feasibility of diastolic function assessment with cardiac ct · the cardiac cycle). for each phase,...

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 . 3 , 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 . 1 1 . 0 1 7

Feasibility of Diastolic Function AssessmentWith Cardiac CTFeasibility Study in Comparison With Tissue Doppler Imaging

Mark J. Boogers, MD,*† Jacob M. van Werkhoven, MSC,*† Joanne D. Schuijf, PHD,*Victoria Delgado, MD,* Heba M. El-Naggar, MD,* Eric Boersma, PHD,‡Gaetano Nucifora, MD,* Rob J. van der Geest, MSC,§ Bernard P. Paelinck, MD, PHD,�Lucia J. Kroft, MD, PHD,§ Johan H. C. Reiber, PHD,§ Albert de Roos, MD, PHD,§Jeroen J. Bax, MD, PHD,* Hildo J. Lamb, MD, MSC, PHD§

Leiden, Utrecht, and Rotterdam, the Netherlands; and Antwerp, Belgium

O B J E C T I V E S This study aimed to demonstrate the feasibility of multidetector row computed

tomography (CT) for assessment of diastolic function in comparison with 2-dimensional (2D) echocar-

diography using tissue Doppler imaging (TDI).

B A C K G R O U N D Diastolic left ventricular (LV) function plays an important role in patients with

cardiovascular disease. 2D echocardiography using TDI has been used most commonly to evaluate

diastolic LV function. Although the role of cardiac CT imaging for evaluation of coronary atherosclerosis

has been explored extensively, its feasibility to evaluate diastolic function has not been studied.

M E T H O D S Patients who had undergone 64-multidetector row CT and 2D echocardiography with

TDI were enrolled. Diastolic function was evaluated using early (E) and late (A) transmitral peak velocity

(cm/s) and peak mitral septal tissue velocity (Ea; cm/s). Peak transmitral velocity (cm/s) was calculated

by dividing peak diastolic transmitral flow (ml/s) by the corresponding mitral valve area (cm2). Mitral

septal tissue velocity was calculated from changes in LV length per cardiac phase. Subsequently, the

estimation of LV filling pressures (E/Ea) was determined.

R E S U L T S Seventy patients (46 men; mean age 55 � 11 years) who had undergone cardiac CT and

2D echocardiography with TDI were included. Good correlations were observed between cardiac CT and

2D echocardiography for assessment of E (r � 0.73; p � 0.01), E/A (r � 0.87; p � 0.01), Ea (r � 0.82;

p � 0.01), and E/Ea (r � 0.81; p � 0.01). Moreover, a good diagnostic accuracy (79%) was found for

detection of diastolic dysfunction using cardiac CT. Finally, the study showed a low intraobserver and

interobserver variability for assessment of diastolic function on cardiac CT.

C O N C L U S I O N S Cardiac CT imaging showed good correlations for transmitral velocity, mitral

septal tissue velocity, and estimation of LV filling pressures when compared with 2D echocardiography.

Additionally, cardiac CT and 2D echocardiography were comparable for assessment of diastolic

dysfunction. Accordingly, cardiac CT may provide information on diastolic dysfunction. (J Am Coll

Cardiol Img 2011;4:246–56) © 2011 by the American College of Cardiology Foundation

From the *Department of Cardiology, Leiden University Medical Center, Leiden, the Netherlands; †Interuniversity CardiologyInstitute of the Netherlands, Utrecht, the Netherlands; ‡Department of Epidemiology and Statistics, Erasmus University,Rotterdam, the Netherlands; §Department of Radiology, Leiden University Medical Center, Leiden, the Netherlands; and the�Department of Cardiology, Antwerp University Hospital, Antwerp, Belgium. Dr. Boogers is supported by the Dutch HeartFoundation (grant 2006T102). Dr. van Werkhoven is financially supported by the Netherlands Society of Cardiology. Dr.Delgado has consulted for St. Jude Medical. Dr. Nucifora received a research grant from the European Society of Cardiology.Dr. van der Geest is a consultant for Medis Medical Imaging Systems. Dr. Bax received research grants from Medtronic, BostonScientific, Biotronik, Edwards Lifesciences, Bristol-Myers Squibb Medical Imaging, St. Jude Medical, and GE Healthcare. Allother authors have reported that they have no relationships to disclose.

Manuscript received June 18, 2010; revised manuscript received November 22, 2010, accepted November 29, 2010.

D

uiif2fi

dmowtr

r1aoToTtdat

asc

tissue Doppler imaging

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 . 3 , 2 0 1 1

M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.

Diastolic Heart Function Assessed With Cardiac CT

247

iastolic left ventricular (LV) function playsan important role in the evaluation ofclinical symptoms, therapeutic options,

and prognosis in patients with cardiovascu-lar disease (1,2). More specifically, it has beenshown that diastolic dysfunction represents an im-portant pathological condition in patients withcoronary artery disease (CAD) (3,4).

See page 257

Doppler echocardiography represents the mostcommonly used approach for evaluation of diastolicfunction (5–10). For the evaluation of diastolicfunction, transmitral velocity has been used fre-quently as a noninvasive alternative to directlymeasured LV filling pressures (5,7). However, it isimportant to note that several confounding factorsmay influence transmitral velocity, and conse-quently, transmitral velocity alone may not be thebest marker for diastolic LV dysfunction (6,7,11).Combined assessment of early (E) peak transmitralvelocity and early peak mitral septal tissue velocity(Ea) may be more accurate for the evaluation ofdiastolic LV function, predominantly in patientswith depressed or increased LV filling pressures(6,10,11). Cardiac computed tomography (CT) hasemerged as a potent noninvasive imaging modalityfor the evaluation of coronary atherosclerosis(12,13). In specific subsets of patients, multiphaseCT imaging may be indicated to ensure diagnosticimage quality for visualization of the coronaryarteries.

