Inhaled Nitric Oxide Augments Left Ventricular Assist Device Capacity by Ameliorating Secondary Right Ventricular Failure

Abstract

Clinical right ventricular (RV) impairment can occur with left ventricular assist device (LVAD) use, thereby compromising the therapeutic effectiveness. The underlying mechanism of this RV failure may be related to induced abnormalities of septal wall motion, RV distension and ischemia, decreased LV filling, and aberrations of LVAD flow. Inhaled nitric oxide (NO), a potent pulmonary vasodilator, may reduce RV afterload, and thereby increase LV filling, LVAD flow, and cardiac output (CO). To investigate the mechanisms associated with LVAD-induced RV dysfunction and its treatment, we created a swine model of hypoxia-induced pulmonary hypertension and acute LVAD-induced RV failure and assessed the physiological effects of NO. Increased LVAD speed resulted in linear increases in LVAD flow until pulse pressure narrowed. Higher speeds induced flow instability, LV collapse, a precipitous fall of both LVAD flow and CO. Nitric oxide (20 ppm) treatment significantly increased the maximal achievable LVAD speed, LVAD flow, CO, and LV diameter. Nitric oxide resulted in decreased pulmonary vascular resistance and RV distension, increased RV ejection, promoted LV filling and improved LVAD performance. Inhaled NO may thus have broad utility for the management of biventricular disease managed by LVAD implantation through the effects of NO on LV and RV wall dynamics.

Mechanical support of severe heart failure with left ventricular assist devices (LVAD) may be associated with detrimental effects on right ventricular (RV) mechanics^1,2 that complicate the management of underlying biventricular impairment and impact patient morbidity and mortality.^2–6 In addition to the therapeutic benefits of LVAD deployment for promoting cardiac output, LVAD use may also cause: 1) abnormal trans-septal forces related to a leftward shift of the intraventricular septum and distorted RV geometry, leading to impaired contractility, 2) increased RV distension causing ischemia resulting from augmented venous return, and 3) increased pulmonary vascular resistance (PVR) because of pulmonary vasculopathy and increased RV afterload.^2,3,7–9 These mechanisms are more pronounced in the setting of pulsatile LVAD systems, but are observed as well in the setting of continuous flow technologies.^7,9 Accordingly, LVAD deployment in patients with the most severe cardiomyopathies, and therefore the patients with the most pronounced need for mechanical circulatory support, may paradoxically worsen heart failure with increased LVAD pump speed.

Inhaled nitric oxide (NO) may have utility in the treatment of such RV dysfunction,^10–13 presumably because of its ability to reduce pulmonary hypertension, RV distension, wall tension and oxygen consumption, and promote improved RV flow.^14–16 However, its clinical utility has not been consistently demonstrated. Moreover, in a recent randomized, double-blinded study in which patients undergoing LVAD placement received inhaled NO (40 ppm) or placebo until clinical resolution or RV failure, the impact of NO on RV function was not definitive.¹³ Given the paucity of data in large animal models supporting the mechanisms of RV impairment in the setting of LVAD deployment, we sought to test the effects of inhaled NO in an acute swine model with consistent PVR and LVAD deployment with resulting RV dysfunction. In this paradigm, we specifically examined the impact of NO on LVAD flow, cardiac output (CO), and ventricular dimensions in an acute swine model of hypoxemia-induced pulmonary hypertension. Our overall intent was to define conditions whereby the effects of RV dysfunction could be mitigated, and the therapeutic capacity of the LVAD device increased.

