Right ventricular dynamics during left ventricular assistance in closed-chest dogs

Right Ventricular Dynamics During Left Ventricular Assistance in Closed-Chest Dogs Marc R. Moon, MD, Luis J. Castro, MD, Abe DeAnda, MD, Yasuko Tomizawa, MD, PhD, George T. Daughters 11, MS, Neil B. Ingels Jr, PhD, and D. Craig Miller, MD Department of Cardiovascular and Thoracic Surgery, Stanford University School of Medicine, Stanford, California; Cardiac Surgery Section, Department of Veterans Affairs Medical Center, Palo Alto, California; and Research Institute of the Palo Alto Medical Foundation, Palo Alto, California

To determine the effects of left ventricular assist device (LVAD) support on global right ventricular (RV) systolic mechanics, 8 closed-chest, conscious, sedated dogs were studied after placement of an LVAD (left ventricle to femoral artery bypass) and implantation of 27 tantalum markers into the left ventricular and RV walls for computation of biventricular volumes and geometry. Biplane cinefluoroscopic marker images and hemodynamic parameters were recorded during transient vena caval occlusion at various levels of LVAD support. Right ventricular contractility was assessed using end-systolic elastance and preload recruitable stroke work, and the myocardial (pump) efficiency of converting mechanical energy to external work (stroke workhotal pressure volume area) was calculated. With full LVAD support, RV end-diastolic volume increased from 60 f 15 to 62 f

17 mL ( p < 0.002), pulmonary artery input impedance decreased from 940 2 636 to 587 f 347 dyne s/cm5 ( p < 0.007), and measurement of RV and left ventricular septal-free wall dimensions demonstrated a significant leftward septal shift (p < 0.0005). Global RV end-systolic elastance and preload recruitable stroke work decreased from 2.4 -C 1.0 to 1.7 f 0.7 mm Hg/mL @ < 0.004) and 14.1 2 3.3 to 12.1 f 3.9 mm Hg ( p < 0.021, respectively; however, RV power output and myocardial efficiency did not change significantly ( p > 0.74 and p > 0.33, respectively). Therefore, during LVAD support, global RV contractility is impaired with leftward septal shifting, but RV myocardial efficiency and power output are maintained through a decrease in RV afterload and an increase in RV preload. (Ann Thorac Surg 1993;56:54-67)

T

to investigate this phenomenon and the major difficulties inherent in estimating true three-dimensional RV geometry. Miyamoto, Farrar, and their associates [9, 101 initially evaluated RV "contractility" in open-chest animals (as assessed by RV rate of change of pressure [dP/dt] and ejection fraction), but did not directly measure right or left ventricular volume. These load-dependent measures do not accurately quantify contractility when significant changes in ventricular loading conditions occur; because both RV afterload and preload may change with LVAD support, changes in RV ejection fraction and maximum dP/dt are difficult to interpret. On the other hand, Elbeery and associates [13] applied the shell subtraction method (measuring total cardiac volume with ultrasonic crystals) to estimate relative changes in RV volume and evaluate load-independent measures of systolic function. Although no change in RV contractility with partial LV assist was observed, they were able to achieve only 50% LVAD support with a single left atrial cannula, and, therefore, did not fully unload the left ventricle. We performed this study to determine the effects of both partial and full LVAD support on global RV systolic mechanics in closedchest, conscious dogs. Using myocardial markers to measure accurately both right and left ventricular volumes and geometry, we found that LVAD support decreased RV afterload, increased RV preload, and reduced RV

he left ventricular assist device (LVAD) is commonly used for temporary support of patients with postoperative cardiogenic shock and as a bridge to transplantation [ 1 4 ] , but its use has been associated with right ventricular (RV) failure in humans [5-81. Although the etiology of right-sided failure occurring during LVAD support remains unclear, the magnitude of this problem will most likely increase when the implantation of permanent, "wearable" LVADs for chronic congestive heart failure becomes a reality. Previous studies using openchest, anesthetized animals have shown that left ventricular (LV) unloading with an LVAD decreases RV afterload but causes a progressive decline in RV contractility as the level of support is increased [9-121; however, other investigators have found no change in RV afterload or contractility with partial LVAD support, implying that preexistent RV dysfunction is simply unmasked by increased venous return to the right heart [13]. It has also been suggested that the LVAD can actually improve RV function and output in some cases [12, 14-17]. These discordant results might be due to the variety of methods used Presented at the Twenty-ninth Annual Meeting of The Society of Thoracic Surgeons, San Antonio, TX, Jan 25-27, 1993. Address reprint requests to Dr Miller, Department of Cardiovascular and Thoracic Surgery, Falk Cardiovascular Research Center, Stanford University School of Medicine, Stanford, CA 94305.

0 1993 by The Society of Thoracic Surgeons

-

0003-4975/93/$6.00

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

Ann Thorac Surg 1993;565467

Ao Valve

Pulmonary

contractility (associated with a leftward septal shift), but did not decrease RV pump efficiency or power output.

I

\

Valve

55

Mitral

/ Valve

Material and Methods

Surgical Preparation Under anesthesia induced with pentobarbital (25 mg/kg), 8 healthy dogs (30.1 ? 7.2 kg) were intubated and ventilated (Ohio Anesthesia Ventilator, Madison, WI) with supplemental inhalational isoflurane (1%to 2%). A median stemotomy was performed, and the heart was suspended in a pericardial cradle. Twenty-seven miniature tantalum radiopaque helices (inner diameter = 0.8 mm; outer diameter = 1.3 mm; length = 1.5 to 3.0 mm [with variable small extensions or "tails" to facilitate subsequent radiographic identification]) were then inserted into the right and left ventricular walls. Markers were placed on the obturator of a modified spinal needle (20 gauge), inserted through the epicardial surface, and deposited in the myocardium by withdrawal of the obturator from the sheath. Seventeen markers were placed into the right ventricle and septum (Fig 1) [18]. Four were inserted along the anterior interventricular sulcus and three along the right atrioventricular groove evenly spaced from the pulmonary valve to the posterior interventricular sulcus. Four markers were implanted in the RV free wall midway between the atrioventricular and anterior interventricular landmarks, and four were placed in the interventricular septum directly opposite these corresponding RV free wall markers. Finally, two RV markers were placed in the septum directly below the right atrioventricular groove, superior and inferior to the tricuspid valve. Ten markers were placed in the LV subepicardial layer: three on the anterior wall, three on the inferior wall along the posterior descending artery, three along the obtuse margin midway between the anterior and posterior markers, and one at the apex. This arrangement allowed a three-dimensional representation of the right and left ventricles with combined 45-degree right anterior oblique and 45-degree left anterior oblique videofluoroscopicprojections (Fig 2). Marker position was verified with intraoperative fluoroscopy and confirmed subsequently at postmortem examination. Superior and inferior vena caval pneumatic occluders (In Vivo Metric Systems, Healdsburg, CA), a pulmonary artery (PA) ultrasonic flow probe (12- to 16-mm perivascular probes with a T201D Flowmeter; Transonic Systems, Inc, Ithaca, NY), and biatrial pacing electrodes were placed. Micromanometer-tipped pressure catheters (MPCSOO; Millar Instruments Inc, Houston, TX) were zeroed in a 37°C water bath and placed directly into the PA to measure PA pressure and through a left pulmonary vein to measure left atrial pressure. The tip of the PA catheter was positioned in the main PA distal to the PA flow probe. Long micromanometer-tipped catheters (Millar model SPC-350) were also zeroed and placed into the left and right ventricles through introducers in the left femoral artery and vein, respectively. Systemic arterial pressure was also measured through the side port of the left femoral introducer using a micromanometer-tipped

----

A

Lateral

( 0 ) /

Posterior

n

4

Anterior

Tricuspid / Valve . -..-

RV Free Wall

B Fig I . Marker array inserted to measure right ventricular (RV) and left ventricular (LV)volume. (A) Anterior view with the septum split. Ten septal markers were placed: four anteriorly along the interventricular groove (O),four in the mid-septum (A),and two in the posterior septum directly below the atrioventricular groove (0).Four markers were also placed in the RV free wall (W, and three were inserted along the anterior atrioventricular groove (A [not seen in this projection]). In the LV, three markers were placed in each of the anterior, lateral, and posterior walls with one in the apex. ( B ) Cross-sectional biventricular view taken from the LV equatorial plane (dotted line in A). (Ao = aortic.)

catheter. A 32F to 36F angled LVAD inflow cannula (Research Medical, Inc, Midvale, UT) was then introduced through the left atrial appendage, across the mitral valve, and into the left ventricle. This was connected to a centrifugal pump primed with crystalloid (model 540 Bio-Console with a BP-80 Bio-Pump; Medtronic BioMedicus, Inc, Eden Prairie, MN), and blood was returned through a 15F to 19F Bio-Medicus arterial cannula inserted in the right femoral artery. All tubing and cannulas were coated with the Medtronic Carmeda Bioactive Surface (Medtronic Cardiopulmonary, Anaheim, CA) to prevent clot formation during normal pump function [19], and low-dose heparin (1,000 IU intravenously) was given to

56

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

Ann Thorac Surg 1993;56:5467

stasis, and meperidine (1.0 to 2.0 mg/kg intravenously) and diazepam (5 to 10 mg intravenously) were given as needed for analgesia and sedation.