Thus far, multiphase CT studies have beenrestricted to LV systolic function analysis (14), andno information is available on the feasibility ofcardiac CT imaging to assess diastolic LV function.Accordingly, the present study aimed to evaluatethe feasibility of cardiac CT for assessment ofdiastolic function in a direct comparison with2-dimensional (2D) echocardiography using tissueDoppler imaging (TDI).

M E T H O D S

Patient population and study design. Seventy consec-tive patients who had been referred for 64-CTmaging were retrospectively selected from our clin-cal registry. Cardiac CT imaging had been per-ormed to evaluate known or established CAD, andD echocardiography with TDI had been per-ormed to evaluate therapeutic options. Both exam-
nations had been performed sequentially, in ran-
om order. Known CAD was defined as previousyocardial infarction, revascularization, or evidence

f CAD on previous diagnostic tests. Patientsithout evidence of CAD on previous diagnostic

ests were suspected to have CAD (and thereforeeferred for CT angiography).

Patients were chosen from our ongoing clinicalegistry if they met the following selection criteria:) presence of multiphase CT imaging (informationcquired of the entire cardiac cycle); 2) availabilityf multiphase CT and 2D echocardiography withDI within 3 months; 3) diagnostic image qualityf multiphase CT and 2D echocardiography withDI; 4) sinus rhythm; and 5) absence of valvulopa-

hy (aortic or mitral valvulopathy, grade �I). Ad-itionally, patients with unstable angina pectoris orcute coronary syndrome were excluded from fur-her analysis.

Our institutional review board does not requirepproval for retrospective technical analy-es of clinically obtained data, as was thease in this study.Cardiac CT data acquisition and analysis.Multidetector row CT imaging was per-formed with a 64-slice CT scanner (Aquil-ion 64, Toshiba Medical Systems, Otawara,Japan). Before MDCT imaging, patientswere monitored for blood pressure and heartrate. Patients with heart rates �65 beats/min were given metoprolol 50 or 100 mgorally, unless contraindicated. Patients didnot receive nitroglycerin before cardiac CTimaging.

For the contrast-enhanced helical scan,collimation was 64 � 0.5 mm with a rota-tion time of 400 ms. Tube current and voltage were350 mA and 120 kV, respectively. At an injection rateof 5 ml/min, 95 to 130 ml of nonionic contrastmedium (Iomeron 400, Bracco, Milan, Italy) wasinfused in the antecubital vein. After the start ofcontrast infusion, recurrent low-dose examinationswere performed to monitor contrast arrival within theregion of interest, in the descending aorta. An elec-trocardiogram (ECG) was started simultaneously forretrospective gating of the data. The entire cardiaccycle was scanned without dose modulation. TheECG-gated helical scan was automatically triggeredonce the pre-determined threshold level of baseline �100 Hounsfield units was reached. After a preset delayof 2 s, scanning was performed during an inspiratorybreath hold of 8 to 12 s.

Data were reconstructed with a slice thickness of

A B B

A N D

A � p

atrial

CAD �

CMR �

reson

CT �

E � p

early

Ea �

veloci

LV �

TDI �

1 mm and reconstruction interval of 1 mm.

R E V I A T I O N S

A C R O N YM S

eak transmitral velocity at

contraction

coronary artery disease

cardiac magnetic

ance

computed tomography

eak transmitral velocity in

diastole

peak mitral septal tissue

ty in early diastole

left ventricle/ventricular

With

tlsFtsvps4sfa4tat

lptvpwsLvmFEv

mibriI0

opl24sf


M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.


248

the use of half reconstruction algorithms, the actualtemporal resolution was 200 ms. Segmented recon-struction algorithms yielded a temporal resolutionof up to 50 ms, depending on the actual imagingacquisition conditions (pitch, rotation time, andheart rate). An ECG-gated post-processing soft-ware was used to reconstruct data in short-axisorientation. Images were reconstructed at 20 inter-vals (0% to 95% of the R-R interval) and transferredto a separate workstation with dedicated cardiacfunction analysis software (Mass V2008-EXP,LKEB, Leiden, the Netherlands) (15). Contrast-enhanced scans were analyzed by an independentobserver who was blinded to all other data. Signif-icant coronary artery stenosis was defined as �50%luminal narrowing, whereas nonsignificant stenosiswas defined as �50% luminal narrowing.

The evaluation of LV diastolic function wasbased on the assessment of transmitral velocity (15min per patient) and mitral septal tissue velocity (5min per patient). Accordingly, the assessment ofLV diastolic function was performed within 20 minper patient.Cardiac CT transmitral velocity. Peak transmitral ve-locity (cm/s) was measured in E and late (A)diastole. The peak value represents the highestmean value of the measurements obtained during Eand A diastole. Late peak transmitral velocity wasmeasured at atrial contraction. Transmitral velocitymeasurements were based on several processingsteps (Fig. 1). At first, LV volumes were calculatedfor 20 cardiac phases (each phase represented 5% ofthe cardiac cycle). For each phase, automatic con-tour detection was performed on 1-mm slicedreconstructed short-axis images ranging from mitralvalve annulus to the apex (Fig. 1A, left panel).Manual corrections could be made to improvecontour detection. Papillary muscles were regardedas part of the LV cavity and were included in theLV volume analyses (16). Automatic contour de-tection was performed using dedicated in-housedeveloped Mass research software package (MassV2008-EXP, LKEB, Leiden, the Netherlands)(15). Next, LV volumes were plotted in a volumeversus time curve (Fig. 1A, right upper curve). Inaddition, changes in LV volumes between 2 con-secutive phases (first derivative) were derived andused to calculate the transmitral flow (ml/s) perphase (Fig. 1A, right lower curve). Subsequently,the maximal transmitral flow in E and A diastolewas derived using the transmitral flow versus timecurve. To allow direct comparison with 2D echo-
cardiography, the maximal transmitral flow in E l
and A diastole was divided by their correspondingmitral valve area (which was measured during E andA diastole, as described below), yielding E and Apeak transmitral velocities and the E/A.Cardiac CT mitral septal tissue velocity. Myocardialissue velocity (cm/s) was measured at the septalevel of the mitral valve annulus attachment. Mea-urements of Ea (cm/s) are illustrated in Figure 1B.or 20 phases, LV length (mm) was calculated as

