K

C

Ma

Investigation of a novel algorithm for synchronized left-ventricular pacing and ambulatory optimization of cardiacresynchronization therapy: Results of the adaptive CRT trialDavid O. Martin, MD, MPH,* Bernd Lemke, MD,† David Birnie, MD, MB, ChB,‡

Henry Krum, MBBS, PhD,§ Kathy Lai-Fun Lee, MD,� Kazutaka Aonuma, MD, PhD,¶

Maurizio Gasparini, MD,# Randall C. Starling, MD, MPH,* Goran Milasinovic, MD,**Tyson Rogers, MS,†† Alex Sambelashvili, PhD,†† John Gorcsan III, MD,§§

Mahmoud Houmsse, MD, FHRS,‡‡ Adaptive CRT Study Investigators

From the *Department of Cardiovascular Medicine, Cleveland Clinic, Cleveland, Ohio, †Department of Cardiolgy,linikum Lüdenscheid, Lüdenscheid, Germany, ‡Divison of Cardiology, University of Ottawa Heart Institute, Ottawa,

Ontario, Canada, §Centre of Cardiovascular Research & Education in Therapeutics, Monash University, Melbourne,Australia, �Cardiology Division, Queen Mary Hospital, University of Hong Kong, Hong Kong, ¶Graduate School of

omprehensive Human Sciences, University of Tsukuba, Tsukuba, Japan; #IRCCS Istituto Clinico Humanitas, Rozzano,Italy, **Referral Pacemaker Center, Clinical Center of Serbia, Belgrade, Serbia, ††Cardiac Rhythm Disease Management, Medtronic,

ounds View, Minnesota, §§Cardiology Division, Presbyterian University Hospital, University of Pittsburgh, Pittsburgh, Pennsylvania,nd ‡‡Division of Cardiovascular Medicine, Ohio State University Medical Center, Columbus, Ohio.

d0rawhew

CB

Kt

A

BACKGROUND In patients with sinus rhythm and normal atrioventric-ular conduction, pacing only the left ventricle with appropriate atrioven-tricular delays can result in superior left ventricular and right ventricularfunction compared with standard biventricular (BiV) pacing.

OBJECTIVE To evaluate a novel adaptive cardiac resynchroniza-tion therapy ((aCRT) algorithm for CRT pacing that provides au-tomatic ambulatory selection between synchronized left ventric-ular or BiV pacing with dynamic optimization of atrioventricularand interventricular delays.

METHODS Patients (n � 522) indicated for a CRT-defibrillator wererandomized to aCRT vs echo-optimized BiV pacing (Echo) in a 2:1ratio and followed at 1-, 3-, and 6-month postrandomization.

RESULTS The study met all 3 noninferiority primary objectives:(1) the percentage of aCRT patients who improved in their clinicalcomposite score at 6 months was at least as high in the aCRT armas in the Echo arm (73.6% vs 72.5%, with a noninferiority marginof 12%; P � .0007); (2) aCRT and echo-optimized settings re-sulted in similar cardiac performance, as demonstrated by a highconcordance correlation coefficient between aortic velocity time

boards for Medtronic and Boston Scientific. Dr Starling has received

1547-5271/$ -see front matter © 2012 Heart Rhythm Society. All rights reserved

ance correlation coefficient � 0.93; 95% confidence interval.91–0.94) and at 6-month postrandomization (concordance cor-elation coefficient � 0.90; 95% confidence interval 0.87–0.92);nd (3) aCRT did not result in inappropriate device settings. Thereere no significant differences between the arms with respect toeart failure events or ventricular arrhythmia episodes. Secondarynd points showed similar benefit, and right-ventricular pacingas reduced by 44% in the aCRT arm.

ONCLUSIONS The aCRT algorithm is safe and at least as effective asiV pacing with comprehensive echocardiographic optimization.

EYWORDS Cardiac resynchronization therapy; Fusion pacing; Op-imization; LV pacing; Heart failure

BBREVIATIONS aCRT � adaptive CRT; AoVTI � aortic velocitytime integral; AV � atrioventricular; BiV � biventricular;CCS � clinical composite score; CRT � cardiac resynchronizationtherapy; HF � heart failure; LV � left ventricular; RV � right ventric-ular; VT/VF � ventricular tachycardia/ventricular fibrillation

(Heart Rhythm 2012;9:1807–1814) © 2012 Heart Rhythm Society. All

integrals at aCRT and Echo settings at randomization (concor- rights reserved.

honoraria from Novartis. Dr Milasinovic has received honoraria fromMedtronic. Dr Gorcsan has consulted for or has received research grantsfrom Biotronik, Medtronic, St Jude Medical, GE, and Toshiba Medical. T.Rogers is a statistician employed by Medtronic. A. Sambelashvili is ascientist employed by Medtronic. Address for reprint requests andcorrespondence: Dr David O. Martin, MD, MPH, The Cleveland ClinicFoundation, 9500 Euclid Avenue, J2-2, Cleveland, OH 44195. E-mail

The trial was sponsored by Medtronic, Mounds View, Minnesota. DrMartin serves on a Medtronic advisory board. Dr Lemke has receivedhonoraria and speaker’s fees from Medtronic and Saint Jude Medical andspeaker’s fees from Boston Scientific. Dr Birnie has received honoraria andresearch grants from Medtronic. Dr Krum has received honoraria fromMedtronic. Dr Lee has received research grants from Medtronic. DrAonuma has received honoraria, speaker’s fees, and research grants fromMedtronic. Dr Gasparini has received honoraria and served on advisory

address: martind3@ccf.org.

