J A C C : C A R D I O V A S C U L A R I M A G I N G V O L . 9 , N O . 6 , 2 0 1 6

ª 2 0 1 6 B Y T H E AM 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 UN DA T I O N I S S N 1 9 3 6 - 8 7 8 X / $ 3 6 . 0 0

P U B L I S H E D B Y E L S E V I E R h t t p : / / d x . d o i . o r g / 1 0 . 1 0 1 6 / j . j c m g . 2 0 1 5 . 0 6 . 0 2 0

ORIGINAL RESEARCH

Accurate Quantification Methods forAortic Insufficiency Severity in PatientsWith LVADRole of Diastolic Flow Acceleration and Systolic-to-DiastolicPeak Velocity Ratio of Outflow Cannula

Jonathan Grinstein, MD,a Eric Kruse, BS, RDCS,a Gabriel Sayer, MD,a Savitri Fedson, MD,a Gene H. Kim, MD,a

Ulrich P. Jorde, MD,b Colleen Juricek, RN,a,c Takeyoshi Ota, MD, PHD,a,c Valluvan Jeevanandam, MD,a,c

Roberto M. Lang, MD,a Nir Uriel, MDa

ABSTRACT

Fro

Me

Dr

Re

Ma

OBJECTIVES The aim of this study was to develop a technique to measure regurgitant flow throughout the entire

cardiac cycle that would more accurately measure aortic insufficiency (AI) severity in patients with continuous-flow left

ventricular assist devices (CF-LVADs).

BACKGROUND AI is a common problem after CF-LVAD implantation. Current echocardiographic evaluation of AI does

not take into account the unique flow properties present in patients with CF-LVADs.

METHODS In this prospective study, patients with LVADs who had varying degrees of AI (N ¼ 20) underwent simul-

taneous right-sided heart catheterization (RHC) and transthoracic echocardiography (TTE). Regurgitant fraction (RF)

across the aortic valve was calculated by subtracting the cardiac output obtained using the Fick method from the total

systemic flow measured using the sum of the product of the velocity time integral and the cross-sectional area of the

LVAD outflow cannula and aortic valve, respectively. The RFs were then compared with the following: 1) traditional TTE

grading parameters; and 2) new TTE parameters unique to LVAD physiology, namely the diastolic flow acceleration and

the systolic-to-diastolic peak velocity (S/D) ratio of the LVAD outflow cannula.

RESULTS Patients without evidence of AI had an RF approaching zero (2.4 � 4.6%). Patients with trace and mild AI

had an RF of 31.0 � 5.4%, whereas patients with moderate or severe AI had an RF of 45.8 � 3.6%. RF correlated better

with pulmonary capillary wedge pressure (PCWP) than with vena contracta (correlation coefficient [R] ¼ 0.73 vs. 0.56).

The new TTE parameters (S/D ratio and diastolic acceleration) highly correlated with RF (R ¼ 0.91 and 0.94, respectively)

and more strongly correlated with PCWP than did vena contracta (R ¼ 0.82 and 0.65 vs. 0.56).

CONCLUSIONS RF measured by simultaneous RHC and TTE better correlates with clinical filling pressures than do

traditional TTE parameters and may identify significant AI that could be underestimated using conventional measures.

Novel TTE parameters, unique to CF-LVAD physiology, better correlate with RF and filling pressures than do our current

TTE measurements. (J Am Coll Cardiol Img 2016;9:641–51) © 2016 by the American College of Cardiology Foundation.

m the aDepartment of Medicine, University of Chicago Medical Center, Chicago, Illinois; bDivision of Cardiology, Montefiore

dical Center, New York, New York; and the cDepartment of Surgery, University of Chicago Medical Center, Chicago, Illinois.

. Uriel is a consultant to HeartWare and Thoratec. Dr. Jeevanandam is a scientific advisor to Thoratec, HeartWare, and

liant Heart. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.

nuscript received May 27, 2015; revised manuscript received June 23, 2015, accepted June 25, 2015.

ABBR EV I A T I ON S

AND ACRONYMS

AI = aortic insufficiency

CF-LVAD = continuous-flow

left ventricular assist device

CSA = cross-sectional area

LVAD = left ventricular assist

device

PCWP = pulmonary capillary

wedge pressure

RF = regurgitant fraction

RHC = right-sided heart

catheterization

S/D ratio = systolic-to-

diastolic peak velocity ratio

TTE = transthoracic

echocardiography

VTI = velocity time integral

Grinstein et al. J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6

Aortic Insufficiency Quantification With LVAD J U N E 2 0 1 6 : 6 4 1 – 5 1

642

A ortic insufficiency (AI) is a commoncomplication after continuous-flowleft ventricular assist device

(CF-LVAD) implantation; in approximately1 in 4 patients, at least mild to moderateAI will develop within 1 year of implanta-tion (1–3). It is hypothesized that AI de-velops from failure of aortic valve openingduring LVAD support that leads to aorticvalve commissural fusion and leaflet deteri-oration (1,2,4). Furthermore, the constantunloading of the left ventricle by the LVADresults in a reversed transaortic valvularpressure gradient that predisposes to pro-gressive worsening of AI from increasedshear stress (5). Left untreated, de novo AIafter LVAD implantation can lead to clinicalheart failure and the need for aortic valve

repair, replacement, or closure or urgent cardiactransplantation in up to 50% of patients within 6months of development of symptomatic moderateor greater AI (1,6,7).

