`
`Transcatheter aortic valve replacement versus surgical valve replacement
`in intermediate-risk patients: a propensity score analysis
`Published online in The Lancet on April 3, 2016
`
`This article discusses a clinical study of Edwards Lifesciences’ SAPIEN 3 transcatheter heart valve.
`The PARTNER II S3i study was sponsored by Edwards in support of the FDA approval of the use of
`the SAPIEN 3 valve for transcatheter aortic valve replacement (TAVR) in intermediate-risk patients.
`
`Some of the authors have a consulting or other financial relationship with Edwards and are
`compensated in excess of the amount set forth in 21 U.S.C. §54.2, for services including
`training and proctoring of physicians in the safe and e(cid:30)ective use of Edwards’ products,
`speaking engagements, advice on product development, and similar activities. This includes
`Drs. Herrmann, Lim, Webb, and Whisenant.
`
`Edwards is not aware of publications reaching conclusions di(cid:30)erent from those described in
`this article.
`
`Also enclosed is the approved Instructions for Use for the SAPIEN 3 valve.
`
`Edwards Lifesciences
`One Edwards Way · Irvine, CA USA 92614
`Phone: 949.250.2500 · Fax: 949.250.2525 · www.edwards.com
`
`Page 1 of 54
`
`Edwards Lifesciences v. Boston Scientific Scimed
`IPR2017-00060 U.S. Patent 8,992,608
`Exhibit 2003
`
`
`
`Articles
`
`Transcatheter aortic valve replacement versus surgical valve
`replacement in intermediate-risk patients: a propensity
`score analysis
`
`Vinod H Thourani, Susheel Kodali, Raj R Makkar, Howard C Herrmann, Mathew Williams, Vasilis Babaliaros, Richard Smalling, Scott Lim,
`S Chris Malaisrie, Samir Kapadia, Wilson Y Szeto, Kevin L Greason, Dean Kereiakes, Gorav Ailawadi, Brian K Whisenant, Chandan Devireddy,
`Jonathon Leipsic, Rebecca T Hahn, Philippe Pibarot, Neil J Weissman, Wael A Jaber, David J Cohen, Rakesh Suri, E Murat Tuzcu, Lars G Svensson,
`John G Webb, Jeffrey W Moses, Michael J Mack, D Craig Miller, Craig R Smith, Maria C Alu, Rupa Parvataneni, Ralph B D’Agostino Jr, Martin B Leon
`
`Summary
`Background Transcatheter aortic valve replacement (TAVR) with the SAPIEN 3 valve demonstrates good 30 day clinical
`outcomes in patients with severe aortic stenosis who are at intermediate risk of surgical mortality. Here we report
`longer-term data in intermediate-risk patients given SAPIEN 3 TAVR and compare outcomes to those of intermediate-
`risk patients given surgical aortic valve replacement.
`
`Methods In the SAPIEN 3 observational study, 1077 intermediate-risk patients at 51 sites in the USA and Canada were
`assigned to receive TAVR with the SAPIEN 3 valve [952 [88%] via transfemoral access) between Feb 17, 2014, and
`Sept 3, 2014. In this population we assessed all-cause mortality and incidence of strokes, re-intervention, and aortic
`valve regurgitation at 1 year after implantation. Then we compared 1 year outcomes in this population with those for
`intermediate-risk patients treated with surgical valve replacement in the PARTNER 2A trial between Dec 23, 2011,
`and Nov 6, 2013, using a prespecified propensity score analysis to account for between-trial differences in baseline
`characteristics. The clinical events committee and echocardiographic core laboratory methods were the same for both
`studies. The primary endpoint was the composite of death from any cause, all strokes, and incidence of moderate or
`severe aortic regurgitation. We did non-inferiority (margin 7·5%) and superiority analyses in propensity score
`quintiles to calculate pooled weighted proportion differences for outcomes.
`
`Findings At 1 year follow-up of the SAPIEN 3 observational study, 79 of 1077 patients who initiated the TAVR procedure
`had died (all-cause mortality 7·4%; 6·5% in the transfemoral access subgroup), and disabling strokes had occurred in
`24 (2%), aortic valve re-intervention in six (1%), and moderate or severe paravalvular regurgitation in 13 (2%). In the
`propensity-score analysis we included 963 patients treated with SAPIEN 3 TAVR and 747 with surgical valve
`replacement. For the primary composite endpoint of mortality, strokes, and moderate or severe aortic regurgitation,
`TAVR was both non-inferior (pooled weighted proportion difference of –9·2%; 90% CI –12·4 to –6; p<0·0001) and
`superior (–9·2%, 95% CI –13·0 to –5·4; p<0·0001) to surgical valve replacement.
