- Email: [email protected]
Frank-Starling mechanism contributes modestly to ventricular performance during atrial fibrillation
Frank-Starling mechanism contributes modestly to ventricular performance during atrial fibrillation Zoran B. Popovic´, MD, PhD, Hirotsugu Yamada, MD, Kent A. Mowrey, MS, Youhua Zhang, MD, PhD, Don W. Wallick, PhD, Richard A. Grimm, DO, James D. Thomas, MD, Todor N. Mazgalev, PhD From the Department of Cardiovascular Medicine, The Cleveland Clinic Foundation, Cleveland, Ohio. OBJECTIVES The aim of this study was to assess whether Frank-Starling mechanism has an independent effect on left ventricular (LV) performance in atrial fibrillation (AF). BACKGROUND Ventricular performance in AF depends on variable contractility through the interval-force mechanism based on the ratio of preceding and pre-preceding RR intervals (RRp/RRpp). The impact of end-diastolic volume (EDV) variability, through the Frank-Starling mechanism, is not well understood. METHODS We induced AF in 16 open chest dogs. RR intervals, LV pressure, LV volume, and aortic flow were collected for ⬎400 beats during rapid AF (ventricular cycle length 292 ⫾ 66 ms). In six of the dogs, additional data were collected while average ventricular cycle length was prolonged from 258 ⫾ 34 ms to 445 ⫾ 80 ms by selective vagal nerve stimulation of the AV node. RESULTS The relations of maximal LV power (LVPower) and peak LV pressure derivative (dP/dt) versus RRp/RRpp were fitted to the equation y ⫽ A * (1 ⫺ EXP (RRp/RRppmin ⫺ RRp/RRpp)/C) and the residuals (RES) of these relations were analyzed. LVPower and dP/dt strongly correlated with RRp/RRpp (r2 ⫽ 0.67 ⫾ 0.12 and 0.66 ⫾ 0.12, P ⬍ .0001 for all correlations). Importantly, RES-LVPower and RES-dP/dt showed linear correlation with EDV (r2 ⫽ 0.20 ⫾ 0.14 and r2 ⫽ 0.24 ⫾ 0.17, P ⬍ .01 for all correlations). In the six dogs with slowed average ventricular rate, the slope of both residual relationships (RES-LVPower vs EDV and RES- dP/dt vs EDV) decreased (P ⬍ .03 for both). CONCLUSIONS The Frank-Starling mechanism contributes to ventricular performance in AF independently of the interval-force effects of the beat-to-beat variability in cardiac contractility. The FrankStarling mechanism is sensitive to the average ventricular rate. KEYWORDS Atrial fibrillation; Hemodynamics; Preload; Tachyarrhythmia; Contractility; Frank-Starling mechanism © 2004 Heart Rhythm Society. All rights reserved.
Introduction The increased incidence of atrial fibrillation (AF) observed in the last decade has been described as a new epidemic.1 This has renewed the interest in assessment of left ventricular (LV) performance during AF and in interventions that may improve it.2,3 Efforts have been made to elucidate whether the seemingly random LV performance in AF is due to variable LV contractility2,4 or to beat-to-beat changes of preload (i.e., This study was supported in part by Grant NHLBI RO1 HL60833 from the National Institutes of Health. Address reprint requests and correspondence: Dr. Todor N. Mazgalev, Research Institute FF1-02, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195. E-mail address: [email protected]. (Received May 6, 2004; accepted June 29, 2004.)
Frank-Starling mechanism).5,6 Contractility in AF may vary through the restitution-potentiation mechanism, that is, decreased LV contractility is associated with a short-coupled beat (extrasystolic restitution), whereas a subsequently augmented response may result from so-called postextrasystolic potentiation.7 Quantitatively, these processes are linked to the durations of the RR intervals preceding (RRp) and pre-preceding (RRpp) the current beat and their ratio (RRp/RRpp).7 In contrast, the Frank-Starling mechanism utilizes an initial preload (i.e., end-diastolic volume [EDV]) to adjust the ventricular performance of the following beat, without changes of contractility. In AF, EDV values can fluctuate as uneven filling occurs due to varying diastolic time intervals.8 Recent reports imply that the pseudo-random behavior of LV function during AF may largely be attributed to uneven LV contractility caused by RRp/RRpp variability.7,9 –11 However,
1547-5271/$ -see front matter © 2004 Heart Rhythm Society. All rights reserved. doi:10.1016/j.hrthm.2004.06.016
Heart Rhythm (2004) 4, 482– 489
Popovic´ et al
Frank-Starling Mechanism and AF
483
whether that leaves room for an independent action of FrankStarling mechanism remains controversial. This is a difficult issue to resolve because larger EDVs are correlated with longer cycle length durations (i.e., filling times),8 thus masking the potential contribution of EDV (through the Frank-Starling mechanism) by the recognized role of contractility (through RRp/RRpp). Moreover, LV function-RRp/RRpp relationships are nonlinear, which precludes standard multiple regression analysis.11 For this reason, we conducted a series of hemodynamic experiments in a canine AF model during which we assessed LV volumes by conductance catheter. The primary working hypothesis of the study was that EDV, and thus the Frank-Starling mechanism, influences LV systolic performance in AF independent of cycle length irregularity (RRp/ RRpp). Our second aim was to determine whether the impact of EDV is affected by changes of the average ventricular rate during AF.
Methods The investigation was approved by the Institutional Animal Research Committee and conforms to the “Guide for the Care and Use of Laboratory Animals” published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996).
Experimental procedures We studied 16 healthy, open chest mongrel dogs (body weight 25–30 kg). In 11 of the 16 dogs, acute AF was induced by rapid atrial pacing during the experiments, as previously described.12 The remaining five dogs received high-rate atrial pacemakers, and AF was maintained for 2 months preceding the study. The study protocol was similar to one previously reported.3 Briefly, anesthesia was maintained with 1–2% isoflurane, while volume status, arterial blood gases, and body temperature were monitored. Standard ECG was obtained by subcutaneous needle electrodes. Quadripolar, custom-made, Ag-AgCl plate electrodes were sutured to the high right atrium and right ventricular apex for recording of local atrial and ventricular electrical activity and for pacing. We defined the number of beats and the duration of time intervals (RRp and RRpp) for an individual beat using the right ventricular electrogram (Figure 1). A flowmeter probe (16A/20A, HT 207, Transonic System Inc., Ithaca, NY, USA) was placed around the ascending aorta. A combined conductance-pressure sensor catheter (Millar) was inserted through the left carotid artery and advanced into the LV for pressure/volume measurements. Hemodynamic data were digitized at 1 kHz per channel for up to 1000 ventricular beats, and 400 beats were analyzed in each AF period. For volume calibration, the difference between LV end-diastolic and end-systolic volume points (Figure 1) was divided by the stroke volume obtained by integration of the aortic flow sig-
Figure 1 Typical tracings recorded in one dog during rapid AF (A) and slow AF (B). For illustration of the measuring routine, two representative beats (stars) are shown, along with the corresponding RRp and RRpp intervals, end-diastolic (d), and end-systolic (s) times. d ⫽ point at which d(LVP)/dt reached a threshold of 10% of its peak dP/dt; s ⫽ point of minimal LV volume signal. The nerve stimulation (used for rate slowing) was synchronized with the ventricular electrograms, and brief artifacts can be seen following each RV in panel B. Note that frequent beats (arrows) produce insufficient LV pressure, near-zero dP/dt, and no aortic flow. Such “abortive” beats were less frequent during slow AF. AoF ⫽ aortic flow; dP/dt ⫽ first time-derivative of the left ventricular pressure signal; LVP ⫽ left ventricular pressure; LVV ⫽ left ventricular volume; RV ⫽ right ventricular electrogram.
nal.13,14 The EDV (in milliliters) was determined as a ratio (stroke volume)/(ejection fraction). Ejection fraction was calculated with biplane Simpson equation from epicardial echocardiographic data.11,14 –16 All signals were amplified, displayed, and stored on a dedicated recording system (CardioLab, GE Marquette Medical Systems, Houston, TX). After recording data during sinus rhythm, AF was induced by high-rate atrial pacing and data collection continued after a 15-minute stabilization period. This part of the protocol is referred to as “rapid AF.” In addition, in six dogs selective AV nodal vagal nerve stimulation12 was used to achieve a slower average ventricular rate (“slow AF”), close to the normal sinus rate. To test directly the impact of preload on LV performance in AF, data also were collected after a rapid saline infusion (20 mL/kg) in three dogs.
