From the Dep

(E.H., J.V., T.

Trondheim, N

of Circulation

Technology, T

This study wa

and the Helse

Reprint reque

Cardiology, 7

0894-7317/$3

Copyright 201

http://dx.doi.o

Transthoracic Doppler Echocardiographyfor Detection of Stenoses in the Left Coronary Artery

by Use of Poststenotic Coronary Flow Profiles:AComparisonwithQuantitative Coronary Angiography

and Coronary Flow Reserve

Espen Holte, MD, Johnny Vegsundv�ag, MD, Knut Hegbom, MD,Torstein Hole, MD, PhD, and Rune Wiseth, MD, PhD, �Alesund, Trondheim, Norway

Background: The aim of this study was to determine whether poststenotic diastolic-to-systolic velocity ratio(DSVR) assessed by transthoracic Doppler echocardiography could accurately identify significant stenosesin the left coronary artery.

Methods: A total of 108 patients scheduled for coronary angiography because of chest pain or acute coronarysyndromes were studied.

Results: The success rates of peak DSVR (pDSVR) measurements in the distal to mid left anterior descend-ing coronary artery and marginal branches of the left circumflex coronary artery were 85% and 32%,respectively. With peak coronary flow velocity reserve as a reference, pDSVR was significantly higher inarteries with normal coronary flow reserve (peak coronary flow velocity reserve $ 2.0) compared witharteries with reduced coronary flow reserve (peak coronary flow velocity reserve < 2.0) (1.86 6 0.32 vs1.53 6 0.31, P < .0001). In comparison with quantitative coronary angiography, pDSVR was significantlyhigher in lesions with diameter stenosis < 50% compared with those with diameter stenosis of 50% to75% (1.92 6 0.32 vs 1.53 6 0.18, P < .0001) or diameter stenosis of 76% to 100% (1.43 6 0.13,P < .0001). Receiver operating characteristic curves showed pDSVR < 1.68 to be the optimal cutoff valuefor identifying both functionally significant stenoses and diameter stenoses $ 50%, with sensitivity of86% and 90%, specificity of 74% and 84%, positive predictive value of 51% and 71%, and negative pre-dictive value of 94% and 95%, respectively.

Conclusions: Transthoracic pDSVRmeasurements in the distal to mid left anterior descending coronary arteryand marginal branches of the left circumflex coronary artery had high accuracy for excluding functionally sig-nificant stenoses in the left coronary artery, as well as for identifying angiographic significant stenoses. (J AmSoc Echocardiogr 2013;26:77-85.)

Keyword: Transthoracic Doppler echocardiography

Quantitative coronary angiography (QCA) has traditionally been thegold standard for assessing coronary artery disease, with a significantcoronary stenosis generally defined as luminal diameter reduction $50%. However, it can be difficult to determine on angiography

artment of Internal Medicine, �Alesund Hospital, �Alesund, Norway

H.); the Department of Cardiology, Trondheim University Hospital,

orway (K.H., R.W.); and the Medical Faculty (T.H.) and Department

and Medical Imaging (R.W.), Norwegian University of Science and

rondheim, Norway.

s funded by grants from the Sunnmore Health Trust Research Fund

Midt-Norge Regional Health Trust Research Fund.

sts: Espen Holte, MD, Trondheim University Hospital, Department of

006 Trondheim, Norway (E-mail: espen.holte@stolav.no).

6.00

3 by the American Society of Echocardiography.

rg/10.1016/j.echo.2012.10.001

whether a stenosis causes ischemia or not. This distinction is impor-tant, because anatomically significant but functionally nonsignificantstenoses have a good prognosis without invasive treatment.1,2

Functionally nonsignificant stenoses are common in the borderlinestenosis group (diameter stenosis, 50%–75%) and may as well befound among high-grade stenoses (diameter stenosis, 76%–100%),with a recent study showing that 65% of lesions with diameter steno-sis of 50% to 70% and 20% of lesions with diameter stenosis of 71%to 90% were without functional significance.3 Other testing optionsare therefore important in the selection for coronary angiographyand to distinguish functionally significant from nonsignificant coro-nary stenoses,3-9 with fractional flow reserve as a probablereference for the functional evaluation of a coronary stenosis.3

Coronary flow velocity reserve (CFVR) assessed using Doppler trans-thoracic echocardiography (TTE) has recently been shown to bea noninvasive surrogate of fractional flow reserve, especially in the de-tection of a functionally nonsignificant lesion.8 Normal coronary

77

Abbreviations

CFVR = Coronary flowvelocity reserve

Cx = Left circumflex coronaryartery

CxMb =Marginal branches of

the left circumflex coronary

artery

DSVR = Diastolic-to-systolic

velocity ratio

LAD = Left anterior

descending coronary artery

LM = Left main coronary

artery

NPV = Negative predictivevalue

pCFVR = peak coronary flowvelocity reserve

pDSVR = peak diastolic-to-systolic velocity ratio

PDV = Peak diastolic flow

velocity

PPV = Positive predictive

value

PSV = Peak systolic flow

velocity

QCA = Quantitative coronary

angiography

RCA = Right coronary artery

ROC = Receiver operating

characteristic

TTE = Transthoracic

echocardiography

78 Holte et al Journal of the American Society of EchocardiographyJanuary 2013

arteries display a predominantdiastolic blood flow pattern,which is less marked in the distalright coronary artery (RCA),probably because of lower intra-myocardial systolic contractionpressure in the right ventricle.10

Several studies have shown thatin the presence of a significantcoronary stenosis, the ratio be-tween the diastolic and systoliccoronary blood flow velocities,diastolic-to-systolic velocity ratio(DSVR), is significantly reducedwhen invasively measureddistally to the stenosis.10-13 Thisreduction is postulated to becaused by a combinedpoststenotic decrease of diastolicflow and an increased systolicflow from an intramyocardialsystolic contraction pumpacting on the intramyocardialcapacitance vessels.14 Modernhigh-end echocardiographicequipment permits excellent im-aging of coronary artery bloodflow, allowing measurements ofcoronary blood flow pro-files.6,7,9,15-22 Recent reportshave indicated that findings ofreduced DSVR measured byTTE in the distal left anteriordescending coronary artery(LAD) may be a simple,noninvasive method for thedetection of high-grade coronarystenoses located more proxi-mally in the LAD.18,20,21 This isdemonstrated for patients withor without wall motion

abnormalities of the left ventricle,20 and peak DSVR (pDSVR) values< 1.6 to 1.8 are proposed to represent high-grade LAD steno-ses.18,20,21 However, there is a paucity of studies comparing LADpDSVR obtained by TTE with various degrees of stenosis in theLAD and the left main coronary artery (LM) as defined by QCA,and pDSVR measurements in marginal branches of the leftcircumflex coronary artery (CxMb) might be used for evaluatingcoronary disease in the left circumflex coronary artery (Cx) andLM. Furthermore, the functional significance of pDSVR requiresfurther validation, and this parameter could be useful in theevaluation of borderline coronary stenoses. Because the normalpDSVR in the distal RCA is low and probably close to pathologicvalues, the potential utility of distal pDSVR measurements seemsprimarily to be in the evaluation of possible upstream stenoses inthe LAD and Cx.