Experimental Protocol

B Fig 2. Representative simultaneous biplane cinefluoroscopic marker images in the right anterior oblique (A) and left anterior oblique (B) projections of one heart. Note how the tails on specific myocardial markers facilitate their radiographic identification. The left ventricular assist device cannula and multiple micromanometers are also seen.

prevent clotting during periods when pump flow was negligible. Left ventricular assist device flow was determined on-line with a digital flowmeter (TX-40 Bio-Probe; Bio-Medicus). After the LVAD, caval occluders, and pacing wires were tested, all cannulas and monitoring devices were exteriorized, chest tubes were placed, and the chest was closed. The animal was then allowed to recover from anesthesia for 1 to 2 hours and transferred to the cardiac catheterization laboratory for biplane videofluoroscopic studies. During this time, LVAD flow was maintained at approximately 300 to 500 mL/min to prevent

In the cardiac catheterization laboratory, mild sedation with diazepam and supplemental ketamine (5 mg/kg intravenously) was continued as necessary. To minimize reflex sympathetic and parasympathetic responses that may have occurred in these conscious animals, complete autonomic blockade was achieved with intravenous esmolol (0.3 mg * kg-’ * min-’ infusion) and atropine (0.1 mgkg). The dogs were atrially paced at 110 to 120 min-’. With the LVAD off (0% flow), baseline hemodynamic and biplane cinefluoroscopic recordings were obtained during steady-state conditions and over a physiologic range of peak LV and RV systolic pressures during caval occlusion (to reduce arterial systolic pressure by at least 40 mm Hg). After release of the occluders, the dog was allowed to recover for 3 to 5 minutes before continuing, and all data acquisition runs containing premature ventricular contractions were repeated. Recordings were then performed during maximum LVAD support (100% flow) with vena caval occlusion. The maximum LVAD flow rate was determined by flattening of the systemic arterial pressure trace and minimization of LV systolic pressure. Based on this maximum flow rate, recordings were repeated after LVAD flow was decreased to 50%, ie, 50% (partial) unloading. At the conclusion of the study, the dogs were sacrificed using B-Euthanasia (0.2 mgikg intravenously; Schering-Plough Animal Health Corp, Kenilworth, NJ), and proper position of the cannulas and catheters was confirmed. All animals received humane care in compliance with the ”Principles of Laboratory Animal Care” formulated by the National Society for Medical Research and the ”Guide for the Care and Use of Laboratory Animals” prepared by the National Academy of Sciences and published by the National Institutes of Health (DHEW [NIH] publication 85-23, revised 1985). The study was approved by the Stanford Medical Center Laboratory Research Animal Review Committee and conducted according to Stanford University policy.

Data Acquisition All cinefluoroscopic studies were conducted with the animal in the supine position using a General Electric MLX biplane L-U arm system (General Electric Medical Division, Milwaukee, WI) with the image intensifiers in the 6-inch boost fluoroscopic mode. The 45-degree right anterior oblique and 45-degree left anterior oblique biplane images were recorded on Sony U-Matic 5800 %-inch video cassette recorders and synchronized with the x-ray pulses by a master synchronization oscillator at 60 Hz. The analog electrocardiographic (ECG) signal and LV pressure signal were digitized and recorded in digital format on each individual video image using a custom intelligent video controller (Control Video Corp, Campbell, CA); the peak ECG R-wave was detected electronically and digitally encoded as an end-diastolic timing

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

Ann Thorac Surg 1993;5654-67

marker. At the completion of the study, images of grids containing 1-cm squares and biplane images of a threedimensional spiral phantom of known dimensions were then obtained to determine radiographic distortion and magnification factors. The two-dimensional coordinates of each marker in each projection were digitized frame by frame using a semiautomated, computerized myocardial marker detection system (Hewlett-Packard RS/20 [Palo Alto, CAI equipped with Matrox MVP-/AT/NP image processing boards [Dorval, Quebec, Canada]) using custom-designed image processing and digitization software developed in our laboratory [20]. The data from the two views were then corrected for magnification and distortion; the marker coordinates were then merged using custom software to yield the three-dimensional (x,y,z) coordinates of each marker every 16.7 milliseconds as previously described [21]. Using this system, marker position determinations are accurate and reproducible with a mean overall error of 0.3 2 1.3 mm [21]. During each data acquisition run, seven channels of analog data were simultaneously recorded on a HewlettPackard multichannel recorder (model 7758B; Hewlett Packard Corp, Andover, MA) at a paper speed of 25 m d s , including systemic arterial (distal aortic) pressure, LV pressure (LVP), RV pressure (RVP), PA pressure (PAP), left atrial pressure, PA flow, and surface lead ECG. All analog data were also acquired and digitized simultaneously at 240 Hz using a 486-based microcomputer (486-33MHz; JDR Microdevices Inc, San Jose, CA) with a high-speed data acquisition card (DT 3831-G; Data Translation Inc, Marlboro, MA) controlled by commercially available software (Labtech Control 3.2.0; Laboratory Technology Corp, Wilmington, MA). The 240 Hz pressure and flow data were then merged with the 60 Hz markerderived geometry data by aligning the LVP waves (or ECG R-waves) from the two data sources and calculating a convolution of the two signals to be matched, and using straight-line interpolation between the 60 Hz samples.

57

POSTERIOR BASAL o.ya1)

I

ANTERIOR

:

T Hl2

APEX Fig 3. Multiple cylindrical model for calculating left ventricular volume. (AP = anterior-posterior dimension; H = height between layers; SL = septal-lateral dimension.)

1 Y z 22 1

XI Y 1 2 1 x2

x3 Y 3 23 1 x4 Y4 24 1

I

24

RVVOL=

2 VOLi. i = l

Data Analysis HEMODYNAMICS. To minimize the effects of intrathoracic pressure variation due to respiration, the respirator was temporarily discontinued during vena caval occlusion (approximately 10 to 15 seconds), and only end-expiratory beats were selected for analysis. For each cardiac cycle, end-diastole was defined as the time of the ECG R-wave and end-systole was defined according to Suga and associates [22] as the time of the maximum pressure/volume ratio (for the RV, end-ejection follows end-systole by a considerable period of time because RVP decreases - Vmin) throughout ejection). Stroke volume (SV = V,, and ejection fraction (SV/V,,,) were calculated for both ventricles, where V, and Vminare the maximum and minimum ventricular volumes (see below). For each cardiac cycle, the LV and RV pressure signals were differentiated with respect to time, and LV and RV dP/dt,,, were determined. RV AND LV VOLUME. Right ventricular volume was calculated using a multiple tetrahedral model [18]. The seven-

Left ventricular volume was calculated using a multiple cylindrical model [23]. As there were 13 LV wall markers, three on each of the four walls and one apical marker, three cross-sectional marker levels were identified from base to apex (Fig 3). Three epicardial anterior-posterior (AP,, AP,, AP,) and septal-lateral (SL,, SL,, SL,) diameters were identified with three longitudinal heights (H,,, H,,, H,J connecting the midpoints of adjacent crosssectional levels. Two segmental volumes were computed using the formula for the frustum of a cone: V12

= H12 X

Vz3 = H a

X

(A1 +

fi& + Ad/3,

(A2 +

+ A3)/3,

where A, = T x (AP,/2) x (SLi/2).The volume of the apical segment was calculated using the formula for a cone: V,, = ~ / x3 A, x H34. Total LV volume (including some LV myocardial volume) was taken as the sum of all three segments. Right ventricular and LV volume calculations using these algorithms when compared with known

58

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

biventricular intracavitary balloon volumes have been shown to correlate in a highly linear fashion in excised canine hearts (for the RV: Y = 0.93 0.05, p < 0.0001, standard error of the estimate = 1.0 0.5 mL; for the LV: r = 0.96 k 0.07, p < 0.0001, standard error of the estimate = 1.4 1.2 mL) [24].

