Since 2005, we have been developing and evaluating a cardiac replacement device consisting of two continuous-flow ventricular assist devices (VADs). Various volume-displacement total artificial hearts (TAHs) have been developed over the past 50 years.1 However, the long-term clinical application of these pulsatile devices has been restricted, because of their large size, complexity, and limited durability.2 Continuous-flow implantable VADs have been in development since the 1980s and are small enough to implant in a greater percentage of patients.3 These VADs are robust; they have only one moving part and do not have valves or contain flexible membranes, so mechanical wear is minimized.4,5 Furthermore, continuous-flow VADs have some sensitivity to inflow pressure, which enables them to imitate the native heart by autonomously increasing pump flow in response to an increasing preload. This feature facilitates the balance of systemic and pulmonary flow when two continuous-flow VADs are used to replace the heart, and it may decrease the need for moment-to-moment pump speed adjustment.6–8
In developing and evaluating our continuous-flow total artificial heart (CFTAH) model over the past 8 years, we made numerous refinements in surgical technique and in postoperative care and treatment. Progressive improvements were made in the design of the CFTAH inflow atrial cuffs. We also added a tracheostomy to the surgical procedure, thus enhancing postoperative respiratory care. Other refinements in postoperative care included close monitoring of volume status to prevent fluid accumulation, provision of nutritional supplements to stimulate appetite, performance of routine blood-culture studies to monitor infection status, hourly flushing of indwelling pressure catheters to prevent clots, and modulation of right pump speed to reduce thrombus entrainment and formation. Each of these refinements aided postoperative recovery and prolonged the animals’ survival.
Thus far, we have implanted the CFTAH in 65 calves. The primary endpoint of these experiments has been survival to 90 days, allowing us to evaluate the degree to which continuous flow is compatible with the physiologic needs of the animals. This report describes our experience with the 29 calves that survived for 7–92 days.
Materials and Methods
Corriente Crossbred calves (weighing 70–126 kg) were used for the CFTAH implants. All animals received humane care in compliance with the Principles of Laboratory Animal Care (National Society of Medical Research) and the Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication no. 85-23, revised 1996). Our Institutional Animal Care and Use Committee approved all protocols used for the experiments.
Continuous-Flow Total Artificial Heart
Five types of rotary VAD pairs were configured for integration into a CFTAH (Figure 1): the Jarvik 2000 (Jarvik Heart, Inc., New York, NY), the HeartMate II (Thoratec Corporation, Rancho Cordova, CA), the HeartMate III (Thoratec Corporation), the HeartWare HVAD (HeartWare International, Inc., Framingham, MA), and the MicroMed DeBakey/Heart Assist 5 (MicroMed Cardiovascular, Inc., Houston, TX). The original curved pump inlet cannulas of the HeartMate II and the MicroMed pumps were replaced with short, straight inlets designed to attach to our custom-fabricated atrial cuffs made of silicone-impregnated Dacron. The atrial cuffs were specifically sized to fit the inlets of the Jarvik 2000, the HeartMate III, and the HeartWare HVAD. Fluid-filled pressure lines were placed into the atria or generally attached directly to each atrial cuff to measure the left atrial pressure (LAP) and right atrial pressure (RAP). The outflow portion of each pump consisted of a 10–16 mm collagen- or gelatin-impregnated polyester graft (such as GelWeave; Vascutek, Inchinnan, Scotland, or Hemashield; Maquet, Wayne, NJ) that was externally supported by a plastic casing to prevent kinking. With each pump, a Transonic flow probe (Transonic Systems, Inc., Ithaca, NY) or a custom, integrated flow probe (MicroMed pumps only) was placed on the outflow graft to measure systemic and pulmonary blood flow. Each pump also had its own driveline, which was brought out through the chest wall and was connected to a controller, power supply, and system monitor.