`
`Interpretation TAVR with SAPIEN 3 in intermediate-risk patients with severe aortic stenosis is associated with low
`mortality, strokes, and regurgitation at 1 year. The propensity score analysis indicates a significant superiority for our
`composite outcome with TAVR compared with surgery, suggesting that TAVR might be the preferred treatment
`alternative in intermediate-risk patients.
`
`Funding None.
`
`Introduction
`is
`(TAVR)
`Transcatheter aortic valve replacement
`established for treatment of severe symptomatic aortic
`stenosis in patients deemed to be at high risk of surgical
`mortality or who are not suitable for surgery.1,2 The
`encouraging clinical outcomes with earlier-generation
`TAVR systems in cohorts of high-risk patients,3–8 as well
`as rapid device refinements that led to improved clinical
`outcomes,9–11 have generated interest in use of these
`devices in intermediate-risk patients. A 2015 report from
`an observational study using
`the
`latest-generation
`SAPIEN 3 TAVR system (Edwards Lifesciences, Irvine,
`CA, USA) indicated good 30 day outcomes in both high-
`risk and intermediate-risk patients.10 Mortality and the
`
`intermediate-risk
`in
`incidence of disabling strokes
`patients were both about 1% and moderate or severe
`paravalvular regurgitation was recorded in about 4%. The
`SAPIEN 3 valve system differs from previous versions
`through the following factors: improved geometry of the
`trileaflet bovine pericardial valve; different cobalt alloy
`frame, which is longer than the early version of the
`balloon-expandable valve system (SAPIEN XT valve;
`Edwards Lifesciences) with more open outlet cells and
`denser inlet cells; a polyethylene terephthalate fabric skirt
`sewn to the bottom portion of the interior and exterior of
`the frame (providing an external circumferential seal to
`reduce paravalvular leak); four valve sizes (20 mm,
`23 mm, 26 mm, and 29 mm diameters); and lower-profile
`
`Published Online
`April 3, 2016
`http://dx.doi.org/10.1016/
`S0140-6736(16)30073-3
`See Online/Comment
`http://dx.doi.org/10.1016/
`S0140-6736(16)30122-2
`Emory University School of
`Medicine, Atlanta, GA, USA
`(Prof V H Thourani MD,
`V Babaliaros MD,
`C Devireddy MD); Columbia
`University Medical Center,
`New York, NY, USA
`(S Kodali MD, R T Hahn MD,
`Prof J W Moses MD,
`Prof C R Smith MD, M C Alu MM,
`Prof M B Leon MD); Cedars-Sinai
`Medical Center, Los Angeles,
`CA, USA (R R Makkar MD);
`University of Pennsylvania,
`Philadelphia, PA, USA
`(Prof H C Herrmann MD,
`W Y Szeto MD); NYU Langone
`Medical Center, New York, NY,
`USA (M Williams MD);
`University of Texas Memorial
`Hermann Heart and Vascular
`Institute, Houston, TX, USA
`(Prof R Smalling MD); University
`of Virginia, Charlottesville, VA,
`USA (S Lim MD, G Ailawadi MD);
`Northwestern University,
`Chicago, IL, USA
`(S C Malaisrie MD); Cleveland
`Clinic, Cleveland, OH, USA
`(Prof S Kapadia MD,
`W A Jaber MD, R Suri MD,
`Prof E M Tuzcu MD,
`Prof L G Svensson MD); Mayo
`Clinic, Rochester, MN, USA
`(K L Greason MD); The Christ
`Hospital, Cincinnati, OH, USA
`(Prof D Kereiakes MD);
`Intermountain Medical Center,
`Salt Lake City, UT, USA
`(B K Whisenant MD); St Paul’s
`Hospital, Vancouver, BC,
`Canada (J Leipsic MD,
`Prof J G Webb MD); Laval
`University, Quebec, QC, Canada
`(Prof P Pibarot PhD); Medstar
`Health Research Institute and
`Georgetown University,
`Washington, DC, USA
`
`www.thelancet.com Published online April 3, 2016 http://dx.doi.org/10.1016/S0140-6736(16)30073-3
`
`1
`
`Page 2 of 54
`
`
`
`Articles
`
`(Prof N J Weissman MD); Saint
`Luke’s Mid America Heart
`Institute, Kansas City, MO, USA
`(Prof D J Cohen MD); Baylor
`Scott and White Healthcare,
`Plano, TX, USA
`(Prof M J Mack MD); Stanford
`University, Stanford, CA, USA
`(Prof D C Miller MD);
`Cardiovascular Research
`Foundation, New York, NY,
`USA (R Parvataneni MS); and
`Wake Forest School of
`Medicine, Winston-Salem, NC,
`USA (Prof R B D’Agostino Jr PhD)
`Correspondence to:
`Dr Vinod H Thourani, Emory
`Hospital Midtown, Atlanta,
`GA 30308, USA
`vthoura@emory.edu
`
`2
`
`Research in context
`
`Evidence before this study
`Before we did the PARTNER 2A and SAPIEN 3 studies, clinical
`trial evidence comparing transcatheter aortic valve replacement
`(TAVR) to surgery was mainly limited to patients at high risk of