*

*

RV AFTERLOAD AND POWER OUTPUT. To quantify RV afterload and pump performance, pulmonary vascular impedance and RV power output were calculated using Fourier analysis (see Appendix 1 for mathematical methods) [25, 261. Unlike pulmonary vascular resistance, which describes only the nonpulsatile arterial load, pulmonary vascular impedance takes into account the pulsatile circulation in the pulmonary vascular bed. Pulmonary vascular impedance, therefore, represents the total opposition to flow seen by the right ventricle and is quantified by input resistance and characteristic impedance [25-271. Input resistance describes the ratio of oscillatory pressure to oscillatory flow at the origin of the system (ie, PA), and depends on the cross-sectional area of the pulmonary bed and left atrial pressure. Characteristic impedance represents the pulsatile component of pulmonary flow and is influenced by vascular wall compliance and blood inertia. To calculate the total external energy generated by the right ventricle, one needs to determine both steady flow power (the energy required to move blood in a nonpulsatile stream with a steady pressure and flow) and pulsatile (oscillatory) power, the energy cost of the pulsations per se [26, 271. Total RV power output is thus the sum of steady flow power and pulsatile power. BIVENTRICULAR GEOMETRY. To evaluate septal shifting during LV unloading, RV and LV septal-free wall dimensions were measured in the low-septum (RV inflow and LV apical regions [low-SFW,,, low-SFW,,]), mid-septum (RV sinus and LV equatorial regions [mid-SFW,,, midSFW,,]), and high-septum (RV infundibulum and LV basal regions [high-SFW,,, high-SFW,,]). This allowed us to estimate the relative regional changes in septal position that occurred during LVAD support.

RV CONTRACTILITY. Global RV contractility was assessed using three load-insensitive indexes: end-systolic pressure-volume relationship [28, 291, preload recruitable stroke work [30, 311, and dP/dt,,,end-diastolic volume (EDV) relationship [32]. To define the end-systolic pressure-volume relationship, end-systolic pressure (P,J and volume (V,J points were determined for each cardiac cycle analyzed during abrupt preload reduction using an iterative technique [33, 341. By least-squares linear regression, a straight line was fitted to these points, yielding the equation:

where E,, and Vo are the slope and volume axis intercept, respectively. In addition, external LV and RV stroke work (SW) were computed as the integral of LV or RV pressure and volume for each cardiac cycle:

Ann Thorac Surg 1993;56:5447

E3

to to

E

n

Volume Fig 4. The total energy generated by the ventricle during systole (total pressure-volume area) includes the area below the end-systolic pressure-volume line and above the end-diastolic pressure-volume curve. It, therefore, includes both external stroke work and the potential energy stored in the myocardium at end-systole.

SW=

I

P

*

dV.

The RV preload recruitable SW and RV dP/dt,,,EDV relationships were obtained by linear regression: SW = M, (EDV - V,),

where M, and V,, and Mdpldtand V,,, are the slope and volume axis intercepts for each relationship, respectively. Although the isovolumic parameter dP/dt,,, is preload-dependent, the dP/dt,,,-EDV relationship normalizes for changes in preload and can be used to quantify ventricular contractility in intact animals [32]. DERIVED RV ENERGETICS. The total energy generated by the right ventricle was approximated as the total pressurevolume area (RV PVA), bounded by the end-systolic pressure-volume relationship (the systolic segment of the pressure-volume trajectory) and the end-diastolic pressure-volume curve, including the triangular area to the left of the pressure-volume loop (Fig 4). The PVA is directly related to myocardial oxygen consumption in the left ventricle, and reflects combined potential energy and external mechanical work [35]. The PVA was calculated for each beat analyzed during caval occlusion, and the efficiency of energy transfer from PVA to mechanical energy (external pressure - volume work = SW) was mathematically derived: pump (work) efficiency = SW/ PVA [36]. This represents the ability of the ventricle to convert total energy generated during systole to mechanical energy or external work. Comparisons of SW/PVA between different levels of LVAD support were performed at matched RV end-diastolic volumes. In addition, effectivearterial elastance (RV E,) of the pulmonary vas-

59

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

Ann Thorac Surg 1993;56.5447

cular bed, a function of end-systolic RVP (RVP,,) and SV, was calculated for the steady-state beats using a modification of the definition of Sunagawa and associates [37], ie, E, = RVP,,/SV. Effective arterial elastance estimates RV afterload, as it represents a steady-state approximation of the maximum pulmonary vascular impedance as described above. The right ventricular-arterial coupling ratio, E,/E,,, was then computed, which reflects the (elastance) matching status between the right ventricle and the pulmonary arterial load in the pressure-volume plane framework. Previous investigators have demonstrated that for the LV, stroke work is maximal when arterial elastance equals ventricular elastance (E,/E,, = 1); however, pump efficiency is maximized when arterial elastance is one-half the ventricular elastance (E,/E,, = 0.5) [37, 381. All data are reported as mean ? one standard deviation. Data obtained at all three levels of LVAD support were compared using repeated-measures analysis of variance, and when indicated by a significant F statistic, significant differences were isolated using Fisher's protected least significant difference test [39]. Significance was inferred with a p value less than 0.05. STATISTICAL ANALYSIS.

Results Hemodynamics Hemodynamic data are shown in Table 1. Figure 5 demonstrates the typical changes in hemodynamics seen with increasing LVAD support in 1animal. With maximum LV unloading, mean distal aortic pressure and PA flow increased 17% ( p < 0.002) and 16% ( p < 0.02), respectively, LVP,,, and LV dP/dt,,, both declined 80%, and the aortic pressure trace flattened (progressive decrease in the systemic pulse pressure). As LVAD support was increased from 0% to 50% to loo%, LV EDV fell by 7% ( p < 0.02) and 23% ( p < O.OOOl), whereas LV SV dropped by 27% ( p < 0.02) and 65% ( p < 0.0001) and LV SW declined by 38% ( p < 0.007) and 90% ( p < 0.0001). Although at 50% LVAD flow all pressures proximal to the left ventricle were unchanged, mean left atrial pressure, mean PAP, and RVP,,, fell significantly by 47% ( p < O.OOOZ), 21% ( p < 0.0003), and 9% ( p < 0.008), respectively, at 100% LV unloading. These changes were associated with a 7% decrease in RV dP/dt,,, at 50% LVAD flow ( p < 0.04) and a 20% decline at 100% flow ( p < 0.0001 versus 0% flow). Although RV EDV did not significantly change with partial LVAD support, there was a small, but significant increase in RV EDV with full (100%) support when compared with both 0% ( p < 0.007) and 50% flow ( p < 0.02). Right ventricular afterload fell significantlywith full LVAD support. Although there was no change in characteristic impedance (0%:222 ? 93 dyne s/cm5,50%:231 h 59 dyne * s/cm5; 100%: 203 2 58 dyne s/cm5; p > 0.50), pulmonary artery input impedance declined significantly from 940 k 636 and 904 ? 643 dyne s/cm5 with LVAD flows of 0% and 50%, respectively, to 587 h 347 dyne s/ cm5 with full LVAD support ( p < 0.007 versus 0%; p < 0.02 versus 50%).Right ventricular E,, which also reflects

-

-

-

Table 1. Hemodynamics" Variable Heart rate (beats/ min) Aortic pulse pressure (mm Hg) Mean AoP (mm Hg) Mean LAP (mm Hg) Mean PAP (mm Hg) Mean PA flow (L/min) Left Ventricle LV EDP (mm Hg) LVPmax (mm Hg) LV dPldt,,, (mm Hg/s) LV EDV (mL) LV SV (mL) LV sw (mm Hg . mL) LV E, (mm Hg/mL) Right Ventricle RV EDP (mm Hg) RVP,,, (mm Hg) RV dMdt,,, (mm Hg/s) RV EDV (mL) RV SV (mL) RV SW (mm Hg * mL) RV E, (mm Hg/mL)

0% (off)

107.7 f 10.6 107.4 f 10.6 35.2

f

3.3

100%

50%

19.0 2 5.6b

107.6 f 10.8 4.4

f

2.6b!d

101.6 f 25.4

97.8

rtr

31.3 119.2 f 32.9brd

12.1 t 4.7

11.4

f

3.7

17.5 f 6.6

16.9 2 6.2

1.4 t 0.7

1.4 2 0.7

8.4

7.7

f

6.1

-1.5

86.6

f

24.6

1,305 f 287

879 2 323b

94.3 t 25.4 13.9 f 5.3 1,158 f 440

88.0 f 23.7' 10.2 f 3.2' 717 f 357b

f 4.5

106.5 f 25.2

7.0

f

3.4b,d

13.9 f 4.7b*d 1.7 f 0.7','

rtr

4.2b,d

21.4

f

31.9bzd

288

f

260b,d

72.6 f 19.2b,d 4.9 f 2.9b,d 115 f 200b*d

8.8 f 3.0

8.7

f

2.5

2.5

f

3.7b,d

4.2

f

3.1

3.5

f

2.5

3.3

f

3.0'