Anesthesia and Surgical Technique
In each case, glycopyrrolate, diazepam, and ketamine were intravenously administered to induce sedation. After being intubated, the calf was given isoflurane (0.5–3.0%) in oxygen (40–100%) to maintain general anesthesia. The animal was placed in the dorsal or right lateral recumbent position. A tracheostomy was performed in 25 of the 29 animals. The chest was exposed via either a median sternotomy (n = 4) or a left thoracotomy (n = 25), and a catheter was placed in the internal thoracic artery to measure the arterial pressure (AoP). After heparinization, bypass cannulas were inserted, and cardiopulmonary bypass (CPB) support was initiated. Specifically, venous cannulas were placed in the left jugular vein and the inferior vena cava, and arterial cannulas were inserted into the descending thoracic aorta and the left carotid artery. Dual arterial cannulation was performed to allow clamping of the aortic arch and innominate artery to provide an adequate length of ascending aorta for the aortic anastomosis. The ascending aorta was cross-clamped and the pulmonary artery and aorta were transsected above their respective ventriculoarterial valves. The right and left atria were then transected cephalad to the mitral and tricuspid annuli, respectively, and all ventricular and valvular tissue was removed. Customized cuffs were attached to the left and right atrial remnants by using a continuous suture technique and Teflon strip supports. The pump inflow cannulas were positioned through the sewing cuffs and secured with band ties. The outflow grafts of the left and right pumps were anastomosed in end-to-end manner to the aorta and pulmonary artery, respectively. All lines were tunneled to exit the animal’s left side. The pumps were de-aired, and pumping was initiated; pump speeds were adjusted as necessary to achieve targeted pressures and flow, with the left pump generally set higher to account for differences in the pulmonary and systemic vasculature. The animal was weaned from CPB, and the bypass cannulas were removed. Chest tubes were placed in the left and right pleural cavities for postoperative fluid drainage, and all incisions were closed routinely.
Postoperative Care and Data Collection
Each calf was continuously monitored throughout the postoperative period and was given analgesics and antibiotics as needed for pain management and prevention of infection. Prophylactic gastrointestinal protectant medications, including omeprazole and a probiotic gel, were administered daily throughout the case. Total parenteral nutrition and megestrol acetate were given if needed to provide further nutritional supplementation and appetite stimulation. Anticoagulation was achieved with intravenous heparin and/or warfarin, based on the animal’s coagulation profile. Heparin therapy was generally maintained until all chest tubes and the tracheostomy tube were removed.
Physiologic pressures (AoP, RAP, LAP, and pulmonary artery pressure [PAP]) were continuously recorded by a computerized data-acquisition system (Data Sciences International, St. Paul, MN). Device parameters—including but not limited to pump speed, pump flow, and power consumption—were recorded on an hourly basis. Pump speeds were adjusted as needed, based on the calf’s overall condition, to maintain LAP and RAP in the range of 10–20 mm Hg and systemic pressure in the range of 80–100 mm Hg. Animals with a tracheostomy underwent routine care and cleaning of the site, including frequent replacement of the inner tracheostomy cannula to prevent accumulation of secretions. Ventilatory settings were adjusted as needed to maintain normal oxygenation and acid–base balance. Volume status was monitored and controlled, and systemic and pulmonary vascular resistance was managed pharmacologically with vasoactive agents and diuretics, if needed to maintain targeted pressures. Fluid restriction was frequently necessary to maintain balance in spontaneously drinking animals. Hematologic and biochemistry parameters and blood-culture specimens were drawn at least weekly and assessed for signs of bleeding, hemolysis, end-organ dysfunction, or infection.
Eight animals underwent treadmill exercise studies. The calf was transferred to a motorized treadmill and exercised for various durations at varying speeds. Pressures (AoP, RAP, LAP, and PAP) were monitored, and arterial and/or venous blood-gas values were analyzed at baseline, at each treadmill speed, and at the conclusion of exercise to determine the animal’s oxygen consumption (VO2) and hemodynamic response to exercise.
