Left ventricular assist devices (LVADs) improve the quality of life and the life expectancy of patients with advanced heart failure.1 Despite these benefits, pump thrombosis and subsequent embolic stroke have devastating complications, including a 24% probability of death within 3 months of pump thrombosis.2 In initial trials leading to Food and Drug Administration approval, ischemic stroke occurred in 6–8% of patients.3,4 A systematic review of antithrombotic therapy and outcomes showed that on average, LVAD thrombosis occurs in 5% of patients.5 An increased incidence of HeartMate II (HMII) LVAD thrombosis was reported starting in 2011, with incidence up to 12% at 2 years.6 Stroke remains a leading cause of death in LVAD patients, and event rates remain comparable between axial and centrifugal devices, including the HeartMate 3.7,8
LVAD patients are treated with anticoagulant or antiplatelet medications to reduce the risk of stroke and pump thrombosis. Long-term anticoagulation is managed with vitamin K antagonist therapy, which is usually targeted to a goal international normalized ratio (INR) between 2.0 and 3.0 with some variations based on institutional experience.9 There are significant differences between centers whether antiplatelet therapy is recommended, the dose of aspirin, or if aspirin is combined with dipyridamole or P2Y12 inhibitors.5,10 A multistep approach to reduce the risk of pump thrombosis with HMII was undertaken in the prevention of heartmate II pump thrombosis (PREVENT), trial including use of aspirin and warfarin with goal INR 2–2.5. Lower rates of pump thrombosis and ischemic stroke were reported especially when adhering to recommendations regarding surgical implantation, heparin bridging, and pump speeds.11 Medical therapy for suspected LVAD thrombosis has been attempted with thrombolytic therapy, anticoagulation, and glycoprotein IIb/IIIa inhibitors.12 Rates of thrombus resolution are similar between thrombolytic and non-thrombolytic therapies, but major bleeding has been reported in 29% of patients treated with thrombolysis.12 Improved strategies are needed to prevent and treat pump thrombosis.
The mechanism of LVAD thrombosis initiation and propagation is unknown. Pathological analysis of the LVAD thrombi to date has been limited. A report of 19 cases of HMII thrombosis described thrombi as a mixture of white and red areas interpreted to be fibrin and blood components.13 Analysis of 11 HMII thrombi by hematoxylin and eosin (H&E) staining showed layered thrombi of eosinophilic material thought to be fibrin around the inlet bearing. Small amounts of cellular infiltrate were noted.14 Immunohistochemistry has not been performed to identify platelets or other thrombi components. Correctly identifying the constituents of the clot and location of various components could aid in directing therapy for patients with suspected LVAD thrombosis.
Explanted HMII LVADs were returned to the manufacturer for analysis, where thrombi were removed and stored in formalin. We obtained deidentified thrombi from HMII LVADs explanted at the University of Louisville for pathological analysis. Patients at the University of Louisville are routinely treated with warfarin with an INR goal of 2–3 and aspirin 325 mg orally daily. The study was approved by the Medical College of Wisconsin (MCW) Institutional Review Board.
Paraffin-embedded sections were deparaffinized in xylene and rehydrated in a decreasing concentration of ethanol. Heat antigen retrieval was performed with Epitope Unmasking Buffer (Bioworld, Dublin, OH) for 30 minutes. Slides were washed in Amplifying Wash Buffer (Bioworld). Endogenous peroxidases were inhibited using 50:50 methanol and hydrogen peroxide and then blocked with 2.5% normal goat serum (Vector laboratories, Burlingame, CA) for 1 hour. After thorough washing, the primary antibody or antibody control were added (mouse or rabbit IgG, 5 μg/ml diluted in 0.1% BSA, Vector Laboratories). Slides were incubated overnight at 4°C. After washing in PBS, slides were incubated for 30 minutes with peroxidase labeled anti-mouse or anti-rabbit Ig (ImmPRESS Reagent, Vector Laboratories). Peroxidase substrate was added for 5–15 minutes (ImmPACT NovaRed or DAB Peroxide Substrate Kit; Vector Laboratories). Samples were counterstained with Gill’s hematoxylin for 5 minutes (Vector Laboratories) and dehydrated using an increasing gradient of alcohol and xylene. DPX mounting media (Electron Microscopy Sciences) was applied before the cover slip. Samples were analyzed using the Nikon Eclipse E600. Photos were taken with SPOT software 5.1 with gamma and saturation of 1 applied.
