Creation of a Controlled Venoarterial Shunt
A Surgical Intervention for Right-Side Circulatory Failure
Background Right-side circulatory failure (RSCF), a common complication of heart transplant and left ventricular assist device recipients, results in decreased cardiac output because of diminished flow across the pulmonary circuit. We hypothesized that creation of a controlled venoarterial shunt would result in decompression of the right ventricle and improved systemic cardiac output at tolerable oxygen saturations.
Methods and Results A venoarterial shunt was created in a large-animal model (calf, n=6). RSCF was induced by banding the pulmonary artery. Hemodynamic measures and blood gas determinations were obtained during nonshunted and shunted states. Pulmonary artery banding increased mean right ventricular systolic pressure from 44.9±2.1 mm Hg (mean±SEM) to 85.9±6.9 mm Hg (P<.05, paired t test) and decreased mean aortic flow from 7.8±1.0 to 4.2±1.1 L/min (P<.05). Flow through a venoarterial shunt at approximately 40% of cardiac output resulted in a decrease in right ventricular end-systolic pressure from 85.9±6.9 to 72.1±5.6 mm Hg (P<.01, ANOVA), a decrease in mean pulmonary artery pressure from 42.9±5.0 to 37.2±3.8 mm Hg (P<.01), and an increase in aortic flow from 4.2±.05 to 5.1 L/min (P<.01). Left ventricular stroke work decreased from 2.22±0.28 to 1.55±0.88 (P<.05). Carotid artery oxygen saturation did not change significantly (99.9±.02 to 97.6±1.7) during shunting.
Conclusions A controlled venoarterial shunt improved hemodynamics and cardiac output in a large animal model with RSCF. This strategy may be useful in the management of transplant and left ventricular assist device recipients with perioperative RSCF.
Right-side circulatory failure (RSCF) is a multifactorial entity in which the final common pathway is diminished blood flow across the pulmonary circuit, resulting in a lower systemic cardiac output. RCSF is the primary cause of death in 10% to 20% of early mortality (<30 days) after cardiac transplantation,1 2 usually owing to increased pulmonary vascular resistance in the recipient or donor organ dysfunction.3 Perioperative RSCF is also a complication in 20% to 40% of the increasing number of patients receiving left ventricular assist devices and is the leading cause of perioperative mortality.4 5 Causes in this population include altered ventricular interdependence and acute increased pulmonary vascular resistance in coagulopathic patients after prolonged cardiopulmonary bypass.5 6 7
Pharmocological therapy for RSCF includes pulmonary vasodialators and inotropic support for the right ventricle (RV). RV assist devices can provide mechanical assistance for RSCF refractory to drug therapy. RV assist devices are capable of maintaining adequate cardiac output in most patients but may fail when pulmonary vascular resistance is extremely elevated and increased flow to the pulmonary artery results in RV distension and a leftward shift of the ventricular septum, which limits left ventricular filling.8
In an effort to identify a new and complementary approach to treating RSCF, we studied venoarterial shunting in a large animal model with induced RSCF. We hypothesized that this shunt would decompress the RV and improve systemic hemodynamic stability, with tolerable levels of arterial desturation.
All animals received humane care in compliance with the Principles of Laboratory Animal Care formulated by the Institute of Laboratory Animal Resources and the Guide for the Care and Use of Laboratory Animals prepared by the National Institutes of Health (NIH Publication 86-23, revised 1985).
Six Holstein calves weighing between 80 and 100 kg were preanesthetized with midazolam (0.1 mg/kg IM) and ketamine (10 mg/kg IM). Anesthesia was induced with thiamylal (7 to 10 mg/kg IV), and after intubation, anesthesia was maintained with inhaled isoflurane (1.5% to 1.75% in 2 to 3 L O2).
In the left lateral decubitus position, a left thoracotomy was performed with excision of the fifth rib. The inferior pulmonary ligament was divided, and the lung was retracted dorsally. The pericardium was incised longitudinally. The aorta was dissected free to the ductus arteriosis, dividing the azygous vein. The ductus was clamped to eliminate the possibility of a naturally occurring left-to-right shunt. The pulmonary artery was isolated proximal to its bifurcation.
