Circulatory support devices have progressed to the state of being therapeutic options for destination therapy. Left ventricular assist devices (LVADs) and total artificial hearts are now totally implantable with no percutaneous leads and are capable of functioning reliably in excess of 2 years.1 In addition, fully implantable devices provide a good quality of life and a significantly reduced level of sepsis. However, thromboembolic complications associated with types of devices remain a major obstacle to be overcome to make these devices function in a manner similar to modern heart valves and vascular prostheses.2
Thromboembolic complications can be reduced by two approaches. The first approach involves pharmacologically modifying the coagulation system. The second approach involves the modification of the blood-contacting material within the device. This latter approach has the advantage of fewer systemic effects than those arising from lifelong anticoagulation therapy, such as excessive bleeding and the problems associated with maintaining the proper amount of anticoagulant.3,4
Textured blood-contacting materials have been used clinically in circulatory assist devices since the early 1960s.5–7 These materials are designed to encourage the formation, development, and adherence of a stable, biological lining (neointimal layer) by entrapping platelets, red blood cells, fibrin, and other blood-borne materials.8 An example of a textured-blood contacting material is illustrated in Figure 1. This type of material is referred to as an integrally textured fiber surface and is composed of small, attached fibers that form the boundaries of connected cavities. This surface is incorporated as a blood-contacting surface in the HeartMate (Thoratec Corporation, Pleasanton, CA), a pulsatile diaphragm pump that is available as either an implanted pneumatic (IP) or a vented electric (VE) version. The integrally textured fiber surface is believed to be the key to the low thromboembolic rates that are reported with these devices (2.7–6%). These rates (obtained without anticoagulation) are clinically lower than the thromboembolic rates in other cardiac assist devices that require anticoagulation.6,9,10 However, there is a lack of understanding of how the size and distribution of the cavities created by these fibers affect the growth and stability of the neointimal layer.
Although circulatory support devices based on this principle are in use clinically, there is still a critical gap in understanding how the topography of the blood-contacting surface affects the formation of this neointimal layer. In addition, there is limited published information that defines the critical size of the topographic features for the formation of the neointimal layer.
We have developed a particle-cast surface that uses dissolvable particles to form the cavities. These surfaces were incorporated in a circulatory support device and implanted in calves. The preliminary results are presented here.
Materials and Methods
The microtextured materials were fabricated and incorporated within a circulatory support device and the blood pump was then implanted in a calf. Following necropsy, the microtextured material was examined visually and with light microscopy and scanning electron microscopy techniques.
Fabrication of Microtextured Materials
The microtextured materials were fabricated with microcasting technology that has been developed at the Penn State College of Medicine. The materials were fabricated by first repeatedly dipping a diaphragm mold into a solution of BioSpan (The Polymer Technology Group, Berkeley, CA) MS/0.4 segmented polyurethane and dimethyl acetamide (DMAC) to achieve the desired diaphragm thickness. Dissolvable salt particles were then deposited onto the wet polyurethane immediately after the last coating. After curing was complete, the diaphragm was rinsed with water to dissolve the salt particles and separated from the mold. The thickness of the porous layer can be controlled by altering the proportion of BioSpan MS/0.4 to DMAC.
The dissolvable particles were serially sifted through an array of sieves to isolate the desired particle size, which directly affects the size of the pores (cavities) in the finished surface. For instance, the surface shown in Figures 2A and 2B was fabricated by using particles retained between the 106 and 125 μm sieves, whereas the surface shown in Figures 2C and 2D was fabricated by using particles retained between the 212 and 250 μm sieves.
The main advantage of the particle-cast blood-contacting surface over integrally textured fiber surfaces (as illustrated in Figure 1) is greater control over the size and depth of the cavities through the selection of the salt particles. In order to demonstrate this control, the pore areas and porosities were measured for two types of salt-cast surfaces: 106 to 125 μm and 212 to 250 μm. The use of larger particles resulted in larger pore areas, as demonstrated by the increase in pore area size from the 106 to 125 μm samples to the 212 to 250 μm samples (data not shown).