. http://dx.doi.org/10.1016/j.hrthm.2012.07.009

c

p

vp

RmnapvovAappp

dawltgtpstewcpoectc

wP

1808 Heart Rhythm, Vol 9, No 11, November 2012

IntroductionCardiac resynchronization therapy (CRT) is an establishedtherapy for patients with heart failure (HF) symptoms, leftventricular systolic dysfunction, and a wide QRS.1,2 How-ever, the magnitude of clinical and hemodynamic benefit ofCRT varies significantly among its recipients with no clin-ical improvement in approximately one third.1 Patient-spe-cific characteristics, such as severity and type of electricalconduction abnormalities, dyssynchrony, and scar burden,have been associated with the degree of CRT benefit.3,4

While CRT is most commonly achieved by using biven-tricular (BiV) pacing, multiple acute5 and randomizedhronic6 studies have demonstrated that left-ventricular

(LV) pacing can be at least as efficacious as BiV pacing. Inpatients with sinus rhythm and normal atrioventricular (AV)conduction, pacing only the left ventricle with appropriateAV intervals can result in even superior LV5,7 and right-ventricular (RV)8,9 function compared with standard BiVacing.

Optimization of the AV and interventricular (VV) inter-als during BiV pacing is another option to maximize theositive effects of CRT.10,11 Optimization is usually accom-

plished by using echocardiography or other modalities.However, these methods can be resource-intensive, andonly a minority of clinicians routinely optimize AV and VVdelays.

Based on published research,7,8,12,13 a novel adaptiveCRT (aCRT) algorithm14 has been developed to provide

V-synchronized LV pacing when AV conduction is nor-al, or BiV pacing otherwise. AV conduction is considered

ormal when the Asense to RVsense interval is �200 ms. Thelgorithm also adjusts AV and VV delays on the basis oferiodic automatic evaluation of intrinsic conduction inter-als. The algorithm is intended to provide ambulatory CRTptimization and allow more physiologic ventricular acti-ation and greater device longevity in patients with normalV conduction by reducing unnecessary RV pacing. The

im of the aCRT trial was to determine whether the incor-oration of this algorithm into the management of CRTatients is safe and efficacious when compared with BiVacing with echocardiographic AV and VV optimization.

MethodsStudy designDetails of the algorithm and the aCRT study design havebeen published previously.14 Briefly, this was a prospective,multicenter, randomized, double-blind, noninferiority clini-cal trial comparing CRT with settings dynamically adjustedby an investigational aCRT algorithm (aCRT or treatmentarm) with standard BiV pacing with AV and VV settingsoptimized by using a standardized echocardiographic pro-tocol (Echo or control arm). The study enrolled patients whodid not have permanent atrial tachyarrhythmias and wereclinically indicated for implantation of a de novo CRT-defibrillator system. The clinical indication was New YorkHeart Association functional class III or IV HF symptoms,

left ventricular ejection fraction of �35%, and QRS of any

morphology with a duration of �120 ms while on optimalmedical therapy.

Within 30 days prior to the implant, patients underwentfull echocardiographic evaluation of cardiac function,global clinical status, Minnesota Living with Heart FailureQuality of Life assessment, and 6-minute hall walk test.Patients were then implanted with CRT-defibrillator devices(Medtronic D224TRK, D284TRK, and D294TRK) accord-ing to the centers’ standard practices. Within 2 weeks of theimplant, AV and VV delays optimized by using echocardi-ography and delays recommended by the algorithm weredetermined in all patients. Echocardiographic optimizationrequired optimal VV delay to produce the greatest aorticvelocity time integral (AoVTI), and subsequent AV optimi-zation was done by using the iterative method.15 Echocar-iographic aortic flow was acquired at the echo-optimizednd algorithm-suggested settings. The order of the settingsas randomized by the study center, and the images were

abeled in a manner that ensured that the Echo Core Lab athe University of Pittsburgh was blinded to the device pro-ramming. The patients were then randomized in a 2:1 ratioo CRT by using the aCRT algorithm vs standard BiVacing with echo-optimized AV and VV delays. Clinicaltatus was assessed at 1-, 3-, and 6-month postrandomiza-ion. At 6-month postrandomization, all patients underwentchocardiographic imaging of cardiac function at thealk-in device settings and 6-minute hall walk test. Echo-

ardiographic optimization of the AV and VV delays waserformed and aortic flow was assessed at these echo-ptimized settings. All study procedures were the same forcho-optimized and aCRT patients. Assessment of globallinical status and administration of the quality-of-life ques-ionnaire and 6-minute hall walk test were conducted bylinicians blinded to the randomization assignment.

The study had 3 primary end points. The first end pointas the clinical composite score (CCS) developed byacker16 and used in multiple CRT trials1,2 (Figure 1). The

study sought to demonstrate that over 6-month follow-up,the proportion of patients with improved CCS in the aCRT

Figure 1 Clinical composite score definition. HF � heart failure;

NYHA � New York Heart Association.

bsDrpe

peoap

reNwp

1809Martin et al Adaptive CRT Study Results

arm was at least as high as in the Echo arm. The second endpoint required that the concordance correlation coefficientbetween the AoVTI values measured for each patient underecho-optimized and algorithm-recommended programmingexceeded 0.82 both at randomization and at 6-month post-randomization visits. AoVTI is an echocardiographic sur-rogate of stroke volume, and the objective was to demon-strate that cardiac function was similar when using aCRT vsecho-optimized settings. The third end point was designedto demonstrate that the ambulatory adjustments of the CRTsettings provided by the algorithm were safe. If any of theAV or VV delays, stored by the algorithm in the devicememory, varied by more than 60 ms within a period of 28days, that period was to be evaluated by an unblindedsubcommittee of the Adverse Event Adjudication Commit-tee for safety events and appropriateness of the variation insettings. The study sought to show that the proportion oftreatment patients experiencing any inappropriate AV andVV delay changes within 6 months postrandomization was�5%.

