Listen to this manuscript’s

audio summary by

JACC Editor-in-Chief

Dr. Valentin Fuster.

J O U R N A L O F 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 V O L . 6 8 , N O . 1 8 , 2 0 1 6

ª 2 0 1 6 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

P U B L I S H E D B Y E L S E V I E R

I S S N 0 7 3 5 - 1 0 9 7 / $ 3 6 . 0 0

h t t p : / / d x . d o i . o r g / 1 0 . 1 0 1 6 / j . j a c c . 2 0 1 6 . 0 8 . 0 3 0

Strain Echocardiography andFunctional Capacity in AsymptomaticPrimary Mitral Regurgitation WithPreserved Ejection Fraction

Amgad Mentias, MD, Peyman Naji, MD, A. Marc Gillinov, MD, L. Leonardo Rodriguez, MD, Grant Reed, MD,Tomislav Mihaljevic, MD, Rakesh M. Suri, MD, DPHIL, Joseph F. Sabik, MD, Lars G. Svensson, MD, PHD,Richard A. Grimm, DO, Brian P. Griffin, MD, Milind Y. Desai, MD

ABSTRACT

Fro

a p

At

an

act

the

Ma

BACKGROUND The potential additive utility of baseline resting left ventricular global longitudinal strain (LV-GLS) and

exercise stress testing in risk stratification of patients with significant mitral regurgitation (MR) has not been studied.

OBJECTIVES The goal of this study was to determine whether resting LV-GLS and exercise testing provide incremental

prognostic utility in asymptomatic patients with $3þ primary MR and preserved left ventricular ejection fraction.

METHODS Between 2000 and 2011, resting and exercise echocardiography data, Society of Thoracic Surgeons (STS)

scores, and death were recorded in 737 patients (mean age 58 � 13 years; 68% men).

RESULTS Coronary artery disease and flail leaflet were seen in 10% and 28% of patients, respectively. STS score,

resting left ventricular ejection fraction, mitral effective regurgitant orifice, resting right ventricular systolic pressure

(RVSP), exercise metabolic equivalents (METs), and percentage of age-/sex-predicted METs were 1.5 � 1%, 62 � 2%,

0.45 � 0.2 cm2, 31 � 12 mm Hg, 9.8 � 3, and 115 � 27, respectively. Median LV-GLS was –21.7%. Within 3 months

(interquartile range: 1 to 15 months), 65% underwent mitral valve surgery. At 8.3 � 3 years, 64 (9%) patients died

(0% 30-day post-operative deaths). On multivariable Cox survival analysis, higher STS score (hazard ratio [HR]: 1.14),

more abnormal resting LV-GLS (HR: 1.60), higher baseline RVSP (HR: 1.35), and lower percentage of age-/sex-predicted

METs (HR: 1.13) were associated with higher mortality, whereas mitral valve surgery (HR: 0.82) was associated with

improved survival (all p < 0.01). Addition of predicted METs and resting LV-GLS to STS, resting RVSP, left ventricular

end-systolic dimension, and mitral effective regurgitant orifice increased the C-statistic for longer-term mortality from

0.61 to 0.69 and 0.78, respectively (all p < 0.01). On quadratic spline analysis, the risk of death progressively increased

as resting LV-GLS worsened below –21%.

CONCLUSIONS Reduced exercise capacity and worsening resting LV-GLS were associated with mortality, providing

additive prognostic utility. (J Am Coll Cardiol 2016;68:1974–86) © 2016 by the American College of Cardiology

Foundation.

C urrently, myxomatous mitral valve (MV)degeneration is the most common etiologyof primary mitral regurgitation (MR) in the

developed world (1,2). For patients with severe

m the Heart Valve Center, Heart and Vascular Institute, Cleveland Clinic, C

hilanthropic gift from the Reginald and Jamie Baxter family to Dr. Desai

riCure, Edwards, Medtronic, and St. Jude Medical; is a consultant to Edward

d reports equity stake in PleuraFlow. Dr. Sabik is a consultant for Medtron

ed as Principal Investigator for Edwards and Abbott. All other authors have

contents of this paper to disclose. Drs. Mentias and Naji contributed equa

nuscript received June 21, 2016; revised manuscript received July 22, 201

primary MR, MV surgery results in significantimprovement in the hemodynamic consequences ofthe regurgitant valve (3). According to the mostrecent guidelines, MV surgery is recommended for

leveland, Ohio. The study was supported, in part, by

. Dr. Gillinov has served on the Speakers Bureau for

s Lifesciences, Abbott Vascular, On-X, and Medtronic;

ic and Sorin; and has received research funding and

reported that they have no relationships relevant to

lly to this work.

6, accepted August 1, 2016.

AB BR E V I A T I O N S

AND ACRONYM S

AF = atrial fibrillation

CI = confidence interval

HR = hazard ratio

IQR = interquartile range

IVC = inferior vena cava

LV = left ventricular

LVEF = left ventricular

ejection fraction

MET = metabolic equivalent

MR = mitral regurgitation

MV = mitral valve

RVSP = right ventricular

systolic pressure

STS = Society of Thoracic

Surgeons

J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6 Mentias et al.N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6 Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography

1975

patients with symptomatic severe primary MR orthose with asymptomatic left ventricular (LV) systolicdysfunction, new-onset atrial fibrillation (AF), andpulmonary arterial hypertension (4,5). However,appropriate timing of surgical intervention in asymp-tomatic severe primary MR remains challenging (6–8).Advances in MV repair have led to excellent long-term results, with low morbidity and mortality atexperienced centers (9,10), such that the guidelineshave lowered the threshold to recommend surgicalintervention, especially when there is a high chanceof MV repair (4). However, many patients are un-aware of symptoms because of gradual onset andtheir unwitting ability to subtly reduce exercise toprevent symptoms from occurring (11). Furthermore,LV dysfunction is difficult to assess in severe MR,and LV contractility may already be irreversiblyimpaired, even though left ventricular ejection frac-tion (LVEF) remains in the usual normal range. Previ-ous reports have shown that outcome after MVsurgery is impaired once preoperative LVEF is <60%(3). In addition, impaired LVEF post-operatively(<50%) has been shown to predict poorer long-termsurvival (12).

SEE PAGE 1987

In patients with significant primary MR, previousreports have shown that exercise echocardiographyaids in symptom evaluation, risk stratification, anddecision-making (for delaying of surgery, including inthose with mid–late systolic MR) (11,13,14). In addi-tion, decreased functional capacity on stress testing(15), exercise-induced pulmonary hypertension (16),and MR severity (17) have been associated with worseoutcomes. On the basis of the available data, exerciseechocardiography was given a Class IIa recommen-dation by the most recent guidelines (4). Assessmentof left ventricular global longitudinal strain (LV-GLS)using speckle-tracking echocardiography providesanother quantitative assessment of LV contractilefunction that may be more sensitive to early changethan LVEF (18). However, prognostic data regardingthe role of LV-GLS in MR patients are sparse (19,20).Furthermore, to the best of our knowledge, thepotential additive prognostic utility of baselineresting LV-GLS and exercise stress testing in the riskstratification of patients with significant MR has notbeen previously studied.

In the present study, we hypothesized that base-line resting LV-GLS and exercise stress testing couldprovide incremental prognostic utility in patientswith significant MR. Thus, in asymptomatic patientswith $3þ primary MR and preserved LVEF whounderwent rest and post-exercise echocardiography,

our goal was to determine whether: 1) acombination of baseline resting LV-GLS andexercise stress testing provided incrementalprognostic utility; and 2) there is a thresholdbelow which baseline resting LV-GLS aids inidentifying patients who are at a higher riskof longer-term mortality.

