Downloaded from
`
`http://eurheartj.oxfordjournals.org/
`
` by guest on August 3, 2016
`
`
`
` European Heart Journal Advance Access published October 1, 2014
`
`European Heart Journal
`doi:10.1093/eurheartj/ehu384
`
`CLINICAL RESEARCH
`Valvular heart disease
`
`Paravalvular Regurgitation after Transcatheter
`Aortic Valve Replacement with the Edwards
`Sapien Valve in the PARTNER trial: characterizing
`patients and impact on outcomes
`Susheel Kodali1*, Philippe Pibarot2, Pamela S. Douglas3, Mathew Williams1, Ke Xu4,
`Vinod Thourani5, Charanjit S. Rihal6, Alan Zajarias7, Darshan Doshi1,
`Michael Davidson8, E. Murat Tuzcu9, William Stewart9, Neil J. Weissman10,
`Lars Svensson9, Kevin Greason6, Hersh Maniar7, Michael Mack11, Saif Anwaruddin12,
`Martin B. Leon1, and Rebecca T. Hahn1
`
`1Herbert and Sandi Feinberg Interventional Cardiology, Heart Valve Center at Columbia University Medical Center/New York-Presbyterian Hospital, 177 Fort Washington Avenue,
`New York, NY 10032, USA; 2Laval University, Quebec, QC, USA; 3Duke Clinical Research Institute, Durham, NC, USA; 4Cardiovascular Research Foundation, New York, NY, USA;
`5Emory University School of Medicine, Atlanta, GA, USA; 6Mayo Clinic, Rochester, MN, USA; 7Washington University School of Medicine, Saint Louis, MO, USA; 8Brigham and Women’s
`Hospital, Boston, MA, USA; 9Cleveland Clinic, Cleveland, OH, USA; 10Medstar Health Research Institute, Washington, DC, USA; 11Baylor Healthcare System, Plano, TX, USA; and
`12Hospital of the University of Pennsylvania, Philadelphia, PA, USA
`
`Received 13 March 2014; revised 30 July 2014; accepted 21 August 2014
`
`Aim
`
`The impact of paravalvular regurgitation (PVR) following transcatheter aortic valve replacement (TAVR) remains
`uncertain. In this analysis, we sought to evaluate the impact of varying degrees of PVR on both mortality and changes
`in ventricular geometry and function.
`.... ..... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... .... ..... ..... .....
`Methods and
`Clinical and echocardiographic outcomes of patients who underwent TAVR from the randomized cohorts and continued
`results
`access registries in the PARTNER trial were analysed after stratifying by severity of post-implant PVR, which was graded as
`none/trace in 52.9% (n ¼ 1288), mild in 38.0% (n ¼ 925), and moderate/severe in 9.1% (n ¼ 221). There were significant
`differences in baseline clinical and echocardiographic characteristics. After TAVR, all the patients demonstrated increase
`in left ventricular (LV) function and reduction in the LV mass index, although the magnitude of mass regression was lower
`in the moderate/severe PVR group. The 30-day mortality (3.1 vs. 3.4 vs. 4.5%, P ¼ 0.56) and stroke (3.4 vs. 3.7 vs. 2.3%,
`P ¼ 0.59) were similar in all groups (none/trace, mild, and moderate/severe). At 1 year, there was increased all-cause mor-
`tality (15.9 vs. 22.2 vs. 35.1%, P , 0.0001), cardiac mortality (6.1 vs. 7.4% vs. 16.3%, P , 0.0001) and re-hospitalization
`(14.4 vs. 23.0 vs. 31.3%, P , 0.0001) with worsening PVR. A multivariable analysis indicated that the presence of moder-
`ate/severe PVR (HR: 2.18, 95% CI: 1.57–3.02, P , 0.0001) or mild PVR (HR: 1.37, 95% CI: 1.14–1.90, P ¼ 0.012) was
`associated with higher late mortality.
`.... ..... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... ..... .... ..... ..... ..... .... ..... ..... ..... .... ..... ..... .....
`Conclusion
`Differences in baseline characteristics in patients with increasing severities of PVR may increase the risk of this compli-
`cation. Despite these differences, multivariable analysis demonstrated that both mild and moderate/severe PVR predicted
`higher 1-year mortality.
`- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
`Transcatheter aortic valve replacement † Aortic stenosis † Paravalvular regurgitation
`Keywords
`
`* Corresponding author. Tel: +1 2123420444, Fax: +1 2123423660; Email: sk2427@columbia.edu
`Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2014. For permissions please email: journals.permissions@oup.com.
`
`Page 1 of 9
`
`

`

`Downloaded from
`
`http://eurheartj.oxfordjournals.org/
`
` by guest on August 3, 2016
`
`Page 2 of 9
`
`S. Kodali et al.
`
`Introduction
`Over the past decade, transcatheter aortic valve replacement
`(TAVR) has rapidly emerged as an alternative to surgical aortic
`valve replacement (SAVR) in high-risk patients and the treatment
`of choice for inoperable patients with severe, symptomatic aortic
`stenosis (AS).1 – 6 In recent years, there has been explosive growth
`in the clinical adoption of TAVR worldwide. With this increasing
`role, research efforts have focused on understanding and reducing
`procedural complications, such as paravalvular regurgitation (PVR).
`Numerous studies have shown significant rates of moderate-to-
`severe PVR following TAVR ranging from 0 to 24%.7 – 19 In addition,
`the presence of moderate or severe PVR has been associated
`with higher 1-year mortality.20 – 22 Several recent studies have also
`suggested that mild PVR may also be an important predictor of mor-
`tality.7,10,12,22 Mortality in this patient population is complex and
`whether PVR is causative or is simply associated with it remains to
`be seen. Characterizing clinical and echocardiographic differences
`between patients with varying degrees of PVR, as well as determining
`the effect of PVR on remodelling, could further increase our under-
`standing of the relationship between PVR and mortality. In this study,
`we present an in-depth analysis of patients from the PARTNER trial
`evaluating PVR and its impact on clinical and echocardiographic
`outcomes.
`
`Methods
`The PARTNER trial was a multicentre, randomized, clinical trial compar-
`ing TAVR with SAVR for high-risk patients (cohort A)2 and TAVR with
`medical therapy for inoperable patients (cohort B).1 Following comple-
`tion of the randomized trial and prior to commercial approval of the
`Edwards SAPIEN valve, additional patients were treated in a randomized
`continued access registry (RCA), as well as in a non-randomized contin-
`ued access (NRCA) registry, with the same inclusion and exclusion cri-
`teria as the randomized trial. All the patients had severe native trileaflet
`AS documented on a screening transthoracic echocardiogram (TTE)
`within 30 days of enrolment and were evaluated by two surgeons for
`the assessment of risk with SAVR. Important exclusion criteria included
`bicuspid disease, ejection fraction ,20%, renal failure, severe mitral
`regurgitation (MR), severe aortic regurgitation (AR), recent gastrointes-
`tinal bleeding, or recent neurological event. Complete inclusion and
`exclusion criteria have been presented in the supplementary appendix
`to a previous publication.2
`All the patients undergoing TAVR received either a 23 or 26 mm
`balloon-expandable Edwards SAPIEN transcatheter heart
`valve
`(Edwards Lifesciences, Irvine, CA, USA) from either the transfemoral
`(TF) or transapical (TA) approach based on vascular access. Annular
`assessments to determine valve size required were site determined util-
`izing TTE, transoesophageal echo (TEE), or multi-slice CT scans (MSCT).
`All the patients underwent TTE prior to discharge and at clinical follow-
`up time-points including 1 month, 6 months, and 1 year. All echocardio-
`grams were analysed at an independent core lab with methodology
`described previously.23 Important clinical events (including death,
`stroke, and re-hospitalization) were adjudicated by an independent clin-
`ical events committee (CEC).
`This analysis utilized an as-treated population of patients with either
`discharge or 30-day echoes evaluable for PVR severity. Paravalvular
`regurgitation was graded as none/trace, mild, moderate, or severe utiliz-
`ing semi-quantitative criteria previously described.19 Briefly, PVR after
`
`TAVR/SAVR was graded in accordance with the ASE recommendations
`for native valves24 with one exception. Because of the often eccentric,
`irregular jet and the frequent non-cylindrical ‘spray’ of the paravalvular
`jet contour, the parasternal short-axis view(s) was weighted more
`heavily than other signals in providing an integrated assessment, as
`follows: None, no regurgitant colour flow; Trace, pinpoint jet in AV; Mild,
`jet arc length is ,10% of the annulus circumference; Moderate, jet arc
`length is 10–30% of the annulus circumference; Severe, jet arc length is
`.30% of the annulus circumference. The cover index is defined as:
`100 × [(THV diameter 2 TEE annulus diameter)/THV diameter].8
`
`Statistical methods
`Patients were stratified by severity of PVR to evaluate impact on clinical
`outcomes. Multivariable analysis was performed to evaluate impact of
`PVR on 1-year mortality. Only paired data for patients with discharge
`or 30-day and 1-year echoes were evaluated for changes in the following:
`indexed effective orifice area (iEOA), LV diastolic volume (LVDV), LV
`systolic volume (LVSV), LV ejection fraction (EF), LV mass index.