A shunt circuit was created from the right atrium (32-mm cannula, Bard Co) to the right femoral artery (18-mm Bard femoral cannula) (Fig 1⇓). The right atrial venous line was connected to a centrifugal pump (Biomedicus, Inc). Flow was returned to the right femoral artery. An in-line OxySAT 0200 (Baxter Health Care Corp) oxygen saturation meter was placed in the venous return line.
Ultrasonic flow probes (Transonic Systems Inc) were placed around the pulmonary artery, descending aorta, and left carotid and left femoral arteries. Micromanometer pressure transducers (Millar Instruments Inc) were placed in the RV, pulmonary artery, and left ventricle after calibration with a water column. A Swan-Ganz catheter in the pulmonary artery and catheters in the superior and inferior venae cavae and carotid and femoral arteries allowed measurement of blood pH and oxygen saturation.
An umbilical tape placed around the pulmonary artery was fastened to a ratcheted snare and used to band the pulmonary artery in an incremental fashion.
To establish an upper limit of tolerable shunt flow, mixed venous oxygen saturation was measured while increasing shunt flow from 0 to 5000 mL/min by increments of 500 mL/min. After a 4-minute stabilization period at each level, pulmonary artery samples were obtained. The procedure was repeated for each limb of the shunt circuit.
The optimal shunt flow rate was defined as the maximal flow with a venous oxygen saturation of >60%. This produced an arterial oxygen saturation of >90%. Once the optimal shunt flow rate was determined, it was used consistently for that animal throughout the remainder of the experiment.
The experimental protocol required right-side circulatory failure to be established, after which right-to-left shunting was initiated to improve hemodynamic stability. RSCF was defined by a >50% increase in RV and pulmonary artery pressures with a simultaneous >50% decrease in aortic flow (cardiac output). After baseline data were obtained, the pulmonary artery was banded to produce RSCF. The calf was allowed to stabilize for 4 minutes before all measures were repeated. Venoarterial shunt flow was then instituted at the optimal shunt flow rate through the shunt circuit. Four minutes later, measurements were repeated. Shunt flow was stopped, and the pulmonary artery band was removed. The animal was allowed to stabilize before the pulmonary artery band was reapplied and the above protocol was repeated.
Data were digitized in real-time at a speed of 200 Hz by an analog-to-digital converter (MacLab MkIII, MacLab/8, ADInstruments, Pty Ltd), filtered with a 50-Hz low-pass filter and recorded with a Macintosh computer (Macintosh IIcx, Apple Computers Inc). Blood gas determinations were performed with a Nova Stat 3 Profile (Nova Biomedical Inc).
Hemodynamic variables were measured for each experimental condition, including RV end-diastolic and peak systolic pressures, pulmonary artery pressure (PAP), left ventricular end-diastolic pressure (LVEDP), pulmonary artery flow (PAF), descending aortic flow (AoF), and carotid artery flow (CF). During left atrial–to–femoral artery shunting, aortic flow was calculated as measured flow plus shunted flow, as shunted blood enters the arterial circulation distal to the flow probe. Oxygen saturation was obtained simultaneously from the carotid, femoral, and pulmonary arteries and the inferior and superior venae cavae. Last, pH determinations were performed at the carotid artery and femoral vein.
Right and left ventricular stroke work (SW) calculations were performed for all experimental states (SW=stroke volume×mean ventricular pressure). RV stroke volume was calculated by dividing PAF by heart rate. Left ventricular stroke volume during shunt flow was calculated as stroke volume=(measured AoF+measured CF+shunt flow)/heart rate.
An unpaired Student’s t test was used to compare mean values during baseline and right circulatory failure conditions. A paired ANOVA with the Bonferroni multiple-comparisons test was used to compare the shunted state with the RSCF condition.