METHODS

STUDY POPULATION. This observationalcohort study included 737 consecutiveasymptomatic patients with $3þ primarymyxomatous MR, a nondilated left ventricle(indexed LV end-diastolic dimension <3.3cm/m2 and absolute LV end-systolicdimension <4 cm), and preserved LVEF($60%) who were seen and evaluated at ourtertiary care center between January 2000

and December 2011. All patients included in the studyhad comprehensive resting and treadmill exerciseechocardiography performed at the time of initialclinical evaluation (within 24 h of each other). In thestudy time frame, there were 1,694 patients with $3þMR who underwent treadmill exercise echocardiog-raphy. The following 810 patients were excluded:those with symptoms, any documented baseline AF,more than mild aortic valve disease or mitral stenosis,functional MR, rheumatic heart disease, hypertrophicobstructive cardiomyopathy, or history of MV sur-gery. We also excluded patients in whom poor imagequality precluded strain assessment (n ¼ 147).

The study was approved by the institutional reviewboard. Baseline data were prospectively entered intothe electronic medical record at the time of the initialencounter and manually extracted for the presentstudy. The Society of Thoracic Surgeons (STS) scorewas calculated in all patients. In patients whounderwent MV surgery, the type of MV surgery(repair vs. replacement) was recorded, along with anyconcomitant surgical procedure (coronary arterybypass grafting, maze, pulmonary vein isolation, orleft atrial appendage ligation/excision).

OUTCOMES ASSESSMENT. The date of the patient’sbaseline resting echocardiography at our institutionwas defined as the beginning of the observationalperiod. Patients were followed up by using chart re-view, and deaths were recorded. Mortality data wereobtained from review of medical records or state andnationally available databases (last inquiry inSeptember 2015). In addition, stroke/transientischemic attack during follow-up was recorded. Wealso recorded admissions for congestive heart failureat our institution. The primary endpointwasmortality.

Mentias et al. J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6

Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6

1976

RESTING AND EXERCISE ECHOCARDIOGRAPHY. Allpatients underwent comprehensive echocardio-grams by using commercial instruments (PhilipsMedical Systems, N.A., Bothell, Washington, andSiemens Medical Solutions USA, Inc., Malvern,Pennsylvania). LVEF, indexed LV dimensions, andleft atrial area were measured at rest, according toguidelines (21). Severity of MR was ascertained byusing multiple previously described techniques (22),including qualitative visual assessment (as a per-centage of left atrial size) and measurement ofeffective regurgitant orifice area and vena contracta

FIGURE 1 An Illustrative Example of Resting Left Ventricular Globa

Study Population

An example of resting left ventricular global longitudinal strain obtaine

Solutions, Mountain View, California). Global strain was obtained by ave

strain was defined as the peak negative value on the strain curve during

width. Because of the severity of MR, diastolicfunction was not reported. Presence of flail mitralleaflet was recorded. Right ventricular systolicfunction was measured qualitatively (normal, mild,moderate, or severe). Right ventricular systolicpressure (RVSP) was measured at rest, according toguidelines, by using the following formula: 4�(tricuspid regurgitant velocity [m/s])2 þ right atrialpressure (mm Hg) (23). Right atrial pressure wasestimated from the inferior vena cava (IVC) diam-eter and respiratory changes as follows: 3 mm Hg(IVC #2.1 cm, collapse with sniff >50%), 8 mm Hg

l Longitudinal Strain Assessment in a Patient From the

d by using Velocity Vector Imaging (Syngo VVI, Siemens Medical

raging segmental strain values from all 3 apical views. Peak global

the entire cardiac cycle.

TABLE 1 Baseline Characteristics of the Study Population

(N ¼ 737)

No MV Surgery(n ¼ 260)

MV Surgery(n ¼ 477) p Value

Age, yrs 57 � 14 58 � 13 0.12

Male 160 (62) 338 (71) 0.007

Body mass index, kg/m2 26 � 4 25 � 3 0.38

Hypertension 127 (50) 228 (49) 0.74

Diabetes mellitus 9 (4) 17 (4) 0.99

Coronary artery disease 30 (12) 46 (10) 0.11

Hyperlipidemia, % 114 (45) 203 (43) 0.32

Stroke 10 (4) 12 (3) 0.54

Smoker 107 (41) 190 (40) 0.62

ACE inhibitor or ARB use 99 (38) 139 (39) 0.52

Beta-blockers 88 (34) 139 (29) 0.18

STS score, % 1.6 � 1 1.5 � 0.7 0.11

Serum creatinine, mg/dl 1.2 � 0.2 1.1 � 0.4 0.21

Values are mean � SD or n (%).

ACE ¼ angiotensin-converting enzyme; ARB ¼ angiotensin-receptor blocker;MV ¼ mitral valve; STS ¼ Society of Thoracic Surgeons.

J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6 Mentias et al.N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6 Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography

1977

(IVC >2.1 cm, collapse >50%), and 15 mm Hg (IVC>2.1 cm, collapse with sniff <50%).

Subsequently, patients underwent treadmill echo-cardiography by using Bruce or modified Bruce pro-tocols, as appropriate. The test was terminated due tosymptoms and not at achievement of target heart rate.Blood pressure, heart rate, and electrocardiographicmeasurements were made at rest, at 1-min intervals,and for $6 min in recovery. The maximal predictedheart rate (220 � age), percentage of predictedmaximal heart rate, and number of metabolic equiva-lents (METs) achieved were recorded. Expected METswere calculated on the basis of age and sex, as previ-ously described (11), using the Veterans Affairs cohortformula (predictedMETs¼ 18� [0.15�age]) and the St.James Take Heart Project formula (predicted METs ¼14.7 � [0.13 � age]) for men and women, respectively.We subsequently calculated the following ratio: (METSachieved/age-sex expected METS) � 100. Immediatelyafter exercise, peak stress echocardiographic imageswere acquired, according to guidelines (24), and thefollowing parameters were assessed: regional wallmotion abnormalities for evaluation of ischemia andpeak stress RVSP. We also evaluated for changes in LVcavity size (increase, decrease, or no change), sugges-tive of the presence or absence of contractile reserve.Major (sustained ventricular or atrial arrhythmiasassociated with severe symptoms, hemodynamiccompromise, or need for cardioversion) and minor(decrease in blood pressure, transient symptoms, ornonsustained arrhythmias) events were recorded.

LV-GLS MEASUREMENTS. LV-GLS measurementswere obtained from baseline resting transthoracicechocardiograms using grayscale images recorded inapical 2-, 3-, and 4-chamber views. LV-GLS wasanalyzed offline by using Velocity Vector Imaging(Syngo VVI, Siemens Medical Solutions, MountainView, California), as previously described (25). All rawdata were stored in DICOM format without compres-sion. The frame rate was at least 30 frames/s. Aftermanual definition of the LV endocardial border, theendocardium was automatically tracked throughoutthe cardiac cycle. Global LV strain was obtained byaveraging all segmental strain values from all 3 apicalviews. Peak global strain was defined as the peaknegative value on the strain curve during the entirecardiac cycle. All measurements were made offlineby an investigator blinded to all clinical and demo-graphic information (A.M.). Measurements were per-formed and averaged over 3 cardiac cycles. Becausethe reported LV-GLS values are negative, a lowerabsolute number represented a worse value than ahigher number (Figure 1).

