A Cavopulmonary Assist Device for long-term Therapy of Fontan Patients

Please cite this article as: Andreas Escher MSc , Carsten Strauch MSc , Emanuel J. Hubmann MSc , Prof. Michael Hübler MD , Dominik Bortis PhD , Bente Thamsen PhD , Marc Mueller PhD , Ulrich Kertzscher PhD , Prof. Paul U. Thamsen PhD , Prof. Johann W. Kolar PhD , Prof. Daniel Zimpfer MD , Marcus Granegger PhD , A Cavopulmonary Assist Device for long-term Therapy of Fontan Patients, Seminars in Thoracic and Cardiovascular Surgery (2021), doi: https://doi.org/10.1053/j.semtcvs.2021.06.016


Perspective Statement
Long-term cavopulmonary support remains in its infancy. This study presents a novel cavopulmonary assist device for chronic support in an inclusive Fontan patient population.
In-silico and in-vitro analysis delivered the preclinical proof for a fully implantable, hemocompatible device design. Acute and chronic in-vivo trials are proposed to support laboratory findings.

Structured Abstract
Objective: Treatment of univentricular hearts remains restricted to palliative surgical corrections (Fontan pathway). The established Fontan circulation lacks a sub-pulmonary pressure source and is commonly accompanied by progressively declining hemodynamics. A novel cavopulmonary assist device (CPAD) may hold the potential for improved therapeutic management of Fontan patients by chronic restoration of biventricular equivalency. This study aimed at translating clinical objectives towards a functional CPAD with preclinical proof regarding hydraulic performance, hemocompatibility and electric power consumption.
Methods: A prototype composed of hemocompatible titanium components, ceramic bearings, electric motors, and corresponding drive unit was manufactured for preclinical benchtop analysis: hydraulic performance in general and hemocompatibility characteristics in particular were analyzed in-silico (computational fluid dynamics) and validated in-vitro. The CPAD's power consumption was recorded across the entire operational range. Introduction Univentricular hearts (UVHs) account for approximately 10% of all congenital heart defects. 1 The majority of patients with UVH undergo a Fontan type palliation with total cavopulmonary connection (TCPC). 2 Given the absence of a sub-pulmonary ventricle after TCPC completion, pulmonary perfusion is driven by elevated central venous pressures. This is associated with several long-term complications directly related to chronic venous congestion including lymphatic dysfunction, reduced cardiac output and liver fibrosis. [3][4][5] These complications ultimately result in a failing Fontan circulation 6,7 that presents a primary source of mortality in patients with UVH.
Currently, cardiac transplantation is the only long-term treatment option for patients with failing Fontan circulation. However, due to the limited availability of donor hearts and the complexity of the procedure, 8 cardiac transplantation remains controversially discussed.
Meanwhile, the living Fontan population is predicted to double within the next 20 years, 9 underpinning the medical need for alternative long-term treatment strategies.
Mechanical circulatory support (MCS) in failing Fontan patients is challenging and has only been anecdotally reported, [10][11][12][13][14][15][16][17][18][19] with poorer results than in biventricular patients with heart failure. 15 Recent advances in the field led to the introduction of MCS devices that are specifically intended for chronic cavopulmonary support. 20,21 Despite the seminal potential of those concepts, it remains unclear whether they meet the clinical demands to adequately support the heterogeneous Fontan population ranging from pediatric to adult patients with individual pursuits of physical activity and potentially accompanying secondary disorders (e.g. elevated pulmonary vascular resistance (PVR) or systemic ventricular insufficiency).
Within an interdisciplinary initiative to meet the medical need for a durable MCS option accessible to an inclusive Fontan population we recently introduced a cavopulmonary assist device (CPAD) specifically designed to substitute the missing sub-pulmonary ventricle. 22 The aim of the present study was to translate clinical objectives into a corresponding functional, hemocompatible CPAD with subsequent preclinical evaluation. Focus was laid on clinically relevant aspects including the interaction between the CPAD and the cardiovascular system, hemocompatibility as well as electric power consumption.

