The Fontan operation has provided palliation for children born with cardiac malformations resulting in a single functional ventricle. The Fontan circulation enables systemic venous blood to bypass the right ventricle and flow directly into the pulmonary arteries, which allows the use of the single ventricle as the left heart. Because of improvements in operative procedures, timing, staging, and management, operative mortality is now less than 5% and survival at 20 years is 84%.1 As a steadily increasing population of these children are surviving into adulthood they essentially suffer from right heart failure with elevated central venous pressure. Morbidity is significant because of venous congestion and elevated pulmonary vascular resistance (PVR), progressive ventricular dysfunction, dysrhythmias, hypoxemia, and protein-losing enteropathy.2 Heart transplantation is the only successful treatment for these patients. However, many failed Fontan patients are not eligible for heart transplantation because of organ failure and other complications from their chronic right heart failure. There are a number of mechanical blood pumps under development for short-term Fontan support.3–6 These are usually catheter-based devices that are acceptable in an acute hospital setting but not viable as a long-term option.
Continuous flow left ventricular assist devices are increasingly used to provide long-term support for adults with end-stage heart failure. We adapted this technology to the design of a continuous flow blood pump to serve the requirements of the failed Fontan circulation and function as a permanent right heart replacement device. The purpose of the study is to evaluate the right heart replacement pump in a 30 day sheep study.
A CAD model of the pump design is shown in Figure 1A and the final assembled pump used in the implant is shown in Figure 1B. The pump consists of two inlet ports for direct insertion between the superior vena cava (SVC) and the inferior vena cava (IVC). The one pump outlet connects to the main pulmonary artery (PA). The diameters of the IVC and SVC inlet cannulae are 20 and 16 mm, respectively. The inlets taper to 10.2 mm for the IVC and 7.6 mm for the SVC. The diameter of the volute outlet is 12 mm, which matches the diameter of the PA cannula. The larger IVC versus SVC port allows for a 60:40 split in pump flow that matches the requirements for the Fontan circulation.7 The bladed rotor is supported by a hydrodynamic journal bearing with 28 µm side clearance and 760 µm end-gap spacing on both the IVC and SVC.
Multiple design iterations on the number and shape of the rotor blades were evaluated using computational fluid dynamics (CFD) simulations and in vitro performance testing. Initial testing found that a six-bladed rotor was overpowered for the low head pressures of the right-sided circulation and required low rotational speeds that compromised the hydrodynamic journal bearing. A three bladed rotor was selected to meet the performance requirements at a sufficient rotational speed. Figure 2 shows the design iteration of the rotor blade shape: curved (Figure 2A), straight (Figure 2B), and straight with closed backside (Figure 2C). The three rotor shapes were evaluated for hemolysis and the latter rotor (Figure 2C) was used in the chronic animal study.
A custom in-house designed embedded motor controller (STMicroelectronics, Geneva, Switzerland) was used to control pump speed. A field oriented control algorithm with sinusoidal commutation was implemented to provide precise control of rotor speed with minimal torque ripple. The controller was based on a system developed in our group for control of continuous flow blood pumps.8
Materials and Methods
In Vitro Testing
Pump hemolysis was tested in vitro using a small volume (400–500 ml) mock circulatory loop, filled with heparinized, freshly drawn ovine blood. The loop was configured with separated IVC and SVC inlets fed from a common reservoir with an adjustable resistance to set the operating condition. Pressure was measured using disposable pressured transducers (Argon Medical Devices, Inc., Plano, TX). Pump head pressure (H) was defined as the difference between the outlet pressure and the average of the two inlet pressures. Pump flow (Q) was measured on both inputs and the common output using three inline flow probes (ME13PXN, Transonic System Inc., Ithaca, NY). Three operating conditions were tested: 1) Q = 5 lpm, H = 15 mm Hg; 2) Q = 5 lpm, H = 20 mm Hg; and 3) Q = 5 lpm, H = 30 mm Hg. At each operating condition, plasma-free hemoglobin (PlHb) was measured every 25–30 minutes, and the rate of PlHb increase was used to calculate the normalized index of hemolysis (NIH, units of mg/dl) as follows:9
where, d(plHb)/dt is the time rate change of the plasma-free hemoglobin [(mg/dl)/min], Hct is the average hematocrit (%) at the start and end of each operating condition, V is the loop blood volume (L), and Q is the pump flow rate (lpm).