Materials and Methods

Nitric Oxide Generation and Delivery System

We previously developed methods to generate and monitor highly purified NO (with near zero concentration of contaminating NO₂), from both NO₂ gas and its liquid form, dinitrogen tetroxide (N₂O₄), based on the reduction of NO₂ by ascorbic acid.^17,18 The purity of the NO/O₂ gas generated was determined by mass spectroscopy, gas chromatography (GC), GC-flame ionization detection, and cavity attenuated phase shift spectroscopy.¹⁷ In these experiments, NO (800 ppm, Air Liquide) flowed through a mass flow controller and into a ventilator circuit (Biomed Crossvent 4+). An ascorbic acid cartridge (GeNO, LLC, Cocoa, FL) converted any residual or formed NO₂ into NO before inhalation (Figure 1A and B). The flow rate of NO was adjusted so that the inhaled concentration was 20 ppm as determined by in-line electrochemical detectors (GeNO, LLC; Figure 1B).

Figure 1: Demonstration of experimental methodology. A: Reduction of NO₂ by ascorbic acid into highly purified NO. B: Schematic diagram depicting the mechanism for delivering a controlled concentration of NO through a ventilator. The system features a blender for adjusting FiO₂ and ascorbic acid low-resistance cartridges to convert NO₂ back to NO. Photos of (C) cannula instrumentation into the ascending aorta and the LV apex of swine, and (D) the Biomedicus BPX80 centrifugal pump. FiO₂, fraction of oxygen in inspired gas; LV, left ventricular; NO, nitric oxide.

Animal Preparation

All procedures were approved by the Institutional Animal Care and Use Committee of Steward St. Elizabeth’s Medical Center. Five adolescent Yorkshire swine were anesthetized and catheterized as previously described.¹⁷ Swine (35–39 kg) were sedated with telazol 250 mg, ketamine 125 mg, and xylazine 125 mg (IM) and restrained supine on heating pads and covered except for the operative fields. The airway was secured with a 6.0 mm cuffed endotracheal tube and ventilated at 10 bpm with a tidal volume of 500 ml. The respiratory rate was adjusted so that expired end-tidal carbon dioxide was between 35 and 40 mm Hg (Criticare Systems, Issaquah, WA, INC PoetIQ gas monitor). Femoral artery and right internal jugular vein catheters (9 Fr) were placed as well as a pulmonary artery catheter (746HF8, Edwards Life Sciences) which continuously measured RV-CO (Vigilance II (vig 2), Edwards Life Sciences, Irving, CA). Anesthesia was maintained with midazolam 0.125 mg/kg/hr, ketamine 5 mg/kg/hr, propofol 25–50 mcg/kg/min, and fentanyl at 1 mcg/kg/hr. Lidocaine 2 mg/kg/hr and amiodarone 2 mg/kg/hr were given for arrhythmia prophylaxis. A bilateral thoracotomy with superior median sternotomy exposed the pericardium and aorta. Cannulas were placed in the ascending aortic (16 Fr; Terumo Sarns Softflow, Shibuya, Tokyo, Japan) and LV apex (15 Fr; Biomedicus) and (Figure 1C) connected to a Biomedicus BPX80 centrifugal pump (Figure 1D). The LV catheter was positioned below the mitral valve under epicardial ultrasound guidance. The LVAD circuit (approximately 400 ml) was primed with heparinized (2 U/ml) Ringer’s Lactate solution. The animal was stabilized (fraction of oxygen in inspired gas (FiO₂) 0.30) for 30 minutes before induction of hypoxemia (FiO₂ 0.15), which we have previously shown increases pulmonary artery pressures and vascular resistance in swine.^17,18

Characterization of Left Ventricular Assist Device Effects on Right Ventricular Mechanical Function

The speed of the LVAD was increased in 500 rpm increments until the pulse pressure decreased. Thereafter, the LVAD flow was increased in 250 rpm increments. At each flow the animal was allowed to stabilize for 5 minutes before incrementing the LVAD speed. Left ventricular assist device settings, flows, and RV-CO were recorded as well as heart rate (HR), femoral, and pulmonary arterial pressures. Ventricular dimensions were measured at peak systole and diastole on epicardial echocardiographic images taken from the RV. Pulmonary vascular resistance and systemic vascular resistance were calculated as: PVR = (mPAP - PAOP)/CO, where PAOP is the pulmonary artery occlusion pressure and mPAP is the mean pulmonary arterial pressure.

