Right Ventricle-Sparing Left Ventricular Resection and Replacement with a Continuous-Flow Rotary Blood Pump

Abstract

Despite recent advances in left ventricular assist device and total artificial heart technologies, these devices are still so large that they pose a significant problem in small patients with refractory heart failure. Excising the left ventricle while preserving the right ventricle—and then replacing the left ventricle with a mechanical pump—has been proposed as an alternative approach to this problem. We conducted a pilot study to evaluate possible surgical techniques and the hemodynamic effects of right ventricle-sparing left ventricular resection and replacement with a continuous-flow rotary blood pump in a healthy bovine model.

Left ventricular assist devices (LVADs) have proved useful both in supporting patients who have chronic refractory heart failure until an appropriate donor heart can be obtained and in enabling patients who manifest acute decompensation to become, over time, better candidates for cardiac transplantation.¹ Similarly, the pneumatically actuated total artificial heart has been shown to benefit these same patients.¹ More recently, LVADs have been implanted as permanent, or destination, therapy in patients who were deemed not to be transplantation candidates. Despite substantial advances in LVAD and total artificial heart technology over the past several years, these devices are still so large that they pose a significant problem in small patients with refractory heart failure.^1-4 Surgically excising the left ventricle (LV) and placing an LVAD-like pump in the left atrial-to-aortic position might mitigate some of the difficulties associated with size constraints by removing the markedly enlarged LV and, as an added benefit, might eliminate a potential source of emboli and right ventricular (RV) dysrhythmias.

In our previous studies,^2,3 we demonstrated in a bovine model the feasibility of long-term LV replacement with a pulsatile, pneumatically actuated volume-displacement pump. After LV excision and replacement, the calf maintained normal RV function and systemic hemodynamics for 4 days. In the current pilot study, we attempted to repeat this experiment with a continuous-flow rotary pump. By exploiting some of the advantages of the newer rotary pumps, namely smaller size and better wear resistance, we hoped to demonstrate the viability of a potentially beneficial treatment option for some patients with end-stage heart failure.

Materials and Methods

An 87-kg calf used in this study 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 in the present study.

The pump chosen for the experiment was the HeartWare HVAD™, which is described in detail elsewhere.⁵ Briefly, the HeartWare HVAD (HeartWare International, Inc.; Framingham, Mass) is a small, magnetically levitated, bearingless, centrifugal blood pump. It has a displaced volume of 45 cc, weighs 145 g, and can deliver flows of up to 10 L/min when driven at 4,800 revolutions per minute (rpm) against 100 mmHg of arterial pressure.

After inducing general endotracheal anesthesia by our standard technique,⁵ we placed the calf in the right lateral decubitus position. The left carotid artery and jugular vein were exposed for cannulation. We performed a left thoracotomy in the 5th intercostal space and exposed the heart. The left internal mammary artery was cannulated for arterial pressure monitoring. After heparinization, a 19F arterial cannula was placed in the right carotid artery, and wire-reinforced venous cannulae were placed in the right internal jugular vein and inferior vena cava. Cardiopulmonary bypass (CPB) was instituted, and normothermia was maintained.

The free wall of the LV was sharply excised, with care taken to preserve the left anterior descending and posterior descending coronary arteries and their septal branches. The vigorously bleeding transected diagonal and obtuse marginal coronary branches were suture-ligated, and the cut edge of the myocardium was oversewn to ensure hemostasis (Fig. 1A). The RV, the perfusion of which was not affected by LV resection, continued to beat vigorously throughout the procedure. The competent aortic valve remained closed while the calf was on CPB after the LV was excised.

Fig. 1 A) The left ventricular free wall is excised. B) The inflow sewing ring is attached to the mitral annulus. C) The HeartWare HVAD is inserted into the sewing ring, and the outflow graft is anastomosed to the descending aorta. Adapted by William E. Cohn, MD, from illustrations originally published in Frazier OH, Colón R, Taenaka Y. Surgical technique and hemodynamic characteristics of partial cardiac replacement with an artificial left ventricle. Tex Heart Inst J 1986;13(3):345–51. Illustrations by Bill Andrews. Permission to adapt granted by editors and artist.

Ao = aorta; L. = left; LAD = left anterior descending coronary artery; MV = mitral valve; PA = pulmonary artery; Vent. = ventricular

An adapter had been fabricated to enable the HeartWare HVAD to be implanted with the inlet in the left atrium. The adapter consisted of a shallow hollow cone, 6 cm in diameter at the base and 3 cm tall, made of fiberglass-reinforced silicone and lined with Dacron polyester. The apex of the cone had been transected to create a hole of a size appropriate to accommodate the inflow cannula of the HeartWare pump. The locking sewing ring that is integral to the HeartWare system had been incorporated into the apical end of the adapter so that the pump inlet could be locked in position relative to the adapter in a hemostatic fashion.

The base of the hollow cone was sutured circumferentially to the mitral annulus with running 2-0 polypropylene suture (Fig. 1B). We took care to ensure precise hemostasis between the conical adapter and the left atrium. The HeartWare pump inlet was positioned across the locking sewing ring, and the small nut was tightened to secure the pump in place. The 10-mm outlet graft was then measured, divided, and sutured to the descending thoracic aorta in an end-to-side fashion with a running 4-0 polypropylene suture and a partial occluding clamp (Fig. 1C). Residual air was vented, after which the calf was weaned from CPB while the rotational speed of the pump was gradually increased.

A 10-mm flow probe was placed on the outflow graft. During the 2-hour follow-up period, hemodynamic values (LVAD flow and mean aortic pressure) were recorded at varying pump speeds.