he distance between 2 anatomic markers: 1) mitraleptal annulus (annular attachment of septal mitralalve leaflet) and 2) cardiac apex (Fig. 1B, leftanel). Anatomic markers were positioned at recon-tructed 4-chamber views. Reconstruction of a-chamber view was based on several reconstructionteps. At first, a 2-chamber view was reconstructedrom axial slices, directing the image slice (cardiacxis) through the cardiac apex. Consecutively, a-chamber view was reconstructed by positioninghe image slice at two-thirds of the mitral valvennulus (perpendicular to the interventricular sep-um) using the 2-chamber view.

The LV length per phase was plotted in an LVength versus time curve (Fig. 1B, right upperanel). Changes in LV length between 2 consecu-ive phases were calculated and used to generate aelocity versus time curve (Fig. 1B, right loweranel). In this curve, mitral septal tissue velocitiesere plotted against time. For each phase, mitral

eptal tissue velocity was computed using changes inV length and heart rate. The maximal tissueelocity during E diastole represented Ea. Measure-ents were performed using Mass software (15).inally, the estimation of LV filling pressures (E/a) was calculated by dividing early transmitral

elocity by mitral septal tissue velocity.Cardiac CT mitral valve area. Mitral valve area was

easured to enable direct comparison of volumetricndexes derived from cardiac CT with velocity-ased parameters as assessed with 2D echocardiog-aphy. In Figure 1C, the processing steps involvedn mitral valve area measurements are illustrated.mages were reconstructed with a slice thickness of.5 mm and a reconstruction interval of 0.3 mm.Mitral valve area measurements were determined

n the basis of different steps: the LV axis wasositioned perpendicular to mid–mitral valve annu-us on sagittal and coronal views, yielding a-chamber view (Fig. 1C, panel 1). Subsequently, a-chamber view (Fig. 1C, panel 2) was recon-tructed, and manual contour detection was per-ormed at the most distal level of the mitral valve
eaflets in short-axis views (Fig. 1C, panels 3 and 4).


M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.


249

Figure 1. Diastolic LV Function Assessed With MDCT

(A) Transmitral flow: left ventricular (LV) volumes (ml) were measured for 20 phases per cardiac cycle, using short-axis images by outliningendocardial contours in each phase. The volumes were plotted in a volume versus time curve (right upper). These curves were used to definethe diastole, ranging from end-systolic (ES) to end-diastolic (ED) phase. Changes in LV volumes between 2 consecutive phases were plottedagainst time (transmitral flow vs. time curve) (right lower). Subsequently, early and late peak transmitral flows (ml/s) were derived. (B) Mitralseptal tissue velocity: anatomic markers were positioned at the mitral septal annulus (MS) and cardiac apex (AP). The LV length (mm; distancebetween anatomic markers) was calculated for each phase (left) and plotted in a LV length versus time curve (right upper). Next, changes inLV length between 2 consecutive phases were calculated. Based on these numbers, mitral septal tissue velocities (cm/s) were calculated foreach phase (velocity vs. time curve) (right lower). The early peak mitral septal tissue velocity (cm/s) represented the maximal tissue velocityduring early diastole. (C) Mitral valve area: measurements were taken at the most distal level of the mitral valve leaflets (smallest mitral valvearea) using reconstructed images at peak early and late transmitral velocities. The LV axis was positioned perpendicular to mid–mitral valveannulus on sagittal and coronal views, yielding a 2-chamber view (panel 1). Then the 4-chamber view was reconstructed (panel 2), and themitral valve area was measured at the tip of the leaflets (panels 3 and 4) on short-axis views. For direct comparison, transmitral velocity (cm/s)was calculated using the following formula: peak diastolic transmitral flow divided by the corresponding mitral valve area. MDCT � multidetec-
tor row computed tomography.

tasfw1so

pv1saepm

pwrttiom


M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.


250

Measurements were taken during early and latepeak transmitral flow using a dedicated workstation(Vitrea 2, Vital Images, Minnetonka, Minnesota).

The mitral valve area was calculated to enabledirect comparison with 2D echocardiography.Transmitral velocity was calculated by dividingtransmitral flow by corresponding mitral valve area(Fig. 1).Intraobserver and interobserver reproducibility. In-raobserver and interobserver reproducibility forssessment of transmitral velocity was evaluated in aubset of 15 patients who were randomly selectedrom the patient population. Transmitral velocityas measured twice by the same observer in these5 patients. Subsequently, transmitral velocity mea-urements were performed by a second independentbserver in the same subset of patients.In addition, intraobserver and interobserver re-

roducibility for assessment of mitral septal tissueelocity was evaluated in another random sample of5 patients. Mitral septal tissue velocity was mea-ured twice by the same observer in these patientsccording to the standardized protocol, as describedarlier. Additionally, a second independent observererformed the mitral septal tissue velocity measure-ents in the same subset of patients.Finally, the intraobserver and interobserver re-

roducibility of the mitral valve area measurementsas evaluated in a subset of 20 patients who were

andomly selected from the patient population. Inhese patients, the mitral valve area was measuredwice by the same observer using the same process-ng steps, as described earlier. Furthermore, a sec-nd blinded observer performed the mitral valueeasurements in the same subset of 20 patients.