`Categorical variables were compared using the x2 test. Since regurgi-
`tation is an ordinal variable, most comparisons involving this variable use
`the exact Jonckheere–Terpstra test. However, in survival models, PVR
`was used as a categorical variable. Continuous variables were presented
`as means (+ SD) and compared using Student’s t-test; comparisons with
`baseline values used the paired sample t-test. For multiple comparisons,
`Bonferroni correction was employed. The impact of PVR severity on
`mortality was evaluated using a Cox proportional hazards model, and
`all Cox models were tested to assure that the proportional hazards
`assumption was met. Log-rank tests were performed to compare survival
`distributions. P , 0.05 was used to declare statistical significance, unless
`multiple comparison adjustments were used. All statistical tests were
`two-sided.
`Stepwise multivariable analysis was performed for 1-year mortality using
`the baseline variables that differed between PVR groups (P ≤ 0.10) as well
`as baseline variables that were predictors of 1-year mortality on univariate
`analysis. Age, gender, PVR severity, and mode of access were forced into
`the model. Variables were entered with entry/stay criteria of 0.1/0.1 in a
`forward stepwise fashion.
`Data are based on an extract date of 18 February 2014. All statistical
`analyses were performed in SASw, version 9.2.
`
`Results
`Patient population and baseline
`characteristics
`A total of 2515 patients underwent TAVR with valve implantation as
`part of the randomized trial (n ¼ 496), RCA registry (n ¼ 40), or the
`NRCA registry (n ¼ 1979). Eighty-one patients were excluded from
`this analysis due to missing echocardiograms at discharge and at 30
`days, which resulted in a total population of 2434 patients. In this
`population, PVR was graded as none/trace in 52.9% (n ¼ 1288),
`mild in 38.0% (n ¼ 925), and moderate/severe in 9.1% (n ¼ 221).
`Baseline clinical characteristics stratified by severity of post-
`implant PVR are shown in Table 1. There were differences in
`several important baseline characteristics. Patients with none/trace
`PVR were more often female and had a smaller body surface area
`but higher BMI. The logistic EuroSCORE was higher in the moder-
`ate/severe PVR group, but there was no difference in the STS score
`between PVR groups. Patients with moderate/severe PVR were
`
`Page 2 of 9
`
`

`

`Downloaded from
`
`http://eurheartj.oxfordjournals.org/
`
` by guest on August 3, 2016
`
`Paravalvular regurgitation following TAVR
`
`Page 3 of 9
`
`Table 1 Baseline clinical parameters of patients by severity of paravalvular regurgitation
`
`P-value (all groups)
`
`Baseline parameters
`
`Severity of paravalvular regurgitation
`.. .. .. ... .. ... .. ... .. .. ... .. ... .. ... .. ... .. .. ... .. ... .. ... .. ... .. .. ... .. ... .. ... .. .. ... .. ...
`None/trace (n 5 1288) Mild (n 5 925) Moderate/severe (n 5 221)
`. ............ ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ .........