`death during surgery. Data from large national registries have
`now indicated a global trend towards TAVR being used in
`lower-risk populations despite little rigorous clinical trial
`evidence for this practice. We searched MEDLINE on
`Jan 31, 2016, with the terms “transcatheter aortic valve
`implantation”, “transcatheter aortic valve implantation in low
`risk patients”, “transcatheter aortic valve implantation in
`intermediate risk patients”, “transcatheter aortic valve
`replacement in low risk patients”, “transcatheter aortic valve
`replacement in intermediate risk patients”, “surgical aortic valve
`replacement”, and “surgical aortic valve replacement in
`intermediate risk patients” in English with no date limitations.
`The published studies include a small randomised trial and
`several non-adjudicated comparisons between TAVR and
`surgery in intermediate-risk patients. These studies did not use
`
`neurologists for stroke assessment, a clinical events committee
`to adjudicate outcomes, or core laboratories for analysis of
`imaging studies.
`
`Added value of this study
`We show TAVR with SAPIEN 3 to be superior to surgery at 1 year
`follow-up with lower rates of all-cause mortality, stroke, and
`the composite endpoint of mortality, stroke, and moderate or
`severe aortic regurgitation, but higher rates or moderate or
`severe regurgitation. Our analysis is the first rigorously
`designed and carried out clinical study to compare TAVR with
`the SAPIEN 3 device with surgery in intermediate-risk patients.
`The prespecified propensity analysis allows for meaningful
`comparisons between the two groups.
`
`Implications of all the available evidence
`TAVR should be considered as the preferred alternative to
`surgery in intermediate-risk patients and future should
`consider expanding the indications for TAVR.
`
`delivery catheters with more precise valve positioning
`inserted through 14 or 16 French expandable sheaths for
`transfemoral access.
`Here we aimed to report 1 year outcomes with
`SAPIEN 3 TAVR in intermediate-risk patients from this
`observational study and
`then use a prespecified
`propensity score analysis to compare these outcomes
`with those for similar patients given surgical aortic valve
`replacement in the PARTNER 2A randomised trial.
`
`Methods
`Study design and participants
`In this analysis we used populations from the PARTNER
`2 SAPIEN 3 intermediate risk observational study10 and
`the PARTNER 2A randomised trial (NCT01314313).12
`These two prospective multicentre studies enrolled
`patients with symptomatic severe aortic stenosis who
`were considered to be at intermediate risk for 30 day
`surgical mortality. Risk status was evaluated by a Heart
`Team that included cardiac surgeons. Patients were
`deemed intermediate risk via clinical assessment or if
`their Society of Thoracic Surgeons (STS) score was 4% or
`higher. In those with an STS score lower than 4%, the
`Heart Team deemed the patient intermediate risk if they
`had risk factors not present within the predictive score
`(eg, liver disease, frailty, and pulmonary hypertension).
`In PARTNER 2A, patients were randomly assigned to
`receive either surgical valve replacement or TAVR using
`SAPIEN XT; here we analyse only the patients assigned
`to surgery.12 In the SAPIEN 3 study, all TAVR patients
`who were eligible to receive a valve had mandated
`multidetector computed tomography (MDCT) analysed
`by the study core laboratory and were presented on a
`conference call in which a screening committee reviewed
`
`imaging and clinical data and approved patients prior to
`enrolment.
`Inclusion and exclusion criteria for SAPIEN 310 and
`PARTNER 2A12 were the same. Key exclusion criteria
`were a congenitally bicuspid aortic valve, severe aortic
`regurgitation, left ventricular ejection fraction lower than
`20%, severe renal insufficiency, and estimated life
`expectancy of less than 2 years. Patients with non-
`complex coronary disease requiring revascularisation
`could be enrolled if a treatment plan for the coronary
`disease (medical therapy or revascularisation) was agreed
`on before enrolment. Both trials were approved by the
`institutional review boards of each participating site and
`written informed consent was provided by all patients.