Mechanical Design and Vascular Connection
Clinical requirements regarding the anatomical compliance of a CPAD include the demand for (i) small-sized conception to prevent squeezing of surrounding sensible structures, (ii) high durability for chronic application as well as (iii) versatile and stable vascular connection to fit an inclusive range of patients.
Above requirements were translated into the CPAD design ( Figure 1D). Once implanted ( Figure 1A), blood from the inferior (IVC) and superior vena cava (SVC) is entering the flow chamber ( Figure 1B-C) via two inflow cannulae (Ø11mm), while being rerouted into the left (LPA) and right pulmonary artery (RPA), respectively, along two outflow cannulae (Ø12mm). Distal ends of the cannula in-and outlets are spaced by 34mm and 40mm, respectively. Pressure rise is generated by a four-bladed impeller (Ø19mm, h=9.5mm, medical-grade titanium). The impeller is supported within a circular flow chamber (Ø30mm, h=19mm, medical grade titanium) using blood-immersed mechanical ball-cup bearings (ball: ruby, cup: silicon-carbide whiskers reinforced aluminum oxide) ( Figure 1B).
Versatile anastomosis of the CPAD to the patient-specific vasculature is realized with custom-made conical grafts. Designed for a diameter evolving from 20 to 11mm on the inflow (IVC, SVC), and from 12 to 20mm on the outflow side (LPA, RPA), respectively ( Figure 1C, bottom), the grafts are to be surgically secured on the respective in-and outflow cannulae. The conical shape permits graft shortening to the required vessel diameter facilitating optimal vascular anastomosis. The grafts were manufactured with a previously developed electrospinning device 23 22 Blood is entering the flow chamber through its inflow cannulae that are connected to the IVC and SVC, and radially ejected through its outlet cannulae which are anastomosed to the LPA and RPA, respectively. B. CAD explosion view of the CPAD that is composed of a four-bladed impeller which is suspended within the circular flow chamber using mechanical ball-cup bearings. Via driveline, the impeller is actuated by the electromagnetic force that is generated by the coupling between motor rotor and motor stator. The stators are protected from corrosion by means of 3D-printed stator cap prototypes. C. Functional prototype of the CPAD (top) including the setting with both its in-and outlets connected to the custom-made electrospun conical grafts (bottom). Distal ends of the cannula in-and outlets are spaced by 34mm and 40mm, respectively. The total volume of the CPAD amounts to a magnitude of 17.

Hydraulic Design
The support of a heterogenous Fontan population with potentially accompanying secondary disorders (e.g. elevated PVR or systemic ventricular insufficiency) at distinct physical activities may require diversified magnitudes of flow and cavopulmonary pressure rises.
Accordingly, clinical requirements for a CPAD include the demand for (i) efficient operation across a broad range of flow (Q=0-10L/min) and pressure heads (H=0-50mmHg) to provide freedom for physiologically-controlled support of pediatric and adult patients across all clinically relevant conditions, (ii) to increase blood flow with rising venous return, (iii) to exhibit low resistance towards venous return in the event of stalled pump condition, (iv) to operate at low traumatic and thrombogenic potential and (v) to deliver a homogenous mixture of the hepatic factor to the left and right lung. Loss of the hepatic factor may lead to the degeneration of the pulmonary vasculature. 24 Above demands were accounted for in the hydraulic conception of the CPAD. Based on turbomachinery principles the CPAD was hydraulically designed for a rotational speed (n) of 2500rpm and flow rates of 4L/min (design point). An imbalanced inflow ratio (IR) of Q IVC /Q SVC =2:1 was deemed representative for a typical condition in young adolescent Fontan patients. 25 Gap dimensions (w=500μm) were designed as a trade-off accounting for efficient operation, appropriate motor cooling, prevention of pump occlusion (passage of floating thrombi) and reduction of flow obstruction during dysfunctional condition.