484
Assessing the impact of Frank-Starling mechanism versus contractility in AF Assessment of EDV and RRp/RRpp as independent predictors of LV performance by standard multiple regression is difficult, because two assumptions are violated. First, LV performance is nonlinearly related to RRp/RRpp.10,11 Second, EDV and RRp/RRpp show strong collinearity due to tight coupling between RRp and LV filling.8 To overcome this, we used the following approach. First, we selected preload-dependent and preload-independent indices for concurrent evaluation. We sought to detect the impact of the Frank-Starling mechanism on the preload-dependent indices: maximal LV power (LVPower, a product of the peak aortic flow and peak LV pressure)1 and peak of LV pressure derivative (dP/dt). In contrast, the LV preload-adjusted maximal power (adj.LVPower ⫽ LVPower/EDV2), which correlates well with the theoretical contractility index Emax,9,17 was chosen as a preload-independent index. Second, we fitted each of the indices as a function of RRp/RRpp using the following equation11: LV Performance index ⫽ Amax ⫻ 关1 ⫺ e共RRp ⁄RRppmin⫺RRp ⁄RRpp兲⁄C兴, where Amax is a maximum response (plateau of the relationship), RRp/RRppmin is minimal RRp/RRpp at which an LV response could be observed, and C is the curvature of the relationship. We expected that not only the adj.LVPower (a contractility index) but also LVPower and dP/dt would exhibit strong correlation with RRp/RRpp, confirming the role of varying contractility during AF. Third, we mathematically eliminated the portion of an LV performance index that was predictable based on its relationship to RRp/RRpp. Thus, we obtained residuals (RES) of each of the relationships by calculating the difference between the observed index values and the values estimated by the nonlinear equations. We assumed that RES reflects the contribution of the preload (EDV) and established that these RES show no first-order dependence on RRp/RRpp. Finally, we calculated the correlation between each RES and EDV. A positive correlation for the RESLVPower and RES-dP/dt would support the independent contribution of the preload, i.e., the Frank-Starling mechanism during AF. To illustrate the theoretical (though unrealistic) extremes, the coefficient of determination r2 ⫽ 1 for the relationship LVPower-RRp/RRpp along with r2 ⫽ 0 for RES LVPower-EDV would indicate that LV performance variability in AF is due entirely to an uneven contractility. In contrast, r2 ⫽ 0 for the relationship LVPower-RRp/RRpp along with r2 ⫽ 1 for RES LVPower-EDV would indicate that LV performance variability is due entirely to changes in preload.
Heart Rhythm, Vol 1, No 4, October 2004
Statistics Data are represented as average ⫾ SD. By calculating the coefficients of determination r2 of individual datasets, we determined the proportion of variance in one variable (e.g., LVPower) that can be accounted for by knowing the second variable (e.g., RRp/RRpp).18 The significance of the nonlinear fit was assessed by computing standard F statistics.19 Standard linear regression methods were applied to obtain 2 r , the slope, and the significance of fit of RES-LVPower, RES-adj.LVPower, and RES-dP/dt to EDV in individual datasets. The correlations between these r2 and the average heart rate were assessed by Spearman’s nonparametric correlation coefficient . The paired data obtained during rapid AF and slow AF were compared by Wilcoxon matched pairs test. P ⬍ .05 was considered significant in all analyses.
Results We recorded and analyzed a total of 27 hemodynamic datasets (16 sets during rapid AF, 8 sets during slow AF, and 3 sets recorded after saline infusion), with the average ventricular cycle length ranging from 220 to 662 ms. During rapid AF, the average ventricular cycle length recorded in all 16 dogs was 292 ⫾ 66 ms (corresponding to 206 bpm). In the six dogs with additional vagal nerve stimulation, the average ventricular cycle length during rapid AF was 258 ⫾ 34 ms (233 bpm) but prolonged to 445 ⫾ 80 ms (135 bpm) during slow AF. Figure 1 illustrates typical raw data from one experiment. Note that during rapid AF (panel A), a relatively large number of beats (as identified by right ventricular electrogram) produced no aortic flow due to insufficient LV pressure. For these beats, both dP/dt and LVPower were exceedingly small (or even zero). The number of such “abortive” beats was greatly reduced during slow AF ( panel B).
Cycle length irregularity and LV systolic performance Figure 2 illustrates examples of the nonlinear relationships LVPower versus RRp/RRpp and dP/dt versus RRp/ RRpp during rapid (panels A and C) and slow AF (panels B and D), respectively. For all experiments, the average r2 was 0.67 ⫾ 0.12 for LVPower and 0.66 ⫾ 0.12 for dP/dt (P ⬍ .0001 for all correlations). Figure 3, panels A and B, illustrate examples of the relationship adj.LVPower versus RRp/ RRpp during rapid and slow AF, respectively. The average r2 of these relationships was 0.60 ⫾ 0.14 (P ⬍ .0001 for all correlations). The data indicate that about two thirds of LV performance variability during AF depended on RRp/RRpp, i.e., on the restitution/potentiation mechanisms that govern the contractility.
Popovic´ et al
Frank-Starling Mechanism and AF
Figure 2 Example of the relationships LVPower versus RRp/ RRpp during rapid AF (A) and slow AF (B) recorded in a single animal. C, D: Examples of relationships dP/dt versus RRp/RRpp during rapid AF (C) and slow AF (D) in the same animal. AF ⫽ atrial fibrillation; dP/dt ⫽ maximum of dP/dt; LVPower ⫽ maximal left ventricular power; RRp ⫽ preceding RR interval; RRpp ⫽ pre-preceding coupling interval. y-axis labels for panels B and D are identical to those for panels A and C, respectively.
Effect of EDV on LV performance indices
485
Figure 4 Scatterplot of RES-LVPower versus EDV during rapid AF (A) and slow AF (B). C, D: RES-dP/dt versus EDV during rapid AF (C) and slow AF (D). Data are from the experiment illustrated in Figure 1. RES ⫽ residuals. Other abbreviations as in previous figures. y-axis labels for panels B and D are identical to those for panels A and C, respectively.
experimental datasets, no significant positive correlation was detected. In the remaining 15 experimental datasets (one illustrated in Figure 5), there was weak correlation of RES-adj.LVPower versus EDV at rapid AF that was further attenuated during slow AF (Table 1). The data confirmed that, after removing the direct dependence on contractility, LV performance indices revealed modest but statistically significant relationship to EDV. Moreover, as expected from the working hypothesis, the Frank-Starling mechanism was evident in the case of LVPower and dP/dt (Figure 4), whereas adj.LVPower was not dependent on EDV (Figure 5).
Figure 4, panels A and B, present scatterplots and linear fits of the residuals RES-LVPower versus EDV obtained from Figure 2, panels A and B, for rapid and slow AF, respectively. Similarly, Figure 4, panels C and D, present RES-dP/dt versus EDV obtained from Figure 2, panels C and D, respectively. In all cases, the illustrated correlations were significant (P ⬍ .005 for all). The composite data for r2 presented in Table 1 indicate that 20 to 25% of the residual LV performance (after accounting for the variable contractility) was linked to the effects of the variable EDV. Figure 5, panels A and B, present scatterplots of RESadj.LVPower versus EDV obtained from Figure 3, panels A and B, for rapid and slow AF, respectively. In nine of 24
Impact of cycle length prolongation on FrankStarling mechanism and contractility
Figure 3 Example of the relationship adj.LVPower versus RRp/ RRpp during rapid AF (A) and slow AF (B), recorded in a single animal. adj.LVPower ⫽ preload-adjusted maximal left ventricular power. Other abbreviations as in previous figures.