At high LVAD speeds the RV distends, the LVAD hoses chatter, and mean arterial pressure (MAP) oscillates. The RV eventually fails completely with MAP falling precipitously over seconds (Figure 2). The LVAD was turned off immediately and resuscitative measures given if needed, including increasing FiO₂ to one, defibrillation and open heart massage. Following each failure event and resuscitation, the animal was allowed to stabilize for 30 minutes (FiO₂ 0.3, with the LVAD turned off) before induction of hypoxemia, which was maintained for 10 minutes before starting another LVAD trial. A phenylephrine infusion was used to keep the MAP greater than 60, and Ringer’s lactate solution given to keep the CVP greater than 8 mm Hg before starting the LVAD flow in each trial.

Figure 2: Effect of increased LVAD speed (rpm) on arterial pressure and ventricular wall morphology.Increasing LVAD speed narrows arterial pulse pressure as aortic valve flow is taken over by LVAD flow (A) without and (B) with inhaled NO. At the highest speeds, a flow instability is reached and blood pressure falls precipitously. The administration of inhaled NO significantly extends the maximal achievable LVAD RPM. Epicardial echocardiograph in diastole at LVAD speeds of (C) 1,000 rpm showing normal morphology and (D) 3,250 rpm showing LV evacuation, RV distension, and septal bowing toward the left ventricle. LVAD, left ventricular assist devices; NO, nitric oxide; RPM, revolutions per minute; RV, right ventricular.

The three experimental conditions were A) FiO₂ 0.3, NO = 0; B) FiO₂ 0.15, NO = 0; and C) FiO₂ 0.15, NO = 20 ppm. In each experiment, LVAD flow was incremented until RV failure. In each animal, condition A was first, and then alternated between conditions B and C. In half the animals, condition B was the second experiment and in the other half condition C was the second experiment. Experimentation continued until the animal could not be successfully resuscitated after an episode of failure, which was uniformly because of persistent ventricular fibrillation (VF). A summary of the sequence of experimental conditions for each animal is shown in Table 1. Datasets were compared with one-tailed Mann-Whitney U tests using Prism 5.0 (GraphPad Software, Inc.) and were considered distinct when the p value was less than 0.05.

Table 1: Sequence of Experiments in Each Animal

Results

Increasing LVAD flow decreased pulse pressure and characteristically altered ventricular geometry (Figure 2). Altered LVAD flow simultaneously drew down LV volume whereas it increased venous return with RV distension, thereby pushing the interventricular septum toward the LV in diastole. The decreased filling of the LV led to reduced native stroke volume and loss of pulse pressure. As LVAD flow increased further, the right heart distended further and LV evacuation limited flow into the LVAD. Chattering of the LVAD hoses ensued. In most cases, this led to acute right heart failure with a precipitous fall in blood pressure. Without immediate cessation of LVAD flow, the heart rhythm typically converted to VF, requiring aggressive resuscitative measures.

The relation between LVAD speed and flow was linear up to 3,000 rpm, indicating that the vascular resistance was not impacted by the change in FiO₂ from 30% to 15% or by inhaled NO treatment (Figure 3A). At higher LVAD speeds, the heart started to fail and the data became more scattered; blood pressure dropped requiring immediate intervention and interruption of data collection. Above 3,000 rpm, the number of data points in the analysis declined (Figure 3B). Hypoxemia limited the maximal LVAD speed in each experiment and inhaled NO allowed higher LVAD speeds before failure.