Although the primary end points are defined at 6-monthfollow-up, patients will continue to have follow-up visitsevery 6 months until study closure. All adverse events werereviewed and classified by the Adverse Event AdjudicationCommittee,14 which consisted of 2 subcommittees, onelinded and the other unblinded to the randomization as-ignment. The trial was conducted in compliance with theeclaration of Helsinki and the requirements of the local

egulatory authorities and ethics committees of the partici-ating centers. All patients signed informed consent prior tonrollment.

Statistical analysis and sample sizeThe first primary end point of CCS improvement at 6-monthfollow-up was evaluated by using the Farrington-Manning

Figure 2 Flow diagram for the clinicalcomposite score (CCS) end point. Data onCCS were available for all randomized pa-tients, since CCS uses last observation car-ried forward (LOCF) for missing data(LOCF was used for 2.3% of the subjects).For the second primary objective of car-diac performance, the analyzed data in-cluded 399 aCRT patients at randomiza-tion and 235 aCRT patients at follow-up.For the third primary objective of changesin AV and VV delays, data from 314aCRT patients were analyzed. AV � atrio-ventricular; exc � exclusive; inc � inclu-sive; opt. � optimized; rand. � random-ized.

test of 2 independent proportions with a 12% noninferioritymargin. Assuming that the proportion of improved patientswould be 72% and 70% in the aCRT and Echo arms,respectively, a total of 470 patients (313 in the aCRT armand 157 in the Echo arm) were needed to evaluate the testwith 1-sided alpha level of .025 and power of 0.90. If thenoninferiority hypothesis test was passed, a subsequent testfor superiority of aCRT would be performed at the samealpha level. For the second primary end point, the minimalboundary of 0.82 was determined from the data on AoVTImeasurement error17 and Pearson correlation observed inrevious studies.12 The paired concordance correlation co-fficient was evaluated by using a z test with 1-sided alphaf .025 and power of 0.949 separately for the randomizationnd 6-month AoVTI measurements to result in the overallower of 0.90. For the third primary end point, a z test of

proportions was used to evaluate the hypothesis with1-sided alpha of .025, power of 0.90, assuming that 1.5% orless of aCRT subjects were expected to experience inappro-priate AV or VV delays. The sample size for the trial wasthe maximum of the sample sizes for the 3 primary hypoth-eses, which was 470 patients.

If all 3 primary hypotheses were passed, the secondaryhypothesis that RV pacing percentage would be reducedfrom implant to 6-month postrandomization in the aCRTarm was evaluated by using a 2-sample t test with 1-sidedalpha of .025. If this hypothesis test was passed, all othersecondary end points were evaluated for noninferiority in ahierarchical fashion by using the same 2-sample t test. Theest of the secondary end points included changes in LVnd-systolic volume index, left ventricular ejection fraction,ew York Heart Association classification, 6-minute hallalk distance, and quality of life from baseline to 6-monthostrandomization.14 Following the hierarchical analysis, a

A

N

Q

NQLHRCCHAABP

e viation

1810 Heart Rhythm, Vol 9, No 11, November 2012

2-sample t test for superiority with 1-sided alpha of .025was performed for each noninferiority hypothesis that waspassed under the terms of the hierarchical analysis.

Additional prespecified analyses included risk of deathor HF hospitalization, and adverse events. The occurrenceof true ventricular tachycardia and ventricular fibrillation(VT/VF) episodes, collected by the device and adjudicatedby the Episode Review Committee, was compared betweenarms.

ResultsStudy populationA total of 522 patients were enrolled in 94 sites in the UnitedStates, Europe, Central Asia, Australia, Canada, Japan, andHong Kong between November 2009 and December 2010;478 patients were randomized (318 to the aCRT arm and 160to the Echo arm). The flow of study patients up to the 6-monthpostrandomization is shown in Figure 2. The mean follow-upduration was 9.7 � 3.0 months (range 0.2–19.4 months). Bothgroups demonstrated similar and acceptable visit compliancethroughout the study, with more than 90% of the visits com-pleted in both aCRT and Echo arms. Baseline demographiccharacteristics of the 478 patients who were randomized aregiven in Table 1.

Primary end pointsFigure 3 shows the proportion of patients who improved,worsened, or remained unchanged in their CCS at 6-monthpostrandomization. In the aCRT arm, 73.6% (234 of 318) of

Table 1 Baseline demographics and clinical history of the stud

ge (y), mean � SDSex: Men, % (n)BMI (kg/m2), mean � SDYHA class, % (n)IIIIIIIV

ualifying LVEF (%),mean � SDIschemic origin, % (n)onischemic origin, % (n)RS duration (ms), mean � SDBBB, % (n)ypertension, % (n)enal dysfunction, % (n)OPD, % (n)ABG, % (n)eart valve surgery, % (n)trial fibrillation (paroxysmal, persistent, or permanent), % (n)CE inhibitors or ARBs, % (n)eta-blockers, % (n)revious device, % (n)

ACE � angiotensin-converting enzyme; aCRT � adaptive cardiac resyindex; CABG � coronary artery bypass grafting; COPD � chronic obstructijection fraction; NYHA � New York Heart Association; SD � standard de

the patients had an improved CCS vs 72.5% (116 of 160) in

the Echo arm. The 95% confidence interval for the differ-ence was �6.9% to 9.1%, and the noninferiority objectivewas met with a P value of .0007. Since the lower bound ofthe confidence interval did not exceed 0%, superiority wasnot demonstrated. Details on the individual components ofthe CCS for 6-month postrandomization are provided inTable 2.