The aim of this study was to assess the feasibility and accuracy ofpDSVRmeasurements on TTE as a simple method for diagnosing sig-nificant stenoses in the LM, LAD, and Cx, using QCA and CFVRmeasured by TTE as the anatomic and functional references,respectively.

METHODS

Study Population

Patients were included in the study if they fulfilled the following crite-ria: (1) already scheduled for coronary angiography because of docu-mented or suspected stable or unstable coronary disease, (2) age > 18years, and (3) met no exclusion criteria. The exclusion criteria were(1) previous aortocoronary bypass surgery, (2) presumed insufficientacoustic windows because of severe emphysema or severe over-weight, (3) significant valvular disease, (4) atrial fibrillation, and (5) ad-ministrative reasons.

The study protocol was approved by the Regional Committee forMedical and Health Research Ethics and the Norwegian DataInspectorate. All participants gave written informed consent. Thisstudy is registered at ClinicalTrials.gov under identifierNTC00281346.

A total of 108 patients were included in the study, and this patientcohort has been previously presented.7 Clinical characteristics of thepatients are presented in Table 1. All patients took their medicationson the day of the echocardiographic study. Standard 12-lead electro-cardiograms were recorded in all patients.

Transthoracic Coronary Flow Measurements

Patients were examined using an Acuson Sequoia C512 (SiemensMedical Solutions USA, Inc., Mountain View, CA) ultrasound systemconnected to standard 4V1C and 7V3C transthoracic transducers.Contrast agent was not used. TTE with CFVR measurements wasnot performed earlier than the day after hospital admission andonly after the patients were clinically stable. The coronary arterieswere investigated using color Doppler mapping with the data post-processing mix function, which makes the colors transparent, as de-scribed previously.16 The velocity scale of color Doppler was set to0.24 m/sec but was actively changed to provide optimal images.With the patient in the left lateral decubitus position, the course ofthe mid and distal LAD could be seen from modified parasternalshort-axis and long-axis views or from modified apical two-chamberand three-chamber views, focusing on the anterior interventricularsulcus (Figure 1A1). From modified left parasternal short-axis andlong-axis views, the Cx could be imaged leaving the LM and furthercoursing caudally in the atrioventricular sulcus to the inferior marginof the sulcus.16 However, measurements of flow velocities in themaintrunk of the Cx are difficult because of cyclic cardiacmotion.Marginalbranches of the Cx leave the artery at various levels, and from mod-ified apical four-chamber views focusing on different levels of the lat-eral wall of the left ventricle, marginal branches could be visualizedcoursing in the distal direction on the epicardial surface toward thetransducer (Figure 1B1). Measurements of coronary flow velocitiesin marginal branches are less influenced by cardiac cyclic motion.Whenever possible, the most inferior marginal branch viewed wasused for measurements. Echocardiographic evaluation of regionalleft ventricular contractility was not part of the study protocol, but re-gional contractility was assessed in 30 of 35 patients (86%) with andin 32 of 73 patients (44%) without unstable coronary disease.

The coronary flow velocity waveform appears as a complex ofa small wave in systole and a large trapezoid wave in diastole(Figures 1A2 and 1B2). Blood flow velocities in the distal LAD andCxMb were measured using pulsed-wave Doppler at frequencies of1.75 to 3.5 MHz in a sample volume of 1.5 to 5 mm, with the samplevolume positioned on the laminar color flowDoppler signal. The sam-ple volume was positioned distally to any visualized turbulent color

Table 1 Baseline characteristics of the study cohort (n = 108)

Variable Value

Age (y) 63.1 6 9.5Heart rate (beats/min) 63 6 7.4

BMI (kg/m2) 26 6 3.6Men 79 (73%)

Total cholesterol (mmol/L) 4.9 6 1.1

Blood pressure (mm Hg)

Systolic 142 6 20

Diastolic 82 6 12

Medical history

Hypertension (>140/90 mm Hg) 61 (55%)

Current smoking 29 (27%)

Diabetes 11 (10%)

Previous CAD 23 (21%)

ACS 35 (32%)

Cardiac medication

Aspirin 97 (90%)

Thienopyridine 38 (35%)

Low–molecular weight heparin 30 (28%)

b-blockers 85 (79%)

Statins 87 (81%)

Calcium antagonists 21 (19%)

ACE inhibitors/ARBs 24 (22%)

Organic nitrates, daily maintenance 13 (12%)

ACE, Angiotensin-converting enzyme; ACS, acute coronary syn-

dromes; ARB, angiotensin II receptor blocker; BMI, body mass

index; CAD, coronary artery disease.

Journal of the American Society of EchocardiographyVolume 26 Number 1

Holte et al 79

flow Doppler signal, because turbulent color flow Doppler signalsmight represent stenosis.23 The level of the left ventricular papillarymuscles marked the division between the mid and distal segmentsof the LAD, and the distal portion of the mid segment of the LADwas used for DSVR measurements if the distal LAD could not be vi-sualized. We tried to find at least three consecutive cardiac cycles formeasurements of diastolic and systolic peak flow velocities. Angle cor-rection was used during velocity measurements to keep the angle be-tween coronary blood flow and the Doppler beam as small aspossible. The ratio between the peak diastolic and peak systolic veloc-ities was measured in each cardiac cycle (Figures 1A2 and 1B2), andthe average of these peak velocity ratios measured in the consecutivecycles was the pDSVR.