Actuation Design
Clinical requirements for the electric actuation of a CPAD include the demand for (i) efficient operation to prevent local blood temperature rises above 2°C due to motor heat losses (ISO14708-1), (ii) low power consumption to enable the integration of transcutaneous energy transfer (TET) technologies, (iii) a failsafe design to prevent device dysfunction, and (iv) minimal dimensions to fit within the small-sized CPAD.
An electric motor complying with above requirements was previously optimized for its specific application in this CPAD. 26 To ensure a failsafe design, the motor concept was realized by a redundant axial-flux three-phase synchronous motor configuration with stators in the upper and lower flow chamber casing, respectively, and permanent magnets integrated at the top and bottom of the impeller ( Figure 1B). Motor stators were sealed with epoxide resin, each of which covered with a 3D-printed stator cap prototype (Formlabs, Massachusetts, USA) (Supplementary Section 1).

In-silico Hemocompatibility Prediction
Compliance of the CPAD with stipulated clinical demands regarding hydraulic performance was verified using computational fluid dynamics (CFD) with the package Star CCM+ (Siemens, Munich, Germany) ( Figure 2A, Supplementary Section 2).
As a measure of hemocompatibility, normalized indices of hemolysis (NIH) were computed 27 and volume portions exposed to shear stresses above 9, 50 and 150Pa, respectively, identified. 28,29 The corresponding analysis was performed for an IR of 2:1 across the operational range.
In addition, a passive scalar transport model was incorporated for virtual pump washout analysis during design point operation. The same routine was followed to evaluate the distribution of the hepatic factor to both the LPA and RPA, respectively, for IR's of 2:1 and 3:1. Blood stagnation was defined for velocities below 0.1m/s, 28,29 thus complementing the washout analysis to predict the thrombotic potential within the pump.

In-vitro Hydraulic Characterization and CFD Validation
For in-vitro validation of CFD data, a hydraulic testbench was recently realized, 30 specifically tailored to accurately characterize hydraulic performance of the CPAD (Supplementary Section 3). Via 3/8'' silicon tubings the CPAD was integrated into the four TCPC-mimicking flow paths of the in-vitro testbench ( Figure 2B). The testbench was filled with blood analogue (water-glycerol mixture, ρ=1110kg/m 3 , μ=3.0mPa s, T=37°C) while pressure heads and flow rates were recorded for pump operation at rotational speeds of n=1500-3900rpm for both balanced (Q IVC /Q SVC =1:1) and imbalanced (Q IVC /Q SVC =2:1, Q IVC /Q SVC =3:1) IR's. Potential flow obstruction imposed by a failing CPAD was furthermore inspected by recording the pressure drop across the pump during stalled pump condition (n=0rpm).

In-vitro Hemolysis Assessment
Hemolysis experiments were conducted as previously described, 28