Von Willebrand factor (VWF) was identified using polyclonal rabbit ant-human antibody (Dako A0082). Staining from a polyclonal rabbit anti-human antibody to fibrin(ogen) (Dako A0080) was compared with 59D8, a mouse monoclonal antibody specific for the β-chain of fibrin.15 Platelets were identified using AP5 and 314.5 (gift of Richard Aster). AP5 binds to the first 6 amino acids of β3 and recognizes when the receptor is engaged with fibrinogen.16 Antibody 314.5 binds to the light chain of glycoprotein IIb. platelet endothelial cell adhesion molecule (PECAM) is expressed on platelets after alpha granule release and was detected using CD31 (Dako JC70A). Leukocytes were stained using monoclonal mouse anti-human to CD45 (Dako 2B11+ PD7/26) to distinguish between PECAM expression on leukocytes or platelets.
We obtained 28 thrombi that were removed from 17 HMIIs for suspected pump thrombus. Ten pumps had a single thrombus and seven pumps had clots in multiple locations. In pumps with a single thrombus, four thrombi were at the inlet bearing, three were at the rotor vanes, two were at the outlet bearing, and one was at the outlet elbow. In total, 72% of the thrombi (20/28) occurred on the rotor: nine were located at the inlet bearing, five at the rotor vanes, and six at the outlet bearing (Table 1). Thrombi were found less frequently in other areas including three in the inlet flex section, three in the outlet elbow, one in the inlet tube, and one in the outflow graft (Table 1).
Gross examination showed that clots tended to be homogenous and pale to dark red in color except for one white, thin inlet bearing thrombus. H&E staining was completed on all specimens, and representative clots from various locations within the device are shown in Figure 1. Histologically, the six clots found at the outlet elbow (Figure 1A), and inlet flex section (Figure 1E) had a similar appearance and were composed of loose layers of eosinophilic protein and leukocytes. In contrast, 17 of the 20 thrombi formed on the rotor (inlet bearing [Figure 1D], rotor vane [Figure 1C], and outlet bearing [Figure 1B]) were composed of dense laminated clots. Of the remaining three rotor thrombi, one thrombus on the rotor vane and one thrombus on the outlet bearing were histologically similar to the inlet tube thrombi. One thrombus, located on the outflow bearing, looked like the inlet flex thrombi. The different structures of these three clots raise the possibility that they were formed elsewhere, suctioned into the device, and found on the rotor at explant.
The thrombi from the inlet bearing were analyzed in detail because orientation of the clot to the device is known. Inlet bearing thrombi contained laminated rings measuring 150–2618 μm thick. Both H&E (Figure 1D) and Martius, Scarlet, Blue staining (MSB; Figure 2) showed that the inlet bearing thrombi had distinct layers that varied in composition (median 3 rings [range 1–7]). MSB uses three dyes to differentiate between erythrocytes (yellow), fibrin (red), and larger fibers such as collagen (blue).17 MSB suggested that the rings consisted of fibrin because of the deep red color (Figure 2A). Immunohistochemistry revealed that the inner rings were composed of dense fibrin(ogen) and VWF (Figure 2B–D). Outer rings had fibrin, fibrinogen, and VWF with less organized structure. Clumps of platelets were identified in the outer rings of the thrombi (Figure 3) by PECAM, β3, and glycoprotein IIb staining. PECAM is found on the platelet membrane, and expression increases significantly upon alpha granule release,18 suggesting that the platelets seen in the thrombi were activated (Figure 3C). White blood cells were not found in the inlet bearing thrombi. The thin ring thrombus that was white on gross examination also contained fibrin(ogen), VWF, and small numbers of platelets, similar to the other ring thrombi. The reason for color variation on gross examination without a difference in composition was not determined. Although most clots appeared red, intact erythrocytes and evidence of hemoglobin on iron stain were not found. All ring thrombi had similar structure and composition as to the representative clots in Figures 2 and 3.