The pump performance curve (head pressure versus flow) was obtained at pump speeds of 2,500, 3,000, 3,500, 3,900, and 4,500 rpm in the mock circulatory loop with ovine blood. At each pump speed, the resistance was adjusted to control flow from 1 lpm to max flow in 1 lpm intervals, while the pump head pressure and power were measured.
In Vivo Implant Procedure
General anesthesia and endotracheal intubation was provided. The sheep (93 kg, 3 year old male) was positioned right side up and a right thoracotomy was performed in approximately the fifth intercostal space. The right internal mammary artery was exposed and cannulated with a catheter for arterial pressure monitoring. The right lung was packed away posteriorly with moist laparotomy sponges and malleable retractors. The pericardium was opened in a longitudinal fashion anterior to the phrenic nerve to expose the SVC and IVC. The ascending aorta was exposed by rightward retraction of the SVC. Following heparin dosing, the aorta, SVC, and IVC were cannulated and connected to the bypass circuit. Cardiopulmonary bypass was initiated and the right heart decompressed. Ventilation was continued during bypass.
The right heart was disconnected from the return circulation by cutting the IVC and SVC at the right atrium, but remained connected to the PA to eject return blood from the coronary sinus. After clamping and transecting the SVC and IVC, the venae cavae stumps near the right atrium were oversewn. The pump was placed in the right chest. The two polytetrafluoroethylene (PTFE) inlet grafts to the pump were anastomosed end-to-end to the SVC and IVC, respectively. A ringed PTFE outlet graft was anastomosed end-to-side to the PA. Pump flow was initiated and increased as bypass flow was simultaneously weaned and eventually discontinued. Surgical hemostasis was obtained and the bypass cannulae were removed. Fluid filled pressure monitoring lines were placed in the SVC, IVC, and PA and brought out individually through the posterior chest wall.
Starting on postoperative day (POD) 0, the animal was given 3,000 mg Kefzol Q8 for 7 days, 278 mg carprofen Q24 for 3 days, 280 mg famotidine Q24 for 3 days, buprenorphine Q4 (0.5–0.9 mg) for 11 days, and a trifusion of butorphanol, lidocaine, and ketamine for 24 hours. Global coagulation was assessed using standard assays for thrombolestography (TEG) R-time, prothrombin time (PT), partial thromboplastin time (PTT), activate clotting time (ACT), and fibrinogen concentration. On POD 1, a standard anticoagulation protocol was started beginning with unfractionated heparin and transitioning to Coumadin. Prothrombin time did not increase during the study, and therefore, we chose to maintain a low heparin dose throughout the study, rather than risk increasing the Coumadin dose. Heparin activity was measured using a chromogenic anti-Xa assay. Chest tubes were pulled on POD 10. The animal was euthanized after 30 days and the pump was removed and examined for wear and thrombosis.
In Vitro Testing
The in vitro hemolysis results are provided in Table 1 for the three rotors shown in Figure 2. Both rotors with straight blades had superior hemolysis results compared with the curved bladed rotor. The straight bladed rotor with closed backside (Figure 2C) that was used in the chronic implant had NIH values at or below 0.001 mg/dl for the three operating conditions tested. For the same rotor, the performance curves for head pressure versus flow and power versus flow are shown in Figure 3, A and B, respectively. The slopes of the head pressure versus flow relationships were −4 mm Hg/lpm at pump speeds of 2,500 and 3,000 rpm, −5 mm Hg/lpm at 3,500 rpm, and −6 mm Hg/lpm at 3,900 and 4,500 rpm. The slopes of the pump power versus flow curve were 0.1 W/lpm at pump speeds of 2,500 and 3,000 rpm, 0.2 W/lpm at 3,500 and 3,900 rpm, and 0.3 W/lpm at 4,500 rpm. The hematocrit was 28% during the in vitro performance testing.