Figure 3: Relationship of LVAD speed (rpm) to LVAD flow (L/min). A: LVAD flow–speed relation for normoxia (FiO₂ 0.3, solid line), hypoxemia (FiO₂ 0.15, course dashes), and hypoxemia with inhaled NO 20 ppm (fine dashes). Left ventricular assist device flow is proportional to LVAD speed for all conditions, indicating that systemic vascular resistance is not changed by hypoxemia, inhaled NO, or LVAD speed. The linearity breaks down above 3,000 rpm as critical speed is achieved and RV failure causes cardiovascular collapse. B: The number of viable data points remaining at each LVAD speed (N) in (A) indicating whether the RV had failed before data collection. LVAD, left ventricular assist device; NO, nitric oxide; RPM, revolutions per minut; RV, right ventricular.

Hypoxemia increased PVR by 124%. The addition of NO (20 ppm) decreased the PVR by 45% toward normal levels (Figure 4). Inhaled NO allowed increases of LVAD speed to higher levels, with resulting higher LVAD flows (Figure 5A). Although hypoxemia decreased the terminal LVAD speed, this change was not statistically significant (p = 0.08). The addition of inhaled NO significantly increased terminal LVAD speed by 21%. Hypoxemia decreased the maximum LVAD flow by 29% (Figure 5B), whereas NO increased maximum LVAD flow by 39%. The RV-CO before running the LVAD was statistically identical for all groups, although there was a trend to lower COs with hypoxemia-induced pulmonary hypertension (Figure 5C). The maximal RV-CO before failure did not significantly decrease when made hypoxemic from normoxemic conditions (Figure 5D). The addition of inhaled NO increased maximal RV-CO before failure by 23%.

Figure 4: Pulmonary vascular resistance as a function of oxygen and NO delivery. Pulmonary vascular resistance with normoxia (FiO₂ 0.3, black), hypoxemia (FiO₂ 0.15, gray), and hypoxemia with inhaled NO 20 ppm (white). AVG ± SD. *p < 0.05 as compared with normoxia, **p < 0.05 as compared with hypoxia with NO of 0 PPM. FiO₂, fraction of oxygen in inspired gas; NO, nitric oxide.

Figure 5: Variation of physiological outcome as a function of oxygen and NO delivery. All outcomes were assessed under normoxia (FiO₂ 0.3, black), hypoxemia (FiO₂ 0.15, gray), and hypoxemia with inhaled NO 20 ppm (white). A: LVAD speed (rpm) at the point of LVAD-induced cardiovascular failure, demonstrating that hypoxic animals treated with NO tolerated significantly higher speeds than untreated animals, (B) highest achievable LVAD flow rate, demonstrating that maximum LVAD flows in hypoxic animals (without NO) were 71% of those reached by normoxic animals, whereas inhaled NO restored the maximal LVAD flow to baseline. C: Baseline right ventricular cardiac output, demonstrating that in the absence of LVAD flow, RV-CO was not significantly impacted by hypoxia and inhaled NO treatment, (D) maximum right ventricular cardiac output, demonstrating that in the absence of LVAD flow, RV-CO was not significantly affected hypoxia and inhaled NO treatment, (E) left ventricular internal diameter in diastole (LVIDd), and (F) RV short axis internal dimension in diastole (RVIDd) at 1,500 rpm. It was demonstrated that at this modest LVAD speed, hypoxemia decreases LVIDd by 13% and increases RVIDd by 42% as compared with normoxic animals. Treatment with inhaled NO 20 ppm at this LVAD speed expands the LVID by 15%. The change in RVID with NO treatment was not statistically significant. *p < 0.05 as compared with normoxia, **p < 0.05 as compared with hypoxia with NO of 0 PPM. FiO₂, fraction of oxygen in inspired gas; LVAD, left ventricular assist device; NO, nitric oxide; RV-CO, right ventricular cardiac output.