Thrombi found on other areas of the rotor had similar composition to the inlet bearing thrombi. Thrombi formed along the rotor vanes formed layers analogous to the inlet bearing thrombus as they contained fibrin(ogen) and VWF in dense bands. Orientation of the clot to the device could not be confirmed to determine areas touching the device surface, however. Four of the six outlet bearing thrombi were also composed of layers of fibrin(ogen) and VWF but had less organization and increased areas of nuclear material in comparison to the inlet bearing thrombi (Figure 1B).
In contrast to the thrombi formed on the rotor, thrombi from the inlet flex section and outlet elbow were composed of loose layers of eosinophilic protein, shown to be fibrin(ogen) on MSB staining and immunohistochemistry (Figure 4). Thin layers of fibrin (red staining) were found without erythrocytes. On immunohistochemistry, VWF was seen in areas similar to fibrin(ogen). More platelets and leukocytes were found in the inlet flex sections in comparison to the rotor thrombi (Figure 4).
A detailed immunohistochemical analysis of HMII thrombi was performed as an initial step to identify how thrombi form. Despite differences in the gross appearance of the thrombi (red versus. white), HMII clots were composed primarily of fibrin(ogen) and VWF. Staining for platelet and leukocyte antigens were seen infrequently, and intact red blood cells were not found. The histologically distinct layers of the inlet bearing thrombi suggest that the clots form over time by initiating at the device surface and adding layers circumferentially. This is consistent with clinical experience; patients have intermittent hemolytic episodes treated medically before requiring pump removal. The presence of multiple rings in all but one laminated inlet bearing thrombi suggests there is a size threshold for clinically apparent clots. Additionally, the thrombus growth is perpendicular to the flow of blood. Experimental studies have identified disturbed pathlines around the inlet bearing at low flow rates, which may provide some mechanistic insights into thrombus formation.19
The LVAD thrombi have distinct histopathology when compared with coronary artery thrombi,20 mechanical valve thrombi, and venous thromboembolism.21 Plaque rupture leads to exposure of the cholesterol deposits, foamy macrophages, and necrotic core to the bloodstream. Blood flow is obstructed by platelet aggregation and then stabilized with fibrin formation.22 Only minimal platelets were found within the LVAD thrombi, highlighting a distinct difference. Lines of Zahn form in arterial thrombi and contain alternating layers of platelets and fibrin with red blood cells.23 Although laminated thrombi were seen on clots found on the rotor, the layers were composed of fibrin(ogen) and VWF instead of platelets. Dense fibrin(ogen) has been shown in a mechanical valve thrombus attached to the pyrolytic carbon of the valve leaflet.24 Other pathological analysis of mechanical valve thrombi have shown growth of a cellular infiltrate into the sewing ring,25 but the LVAD thrombi did not contain a significant amount of leukocytes. Organizing areas of venous thrombi contain platelet-rich zones associated with neutrophils,21 but the organizing thrombus in the outermost rings of the inlet bearing thrombi did not contain neutrophils.