In Vivo Implant
Following device implant, pump speed was set at 3,900 rpm for the duration of the study. In the first 2 days, postoperatively pump power was 4.4 W, dropped to 4.3 W on POD 3, and then steadily increased to 4.6 W at the end of the study (Figure 4A). The pump inlet pressures for the SVC and IVC were 14 ± 15 and 11 ± 15 mm Hg, respectively, over the duration of the study (Figure 4B). The PA catheter migrated into the right ventricle during the study and could not be used as a measure of pump outlet pressure.
The animal had hypoproteinemia-induced subcuticular edema of the cranial chest that resolved over several days with diuretics. An initial subcutaneous emphysema over the left lateral thorax and up to the drive line resolved over several days. The coagulation data are shown in Figure 5. Thrombolestography R-time peaked at 194 min on POD 1 (not shown on scale of Figure 5A) and remained at 37 ± 19 min for the remainder of the study. Partial thromboplastin time steadily increased from 30 s preoperatively to a high of 59 s on POD 20, while PT remained at 20 ± 2 s for the duration of the study. As shown in Figure 5B, ACT peaked at 335 s at the time of implantation and remained stable at 152 ± 12 s postoperatively. Fibrinogen concentration remained between 150 and 400 mg/dl. Heparin was started at 10 U/kg/hr for the first 15 days postoperatively (Figure 5C). The heparin dosage was temporarily dropped on PODs 3 and 9 because of suspected bleeding events that were resolved. Between PODs 13 and 15, the TEG R-times dropped to the preoperative range of 10–20 min, and therefore, on heparin was increased to 20 U/kg/hr for the remainder of the study. Heparin anti-Xa values were less than 0.2 UI/ml, indicating minimal heparin activity. Coumadin was started on POD 1 at 5 mg and increased to 20 mg on POD 9 and maintained at 20 mg for the duration of the study.
Liver and kidney function assay results are shown in Figure 6. Alkaline phosphatase levels remained below preoperative levels, serum glutamic oxaloacetic transaminase (SGOT) were elevated immediately after implantation then returned to baseline, and alkaline phosphatase decreased following surgery and were slightly elevated at the end of the study (Figure 6A). Total protein and albumin levels decreased and bilirubin increased after surgery, and all levels returned to baseline at the end of the study (Figure 6B). Creatine and blood urea nitrogen (BUN) levels remained stable for the duration of the study (Figure 6C).
Hematology data are shown in Figure 7. Following surgery, hematocrit dipped to 25% on POD 7 and returned to the preoperative level of 35% by the end of the study (Figure 7A). White blood cell count, platelet count, and PlHb remained at baseline levels throughout the study (Figure 7A). Creatine phosphokinase (CPK) increased immediately following surgery before returning to baseline and lactate dehydrogenase (LDH) remained at baseline for the duration of the study (Figure 7B).
Necropsy was uneventful with no signs of complications. The lungs and kidneys were free of infarctions or emboli. Photographs from the pump teardown are shown in Figure 8. The IVC and SVC cannulae and pump inlet connectors were clean and clear of any deposits (Figure 8, A and B). The PA outlet connector junction had red thrombi around 40% of the joint circumference (Figure 8C). The outlet cannula showed a hard bend next to the pump outlet port and had red adhesions on the internal wall distal to the bend. The inside surfaces of the IVC and SVC inlets were completely clear of any deposits or defects (Figure 8, D and E). All rotor surfaces were clear of deposits except for small, approximately 0.5 mm diameter, thrombi between each of the three rotor blades (Figure 8F). There were thrombi formations along 20% of the parting line of the two volute halves.
This study was the first successful test of a right heart replacement device in a chronic animal study. Although we have performed hundreds of pump implants in animals,10–13 including total artificial hearts, the Fontan model is considered surgically challenging. We chose an approach by right thoracotomy because of the difficult recovery for ruminants with the sternal approach. The right heart was disconnected from the return circulation except for coronary sinus drainage, resulting in complete flow support through the test pump. The 30 day study was uneventful, and the pump was able to provide necessary support for the animal over the duration of the study. This study was highlighted by a successful implantation and postoperative recovery that demonstrated normal physiologic pulmonary and venous pressures and cardiac output.