For each segment (distal to mid LAD, CxMb), four different out-comes were defined: (1) the coronary segment was defined with ret-rograde flow; (2) the coronary segment was not visualized; (3) thecoronary artery was defined with antegrade flow, with pDSVR impos-sible to measure; or (4) pDSVR was satisfactorily measured. Stop-motion frames and clips were digitally recorded for offline analysis.

CFVR was measured in the distal to mid LAD and in the CxMb, asdescribed previously.7 In brief, blood flow velocities weremeasured us-ingpulsed-waveDoppler,withmeasurements both at baseline and dur-ing the intravenous infusionof adenosine (0.14mg/kg/minover 2min).CFVR was calculated as the ratio of hyperemic to basal peak (pCFVR)diastolic flow velocities. A predefined pCFVR value < 2.0 was used asa cutoff value for functionally significant stenosis, in accordance withearlier studies.24-26 Peak CFVR findings were categorized in one oftwo groups: (1) pCFVR $ 2.0, indicating no functionally significantstenosis, or (2) pCFVR < 2.0, indicating functionally significantstenosis. The same CxMb were used for pDSVR and CFVR

measurements. If retrograde coronary artery flow was found, weconcluded that the artery was occluded more proximally,27,28 andpDSVR and CFVR measurements were not performed.

Coronary Angiography

Coronary angiography was performed using standard techniques. Allangiographic studies were digitally stored for later offline reviewingand measurements, blinded to the findings on TTE. Disagreementsin interpretation were resolved by consensus between the two cardi-ologists responsible for the angiographic readings. The angiogramswere analyzed using a 16-segment model of the coronary arteries.29

All angiograms were classified according to left or right dominance.The severity of coronary stenoses in the LM, LAD, and Cx was deter-mined by QCA, with each segment categorized in one of threegroups: (1) diameter stenosis of 0% to 49%, (2) diameter stenosisof 50% to 75%, or (3) diameter stenosis of 76% to 100%.Collateral flow to the occluded LAD and Cx was graded accordingto the Rentrop classification.30

Reproducibility

To assess interobserver measurement variability, two experienced ob-servers (E.H. and J.V.) evaluated data from13 randomcases in a blindedmanner. Intraobserver variability was similarly tested by one experi-enced observer (E.H.) 2 weeks apart, blinded to previous results. Forthe reproducibility studies, four separate coronary flow velocity wave-form recordings were selected for measurements in each patient.Interobserver and intraobservermeasurement variabilitywas expressedas themeandifference as apercentage and the coefficient of variationofthe differences between the measurements for each parameter.

Statistical Analysis

Continuous variables are presented as mean 6 SD and categoricalvariables as fractions and percentages. Comparisons of mean valueswere performed using Student’s t tests for normally distributed param-eters. Logistic regression analyses were used to explore relationshipsbetween the success rate for pDSVRmeasurements and demographicand clinical variables. Linear regression analyses were used to explorethe relationships between coronary flow measurements (systolic anddiastolic peak flow velocities and pDSVR) and baseline characteristics.Receiver operating characteristic (ROC) curve analysis was used to as-sess the optimal cutoff value of pDSVR. Sensitivity, specificity, positivepredictive value (PPV), and negative predictive value (NPV) to detectsignificant coronary disease defined as diameter stenosis $ 50% orpCFVR < 2.0 in the presence of the optimal pDSVR cutoff valuewere assessed using standard formulas. P values < .05 were consid-ered statistically significant. All analyses were performed with SPSSfor Windows version 15.0 (SPSS, Inc., Chicago, IL).

RESULTS

QCA

Therewere 35patientswith and 73 patientswithout unstable coronarydisease (Table 1). For stable patients, the mean time from echocardio-graphic examination to angiographywas 24.7631.7days. In the groupwith unstable coronary disease, themean time from echocardiographicexamination to coronary angiography was 9.56 23.1 days. However,two patients in this group had longer delays for coronary angiographybecause of intercurrent disease, and the mean time from echocardio-graphic examination to angiography was 4.3 6 3.4 days for the

Figure 1 Examples of antegrade coronary flow in the distal LAD (dLAD) and CxMb, with pDSVR in nonstenosed arteries. (A1) In themodified apical two-chamber view, the dLAD is seen coursing toward the apex in the anterior interventricular sulcus. (A2) SpectralDoppler tracings of blood flow in the dLAD, with pDSVR of 1.94. (B1) In the modified apical four-chamber view, a marginal branch ofthe Cx is seen coursing toward the apex on the epicardial surface of the left ventricle (LV). (B2) Spectral Doppler tracings of blood flowin the Cx marginal branch, with pDSVR of 1.87. D, Spectral Doppler tracings of diastolic coronary blood flow; IVS, interventricularseptum; LA, left atrium; RA, right atrium; RV, right ventricle; S, spectral Doppler tracings of systolic coronary blood flow.

Table 2 Number of stenoses by QCA in the LM, LAD, and Cx

Segment

Stenosis group 2 (DS,

50%–75%)

Stenosis group 3 (DS,

76%–100%)

LM 4 0

Proximal LAD 21 12

Mid LAD 15 5

Distal LAD 4 2

Proximal Cx 6 4

Mid Cx 5 5

Distal Cx 3 2

DS, Diameter stenosis.

Table 3 Summary of the study design

Segment Outcome 1* Outcome 2† Outcome 3‡ Outcome 4§ Total

Distal to

mid LAD

2 1 15 90 108

CxMb 1 39 34 34 108

*The coronary segment was defined with retrograde flow.†The coronary segment was not visualized.‡The coronary artery was defined with antegrade flow, with pDSVR

impossible to measure.§Peak DSVR was satisfactorily measured.

80 Holte et al Journal of the American Society of EchocardiographyJanuary 2013

remaining 33 patients. Table 2 lists findings onQCA for stenoses in theLM, LAD, andCx (segments 5–8, 11, 13, and 16 in theAmericanHeartAssociation’s 16-segment model29) in groups 2 and 3.

Feasibility of pDSVR Measurements

For patients with unstable coronary disease, the mean time fromsymptom onset to pDSVR measurements was 3.9 6 3.0 days.Table 3 presents the artery-based study protocol and results.Retrograde flow was identified by TTE in the distal to mid LAD intwo patients and in the CxMb in one patient. Coronary angiographyconfirmed the findings of retrograde flow, with Rentrop grade $ 2collateral circulation in all patients. Peak DSVR was satisfactorilymeasured in the distal to mid LAD and CxMb in 90 (85%) and 34(32%) patients, respectively, with numbers in parentheses denotingthe percentages of patients with pDSVR measurements among pa-tients without retrograde flow in the relevant coronary artery seg-ment. Peak DSVR was satisfactorily measured in both coronary

territories in 28 patients (26%). The only baseline variables(Table 1) related to the success rate of pDSVR measurements wereage (positive relation to LAD measurements) and current smoking(inverse relation to CxMb measurements). The mean angle correc-tions used for pDSVR measurements in the distal to mid LAD andCxMb were 26 6 11� and 19 6 12�, respectively.

