1-Year Survival After TAVR of Patients With Low-Flow, Low ... · OBJECTIVES This study sought to examine whether the prognosis of patients with severe aortic stenosis (AS) having

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 2 , N O . 8 , 2 0 1 9

ª 2 0 1 9 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

STRUCTURAL

1-Year Survival After TAVR of PatientsWith Low-Flow, Low-Gradient andHigh-Gradient Aortic Valve Stenosis inMatched Study Populations
Ulrich Fischer-Rasokat, MD, PHD,a,b Matthias Renker, MD,a,c,d Christoph Liebetrau, MD, PHD,a,b,d
Arnaud van Linden, MD,c,e Mani Arsalan, MD,c,e Maren Weferling, MD,a Andreas Rolf, MD, PHD,a,b,d

Mirko Doss, MD, PHD,c Helge Möllmann, MD, PHD,f Thomas Walther, MD, PHD,c,d,e Christian W. Hamm, MD, PHD,a,b,d

Won-Keun Kim, MDa,b,c

ABSTRACT

ISS

Fro

An

Na

of

an

Ke

OBJECTIVES This study sought to examine whether the prognosis of patients with severe aortic stenosis (AS) having

high versus low transvalvular mean pressure gradients (MPGs) is intrinsically different after transcatheter aortic valve

replacement (TAVR), even after strict matching of baseline parameters.

BACKGROUND Patients with low MPG are characterized by higher cardiovascular risk and more comorbidities than

other AS patients are.

METHODS In this retrospective, single-center study involving 2,282 patients, 3 groups were derived according to

the following criteria: 1) high-gradient AS (HG-AS) (MPG $40 mm Hg); 2) low-flow, low-gradient AS (LFLG-AS)

(MPG <40 mm Hg, ejection fraction [EF] #40%, stroke volume index #35 ml/m2); 3) paradoxical LFLG-AS (pLFLG-AS)

(similar to LFLG-AS but with EF$50%). Propensity score matching that included EF was used to compare 1-year survival.

RESULTS A total of 136 patients with HG-AS or LFLG-AS were identified. Kaplan-Meier survival curves were significantly

different (p ¼ 0.039), with death occurring in 11 versus 21 patients (hazard ratio: 2.12; 95% confidence interval: 1.02 to

4.39; p ¼ 0.044), respectively. A total of 226 patients with HG-AS or pLFLG-AS were identified and here the curves were

identical (p ¼ 0.468), with death occurring in 18 versus 21 patients (hazard ratio: 1.26; 95% confidence interval: 0.67 to

2.38; p ¼ 0.469).

CONCLUSIONS This is the first study comparing survival after TAVR of patients with high versus low MPG in

matched study populations. Mortality in patients with LFLG-AS was twice that of HG-AS patients. However, it appears

that patients with pLFLG-AS might benefit from TAVR to the same extent as patients with HG-AS. There must

be still unmasked factors that influence mortality of patients with LFLG-AS. (J Am Coll Cardiol Intv 2019;12:752–63)

© 2019 by the American College of Cardiology Foundation.

N 1936-8798/$36.00 https://doi.org/10.1016/j.jcin.2019.01.233

m the aDepartment of Cardiology, Kerckhoff Heart Center, Bad Nauheim, Germany; bMedical Clinic I (Cardiology and

giology), University Hospital of Giessen, Giessen, Germany; cDepartment of Cardiac Surgery, Kerckhoff Heart Center, Bad

uheim, Germany; dGerman Centre for Cardiovascular Research, Partner Site Rhein-Main, Bad Nauheim, Germany; eDepartment

Cardiac, Thoracic and Thoracic Vascular Surgery, University Hospital of the Goethe University, Frankfurt am Main, Germany;

d the fDepartment of Cardiology, Medical Clinic I, St. Johann Hospital, Dortmund, Germany. This work was supported by the

rckhoff Heart Research Institute together with the Justus Liebig University. Dr. Renker has received speaker fees from St. Jude

https://doi.org/10.1016/j.jcin.2019.01.233

http://crossmark.crossref.org/dialog/?doi=10.1016/j.jcin.2019.01.233&domain=pdf

AB BR E V I A T I O N S

AND ACRONYM S

AS = aortic stenosis

CI = confidence interval

EF = ejection fraction

HG-AS = high-gradient aortic

stenosis

HR = hazard ratio

IQR = interquartile range

LFLG-AS = low-flow, low-

gradient aortic stenosis

LV = left ventricular

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 2 , N O . 8 , 2 0 1 9 Fischer-Rasokat et al.A P R I L 2 2 , 2 0 1 9 : 7 5 2 – 6 3 Survival in Matched Low-Flow, Low-Gradient Patients After TAVR

753

P atients with severe aortic stenosis (AS) andimpaired systolic or diastolic left ventricular(LV) function, leading to low stroke volume

and low transvalvular gradients (low flow, lowgradient [LFLG]), represent a high-risk patient popu-lation (1,2). Transcatheter aortic valve replacement(TAVR) evolved as a treatment option for these pa-tients that has clearly demonstrated better outcomesthan conservative management (3,4). However, initialstudies report that midterm mortality of patients withLFLG-AS—even after successful TAVR—is still consid-erably higher than that of patients with high-gradientAS (HG-AS) (4–6).