SEE PAGE 652

Whereas in a native heart, AI occurs in diastole,in an LVAD-implanted heart, AI is pancyclic, occur-ring throughout systole and diastole in response tothe constant positive transaortic pressure gradient(1,8–10). Traditional echocardiographic indices of AIseverity (i.e., vena contracta, jet width/left ventricularoutflow tract [LVOT] diameter, and proximal iso-velocity surface area) do not take into considerationthe pancyclic nature of AI jets in patients with LVADsand are less reliable with eccentric regurgitantjets, which are commonly encountered in LVAD-associated AI (11,12). Furthermore, the current echo-cardiographic parameters used for grading AI severityhave not been prospectively evaluated in theunique flow patterns that are associated with LVADimplantation.

To our knowledge, this is the first study to evaluatethe accuracy of traditional echocardiography mea-sures of AI severity in patients with LVADs. Here, wedirectly compare the traditional echocardiographyparameters of vena contracta and visual estimationwith RF measured by right-sided heart catheterization(RHC) and Doppler echocardiography and then corre-late these parameters with pulmonary capillary wedgepressure (PCWP). Furthermore, we test the accuracy of2 new, noninvasive methods for measuring AI in pa-tients with LVADs that were developed to reflect thepancyclic nature and constant volume load of aorticregurgitation more accurately in this unique patientpopulation.

METHODS

PATIENT POPULATION. Between September 2014and February 2015, 20 patients who previously un-derwent LVAD implantation with either a HeartMate II(Thoratec Corp., Pleasanton, California) or HeartWareHVAD (HeartWare International Inc., Framingham,Massachusetts) and who had varying degrees of AIwere prospectively enrolled at the University of Chi-cago Medical Center in Chicago, Illinois. The Institu-tional Review Board approved this study, and allpatients provided informed consent. Patients wereexcluded if they had poor echocardiographic windowsor known left-to-right or right-to-left shunting. Pa-tients were also excluded if they had known pumpmalfunction, inlet or outlet cannulamalpositioning, orclinical suspicion of thrombosis. Patients’ baselinecharacteristics were collected and stored.RIGHT-SIDED HEART CATHETERIZATION AND

ECHOCARDIOGRAPHIC IMAGING. All patients un-derwent simultaneous RHC and transthoracic echo-cardiography (TTE) in the catheterization laboratoryusing a 7-F Swan-Ganz Catheter (Edwards Life-sciences, Irvine, California). RHC was performedthrough the right internal jugular vein or right femoralvein. The following values were measured: centralvenous pressure, systolic pulmonary artery pressure,diastolic pulmonary artery pressure, mean pulmonaryartery pressure, PCWP, and pulmonary artery satura-tion. Cardiac output and cardiac index were calcu-lated by the indirect Fick equation with estimatedoxygen consumption of 125 ml/min/m2. Hemoglobinwas measured from venous blood gas, and arterialoxygen saturation was measured by pulse oximetry.In a subset of 5 patients, repeat hemodynamic valueswere measured during a ramp study. The hemoglobinand arterial oxygen saturation values obtained duringthe baseline RHC were used for all subsequent Fickcalculations during the ramp study.

Data were initially acquired at the patient’s pre-senting pump speed. From the parasternal window,linear measurements of left ventricular chambersize, aortic valve opening, and traditional AI severitywere measured using aortic regurgitation vena con-tracta width and qualitative estimation of AI severityaccording to guideline recommendations (11,12) Froma modified, right-sided parasternal view, the LVADoutflow cannula diameter and pulse-wave Dopplersignal were acquired as previously described (13).The diameter of the LVAD outflow cannula wasmeasured at the point of acquisition of the pulsedDoppler signal. The aortic valve opening wasassessed by M-mode echocardiography through theaortic valve from a parasternal short-axis view and

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6 Grinstein et al.J U N E 2 0 1 6 : 6 4 1 – 5 1 Aortic Insufficiency Quantification With LVAD

643

averaged over 10 cardiac cycles to determine thefraction of valve opening. Finally, for patients withat least intermittent antegrade flow across the aorticvalve, pulse-wave Doppler of the LVOT wasmeasured from the apical window (12). The aortic RFwas calculated by subtracting the venous return tothe right side of the heart obtained by RHC by theFick method from the total left-sided systemic flowmeasured using TTE (Figure 1). Total left-sided sys-temic flow was calculated by measuring the flowthrough the LVAD outflow cannula and flow acrossthe aortic valve in those patients with at leastintermittent aortic valve opening. Flow across theLVAD outflow cannula was calculated by multiplyingthe velocity time integral (VTI) across the entiretyof the cardiac cycle by the cross-sectional area (CSA)of the LVAD outflow cannula and heart rate (HR).Flow across the aortic valve was calculated bymultiplying the VTI during systole in the LVOT by