Peak DSVR Compared with Angiography

On QCA, each main coronary artery could have more than one ste-nosis, with themost tight stenosis defining the degree of stenosis. PeakDSVR in the distal to mid LAD evaluates both the LAD and the LMbecause of the functional unity of the LAD and the LM, and findingsfor these two vessels were therefore analyzed together. Distalbranches from the Cx were more difficult to visualize by TTE thanthe more proximal branches, thus implying that the CxMB pDSVRmeasurements primarily evaluated stenoses in the proximal andmid segments of the Cx and less often stenoses in the distal segmentof the Cx. Hence, stenoses in the distal segment of the Cx (segment

Table 4 Peak diastolic and systolic coronary flow velocitiesand pDSVR in stenosis groups 1, 2, and 3

Group n PDV (m/sec) PSV (m/sec) pDSVR

Group 1 (DS, 0%–49%) 85 0.36 6 0.12 0.19 6 0.06 1.92 6 0.32

Group 2 (DS, 50%–75%) 26 0.36 6 0.10 0.23 6 0.06 1.53 6 0.18

Group 3 (DS, 76%–100%) 13 0.34 6 0.11 0.24 6 0.08 1.43 6 0.13

DS, Diameter stenosis.

Figure 2 ROC curves for pDSVR in the diagnosis of angiograph-ically and functionally significant stenoses in the LM, LAD, andCx. The red line is the ROC curve and area under the curve(AUC) for pDSVR in the diagnosis of significant coronary diseasein the LM, LAD, and Cx as defined by QCA (diameter stenosis[DS], 50%–100%), with optimal pDSVR < 1.68 (sensitivity,90%; specificity, 84%). The blue line is the ROC curve andAUC for pDSVR in the diagnosis of functionally significant steno-ses in the LM, LAD, and Cx as defined by pCFVR (pCFVR < 2.0),with optimal pDSVR < 1.68 (sensitivity, 86%; specificity, 74%).CI, Confidence interval.

Journal of the American Society of EchocardiographyVolume 26 Number 1

Holte et al 81

15 in the American Heart Association’s 16-segment model29) wereexcluded from pDSVR CxMb analyses, with the proximal and midsegments of the Cx and the LM analyzed together because of func-tional unity.

Among coronary arteries with pDSVR measurements, 26 arteries(23 LADs and LMs and three proximal and mid segments of the Cxand LM) had stenoses in group 2, and 13 arteries (11 LADs and LMsand two proximal and mid segments of the Cx and LM) had stenosesin group 3. Systolic and diastolic peak flow velocities are listed inTable 4, with corresponding pDSVR measurements. Because of thesmall number of CxMb measurements, the results are presented forboth arteries together. No statistical differences were found betweenpeak diastolic flow velocities (PDVs) in groups 1 (0.366 0.12 m/sec),2 (0.366 0.10 m/sec), and 3 (0.346 0.11 m/sec). Peak systolic flowvelocity (PSV) was significantly lower in group 1 compared withgroup 2 (0.19 6 0.06 vs 0.23 6 0.06 m/sec, P = .002) and group3 (0.24 6 0.08 m/sec, P = .008), with no statistical difference be-tween groups 2 and 3 (P= .75). Peak DSVRwas 1.926 0.32 in group1, significantly higher than in groups 2 (1.53 6 0.18) and 3 (1.43 60.13) (P < .0001 for both). The difference between groups 2 and 3did not reach statistical difference (P = 0.077). The ROC curve forpDSVR for the detection of diameter stenosis $ 50% (groups 2and 3 combined) is shown in Figure 2.

ROC curve analysis demonstrated that the optimal pDSVR cutoffvalue of 1.68 had sensitivity of 90%, specificity of 84%, PPVof 71%,and NPVof 95% for the detection of diameter stenosis $ 50%. Thesame cutoff value had sensitivity of 100%, specificity of 75%, PPVof49%, and NPVof 100% for the detection of diameter stenosis of 76%to 100%. Excluding the Cx, the same pDSVR cutoff value had sensi-tivity of 91%, specificity of 84%, PPVof 78%, andNPVof 94% for thedetection of diameter stenosis $ 50% in the LAD and LM. Figure 3shows the spread of pDSVR measurements for the different stenosisgroups. The Cx was the dominant artery in three patients withpDSVR measurements in the CxMb, with pDSVR > 1.68 and no an-giographic stenoses in any patient. Both PSV and PDV showed posi-tive correlations with systolic blood pressure (P = .026 and P = .025,respectively), with no correlations with other baseline characteristics(including diastolic blood pressure, heart rate, and rate-pressure prod-uct; Table 1). Peak DSVR showed no correlations with any baselinecharacteristics.

In patients without unstable coronary disease but with significantcoronary stenoses (diameter stenosis, 50%–100%) in the LAD, LM,or Cx, regional myocardial contractility was assessed by echocardiog-raphy in 15 of 26 patients (58%), with no contractility reductionfound in the corresponding coronary territories. The remaining 11patients without contractility evaluation showed no electrocardio-graphic changes that could indicate previous myocardial infarction.Regional contractility was assessed in 26 of 30 patients (87%) withunstable coronary disease and pDSVR measurements, and regionalcontractility reduction in the LAD and Cx territories secondary to

significant upstream stenoses was found in four patients. In arterieswith significant upstream stenoses, the group with reduced regionalcontractility showed significantly decreased pDSVR compared withthe group without downstream contractility reduction (1.42 6 0.10vs 1.60 6 0.21 m/sec, P = .02), but without statistical differences insystolic or diastolic peak velocities between the groups. In the four un-stable patients with regional contractility reduction, two patients hadgroup 2 stenoses, and the other two had group 3 stenoses, while allupstream stenoses in the group with no impairment of contractilitywere group 2 stenoses.