Hydraulic and Hemocompatibility Characteristics
In-silico Hemocompatibility Properties CFD data showed the CPAD to operate at hydraulic efficiencies (η hyd ) (Supplementary Section 4) above 30% across a broad range (Q=2-8L/min). Peak efficiencies were identified around design point operation (η hyd =45.79%) ( Figure 3A). Figure 3. In-silico hydraulic performance and blood trauma prediction. A. Numerically computed hydraulic efficiency of the CPAD with three exemplary pressure-flow curves (n=1800, 2500, 3200rpm, IR=2:1) indicating a comprehensive operational range of high hydraulic efficiency (η hyd >30%) performance. B. Numerically predicted NIH for the CPAD operated at rotational speeds of 1800, 2500 and 3200rpm with an IR of 2:1 indicating an increase towards low-flow operation, however with low-level values (NIH<1.43mg/100L) across a broad clinically relevant range. C. Computed blood volume portions exposed to shear stresses above 9, 50 and 150Pa across the pump's operational range (IR=2:1), expressed as percentage of the pump's priming volume. Volumes exposed to the respective shear stress levels tend to increase with increasing flow rate. CPAD: cavopulmonary assist device; n: rotational speed; IR: inflow ratio; NIH: normalized index of hemolysis.
The numerically predicted NIH increased during low-flow, low-efficiency operation (Q=1-3L/min, Figure 3B). However, peak values remained below 1.43mg/100L across the clinically relevant range. Blood volumes exposed to shear stresses above 9, 50 and 150Pa remained below 10, 0.4 and 0.01% of the priming volume in the CPAD, respectively, with noticeable rise towards increasing flow rates ( Figure 3C).
At design point operation, 90 and 95% of the old blood was washed out within t 90 =0.21s (8.7 revolutions) and t 95 =0.26s (11 revolutions), respectively ( Figure 4A). Further, blood stagnation with velocities below 0.1m/s was observed in 1.77% of the entire priming volume.
In the event of imbalanced IR's, the CPAD delivered well-mixed homogeneous outflow to both LPA (50.2% and 49.6% of IVC blood during 2:1 and 3:1 IR condition, respectively) and RPA (49.8% and 50.4% of IVC blood during 2:1 and 3:1 IR condition, respectively) ( Figure   4B). A. In the process of the virtual washout experiment the CPAD is run with design point settings (n=2500rpm, Q=4L/min, H=12.55mmHg, IR=2:1), while old blood (visualized in red, t 1 ) is continuously replaced with the newly entering blood represented in blue. After 3 revolutions (t 2 =0.072s) the new blood is increasingly mixing with the old blood consequently displacing the old blood towards the LPA and RPA outlets. In-vitro Hydraulic Properties and CFD Validation

Discussion
This study aimed at (i) the translation of key clinical demands for cavopulmonary support towards a functional prototype and (ii) the verification of its compliance with clinical demands regarding hydraulic performance, hemocompatibility and electric power consumption.
We delivered the proof-of-feasibility to manufacture the previously proposed novel CPAD 22 in an advanced prototype that meets key demands regarding hemocompatibility, robustness and vascular connection. Pump components complied with material selections of widespread application in implantable blood pumps with a bearing design that comes with long-term experience in both hemocompatibility and durability given its low wear profile. 31 Previous results pointed towards low-level bearing forces combined with a well-washed bearing configuration, which may mitigate the risk of heat generation and thrombus deposition. 22 Further, the electrospun grafts provide the potential for versatile, leak-tight and tear-resistant anastomosis of the CPAD to patient-specific vasculatures. In-silico and in-vitro findings revealed the CPAD to operate at low traumatic and thrombogenic potential with little electric power consumption across a comprehensive range of clinically relevant hemodynamic conditions ( Figure 8). In recent years, the scientific community has witnessed subtle progress in the field of longterm mechanical cavopulmonary support. Rodefeld et al. 10 were the first to describe the potential for cavopulmonary support using a double-inlet, double-outlet rotary blood pump (RBP) -a seminal work that was continued in the past decade. 20 Their actual device design aims at modest pressure step-ups around 6mmHg arguing such low head pressure levels to potentially suffice for chronic restoration of biventricular equivalency under physiologic conditions. To date, it remains controversially discussed whether such low cavopulmonary pressure rise is reliably sufficient, particularly in a heterogenous Fontan population with distinct pursuits of physical activity and potentially accompanying secondary disorders (e.g.