Results from a representative dog illustrated in Figures 4 and 5 indicated that the studied residuals exhibited stronger correlations with EDV during rapid AF. The summarized data for average r2 of LV performance residuals in all 24 analyzed datasets during rapid and slow AF are presented in Table 1. In addition, the average heart rate during AF correlated with both r2 of RES-LVPower (Spearman’s ⫽ 0.42, P ⫽ .04) and r2 of RES-dP/dt (Spearman’s ⫽ 0.64, P ⬍ .001; Figure 6), that is, faster fibrillatory rates were associated with larger r2. Furthermore, in the six dogs in which the average ventricular rate was significantly slowed by vagal nerve stimulation, there was a decrease in the slope of RES-LVPower versus EDV. The results for each of these six dogs are shown in Figure 7A. The mean slopes for RES-LVPower were 0.055 ⫾ 0.027 W/mL during rapid AF and 0.026 ⫾
486 Table 1
Heart Rhythm, Vol 1, No 4, October 2004 Average correlation values (r2) for residuals of LV parameters v EDV
Datasets
RES-dP/dt
RES-LVPower
RES-adj.LVPower
Rapid AF (n ⫽ 16) Slow AF (n ⫽ 8) Total (n ⫽ 24)
0.28 ⫾ 0.16 0.11 ⫾ 0.12* 0.24 ⫾ 0.17
0.26 ⫾ 0.14 0.12 ⫾ 0.10* 0.20 ⫾ 0.14
0.13 ⫾ 0.12† 0.02 ⫾ 0.10*† 0.10 ⫾ 0.09†
AF ⫽ atrial fibrillation; EDV ⫽ end-diastolic volume; LV ⫽ left ventricular; RES-adj.LVPower ⫽ residuals of preload-adjusted maximum LV power; RES-dP/dt ⫽ residuals of maximum derivative of LV pressure, RES-LVPower-residuals ⫽ residuals of maximal LV power. *P ⬍ .05 vs Rapid AF. †P ⱕ .01 vs RES-LVPower.
0.014 W/mL during slow AF (P ⫽ .028). Similarly, slow AF decreased the mean slope of the relationship RES-dP/dt versus EDV (30.3 ⫾ 20.1 versus 10.3 ⫾ 730 mmHg/s/mL, P ⫽ .028; Figure 7B). A possible explanation for these findings is that slowing of the ventricular rate during AF enhanced contractility and therefore masked the relative contribution of the Frank-Starling mechanism to LV performance. To test this explanation, we compared the values of adj.LVPower (a preload-independent, contractility index) obtained before and after the slowing of the ventricular rate. We used a previously described technique that calculates the values of adj.LVPower at RRp/RRpp ⫽ 1 in the regression line.11,14,16 As seen in Figure 8, adj.LVPower increased during slow AF (0.09 ⫾ 0.04 versus 0.11 ⫾ 0.05 W/mL2 ⫻ 10⫺2, P ⫽ .028). This finding suggested that slowing of the average ventricular rate during AF increased the relative contribution of LV contractility as a major determinant of LV performance.
Impact of preload augmentation on LV performance In each of the three animals, saline infusion produced an increase of EDV (from 68 ⫾ 26 to 88 ⫾ 24 mL) and an increase of dP/dt (from 1,110 ⫾ 193 to 1,303 ⫾ 328 mmHg/s) and LVPower (from 1.7 ⫾ 0.3 to 2.4 ⫾ 0.9 W). This indicated that direct EDV augmentation accentuated the Frank-Starling mechanism in the presence of AF. Interestingly, a slight decrease of the average heart rate (from 159 ⫾ 9.6 to 148 ⫾ 4.2 bpm) also was observed.
Figure 5 Scatterplot of RES- adj.LVPower versus EDV during rapid AF (A) and slow AF (B). Data are from the experiment illustrated in Figure 2. Abbreviations as in previous figures.
Discussion Major findings We showed that contractility and the Frank-Starling mechanism were operative during AF, but their relative contributions were substantially different and further modified by the prevailing average ventricular rate. In particular, ventricular rate slowing accentuated the dominant role of contractility and minimized the independent role of FrankStarling mechanism in determining LV performance.
Previous studies The physiologic basis of the beat-to-beat variability in LV performance during AF has been a matter of continuous debate. Previous reports have shown that EDV, along with contractility, is an independent determinant of the ejection fraction during AF.6,20 Specifically, increased EDV values were associated with higher ejection fraction, although this effect diminished with the prolongation of the preceding cycle length. However, ejection fraction is afterload sensitive (through the end-systolic volume, i.e., 1 ⫺ ESV/ EDV)21 and, therefore, is not well suited for evaluation of the Frank-Starling mechanism.2 In fact, some authors argue that the Frank-Starling mechanism is of little importance during AF,4 whereas other authors conclude that both the Frank-Starling mechanism and the varying contractility probably contribute to LV functional performance.10 However, no quantification of the proposed independent roles has been provided.
Figure 6 Correlation of average ventricular rate with r2 of RESLVPowerversus EDV (A) and r2 of RES-dP/dtversus EDV (B). HR ⫽ heart rate; r2 ⫽ coefficient of determination. Other abbreviations as in previous figures.
Popovic´ et al
Frank-Starling Mechanism and AF
Figure 7 Impact of average ventricular rate on the slope of the relationships RES-LVPower versus EDV (A) and RES-dP/dt versus EDV (B) in each of six dogs. Abbreviations as in previous figures.
Control of LV function by Frank-Starling mechanism Even though more than a century has passed since the initial observations,22 the Frank-Starling mechanism is incompletely understood at molecular level. It is known that the larger EDV results in individual sarcomere stretch that increases crossbridge Ca2⫹ affinity.23 However, the exact “mechanical transducer” that links sarcomere stretch and Ca2⫹ affinity is unknown.24 On the other hand, the force-interval relationships underlying contractility depend on sarcolemmal channels that regulate Ca2⫹ release and uptake.25,26 This implies that the two mechanisms may independently affect LV function, the former by increasing Ca2⫹ affinity, the latter by changing Ca2⫹ concentration through sarcolemmal channel activity. It should be pointed out that the impact of Frank-Starling mechanism on LV function in normal conscious subjects is attenuated, because increased preload activates reflex increase of the heart rate, which in turn blunts EDV augmentation.27,28 In all three animals in which volume loading was performed, however, both dP/dt and LVPower increased during the intervention, indicating accentuated recruitment of the Frank-Starling mechanism during ongoing AF. It is conceivable that increased variability of EDV due to variable LV filling8 is a prerequisite for manifestation of the Frank-Starling mechanism during AF.
487 inant mechanism linking the ratio RRp/RRpp and LV performance variability during AF. Importantly, when we mathematically subtracted the effect of contractility and examined the residuals, the experimental data confirmed that the Frank-Starling mechanism had an impact on LV performance that was mediated by the preload (EDV) and was additive to the direct effects of RRp/RRpp variability. It should be stressed, however, that EDV during AF might depend not only on the filling times but also on factors such as respiration and preceding relaxation.
Role of ventricular rate slowing during AF The dependence of LV performance on EDV was decreased by ventricular rate slowing during AF. At the same time, contractility slightly but significantly increased, implying more prompt restitution.25 This is in line with our previous experimental observations demonstrating improvement of LV performance during vagally induced rate slowing.3 These effects are mostly cycle length dependent, because selective stimulation of vagal postganglionic neurons to the AV nodal region has no or minimal effect on cardiac inotropism.29 –31 It should be noted that these important findings apply to the range of high ventricular rates typically associated with AF. Excessive rate slowing (below the normal sinus rate) has an opposite effect: it produces negative inotropic effects through the force-frequency relationships.32-34
Clinical implications Both others and we have previously shown that representative values of systolic performance parameters during
Contractility versus the Frank-Starling mechanism as determinants of irregular LV performance during AF We have confirmed that LV performance in AF strongly depends on ventricular rate irregularity. This can be explained by varying contractility,7 preload variability, the Frank-Starling mechanism,8,10 or to their combination. To discern the important of contractility variability, we attenuated the effects of preload on LV performance by calculating adj.LVPower. The results indicated that the relationship between adj.LVPower and RRp/RRpp was strong (Figure 3), confirming that varying contractility is the dom-
Figure 8 Impact of average ventricular rate on adj.LVPower calculated for RRp/RRpp ⫽ 1 in the regression line. Abbreviations as in previous figures.
488 AF can be obtained from a relatively small number of cardiac cycles if correction for the ratio RRp/RRpp is performed.11,14,16 Our new findings suggest that adjustments taking into consideration both RRp/RRpp and EDV would be superior during rapid AF because they would account for both contractility and the Frank-Starling mechanism.
