Mechanical circulatory support (MCS) is becoming a standard treatment option for heart failure patients. Because of an enduring shortage of donor organs, MCS becomes increasingly important as a bridge to transplant but also as bridge to destination.1,2 Partial MCS is a recent promising treatment option for heart failure patients, especially for those not qualified for a full support device (e.g., left ventricular assist devices [LVAD]).
The Circulite Synergy Micro-pump (HeartWare) is the first clinically approved partial MCS device in Europe. It connects the left atrium to the right subclavian artery (RSA) from where the blood is intended to flow retrograde through the innominate artery into the aorta (Figure 1). This device is implanted minimal-invasively (off-pump) requiring only a minithoracotomy,3 thereby lowering perioperative and operative morbidity and mortality. The device pumps between 3 and 4.25 L/min, and remnant native cardiac output (CO) is needed to maintain a sufficient circulation resulting in interacting blood flows from the device and the heart (Figure 1).
Several studies in the past have shown an improvement in hemodynamics after implantation of the Circulite Synergy Micro-pump, but induced changes in the hemodynamics are not yet well understood. Partial support was initially intended for patients not qualifying to receive a full support LVAD. Studies by Meyns et al.4 and Sabashnikov et al.5 showed promising results in New York Heart Association (NYHA) IIIB und early NYHA IV patients, i.e., improvements in cardiac index and decrease in pulmonary pressure along with an increase in exercise testing (6-minute walking test and VO2max). Following these initial results, also inotropic dependent patients (interagency registry for mechanically assisted circulatory support (INTERMACS) 1–3) were reported to benefit from partial support.5 Barbone et al.6 demonstrated that elderly patients, like younger ones, benefitted from the minimal invasive approach and showed improvement in exercise tolerance (i.e., increase in 6-minute walk test and VO2max).
Although partial circulatory support is thought to be a promising approach, several problems occurred during the use of this particular partial support device including common adverse events in MCS, such as thrombosis, hemolysis, infection, and fracture of the inflow graft, which led to a temporary market withdrawal in 2013.7 More insight in hemodynamic alterations would be beneficial for improving device design by identifying unfavorable hemodynamic alterations that cause adverse events, such as thrombosis, stroke, or hemolysis, with the ultimate goal of impacting patient outcome.
In vivo flow information is difficult to obtain in this patient group. Direct measurements of the blood flow are only available from Doppler ultrasound techniques because magnetic resonance imaging (MRI) is contraindicated due to the implanted device.8 A comprehensive three-dimensional (3D) understanding of the hemodynamics during the entire cardiac cycle is yet not available by standard imaging techniques. Computational fluid dynamics (CFD) simulations present an alternative, noninvasive method to gain insight into basic principles of flow phenomena. This method was already successfully applied for full support LVAD devices,9,10 aortic dissections,11 and cerebral aneurysms.12 The use of CFD simulations in the medical field started in neurosurgical research where simulations were run for cerebral aneurysm to analyze the likelihood of rupture. It then expanded into the cardiovascular research field, especially but not only to simulate alterations of hemodynamics in the use of mechanical circulatory assist devices and their theoretical testing and development. Nowadays CFD simulations also find a place in the clinical use in cardiovascular medicine, e.g., in the evaluation of coronal artery disease using fractional flow reserve (FFR).13,14 The CFD simulations calculate the blood flow in the arterial tree (aorta and supraaortic vessels) based on a computational mesh derived from segmented computed tomography angiographic (CTA) images and on estimated or measured inflow and outflow conditions. With this approach, a complete 3D description of hemodynamic parameters during the whole cardiac cycle is achieved.
In this study, CFD simulations were carried out for 10 patients with an implanted Circulite Synergy Micro-pump for the entire cardiac cycle to quantify changes in hemodynamics caused by the two competing blood flows from the heart and the device. A pilot CFD study in two patients with partial circulatory support demonstrated competing blood flows between the device and native CO.15 With our findings, we hope to promote the development of an improved partial assist device so that heart failure patients can benefit from this promising concept in the future.