25.7

f

3.6

25.2

f

4.2

23.3

f

4.4b,e

349 t 84

324

f

87'

279

f

81b,d

59.6 f 15.3 10.6 f 4.0 169 f 66

2.6 f 0.7

60.4 f 16.1 9.8 f 3.9 150 t 62

2.8

f

0.8

62.4 f 17.1bsd 11.2 f 4.2d 165 f 81

2.3 f 0.7',d

Values are mean + 1 standard deviation. Repeated measures analysis of variance and Fisher's protected least significant difference tesi: p < 0.01, p < 0.05 versus 0% flow (off); p < 0.01, p < 0.05 versus 50% flow.

a

AoP = distal aortic pressure; dl'ldt,,, = maximum rate of change of pressure; E, = effective arterial elastance; EDP = end-diastolic pressure; EDV = end-diastolic volume; LAP = left atrial pressure; LV = left ventricular; LVP,,, = maximum left ventricular pressure; PA = pulmonary artery; RV = PAP = pulmonary artery pressure; right ventricular; RVP,,, = maximum right ventricular pressure; SW = stroke work. SV = stroke volume;

RV afterload, fell significantlyby 13%during 100%flow ( p < 0.04).

Septal Shifting As LVAD flow increased from 0% to 100% the septum progressively shifted leftward as LV pressure fell relative to RV pressure (Table 2). This shift was present in the high-septum and low-septum, but was most pronounced in the mid-septum. The mid-SFW,, dimension signifi-

60

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

0% Flow (Off)

Ann Thorac Surg 1993;56:54-67

50% Flow

100% Flow

Table 3. Right Ventricular Systolic Mechanics" Variable

30 I

30 I

30 I

Fig 5. The hemodynamic response of one animal to partial (50%)and full (100%) left ventricular assist device support. As left ventricular assist device pow increased, the systemic pulse pressure, peak left ventricular pressure (LVP), mean left atrial and pulmonary artery pressures (LAP, PAP), and peak right ventricular pressure (RVP) all progressively declined. However, pulmonary artery (PA) flow did not change. (AoP = aortic pressure.)

cantly decreased by 14% with full LVAD support (p < O.OOOl), whereas mid-SFW,, significantly increased by 10%(p < 0.0005). The mid-SFW,, dimension at 50% flow was unchanged compared with control conditions ( p > 0.37), but thereafter increased significantly when LVAD flow was taken to 100% (p < 0.0005).

ESPVR E, (mm Hgl mL) v, (mL) r PRSW M, (mm Hg) v w (mu r dPIdt,, versus EDV MdP/d\(mm Hg * s - .mL-*) VdP/dt (mL) r

0% (off)

50%

2.4 f 1.0

2.1 +. 1.2

41 f 11 0.96 f 0.04

40 f 11 0.98 f 0.01

41 f 14 0.98 f 0.03

14.1

f 3.3 48 f 12 0.99 -t 0.00

12.3 f 3.8' 48 f 13 0.98 2 0.02

12.1 f 3.F 14 49 0.98 f 0.03

l7

12 +. 7

9 5 4b

34 f 11 0.94 t 0.10

27 f 15 0.88 f 0.18

23 f 24 0.93 f 0.07

100% 1.7

f

0.7b

*

Values are mean 2 1 standard deviation. Repeated measures analysis of variance and Fisher's protected least significant difference test p < 0.01, 'p < 0.05 versus 0% flow (off). a

dP/dt,,, = maximum rate of change of pressure; EDV = end-diastolic E,, = end-systolic elastance; ESPVR = end-systolic presvolume; M,,,, = slope of dP/dt; sure-volume relationship; M, = slope of PRSW; PRSW = preload recruitable stroke work; r = correlation V, = volumecoefficient; Vdp,dt= volume-axis intercept of dP/dt; V, = volume-axis intercept of ESPVR. axis intercept of PRSW;

animal (the same one illustrated in Figure 5), as peak LV systolic pressure fell from 81 to 73 to 32 mm Hg with 0%, 50%, and 100%LVAD support, RV E,, fell from 3.2 to 2.4 Although the fall in global RV end-systolic elastance was to 2.0 mm Hg/mL, respectively. Figure 7A illustrates the not significant at 50% LVAD flow (p > 0.17), RV E,, changes in RV E,, with partial and full LVAD support in declined significantly by 29% (p < 0.004) with full LVAD all animals. (One animal experienced sustained pulsus support (Table 3). The volume intercept (V,) did not alternans at 100% LVAD flow with two alternating conchange at any level of LVAD support (p > 0.77). Figures 6 tractile states 140, 411. The slopes of the end-systolic and 7A demonstrate these changes. Figure 6 illustrates pressure-volume relationship, preload recruitable stroke representative RV pressurevolume loops from 1 animal work, and dP/dt,,,-EDV relationship of the strong beats at 0%, 50%, and 100% LVAD flow. In this particular were greater than those of the weak beats 11.9 versus 1.7 mm Hg/mL, 11.4 versus 8.4 mm Hg, and 12 versus Table 2. Biventricular End-Diastolic Septal-Free Wall 8 mm Hg s-l mL-', respectively], whereas the interDimensions" cepts were similar 130 versus 31 mL, 38 versus 39 mL, and 26 versus 22 mL, respectively]. The two states were, Region 0% (off) 50% 100% therefore, averaged to yield a single value for statistical High-septum (cm) comparison.) Similarly, the slope of the preload re4.14 f 1.1' 3.84 f l.lb,d cruitable stroke work relationship fell by 14% (p < 0.02) LV (basal 4.21 2 1.1 region) with full LVAD support (Fig 7B), whereas the volume 3.05 f 0.8 3.09 f 0.8 3.13 f 0.8b*e RV (RVOT) intercept did not change (p > 0.13). The slope of the Mid-septum (cm) dP/dt,,,-EDV relationship fell by 47% during full LVAD LV (equatorial 4.81 2 0.7 4.64 +. 0.T 4.13 f 0.6b,d support (p < 0.006), also reflecting a decrease in global RV region) contractility, whereas the intercept did not change (p > 3.29 f 0.5 3.34 +. 0.5 3.65 +. 0.6b*d RV (sinus 0.17) (Fig 7C). region)

RV Contractility

-

Low-septum (cm) LV (apical region) RV (inlet region)

4.67 2 1.2

4.47

1.2'

4.02 f 1.3brd

2.48 f 0.5

2.56 f 0.5

2.78 5 0.6brd

f

Values are mean k 1 standard deviation. Repeated measures analysis of variance and Fisher's protected least significant difference test: p < 0.01, p < 0.05 versus 0% flow (off); p < 0.01, p < 0.05 versus 50% flow. RV = right ventricle. LV = left ventricle; a

-

RV-Arterial Coupling, Derived Energetics, and Power output Although there was a significant decline in RV E,, (contractility), the simultaneous decline in E, (afterload) resulted in no significant change in the ventricular-arterial coupling ratio (Ea/Ees) (Table 4). These EJE,, ratios, however, were higher than that favoring optimal power output (E,/E,, = 1.0) and substantially greater than that

Ann Thorac Surg 1993;5654-67

50% Flow

0% Flow (Off) 30

'

I

100% Flow

I

30

0

40

50

60

50

40

RV Volume Iml)

60

50

40

RV Volume (ml)

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

61

Fig 6 . Pressure-volume loops for the right ventricle (RV) obtained with caval occlusion in 1 dog during 0%, 50%, and 100% left ventricular assist device support. The corresponding end-systolic pressurevolume relations were highly linear within the physiologic range of pressures studied. The slope (E), progressively decreased in this example while the volumeaxis intercept (VO, did not change. (RVP = right ventricular pressure.)