Peak DSVR Compared with CFVR

Among coronary arteries with DSVR measuremens, pCFVR wasmeasured in the distal to mid LAD and the CxMb in 89 (99%) and31 (91%) arteries, respectively. Twenty-five arteries (23 LADs andLMs and two proximal and mid segments of the Cx and LM) had ste-noses in group 2, and 13 arteries (11 LADs and LMs and two proximaland mid segments of the Cx and LM) had stenoses in group 3. PeakDSVR was 1.866 0.32 in pCFVR group 1 (pCFVR $ 2.0) and 1.536 0.31 in pCFVR group 2 (pCFVR < 2.0) (P < .0001). The ROCcurve for pDSVR for the detection of pCFVR < 2.0 is shown inFigure 2. ROC curve analysis demonstrated that the optimalpDSVR cutoff value of < 1.68 had sensitivity of 86%, specificity of74%, PPV of 51%, and NPV of 94% for the detection of pCFVR <2.0. Figure 4 shows the spread of pDSVR measurements for the dif-ferent pCFVR groups. Twelve of 25 stenoses in group 2 (diameter ste-nosis, 50%–75%) and two of 13 stenoses in group 3 (diameterstenosis, 76%–100%) showed pCFVR $ 2.0.

Observer Variability

There were no significant differences in mean values between the ob-servers. The biases for peak diastolic and systolic coronary flow veloc-ities and pDSVR were 1.2%, 0.7%, and 0.7% for intraobserver and

Figure 4 Relationships between pDSVR values and differentstenosis degrees in the LAD and Cx as defined by pCFVR.

Figure 3 Relationships between pDSVR values and differentstenosis degrees in the LAD and Cx as defined by QCA. DS, Di-ameter stenosis by QCA.

82 Holte et al Journal of the American Society of EchocardiographyJanuary 2013

8.4%, 2.1%, and 6.3% for interobserver measurements, respectively.The intraobserver and interobserver coefficients of variation were3.2% and 4.4% for peak diastolic coronary flow velocities, 4.7%and 7.9% for peak systolic coronary flow velocities, and 4.2% and12.8% for pDSVR, respectively.

DISCUSSION

In this study of consecutively included patients with suspected or de-finitive coronary artery disease, the feasibility and accuracy of diag-

nosing significant stenoses in the LM, LAD, and Cx using pDSVRmeasurements on TTE were examined and compared with findingsby QCA and CFVR measured by TTE. The main findings were that(1) pDSVR measurements in the distal to mid LAD were feasible inthe majority of patients, while corresponding measurements in theCxMb were feasible in one third of patients; (2) pDSVR < 1.68had high accuracy for the detection of diameter stenosis$ 50% as de-fined by QCA in the LM, LAD, and proximal and mid segments ofCx; and (3) pDSVR$ 1.68 had high accuracy for excluding function-ally significant stenoses, as defined by pCFVR.

DSVR is ameasure of the phasic blood flow in the coronary arteriesand is determined by the interaction of several parameters influencingcoronary artery flow. Blood pressure, intramyocardial systolic contrac-tion pressure, epicardial stenoses, and resistances in the peripheralcoronary circulation are major determinants of DSVR. DSVR is alsoinfluenced by the position at which measurements are performed.DSVR can be reduced by 15% to 20% in distal coronary locationscompared with proximal locations.10 Normal coronary arteries dis-play a predominant diastolic blood flow pattern, which is less markedin the RCA, probably because of lower intramyocardial systolic con-traction pressure in the right ventricle.10 In contrast to the LAD andCx, the normal DSVR in the RCA is close to pathologic valueswhen measured in distal locations, thus likely limiting the utility ofDSVR measurements to the LAD and Cx, with similar DSVRs inthe LAD and Cx.10 In the presence of a significant coronary stenosis,the poststenotic difference between the diastolic and systolic bloodflow velocities is reduced, with a lower DSVR.10,21 Spaan et al.14

found, downstream to a significant stenosis, a reduction of diastolicflow in combination with an increased systolic flow component.The increased systolic flow was attributed to an intramyocardial sys-tolic pump acting on the intramyocardial capacitance vessels, withthis systolic pump becoming increasingly predominant with progres-sive upstream stenosis. Several studies, both invasive and by TTE,have demonstrated that angiographic significant LAD stenoses re-duce the poststenotic pDSVR to pathologic values of <1.6 to1.8.10,12,18,20,21 Furthermore, studies have shown that LAD pDSVR< 1.6 to 1.7 is in agreement with findings of reversible ischemia bymyocardial perfusion imaging.11,19 Invasive measurements areexpensive and available only during cardiac catheterization, whileTTE is more attractive because it is noninvasive, is widely available,and may offer an opportunity for long-term follow-up of patients.

The high feasibility (85%) of pDSVRmeasurements in the distal tomid LAD in our study is comparable with that in other studies.18-21

The feasibility of pDSVR measurements in the CxMb wasconsiderably lower (32%). This was mainly caused by failure tovisualize the CxMb for DSVR measurements or inadequate spectralDoppler recordings of coronary flow through both systole anddiastole, with the low-velocity systolic flow component the most dif-ficult to measure (Table 3). To the best of our knowledge, this is thefirst study using TTE to investigate the feasibility and value ofDSVR measurements for diagnosing stenoses in the Cx.

In our study, findings of pDSVR < 1.68 by TTE showed high accu-racy for identifying angiographically significant stenoses in the LM,LAD, and proximal and mid Cx, with sensitivity of 90%, specificityof 84%, PPVof 71%, and NPVof 95% (Figures 1–3 and 5). This cutoffvalue corresponds well with cutoff values proposed by others.18,20,21

However, the definition of significant stenosis differs significantlyamong the various studies, with significant stenosis defined asdiameter stenosis $ 50% in our study and >75% to 90% in otherstudies using TTE.18,20,21 Our study has the advantage of exploringthe pDSVR cutoff value in relation to the traditional definition of

Figure 5 Examples of pDSVR measurements by Doppler TTE in patients with significantly stenosis in the LAD and Cx. (A1) QCAshows proximal LAD stenosis (diameter stenosis, 75%) (arrow), (A2) with spectral Doppler tracings of blood flow in the distal LAD(dLAD). Peak DSVRmeasured in the dLADwas 1.11, noticeably below the cutoff value of 1.68, indicating significant coronary stenosislocated more proximally in the LAD or LM. (B1) QCA shows mid Cx stenosis (diameter stenosis, 70%) (arrow), (B2) with spectralDoppler tracings of blood flow in a marginal branch of the CxMb (with enveloped spectral Doppler tracings of blood flow to discernthese tracings from the noise in the Doppler recording). Peak DSVR measured in the Cx marginal branch was 1.50, noticeably belowthe cutoff value of 1.68, indicating significant coronary stenosis located more proximally in the Cx marginal branch, Cx, or LM. PeakCFVR measured in the dLAD was 1.93 in the patient presented in (A), and peak CFVR measured in CxMb was 1.54 in the patientpresented in (B) (peak CFVR measurements not shown). D, Spectral Doppler tracings of diastolic coronary blood flow; S, spectralDoppler tracings of systolic coronary blood flow.