Transthoracic 2D echocardiography using TDIacquisition. Transthoracic 2D echocardiographywas performed in left lateral decubitus positionusing a commercially available system (VingmedVivid-7, General Electric, Horten, Norway). Stan-dard parasternal (long- and short-axis) and apicalviews (2- and 4-chamber) were obtained. In addi-tion, continuous-wave and pulsed-wave Dopplerexaminations were performed. From the 4-chamberview, TDI was obtained with color Doppler framerates exceeding 115 frames/s, depending on thesector width of the range of interest. Aliasingvelocities varied between 16 and 32 cm/s andresulted from pulse repetition frequencies rangingfrom 500 to 1,000 Hz. The echocardiographicanalyses were performed by an independent and
blinded observer.
2D echocardiography transmitral velocity. Transmi-tral velocity was recorded at the end of respiratoryexpiration (Fig. 2, upper panel). Transmitral veloc-ity measurements were taken using dedicated offlinesoftware (EchoPAC 7.0.0, General Electric). Stan-dard pulsed-wave Doppler imaging was performedto assess E and A peak transmitral velocities. E andA peak transmitral velocities were used to calculatethe E/A. Doppler sample volume was placed at thetip of the mitral valve leaflets, on a 4-chamber view(17). Subsequently, E and A peak transmitral ve-locities were obtained in diastole.2D echocardiography mitral septal tissue velocity.Early peak mitral septal tissue velocity was assessedusing color-coded TDI on a 4-chamber view. Im-ages were obtained in end-expiration in a patient inleft lateral decubitus position. Doppler velocitieswere measured from the apical 4-chamber view

Figure 2. Evaluation of Diastolic Function With 2DEchocardiography Using TDI

(A) Two-dimensional (2D) echocardiographic assessment ofpulsed-wave Doppler of early (E) transmitral velocity (cm/s)(white arrow). (B) Early diastolic peak mitral septal tissue veloc-ity (Ea) (cm/s) (white arrow) at basal septal segment by tissueDoppler imaging (TDI).

using a 6 � 6 mm sample volume positioned at the

tdcdigpa(�pndp

maSoe(waawap

rlccttCTsiaDdadpCw

dfewIcasS

ddpan(

iCfcN


M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.


251

basal septal mitral valve annulus, as illustrated inFigure 2 (lower panel) (17).

Color-coded images from 3 consecutive heart-beats were analyzed using dedicated offline software(EchoPAC 7.0.0., General Electric). Reliable TDIcurves were obtained in 67 patients.Detection of diastolic dysfunction. For evaluation ofhe accuracy of cardiac CT to detect diastolicysfunction, diastolic function was graded in 4ategories using the following criteria: 1) normaliastolic function (�1 E/A �2 and E/Ea �8); 2)

mpaired relaxation pattern (diastolic dysfunctionrade I; E/A �1 and E/Ea �8); 3) pseudonormalattern (diastolic dysfunction grade II; �1 E/A �2nd �9 E/Ea �12); and 4) restrictive filling patterndiastolic dysfunction grade III; E/A �2 and E/Ea13) (18). Based on these criteria, the patient

opulation was divided into 2 groups: patients withormal diastolic function and patients with diastolicysfunction (including impaired LV relaxation andseudonormal and restrictive LV filling patterns).

Statistical analysis. Continuous data are presented asean � SD, and categoric data are presented as

bsolute numbers or percentages. Kolmogorov-mirnov tests were used to evaluate the distributionf the data. All variables were normally distributedxcept for the estimation of LV filling pressuresE/Ea) on cardiac CT and 2D echocardiographyith TDI. Data for E/Ea were presented as medi-

ns and 25th and 75th percentiles. When appropri-te, paired t tests or Wilcoxon signed rank testsere used to compare diastolic function parameters

s derived from cardiac CT and 2D echocardiogra-hy with TDI.Comparison of cardiac CT and 2D echocardiog-

aphy with TDI was performed using Pearsoninear regression analysis or the Spearman rhoorrelation. The 95% limits of agreement werealculated using Bland-Altman analysis that plottedhe mean value of differences of each pair againsthe average value of a similar pair of data. CardiacT was subtracted from 2D echocardiography withDI because the latter was considered the clinical

tandard. Reproducibility was evaluated by calculat-ng the intraclass correlation coefficients (ICC), andn excellent agreement was defined as ICC � 0.8.iagnostic accuracy of cardiac CT for detection of

iastolic dysfunction was assessed using a binarypproach; normal diastolic function and diastolicysfunction (including impaired LV relaxation andseudonormal and restrictive LV filling patterns).orresponding sensitivity and specificity values
ere calculated. For these values, the 95% confi- b
ence intervals (CIs) were calculated using theollowing formula: p � 1.96 � SE. The SE wasstimated by �(p[1�p]/n). Statistical analysesere performed with SPSS (version 16.0, SPSS

nc., Chicago, Illinois). A value of p � 0.05 wasonsidered statistically significant. Bland-Altmannalyses were performed with GraphPad Prismoftware (version 5.01, GraphPad Software Inc.,an Diego, California).

R E S U L T S

Patient population. A total of 80 patients had un-ergone multiphase CT imaging and 2D echocar-iography with TDI within 3 months. Of these 80atients, 10 patients were excluded owing to thebsence of sinus rhythm (n � 2) or the presence ofondiagnostic image quality of either cardiac CTn � 7) or 2D echocardiography with TDI (n � 1).