Calves tolerated shunt flows approaching 50% of cardiac output. Representative data from one animal are presented in Fig 2⇓. This animal had a baseline cardiac output of 8.0 L/min. Shunt flow is represented as a percentage of cardiac output versus mixed venous oxygen saturation.
A representative example of the effects of pulmonary artery banding and venoarterial shunting is presented in Fig 3⇓. Pulmonary artery banding caused an increase in right-side pressures and a decrease in left-side pressures. Flow through the pulmonary artery, aorta, and carotid artery were also decreased. Shunting resulted in decreased right-side pressures and improved flow through both pulmonary and systemic circulations.
Mean hemodynamic effects of banding the pulmonary artery to create RSCF are presented in the Table⇓. The RV end-diastolic and peak systolic pressures were increased markedly (P<.05), and LVEDP decreased (P<.05) after placement of the pulmonary artery band. PAF, AoF, and CF decreased (P<.05).
Arterial oxygen saturation was unchanged at all sites measured, but venous saturation decreased (P<.05). Neither arterial pH from the carotid artery nor venous pH from the femoral artery changed significantly.
Effects of Shunt
The mean effects of a venoarterial shunt on pulmonary and systemic hemodynamic parameters, oxygen saturation, and blood pH are presented in the Table⇑ and Fig 4 through 7⇓⇓⇓⇓. RV pressures and PAP decreased with shunting (P<.01) (Fig 4⇓). LVEDP also decreased during shunting (P<.05) (Fig 4⇓). PAP and AoF increased during shunt flow (P=.07 and P<.01, respectively) (Fig 5⇓). CF did not change significantly during shunt flow (Fig 5⇓). Left ventricular SW decreased during shunt flow (P<.05), whereas RV SW was unchanged (Fig 6⇓).
Effects of shunting on oxygen saturation are illustrated in Fig 7⇑. Carotid artery and superior vena cava oxygen saturation were not affected by shunting. Femoral artery and inferior vena cava oxygen saturation markedly decreased (P<.01). Mixed venous saturation trended lower during shunting; this was not statistically significant. Carotid artery pH decreased during shunting (P<.01), as did femoral vein pH (P<.01) (Table⇑).
In 1964, Austen et al9 described the use of a venoarterial shunt as treatment for pulmonary hypertension in an animal model.9 Experiments were based on the observation that patients with primary pulmonary hypertension often died suddenly from right-side heart failure during physical or emotional stress, but patients with Eisenmenger’s syndrome (pulmonary vascular disease and intercardiac shunt) rarely died from sudden death. In an acute canine preparation, a nonoxygenated right-to-left shunt was well tolerated in in the presence of increased pulmonary vascular resistance. In a long-term preparation, animals with pulmonary hypertension and an atrial septal defect were able to tolerate strenuous exercise, but animals with pulmonary hypertension and no atrial septal defect died suddenly after the same level of exercise. After the report of Austen et al, there were case reports and one small series in which right-to-left shunts were used clinically to treat patients with end-stage primary pulmonary hypertension.10 11 12 13 14 In the more recent reports, this strategy has been used as a temporizing measure before lung transplantation. Within this small group of patients, there have been some striking successes.
In the present study, in healthy animals, shunt flow rates approaching 50% of cardiac output resulted in a mixed venous oxygen saturation of 60%. This suggests that an animal without pulmonary disease is capable of maintaining adequate oxygenation when only half of the cardiac output is flowing through the pulmonary circulation.
We found that pulmonary artery banding in calves reproducibly increased RV and mean PAP, whereas LVEDP, PAF, AoF, and CF decreased. The effect of decreased systemic flow was detectable as a drop in the mixed venous oxygen saturation. This constellation of changes, consistent with RSCF, was useful for testing hemodynamic effects of a venoarterial shunt.