elevated PVR or systemic ventricular insufficiency).
We believe that chronic cavopulmonary support should be accessible to a heterogenous population including pediatric and adult patients at distinct health states and individual pursuits of physical activity. Consequently, the aspiration of this study was to develop a CPAD operating at wide ranges of pressure step-up's (H=0-50mmHg) and flow rates (Q=0-10L/min) to permit sufficient freedom for physiologically controlled, comprehensive destination therapy in an inclusive Fontan population. To enhance hemodynamic condition and avoid impaired venous return during physical activity, a physiologic control algorithm for automated speed modulation is anticipated as either sensor-based 32 or sensor-less 33 strategy.
An assistive device capable to work in a similarly broad operational range was recently introduced for long-term mechanical support of Fontan patients. 21 Accomplishing the first successful completion of a chronic in-vivo trial with a right heart substitute, feasibility for chronic CPAD implantation was underpinned. Yet, this CPAD is currently constrained by considerably larger dimensions and markedly elevated electric power consumption as compared to the herein presented device.
The demand for low traumatic operation covering all clinically relevant hemodynamic conditions imposes intricate challenges: RBPs are designed for a distinct operating condition, accompanied by adverse flow conditions during off-design operation. 28 Nevertheless, in-silico simulations disclosed suitability of the CPAD for efficient support across an inclusive patient population of distinct health states (η hyd >30% at Q=3-8L/min).
With 1.77% and 4.15% of the blood volume being exposed to velocities below 0.1m/s and 0.2m/s, respectively, during design point operation, the CPAD can be attested similar blood stagnation potential as compared to the frequently implanted HeartWare Ventricular Assist Device (HVAD, Medtronic, Minneapolis, USA) 28 and the HM3. 29 Additionally, the CPAD proved effective pump washout with periods for 90% and 95% replacement of old blood remaining below values reported for the HVAD 28 and the HM3, 29 respectively. In-silico hemocompatibility findings were complemented by in-vitro hemolysis experiments.
Experimentally determined values of NIH were more than an order of magnitude smaller than values presented by Giridharan et al. 34  Thus, we realized a CPAD that accounts for a trade-off between efficient, low-traumatic broad-range operation in functional state and low obstructive behavior in the event of pump malfunction (flow resistance: 0.63-3.15mmHg/(L/min) (Wood Units)). Yet, it remains to be addressed, whether such flow resistance is acceptably tolerated by the patient in case of device dysfunction.
To mitigate the risk of flow obstruction due to device dysfunction, we focused on a failsafe motor conception by redundant dual motor configuration. In addition, the experience with the HM3 may indicate device malfunction (1.6% at 2 years) 35 and pump thrombosis (<0.01 events per patient-year) 36 to be considered a rare event in RBPs with similar design characteristics. Thus, features which may contribute to this excellent failure rate and outcomes of the HM3 (large gap design of 500µm may prevent occlusive pump thrombosis) were integrated in the proposed CPAD. This contrasts with other devices presented for the same application that incorporate tiny clearance gaps. 20,21 Given the large gap dimensions within the CPAD, efforts were taken to optimize the size and efficiency of the electric motors. Around its clinically most relevant operating points, the electric power expenditure of the CPAD (<1.5W) was substantially below values reported for contemporary MCS devices and the long-term cavopulmonary support systems introduced above. 20,21 Further, the monotonic relationship between power consumption and pump flow with discernible dependence on imbalanced IR's revealed the potential for robust and reliable pump flow estimation based on intrinsic pump parameters even in light of varying IR's. 37 Omitting the need for additional sensors, such neat approaches could create new avenues for informed monitoring of the cardiovascular system's state in response to device support.
The integration of TET technologies requires the CPAD to accommodate an internal TET component whose implantable battery size proportionally scales with the pump's power demand. Thus, considering a usable volumetric energy density of 0.125 Wh/mL, 38 the herein presented low electric power consumption may permit the battery size to be reduced by approximately 70% compared to equivalent application in current LVADs which run at higher power expenditures. Hence, the design of a fully implantable system with TET technologies seems promising and may substantially facilitate long-term support of Fontan patients at high patient mobility with eliminated risk for driveline associated adverse events.

Limitations
Except for the in-silico analysis of the IVC/SVC mixing behavior during IR's of 2:1 and 3:1, respectively, the numerical simulations presented herein are limited to the representative consideration of a typical IR of 2:1.
Further, the reliable numerical estimation of NIH across the operational range is hampered by substantial discrepancies in magnitudes among in-silico and in-vitro data. This constitutes the well-known limitation of current in-silico hemolysis predictions being constrained to comparative evaluations, while failing to replicate absolute measures. Further, direct comparison of NIH computations with in-vitro measurements is hindered given the assumption of smooth surfaces in the simulation setup as opposed to the surfaces in the current prototype that lacks optimal surface finish.
Device implantability was yet solely investigated in a virtual fitting study of a 11-year-old patient. 22 Feasibility to implant the herein presented device in a heterogenous Fontan population remains to be confirmed in a larger virtual fitting study that also accounts for younger patients with strongly limited anatomical space.