Analysis of the inlet bearing thrombi provides a unique opportunity to understand the components of the clot nidus that formed near the device from which subsequent layers grow. The inner rings of the inlet bearing thrombi are composed of fibrin(ogen) and VWF, suggesting that thrombi are initiated through the coagulation system. Warfarin therapy decreases the amount of coagulation factors II, VII, IX, and X, resulting in a decreased propensity to form thrombi.26 Artificial surfaces can directly activate factors XI and XII, proteins of the intrinsic pathway which are not targeted by warfarin.27,28 A 40–50% absolute decrease in factor XI and factor XII was found after implantation of a pulsatile LVAD, suggesting activation of the intrinsic pathway of coagulation.29 Strategies to inhibit the intrinsic pathway for prevention of venous thrombosis after orthopedic surgery have been studied in phase II clinical trials.30 A comprehensive understanding of how the coagulation system becomes activated after LVAD implantation may identify alternative anticoagulant strategies, such as targeting the intrinsic pathway to prevent LVAD thrombosis.
LVADs create high shear stress which leads to a loss of high molecular weight multimers of VWF.31 Because all LVAD patients have acquired von Willebrand syndrome but less than half experience bleeding, the clinical implications of acquired von Willebrand syndrome is uncertain.31,32 VWF was identified in all HMII thrombi, suggesting that despite the loss of high molecular weight multimers, VWF was able to participate in clot formation. Immunohistochemistry cannot identify fiber length or interactions, but VWF was seen in areas with fibrin(ogen), suggesting a potential target for future investigation.
LVAD patients at the University of Louisville are treated with 325 mg aspirin. Despite aspirin therapy, platelets were identified in the outer rings of inlet bearing thrombi and, thus, may be involved with clot propagation or new thrombus formation. The presence of platelets in aggregates suggests that platelets participate in clot development and are not caught in the fibrin(ogen) and VWF network as a bystander. Because we cannot identify when the clot started, platelets could be part of initial clot development and then become replaced or degraded as the thrombus organizes. An in vitro LVAD thrombosis model would be needed to outline the time course of thrombus development.
Our study has several limitations. Samples of HMII thrombi were analyzed because preserved samples were available. Although a direct comparison was not completed with centrifugal flow device thrombi, HVAD thrombi have been reported to contain a homogenous eosinophilic stranded material, similar to the middle rings of the inlet bearing thrombi.33 Histopatholgic analysis of HMII thrombi will serve as a baseline for comparison to other devices. Although early studies of the HeartMate 3 device have not shown device thrombosis, ischemic stroke has occurred at rates equal to other devices.8 Additionally, the concerning rates of pump thrombosis were not seen in the initial clinical trials of the HMII device34,35; thus, it is possible that with continued use of new devices, differences in the adverse event profile could occur. Because the thrombi analyzed were from a single center, we could not identify whether differences in antithrombotic therapy management could affect the composition of the thrombi. However, the appearance of the thrombi on H&E is similar to that in previous reports.14 After explantation, pumps are sent to the manufacturer and then disassembled, a process that takes 24–72 hours. Therefore, protein and cellular degradation may have occurred before fixation. Immunohistochemistry provides qualitative information about absence or presence of a protein. Attempts to use immunohistochemistry to quantify protein amounts between samples are limited by lack of intra- and interobserver reliability and remain semiquantitative.36 Because of differences in binding affinity of antibodies for different proteins, immunohistochemistry cannot be used to compare between proteins either. Because banked samples were used, we were not able to confirm whether the clots were tightly adherent suggesting thrombus formation in situ or were suctioned into the device. Finally, samples were deidentified, so clinical conditions associated with LVAD thrombosis, level of antithrombotic therapy, and treatment before LVAD explantation could not be correlated to thrombi composition.
The majority of LVAD thrombi form in the rotor in pathologically distinct layers, suggesting development over time. Clots formed on the rotor do not appear to be platelet-driven as platelets were found in limited numbers only in the outer rings of inlet bearing thrombi. Initiating events require further investigation, but the fibrin-rich structure suggests that alternative anticoagulation strategies perhaps targeting the intrinsic pathway are needed to prevent thrombosis in our LVAD patients.