Journal of the American Society of EchocardiographyVolume 26 Number 1

Holte et al 83

angiographically significant stenosis (diameter stenosis $ 50%). Thesensitivity and specificity of identifying significant stenoses in ourstudy compared favorably with that in other studies using TTE,which have reported sensitivity and specificity of 77% to 100% and76% to 82%, respectively.18,20,21 Interestingly, all group 3 stenoses(diameter stenosis, 76%–100%) in our study showed pDSVRs belowthe cutoff value of 1.68 (Figure 3). Crowley and Shapiro17 demon-strated that pDSVR was significantly reduced when comparing LADlesions with diameter stenosis 30% to 70% with lesions with diameterstenosis < 30%, and pDSVR was further significantly reduced whencomparing the group with diameter stenosis > 70% with that with di-ameter stenosis of 30% to 70%. This suggests a gradual reduction ofpDSVR with increasing degree of stenosis, with pDSVR reductionstarting relatively early in the stenotic process. Our study confirmsthat both borderline and high-grade stenoses significantly reducepDSVR compared with arteries with no stenosis, as defined byQCA. In our study, pDSVR was lower in high-grade stenoses(pDSVR, 1.43 6 0.13) compared with borderline stenoses (pDSVR,1.53 6 0.18), but this difference did not reach statistical significance(P = .077; Figure 3), probably because of the small number of patients.

In our study, pDSVR < 1.68 corresponded reasonably well withfindings of pCFVR < 2.0 measured in the same artery, with sensitivityof 86%, specificity of 74%, PPVof 51%, and NPVof 94% (Figures 2and 4). However, pDSVR $ 1.68 had high accuracy for excludingfunctionally significant stenoses (Figures 2 and 4). These findings sup-port earlier reports of pDSVR being able to identify or exclude func-tionally significant LAD stenoses.11,19 In this study, only four pDSVRvalues were false-negatives related to functional stenosis (pDSVR $1.68, pCFVR < 2.0; Figure 4). In these cases, no stenoses were found

by QCA in the respective arteries, and these discrepancies could becaused by limitations of CFVR measurements. Twenty-four arterieswith normal pCFVRs ($2.0) showed pDSVRs below the cutoff value(Figure 4). Nine of these arteries had borderline stenoses (diameterstenosis, 50%–75%), and two arteries had high-grade stenoses (diam-eter stenosis, 76%–100%). Accordingly, these stenoses were angio-graphically but not functionally significant. In the remaining 13arteries, no stenoses were found by QCA, and these discrepanciescould be caused by limitations of pDSVR measurements. We suspectthat our findings of lower sensitivity and specificity of pDSVR forfunctional evaluation (as defined by CFVR) compared with anatomicevaluation (as defined by QCA) of coronary stenoses are explainedmainly by the fact that this index is a velocity ratio is less related tothe functional compared with the anatomic degree of stenosis.Discrepancies between pDSVR and pCFVR may also be caused bymeasurements in different locations in the artery, with a differentnumber of stenoses upstream from the measuring site. However, inour study, we were meticulous in using the same site for measure-ments of pDSVR and pCFVR. Differences between pCFVR andpDSVR might also be because the latter possibly is less influencedby functional abnormalities of the resistance vessels or because mul-tiple upstream stenoses could affect pDSVR and pCFVR differently.The same optimal pDSVR cutoff value of <1.68 identified for bothangiographically and functionally significant stenoses may be an inci-dental finding related to the study cohort. Nevertheless, this findingsuggests that the pDSVR cutoff values for angiographically and func-tionally significant stenoses are close.

The detailed mechanisms of poststenotic regulation of coronaryblood flow and changes in flow profiles may not seem fully

84 Holte et al Journal of the American Society of EchocardiographyJanuary 2013

understood. Several investigators have explained the poststenotic de-crease in pDSVR in the presence of a significant epicardial stenosis byfindings of significant reduction of PDV,10,17,18 while others havefound increased PSV as the main cause of the reduced pDSVR.20,21

In our study, the decrease in pDSVR in the presence of significantstenosis (groups 2 and 3) was caused mainly by a significantincrease in PSV (Table 4). Table 4 also demonstrates a minor, nonsig-nificant decrease in PDV when comparing the group without stenosis(group 1) with the group with stenosis > 75% (group 3), and therewas a significant increase in PSV between group 1 and 3.Accordingly, our data indicate that the decrease in pDSVRwith signif-icant coronary stenosis may be caused by both a decrease in diastolicvelocity and an increase in systolic velocity, supporting the findings ofSpaan et al.14 In our study, both PSV and PDV showed positive corre-lations with systolic blood pressure but not with other parameters(heart rate, rate-pressure product) that reflect oxygen demand. Thelack of a correlation of pDSVR with systolic blood pressure may becaused by the positive correlations of both PSV and PDV to this pres-sure. Peak DSVR showed no correlations with other parameters re-flecting oxygen demand or other baseline characteristics. In thepatient group with unstable coronary disease undergoing additionalcontractility evaluation, we found PSV and PDV not to be statisticallydifferent when comparing the groups with and without poststenoticcontractility reduction, with the latter group having significantly lessreduced pDSVR. From Spaan et al.’s pump model, one might expectto find an inverse correlation between poststenotic contractility andpDSVR because of reduced systolic flow with reduced contractility.We suspect that our finding of an apparent positive relation betweenpoststenotic contractility and pDSVR may be due to the small num-ber of patients and differences in the degree of stenosis in the twogroups (as shown in the ‘‘Results’’ section), and our findingmust there-fore be interpreted with caution.

