J A C C : C A R D I O V A S C U L A R I M A G I N G V O L . 4 , N O . 1 , 2 0 1 1

© 2 0 1 1 B Y T H E A M E R I C A N C O L L E G E O F C A R D I O L O G Y F O U N D A T I O N I S S N 1 9 3 6 - 8 7 8 X / $ 3 6 . 0 0

P U B L I S H E D B Y E L S E V I E R I N C . D O I : 1 0 . 1 0 1 6 / j . j c m g . 2 0 1 0 . 0 9 . 0 2 1

Real-Time 3D Echo in Patient Selection forCardiac Resynchronization Therapy

Stamatis Kapetanakis, MBBS,* Amit Bhan, MBBS,* Francis Murgatroyd, MA,*Mark T. Kearney, MD, MBCHB,† Nicholas Gall, MSC, MD,* Qing Zhang, BM, MM,‡Cheuk-Man Yu, MD,§ Mark J. Monaghan, PHD*

London and Leeds, United Kingdom; and Chengdu and Hong Kong, China

O B J E C T I V E S This study investigated the use of 3-dimensional (3D) echo in quantifying left ventricular

mechanical dyssynchrony (LVMD), its interhospital agreement, and potential impact on patient selection.

B A C K G R O U N D Assessment of LVMD has been proposed as an improvement on conventional

criteria in selecting patients for cardiac resynchronization therapy (CRT). Three-dimensional echo offers

a reproducible assessment of left ventricular (LV) structure, function, and LVMD and may be useful in

selecting patients for this intervention.

M E T H O D S We studied 187 patients at 2 institutions. Three-dimensional data from baseline and

longest follow-up were quantified for volume, left ventricular ejection fraction (LVEF), and systolic

dyssynchrony index (SDI). New York Heart Association (NYHA) functional class was assessed indepen-

dently. Several outcomes from CRT were considered: 1) reduction in NYHA functional class; 2) 20%

relative increase in LVEF; and 3) 15% reduction in LV end-systolic volume. Sixty-two cases were shared

between institutions to analyze interhospital agreement.

R E S U L T S There was excellent interhospital agreement for 3D–derived LV end-diastolic and end- systolic

volumes, EF, and SDI (variability: 2.9%, 1%, 7.1%, and 7.6%, respectively). Reduction in NYHA functional class

was found in 78.9% of patients. Relative improvement in LVEF of 20% was found in 68% of patients, but

significant reduction in LV end-systolic volume was found in only 41.5%. The QRS duration was not predictive

of any of themeasures of outcome (area under the curve [AUC]: 0.52, 0.58, and 0.57 for NYHA functional class,

LVEF, and LV end-systolic volume), whereas SDI was highly predictive of improvement in these parameters

(AUC: 0.79, 0.86, and 0.66, respectively). For patients not fulfilling traditional selection criteria (atrial fibrillation,

QRS duration �120 ms, or undergoing device upgrade), SDI had similar predictive value. A cutoff of 10.4%

for SDI was found to have the highest accuracy for predicting improvement following CRT.

C O N C L U S I O N S The LVMD quantification by 3D echo is reproducible between centers. SDI was an

excellent predictor of response to CRT in this selected patient cohort and may be valuable in identifying

a target population for CRT irrespective of QRS morphology and duration. (J Am Coll Cardiol Img 2011;

4:16–26) © 2011 by the American College of Cardiology Foundation

From the *Department of Cardiology, King’s College Hospital, Denmark Hill, London, United Kingdom; †Division ofCardiovascular and Diabetes Research, Leeds Multidisciplinary Cardiovascular, University of Leeds, Leeds, United Kingdom;‡Department of Cardiology, West China Hospital, Sichuan University, Chengdu, China; and the §Prince of Wales Hospital,Hong Kong, China. Drs. Kapetanakis and Bhan have received honoraria from Philips Medical Systems for teaching. Dr.Murgatroyd is a member of the advisory boards for Medtronic, Boston Scientific, and Sorin, and has received honoraria fromMedtronic and Sorin as well as research support from Medtronic. Dr. Kearney has received research support from Medtronic.Dr. Gall has received honoraria and unrestricted research grants from St. Jude, Medtronic, Sorin ELA, and Boston Scientific.Dr. Zhang has reported that he has no relationships to disclose. Prof. Yu has received minor research support from Philips. Dr.Monaghan has received honoraria from Philips Medical Systems for teaching and research support from Philips, GEHealthcare, and TomTec.

Manuscript received June 7, 2010; revised manuscript received September 14, 2010, accepted September 16, 2010.

CpeocaSttAof((b(tn

aimoctegocShsit1(pL(tqwc

uaTadr

2Rvt

M

WidCdUCfCwfPCnstm(AfscbmpsaCufafisvivEwMdwoiwt

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

17

ardiac resynchronization therapy (CRT)has emerged as an important therapeuticintervention in the treatment of chronicheart failure. In multicenter trials, selected

atients with chronic heart failure undergoing CRTxperience significant improvement in both markersf subjective well-being and echo parameters ofardiac function as well as clinical end points suchs hospitalization for heart failure and death (1–6).election criteria in trials of CRT have been rela-ively uniform in identifying patients with symp-omatic chronic heart failure (New York Heartssociation [NYHA] functional class III or IV onptimal medical therapy, left ventricular ejectionraction [LVEF] �35%) and electrocardiographicECG) evidence of dyssynchronous LV contractionprolonged QRS duration [QRSd] with left bundleranch block [LBBB] pattern on surface ECG)7–9). These widely accepted patient selection cri-eria are, however, associated with a 20% to 30%onresponder rate.Nonresponse to CRT may reflect intrinsic char-

cteristics of the heart (i.e., synchrony cannot bemproved) or inadequacies in conventional CRT

ethods (i.e., synchrony does not improve becausef inability to pace at an appropriate location withurrent technology). In attempting to identify po-ential responders to CRT more accurately, studiesxamining various imaging modalities have sug-ested that there is a correlation between presencef significant left ventricular mechanical dyssyn-hrony (LVMD) and positive outcome from CRT.everal echocardiographic methods in particularave appeared promising in a large number ofingle-center trials. However, a lack of reproduc-bility in multicenter trials has cast some doubt overhe clinical applicability of these techniques (10–2). Real-time 3-dimensional echocardiographyRT3DE) has been proposed as an alternative andotentially more accurate method for quantifyingVMD and identifying patients suitable for CRT