STATISTICAL ANALYSIS. Continuous variables areexpressed as mean � SD or median (interquartilerange) for skewed distributions, and they werecompared by using the Student t test or analysis ofvariance (for normally distributed variables) or theMann-Whitney U test (for non-normally distributedvariables). Categorical data are expressed as a per-centage and were compared by using the chi-squaretest or Fisher exact test, as appropriate. TheSpearman correlation coefficient was used to assesscorrelation between continuous variables. Intra-observer and interobserver variability for LV-GLSmeasurements was assessed by using intraclass cor-relation coefficients. To assess outcomes, a Cox pro-portional hazards analysis was performed. Wecreated a parsimonious model in which pre-specifiedrelevant variables associated with adverse outcomesin patients with primary MR were included (includingMV surgery). As we found in a previous study of pa-tients with primary MR, even though the STS scorehas only been validated to predict 30-day post-operative mortality, we used it in the survival anal-ysis because it is a composite of many factors that areknown to be associated with adverse post-operativeevents in the longer term (26).

We also assessed the reclassification of mortalityrisk by using a category-free integrated discrimina-tion index and continuous net reclassificationimprovement (27). In addition, the discriminativeabilities of various survival models were compared byusing the C-statistic. The functional relationshipbetween LV-GLS and the percentage of predictedMETs with the risk of death was assessed by using a

TABLE 2

Study Pop

Resting tr

LVEF, %

Indexed

Indexed

Indexed

Mitral E

Mitral le

RVSP, m

Tricuspi

None

Mild

Mode

Right ve

None

Mild

LV-GLS,

Exercise e

Maximu

>85% m

METs ac

% Age-

Age-/se

<85%

86%–

>100

Stress-i

Change

Decre

Uncha

Increa

Peak str

Values are m

ERO ¼end-diastolend-systolicequivalent;

Mentias et al. J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6

Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6

1978

parametric multiphase hazard model, as previouslydescribed (26,28,29). To assess the possible nonlinearrelationship between the percentage of predictedMETs, LV-GLS, and risk of death, covariate-predictedMETs and LV-GLS were modeled as a quadratic splinewith 6 knots at the 5th, 23rd, 41st, 59th, 77th, and 95thpercentile values of predicted METs and LV-GLS,respectively. Nomograms were used to depict theestimated hazard rate at specified time points across arange of covariate values.

Statistical analysis was performed by using SPSSversion 11.5 (IBM SPSS Statistics, IBM Corporation,Armonk, New York), Stata version 10.0 (StataCorp,College Station, Texas), SAS version 9.4 (SAS Insti-tute, Inc., Cary, North Carolina), and R 3.0.3

Resting and Post-Exercise Echocardiographic Data in theulation

No MV Surgery(n ¼ 260)

MV Surgery(n ¼ 477) p Value

ansthoracic echocardiography

61 � 2 62 � 2 0.52

LV-ESD, cm/m2 1.58 � 0.3 1.62 � 0.4 0.09

LV-EDD 2.51 � 0.6 2.53 � 0.3 0.13

left atrial area, cm/m2 13.2 � 4.0 13.7 � 4.0 0.22

RO area, cm2 0.41 � 0.3 0.47 � 0.2 <0.01

aflet flail 12 (5) 198 (42) <0.001

m Hg 30 � 11 31 � 12 0.21

d regurgitation 0.23

–trivial 152 (58) 302 (63)

59 (23) 119 (25)

rate or higher 43 (17) 62 (13)

ntricular dysfunction 0.54

257 (99) 472 (99)

3 (1) 5 (1)

% –21.5 � 2.0 –21.7 � 2.0 0.34

chocardiography

m predicted heart rate, % 95 � 12 96 � 21 0.41

aximum predicted heart rate 228 (88) 435 (89) 0.62

hieved 10.1 � 4.0 9.7 � 3.0 0.04

/sex-predicted METs 116 � 28 114 � 27 0.31

x-predicted METs 0.74

26 (10) 50 (11)

100% 42 (16) 88 (18)

% 192 (74) 339 (71)

nduced ischemia, % 11 (4) 19 (4) 0.5

in LV cavity size with stress 0.53

ased 239 (92) 432 (91)

nged 16 (6) 35 (7)

sed 5 (2) 10 (2)

ess RVSP, mm Hg 42 � 14 48 � 18 0.03

ean � SD or n (%).

effective regurgitant orifice; LV ¼ left ventricular; LV-EDD ¼ left ventricularic dimension; LVEF ¼ left ventricular ejection fraction; LV-ESD ¼ left ventriculardimension; LV-GLS ¼ left ventricular global longitudinal strain; MET ¼ metabolic

MV ¼ mitral valve; RVSP ¼ right ventricular systolic pressure.

(R Foundation for Statistical Computing, Vienna,Austria). A p value <0.05 was considered significant.

RESULTS

Baseline clinical characteristics for the study popula-tion are shown in Table 1. The mean STS score was verylow (1.5� 1.1). Resting and treadmill echocardiographicdata are shown in Table 2. According to the studydesign, all patients were asymptomatic with $3þ MRand had normal LV dimensions and preserved LVEF.Within the population, a majority (n ¼ 468 [64%]) ofthe patients had a dilated left atrium (area >20 cm2),and 258 (35%) patients had resting RVSP $35 mm Hg.The median LV-GLS in the study population was–21.7% (interquartile range [IQR]: –20.63% to –22.98%;maximal value –27.6%; minimum value –17.2%). Rele-vant clinical and imaging characteristics of the studypopulation, separated on the basis of LV-GLS better orworse than median, are given in Online Table 1. Theintraclass correlation coefficient for intraobserver(M.Y.D.) and interobserver (A.M. and M.Y.D.) repro-ducibility of resting LV-GLS measurements were0.89 (0.81 to 0.96) and 0.86 (0.71 to 0.92), respectively(both p<0.001). Therewas aweak correlation betweenLV-GLS and LVEF (r ¼ –0.10; p ¼ 0.04).

Patients terminated the stress test due to gener-alized fatigue. At peak stress, 40 patients also hadchest discomfort (none experienced dizziness orlightheadedness). There were no significant arrhyth-mias, syncope, or deaths during exercise. Of the30 patients with a segmental ischemic LV response,17 (57%) were in the left anterior descending terri-tory, 2 (7%) in the left circumflex territory, 7 (23%) inthe right coronary territory, and 4 (13%) in a multi-vessel distribution. In addition, 117 (16%) hadpost-stress RVSP $60 mm Hg. There was a weakbut significant correlation between LV-GLS andmaximum METS achieved (beta ¼ –0.26; p < 0.001) orthe percentage of age-/sex-predicted METs achieved(beta ¼ –0.21; p < 0.001).FOLLOW-UP DATA. In the study, 477 (65%) patientsunderwent MV surgery (439 [92%] underwent MVrepair and 38 [8%] underwent MV replacement), withmedian time to surgery (from stress) of 3 months(IQR: 1 to 15 months). Each of these patients had atleast a primary Class IIa indication according toguidelines prevalent at the time of surgery (post-exercise RVSP $60 mm Hg [n ¼ 117], flail mitral leaflet[n ¼ 198], and severely dilated left atrium [n ¼ 162]).In addition, 206 patients achieved 100% of theirage-/sex-predicted METs.

According to the study design, no patienthad preoperative AF as a driver of surgical decision

TABLE 3 Multivariable Cox Proportional Hazard Analysis in the

Study Population for Longer-Term Mortality

Hazard Ratio (95% CI) p Value

STS score (for every % increase) 1.14 (1.04–1.25) <0.001

Resting RVSP(for every 10–mm Hg increase)

1.35 (1.10–2.32) <0.001

LV-GLS (for every unit worsening) 1.60 (1.47–1.73) <0.001

% Age-/sex-predicted METs(for every 10% decrease)

1.13 (1.04–1.22) <0.001

MV surgery 0.82 (0.70–0.96) 0.01

Individual predictors that comprise the STS score (including age, sex, LVEF, andrenal function, among others) were not separately entered into the multivariableanalysis model. Due to exclusion criteria, baseline atrial fibrillation was notconsidered for multivariable analysis. When mitral effective regurgitant orifice areaor flail leaflet were forced into the multivariable model, the results were similar,and these variables were not significant. When LV-ESD was forced into the model,it was not significant (due to exclusion of patients with a dilated left ventricle,according to study design), and the results of the multivariable were similar.Similarly, when ischemic LV response was forced into the model, the results weresimilar. When peak stress RVSP was substituted for resting RVSP in multivariableanalysis, it was not significantly associated with mortality.