Conclusion
Given the inclusive operational range, the promising hemocompatibility properties and the low electric power consumption the proposed cavopulmonary assist may offer a promising option for the long-term therapy of Fontan patients. These findings underpin the rational for further development by means of acute and chronic in-vivo investigations to confirm hemodynamic benefit in chronic disease associated with the Fontan circulation.
Legend Section Figure 1. Mechanical design of the CPAD for implantation in TCPC position. A. CPAD implanted in TCPC location between IVC, SVC, LPA and RPA in a heart with single ventricle (SV). 22 Blood is entering the flow chamber through its inflow cannulae that are connected to the IVC and SVC, and radially ejected through its outlet cannulae which are anastomosed to the LPA and RPA, respectively. B. CAD explosion view of the CPADthat is composed of a four-bladed impeller which is suspended within the circular flow chamber using mechanical ball-cup bearings. Via driveline, the impeller is actuated by the electromagnetic force that is generated by the coupling between motor rotor and motor stator. The stators are protected from corrosion by means of 3D-printed stator cap prototypes. C. Functional prototype of the CPAD (top) including the setting with both its in-and outlets connected to the custom-made electrospun conical grafts (bottom). Distal ends of the cannula in-and outlets are spaced by 34mm and 40mm, respectively. The total volume of the CPAD amounts to a magnitude of 17.   In-silico hydraulic performance and blood trauma prediction. A. Numerically computed hydraulic efficiency of the CPAD with three exemplary pressure-flow curves (n=1800, 2500, 3200rpm, IR=2:1) indicating a comprehensive operational range of high hydraulic efficiency (η hyd >30%) performance. B. Numerically predicted NIH for the CPAD operated at rotational speeds of 1800, 2500 and 3200rpm with an IR of 2:1 indicating an increase towards low-flow operation, however with low-level values (NIH<1.43mg/100L) across a broad clinically relevant range. C. Computed blood volume portions exposed to shear stresses above 9, 50 and 150Pa across the pump's operational range (IR=2:1), expressed as percentage of the pump's priming volume. Volumes exposed to the respective shear stress levels tend to increase with increasing flow rate. CPAD: cavopulmonary assist device; n: rotational speed; IR: inflow ratio; NIH: normalized index of hemolysis. Figure 4. Numerical prediction of old blood washout and IVC/SVC blood mixing behavior. A. In the process of the virtual washout experiment the CPAD is run with design point settings (n=2500rpm, Q=4L/min, H=12.55mmHg, IR=2:1), while old blood (visualized in red, t 1 ) is continuously replaced with the newly entering blood represented in blue. After 3 revolutions (t 2 =0.072s) the new blood is increasingly mixing with the old blood consequently displacing the old blood towards the LPA and RPA outlets. After 8 revolutions (t 3 =0.192s), 87.83% of the old blood is replaced with new blood. B. During design point operation (n=2500rpm, Q=4L/min, H=12.55mmHg), however, with highly imbalanced IR of 3:1 blood that is entering through the IVC (denoted as dark blue) is homogeneously mixed with the inflowing blood of the SVC, equalizing outflow distribution of IVC-blood to both LPA and RPA outlet. IVC: inferior vena cava; SVC: superior vena cava; LPA: left pulmonary artery; RPA: right pulmonary artery; CPAD: cavopulmonary assist device; n: rotational speed; Q: flow rate; H: pressure head; IR: inflow ratio.    Video 1. This video provides an illustration of the clinical challenges that are accompanying patients with established Fontan circulation. It indicates the unmet medical need of an appropriate long-term treatment option for the heterogeneous Fontan population and proposes a novel cavopulmonary assist device as a potential solution for chronic restoration of biventricular equivalency. Further, this video provides insight into key aspects of the device design, preclinical evaluation and corresponding clinical implications.