The objective of this study was to examine the hypothesis that long-term, rapid atrial pacing produces a model of atrial systolic and diastolic dysfunction but does not alter ventricular function. Eight dogs were atrially paced at 400 beats/min (3:1–5:1 ventricular response) for 6 wk and subsequently instrumented with left atrial (LA) and left ventricular (LV) sonomicrometers and micromanometers. Data were compared with those from six sham-operated controls at matched heart rates and mean LA pressures of 10 mmHg. Dogs with rapid pacing had slightly greater LA volume (10.3 ± 4.0 vs. 7.9 ± 4.4 ml) and reduced ejection fraction (2.2 ± 1.4 vs. 13.0 ± 4.0,P < 0.05), systolic ejection rate (0.3 ± 0.1 vs. 2.8 ± 1.2 vol/s,P < 0.05), and reservoir fraction (0.07 ± 0.04 vs. 0.35 ± 0.06,P < 0.05) compared with controls. LA diastolic chamber stiffness was greater after rapid atrial pacing than before (stiffness constantk c, 5.7 ± 2.3 vs. 3.4 ± 0.6, P < 0.05), and the ratio of transesophageal echo-determined pulmonary venous systolic to diastolic integrated flow (a measure of relative reservoir to conduit function of the LA) was less in rapidly paced dogs compared with control dogs (0.41 ± 0.19 vs. 0.68 ± 0.23,P < 0.05). In contrast, rapid atrial pacing did not influence LV systolic performance or lusitropy, because the LV pressure time derivative and the time constant of LV relaxation were similar in both groups. In this model of isolated atrial myopathy, increased atrial stiffness and enhanced conduit function compensate for impaired atrial booster pump and reservoir functions.
- atrial function
- left atrium
- cardiac mechanics
- heart failure
experimental models of both sustained (e.g., 6 wk) and short-term paroxysmal (e.g., 5 and 30 min) atrial fibrillation induced by rapid pacing (≥400 beats/min) have been shown to result in impaired atrial contractility (16, 17, 19). Although we recently demonstrated that after 1 wk of rapid atrial pacing, noninvasively determined indexes of atrial booster pump function were markedly impaired (9), the more chronic effects of atrial tachyarrhythmia on invasively and noninvasively determined measures of atrial systolic and diastolic function are unknown. This gap in our knowledge is particularly relevant insofar as chronic rapid atrial pacing is a potentially important model for the study of both the electrophysiological and mechanical consequences of atrial arrhythmias. For example, the model has been used to demonstrate atrial remodeling and sinus node dysfunction owing to chronic atrial fibrillation (4, 9, 19). Moreover, largely for methodological reasons, the contributions of atrial function toward cardiovascular performance and ventricular filling remain controversial (2, 7, 23). Thus the objective of the current study was to examine the hypothesis that long-term (6 wk), rapid (400 beats/min) atrial pacing myopathy produces a model of isolated atrial myopathy characterized by impaired atrial systolic and diastolic function and unaltered ventricular function.
Fourteen conditioned mongrel dogs were used in these experiments. Eight dogs weighing 21–25 kg (23.7 ± 1.7 kg) were studied 6 wk after rapid (400 beats/min) atrial pacing. Six dogs weighing 19–25 kg (22.8 ± 2.4 kg) were instrumented with a right atrial pacemaker (see below) but were unpaced; these sham-operated dogs served as controls.
Animals were anesthetized with pentobarbital sodium (30 mg/kg iv). A unipolar pacemaker (Medtronic; Bloomington, MN) lead was sutured to the right atrial appendage through a right lateral thoracotomy and small pericardiotomy. In eight dogs, a pulse generator programmed at 400 beats/min was implanted in the subcutaneous tissue over the back of the neck. Six weeks after we instituted rapid pacing, the pacemaker was reprogrammed to 30 beats/min, and the animals underwent hemodynamic study. The remaining six dogs had the pacing lead and pulse generator (set at 30 beats/min) implanted and served as controls. Sham dogs were studied after a minimum of 2 wk after surgery. Long-acting diltiazem (Cardizem 300 mg) was given daily to slow the ventricular response (3:1 to 5:1) in all animals.
