Ventricular assist devices (VADs) have proven successful in extending and improving heart failure patients’ quality of life. Although cardiac transplantation remains the preferred treatment for end-stage heart failure, the marked disparity between the number of donor hearts and potential transplant recipients1 has led to the increased usage of VADs over the past few decades.2 With each generation of devices, VAD developers and researchers are presented with new challenges to make these devices safer and more effective than the last. Although each generation of VAD represents advancements in device technology, the underlying issue of thrombosis in all blood-contacting devices is a consistent problem.3–11
Three major phenomena contribute to thrombosis: alterations in blood composition, alteration in blood-contact surfaces, and alterations in blood flow.12 Implantation of VADs inherently provides opportunities for each of these three components to manifest. During implantation procedures, patients are typically put on anticoagulation therapy (heparin, warfarin sodium, etc.) to decrease the risk of creating a thrombus during and immediately after the surgical procedure.13 In addition, the introduction of the device will cause alterations of blood flow, introduce new areas of stasis and turbulence, and introduce a nonendothelial surface. The rotating components in these devices provide a shear force on the blood components,14 and the device itself provides a surface for platelets to adhere and aggregate on.15 Additional factors such as altered endothelial surfaces, infections, and other patient variables4 may add to the causes of thrombosis in patients with VADs. Regardless of thrombus size or origin, if embolized to the pump, the brain or other down-stream organs, the outcomes could include pump failure, disability, or death.4 Furthermore, post-VAD thrombosis could lead to VAD exchange surgery.
During pathology evaluations of explanted clinical VADs, all external and internal surfaces are closely examined for any evidence of thrombi.16 Thrombi along the internal components of VADs (e.g., impeller, flow straightener, diffuser, valve leaflets, etc.) exhibit morphological characteristics different than thrombi formed in non-VAD regions (e.g., atria or ventricles, Figure 1). Rather than appearing as loose, randomly oriented, eosinophilic strands with abundant enmeshed erythrocytes, thrombi in VADs with rotating components contain irregular bands of eosinophilic, poorly cellular material with rare erythrocytes. These changes are primarily caused by the rotating component within the device (i.e., impeller) applying shear forces on the thrombus. Thrombus origin is identified by histologic examination of the specimen; however, in some cases, it is not possible to determine (based on integrity of specimen after retrieval). Thrombi may form along the wall of the ventricle (mural thrombus), which is common among patients with dilated cardiomyopathy, in the atrial appendage, or in some cases within the VAD. In the case of thromboembolism, thrombi formed outside the device travel into the pump and remain there or pass through and cause ischemic damage downstream. Whether or not the thrombus formed inside the device, if it remains it will be exposed to the rotational forces of the impeller and have a different microscopic morphology than thrombi formed in non-VAD regions.
Clot and thrombus formation have been studied in vitro with both continuous and pulsatile flow conditions.17–19 These studies characterize the thrombus formation morphology in specific locations and geometries based on length, speed, and type of flow and material surface. Taylor et al. demonstrated that areas of high wall shear stresses leads to larger thrombus formation, and multiple studies have shown areas of low flow are more likely to cause platelet activation leading to formation of loosely adhered thrombi.14,15,18,20 Loosely adhered thrombi in vivo have a greater potential to dislodge and introduce thromboembolism. Introducing a textured surface in the blood has been shown to reduce platelet adhesion and thus reduce thrombus formation; however, thrombus formation is not completely diminished.15,18
Ventricular assist devices are typically tested before preclinical trials with in vitro circulatory loops;21,22 however, these are designed to specifically test the hemodynamic conditions of the device. Precise VAD response to thromboembolism needs to be studied in a controlled in vitro setting where specific pump parameters (i.e., power consumption, flow rates, impeller RPM) can be monitored while various types of thrombi are introduced. In this article, we propose a novel method for creating standardized fibrin thrombi that could be introduced into a mock circulatory loop for testing VAD response to pass-through thromboembolism.
Materials and Methods
Normal equine blood was originally collected for transfusion in commercial 450 ml blood storage bags containing sodium citrate and stored at 5°C. Blood stored beyond its usable date for transfusions was made available to this experiment. The blood in all experiments mentioned in this article was collected from the same donor approximately 3 months before use.