Study limitations We stress that the design of the present in situ study precludes direct quantitative elucidation of the separate contributions of restitution/potentiation (contractility) and the Frank-Starling mechanism (preload) to ventricular performance during AF. This would require an isolated heart model with independent control of contractility (e.g., via pacing with programmed RRp and RRpp intervals) and preload (e.g., via isovolumic EDV manipulations). However, the relevance of such a model to AF observed in the intact heart requires separate validations. This was an experimental, acute, open chest animal study. Efforts were made to use an AF model with minimal additional perturbations, but we could not avoid some obvious limitations. Along with the use of anesthesia, the animals were not decentralized, leaving open the possibility for some reflex modulation of contractility secondary to alterations of the ventricular rate. Also, we did not evaluate potential effects of afterload on LV performance in AF20; however both dP/dt and LVPower are preload-sensitive, but not afterload-sensitive, parameters.17,35 Our findings are not necessarily applicable to patients with chronic AF with associated cardiovascular pathologies. Previous studies imply that the impact of both ventricular rate irregularity and the Frank-Starling mechanism may be observed in AF patients,2,6 but further clinical evaluations are needed.
Heart Rhythm, Vol 1, No 4, October 2004
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
References 1. Braunwald E. Shattuck lecture: cardiovascular medicine at the turn of the millennium: triumphs, concerns, and opportunities. N Engl J Med 1997;337:1360 –1369. 2. Noble MI. Beat to beat left ventricular performance in spontaneous atrial fibrillation does not depend on afterload. Heart 2000;84:89. 3. Zhang Y, Mowrey KA, Zhuang S, Wallick DW, Popovic ZB, Mazgalev TN. Optimal ventricular rate slowing during atrial fibrillation by feedback AV nodal-selective vagal stimulation. Am J Physiol Heart Circ Physiol 2002;282:H1102–1110. 4. Meijler FL, Strackee J, Van Capelle JL, Du Perron JC. Computer analysis of the RR interval-contractility relationship during random stimulation of the isolated heart. Circ Res 1968;22:695–702. 5. Braunwald E, Frye RL, Aygen MM. Studies on Starling’s law of the heart. III. Observations in patients with mitral stenosis and atrial fibrillation on the relationships between left ventricular end-diastolic segment length, filling pressure, and the characteristics of ventricular contraction. J Clin Invest 1960;39:1874 –1884. 6. Gosselink AT, Blanksma PK, Crijns HJ, Van Gelder IC, de Kam PJ, Hillege HL, Niemeijer MG, Lie KI, Meijler FL. Left ventricular beat-to-beat performance in atrial fibrillation: contribution of Frank-
18. 19. 20.
21.
22. 23. 24.
25.
Starling mechanism after short rather than long RR intervals. J Am Coll Cardiol 1995;26:1516 –1521. Suzuki S, Araki J, Morita T, Mohri S, Mikane T, Yamaguchi H, Sano S, Ohe T, Hirakawa M, Suga H. Ventricular contractility in atrial fibrillation is predictable by mechanical restitution and potentiation. Am J Physiol 1998;275:H1513–1519. Cieslinski A, Hui WK, Oldershaw PJ, Gregoratos G, Gibson D. Interaction between systolic and diastolic time intervals in atrial fibrillation. Br Heart J 1984;51:431– 437. Takagaki M, McCarthy PM, Chung M, Connor J, Dessoffy R, Ochiai Y, Howard M, Doi K, Kopcak M, Mazgalev TN, Fukamachi K. Preload-adjusted maximal power: a novel index of left ventricular contractility in atrial fibrillation. Heart 2002;88:170 –176. Brookes CI, White PA, Staples M, Oldershaw PJ, Redington AN, Collins PD, Noble MI. Myocardial contractility is not constant during spontaneous atrial fibrillation in patients. Circulation 1998;98:1762– 1768. Popovic ZB, Mowrey KA, Zhang Y, Zhuang S, Tabata T, Wallick DW, Grimm RA, Thomas JD, Mazgalev TN. Slow rate during AF improves ventricular performance by reducing sensitivity to cycle length irregularity. Am J Physiol Heart Circ Physiol 2002;283:H2706 – 2713. Wallick DW, Zhang Y, Tabata T, Zhuang S, Mowrey KA, Watanabe J, Greenberg NL, Grimm RA, Mazgalev TN. Selective AV nodal vagal stimulation improves hemodynamics during acute atrial fibrillation in dogs. Am J Physiol Heart Circ Physiol 2001;281:H1490 –1497. Baan J, van der Velde ET, de Bruin HG, Smeenk GJ, Koops J, van Dijk AD, Temmerman D, Senden J, Buis B. Continuous measurement of left ventricular volume in animals and humans by conductance catheter. Circulation 1984;70:812– 823. Tabata T, Grimm RA, Greenberg NL, Agler DA, Mowrey KA, Wallick DW, Zhang Y, Zhuang S, Mazgalev TN, Thomas JD. Assessment of LV systolic function in atrial fibrillation using an index of preceding cardiac cycles. Am J Physiol Heart Circ Physiol 2001;281:H573–580. Ratcliffe MB, Wallace AW, Salahieh A, Hong J, Ruch S, Hall TS. Ventricular volume, chamber stiffness, and function after anteroapical aneurysm plication in the sheep. J Thorac Cardiovasc Surg 2000;119: 115–124. Sumida T, Tanabe K, Yagi T, Kawai J, Konda T, Fujii Y, Okada M, Yamaguchi K, Tani T, Morioka S. Single-beat determination of Doppler-derived aortic flow measurement in patients with atrial fibrillation. J Am Soc Echocardiogr 2003;16:712–715. Kass DA, Beyar R. Evaluation of contractile state by maximal ventricular power divided by the square of end-diastolic volume. Circulation 1991;84:1698 –1708. Dawson-Saunders B, Trapp RG. Basic and Clinical Biostatistics. East Norwalk, CT: Appleton & Lange, 1990. Glantz SA, Slinker BK. Analysis of Variance and Multiple Regression. New York: MCGraw-Hill, 1990:491– 493. Muntinga HJ, Gosselink AT, Blanksma PK, De Kam PJ, Van Der Wall EE, Crijns HJ. Left ventricular beat to beat performance in atrial fibrillation: dependence on contractility, preload, and afterload. Heart 1999;82:575–580. Kass DA, Maughan WL, Guo ZM, Kono A, Sunagawa K, Sagawa K. Comparative influence of load versus inotropic states on indexes of ventricular contractility: experimental and theoretical analysis based on pressure-volume relationships. Circulation 1987;76:1422–1436. Katz AM. Ernest Henry Starling, his predecessors, and the “Law of the Heart.” Circulation 2002;106:2986 –2992. Fuchs F, Smith SH. Calcium, cross-bridges, and the Frank-Starling relationship. News Physiol Sci 2001;16:5–10. Konhilas JP, Irving TC, de Tombe PP. Myofilament calcium sensitivity in skinned rat cardiac trabeculae: role of interfilament spacing. Circ Res 2002;90:59 – 65. Prabhu SD, Freeman GL. Postextrasystolic mechanical restitution in closed-chest dogs. Effect of heart failure. Circulation 1995;92:2652– 2659.
Popovic´ et al
Frank-Starling Mechanism and AF
26. Wier WG, Yue DT. Intracellular calcium transients underlying the short-term force-interval relationship in ferret ventricular myocardium. J Physiol 1986;376:507–530. 27. Boettcher DH, Vatner SF, Heyndrickx GR, Braunwald E. Extent of utilization of the Frank-Starling mechanism in conscious dogs. Am J Physiol 1978;234:H338 –H345. 28. Vatner SF, Pagani M. Cardiovascular adjustments to exercise: hemodynamics and mechanisms. Prog Cardiovasc Dis 1976;19:91–108. 29. Dickerson LW, Rodak DJ, Fleming TJ, Gatti PJ, Massari VJ, McKenzie JC, Gillis RA. Parasympathetic neurons in the cranial medial ventricular fat pad on the dog heart selectively decrease ventricular contractility. J Auton Nerv Syst 1998;70:129 –141. 30. Massari VJ, Dickerson LW, Gray AL, Lauenstein JM, Blinder KJ, Newsome JT, Rodak DJ, Fleming TJ, Gatti PJ, Gillis RA. Neural control of left ventricular contractility in the dog heart: synaptic interactions of negative inotropic vagal preganglionic neurons in the
489
31.
