The transition from pulsatile devices to continuous-flow left ventricular assist devices (cf-LVADs) contributed to improved clinical outcomes with better survival and lower adverse events in patients with advanced heart failure.1 Even after cf-LVADs became the standard in mechanical circulatory support, miniaturization of blood pumps remained a market trend to minimize surgical trauma, resulting in shorter surgical times, less bleeding, and faster postoperative recovery.
EVAHEART (Sun Medical Technology Research Corp., Nagano, Japan) is a centrifugal-type cf-LVAD with a hydraulically levitated impeller2 that has been used clinically in Japan since 2005 as a bridge to heart transplantation.3 Sterile water for injection is continuously circulated through the driveline and around the rotating impeller shaft and bearing interface to apply a hydraulic levitation force (Figure 1). This unique hydraulic circulation system, called the “cool seal unit,” has demonstrated durability of up to 10 years in clinical use without any adverse infectious events. The impeller features a unique open-vane design with large gaps to the pump housing wall (700–1,000 µm), minimizing wall shear stress. A computational fluid dynamic analysis of the pump was reported previously.4
Following the trend in pump miniaturization, the original EVAHEART left ventricular assist system (LVAS; EVAHEART 1) was miniaturized and developed as the EVAHEART 2 LVAS. The EVAHEART 1 required thick pump housing walls for a screw bolt assembly. To miniaturize the pump size without compromising the pump flow characteristics, the pump’s external walls were made thinner and were assembled through a press-fit. All of the surfaces that contact blood (including the impeller) and the blood pathway dimensions remained unchanged. Compared with the EVAHEART 1, the EVAHEART 2 was miniaturized by 30% in weight and 26% in displacement volume. The driveline diameter decreased from 9.8 to 7.8 mm (Figure 2). Because the blood pathway dimensions are identical, the pump performance curves reported previously remain unchanged.3 The hydraulic pump performance curves (generally known as the head pressure and pump flow quantity relationship curve) of the EVAHEART 2 are relatively “flat,” meaning the flow is sensitive to delta pressure. Additionally, the EVAHEART 2 is capable of high end-systolic peak flow, which contributes to preserving the arterial pulsatility. The average pressure gradient (mm Hg/litter per minute) for a pump speed required for a head pressure of 80 mm Hg and a pump flow of 5 L/min is −2.67.
Current LVADs in clinical use adopt a tubular-type inflow cannula that is inserted through the left ventricular (LV) apex wall and that protrudes inside the LV. This protrusion of the inflow cannula alters the blood flow patterns in the LV, which is associated with stagnation caused by interference of washout flow between the cannula and the endocardium.5 This can result in “wedge thrombus” formation. Conventional inflow cannulae are also susceptible to intraventricular migration and are prone to causing ventricular suction and inflow occlusion during surgery, after cardiac reverse remodeling, or with hypovolemia. Pump pocket position (i.e., depth) and the pocket healing process can worsen inflow cannula malposition. A malpositioned inflow tip touching the myocardial wall can trigger thrombus formation, microemboli, and pannus formation over the inflow tip (tissue ingrowth originating from the adjacent myocardium), which can contribute to post-LVAD stroke.6 Such malposition can also result in ventricular wall suction, increasing the risk of arrhythmia or low pump flow, followed by possible pump thrombosis.7,8
To overcome the issues associated with the tubular-type inflow cannula, we developed the “double-cuff tipless” (DCT) inflow cannula, which does not protrude into the LV chamber. Instead, the DCT cannula is secured to the endocardial surface through a mesh suture cuff. EVAHEART 2 adopted the DCT inflow cannula specifically to mitigate the risk of stroke and inflow suction events. The DCT inflow offers a more forgiving design, avoiding inflow malposition within the LV. Figure 3 shows the following components of the DCT inflow design: 1) polished, methacryloyloxyethyl phosphorylcholine–coated titanium conduit for blood drainage; 2) polyester surgical mesh for approximating cored apical myocardium; 3) polytetrafluorethylene (PTFE) suture cuff; and 4) O-ring sealed connector.
