prosthesis-patient mismatch in 62,125 patients following transcatheter aortic valve ... · 1...
TRANSCRIPT
1
Prosthesis-Patient Mismatch in 62,125 Patients Following Transcatheter Aortic Valve
Replacement: From the STS/ACC TVT Registry
Howard C. Herrmann MD1, Samuel A. Daneshvar MD2, Gregg C. Fonarow, MD2, Amanda
Stebbins3, Sreekanth Vemulapalli MD3, Nimesh D. Desai MD1, David J. Malenka MD4, Vinod
H. Thourani MD5, Jennifer Rymer MD3, Andrzej S. Kosinski PhD3
1Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania; 2University of California Los Angeles, Los Angeles California; 3Duke Clinical Research
Institute, Durham North Carolina; 4Dartmouth-Hitchcock, Lebanon New Hampshire; 5MedStar
Heart and Vascular Institute and Georgetown University, Washington, DC
Brief Title: Outcomes of PPM Following TAVR
Disclosures: HCH reports institutional research funding from Abbott Vascular, Bayer, Boston
Scientific, Edwards Lifesciences, Medtronic, and St Jude Medical and consulting for Edwards
Lifesciences, Medtronic, and Siemens Healthineers. GCF reports consulting for Abbott Vascular
and Medtronic. VHT reports consulting for Abbott Vascular, Boston Scientific, Edwards
Lifesciences, and Gore Vascular. NDD reports institutional research funding from Abbott
Vascular, Medtronic, and Gore and consulting for Edwards Lifesciences, Medtronic, Abbott
Vascular, and Gore.
Address for correspondence:
Howard C. Herrmann, MD
University of Pennsylvania
PCAM South Pavilion 11-107
3400 Civic Center Boulevard
Philadelphia, Pennsylvania 19104
Telephone: (215) 662-2180
E-mail: [email protected]
Twitter: @Penn | @gcfmd
2
Abstract
Background: Prosthesis-patient mismatch (PPM) after surgical aortic valve replacement (AVR)
for aortic stenosis is generally associated with worse outcomes. Transcatheter AVR (TAVR) can
achieve a larger valve orifice and the effects of PPM after TAVR are less well studied.
Objective: We utilized the STS/ACC Transcatheter Valve Therapy (TVT) Registry to examine
the frequency, predictors, and association with outcomes of PPM after TAVR in 62,125 patients
enrolled between 2014 and 2017.
Methods: Based on the discharge echocardiographic effective valve area indexed to body
surface area (EOAI), PPM was classified as severe (<0.65 cm2/m2), moderate (0.65-0.85
cm2/m2), or none (>0.85 cm2/m2). Multivariable regression models were utilized to examine
predictors of severe PPM as well as adjusted outcomes, including mortality, heart failure (HF)
rehospitalization, stroke, and quality of life (QOL), at 1 year in 37,470 Medicare patients with
claims linkage.
Results: Severe and moderate PPM were present following TAVR in 12% and 25% of patients,
respectively. Predictors of severe PPM included small (<23 mm diameter) valve prosthesis,
valve-in-valve procedure, larger BSA, female sex, younger age, non-White/Hispanic race, lower
ejection fraction, atrial fibrillation, and severe mitral or tricuspid regurgitation. At 1 year,
mortality was 17.2%, 15.6%, and 15.9% in severe, moderate, and no PPM patients, respectively
(p=0.02). Heart failure (HF) re-hospitalization had occurred in 14.7%, 12.8%, and 11.9% of
patients with severe, moderate, and no PPM, respectively (p<0.0001). There was no association
of severe PPM with stroke or QOL score at 1 year.
Conclusions: Severe PPM after TAVR was present in 12% of patients and was associated with
higher mortality and HF rehospitalization at 1 year. Further investigation is warranted into the
prevention of severe PPM in patients undergoing TAVR.
Condensed Abstract: We examined the outcomes of PPM in 62,125 patients receiving TAVR
and enrolled in the STS/ACC TVT Registry between 2014 and 2017. Severe and moderate PPM
were present in 12% and 25% of patients, respectively. Patients with severe PPM were more
likely female, younger, non-White/Hispanic, received a small prosthesis (<23 mm diameter) or
underwent a valve-in-valve procedure. Severe PPM was associated with greater 1-year mortality
(HR 1.19) and HF re-hospitalization (HR 1.12). Further investigation is warranted into
prevention of severe PPM in patients undergoing TAVR.