SEE PAGE 764

MPG = transvalvular mean

pressure gradient

pLFLG-AS = paradoxical low-

flow, low gradient aortic

stenosis

SVI = stroke volume index

TAVR = transcatheter aortic

replacement

The consistently worse outcome of patients withLFLG-AS is usually associated with increasedbaseline cardiovascular risk (6,7). The largest reg-istry of aortic valve procedures, GARY (GermanAortic Valve Registry), reported significant differ-ences in the baseline characteristics between pa-tients with HG-AS and patients with lowtransvalvular mean pressure gradient (MPG) andreduced ejection fraction (EF) with respect to sex,cardiovascular risk factors, existing cardiovasculardisease, and EF, which was reflected by a signifi-cantly higher logistic EuroSCORE (European Systemfor Cardiac Operative Risk Evaluation) in the latterpatient population.

Given that most of the baseline characteristicsmentioned previously have been shown to bepredictors of worse outcome, we speculated thatsurvival of patient populations with similar baselinecharacteristics could be identical. Therefore, wematched patient populations and analyzed 1-yearsurvival. If survival rates were found to bedifferent between matched groups, other as-yet-unmasked factors would have to play a role in theoutcome of aortic valve patients with thesecharacteristics.

METHODS

STUDY DESIGN. This study was designed as a retro-spective analysis of data from an ongoing TAVRregistry at a single TAVR center. The study protocol

Medical/Abbott. Dr. Liebetrau has received speaker fees from Abbott. Dr. Dos

Möllmann has received proctor/speaker fees from Abbott, Biotronik, Edwar

and Symetis SA. Dr. Hamm has served on the advisory board for Medtron

Abbott, Symetis, St. Jude Medical, and Edwards Lifesciences. All other aut

relevant to the contents of this paper to disclose.

Manuscript received December 3, 2018; revised manuscript received Januar

was approved by the institutional reviewboard. From January 2011 until September2017, a total of 2,282 patients were treated byTAVR. The decision to perform TAVR wasmade by a heart team based onthe established criteria. Only patients withsevere, symptomatic native AS (aortic valvearea <0.6 cm2/m2 body surface area) andfollow-up information from at least a 30-dayperiod were included. Patients with accessroutes other than transfemoral or transapicalwere excluded. For further analysis, patientswere assigned to 1 of 3 subgroups according tothe flow chart shown in Figure 1: 1) HG-AS:MPG $40 mm Hg; 2) LFLG-AS: MPG <40mm Hg, stroke volume index (SVI) #35 ml/m2,EF #40%; and 3) paradoxical LFLG-AS(pLFLG-AS): MPG <40 mm Hg, SVI#35 ml/m2, EF $50%.

Clinical events were evaluated according
to updated Valve Academic Research Consortium-2criteria (8). Follow-up time was defined as the timefrom the procedure to the last documented contactwith the patient (alive) or to the time of documenteddeath. The primary endpoint was death from anycause. Patients with follow-up time longer than 1 yearwere censored as alive after 365 days.
ECHOCARDIOGRAPHIC MEASUREMENTS. All pa-tients underwent standard transthoracic echocardi-ography at rest in a left decubitus position byexperienced operators in accordance with establishedguidelines. EF was estimated visually. Aortic valvearea was calculated according to the continuityequation. Patients with severe LFLG-AS and calcu-lated SVI <35 ml/m2 were examined by trans-esophageal echocardiography to confirm aortic valvearea indexed <0.6 cm2/m2 by planimetry. Furtherechocardiographic exams were planned beforereferral and at the 30-day follow-up visit.

MULTIDETECTOR COMPUTED TOMOGRAPHY

MEASUREMENTS. Noncontrast computed tomogra-phy was performed using a multidetector computedtomography system (Somatom Force, SiemensHealthineers, Forchheim, Germany) in 131/136(HG-AS and LFLG-AS) and 220/226 (HG-AS and

valve

s has received proctor fees from Boston Scientific. Dr.

ds Lifesciences, St. Jude Medical, Boston Scientific,

ic. Dr. Kim has received proctor/speaker fees from

hors have reported that they have no relationships

y 9, 2019, accepted January 22, 2019.

FIGURE 1 Flow Chart Illustrating the 3 Groups Derived From the Entire Patient Population

AVA ¼ aortic valve area; BSA ¼ body surface area; MPG ¼ transvalvular mean pressure gradient; HG-AS ¼ high-gradient aortic stenosis; LFLG-

AS ¼ low-flow, low-gradient aortic stenosis; LVEF ¼ left ventricular ejection fraction; pLFLG-AS ¼ paradoxical low-flow, low-gradient aortic

stenosis; SV ¼ stroke volume; TA ¼ transapical; TF ¼ transfemoral; TAVR ¼ transcatheter aortic valve replacement.

Fischer-Rasokat et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 2 , N O . 8 , 2 0 1 9

Survival in Matched Low-Flow, Low-Gradient Patients After TAVR A P R I L 2 2 , 2 0 1 9 : 7 5 2 – 6 3

754

pLFLG-AS) patients. Aortic valve calcification (AVC)measurements were performed offline on dedicatedworkstations using validated software (syngo.via,Siemens Healthineers, Forchheim, Germany) by theAgatston method and are expressed in arbitraryunits (AU). AVC density (AVCd) was calculated asAVC indexed to the cross-sectional area of the aorticannulus, which was calculated from the diameter ofthe LV outflow tract measured by echocardiographyaccording to Clavel et al. (9). Anatomically severe ASaccording to the sex-specific cutoff values weredefined for AVC scores $1,200 AU in womenand $2,000 AU in men (10) and for AVCdvalues $300 AU/cm2 in women and $500 AU/cm2 inmen (9).

STATISTICS. Data are expressed as median (inter-quartile range [IQR]). Categorical variables are pre-sented as numbers and percentages. Continuousvalues were compared by the Mann-Whitney U orKruskal-Wallis test. The comparison of categoricalvariables was performed by the chi-square test.