32.
33.
34.
35.
nucleus ambiguus with tyrosine hydroxylase immunoreactive terminals. Brain Res 1998;802:205–220. Fee JD, Randall WC, Wurster RD, Ardell JL. Selective ganglionic blockade of vagal inputs to sinoatrial and/or atrioventricular regions. J Pharmacol Exp Ther 1987;242:1006 –1012. Miura T, Miyazaki S, Guth BD, Kambayashi M, Ross J Jr. Influence of the force-frequency relation on left ventricular function during exercise in conscious dogs. Circulation 1992;86:563–571. Freeman GL, Little WC, O’Rourke RA. Influence of heart rate on left ventricular performance in conscious dogs. Circ Res 1987;61: 455– 464. Miura T, Miyazaki S, Guth BD, Indolfi C, Ross J Jr. Heart rate and force-frequency effects on diastolic function of the left ventricle in exercising dogs. Circulation 1994;89:2361–2368. Little WC. The left ventricular dP/dtmax-end-diastolic volume relation in closed-chest dogs. Circ Res 1985;56:808 – 815.
Introduction The increased incidence of atrial fibrillation (AF) observed in the last decade has been described as a new epidemic.1 This has renewed the interest in assessment of left ventricular (LV) performance during AF and in interventions that may improve it.2,3 Efforts have been made to elucidate whether the seemingly random LV performance in AF is due to variable LV contractility2,4 or to beat-to-beat changes of preload (i.e., This study was supported in part by Grant NHLBI RO1 HL60833 from the National Institutes of Health. Address reprint requests and correspondence: Dr. Todor N. Mazgalev, Research Institute FF1-02, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195. E-mail address: [email protected]. (Received May 6, 2004; accepted June 29, 2004.)
Frank-Starling mechanism).5,6 Contractility in AF may vary through the restitution-potentiation mechanism, that is, decreased LV contractility is associated with a short-coupled beat (extrasystolic restitution), whereas a subsequently augmented response may result from so-called postextrasystolic potentiation.7 Quantitatively, these processes are linked to the durations of the RR intervals preceding (RRp) and pre-preceding (RRpp) the current beat and their ratio (RRp/RRpp).7 In contrast, the Frank-Starling mechanism utilizes an initial preload (i.e., end-diastolic volume [EDV]) to adjust the ventricular performance of the following beat, without changes of contractility. In AF, EDV values can fluctuate as uneven filling occurs due to varying diastolic time intervals.8 Recent reports imply that the pseudo-random behavior of LV function during AF may largely be attributed to uneven LV contractility caused by RRp/RRpp variability.7,9 –11 However,
1547-5271/$ -see front matter © 2004 Heart Rhythm Society. All rights reserved. doi:10.1016/j.hrthm.2004.06.016
Heart Rhythm (2004) 4, 482– 489
Popovic´ et al
Frank-Starling Mechanism and AF
483
whether that leaves room for an independent action of FrankStarling mechanism remains controversial. This is a difficult issue to resolve because larger EDVs are correlated with longer cycle length durations (i.e., filling times),8 thus masking the potential contribution of EDV (through the Frank-Starling mechanism) by the recognized role of contractility (through RRp/RRpp). Moreover, LV function-RRp/RRpp relationships are nonlinear, which precludes standard multiple regression analysis.11 For this reason, we conducted a series of hemodynamic experiments in a canine AF model during which we assessed LV volumes by conductance catheter. The primary working hypothesis of the study was that EDV, and thus the Frank-Starling mechanism, influences LV systolic performance in AF independent of cycle length irregularity (RRp/ RRpp). Our second aim was to determine whether the impact of EDV is affected by changes of the average ventricular rate during AF.
Methods The investigation was approved by the Institutional Animal Research Committee and conforms to the “Guide for the Care and Use of Laboratory Animals” published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996).
Experimental procedures We studied 16 healthy, open chest mongrel dogs (body weight 25–30 kg). In 11 of the 16 dogs, acute AF was induced by rapid atrial pacing during the experiments, as previously described.12 The remaining five dogs received high-rate atrial pacemakers, and AF was maintained for 2 months preceding the study. The study protocol was similar to one previously reported.3 Briefly, anesthesia was maintained with 1–2% isoflurane, while volume status, arterial blood gases, and body temperature were monitored. Standard ECG was obtained by subcutaneous needle electrodes. Quadripolar, custom-made, Ag-AgCl plate electrodes were sutured to the high right atrium and right ventricular apex for recording of local atrial and ventricular electrical activity and for pacing. We defined the number of beats and the duration of time intervals (RRp and RRpp) for an individual beat using the right ventricular electrogram (Figure 1). A flowmeter probe (16A/20A, HT 207, Transonic System Inc., Ithaca, NY, USA) was placed around the ascending aorta. A combined conductance-pressure sensor catheter (Millar) was inserted through the left carotid artery and advanced into the LV for pressure/volume measurements. Hemodynamic data were digitized at 1 kHz per channel for up to 1000 ventricular beats, and 400 beats were analyzed in each AF period. For volume calibration, the difference between LV end-diastolic and end-systolic volume points (Figure 1) was divided by the stroke volume obtained by integration of the aortic flow sig-
Figure 1 Typical tracings recorded in one dog during rapid AF (A) and slow AF (B). For illustration of the measuring routine, two representative beats (stars) are shown, along with the corresponding RRp and RRpp intervals, end-diastolic (d), and end-systolic (s) times. d ⫽ point at which d(LVP)/dt reached a threshold of 10% of its peak dP/dt; s ⫽ point of minimal LV volume signal. The nerve stimulation (used for rate slowing) was synchronized with the ventricular electrograms, and brief artifacts can be seen following each RV in panel B. Note that frequent beats (arrows) produce insufficient LV pressure, near-zero dP/dt, and no aortic flow. Such “abortive” beats were less frequent during slow AF. AoF ⫽ aortic flow; dP/dt ⫽ first time-derivative of the left ventricular pressure signal; LVP ⫽ left ventricular pressure; LVV ⫽ left ventricular volume; RV ⫽ right ventricular electrogram.
nal.13,14 The EDV (in milliliters) was determined as a ratio (stroke volume)/(ejection fraction). Ejection fraction was calculated with biplane Simpson equation from epicardial echocardiographic data.11,14 –16 All signals were amplified, displayed, and stored on a dedicated recording system (CardioLab, GE Marquette Medical Systems, Houston, TX). After recording data during sinus rhythm, AF was induced by high-rate atrial pacing and data collection continued after a 15-minute stabilization period. This part of the protocol is referred to as “rapid AF.” In addition, in six dogs selective AV nodal vagal nerve stimulation12 was used to achieve a slower average ventricular rate (“slow AF”), close to the normal sinus rate. To test directly the impact of preload on LV performance in AF, data also were collected after a rapid saline infusion (20 mL/kg) in three dogs.