Material and Methods
The investigation conforms with the principles outlined in the Declaration of Helsinki. After obtaining permission from the institutional ethics committee, CTA images of 10 patients (average age 50.6 ± 8.8 years, 5 male, 5 female) after implantation of the Circulite Synergy Micro-pump (average time between implantation and CTA: 15 ± 240 days with a minimum of 5 days and a maximum of 618 days) were selected for this retrospective study. Demographic and basic clinical data are summarized in Table 1.
Computational fluid dynamics is a widely used method to simulate the behavior of fluids in different settings. The calculations are based on the Navier-Stokes equations that provide the mathematical description of fluid flow behavior. This set of four partial differential equations is numerically solved in CFD software using in our case (STAR CCM+ software) the finite volume method on a polyhedral grid (mesh). The polyhedral mesh is based on individual 3D reconstructions of patients’ vessels. Several hemodynamic parameters, including velocity, wall shear stress (WSS), and pressure differences, are calculated for every single time step for each mesh volume.
Therefore, the lumina of the aorta, supraaortic vessels, and Circulite outflow cannula were manually segmented from CTA images (ImageJ 1.48v, National Institutes of Health) and used to reconstruct a 3D surface of the luminal boundaries (ParaView 4.3.1, Kitware Inc).
These reconstructions were imported into STAR CCM+ (version 10.6.010, cd-adapco) and used to create polyhedral meshes. A mesh independency study was performed (over a range of 100,000–2,500,000 cells), and an average mesh size of 500,000 cells was established for this study (mesh size range: 513,670–801,005).
Because of technical reasons, CFD simulations were carried out for two cardiac cycles, and data from the second cycle were analyzed. Eighteen time steps were recorded per cardiac cycle (duration of cardiac cycle was 0.85 s in steps of 0.05 s). Constant inflow from the partial assist device into the subclavian artery was set to 3 L/min for all 10 patients.16 Because transthoracic Doppler echocardiographically measured CO was only available for two of our patients,15 we calculated an estimated CO for the remaining eight patients from echo data with the Teichholz formula17 and then transferred into a cardiac inflow waveform using a normalized aortic inflow waveform. Average CO in our patient population was 4.0 ± 2.2 L/min (ranging from 1.0 to 7.3 L/min).
The outflow conditions for all outlets (supraaortic vessels, descending aorta) were set as a pressure outlet with 0.0 Pa as commonly used in CFD simulations of this kind.
All simulations were carried out with a non-Newtonian fluid model (Carreau-Yasuda model).18 To include turbulence effects of the hemodynamic changes, a turbulent flow model (k-epsilon-model) was used.
Postprocessing of Simulation Results
Patients were divided into two groups depending on their native CO after implantation of the partial assist device (low CO [<3 L/min]: 1.7 ± 0.7 L/min, n = 4 as opposed to high CO [>4 L/min]: 5.5 ± 1.1 L/min, n = 6; p < 0.001; Table 1) to compare hemodynamic alterations in these two groups of patients.
Simulation results were visualized in Paraview (version 5.0, Kitware Inc.) and exported to Matlab (R2015a, The MathWorks, Inc.) for further statistical analysis.
Velocity magnitudes, pressures differences, and WSS were averaged for each time step in the ascending and descending aorta, the aortic arch, and the supraaortic vessels. Flow values were calculated by multiplying average velocity values with the cross-section of the respective artery segments. All data were expressed as mean ± standard deviation. For each case, all 18 time steps were quantitatively and qualitatively analyzed. Pressure values mentioned in this study all represent pressure differences.
The Student’s t-test (significance level p < 0.05) was used to test for significance of differences in hemodynamic parameters between the two groups and different times of the cardiac cycle. Pearson’s R correlations coefficient was used to quantify correlations between different parameters. Pseudo-color representations were used to illustrate changes in velocity magnitude, total pressure, and WSS.