FIGURE 1 Derivation of Regurgitant Fraction

Derivation of regurgitant fraction by flow measurements using echocard

ventricular assist device (LVAD) outflow cannula, across the aortic valve

fraction of aortic valve opening; Ca ¼ arterial oxygen content; CO ¼ card

HR ¼ heart rate; LVAD ¼ left ventricular assist device; LVOT ¼ left ven

the CSA of the LVOT and HR, as well as the fractionof aortic valve opening according to the equation inFigure 1. The RFs measured with this novel methodwere compared with traditional indices includingvena contracta and qualitative assessment (12). Re-sults were clinically compared with left-sided fillingpressures obtained during RHC.NOVEL MEASUREMENTSOF AORTIC INSUFFICIENCY. Twonew, totally noninvasive indices for grading AIseverity in patients with LVADs are proposed:1) LVAD outflow cannula diastolic acceleration; and2) the outflow LVAD cannula systolic-to-diastolicpeak velocity ratio (S/D ratio) (Figures 2A to 2C). Themodified, right-sided parasternal window was used toacquire the pulse-wave Doppler signal of the outflowLVAD cannula. Care was taken to align the LVADoutflow cannula with the transducer at the point ofanastomosis with the ascending aorta. The LVADoutflow cannula diastolic acceleration was obtained

iography and right-sided heart catheterization (RHC) through the left

, and to the right side of the heart. AI ¼ aortic insufficiency; AoVOF ¼iac output; CSA ¼ cross-sectional area; Cv ¼ venous oxygen content;

tricular outflow tract; VTI ¼ velocity time integral.

FIGURE 2 Novel TTE Parameters: Diastolic Acceleration and S/D Ratio

Measurement of left ventricular assist device outflow cannula diastolic acceleration and left ventricular assist device outflow cannula systolic-

to-diastolic peak velocity ratio (S/D ratio) in a representative patient with (A) no aortic insufficiency (AI), (B) trace-mild AI, and (C) moderate-

severe AI. AI severity was graded using vena contracta. D ¼ peak diastolic velocity; S ¼ peak systolic velocity; TTE ¼ transthoracic

echocardiography.

Grinstein et al. J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6

Aortic Insufficiency Quantification With LVAD J U N E 2 0 1 6 : 6 4 1 – 5 1

644

by measuring the diastolic slope from the onset to theend of diastole (Figures 2A to 2C). The S/D ratiowas obtained by dividing the peak systolicvelocity by the end diastolic peak velocity of theLVAD outflow cannula (Figures 2A to 2C). Theperformance of these indices was compared withtraditional vena contracta width and aortic regurgi-tation fraction, measured as described earlier, as wellas invasively acquired left-sided filling pressuresmeasured by PCWP at the time of simultaneous heartcatheterization.

To assess the performance of these indices atdifferent AI severities induced by different LVADspeeds, parameters were acquired at different stagesof a ramp study in a subset of patients (n ¼ 5). We re-ported our ramp test protocol previously (14); in brief,patients had LVAD speeds increased by 400-rpm in-crements for the HeartMate II device. Left ventricularlinear diastolic dimensions, frequency of aorticvalve opening, severity of AI and mitral valvularinsufficiency, blood pressure, and LVAD parameterswere measured and recorded at each stage.

STATISTICAL METHODS. Data were collected usingExcel software (2007 Microsoft Corp., Redmond,Washington) and were analyzed using GraphPadPrism (GraphPad Software Inc., San Diego, Califor-nia). Continuous variables were evaluated for nor-mality by using the D’Agostino-Pearson test fornormal distribution. The Student t test was used todetermine differences in normally distributed data.To determine the relationships among the differentAI indices, as well as with clinical filling pressures,the Pearson coefficient of correlation was testedwith linear regression analysis for each scoringparameter. For RF, patients without AI wereexcluded from correlation analysis, and for the S/Dratio, patients without AI were excluded fromanalysis. To assess the reproducibility of the S/Dratio and diastolic slope measurements, all mea-surements were repeated by a second observer, andinterobserver variability was assessed by intraclasscorrelation measurement. Each patient was analyzedseparately and given equal weight throughout allanalyses.

TABLE 1 Baseline Characteristics

All Patients(n ¼ 20)

No AI on TTE(n ¼ 7)

Trace or MildAI on TTE(n ¼ 7)

Moderate orGreater AI on TTE

(n ¼ 6)