60

RV Volume (ml)

tethering to the contiguous cardiac structures. As the septum shifts leftward, its contribution to RV contraction decreases. Damiano and associates [42] elegantly demonstrated that RV function is highly dependent on contraction of the left ventricle and septum. To evaluate the relative contributions of the RV free wall and septum to RV function, they electrically isolated the RV free wall from the LV by surgcally detaching it from the septum at the interventricular groove and then resuturing the free wall back into its normal anatomic position. Pacing each ventricle independently, they found that contraction of the LV was responsible for 64% of RV developed pressure and 68% of pulmonary artery flow. With increasing LVAD support, as the septum shifts away from the RV and developed pressure and output of the LV itself decreases, one would expect a decline in RV contractility and output based solely on systolic ventricular interdependence (direct interaction) due to the anatomic coupling of the ventricles. However, although we saw a significant decline in RV contractility with full LVAD support, there was no significant change in RV power output due to simultaneous changes in RV preload and afterload. At 100% LVAD flow, pressure unloading of both the left and right ventricles occurred. As LV pressure decreased, left atrial, PA, and, consequently, RV pressure passively declined. This resulted in a decrease in RV afterload as reflected by a 38% decline in PA input impedance and a 13% reduction in PA elastance (RV Ea). A small, but significant, increase in RV preload (EDV)was

associated with optimal energetic efficiency (E,/E,, = 0.5); this reflects the adrenergically blocked state under which these dogs were studied. In addition, there was no change in the RV pump efficiency of converting total energy (PVA) to external pressure-volume work (RV SWPVA). The maintenance of myocardial efficiency reflects the constant relationship between RV and PA elastance (consistent ventricular-arterial coupling). Furthermore, RV pump performance, as quantified by RV power output, was unchanged with LVAD support (0%: 85.0? 42.2mW, 50%: 81.8 k 42.1 mW, 100%:83.4? 46.7 mW; p > 0.74) (Fig 8). Therefore, although LV support impaired RV myocardial contractility, RV pump performance was not impaired as evidenced by no change in RV total power output or stroke work ( p > 0.17) (see Table 1).

Comment These results reveal that left ventricular assistance can significantly alter RV systolic mechanics. With decreasing transseptal pressures, septal position progressively shifted toward the left ventricle, and this leftward septal shift was associated with a significant decline in RV contractility as evidenced by a 26% fall in E,,, 15% decline in preload recruitable stroke work, and a 47% decrease in the slope of the dP/dt,,,-EDV relationship. The septal shift was progressive as LVAD support increased from 0% to 50% to loo%, and was most notable in the midventricular septum, where one would expect the least

Fig 7. Right ventricular global conI

\

*

3

=

I

10 5 0%

Wh 100%

0%

LVAD Flow

A

B

tractility. Data for the 8 individual animals at 0%, 50%, and 100% left ventricular assist device (LVAD) flow are demonstrated by the solid circles, and group means one standard deviation are represented by the open circles with error bars at 0% and 100% support. With full LVAD support, there was a significant decline in (A) E,, (slope of the end-systolic pressure-volume relationship; p < 0.004 versus 0% flow), (B) M, (slope of the preload recruitable stroke work relationship; p < 0.02 versus 0% flow), and (C) M,,,, (slope of the dP/dt,,,EDV relationship; p < 0.006).

0%

50%

100%

LVAD flow

C

62

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

Ann Thorac Surg 1993;56:5&67

Table 4. Right Ventricular Energetics and Ventricular-Arterial Coupling" 0% (off)

Variable

50%

~

100% ~~

RV SWIPVA RV E a K s

0.67 2 0.16 1.2 f 0.4

0.59 2 0.19 1.6 f 0.8

*

0.22 1.5 2 0.5

0.63

a Values are mean f 1 standard deviation; p = not significant for all comparisons between all levels of support by repeated-measures analysis of variance. EJE,, = right ventricular-arterial coupling ratio; PVA = pressurevolume area; RV = right ventricular; SW = stroke work.

also observed as the LVAD became responsible for all systemic flow. At constant heart rate, ventricular output or pump performance is a function of preload, afterload, and contractility. In our preparation, the decrease in RV contractility secondary to changes in septa1 position was balanced by an increase in RV preload and decrease in RV afterload. The net result was maintenance of RV power output and overall systolic pump performance. In addition, there was no decline in myocardial efficiency of the RV as a result of these changes in systolic mechanics. In the time-varying elastance model, PVA represents the total mechanical energy generated by the ventricle during contraction [35]. The PVA includes both the external mechanical work performed by the ventricle during systole (stroke work) and the elastic potential energy stored in the myocardium at end-systole. The mechanical or pump efficiency of the RV, estimated by RV SWPVA, was unaffected by changes in LVAD flow; because PVA parallels myocardial oxygen consumption, the ability of the ventricle to transfer total energy from myocardial oxygen consumption into external work presumably also did not change [35, 361. Although RV contractility was impaired, the concurrent decrease in afterload and increase in preload allowed maintenance of both its efficiency and output. This also reflects constant RV ventricular-arterial matching during increasing LVAD support. While RV E,, decreased, effective E, decreased simulta-

T

0%

50%

T

I

OOscillatory =Steady-Flow

100%

LVAD Flow Figure 8. Right ventricular power output with 0%, 50%, and 100% left ventricular assist device (LVAD) pow. No significant difference was seen between the groups in steady flow power (dark bars), oscillatory (pulsatile) power (open bars), or total right ventricular power (steady pow + oscillatory power). The error bars represent one standard deviation above (oscillatory) or below (steady flow) the mean.

neously, due to changes in left atrial and PA pressures decreasing PA input impedance. Thus, although there is impairment of the contractile state of the right ventricle with LV unloading, the LVAD does not impair RV energetics, RV-arterial coupling, or RV power output. Use of the LVAD has been associated with RV failure in humans, occurring in 20% to 25% of patients requiring LV support, and the population at risk for this problem will most likely increase with the impending Food and Drug Administration approval of implantable, permanent LVADs for chronic congestive heart failure [ M I . Clinically, RV failure exists when cardiac index falls to less than 1.8 L * min-' m-* and there is an inability to completely fill the left atrium or LV despite a right atrial pressure greater than 20 mm Hg. The major determinant of survival after LVAD placement rests in the ability of the RV to provide sufficient output to fill the LVAD itself. Initial supportive measures should include volume loading and inotropic support with isoproterenol, dobutamine, or dopamine, perhaps even followed by PA balloon counterpulsation in selected cases [43]. Refractory RV failure after LVAD insertion, however, requires simultaneous right ventricular assist device mechanical support to avert impending cardiovascular collapse. Many studies using both animals and humans have been performed to determine the effects of LV unloading on RV function and to elucidate the mechanism of RV failure during LVAD support, but their results have been confounding due to the diverse conditions under which they were performed [9-171. Miyamoto and associates [9] evaluated RV function during LV assist in open-chest dogs using load-dependent measures of cardiac contractility and found that as LVAD support increased to 60% and then loo%, RV dP/dt,,, fell significantly by 9% and 18%compared with control (LVAD off), but peak RVP did not significantly change. Farrar and colleagues [lo] also studied the effects of LV assist on RV function in openchest dogs, but used thermodilution techniques to estimate RV volume. They found no change in RV output and a decrease in RV dPIdt,,,, but, unlike Miyamoto and associates, they also observed a 10% to 30% decrease in maximum RVP and mean PAP, which, when coupled with constant stroke volumes and PA flows, represented a decrease in RV afterload. In a later study, Chow and Farrar [ l l ] used ultrasonic crystals to measure RV and LV septal-free wall dimension and RV anterior-posterior dimension. Santamore and co-workers [44] have shown that RV contraction is the result of a decrease in the surface area of the RV free wall and the RV septal-free wall distance during systole. In Chow and Farrar's study, as LVP,,, fell by 89%, LV septal-free wall end-diastolic distance decreased 19% while RV septal-free wall distance increased 18%, but the RV anterior-posterior dimension did not change. Although they reported no significant change in the slope of the RV septal-free wall dimension-end-systolic pressure relationship (an estimate of regional RV contractility) or the regional stroke work end-diastolic dimension relationship (regional preload recruitable stroke work), they observed a 17% increase in the dimension-axisintercept of both relations. In