Anticoagulation in patients with mechanical circulatory support devices is usually by warfarin (international normalized ratio [INR] range 2.5–3.5) and low dose aspirin. Aspirin is known to be ineffective in ruminants. Coumadin has been used routinely in bovines for mechanical circulatory support testing, but higher doses are required in ovines and a target INR can be difficult to maintain. We planned on using a standard protocol of beginning with unfractionated heparin and transitioning to Coumadin. However, the PT did not increase and therefore, we decided to maintain a low heparin dose throughout the study, rather than risk increasing the Coumadin dose.
At the conclusion of the study, the rotor showed no signs of wear and the rotor bore measured the same as before implant, indicating that the hydrodynamic journal bearing performed satisfactorily. There was little evidence of thrombogenicity in the pump. The pump connections and inside surfaces were clear of deposits. The only regions of concern were the thrombi deposited on the rotor and along the volute parting line. On examination of the volute, there was a gap in the split line for the two volute halves and the edges were not as sharp as intended. The pump manufacturing process has been improved to eliminate the gap, and we expect in future implants that the area will be clear of thrombus formations. Concerning the thrombi on the rotor (Figure 8F), we believe these are because of the shape of the blades and that improvements can be made in the blade design. The current blade shape was selected based on design iterations using CFD and in vitro hemolysis testing (Figure 2). The initial curved blade shape demonstrated high hemolysis that did not occur with the straight blade shape (Table 1). However, CFD simulations found a significant stagnation region behind the straight blade that was eliminated by extending the backside of the blade (Figure 2C). There were regions of low shear stress behind the rotor blades in the extended blade design that correspond to the locations of the three thrombi (Figure 8F). We are continuing to refine the rotor shape to eliminate this region of low shear stress and expect the improved rotor design to have better washing and reduced risk of thrombus formation.
The number of patients who would benefit from mechanical support can be estimated based on the number of failed Fontan circulations. The total number of Fontan cases can be estimated from the STS Congenital Heart Surgery Database, which reports from approximately 77% of the pediatric heart surgery programs. On average, 776 operations are performed each year. Extrapolating to include nonreporting centers, just over 1000 Fontan procedures are performed per year in the United States and Canada. Because of improvements in Fontan surgery, survival at 20 years is 84%.1 These adult patients suffer from complications because of elevated central venous pressures and chronic low cardiac output,2 and 16% suffer from failed Fontan circulation14 that require heart transplantation. Low PVR is required for adequate pulmonary blood flow in the Fontan circulation, and increased PVR is an indicator of a failed Fontan. Our pump has the potential to reduce morbidity in this patient cohort by reducing venous pressures and eliminating the need for heart transplantation. However, Fontan patients with chronic low cardiac output can develop ventricular dysfunction that would not be treatable with our device.2 For these patients, a heart transplant or some form of biventricular support or a total artificial heart would be necessary.
Efforts to develop mechanical support devices for failing Fontan patients have focused mainly on short-term support via intravascular, catheter-based rotary pumps.3–6 These devices are at an early stage of development, and are intended to provide only short-term, nonambulatory support, because they require a drive cable or catheter exiting through the femoral or jugular vein. The use of the HeartMate II LVAD (Abbot-Thoratec Co., Chicago, IL) has been investigated in acute animal studies15; however, rotary pumps designed as left ventricular assist devices (LVADs) are not configured for the Fontan anatomy and are not optimized for low head pressures, which may cause stagnation regions that could potentially induce thrombosis. The SynCardia Total Artificial Heart (SynCardia Systems, Inc, Tucson, AZ) has been used for failing Fontan support,16 but that device is primarily a bridge-to-transplant device and is not a preferred option for destination therapy. Our proposed device is specifically designed for the two inputs and low pressures of the Fontan circulation, and to provide long-term support that can be used as either bridge-to-transplant or destination therapy.