General characteristics

Age, yrs (mean) 58.2 54.3 55.9 65.5

Male, % 75 86 71 67

LVAD characteristics

Duration of LVAD, months 19.1 12.9 15.6 30.5

Destination, %

BTT 30 57 14 17

DT 70 43 86 83

Type of LVAD, %

HMII 75 86 71 83

HVAD 25 14 29 17

Average speed, rpm

HMII 9,253 9,240 9,200 9,320

HVAD 2,724 2,780 2,650 2,760

Origin of cardiomyopathy

Ischemic, % 40 29 57 33

Nonischemic, % 60 71 43 67

Medical history

Hypertension, % 75 71 86 67

Average Doppler bloodpressure, mm Hg

85.4 80.0 93.6 82.0

Hyperlipidemia, % 25 29 14 33

Atrial fibrillation, % 30 29 29 33

Diabetes mellitus, % 45 57 29 50

COPD, % 15 14 0 33

PAD, % 10 0 14 17

CVA, % 20 14 29 17

TTE measurements

Vena contracta, cm 0.19 0 0.20 0.40

LVEDD, cm 6.0 6.5 5.3 6.3

LVESD, cm 5.7 6.2 4.9 6.1

Aortic root (SOV), cm 3.1 3.3 2.9 3.2

Aortic valve opening, % 35 14 43 40

No opening, n 13 6 4 3

Intermittent, n 0 0 0 0

Regular, n 7 1 3 3

AI ¼ aortic insufficiency; BTT ¼ bridge to transplantation; COPD ¼ chronic obstructive pulmonary disease; CVA ¼cerebrovascular accident; DT ¼ destination therapy; HMII ¼ HeartMate II; HVAD ¼ HeartWare ventricular assistdevice; LVAD ¼ left ventricular assist device; LVEDD ¼ left ventricular end-diastolic diameter; LVESD ¼ leftventricular end-systolic diameter; PAD ¼ peripheral arterial disease; SOV ¼ sinus of Valsalva; TTE ¼ transthoracicechocardiography.

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6 Grinstein et al.J U N E 2 0 1 6 : 6 4 1 – 5 1 Aortic Insufficiency Quantification With LVAD

645

RESULTS

BASELINE CHARACTERISTICS. Simultaneous RHCand TTE were performed in 20 patients with variousdegrees of AI ranging from absent to moderately se-vere AI. Baseline characteristics of the cohort are re-ported in Table 1. Patients’ ages ranged from 45 to 76years (mean age 58 years), and 75% of these patientswere men. Of these patients, 75% had a HeartMate IIdevice, and 25% had a HeartWare HVAD. The LVADwas implanted as destination therapy in most of thepatients (70%).

PERFORMANCE OF REGURGITANT FRACTION. RFwas calculated in 15 of 20 (75%) patients. In the 5patients in whom RF could not be calculated, eitherthe LVAD outflow cannula could not be imaged (n ¼ 2)or the Doppler flow recorded was orthogonal to thetransducer and was therefore incapable of yielding anaccurate Doppler signal. Patients without AI had anRF that was nearly zero (2.4 � 4.6%). Patients withqualitatively estimated trace or mild AI had an RF of31.0 � 5.4%, a value that is well within the lowermoderate range for RF, according to American Societyof Echocardiography guidelines. The RF of patientswith qualitatively estimated moderate or greater AIwas higher (45.8 � 3.6%), falling within the highrange of moderate regurgitation fraction for AI(Figure 3A). RF correlated better with PCWP than didvena contracta (correlation coefficient [R] ¼ 0.73 vs.0.56) in all patients with AI (Figure 3B).

NOVEL TRANSTHORACIC ECHOCARDIOGRAPHY

PARAMETERS. The S/D ratio of the LVAD outflowcannula and LVAD outflow cannula diastolic acceler-ation strongly correlated with RF (R ¼ 0.91 and 0.94,respectively) (Figures 4A and 4B). Both these indiceswere successfully acquired in 15 of 15 (100%) of pa-tients in whom the LVAD cannula was well visualized.Both the S/D ratio and diastolic acceleration of theLVAD cannula correlated better with PCWP thandid the vena contracta (R ¼ 0.82 and 0.65 vs. 0.56)(Figures 4 C and D). The interobserver variability waslow for both the S/D ratio and the diastolic slope asrepresented by a high intraclass correlation coeffi-cient for both variables (r ¼ 0.91 and 0.94,respectively).

The mean S/D ratio and diastolic accelerationstratified by RF along with proposed thresholds formoderate and severe AI stratified at the 30% and 50%RF thresholds, respectively, are reported in Table 2.

In a subset of 5 patients, the S/D ratio and diastolicacceleration of the LVAD cannula was measured ateach stage of a ramp test (Figures 5A and 5B). Onepatient had no AI, 2 patients had mild AI, and 2

patients had moderate to severe AI. An average of6 data points was obtained (range 5 to 9) in eachpatient. Given the augmented pressure gradient be-tween the ascending aorta and the left ventricle atfaster LVAD speeds, it would be expected that thedegree of AI would increase at faster LVAD speeds.In keeping with this concept, in all patients the S/Dratio decreased and the diastolic slope of the LVADoutflow cannula increased at the augmented ramptest speeds, findings supporting the performanceof these TTE parameters across a variety of LVADsettings (Figures 5A and 5B).

FIGURE 3 Performance of RF Versus VC

50%

40%

30%

20%

10%

0%

*P < 0.0001

#P = 0.0004

None

4N =

Trace/Mild

RF vs. Visual EstimationA

*

Moderate/Severe

5

*#

6

R = 0.56

R = 0.73

0.6

0.5

0.4

0.3

0.2

0.1

0.0

RF and VC vs. PCWPB

252015105 030

Pulmonary Capillary Wedge Pressure

Regurgitant Fraction Vena Contracta

(A) Aortic insufficiency severity measured by mean regurgitant fraction (RF) compared with standard transthoracic echocardiography grading

using a combination of vena contracta (VC) and visual estimation. (B) Performance of RF compared with VC as a function of left-sided filling

pressure for patients with trace or greater aortic insufficiency. PCWP ¼ pulmonary capillary wedge pressure.