Accordingly, 70 patients (46 men [66%]; mean age55 � 11 years) were included. Baseline character-stics of the patient population are listed in Table 1.ardiac CT and 2D echocardiography were per-

ormed within 3 months, during which no acuteoronary events or worsening of angina occurred.o changes in the use of medication occurred

Table 1. Baseline Characteristics of Study Population (N � 70)

Men 46 (66)

Age (yrs) 55 � 11

Suspected CAD 58 (83)

Known CAD 12 (17)

Significant coronary stenosis 21 (30)

Cardiovascular risk factors

Diabetes mellitus 35 (50)

Systemic hypertension 43 (61)

Hypercholesterolemia 40 (57)

Current smoking 11 (16)

Positive family history 27 (39)

Medication use

Beta-blockers 24 (34)

ACEI/AT II antagonists 35 (50)

Statins 29 (41)

Diuretics 15 (21)

Anticoagulants 30 (43)

Cardiac CT

Heart rate (beats/min) 58 � 10

LV end-systolic volume (ml) 70 � 50

LV end-diastolic volume (ml) 149 � 52

LV ejection fraction (%) 56 � 13

Data are presented as mean � SD or n (%).ACEI � angiotensin-converting enzyme inhibitor; AT � angiotensin; CAD �coronary artery disease; CT � computed tomography; LV � left ventricular.

etween both examinations. The mean duration

d

ce5

cfmtcwvl(

velocity in early diastole; E


M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.


252

between cardiac CT and 2D echocardiography was34.1 � 25.0 days. Clinical referral for cardiac CTwas based on suspected CAD in 58 patients andknown CAD in 12 patients. Patients with knownCAD included patients with previous myocardialinfarction (n � 10), percutaneous coronary inter-vention (n � 7), and indications of CAD on earlier

iagnostic tests (n � 3). Significant coronary arterystenosis (�50% luminal narrowing) was reported in21 patients (30%). In total, 31 (44%) patientsreceived beta-blocking therapy before cardiac CTimaging.Transmitral velocity. Transmitral flow versus timeurves were obtained in all patients. Mean LVnd-systolic and end-diastolic volumes were 70 �0 ml and 149 � 52 ml, respectively, on cardiac

CT. Accordingly, the mean LV ejection fractionwas 56% � 13% (Table 1). Mean values for E andA peak transmitral velocities are shown in Table 2.The mean diastolic transmitral velocity was 24.3 �8.5 cm/s. Pearson correlation showed a good cor-

Figure 3. Comparison Between 2D Echocardiography and MDCT

(A) Good correlation was observed between both techniques. (B) Blan

ction Parameters for Cardiac CT and 2D Echocardiography

Cardiac CT 2D Echocardiography

59.0 � 16.6 61.8 � 14.5*

56.2 � 17.4 64.8 � 18.2*

1.1 � 0.4 1.1 � 0.5*

ocity

6.6 � 2.7 6.2 � 2.3*

8.8 (6.4–13.1) 9.7 (7.5–14.0)**

an � SD or median (25th–75th percentiles). Data for E/Ea are presented asng 25th and 75th percentiles. *Paired t test showed a value of p � 0.05.st showed a value of p � 0.05.ocity at atrial contraction; CT � computed tomography; E � peak transmitrala � peak mitral septal tissue velocity in early diastole; 2D � 2-dimensional.

average value of the same pair of values. Dotted lines represent 95% limit

relation for E (r � 0.73; p � 0.01) and E/A (r �0.87; p � 0.01) (Figs. 3A and 4A). Bland-Altmananalysis for E showed a mean difference of 2.4 �12.0 cm/s, with 95% limits of agreement rangingfrom �21.2 to 26.0 cm/s, whereas the mean valueof difference was �0.1 � 0.2 cm/s for E/A, with95% limits of agreement ranging from �0.5 to 0.4cm/s (Figs. 3B and 4B).

A low intraobserver and interobserver variabilitywas observed for assessment of E (ICC: 0.97, 95%CI: 0.91 to 0.99, and ICC: 0.93, 95% CI: 0.78 to0.97, respectively) and A (ICC: 0.98, 95% CI: 0.94to 0.99, and ICC: 0.91, 95% CI: 0.51 to 0.98,respectively) peak transmitral velocities. Moreover,the assessment of E/A showed a low intraobserverand interobserver variability (ICC: 0.95, 95% CI:0.84 to 0.98, and ICC: 0.95, 95% CI: 0.85 to 0.98,respectively).

Finally, the intraobserver and interobserver re-producibility of mitral valve area measurements wasevaluated. In a subset of 20 patients, an excellentintraobserver and interobserver reproducibility wasobserved for assessment of mitral valve area (ICC:0.94, 95% CI: 0.83 to 0.98, and ICC: 0.92, 95% CI:0.80 to 0.97, respectively).Mitral septal tissue velocity. Velocity versus timeurves were obtained for all patients. Mean valuesor Ea are shown in Table 2. In addition, theedians and corresponding 25th and 75th percen-

iles for E/Ea are shown in Table 2. A goodorrelation (r � 0.82; p � 0.01) (Fig. 5A) for Eaas found. Bland-Altman analysis showed a meanalue of difference of �0.5 � 1.6 cm/s, with 95%imits of agreement ranging from �3.6 to 2.5 cm/sFig. 5B). In addition, good correlation (r � 0.81;

Assessment of Early Maximal Diastolic Transmitral Velocity

tman analysis showing the difference of each pair plotted against the

for

d-Al

Table 2. Diastolic Fun

Transmitral velocity

E (cm/s)

A (cm/s)

E/A

Mitral septal tissue vel

Ea (cm/s)

E/Ea

Data are presented as memedians and correspondi**Wilcoxon signed rank teA � peak transmitral vel

s of agreement. Abbreviations as in Figures 1 and 2.

ml(

ps0tI9


M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.