CI ¼ confidence interval; other abbreviations as in Tables 1 and 2.

J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6 Mentias et al.N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6 Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography

1979

making. The decision not to operate was made on thebasis of a discussion between the patient and theevaluating cardiologist/cardiac surgeon. No patienthad documented noncardiac comorbidities thatprevented their having MV surgery. There was nocorrelation between time to surgery and LV-GLS(beta ¼ 0.01; p ¼ 0.30); however, there was aweakly significant correlation with METs achieved(beta ¼ 0.10; p ¼ 0.01). Additional procedures per-formed at the time of MV surgery included coronaryartery bypass grafting (66 [9%]), left atrial appendageexcision (106 [14%]), tricuspid valve repair (22 [3%]),pulmonary vein isolation (28 [4%]), and the mazeprocedure (35 [5%]). During follow-up, 3 patientsneeded redo cardiac surgery (MV replacements).OUTCOMES AND SURVIVAL DATA. During a meanfollow-up of 8.3 � 3 years, 64 patients (9%) died. Inthis group, there were no documented comorbidities(e.g., cancer, renal dysfunction, pulmonary, neuro-logical) that could result in noncardiac deaths. Inaddition, there were 9 (1%) strokes, 11 (1%) transientischemic attacks, 6 (0.7%) myocardial infarctions, and24 (3%) patients with progression to congestive heartfailure. In the surgical group, there were no deaths atday 30 post-operatively, whereas 1 patient had astroke at day 29 post-operatively. In the total group,96 patients (13%) had new-onset AF during long-termfollow-up (beyond the 30-day post-operative period).The breakdown of New York Heart Association func-tional class at final follow-up was as follows: 622(84%) in Class I; 84 (10%) in Class II; 4 (0.5%) in ClassIII; and 1 (0.1%) in Class IV.

We subsequently performed Cox proportional haz-ards analysis for longer-term deaths. Of note, flailmitral leaflet (hazard ratio [HR]: 1.06; 95% confidenceinterval [CI]: 0.68 to 1.89; p¼0.6) and indexed LV end-systolic dimension (HR: 1.05; 95% CI: 0.63 to 1.82;p ¼ 0.80) according to study design were not associ-ated with mortality on univariable analysis. Onmultivariable Cox analysis, higher STS score (HR: 1.14;95% CI: 1.04 to 1.25), more abnormal LV-GLS (HR: 1.60;95% CI: 1.47 to 1.73), higher baseline RVSP (HR: 1.35;95% CI: 1.10 to 2.32), and lower percentage of age-/sex-predicted METs (HR: 1.13; 95% CI: 1.04 to 1.22) wereindependently associated with higher mortality,whereas MV surgery (HR: 0.82; 95% CI: 0.70 to 0.96)was associated with improved survival (all p < 0.01)(Table 3). Neither quadratic nor cubic transformationsof the percentage of age-/sex-predicted METs or LV-GLS were significant predictors of outcomes whenforced into the Cox model that already included thesevariables in a nontransformed form.

We subsequently tested whether the addition of thepercentage of age-/sex-predicted METs and LV-GLS

significantly improved risk stratification for longer-termmortality. The change in probabilities (expressedby integrated discrimination index and net reclassifi-cation improvement), presented as the difference inthe change in the probability of death among thosewho died and those who did not, is shown in Table 4.Addition of the percentage of age-/sex-predictedMETs and LV-GLS to the clinical model resulted insignificant reclassification of longer-term mortalityrisk. Similarly, addition of the percentage of age-/sex-predicted METs to the clinical model increased the C-statistic for longer-term mortality from 0.61 (0.53 to0.70) to 0.69 (0.59 to 0.80) (p < 0.01). Further additionof baseline LV-GLS increased the C-statistic to 0.78(0.65 to 0.90) (p < 0.01).

During follow-up, the breakdown (as number [%]of events) of longer-term mortality in the studypopulation, on the basis of achieving 100% age-/sex-predicted METs, was as follows: $100% of age-/sex-predicted METs, 37 (7%) of 531; and <100% of age-/sex-predicted METs, 27 (13%) of 206. Figure 2A dis-plays the Kaplan-Meier curves of longer-term out-comes in the study population, separated on the basisof age-/sex-predicted METs above or below 100%.Similarly, the breakdown of longer-term mortalityin the study population on the basis of the LV-GLSmedian was as follows: LV-GLS worse than median(–21.7%), 59 (16%) of 367 events; and LV-GLS betterthan median, 5 (1.4%) of 370 events. Figure 2B dis-plays the Kaplan-Meier curves of the longer-termoutcomes in the study population, separated on thebasis of LV-GLS above or below the median value.

Subsequently, to understand the interplay betweenLV-GLS and the percentage of age-/sex-predicted

TABLE 4 Reclassification of Patients Who Survived and Who Died During Follow-Up

A Reclassification on the basis of the clinical model (STS score, indexed LV-ESD, RVSP, and mitral ERO) plus LV-GLS

Patients who were alive during longer-term follow-up

Initial clinical model

Updated Model (Clinical þ LV-GLS)

(0-0.0578) (0.0578-0.067) (0.067-0.0827) (0.0827-1) % Reclassified

(0-0.0578) 133 10 18 20 27

(0.0578-0.067) 108 20 17 30 89

(0.067-0.0827) 97 15 26 36 85

(0.0827-1) 41 10 21 71 50

Patients who died during longer-term follow-up

Initial clinical model

Updated Model (Clinical þ LV-GLS)

(0-0.0578) (0.0578-0.067) (0.067-0.0827) (0.0827-1) % Reclassified

(0-0.0578) 1 2 0 0 67

(0.0578-0.067) 2 1 1 5 89

(0.067-0.0827) 2 0 0 8 100

(0.0827-1) 6 0 2 34 19

IDI, 0.11 (CI: 0.07-0.15); continuous NRI, 0.79 (CI: 0.56-1.03); both p < 0.001

B Reclassification on the basis of clinical model (as above) plus % predicted METs

Patients who were alive during longer-term follow-up

Initial clinical model

Updated Model (Clinical þ % Predicted METs)

(0-0.0652) (0.0652-0.0727) (0.0727-0.0847) (0.0847-1) % Reclassified

(0-0.0652) 105 12 15 46 41

(0.0652-0.0727) 104 10 12 49 94

(0.0727-0.0847) 105 9 11 45 94

(0.0847-1) 80 7 11 52 65

Patients who died during longer-term follow-up

Initial clinical model

Updated Model (Clinical þ % Predicted METs)

(0-0.0652) (0.0652-0.0727) (0.0727-0.0847) (0.0847-1) % Reclassified

(0-0.0652) 1 1 0 4 83

(0.0652-0.0727) 3 0 1 5 100

(0.0727-0.0847) 2 1 1 10 93

(0.0847-1) 6 2 2 25 29

IDI, 0.08 (CI: 0.4-0.11); continuous NRI, 0.60 (CI: 0.36-0.85); both p < 0.001

C Reclassification on the basis of the clinical model (as noted in Table 1) plus LV-GLS and % predicted METs