Dr Baumann Kreuziger developed the study, performed the analysis, drafted, and approved the manuscript. Dr Slaughter was involved with study design, interpretation of results, and approved the manuscript. Dr Sundareswaran assisted with sample acquisition, data interpretation, and approved the manuscript. Dr Mast was involved with study design, sample analysis, interpretation of results, and approved the manuscript.
1. Fang JC. Rise of the machines–left ventricular assist devices as permanent therapy for advanced heart failure. N Engl J Med 2009.361: 22822285.
2. Smedira NG, Blackstone EH, Ehrlinger J, et al. Current risks of HeartMate II pump thrombosis: non-parametric analysis of Interagency Registry for Mechanically Assisted Circulatory Support data. J Heart Lung Transplant 2015.34: 15271534.
3. Aaronson KD, Slaughter MS, Miller LW, et al.; HeartWare Ventricular Assist Device (HVAD) Bridge to Transplant ADVANCE Trial Investigators: Use of an intrapericardial, continuous-flow, centrifugal pump in patients awaiting heart transplantation. Circulation 2012.125: 31913200.
4. Slaughter MS, Rogers JG, Milano CA, et al.; HeartMate II Investigators: Advanced heart failure treated with continuous-flow left ventricular assist device. N Engl J Med 2009.361: 22412251.
5. Baumann Kreuziger LM, Kim B, Wieselthaler GM. Antithrombotic therapy for left ventricular assist devices in adults: a systematic review. J Thromb Haemost 2015.13: 946955.
6. Starling RC, Moazami N, Silvestry SC, et al. Unexpected abrupt increase in left ventricular assist device thrombosis. N Engl J Med 2014.370: 3340.
7. Kirklin JK, Naftel DC, Pagani FD, et al. Seventh INTERMACS annual report: 15,000 patients and counting. J Heart Lung Transplant 2015.34: 14951504.
8. Mehra MR, Naka Y, Uriel N, et al.; MOMENTUM 3 Investigators: A fully magnetically levitated circulatory pump for advanced heart failure. N Engl J Med 2017.376: 440450.
9. Feldman D, Pamboukian SV, Teuteberg JJ, et al.; International Society for Heart and Lung Transplantation: The 2013 International Society for Heart and Lung Transplantation Guidelines for mechanical circulatory support: executive summary. J Heart Lung Transplant 2013.32: 157187.
10. Katz JN, Adamson RM, John R, et al.; TRACE study: Safety of reduced anti-thrombotic strategies in HeartMate II patients: a one-year analysis of the US-TRACE Study. J Heart Lung Transplant 2015.34: 15421548.
11. Maltais S, Kilic A, Nathan S, et al.; PREVENT Study Investigators: PREVENtion of HeartMate II Pump Thrombosis Through Clinical Management: The PREVENT multi-center study. J Heart Lung Transplant 2017.36: 112.
12. Dang G, Epperla N, Muppidi V, et al. Medical management of pump-related thrombosis in patients with continuous-flow left ventricular assist devices: a systematic review and meta-analysis. ASAIO J 2017.63: 373385.
13. Uriel N, Han J, Morrison KA, et al. Device thrombosis in HeartMate II continuous-flow left ventricular assist devices: a multifactorial phenomenon. J Heart Lung Transplant 2014.33: 5159.
14. Prasad SM, Robertson JO, Itoh A, Joseph SM, Silvestry SC. Histologic analysis of clots in explanted axial continuous-flow left ventricular assist devices. J Heart Lung Transplant 2015.34: 616618.
15. Hui KY, Haber E, Matsueda GR. Monoclonal antibodies to a synthetic fibrin
-like peptide bind to human fibrin
but not fibrinogen. Science 1983.222: 11291132.
16. Honda S, Tomiyama Y, Pelletier AJ, et al. Topography of ligand-induced binding sites, including a novel cation-sensitive epitope (AP5) at the amino terminus, of the human integrin beta 3 subunit. J Biol Chem 1995.270: 1194711954.