Hemodynamic studies were performed as described previously (14). Briefly, animals were anesthetized with pentobarbital sodium (30 mg/kg) and morphine sulfate (3 mg/kg sc) and then intubated and ventilated with a positive-pressure respirator (Harvard Apparatus, South Natick, MA). The heart was exposed with a left lateral thoracotomy at the fourth intercostal space and was suspended in a pericardial cradle. A 7-Fr micromanometer with lumen (Millar Instruments, Houston, TX) was advanced into the left atrium via a pulmonary vein; a second 7-Fr micromanometer with lumen (Millar Instruments) was advanced into the left ventricle through an apical stab wound. A femoral vein was cannulated to administer intravenous fluids. An 8-Fr catheter with a 10-ml capacity balloon was advanced from the femoral vein into the inferior vena cava just below the right atrium, and a second balloon catheter was advanced from the jugular vein into the superior vena cava.
Pairs of 3-MHz sonomicrometers (6-mm diameter, Triton Technology, San Diego, CA) were sewn to the epicardial anterior and posterior walls (long axis) and to the medial and lateral walls (short axis) of the left atrium as previously described (11, 14, 15). A pair of 3-MHz (7-mm diameter) sonomicrometers were sewn to the anterior and posterior left ventricular epicardium to measure the minor axis dimension. The transit time of ultrasound between each crystal pair was measured with a multichannel sonomicrometer (Triton Technology).
The pressure waveforms from the micromanometers were matched with those of the fluid-filled catheters. Analog signals for left ventricular (LV) and left atrial (LA) pressures and atrial dimensions were digitized through an analog-to-digital board (Data Translation, Marlboro, MA) interfaced to an IBM AT computer with a 2-ms sampling frequency and were stored on floppy disk.
Fluid-filled catheters were connected to Statham 23 dB pressure transducers with zero pressure set at the level of the mid right atrium. The electrocardiogram and analog signals for pressures and dimensions were recorded on-line at slow and rapid paper speeds (5, 25, and 100 mm/s) on a multichannel physiological recorder (Gould, Cleveland, OH).
A Hewlett-Packard biplane transesophageal imaging transducer was lubricated and advanced into the esophagus behind the atrium. The left upper pulmonary vein and mitral valve were visualized in transverse and orthogonal longitudinal views. Color flow-directed Doppler identification of intracavitary flow was used in all instances. Transmitral flow was obtained from apical four-chamber and orthogonal two-chamber views with the sample volume placed within the left ventricle between the opened mitral leaflet tips. Pulmonary venous flow velocity was sampled from a left superior pulmonary vein, 1–2 cm proximal to the left atrium. Attempts were made to maintain the angle between the ultrasound beam and various flows within 30°.
Experiments were conducted in accordance with institutional guidelines and the Guide for the Care and Use of Laboratory Animals put forth by the United States Department of Health and Human Services. The experimental protocol was approved by the Institutional Animal Care and Use Committee at the University of Cincinnati.
Hemodynamic studies. The heart rate was slowed with 10–20 mg DKAH 0264 (Boehringer Ingelheim), an agent that inhibits the sinoatrial (SA) nodeI f current without effects on myocardial contractility (8). Constant heart rate was achieved by right atrial pacing with the pacing rate selected to eliminate competing rhythms and to permit separation of active from passive atrial emptying. Atrioventricular (AV) conduction, measured from the onset of the atrial pacing spike to the R wave, was similar in atrial failure animals compared with sham animals (179 ± 17 vs. 174 ± 15 ms). Hemodynamic and dimension data were recorded under steady-state baseline conditions and during abrupt increases in LA pressure and volume generated by either bolus infusion of phenylephrine (200–400 μg iv) or vena caval occlusion. Cholinergic blockade was accomplished using atropine (0.02–0.04 μg/kg iv).