Patients who are alive during longer-term follow-up

Initial model(clinical þ LV-GLS)

Updated Model (Clinical þ LV-GLS þ % predicted METs)

(0-0.0475) (0.0475-0.0684) (0.0684-0.0945) (0.0945-1) % Reclassified

(0-0.0475) 137 14 10 14 22

(0.0475-0.0684) 118 18 11 28 90

(0.0684-0.0945) 88 28 23 37 87

(0.0945-1) 62 20 14 51 65

Patients who died during longer-term follow-up

Initial model(clinical þ LV-GLS)

Updated Model (Clinical þ LV-GLS þ % Predicted METs)

(0-0.0475) (0.0475-0.0684) (0.0684-0.0945) (0.0945-1) % Reclassified

(0-0.0475) 3 0 0 6 67

(0.0475-0.0684) 1 1 0 7 89

(0.0684-0.0945) 1 2 0 5 100

(0.0945-1) 3 3 0 32 16

IDI, 0.17 (95% CI: 0.14-0.22); continuous NRI, 0.78 (95% CI: 0.54-1.02); both p < 0.001

Numbers in parentheses represent probabilities from 0 to 1. Numbers in columns (under the probabilities) represent numbers of patients who were reclassified.

IDI ¼ integrated discrimination improvement; NRI ¼ net reclassification improvement; other abbreviations as in Tables 1, 2, and 3.

Mentias et al. J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6

Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6

1980

FIGURE 2 Kaplan-Meier Survival Curves for the Study Population

Age-gender predicted METs achieved

Age-gender predicted METs achieved

≥ 100%<100%

≥ 100%<100%

p=0.002

Blue: LV-GLS better thanmedian (-21.7%)Orange: LV-GLS worse than medianp<0.001

0.4

0.3

0.2

0.1

0.0

0.4

0.3

0.2

0.1

0.0

.00 3.00 6.00 9.00 12.00Follow Up (Years)

.00 3.00 6.00 9.00 12.00Follow Up (Years)

Numbers at risk Numbers at riskResting LV-GLSBetter than medianWorse than median

531206

522 370367

367350

238261

131181

6180195

361138

22389

9150

Cum

ulat

ive

Prop

ortio

n of

Dea

th

Cum

ulat

ive

Prop

ortio

n of

Dea

th

A B

The figure shows Kaplan-Meier survival curves separated on the basis of (A) age-/sex-predicted exercise metabolic equivalents (METs)

higher or lower than 100% and (B) according to left ventricular global longitudinal strain (LV-GLS) better or worse than median (–21.7%).

A significantly lower longer-term mortality was noted in patients who achieved $100% age-/sex-predicted METs versus those who did not

(7% vs. 13%; p < 0.01) and whose LV-GLS was better than median versus not better than median (1.4% vs. 16%; p < 0.001).

J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6 Mentias et al.N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6 Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography

1981

METs, we created 4 subgroups. The proportion ofdeaths, on the basis of these 4 subgroups, was as fol-lows: 1) LV-GLS better than median and $100% ofage-/sex-predicted METs, 2 (0.7%) of 282; 2) LV-GLSbetter than median and <100% of age-/sex-predictedMETs, 3 (3%) of 88; 3) LV-GLS worse than median and$100% of age-/sex-predicted METs, 35 (14%) of 249;and 4) LV-GLS worse than median and <100% ofage/sex-predicted METs, 24 (20%) of 118. Figure 3illustrates the survival curves according to LV-GLSand achieving 100% of age-/sex-predicted METs(p < 0.001). Similarly, to understand the interplaybetween LV-GLS and MV surgery, we created 4 sub-groups. The proportion of deaths on the basis of these4 subgroups was as follows: 1) LV-GLS better thanmedian and MV surgery, 4 (1.6%) of 244; 2) LV-GLSbetter than median and no MV surgery, 1 (0.8%) of126; 3) LV-GLS worse than median and MV surgery, 32(14%) of 233; and 4) LV-GLS worse than median and noMV surgery, 27 (20%) of 134. Figure 4 illustrates thesurvival curves according to LV-GLS and MV surgery(p < 0.001).

The Central Illustration presents data on 5-yearmortality by using a quadratic spline with 6 knots.Based on visual analysis of the data, we showedthat patients with LV-GLS better than –21% hadgood 5-year survival. However, the risk of death

significantly and continuously increased as LV-GLSworsened below –21%. Similarly, patients who ach-ieved age-/sex-predicted METs approximately 100%or better had much better survival at 5 years.

When patients with CAD were excluded from themultivariable Cox proportional hazards survivalanalysis (total n ¼ 661; number of deaths ¼ 51), theresults were similar. Higher STS score (HR: 1.10; 95%CI: 1.02 to 1.29), higher RVSP (HR: 1.29; 95% CI: 1.07 to2.47), more abnormal LV-GLS (HR: 1.56; 95% CI: 1.42 to1.75), and lower percentage of age-/sex-predictedMETs (HR: 1.10; 95% CI: 1.02 to 1.24) were associatedwith increased mortality, whereas MV surgery (HR:0.83; 95% CI: 0.73 to 0.97) (p ¼ 0.01) was associatedwith improved survival.

DISCUSSION

In the present study of asymptomatic patients withsignificant primary MR, normal LV dimensions, andpreserved LVEF who underwent rest and post-exercise echocardiography, we found that worseningbaseline resting LV-GLS and reduced exercise capac-ity were independently associated with mortality,providing additive (rather than duplicative) prog-nostic utility to previously known predictors. Inaddition, exercise capacity and LV-GLS sequentially

FIGURE 3 Kaplan-Meier Survival Curves for the Study Population, Separated Into

4 Subgroups Created on the Basis of Resting LV-GLS Better or Worse Than Median

and Age-/Sex-Predicted METs Higher or Lower Than 100%

Blue: Resting LV-GLS better than median, % predicted METs ≥100Orange: Resting LV-GLS better than median, % predicted METs <100

Red: Resting LV-GLS worse than median, % predicted METs <100p<0.001

Gray: Resting LV-GLS worse than median, % predicted METs ≥100

0.4

0.3

0.2

0.1

0.0

.00 3.00 6.00 9.00 12.00Follow Up (Years)

Numbers at riskBlue: Group 1 282 279 176 97 43Orange: Group 2 88 88 62 34 18Gray: Group 3 249 243 185 126 48Red: Group 4 118 107 76 55 32

Cum

ulat

ive

Prop

ortio

n of

Dea

th

Group 1, resting LV-GLS better than median and predicted METs $100%; Group 2, resting

LV-GLS better than median and predicted METs <100%; Group 3, resting LV-GLS worse

than median and predicted METs $100%; and Group 4, resting LV-GLS worse than median

and predicted METs <100%. According to Kaplan-Meier survival curves, longer-term

mortality was the lowest in the subgroup with LV-GLS better than median and $100%

age-/sex-predicted METs (0.7%), whereas it was the highest in those with LV-GLS worse

than median and <100% age/sex-predicted METs (20%). Abbreviations as in Figure 2.

Mentias et al. J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6

Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6

1982

improved risk classification in these patients, inde-pendent of known prognostic variables. Furthermore,patients who had a combination of resting LV-GLSvalues worse than median and achieved <100% ofage-/sex-predicted METs had significantly worsesurvival during longer-term follow-up. Similarly, pa-tients with resting LV-GLS values worse than medianwho did not undergo MV surgery had significantlyworse survival during longer-term follow-up. We alsofound that in the study population, at (and worsethan) an approximate LV-GLS of –21% and less thanapproximately 100% of age-/sex-predicted METs,there was a progressive increase in longer-term mor-tality. However, as shown in the Central Illustration,the impact of resting LV-GLS on mortality was moreprominent than that of METs achieved.