Echocardiographic AoVTI was obtained for both echo-optimized and aCRT settings in 399 (83.5%) patients at therandomization visit and in 235 (73.9%) aCRT patients at the6-month visit. The main reason for not obtaining the mea-surements was unreadable echocardiographic images. As

nts

aCRT(n � 318)

Echo(n � 160) P

65.4 � 11.2 66.2 � 9.7 .4069 (221) 68 (109) .7629.1 � 5.8 30.1 � 7.1 .11

.420 (0) �1 (1)1 (4) �1 (1)

94 (300) 96 (153)4 (14) 3 (5)24.7 � 6.6 24.9 � 6.5 .74

45 (143) 51 (81) .2447 (151) 41 (65) .16154.3 � 21.0 155.7 � 21.4 .4775 (240) 80 (128) .2764 (202) 69 (110) .2622 (70) 20 (32) .6115 (48) 19 (31) .2324 (76) 31 (50) .094 (13) 7 (11) .19

18 (56) 19 (30) .7686 (274) 89 (143) 0.3291 (289) 91 (146) 0.8923 (72) 18 (29) 0.25

ization therapy; ARB � angiotensin-receptor blocker; BMI � body massonary disease; LBBB � left bundle branch block; LVEF � left ventricular.

Figure 3 Clinical composite score at 6-mo postrandomization by ran-domization arm. The proportion of subjects improved, worsened, andunchanged is shown for the aCRT (light gray bars) and Echo (white bars)arms. The first primary end point compared the proportion of improvedsubjects from baseline to 6-mo postrandomization (noninferiority P �

y patie

nchronve pulm

.0007). aCRT � adaptive cardiac resynchronization therapy.

ps(

sadtt

ctpattDL

C

I

a

1811Martin et al Adaptive CRT Study Results

seen in Figure 4, the paired AoVTI measurements followthe line of equality closely. The lower confidence boundsfor the concordance correlation coefficient both at random-ization and at 6-month postrandomization exceeded the pre-specified threshold of 0.82 (P � .0001); thus, the secondprimary objective was met. At randomization, the meanAoVTI for aCRT and echo-optimized settings was 17.8 �6.0 and 18.0 � 6.2 cm, respectively (P � .18). At 6-monthostrandomization, AoVTI for aCRT and echo-optimizedettings was 17.8 � 5.2 and 18.0 � 5.2 cm, respectivelyP � .20).

Complete daily algorithm data over the entire 6-monthtudy period were recorded for 301 (94.7%) patients in theCRT arm. The reasons for missing or incomplete deviceata were that the device was not interrogated after a pa-ient’s death (n � 6) or after a missed visit (n � 2) or prioro exit (n � 2) or because of human error (n � 7). Based on

the available device data, no patients had wide ranges of AVor VV delays (�60 ms) in any 28-day period. Thus, thethird primary objective was met. During review of theadverse events, the unblinded subcommittee of the Adverse

Table 2 Clinical composite response details in the aCRT and Ec

linical composite score

mprovedModerately or markedly improved patient global assessment anImproved NYHA class onlyModerately or markedly improved patient global assessment on

UnchangedWorsened

DeathHospitalized because of or associated with worsening HFCrossover due to worsening HFModerately or markedly worse patient global assessment and wWorsened NYHA classModerately or markedly worse patient global assessment

aCRT � adaptive cardiac resynchronization therapy; CCS � clinical co*Note that a subject is indicated only once according to the CCS definitionin the “Death” row).

Figure 4 Comparison of cardiac performance at the aCRT and Echo-oppoint compared echocardiographic aortic velocity time integral (AoVTI)echocardiographic optimization protocol for all subjects at randomization

CRT � adaptive cardiac resynchronization therapy; CCC � concordance correl

Event Adjudication Committee also reviewed device andalgorithm data for each adverse event and found no adverseevents related to the aCRT algorithm.

Secondary end pointsSince all primary hypotheses were satisfied, the prespecifiedhierarchical analysis was conducted on the secondary endpoints. Ventricular pacing percentage data were collectedvia device interrogations. A total of 314 of the 318 patientsin the aCRT group and all 160 patients in the control grouphad at least 1 interrogation following randomization. Thepercentage of ventricular pacing through 6 months in theaCRT arm was 95.5% � 5.7% (range 35.6%–100.0%) asompared with 95.1% � 10.5% (range 2.9%–100.0%) inhe Echo arm. In the Echo arm, virtually all ventricularacing was BiV, whereas in the aCRT arm BiV pacingccounted for a median of 50.9% of all ventricular pacing,he rest being LV-only pacing. Figure 5 shows the distribu-ion of LV-only and BiV pacing in the aCRT patients.epending on a patient’s intrinsic AV conduction, percentV-only pacing could range from 0.0% to 97.9%. Forty-

ups at 6-mo postrandomization*

% (n)

aCRT Echo

73.6 (234) 72.5 (116)oved NYHA class 50.9 (162) 46.3 (74)

14.8 (47) 17.5 (28)7.9 (25) 8.8 (14)

12.3 (39) 16.3 (26)14.2 (45) 11.3 (18)4.1 (13) 1.3 (2)8.5 (27) 10.0 (16)0.0 (0) 0.0 (0)

d NYHA class 0.3 (1) 0.0 (0)0.3 (1) 0.0 (0)0.9 (3) 0.0 (0)

score; HF � heart failure; NYHA � New York Heart Association.subject who died and had a hospitalization for worsening HF is listed only

settings at randomization and 6-mo follow-up. The second primary endsettings calculated by the aCRT algorithm and settings obtained using

anel) and only aCRT subjects at 6-mo postrandomization (right panel).

ho gro

d impr

ly

orsene

mposite(eg, a

timizedat the(left p

ation coefficient.

pltiSt

L

N

6

M

1812 Heart Rhythm, Vol 9, No 11, November 2012

seven percent of aCRT patients experienced LV-only pac-ing at least 50% of the time. Patients with aCRT had, onaverage, a 43.8% absolute reduction in RV pacing over 6months postrandomization compared with control patients.

Table 3 lists the results for the remaining secondary endoints. Both arms demonstrated an improvement from base-ine to 6-month postrandomization, and the differences be-ween the aCRT and Echo arms were within the correspond-ng noninferiority margins for all secondary end points.uperiority testing did not reach significance for any of

hese end points.