Approximately half of the citrated whole blood was collected in 5 ml aliquots. The remaining blood was transferred from the blood bag to a plastic container with a screw-top lid and stored at 5°C. Because of the high sedimentation rate of equine blood,23,24 the cells collected at the bottom of the container and the cell-free plasma top layer was collected using pipettes and stored in 5 ml aliquots. For superior red blood cell sedimentation, it is best to let the blood sit undisturbed for approximately 24 hours before removing the plasma layer.25 To preserve the blood components and ensure success of creating consistent thrombi, 5 ml volumes of whole blood and plasma were put into separate vials and stored in a 5°C refrigerator and −20°C freezer, respectively. Approximately 24 hours before plasma clot creation, the plasma was transferred to the 5°C refrigerator and allowed to thaw.
Titration of CaCl2
To initiate clotting by counteracting the sodium citrate effect, calcium chloride (CaCl2) was added to the blood. Approximately 5 ml each of either equine whole blood or plasma were collected into separate test tubes. Granular, dihydrate CaCl2 (Macron Fine Chemicals, Center Valley, PA) was dissolved in distilled water at a concentration of 12 mg/ml.26 Fifty microliter aliquots of CaCl2 solution were added to each test tube, once per minute. After each aliquot was added, the test tube was tilted to test the coagulation of the blood and plasma. Once the test tube could be tilted at a 45° angle without disruption of the liquid, the titration was complete and the time and total volume of CaCl2 added were recorded and used for subsequent experiments.
Adding the Rotational Component
To mimic the forces and conditions inside a VAD in vivo, small plastic Petri dishes (Corning, Inc., 38 mm diameter) were affixed to the shafts of electric and gear motors (four total) with 3M Dual Lock Reclosable Fastener. This apparatus was arranged so the Petri dish rotated inside a larger stationary Petri dish (Corning, Inc., 50 mm diameter) with an adjustable gap between the dishes (Figure 2). Each motor was powered by a variable power supply ranging from 0 to 12 V and produced speeds ranging from 75 to 300 RPM.
Creation of Thrombi
Multiple clots were created with the setup shown in Figure 2 testing the following parameters: 1) CaCl2 concentration, 2) gap size between rotating disk and stationary reservoir, 3) RPM, and 4) duration of rotation. For all experiments, both the rotating and stationary Petri dishes were lined with disposable aluminum foil. After each experiment, clots remained stationary for approximately 18 hours to maximize clot condensation. Clots were then carefully removed from the aluminum foil and placed in formalin for at least 24 hours. Formalin-fixed clots were then sectioned to fit in a 2.5 × 2.5 cm cassette and processed for traditional paraffin histology and scanning electron microscopy (SEM) to qualitatively characterize the microstructure.27 Histology sections were stained with hematoxylin and eosin (H&E). Clots examined by SEM were dehydrated in progressively increasing concentrations of alcohol (after 24 hour formalin fixation), coated with gold using a Ted Pella 108 Manual Sputter Coater, and finally imaged using a Tabletop Hitachi TM3000 Scanning Electron Microscope.
Adjusting the Concentration of CaCl2
To determine the optimal volume and concentration of CaCl2 necessary for clotting of citrated whole blood and plasma, thrombi were created using two concentrations of CaCl2: 12 and 24 mg CaCl2/ml distilled water. This will also reveal if different concentrations of CaCl2 introduce artifactual histologic differences in the morphology of the clots. Approximately 5 ml of whole blood and plasma were collected in stationary test tubes; 0.15 ml of CaCl2 was added every 3 minutes until clotting was observed. The clotting blood and plasma remained stationary for approximately 18 hours to maximize clot condensation before collection and formalin fixation.
Adjusting the Gap Height
The distance between the rotating disk and stationary reservoir was set at approximately 6 mm (±1 mm) and 4 mm (±0.2 mm). The distance was set by raising the rotating component and measured with calipers. The motors ran for approximately 6 hours, then the resulting thrombi were collected and fixed in formalin.
Adjusting the Speed and Duration of Rotation
In this set of experiments, the motor speed and total time rotating were varied. For this set, all four motors were attached to power sources with similar distances between the rotating disk and stationary reservoir (4 mm ± 0.2 mm). Motors were set to 75, 225, 240, and 300 RPM and ran for approximately 6–48 hours. After the allotted time period, the clots were left undisturbed for approximately 4–6 hours before collection and formalin fixation.
Clot Image Analysis
H&E-stained sections of clot were photographed using a microscope camera and images were read by a program in MATLAB (MathWorks, Inc., Natick, MA) to analyze the clot density by determining the percentage of pixels in the picture that contained clot. Each image analyzed contained a consistent square area of clot. In addition, area and length measurements were collected using open-source imaging software (ImageJ, National Institute of Health).