-

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

Ann Thorac Surg 1993;565447

a similar preparation, Fukamachi and associates [12] integrated PA flow to estimate relative RV stroke volume: as the LVAD-assisted flow ratio (bypass flow1PA flow) exceeded 75%, Em,, decreased 14% while its volume-axis intercept increased more than lo%, suggesting impairment of RV contractility with LVAD support. Elbeery and co-workers [13] first used ultrasonic crystals and the shell subtraction technique to estimate RV volume and evaluate load-independent measures of RV contractility during LVAD support. The shell subtraction technique involves the subtraction of LV elliptical volume from total cardiac volume to yield an estimate of RV volume [45]. Although this approach is based on strong mathematical foundations, it assumes that the right and left ventricles maintain an elliptical shape throughout the cardiac cycle, which may not always be the case. The calculation of RV volume using this technique has recently been shown to be influenced by changes in LV geometry, and, therefore, it may be inconsistent during LVAD support [24]. In Elbeery and co-workers’ [13] surgical preparation, the LV was unloaded using a 38F cannula in the left atrium of open-chest dogs. Although they found a small but significant increase in RV SFW distance (6%;p < 0.05), they did not find any change in RV EDV, RVP,,,, RV SW, or the preload recruitable stroke work slope or intercept; however, the degree of LV unloading they were able to achieve with the left atrial cannula was not relatively high. With the LVAD on, LVP,,, fell only 19% (108 & 15 versus 87 29 mm Hg) and LV dP/dt,,, remained relatively high (LVAD off 2,335 698 mm Hg/s versus LVAD on: 1,565 & 624 mm Hg/s). Although these decrements were significant, they did not represent much more than 50% unloading of the left ventricle, which we and others have shown only minimally perturbs RV systolic function [9, 11, 131. With greater levels of LV unloading, they may have seen more of a decline in RV contractility during LV assist. In human subjects, Morita and associates [16] studied the effects of LV unloading shortly after implantable LVAD (Novacor)placement in the operating room. Transesophageal echocardiography was used to measure RV cross-sectional area and RV fractional area change (FAC) as an approximation of RV ejection fraction or output. This allowed calculation of the RV end-systolic pressurearea relationship and RV afterload (RV E,) in 8 patients before and after LVAD implantation (as a bridge to transplantation). They found a 31% increase in FAC, 59% decrease in E,, and a significant shift of the end-systolic pressure-area relationship to the right (increase in its area-axis intercept [A,]). Multiple linear regression analysis was used to determine the relative influence of changes in contractility and afterload on FAC. The concept of ventricular-arterial coupling predicts that an increase in A, (reduced RV systolic mechanics) would decrease FAC, whereas a reduction in E, would increase FAC. In this study, Morita and associates found that the positive influence of a reduction in afterload on FAC was larger than the negative influence of a decrease in contractility; the improvement in FAC was due to lower RV afterload, which overwhelmed the negative influence of

*

*

63

the rightward shift of the end-systolic pressure-area relationship. In a follow-up report, this group reported that RV mechanical efficiency (SWIPVA) improved after LVAD placement [17]. Although we did not see a change in RV efficiencyin these healthy animals, a major cause of RV failure in patients with impaired LV function is high RV afterload secondary to pulmonary hypertension; therefore, a major decrease in RV afterload could outweigh a minor decrease in RV contractility when the right ventricle has been pressure overloaded. The two goals of mechanical left heart bypass are to maintain satisfactory systemic flow while decreasing LV oxygen demand. The adverse perturbations in RV function associated with LV unloading become more substantial as the level of LVAD support increases [9, 11, 121. Therefore, one needs to balance the possible detrimental effects of LVAD support on RV systolic mechanics with the potential benefits of totally unloading a failing left ventricle. It has been suggested that because total cardiac functional recovery (RV and LV) is the ultimate goal of temporary cardiac support, biventricular support should probably be considered if flow rates greater than 80% to 90% are required to maintain the systemic circulation [9, 461. In this study using hearts with normal right ventricles, RV myocardial efficiency and pump performance were not impaired, even at flow rates of 100%;therefore, this clinical recommendation may only be applicable in patients with underlying cardiac or pulmonary disease. Obviously, further studies need to be performed to determine the effects of LVAD support on RV function in animals or patients with ischemic or dilated right ventricles, as the effects of LV assist may be different. Mechanical support of only one ventricle in a patient with biventricular failure may unmask preexisting dysfunction of the other ventricle, whereas LVAD support in a patient with a normal RV and low cardiac output may actually improve RV contractility and function [8, 12, 14-17]. In a low cardiac output state, aortic pressure and pari pasu coronary perfusion pressure may be low, resulting in global myocardial ischemia. With LVAD support, coronary flow may increase, ameliorating RV ischemia if present and improving RV contractility. The resultant salutary effect would, therefore, be a result of increased coronary perfusion. In a severely ischemic failing heart, this increase in RV contractility could potentially outweigh the physiologic sequelae of any leftward septal shift. Furthermore, with a dilated myopathic left ventricle, the septum may abnormally encroach on the right ventricle before LVAD support, and LV unloading, which causes a leftward septal shift, may significantly improve RV compliance, thereby augmenting RV filling and forward output (12,471. Finally, although LV unloading may relieve passive pulmonary congestion and decrease RV afterload in subjects with a failing left ventricle, patients with obstructive pulmonary disease and elevated pulmonary vascular resistance may not benefit from this reduction in left-sided filling pressures [6, 81. The increase in PA flow associated with an increase in RV output may actually increase RV afterload due to the poor compliance of the pulmonary vascular system. Obviously, the effects

64

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

of LV assist will depend on the underlying pathophysiologic status of the patient, and further controlled experiments need to be performed in animals and human subjects with dilated or ischemic ventricles. This will allow determination of which physiologic changes are due to the LVAD itself, and which are due to secondary alterations involving the underlying cardiopulmonary status. This might help identify which patients would benefit most from biventricular versus univentricular assist systems. Technical assistance was provided by Geraldine C. Derby, RN, BS, Cynthia E. Handen, BA, Mary K. Zasio, BA, Carol W. Mead, BA, and Erin K. Schultz, BS. We gratefully acknowledge Phoebe E. Taboada for her assistance in the preparation of the manuscript. In addition, we would like to thank Medtronic Bio-Medicus, Inc, for donating the left ventricular assist device and cannulas used in this experiment. Supported by grant HL-29589 from the National Heart, Lung, and Blood Institute and the Veterans Administration Medical Research Service. Doctor Moon was supported by Individual National Research Service Award HL-08532 from the National Heart, Lung, and Blood Institute, and Drs Moon, Castro, and DeAnda are Carl and Leah McConnell Cardiovascular Surgical Research Fellows.

References 1. DeBakey ME. Left ventricular bypass pump for cardiac assistance: clinical experience. Am J Cardiol 1971;27%11. 2. Golding LR, Jacobs G, Groves LK, Gill CC, Nose Y, Loop FD. Clinical results of mechanical support of the failing left ventricle. J Thorac Cardiovasc Surg 1982;83:597-601. 3. Pennock JL, Pierce WS, Wisman CB, Bull AP, Waldhausen JA. Survival and complications following ventricular assist pumping for cardiogenic shock. Ann Surg 1983;198:469-78. 4. Pennock JL, Pierce WS, Campbell DB, et al. Mechanical support of the circulation followed by cardiac transplantation. J Thorac Cardiovasc Surg 1986;92:994-1004. 5. Pae WE, Rosenberg G, Donachy JH, et al. Mechanical circulatory assistance for postoperative cardiogenic shock a three year experience. Trans Am SOCArtif Intern Organs 1980;26 256-61. 6. Richenbacher WE, Pierce WS. Right ventricular failure following implantation of a left ventricular assist device. Curr Surg 1983;40:274-7. 7. Pennington DG, Merjavy JP, Swartz MT, et al. The importance of biventricular failure in patients with postoperative cardiogenic shock. Ann Thorac Surg 1985;39:16-26. 8. Farrar DJ, Compton PG, Hershon JJ, Fonger JD, Hill JD. Right heart interaction with the mechanically assisted left heart. World J Surg 1985;9:89-102. 9. Miyamoto AT, Tanaka S, Matloff JM. Right ventricular function during left heart bypass. J Thorac Cardiovasc Surg 1983;85:49-53. 10. Farrar DJ, 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;70 708-16. 11. Chow E, Farrar DJ. Effects of left ventricular pressure reductions on right ventricular systolic performance. Am J Physiol 1989;257:H1878-85. 12. Fukamachi K, Asou T, Nakamura Y, et al. Effects of left heart bypass on right ventricular performance: evaluation of the right ventricular end-systolic and end-diastolic pressurevolume relation in the in situ normal canine heart. J Thorac Cardiovasc Surg 1990;99:725-34.