13). Several studies have proven this modality to behe most accurate echocardiographic method foruantifying LV volumes and function (14–21),hich in itself may have an important impact on

linical practice (22).We previously demonstrated that RT3DE can be

sed to examine and quantify endocardial motion ofll myocardial segments across their entire surface.his has allowed us to derive a potentially more

ccurate measure of LVMD called the systolicyssynchrony index (SDI) (13), which has been

eproduced in multiple later studies (23–30). In this o

-center study, we investigate in potential value ofT3DE in patients undergoing CRT, intercenter

ariability in quantification, and the ability of theechnique to identify responders to this therapy.

E T H O D S

e investigated 187 patients undergoing primarymplant or upgrade to biventricular pacemakers orefibrillators between 2004 and 2008 at King’sollege Hospital (KCH), London, United King-om and the Prince of Wales Hospital, Chineseniversity of Hong Kong (CUHK), Hong Kong,hina. Data from 62 cases were shared across sites

or interhospital variability assessment.linical and echocardiographic outcomesere recorded at baseline and longest

ollow-up.atient selection. Patient selection forRT was guided by traditional criteria:ormal sinus rhythm with LVEF �35%,ymptomatic heart failure (NYHA func-ional class III/IV) on maximal achievableedical therapy, and QRS prolongation

�120 ms) with LBBB pattern on ECG.t the implanter’s discretion, patients not

ulfilling all these criteria were also con-idered for biventricular pacing, thus in-luding some patients with right bundleranch block, paced rhythm, QRSd �120s, or persistent atrial fibrillation. In these

atients, baseline echocardiographic as-essment of LVMD was also taken intoccount.linical assessment. All patients were reg-larly reviewed by nurse specialists in heartailure. Functional capacity was assessedccording to the NYHA functional classi-cation. NYHA functional class was as-igned before implantation and at each subsequentisit, without reference to echo findings. Clinicalnformation was retrieved at the end of the obser-ation period.chocardiography. RT3DE studies were performedith the Sonos 7500 or iE33 systems (Philipsedical Systems, Andover, Massachusetts). Three-

imensional datasets were acquired in the apicalindow with an ECG-gated acquisition as previ-usly described (13). In patients with arrhythmia,ncluding atrial fibrillation, multiple acquisitionsere performed until a dataset with no appreciable

ranslation artifacts could be obtained, as we previ-

A B B

A N D

AUC �

CRT �

therap

ECG �

ICC �

coeffi

LBBB

LVEF

fractio

LVESV

systol

LVMD

mecha

NYHA

Assoc

QRSd

ROC �

chara

RT3D

dimen

SDI �

index

usly described (13).

R E V I A T I O N S

A C R O N YM S

area under the curve

cardiac resynchronization

y

electrocardiography

intraclass correlation

cient

� left bundle branch block

� left ventricular ejection

n

� left ventricular end-

ic volume

� left ventricular

nical dyssynchrony

� New York Heart

iation

� QRS duration

receiver-operator

cteristic

E � real-time 3-

sional echocardiography

systolic dyssynchrony

Qt(Ttmssv(fspDjop1ptdOrcc2r1(torctIhcbtegtpwtafbeSmn

ttSmysvcFl

ccQt

sa

caaaJCIa

R

DatTlvlpwiffT2rNfC7Pc2

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

18

uantification of dyssynchrony. RT3DE acquisi-ions were analyzed with 4-dimensional LV analysisResearch Arena 2.0, TomTec, Munich, Delaware).his software is performs 3D endocardial border

racking throughout the cardiac cycle, to provide aathematical model of the LV volume. This is

egmented into 16 subvolumes corresponding to thetandard myocardial segments to derive time-olume curves for each. Time to peak contractionminimum volume) in each segment is normalizedor the R-R duration, and SDI is defined as thetandard deviation of these timings, expressed as aercentage of cardiac cycle duration.evice optimization. Atrioventricular delay was ad-

usted 2 � 1 day after implantation by examinationf LV inflow with pulsed Doppler at the mitralosition. The average atrioventricular delay was 115 �5 ms. All patients received synchronous biventricularacing as default after implantation and no adjustment ofhe interventricular delay was performed beforeischarge.utcomes. Several outcomes were taken as positive

esponse to CRT. A reduction in NYHA functionallass by at least 1 class was considered a positivelinical response. A relative increase in LVEF by0% was considered a positive echocardiographicesponse. Reverse remodeling was defined as at least5% reduction in LV end-systolic volumeLVESV). Empirically, patients often report subjec-ive improvement despite not reaching these thresh-lds. A more sensitive measure of outcome—�10%elative increase in LVEF—was therefore also in-luded on an exploratory basis, though this approacheshe limit of reproducibility of the technique.nterhospital variability. Three-dimensional echoas been shown to be highly reproducible in single-enter studies (13,23,26,27,30). To assess variabilityetween centers, 62 datasets were shared betweenhe 2 participating specialist cardiology centers. Tonsure that methodology was comparable, investi-ators at CUHK underwent 1 week of intensiveraining with an experienced operator from KCHrior to the study. For this purpose, 20 datasetsith a variety of pathologies were analyzed, and at