Calcium Chloride Concentration
The volumes and concentrations of CaCl2 needed to initiate clotting in whole blood and plasma are represented in Table 1.
Using a relatively dilute CaCl2 solution required the addition of a greater volume of solution to the blood and plasma to facilitate clotting. The liquid completely clotted in the test tube after addition of CaCl2. Grossly, the clots were cylindrically shaped, gelatinous, either yellow (plasma) or dark red (whole blood) material. Figure 3 shows the H&E-stained sections of each clot (whole blood and plasma) at different concentrations of CaCl2. Figure 3A, B (whole blood) shows a dominance of red blood cells, whereas Figure 3C, D (plasma) shows primarily a uniform network of fibrin. Histologically, there are no apparent differences between the clots created from different CaCl2 concentrations. Computational image analysis in MATLAB also reveals no significant differences between the two clotting methods; images of randomly selected areas of each clot had a density (measured as the percentage of colored pixels) that only varied by less than 3.10% for both plasma and whole blood clots. Thus, a higher concentration of CaCl2 was used for the remaining experiments because the smaller stationary reservoirs limited the amount of liquid that could be added.
When using whole blood in the setup shown in Figure 2, the blood tended to clot in separate sections loose within the stationary reservoir. The largest clot section created in each trial ranged in size from 6–7 mm × 10–14 mm regardless of height of the rotating component. Figure 4 shows the resulting histology of whole blood clots created with varied distances between the rotating and stationary components. Figure 4A (6 mm gap) and Figure 4B (4 mm gap) show two distinct areas, one consisting of abundant erythrocytes and the other of fibrin enmeshed erythrocytes and leukocytes. These clots, formed at low shear stresses (75 RPM), are relatively isotropic, and there are no apparent differences between the two gap widths. Figure 4C (6 mm gap) shows a more uniform section of a fibrin mesh with embedded erythrocytes, whereas Figure 4D (4 mm gap) shows fibrin strands arranged into compact layers of common directionality with scattered erythrocytes. These clots, formed at higher shear stresses (225 RPM), indicate a smaller clearance between the rotating surfaces will create greater morphological changes that more closely resemble ante-explant thrombi found in clinically used VADs (Figure 1).
Speed and Duration of Rotation
When using plasma in the setup shown in Figure 2, the blood tended to clot as two pieces: one circular gelatinous yellow clot loose within the stationary reservoir and one or more sections adhered to the rim of the stationary reservoir. In rare cases, the clot adhered to the rotating component. As the speed of the rotating component increased, the size of the clot created decreased from approximately 40 to 20 mm in largest dimension. As the time of rotation increased, the gross appearance of the clot changed from a gelatinous, glistening material of larger size to a dry, brittle, dull material smaller in size.
Figure 5 shows H&E-stained histology sections of plasma clots created with varied speeds of the rotating component and increasing lengths of duration. As the motors’ run-time and speed increased, the resulting thrombi became more layered and compact. When the motor ran less than 10 hours, the shear forces from the motor at low RPMs did not have a great effect on the thrombi, which resulted in layers spread further apart (maximum width of 33 μm; Figure 5A, D). However, as the device rotated for extended periods of time, the layers of the thrombi were compacted. This resulted in a tight, more uniform, and acellular appearance as seen in the 24 hour (Figure 5B, E, H) and 48 hour runs (Figure 5C, F, I). By increasing the time of rotation from 24 to 48 hours at 225 RPM (Figure 5E, F), the clot density increased by 3.45% (measured by percentage of colored pixels). The gaps between the layers of fibrin strands decreased from an average length of 11 μm (Figure 5B) to 5–8 μm (Figure 5H, C) until they were essentially nonexistent (Figure 5F, I) as the time of rotation and RPM increased.
Figure 6 shows SEM images of plasma clots created from slow speeds with short duration (Figure 6A, B) and faster speeds with longer duration (Figure 6C, D). In thrombi created at lower speeds and in shorter lengths of time (Figure 6A, B), fibrin strands begin to align into layers. As the motor speed and length of time increased (Figure 6C, D), these layers became more compact.
In Vivo Thrombi Versus In Vitro Clots
The implantation of a VAD inherently poses a risk for thrombus formation. Because thrombi can damage the pump or cause infarction, when a VAD controller log indicates a suspected thrombus,5,28,29 the physician must pursue either anticoagulation therapy or VAD exchange surgery.