In five rapidly paced dogs, data were obtained before and at the peak effect of a rapid intravenous infusion of 1–2 g CaCl2. Baseline measurements were taken after the animal had recovered from preceding protocols, and peak effects were taken at a steady hemodynamic state.
Steady-state hemodynamic, dimensional, and Doppler echocardiographic data were acquired at a mean LA pressure of ∼10 mmHg for purposes of comparisons at a common atrial pressure. Data were obtained both at matched LA pressure and paced heart rates. All data were acquired with the respirator turned off at end expiration.
In an additional four dogs, AV nodal block was produced by transvenous radiofrequency ablation (Radionics, Burlington, MA), and right atrial and ventricular pacemakers were subsequently placed via a small pericardiotomy (see above) and the transvenous route, respectively. In two dogs, the atrial pacemaker was programmed to 400 beats/min and the ventricular pacemaker was programmed to 100 beats/min; the remaining animals served as sham controls and were unpaced. At the end of 6 wk, the animals were studied with an AV sequential pacemaker set at 72 beats/min with an AV interval of 200 ms (Medtronic, Minneapolis, MN) in the manner described for the larger study.
At the end of the experiment, animals were euthanized with a pentobarbital sodium overdose (65 mg/kg iv). The hearts were immediately removed, divided into right and left atria and right and left ventricles, and weighed on a balance (Galaxy 400, Ohaus, Florham Park, NJ).
LA variables. The LA dimension signals were analyzed as previously described (11, 14, 15). Maximum LA dimension (LAmax) was taken as the largest atrial dimension corresponding to the V wave on the LA pressure tracing, LA end-diastolic dimension (LAed) was taken as the largest diameter immediately preceding the A wave of the LA pressure tracing, and LA end systole (LAes) was taken as the smallest dimension at the end of LA contraction. Relative LA volume was estimated as a general ellipsoid of revolution (14): LA volume = (π/6)(SAX)2 ⋅ (LAX), where SAX is the short or mediolateral axis, and LAX is the long or anteroposterior axis of the left atrium.
LA stroke volume was calculated as LA end-diastolic volume minus end-systolic volume. LA ejection fraction was calculated as 100 × (LA stroke volume/LA end-diastolic volume). The heart rate-corrected mean normalized systolic ejection rate was calculated as the LA ejection fraction divided by the heart rate-corrected LA ejection time; the latter was defined as the time from LAed to LAes, divided by the square root of the R-R interval (3).
LA pressure-volume loops were generated off-line by plotting instantaneous LA pressure and volume data from phenylephrine runs every 2 ms. The LA “A” loop represents active LA contraction, and the “V” loop represents passive LA filling. LA A and V pressure-volume loop areas were determined by planimetry (Sigma plot, Jandel Scientific, San Rafael, CA) at a common LA pressure of ∼10 mmHg. The net work by the atrium was computed by subtracting the V loop area from the A loop area (14).
The atrial diastolic dynamic chamber stiffness constant (k) (modulus of chamber stiffness) was determined by fitting LA pressure-volume data at LAmax to the exponential curve equation, P =Ae k V(Delta Graph, Deltapoint, Monterey, CA), where P is LA pressure, the constant A is they intercept,e is the base of the natural logarithm, and V is LA volume. A range of atrial pressure and volume was derived from either phenylephrine boluses or vena caval occlusion. To normalize for the effects of LA volume and pressure on atrial compliance, stiffness constants (k c) were calculated as the slope of the linear relationship between V(dP/dV) versus P, where dP/dV = kP (18). Atrial stiffness constants were also calculated using an offset term Po; thus P = Po+ Ae kV (Origin, Microcal Software).
In addition to booster pump function, the left atrium serves as a reservoir for LV filling during ventricular systole. Therefore, the increase in LA volume from minimum (at the time of mitral valve closure) to maximum (at the time of mitral valve opening) represents the reservoir volume of the atrium (6). The LA reservoir function was assessed by the reservoir volume fraction at a LA pressure of 10 mmHg, which was computed as the maximum minus minimum LA volume divided by the minimum LA volume.