Before clinical use, the EVAHEART 2 DCT inflow cannula was evaluated in in vivo animal implant studies for system safety and feasibility.
Materials and Methods
Implantation techniques for the DCT inflow cannula and myocardial anatomical fitting are illustrated in Figure 4. The DCT inflow was designed to pull the endocardial layer into contact with the proximal mesh suture cuff. This technique, together with the wide distal PTFE felt suture cuff, allows the DCT inflow to adapt to various ventricular wall thicknesses (from thin myocardium to hypertrophic myocardium of up to 2.5 cm) at the implantation site (Figure 4, A–C).
To evaluate the DCT inflow cannula in vivo, a clinical model of the DCT inflow cannula (test article) was manufactured for chronic animal implantation. Eight animal studies were scheduled at the Texas A&M Institute for Preclinical Studies (TIPS) (College Station, TX; N = 6) and the National Cerebral and Cardiovascular Center (NCVC) Artificial Organ Laboratory (Osaka, Japan; N = 2). Animal use protocols were approved by Institutional Animal Care and Use Committees at Texas A&M University and NCVC with a harmonized study protocol design. Test protocol numbers are 01216/00916 at TIPS and JP152946662.9/152485737 at NCVC. Texas A&M Cardiovascular Pathology Lab performed necropsy and pathology services at TIPS. The investigator group performed the necropsy at NCVC; however, histopathology analysis was outsourced to institutional pathology laboratory.
Healthy bovine animals (approximately 6 months of age and weighing from 90 to 140 kg, 6 males and 2 females) were used. Animals were given a unique identifying number to ensure traceability and data integrity in the study.
After delivery to the study site, animals were quarantined for a minimum of 14 days before study assignment. Animals were weighed upon arrival, and a visual exam was performed by a veterinarian staff. Daily observations, vaccinations, examinations, weights, treatments, clinical health problems, and any diagnostic tests were documented in the animal’s records.
For the surgical procedure, ketamine (3–8 mg/kg) was administered intravenously for induction. The animals were intubated and maintained with isoflurane for a fifth or sixth intercostal left thoracotomy surgical approach. The descending aorta was partially clamped, and the outflow graft was anastomosed by use of an end-to-side technique. A perfusion cannula (19 Fr) was inserted into the external carotid artery, and venous blood was drained by use of a two-staged cannula (34 Fr/46 Fr) through the right atrial appendage. After establishment of cardiopulmonary bypass (CPB), the LV apex was cored, and the DCT inflow cannula was implanted under beating heart condition without aortic cross-clamping. In all cases, inflow cannulae were tilted toward the ventricular septal wall to simulate a worst-case malposition scenario. After the animal was taken off CPB and the chest was closed, the animal was humanely managed at the animal intensive care unit to monitor vital signs and pump parameters. Before the chest was closed, an alcohol nerve block technique was applied to control rib cage pain around the left thoracotomy area. For perioperative analgesia, butorphanol (0.1–0.3 mg/kg, every 6 hours up to 7 days) and flunixin (1.1–2.2 mg/kg, every 12–24 hours up to 4 days) were administered intravenously. Adequate hemostasis was confirmed by chest drain output, and the animal was anticoagulated by warfarin administration to maintain a target international normalized ratio (INR) range between 2 and 3. The primary endpoint was survival either at 20 ± 5 days (short-term study; 5 cases) or 65 ± 5 days (long-term study; 3 cases); secondary endpoints were major adverse events or device malfunction.
After device implantation, the animals were maintained in the cage at the animal intensive care unit (ICU) with controlled temperature, humidity, and the same light cycle. Animals were fed a diet of Rumilab Maintenance Diet 5510 from Lab Diet (a certified ruminant feed) and Coastal Bermuda hay at least twice daily. All animals had free access to water. Animals were monitored by study personnel continuously for the duration of the study.
