4. Stambler BS, Wood MA, Ellenogen KA. Comparative efficacy of intravenousibutilide versus procainamide for enhancing termination of atrial flutter by atrialoverdrive pacing.Am J Cardiol1996;77:960–966.5. Mattioni T, Denker S, Riggio D, Smith J, Welch S, Perry K, Wakefield L,Vanderlugt J. Efficacy and safety of intravenous ibutilide in the treatment of atrialflutter and atrial fibrillation following valvular or coronary artery bypass surgery(abstr).PACE1997;20:1059.6. Fozzard H, Sheets M. Cellular mechanism of action of cardiac glycosides.J Am Coll Cardiol1985;5(suppl A):10A–15A.

7. Rosen MR. Cellular electrophysiology of digitalis toxicity.J Am Coll Cardiol1985;5(suppl A):22A–34A.8. Lee KS. Ibutilide, a new compound with potent class III antiarrhythmicactivity, activates a slow inward Na1 current in guinea pig ventricular cells.J Pharmacol Exp Ther1992;262:99–108.9. Yang T, Snyders DJ, Roden DM. Ibutilide, a methanesulfonanilide antiar-rhythmic, is a potent blocker of the rapidly activating delayed rectifier K1 current(IKr) in AT-1 cells. Concentration-, time-, voltage-, and use-dependent effects.Circulation 1995;91:1799–1806.

Effect of Rhythm Regularization on Left VentricularContractility in Patients With Atrial Fibrillation

Mohamed Effat, MD, Edgar C. Schick, MD, David T. Martin, MD, andWilliam H. Gaasch, MD

C linical and experimental data support the notionthat atrial fibrillation (AF) with persistent tachy-

cardia can have a negative impact on left ventricular(LV) function, and that rate control may reversetachycardia-mediated ventricular dysfunction.1–3 Theindependent effect of rhythm regularization also ap-pears to have a beneficial hemodynamic effect, suchthat, at equal heart rates, cardiac output appears to behigher when the ventricular rhythm is regular.3–5

However, whether an irregular rhythm, per se, has anegative effect on LV intrinsic contractility is un-known. Accordingly, we designed the present study toassess contractility in patients with AF before andimmediately after direct-current cardioversion. Byperforming the studies immediately after cardiover-sion, we were able to obtain data before the return ofan effective atrial contraction, thereby excluding thebeneficial hemodynamic effect of atrial systole. Inaddition, comparisons (before and after cardioversion)were made at equal preceding RR intervals, so as tolimit differences in LV preload. Thus, we tested thehypothesis that rhythm regularization is associatedwith a higher LV contractile state than is seen whenrhythm is irregular.

• • •Ten patients undergoing elective direct-current car-

dioversion for AF were studied. All patients wereclinically stable, having had AF for approximately 6to 8 weeks. All had undergone anticoagulation withwarfarin for a duration exceeding 3 weeks, and allwere being treated with digitalis and/or ab-adrenergicreceptor blocking agent. Patients with valvular heartdisease were excluded. Echocardiography was per-formed using previously described equipment andmethodology.6 In this group of 10 patients, mean LVend-diastolic diameter was 506 6 mm, mean ejectionfraction was 566 12%, and mean left atrial dimen-

sion was 476 5 mm. Direct-current cardioversion,using anteroposterior paddles, was performed afterinduction of general anesthesia with intravenousmethohexital sodium. In 6 patients, sinus rhythm wasachieved with a single shock (100 J); in the remaining4 patients a second shock (200 J) was required. Be-cause the purpose of the study was to assess theventricle independent of atrial contraction, we in-cluded only patients with an ineffective atrial contrac-tion after cardioversion.7,8 Postcardioversion datawere obtained approximately 5 to 10 minutes aftercardioversion; at this time the A wave (Doppler echo-cardiography) was,0.3 m/s in all 10 patients. Peakaortic ejection velocity (Vpe) was obtained before andimmediately after cardioversion by continuous-waveDoppler interrogation using a nonimaging transducerfrom the apical window. In each patient, 80 to 100consecutive beats were recorded for measurement andanalysis.

Measurements of Vpe and RR intervals were ob-tained directly from strip-chart recordings, and theanalysis was similar to that previously described.8 Vpewas normalized by expressing the velocity as apercentage of the maximum observed value in the80 to 100 recorded beats. We then plotted normalizedVpe against the preceding RR interval for intervals.500 ms. The velocity-interval coordinates from eachpatient were then divided into quartiles according tothe prepreceding RR intervals. Thus, quartiles with theshortest and longest prepreceding intervals were de-fined. This method provides approximately 20 coor-dinates with short and another 20 coordinates withlong prepreceding intervals. Velocity interval plotswere examined and the slope (linear best fit) of therelation between Vpe and the preceding interval wasdetermined for each of these 2 quartiles. Such plotsprovide a reference for comparison of multiple veloc-ity-interval coordinates before and a single velocity-interval coordinate after cardioversion (Figure 1). Val-ues for Vpe were then interpolated from the 2 slopevalues at a preceding RR interval that was equal tothe postcardioversion RR interval. This allows an as-sessment of the postcardioversion Vpe relative to thehighest and lowest velocity quartiles that were presentduring AF. Differences were assessed with 1-way

From the Department of Medicine (Cardiovascular Section), LaheyClinic, Burlington, Massachusetts. Dr. Gaasch’s address is: LaheyClinic Medical Center, 41 Mall Road, Burlington, Massachusetts01805. Manuscript received July 5, 1999; revised manuscript re-ceived August 9, 1999, and accepted August 10, 1999.