Ventricular dimensions were impacted by the presence of pulmonary hypertension and treatment with inhaled NO. Left ventricular internal diameter at end diastole (LV_ID) at modest LVAD speed (1,500 rpm) was decreased by 13% by the addition of experimental pulmonary hypertension by hypoxemia (Figure 5E). Treatment with inhaled NO increased the LV_ID at this speed by 15%. Although hypoxemia significantly increased the RV short-axis dimension (Figure 5F), no statistical impact of inhaled NO on RV size could be appreciated.

Discussion

Left ventricular assist device–associated RV dysfunction is frequent (20–44%), negatively affects morbidity and mortality, and requires inotropic drug therapy or RVAD placement.^7–9 Right ventricular dysfunction after LVAD deployment may be related to a combination of underlying myocardial disease, leftward shift of the intraventricular septum, distorted RV geometry, and increased RV afterload from underlying pulmonary hypertension.^2,3,6 These mechanisms are more pronounced with pulsatile flow systems than in newer continuous flow devices^7,9; however, RV failure requiring either an RVAD placement or inotropic therapy is still significant even with continuous flow LVAD devices.^21–23 In the current report, we demonstrate in a swine model of hypoxia-induced pulmonary hypertension combined with acute LVAD-induced RV failure that inhaled NO significantly decreases PVR and RV distension, increases RV ejection, and promotes LV filling and improved LVAD performance. As a consequence, the effects of NO on ventricular wall dynamics should allow more robust treatment paradigms to be implemented in patients with the most serious clinical need.

Left ventricular assist devices characteristically take blood out of the LV and pump it into the ascending aorta, thus restoring circulation to a patient with left heart failure while resting the LV for recovery or providing a bridge-to-transplantation. At high LVAD speeds, negative pressure is applied to the LV and augmented venous return enlarges the RV so that the intraventricular septum shifts to the left, with several physiological consequences. First, the perfusion pressure of the RV myocardium is inversely dependent on the tissue pressure outside the myocardial capillaries. As wall tension in the RV increases coincident with distension, the downstream limiting pressure of myocardial perfusion is increased. At the same time, the myocardial work to mobilize this larger volume of blood in the RV increases, setting up an unfavorable oxygen balance and leading to RV ischemia. Second, the leftward shift of the intraventricular septum alters the geometry of both ventricular chambers, thereby reducing their overall contractility. A transient increase in flow at the LVAD inlet not only results in further RV distension with septal shift but also causes dynamic collapse of the myocardium around the inflow cannula from corresponding local loss of pressure. Such Venturi effects result in flow instabilities, where the inlet flow becomes limited or choked. Once this instability is established, MAP starts to oscillate with low frequency accompanied by chatter in LVAD hoses. If left untreated, MAP falls precipitously in an unstable spiral of events including loss of right coronary perfusion pressure and potential progression to VF. The only course of action that can interrupt this cascade is immediate cessation of LVAD flow. Indeed, many modern implantable LVAD systems have algorithms to detect abrupt negative changes in inflow and automatically turn down VAD speed.

Many patients in need of LVAD support have some element of pulmonary hypertension and RV dysfunction.⁸ In many clinical instances, the patient may require either inotropic support or a placement of an RVAD.^7–9,21 In our acute model of hypoxia-induced pulmonary hypertension, the RV distends from the VAD flow and the intraventricular septum is dysfunctional through altered geometric relationships. Oxygen delivery is limited from both hypoxemia and RV distension, limiting flow to its myocytes. The resulting RV ischemia coupled with both increased afterload and geometric loss of contractility pushes it toward failure. Thus, experimental pulmonary hypertension is a relevant acute model for studying the impact of LVAD-associated RV dysfunction and the potential use of NO treatment.