The EVAHEART pump parameters (pump speed and pump power) were automatically stored every 5 seconds in an external controller system in addition to manual recording by the ICU staff every hour throughout the study period. Stored data were exported from the controller, and trend data for pump parameters were analyzed periodically.
For blood samples, an appropriately sized needle was used to enter the jugular vein, when an existing catheter was not in place, and blood was collected into a syringe or blood collection tub. Approximately 2–12 ml of blood was withdrawn for analysis at predetermined time points (preoperative baseline; postoperative days 2, 3–6, 7; and every 7 day intervals until study termination). Sample blood was used to measure complete blood count (i.e., white blood cells, hemoglobin, etc.), blood chemistry panel including liver function (i.e., aspartate aminotransferase, alanine aminotransferase), renal function (blood urea nitrogen, creatinine), hemocompatibility parameters (plasma free hemoglobin, lactate dehydrogenase), and blood coagulation panel (INR, fibrinogen, prothrombin time). These blood tests were performed by trained study personnel with properly calibrated analytical equipment or were performed by the accredited blood testing laboratory. Rectal temperature of the animal was recorded once or twice daily.
Antibiotic (cefazolin, 15–25 mg/kg) was administered intravenously before opening the chest, during surgery for prophylactic purpose, and every 12 hours postoperatively until chest tube drainage tubing was removed.
If the animal was physically compromised and the condition was deleterious to the animal or was intractable to treatment, the animal was sedated, heparinized, and euthanized. If the animal reached the study endpoint, the scheduled termination was performed in the same manner. After allowing the heparin to circulate for at least 5 minutes, the animal was euthanized using potassium chloride. The pump was turned off after the euthanasia solution had been given.
A gross necropsy was performed by the Cardiovascular Pathology Laboratory at TIPS or by the necropsy team at NCVC. All explanted tissues, organs, and the test article were fixed with 10% formalin. The heart, lungs, kidneys, liver, spleen, adrenal glands, brain, and any thrombi were evaluated and photographed. Samples were stored in 10% neutral buffered formalin. Selected samples underwent histology processing including paraffin embedding, sectioning, and staining with hematoxylin and eosin, phosphotungstic acid-hematoxylin, and/or Masson trichrome. Slides were evaluated by the study pathologists. Gross and histological findings were included in the contributing scientist’s pathology report.
The consecutive study results are summarized in Table 1. The inflow implantation procedures were uneventful. However, one animal (animal ID #327) experienced ventricular fibrillation caused by an air embolism after the inflow implantation (air suction from a connector component designed only for the animal study). Average CPB time was 56 minutes (range, 41–86 minutes). Six animals reached the primary study endpoint (2 long-term and 4 short-term). One long-term case (animal ID #B26) did not reach the primary endpoint (terminated on postoperative day 42) because of an accidental left front radial fracture when the calf’s leg got stuck in the metal cage. During the course of the studies, no clinical adverse events were identified. Pump speed and power trend data remained stable and stayed within normal clinical ranges (Figure 5). One short-term animal (animal ID #1038) showed remarkable warfarin resistance (average INR was 1.1 despite the escalation of warfarin administration to more than 10 mg/day; Figure 6), and thrombus formation was identified at both the outflow graft anastomosis and the inflow ostium.
Necropsy was performed on all eight test articles, and seven cases were evaluated by histopathology analysis (animal ID #327 was omitted from histopathology because of intraoperative death as described above). Mean malposition angle of the inflow cannula was 58 degrees from the optimal inflow position (range, 30–77 degrees), but there was no evidence of ventricular suction according to either intrinsic pump data or necropsy findings.
Lactate dehydrogenase and plasma free hemoglobin were elevated immediately after surgery because of the effect of CPB but returned to values within the normal range within 7 days (Figure 6). After recovery from surgical and CPB insult, hematological data and rectal temperatures returned to values within the normal range (Figure 7), and no end-organ dysfunction was identified (Figure 8). No device malfunctions occurred throughout the study period.