Fourier transform infrared attenuated total reflection (FTIR-ATR) analyses were conducted on samples of the particle-cast materials and on samples from a smooth polyurethane material (control) that is typically used in circulatory support devices developed by our group.11–13 The spectra showed no differences between the particle-cast materials and the smooth polyurethane. This demonstrated that no salt residue remained on the particle-cast surface.
The blood pump was a modification of a previous design that was developed by our group under the Innovative Ventricular Assist System program sponsored by NHLBI (NHLBI-HV-94-25). The pump is actuated by a pusherplate.14 The energy converter consists of a brushless DC motor that rotates the nut of a custom rollerscrew. The nut rotates resulting in translation of the rollerscrew screw that is attached to the pusherplate. The pusherplate compresses the diaphragm during pump systole. During pump diastole, the pump fills passively, and the blood pump diaphragm moves toward the pusherplate. Prosthetic heart valves regulate flow in and out of the blood pump. Tilting disk mechanical heart valves (Björk-Shiley Monostrut with Delrin occluders, Arrow International, Reading, PA) are used to control flow direction through the pump. A size 27 valve is used in the inlet position and a size 25 valve is used in the outlet position. These mechanical valves were chosen for this pump because of their successful clinical use in circulatory support devices.11,15,16 The particle-cast diaphragm was placed between the pusher plate and the pump case. The pump case was fabricated from titanium, and three types of blood-contacting surfaces were used on the titanium case for the in vivo studies: (1) smooth, unpolished (machined) titanium surface; (2) smooth, polished (high gloss finish) titanium surface; and (3) textured surface composed of sintered titanium microspheres (75 to 100 μm) (used by the HeartMate LVAS17,18).
In Vivo Evaluation
The in vivo evaluation of neointimal formation upon microscopically textured surfaces was accomplished by implanting the blood pump as an apical left ventricular to descending aorta assist device with a particle-cast diaphragm. The device was placed in male Holstein calves. This animal model has been extensively used for over 20 years by our group to evaluate LVADs.12,14
The animals were maintained on postoperative anticoagulant therapy. Animals received heparin postoperatively to maintain an activated clotting time (ACT) between 225 and 300 seconds until Coumadin became effective. The Coumadin dose was adjusted so that the prothrombin time (PT) was 1.5 to 2.0 times greater than the control for the specific animal.
The majority of the animals were electively killed. While the animal was alive and the blood pump was pumping, the animal was anesthetized. A catheter was surgically placed in the inlet cannula to the assist pump. While the pump continued to run without interruption, the blood sac was flushed with a saline solution. The animal was then euthanized. The pump and its various cannula were rinsed gently with saline and disassembled. The blood pump was then removed, and carefully disassembled, opened, and photographed. The diaphragm was then placed in 2.5% glutaraldehyde for fixation after visual inspections were performed. After fixation was complete, samples from the microtextured diaphragm and neointima were obtained from different regions of the diaphragm for microscopic analysis.
The structure of the neointima was evaluated using light and scanning electron microscopy (SEM). Histological sections were stained with hematoxylin and eosin (H&E) and viewed by conventional light microscopy. Mason’s trichrome stain was used to identify the presence of collagen. High-resolution imaging of the explanted neointimal surfaces was performed by SEM. Small samples were cut from the explanted diaphragms and subjected to ethanol dehydration sequentially in 50%, 60%, 70%, 80%, 90%, and 100% ethanol solutions for 10 minutes each. Samples were allowed to air dry and then mounted to SEM sample stubs using carbon tape. Samples were sputter-coated with 10 nm of either gold or gold-palladium. A small drop of liquid silver was placed on one edge of the sample to ensure that there was electrical contact between the sample surface to be imaged and the SEM stub. Imaging was performed on either an AmRay 3200C EcoSEM (Bedford, MA) or a JEOL 6700F (Tokyo, Japan).
A total of five implants were performed with the particle-cast surfaces. These implants are summarized in Table 1.