LV variables. The time constant of LV relaxation was derived from the high-fidelity LV pressure tracing using the method of Weiss et al. (26). Data from five end-expiratory beats were digitized by hand (Sigma Plot) and averaged. LV dP/dt was obtained by electronic differentiation of the high-fidelity LV pressure signal. LV end-diastolic (LVEDD) and end-systolic (LVESD) dimensions were taken as the maximum and minimum LV external sonomicrometer dimensions, respectively. LV fractional shortening (LVFS) was defined as [(LVEDD − LVESD)/LVEDD]. LV volume was estimated from the LV epicardial sonomicrometer diameter (D) as volume = (D 3)(π/6). LV diastolic chamber compliance was estimated by fitting end-diastolic LV pressure and volume data to the exponential curve equation, P =Ae k V.
Echo Doppler variables. Diastolic transmitral waveforms were analyzed for the peak and integral early (E) and late (A) velocities and their ratios (E/A); owing to the relatively slow heart rates, the early and late waveforms did not overlap. Fractional early (E Fx) and fractional late velocity (A Fx) were derived asE/(E+ A) andA/(E+ A), respectively.
Pulmonary venous waveforms were analyzed for the peak and integrated systolic (J) and diastolic (K) velocities, and their respective ratios (J/K) were derived. When necessary, integrals were calculated by dropping a vertical line to the baseline from the intersection of the systolic and diastolic waveforms. The fractional pulmonary venous velocity time integral during ventricular systole (J vti/[J vti+ K vti]) was taken as the reservoir fraction, and the fractional pulmonary venous velocity time integral during diastole (K vti/[J vti+ K vti]) was taken as the conduit fraction (12, 21).
Hemodynamic, dimension, and echocardiographic data in atrial myopathic and sham-operated dogs were compared with unpairedt-tests. Calcium chloride data were compared with paired t-tests. All data are means ± SD. A P value <0.05 was taken to indicate statistical significance.
Hemodynamic, LV, and LA Functional Variables
Hemodynamic and LA functional data are presented in Tables1 and 2 and representative examples of each are shown in Fig.1. When compared with sham-operated controls at matched heart rates, rapidly paced (i.e., 400 beats/min for 6 wk) dogs had a slightly greater LAmax dimension and diastolic stiffness constant and significantly reduced reservoir volume; importantly, atrial systolic shortening was absent. In contrast, LV systolic pressure, peak positive dP/dt, the LV diastolick c, and the time constant of LV isovolumic relaxation were similar. Thus, despite maintenance of LV systolic and diastolic function, LA systolic function was markedly impaired after 6 wk of rapid pacing.
LA pressure volume loops in rapidly paced dogs were characterized by a single clockwise V loop, with an area of 4.0 ± 3.7 mmHg/ml (Fig.2 B). In contrast, a typical figure-eight shape, with a counterclockwise A loop and a clockwise V loop, was seen in sham control dogs (Fig.2 B); the A and V loop areas in these animals were 3.2 ± 2.8 and 3.9 ± 2.5 mmHg/ml, respectively. V loop areas were similar in rapidly paced dogs and sham control dogs.
The results for the four animals with AV nodal ablation and right atrial and ventricular pacemakers (Table 3) confirm those reported for dogs receiving calcium channel blockers.
Diastolic transmitral velocities after 6 wk of rapid atrial pacing and in control dogs are summarized in Table 4. Late transmitral (A) peak and integral velocities were both significantly smaller in rapidly paced dogs than control dogs; however, the early transmitral (E) peak and integral velocities were similar. Accordingly, the early-to-late ratio of transmitral flow was significantly increased. When expressed as fractional filling, there was a significantly greater contribution of early velocity (E Fx) and a smaller contribution of late, systolic (A Fx) transmitral flow velocity in rapidly paced dogs than control dogs. In addition, the sum ofE vti andA vti of mitral flow (a surrogate of stroke volume) was significantly less in the rapidly paced than sham-operated animals.