Additional analysesDeath and HF hospitalizations at 6-month follow-up arepresented in Table 2. Further examination of all follow-updata available at the time of the 6-month analysis data lock,

Table 3 Structural and functional secondary end points

aCRT (n � 318) E

n Mean � SD n

LVESVi (mL/m2)Baseline 291 71.7 � 28.3 16-mo postrandomization 268 63.5 � 31.9 1Paired difference at 6 mo 250 �8.3 � 23.3 1

VEF (%)Baseline 291 29.6 � 9.2 16-mo postrandomization 268 33.6 � 10.4 1Paired difference at 6 mo 250 3.9 � 10.0 1

YHABaseline 318 3.0 � 0.2 16-mo postrandomization 296 2.0 � 0.8 1Paired difference at 6 mo 296 �1.0 � 0.8 1

-min walk distance (m)Baseline 312 276.8 � 127.5 16-mo postrandomization 288 325.5 � 130.4 1Paired difference at 6 mo 284 42.4 � 103.3 1

LWHF QOLBaseline 286 48.5 � 24.1 16-mo postrandomization 263 28.2 � 22.0 1Paired difference at 6 mo 261 �19.3 � 20.7 1

aCRT � adaptive cardiac resynchronization therapy; CI � confidence inindex; MLWHF � Minnesota Living With Heart Failure; NYHA � New York

*P value for noninferiority between aCRT and Echo arms.

which includes follow-up after the 6-month visit, was done.There were a total of 18 deaths and 71 HF hospitalizations(46 patients) in the aCRT arm and 7 deaths and 34 HFhospitalizations (21 patients) in the Echo arm. At the meanfollow-up of 9.7 months, the mortality rate was 7.5% in theaCRT arm and 8.8% in the Echo arm and the rate of first HFhospitalization was 18.6% in the aCRT arm and 16.1% inthe Echo arm. Neither mortality (log-rank P � .47) nor firstHF hospitalization (log-rank P � .61) rates were signifi-cantly different between the arms.

There was no difference in time to first occurrence ofVT/VF between the aCRT and Echo arms (log-rank P �.87). The rate of VT/VF per patient-year was 1.68 in theaCRT arm and 0.88 in the Echo arm and was not signifi-cantly different (generalized estimating equations [GEE]

Figure 5 Distribution of LV-only and biventricular pacingin the aCRT arm. Percentage of total ventricular pacing con-sists of percent adaptive BiV pacing (light gray) and percentLV-only pacing (dark gray) and is displayed as 1 stacked barfor each patient. aCRT � adaptive cardiac resynchronizationtherapy; LV � left ventricular.

� 160)

Difference (95% CI) P* (margin)Mean � SD

74.0 � 30.964.7 � 32.7

�10.5 � 24.2 2.3 �(2.8 to 7.4) �.0001 (15)

30.3 � 8.432.9 � 10.12.9 � 9.8 1.0 �(1.2 to 3.1) 0.0009 �(2.5)

3.0 � 0.32.2 � 0.8

�0.8 � 0.8 �0.15 (0.3 to 0.0) �.0001 (0.3)

277.7 � 137.8311.4 � 152.029.0 � 123.0 13.4 �(8.9 to 35.7) 0.0002 �(30)

46.3 � 23.628.4 � 23.0

�17.6 � 23.8 �1.7 �(6.3 to 2.8) 0.002 (5.1)

LVEF � left ventricular ejection fraction; LVESVi � LV end-systolic volumessociation; QOL � quality of life; SD � standard deviation.

cho (n

403723

403723

605353

564642

423935

terval;Heart A

dtp

marLrw

fpaLn

sdagL

p

1813Martin et al Adaptive CRT Study Results

P � .46). There were 3 patients (2 aCRT, 1 Echo) who hadVT/VF episode counts exceeding 3 standard deviationsfrom the mean. The 2 aCRT patients had 19 and 128episodes, whereas the Echo patient had 19 episodes. Toaccount for these outliers, a sensitivity analysis was per-formed by substituting the number of episodesin the outliers with the median number of episodes (2)among the rest of the subjects with VT/VF. After thisadjustment, the aCRT group had 0.53 events per patient-year vs 0.66 events per patient-year in the Echo group.

The odds of an improved CCS were compared betweenaCRT and Echo arms for typically considered subgroups(Figure 6). For left ventricular ejection fraction and QRSduration, the subgroups were defined by the quartiles andsubgroup analyses tested for a linear trend. The analysisdemonstrated comparable results across subgroups.

We also analyzed the subgroup of patients in sinusrhythm with normal AV conduction and left bundle branchblock before randomization, as one would expect this sub-group to benefit from LV fusion pacing the most. Withinthis subgroup (n � 219; 45.8%), there were no significantifferences between the arms in baseline demographics andotal percent atrial and ventricular pacing, but the aCRTatients received LV-only pacing 64.0% � 32.8% of the

time. In this subgroup, more aCRT patients improved inCCS compared with the Echo arm (80.7% vs 68.4%; P �

Figure 6 Improved clinical composite score by subgroups. CRT �cardiac resynchronization therapy; LBBB � left bundle branch block;LVEF � left ventricular ejection fraction; NYHA � New York Heart

mAssociation.

.04), with an odds ratio of 1.94 for improved CCS (95%confidence interval 1.03–3.65).

DiscussionThe aCRT trial evaluated a novel pacing algorithm for CRT.The design of the algorithm was based on the hypothesisthat CRT benefit can further be increased through (1) avoid-ance of RV pacing and greater recruitment of intrinsicconduction in patients with normal conduction into the rightventricle and (2) dynamic adjustment of AV and VV delaysbased on the electrical conduction intervals. The aCRTalgorithm resulted in a 44% absolute reduction in the per-centage of RV pacing. The trial demonstrated that optimi-zation provided by the aCRT algorithm is safe and nonin-ferior to echocardiographic optimization with respect toclinical, structural, and functional improvement at 6-monthpostrandomization.