`
`Procedures
`Preprocedural valve sizing for TAVR was determined
`through MDCT or 3D transoesophageal echo cardio-
`graphy. Access was via transfemoral, trans apical, or
`trans aortic
`routes, depending on preprocedural
`peri pheral vascular assessments. Postoperative dual
`antiplatelet therapy with aspirin and clopidogrel was
`recommended for at least 1 month, at the Heart Team’s
`discretion. Warfarin was also recommended in patients
`with atrial fibrillation, based on patient tolerance.
`The co-principal investigators and other members of
`the executive committee had access to the data after the
`database was locked and prepared the manuscript. The
`same executive committee was used for both trials and
`attest to the completeness and accuracy of the data and
`adherence of
`the studies
`to
`the protocol. All
`echocardiograms in patients given TAVR were analysed
`independently by a consortium of echocardiography
`core
`laboratories. The grading of paravalvular
`
`Page 3 of 54
`
`
`
`Articles
`
`regurgitation was based on an expanded and more
`granular classification scheme, which was then reduced
`to the standard classification scheme.13 Clinical events
`were independently adjudicated by a clinical events
`committee and neurologists examined all patients to
`ascertain any changes in neurological status after the
`procedures. The clinical events committee and echo-
`cardiographic core laboratory methods were the same
`for both studies. Clinical outcomes were reported as
`defined by Valve Academic Research Consortium
`(VARC)-2 definitions.14
`
`Statistical analysis
`For our analysis of 1 year clinical outcomes after the
`SAPIEN 3 TAVR procedure we included only patients
`who
`initiated
`the procedure—ie,
`the as-treated
`population. Echocardiography outcomes are based on
`the valve-implanted cohort (ie, patients who received the
`valve as assigned and excluding those who did not
`complete or died during the procedure). For statistical
`comparisons between baseline, 30 day, and 1 year values
`we used McNemar’s test.
`In a prespecified analysis we used propensity score
`analysis methodology to compare outcomes from the
`SAPIEN 3 TAVR patient cohort with those of similar
`intermediate-risk patients in the surgical arm of the
`PARTNER 2A
`trial who received surgical valve
`replacement (appendix).15,16 This analysis was done in the
`valve-implanted analysis population. The primary
`endpoint for the propensity score analysis was the 1 year
`non-hierarchical composite event of death from any
`cause, all strokes, and post-treatment aortic regurgitation
`(moderate or greater [severe]). Secondary endpoints
`included each of the individual components of the
`composite primary endpoint.
`Propensity score methodology was used to reduce the
`confounding in the statistical comparison of outcomes
`of two treatment groups from the two different studies
`by accounting for differences in baseline patient
`characteristics. First, a logistic regression model was
`performed on the prespecified baseline characteristic
`variables to calculate the propensity score for each
`patient (appendix). Propensity scores represent the
`likelihood that the patient was in the TAVR arm. All
`patients with propensity scores were partitioned into
`five quintiles (1–5; patients in quintile 1 represent the
`lowest 20% of propensity scores while patients in
`quintile 5 represent the highest 20% of scores) based on
`their propensity scores. The variable balance was then
`assessed to confirm the adequacy of the propensity
`model. Within each quintile, patients in both groups
`had similar degrees of residual bias from randomness.
`An independent biostatistician blinded to the treatment
`groups and with no knowledge of the clinical outcome
`data carried out the propensity score analyses. The issue
`of missing baseline covariates was addressed using
`multiple imputation techniques before the propensity
`
`score model was finalised and propensity score and
`quintiles derived for the outcome analyses adjustments.
`The analyses of the primary and secondary endpoints
`were based on the proportion difference between the two
`treatment groups, and were done as a non-inferiority
`analysis with designated non-inferiority margins. To
`estimate the overall treatment effect and confidence
`limits adjusted for the propensity score quintiles, we
`calculated the weighted proportion diff erence using the
`average treatment effect on the treated methods with
`weights derived based on the sample size of the SAPIEN 3
`cohort within each quintile. If the upper bound of the
`two-sided 90% CI was 7·5% or lower (absolute margin),
`then non-inferiority was fulfilled. Superiority between
`the groups in this study was established if the upper
`bound of the two-sided 95% CI for the difference in
`proportion was lower than 0%.