Propensity score matching was performed to identifypatient populations with similar baseline character-istics. The following parameters were included: age,sex, body mass index, renal function, EF, the pres-ence of diabetes mellitus, atrial fibrillation, orchronic obstructive lung disease and the accessroute. Kaplan-Meier survival curves were comparedusing the log-rank test. Hazard ratios were deter-mined by Cox regression analysis and were correctedfor variables that revealed significant differencesbetween groups despite group matching. All statis-tical analyses were performed using the SPSS sta-tistical package version 24 (IBM Corporation,Armonk, New York).

RESULTS

In the entire study population, we identified 1,052patients with HG-AS, 166 patients with LFLG-AS, and244 patients with pLFLG-AS (Online Table 1).Compared with patients with HG-AS, patients with

https://doi.org/10.1016/j.jcin.2019.01.233

FIGURE 2 Survival Curves Based on All-Cause Mortality for Patients in the Overall (Unmatched) Study Population

Kaplan-Meier survival estimates for all-cause mortality. Abbreviations as in Figure 1.


755

LFLG-AS were characterized by a lower percentageof women (25% vs. 60%), worse renal function, and asignificantly higher number of cardiovascular riskfactors and manifest cardiovascular disease (e.g.,3-vessel coronary artery disease [52% vs. 21%], priormyocardial infarction [34% vs. 12%], prior CABG[37% vs. 13%], or atrial fibrillation [59% vs. 33%]),resulting in significantly higher Society of ThoracicSurgeons scores of 6.1 (IQR: 4.1 to 9.5) versus 4.7(IQR: 3.2 to 6.9) (p < 0.001 for all comparisons). EFand SVI were substantially lower in LFLG-AS pa-tients. These differences were associated with ahigher percentage of patients of New York HeartAssociation functional class IV in LFLG-AS patients.Patients with pLFLG-AS had values for baselinecharacteristics that were generally between those ofpatients with HG-AS and patients with LFLG-AS andhad an intermediate Society of Thoracic Surgeonsscore of 5.0 (IQR: 3.5 to 7.6). There were significantdifferences between groups with respect to the

percentage of transfemoral access (69% vs. 69% vs.61%; p ¼ 0.038 for patients with HG-AS, LFLG-AS, orpLFLG-AS, respectively) as well as with respect tothe use of balloon-expandable valves (44% vs. 56%vs. 44%; p < 0.001). Median follow-up time was 363(IQR: 105 to 365) days versus 330 (IQR: 79 to 365)days versus 350 (IQR: 91 to 365) days (p ¼ 0.064) forHG-AS, LFLG-AS, and pLFLG-AS patients,respectively.

The differences in baseline risks between groupstranslated to significantly (p < 0.001) differentKaplan-Meier survival curves (Figure 2), with HG-AS patients having the best prognosis and pa-tients with LFLG-AS having the worst prognosis.Mortality rates were 12.8%, 29.5%, and 20.1% forHG-AS, LFLG-AS, and pLFLG-AS patients,respectively.

We then performed propensity score matching tocreate groups with identical baseline characteristics.A total of 136 patients with HG-AS or LFLG-AS were

TABLE 1 Clinical Data: HG-AS Versus LFLG-AS

HG-AS(n ¼ 68)

LFLG-AS(n ¼ 68) p Value

Demographic data

Female 21 (31) 20 (29) 0.852

Age, yrs 81 (77–85) 80 (76–83) 0.407

BMI, kg/m2 26.0(23.7–30.6)

26.6(23.9–31.1)

0.689

GFR, ml/min/1.73 m2 59 (49–74) 60 (39–73) 0.512

Dialysis 3 (4.4) 4 (5.9) 0.698

NYHA functional classIII/IV

48 (71)/14 (21) 44 (65)/15 (22) 0.503

Risk factors

Arterial hypertension 66 (97) 64 (94) 0.404

Diabetes mellitus 25 (37) 29 (43) 0.483

Dyslipidemia 29 (43) 27 (40) 0.727

COPD 13 (19) 11 (16) 0.653

STS score 6.6 (4.1–9.4) 5.9 (3.7–10.3) 0.970

Cardiovascular disease

CAD I 13 (19) 7 (10) 0.146

CAD II 13 (19) 10 (15) 0.493

CAD III 22 (32) 31 (46) 0.114

CABG 14 (21) 23 (34) 0.083

Prior myocardialinfarction

15 (22) 14 (21) 0.834

Atrial fibrillation 35 (52) 37 (54) 0.731

Prior stroke 7 (10) 7 (10) NS

Peripheral arterial disease 16 (24) 18 (27) 0.692

Values are n (%) or median (interquartile range).

BMI ¼ body mass index; CABG ¼ coronary artery bypass graft; CAD ¼ coronaryartery disease; COPD ¼ chronic obstructive pulmonary disease; CT ¼ computedtomography; GFR ¼ glomerular filtration rate; HG-AS ¼ high-gradient aorticstenosis; LFLG-AS ¼ low-flow, low-gradient aortic stenosis; NYHA ¼ New YorkHeart Association; STS ¼ Society of Thoracic Surgeons.