Inflow cannula malposition was evaluated by left ventriculography (Figure 9, A) before study termination or was measured during necropsy. During the necropsy, the blood pumps were disassembled, and no pump thrombosis was observed. Figure 9 shows necropsy pictures of all DCT inflow cannulae implanted. Animal 1038 was warfarin resistant and presented with multiple clot formations around the border of the inflow ostium and the outflow graft anastomosis site. Animal N28 showed tissue overgrowth around the mesh suture cuff, revealing that the cannula tip compressed the ventricular septum. Hyperplasia of inflammatory granulation tissue was observed around the gap between the myocardium and the inflow cannula cuff. This hyperplastic tissue was mainly composed of white fibrotic tissue, which was friable and extended into the cannula ostium. Histopathology study indicated that infiltration of inflammatory cells, hypergranulomatous cells, and myocardial necrosis appeared to be from inappropriate suturing techniques. Myocardial necrosis was found around almost the entire circumference of the cored myocardium section at the LV apex in Animal N28 as opposed to other animals without remarkable myocardial necrosis and tissue hyperplasia.
Although originally conceived as a means of bridging heart failure patients to transplantation, more than 50% of LVAD implantations are currently performed as destination therapy and result in an 81% survival rate 1 year after implantation.9 Despite recent state-of-the-art cf-LVAD technologies, adverse events do occur. In particular, post-LVAD stroke remains a major concern.10 Left ventricular assist device inflow cannulae, the cannula–myocardium interface, and cannulation techniques have been implicated as contributing sources of thrombus formation through distortion of blood flow patterns within the ventricular chamber. The distorted patterns result in nonphysiologic focal areas of turbulence, shear stress, and stasis, which in turn activate coagulation pathways.11–14
The inflow cannula can create conditions for early wedge thrombus formation as a result of its positioning within the ventricular chamber and the relationship to the apical geometry at the insertion site.15 The risk for thrombosis appears to peak within 3 months after implantation and may contribute to a high embolic stroke rate.9 The inflow cannula itself can also serve as a site for pannus establishment and overgrowth.16
After the initial postoperative phase, the hazard rate of LVAD stroke generally decreases and plateaus around 1 year postoperatively and then gradually increases over time.17 This may be because of the healing process of the LV apical core. After myocardial coring for inflow cannula placement, the current clinically approved LVAD inflow cannulae provide hemostatic control only at the epicardial surface. The cored myocardium is left exposed to blood components and is prone to increased local platelet adhesion and thrombus formation as part of the natural wound-healing process. As this exposed myocardium heals and new endothelium encapsulates the side of the inflow cannula, emboli may be generated, possibly leading to pump thrombosis and neurologic events.
Malposition of the inflow cannula tip can occur during LVAD implantation as a result of technical failure or migration of the cannula after cardiac remodeling.7 A malpositioned inflow tip that touches the myocardial wall can trigger thrombus, microemboli generation, and pannus formation that can occlude the inflow tip, further contributing to hemolysis, platelet activation, and post-LVAD stroke.6 Inflow cannula malposition can also result in ventricular wall suction, increasing the risk of arrhythmia or low pump flow followed by possible pump thrombosis.18 The PREVENtion of heartmate II pump thrombosis study investigated adherence to optimal surgical procedures in HeartMate II (HMII) patients and recommended medical management in relationship with clinical outcomes including stroke-free survival.19 The evidence showed that optimization of the LVAD implantation technique significantly impacts clinical outcomes.