Currently, the definition of “normal” LV-GLSvalues in subjects free of cardiovascular diseaseremains to be fully elucidated. On the basis of a meta-analysis of healthy subjects free of cardiovasculardisease, the cutoff for normal LV-GLS value was

–19.7% (21,30), which is slightly lower than the cutoff(approximately –21%) observed in the present study.Our cutoff was also significantly higher than thenormal LV-GLS values (–17.3%) reported in an earlierstudy of healthy individuals free of cardiovasculardisease in which only Velocity Vector Imaging soft-ware was used (similar to the present study) (31). Thisprevious study also demonstrated that LV-GLS valuesobtained by using Velocity Vector Imaging softwarewere similar for different echocardiography vendors(similar to our study). However, a recent study foundthat although there is a very strong correlation be-tween LV-GLS values obtained by using differentstrain analysis packages in the same subjects, theabsolute mean values vary between –18% and –21.5%(with a mean of –20% for the Siemens strain software)(32); however, this study included patients with awide spectrum of LV systolic function, rather thanhealthy individuals free of cardiovascular disease.The higher-than-normal LV-GLS values in chronicsevere MR seem intuitive, as the pathophysiologicalstate (reduced afterload and increased preload)results in a state of hypernormal LV function. Thisfinding also suggests that LV-GLS is likely load-dependent and might have to be corrected for LVvolumes in future analyses. In the present study,when LV end-systolic dimension was forced into thesurvival analysis, it was not significant, likely due toexclusion of patients with a dilated left ventricle.

As shown in the present study of asymptomaticpatients with preserved LVEF, there is perceivedutility in studying more sensitive markers of earlyregional myocardial function, such as LV-GLS, as ithas been shown that subtle LV dysfunction may occurin patients before any drop in ejection fraction isevident. A previous study reported that LV biopsyfindings in asymptomatic patients with chronic severeMR and normal ejection fraction at the time of MVsurgery demonstrate more myofibrillar degeneration,and higher xanthine oxidase and lipofuscin deposi-tion, indicating more oxidative stress compared withnormal subjects (33). Indeed, in several previous re-ports, impaired LV-GLS in asymptomatic patientswith severe MR and preserved ejection fraction hasbeen shown to predict post-operative LV dysfunctionafter MV surgery (34–37). Similar to the present study,a common finding in these reports was that baselineLV-GLS was higher than has been reported in normalsubjects, with the median ranging from –19.9% to–21.7%. However, these reports had relatively smallpopulation sizes and short follow-up durations,without survival analysis or stress-testing data. Aprevious study reported resting and stress LV-GLSand found a strong correlation of GLS indexed to LV

FIGURE 4 Kaplan-Meier Survival Curves for the Study Population, Sepa-

rated Into 4 Subgroups Created on the Basis of Resting LV-GLS Better

or Worse Than Median and Mitral Surgery During Follow-up Versus Not

Blue: Resting LV-GLS better than median, surgeryOrange: Resting LV-GLS better than median, no surgery

Red: Resting LV-GLS worse than median, no surgeryp<0.001

Gray: Resting LV-GLS worse than median, surgery

0.4

0.3

0.2

0.1

0.0

.00 3.00 6.00 9.00 12.00Follow Up (Years)

Numbers at riskBlue: Group 1 244 241 164 95 36Orange: Group 2 126 126 74 36 25Gray: Group 3 233 225 180 131 59Red: Group 4 134 125 81 50 25

Cum

ulat

ive

Prop

ortio

n of

Dea

th

Group 1, resting LV-GLS better than median and surgery; Group 2, resting LV-

GLS better than median and no surgery; Group 3, resting LV-GLS worse than

median and surgery; and Group 4, resting LV-GLS worse than median and no

surgery. According to Kaplan-Meier survival curves, longer-term mortality was

lowest in the subgroup with LV-GLS better than median that underwent MV

surgery (1.6%), whereas it was highest in the group with LV-GLS worse than

median that did not undergo MV surgery (20%). Abbreviations as in Figure 2.

J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6 Mentias et al.N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6 Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography

1983

end-systolic dimension with post-operative ejectionfraction; however, no survival analysis was reported(38). A large study reported the association betweenresting LV-GLS and mortality, but no stress testingdata were reported (20); the study also included avariety of patients undergoing cardiac surgery,not just MV surgery, with only 96 patients who hadsurgical MV disease, and the outcome was earlypost-operative death, not long-term mortality. Onestudy showed the utility of rest and post-exerciseechocardiography, reporting a higher event rate inpatients with an abnormal GLS contractile reserve;however, the population size was small (n ¼ 115), andthe investigators reported a composite outcome ofdeath, need for MV surgery, and heart failure admis-sion (19). Another study from the same group reporteda higher composite event rate in the setting of anabnormal B-type natriuretic peptide and LV-GLS (39).The present study had a significantly larger popula-tion size, a much longer follow-up, and a hardendpoint of mortality, along with studying thesequentially incremental impact of LV-GLS and exer-cise capacity.

In symptomatic patients or those with AF/overt LVdysfunction, the decision regarding MV surgicaltiming is already clear. As a result, we only includedasymptomatic patients with preserved LVEF in thepresent study, as we intuited that this is the popula-tion in whom additional information may have themost clinical utility in determining long-term risk, dueto conflicting data (6–8). It could also aid in helpingdetermine the timing of surgery, especially in thosepatients who are risk-averse and wish to delay surgeryfor as long as possible. In the absence of randomizedcontrolled trials, the guidelines rely mainly on obser-vational outcome studies (4,5). Factors such as symp-tomatic deterioration (3,40), severity of disease (41),increased LV or left atrial dimensions (42,43), anddevelopment of AF (44), pulmonary hypertension(45), or LV dysfunction (3,46) have all been associatedwith worse clinical outcomes. Nevertheless, thesestudies vary in the definition of outcome (3,7,43,45)and have included limited numbers of patients (7),only patients with flail leaflets (42,43,45), and patientswith various etiologies (40,46) or severities (3,45) ofMR. In addition to the previously mentioned restingvariables, exercise echocardiography can aid insymptom evaluation, risk stratification, and decision-making (for appropriate delaying of surgical timing,including in those patients with mid–late systolic MR)(11,13,14). In addition, decreased functional capacityduring exercise testing (15), exercise-induced pulmo-nary hypertension (16), and MR severity (17) have beenassociated with worse outcomes.

On the basis of our results, it seems that reductionin exercise capacity and an abnormal resting LV-GLSmay provide incremental and additive prognosticvalue, as well as aid in reclassification of risk inpatients with significant MR before the onset ofovert LV systolic dysfunction or symptoms. RestingLV-GLS worse than –21% was associated with pro-gressively higher longer-term mortality. Further-more, patients with resting LV-GLS values worse thanmedian had significantly worse longer-term survival,especially without MV surgery. These noninvasiveparameters could potentially aid in further optimi-zation of surgical timing. However, the presentfindings are hypothesis-generating and need to bevalidated prospectively in the form of a multicentertrial.

STUDY LIMITATIONS. This observational studyinvolved a single center, with its inherent selectionbias. We only included asymptomatic patientswith $3þ primary MR who subsequently underwent

CENTRAL ILLUSTRATION Mitral Regurgitation, Resting Strain, and Stress Echocardiography:Nomogram of Estimated Risk of Death at 5 Years

Mentias, A. et al. J Am Coll Cardiol. 2016;68(18):1974–86.