`84.17 + 7.05
`84.71 + 7.22
`85.04 + 7.52
`Age
`0.10
`Male, %
`44.3
`59.8
`70.6
`,0.0001
`1.78 + 0.24
`1.83 + 0.25
`1.82 + 0.24
`Body surface area
`,0.0001
`27.17 + 6.45
`26.53 + 6.09
`25.12 + 5.52
`Body mass index
`,0.0001
`25.74 + 16.06
`26.42 + 16.24
`29.79 + 17.05
`Logistic EuroSCORE
`0.004
`11.56 + 4.32
`11.31 + 3.85
`11.10 + 3.50
`STS score
`0.17
`Diabetes, %
`37.6
`37.8
`30.9
`0.14
`Carotid disease, %
`27.1
`26.4
`17.1
`,0.01
`Prior coronary artery bypass grafting, %
`43.3
`41.5
`44.5
`0.60
`Prior balloon aortic Valvuloplasty, %
`24.1
`23.0
`19.6
`0.34
`Renal disease (Cr ≥ 2), %
`14.8
`18.3
`18.6
`0.059
`Major arrhythmia, %
`46.9
`54.7
`60.0
`,0.0001
`Permanent pacemaker, %
`20.1
`22.4
`26.4
`0.08
`Smoking, %
`47.8
`50.9
`44.5
`0.15
`Chronic obstructive pulmonary disease, %
`43.9
`45.6
`40.3
`0.33
`Pulmonary hypertension, %
`37.6
`38.6
`48.0
`0.02
`
`Table 2 Baseline echocardiographic characteristics of patients by severity of paravalvular regurgitation
`
`Baseline parameters
`
`P-value (all groups)a
`
`Severity of paravalvular regurgitation
`....... ........... ........... ........... ........... ........... ............ ........... ........... ........... ..
`(a) None/trace (n 5 1288)
`(b) Mild (n 5 925)
`(c) Moderate/severe (n 5 221)
`.. ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ .........
`4.41 + 0.74
`4.60 + 0.77
`4.68 + 0.74
`LVEDD (cm)
`,0.0001
`3.20 + 0.92
`3.35 + 0.94
`3.51 + 0.92
`LVESD (cm)
`,0.0001
`64.2 + 19.6
`68.5 + 21.4
`67.6 + 25.0
`Stroke volume (cc)
`0.01
`4.38 + 1.41
`4.62 + 1.54
`4.57 + 1.59
`Cardiac output
`0.08
`53.7 + 12.4
`51.4 + 13.2
`50.2 + 13.9
`LV EF (%)
`,0.0001
`238.7 + 74.1
`260.3 + 78.3
`267.2 + 73.6
`LV mass (g)
`,0.0001
`1.98 + 0.18
`2.04 + 0.18
`2.06 + 0.19
`LVOT diameter (cm)
`,0.0001
`21.27 + 1.86
`21.64 + 1.83
`21.91 + 1.88
`Annulus diameter (cm)
`,0.001
`EOA (cm2)
`0.65 + 0.19
`0.66 + 0.19
`0.65 + 0.19
`0.25
`.. ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ .........
`Aortic regurgitation
`0.02
`34.2%
`42.8%
`44.7%
`None/trace
`0.36
`41.2%
`46.8%
`46.5%
`Mild
`Moderate/severe
`8.6%
`10.3%
`24.4%
`,0.0001
`.. ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ .........
`Mitral regurgitation
`None/trace
`Mild
`Moderate/severe
`
`29.9%
`50.7%
`19.5%
`
`25.8%
`51.7%
`22.5%
`
`17.8%
`46.1%
`36.1%
`
`0.001
`0.37
`,0.0001
`
`aComparisons performed using an exact Jonckheere–Terpstra test.
`
`less likely to have significant carotid artery disease, but more likely to
`have pulmonary hypertension.
`Baseline echocardiographic parameters also differed between the
`three groups (Table 2). Compared with patients with none/trace,
`patients with mild or moderate/severe PVR had larger LV
`
`end-diastolic dimensions (LVEDD), LV end-systolic dimensions
`(LVESD), and LV mass but lower EF at baseline. Patients with moder-
`ate/severe PVR post-implant were more likely to have moderate/
`severe baseline AR (P , 0.001) and MR (P , 0.01). Interestingly,
`the systolic LV outflow tract (LVOT) and aortic annular diameters,
`
`Page 3 of 9
`
`

`

`Downloaded from
`
`http://eurheartj.oxfordjournals.org/
`
` by guest on August 3, 2016
`
`Page 4 of 9
`
`S. Kodali et al.
`
`Table 3 Procedural characteristics
`
`P-value (all groups)a
`Characteristic
`None/trace (n 5 1288)
`Mild (n 5 925)
`Moderate/severe (n 5 221)
`.... ............ ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ ......
`Approach (%)
`,0.0001
`75.1
`67.1
`48.8
`Transfemoral
`Transapical
`51.2
`32.9
`24.9
`,0.0001
`.... ............ ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ ......
`Valve size
`0.01
`45.9
`49.5
`54.7
`23 mm
`26 mm
`45.3
`50.5
`54.1
`0.01
`.... ............ ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ ......