Optimizing inflow cannula design is, therefore, crucial for minimizing adverse events and stroke risk by eliminating early thrombus formation and improving long-term survival. The EVAHEART DCT inflow cannula (Figure 3) prevents protrusion into the LV chamber and eliminates myocardial exposure to blood components by moving the hemostatic area to the endocardial surface. This unique double-cuff design eliminates the protruding intraventricular inflow tip, which is sensitive to malposition and is a possible root cause of wedge thrombus formation. The wide distal suture cuff accommodates a range of myocardial thicknesses from a thin apex wall to a thick hypertrophic myocardium. The DCT inflow cannula can theoretically tolerate malposition, thus avoiding ventricular suction and septal impingement owing to its tipless design.
This series of animal studies was considered a “worst-case” condition regarding inflow malposition and hypertrophic apex (healthy bovine myocardium is generally hypertrophic), yet there was no evidence of wedge thrombus or ventricular wall suction judging from the necropsy data and trend data for pump parameters such as pump power wave forms, which were continuously monitored. This result implies forgiveness of the DCT inflow cannula against malposition and adaptability to myocardial variability. Meticulous anticoagulation therapy is important, particularly in the acute phase (the initial few weeks up to 2 months after implantation) until the inflow suture cuff is endothelialized. If warfarin is not managed within the target therapeutic range during the postoperative acute phase, continuous heparin administration may be required.
Despite forgiveness against inflow malposition, there may be a possible surgical pitfall associated with the DCT inflow cannula suturing technique. Pledgeted mattress sutures that penetrate the cored myocardial wall from the epicardium into the LV should securely anchor the endocardium to avoid exposure of the cut myocardium cross-section to the bloodstream. It is important to pull the anchored endocardium into the mesh suture cuff. If the suture is threaded through the cut myocardial surface without properly anchoring the endocardium, the suture may tear and expose the myocardium, which may activate blood coagulation factor, platelet aggregation, and eventually thrombus formation and hyperinflammation around the suture line.
The current DCT design and surgical technique require CPB; however, clinical outcomes with the DCT inflow may be expected to outweigh any risks associated with CPB during implantation. To create an adequate surgical view in the LV apex, a good venous drainage strategy is necessary with CPB. Vacuum-assisted drainage can be an option to achieve empty beating condition in the LV. A suture guide can also help to organize the 12 mattress sutures and stabilize the landing zone at the myocardial cut edge for proper suture positioning. Another surgical technique is the needle entry point at the DCT mesh cuff; it is advisable to aim the threading point 2–3 mm below the inflow ostium line of the mesh. This helps to keep the polished metal surface at the inflow ostium away from the pulled endocardium. Rigorous surgical training is required to master these key techniques and prevent pannus formation.
Concerning the duration of the animal study, we rationalized that a 20 day study was an appropriate duration to confirm hematological adaptation in the early phase after LVAD implantation compared with a 65 day study. However, a long-term study (65 day) is still crucial for confirming tissue adaptation. According to current reports on adult heart failure cohorts, the initial stroke event risk peaks at 30 days after LVAD implantation.9,17 Therefore, it is important to investigate the presence of red thrombus and other biologic depositions (i.e., platelet-driven white thrombus and fibrin deposition) at the interface between the inflow cannula (or inflow cuff) and the myocardial tissue in the LV before tissue ingrowth is established around the inflow cannula (65 day study). Using this rationale, we proposed a combined 20 day and 65 day study for the preclinical animal study protocol.
Pannus formation at the inflow ostium remains a long-term concern. It is challenging to ascertain the optimal study duration for long-term cardiac assist devices. Although bovines have been widely used for preclinical safety validation, the overgrowth tendency of the experimental animal limits the extended study duration beyond 60 days. In this study, a 60 day duration was used to evaluate chronic tissue adaptation and antithrombogenicity. But a 60 day study is not long enough to preclude the potential risk for pannus formation, which may obstruct the inflow ostium over time. Therefore, the inflow cannula (test article) was implanted with marked malposition at the hypertrophic myocardial wall, defined as the “worst-case scenario.” Mid-sized animal models such as goats or sheep may be alternatives for evaluating long-term pannus. However, there are practical challenges to validate similar long-term implantation models for human bridge to heart transplantation (>1 year) to an even more extended application such as destination therapy. Further investigation into the DCT inflow cannula’s longer-term behavior vis-a-vis pannus formation is warranted.