756

identified (Table 1). The adequacy of the matchingof the baseline characteristics of these 2 groups wasconfirmed by their similar Society of Thoracic Sur-geons scores (p ¼ 0.970). Whereas median EFwas <35% in both groups (p ¼ 0.582), SVI wassignificantly higher in patients with HG-AS than inpatients with LFLG-AS (29 [IQR: 22 to 37] ml/m2 vs.26 [IQR: 21 to 29] ml/m2; p ¼ 0.001) (Table 2). AVCand AVCd were higher in patients with HG-AS thanin patients with LFLG-AS. In the patients withLFLG-AS, the criterion for severe AVC was met by72% and 78% of the male and female patients,respectively, and the criterion for severe AVCd by76% and 100%, respectively. There was a slightlylower rate of transfemoral access used amongpatients with HG-AS (57% vs. 68%; p ¼ 0.215); the useof balloon-expandable valves (46% vs. 47%; p¼ 0.863)and residual aortic regurgitation post-implantation(8% vs. 11%; p ¼ 0.638) was the same. The

median follow-up time for HG-AS was 365 (IQR: 132to 365) days versus 323 (IQR: 78 to 365) days forLFLG-AS (p ¼ 0.082). Kaplan-Meier survival curves(Figure 3) for these 2 groups were significantlydifferent (p ¼ 0.039), with death occurring (within1 year) in 11 (HG-AS) versus 21 (LFLG-AS) patients(hazard ratio [HR]: 2.12; 95% confidence interval[CI]: 1.02 to 4.39; p ¼ 0.044). When all patientswith HG-AS but only patients with LFLG-AS andpositive criteria for severe anatomical AS accordingto AVC criteria were included, the HR was 1.76(95% CI: 0.78 to 3.99; p ¼ 0.176). Inclusion ofLFLG-AS patients according to positive AVCdcriteria resulted in a HR of 2.15 (95% CI: 1.00 to4.65; p ¼ 0.050).

A total of 226 patients with HG-AS or pLFLG-ASwere identified through propensity score matching(Table 3). SVI was significantly higher in patients withHG-AS than in those with pLFLG-AS (p < 0.001).Again, AVC and AVCd were higher in patientswith HG-AS than in those with pLFLG-AS. AmongpLFLG-AS patients, 68% and 74% of the male andfemale patients, respectively, met the criterion forsevere AVC, and 83% and 86% met the criterion forsevere AVCd (Table 4). The percentage of patientstreated by transfemoral access (69% vs. 67%;p ¼ 0.775) and balloon-expandable valves (47% vs.47%; p ¼ NS) was the same in the 2 AS groups, andthere were no significant differences in residualaortic regurgitation (9% vs. 9%; p ¼ 0.521). Themedian follow-up time was 365 (IQR: 162 to 365)days versus 352 (IQR: 85 to 365) days (p ¼ 0.152).Kaplan-Meier survival curves (Figure 4) were iden-tical (p ¼ 0.468), with 18 versus 21 patients dyingwithin 1 year (HR corrected for EF: 1.26; 95% CI:0.67 to 2.38; p ¼ 0.469). Inclusion of only pLFLG-ASpatients with severe AVC yielded a HR of 0.93 (95%CI: 0.44 to 1.98; p ¼ 0.858) and inclusion of onlythose with severe AVCd gave a HR of 1.22 (95% CI:0.62 to 2.40; p ¼ 0.575).

Clinical outcome data are presented in Table 5.Although the rate of major strokes or severe kidneyinjury during 30-day follow-up tended to be higherin patients with low-gradient AS compared withtheir matched HG-AS patients, the overall eventrates according to the Valve Academic ResearchConsortium-2 criteria were the same betweengroups. EF was the same or improved betweenbaseline and 30-day follow-up in 79% of survivingLFLG-AS patients and in 69% of pLFLG-AS patients.However, no association was detected betweenchanges in EF and any further clinical events. One

TABLE 2 Doppler Echocardiographic and MDCT Data: HG-AS Versus LFLG-AS

HG-AS LFLG-AS p Value

Echocardiographic data

Ejection fraction, % 31 (25–38) 34 (30–40) 0.582

LV mass index, g/m2 146 (126–184) 143 (118–169) 0.135

SVI, ml/m2 29 (22–37) 26 (21–29) 0.001

MPG, mm Hg 46 (42–51) 24 (18–30) <0.001

AVAi, cm2/m2 0.29 (0.22–0.34) 0.35 (0.31–0.41) <0.001

MDCT data

AVC

Men, AU 3,537 (2,687–4,691) 2,576 (1,880–3,061) <0.001

Meeting criteria for severe AS 96 72

Women, AU 2,751 (2,033–3,728) 1,874 (1,159–2,140) 0.003


AVCd

Men, AU/cm2 989 (784–1,281) 702 (499–852) <0.001


Women, AU/cm2 876 (694–1,139) 573 (407–822) 0.005


Values are median (interquartile range) or %.

AU ¼ arbitrary units; AVAi ¼ aortic valve area indexed; AVC ¼ aortic valve calcification; AVCd ¼ aortic valvecalcification density; LV ¼ left ventricular; MDCT ¼ multidetector computed tomography; MPG ¼ transvalvularmean pressure gradient; SVI ¼ stroke volume index; other abbreviations as in Table 1.


757

year after TAVR the rate of major strokes still ten-ded to be higher in patients with low-gradient ASwithout reaching significance. There was also nodifference with respect to documented episodes ofcardiac decompensation within 1 year after TAVRbetween patients with HG-AS and LFLG-AS orpLFLG-AS. The type of valve (balloon-expandablevs. non–balloon expandable) and the final effectiveorifice area after TAVR had no influence on mor-tality in either of our matched study populations.The number of patients with significant residualaortic regurgitation was too small to allow for anymeaningful comparisons.