During diastole, retrograde flow through the innominate artery into the aortic arch was sustained by the partial assist device, essentially supporting the CO from the systemic circulation. During systole, this flow was reversed in 9 of the 10 cases due to the flow from the heart. Mean retrograde flow during diastole was 1.9 ± 0.6 L/min (mean velocity: 0.3 ± 0.2 m/s) while mean reversed (antegrade) flow during peak systole was −1.2 ± 1.5 L/min (mean velocity: −0.2 ± 0.2 m/s) (negative values indicate reversed flow).
In these nine cases, a collision of blood flow around the origin of the right common carotid was observed in the innominate artery during peak systole. Native CO reversed the flow from the device in the innominate artery (Figure 2). This flow reversal effect was significantly higher in the group with higher CO (above 4 L/min; n = 6) (mean reversed flow: −2.2 ± 1.1 L/min) than in the cases with a CO lower than 3 L/min (n = 4) (mean flow: 0.2 ± 0.6 L/min) (p = 0.002) (Figure 3). Reversed flow during peak systole inversely correlated with the remaining native CO (R = −0.7172; p = 0.0196). One patient in the low CO group (CO: 2.3 L/min) did not show any reversed flow over the entire cardiac cycle. This was probably because of a combination of low native CO and a wide medial anastomosis angle, which allowed more blood flow from the partial assist device toward the systemic circulation. In addition, a closer look at the flow patterns during peak systole in this case suggested a weak collision with the native CO at the origin of the innominate artery (Figure 4). In all other cases, areas of oscillating WSS were found in the innominate artery at the location of systolic flow collision. Also significantly elevated pressures were observed there as well (1,504.8 ± 1,034.9 Pa in peak systole vs. 87.08 ± 119.29 Pa during diastole; p = 0.0019) when compared with total pressures in diastole during which no flow collisions were observed. This phenomenon was also more pronounced in the high CO group (mean pressure elevation in high CO group: 2,063 ± 613 Pa compared with 450 ± 300 Pa; p = 0.0006) (Figure 5). A significant pressure elevation during peak systole was also found in the Circulite outflow cannula (1,785 ± 1,059 Pa in peak systole vs. 677 ± 574 Pa in diastole; p = 0.0115) while flow was steady at 3 L/min. Pressure elevation was more distinct in the high CO group as opposed to the low CO group (mean total pressure elevation in high CO group: 1,639 ± 842 Pa; low CO group: 311 ± 187 Pa; p = 0.0107).
In our simulation settings, the partial assist device increased blood flow in the descending aorta and supraaortic vessels by an average of 1.54 ± 0.78 L/min compared with CO before implantation of the partial assist device. More detailed analysis of the cardiac cycle showed that circulatory support was highest during diastole (mean net additional flow: 1.8 ± 0.6 L/min) and markedly lower during systole (mean net additional flow 0.5 ± 1.9 L/min) (p = 0.084). The high CO group showed a trend toward a higher decrease of support during systole than the low CO group (mean decrease of net additional flow: 1.7 ± 2.0 L/min [high CO group] vs. 0.5 ± 0.2 L/min [low CO group]; p = 0.226; Figure 6). All results are summarized in Table 2.
A better understanding of hemodynamic alterations in partial circulatory support is necessary to further improve this concept in heart failure treatment. The results from our CFD study demonstrated significant basic hemodynamic changes during the cardiac cycle.
Partial support of an average of 1.5 ± 0.8 L/min was found which is consistent with in vivo findings in previous clinical studies regarding the Circulite Synergy Micro-pump.4 Our results also demonstrate that the efficiency of partial support is not yet optimal throughout the entire cardiac cycle. During systole, native CO works against the flow from the Circulite Synergy Micro-pump causing a collision of blood flow which leads to areas of unordered blood flow with oscillating WSS and elevated pressures. Disturbed flow has previously been shown to induce and promote endothelial dysfunction.19 Furthermore, this antagonism between the two blood flows leads to elevated pressure in the Circulite outflow graft during systole, which might increase stress on the pump and potentially increase the risk of device related complications.