The DCT inflow was adapted in the EVAHEART 2 pump, which was miniaturized in size, although the surfaces that contact blood and the blood pathways remained identical. Pump hydraulic performance and hemocompatibility are unchanged between EVAHEART 1 and 2. Bartoli et al. reported characterization of von Willebrand factor (vWF) degradation in a comparison of the EVAHEART LVAS and the HMII by use of a mock circulatory loop with human blood. Both pumps were run at speeds near the higher end of their clinical ranges, 2,300 rpm (5.7 ± 0.1 L/minute) and 11,400 rpm (6.3 ± 0.8 L/minute), respectively. HeartMate II caused significantly higher degradation of vWF high-molecular-weight multimer than did the EVAHEART pump.20 This was likely because of the design of the EVAHEART LVAS impeller, which requires a lower pump speed and larger flow gaps (>700 μm compared with 50 μm in the HMII), creating less shear stress (22 Pa of wall shear stress compared with 55 Pa of wall shear stress in the HMII). In another study, May-Newman et al.5,21 reported that the length of the protruding inflow cannula and inflow angle in the LV alter clockwise and counterclockwise voltex flow pattern. A protruding cannula diminishes the washout at the base of the protruding top, which can lead to blood stagnation and wedge thrombus between the inflow wall and the myocardial wall. Furthermore, a protruding inflow cannula may induce superphysiological shear stress and impact total system hemocompatibility in addition to the blood pump. Further investigation may be required to characterize how the DCT inflow can impact hemocompatibility, specifically hemolysis, and platelet factors including vWF-related parameters.
As opposed to conventional inflow cannulae, with either blunt or beveled tips, which may be predisposed to inflow obstruction because of the inflow tip geometry in the LV chamber, a trumpet-type inflow tip may be a possible design solution to achieve nonprotrusion or minimal protrusion into the LV.22,23 Although inserting the trumpet mouth into the ventricular chamber is technically challenging because of the size mismatch between the trumpet mouth diameter and the cored apex, this concept appears to be feasible for short-term application. But the most critical area susceptible to thrombus formation or pannus formation is the interface between the endocardium and the flange of the inflow. For long-term use, it may be advisable to surgically secure the endocardium plane to the suture cuff without exposing the cut myocardium. The proximal suture mesh cuff of the EVAHEART 2 DCT inflow allows 12 mattress sutures to pull and engage the endocardium, providing the benefit of suture line immobilization and enhanced tissue ingrowth onto the mesh surface.
Inflow malposition may be well tolerated by the new EVAHEART 2 DCT inflow cannula. This tipless inflow design may mitigate the risk of cannula wedge thrombus formation and ventricular wall suction. Further investigation with additional animal studies may be required to confirm long-term pannus formation around the inflow ostium.
The authors specially thank Erin Zebrowski for her data analysis and editorial contributions to this article.
1. 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: 2241–2251.
2. Yamazaki K, Kihara S, Akimoto T, et al. EVAHEART: an implantable centrifugal blood pump for long-term circulatory support. Jpn J Thorac Cardiovasc Surg 2002.50: 461–465.
3. Saito S, Yamazaki K, Nishinaka T, et al. J-MACS Research Group: Post-approval study of a highly pulsed, low-shear-rate, continuous-flow, left ventricular assist device, EVAHEART: a Japanese multicenter study using J-MACS. J Heart Lung Transplant 2014.33: 599–608.
4. Yamane T, Nishida M, Kawamura H, Miyakoshi T, Yamazaki K. Flow visualization for the implantable ventricular assist device EVAHEART®. J Artif Organs 2013.16: 42–48.
5. May-Newman K, Moon J, Ramesh V, et al. The effect of inflow cannula length on the intraventricular flow field: an in vitro
flow visualization study using the Evaheart left ventricular assist device. ASAIO J 2017.63: 592–603.