253

p � 0.01) (Fig. 6A) was reported for E/Ea, with aean value of difference of 1.0 � 2.9 cm/s and 95%

imits of agreement ranging from �4.6 to 6.7 cm/sFig. 6B).

An excellent intraobserver and interobserver re-roducibility was observed for assessment of mitraleptal tissue velocity (ICC: 0.95, 95% CI: 0.85 to.98, and ICC: 0.89, 95% CI: 0.69 to 0.96, respec-ively) and estimation of LV filling pressures (E/Ea;CC: 0.96, 95% CI: 0.87 to 0.99, and ICC: 0.93,5% CI: 0.80 to 0.98, respectively).

Detection of diastolic dysfunction. The diagnostic ac-curacy of cardiac CT to detect diastolic dysfunction incomparison with 2D echocardiography with TDI wascalculated (18). In total, 19 patients (27%) showednormal diastolic function, whereas 51 patients (73%)


(A) Good correlation was observed between both techniques. (B) Bagainst the average value of the same pair of values. Dotted linesatrial contraction; other abbreviations as in Figures 1 and 2.


(A) Good correlation was observed between both imaging techniques
against the average value of the same pair of values. Dotted lines represe
showed diastolic dysfunction using 2D echocardiog-raphy. Of the patients with diastolic dysfunction on2D echocardiography, 40 patients were scored simi-larly using cardiac CT, yielding a sensitivity of 78%(95% CI: 67% to 89%). Normal diastolic function wasfound in 15 of the 19 patients using cardiac CT,yielding a specificity of 79% (95% CI: 61% to 97%).Overall, diagnostic accuracy for assessment of diastolicdysfunction was 79% (95% CI: 69% to 89%).

D I S C U S S I O N

This study demonstrated good correlations fortransmitral velocity (E and E/A) and mitral septaltissue velocity (Ea). Additionally, combined assess-ment of transmitral and mitral septal tissue velocity

E/A

-Altman analysis showing the difference of each pair plottedesent 95% limits of agreement. A � peak transmitral velocity at

Assessment of Ea

Bland-Altman analysis showing the difference of each pair plotted

for

landrepr

for

. (B)
nt 95% limits of agreement. Abbreviations as in Figures 1 and 2.

smfacdicmsm

rese


M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.


254

(E/Ea), representing an estimation of LV fillingpressures, showed good correlation between cardiacCT and 2D echocardiography with TDI. Finally,the study showed that cardiac CT and 2D echocar-diography were comparable for assessment of dia-stolic dysfunction. Accordingly, cardiac CT mayprovide information on diastolic dysfunction.

The importance of diastolic function in patientswith coronary atherosclerosis has been demonstratedin several studies (3,4). A recent meta-analysis pooled3,396 patients with documented myocardial infarctionfrom 12 prospective studies and demonstrated thatpatients with a restrictive LV filling pattern had asignificantly higher mortality rate than patients with anonrestrictive LV filling pattern (28.7% vs. 11.3%,respectively; p � 0.01) (3,4). Although invasive mea-urements of LV filling pressure are considered theost accurate approach for evaluation of diastolic LV

unction, they are not ideal for widespread applicationnd follow-up examinations. Consequently, severalardiac imaging techniques (particularly 2D echocar-iography) have been used to assess transmitral veloc-

ty as a noninvasive alternative (5,7). Even thoughomplex interacting pathophysiologic mechanismsay underlie diastolic dysfunction, evaluation of dia-

tolic LV function is most frequently based on trans-itral velocity measurements alone (5,7,11).

Transmitral velocity. Doppler echocardiography hasbeen validated for the assessment of transmitral veloc-ity as a noninvasive alternative of direct LV fillingpressures (5). Additionally, Doppler echocardiographyhas been compared with cardiac magnetic resonance(CMR) for assessment of transmitral velocity (9,10).The present study demonstrated that cardiac CT can


(A) Good correlation for E/Ea was observed between both techniques.against the average value of the same pair of values. Dotted lines rep

be used for assessment of LV diastolic function, as was

indicated by good correlations between cardiac CTand Doppler echocardiography for E transmitral ve-locity (r � 0.73; p � 0.01) and E/A (r � 0.87; p �0.01). In line with the study by Hartiala et al. (9), asystematic underestimation of transmitral velocity wasobserved when compared with Doppler echocardiog-raphy (Table 2). One of the potential explanations ofthis underestimation could be related to the samplingrate of cardiac CT. Because the datasets were sampledin 20 cardiac phases, each sampling took place in 5%steps of the R-R interval. Accordingly, the samplingrate of cardiac CT was markedly lower as comparedwith 2D echocardiography, and this may have resultedin an underestimation of transmitral velocity on CTbecause the peak transmitral velocity as derived from2D echocardiography could be present between 2sampled cardiac phases.

In both studies, correlations were not excellent fortransmitral velocity, and this may be related to otherparameters that could influence transmitral velocitymeasurements, including filling pressures, degree ofLV relaxation, and myocardial elastic recoil and stiff-ness (7,8). For overcoming these limitations, addi-tional measurements have been proposed, includingthe evaluation of pulmonary venous velocity, M-modeechocardiography flow velocity curves, and alteringpre-load and afterload conditions (Vasalva maneuveror nitroglycerin administration) (19,20). In the currentstudy, however, these measurements were not per-formed because this study was only conducted toevaluate the feasibility of cardiac CT.

Additionally, it has been suggested to combinetransmitral velocity and mitral septal tissue velocitymeasurements when evaluating diastolic heart func-

E/Ea

Bland-Altman analysis showing the difference of each pair plottednt 95% limits of agreement. Abbreviations as in Figures 1 and 2.

for

(B)

tion. Importantly, the combined assessment of trans-

pmwTvsftCdsrm

ddfitpdnsnsppfo

tfinni

cpCutcigvfiateca

qira


M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.