Among patients with intact AV conduction, LV dysfunc-tion, and a narrow QRS, it is well accepted that RV pacingis to be avoided as it results in a higher incidence of HF,presumably due to iatrogenic ventricular dyssynchronycaused by RV pacing.18 One can speculate that substitutingintrinsic RV activation with RV-paced activation may alsobe unnecessary in patients with left bundle branch block andLV dysfunction. Multiple acute studies proposed incremen-tal superiority of appropriately timed LV pacing over si-multaneous BiV pacing in the CRT subpopulation withsinus rhythm and intact AV conduction5. For instance, vanGelder et al7 compared invasive hemodynamic measure-

ents under simultaneous BiV pacing and LV-only pacingt a variety of AV delays in 34 CRT patients who had sinushythm and intact AV conduction. The authors found thatV-only pacing produced a higher maximum LV pressure

ise (dP/dtmax) than BiV pacing, but only when LV pacingas timed to fuse with intrinsic ventricular activation.RV pacing may also have deleterious effects on RV

unction. In an acute hemodynamic study using 17 CRTatients with sinus rhythm and intact AV conduction, Lee etl8 found that BiV pacing suppressed RV dP/dtmax, whereasV pacing synchronized to preempt intrinsic conduction didot. To evaluate the mechanism of this further, Varma et al9

used electrocardiographic mapping to assess ventricular ac-tivation by using RV-only, LV-only, and BiV pacing atoptimized AV intervals in 14 CRT recipients. RV pacing,whether alone or as part of BiV pacing, produced RVactivation delays, which were avoided with LV pacing.

The previous studies suggested that LV fusion pacing ismost likely to benefit patients with left bundle branch blockand normal AV conduction.7 In the present trial, this patientubgroup primarily received synchronized LV pacing andemonstrated a higher improvement in CCS with the aCRTlgorithm. Further investigation of clinical outcomes over lon-er follow-up is needed to support the benefit of synchronizedV pacing.

The present trial compared the aCRT algorithm to BiVacing with echocardiographic optimization, utilized in

ultiple CRT trials and recommended by the American

t

ttOnC

1

1

1

1

1

1

1

1

1

1

2

2

1814 Heart Rhythm, Vol 9, No 11, November 2012

Society of Echocardiography.15 Although the clinical ben-efit of echocardiographic optimization was not demon-strated in 1 multicenter trial,19 smaller studies have sug-gested a positive impact on clinical condition10 or exerciseolerance.11 In contrast to previous device optimization tri-

als,19, 20 the control arm in the present study had both AVand VV delays optimized by using a mandatory standard-ized protocol.

Electrocardiographic CRT optimization algorithms havepreviously been proposed and evaluated.19,20 However, thesealgorithms reside in a device programmer and provide onlyin-office optimization, which may be insufficient, since opti-mal settings may change with time and patient activity.21 TheaCRT algorithm is unique in that it permits dynamic AV andVV optimization with the option of synchronized LV pacing.14

Based on bench-testing, the algorithm itself has negligibleimpact on battery life. The synchronized LV-pacing optionshould improve the longevity of the device and may benefitpatients with normal AV conduction. Interestingly, the recentGREATER-EARTH trial,6 which utilized a cross-over designo compare LV pacing with BiV pacing, showed that 17.1% ofhe nonresponders to BiV pacing improved with LV pacing.ur findings suggest that the aCRT algorithm may offer cli-icians and patients enhanced options to improve response toRT.

Study limitationsThe specific echocardiographic optimization approach cho-sen for the treatment arm may be no better than otheroptimization methods or no optimization at all. The studywas conducted in a population of New York Heart Associ-ation HF class III and IV patients without permanent atrialtachyarrhythmias, and so the results cannot be generalizedto patients with permanent atrial fibrillation or to less symp-tomatic patients. Longer-term follow-up is needed to con-firm the safety of the algorithm with respect to mortality andhospitalizations. Inappropriate AV and VV delay changescaused by the aCRT algorithm were defined as changes of atleast 60 ms. It is possible that smaller variations could havean adverse impact on clinical condition. Missing deviceinterrogation data from aCRT patients (5.3% of the patientshad at least 1 interrogation missing) could have had safety-related information.

ConclusionsThe aCRT algorithm was safe and at least as effective asBiV pacing with comprehensive echocardiographic optimi-zation across a variety of primary and secondary end points.

SupplementA complete list of primary investigators in the Adaptive

CRT Clinical Trial is available online.

References1. Abraham WT, Fisher WG, Smith AL, et al. Cardiac resynchronization in chronic

heart failure. N Engl J Med 2002;346:1845–1853.2. Young JB, Abraham WT, Smith AL, et al. Combined cardiac resynchronization

and implantable cardioversion defibrillation in advanced chronic heart failure:the MIRACLE ICD trial. JAMA 2003;289:2685–2694.

3. Zareba W, Klein H, Cygankiewicz I, et al. Effectiveness of cardiac resynchro-nization therapy by QRS morphology in the Multicenter Automatic DefibrillatorImplantation Trial-Cardiac Resynchronization Therapy (MADIT-CRT). Circu-lation 2011;123:1061–1072.

4. Ypenburg C, Roes SD, Bleeker GB, et al. Effect of total scar burden oncontrast-enhanced magnetic resonance imaging on response to cardiac resyn-chronization therapy. Am J Cardiol 2007;99:657–660.

5. Kass DA, Chen CH, Curry C, et al. Improved left ventricular mechanics fromacute VDD pacing in patients with dilated cardiomyopathy and ventricularconduction delay. Circulation 1999;99:1567–1573.

6. Thibault B, Ducharme A, Harel F, et al. Left ventricular versus simultaneousbiventricular pacing in patients with heart failure and a QRS complex �120milliseconds. Circulation 2011;124:2874–2881.