Grinstein et al. J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6

Aortic Insufficiency Quantification With LVAD J U N E 2 0 1 6 : 6 4 1 – 5 1

646

DISCUSSION

In the current study we evaluated the traditionalechocardiographic methods of assessing the severityof AI and compared them with a novel, semi-invasivemethod of acquiring RF in patients with LVADs.Furthermore, we described 2 new noninvasive indicesto assess pancyclic AI. The main findings of thisstudy are as follows: 1) traditional echocardiographicmethods used to assess AI in patients with LVADunderestimate the severity of AI compared with RF;2) vena contracta fails to correlate accurately withleft-sided filling pressures; and 3) the LVAD outflowcannula diastolic acceleration and the LVAD outflowcannula S/D ratio correlated well with RF andleft-sided filling pressures.

It is now well established that de novo AI is acommon complication after CF-LVAD implantation(1,2,15). The AI tends to be progressive and can leadto worsening heart failure and end-organ hypo-perfusion with adverse effects on long-term survival(3,16). Initial management often involves adjustingLVAD speeds. Reduction of the LVAD speed de-creases the transaortic pressure gradient and thusdecreases AI severity, which often occurs at theexpense of an elevation of left-sided filling pressuresand worsening end-organ perfusion (5). Alterna-tively, LVAD speeds can be augmented to partiallyovercome the regurgitant flow and improve end-organ perfusion while simultaneously reducing clin-ical congestion (1). However, the increased pump

speed further accelerates the destructive forcesresponsible for the AI. An invasive ramp study usingcombined real-time hemodynamics and echocardio-graphic structural assessment can help determinethe optimal LVAD speed setting to temporize thedisease (14,17). Ultimately, definitive managementwith valve repair, replacement, or closure or, alter-natively, urgent transplantation is required forsymptomatic AI. To date, there are no guidelines toassess the severity of AI in patients with LVADs,and instead, most centers continue to use AmericanSociety of Echocardiography guidelines that weredescribed for patients with normal cardiac physi-ology and pulsatility.

Echocardiography has been shown to be useful inevaluating LVAD pump function, ventricular func-tion, and hemodynamics in patients with CF-LVADs(14,18,19). Based on the continuity equation, totalleft-sided systemic flow must equal right-sidedcardiac flow unless a shunt or AI is present. Usingthis principle, investigators have proposed thatpump flow can be estimated by subtracting TTE-derived flow across the LVOT from flow across theright ventricular outflow tract (RVOT) (17,20). Simi-larly, Estep et al. (18) have suggested that a dropin flow across the RVOT as measured by TTE withpreservation of cannula inflow or outflow Dopplerprofiles may indicate significant AI (18). In thisstudy, we report for the first time a novel methodthat combines RHC and Doppler echocardiographyto quantify AI severity based on the continuity

FIGURE 4 Performance of S/D ratio and Diastolic Acceleration

R = 0.65

100

80

60

40

20

0

Diastolic Slope vs. PCWPD

252015105 030

PCWP (mm Hg)

Dia

sto

lic S

lop

e (c

m/s

2 )

R = 0.94

100

80

60

40

20

0

Diastolic Slope vs. RFB

50%40%30%20%10% %06%0

Regurgitant Fraction (%)

Dia

sto

lic S

lop

e (c

m/s

2 )

R = 0.91

S/D

Vel

oci

ty

7

6

5

4

3

2

1

0

S/D Velocity of Outflow Cannula vs. RFA

50%40%30%20%10% %06%0

Regurgitant Fraction (%)

Sys

tolic

to

Dia

sto

lic V

elo

city

R = 0.82

7

6

5

4

3

2

1

0

S/D Velocity Outflow vs. PCWPC

252015105 030

PCWP (mm Hg)

(A) Left ventricular assist device outflow cannula systolic-to-diastolic peak velocity ratio (S/D ratio) compared with regurgitant fraction (RF) for

patients with trace or greater aortic insufficiency. (B) Left ventricular assist device outflow cannula diastolic acceleration compared with RF. (C)

Performance of left ventricular assist device outflow cannula S/D ratio compared with left-sided filling pressures for patients with trace or

greater aortic insufficiency. (D) Performance of left ventricular assist device outflow cannula diastolic acceleration compared with left-sided

filling pressures. PCWP ¼ pulmonary capillary wedge pressure.

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6 Grinstein et al.J U N E 2 0 1 6 : 6 4 1 – 5 1 Aortic Insufficiency Quantification With LVAD

647

of flow principle. After excluding patients withknown shunts, we demonstrate that AI RF can bemeasured by subtracting right-sided cardiac outputmeasured by RHC from left-sided total systemiccardiac output (flow across the LVAD added to flowacross the aortic valve when present) measured byechocardiography.