`12.74% + 5.40
`11.73% + 5.57
`10.82% + 5.73
`Cover index
`,0.0001
`
`aComparisons performed using an exact Jonckheere–Terpstra test.
`
`as well as diastolic measurements of the aorta were progressively
`larger with increasing PVR severity.
`Patients undergoing valve implantation via the TA approach com-
`pared with the TF approach had significantly less PVR (P , 0.0001).
`Patients with moderate-to-severe PVR were more likely to have
`received a 26 mm valve and had a lower cover index then those
`with less PVR (Table 3).
`
`Clinical outcomes
`In patients with none/trace, mild or moderate/severe PVR, the 30-day
`or in-hospital mortality (3.1 vs. 3.4 vs. 4.5%, P ¼ 0.56), and stroke (3.4
`vs. 3.7 vs. 2.3%, P ¼ 0.59) were similar in all groups. At 1-year, there
`was increased all-cause mortality (15.9 vs. 22.2 vs. 35.1%,
`P , 0.0001), cardiac mortality (6.1 vs. 7.4 vs. 16.3%, P , 0.0001),
`and re-hospitalization (14.4 vs. 23.0 vs. 31.3%, P , 0.0001) with
`higher grades of PVR (Figure 1). Patients with moderate/severe PVR
`showed less improvement in NYHA class at 6months compared
`with those with none/trace or mild PVR (Figure 2).
`Multivariable analysis (Table 4) evaluating the impact of PVR on
`1-year all-cause mortality was performed using the following covari-
`ates: age, sex, BMI, STS score, diabetes, smoking history, prior CABG,
`prior BAV,
`frailty, renal disease, major arrhythmia, pacemaker,
`chronic obstructive pulmonary disease, anaemia, 6 min walk distance,
`LV ejection fraction, LV mass, LVED, LVES, AV annulus diameter, and
`AV mean gradient. In addition, the following variables were forced
`into the model: PVR, TF vs. TA, and baseline moderate/severe AR.
`The presence of moderate/severe PVR (HR: 2.18, 95% CI: 1.69–
`3.35, P , 0.0001) or mild PVR (HR: 1.37, 95% CI: 1.14–1.90,
`P ¼ 0.012) were each associated with higher 1-year mortality.
`
`Changes in ventricular size and function
`Table 5 shows changes in ventricular size and function between base-
`line and 1-year using paired data in patients stratified by post-implant
`PVR. As expected, following valve implantation, iEOA significantly
`increased from baseline to 1-year, with no significant between
`group differences. Compared with baseline, there was a significant
`decrease in LVES and LVED in none/trace and mild PVR groups
`with no change in LVES and an increase in LVED in the moderate/
`severe PVR group. Left ventricular ejection fraction significantly
`increased over time in all groups. The left ventricular mass index
`
`improved significantly in all group over time, however, compared
`with moderate/severe PVR (28.4 + 26.6 g/m2) greater reductions
`were seen in none/trace (219.6 + 32.8 g/m2, P ¼ 0.001) and mild
`PVR groups (217.3 + 33.0 g/m2, P ¼ 0.01).
`
`Discussion
`This report from the PARTNER trial represents the largest published
`single study to evaluate the impact of PVR following TAVR with the
`Edwards Sapien valve on clinical and echocardiographic outcomes.
`It is particularly notable for utilizing echocardiography core labora-
`tory data as well as CEC adjudication for important endpoints. The
`principle findings of this study are the following: (i) there were signifi-
`cant differences in the baseline clinical and echocardiographic charac-
`teristics of patients with none/trace, mild, or moderate/severe PVR;
`(ii) patients with moderate/severe PVR demonstrated increases in
`LVED and less reduction in the LV mass index when compared
`with patients with less PVR; (iii) the presence of greater severity
`PVR was associated with reduced improvement in NYHA class and
`higher rates of re-hospitalization; (iv) On multivariable analysis, the
`presence of either mild (HR: 1.37) or moderate/severe (HR: 2.18)
`PVR resulted in significantly higher 1-year mortality.