Instituting flow across a venoarterial shunt increased effective cardiac output. In addition, RV pressure and PAP were decreased as a result of diverting up to 50% of venous return away from the overloaded RV. Surprisingly, we observed a near–statistically significant (P=.07) increase in PAF despite the fixed obstruction and large shunt. This increase may result from increased RV perfusion pressure secondary to decreased RV pressures. A greater perfusion pressure will translate to improved RV performance.8 Similarly, we observed a decrease in LVEDP. Shunting blood directly into the distal aorta increases aortic root pressure, resulting in improved coronary perfusion pressure, and may improve left and right ventricular performance. The decrease in LVEDP may further decrease RV afterload, which may also contribute to increased PAF. A reduced LVEDP has the theoretical benefit of decreasing the work of the left ventricle by lowering filling pressures. This was observed as a decrease in left ventricular SW during shunted flow with a simultaneous increase in AoF. An increase in effective cardiac output while lowering LV SW represents a significant hemodynamic advantage to the left ventricle.
The physiological cost of a right-to-left shunt is decreased systemic blood oxygen saturation. Oxygen saturation was most affected at the level where the admixture of deoxygenated and oxygenated blood occurred. Venous desaturation paralleled arterial desaturation. Mixed venous oxygen saturation, an approximate index of mean tissue oxygen levels,15 was not significantly decreased during shunt flow. Femoral artery and inferior vena cava oxygen saturations were markedly decreased with shunting, but blood flowing to the head remained fully oxygenated. The risks of shunting blood peripherally, in the manner described, are ischemic damage to the lower extremities and local lactic acid production, resulting in a systemic acidosis. Arterial blood gas determinations at the carotid artery and femoral vein revealed statistically significant but not clinically significant changes in pH (>7.3 and >7.2, respectively). This suggests no profound systemic or local acidosis as a result of the venoarterial shunt.
These initial studies indicate that a venoarterial shunt may be useful in the treatment of perioperative RSCF. The major limitation of the present study is the relatively short duration of observed shunt flow. The effects of shunting on arterial oxygen saturation and blood pH may not fully manifest during the time course studied. This study demonstrates that a controlled venoarterial shunt has a beneficial effect on the hemodynamic profile of a large animal with induced RSCF. These effects can be obtained at flow rates that do not result in significant oxygen desaturation. These results may lead to an improved short-term treatment strategy for transplant and left ventricular assist device recipients who develop RSCF in the perioperative period.
This work was supported in part by Irving Assistant Professorship (Dr Oz) and a clinical research award (Dr Slater), Columbia University, New York, NY. Dr Oz is an Irving Research Scholar. We thank Alan Weinberg, PhD, for advice and statistical analysis. We also thank the members of the Institute of Comparative Medicine at Columbia University and the Perfusion Team at Columbia-Presbyterian Hospital, without whom completion of the study would not have been possible.
Presented in part at the 67th Scientific Sessions of the American Heart Association, Dallas, Tex, November 14-17, 1994, and published in abstract form (Circulation. 1994;90[pt 2]:I-161).
- Copyright © 1995 by American Heart Association
Terasaki PI. Clinical Transplants. Los Angeles, Calif: UCLA Tissue Typing Laboratory; 1990;21-27.
Farrar JD, Compton PG, Dajee H, Fonger JD, Hill JD. Right heart function during left heart assist and the effects of volume loading in a canine preparation. Circulation. 1984;70:708-716.
Kelly DT, Spotnitz HM, Beiser GD, Pierce JE, Epstein SE. Effects of chronic right ventricular volume and pressure loading on left ventricular performance. Circulation. 1971;44:403-412.
Austen WG, Morrow AG, and Berry WB. Experimental studies of the surgical treatment of primary pulmonary hypertension. J Thorac Cardiovasc Surg. 1964;48:448-455.
Nihill MR, O’Laughlin MP, Mullins CE. Effects of atrial septostomy in patients with terminal cor pulmonale die to pulmonary vascular disease. Cathet Cerebrovasc Diagn. 1991;24:166-172.
Kirklin JW, Barret-Boyes BG, eds. Cardiac Surgery, 2nd ed. New York, NY: Churchill-Livingston; 1990