Partial support is used in a wide range of patients. In our simulations, the antagonism between partial support and native CO is found to be more distinct in patients with a higher CO. Because partial support is mostly an option used patients in earlier stages of heart failure, our simulations indicate that a synchronization of the supporting flow with the native cardiac cycle is therefore necessary to optimize support and reduce unfavorable hemodynamic alterations. Decreasing pump speed during systole to reduce likeliness of flow collision and increasing speed during diastole to maintain a sufficient support overall could be an option to improve partial support devices.
First results on a partial circulatory assist device that is synchronized with the cardiac cycle have been reported in a case study.20 Results from the first in-man use appeared promising despite difficulties encountered with placement of the device in relation to the subclavian artery.
In the development of new partial assist devices or the use of full support LVADs for partial support, the findings of this study support the concept of synchronization with the cardiac cycle leading to a more favorable hemodynamics in the aorta and supraaortic vasculature.
Cardiac recovery is a desirable but rare event during MCS.21 Even though unloading of the heart in principle supports the recovery of the myocardium, several studies raised the idea of a disuse myocardial atrophy in patients with full support MCS devices, decreasing the likelihood of recovery. Partial unloading (via partial support) may play a favorable role in this process due to a training effect of the heart as studies suggested.22,23 Antagonistic flow patterns may decrease or even completely prevent cardiac recovery by increasing cardiac strain during systole. Further studies on this topic are warranted.
This study has several limitations. Although CFD simulation data have been validated with in vivo data in several studies, our data derived from the simulations are still only a mathematically calculated model of in vivo hemodynamics. However, our methods have been established in prior studies.24–27In vitro validation using particle image velocimetry (PIV) showed that hemodynamic values derived from CFD simulations are within 10% accuracy.26,28,29In vivo validation using four-dimensional (4D) MRI measurement showed statistically significant correlations between measurements from CFD models and 4D MRIs (p < 0.00000001).30 We already included turbulent flow and the non-Newtonian behavior of human blood but omitted wall movement and blood damage models. Hemodynamic information from the supraaortic vessels in this model is very basic because we did not simulate cerebral autoregulation of the perfusion and used zero pressure outlets. Flow from the partial assist device was set to a continuous flow of 3 L/min disregarding fluctuations in flow due to pressure drops across the pump. However, in the clinical setting, the pump flow averaged around 3 L/min. In the model applied in this study, we observed backflow in the supraaortic vessels during systole because of the Bernoulli effect, which has been previously described in CFD studies of the subclavian cannulation.31,32 Another limitation of this study is the small sample size due to available CTA images after implantation of the Circulite Synergy Micro-pump. Although CFD simulations have their limitations, they have become an additional instrument for clinical imaging, e.g., in noninvasive FFR diagnostics.33 In our 3D model, we only included the section of outflow graft anastomosis with the subclavian artery joining the aorta. Therefore, we speculate that the geometry of the pump itself or reduction of preload through the apex in potential future models would not significantly influence the results of our simulations. The findings might be different if the outflow graft was anastomosed to the ascending aorta in future pumps.
Our CFD study demonstrates that the partial circulatory assist device Circulite Synergy Micro-pump provides sufficient support of the systemic circulation with a mean net additional flow of 1.5 ± 0.8 L/min in our simulations. Most importantly, blood flow from the heart and from the partial assist device collide in the innominate artery during systole leading to an area of oscillating WSS and elevated pressure. Considering the eventual introduction of the next generation of partial assist devices, we propose synchronization of partial support VADs with the cardiac cycle to increase effectiveness of future partial assist devices and reduce unfavorable hemodynamic alterations.