The peak and integrated systolic flow velocities were significantly less and the peak and integrated diastolic flow velocities were similar in rapidly paced and control dogs. Thus the ratios of peak and integrated systolic to diastolic pulmonary vein flow velocities (estimates of relative reservoir to conduit function) were significantly less in atrial myopathic than control dogs; in addition, the reservoir fraction was less and the conduit fraction was greater.
Calcium Chloride Protocol
The effects of calcium chloride infusion in dogs with rapid pacing-induced atrial myopathy and in normal historical control dogs are shown in Tables 5 and6. At similar heart rates and mean LA pressures, calcium produced a significant increase in LV systolic pressure, dP/dt max, and fractional shortening but did not change the time constant of LV isovolumic relaxation. In contrast to its effects on ventricular systolic function, calcium infusion produced only a small, albeit significant, increase in atrial systolic shortening.
LA and Ventricular Mass
LA mass was 18.4 ± 5.1 g in the rapidly paced dogs and 16.7 ± 4.3 g in sham control dogs (P = not significant). In addition, the LA mass, either corrected for body weight (0.78 ± 0.20 vs. 0.71 ± 0.11 g/kg) or LV mass (0.16 ± 0.04 vs. 0.14 ± 0.02), was similar in rapidly paced dogs vs. sham control dogs.
The principal finding of this study is that 6 wk of rapid (400 beats/min) atrial pacing and a normal ventricular rate response produce an isolated atrial cardiomyopathy characterized by impaired booster pump and reservoir functions and increased chamber stiffness and relative conduit function. The model mimics the effects of rapid atrial tachycardias, such as atrial flutter and fibrillation independently of ventricular function; this is important insofar as in our previous study (14), ventricular dysfunction produced compensatory increases in LA function. Further studies are needed to determine whether the atrial myopathy observed in this study may be modulated by the adaptive (and directionally opposite) LA functional changes that occur in response to coexisting LV dysfunction.
The loss of atrial booster pump function at the paced heart rates as assessed by sonomicrometry precluded measurement of atrial “A” wave pressure-volume loop areas and atrial systolic elastance; indeed, atrial ejection phase indexes indicated that the LA body failed to shorten during atrial systole. Thus net atrial work (A-V loop area) was negative in atrial myopathic dogs; these data indicate that net work was performed on the left atrium and suggest that atrial efficiency is considerably reduced in this model (24). The altered Doppler filling indexes confirm the severe impairment of atrial contractile function in these animals. However, further work is needed to determine the effect of atrial inefficiency on LV systolic performance.
The present study also indicates that decreased diastolic LA compliance (increased stiffness constant), impaired atrial reservoir function (decreased reservoir fraction), and relative increases in the conduit-to-reservoir capacity as assessed from the pulmonary vein Doppler occur in the absence of LA hypertrophy. To compare chamber stiffnesses of left atria with different unstressed volumes, we performed a volume normalization and compared stiffness over a common level of LA pressure. Thus the observed differences did not result from differences in operative compliance. The contributions of abnormalities intrinsic and extrinsic to atrial myocardium that are responsible for the reduced distensibility of the left atrium await further study.
We previously showed that calcium infusion causes significant increases in atrial systolic elastance, stroke volume, ejection fraction, and mean ejection rate (14). However, in the present study, atrial shortening indexes during maximal calcium infusion were markedly attenuated. These data are consistent with those from a previous study in which impaired responsiveness to calcium was demonstrated in ventricular trabeculae from dogs with pacing-induced ventricular failure (22). The positive inotropic effectiveness of calcium in our study is evident insofar as indexes of ventricular systolic function were increased and indicates that rapid pacing produces impaired atrial but not ventricular inotropic function to noncatecholamine-mediated stimulation.