484
Assessing the impact of Frank-Starling mechanism versus contractility in AF Assessment of EDV and RRp/RRpp as independent predictors of LV performance by standard multiple regression is difficult, because two assumptions are violated. First, LV performance is nonlinearly related to RRp/RRpp.10,11 Second, EDV and RRp/RRpp show strong collinearity due to tight coupling between RRp and LV filling.8 To overcome this, we used the following approach. First, we selected preload-dependent and preload-independent indices for concurrent evaluation. We sought to detect the impact of the Frank-Starling mechanism on the preload-dependent indices: maximal LV power (LVPower, a product of the peak aortic flow and peak LV pressure)1 and peak of LV pressure derivative (dP/dt). In contrast, the LV preload-adjusted maximal power (adj.LVPower ⫽ LVPower/EDV2), which correlates well with the theoretical contractility index Emax,9,17 was chosen as a preload-independent index. Second, we fitted each of the indices as a function of RRp/RRpp using the following equation11: LV Performance index ⫽ Amax ⫻ 关1 ⫺ e共RRp ⁄RRppmin⫺RRp ⁄RRpp兲⁄C兴, where Amax is a maximum response (plateau of the relationship), RRp/RRppmin is minimal RRp/RRpp at which an LV response could be observed, and C is the curvature of the relationship. We expected that not only the adj.LVPower (a contractility index) but also LVPower and dP/dt would exhibit strong correlation with RRp/RRpp, confirming the role of varying contractility during AF. Third, we mathematically eliminated the portion of an LV performance index that was predictable based on its relationship to RRp/RRpp. Thus, we obtained residuals (RES) of each of the relationships by calculating the difference between the observed index values and the values estimated by the nonlinear equations. We assumed that RES reflects the contribution of the preload (EDV) and established that these RES show no first-order dependence on RRp/RRpp. Finally, we calculated the correlation between each RES and EDV. A positive correlation for the RESLVPower and RES-dP/dt would support the independent contribution of the preload, i.e., the Frank-Starling mechanism during AF. To illustrate the theoretical (though unrealistic) extremes, the coefficient of determination r2 ⫽ 1 for the relationship LVPower-RRp/RRpp along with r2 ⫽ 0 for RES LVPower-EDV would indicate that LV performance variability in AF is due entirely to an uneven contractility. In contrast, r2 ⫽ 0 for the relationship LVPower-RRp/RRpp along with r2 ⫽ 1 for RES LVPower-EDV would indicate that LV performance variability is due entirely to changes in preload.
Heart Rhythm, Vol 1, No 4, October 2004
Statistics Data are represented as average ⫾ SD. By calculating the coefficients of determination r2 of individual datasets, we determined the proportion of variance in one variable (e.g., LVPower) that can be accounted for by knowing the second variable (e.g., RRp/RRpp).18 The significance of the nonlinear fit was assessed by computing standard F statistics.19 Standard linear regression methods were applied to obtain 2 r , the slope, and the significance of fit of RES-LVPower, RES-adj.LVPower, and RES-dP/dt to EDV in individual datasets. The correlations between these r2 and the average heart rate were assessed by Spearman’s nonparametric correlation coefficient . The paired data obtained during rapid AF and slow AF were compared by Wilcoxon matched pairs test. P ⬍ .05 was considered significant in all analyses.
Results We recorded and analyzed a total of 27 hemodynamic datasets (16 sets during rapid AF, 8 sets during slow AF, and 3 sets recorded after saline infusion), with the average ventricular cycle length ranging from 220 to 662 ms. During rapid AF, the average ventricular cycle length recorded in all 16 dogs was 292 ⫾ 66 ms (corresponding to 206 bpm). In the six dogs with additional vagal nerve stimulation, the average ventricular cycle length during rapid AF was 258 ⫾ 34 ms (233 bpm) but prolonged to 445 ⫾ 80 ms (135 bpm) during slow AF. Figure 1 illustrates typical raw data from one experiment. Note that during rapid AF (panel A), a relatively large number of beats (as identified by right ventricular electrogram) produced no aortic flow due to insufficient LV pressure. For these beats, both dP/dt and LVPower were exceedingly small (or even zero). The number of such “abortive” beats was greatly reduced during slow AF ( panel B).
Cycle length irregularity and LV systolic performance Figure 2 illustrates examples of the nonlinear relationships LVPower versus RRp/RRpp and dP/dt versus RRp/ RRpp during rapid (panels A and C) and slow AF (panels B and D), respectively. For all experiments, the average r2 was 0.67 ⫾ 0.12 for LVPower and 0.66 ⫾ 0.12 for dP/dt (P ⬍ .0001 for all correlations). Figure 3, panels A and B, illustrate examples of the relationship adj.LVPower versus RRp/ RRpp during rapid and slow AF, respectively. The average r2 of these relationships was 0.60 ⫾ 0.14 (P ⬍ .0001 for all correlations). The data indicate that about two thirds of LV performance variability during AF depended on RRp/RRpp, i.e., on the restitution/potentiation mechanisms that govern the contractility.
Popovic´ et al
Frank-Starling Mechanism and AF
Figure 2 Example of the relationships LVPower versus RRp/ RRpp during rapid AF (A) and slow AF (B) recorded in a single animal. C, D: Examples of relationships dP/dt versus RRp/RRpp during rapid AF (C) and slow AF (D) in the same animal. AF ⫽ atrial fibrillation; dP/dt ⫽ maximum of dP/dt; LVPower ⫽ maximal left ventricular power; RRp ⫽ preceding RR interval; RRpp ⫽ pre-preceding coupling interval. y-axis labels for panels B and D are identical to those for panels A and C, respectively.
Effect of EDV on LV performance indices
485
Figure 4 Scatterplot of RES-LVPower versus EDV during rapid AF (A) and slow AF (B). C, D: RES-dP/dt versus EDV during rapid AF (C) and slow AF (D). Data are from the experiment illustrated in Figure 1. RES ⫽ residuals. Other abbreviations as in previous figures. y-axis labels for panels B and D are identical to those for panels A and C, respectively.
experimental datasets, no significant positive correlation was detected. In the remaining 15 experimental datasets (one illustrated in Figure 5), there was weak correlation of RES-adj.LVPower versus EDV at rapid AF that was further attenuated during slow AF (Table 1). The data confirmed that, after removing the direct dependence on contractility, LV performance indices revealed modest but statistically significant relationship to EDV. Moreover, as expected from the working hypothesis, the Frank-Starling mechanism was evident in the case of LVPower and dP/dt (Figure 4), whereas adj.LVPower was not dependent on EDV (Figure 5).
Figure 4, panels A and B, present scatterplots and linear fits of the residuals RES-LVPower versus EDV obtained from Figure 2, panels A and B, for rapid and slow AF, respectively. Similarly, Figure 4, panels C and D, present RES-dP/dt versus EDV obtained from Figure 2, panels C and D, respectively. In all cases, the illustrated correlations were significant (P ⬍ .005 for all). The composite data for r2 presented in Table 1 indicate that 20 to 25% of the residual LV performance (after accounting for the variable contractility) was linked to the effects of the variable EDV. Figure 5, panels A and B, present scatterplots of RESadj.LVPower versus EDV obtained from Figure 3, panels A and B, for rapid and slow AF, respectively. In nine of 24
Impact of cycle length prolongation on FrankStarling mechanism and contractility
Figure 3 Example of the relationship adj.LVPower versus RRp/ RRpp during rapid AF (A) and slow AF (B), recorded in a single animal. adj.LVPower ⫽ preload-adjusted maximal left ventricular power. Other abbreviations as in previous figures.
Results from a representative dog illustrated in Figures 4 and 5 indicated that the studied residuals exhibited stronger correlations with EDV during rapid AF. The summarized data for average r2 of LV performance residuals in all 24 analyzed datasets during rapid and slow AF are presented in Table 1. In addition, the average heart rate during AF correlated with both r2 of RES-LVPower (Spearman’s ⫽ 0.42, P ⫽ .04) and r2 of RES-dP/dt (Spearman’s ⫽ 0.64, P ⬍ .001; Figure 6), that is, faster fibrillatory rates were associated with larger r2. Furthermore, in the six dogs in which the average ventricular rate was significantly slowed by vagal nerve stimulation, there was a decrease in the slope of RES-LVPower versus EDV. The results for each of these six dogs are shown in Figure 7A. The mean slopes for RES-LVPower were 0.055 ⫾ 0.027 W/mL during rapid AF and 0.026 ⫾
486 Table 1
Heart Rhythm, Vol 1, No 4, October 2004 Average correlation values (r2) for residuals of LV parameters v EDV
Datasets
RES-dP/dt
RES-LVPower
RES-adj.LVPower
Rapid AF (n ⫽ 16) Slow AF (n ⫽ 8) Total (n ⫽ 24)
0.28 ⫾ 0.16 0.11 ⫾ 0.12* 0.24 ⫾ 0.17
0.26 ⫾ 0.14 0.12 ⫾ 0.10* 0.20 ⫾ 0.14
0.13 ⫾ 0.12† 0.02 ⫾ 0.10*† 0.10 ⫾ 0.09†
AF ⫽ atrial fibrillation; EDV ⫽ end-diastolic volume; LV ⫽ left ventricular; RES-adj.LVPower ⫽ residuals of preload-adjusted maximum LV power; RES-dP/dt ⫽ residuals of maximum derivative of LV pressure, RES-LVPower-residuals ⫽ residuals of maximal LV power. *P ⬍ .05 vs Rapid AF. †P ⱕ .01 vs RES-LVPower.