7. van Gelder BM, Bracke FA, Meijer A, Pijls NH. The hemodynamic effect ofintrinsic conduction during left ventricular pacing as compared to biventricularpacing. J Am Coll Cardiol 2005;46:2305–2310.

8. Lee KL, Burnes JE, Mullen TJ, Hettrick DA, Tse HF, Lau CP. Avoidance ofright ventricular pacing in cardiac resynchronization therapy improves rightventricular hemodynamics in heart failure patients. J Cardiovasc Electrophysiol2007;18:497–504.

9. Varma N, Jia P, Ramanathan C, Rudy Y. RV electrical activation in heart failureduring right, left, and biventricular pacing. JACC Cardiovasc Imaging 2010;3:567–575.

0. Sawhney NS, Waggoner AD, Garhwal S, Chawla MK, Osborn J, Faddis MN.Randomized prospective trial of atrioventricular delay programming for cardiacresynchronization therapy. Heart Rhythm 2004;1:562–567.

1. Vidal B, Sitges M, Marigliano A, et al. Optimizing the programation of cardiacresynchronization therapy devices in patients with heart failure and left bundlebranch block. Am J Cardiol 2007;100:1002–1006.

2. Jones RC, Svinarich T, Rubin A, et al. Optimal atrioventricular delay in CRTpatients can be approximated using surface electrocardiography and deviceelectrograms. J Cardiovasc Electrophysiol 2010;21:1226–1232.

3. Khaykin Y, Exner D, Birnie D, Sapp J, Aggarwal S, Sambelashvili A. Adjustingthe timing of left-ventricular pacing using electrocardiogram and device elec-trograms. Europace 2011;13:1464–1470.

4. Krum H, Lemke B, Birnie D, et al. A novel algorithm for individualized cardiacresynchronization therapy: rationale and design of the adaptive CRT trial. AmHeart J 2012;163:747–752.e.1.

5. Gorcsan J III, Abraham T, Agler DA, et al. Echocardiography for cardiacresynchronization therapy: recommendations for performance and reporting–areport from the American Society of Echocardiography Dyssynchrony WritingGroup endorsed by the Heart Rhythm Society. J Am Soc Echocardiogr 2008;21:191–213.

6. Packer M. Proposal for a new clinical end point to evaluate the efficacy of drugsand devices in the treatment of chronic heart failure. J Card Fail 2001;7:176–182.

7. Chung ES, Leon AR, Tavazzi L, et al. Results of the Predictors of Response toCRT (PROSPECT) trial. Circulation 2008;117:2608–2616.

8. Wilkoff BL, Cook JR, Epstein AE, et al. Dual-chamber pacing or ventricularbackup pacing in patients with an implantable defibrillator: the Dual Cham-ber and VVI Implantable Defibrillator (DAVID) trial. JAMA 2002;288:3115–3123.

9. Ellenbogen KA, Gold MR, Meyer TE, et al. Primary results from the SmartDelaydetermined AV optimization: a comparison to other AV delay methods used incardiac resynchronization therapy (SMART-AV) trial: a randomized trial com-paring empirical, echocardiography-guided, and algorithmic atrioventricular de-lay programming in cardiac resynchronization therapy. Circulation 2010;122:2660–2668.

0. Abraham WT, Gras D, Yu CM, et al. Results from FREEDOM trial—assess thesafety and efficacy of frequent optimization of cardiac resynchronization ther-apy. HRS 2010 Late-Breaking Clinical Trials. May 13, 2010

1. Scharf C, Li P, Muntwyler J, et al. Rate-dependent AV delay optimization in

cardiac resynchronization therapy. Pacing Clin Electrophysiol 2005;28:279–284.

1814.e1Martin et al Adaptive CRT Study Results

AppendixThe authors would like to congratulate the achievementsand acknowledge the contributions made by the follow-ing primary investigators in the Adaptive CRT ClinicalTrial: Dr. Vikram Nangia, Aurora Sinai Medical Center,USA; Dr. Peter Wells, Baylor Jack and Jane Hamilton Heartand Vascular Hospital, USA; Dr. Andrew Merliss, Bryan-LGH Heart Institute, USA; Dr. Irfan Khan, Buffalo HeartGroup LLC, USA; Dr. James Stone, Cardiology Associatesof North Mississippi, USA; Dr. David Rodak, SpartanburgRegional Healthcare System, USA; Dr. Sung Lee, Wash-ington Adventist Hospital, USA; Dr. Simon Milstein, Cen-tral Minnesota Heart Center, USA; Dr. David Martin,Cleveland Clinic Foundation, USA; Dr. Thomas Svinarich,Colorado Heart and Vascular, PC, USA; Dr. Daniel Lust-garten, Fletcher Allen Health Care, USA; Dr. Robert Sor-rentino, Georgia Health Sciences University, USA; Dr. JohnMcKenzie, Glendale Memorial Hospital & Health Center,USA; Dr. Jonathan Hobson, Heart & Vascular Institute ofFlorida, USA; Dr. Jeffrey Scott Allison, Heart Center Re-search LLC, USA; Dr. Van De Bruyn, Heart Clinic Arkan-sas PA, USA; Dr. Timothy Lessmeier, Heart Clinics North-west PS, USA; Dr. W. Ben Johnson, Iowa Heart Center PC,USA; Dr. Satish Goel, Jacksonville Heart Center, USA; Dr.Martin Emert, Kansas University Medical Center ResearchInstitute Inc, USA; Dr. Robert Lerman, LA CardiologyAssociates, USA; Dr. Muqtada Chaudhry, Lahey ClinicHospital Inc, USA; Dr. Steven Klein, LeBauer Cardiovas-cular Research Foundation, USA; Dr. Michael Imburgia,Louisville Cardiology Medical Group PSC, USA; Dr. Mi-chael Gold, Medical University of South Carolina, USA;Dr. Liaqat Zaman, Michigan Cardiovascular Institute, USA;Dr. Brian Ramza, Mid America Heart Institute, USA; Dr. J.Russell Bailey, Mid Carolina Cardiology, USA; Dr. DavidBello, Mid Florida Cardiology, USA; Dr. Blair Foreman,Midwest Cardiovascular Research Foundation, USA; Dr.Raymond Kawasaki, Midwest Heart Foundation, USA; Dr.John Lobban, Morgantown Internal Medicine Group, USA;Dr. Jay Curwin, Morristown Memorial Hospital, USA; Dr.Gioia Turitto, New York Methodist Hospital, USA; Dr.Jonathan Lowy, North Cascade Cardiology, USA; Dr. SreeKaranam, Northern Indiana Research Alliance, USA; Dr.Eric Putz, Northwest Cardiovascular Institute, USA; Dr.Jack Collier, Oklahoma Cardiovascular Research Group,USA; Dr. Jay Simonson, Park Nicollet Institute, USA; Dr.Brad Mikaelian, Pikes Peak Cardiology, USA; Dr. StevenCompton, Providence Alaska Medical Center, USA; Dr.Luis Pires, Saint John Hospital and Medical Center, USA;Dr. J. Timothy Walsh, Saint Vincent’s Ambulatory CareInc, USA; Dr. Manish Wadhwa, San Diego ArrhythmiaAssociates, USA; Dr. John Herre, Sentara Norfolk GeneralHospital, USA; Dr. Robert Canby, Texas Cardiac Arrhyth-mia Research, USA; Dr. Keith Bruce, The ChattanoogaHeart Institute, USA; Dr. Mahmoud Houmsse, The OhioState University Medical Center, USA; Dr. John Sims and