The pumps were implanted in male Holstein calves weighing between 160 and 240 pounds. The animals typically began receiving therapeutic doses of Coumadin on postoperative day 1. All the calves appeared to be normal, healthy, growing animals with no health problems at the time of the elective termination for implants 2 to 5.
No thrombus was noted on the inlet cannula for any of the implants. The aortic anastomosis was always patent with no thrombus. The aortic graft and the suture line on the graft-to-graft anastomosis were generally free of thrombus. Multiple renal infarcts were grossly observed in both kidneys for all of the implants. The large and small intestines, lungs, liver, pancreas and spleen appeared to be normal for all of the implants.
As stated earlier, three types of surfaces were used on the titanium case for these preliminary in vivo studies. Implant 1 used a smooth, unpolished (machined) titanium surface. However, red thrombus formation was observed on this surface at explant. This problem was addressed by polishing the pump surface to a high gloss finish for implants 2 to 4. No thrombus formation was observed on this polished surface. Implant 5 used a textured surface composed of sintered titanium balls. This type of surface is used by the HeartMate (as described previously).
Figures 3 to 12 are representative of the five particle-cast implants. Figure 3 shows the interior of the pump from implant 3 (106 to 125 μm particles) with the diaphragm and case exposed. No thrombus was observed on the polished case. The color of the neointima was pale-red and white. A red ring was typically observed around the junction of the particle-cast diaphragm and the titanium pump case. Although thrombus was observed on the unpolished pump case for implant 1, no thrombus was observed on the polished case for implants 2 to 4.
Figure 4 shows the neointimal formation from implant 5 (106 to 125 μm particles) during disassembly of the blood pump. The color of the neointima was pale-red and white. This implant was the first one to incorporate a sintered titanium case. The formation of a continuous neointimal surface between the particle-cast diaphragm and the sintered case was observed (as indicated by the arrow).
Figure 5 illustrates neointimal formation on the sintered titanium case from implant 5. The pale-red and white neointima was generally confined to the areas adjacent to the particle-cast diaphragm. The remainder of the sintered case was covered with a thin, semitransparent neointima. White patches were also periodically observed. Similar observations have been reported on calf and human implants of sintered cases.8,17
Each of the particle-cast diaphragms was entirely covered with a neointimal surface at explant. The thickness of the neointima was typically thinner (125–250 μm) on the periphery of the diaphragm and thicker (1,000–2,500 μm) toward the center of the diaphragm. The neointimal adherences for implants 1, 2, and 3 were classified qualitatively as “moderately adherent.” A “moderately adherent” neointima was defined as one that could be separated from the underlying microtextured surface if moderate force was applied. The neointimal adherences for implants 4 and 5 were qualitatively classified as “significantly adherent.” A “significantly adherent” neointima could not be separated from the underlying surface without destroying the neointima.
The neointima was thinner in the high-flex regions for each of the particle-cast implants, as shown in Figure 6 for implant 2. Figure 7 shows the laminated nature of the neointima for implant 3. This structure is similar to that observed for integrally textured fiber implants.17–19
Samples were taken from the neointima and fixed as described in the previous section. Light microscopy examinations using standard paraffin, embedding, sectioning, H&E, and Mason’s trichrome staining techniques were made of these samples. An example using H&E techniques is shown in Figure 8 for implant 2. Fibrin, neutrophils, and red blood cells were observed within this layer. Platelets are typically difficult to observe under light microscopy. Focal mineralization associated with some calcification was also observed in these regions for all implants except implant 1. No calcification was observed for implant 1 owing to its brief duration (10 hours).
Figure 9 shows a sample from implant 5 using Mason’s trichrome staining techniques. The blood-contacting side of the sample is at the bottom of this Figure. Three layers composed primarily of endothelial cells, collagen, and fibrin were observed. The presence of endothelial cells on the blood-contacting side of the neointimal layer has also been observed on human implants of integrally textured fiber surfaces.20 The presence of endothelial cells suggests the continuous development of the neointimal layer.