Ann Thorac Surg 1993;56:54-67

13. Elbeery JR, Owen CH, Savitt MA, et al. Effects of the left ventricular assist device on right ventricular function. J Thorac Cardiovasc Surg 1990;99:809-16. 14. Farrar DJ, Compton PG, Hershon JJ, Hill JD. Right ventricular function in an operating room model of mechanical left ventricular assistance and its effects in patients with depressed left ventricular function. Circulation 1985;72: 1279-85. 15. Kormos RL, Borovetz HS, Gasior T, et al. Experience with univentricular support in mortally ill cardiac transplant candidates. Ann Thorac Surg 1990;49:261-72. 16. Morita S, Kormos RL, Mandarin0 WA, et al. Right ventriculadarterial coupling in the patient with left ventricular assistance. Circulation 1992;86(Suppl 2):316-25. 17. Morita S, Kormos RL, Eishi K, et al. Left ventricular assistance improves mechanical efficiency of the right ventricle. Surg Forum 1991;42:3114. 18. Schwiep F, Cassidy SS, Ramanathan M, Johnson RL. Rapid in vivo determinations of instantaneous right ventricular pressure and volume in dogs. Am J Physiol 1988;254: H622-30. 19. Von Segesser L, Weiss B, Gallino A, et al. Superior hemodynamics in left heart bypass without systemic heparinization. Eur J Cardiothorac Surg 1990;4:384-9. 20. Niczyporuk MA, Miller DC. Automatic tracking and digitization of multiple radiopaque myocardial markers. Comput Biomed Res 1991;2412942. 21. Daughters GT, Sanders WJ, Miller DC, Schwarzkopf A, Mead CW, Ingels NB. A comparison of two analytical systems for three-dimensional reconstruction from biplane videoradiograms. Proc Comp Cardiol (IEEE) 19887942. 22. Suga H, Sagawa K, Shoukas AA. Load independence of the instantaneous pressure-volume ratio of the canine left ventricle and effects of epinephrine and heart rate on the ratio. Circ Res 1973;32:314-22. 23. Yun KL, Rayhill SC, Niczyporuk MA, et al. Mitral valve replacement in dilated canine hearts with chronic mitral regurgitation: importance of the mitral subvalvular apparatus. Circulation 1991;84(Suppl3):112-24. 24. Moon MR, Castro LJ, Derby GC, et al. Calculation of biventricular volume: myocardial marker vs. sonomicrometric shell subtraction technique. Circulation 1992;86(Suppl1):553. 25. Bergel DH, Milnor WR. Pulmonary vascular impedance in the dog. Circ Res 1965;16:401-15. 26. Milnor WR, Conti CR, Lewis KB, ORourke h4F.Pulmonary arterial pulse wave velocity and impedance in man. Circ Res 1969;25:637-49. 27. Calvin JE, Baer RW, Glantz SA. Pulmonary artery constriction produces a greater right ventricular dynamic afterload than lung microvascular injury in the open chest dog. Circ Res 1985;56:40-56. 28. Sagawa K, Suga H, Shoukas AA, Bakalar KM.End-systolic pressure-volume ratio: a new index of contractility. Am J Cardiol 1977;40:748-53. 29. Maughan WL, Shoukas AA, Sagawa K, Weisfeldt ML. Instantaneous pressure/volume relationship of the canine right ventricle. Circ Res 1979;44:309-15. 30. Glower DD, Spratt JA, Snow ND, et al. Linearity of the Frank-Starling relationship in the intact heart: the concept of preload recruitable stroke work. Circulation 1985;71:9941009. 31. Karunanithi MK, Michniewicz J, Copeland SE, Feneley MP. Right ventricular preload recruitable stroke work, endsystolic pressure-volume, and dP/dt,,,-end-diastolic volume relations compared as indexes of right ventricular contractile performance in conscious dogs. Circ Res 1992;70:1169-79. 32. Little WC. The left ventricular dP/dt,,,-end-diastolic volume relation in closed-chest dogs. Circ Res 1985;56:808-15. 33. Kono A, Maughan WL, Sunagawa K, Hamilton K, Sugawa K, Weisfeldt ML. The use of left ventricular end-ejection pressure and peak pressure in the estimation of the endsystolic pressure-volume relationship. Circulation 1984;70: 1057-65.

65

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

Ann Thorac Surg 1993;565&67

34. Alyono D, Larson VE, Anderson RW. Defining end-systole for end-systolic pressure-volume ratio. J Surg Res 1985;39: 344-450. 35. Suga H, Hayashi T, Shirahata M, Suehiro S, Hisano R. Regression of cardiac oxygen consumption on ventricular pressure-volume area in dog. Am J Physiol1981;240H320-5. 36. Nozawa T, Yasumura Y, Futaki S, Tanaka N, Uenishi M, Suga H. Efficiency of energy transfer from pressure-volume area to external mechanical work increases with contractile state and decreases with afterload in the left ventricle of the anesthetized closed-chest dog. Circulation 1988;771116-24. 37. Sunagawa K, Maughan WL, Burkhoff D, Sagawa K. Left ventricular interaction with arterial load studied in isolated canine ventricle. Am J Physiol 1983;245:H773-80. 38. Burkhoff D, Sagawa K. Ventricular efficiency predicted by an analytical model. Am J Physiol 1986;250:R1021-7. 39. Canner SG, Swanson MR. An evaluation of ten painvise multiple comparison procedures by Monte Carlo methods. J Am Stat Assoc 1973;68:66-74. 40. McGaughey MD, Maughan WL, Sunagawa K, Sagawa K. Alternating contractility in pulsus alternans studied in the isolated canine heart. Circulation 1985;71:357-62. 41. Adler D, Nikolic S, Sonnenblick EH, Yellin EL. Mechanism of sustained mechanical alternans: effect of variations in ventricular filling volume. Circ Res 1991;69:26-38. 42. Damiano RJ, LaFollette P, Cox JL, Lowe JE, Santamore WP. Significant left ventricular contribution to right ventricular systolic function. Am J Physiol 1991;261:H151&24. 43. Miller DC, Moreno-Cabral RJ, Stinson EB, Shinn JA, Shumway NE. Pulmonary artery balloon counterpulsation for acute right ventricular failure. J Thorac Cardiovasc Surg 1980;80:76C-3. 44. Santamore WP, Meier GD, Bove AA. Effects of hemodynamic alterations on wall motion in the canine right ventricle. Am J Physiol 1979;236:H25462. 45. Feneley MP, Elbeery JR, Gaynor JW, Gall SA, Davis JW, Rankin JS. Ellipsoidal shell subtraction model of right ventricular volume: comparison with regional free wall dimensions as indexes of right ventricular function. Circ Res 1990;67142736. 46. Yada I, Wei CM, Hattori R, et al. Right ventricular function during left heart bypass evaluated by two-dimensional echocardiography. Trans Am SOCArtif Intern Organs 1985;31: 17-9. 47. Damiano RJ, Asano T, Smith PK, Ferguson TB, Cox JL. Functional consequences of the right ventricular isolation procedure. J Thorac Cardiovasc Surg 1990;100569-79.

Appendix 1 impedance and RV power Output were determined using Fourier analysis [25-271. With Fourier analysis, a periodic function, such as a pressure or flow signal, is described as the sum of a series of sines and cosines, where each harmonic Of the series is a sinusoidal wave represented by its (amp1itude) and phase. For RVPt and flow, harmonics for each cardiac cycle were calculated and each harmonic was averaged over 9.4 2 2.8 steady-state beats. Calculations were limited to harmonics less than 12 Hz to minimize high-frequency noise [26]. Total pulmonary flow was expressed as:

+ O%Flow(Ofl) 1200

800

- 0-100% Flow

Ra\

Characteristic --1 +

400

0 0

1

2

3

4

c Qnsin ( n o t + 4,J, n

=

1

where Q, is mean flow, and Q, and 4, represent the modulus and phase angle of the nth harmonic, respectively, and w is the fundamental frequency of pulsation (rads/s). Similarly, the RV and PA pressure signals were expressed as:

5

6

7

8

9

10

Frequency (Hz) Fig 9. Pooled pulmonary vascular impedance spectra (mean f one standard deviation at each harmonic) for all 8 animals with 0% left ventricular assist device flow (solid circles, solid line) and 100% flow (open circles, dotted line). The modulus at 0 Hz, representing pulmonary input impedance, is significantly lower with full left ventricular assist device support (p < 0.007), whereas the higher frequency data, representing characteristic impedance of the pulmona y vascular bed, are superimposed.

where Po is mean pressure, and P, and p, represent the modulus and phase angle of the nth harmonic. All harmonics with a pressure amplitude less than 0.1 mm Hg or a flow amplitude less than 0.5 mL/s were discarded, as these values were at the noise level of our measurement techniques. At each harmonic, the impedence modulus (Z,) was defined as the ratio of the PAP and PA flow moduli (2, = PAP,/Q,), and the phase angle (0,) was the difference between their phases (0, = p, - 4,).The impedance phase is negative if flow precedes pressure, whereas a positive phase occurs when pressure precedes flow. Pulmonary input impedance (Z,) was defined as vascular impedance at 0 Hz (Z, = PAPJQ,), and the average impedance between the 5th and 11th harmonics quantified characteristic impedance (ZJ. 9 . Figure illustrates the pooled impedance spectra for all animals at 0% and 100% LVAD flow. Pulmonary input impedance is the impedance at 0 Hz, whereas 2, is the average impedance between 5 and 11 H ~H~~~ , it can be Seen that Z, decreases with 100% LVAD support, while Z, does not change, To calculate the total external work of the right ventricle, power output is separated into steady-flow power (W,) and pulsatile power (wP),steady-flow power (w,) is the product (osci~~atory) of mean Rvp and mean PA flow (w, = RVP, Q,), and pulsatile power (w,) is calculated using the oscillatory terms of the pressure and flow F~~~~~~series:

.