Thrombi may form inside the VAD because of a combination of shear forces from the moving components, areas of stasis within the device, or the blood–device interaction. In addition, thrombi may form outside the VAD, such as around the sewing ring, along the ventricular wall, in the atrial appendage, etc. In some cases, thrombi that developed outside the device may travel into the VAD and become a nidus for further thrombus formation inside the VAD. Regardless of origin, if a thrombus adheres to some part of the VAD, it will cause an increase in friction, which leads to the increase in power consumption denoted on the controller log. A histological analysis of the material within the device can distinguish postexplant from ante- and peri-explant materials and, based on the morphology, determine the potential origin of the material.16 Typically in postexplant VAD evaluations, the thrombi discovered are clots that either originated upstream from and traveled through the device or formed inside the device (or some combination of the two). The creation of artificial thrombi with similar features and morphologies of in vivo thrombi as discussed in this article will help elucidate the interaction of VADs and “pass-through” thromboemboli, which will potentially help VAD manufacturers and physicians create and implement devices that can reduce the risks associated with VAD therapy.
Many histological similarities are observed when comparing the clots produced using the technique discussed in this article to the thrombi seen in vivo. The two most influential factors for recreating thrombi with abundant layers were time and speed of rotation. No rotation applied to the blood while it was clotting resulted in a postexplant blood clot (Figure 7A, B). The greatest difference between these two clots is the amount of erythrocytes per 243 mm2 area, which differs by less than 10%; this variation is expected because of the innate differences between equine and human blood. A small duration of rotation resulted in a clot that resembled the periexplant fibrin thrombus with similar fibrinous layers (Figure 7C, D) that varied in density by less than 7%. A faster rotational speed resulted in a clot with dense layers, which resembled the organized thrombus seen in vivo (Figure 7E, F). These specimens are histologically similar; the only noticeable difference is that the eosinophilic material in the in vitro clot (Figure 7E) stained more vividly than the in vivo thrombus (Figure 7F).
Plasma Versus Whole Blood
In this article, we described a method to create both plasma and whole blood-derived clots. Although it would be ideal to use whole blood exclusively, plasma has benefits that might make in vitro experiments more attainable. Plasma, as opposed to whole blood, created clots that showed greater histological uniformity and were more consistently reproducible. Plasma also allowed for better visualization of the fibrin strands in comparison to whole blood (i.e., lacked enmeshed erythrocytes). In addition, because of plasma’s ability to be frozen without losing its utility, it is much easier to preserve long-term, which readily allows for future experiments without acquiring and titrating new blood samples. However, for in vitro experiments that study biochemical reactions between the device and the blood, whole blood clots would be necessary to ascertain the reaction of blood components (i.e., platelets, leukocytes, and RBCs).
Platelets play an important role in coagulation, particularly inside a VAD. The shear stress from device-induced flow activates platelets, which leads to aggregation of platelets, and eventually thrombus formation. The plasma used in these experiments contained little to no platelets, meaning the coagulation came from the plasma proteins. The lack of platelets caused the clot that formed to be mostly gelatinous, and in a flat circular shape mostly between the two plates. However, in the whole blood samples, the platelets were activated and the resulting clot was irregular in shape and formed in clumps rather than flat disks.
This Method Versus Others
The Chandler Loop is a historic method for creating thrombi in vitro.19 This method uses an inclined rotating closed loop that circulates blood until it coagulates, thus producing a thrombus. These thrombi are grossly and histologically comparable with vascular thrombi,19 whereas the method described in this article focuses on creating clots similar to what would be seen within a medical device that has a rotating component. The rotating mechanism within the device applies different forces to the blood components not seen in the body naturally.
Other previous studies that created clots in vitro studied the adhesion characteristics of certain materials used in medical devices,18 the potential for thrombi to form on these surfaces, how the shear rates applied by the rotating component affected the thrombus formation,15,18 and how certain geometries created a potential for thrombus formation.14 In the method described in this article, the end goal is to have an artificial clot that can be introduced into a mock circulatory loop for the purposes of studying thromboembolism in medical devices, specifically VADs.
Only equine blood was made available for this experiment, whereas bovine, porcine, or ovine blood is typically used for VAD development. Equine blood has a higher sedimentation rate23,24 compared with that of humans; however, a high sedimentation rate in humans can be indicative of an inflammatory reaction in the body, which might be possible in VAD patients. Porcine blood is most similar to human blood in adhesive forces,30 whereas bovine, ovine, and porcine specimens are most often used for VAD testing because of size and other physiologic similarities. Although equine blood has differences to human blood, the authors believe using blood from different species would yield similar results, with potential differences in the titration curve.