17. Lendrum AC, Fraser DS, Slidders W, Henderson R. Studies on the character and staining of fibrin
. J Clin Pathol 1962.15: 401413.
18. Cramer EM, Berger G, Berndt MC. Platelet alpha-granule and plasma membrane share two new components: CD9 and PECAM-1. Blood 1994.84: 17221730.
19. Yang F, Kormos RL, Antaki JF. High-speed visualization of disturbed pathlines in axial flow ventricular assist device under pulsatile conditions. J Thorac Cardiovasc Surg 2015.150: 938944.
20. Falk E, Nakano M, Bentzon JF, Finn AV, Virmani R. Update on acute coronary syndromes: the pathologists’ view. Eur Heart J 2013.34: 719728.
21. Savchenko AS, Martinod K, Seidman MA, et al. Neutrophil extracellular traps form predominantly during the organizing stage of human venous thromboembolism development. J Thromb Haemost 2014.12: 860870.
22. Falk E. Dynamics in thrombus formation. Ann N Y Acad Sci 1992.667: 204223.
23. Lee R, Adlam D, Clelland CA, Channon KM. Lines of Zahn in coronary artery thrombus. Eur Heart J 2012.33: 1039.
24. Scheidmann R, Foth R, Sigler M. Thrombosis of a mechanical prosthetic aortic valve in early pregnancy: histopathological findings. Cardiovasc Pathol 2017.27: 3536.
25. Kurian MV, Sabareeswaran A, Kanchanamala M, Balachandran S, Mogileswari D, Subban V. The role of the sewing ring in tissue healing: a pathologic study of explanted mechanical heart valves. J Heart Valve Dis 2012.21: 253259.
26. Hamaguchi S, Hirose T, Akeda Y, et al. Identification of neutrophil extracellular traps in the blood of patients with systemic inflammatory response syndrome. J Int Med Res 2013.41: 162168.
27. Jaffer IH, Fredenburgh JC, Hirsh J, Weitz JI. Medical device-induced thrombosis: what causes it and how can we prevent it? J Thromb Haemost 2015.13(suppl 1): S72S81.
28. Jaffer IH, Stafford AR, Fredenburgh JC, Whitlock RP, Chan NC, Weitz JI. Dabigatran is less effective than warfarin at attenuating mechanical heart valve-induced thrombin generation. J Am Heart Assoc 2015.4: e002322.
29. Himmelreich G, Ullmann H, Riess H, et al. Pathophysiologic role of contact activation in bleeding followed by thromboembolic complications after implantation of a ventricular assist device. ASAIO J 1995.41: M790M794.
30. Büller HR, Bethune C, Bhanot S, et al.; FXI-ASO TKA Investigators: Factor XI antisense oligonucleotide for prevention of venous thrombosis. N Engl J Med 2015.372: 232240.
31. Meyer AL, Malehsa D, Bara C, et al. Acquired von Willebrand syndrome in patients with an axial flow left ventricular assist device. Circ Heart Fail 2010.3: 675681.
32. Crow S, Chen D, Milano C, et al. Acquired von Willebrand syndrome in continuous-flow ventricular assist device recipients. Ann Thorac Surg 2010.90: 12631269; discussion 1269.
33. Jessen SL, Weeks BR, Jackson ND, et al. Pathology of ventricular assist devices, in ASAIO. 2013, 29.Chicago, IL.
34. Miller LW, Pagani FD, Russell SD, et al.; HeartMate II Clinical Investigators: Use of a continuous-flow device in patients awaiting heart transplantation. N Engl J Med 2007.357: 885896.
35. Pagani FD, Miller LW, Russell SD, et al.; HeartMate II Investigators: Extended mechanical circulatory support with a continuous-flow rotary left ventricular assist device. J Am Coll Cardiol 2009.54: 312321.
36. Becker RL Jr. Standardization and quality control of quantitative microscopy in pathology. J Cell Biochem Suppl 1993.17G: 199204.