114 ©2000 by Excerpta Medica, Inc. All rights reserved. 0002-9149/00/$–see front matterThe American Journal of Cardiology Vol. 85 January 1, 2000 PII S0002-9149(99)00620-7

analysis of variance and the Bonferroni test. Data arepresented as mean6 SD.

Examples of the results from 3 representative pa-tients are shown in Figure 1; in these examples, thevelocity-interval plots demonstrate the range seen dur-ing AF and after cardioversion. Figure 2 shows theaverage interpolated Vpevalues at preceding intervals of500, 750, and 1,000 ms and the regression lines for thehigh- and low-velocity quartiles; the average postcar-dioversion coordinate is also shown. Vpe datawith short prepreceding RR intervals consistently ex-ceeded those with long prepreceding RR intervals (asseen in these figures). This result, demonstrating thepositive inotropic effect of a short prepreceding inter-val, is similar to that described previously.8

After cardioversion, the mean cycle length was850 6 172 ms. Values for Vpe at a preceding intervalof 850 ms were interpolated (for both the long andshort quartiles during AF) and compared with the Vpe

after cardioversion (Table I). The average Vpe after car-dioversion was not significantly different from theVpe derived from the long quartile data; it was signifi-cantly less than the Vpe derived from the short quartiledata (p,0.01).

• • •The primary finding in this study is that regular-

ization of ventricular rhythm (i.e., cardioversion ofAF) does not augment LV contractility. In fact, theaverage ejection velocity after cardioversion was sig-nificantly lower than that derived from the upper quar-tile data, and it was not significantly different from thelower quartile data. This might suggest a tendencytoward a lower contractile state after cardioversion.However, it should be recognized that 2 patients ex-hibited marked bradycardia with relatively low Vpe

after cardioversion, and that this likely contributed to

the low average velocity-interval coordinate. If theanalysis is limited to patients with a heart rate.60beats/min (i.e., by excluding patients 3 and 4), theaverage postcardioversion Vpe is 74% at an interval of778 ms; this coordinate is midway between the upperand lower quartiles. It appears, therefore, that cardio-version does not augment Vpe (when the comparisonis made at common RR intervals).

If there is no beneficial effect of rhythm regular-ization on intrinsic contractility, how can an irregularrhythm contribute to depressed LV pump performance(i.e., depressed cardiac output)? A likely explanationinvolves the relative contributions of very short andvery long cycle lengths. When the RR interval is very

FIGURE 1. Velocity-interval coordinates before and after cardioversion (post CV) of atrial fibrillation. Coordinates from the highest(short RR, triangles) and lowest (long RR, circles) Vpe quartiles are shown, and the best-fit lines for each quartile are plotted. Aftercardioversion the single velocity-interval coordinate ranged from high (left panel) to low (right panel) to intermediate (middle panel).These 3 examples illustrate the ventricular response to cardioversion in 3 representative patients.

FIGURE 2. Average velocity-interval data from 10 patients. Thisanalysis shows the average regression lines for the high and lowquartile data (during atrial fibrillation); the average interpolated Vpevalues at preceding intervals of 500, 750, and 1,000 ms are alsoshown. The postcardioversion (post CV) coordinate lies within therange defined by highest and lowest velocity-interval lines obtainedduring atrial fibrillation.

BRIEF REPORTS 115

short (i.e.,,500 ms), mechanical restitution is incom-plete and LV filling is limited.9,10 As a result, strokevolume falls dramatically in beats, with a very shortpreceding cycle length. If compensatory mechanismswere not available, cardiac output would fall in directrelation to the frequency of such early beats. Thus, if15% of the beats are very early, cardiac output wouldbe expected to fall by 15%. However, several com-pensatory mechanisms attenuate such an effect. First,the mechanism of postextrasystolic potentiation(thought to be responsible for interval-dependentchanges in contractility in AF) contributes to an in-crement in stroke volume.8,11,12 Second, a long dia-stolic interval following a beat with a short cyclelength is accompanied by a time-dependent decreasein aortic diastolic pressure that could effectively un-load the subsequent beat and contribute to a compen-satory increment in stroke volume. Third, increasedLV filling in the pause following an early beat couldcontribute to a compensatory increment in stroke vol-ume; this effect is probably the least important. In-deed, lengthening the RR interval from 700 to$1,000ms does not appreciably augment filling.9 Although it islikely that the first of these 3 mechanisms is the mostimportant, their relative importance remains speculative.In any case, these mechanisms apparently fail to providea complete compensation for very short cycle lengths, sothat cardiac output decreases by.10% in most patientswith AF.3,5Clark et al5 found a 10% difference in cardiacoutput when irregular and regular rhythms were com-pared, and they suggested that the short cycle lengthscontributed relatively more to a decrement in output thanthe compensatory effects of long cycle lengths contrib-uted to an increment. Therefore, regularization of rhythmor reduction in the variability of RR intervals may aug-ment cardiac output by limiting the number of shortcycle lengths. Reduction in the variability of RR inter-vals using rate-smoothing algorithms and ventricularpacing could provide improved hemodynamics over thatseen in AF, with wide variation in RR intervals.13

In summary, function-interval (preceding RR)coordinates after cardioversion were found to bewell within the range defined by the AF coordi-nates. We conclude that regularization of ventric-ular rhythm by direct-current cardioversion doesnot augment contractility above that seen duringAF.