`There were important differences in the baseline clinical and echo-
`cardiographic characteristics between the three groups. Whether
`these differences are responsible for the severity of PVR remains
`uncertain. Unfortunately, patients in the PARTNER trial did not
`have routine 3D imaging of the aortic annulus that would have
`allowed a more complete analysis evaluating predictors of PVR
`such as LVOT calcification and annular area. Nevertheless, less over-
`sizing (i.e. lower cover index) as assessed by a 2D measurement of the
`annulus was associated with in more PVR. Interestingly, patients
`undergoing TAVR via the TA approach had less severe PVR. The
`reasons for this are uncertain and may be related to procedural differ-
`ences as well as important differences in baseline characteristics
`between the two groups. Despite less PVR, the TA approach resulted
`in higher 1-year mortality in the multivariable analysis.
`Several prior reports have suggested that PVR could negatively
`impact mid- and long-term prognosis following TAVR.25 – 28 A recent
`meta-analysis by Athappan et al. with 1620 patients demonstrated
`increased 1-year mortality in patients with either moderate/severe
`
`Page 4 of 9
`
`

`

`Downloaded from
`
`http://eurheartj.oxfordjournals.org/
`
` by guest on August 3, 2016
`
`Paravalvular regurgitation following TAVR
`
`Page 5 of 9
`
`Figure 1 Differences in all-cause mortality (A), cardiac mortality (B) and repeat hospitalization (C) in patients following transcatheter aortic valve
`replacement stratified by severity of post-implant paravalvular regurgitation: none/trace (group A), mild (group B), and moderate/severe (group C).
`
`(HR: 2.27) or mild PVR (HR: 1.83). However, on sensitivity analysis, the
`clinical impact of mild PVR was less certain. Our study confirms that
`moderate/severe PVR results in higher 1-year mortality with a multi-
`variate hazard ratio (HR: 2.18) similar to that seen in the meta-analysis.
`In addition, our analysis demonstrates that mild PVR also results in sig-
`nificantly higher 1-year mortality. There are several key differences
`between the current study and the prior ones. First, in contrast to
`ours, the studies used in the meta-analysis consisted of both
`Edwards Sapien and Medtronic CoreValve implants. Also the assess-
`ment of PVR in prior studies was variable and included both angio-
`graphic and echocardiographic assessment with only one study29
`relying on a core laboratory for grading. Finally, our study with 2434
`patients represents one of the largest experiences published to date.
`The impact of mild PVR on mortality remains controversial. As
`noted above, prior studies have not demonstrated a clear association.
`In the recently presented FRANCE2 registry30 in which site-graded
`PVR was analysed, patients with grade 1 or mild PVR did not have
`increased 1-year mortality when compared with none/trace. One
`potential explanation for this difference is variability in the assess-
`ment of PVR severity. It is often challenging to characterize those
`patients with mild-to-moderate PVR. In this grey zone, one individual
`may downgrade the PVR to mild while another may call it moderate.
`
`These differences may result in different patient populations between
`various studies. As an example, in the FRANCE 2 registry, 13% of
`patients receiving a balloon-expandable valve were graded as
`having PVR grade 2 (moderate) or greater based on site assessed pre-
`discharge TTE. In our study, moderate or greater PVR was seen in
`only 9.1% of patients. It is conceivable that these differences in fre-
`quency may be due to different thresholds for what is graded as mod-
`erate and whether the grading is performed at the site or by a core
`laboratory. In our study, using core lab assessed PVR, the differential
`LVED response of each PVR group (20.16 + 0.60 cm in none/trace,
`20.04 + 0.60 cm in mild, and +0.09 + 0.64 cm in mod/severe) as
`well as the differential LV mass response of each PVR group,
`suggest that these groups represent truly different volume loads on
`the ventricle and support the grading scheme used by the core lab.
`Nonetheless, a standardized and comprehensive system for assessing
`PVR, including quantitative assessment as proposed by the recently
`published VARC-2 guidelines,31 may help elucidate the true impact
`of varying severities of PVR.
`Another question is whether PVR results in higher mortality or is
`simply associated with other factors leading to late mortality. Prior
`studies have suggested that higher PVR rates are seen in patients
`with certain clinical characteristics such as severe aortic valve
`
`Page 5 of 9
`
`

`

`Downloaded from
`
`http://eurheartj.oxfordjournals.org/
`
` by guest on August 3, 2016
`
`Page 6 of 9
`
`S. Kodali et al.
`
`Figure 2 Improvement in NYHA symptom class at 6 months following transcatheter aortic valve replacement stratified by severity of post-implant
`paravalvular regurgitation: none/trace vs. mild vs. moderate/severe. P-value represents comparison between groups indicated.