Interestingly, when traditional AI indices werecompared with RF measured by the proposed

TABLE 2 Aortic Insufficiency Severity Stratified by Regurgitant Fracti

as Well as Proposed Thresholds for Moderate and Severe AI

No AI Trace/Mild Moder

N 4 2

RF, % 2.4 26.6

S/D NA 6.0

Diastolic acceleration, cm/s2 4.9 35.7 6

PCWP, mm Hg 10.2 10.5

AI ¼ aortic insufficiency; NA ¼ not applicable; RF ¼ regurgitant fraction; PCWP ¼ pulm

method, traditional indices appeared to underesti-mate AI severity, particularly among patients withless than moderate regurgitation. Similarly, RFcorrelated better with left-sided filling pressures thandid tradition indices. Because AI increases the vol-ume load of the left ventricle, left-sided filling pres-sures serve as a strong clinical correlate of AI severity.Accordingly, it appears that traditional indicesmay lead to underestimation of hemodynamically

on With Corresponding S/D Ratio, Diastolic Acceleration, and PCWP,

ate/SevereExtrapolated Mild toModerate Threshold

Extrapolated Moderate toSevere Threshold

9 NA NA

41.8 30.0 50.0

2.6 5.0 1.1

8.7 49.0 82.0

15.0 NA NA

onary capillary wedge pressure; S/D ratio ¼ systolic-to-diastolic peak velocity ratio.

FIGURE 5 Performance of Novel Parameters During a Ramp Study

7

6

5

4

3

2

1

0

LVAD Speed (RPM)

7000 8000 9000 10000 11000 12000

S/D Ratio During Ramp StudyA

S/D

Rat

io

No AI Trace/Mild Moderate/Severe

140

120

100

80

60

40

20

0

LVAD Speed (RPM)

6000 7000 8000 9000 10000 1200011000

Diastolic Acceleration During Ramp Study B

Dia

sto

lic S

lop

e cm

/sec

2

(A) Left ventricular assist device (LAVD) outflow cannula systolic-to-diastolic (S/D) peak velocity ratio and (B) LVAD outflow cannula diastolic

acceleration during various stages of an LVAD ramp test. Aortic insufficiency (AI) severity was graded using vena contracta.

Grinstein et al. J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6

Aortic Insufficiency Quantification With LVAD J U N E 2 0 1 6 : 6 4 1 – 5 1

648

significant AI and to diagnosis in advanced clinicalpresentations. Unfortunately, with more advanced AIand clinical disease, patients are deemed high-surgical risk candidates and therefore frequentlyundergo high-risk aortic valve surgical proceduresor, alternatively, undergo percutaneous aortic valveclosure with less favorable long-term outcomes (21).It is hoped that more timely recognition of moresevere AI may allow for disease modification or sur-gical intervention with improved outcomes. Howev-er, there are no published data to prove that earlyrepair of AI will change the course of the diseaseand result in better outcomes.

Over the course of our study, we gained valuableinsight and experience with imaging the LVADoutflow cannula. It became apparent during thisstudy that the LVAD outflow cannula area was morevariable than initially appreciated. Although theoutflow cannula has a predefined ex vivo diameter asdefined by the respective manufacturer (1.4 cm forthe HeartMate II and 1.0 cm for the HeartWareHVAD), in vivo several cannulas had considerablysmaller diameters at the point of Doppler signalacquisition related to the angle of anastomosis to theaorta, the degree of sewing at the anastomosis,kinking of the outflow cannula, or possible externalcompression of the pliable cannula. Outflow cannula

obstruction can impair the efficiency of the LVADand can predispose the patient to clinical signs ofheart failure. Additional imaging studies focusingon distortion of LVAD outflow cannula anatomy areneeded to define the scope of this problem moreclearly.

We also describe 2 novel, completely noninvasivetechniques to assess AI severity using Doppler echo-cardiography that correlate more favorably with RFand left-sided filling pressures than vena contractadoes. We demonstrated that LVAD outflow cannuladiastolic acceleration and LVAD outflow cannula S/Dratio can be acquired in the majority of patients withLVADs, although such indices may be difficult tomeasure in patients with poor acoustic windowsor tachycardia. These new indices strongly correlatewith RF and better correlate with PCWP than doesvena contracta. Flow through the LVAD outflow can-nula is directly proportional to preload in the leftventricle and is inversely proportional to afterload inthe ascending aorta. During systole, afterload in theascending aorta augments from enhanced flow eitherthrough the LVAD or across the aortic valve (in pa-tients who have some degree of aortic valve opening)(Figure 6A). During diastole, regurgitant flow intothe left ventricle across the aortic valve coupledwith forward downstream flow into the aortic arch

FIGURE 6 Physiology of Novel Parameters

Aortic insufficiency (AI)–associated changes in preload, afterload, and flow in (A) systole and (B) diastole. LVAD ¼ left ventricular assist device;

S/D ¼ systolic to diastolic peak velocity ratio.

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6 Grinstein et al.J U N E 2 0 1 6 : 6 4 1 – 5 1 Aortic Insufficiency Quantification With LVAD

649

and ascending aorta leads to progressive reductionin afterload in the ascending aorta (Figure 6B). Fur-thermore, the regurgitant flow across the aorticvalve adds to the left ventricular preload. Thus, withincreasing AI severity, an acceleration of diastolicoutflow cannula flow is expected. Similarly, the flowratio through the LVAD cannula from peak systole(when afterload is the highest and preload is thelowest) to end-diastole (when afterload is the lowestand preload the highest) is expected to decrease asAI severity worsens.