Peak DSVR measurements in the distal to mid LAD or CxMb aresimple, are noninvasive, and do not involve extra costs (eg, comparedwith measurements of CFVR or stress echocardiography). In our ex-perience, obtaining a series of pDSVRmeasurements takes 3 to 5minfor each coronary artery. A finding of pDSVR < 1.68 most likely indi-cates significant upstream coronary stenosis, while a finding ofpDSVR$ 1.68 most likely excludes functionally significant upstreamstenosis. All this implies that these measurements are advantageouswhen feasible, even in the CxMb, which is less feasible. Another pos-sible clinical consequence is that repeated angiography to assess reste-nosis might be avoided when pDSVR is $1.68. The use ofechocardiographic contrast agent and the further development ofechocardiographic technology may both increase the feasibility ofpDSVR measurements in the LAD and Cx territories. It must bekept in mind, however, that pDSVR measurements in the distal tomid LAD and CxMb will not reveal stenoses in the branches of theLAD and Cx (except in the CxMb used for measurements) or steno-ses in the RCA territory.

Study Limitations

There were several limitations to our study. Because of larger distanceto the echocardiographic probe, the CxMb were more difficult to vi-sualize than the distal to mid LAD. For the same reason, recordingspectral Doppler signals was more complicated in the CxMb. Theuse of ultrasound contrast agent might have improved the feasibilityand quality of pDSVR and pCFVR measurements.19,31 Because oflimited clinical experience in patients with acute coronarysyndromes, we chose not to use ultrasound contrast when planning

the study. Left ventricular regional contractility was not assessed inall patients, and this might have limited the study results. Otherfactors, such as filling pressure and hypertrophy, might influencepDSVR measurements, and the lack of estimation of theseparameters might also have limited the study results. We cannotexclude that nearby coursing LAD branches may on occasion havebeen mistaken as the main LAD trunk, resulting in measurementsof pDSVR and pCFVR in this branch instead of the main LAD.Moreover, we did not take collateral perfusion into account.Stenoses may have eluded detection if pDSVR and pCFVR weremeasured proximally to the stenoses. Stenoses in distal segments ofthe main coronary arteries are, however, usually clinically lessimportant than stenoses in the mid or proximal segments, becausedistal segments supply a smaller amount of myocardium. ThepDSVR is probably not a binary variable (normal or abnormal) orabsolute for the cutoff value of 1.68, implying that a pDSVR valueclose to the cutoff value might give uncertain estimates of theanatomic and functional significance of a coronary lesion. Thefunctional significance of a stenosis may be overestimated orunderestimated by QCA, and the functional aspect of a stenosismay be incorrectly assessed by pCFVR measurements. The use ofadditional stress tests such as dobutamine stress echocardiographyor myocardial perfusion imaging might have further clarified thefunctional significance of angiographic findings. Finally, we cannotexclude selection bias in our results, because our study cohortincluded only patients planned for coronary angiography andexcluded those with previous coronary bypass surgery, presumedinsufficient acoustic windows, significant valvular disease, or atrialfibrillation.

CONCLUSIONS

Peak DSVR measurements by TTE in the distal to mid LAD werefeasible in most patients and in the CxMb in one third of patients.To the best of our knowledge, this is the first study using TTE to in-vestigate the value of pDSVR measurements in the Cx. A pDSVRvalue < 1.68 showed high accuracy to detect angiographically signif-icant stenoses in the LAD, LM, and proximal and mid segments ofthe Cx. A pDSVR value < 1.68 showed moderate ability to detectfunctionally significant stenoses, while pDSVR $ 1.68 had high ac-curacy to exclude these stenoses. When feasible, assessing pDSVR inthe distal to mid LAD and CxMb provides a simple, noninvasivemethod to identify and rule out angiographic and hemodynamic sig-nificant stenoses in the LAD, LM, and proximal and mid segmentsof Cx.

REFERENCES

1. Pijls NHJ, Fearon WF, Tonino PAL, Siebert U, Ikeno F, Bornschein B,et al. Fractional flow reserve versus angiography for guiding percutane-ous coronary intervention in patients with multivessel coronary arterydisease. 2-Year follow-up of the FAME (Fractional Flow Reserve VersusAngiography for Multivessel Evaluation) study. J Am Coll Cardiol2010;56:177-84.

2. Meimoun P, Benali T, Elmkies F, Sayah S, Luycx-Bore A, Doutrelan L, et al.Prognostic value of transthoracic coronary flow reserve in medicallytreated patients with left anterior descending coronary disease with steno-sis 51% to 75% in diameter. Eur J Echocardiogr 2009;10:127-32.

3. Tonino PAL, Fearon WF, De Bruyne B, Oldroyd KG, Leesar MA, VerLee PN, et al. Angiographic versus functional severity of coronary artery

Journal of the American Society of EchocardiographyVolume 26 Number 1

Holte et al 85

stenoses in the FAME study. Fractional flow reserve versus angiography inmultivessel disease. J Am Coll Cardiol 2010;55:2816-21.

4. Sicari R, Nihoyannopoulos P, Evangelista A, Kasprzak J, Lancellotti P,Poldermans D, et al. Stress echocardiography expert consensus statement.European Association of Echocardiography (EAE) (a registered branch ofthe ESC). Eur J Echocardiogr 2008;9:415-37.

5. Gibbons RJ, Abrams J, Chatterjee K, Daley J, Deedwania PC, Douglas JS,et al. ACC/AHA 2002 guideline update for the management of patientswith chronic stable angina—summary article: a report of the AmericanCollege of Cardiology/American Heart Association Task Force on PracticeGuidelines (Committee on theManagement of PatientsWith Chronic Sta-ble Angina). J Am Coll Cardiol 2003;43:159-68.

6. Voci P, Pizzuto F, Romeo F. Coronary flow: a new asset for the echo lab?Eur Heart J 2004;25:1867-79.

7. Vegsundv�ag J, Holte E, Wiseth R, Hegbom K, Hole T. Coronary flow veloc-ity reserve in the three main coronary arteries assessed with transthoracicDoppler: a comparative study with quantitative coronary angiography.J Am Soc Echocardiogr 2011;24:758-67.

8. Meimoun P, Sayah S, Luycx-Bore A, Boulanger J, Elmkies F, Benali T, et al.Comparison between non-invasive coronary flow reserve and fractionalflow reserve to assess the functional significance of left anterior descendingartery stenosis of intermediate severity. J Am Soc Echocardiogr 2011;24:378-81.