`
`Table 4 Multivariable predictors of all-cause 1-year
`mortality
`
`Multivariable analysis: baseline and procedural predictors of
`1-year mortality
`... ..... ...... ...... ...... ...... ...... ...... ...... ..... ...... ...... ...... ...... .
`P-valuea
`Variable
`Hazard
`95% Confidence
`ratio
`interval
`... ..... ...... ...... ...... ...... ...... ...... ...... ..... ...... ...... ...... ...... .
`Major arrhythmia
`1.41
`1.14–1.75
`0.002
`TF vs. TA
`0.73
`0.59–0.91
`0.005
`AV annulus diameter
`1.07
`1.03–1.11
`0.001
`(per 1 mm increase)
`BMI (per 1 kg/m2 increase) 0.95
`Total distance walked
`0.97
`(per 10 m increase)
`AV mean gradient
`(per 1 mmHg)
`Paravalvular regurgitation
`None/trace
`Mild
`Moderate/severe
`Renal disease (CR ≥2)
`
`0.93–0.97
`0.96–0.98
`
`,0.0001
`,0.0001
`
`0.98
`
`0.97–0.99
`
`,0.0001
`
`Referent
`1.35
`2.20
`1.35
`
`–
`1.07–1.72
`1.60–3.03
`1.04–1.74
`
`–
`0.013
`,0.0001
`0.023
`
`Potential covariates: age, sex, BMI, STS score, DM, smoking, prior CABG, prior BAV,
`frailty, renal disease, access route, major arrhythmia, pacemaker, chronic obstructive
`pulmonary disease, anaemia, 6 min walk, total distance walking (imputing 0 time for
`those who do not walk), LV ejection fraction, LV mass, LVED, LVES, AV annulus
`diameter, AV mean gradient, and baseline moderate/severe total AR.
`aAll results are from Cox regression.
`
`calcification.9,27,32 – 37 Some investigators have argued that these
`other factors may be markers for a higher risk population and thus
`PVR is simply associated with higher late mortality. The current
`
`analysis does demonstrate that the baseline clinical and echocardio-
`graphic characteristics of the patients with none/trace, mild, and
`moderate/severe PVR are different. A larger BSA and a higher per-
`centage of males are seen with a greater severity of PVR. In addition,
`patients with moderate/severe PVR had larger ventricular dimen-
`sions and mass, and more baseline regurgitant valvular disease,
`worse ejection fraction, and lower stroke volumes. However, even
`after accounting for these differences, the multivariable analysis
`demonstrates that both mild and moderate/severe PVR result in
`higher 1-year mortality.
`The current analysis is the first to investigate the impact of PVR
`on left ventricular (LV) geometry and function following TAVR.
`Although some studies have suggested no adverse outcome asso-
`ciated with lack of LV mass regression, Ali et al.38 showed that on
`multivariable analysis, LV mass reduction of .150 g remained an in-
`dependent predictor of improved long-term survival. Other studies
`have shown that early regression of LV mass following TAVR39 is
`associated with improved diastolic function. Ours is the first study
`to show that LV mass regression following TAVR may be blunted
`by the presence of PVR. However, this analysis is confounded by
`baseline differences in important echocardiographic parameters
`such as LV dimensions and EF. The mechanism by which the degree
`of LV mass regression and changes in geometry relate to PVR
`deserves further study.
`
`Limitations
`Although all echocardiograms were analysed in a core laboratory, a
`number of limitations exist for the semi-quantitative criteria used
`for grading. First, the qualitative grading scheme for transcatheter
`valves is adapted from that used for native valves. Interpretation is
`made more difficult because the intact calcified cusps and annulus
`
`Page 6 of 9
`
`

`

`Downloaded from
`
`http://eurheartj.oxfordjournals.org/
`
` by guest on August 3, 2016
`
`Paravalvular regurgitation following TAVR
`
`Page 7 of 9
`
`Table 5 Echocardiographic parameters at baseline, 1 year, and the mean difference (D) between these two time-points
`utilizing paired data
`
`D M
`
`PVR severity
`
`P-value baseline
`1-year
`Baseline
`D P-value
`. .. .. . .. .. .. . .. .. .. .. . .. .. .. . .. .. .. . .. .. ..