1. Phillips E, Levine SA. Auricular fibrillation without other evidence of heartdisease: a cause of reversible heart failure.Am J Med1949;7:478–489.2. Grogan M, Smith HC, Gersh BJ, Wood DL. Left ventricular dysfunction dueto atrial fibrillation in patients initially believed to have idiopathic dilated car-diomyopathy.Am J Cardiol1992;69:1570–1573.3. Schumacher B, Luderitz B. Rate issues in atrial fibrillation: consequences oftachycardia and rate control.Am J Cardiol1998;82:29N–36N.4. Daoud EG, Weiss R, Bahu M, Knight BP, Bogun F, Goyal R, Harvey M,Strickberger SA, Mann KC, Morady F. Effect of an irregular rhythm on cardiacoutput.Am J Cardiol1996;78:1433–1436.5. Clark DM, Plumb VJ, Epstein AE, Kay GN. Hemodynamic effects of anirregular sequence of ventricular cycle lengths during atrial fibrillation.J Am CollCardiol 1997;30:1039–1045.6. Schneider F, Martin DT, Schick EC, Gaasch WH. Interval-dependent changesin left ventricular contractile state in lone atrial fibrillation and in atrial fibrillationassociated with coronary artery disease.Am J Cardiol1997;80:586–590.7. Manning WJ, Silverman DI, Katz SE, Riley, MF, Come PC, Doherty RM,Munson JT, Douglas PS. Impaired let atrial mechanical function after cardiover-sion: relation to the duration of atrial fibrillationJ Am Coll Cardiol1994;23:1535–1540.8. Harjai KJ, Mobarek SK, Cheirif J, Boulos LM, Murgo JP, Abi-Samra F.Clinical variables affecting recovery of left atrial mechanical function aftercardioversion from atrial fibrillation.J Am Coll Cardiol1997;30:481–486.9. Freeman GL, Colston JT. Evaluation of left ventricular mechanical restitutionin closed-chest dogs based on single beat elastance.Circ Res1990;67:1437–1445.10. Hardman SMC, Noble MIM, Biggs T, Seed WA. Evidence for an influenceof mechanical restitution on beat-to-beat variations in hemodynamics duringspontaneous atrial fibrillation in patients.Cardiovasc Res1998;38:82–90.11. Gosselink AT, Blanksma PK, Crijns HJ, Van Gelder IC, de Kam PJ, HillegeHL, Niemeijer MG, Lie KI, Meijler FL. Left ventricular beat-to-beat performancein atrial fibrillation: contribution of Frank-Starling mechanism after short ratherthan long RR intervals.J Am Coll Cardiol1995;26(6):1516–1521.12. Brookes CIO, White PA, Staples M, Oldershaw PJ, Redington AN, CollinsPD, Noble MIM. Myocardial contractility is not constant during spontaneousatrial fibrillation in patients.Circulation 1998;98:1762–1768.13. Duckers HJ, van Hemel NM, Kelder JC, Bakema H, Yee R. Effective use ofa novel rate-smoothing algorithm in atrial fibrillation by ventricular pacing.EurHeart J 1997;18:1951–1955.

TABLE I Velocity–Interval Data Before and After Cardioversion

Atrial Fibrillation

After Cardioversion

BP (mm Hg)

Vpe(%) @ RR1 5 850 ms

BP (mm Hg) RR (ms) Vpe (%)Long RR2 Short RR2

1 143/66 62 71 154/76 660 662 130/80 69 80 141/84 960 673 100/60 69 87 105/55 1150 704 130/80 79 90 130/75 1130 755 118/70 82 90 132/75 700 776 160/80 78 88 158/80 760 807 120/70 59 74 120/70 780 668 143/84 74 98 135/80 830 759 130/75 74 83 142/80 770 77

10 120/80 75 86 117/75 760 84

Average 6 SD 129/75 6 17/8 72 6 7 85 6 8 133/75 6 16/8 850 6 172 74* 6 6

*p ,0.01 versus Vpe at short RR2.BP 5 blood pressure; RR1 5 preceding RR interval or cycle length; RR2 5 prepreceding RR interval; Vpe 5 peak ejection velocity. During atrial fibrillation the Vpe

values were interpolated from the best-fit velocity-interval line at a preceding interval of 850 ms; after atrial fibrillation the Vpe value was measured.

116 THE AMERICAN JOURNAL OF CARDIOLOGYT VOL. 85 JANUARY 1, 2000