Solid line represents the 5-year parametric estimates of instantaneous risk of death. Solid lines are enclosed by a 68% confidence interval

(dotted lines). To assess the possible nonlinear relationship between percent predicted exercise metabolic equivalents (METs), left ventricular

global longitudinal strain (LV-GLS), and risk of death, covariate predictedMETs and LV-GLSwasmodeled as a quadratic spline with 6 knots at the

5th, 23rd, 41st, 59th, 77th, and 95th percentile values of predicted METs and LV-GLS, respectively. From visual analysis of the curves, patients

with LV-GLS better than –21% had good 5-year survival, and the risk of death significantly and continuously increased as LV-GLSworsened below

–21%. Similarly, patients who achieved age-/sex-predicted METs approximately 100% or better had much better survival at 5 years.

Mentias et al. J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6

Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6

1984

PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: In asymptomatic

patients with significant primary MR and preserved LVEF,

reduced exercise capacity and worsening LV-GLS were indepen-

dently associated with mortality, independent of and providing

added prognostic value over known predictors.

TRANSLATIONAL OUTLOOK: More research is needed to

develop and validate algorithms that incorporate resting LV

strain and exercise capacity in selecting the optimum time for

surgical intervention in asymptomatic patients with primary MR

and preserved LVEF.

J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6 Mentias et al.N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6 Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography

1985

rest and post-exercise echocardiography. At our cen-ter, during the time frame of this study, approxi-mately 8,500 patients were evaluated for significantprimary MR, of whom only approximately 10%underwent treadmill echocardiography, as there wasambiguity in terms of symptoms and more informa-tion was needed to aid in timing of MV surgery, asreported previously (11). Therefore, patients whowere more symptomatic at baseline did not undergothe test and are not included in the study. However,this approach has inherently helped us to focus on apopulation in whom uncertainties exist regardingoptimal timing of surgical management. We includedpatients over a broad time frame, and not alldata (biomarker and quantitative right ventricularmeasurements) were obtained in all patients. Inaddition, we do not report results of LV-GLS at peakstress. However, it can be argued that LV-GLS mea-surement at peak stress is difficult and would resultin many more excluded patients (due to nontrackedsegments), making an outcomes study less meaning-ful. In addition, exercise capacity is the output ofoverall function of the cardiovascular, respiratory,and musculoskeletal systems. Other possible etiol-ogies of impaired exercise capacity could have playedan important role in predicting outcomes in thesepatients. The study is potentially underpowered tomake conclusive assertions about subgroup analyses,and a larger, prospective study is needed. Our insti-tution is well experienced in MV surgery, with anextremely low rate of adverse events, and resultsfrom this cohort might not be generalizable across allother centers. We report all-cause mortality as theprimary endpoint, as opposed to cardiac mortality.However, it has been shown previously that all-causemortality is more objective and unbiased thancardiac mortality (47). In any case, in this study ofasymptomatic patients with significant MR, asmentioned previously, there were no patients with anovert noncardiac comorbidity that could account fordeath.

CONCLUSIONS

In asymptomatic patients with significant primary MRand preserved LVEF who underwent rest and stressechocardiography, we showed that reduced exercisecapacity and worsening LV-GLS were independentlyassociated with mortality, providing additive (ratherthan duplicative) prognostic utility to previouslyknown predictors. In addition, exercise capacity andLV-GLS sequentially improved risk classification inthese patients, independent of known prognosticvariables. Also, resting LV-GLS worse than –21%seems to be associated with progressively higherlonger-term mortality. However, these findings needprospective validation.

ACKNOWLEDGMENTS The authors acknowledgePenny Houghtaling, PhD, Department of QuantitativeHealth Sciences, for her statistical support for thequadratic spline analysis.

REPRINT REQUESTS AND CORRESPONDENCE: Dr.Milind Y. Desai, Heart and Vascular Institute,Department of Cardiovascular Medicine, ClevelandClinic, 9500 Euclid Avenue, Desk J1-5, Cleveland,Ohio 44195. E-mail: desaim2@ccf.org.

RE F E RENCE S

1. Freed LA, Levy D, Levine RA, et al. Prevalenceand clinical outcome of mitral-valve prolapse.N Engl J Med 1999;341:1–7.

2. Iung B, Baron G, Butchart EG, et al.A prospective survey of patients with valvularheart disease in Europe: the Euro Heart Surveyon Valvular Heart Disease. Eur Heart J 2003;24:1231–43.

3. Ling LH, Enriquez-Sarano M, Seward JB, et al.Clinical outcome of mitral regurgitation due to flailleaflet. N Engl J Med 1996;335:1417–23.

4. Nishimura RA, Otto CM, Bonow RO, et al. 2014AHA/ACC guideline for the management of pa-tients with valvular heart disease: a report of theAmerican College of Cardiology/American HeartAssociation Task Force on Practice Guidelines.J Am Coll Cardiol 2014;63:e57–185.

5. Vahanian A, Baumgartner H, Bax J, et al.Guidelines on the management of valvular heartdisease. Eur Heart J 2007;28:230–68.

6. Topilsky Y, Suri R, Schaff HV, et al. When tointervene for asymptomatic mitral valve

regurgitation. Semin Thorac Cardiovasc Surg 2010;22:216–24.

7. Rosenhek R, Rader F, Klaar U, et al. Outcome ofwatchful waiting in asymptomatic severe mitralregurgitation. Circulation 2006;113:2238–44.

8. Kang DH, Kim JH, Rim JH, et al. Comparison ofearly surgery versus conventional treatment inasymptomatic severe mitral regurgitation. Circu-lation 2009;119:797–804.

9. Suri RM, Vanoverschelde JL, Grigioni F, et al.Association between early surgical intervention vs

Mentias et al. J A C C V O L . 6 8 , N O . 1 8 , 2 0 1 6

Mitral Regurgitation, Resting Strain, and Post-Exercise Echocardiography N O V E M B E R 1 , 2 0 1 6 : 1 9 7 4 – 8 6

1986

watchful waiting and outcomes for mitral regur-gitation due to flail mitral valve leaflets. JAMA2013;310:609–16.

10. Yazdchi F, Koch CG, Mihaljevic T, et al.Increasing disadvantage of “watchful waiting” forrepairing degenerative mitral valve disease. AnnThorac Surg 2015;99:1992–2000.

11. Naji P, Griffin BP, Asfahan F, et al. Predictors oflong-term outcomes in patients with significantmyxomatous mitral regurgitation undergoingexercise echocardiography. Circulation 2014;129:1310–9.

12. Quintana E, Suri RM, Thalji NM, et al. Left ven-tricular dysfunction after mitral valve repair—thefallacy of “normal” preoperative myocardial func-tion. J Thorac Cardiovasc Surg 2014;148:2752–60.

13. Naji P, Griffin BP, Barr T, et al. Importance ofexercise capacity in predicting outcomes anddetermining optimal timing of surgery in signifi-cant primary mitral regurgitation. J Am HeartAssoc 2014;3:e001010.

14. Naji P, Asfahan F, Barr T, et al. Impact ofduration of mitral regurgitation on outcomes inasymptomatic patients with myxomatous mitralvalve undergoing exercise stress echocardiogra-phy. J Am Heart Assoc 2015;4:e001348.

15. Messika-Zeitoun D, Johnson BD, Nkomo V,et al. Cardiopulmonary exercise testing determi-nation of functional capacity in mitral regurgita-tion: physiologic and outcome implications. J AmColl Cardiol 2006;47:2521–7.

16. Magne J, Lancellotti P, Piérard LA. Exercisepulmonary hypertension in asymptomatic degen-erative mitral regurgitation. Circulation 2010;122:33–41.