`mean +++++ SD
`mean +++++ SD
`ean +++++ SD
`vs. 1 year
`(a) vs. (b)
`(a) vs. (c)
`(b) vs. (c)
`. ............ ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ .........
`iEOA (cm2/m2)
`0.37 + 0.10
`0.92 + 0.27
`0.55 + 0.27 ,0.0001
`(a) None/trace
`0.46
`0.22
`0.42
`(n ¼ 759)
`(b) Mild (n ¼ 536)
`(c) Moderate/
`severe (n ¼ 104)
`. ............ ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ .........
`125.0 + 48.6
`123.4 + 46.9
`21.6 + 34.4
`LVED (mL)
`(a) None/trace
`0.72
`0.09
`0.03
`0.23
`(n ¼ 161)
`(b) Mild (n ¼ 141)
`(c) Moderate/
`severe (n ¼ 30)
`. ............ ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ .........
`63.5 + 39.8
`57.8 + 34.6
`25.8 + 25.6
`0.03
`0.12
`0.06
`0.33
`LVES (mL)
`(a) None/trace
`(n ¼ 161)
`(b) Mild (n ¼ 141)
`(c) Moderate/
`severe (n ¼ 30)
`. ............ ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ .........
`2.2 + 9.6
`53.7 + 12.4
`56.0 + 9.8
`LVEF (%)
`(a) None/trace
`,0.0001
`0.21
`0.87
`0.61
`(n ¼ 831)
`(b) Mild (n ¼ 567)
`(c) Moderate/
`severe (n ¼ 114)
`. ............ ............ ............ ............ ........... ............ ............ ............ ........... ............ ............ ............ ........... ............ .........
`133.8 + 37.5
`114.1 + 31.7
`219.6 + 32.8 ,0.0001
`LV mass index
`(a) None/trace
`0.22
`0.001
`0.01
`(n ¼ 689)
`(g/m2)
`(b) Mild (n ¼ 480)
`(c) Moderate/
`severe (n ¼ 101)
`
`0.36 + 0.10
`0.35 + 0.12
`
`0.93 + 0.25
`0.93 + 0.29
`
`0.57 + 0.26 ,0.0001
`0.58 + 0.28 ,0.0001
`
`135.4 + 45.8
`145.2 + 51.6
`
`141.2 + 50.3**
`160.4 + 60.7**
`
`5.8 + 39.9
`15.2 + 47.4
`
`0.08
`0.06
`
`66.1 + 37.7
`74.3 + 41.6
`
`65.2 + 36.9
`78.8 + 42.2**
`
`20.9 + 29.1
`4.5 + 29.0
`
`0.93
`0.34
`
`52.0 + 13.1
`51.4 + 13.7
`
`54.9 + 9.8
`53.9 + 9.8
`
`2.9 + 10.3 ,0.0001
`2.4 + 9.9
`0.02
`
`140.8 + 36.8*
`147.9 + 34.9*
`
`123.5 + 34.8** 217.3 + 33.0 ,0.0001
`139.5 + 38.1** 28.4 + 26.6
`0.0015
`
`*P , 0.017 was used to declare statistical significance in order to account for multiple comparisons.
`**P , 0.001 between none/trace vs. mild PVR or none/trace vs. moderate/severe groups at 1 year.
`
`significantly influence the location, shape, and direction of the para-
`valvular jets. In addition, PVR may need to be assessed differently
`for each type of transcatheter valve. In addition, although a multivari-
`able analysis was performed to account for the differences in baseline
`characteristics, there may be other factors that are unaccounted for
`in this analysis including residual confounding from measurement
`errors. And lastly, while the Jonckhere–Terpstera test was used
`for ordering in non-adjusted analyses, it was relaxed for multivariable
`modelling.
`
`Conclusions
`Paravalvular regurgitation after TAVR remains a frequent occurrence
`and significant differences in baseline clinical and echocardiographic
`characteristics exist between severity groups. In spite of comparable
`improvement in the valve area, the ventricular response following
`TAVR is adversely influenced by increasing grades of PVR. Despite
`baseline differences, this large cohort study shows that the presence
`of either mild or moderate/severe PVR significantly impacts 1-year
`mortality, after adjusting for putative confounding variables. Efforts
`should be made to reduce PVR following TAVR with improved
`
`annular sizing (with 3D imaging) and procedural technique to opti-
`mize results.
`
`Acknowledgements
`The authors would like t