`Time-to-event Kaplan-Meier curves were constructed
`for mortality, stroke, and a composite of death and stroke
`at 1 year, using all available follow-up data in the as-
`treated population for both studies. All-cause mortality
`up to 1 year was plotted for patients who received TAVR,
`stratified by paravalvular regurgitation classification and
`
`See Online for appendix
`
`1078 assigned to TAVR in SAPIEN 3
`study
`
`1021 randomly allocated to surgery
`in PARTNER 2A
`
`1 died before treatment
`
`68 withdrew before treatment
`4 ineligible due to aortic calcification
`or deteriorating condition
`5 died before treatment
`
`1077 initiated procedure
`(as-treated cohort)
`
`944 initiated procedure
`(as-treated cohort)
`
`8 did not receive allocated procedure
`3 ineligible as identified by
`intraoperative TEE
`1 inability to gain access
`2 annular ruptures
`1 died during treatment
`1 due to device embolisation
`
`8 did not receive allocated procedure
`5 due to aortic calcification
`1 hypotensive event during
`anaesthesia, BAV only
`2 not treated as assigned
`
`1069 implanted with TAVR valve
`(valve-implanted cohort)
`
`936 implanted with surgical valve
`(valve-implanted cohort)
`
`17 exited study before 1 year without
`death or stroke event
`14 had unknown alive status before
`1 year (no success phone sweep)
`74 with missing 1 year echocardiogram
`1 without propensity score
`
`21 exited study before 1 year without
`death or stroke event
`3 unknown alive status before 1 year
`(no success phone sweep)
`165 with missing 1 year echocardiogram
`
`963 assessed for primary endpoint
`
`747 assessed for primary endpoint
`
`Figure 1: Study profile
`TAVR=transcatheter aortic valve replacement. BAV=balloon aortic valvuloplasty. TEE=transoesophageal
`echocardiogram.
`
`www.thelancet.com Published online April 3, 2016 http://dx.doi.org/10.1016/S0140-6736(16)30073-3
`
`Page 4 of 54
`
`
`
`Articles
`
`with a log-rank p value calculated for differences between
`classes. The SAPIEN 3 observational study was nested
`within the PARTNER 2 trial, which was registered with
`ClinicalTrials.gov, number NCT01314313.
`
`Role of the funding source
`There was no funding source for this analysis; SAPIEN 3
`and PARTNER 2A were sponsored by Edwards Lifesciences.
`The corresponding author had full access to all the data
`in the study and had final responsibility for the decision
`to submit for publication.
`
`Results
`From Feb 17, 2014, to Sept 3, 2014, 1078 intermediate- risk
`patients were enrolled in the SAPIEN 3 observational
`study at 51 sites in the USA and Canada. One patient died
`before treatment, leaving 1077 patients in the as-treated
`analysis population (figure 1). From Dec 23, 2011, to
`Nov 6, 2013, 2032 inter mediate-risk patients were enrolled
`in the PARTNER 2A randomised trial at 57 sites in the
`USA and Canada. 1021 patients were randomly allocated
`to surgery and 944 received surgical valve replacement
`
`TAVR population
`(n=1077)
`
`Surgery population
`(n=944)
`
`p value
`
`Age (years)
`Men
`Body-mass index (kg/m²)
`Society of Thoracic Surgeons score (%)
`NYHA class III or IV
`Coronary artery disease
`Previous myocardial infarction
`Previous CABG
`Previous PCI
`Previous BAV
`Cerebrovascular disease
`Peripheral vascular disease
`COPD
`Any
`Oxygen dependent
`Creatinine ≥177 µmol/L
`Atrial fibrillation
`Permanent pacemaker
`Frail condition
`15 ft walk time >7 s
`Albumin <35 g/L
`Aortic valve area (cm²)
`Mean gradient (mm Hg)
`Left ventricular ejection fraction (%)
`Left ventricular mass index (g/m²)
`Moderate or severe mitral regurgitation
`
`81·9 (6·6)
`665 (62%)
`28·7 (6·1)
`5·2 (4·3–6·3)
`781 (73%)
`750 (70%)
`172 (16%)
`301 (28%)
`344 (32%)
`55 (5%)
`97 (9%)
`304 (28%)
`
`322/1075 (30%)
`54/1070 (5%)
`81 (8%)
`388 (36%)
`142 (13%)
`
`434/1050 (41%)
`138/1056 (13%)
`0·7 (0·17)
`46·1 (12·6)
`58·5 (13·4)
`116·27 (33·5)
`91/1033 (9%)
`
`81·6 (6·76)
`519 (55%)
`28·4 (6·2)
`5·4 (4·4–6·7)
`718/943 (76%)
`628 (67%)
`167 (18%)
`243 (26%)
`254 (27%)
`45 (5%)
`97 (10%)
`304 (32%)
`
`283/938 (30%)
`28/931 (3%)
`51 (5%)
`329 (35%)
`113 (12%)
`
`391/855 (46%)
`138/925 (15%)
`0·7 (0·20)
`44·7 (12·6)
`55·4 (11·8)
`118·7 (32·2)
`153/841 (18%)
`
`0·23
`0·002
`0·32
`0·0002
`0·07
`0·14
`0·31
`0·27
`0·01
`0·76
`0·36
`0·052
`
`0·92
`0·02
`0·058
`0·61
`0·42
`
`0·057
`0·24
`0·45
`0·01
`<0·0001
`0·12
`<0·0001
`
`Data are mean (SD), median (IQR), n (%), or n/N (%). TAVR=transcatheter aortic valve replacement. NYHA=New York
`Heart Association. CABG=coronary artery bypass grafting. PCI=percutaneous coronary intervention. BAV=balloon
`aortic valvuloplasty. COPD=chronic obstructive pulmonary disease.