1. Jorde UP, Kushwaha SS, Tatooles AJ, et al; HeartMate II Clinical Investigators: Results of the destination therapy post-food and drug administration approval study with a continuous flow left ventricular assist device: A prospective study using the INTERMACS registry (Interagency Registry for Mechanically Assisted Circulatory Support). J Am Coll Cardiol 2014.63: 1751–1757.
2. Kirklin JK, Naftel DC, Pagani FD, et al. Seventh INTERMACS annual report: 15,000 patients and counting. J Heart Lung Transplant 2015.34: 1495–504.
3. Meyns B, Klotz S, Simon A, et al. Proof of concept: Hemodynamic response to long-term partial ventricular support with the synergy pocket micro-pump. J Am Coll Cardiol 2009.54: 79–86.
4. Meyns BP, Simon A, Klotz S, et al. Clinical benefits of partial circulatory support
in New York Heart Association Class IIIB and Early Class IV patients. Eur J Cardiothorac Surg 2011.39: 693–698.
5. Sabashnikov A, Popov AF, Bowles CT, et al. First experience with the Synergy Micro-Pump in patients in INTERMACS class 1-2 as a bridge to transplantation: Pushing the limits? Artif Organs 2015.39: 118–125.
6. Barbone A, Pini D, Rega F, Ornaghi D, Vitali E, Meyns B. Circulatory support in elderly chronic heart failure
patients using the CircuLite® Synergy® system. Eur J Cardiothorac Surg 2013.44: 207–212; discussion 212.
7. Schmack B, Gorenflo M, Rengier F, Ruhparwar A. Extempore interventional closure of a broken CircuLite Synergy micropump inflow graft defect. J Heart Lung Transplant 2013.32: 1140–1141.
8. Mohite PN, Bowles CT, Sabashnikov A, et al. Simple and non-invasive techniques to evaluate the function of CircuLite Synergy. Interact Cardiovasc Thorac Surg 2014.19: 738–742.
9. Medvitz RB, Reddy V, Deutsch S, Manning KB, Paterson EG. Validation of a CFD methodology for positive displacement LVAD analysis using PIV data. J Biomech Eng 2009.131: 111009.
10. Karmonik C, Partovi S, Loebe M, et al. Computational fluid dynamics
in patients with continuous-flow left ventricular assist device support show hemodynamic alterations in the ascending aorta. J Thorac Cardiovasc Surg 2014.147: 1326–1333.e1.
11. Karmonik C, Partovi S, Davies MG, et al. Integration of the computational fluid dynamics
technique with MRI in aortic dissections. Magn Reson Med 2013.69: 1438–1442.
12. Karmonik C, Klucznik R, Benndorf G. Blood flow in cerebral aneurysms: Comparison of phase contrast magnetic resonance and computational fluid dynamics
–preliminary experience. Rofo 2008.180: 209–215.
13. Carrizo S, Xie X, Peinado-Peinado R, et al. Functional assessment of coronary artery disease by intravascular ultrasound and computational fluid dynamics
simulation. Rev Port Cardiol 2014.33: 645.e1–645.e4.
14. Taylor CA, Fonte TA, Min JK. Computational fluid dynamics
applied to cardiac computed tomography for noninvasive quantification of fractional flow reserve: Scientific basis. J Am Coll Cardiol 2013.61: 2233–2241.
15. Karmonik C, Partovi S, Rengier F, et al. Hemodynamic assessment of partial mechanical circulatory support
: Data derived from computed tomography angiographic images and computational fluid dynamics
. Cardiovasc Diagn Ther 2015.5: 160–165.
16. Klotz S, Meyns B, Simon A, et al. Partial mechanical long-term support with the CircuLite Synergy pump as bridge-to-transplant in congestive heart failure
. Thorac Cardiovasc Surg 2010.58(suppl 2): S173–S178.