he end of the training period there was excellentgreement. The datasets—43 from KCH and 19rom CUHK—were analyzed independently atoth sites to examine interhospital variability ofnd-diastolic and end-systolic volumes, LVEF, andDI. To illustrate the impact of image quality ofeasurement variability, we assessed this based on

umber of missing segments, endocardial defini- c

ion, and presence of artifacts; datasets were quali-atively assessed as excellent, average, or poor.tatistics. Continuous variables are expressed asean � SD and compared with parametric (anal-

sis of variance) and nonparametric (Wilcoxonigned rank and Kruskal-Wallis) tests. Nominalariables are expressed as absolute count and per-entages and compared with the chi-square orisher exact tests. Correlations were assessed with

inear and polynomial regression.Outcomes were assessed with logistic regression to

reate receiver-operator characteristic (ROC) and cal-ulate probability of response for each level of SDI andRS. Optimal cutoffs were selected as the level with

he highest (sensitivity – [1 – specificity]).Odds ratios were derived by dichotomous analy-

is based on traditional values for QRSd (120 ms)nd the observed best cutoff for SDI (10.4%).

Interobserver agreement was assessed with intra-lass correlation coefficient (ICC), linear regression,nd Bland-Altman plots; variability was expresseds mean difference of measurements as well asdjusted for the mean of each pair of measurements.MP version 7 (SAS Institute Inc., Cary, Northarolina) and PASW Statistics version 18 (SPSS

nc., Chicago, Illinois) were used for statisticalnalyses.

E S U L T S

uring the period from 2004 to 2008, 187 patientst KCH and CUHK, underwent primary implan-ation of, or upgrade to, biventricular pacing.wenty-one patients (11.2%) were excluded due to

imited image quality or rhythm disturbances pre-enting acquisition of analyzable 3D datasets. Base-ine and follow-up data were available for 147atients with 19 patients lost to follow-up. Thereere 116 (78.9%) men, and 103 (70%) patients had

schemic heart disease. The time to longestollow-up echo was 7 � 3 months (medianollow-up 6 months, shortest follow-up 2 months).he 2 cohorts had similar baseline QRSd (137 �3 ms vs. 128 � 30 ms for KCH and CHUK,espectively, p � 0.15) but differed in baselineYHA functional class (KCH patients in NYHA

unctional class II � 9 [6.1%] vs. 9 [50%] forHUK, chi-square p � 0.001) and LVEF (21.5 �.3% vs. 27.6 � 7.3%, respectively, p � 0.0012).atients lost to follow-up had similar baselineharacteristics (QRSd: 131 � 18, p � 0.18, LVEF:4 � 9.1%, p � 0.7). Baseline clinical and echo

haracteristics are shown in Table 1.

Cetaat(

ttItidcr9

cmI1(vaveaiOcl3p

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

19

ompliance with guidelines for patient selection. Sev-nty (47.6%) patients did not fulfill strict implan-ation criteria. Of these patients, 24 (16.3%) hadtrial fibrillation; 23 (15.6%) had QRSd �120 ms;nd 9 (6.1%) had NYHA functional class II symp-oms at the time of implant. Seventeen patients11.6%) underwent upgrade of pacemaker to biven-

Figure 1. Histogram of Patients Fulfilling Standard Criteria

The histogram illustrates the proportion of patients fulfilling standaeach exception is shown (some groups overlap). AF-LBBB � proportraditional implantation criteria; AF-Low NYHA � proportion of paticlass I or II; AF-Moderate LVD � proportion of patients in atrial fibrAF-Paced � proportion of patients in atrial fibrillation with paced rtional class I or II; Moderate LVD � left ventricular ejection fractionPPM Upgrade � proportion of patients with paced rhythm. AF � a

Table 1. Baseline Characteristics for Patients Included inthe Study (n � 166)

NYHA functional class 3.0 � 0.5

Male sex 131 (78.9%)

Age (yrs) 66.4 � 12.0

Ischemic LVD 116 (70.1%)

LBBB 112 (67.4%)

QRS duration (ms) 136.7 � 22.0

QRS �120 ms 138 (83%)

Atrial fibrillation 32 (19%)

LVEDV 206.1 � 73.8

LVESV 164.6 � 64.6

LVEF (%) 21.5 � 7.3

SDI (%) 14 � 5.2

Values are mean � SD or n (%).LBBB � left bundle branch block; LVD � left ventricular dysfunction; LVEDV �

left ventricular end-diastolic volume; LVEF � left ventricular ejection fraction;LVESV � left ventricular end-systolic volume; NYHA � New York Heart Associa-tion; SDI � systolic dyssynchrony index.

lar dysfunction; NYHA � New York Heart Association; PPM � permanen

ricular device (4 of whom also had atrial fibrilla-ion). These groups are summarized in Figure 1.nterhospital variability. The 62 datasets pooled fromhe 2 specialist centers (KCH and CUHK) werendependently analyzed at both sites. Five (8.1%) wereeemed to have poor image quality, of which analysisould not be completed in 4 cases (6.5%). Of theemainder, 48 (77.4%) had average image quality and(14.5%) had excellent image quality.There was excellent correlation between both

enters for both LV end-diastolic volume (r � 0.95,ean difference: 4.61 � 23.2 ml, 2.9% variability,