`
`Table 1: Demographics and baseline characteristics of the patients in the as-treated populations
`
`4
`
`(ie, the as-treated population; figure 1). The main reason
`for non-treatment in the surgery arm (85 patients) was
`withdrawal from the study, most commonly due to a
`decision not to have surgery after randomisation.
`Most baseline characteristics were similar between the
`SAPIEN 3 and the PARTNER 2A patient populations
`(table 1). Patients in the SAPIEN 3 group were more
`frequently male and had more frequent oxygen-
`dependent chronic obstructive pulmonary disease,
`whereas PARTNER 2A surgery patients had higher
`median STS scores, lower mean gradients and left
`ventricle ejection fractions, and had more frequent
`moderate or severe mitral regurgitation (table 1). The
`median postoperative length of hospital stay was shorter
`in the TAVR cohort than in the surgical cohort (4 days
`[range 1·0–122·0] vs 9 days [1·0–77·0]) and a higher
`percentage of patients went home after the procedure
`(912 [85%] vs 436 [46%]).
`79 of the patients who received TAVR with SAPIEN 3
`died within 1 year (all-cause mortality 7·4%; table 2).
`61 (7%) patients died in the subset of 925 who had
`received transfemoral TAVR (appendix). 49 (5%) patients
`had a stroke, with nearly half having disabling strokes
`(table 2). At 1 year, 119 (11%) patients were rehospitalised
`for procedure-related or valve-related reasons, but both
`endocarditis and aortic valve re-interventions were rare
`(table 2). 132 (12%) had a new pacemaker permanently
`implanted. At 1 year after TAVR, cardiac symptoms had
`significantly improved, with 94% of patients in New York
`Heart Association function class I or II (appendix).
`The improvements in mean aortic valve areas and
`gradients after TAVR seen at 30 days were maintained at
`1 year (valve areas 1·7 cm² and gradient 11·4 mm Hg;
`appendix). On both a standard grading system and an
`extended grading
`system, moderate or greater
`paravalvular regurgitation at 1 year was noted only in
`1·5% of patients after TAVR; 40% of patients had mild
`regurgitation on standard grading (appendix). In patients
`with no or trace paravalvular regurgitation at 30 days,
`mortality at 1 year was 4·5%, which was similar to those
`patients with mild paravalvular regurgitation (6·4%).
`However, in patients with moderate or severe paravalvular
`regurgitation, the 1 year mortality was significantly
`higher at 13·3% (log-rank p=0·0184; figure 2).
`In each of the quintiles, patients undergoing TAVR had
`a lower incidence of the composite primary endpoint
`than did the group who received surgery (varying from
`–14·5% in quintile 1 to –4·3% in quintile 5; table 3). The
`non-inferiority analysis was based on the pooled weighted
`proportion difference of –9·2% (90% CI –12·4 to –6)
`favouring TAVR, which was below the 7·5% non-
`inferiority margin (p<0·0001).
