. J Am Chem Soc
119: 2321–2322, 1997.
5. Weiss DJ: Comparative physiology of platelets from different species, in Rao HR (ed), Handbook of Platelet Physiology and Pharmacology
, Boston, Kluwer Academic Publishers, 1999, pp. 379–393.
6. Suckow MA, Douglas FA: The Laboratory Rabbit
, New York, CRC Press, 1997.
7. Rand ML, Packham MA, Mustard JF: Survival of density subpopulations of rabbit platelets: Use of 51
Cr- or 111
In-labeled platelets to measure survival of least dense and most dense platelets concurrently. Blood
61: 362–367, 1983.
8. Hensley FR, Martin DE, Graulee GP (es): A Practical Approach to Cardiac Anesthesia
, 3rd ed (1995). Philadelphia, Lippincott William & Wilkins, 2003.
9. Thakur ML, Welch MJ, Joist J, Coleman RE: Indium-111-labeled human platelets: Improved method, efficacy, and evaluation. Thromb Res
9: 345–357, 1976.
10. Ardlie NG, Packham MA, Mustard JF: Adenosine diphosphate-induced platelet aggregation in suspensions of washed rabbit platelets. Br J Haemost
19: 7–17, 1970.
11. Lamba NMK, Cooper SL: Hemostasis and Thrombosis: Interaction of Blood with Artificial Surfaces
, 4th ed. Philadelphia, Lippincott Williams & Wilkins, 661–672, 2001.
12. Haisch CE, Parker F, Brown P: Vascular, in Townsend: Sabiston Text of Surgery
, 17th Ed. Philadelphia, Saunders, 2004, pp. 2089–2090.
13. Thomson C, Forbes CD, Prentice RM: The potentiation of platelet aggregation and adhesion by heparin in vitro
and in vivo. Clin Sci Mol Med
45: 485–494, 1973.
14. Hennessy VL, Hicks Jr. RE, Nierwiarowski S, et al
: Function of human platelets during extracorporeal circulation. Am J Physiol
232: H622–H628, 1977.
15. Gorbet MB, Sefton MV: Biomaterial-associated thrombosis: roles of coagulation factors, complement, platelets, and leukocytes. Biomaterials
25: 5681–5703, 2004.
16. Poole AW: The study of platelet function in other species, in Watson SP, Authi KS (eds), Platelets: A Practical Approach
, New York, IRL Press at Oxford University Press, 1996, pp. 341–350.
17. Plotz FB, Wildevuur WR, Wildevuur CRH, et al
: Platelet consumption during neonatal extracorporeal life support (ECLS). Perfusion
7: 27–32, 1992.
Copyright © 2006 by the American Society for Artificial Internal Organs
18. Annich GM, Meinhardt JP, Mowery KA, et al
: Reduced platelet activation and thrombosis in extracorporeal circuits coated with nitric oxide release polymers. Crit Care Med
28 4: 915–920, 2000.