6. Milano CA, Simeone AA, Blue LJ, Rogers JG. Presentation and management of left ventricular assist device inflow cannula malposition. J Heart Lung Transplant 2011.30: 838–840.
7. Adamson RM, Mangi AA, Kormos RL, Farrar DJ, Dembitsky WP. Principles of HeartMate II implantation to avoid pump malposition and migration. J Card Surg 2015.30: 296–299.
8. Starling RC, Moazami N, Silvestry SC, et al. Unexpected abrupt increase in left ventricular assist device thrombosis. N Engl J Med 2014.370: 33–40.
9. Kirklin JK, Naftel DC, Pagani FD, et al. Seventh INTERMACS annual report: 15,000 patients and counting. J Heart Lung Transplant 2015.34: 1495–1504.
10. Mehra MR, Goldstein DJ, Uriel N, et al. Two-year outcomes with a magnetically levitated cardiac pump in heart failure. N Engl J Med 2018.378: 1386–1395.
11. Hanke JS, Krabatsch T, Rojas SV, et al. In vitro
evaluation of inflow cannula fixation techniques in left ventricular assist device surgery. Artif Organs 2017.41: 272–275.
12. Chiu WC, Alemu Y, McLarty AJ, Einav S, Slepian MJ, Bluestein D. Ventricular assist device implantation configurations impact overall mechanical circulatory support system thrombogenic potential. ASAIO J 2017.63: 285–292.
13. Wong KC, Büsen M, Benzinger C, et al. Effect of rotary blood pump pulsatility on potential parameters of blood compatibility and thrombosis in inflow cannula tips. Int J Artif Organs 2014.37: 875–887.
14. Ong C, Dokos S, Chan B, et al. Numerical investigation of the effect of cannula placement on thrombosis. Theor Biol Med Model 2013.10: 35
15. Taghavi S, Ward C, Jayarajan SN, Gaughan J, Wilson LM, Mangi AA. Surgical technique influences HeartMate II left ventricular assist device thrombosis. Ann Thorac Surg 2013.96: 1259–1265.
16. Strickland KC, Watkins JC, Couper GS, Givertz MM, Padera RF. Thrombus around the redesigned HeartWare HVAD inflow cannula: A pathologic case series. J Heart Lung Transplant 2016.35: 926–930.
17. Frontera JA, Starling R, Cho SM, et al. Risk factors, mortality, and timing of ischemic and hemorrhagic stroke with left ventricular assist devices. J Heart Lung Transplant 2017.36: 673–683.
18. Sorensen EN, Kon ZN, Feller ED, et al. Quantitative assessment of inflow malposition in two continuous-flow left ventricular assist devices. Ann Thorac Surg 2018.105: 1377–1383.
19. Maltais S, Kilic A, Nathan S, et al. PREVENtion of HeartMate II Pump Thrombosis Through Clinical Management: the PREVENT multi-center study. J Heart Lung Transplant 2017.36: 1–12.
20. Bartoli CR, Kang J, Zhang D, et al. Left ventricular assist device design reduces von Willebrand factor degradation: a comparative study between the HeartMate II and the EVAHEART left ventricular assist system. Ann Thorac Surg 2017.103: 1239–1244.
21. May-Newman K, Marquez-Maya N, Montes R, et al. The effect of inflow cannula angle on the intraventricular flow field of the left ventricular assist device-assisted heart: an in vitro flow visualization study. ASAIO J 2018.XX: XXX–XXX.
22. Liu G, Zhou J, Hu S, et al. [Numerical simulation of LVAD inflow cannulas with different tips]. Sheng Wu Yi Xue Gong Cheng Xue Za Zhi 2013.30: 141–148.
23. Antaki JF, Dennis TJ, Konishi H, et al. An improved left ventricular cannula for chronic dynamic blood pump support. Artif Organs 1995.19: 671–675.