Seven calves were returned to the operating room immediately before study termination, and radiopaque-contrast material was injected through the pressure lines into the CFTAH circuit to assess pump position, inflow and outflow graft position/geometry, graft integrity, and flow distribution. After euthanasia, a necropsy was performed, and the pumps, atrial cuffs, and end-organs were evaluated and photographed.
Figure 2 shows the pumps used and the survival durations. After receiving the CFTAH, the 29 calves survived for 7–92 days (mean, 24 days). The mean survival time was 14 days with the Jarvik 2000 (n = 2), 31 days with the HeartMate II (n = 2), 50 days with the HeartMate III (n = 2), 29 days with the HeartWare HVAD (n = 3), and 20 days with the MicroMed DeBakey/Heart Assist 5 (n = 20).
Two of the 29 calves survived for the targeted elective case duration of 90 days. Sixteen animals were euthanized before 90 days for various problems that included thrombus (which appeared to originate from the periphery and lodge in the right pump) and infection/sepsis. Five animals developed thrombus in the jugular vein, resulting in unstable pressures and flows. The remaining six calves were euthanized for other reasons, including sepsis (n = 2), renal failure (n = 1), an equipment mishap that caused the left pump to become disconnected for 10–15 minutes (n = 1), and a neurologic insult after tracheostomy tube replacement (n = 1); the sixth animal was euthanized after reoperation for bleeding (n = 1). After CFTAH implantation, six animals were returned to the operating room because of suspected bleeding, fluid accumulation within the chest cavity, or the need for catheter replacement to aid postoperative monitoring. Two of the six calves died during the secondary procedure, and the remaining four animals survived for up to 7 days after the procedure.
Throughout each case, pressures were maintained within the acceptable physiologic range (Figure 3). The LAP and RAP were maintained at relatively higher pressures than normal to facilitate flow and pressure balancing during changes in physiologic state (i.e., changes in posture). Right pump flow was maintained at 9.1 ± 1.7 L/minute and left-pump flow at 9.4 ± 1.9 L/minute, with bronchial collateral flow accounting for the left-to-right differential. Pump stoppages occurred in four cases, one of which contributed to the animal’s death, as noted above. In the remaining three cases: 1) the right pump stopped briefly when the driveline accidentally became disconnected; 2) the left pump stopped for approximately 40 seconds when it was temporarily connected to battery power; and 3) the controller became briefly disconnected from power, causing both pumps to decrease their speed to 5,000 rpm. In each of these three instances, the issues were quickly corrected, allowing the pumps to return to normal function before any distress was experienced by the animals.
Hematologic and biochemistry values in the two animals that survived to 90 days remained mostly within the normal range throughout the case, and no significant abnormalities were noted. In a majority of the cases that were terminated before their intended duration, the animal’s condition slowly deteriorated over a 1–3 day period, complicating the assessment of hemodynamic values and hematologic and biochemistry parameters. Before this deterioration occurred, however, hematologic and biochemistry parameters were mostly in the acceptable range. Occasional increases were noted in the white blood cell count (Figure 4A), and 10 calves had one or more positive blood-culture results that necessitated antibiotic treatment. Hematocrit (HCT) and hemoglobin (Hgb) levels decreased within the first postoperative month before increasing steadily thereafter (Figure 4B). Occasional blood products were transfused to offset the decreases in HCT and Hgb in the initial postoperative weeks. Modest transient increases were observed in the total and direct bilirubin values in all the calves during the early postoperative period (Figure 4C). These values gradually returned to baseline after the first postoperative month. Increases in blood urea nitrogen and creatinine levels were occasionally observed because of temporary periods of anuria that were treated with diuretic therapy (Figure 4D). In two animals, the anuria could not be treated solely with diuretic therapy, so continuous venovenous hemodialysis or continuous arteriovenous hemofiltration (CAVH) was required. In these two cases, the average daily difference between volume input and output was minimal to negative. The average daily difference between volume input and output for all the animals was 1,292 ± 1,050 ml.