Figure 10 shows the surface of the neointima that contacted the particle-cast polyurethane surface (Figure 10A) and the blood (Figure 10B) for implant 2. The large-scale roughness observed on the polyurethane-contacting neointimal surface is most likely the result of contact with the particle-cast surface (Figure 2). The smaller-scale roughness appears to be related to the fibrin matrix observed in Figure 8. Many cells (assumed to be red blood cells and monocytes) are observed. In contrast, the blood-contacting surface has only smaller-scale roughness that is presumed to be related to the underlying fibrin matrix. Few blood cells are observed.
Figure 11 shows the integration of the neointima with the underlying particle-cast surface for implant 4. As stated earlier, the adherence for this implant was qualitatively classified as “significantly adherent.” Figure 12 shows integration of the neointima within a pore of the particle-cast surface for implant 4. Mononuclear cells and strands (presumed to be fibrin) are also observed.
The overall neointimal microstructure and cells observed in these implants are similar in substance to that observed upon the integrally textured fiber surfaces from both short- and long-term HeartMate implants.17–19
The preliminary in vivo studies have shown that a stable, durable, and adherent neointimal surface can be formed upon the particle-cast surfaces for periods up to 30 days. The implant of the sintered titanium case showed the formation of a continuous neointimal surface between the particle-cast diaphragm and the sintered case. The presence of calcification in all of the implants suggests that the bovine model may not be suitable for long-term testing of microtextured surfaces.
These preliminary results also suggest that cavity size has an effect on neointimal adhesion. As summarized in Table 1, implant 2 used surfaces cast from 32 to 53 μm particles, implant 3 used surfaces cast from 106 to 125 μm, and implant 4 used surfaces cast from 106 to 250 μm. The difference in adherence between these implants (“moderate” for implants 2 and 3 and “significant” for implant 4) is hypothesized to be associated with the different range of cavity sizes between the implants.
Based on these preliminary in vivo studies, we hypothesize that the integration of the neointima into the pores of the particle-cast surface is responsible for neointimal formation, morphology, and adherence to the particle-cast surfaces. As the neointima develops, it spreads into the pores located at the interface between the neointima and the particle-cast surface. These surface pores are illustrated in the top views of the surfaces illustrated in Figure 2. The neointima then spreads into the underlying pores (Figure 2) and becomes entrapped within these pores. This entrapment results in traction between the neointima and the underlying particle-cast surface.
It is important to note that the mechanism of neointimal adherence is different for the integrally textured fiber and particle-cast surfaces. For the integrally textured fiber surface, the neointima attaches to the underlying surface by entrapping the fibers. The neointima becomes integrated with the underlying pores for the particle-cast surface. Additional research is required to determine if this difference in adherence mechanism between the integrally textured fiber and particle-cast surfaces has an effect on the rate of thromboembolic complications and long-term neointimal durability.
Future implants with particle-cast surfaces will use the particle-cast surfaces in conjunction with the sintered titanium case used in implant 5. This sintered surface will have surface characteristics that are similar to the case used by the HeartMate. The combination of a sintered case with a particle-cast surface creates a pump that is similar to clinically available pumps that use textured blood-contacting surfaces. In addition, we will perform longer implants to determine the effects of particle-cast cavity size on long-term neointimal adherence. Multiple implants will also be performed to demonstrate repeatability.
Testing in animals has shown that a stable neointimal layer can be formed upon particle-cast surfaces when incorporated within circulatory support devices. These preliminary results also indicate that the cavity size on these surfaces may have a significant effect on neointimal adhesion. Future research will identify the key cavity size, depth, and distribution that will minimize thromboembolic complications. These results can be used in the design of future circulatory support devices.
The authors wish to thank Christopher A. Siedlecki, PhD, Kimberly Griffith, Dennis Hicks, John Reibson, and other members of the Division of Artificial Organs for their significant contributions to this work.
1. Rose EA, Gelijns AC, Moskowitz AJ, et al
: Long-term mechanical left ventricular assistance for end-stage heart failure. N Engl J Med
345: 1435–1443, 2001.