In this study, an excellent correlation betweenLVAD outflow cannula diastolic flow acceleration andRF is demonstrated, thus supporting the use of theseindices as sensitive markers of AI severity in patientswith LVADs. We also demonstrate a high inversecorrelation between peak S/D ratios and AI RF. Anelevated S/D ratio has also been previously shown topredict LVAD thrombosis in patients with intravas-cular hemolysis (22). Both the diastolic accelerationof the LVAD outflow cannula and the peak S/D ratioperformed well across a variety of LVAD settings asdemonstrated during ramp testing. AI severity would

be expected to increase at higher LVAD speeds giventhe enhanced gradient between the aorta and the leftventricle with increased LVAD speed. Accordingly,the S/D ratio decreased and the diastolic accelerationof the LVAD outflow cannula increased in all patientsduring ramp testing.

Our study demonstrates that traditional echocar-diographic methods underestimate the severity ofAI and do not correlate as highly with left-sidedfilling pressures compared with RF. As such, thetrue severity of aortic regurgitation as a contributorto worsening heart failure may be underappreciatedin these patients. We describe 2 additional novelnoninvasive methods that correlate well with RFand PCWP. These indices may further help usunderstand the physiology and progression of thedisease and allow us better to study the clinicalconsequences of AI in patients with LVADs. Giventhe lack of a true gold standard for measuring AI inpatents with LVADs, the enhanced correlation withleft-sided filling pressures that was observed withthe novel parameters suggests that these parame-ters may be more clinically relevant. Accordingly,

PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: The

new TTE parameters, diastolic acceleration of the

LVAD outflow cannula and S/D ratio of the LVAD

outflow cannula, allow for reliable and easily repro-

ducible evaluation of AI severity in patients with CF-

LVADs. These novel parameters better correlated with

intracardiac filling pressures and may detect clinically

meaningful regurgitant flow across the aortic valve

sooner than currently used TTE parameters.

TRANSLATIONAL OUTLOOK: The clinical conse-

quence of AI in patients with CF-LVADs remains un-

certain, and future studies are needed to evaluate the

prognostic performance of the S/D ratio and diastolic

acceleration of the LVAD outflow cannula. Further-

more, clinical studies evaluating the efficacy of aortic

valve interventions including valve closure and

replacement are needed to define more clearly when

such interventions are warranted.

Grinstein et al. J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6

Aortic Insufficiency Quantification With LVAD J U N E 2 0 1 6 : 6 4 1 – 5 1

650

the novel parameters may allow us better toevaluate the epidemiology of AI in this patientpopulation.

STUDY LIMITATIONS. This study was a single-center,prospective study with a small cohort size and thusprone to bias related to our own institution’s surgicaltechniques and TTE image acquisition. However, thephysiologic nature of the study together with thestrong correlations described overcome the size lim-itation. In our study, cardiac output was measured bythe indirect Fick method using calculated oxygenconsumption and indirect arterial oxygen saturationusing pulse oximetry. We opted to use a physiologiccalculation of right-sided flow instead of calculatingright-sided flow using the RVOT VTI. In a subset ofpatients, we were able to show a linear increase in AIseverity as measured both by RF and by S/D ratio anddiastolic acceleration of the LVAD cannula during aRAMP study. During the ramp, each patient acts as hisor her own control, thereby effectively limiting biasrelated to cannula anatomy, operator experience, andoxygen consumption assumptions. Furthermore, ourstudy analyzed the limitations of vena contracta andvisual estimation. We did not analyze alternativetraditional parameters including jet width/LVOTdiameter and proximal isovelocity surface area, andtherefore the accuracy of these methods is still un-clear. Additionally, the validity of any noninvasiveparameter, be it traditional or novel, is limited by thelack of a universally agreed on, quantifiable, directmeasurement of AI severity to serve as the goldstandard.

CONCLUSIONS

Traditional echocardiographic methods underesti-mate the severity of AI in LVAD patients compared

with RF and do not correlate as strongly with left-sided filling pressures. The LVAD outflow cannuladiastolic acceleration and the LVAD outflow cannulaS/D ratio had excellent correlations with RF and left-sided filling pressures and should be used to assessAI severity in these patients.

REPRINT REQUESTS AND CORRESPONDENCE: Dr.Nir Uriel, Transplant and Mechanical CirculatorySupport, University of Chicago Medical Center, Car-diovascular Division, 5841 South Maryland Avenue,MC 2016, Chicago, Illinois 60637. E-mail: nuriel@medicine.bsd.uchicago.edu.

RE F E RENCE S

1. Jorde UP, Uriel N, Nahumi N, et al. Preva-lence, significance, and management of aorticinsufficiency in continuous flow left ventricularassist device recipients. Circ Heart Fail 2014;7:310–9.