Our study provides the first detailed characterization of the functional correlates that accompany the electrophysiological and ultrastructural changes in this model. Although similar models have been described in both the goat and dog, atrial mechanical function in these models has not been directly characterized (19, 27). Decreased atrial refractoriness, increased intra-atrial conduction time, increased susceptibility to and maintenance of atrial fibrillation, and evidence of sinus node dysfunction are associated with a variety of atrial ultrastructural abnormalities, including myocyte hypertrophy, fibrosis, and myofiber disarray (4, 19, 27, 28). Although a marked increase in atrial size was previously demonstrated echocardiographically (19), the abnormalities of atrial systolic reservoir and conduit functions that we report have implications for atrial arrhythmia risk factor assessment and management and for understanding the genesis of atrial arrhythmias.
The mechanism responsible for the atrial abnormalities is not completely understood. Atrial stunning and impaired calcium homeostasis have been implicated, but other mechanisms such as neurohormonal and ultrastructural changes resulting from increased atrial pressure and stretch, and the rapid pacing rates, per se, may be responsible (20). In this regard, it is interesting that the functional changes at 6 wk were similar in many respects to those changes occurring after 1 wk of pacing (9). Of interest, a linear relationship between pacing duration and the duration of induced atrial fibrillation was recently reported (4). Therefore, it is likely that the duration of pacing-induced tachycardia produces a spectrum of disease and that the early changes facilitate the mechanical and electrophysiological alterations that eventuate in structural changes in the atrium. This hypothesis warrants additional testing.
There are several potential limitations of this study. First, we used calcium channel blockers to slow the ventricular response. Although these agents depress myocardial contractility, a recent study suggests that impaired atrial contractility after short-term atrial fibrillation is attenuated by calcium channel blockade. Therefore, it is important that both sham-operated control dogs and experimental animals were given diltiazem; verapamil was avoided in these animals because of the frequency-dependent myocardial depression associated with its use (1). Moreover, the findings in the dogs studied with AV nodal blockade and atrial and ventricular pacemakers support the conclusions of the larger study. Second, improper alignment of LV epicardial and LA diameter gauges produce inaccurate estimates of absolute LV and LA volume, respectively. However, the analytic methods we employed are not dependent on absolute volume determinations. Third, because of the complex instrumentation required, terminal hemodynamic studies were performed in anesthetized, open chest animals. Despite the cardiodepression and blunted reflexes associated with anesthesia (25), time-dependent hemodynamic changes are qualitatively similar in conscious and anesthetized animals. A related potential problem relates to the loss of pericardial influences on reservoir and conduit functions in the terminal studies. Although we recently demonstrated that the pericardium reduces atrial compliance and that pericardial restraint increases directly with atrial volume (13), the present study was performed with a widely opened pericardium; thus, pericardial effects cannot account for our findings. Nevertheless, it is important that data are extrapolated cautiously from these animal models in human disease. Finally, the use of a simple monoexponential to describe diastolic pressure and volume behavior is potentially limited because it has little physiological basis (5). However, we were primarily interested in demonstrating that for a similar range of LV diastolic pressures and volumes, there was no difference in a frequently used parameter of chamber stiffness.
In conlusion, despite these limitations, we have shown that rapid pacing produces an atrial myopathy characterized by impaired atrial systolic function, diastolic stiffness, and altered atrial reservoir to conduit functions in the context of normal ventricular function. Simultaneous rapid atrial and ventricular pacing should be feasible with experimentally produced AV block and will allow the investigation of a range of atrial and ventricular functions that accompany human atrial arrhythmias.
We gratefully acknowledge the expert secretarial assistance of Norma Burns and the assistance of Jim Reese and Chris Kirk from Medtronics. DKAH 0264 was kindly donated by Boehringer Ingelheim Pharmaceuticals.
Address for reprint requests: B. D. Hoit, Division of Cardiology, Univ. of Cincinnati Medical Center, PO Box 670542, Cincinnati, OH 45267-0542.
This work was supported in part by American Heart Association Grant-in-Aid 9650695N.
- Copyright © 1998 the American Physiological Society