Finally, we compared SVI in survivors and non-survivors to determine whether SVI is an importantdeterminant of outcome in patients after TAVR(Figure 5). Interestingly, SVI was not different be-tween survivors and nonsurvivors, neither in HG-ASpatients nor in patients with low-gradient AS. Infact, SVI even tended to be higher in HG-AS non-survivors than in HG-AS survivors. Thus, althoughSVI—by definition—is lower in patients with LFLG-ASor pLFLG-AS compared with HG-AS patients, it did

FIGURE 3 Survival Curves Based on All-Cause Mortality for Patients in the Matched Study Populations HG-AS and LFLG-AS


TABLE 3 Clinical Data: HG-AS Versus pLFLG-AS

HG-AS(n ¼ 113)

pLFLG-AS(n ¼ 113) p Value

Demographic data

Female 66 (58) 68 (60) 0.787

Age, yrs 83 (79–86) 82 (79–86) 0.748

BMI, kg/m2 27.1(24.5–31.6)

26.8(23.5–30.4)

0.291

GFR, ml/min/1.73 m2 64 (43–83) 64 (44–83) 0.723

Dialysis 2 (1.8) 1 (0.9) 0.561

NYHA functional classIII/IV

89 (79)/9 (8) 88 (78)/11 (10) 0.524

Risk factors

Arterial hypertension 110 (97) 109 (97) 0.701

Diabetes mellitus 36 (32) 32 (28) 0.562

Dyslipidemia 55 (49) 36 (32) 0.010

COPD 26 (23) 21 (19) 0.413

STS score 5.1 (3.2–7.3) 4.5 (3.2–6.1) 0.165

Cardiovascular disease

CAD I 23 (20) 13 (12) 0.069

CAD II 20 (18) 18 (16) 0.722

CAD III 23 (20) 34 (30) 0.092

CABG 17 (15) 22 (20) 0.379

Prior myocardial infarction 12 (11) 19 (17) 0.176

Atrial fibrillation 43 (38) 47 (42) 0.587

Prior stroke 18 (16) 17 (15) 0.854

Peripheral arterial disease 20 (18) 24 (21) 0.502

Values are n (%) or median (interquartile range).

Abbreviations as in Table 1.

TABLE 4 Doppler Echocardiographic and MDCT Data: HG-AS Versus pLFLG-AS

HG-AS pLFLG-AS p Value

Echocardiographic data

Ejection fraction, % 62 (60–65) 60 (55–65) 0.012

LV mass index, g/m2 134 (113–164) 124 (99–147) 0.007

SVI, ml/m2 36 (30–44) 30 (26–33) <0.001

MPG, mm Hg 53 (46–64) 29 (24–35) <0.001

AVAi, cm2/m2 0.31 (0.26–0.37) 0.35 (0.31–0.41) <0.001

MDCT data

AVC

Men, AU 3,599 (2,821–5,213) 2,655 (1,834–3,574) 0.001


Women, AU 2,955 (2,232–4,263) 1,568 (1,107–2,189) <0.001


AVCd

Men, AU/cm2 1,065 (836–1,527) 754 (547–1,180) 0.002


Women, AU/cm2 1,038 (740–1,414) 556 (399–753) <0.001


Values are median (interquartile range) or %.

pLFLG-AS ¼ paradoxical low-flow, low-gradient aortic stenosis; other abbreviations as in Tables 1 and 2.



758

not discriminate between survivors and nonsurvivorsin any population.

DISCUSSION

Patients with AS and poor systolic or diastolic LVfunction constitute a challenging patient popula-tion. TAVR emerged as a suitable treatment optionfor such patients with high perioperative risk.However, despite the existence of several studieson this topic, it is difficult to clearly establish thebenefit of TAVR for patients with weak LV function.This is due to the plethora of definitions that areused to identify “the patient at high risk,” whichleads to a confusing spectrum of information onoutcome. For instance, conflicting data were re-ported for various groups of high-risk patients whomet the following pre-interventional criteria: low EFalone (4), low MPG alone (6,11), low EF and lowMPG (7), low SVI and preserved EF (12), or a com-bination of EF, MPG, and SVI (13). Finally, studieshave been published that are comparable to ourstudy, including patients with LFLG-AS according tothe current European guidelines for the manage-ment of valvular heart disease (10), but the resultsof these studies are conflicting. In contrast toKataoka et al. (5) and Reinthaler et al. (14), whoreport a worse outcome for patients with pLFLG-ASversus those with HG-AS, Debry et al. (15) describesimilar survival rates of patients with pLFLG-AS andHG-AS. On the other hand, Bavishi et al. (16)observed a similar outcome for patients withLFLG-AS and HG-AS. However, all of the authorsmentioned are in agreement that low SVI may be apredictor of outcome, whereas EF assessment alonemay not be adequate to determine the prognosis ofpatients.

What all of these studies have in common is ahigh degree of variation in demographic parame-ters or parameters of body constitution, concomi-tant (mainly cardiovascular) risk factors, ormorbidities in patients with depressed LV func-tion, which may also be responsible for thediffuse database of current studies. Indeed, suchcharacteristics have not only been demonstratedto have an independent impact on survival inpatients after TAVR, for example, sex (17,18), bodymass index (19), and peripheral vascular disease(20), but may additionally trigger the choice ofthe interventional access site, which, in turn, isassociated with survival rates (21). Usually, thesedifferences in baseline parameters are addressedby statistical maneuvers to eliminate or relativizethe influence of single covariates. Our strategy

FIGURE 4 Survival Curves Based on All-Cause Mortality for Patients in the Matched Study Populations HG-AS and pLFLG-AS



759

was different. We aimed to identify patient pop-ulations with “classical” HG-AS and LFLG-AS thatideally would differ only with respect to the MPGand SVI. Indeed, after matching, patient pop-ulations had comparable baseline characteristicsincluding EF that were mirrored by a similarpredicted 30-day mortality.