CC: 0.97), LVESV (r � 0.95, mean difference:� 19.6 ml, 1% variability, ICC: 0.98), and LVEF

r � 0.84, mean difference: 2 � 4.9%, 7.1%ariability, ICC: 0.91). There was also very goodgreement in quantification of dyssynchrony withariability for SDI of 7.6% (r � 0.76, mean differ-nce: 0.07 � 3.38%, ICC: 0.84). Linear correlationnd Bland-Altman plots demonstrate the excellentnterhospital agreement (Fig. 2).utcomes from CRT. The overall clinical and echo-ardiographic outcomes were consistent with pub-ished findings. At the latest follow-up (mean 7 �

months following implantation), 116 (78.9%)atients were reported as having subjective im-

riteria. In patients not fulfilling standard criteria the proportion ofof patients in atrial fibrillation that would have otherwise fulfilledin atrial fibrillation with New York Heart Association functionalon with left ventricular ejection fraction between 35% and 45%;m; Low NYHA � patients in New York Heart Association func-ween 35% and 45%; Narrow QRS � QRS duration of �120 ms;fibrillation; LBBB � left bundle branch block; LVD � left ventricu-

rd ctionentsillatihythbettrial

t pacemaker; RBBB � right bundle branch block.

pc(Lrwcr

t(ra0wi0clcLcfis�

gpFPp

iNripQR

ba1s(0tacw6c

tnmsS4Ri

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

20

rovement with reduction in NYHA functionallassification by at least 1 class. Ninety-six patients65.3%) demonstrated a �20% relative increase inVEF, but only 61 (41.6%) demonstrated a �15%

eduction in LVESV. A smaller increase in LVEFas found to more closely reflect NYHA functional

lass changes: 123 (83.6%) of patients had a 10%elative increase in LVEF.

In patients with �20% improvement in LVEF,here was a highly significant reduction in SDI�SDI: 5.9 � 4.3%, p � 0.0001) and a similareduction was also seen in patients with symptom-tic improvement (mean �SDI: 5.9 � 0.4%, p �.0001). In patients exhibiting reverse remodelingith reduction in LVESV, there was also a signif-

cant reduction in SDI (�SDI: 6.1 � 0.66%, p �.009). There was a good correlation betweenhange in LVEF and change in SDI (linear corre-ation coefficient: 0.62). There was good linearorrelation between baseline SDI and increase inVEF (r � 0.55, p � 0.0001), whereas the bestorrelation was found to be quadratic (polynomialt degree � 2, r � 0.67, p � 0.0001), and from thecatter plot, it is obvious this is due to a plateau inLVEF with increasing SDI (Fig. 3).No significant difference in clinical or echocardio-

raphic response was noted between male or femaleatients with or without ischemic heart disease (2-tailisher exact test p � 0.1 for all comparisons).redicting outcomes from CRT. Figures 4 and 5 com-

Figure 2. Interhospital Agreement for 3D Echo

Linear regression (top row) and Bland-Altman graphs (bottom rowend-diastolic volume, left ventricular ejection fraction, and SDI betwHong Kong. HK� Prince of Wales Hospital, Hong Kong; LVEDV � lefraction; SDI � systolic dyssynchrony index; 3D� 3-dimensional.

are QRSd and SDI as predictors of response accord- o

ng to 4 parameters: functional improvement (�1YHA functional class), systolic function (�20%

elative improvement in LVEF), remodeling (�15%mprovement in LVESV), and � 10% relative im-rovement in LVEF. Figure 4 shows scatter plots forRSd and SDI according to response, and in Figure 5,OC has been calculated for each predictor.At baseline, QRSd was not significantly different

etween patients that had functional improvementnd those that did not (QRSd: 137.5 � 25.9 ms vs.33.8 � 25.9 ms, p � 0.43), whereas SDI wasignificantly different between groups at baselineSDI: 15 � 4.7 vs. 10.1 � 5.1, respectively; p �.0001). This is reflected in the ROC curves, wherehe area under the curve (AUC) was 0.52 for QRSdnd 0.79 for SDI. Based on ROC curves, the optimalutoff for SDI in predicting response was 10.4%,hich confers a sensitivity of 90% and a specificity of7% for predicting reduction in NYHA functionallassification by 1 class.

Echocardiographic improvement with �20% rela-ive increase in LVEF was seen in 96 patients (29.2%onresponder rate). Baseline QRSd was 139.5 � 23.4s vs. 133.8 � 22.8 ms for responders and nonre-

ponders, respectively (p � 0.19). Between groups,DI at baseline was significantly different (16.2 �.7% vs. 10.6 � 4.4%, respectively, p � 0.0001). InOC analysis, the AUC for QRSd was 0.57, whereas

t was significantly higher for SDI at 0.86. The

monstrating the agreement in quantification of 3-dimensionalKing’s College Hospital, London, and Prince of Wales Hospital,ntricular end-diastolic volume; LVEF � left ventricular ejection

) deeenft ve

ptimum SDI cutoff for this outcome was 11.4%,

wf

ssc

bc0c

n

ction

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

21

hich has a sensitivity of 91% and specificity of 71%or predicting a significant increase in LVEF after CRT.