2. Lazar RM, Shapiro PA, Jaski BE, et al
: Neurological events during long-term mechanical circulatory support for heart failure: the Randomized Evaluation of Mechanical Assistance for the Treatment of Congestive Heart Failure (REMATCH) experience. Circulation
109: 2423–2427, 2004.
3. Bussey HI: An overview of anticoagulants, antiplatelet agents, and the combination in patients with mechanical heart valves. J Heart Valve Dis
13: 319–324, 2004.
4. Koertke H, Zittermann A, Minami K, et al
: Low-dose international normalized ratio self-management: A promising tool to achieve low complication rates after mechanical heart valve replacement. Ann Thorac Surg
79: 1909–1914; discussion 1914, 2005.
5. Frazier OH, Myers TJ, Radovancevic B: The HeartMate left ventricular assist system: Overview and 12-year experience. Texas Heart Inst J
25: 265–271, 1998.
6. Rose EA, Levin HR, Oz MC, et al
: Artificial circulatory support with textured interior surfaces. A counterintuitive approach to minimizing thromboembolism. Circulation
90: II87–91, 1994.
7. Liotta D, Hall CW, Akers WW, et al
: A pseudoendocardium for implantable blood pumps. Tran ASAIO
12: 129–138, 1966.
8. Szycher M, Bernhard WF: Hemocompatible interfaces in artificial heart devices, in Szycher M (ed), Biocompatible Polymers, Metals, and Composites.
Lancaster, PA, TECHNOMIC Publishing Co., pp. 103–163, 1983.
9. Frazier OH, Rose EA, Oz MC, et al
: Multicenter clinical evaluation of the HeartMate vented electric left ventricular assist system in patients awaiting heart transplantation. J Thorac Cardiovasc Surg
122: 1186–1195, 2001.
10. Richenbacher WE, Naka Y, Raines EP, et al
: Surgical management of patients in the REMATCH trial. Ann Thorac Surg
75: S86–92, 2003.
11. El-Banayosy A, Arusoglu L, Kizner L, et al
: Preliminary experience with the LionHeart left ventricular assist device in patients with end-stage heart failure. Ann Thorac Surg
75: 1469–1475, 2003.
12. Weiss WJ, Rosenberg G, Snyder AJ, et al
: In vivo performance of a transcutaneous energy transmission system with the Penn State motor driven ventricular assist device. ASAIO Trans
35: 284–288, 1989.
13. Weiss WJ, Rosenberg G, Snyder AJ, et al
: Steady state hemodynamic and energetic characterization of the Penn State/3M Health Care Total Artificial Heart. ASAIO J
45: 189–193, 1999.
14. Weiss WJ, Rosenberg G, Snyder AJ, et al
: A completely implanted left ventricular assist device. Chronic in vivo testing. ASAIO J
39: M427–432, 1993.
15. Mavroidis D, Sun BC, Pae WE, Jr: Bridge to transplantation: the Penn State experience. Ann Thorac Surg
68: 684–687, 1999.
16. McBride LR, Naunheim KS, Fiore AC, et al
: Clinical experience with 111 Thoratec ventricular assist devices. Ann Thorac Surg
67: 1233–1238; discussion 1238-1239, 1999.
17. Dasse KA, Chipman SD, Sherman CN, et al
: Clinical experience with textured blood contacting surfaces in ventricular assist devices. ASAIO Trans
33: 418–425, 1987.
18. Menconi MJ, Pockwinse S, Owen TA, et al
: Properties of blood-contacting surfaces of clinically implanted cardiac assist devices: Gene expression, matrix composition, and ultrastructural characterization of cellular linings. J Cell Biochem
57: 557–573, 1995.
19. Salih V, Graham TR, Berry CL, et al
: The lining of textured surfaces in implantable left ventricular assist devices: An immunocytochemical and electron microscopic study. Am J Cardiovasc Path
4: 317–325, 1993.
20. Frazier OH, Baldwin RT, Eskin SG, Duncan JM: Immunochemical identification of human endothelial cells on the lining of a ventricular assist device. Texas Heart Inst J
20: 78–82, 1993.