The results of our matched cohorts are inter-esting. On the one hand, we observed that 1-yearmortality in patients with LFLG-AS was twice ashigh as that of HG-AS patients, which is in linewith previous study results. On the other hand, itappears that patients with pLFLG-AS benefit fromTAVR to the same extent as do those with HG-AS(Central Illustration). This makes an importantcontribution to the ongoing discussion about thiscontroversial topic, because numerous studiesreport a worse outcome of patients with pLFLG-AS (5,14,22). However, none of these studies

matched the baseline risks that display evidentdifferences between patients with HG-AS andthose with low-flow hemodynamics. In summary,our results suggest that the presence of a lowflow or a low gradient does not include anintrinsic risk per se in a selected population ofpatients with preserved EF (pLFLG-AS) andwhen baseline risks are adjusted.

A low SVI can explain the discrepancy betweenan aortic valve effective orifice area <1 cm2 deter-mined by the continuity equation in the presence of aMPG <40 mm Hg, and the definition of flow status ismandatory in the stepwise approach to diagnosis ofLFLG-AS in the most recent guidelines (10). Exami-nation of the baseline characteristics of previousstudies suggests that it is only the low SVI in patientswith LFLG-AS that drives mortality in these studypopulations. However, in our study, SVI was notdifferent between survivors and nonsurvivors.

TABLE 5 Clinical Outcomes

HG-AS(n ¼ 68)

LFLG-AS(n ¼ 68)

pValue

HG-AS(n ¼ 113)

pLFLG-AS(n ¼ 113)

pValue

30-day clinical outcomes

Overall mortality 4 (5.9) 9 (13.2) 0.145 8 (7.1) 8 (7.1) NS

In-hospital mortality 3 (4.4) 8 (11.8) 0.116 8 (7.1) 6 (5.3) 0.581

Cardiovascular mortality 4 (5.9) 8 (11.8) 0.281 7 (6.2) 5 (4.4) 0.531

Major stroke 1 (1.5) 3 (4.4) 0.310 3 (2.7) 5 (4.4) 0.472

Major vascular complication 4 (5.9) 4 (5.9) NS 6 (5.3) 9 (8.0) 0.423

New pacemaker implant 11 (16.2) 12 (17.6) 0.819 24 (21.2) 13 (11.5) 0.048

Acute kidney injury

Stage 1 7 (10.3) 3 (4.4) 11 (9.7) 4 (3.5)

Stage 2 1 (1.5) 3 (4.4) 0.352 7 (6.2) 3 (2.7) 0.068

Stage 3 2 (2.9) 4 (5.9) 2 (1.8) 6 (5.3)

Any event according to VARC-2criteria

26 (38.2) 27 (39.7) 0.860 51 (45.1) 42 (37.2) 0.224

EF unchanged or improved* 40 (100) 26 (78.8) 0.002 41 (65.1) 44 (68.8) 0.660

1-yr clinical outcomes

Overall mortality 11 (16.2) 21 (30.9) 0.043 18 (15.9) 21 (18.6) 0.597

Cardiovascular mortality 9 (13.2) 16 (23.5) 0.122 14 (12.4) 16 (14.2) 0.864

Major stroke 1 (1.5) 3 (4.4) 0.310 3 (2.7) 8 (7.1) 0.122

New pacemaker implant 12 (17.6) 18 (26.5) 0.215 25 (22.1) 14 (12.4) 0.053

Decompensation after 30-dayfollow-up

3 (4.4) 5 (7.4) 0.466 2 (1.8) 5 (4.4) 0.249

Values are n (%). *In survivors with echocardiographic exam 30 days after transcatheter aortic valvereplacement.

EF ¼ ejection fraction; VARC-2 ¼ Valve Academic Research Consortium-2; other abbreviations as in Tables 1and 4.



760

Furthermore, mortality was identical in patients withHG-AS and pLFLG-AS, irrespective of the lower SVI inpatients with pLFLG-AS. Finally, in the overall, un-matched study population, prognosis was differentbetween patients with LFLG-AS and pLFLG-ASdespite the fact that SVI was roughly the same.Thus, although a low SVI is mandatory for the defi-nition of patients with low-flow AS and is thereforeassociated with the increased mortality of this patientpopulation overall (as in other studies mentionedpreviously), our data do not support the idea that SVIhas a distinct and specific influence on individualprognosis in our study population. In line with ourconclusion are the results from the recently publishedTOPAS (Multicenter Prospective Study of Low-FlowLow-Gradient Aortic Stenosis) trial (23) in which theauthors demonstrated that the presence or absence ofa contractile reserve in patients with low SVI beforeTAVR has no effect on clinical outcomes or changes inLVEF during follow-up.

Chetcuti et al. (24) very recently published theresults from the prospective, multicenter CoreValveExpanded Use Study that included more than 2,400

patients. Interestingly, patients with low-gradientAS but normal EF had a baseline risk that wassimilar to that of HG-AS patients, which makes theirresults comparable to ours with matched studypopulations. In fact, in accordance with our results,patients with low-gradient AS and normal EFshowed the same rate of all-cause mortality ormajor stroke as HG-AS patients. The similarities toour study results, however, must be consideredwith care, because only 56% of their patients had anSVI <35 ml/m2 and therefore met the definition ofpLFLG-AS. They further analyzed the role of SVI inthe merged study cohort. Although a very lowSVI (<30 ml/m2) was associated with a slightlyhigher 1-year mortality, SVI had no impact on30-day mortality and did not emerge as anindependent predictor of 1-year survival, whichcorroborates the results of our study that did notdemonstrate a causal role of SVI in survival of TAVRpatients.