The parallel plates used in these experiments had a gap clearance of 4–6 mm (±1 mm), whereas distances between the impeller and adjacent VAD housing can be much smaller (<1 mm). In addition, the parallel plate setup described in this manuscript most closely resembles a centrifugal-flow device (as opposed to a uniaxial-flow VAD orientation). The goal of this experiment was to add a rotational force to the blood while it was clotting to show the morphological changes compared with a clot formed under static conditions. Future testing could be performed with real VAD-like conditions (closer gap sizes, and multiple orientations of the rotating component) to further describe thrombus formation in an environment with rotational forces.
Future applications of this method include characterizing flow rate and pressure changes of VADs in response to “pass-through” thromboembolism. Such a study would include a mock circulatory system incorporating a VAD, pressure transducers, and a flow meter. Previously citrated blood specimens would be titrated with CaCl2 as described here, and then clots created in a static environment would then be introduced to the system proximal to the VAD. The resulting clots would then be compared histologically with the clots created with a rotating component as described in this article, as well as those seen in vivo. This test would determine the pressure changes and changes in flow rate as a clot travels into the VAD and potentially correlate the size/age of the clot to magnitude of the power-spike seen on a VAD controller log. This flow loop would require a clear liquid, such as water or a water/glycerin mixture, to be able to visualize the clot as it travels into the device. Future studies could also be performed to learn more about thrombus formation inside of VADs as well. In this case, the flow loop would need to be filled with previously citrated whole blood (or an appropriate blood analog), and varying amounts of CaCl2 could be added to the system (as blood is flowing) to allow blood to clot as it would in vivo according to shear forces inside the VAD and geometrical constraints. Furthermore, in vivo studies could be performed where real-time VAD parameters are monitored while inducing instances of thromboembolism by inserting previously formed clots into the ventricle.
Based on the numerous variables tested, we were able to optimize a method to create artificial thrombi similar to those seen in blood-contacting devices with rotating surfaces. In all experiments, the rotating disk and the stationary reservoir were similar in size, with the rotating disk slightly smaller. We believe this caused the thrombi to form almost completely under the rotating disk where they experienced the most shear force and allowed minimal stagnation. A smaller distance between the rotating dish and stationary reservoir (approximately 4 mm) allowed formation of a clot large enough to withstand processing and handling with histologic evidence of lamination. Because of sample size constraints, 24 mg CaCl2/ml distilled water was used because less volume could be added to achieve the same amount of clotting for whole blood and plasma.
Ultimately, this standardized method for creating in vitro whole blood and plasma clots will be used for additional studies involving physiologic flow loops to further elucidate the interaction of VADs with fibrin thromboemboli. These experiments will establish a basis for better understanding of blood flow, thrombus formation, and embolization within VADs, which will help future VAD designs in limiting thromboembolic complications.
1. Orians CE, Evans RW, Ascher NL: Estimates of organ-specific donor availability for the United States. Transplant Proc 1993.25(1 Pt 2): 15411542.
2. Slater JP, Rose EA, Levin HR, et al.: Low thromboembolic risk without anticoagulation using advanced-design left ventricular assist devices. Ann Thorac Surg 1996.62: 13211327; discussion 1328.
3. Levinson MM, Smith RG, Cork RC, et al.: Thromboembolic complications of the Jarvik-7 total artificial heart: Case report. Artif Organs 1986.10: 236244.
4. Icenogle TB, Smith RG, Cleavinger M, et al.: Thromboembolic complications of the symbion AVAD system. Artif Organs 1989.13: 532538.
5. Eckman PM, John R: Bleeding and thrombosis
in patients with continuous-flow ventricular assist devices. Circulation 2012.125: 30383047.
6. Goldstein DJ, John R, Salerno C, et al.: Algorithm for the diagnosis and management of suspected pump thrombus. J Heart Lung Transplant 2013.32: 667670.
7. Muthiah K, Robson D, Macdonald PS, et al.: Thrombolysis for suspected intrapump thrombosis
in patients with continuous flow centrifugal left ventricular assist device. Artif Organs 2013.37: 313318.
8. Najjar SS, Slaughter MS, Pagani FD, et al.; HVAD Bridge to Transplant ADVANCE Trial Investigators: An analysis of pump thrombus events in patients in the HeartWare ADVANCE bridge to transplant and continued access protocol trial. J Heart Lung Transplant 2014.33: 2334.