Dr. Robert Carney, Tyler Cardiovascular Consultants, USA;

Dr. Alan Bank, United Heart and Vascular Clinic, USA; Dr.Mark Borganelli, University of Mississippi Medical Center,USA; Dr. Walter Clair, Vanderbilt University, USA; Dr.Richard Shepard, Virginia Commonwealth UniversityHealth System, USA; Dr. Jim Cheung, Weill Medical Col-lege of Cornell University, USA; Dr. Darryl Elmouchi,West Michigan Heart, USA; Dr. F. Heath, Aalborg Syge-hus, Denmark; Dr. A. Kypta, Allgemeines Krankenhaus derStadt Linz, Austria; Dr. P.T. Mortensen, Århus Universitet-shospital Skejby, Denmark; Dr. F. Voss, BarmherzigeBrüder Trier e.V. - Krankenhaus der Barmherzigen BrüderTrier – Akademisches Lehrkranken, Germany; Dr. T. Lawo,Berufsgenossenschaft liche Universitätsklinik Bergmann-sheil GmbH, Germany; Dr. L.H.R. Bouwels, Canisius-Wil-helmina ziekenhuis, the Netherlands; Prof. G. Milasinovic,Clinical Center of Serbia, Republic of Serbia; Dr. M.E.Landolina, Fondazione IRCCS Policlinico San Matteo, It-aly; Prof. S. Faerestrand, Helse Bergen HF – HaukelandUniversitetssjukehus, Norway; Dr. M. López Gil, HospitalUniversitario 12 de Octubre, Spain; Prof. D.V. Kovacevic,Institut za Kardiovaskularne Bolesti Vojvodine, Republic ofSerbia; Prof. M. Gasparini, IRCCS Istituto Clinico Humani-tas - Humanitas Mirasole Spa, Italy; Dr. J. Hörnsten, Karo-linska Universitetssjukhuset, Sweden; Prof. Z. Perisic, Klin-icky Centar Niš, Republic of Serbia; Prof. H. Nägele,Krankenhaus Reinbek St. Adolf Stift, Germany; Dr. M.Scheffer, Maasstad Ziekenhuis - Lokatie St. Clara, the Neth-erlands; Prof. B. Lemke, Märkische GesundheitsholdingGmbH & Co. KG – Klinikum Lüdenscheid, Germany; Dr.A. Brandes, Odense Universitetshospital, Denmark; Dr. E.Kongsgård, Oslo Universitetssykehus, Rikshospitalet, Nor-way; Dr. C. Hassager, Rigshospitalet, Denmark; Dr. A.Hersi, The College of Medicine & King Khalid UniversityHospital, King Saud University, Saudi Arabia; Dr. V.A.Kuznetsov, Tyumen Cardiology Center, Russia; Dr. M.A.Aydin, Universitäres Herzzentrum Hamburg GmbH (UHZ),Germany; Dr. R. Borgquist, Universitetssjukhuset i Lund,Sweden; Dr. W. Mullens, Ziekenhuis Oost Limburg – Cam-pus St.-Jan, Belgium; Dr. Vince Paul, Royal Perth Hospital,Australia; Dr. Michael Kilborn, Royal Prince Alfred Hos-pital, Australia; Dr. John Hayes, St. Andrew’s Hospital,Australia; Dr. Bruce Walker, St Vincent’s Hospital Sydney,Australia; Dr. Russell Denman, The Prince Charles Hospi-tal, Australia; Prof. Prashanthan Sanders, Royal AdelaideHospital, Australia; Dr. Raymond Yee, London Health Sci-ences Centre - University Campus, Canada; Dr. YaarivKhaykin, Newmarket Electrophysiology Research Group,Canada; Dr. Glen Sumner, University of Calgary LibinCardiovascular Institute, Canada; Dr. David Birnie, Univer-sity of Ottawa Heart Institute, Canada; Dr. Anthony Tang,Victoria Cardiac Arrhythmia Trials Inc, Canada; Dr. ShiroKamakura, National Cerebral and Cardiovascular Center,Japan; Dr. Kazutaka Aonuma, Tsukuba University Hospital,Japan; Dr. Hung-Fat Tse, Queen Mary Hospital, Hong

Kong.