0.014 W/mL during slow AF (P ⫽ .028). Similarly, slow AF decreased the mean slope of the relationship RES-dP/dt versus EDV (30.3 ⫾ 20.1 versus 10.3 ⫾ 730 mmHg/s/mL, P ⫽ .028; Figure 7B). A possible explanation for these findings is that slowing of the ventricular rate during AF enhanced contractility and therefore masked the relative contribution of the Frank-Starling mechanism to LV performance. To test this explanation, we compared the values of adj.LVPower (a preload-independent, contractility index) obtained before and after the slowing of the ventricular rate. We used a previously described technique that calculates the values of adj.LVPower at RRp/RRpp ⫽ 1 in the regression line.11,14,16 As seen in Figure 8, adj.LVPower increased during slow AF (0.09 ⫾ 0.04 versus 0.11 ⫾ 0.05 W/mL2 ⫻ 10⫺2, P ⫽ .028). This finding suggested that slowing of the average ventricular rate during AF increased the relative contribution of LV contractility as a major determinant of LV performance.
Impact of preload augmentation on LV performance In each of the three animals, saline infusion produced an increase of EDV (from 68 ⫾ 26 to 88 ⫾ 24 mL) and an increase of dP/dt (from 1,110 ⫾ 193 to 1,303 ⫾ 328 mmHg/s) and LVPower (from 1.7 ⫾ 0.3 to 2.4 ⫾ 0.9 W). This indicated that direct EDV augmentation accentuated the Frank-Starling mechanism in the presence of AF. Interestingly, a slight decrease of the average heart rate (from 159 ⫾ 9.6 to 148 ⫾ 4.2 bpm) also was observed.
Figure 5 Scatterplot of RES- adj.LVPower versus EDV during rapid AF (A) and slow AF (B). Data are from the experiment illustrated in Figure 2. Abbreviations as in previous figures.
Discussion Major findings We showed that contractility and the Frank-Starling mechanism were operative during AF, but their relative contributions were substantially different and further modified by the prevailing average ventricular rate. In particular, ventricular rate slowing accentuated the dominant role of contractility and minimized the independent role of FrankStarling mechanism in determining LV performance.
Previous studies The physiologic basis of the beat-to-beat variability in LV performance during AF has been a matter of continuous debate. Previous reports have shown that EDV, along with contractility, is an independent determinant of the ejection fraction during AF.6,20 Specifically, increased EDV values were associated with higher ejection fraction, although this effect diminished with the prolongation of the preceding cycle length. However, ejection fraction is afterload sensitive (through the end-systolic volume, i.e., 1 ⫺ ESV/ EDV)21 and, therefore, is not well suited for evaluation of the Frank-Starling mechanism.2 In fact, some authors argue that the Frank-Starling mechanism is of little importance during AF,4 whereas other authors conclude that both the Frank-Starling mechanism and the varying contractility probably contribute to LV functional performance.10 However, no quantification of the proposed independent roles has been provided.
Figure 6 Correlation of average ventricular rate with r2 of RESLVPowerversus EDV (A) and r2 of RES-dP/dtversus EDV (B). HR ⫽ heart rate; r2 ⫽ coefficient of determination. Other abbreviations as in previous figures.
Popovic´ et al
Frank-Starling Mechanism and AF
Figure 7 Impact of average ventricular rate on the slope of the relationships RES-LVPower versus EDV (A) and RES-dP/dt versus EDV (B) in each of six dogs. Abbreviations as in previous figures.
Control of LV function by Frank-Starling mechanism Even though more than a century has passed since the initial observations,22 the Frank-Starling mechanism is incompletely understood at molecular level. It is known that the larger EDV results in individual sarcomere stretch that increases crossbridge Ca2⫹ affinity.23 However, the exact “mechanical transducer” that links sarcomere stretch and Ca2⫹ affinity is unknown.24 On the other hand, the force-interval relationships underlying contractility depend on sarcolemmal channels that regulate Ca2⫹ release and uptake.25,26 This implies that the two mechanisms may independently affect LV function, the former by increasing Ca2⫹ affinity, the latter by changing Ca2⫹ concentration through sarcolemmal channel activity. It should be pointed out that the impact of Frank-Starling mechanism on LV function in normal conscious subjects is attenuated, because increased preload activates reflex increase of the heart rate, which in turn blunts EDV augmentation.27,28 In all three animals in which volume loading was performed, however, both dP/dt and LVPower increased during the intervention, indicating accentuated recruitment of the Frank-Starling mechanism during ongoing AF. It is conceivable that increased variability of EDV due to variable LV filling8 is a prerequisite for manifestation of the Frank-Starling mechanism during AF.
487 inant mechanism linking the ratio RRp/RRpp and LV performance variability during AF. Importantly, when we mathematically subtracted the effect of contractility and examined the residuals, the experimental data confirmed that the Frank-Starling mechanism had an impact on LV performance that was mediated by the preload (EDV) and was additive to the direct effects of RRp/RRpp variability. It should be stressed, however, that EDV during AF might depend not only on the filling times but also on factors such as respiration and preceding relaxation.
Role of ventricular rate slowing during AF The dependence of LV performance on EDV was decreased by ventricular rate slowing during AF. At the same time, contractility slightly but significantly increased, implying more prompt restitution.25 This is in line with our previous experimental observations demonstrating improvement of LV performance during vagally induced rate slowing.3 These effects are mostly cycle length dependent, because selective stimulation of vagal postganglionic neurons to the AV nodal region has no or minimal effect on cardiac inotropism.29 –31 It should be noted that these important findings apply to the range of high ventricular rates typically associated with AF. Excessive rate slowing (below the normal sinus rate) has an opposite effect: it produces negative inotropic effects through the force-frequency relationships.32-34
Clinical implications Both others and we have previously shown that representative values of systolic performance parameters during
Contractility versus the Frank-Starling mechanism as determinants of irregular LV performance during AF We have confirmed that LV performance in AF strongly depends on ventricular rate irregularity. This can be explained by varying contractility,7 preload variability, the Frank-Starling mechanism,8,10 or to their combination. To discern the important of contractility variability, we attenuated the effects of preload on LV performance by calculating adj.LVPower. The results indicated that the relationship between adj.LVPower and RRp/RRpp was strong (Figure 3), confirming that varying contractility is the dom-
Figure 8 Impact of average ventricular rate on adj.LVPower calculated for RRp/RRpp ⫽ 1 in the regression line. Abbreviations as in previous figures.
488 AF can be obtained from a relatively small number of cardiac cycles if correction for the ratio RRp/RRpp is performed.11,14,16 Our new findings suggest that adjustments taking into consideration both RRp/RRpp and EDV would be superior during rapid AF because they would account for both contractility and the Frank-Starling mechanism.
Study limitations We stress that the design of the present in situ study precludes direct quantitative elucidation of the separate contributions of restitution/potentiation (contractility) and the Frank-Starling mechanism (preload) to ventricular performance during AF. This would require an isolated heart model with independent control of contractility (e.g., via pacing with programmed RRp and RRpp intervals) and preload (e.g., via isovolumic EDV manipulations). However, the relevance of such a model to AF observed in the intact heart requires separate validations. This was an experimental, acute, open chest animal study. Efforts were made to use an AF model with minimal additional perturbations, but we could not avoid some obvious limitations. Along with the use of anesthesia, the animals were not decentralized, leaving open the possibility for some reflex modulation of contractility secondary to alterations of the ventricular rate. Also, we did not evaluate potential effects of afterload on LV performance in AF20; however both dP/dt and LVPower are preload-sensitive, but not afterload-sensitive, parameters.17,35 Our findings are not necessarily applicable to patients with chronic AF with associated cardiovascular pathologies. Previous studies imply that the impact of both ventricular rate irregularity and the Frank-Starling mechanism may be observed in AF patients,2,6 but further clinical evaluations are needed.