Twenty-five of the 29 calves underwent tracheostomy. All of the animals were able to eat, drink, and ruminate while receiving ventilatory support. Fifteen animals had the tracheal cannula in place from implantation to case conclusion. Six calves were able to be permanently removed from the ventilator and to have their tracheal cannula removed; they survived for 3–86 days afterward. In these cases, varying degrees of tracheal stenosis were observed at necropsy, ranging from minimal to severe, with a residual airway diameter of 5 mm in one animal, which survived for 48 days. The remaining four calves had their tracheal cannula removed between 3 and 9 days postimplant, but the cannula later had to be replaced and maintained until the end of the case. In the first case, this was necessary before the initiation of CAVH; in the second case, the cannula was replaced after right pump flow decreased during a triple-lumen catheter exchange resulted in thromboembolus to the right pump; in the third case, recannulation was performed as a precaution after the animal developed profuse nasal bleeding due to prolonged insertion of a nasal oxygen catheter; and in the fourth case, the animal’s overall condition deteriorated, rendering extubation intolerable. The 90- and 92-day survivors had their tracheal cannula permanently removed on postoperative days 8 and 6, respectively.
Treadmill Exercise Studies
Eight of the 29 calves underwent treadmill exercise studies, and three of these animals underwent two separate treadmill studies. In all treadmill studies, the calves walked for a total of 6–30 minutes at various speed settings, ranging from 0.8 to 1.6 mph. All animals generally tolerated the exercise well. Throughout the exercise evaluations, systemic and pulmonary flow remained balanced, and it increased autonomously by 2–4% in response to increased venous return and filling pressures. A normal, steady rise was noted in blood lactate levels, and VO2 was comparable with physiologic total-body VO2 during exercise.
End-organ dysfunction was difficult to evaluate in the calves whose cases were terminated before the intended 90 days of duration because of a 1–3 day decline in their overall condition. In 10 cases, thrombus was discovered in the right pump, cuff, and/or inflow graft and was believed to have contributed to the deaths of these animals. In each of these cases, right pump malfunction was evident with increased pump power and decreased flow before termination. In five of these calves, clotting was also observed in the jugular veins at necropsy, presumably due to an indwelling venous catheter on that side and/or narrowing and trauma of the jugular vein after venous cannulation for CPB and subsequent repair.
Infection appeared to be a contributing factor in four cases. In one calf, vegetative growth was discovered in both pumps, grafts, and atrial cuffs of the CFTAH circuit. Evidence of infection was observed in another calf in the left shoulder area, and acute bronchopneumonia was discovered; in a third animal, Enterobacter organisms were found in the left lung; and in a fourth animal, a mixed bacterial infection was discovered in the right pump, aorta, lungs, kidneys, and cerebral cortex. In the two calves surviving to term, no significant abnormalities were noted in the end-organs, and the pumps, atrial cuffs, and outflow grafts were free of thrombus.
From 2005 to 2012, 29 calves have been maintained with a CFTAH at our institute for 7–92 days. Since performing the initial cases, we have made several refinements to the surgical technique and postoperative care and treatment. Refinements to the surgical procedure have involved progressive modifications in the design of the inflow sewing cuffs, including the use of reinforced Dacron, which provides more uniform mechanical properties and facilitates cuff implantation.8
In the first four calves implanted with a CFTAH, a tracheostomy was not performed. Those animals experienced episodes of respiratory distress and required reintubation; during that period, they were unable to eat, drink, or receive oral medications, and this complicated their recovery. In the 25 subsequent calves, we performed a tracheostomy at the time of CFTAH implantation, and this significantly enhanced postoperative respiratory care. The tracheostomy eliminated the stress of repeated intubation for minor, acute episodes of respiratory distress, allowed rapid regulation of oxygen intake, and facilitated evaluation of pulmonary function.9
Our experience in optimizing surgical technique, perioperative, and postoperative management is reflected in the survival durations of our CFTAH-implanted calves. One of the first 20 animals included in our study lived for longer than 20 days. Of the next nine animals implanted, eight lived for longer than 20 days. In many of the initial cases, fluid accumulation in the cervical region and/or chest cavity necessitated a secondary surgical procedure to drain the fluid; alternatively, the animal was sedated, and additional chest drainage tubes were inserted into the left and right pleural cavities. Both procedures involved further treatment and adversely impacted recovery. Throughout our experience, increasing attention has been devoted to monitoring the volume status of the animals to prevent fluid accumulation by minimizing intravenous volume input and aggressively administering diuretic therapy as needed. In spontaneously drinking animals, this frequently necessitated fluid restriction. This strategy has allowed more rapid removal of chest drainage tubes, increasing the animals’ comfort level. Nutritional supplementation has also been incorporated into the postoperative care regimen. Total parenteral nutrition, which enhances survivability in calves,10 was ultimately initiated in the immediate postoperative period and was shown to nutritionally sustain the animals until resumption of spontaneous appetite.