9. Murata E, Hozumi T, Matsumura Y, Fujimoto K, Sugioka K, Takemoto Y,et al. Coronary flow velocity reserve measurement in three major coro-nary arteries using transthoracic Doppler echocardiography. Echocardiog-raphy 2006;23:279-86.

10. Ofili EO, Labovitz AJ, Kern MJ. Coronary flow velocity dynamics in nor-mal and diseased arteries. Am J Cardiol 1993;71(suppl):3D-9.

11. Deychak YA, Segal J, Reiner JS, Rohrbeck SC, Thompson MA,Lundergan CF, et al. Doppler guide wire flow-velocity indexes measureddistal to coronary stenosis associated with resersible thallium perfusion de-fects. Am Heart J 1995;129:219-27.

12. Donohue TJ, Kern MJ, Aguirre FV, Bach RG, Wolford T, Bell CA, et al. As-sessing the hemodynamic significance of coronary artery stenoses: analy-ses of translesional pressure-flow velocity relations in patients. J Am CollCardiol 1993;22:449-58.

13. Serruys PW, Di Mario C, Meneveau N, de Jaegere P, Strikwerda S, deFeyter PJ, et al. Intracoronary pressure and flow velocity with sensor-tipguidewires: a new methodologic approach for assessment of coronary he-modynamics before and after coronary interventions. Am J Cardiol 1993;71(suppl):41D-53.

14. Spaan JAE, Breuls NPW, Laird JD. Diastolic-systolic coronary flow differ-ences are caused by intramyocardial pump action in the anesthetizeddog. Circ Res 1981;49:584-93.

15. Meimoun P, Benali T, Sayah S, Luycx-Bore A, Boulanger J, Zemir H, et al.Evaluation of left anterior descending coronary stenosis of intermediate se-verity using transthoracic coronary flow reserve and dobutamine stressechocardiography. J Am Soc Echocardiogr 2005;18:1233-40.

16. Vegsundv�ag J, Holte E, Wiseth R, Hegbom K, Hole T. Transthoracic echo-cardiography for imaging of the different coronary artery segments: a fea-sibility study. Cardiovasc Ultrasound 2009;7:8.

17. Crowley JJ, Shapiro LM. Noninvasive analysis of coronary artery postste-notic flow characteristics by using transthoracic echocardiography. J AmSoc Echocardiogr 1998;11:1-9.

18. Higashiue S, Watanabe H, Yokoi Y, Takeuchi K, Yoshikawa J. Simple detec-tion of severe coronary stenosis using transthoracic Doppler echocardiog-raphy at rest. Am J Cardiol 2001;87:1064-8.

19. Daimon M, Watanabe H, Yamagishi H, Kuwabara Y, Hasegawa R,Toyoda T, et al. Physiologic assessment of coronary artery stenosis withoutstress tests: noninvasive analysis of phasic flow characteristics by transtho-racic Doppler echocardiography. J Am Soc Echocardiogr 2005;18:949-55.

20. Okura H, Fuyuki H, Kubo T, Iwata K, Taguchi H, Toda I, et al. Noninvasivediagnosis of ischemic cardiomyopathy using coronary flow velocity mea-surements of the left anterior descending coronary artery by transthoracicDoppler echocardiography. J Am Soc Echocardiogr 2006;19:552-8.

21. Tani T, Tanabe K, Kitai T, Yamane T, Kureha F, Katayama M, et al. Detec-tion of severe stenosis and total occlusion in the left anterior descendingcoronary artery with transthoracic Doppler echocardiography in the emer-gency room. Echocardiography 2009;26:15-20.

22. Hyodo E, Hirata K, Hirose M, Sakanoue Y, Nishida Y, Arai K, et al. Detec-tion of restenosis after percutaneous coronary intervention in three majorcoronary arteries by transthoracic Doppler echocardiography. J Am SocEchocardiogr 2010;23:553-9.

23. Saraste M, Vesalainen RK, Ylitalo A, Saraste A, Koskenvuo JW, Toikka JO,et al. Transthoracic Doppler echocardiography as a non-invasive tool to as-sess coronary artery stenosis—a comparisonwith quantitative coronary an-giography. J Am Soc Echocardiogr 2005;18:679-85.

24. Pizzuto F, Voci P, Mariano E, Puddu PE, Chiavari PA, Romeo F. Noninva-sive coronary flow reserve assessed by transthoracic coronary Doppler ul-trasound in patients with left anterior descending coronary artery stents.Am J Cardiol 2003;91:522-6.

25. Winter R, Gudmundsson P, Willenheimer R. Feasibility of noninvasivetransthoracic echocardiography/Doppler measurement of coronaryflow reserve in left anterior descending coronary artery in patients withacute coronary syndrome: a new technique tested in clinical practice.J Am Soc Echocardiogr 2003;16:464-8.

26. Voci P, Pizzuto F, Mariano E, Puddu PE, Chiavari PA, Romeo F. Measure-ment of coronary flow reserve in the anterior and posterior descendingcoronary arteries by transthoracic Doppler ultrasound. Am J Cardiol2002;90:988-91.

27. Watanabe N, Akasaka T, Yamaura Y, Akiyama M, Koyama Y,Kamiyama N, et al. Noninvasive detection of total occlusion of the left an-terior descending coronary arterywith transthoracic Doppler echocardiog-raphy. J Am Coll Cardiol 2001;38:1328-32.

28. Boschchenko AA, Vrublevsky AV, Karpov RS. Transthoracic echocardiog-raphy in the detection of chronic total coronary artery occlusion.Eur J Echocardiography 2009;10:62-8.

29. Austen WG, Edwards JE, Frye RL, Gensini GG, Gott VL, Griffith LS, et al.A reporting system on patients evaluated for coronary artery disease.Report of the Ad Hoc Committee for Grading of Coronary Artery Dis-ease, Council on Cardiovascular Surgery, American Heart Association.Circulation 1975;51(suppl):5-40.

30. Rentrop KP, Cohen M, Blanke H, Phillips RA. Changes in collateral chan-nel filling immediately after controlled coronary artery occlusion by an an-gioplasty balloon in human subjects. J Am Coll Cardiol 1985;5:587-92.

31. Caiati C, Montaldo C, Zedda N, Bina A, Iliceto S. New non-invasivemethod for coronary flow reserve assessment: contrast-enhanced trans-thoracic second harmonic echo Doppler. Circulation 1999;99:771-8.