Despite using a rigorous diagnostic approachbased mainly on echocardiography, we did notexclude the presence of pseudo-severe AS in all ofour patients by systematically performing stressechocardiography or determination of AVC. At ourcenter, this might be due to the fact that patientswere included before the awareness of LFLG-AS asan important group of patients increased, which ledto new recommendations and cutoff values thatshould now be applied systematically. However, ourmultidetector computed tomography data on AVCdemonstrate that, indeed, the majority of our pa-tients meet the criteria for severe anatomical AS.Furthermore, outcome data were not different whenonly patients with severe AVC or AVCd wereincluded.

STUDY LIMITATIONS. Our study was a single-center,retrospective study in which matching of patientsaccording to baseline parameters led to rather smallstudy populations. All of this may have affectedthe conclusions drawn from our study. Echocardio-graphic measurements were made by differentoperators without a centralized core lab, and EF wasestimated visually. Thus, one cannot exclude thatthe echocardiographic measurements might havebeen biased. In addition, a more detailed character-ization of LV morphology and function (volumes,strain) and ventriculoarterial interplay may havecontributed to a more complete description of LVmechanics. Moreover, SVI was not available in all

CENTRAL ILLUSTRATION Survival Curves of Matched Study Populations

Fischer-Rasokat, U. et al. J Am Coll Cardiol Intv. 2019;12(8):752–63.

Kaplan-Meier survival estimates for all-cause mortality of matched patients with (A) high-gradient aortic stenosis (HG-AS) and low-flow, low-gradient aortic stenosis

(LFLG-AS) and (B) HG-AS and paradoxical low-flow, low-gradient aortic stenosis (pLFLG-AS).

FIGURE 5 Stroke Volume Index of Survivors and Nonsurvivors

Boxplot graphs of the SV index (SVI) from survivors (blue) or nonsurvivors (green) 1 year after TAVR in the HG-AS versus LFLG-AS group (left) or in the HG-AS versus

pLFLGAS group (right). The horizontal line in the box represents the median, the lengths of the box indicate the 25th to 75th interquartile ranges, and the whiskers

indicate minimal and maximal values. Abbreviations as in Figure 1.


761

PERSPECTIVES



762

patients. Finally, only 1-year survival was analyzed;differences in mortality may have changed duringlonger follow-up.

WHAT IS KNOWN? Patients with severe AS but low

transvalvular gradients and low EF (LFLG-AS) or,

paradoxically, normal EF (pLFLG-AS) are character-

ized by a high burden of cardiovascular and noncar-

diovascular disease and a substantially worse overall

outcome than are other patients with AS (HG-AS),

even after TAVR. There is an ongoing discussion

regarding whether these patients have an intrinsically

poor prognosis (classifying all of these patients as

high-risk patients) or whether survival is determined

by already-known parameters of cardiac function and

concomitant disease.

WHAT IS NEW? We identified similar populations of

patients with HG-AS and LFLG-AS or pLFLG-AS by

propensity score matching, thereby eliminating the

influence of prognostic baseline parameters (e.g., sex,

EF, atrial fibrillation, renal function). Notably we

found that patients with LFLG-AS still had a 2-fold

higher 1-year mortality rate than did patients with

HG-AS, although they had a comparable baseline risk.

On the other hand, survival in patients with pLFLG-AS

was identical to that of HG-AS patients.

WHAT IS NEXT? Special care should be given to

patients with LFLG-AS—but not necessarily to patients

with pLFLG-AS—treated by TAVR, as these patients

may inherit an increased risk beyond that engendered

by conventional risk factors. It is of utmost interest to

unmask further, more sensitive, and yet unknown

factors that exert a strong influence on prognosis in

LFLG-AS patients.

CONCLUSIONS

To our knowledge, our study is the first to try toeliminate the intrinsic baseline differences betweenpatients with LFLG-AS and pLFLG-AS by propensityscore matching to compare the prognosis of thesestudy populations. Under these conditions, patientswith LFLG-AS had a 2-fold higher 1-year mortalityrate than patients with HG-AS, whereas survival inpatients with pLFLG-AS was identical to that ofHG-AS patients. SVI was lower in patients withLFLG-AS or pLFLG-AS but failed to discriminatebetween survivors and nonsurvivors in our studypopulations. Our results suggest that patients withpLFLG-AS benefit from TAVR to the same extent aspatients with HG-AS and highlight the need to un-mask further, more sensitive, and yet unknownfactors that exert a strong influence on prognosis inLFLG-AS patients.

ACKNOWLEDGMENT The authors thank ElizabethMartinson, PhD, of the Kerckhoff Heart ResearchInstitute Editorial Office for excellent editorialassistance.

ADDRESS FOR CORRESPONDENCE: Dr. UlrichFischer-Rasokat, Department of Cardiology,Kerckhoff Heart Center, Benekestraße 2-8, 60231 BadNauheim, Germany. E-mail: [email protected].

RE F E RENCE S

1. Monin JL, Quere JP, Monchi M, et al. Low-gradient aortic stenosis: operative risk stratifica-tion and predictors for long-term outcome: amulticenter study using dobutamine stress he-modynamics. Circulation 2003;108:319–24.

2. Clavel MA, Magne J, Pibarot P. Low-gradientaortic stenosis. Eur Heart J 2016;37:2645–57.

3. Aguiar Rosa S, Moura Branco L, Galrinho A, et al.Contribution of dobutamine stress echocardiog-raphy to the diagnosis and prognosis of low-flow/low-gradient aortic stenosis. J Heart Valve Dis2016;25:130–8.