In most cases, deterioration in the overall condition of the animal was observed 1–3 days before euthanasia. As a result, we have taken various steps to diminish or alleviate the underlying conditions that contribute to this type of decline. Blood-specimen cultures are now routinely performed to determine whether an infection is present and if so, to clarify the appropriate antibiotic therapy. In addition, we flush all indwelling pressure-monitoring catheters hourly with heparinized saline to prevent clot formation. Despite anticoagulation, right pump thrombus was a contributing factor in the deaths of 10 of our 29 calves. Because thrombus has not been identified in the left pump in any animal, we believe that the clots seen on the right originated from the periphery in some cases or from the jugular vein in others. To address this issue, we performed experiments to determine whether systemic and/or pulmonary pump speed modulation could reduce or eliminate the likelihood that small thromboemboli would interfere with right pump function by eliminating steady-state flow paths through the pump.11–14 We concluded that by modulating right pump speed, we could reduce thrombus formation while still maintaining pressures and flows in the physiologic range. Accordingly, induced modulation of the right pump speed was maintained in five of the 29 animals throughout their entire postoperative course. Although further experimentation is needed, the initial results of this technique were promising, as four of these five animals had no thromboembolic events or right pump thrombus.
Much remains to be learned regarding the long-term effects of steady-state, pulseless perfusion. However, the survival times of our CFTAH-implanted calves have increased through the above-described ongoing efforts. In turn, this increased survival has allowed us to perform additional evaluations, including treadmill studies, to show that completely continuous flow can support calves during routine activities such as exercise. In our two calves that survived for 90 days or more, normal end-organ histology and function provided further evidence that a large animal can be physiologically and hemodynamically maintained with a CFTAH. In the future, we plan to use the information gained in these cases to improve our design and development efforts. Presently, we have only the ability to set each pump at a fixed rotational speed and manually adjust the pump speed to achieve desired pressure and flow parameters. The use of a single integrated, closed-loop controller with sensor and flow algorithm capability to run both pumps instead of two separate controllers could facilitate RAP/LAP and flow balance. Although the intrinsic sensitivity of pump output to inflow pressure that characterizes all rotary blood pumps will decrease the demand for moment-to-moment adjustment of pump rotational speed to maintain a balance between the systemic and pulmonary flow, additional control algorithms will clearly be useful to accommodate a wider range of physiologic states.
Right pump vulnerability to thrombus or other particulate material from peripheral veins remains problematic. Modulation of the right pump speed seems to have a favorable impact by decreasing steady-state standing waves in the pump, but additional work is necessary. Dedicated pumps designed for the unfiltered blood from the venous system may be useful to optimize the trade-off between efficiency and a larger-area (less restrictive) blood path.
Although this work is still quite preliminary, it seems to clearly show that a large mammal can be sustained by continuous flow without adverse physiologic effects. Given the small size, mechanistic simplicity, and markedly improved durability of continuous-flow pumps, the prospects for a completely implantable CFTAH replacement device are encouraging.