Only 41.6% of patients in our cohort demon-trated a �15% reduction in LVESV. Despiteignificant overlap, there was a statistically signifi-ant difference in baseline SDI but not QRSd

Figure 3. Change in SDI and LVEF After CRT

(A) Scatter plot illustrating the change in systolic dyssynchrony indedyssynchrony index difference denote reduction, p � 0.001). (B) Incnization therapy correlates well with reduction in systolic dyssynchrlates well with systolic dyssynchrony index. The red line illustratesline fit) with dark blue areas showing confidence intervals for meaand lower 95% confidence limits for an individual predicted value,diac resynchronization therapy; EF20 � 20% relative increase in eje

Figure 4. Baseline QRS Duration and SDI in Responders and No

Scatter plots for QRS duration (top row) and systolic dyssynchronysponders by New York Heart Association functional class, 20% relatventricular end-systolic volume, and 10% relative increase in left vedifferentiate between groups, whereas systolic dyssynchrony index

in ejection fraction; ESV15 � 15% reduction in end-systolic volume; oth

etween responders and nonresponders to this out-ome (p � 0.02). AUC for QRSd was 0.57, versus.66 for SDI. The optimal SDI cutoff was 11.4%,onferring a sensitivity of 92% and specificity of 43%.

The greatest difference between responders andonresponders was seen in the exploratory measure

responders versus nonresponders (positive values for systolice in left ventricular ejection fraction following cardiac resynchro-index. (C) Increase in left ventricular ejection fraction also corre-r correlation. The blue line illustrates polynomial correlation (beste, and polynomial fits, and lighter blue areas showing the uppercting variation in the error and parameter estimates. CRT � car-fraction; ∆ � difference; other abbreviations as in Figure 2.

ponders

x (bottom row) illustrating distribution in responders and nonre-ncrease in left ventricular ejection fraction, 15% reduction in leftular ejection fraction. In all comparisons, QRS duration did notsignificantly higher in responders. EF10 � 10% relative increase

x inreasonylinean, linrefle

nres

indeive intricwas

er abbreviations as in Figures 1, 2, and 3.

owNsts2pt8Ppiasocd

cwiwbLn

u1fafhr

twwdHbPairSFr0po

actio

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

22

f outcome of a 10% relative increase in LVEF,hich also appears to more closely correlate toYHA response as the optimal SDI cutoff was

imilar (10.7%). Baseline QRSd was similar be-ween groups (p � 0.69), whereas there was aignificant difference in SDI (15.5 � 4.8 vs. 7.9 �.4% for responders and nonresponders, respectively,� 0.0001) (Fig. 4). For SDI, AUC was 0.96, and

he optimal cutoff of 10.7% conferred a sensitivity of9% and specificity of 99% for this outcome.atients not conforming to traditional criteria. A smallroportion of our cohort had persistent AF atmplantation (n � 19), paced rhythm (n � 17),nd/or QRSd �120 ms (n � 33). Although thetudy design does not allow definitive investigation ofutcomes in these groups, we observed similar out-omes to those stated with respect to clinical echocar-iographic and remodeling responses after CRT.In patients with AF, we observed similar out-

omes to patients with normal sinus rhythm. QRSdas similar in both responders and nonresponders

n all outcomes (p � 0.28 for all comparisons),hereas SDI was significantly higher in respondersy improvement in NYHA functional class and inVEF (p � 0.02 and p � 0.008, respectively), but

Figure 5. Predictive Value of Baseline QRS Duration and SDI in

Receiver-operator characteristic curves for prediction of outcome bytive increase in left ventricular ejection fraction, 15% reduction in leventricular ejection fraction. Systolic dyssynchrony index (solid linetion (dotted line) showed no predictive value with area under theLVESV � left ventricular end-systolic volume; TPF � true positive fr

ot by reduction in LVESV (p � 0.3). s

In patients with paced rhythm undergoing CRTpgrade, QRSd was significantly higher (151.2 vs.34.8 ms, p � 0.006), whereas baseline NYHAunctional class, LV end-diastolic volume, LVEF,nd SDI were similar to nonpaced subjects (p � 0.7or all comparisons), although no paced patientsad an SDI of less that 10.4%. The observedesponse rate was similar across all outcomes.

Baseline QRSd was not statistically different be-ween any of these groups (p � 0.22). Baseline SDIas not significantly different in patients with andithout remodeling (p � 0.48) and showed a trend toifference in those with clinical response (p � 0.07).owever, there was a statistically significant difference

etween echo responders (p � 0.009).robabilistic nature of SDI. Using logistic regressionnd ROC curves for predicting a 20% relativencrease in LVEF, the probability of a positiveesponse was calculated for each level of QRSd andDI and was plotted against these parameters (Fig. 6).or SDI, unit odds ratio was 0.236 and range odds

atio �0.0001. For QRSd, the unit odds ratio was.996 and range odds ratio was 0.599. Though thelot is nearly linear for QRS, illustrating that a widef QRSd have the same probability, there is a

comes After CRT

uction in New York Heart Association functional class, 20% rela-entricular end-systolic volume; and 10% relative increase in leftrformed significantly better in all comparisons, whereas QRS dura-e of close to 0.5 for all comparisons. FPF � false positive fraction;n; other abbreviations as in Figures 1, 2, and 3.

Out

redft v) pecurv

igmoid probability distribution for SDI. From this,

itn7dScf7HApcQwh7et(c

pw7it3t

(ai(on

bL

D

TtCfiLaCepno4sldt

vahchFfqpsa

f

nd 3

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

23

t becomes apparent that above an SDI of 10.4%,here is a high probability of responding that doesot increase significantly with higher SDI. Below.5%, there is a low probability of response, whichoes not reduce further with lower SDIs. UsingDI and QRSd as dichotomous variables with autoff of 10.4% and 120 ms, respectively, odds ratiosor improvement in LVEF were 147 (27.47 to86.67) and 0.62 (0.2 to 1.92), respectively.ow would 3D echo have influenced patient selection?pplying selection criteria retrospectively to thisopulation, if only patients meeting traditionalriteria (sinus rhythm and either QRSd �150 ms orRSd �120 with echo evidence of dyssynchrony)ere included, then this would have resulted in aigher response rate (88.3% functional improvement,0.6% systolic function, and 47.1% remodeling) andxcluding 47.6% of the cohort, of which 48 (68.6% ofhose not fulfilling these criteria), 34 (59.6%), and 2035.1%) would have responded by NYHA functionallass, LVEF, and ESV, respectively.