9. Pratt AK, Shah NS, Boyce SW: Left ventricular assist device management in the ICU. Crit Care Med 2014.42: 158168.
10. Uriel N, Han J, Morrison KA, et al.: Device thrombosis
in HeartMate II continuous-flow left ventricular assist device: A multifactorial phenomenon. J Heart Lung Transplant 2014.33: 5359.
11. Hasin T, Deo S, Maleszewski JJ, et al.: The role of medical management for acute intravascular hemolysis in patients supported on axial flow LVAD. ASAIO J 2014.60: 914.
12. Dickson BC: Venous thrombosis
: On the history of Virchow’s triad. Univ Toronto Med J 2004.81: 166171.
13. Spanier T, Oz M, Levin H, et al.: Activation of coagulation and fibrinolytic pathways in patients with left ventricular assist devices. J Thorac Cardiovasc Surg 1996.112: 10901097.
14. Hochareon P, Manning KB, Fontaine AA, Tarbell JM, Deutsch S: Correlation of in vivo
clot deposition with the flow characteristics in the 50 cc penn state artificial heart: A preliminary study. ASAIO J 2004.50: 537542.
15. Milner KR, Snyder AJ, Siedlecki CA: Sub-micron texturing for reducing platelet adhesion to polyurethane biomaterials. J Biomed Mater Res A 2006.76: 561570.
16. Carpenter BC, Gonzalez CG, Jessen SL, et al.: A brief review of ventricular assist devices and a recommended protocol for pathology evaluations. Cardiovasc Pathol 2013.22: 408415.
17. Taylor JO, Witmer KP, Neuberger T, et al.: In vitro
quantification of time dependent thrombus size using magnetic resonance imaging and computational simulations of thrombus surface shear stresses. J Biomech Eng 2014.136:,111.
18. Navitsky MA, Taylor JO, Smith AB, et al.: Platelet adhesion to polyurethane urea under pulsatile flow conditions. Artif Organs 2014.38: 10461053.
19. Chandler AB: In vitro
thrombotic coagulation of the blood; a method for producing a thrombus. Lab Invest 1958.7: 110114.
20. Topper SR, Navitsky MA, Medvitz RB, et al.: The use of fluid mechanics to predict regions of microscopic thrombus formation in pulsatile VADs
. Cardiovasc Eng Technol 2014.5: 5469.
21. Timms D, Hayne M, McNeil K, Galbraith A: A complete mock circulation loop for the evaluation of left, right, and biventricular assist devices. Artif Organs 2005.29: 564572.
22. Pantalos GM, Koenig SC, Gillars KJ, Giridharan GA, Ewert DL: Characterization of an adult mock circulation for testing cardiac support devices. ASAIO J 2004.50: 3746.
23. Geddes LA, Sadler C: The specific resistance of blood at body temperature. Med Biol Eng Comput1973.11: 336339.
24. Knisely MH, Bloch EH: Microscopic observations of the circulating blood of nine healthy normal horses, all of which had unagglutinated circulating blood cells and high in vitro
erythrocyte sedimentation rates; a contribution to the theory and general understanding of the pathologic circulatory physiology of sludged blood. Am J Med Sci 1950.219: 249267.
25. Eicker SW, Ainsworth DM: Equine plasma banking: Collection by exsanguination. J Am Vet Med Assoc 1984.185: 772774.
26. Roessler FC, Ohlrich M, Marxsen JH, et al.: The platelet-rich plasma clot: A standardized in-vitro clot formation protocol for investigations of sonothrombolysis under physiological flows. Blood Coagul Fibrinolysis 2011.22: 407415.
27. Joy DC: Scanning electron microscopy. Materials Science and Technology. Wiley-VCH Verlag GmbH & Co KGaA, 221249.
28. Chorpenning K, Brown MC, Voskoboynikov N, Reyes C, Dierlam AE, Tamez D: HeartWare controller logs a diagnostic tool and clinical management aid for the HVAD pump. ASAIO J 2014.60: 115118.
29. Bashir J, Cheung A, Kaan A, Kearns M, Ibey A, Ignaszewski A: Thrombosis
and failure of a HeartMate II device in the absence of alarms. J Heart Lung Transplant 2011.30: 11971199.
30. Weng X, Cloutier G, Pibarot P, Durand LG: Comparison and simulation of different levels of erythrocyte aggregation with pig, horse, sheep, calf, and normal human blood. Biorheology 1996.33: 365377.