17. Magne J, Lancellotti P, Piérard LA. Exercise-induced changes in degenerative mitral regurgi-tation. J Am Coll Cardiol 2010;56:300–9.

18. Marwick TH, Leano RL, Brown J, et al. Myocar-dial strain measurement with 2-dimensionalspeckle-tracking echocardiography: definition ofnormal range. J Am Coll Cardiol Img 2009;2:80–4.

19. Magne J, Mahjoub H, Dulgheru R, et al. Leftventricular contractile reserve in asymptomaticprimary mitral regurgitation. Eur Heart J 2014;35:1608–16.

20. Ternacle J, Berry M, Alonso E, et al. Incre-mental value of global longitudinal strain forpredicting early outcome after cardiac surgery. EurHeart J Cardiovasc Imaging 2013;14:77–84.

21. Lang RM, Badano LP, Mor-Avi V, et al. Rec-ommendations for cardiac chamber quantificationby echocardiography in adults: an update from theAmerican Society of Echocardiography and theEuropean Association of Cardiovascular Imaging.J Am Soc Echocardiogr 2015;28:1–39.e14.

22. Zoghbi WA, Enriquez-Sarano M, Foster E, et al.Recommendations for evaluation of the severity ofnative valvular regurgitation with two-dimensionaland Doppler echocardiography. J Am Soc Echo-cardiogr 2003;16:777–802.

23. Rudski LG, Lai WW, Afilalo J, et al. Guidelinesfor the echocardiographic assessment of the rightheart in adults: a report from the American So-ciety of Echocardiography endorsed by the

European Association of Echocardiography, aregistered branch of the European Society ofCardiology, and the Canadian Society of Echo-cardiography. J Am Soc Echocardiogr 2010;23:685–713; quiz 786–8.

24. Pellikka PA, Nagueh SF, Elhendy AA, et al.American Society of Echocardiography recom-mendations for performance, interpretation, andapplication of stress echocardiography. J Am SocEchocardiogr 2007;20:1021–41.

25. Kusunose K, Goodman A, Parikh R, et al. In-cremental prognostic value of left ventricularglobal longitudinal strain in patients with aorticstenosis and preserved ejection fraction. Circ Car-diovasc Imaging 2014;7:938–45.

26. Mentias A, Patel K, Patel H, et al. Effect ofpulmonary vascular pressures on long-termoutcome in patients with primary mitral regurgi-tation. J Am Coll Cardiol 2016;67:2952–61.

27. Pencina MJ, D’Agostino RB Sr.,D’Agostino RB Jr., et al. Evaluating the addedpredictive ability of a new marker: from area underthe ROC curve to reclassification and beyond. StatMed 2008;27:157–72; discussion 207–12.

28. Blackstone EH, Naftel DC, Turner ME Jr. Thedecomposition of time-varying hazard into phases,each incorporating a separate stream of concom-itant information. J Am Stat Assoc 1986;81:615–24.

29. Harrell FE Jr. Regression Modeling Strategies:With Applications to Linear Models, LogisticRegression and Survival Analysis. New York, NY:Springer-Verlag, Inc., 2001.

30. YingchoncharoenT,Agarwal S, Popovi�c ZB, et al.Normal ranges of left ventricular strain: a meta-analysis. J Am Soc Echocardiogr 2013;26:185–91.

31. Fine NM, Shah AA, Han IY, et al. Left and rightventricular strain and strain rate measurement innormal adults using velocity vector imaging: anassessment of reference values and intersystemagreement. Int J Cardiovasc Imaging 2013;29:571–80.

32. Farsalinos KE, Daraban AM, Ünlü S, et al.Head-to-head comparison of global longitudinalstrain measurements among nine different ven-dors: the EACVI/ASE Inter-Vendor ComparisonStudy. J Am Soc Echocardiogr 2015;28:1171–81, e2.

33. Ahmed MI, Gladden JD, Litovsky SH, et al.Increased oxidative stress and cardiomyocytemyofibrillar degeneration in patients with chronicisolated mitral regurgitation and ejection fraction>60%. J Am Coll Cardiol 2010;55:671–9.

34. Pandis D, Sengupta PP, Castillo JG, et al.Assessment of longitudinal myocardial mechanicsin patients with degenerative mitral valve regur-gitation predicts postoperative worsening of leftventricular systolic function. J Am Soc Echo-cardiogr 2014;27:627–38.

35. Witkowski TG, Thomas JD, Debonnaire PJ,et al. Global longitudinal strain predicts left ven-tricular dysfunction after mitral valve repair. EurHeart J Cardiovasc Imaging 2013;14:69–76.

36. Lancellotti P, Cosyns B, Zacharakis D, et al.Importance of left ventricular longitudinal func-tion and functional reserve in patients withdegenerative mitral regurgitation: assessment by

two-dimensional speckle tracking. J Am SocEchocardiogr 2008;21:1331–6.

37. Song JM, Kang SH, Lee EJ, et al. Echocardio-graphic predictors of left ventricular function andclinical outcomes after successful mitral valverepair: conventional two-dimensional versusspeckle-tracking parameters. Ann Thorac Surg2011;91:1816–23.

38. Donal E, Mascle S, Brunet A, et al. Predictionof left ventricular ejection fraction 6 months aftersurgical correction of organic mitral regurgitation:the value of exercise echocardiography anddeformation imaging. Eur Heart J Cardiovasc Im-aging 2012;13:922–30.

39. Magne J, Mahjoub H, Pierard LA, et al. Prog-nostic importance of brain natriuretic peptide andleft ventricular longitudinal function in asymp-tomatic degenerative mitral regurgitation. Heart2012;98:584–91.

40. Tribouilloy CM, Enriquez-Sarano M,Schaff HV, et al. Impact of preoperative symp-toms on survival after surgical correction oforganic mitral regurgitation: rationale for opti-mizing surgical indications. Circulation 1999;99:400–5.

41. Enriquez-Sarano M, Avierinos JF, Messika-Zeitoun D, et al. Quantitative determinants of theoutcome of asymptomatic mitral regurgitation.N Engl J Med 2005;352:875–83.

42. Tribouilloy C, Grigioni F, Avierinos JF, et al.,MIDA Investigators. Survival implication of leftventricular end-systolic diameter in mitral regur-gitation due to flail leaflets: a long-term follow-upmulticenter study. J Am Coll Cardiol 2009;54:1961–8.

43. Rusinaru D, Tribouilloy C, Grigioni F, et al.,Mitral Regurgitation International Database (MIDA)Investigators. Left atrial size is a potent predictor ofmortality in mitral regurgitation due to flail leaflets:results from a large international multicenter study.Circ Cardiovasc Imaging 2011;4:473–81.

44. Grigioni F, Avierinos JF, Ling LH, et al. Atrialfibrillation complicating the course of degener-ative mitral regurgitation: determinants andlong-term outcome. J Am Coll Cardiol 2002;40:84–92.

45. Barbieri A, Bursi F, Grigioni F, et al. Prog-nostic and therapeutic implications of pulmonaryhypertension complicating degenerative mitralregurgitation due to flail leaflet: a multicenterlong-term international study. Eur Heart J 2011;32:751–9.

46. Enriquez-Sarano M, Tajik AJ, Schaff HV, et al.Echocardiographic prediction of survival aftersurgical correction of organic mitral regurgitation.Circulation 1994;90:830–7.

47. Lauer MS, Blackstone EH, Young JB, et al.Cause of death in clinical research: time for areassessment? J Am Coll Cardiol 1999;34:618–20.

KEY WORDS global longitudinal strain,outcomes, post-exercise echocardiography

APPENDIX For a supplemental table, pleasesee the online version of this article.