Applying stricter ECG criteria to include onlyatients with sinus rhythm and QRSd �150 msould have resulted in a similar response rate (80%,5%, and 47.2%, respectively) at the cost of exclud-ng 72.8% of the cohort, of which 84 (78.5% ofhose not fulfilling these criteria), 55 (61.8%), and5 (39.3%) would have responded by NHYA func-ional class, LVEF, or ESV, respectively.

Subselection of this cohort using LVMD onlywith an SDI �10.4%), and including patients withtrial fibrillation and all QRSd, would have resultedn a much higher response rate in all outcomes90.3%, 81%, and 51%, respectively), and excludingnly 28% of the cohort, of which 10 (34.8% of those

Figure 6. Probability of a Positive Response to CRT Predicted b

QRS duration (A) does not predict response, as probabilities are simshows a sigmoid probability curve with values of �7.5% corresponrespond to a very high probability; there is a “gray zone” betweenrespond to large changes in probability and is therefore less reliablUsing inverse prediction, the mean and 95% confidence limits for pprobability of positive response (C). Abbreviations as in Figures 2 a

ot fulfilling this criterion) would have responded p

y NYHA functional class. No responders byVEF or LVESV would have been excluded.

I S C U S S I O N

he current study investigates the value of SDI inhe largest cohort so far of patients undergoingRT to be investigated with RT3DE. The keyndings of this study are: 1) 3D quantification ofV volumes, function, and LVMD is reproduciblecross centers; 2) in patients already selected forRT, QRSd is not predictive of either clinical or

chocardiographic outcomes, whereas SDI is highlyredictive; 3) patients with atrial fibrillation orormal QRSd but with high SDI have comparableutcomes to those selected by traditional criteria; and) the occurrence of reverse LV remodeling, as de-cribed by a 15% reduction in LVESV, was muchower than anticipated based on published studies andid not reflect clinical improvement, suggesting thathis may not be an adequate measure of outcome.

In keeping with previous reports, quantification ofolumes and ejection fraction by 3D echo is accuratend highly reproducible, which is reflected in the veryigh interobserver agreement between centers in theurrent study. The large cohort selected for analysisad variable image quality reflecting real-life results.ewer than 10% of cases had poor image quality and

ewer than 15% had excellent image quality. 3Duantification of LV function would therefore appearreferable to other echocardiographic methods forerial clinical evaluation of individual patients as wells quantification of dyssynchrony.

In the current cohort of patients already selectedor CRT, QRSd was found to have no further

S Duration and SDI Plotted Against These Variables

across the range of its values. Systolic dyssynchrony index (B)to very low chance of responding whereas values of �10.5% cor-e values where small changes in systolic dyssynchrony index cor-e histograms along each axis reflect the distribution of subjects.cted systolic dyssynchrony index were estimated for each level of.

y QR

ilardingthese. Thredi

rognostic value, whereas SDI has highly accurate

io1fipv�

t1LhaktbpdiffosataSwps(essa

sbfptsfpS

wasptptfs

sa(pioipdw3Srto

dipu

C

Rifciurc

Rap

R

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

24

n separating responders from nonresponders for allutcomes. The best cutoff for SDI was found to be0.4%, which confers sensitivity �90% and speci-city of �67% for all outcomes after CRT. SDIrovides a probabilistic assessment of outcome withery high probability of positive response at10.4% and very low probability at SDI �7.5%.A recent observational study comparing 16 pa-

ients with dilated cardiomyopathy and LBBB and6 patients with dilated cardiomyopathy withoutBBB to normal subjects found that all patientsad SDI higher than that of normal subjects (4%)nd that SDI was independent of QRSd (31), ineeping with our previous findings (13). The inves-igators concluded that the lack of differentiationetween the 2 groups of heart failure patientsroved SDI was not a useful discriminator ofyssynchrony. This study did not, however, exam-

ne clinical outcomes following CRT, and the cutoffor SDI used was the upper 95% confidence intervalor normal subjects, which is a “normality” thresh-ld, rather than a “response” threshold. Our currenttudy demonstrates that it is not sufficient to havebove-normal intraventricular dyssynchrony, buthat a significantly higher SDI (10.4%) is associ-ted with positive outcomes. A similar cutoff forDI has also been found by Soliman et al. (29),here an SDI �10% was found to be highlyredictive of positive outcome after CRT. Mar-an et al. (24) identified a much lower cutoff6.4%), but analyses were performed on a differ-nt software platform that applies a differentegmentation model to the mathematical repre-entation of the LV cavity, indicating that cutoffsre specific to the software platform used.

Approximately one-fifth of our cohort did nottrictly conform to established selection criteriaut a high SDI indicated a positive outcomeollowing CRT in most cases in patients withre-implantation paced rhythm, atrial fibrilla-ion, or narrow QRS. Current evidence (32,33)uggests that even patients with low NYHAunctional class may benefit, but numbers of suchatients in our cohort were too low to assess.tudy limitations. Data in this study are retrospec-

ure. Circulation 2003;107:1985–90. Trial. JAMA 2003;

ill inevitably influence the accuracy of sensitivitynd specificity, particularly in groups not fulfillingtandard criteria for CRT implantation. It was notossible to obtain data from other modalities, such asissue Doppler, for comparison, as these were noterformed in a controlled manner in a large propor-ion of patients. Functional testing was not availableor the majority of the patients. Applying “what if”cenarios retrospectively is limited by selection bias.