We and others have shown that inhaled NO decreases PVR in models of pulmonary hypertension.^14,17 Lower PVR decreases RV afterload and augments RV ejection. In this study, inhaled NO decreased PVR as expected (Figure 4). When applied to an acute hypoxemic model of pulmonary hypertension the maximum LVAD speed and flow, and the maximal RV-CO are significantly increased by the addition of inhaled NO in modest doses (Figure 5A, B, D). Inhaled NO also favorably altered the left ventricular dimension in diastole (Figure 5E). Therefore, adding a pulmonary vasodilator delays the point where the LV becomes flow limited or choked, and the LVAD therefore can be run to higher levels of ventricular support. In theory, inhaled NO should also cause a corresponding decrease in RV dimensions; however, our data were unable to show statistical significance (Figure 5F), possibly because of the complex shape of this chamber and artifact from placement of the epicardial ultrasound probe. The loss of PVR and increase in LV dimensions with inhaled NO suggests that RV distension was mitigated by the use of NO. Our data therefore suggest that LVAD-induced acute RV failure may be treated with inhaled NO without the use of inotropic agents or mechanical RV support. This acute model of pulmonary hypertension in LVAD use is highly reproducible, both in terms of altered PVR, altered interventricular geometry, and RV dysfunction and allows rigorous confirmation of mechanisms postulated from clinical experience. It is, however, an acute model with none of the microstructural or myoarchitectural alterations of patients with longstanding right heart dysfunction, either from intrinsic cardiomyopathy or long standing pulmonary hypertension. Nonetheless, this model does allow us to demonstrate the physiological mechanisms for the use of selective pulmonary vasodilators, such as inhaled NO, on cardiac geometric relations, and defines the basis by which LVAD function may be enhanced. In these experiments, the VAD flow was incremented until LV collapse and frank RV failure. Modern LVAD systems are equipped with pressure sensors and algorithms to trigger alarms and automatically turn down LVAD speed. Our data suggest that inhaled NO will increase the maximum VAD speed and flow in which these alarms would trigger in patients with pulmonary hypertension. Our results indicate that when pulmonary arterial hypertension (PAH) complicates LVAD function, inhaled NO, or other pulmonary vasodilating compounds, may be advantageous. Parental means of reducing pulmonary hypertension, such as prostacyclin analogs, could potentially benefit LVAD patients with PAH. They are, however, costly, complex to administer, and are associated with frequent side effects such as hemodynamic compromise with abrupt cessation and rapid tolerance.^24–26 Patient acceptance of subcutaneous infusions is poor due to pain at the injection site.²⁴ Other parental therapies include phosphodiesterase (PDE5) inhibitors and endothelin receptor blocking agents, but these therapies are associated with systemic hypotension, visual disturbances, headache, dyspepsia, facial and leg edema, epistaxis, increasing liver enzymes, nausea, and vomiting.^27–30 By contrast, inhaled NO consistently elicits a reduction of mean pulmonary artery pressure and PVR, as well as an improvement in oxygenation in a selective fashion with few side effects, suggesting that chronic therapy may be useful when pulmonary vasoconstriction plays a pathogenic role. The use of such a system in the ambulatory setting may have great utility. We have shown in a prior study that liquid N₂O₄ substrate effectively generates gas phase NO₂ for ultimate reduction by ascorbic acid into inhaled NO and that this product effectively modulates experimental pulmonary hypertension.¹⁸ Such a system is miniaturizable because of the highly concentrated nature of liquid N₂O₄ and therefore, when reduced to practice, may allow LVAD recipients with pulmonary hypertension to be effectively managed as an outpatient. By expanding the functional capacity of LVADs in patients with severe left ventricular heart failure and by extending utility into the outpatient settings, a significant advance in cardiac therapeutics may be achieved.

Conclusions

Mechanical support of ventricular failure associated with LVAD deployment may paradoxically worsen heart failure through its promotion of RV dysfunction. We present an acute model of LVAD-induced right heart failure that is characterized by chamber deformation along with increased RV afterload. Our data demonstrates that inhaled NO, a selective pulmonary vasodilator, allows unloading of the right heart while promoting the filling of the left ventricle, thus promoting a right to left shift of blood volume and allowing higher levels of circulatory support to be achieved in this acute model.