Pulmonary vascular resistance and systemic venous compliance were not measured in this study, as the primary objective of the study was to evaluate pump placement and function. Pump flow was not measured during the chronic study because ultrasonic transducers are incompatible with PTFE grafts and we did not want to add complications to the procedure with a flow probe on the aorta. Head pressure was expected to provide an estimate of pump flow, but unfortunately the PA pressure sensor migrated into the right ventricle and the readings were not reliable. Power consumption can be used to provide an estimate of pump flow base on the pump performance curve (Figure 3B). The power consumption during the study (Figure 4A) would correspond to a pump flow of approximately 6 lpm, which is reasonable for the size of the animal. Thrombus formation can cause an increase in power consumption, but on necropsy, the pump was clean and free from major deposits. Hematocrit variations can shift the pump power versus flow relationship, and the slight increase in hematocrit (Figure 7A) during the study may explain the observed power increase.
1. d’Udekem Y, Iyengar AJ, Cochrane AD, et al. The Fontan procedure: Contemporary techniques have improved long-term outcomes. Circulation 2007.116(suppl 11): I157–l164.
2. Gewillig M:. The Fontan circulation
. Heart 2005.91: 839–846.
3. Rodefeld MD, Coats B, Fisher T, et al. Cavopulmonary assist for the univentricular Fontan circulation
: Von Karman viscous impeller pump. J Thorac Cardiovasc Surg 2010.140: 529–536.
4. Throckmorton AL, Ballman KK, Myers CD, Frankel SH, Brown JW, Rodefeld MD. Performance of a 3-bladed propeller pump to provide cavopulmonary assist in the failing Fontan circulation
. Ann Thorac Surg 2008.86: 1343–1347.
5. Throckmorton AL, Kapadia J, Madduri D. Mechanical axial flow blood pump to support cavopulmonary circulation. Int J Artif Organs 2008.31: 970–982.
6. Wang D, Plunkett M, Gao G, et al. A practical and less invasive total cavopulmonary connection sheep model. ASAIO J 2014.60: 178–182.
7. Whitehead KK, Sundareswaran KS, Parks WJ, Harris MA, Yoganathan AP, Fogel MA. Blood flow distribution in a large series of patients having the Fontan operation: A cardiac magnetic resonance velocity mapping study. J Thorac Cardiovasc Surg 2009.138: 96–102.
8. Cysyk J, Jhun CS, Newswanger R, Weiss W, Rosenberg G. Rotary blood pump
control using integrated inlet pressure sensor. Conf Proc IEEE Eng Med Biol Soc 2011.2011: 373–376.
9. Shinowara GY. Spectrophotometric studies on blood serum and plasma; the physical determination of hemoglobin and bilirubin. Am J Clin Pathol 1954.24: 696–710.
10. Izraelev V, Weiss WJ, Fritz B, et al. A passively suspended Tesla pump left ventricular assist device. ASAIO J 2009.55: 556–561.
11. Mehta SM, Pae WE Jr, Rosenberg G, et al. The LionHeart LVD-2000: A completely implanted left ventricular assist device for chronic circulatory support. Ann Thorac Surg 2001.71(suppl 3): S156—S161; discussion S83—S84.
12. Rosenberg G, Snyder AJ, Weiss WJ, et al. Dynamic in vitro
and in vivo
performance of a permanent total artificial heart. Artif Organs 1998.22: 87–94.
13. Weiss WJ, Rosenberg G, Snyder AJ, Pae WE, Richenbacher WE, Pierce WS. In vivo
performance of a transcutaneous energy transmission system with the Penn State motor driven ventricular assist device. ASAIO Trans 1989.35: 284–288.
14. Greutmann M, Tobler D, Kovacs AH, et al. Increasing mortality burden among adults with complex congenital heart disease. Congenit Heart Dis 2015.10: 117–127.
15. Riemer RK, Amir G, Reichenbach SH, Reinhartz O. Mechanical support of total cavopulmonary connection with an axial flow pump. J Thorac Cardiovasc Surg 2005.130: 351–354.
16. Rossano JW, Goldberg DJ, Fuller S, Ravishankar C, Montenegro LM, Gaynor JW. Successful use of the total artificial heart in the failing Fontan circulation
. Ann Thorac Surg 2014.97: 1438–1440.