`TAVR was superior to surgery for the composite
`endpoint (weighted difference of proportions –9·2%,
`95% CI –13·0 to –5·4; p<0·0001), and for the individual
`outcomes of death (–5·2%, –8·0 to –2·4; p= 0·0003) and
`stroke (–3·5%, –5·9 to –1·1; p=0·0038). Surgery was
`
`Page 5 of 54
`
`
`
`Articles
`
`30 days
`
`SAPIEN 3 TAVR
`
`Surgical valve replacement
`
`SAPIEN 3 TAVR
`
`Surgical valve replacement
`
`1 year
`
`Events
`(n)
`
`Cumulative KM
`estimates
`
`Numbers
`at risk
`
`Events
`(n)
`
`Cumulative
`KM estimates
`
`Numbers
`at risk
`
`Events
`(n)
`
`Cumulative
`KM estimates
`
`Numbers
`at risk
`
`Events
`(n)
`
`Cumulative
`KM estimates
`
`Numbers
`at risk
`
`12
`10
`2
`
`4
`
`29
`11
`18
`22
`
`1·1%
`0·9%
`0·2%
`
`0·4%
`
`2·7%
`1·0%
`1·7%
`2·0%
`
`1063
`1063
`1063
`
`1059
`
`1035
`1053
`1045
`1053
`
`38
`29
`9
`
`4
`
`57
`41
`16
`75
`
`4·0%
`3·1%
`1·0%
`
`0·4%
`
`6·1%
`4·4%
`1·7%
`8·0%
`
`902
`902
`902
`
`898
`
`852
`868
`886
`868
`
`79
`47
`32
`
`18
`
`49
`24
`25
`90
`
`119
`3
`··
`
`7·4%
`4·5%
`3·1%
`
`1·8%
`
`4·6%
`2·3%
`2·4%
`8·4%
`
`11·4%
`0·3%
`··
`
`963
`963
`963
`
`945
`
`930
`953
`940
`953
`
`865
`961
`··
`
`121
`74
`47
`
`15
`
`75
`54
`22
`155
`
`135
`28
`··
`
`13·0%
`8·1%
`5·3%
`
`1·7%
`
`8·2%
`5·9%
`2·4%
`16·6%
`
`15·1%
`3·1%
`··
`
`795
`795
`795
`
`782
`
`743
`764
`773
`764
`
`697
`780
`··
`
`Death
`From any cause
`Cardiac death
`Non-cardiac death
`Neurological events
`Transient ischaemic
`attack
`Any stroke
`Disabling stroke
`Non-disabling stroke
`Death from any cause
`or disabling stroke
`Rehospitalisation
`Myocardial infarction
`Life-threatening or
`disabling bleeding
`Major vascular
`complication
`Acute kidney injury
`(stage 3)
`New atrial fibrillation
`New permanent
`pacemaker
`Endocarditis
`Aortic valve
`re-intervention
`
`49
`3
`50
`
`66
`
`5
`
`54
`109
`
`2
`1
`
`4·6%
`0·3%
`4·6%
`
`6·1%
`
`0·5%
`
`5·0%
`10·2%
`
`0·2%
`0·1%
`
`1017
`1060
`1018
`
`1000
`
`1058
`
`1012
`955
`
`1061
`1062
`
`62
`18
`440
`
`51
`
`31
`
`265
`68
`
`0
`0
`
`6·8%
`1·9%
`46·7%
`
`5·4%
`
`3·3%
`
`28·3%
`7·3%
`
`0·0%
`0·0%
`
`845
`889
`493
`
`860
`
`879
`
`649
`836
`
`902
`902
`
`··
`
`··
`
`63
`132
`
`8
`6
`
`··
`
`··
`
`5·9%*
`12·4%*
`
`0·8%
`0·6%
`
`··
`
`··
`
`912
`842
`
`958
`958
`
`··
`
`··
`
`272
`85
`
`6
`4
`
`··
`
`··
`
`29·2%
`9·4%
`
`0·7%
`0·5%
`
`··
`
`··
`
`564
`721
`
`792
`794
`
`TAVR=transcatheter aortic valve replacement. KM=Kaplan-Meier. *Site-reported data.
`
`Table 2: Unadjusted clinical outcomes in the as-treated populations
`
`to TAVR for moderate or severe aortic
`superior
`regurgitation (1·2%, 0·2 to 2·2; p=0·0149).
`Time-to-event analyses for the SAPIEN 3 TAVR and
`PARTNER 2A surgery cohorts for all-cause death, all
`strokes, and the composite event of death and strokes are
`shown in the appendix, as are those for transfemoral-
`access TAVR only. Important differences between TAVR
`and surgery for each endpoint are observed in the first
`several months, with the curves then remaining parallel
`(appendix).