255

mitral velocity and mitral septal tissue velocity repre-sents a better estimate of LV filling pressures becauseit is a normalization of LV filling gradient for fillingLV volume (6,10,11).Mitral septal tissue velocity. Ommen et al. (6) studiedthe clinical use of TDI for evaluation of diastolic LVfunction in 100 patients. Comparison between inva-sive LV filling pressures and combined assessment ofE transmitral velocity and Ea showed improved cor-relation (r � 0.64) as compared with transmitralvelocity (r � 0.59) or mitral septal tissue velocity (r �0.36) alone (6). In addition, CMR has been comparedwith TDI for assessment of tissue velocities. Paelincket al. (10) used phase-contrast CMR and Dopplerechocardiography to measure transmitral and mitralseptal tissue velocities in 18 patients with hypertrophiccardiomyopathy. Importantly, combined assessmentof E transmitral velocity and Ea (E/Ea) showed agood correlation between Doppler echocardiographyand CMR (r � 0.89; p � 0.01). Moreover, invasivemeasurements were well correlated to E/Ea derivedfrom Doppler echocardiography (r � 0.85; p � 0.01)and CMR (r � 0.80; p � 0.01). Likewise, the currentstudy reported good correlations for transmitral veloc-ity (E/A; r � 0.87; p � 0.01) and E/Ea (r � 0.81;

� 0.01). Furthermore, both studies showed thatitral septal tissue velocity was slightly overestimatedhen compared with 2D echocardiography withDI. With tissue Doppler echocardiography, tissue

elocities are quantified using changes in Dopplerignal over time. Doppler patterns are only displayedor the region of interest (sample volume), located athe basal septal mitral valve annulus. With cardiacT, however, tissue velocities are measured using aifferent region of interest, ranging from the basaleptal mitral valve annulus to the apex. The differentegions of interest may have caused a slight overesti-ation of tissue velocity using cardiac CT.

Diastolic LV function. Cardiac CT and 2D echocar-iography with TDI were comparable for detection ofiastolic dysfunction. This represents an importantnding because the assessment of diastolic dysfunc-ion provides important diagnostic, therapeutic, andrognostic information in patients with cardiovascularisease, and more specifically, in patients with coro-ary atherosclerosis (1–4). Additionally, it has beenhown that patients with coronary atherosclerosis andormal LV systolic function may already exhibit dia-tolic dysfunction (21). Accordingly, additional post-rocessing for diastolic dysfunction may have theotential to enhance the clinical evaluation derivedrom cardiac CT, particularly in patients with evidence
f coronary atherosclerosis but with normal LV sys-
olic function. Moreover, the feasibility of cardiac CTor assessment of diastolic function is of particularnterest because the number of patients referred foroninvasive evaluation of known or suspected coro-ary atherosclerosis with cardiac CT imaging has

ncreased substantially over recent years.At present, prospective triggering techniques are

ommonly used in patients referred for cardiac CT,articularly for ruling out the presence of significantAD in young patients. However, cardiac CT is alsosed in patients with advanced CAD or elderly pa-ients for detailed characterization of coronary andardiac anatomy. In these patients, multiphase CTmages may still be acquired using retrospective ECGating to ensure diagnostic image quality for coronaryisualization. The additional information derivedrom this examination can be used for other purposes,ncluding the retrospective evaluation of systolic, ands demonstrated in the current study, diastolic func-ion. Accordingly, the information that is needed forvaluation of diastolic function can be derived fromonventional multiphase CT, without additional im-ge acquisition or radiation dose.

Moreover, with the recent developments in ac-uisition and reconstruction algorithms, multiphasemaging may be performed at considerably loweradiation dose as compared with the currentlyvailable imaging protocols.Study limitations. Some limitations of this studyneed to be considered. This was a retrospectivestudy, and a prospective study design would bepreferable. Secondly, transmitral velocity parame-ters were assessed with Doppler echocardiographyand cardiac CT as a noninvasive alternative todirectly measured LV filling pressures. Althoughdirect measurements of LV filling pressures wouldhave been preferred, they are not ideal for routineclinical examination. Furthermore, patients withvalvular regurgitation were excluded. Severe valvu-lar regurgitation may disturb accurate velocity mea-surements, leading to an inaccurate diastolic LVfunction analysis. Future studies are needed toevaluate this potential confounding effect. Addi-tionally, cardiac CT and 2D echocardiography withTDI were performed within a period of 3 monthsmaximum. However, to ensure optimal compara-bility between both examinations, the study onlyincluded patients with stable hemodynamic condi-tions; patients with unstable angina pectoris oracute coronary syndromes were excluded. Also, inthe current study, diastolic function indexes werederived from multiphase CT datasets acquired
without tube current modulation. To what extent


M A R C H 2 0 1 1 : 2 4 6 – 5 6

Boogers et al.


256

tube modulation may influence the measurementsof the mitral valve area and other diastolic functionparameters needs to be addressed in additionalstudies. Finally, it is important to note that theeffect of intravenous infusion of contrast mediaduring cardiac CT angiography on diastolic func-tion indexes is currently unknown (22).

C O N C L U S I O N S

Good correlations were observed for transmitral

8. Oh JK, Appleton CP, Hatle LK,Nishimura RA, Seward JB, Tajik AJ.

1

1

1

1

1

1tification in cardiaReson Imaging 199

of LV filling pressures when compared with 2Dechocardiography using TDI. Accordingly, cardiacCT may provide information on diastolic dysfunc-tion in selected patients imaged by retrospectivelyECG-gated CT.