11

11

Q(t) = Qo +

Irnmance

Wp = 1/2

(Q,)*Z,cosB,,

fl=l

where 2, and 0, are the impedance modulus and phase angle and Q, is the flow modulus at the nth harmonic. Total right ventricular power output (WRv)is the sum of steady-flow power and pulsatile (oscillatory) power (WRv = W, + Wp).

66

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

Ann Thorac Surg 1993;56:5467

DISCUSSION DR ROBERT L. KORMOS (Pittsburgh, PA): You will be pleased to know that the findings that you have demonstrated in this very nicely designed study are comparable with those found in patients with chronic heart failure who undergo left ventricular support. Our findings are very similar to yours in that we find globally impaired right ventricular contractility. However, the afterload reduction of the right ventricle is so dramatic due to the afterload reduction of the left ventricle that these changes override any of the reductions in contractility that you see in the right ventricle. The question that I would like to ask you relates to differential functioning of the septum and right ventricular free wall. I am not sure if you can do this in your study or if you had the opportunity to look at it, but we found that in some circumstances, although septal functioning is impaired in perhaps most of the humans we studied, the free wall is able to take over a significant portion of the contractile performance of the right ventricle. Did you have an opportunity to look at the differential functioning of the septum and right ventricular free wall? DR MOON: We have not looked specifically at differential function of those exact portions of the right ventricle, ie, the septum versus the right ventricular free wall; however, the myocardial marker technique does allow us the opportunity to examine segmental or regional systolic mechanics and geometry in many different parts of the right ventricle. We recently separated the right ventricle into three regions-inlet, sinus, and infundibulum (or right ventricular outflow t r a c t t t o evaluate changes in regional right ventricular systolic dynamics during left ventricular assist device support. Although regional end-systolic elastance fell significantly (by 29%) in the sinus region at full flow, it did not change appreciably in either the right ventricular inlet portion or infundibulum. The sinus component of the chamber (which obviously includes the right ventricular free wall) may be less tethered to the left ventricle and base of the heart and thereby may be affected more by flattening of septal curvature. With respect to your superb clinical studies after implantation of the Baxter Edwards Novacor left ventricular assist device in patients awaiting transplantation, it is indeed gratifying to know that similar changes are seen in patients with diseased hearts. The fact that our current study was performed in animals with normal hearts and pulmonary artery pressures (ie, normal right ventricular afterload) may explain why we observed no change in right ventricular power output with left ventricular assist device support, whereas your group found an increase in right ventricular forward flow. Furthermore, because your patients with ischemic cardiomyopathy had coronary disease, the left ventricular assist device also may have improved right ventricular systolic function by virtue of another mechanism, viz, increased

coronary perfusion. Finally, a leftward septal shift may well augment right ventricular dynamic chamber compliance during diastole and thereby enhance right ventricular filling in patients with the septum abnormally shifted to the right due to chronic left ventricular dilatation. So, I agree that our experimental observations are fully concordant with your clinical findings. We have future plans to study the effects of left ventricular unloading on right ventricular systolic mechanics during myocardial ischemia to determine if hearts with impaired right ventricles will respond differently to left ventricular assist device support.

DR RAY C.-J. CHIU (Montreal, Que, Canada): You mentioned the increased right ventricular preload. Do you actually see that here? Are you basing that statement on the increased right ventricular end-diastolic volume? DR MOON You are quite correct, Dr Chiu. Global right ventricular or left ventricular preload (end-diastolic cardiomyocyte length) is best approximated by end-diastolic volume, or, alternatively, end-diastolic right ventricular wall stress. We did observe a small increase in right ventricular end-diastolic volume. Please remember that right ventricular end-diastolic volume in these autonomically blocked animals averaged approximately 60 mL; it rose by 3 mL with full left ventricular assist device flow, or by an average of 4.6%. DR CHIU: Yes, but these volume changes are associated with shifts of the septum. Do they actually represent increased left ventricular end-diastolic pressure?

DR MOON: Depending on the specific techniques used to estimate right ventricular volume, one can be led astray by apparent changes in right ventricular end-diastolic volume. For example, employing the shell subtraction technique (using sonomicrometrically measured cardiac dimensions), measurement of right ventricular volume is indirect and depends not only on left ventricular volume, but also on septal position. As we showed in a report presented at the American Heart Association Scientific Sessions in 1992, it is possible that changes in left ventricular volume could affect the measurement of right ventricular volume without the volume on the right side actually changing. Using the myocardial marker biplane cinefluoroscopic method, one can precisely model the entire right ventricle in three-dimensional space, independent of left ventricular and septal geometry. Therefore, we feel confident that our calculation of right ventricular volume was not affected by changes in left-sided volume, septal position, or left ventricular end-diastolic pressure. In fact, left ventricular end-diastolic pressure actually fell markedly as left ventricular assist device flow was increased due to progressive unloading of the left ventricle.

INVITED COMMENTARY Our understanding of right heart physiology during mechanical left ventricular support has improved over the last decade. But in the clinical arena we still cannot predict with great confidence which patients will require biventricular support and which can be supported with isolated

left heart assistance. It is clear, however, that adequate right ventricular function is essential for the maintenance of cardiac output and successful use of left ventricular assist devices. One of the major limitations of experimental studies to

MOONETAL RV DYNAMICS DURING LVAD SUPPORT

Ann Thorac Surg 1993;5635447

date has been the use of inadequate methods of determining right and left ventricular volumes. This report by Moon and associates, using implanted tantalum myocardial markers and biplane cinefluoroscopic images, provides the most convincing techniques to date for measuring biventricular volumes during left ventricular unloading. The results confirm that there is an interplay between beneficial and detrimental effects of left ventricular unloading on the determinants of right ventricular function: preload, afterload, and contractility [l].Although these findings with a continuous-flow centrifugal pump may not be representative of pulsatile ventricular assist devices in patients with end-stage heart failure, the results support the concept that there can be a reduction in effective right ventricular contractility during left ventricular pressure unloading due to direct anatomic systolic interaction between the ventricles. In healthy animal models such as in this report, detrimental and beneficial effects tend to balance, resulting in no overall change in right ventricular performance due to left heart unloading. Whereas in models of congestive heart failure, the reduction of right ventricular contractility appears to be dominant, resulting in a net impairment of right heart function during left heart assist [2]. On the other hand, in models of right ventricular ischemia, interactions through the interventricular septum appear to be of minor importance to right ventricular function compared with the overwhelming effects of ischemia [3]. In contrast to animal models, one of the most striking findings in the clinical application of ventricular assist devices for bridging to cardiac transplantation is a marked improvement in passive pulmonary artery hypertension.

67

It is not uncommon to see a 30% to 50% reduction in pulmonary artery pressures and input resistance in patients after left ventricular assist device implantation. This substantial beneficial effect to right heart afterload appears to be the dominant factor in most patients, which easily offsets detrimental effects on systolic ventricular interactions via the interventricular septum. The physiologic concepts developed on systolic and diastolic ventricular interactions and of beneficial and detrimental effects on preload, afterload, and contractility appear to be valid, but the net effect of left ventricular assist device support on overall right ventricular function will differ in different pathophysiologic conditions.

David 1. Farrar, PhD Department of Cardiac Surgery California Pacific Medical Center PO Box 7999, Rm S637 Sun Francisco, C A 94120

References 1. Farrar DJ, Compton PG, Hershon JJ, Fonger JD, Hill JD. Right heart interaction with the mechanically assisted left heart. World J Surg 1985;9:89-102. 2. Chow E, Farrar DJ. Right heart function during prosthetic left ventricular assistance in a porcine model of congestive heart failure. J Thorac Cardiovasc Surg 1992;104:569-78. 3. Farrar DJ, Chow E, Compton PG, Foppiano L, Woodard J, Hill JD. Effects of acute right ventricular ischemia on ventricular interactions during prosthetic left ventricular support. J Thorac Cardiovasc Surg 1991;102:588-95.