`
`Discussion
`TAVR with SAPIEN 3 resulted in excellent clinical
`outcomes after 1 year follow-up, including low rates of
`death, stroke, and cardiac symptoms. Paravalvular
`regurgitation was also low at 1 year and the presence of a
`higher proportion of patients with mild regurgitation
`after TAVR did not affect mortality (figure 2). Based on
`our propensity score analysis, TAVR in intermediate-risk
`patients was superior at 1 year to surgical valve
`replacement, with lower rates of death, strokes, and the
`
`composite endpoint of death, strokes, and moderate to
`severe aortic regurgitation combined.
`The clinical outcomes with SAPIEN 3 TAVR reported
`from the multicentre, endpoint-adjudicated observational
`study at 30-days10 and at 1 year are improved compared
`with previous randomised trials and observational studies
`in both high-risk or extreme-risk and inter mediate-risk
`patients and with either balloon-expandable or self-
`expanding TAVR devices.5–8,17,18 Particularly, the low 1 year
`rates of death from any cause, strokes, and paravalvular
`regurgitation establish a new clinical per formance
`standard (table 2). Moreover, other infrequent but
`nonetheless important procedure-related com plications,
`including coronary occlusion, annulus rupture, and
`device embolisation were also decreased after TAVR at
`30 days.10 The reasons for improved outcomes are
`probably multi factorial and include more strategic case
`selection and planning (including systematic MDCT
`annulus sizing and aortic root assessment), increased
`operator experience, facilitated in-hospital patient care
`pathways, and device enhancements (eg, lower-profile
`
`www.thelancet.com Published online April 3, 2016 http://dx.doi.org/10.1016/S0140-6736(16)30073-3
`
`Page 6 of 54
`
`
`
`TAVR outcomes in lower-risk patients, but no rigorous
`clinical trial validation has been provided. In our analysis
`of a population of intermediate-risk patients, propensity
`score matching suggests that TAVR with SAPIEN 3 is
`superior to surgery to prevent death and strokes at 1 year.
`The use of propensity score analyses to adjust for
`differences in baseline patient characteristics for the
`purpose of comparing non-randomised patient cohorts is
`a well accepted statistical methodology.15,16 The robustness
`of the propensity score analysis in this study is
`strengthened by the fact that patient inclusion and
`exclusion criteria were almost identical between cohorts,
`event adjudication was performed by the same clinical
`events committee, the echocardiography core laboratory
`methods were the same, all trial management processes
`were the same, and most of the clinical sites and
`operators were also the same.
`An early concern about TAVR was the possibility of an
`increased frequency of periprocedural embolic strokes.6,7
`Over time, and with greater operator experience and
`lower-profile TAVR systems, a recent meta-analysis has
`shown fewer strokes after TAVR.28 Despite the presence
`of high neurological scrutiny of all patients to improve
`ascertainment of periprocedural strokes in this study,
`stroke frequency was low (5%) in patients who received
`TAVR and was significantly reduced compared with
`surgery in the propensity score analysis (table 2).
`Previously, a distinguishing characteristic of TAVR
`has been the unpredictable and, at times, disturbingly
`frequent occurrence of procedure-related paravalvular
`regurgitation, especially when compared with rates after
`surgical valve replacement.6,7,27 In our propensity score
`analysis, moderate or severe aortic regurgitation was
`still higher in the SAPIEN 3 TAVR cohort than in the
`surgery cohort. However, at 1 year, the absolute
`frequency of paravalvular regurgitation was lower than
`has been previously reported in TAVR studies with core
`laboratory evaluations,5–8 and unlike in previous studies,
`only moderate or severe, not mild, paravalvular
`regurgitation was associated with
`increased
`late
`mortality (figure 2).29,30 The SAPIEN 3 valve combined
`with improved MDCT valve sizing and more precise
`valve positioning techniques might be responsible for
`the reduction of moderate or severe paravalvular
`regurgitation to sufficiently low levels that treatment of
`lower-risk patients should no longer be discouraged.
`Nevertheless, even mild paravalvular regurgitation
`might be a limitation of TAVR compared with surgery.
`Although we do not predict worsening mortality as a
`consequence of mild paravalvular regurgitation at 1 year,
`longer follow-up will be required to determine the effect
`of it on outcomes.
`Several limitations must be noted to accurately interpret
`the findings from this analysis. First, although a careful
`propensity score analysis justifies strong conclusions, it is
`not a randomised trial and relevant confounders might
`not be represented in the risk-adjustment process, which
`
`Articles
`
`100
`
`Mild vs none or trace log-rank p=0·2306
`Moderate or severe vs none or trace log-rank p=0·0184
`
`None or trace
`Mild
`Moderat