2. Pak SW, Uriel N, Takayama H, et al. Prevalenceof de novo aortic insufficiency during long-termsupport with left ventricular assist devices.J Heart Lung Transplant 2010;29:1172–6.

3. Cowger JA, Aaronson KD, Romano MA, Haft J,Pagani FD. Consequences of aortic insufficiencyduring long-term axial continuous-flow left ven-tricular assist device support. J Heart LungTransplant 2014;33:1233–40.

4. Mudd JO, Cuda JD, Halushka M, Soderlund KA,Conte JV, Russell SD. Fusion of aortic valve com-missures in patients supported by a continuous

axial flow left ventricular assist device. J HeartLung Transplant 2008;27:1269–74.

5. Holtz J, Teuteberg J. Management of aorticinsufficiency in the continuous flow left ventricularassist device population. Curr Heart Fail Rep 2014;11:103–10.

6. Mano A, Gorcsan J, Teuteberg JJ, et al. Inci-dence and impact of de novo aortic insufficiencyfollowing continuous flow LVADs implantation.J Heart Lung Transplant 2012;31:S22.

7. Atkins BZ, Hashmi ZA, Ganapathi AM, et al.Surgical correction of aortic valve insufficiencyafter left ventricular assist device implantation.J Thorac Cardiovasc Surg 2013;146:1247–52.

8. Martina J, de Jonge N, Sukkel E, Lahpor J. Leftventricular assist device–related systolic aorticregurgitation. Circulation 2011;124:487–8.

9. Kato TS, Maurer MS, Sera F, Homma S,Mancini D. Aortic regurgitation during systolic-phase accompanied by mitral regurgitation inpatients with continuous-flow left ventricularassist device. Eur Heart J Cardiovasc Imaging2013;14:1022.

10. Siontis KC, Nkomo VT, Pislaru C, Enriquez-Sarano M, Pellikka PA, Pislaru SV. Mechanismof aortic valve opening: beyond the pressuregradient. J Am Coll Cardiol Img 2014;7:633–4.

11. Lang RM, Bierig M, Devereux RB, et al.Recommendations for chamber quantification:a report from the American Society of Echo-cardiography’s Guidelines and Standards Com-mittee and the Chamber Quantification WritingGroup, developed in conjunction with theEuropean Association of Echocardiography, a

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . 9 , N O . 6 , 2 0 1 6 Grinstein et al.J U N E 2 0 1 6 : 6 4 1 – 5 1 Aortic Insufficiency Quantification With LVAD

651

branch of the European Society of Cardiology.J Am Soc Echocardiogr 2005;18:1440–63.

12. Zoghbi WA, Enriquez-Sarano M, Foster E,et al. Recommendations for evaluation ofthe severity of native valvular regurgitationwith two-dimensional and Doppler echocardi-ography. J Am Soc Echocardiogr 2003;16:777–802.

13. Horton SC, Khodaverdian R, Chatelain P, et al.Left ventricular assist device malfunction: anapproach to diagnosis by echocardiography. J AmColl Cardiol 2005;45:1435–40.

14. Uriel N, Morrison KA, Garan AR, et al. Devel-opment of a novel echocardiography ramp testfor speed optimization and diagnosis of devicethrombosis in continuous-flow left ventricularassist devices: the Columbia ramp study. J Am CollCardiol 2012;60:1764–75.

15. Aggarwal A, Pant R, Kumar S, et al. Incidenceand management of gastrointestinal bleeding with

continuous flow assist devices. Ann Thorac Surg2012;93:1534–40.

16. Toda K, Fujita T, Domae K, Shimahara Y,Kobayashi J, Nakatani T. Late aortic insufficiencyrelated to poor prognosis during left ventricularassist device support. Ann Thorac Surg 2011;92:929–34.

17. Myers TJ, Frazier OH, Mesina HS,Radovancevic B, Gregoric ID. Hemodynamics andpatient safety during pump-off studies of an axial-flow left ventricular assist device. J Heart LungTransplant 2006;25:379–83.

18. Estep JD, Stainback RF, Little SH, Torre G,Zoghbi WA. The role of echocardiography andother imaging modalities in patients with leftventricular assist devices. J Am Coll Cardiol Img2010;3:1049–64.

19. Estep JD, Vivo RP, Krim SR, et al. Echo-cardiographic evaluation of hemodynamics inpatients with systolic heart failure supported by a

continuous-flow LVAD. J Am Coll Cardiol 2014;64:1231–41.

20. Stainback RF, Croitoru M, Hernandez A,Myers TJ, Wadia Y, Frazier OH. Echocardiographicevaluation of the Jarvik 2000 axial-flow LVAD.Tex Heart Inst J 2005;32:263–70.

21. Retzer E, Tabit CE, Estrada JR, et al. Predictorsof survival in patients undergoing trans-catheteraortic valve closure for significant left ventricularassist device associated aortic insufficiency. J AmColl Cardiol 2014;64.

22. Fine NM, Topilsky Y, Oh JK, et al. Role ofechocardiography in patients with intravascularhemolysis due to suspected continuous-flow LVADthrombosis. J Am Coll Cardiol Img 2013;6:1129–40.

KEY WORDS aortic insufficiency,heart failure, LVAD