4. Luo X, Zhao Z, Chai H, et al. Efficacy of trans-catheter aortic valve implantation in patients withaortic stenosis and reduced LVEF. A systematicreview. Herz 2015;40 Suppl 2:168–80.

5. Kataoka A, Watanabe Y, Kozuma K, et al.Prognostic impact of low-flow severe aortic ste-nosis in small-body patients undergoing TAVR: the

OCEAN-TAVI registry. J Am Coll Cardiol Img 2018;11:659–69.

6. Conrotto F, D’Ascenzo F, D’Amico M, et al.Outcomes of patients with low-pressure aorticgradient undergoing transcatheter aortic valveimplantation: a meta-analysis. Catheter Car-diovasc Interv 2017;89:1100–6.

7. Lauten A, Figulla HR, Mollmann H, et al. TAVIfor low-flow, low-gradient severe aortic stenosiswith preserved or reduced ejection fraction: asubgroup analysis from the German Aortic ValveRegistry (GARY). EuroIntervention 2014;10:850–9.

8. Kappetein AP, Head SJ, Genereux P, et al.Updated standardized endpoint definitions fortranscatheter aortic valve implantation: the ValveAcademic Research Consortium-2 consensusdocument (VARC-2). Eur J Cardiothorac Surg2012;42:S45–60.

9. Clavel MA, Messika-Zeitoun D, Pibarot P, et al.The complex nature of discordant severe calcifiedaortic valve disease grading: new insights fromcombined Doppler echocardiographic andcomputed tomographic study. J Am Coll Cardiol2013;62:2329–38.

10. Baumgartner H, Falk V, Bax JJ, et al. 2017ESC/EACTS Guidelines for the management ofvalvular heart disease. Eur Heart J 2017;38:2739–91.

11. Amabile N, Agostini H, Gilard M, et al. Impactof low preprocedural transvalvular gradient oncardiovascular mortality following TAVI: an anal-ysis from the FRANCE 2 registry. EuroIntervention2014;10:842–9.

12. Eleid MF, Michelena HI, Nkomo VT, et al.Causes of death and predictors of survival afteraortic valve replacement in low flow vs. normalflow severe aortic stenosis with preserved ejection

mailto:[email protected]

mailto:[email protected]

http://refhub.elsevier.com/S1936-8798(19)30398-X/sref1


























































763

fraction. Eur Heart J Cardiovasc Imaging 2015;16:1270–5.

13. Puls M, Korte KP, Bleckmann A, et al. Long-term outcomes after TAVI in patients withdifferent types of aortic stenosis: the conundrumof low flow, low gradient and low ejection frac-tion. EuroIntervention 2017;13:286–93.

14. Reinthaler M, Schwabe A, Landmesser U, et al.’Paradoxical’ low-flow, low-gradient severe aorticvalve stenosis: an entity with limited improvementfollowing transcatheter aortic valve implantation.J Heart Valve Dis 2014;23:441–9.

15. Debry N, Sudre A, Amr G, et al. Transcatheteraortic valve implantation for paradoxical low-flowlow-gradient aortic stenosis patients. CatheterCardiovasc Interv 2016;87:797–804.

16. Bavishi C, Balasundaram K, Argulian E.Integration of flow-gradient patterns intoclinical decision making for patients with sus-pected severe aortic stenosis and preservedLVEF: a systematic review of evidence andmeta-analysis. J Am Coll Cardiol Img 2016;9:1255–63.

17. Siontis GC, Praz F, Pilgrim T, et al. Trans-catheter aortic valve implantation vs. surgicalaortic valve replacement for treatment of severeaortic stenosis: a meta-analysis of randomizedtrials. Eur Heart J 2016;37:3503–12.

18. O’Connor SA, Morice MC, Gilard M, et al.Revisiting sex equality with transcatheter aorticvalve replacement outcomes: a collaborative,patient-level meta-analysis of 11,310 patients.J Am Coll Cardiol 2015;66:221–8.

19. Arsalan M, Filardo G, Kim WK, et al. Prog-nostic value of body mass index and body sur-face area on clinical outcomes aftertranscatheter aortic valve implantation. Clin ResCardiol 2016;105:1042–8.

20. Van Mieghem NM, Tchetche D, Chieffo A,et al. Incidence, predictors, and implications ofaccess site complications with transfemoraltranscatheter aortic valve implantation. Am JCardiol 2012;110:1361–7.

21. Chandrasekhar J, Hibbert B, Ruel M, Lam BK,Labinaz M, Glover C. Transfemoral vs non-transfemoral access for transcatheter aortic valve

implantation: a systematic review and meta-anal-ysis. Can J Cardiol 2015;31:1427–38.

22. Mangner N, Stachel G, Woitek F, et al. Pre-dictors of mortality and symptomatic outcome ofpatients with low-flow severe aortic stenosis un-dergoing transcatheter aortic valve replacement.J Am Heart Assoc 2018;7:e007977.

23. Ribeiro HB, Lerakis S, Gilard M, et al. Trans-catheter aortic valve replacement in patients withlow-flow, low-gradient aortic stenosis: the TOPAS-TAVI registry. J Am Coll Cardiol 2018;71:1297–308.

24. Chetcuti SJ, Deeb GM, Popma JJ, et al. Self-expanding transcatheter aortic valve replacementin patients with low-gradient aortic stenosis. J AmColl Cardiol Img 2019;12:67–80.

KEY WORDS low-flow, low-gradient aorticstenosis, propensity score matching,transcatheter aortic valve replacement,1-year survival

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






























































Documents

1-Year Survival After TAVR of Patients With Low-Flow, Low ... · OBJECTIVES This study sought to examine whether the prognosis of patients with severe aortic stenosis (AS) having