Presence, extent, and localization of myocardialcar tissue has been found to be potentially associ-ted with reduction in LVESV following CRT34–36). Although we found SDI to be a strongerredictor of this outcome after CRT than QRSd,ts predictive value is significantly lower than forther outcomes and perhaps inclusion of scarnformation will strengthen this. During the studyeriod, no patient underwent pre-implantation car-iac magnetic resonance imaging, and, therefore, itas not possible to correlate myocardial scar withD dyssynchrony and clinical outcomes from CRT.imilarly, pacing lead position was not recordedoutinely and, therefore, could not be correlated tohe extent of baseline dyssynchrony and clinicalutcomes.Data in this study were derived from 2 indepen-

ent centers; however, large-scale prospective stud-es are needed to establish whether this highlyromising technique can be translated to routinese for patient selection across the world.

O N C L U S I O N S

eal-time 3D echo provides an accurate, reproduc-ble quantification of LVMD. In patients selectedor CRT, 3D SDI is a superior predictor of out-omes therapy compared with QRSd and has a rolen further evaluating and monitoring patientsndergoing this intervention and potentially has aole in patients not fulfilling traditional selectionriteria.

eprint requests and correspondence: Prof. Mark J. Mon-ghan, Department of Cardiology, King’s College Hos-ital, Denmark Hill, London SE22, United Kingdom.

tive and not powered for subgroup analysis, which E-mail: mark.monaghan@nhs.net.

E F E R E N C E S

1. St. John Sutton MG, Plappert T, Abra-ham WT, et al. Effect of cardiac resyn-chronization therapy on left ventricularsize and function in chronic heart fail-

2. Young JB, Abraham WT, Smith AL,et al. Combined cardiac resynchroni-zation and implantable cardioversiondefibrillation in advanced chronicheart failure: the MIRACLE ICD

289:2685–94.

3. Bristow MR, Saxon LA, Boehmer J,et al. Cardiac-resynchronization ther-apy with or without an implantabledefibrillator in advanced chronic heartfailure. N Engl J Med 2004;350:

2140–50.

1

1

1

1

2

2

2

2

2

2

3

3

3

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

25

4. Cleland JG, Daubert JC, Erdmann E,et al. The effect of cardiac resynchroni-zation on morbidity and mortality inheart failure. N Engl J Med 2005;352:1539–49.

5. Sutton MG, Plappert T, Hilpisch KE,Abraham WT, Hayes DL, ChinchoyE. Sustained reverse left ventricularstructural remodeling with cardiac re-synchronization at one year is a func-tion of etiology: quantitative Dopplerechocardiographic evidence from theMulticenter InSync RandomizedClinical Evaluation (MIRACLE).Circulation 2006;113:266–72.

6. Anand IS, Carson P, Galle E, et al.Cardiac resynchronization therapy re-duces the risk of hospitalizations in pa-tients with advanced heart failure: re-sults from the Comparison of MedicalTherapy, Pacing and Defibrillation inHeart Failure (COMPANION) trial.Circulation 2009;119:969–77.

7. Galizio NO, Pesce R, Valero E, et al.Which patients with congestive heartfailure may benefit from biventricularpacing? Pacing Clin Electrophysiol2003;26:158–61.

8. Pires LA, Abraham WT, Young JB,Johnson KM. Clinical predictors andtiming of New York Heart Associa-tion class improvement with cardiacresynchronization therapy in patientswith advanced chronic heart failure:results from the Multicenter InSyncRandomized Clinical Evaluation(MIRACLE) and Multicenter InSyncICD Randomized Clinical Evaluation(MIRACLE-ICD) trials. Am Heart J2006;151:837–43.

9. Richardson M, Freemantle N, CalvertMJ, Cleland JG, Tavazzi L. Predictorsand treatment response with cardiacresynchronization therapy in patientswith heart failure characterized bydyssynchrony: a pre-defined analysisfrom the CARE-HF trial. EurHeart J 2007;28:1827–34.

0. Marcus GM, Rose E, Viloria EM, etal., for VENTAK CHF/CONTAK-CD Biventricular Pacing Study Investi-gators. Septal to posterior wall motiondelay fails to predict reverse remodelingor clinical improvement in patients un-dergoing cardiac resynchronizationtherapy. J Am Coll Cardiol 2005;46:2208–14.

1. Burri H, Muller H, Vieira I, Lerch R.Poor agreement of echographic mea-sures of ventricular dyssynchrony. EurJ Echocardiogr 2008;9:235–40.

2. Chung E, Leon A, Tavazzi L, et al.Results of the Predictors of Responseto CRT (PROSPECT) Trial. Circu-lation 2008;117:2608–16.

3. Kapetanakis S, Kearney MT, Siva A,Gall N, Cooklin M, Monaghan MJ.

Real-time three-dimensional echocar-diography: a novel technique to quan-

tify global left ventricular mechanicaldyssynchrony. Circulation 2005;112:992–1000.

14. Mannaerts HFJ, Van Der Heide JA,Kamp O, et al. Quantification of leftventricular volumes and ejection frac-tion using freehand transthoracicthree-dimensional echocardiography:comparison with magnetic resonanceimaging. J Am Soc Echocardiogr2003;16:101–9.

15. Jenkins C, Bricknell K, Hanekom L,Marwick TH. Reproducibility and ac-curacy of echocardiographic measure-ments of left ventricular parametersusing real-time three-dimensionalechocardiography. J Am Coll Cardiol2004;44:878–86.