Results

The calf had hemodynamic stability and adequate perfusion throughout the 2-hour observation period, after which the calf was electively euthanized. During those 2 hours, pump rotation was set at a variety of speeds between 3,000 and 4,000 rpm, which resulted in aortic pressures between 68 and 165 mmHg and in flow rates from 4.8 to 9.0 L/min. Both arterial pressure and pump flow seemed to increase linearly with increased pump rpm. Right ventricular function remained vigorous throughout the 2-hour period, without inotropic support. Of interest was the observation that, absent the LV, the retained ventricular septum bowed leftward, giving the RV a more axisymmetric, “left ventricle-like” geometry. The aortic valve, relocated to the outside surface of the heart after LV resection, remained competent throughout the study and prevented hemorrhage into the operative field. Right ventricular contractions added a modest degree of pulsatility to the systemic pressure waveform, but this was believed to be due to vigorous rocking of the HeartWare pump by the ventricular septum, rather than to cyclic oscillations in left atrial pressure and instantaneous pump flow. Arterial oxygen saturation, carbon dioxide content, and pH were maintained within physiologic limits throughout the study.

Discussion

In this study in a calf, RV-sparing LV resection and replacement with a rotary pump was shown to provide physiologic systemic and pulmonary pressures and flows for 2 hours. The RV's function was unaffected by the absence of the LV. After LV resection, the ventricular septum bulged leftward, which resulted in a more conical axisymmetric RV geometry. However, this septal bowing was passive; during RV systole, the septum moved briskly to the right again, so that RV and tricuspid valve function remained unimpaired. (This differs from pathophysiologic states that cause LV suction during systole, thereby inhibiting rightward septal motion.)

Although the arrangement we have evaluated is perhaps somewhat similar physiologically to using a rotary blood pump as an LVAD in the presence of a severely hypokinetic LV, there are some significant differences that merit discussion. In both arrangements, flow maintains physiologic pulsatility in the pulmonary circulation and is pulseless (or essentially pulseless, depending on the LV reserve) in the systemic circulation. Similarly, in both arrangements, the aortic valve remains closed throughout the cardiac cycle. However, when the LV is preserved—as it always is when rotary pumps are implanted as LVADs—limited space in the chest can become problematic. The myopathic LV, even when decompressed, occupies a considerable amount of space. When the inlet cannula of the LVAD is inserted into the LV, the pump by necessity must sit below the diaphragm, in the left pleural space, or between the cardiac apex and the pericardium. Despite significant reduction in the size of rotary blood pumps during the last 10 years, anatomic geometry can still make pump placement challenging in many patients.^2,3 If, however, the dilated LV is resected, as we have described in this report, a significant amount of space becomes available for pump placement.

In addition, instances of LV suction, characterized by acute decreases in pump flow despite constant pump rotational speed, are well-recognized and frequent events caused by inadequate LV volume or excessive pump speed. These events occasionally result in mechanical stimulation of the ventricular septum by the inlet cannula and in biventricular tachycardia or fibrillation.⁶ Because patients require RV function to maintain LVAD filling and pump output, ventricular dysrhythmias frequently result in hemodynamic compromise. Should, however, the LV be resected and the pump inlet be positioned across the mitral annulus as we have described it here, this problem might be avoided. Although pump-suction events might still occur, they are not likely to trigger ventricular tachycardia or fibrillation.

Occasionally, ventricular muscle or thrombus around the inlet cannula can cause partial inlet obstruction, hemolysis, or embolus formation. Excising the LV and replacing it with a rotary pump, as we have described, might mitigate these complications.

Last, securing the LVAD inlet to necrotic myocardium in the presence of an acutely infarcted apex can be technically challenging. Acute hemostasis notwithstanding, there may be an associated increased risk of late LV rupture or pseudoaneurysm. Resection and mechanical replacement of the acutely infarcted LV could mitigate these difficulties.

Counterbalancing these challenges are the issues surrounding the absence of native LV function, should the pump malfunction. In patients with severe ventricular dysfunction in whom a rotary pump is implanted as an LVAD, some degree of forward systemic flow can be maintained by the weakened LV for a short period of time in the event of pump malfunction. In contrast, after LV resection, pump function alone impels the forward flow of blood.

In addition, it is essential that the aortic valve remain closed and hemostatic to prevent exsanguinating hemorrhage after weaning from CPB. Even a trivial amount of aortic valvular insufficiency will result in prohibitive bleeding. In this experiment, the aortic valve was perfectly competent. Whether or not this competence would be maintained over a longer period of continuous pressure-loading of the valve throughout the cardiac cycle is unknown. It might be necessary to sew a hemostatic patch to the aortic annulus to prevent late hemorrhage. Alternatively, the outflow graft of the pump could be sutured circumferentially to the aortic annulus from the LV outflow aspect. Whatever approach is used, perfusion of the left and right coronary arteries must be maintained to ensure continued RV function.

This experimental model demonstrates a new and potentially useful method of supporting patients with end-stage LV failure via a continuous-flow blood pump. By excising the enlarged LV, one can mitigate challenges associated with space limitations in smaller patients.

In addition, this approach may serve as a long-term total heart replacement. We have reported a univentricular total heart replacement in a “stone heart” patient who had undergone a Fontan procedure.⁷ It is the opinion of the author (OHF) that a continuous-flow pump that uses the left atrium as a reservoir (by constantly unloading the pulmonary circuit) can function effectively as a total heart replacement. This approach may be applicable even in patients without a functional right ventricle who require long-term cardiac replacement.

Acknowledgment

Stephen N. Palmer, PhD, ELS, contributed to the editing of this manuscript.