Heart Rhythm, Vol 1, No 4, October 2004
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
References 1. Braunwald E. Shattuck lecture: cardiovascular medicine at the turn of the millennium: triumphs, concerns, and opportunities. N Engl J Med 1997;337:1360 –1369. 2. Noble MI. Beat to beat left ventricular performance in spontaneous atrial fibrillation does not depend on afterload. Heart 2000;84:89. 3. Zhang Y, Mowrey KA, Zhuang S, Wallick DW, Popovic ZB, Mazgalev TN. Optimal ventricular rate slowing during atrial fibrillation by feedback AV nodal-selective vagal stimulation. Am J Physiol Heart Circ Physiol 2002;282:H1102–1110. 4. Meijler FL, Strackee J, Van Capelle JL, Du Perron JC. Computer analysis of the RR interval-contractility relationship during random stimulation of the isolated heart. Circ Res 1968;22:695–702. 5. Braunwald E, Frye RL, Aygen MM. Studies on Starling’s law of the heart. III. Observations in patients with mitral stenosis and atrial fibrillation on the relationships between left ventricular end-diastolic segment length, filling pressure, and the characteristics of ventricular contraction. J Clin Invest 1960;39:1874 –1884. 6. Gosselink AT, Blanksma PK, Crijns HJ, Van Gelder IC, de Kam PJ, Hillege HL, Niemeijer MG, Lie KI, Meijler FL. Left ventricular beat-to-beat performance in atrial fibrillation: contribution of Frank-
18. 19. 20.
21.
22. 23. 24.
25.
Starling mechanism after short rather than long RR intervals. J Am Coll Cardiol 1995;26:1516 –1521. Suzuki S, Araki J, Morita T, Mohri S, Mikane T, Yamaguchi H, Sano S, Ohe T, Hirakawa M, Suga H. Ventricular contractility in atrial fibrillation is predictable by mechanical restitution and potentiation. Am J Physiol 1998;275:H1513–1519. Cieslinski A, Hui WK, Oldershaw PJ, Gregoratos G, Gibson D. Interaction between systolic and diastolic time intervals in atrial fibrillation. Br Heart J 1984;51:431– 437. Takagaki M, McCarthy PM, Chung M, Connor J, Dessoffy R, Ochiai Y, Howard M, Doi K, Kopcak M, Mazgalev TN, Fukamachi K. Preload-adjusted maximal power: a novel index of left ventricular contractility in atrial fibrillation. Heart 2002;88:170 –176. Brookes CI, White PA, Staples M, Oldershaw PJ, Redington AN, Collins PD, Noble MI. Myocardial contractility is not constant during spontaneous atrial fibrillation in patients. Circulation 1998;98:1762– 1768. Popovic ZB, Mowrey KA, Zhang Y, Zhuang S, Tabata T, Wallick DW, Grimm RA, Thomas JD, Mazgalev TN. Slow rate during AF improves ventricular performance by reducing sensitivity to cycle length irregularity. Am J Physiol Heart Circ Physiol 2002;283:H2706 – 2713. Wallick DW, Zhang Y, Tabata T, Zhuang S, Mowrey KA, Watanabe J, Greenberg NL, Grimm RA, Mazgalev TN. Selective AV nodal vagal stimulation improves hemodynamics during acute atrial fibrillation in dogs. Am J Physiol Heart Circ Physiol 2001;281:H1490 –1497. Baan J, van der Velde ET, de Bruin HG, Smeenk GJ, Koops J, van Dijk AD, Temmerman D, Senden J, Buis B. Continuous measurement of left ventricular volume in animals and humans by conductance catheter. Circulation 1984;70:812– 823. Tabata T, Grimm RA, Greenberg NL, Agler DA, Mowrey KA, Wallick DW, Zhang Y, Zhuang S, Mazgalev TN, Thomas JD. Assessment of LV systolic function in atrial fibrillation using an index of preceding cardiac cycles. Am J Physiol Heart Circ Physiol 2001;281:H573–580. Ratcliffe MB, Wallace AW, Salahieh A, Hong J, Ruch S, Hall TS. Ventricular volume, chamber stiffness, and function after anteroapical aneurysm plication in the sheep. J Thorac Cardiovasc Surg 2000;119: 115–124. Sumida T, Tanabe K, Yagi T, Kawai J, Konda T, Fujii Y, Okada M, Yamaguchi K, Tani T, Morioka S. Single-beat determination of Doppler-derived aortic flow measurement in patients with atrial fibrillation. J Am Soc Echocardiogr 2003;16:712–715. Kass DA, Beyar R. Evaluation of contractile state by maximal ventricular power divided by the square of end-diastolic volume. Circulation 1991;84:1698 –1708. Dawson-Saunders B, Trapp RG. Basic and Clinical Biostatistics. East Norwalk, CT: Appleton & Lange, 1990. Glantz SA, Slinker BK. Analysis of Variance and Multiple Regression. New York: MCGraw-Hill, 1990:491– 493. Muntinga HJ, Gosselink AT, Blanksma PK, De Kam PJ, Van Der Wall EE, Crijns HJ. Left ventricular beat to beat performance in atrial fibrillation: dependence on contractility, preload, and afterload. Heart 1999;82:575–580. Kass DA, Maughan WL, Guo ZM, Kono A, Sunagawa K, Sagawa K. Comparative influence of load versus inotropic states on indexes of ventricular contractility: experimental and theoretical analysis based on pressure-volume relationships. Circulation 1987;76:1422–1436. Katz AM. Ernest Henry Starling, his predecessors, and the “Law of the Heart.” Circulation 2002;106:2986 –2992. Fuchs F, Smith SH. Calcium, cross-bridges, and the Frank-Starling relationship. News Physiol Sci 2001;16:5–10. Konhilas JP, Irving TC, de Tombe PP. Myofilament calcium sensitivity in skinned rat cardiac trabeculae: role of interfilament spacing. Circ Res 2002;90:59 – 65. Prabhu SD, Freeman GL. Postextrasystolic mechanical restitution in closed-chest dogs. Effect of heart failure. Circulation 1995;92:2652– 2659.
Popovic´ et al
Frank-Starling Mechanism and AF
26. Wier WG, Yue DT. Intracellular calcium transients underlying the short-term force-interval relationship in ferret ventricular myocardium. J Physiol 1986;376:507–530. 27. Boettcher DH, Vatner SF, Heyndrickx GR, Braunwald E. Extent of utilization of the Frank-Starling mechanism in conscious dogs. Am J Physiol 1978;234:H338 –H345. 28. Vatner SF, Pagani M. Cardiovascular adjustments to exercise: hemodynamics and mechanisms. Prog Cardiovasc Dis 1976;19:91–108. 29. Dickerson LW, Rodak DJ, Fleming TJ, Gatti PJ, Massari VJ, McKenzie JC, Gillis RA. Parasympathetic neurons in the cranial medial ventricular fat pad on the dog heart selectively decrease ventricular contractility. J Auton Nerv Syst 1998;70:129 –141. 30. Massari VJ, Dickerson LW, Gray AL, Lauenstein JM, Blinder KJ, Newsome JT, Rodak DJ, Fleming TJ, Gatti PJ, Gillis RA. Neural control of left ventricular contractility in the dog heart: synaptic interactions of negative inotropic vagal preganglionic neurons in the
489
31.
32.
33.
34.
35.
nucleus ambiguus with tyrosine hydroxylase immunoreactive terminals. Brain Res 1998;802:205–220. Fee JD, Randall WC, Wurster RD, Ardell JL. Selective ganglionic blockade of vagal inputs to sinoatrial and/or atrioventricular regions. J Pharmacol Exp Ther 1987;242:1006 –1012. Miura T, Miyazaki S, Guth BD, Kambayashi M, Ross J Jr. Influence of the force-frequency relation on left ventricular function during exercise in conscious dogs. Circulation 1992;86:563–571. Freeman GL, Little WC, O’Rourke RA. Influence of heart rate on left ventricular performance in conscious dogs. Circ Res 1987;61: 455– 464. Miura T, Miyazaki S, Guth BD, Indolfi C, Ross J Jr. Heart rate and force-frequency effects on diastolic function of the left ventricle in exercising dogs. Circulation 1994;89:2361–2368. Little WC. The left ventricular dP/dtmax-end-diastolic volume relation in closed-chest dogs. Circ Res 1985;56:808 – 815.