17. Seitz WS, Spiel MS. Echocardiographic formula for computation of left ventricular volume and stroke volume. Comparison with cardiac catheterization and the Teichholz formula. Jpn Heart J 1983.24: 863–869.
18. Boyd J, Buick JM, Green S. Analysis of the Casson and Carreau-Yasuda non-Newtonian blood models in steady and oscillatory flows using the lattice Boltzmann method. Physics of Fluids 2007.19: 093103.
19. Heo KS, Fujiwara K, Abe J. Disturbed-flow-mediated vascular reactive oxygen species induce endothelial dysfunction. Circ J 2011.75: 2722–2730.
20. Cecere R, Dowling RD, Giannetti N. Initial clinical experience with the Symphony heart assist system. Ann Thorac Surg 2015.99: 298–301.
21. Krabatsch T, Schweiger M, Dandel M, et al. Is bridge to recovery more likely with pulsatile left ventricular assist devices than with nonpulsatile-flow systems? Ann Thorac Surg 2011.91: 1335–1340.
22. Kinoshita M, Takano H, Taenaka Y, et al. Cardiac disuse atrophy during LVAD pumping. ASAIO Trans 1988.34: 208–212.
23. Maybaum S, Epstein S, Beniaminovitz A, et al. Partial loading of the left ventricle during mechanical assist device support is associated with improved myocardial function, blood flow and metabolism and increased exercise capacity. J Heart Lung Transplant 2002.21: 446–454.
24. Karmonik C, Partovi S, Loebe M, et al. Influence of LVAD cannula outflow tract location on hemodynamics in the ascending aorta: A patient-specific computational fluid dynamics
approach. ASAIO J 2012.58: 562–567.
25. Karmonik C, Partovi S, Schmack B, et al. Comparison of hemodynamics in the ascending aorta between pulsatile and continuous flow left ventricular assist devices using computational fluid dynamics
based on computed tomography images. Artif Organs 2014.38: 142–148.
26. Zhang J, Zhang P, Fraser KH, Griffith BP, Wu ZJ. Comparison and experimental validation of fluid dynamic numerical models for a clinical ventricular assist device. Artif Organs 2013.37: 380–389.
27. Morley D, Litwak K, Ferber P, et al. Hemodynamic effects of partial ventricular support in chronic heart failure
: Results of simulation validated with in vivo
data. J Thorac Cardiovasc Surg 2007.133: 21–28.
28. Kaufmann TA, Hormes M, Laumen M, et al. Flow distribution during cardiopulmonary bypass in dependency on the outflow cannula positioning. Artif Organs 2009.33: 988–992.
29. Kaufmann TA, Gregory SD, Büsen MR, Tansley GD, Steinseifer U. Development of a numerical pump testing framework. Artif Organs 2014.38: 783–790.
30. Miyazaki S, Itatani K, Furusawa T, et al. Validation of numerical simulation methods in aortic arch using 4D Flow MRI. Heart Vessels 2017.32: 1032–1044.
31. Kaufmann TA, Hormes M, Laumen M, et al. The impact of aortic/subclavian outflow cannulation for cardiopulmonary bypass and cardiac support: A computational fluid dynamics
study. Artif Organs 2009.33: 727–732.
32. Laumen M, Kaufmann T, Timms D, et al. Flow analysis of ventricular assist device inflow and outflow cannula positioning using a naturally shaped ventricle and aortic branch. Artif Organs 2010.34: 798–806.
33. Nørgaard BL, Leipsic J, Gaur S, et al; NXT Trial Study Group: Diagnostic performance of noninvasive fractional flow reserve derived from coronary computed tomography angiography in suspected coronary artery disease: The NXT trial (Analysis of Coronary Blood Flow Using CT Angiography: Next Steps). J Am Coll Cardiol 2014.63: 1145–1155.
Keywords:Copyright © 2018 by the American Society for Artificial Internal Organs
mechanical circulatory support; heart failure; computational fluid dynamics; partial circulatory support