16. Prakash K, Li X, Hejmadi A, Hashi-moto I, Sahn DJ. Determination ofasymmetric cavity volumes usingreal-time three-dimensional echo-cardiography: an in vitro balloonmodel study. Echocardiography2004;21:257– 63.

17. Sugeng L, Mor-Avi V, Weinert L, etal. Quantitative assessment of left ven-tricular size and function: side-by-sidecomparison of real-time three-dimensional echocardiography andcomputed tomography with magneticresonance reference. Circulation 2006;114:654–61.

18. van den Bosch AE, Robbers-VisserD, Krenning BJ, et al. Comparisonof real-time three-dimensionalechocardiography to magnetic reso-nance imaging for assessment of leftventricular mass. Am J Cardiol2006;97:113–7.

19. Jaochim Nesser H, Sugeng L, CorsiC, et al. Volumetric analysis of re-gional left ventricular function withreal-time three-dimensional echocar-diography: validation by magnetic res-onance and clinical utility testing.Heart 2007;93:572–8.

20. Pouleur A-C, le Polain de WarouxJ-B, Pasquet A, et al. Assessment ofleft ventricular mass and volumes bythree-dimensional echocardiographyin patients with or without wall mo-tion abnormalities: comparisonagainst cine magnetic resonance imag-ing. Heart 2008;94:1050–7.

21. Nesser H-J, Mor-Avi V, Gorissen W,et al. Quantification of left ventricularvolumes using three-dimensionalechocardiographic speckle tracking:comparison with MRI. Eur Heart J2009;30:1565–73.

22. Hare JL, Jenkins C, Nakatani S,Ogawa A, Yu C-M, Marwick TH.Feasibility and clinical decision-making with 3D echocardiography inroutine practice. Heart 2008;94:440–5.

23. Zhang Q, Yu C-M, Fung JW-H, etal. Assessment of the effect of cardiac

resynchronization therapy on intra-ventricular mechanical synchronicityby regional volumetric changes. Am JCardiol 2005;95:126–9.

4. Marsan NA, Bleeker GB, YpenburgC, et al. Real-time three-dimensionalechocardiography permits quantifica-tion of left ventricular mechanical dys-synchrony and predicts acute responseto cardiac resynchronization therapy.J Cardiovasc Electrophysiol 2008;19:392–9.

5. Marsan NA, Henneman MM, ChenJ, et al. Real-time three-dimensionalechocardiography as a novel approachto quantify left ventricular dyssyn-chrony: a comparison study withphase analysis of gated myocardialperfusion single photon emissioncomputed tomography. J Am SocEchocardiogr 2008;21:801–7.

6. Marsan NA, Henneman MM, ChenJ, et al. Left ventricular dyssynchronyassessed by two three-dimensional im-aging modalities: phase analysis ofgated myocardial perfusion SPECTand tri-plane tissue Doppler imaging.Eur J Nucl Med Mol Imaging 2008;35:166–73.

7. van Dijk J, Knaapen P, Russel IK, etal. Mechanical dyssynchrony by 3Decho correlates with acute haemody-namic response to biventricular pacingin heart failure patients. Europace2008;10:63–8.

8. Liodakis E, Al Sharef O, DawsonD, Nihoyannopoulos P. The use ofreal time three dimensional echo-cardiography for assessing mechani-cal synchronicity. Heart 2009;95:1865–71.

9. Soliman OII, Geleijnse ML, TheunsDAMJ, et al. Usefulness of left ven-tricular systolic dyssynchrony by real-time three-dimensional echocardiog-raphy to predict long-term response tocardiac resynchronization therapy.Am J Cardiol 2009;103:1586–91.

0. Soliman OI, van Dalen BM, NemesA, et al. Quantification of left ventric-ular systolic dyssynchrony by real-timethree-dimensional echocardiography.J Am Soc Echocardiogr 2009;22:232–9.

1. Sonne C, Sugeng L, Takeuchi M, etal. Real-time 3-dimensional echo-cardiographic assessment of left ven-tricular dyssynchrony: pitfalls in pa-tients with dilated cardiomyopathy.J Am Coll Cardiol Img 2009;2:802–12.

2. Moss AJ, Hall WJ, Cannom DS,et al., for Investigators M-CT.Cardiac-resynchronization therapy forthe prevention of heart-failure events.

N Engl J Med 2009;361:1329–38.

3

3

Krb

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 4 , N O . 1 , 2 0 1 1

J A N U A R Y 2 0 1 1 : 1 6 – 2 6

Kapetanakis et al.

3D Echo in CRT

26

3. Daubert C, Gold MR, Abraham WT,et al. Prevention of disease progressionby cardiac resynchronization therapy inpatients with asymptomatic or mildlysymptomatic left ventricular dysfunc-tion: insights from the European cohortof the REVERSE (ResynchronizationReverses Remodeling in Systolic LeftVentricular Dysfunction) trial. J AmColl Cardiol 2009;54:1837–46.

4. Ypenburg C, Schalij MJ, Bleeker

GB, et al. Impact of viability and

scar tissue on response to cardiacresynchronization therapy in ischae-mic heart failure patients. EurHeart J 2007;28:33– 41.

35. Ypenburg C, Roes SD, Bleeker GB,et al. Effect of total scar burden oncontrast-enhanced magnetic reso-nance imaging on response to cardiacresynchronization therapy. Am J Car-diol 2007;99:657–60.

36. Taylor AJ, Elsik M, Broughton A, et

al. Combined dyssynchrony and scar d

imaging with cardiac magnetic reso-nance imaging predicts clinical re-sponse and long-term prognosis fol-lowing cardiac resynchronizationtherapy. Europace 2010;12:708–13.

ey Words: cardiacesynchronization therapy y leftundle branch block y 3-

imensional echocardiography.