Reprint requests and correspondence: Dr. Hildo J. Lamb,Department of Radiology, Leiden University MedicalCenter, Albinusdreef 2, 2333 ZA Leiden, the Nether-

velocity, mitral septal tissue velocity, and estimation lands C2-S. E-mail: [email protected].

1

1

1

1

2

2

2

R E F E R E N C E S

1. Zile MR, Brutsaert DL. New conceptsin diastolic dysfunction and diastolicheart failure: part I: diagnosis, progno-sis, and measurements of diastolic func-tion. Circulation 2002;105:1387–93.

2. Bhatia RS, Tu JV, Lee DS, et al. Out-come of heart failure with preserved ejec-tion fraction in a population-based study.N Engl J Med 2006;355:260–9.

3. Perrone-Filardi P, Bacharach SL,Dilsizian V, Bonow RO. Impaired leftventricular filling and regional dia-stolic asynchrony at rest in coronaryartery disease and relation to exercise-induced myocardial ischemia. Am JCardiol 1991;67:356–60.

4. Moller JE, Whalley GA, Dini FL, et al.Independent prognostic importance of arestrictive left ventricular filling patternafter myocardial infarction: an individualpatient meta-analysis: Meta-Analysis Re-search Group in Echocardiography acutemyocardial infarction. Circulation 2008;117:2591–8.

5. Nishimura RA, Appleton CP, Red-field MM, et al. Noninvasive Dopplerechocardiographic evaluation of leftventricular filling pressures in patientswith cardiomyopathies: a simultane-ous Doppler echocardiographic andcardiac catheterization study. J AmColl Cardiol 1996;28:1226–33.

6. Ommen SR, Nishimura RA, AppletonCP, et al. Clinical utility of Dopplerechocardiography and tissue Dopplerimaging in the estimation of left ven-tricular filling pressures: a comparativesimultaneous Doppler-catheterizationstudy. Circulation 2000;102:1788–94.

7. Nishimura RA, Tajik AJ. Evaluation ofdiastolic filling of left ventricle in healthand disease: Doppler echocardiographyis the clinician’s Rosetta Stone. J AmColl Cardiol 1997;30:8–18.

The noninvasive assessment of leftventricular diastolic function withtwo-dimensional and Doppler echo-cardiography. J Am Soc Echocardiogr1997;10:246–70.

9. Hartiala JJ, Mostbeck GH, Foster E,et al. Velocity-encoded cine MRI inthe evaluation of left ventricular dia-stolic function: measurement of mitralvalve and pulmonary vein flow veloc-ities and flow volume across the mitralvalve. Am Heart J 1993;125:1054–66.

0. Paelinck BP, de Roos A, Bax JJ, et al.Feasibility of tissue magnetic reso-nance imaging: a pilot study in com-parison with tissue Doppler imagingand invasive measurement. J Am CollCardiol 2005;45:1109–16.

1. Nagueh SF, Middleton KJ, KopelenHA, Zoghbi WA, Quinones MA.Doppler tissue imaging: a noninvasivetechnique for evaluation of left ven-tricular relaxation and estimation offilling pressures. J Am Coll Cardiol1997;30:1527–33.

2. Raff GL, Gallagher MJ, O’NeillWW, Goldstein JA. Diagnostic accu-racy of noninvasive coronary angiog-raphy using 64-slice spiral computedtomography. J Am Coll Cardiol 2005;46:552–7.

3. Mollet NR, Cademartiri F, vanMieghem CA, et al. High-resolutionspiral computed tomography coronaryangiography in patients referred for di-agnostic conventional coronary angiog-raphy. Circulation 2005;112:2318–23.

4. Henneman MM, Bax JJ, Schuijf JD,et al. Global and regional left ventric-ular function: a comparison betweengated SPECT, 2D echocardiographyand multi-slice computed tomogra-phy. Eur J Nucl Med Mol Imaging2006;33:1452–60.

5. van der Geest RJ, Reiber JH. Quan-
c MRI. J Magn9;10:602–8. e
6. Pattynama PM, Lamb HJ, van derVelde, van der Wall EE, de Roos A.Left ventricular measurements withcine and spin-echo MR imaging: astudy of reproducibility with variancecomponent analysis. Radiology 1993;187:261–8.

7. Lester SJ, Tajik AJ, Nishimura RA,Oh JK, Khandheria BK, Seward JB.Unlocking the mysteries of diastolicfunction: deciphering the RosettaStone 10 years later. J Am Coll Car-diol 2008;51:679–89.

8. Nagueh SF, Appleton CP, GillebertTC, et al. Recommendations for theevaluation of left ventricular diastolicfunction by echocardiography. J AmSoc Echocardiogr 2009;22:107–33.

9. Nishimura RA, Abel MD, Hatle LK,Tajik AJ. Relation of pulmonary vein tomitral flow velocities by transesophagealDoppler echocardiography. Effect ofdifferent loading conditions. Circulation1990;81:1488–97.

0. Takatsuji H, Mikami T, Urasawa K,et al. A new approach for evaluation ofleft ventricular diastolic function: spatialand temporal analysis of left ventricularfilling flow propagation by colorM-mode Doppler echocardiography.J Am Coll Cardiol 1996;27:365–71.

1. Garcia-Fernandez MA, Azevedo J,Moreno M, et al. Regional diastolicfunction in ischaemic heart diseaseusing pulsed wave Doppler tissueimaging. Eur Heart J 1999;20:496 –505.

2. Stern L, Firth BG, Dehmer GJ, et al.Effect of selective coronary arteriogra-phy on left ventricular volumes andejection fraction in man. Am J Cardiol1980;46:827–31.

Key Words: computedtomography y diastole y
chocardiography y imaging.
mailto:[email protected]

feasibility of diastolic function assessment with cardiac ct · the cardiac cycle). for each phase,...

Documents