Key Words: aortic stenosis, prosthesis-patient mismatch, transcatheter aortic valve
replacement
Abbreviations
AS Aortic Stenosis
BMI Body Mass Index
BSA Body Surface Area
CABG Coronary Artery Bypass Graft
EOAI Effective Orifice Area Indexed
PCI Percutaneous Coronary Intervention
PPM Prosthesis-Patient Mismatch
SAVR Surgical Aortic Valve Replacement
TAVR Transcatheter Aortic Valve Replacement
VIV Valve-in-Valve
3
Introduction
Prosthesis-patient mismatch (PPM) was first defined by Rahimtoola in 1978 to describe the
mismatch between the hemodynamics of a valve prosthesis and the patient requirements for
cardiac output (1,2). It is defined based on the effective valve orifice area indexed to body
surface area (EOAI). Standard values for moderate and severe PPM after aortic valve
replacement (AVR) for severe aortic stenosis (AS) have been suggested and validated in
numerous studies over decades (3-5).
Many studies have investigated PPM after surgical AVR (SAVR). In a large meta-
analysis including 34 of these studies, Head and colleagues demonstrated PPM in 44% of
patients with a statistically significant association with all-cause mortality (6). More recently,
Fallon and colleagues, utilizing the Society of Thoracic Surgery (STS) Adult Cardiac Surgical
Database, also showed that both moderate and severe PPM following SAVR were associated
with reduced 10-year survival and an increased risk for hospital readmission (7). Other surgical
series have suggested that PPM adversely affects functional improvement and exercise tolerance,
left ventricular mass regression, and late structural valve deterioration (8, 9).
Transcatheter AVR (TAVR) has been shown to result in larger EOA compared with
SAVR (10, 11). The associations of PPM with outcomes following TAVR have been studied in
small series with limited follow up (10-17). In this report from the Society of Thoracic
Surgeons/American College of Cardiology (ACC) Transcatheter Valve Therapy (TVT) Registry,
we report the incidence, predictors, and 1-year outcome of PPM in 62,125 patients undergoing
TAVR in the US between 2014 and 2017.
Methods
The TVT Registry.
4
The STS/ACC TVT Registry is a joint initiative of the STS and ACC with primary goals of
facilitating device and procedure surveillance, promoting quality assurance and improvement,
and conducting studies that help with access to new therapies and expand device labelling
through evidence development (18). Participating centers in the registry use standardized
definitions to collect and report patient-specific data on demographics, morbidities, functional
status, quality of life, hemodynamics, procedural details and outcomes (in-hospital, 30-day, and
1-year). The ACC National Cardiovascular Data Registry warehouse and the Duke Clinical
Research Institute Data Analysis Center both implement data quality checks, including feedback
reports, and examine data ranges and consistency to optimize completeness and accuracy. In
addition, TVT registry audits are performed by a third party at randomly selected sites (ten
percent yearly) and are designed to complement internal quality controls by examining the
accuracy, consistency, and completeness of the data collected within the database. A central
institutional review board (Chesapeake Research Review Inc.) approves activities of the TVT
registry. The present study has been granted a waiver of informed consent.
Study Cohort.
Transcatheter AVR was commercially approved for use in high and extreme risk surgical
patients in the United States on November 2, 2011. Subsequent approvals for a second device as
well as expanding indications including intermediate risk patients have resulted in a further
increase in its use. All patients receiving TAVR since the initial US Food and Drug
Administration approval in 2011 through the 1st quarter of 2017 are enrolled in the TVT registry.
The analysis set for the present investigation included all patients enrolled between January 1,
2014 and March 31, 2017. TVT enrollees > 65 years of age at the time of their procedure were
5
linked to fee-for-service Medicare administrative claims data by CMS using unique patient
identifiers (name and social security number) as previously described (19).
Study Outcomes and Definitions.
Procedural and in-hospital outcomes were determined from data in the TVT registry. Standard
definitions, in accordance with the Valve Academic Research Consortium and American Society
of Echocardiography guidelines, were used for collection of data elements in the registry as
instructed in the data dictionary supplied to sites and harmonized with the STS national database,
whenever possible. Prosthesis-Patient Mismatch was classified based on the discharge
echocardiographic effective valve orifice area (calculated with the continuity equation) indexed
to body surface area (EOAI) as severe (<0.65 cm2/m2), moderate (0.65-0.85 cm2/m2), or none
(>0.85 cm2/m2) (3, 4). In order to account for data entry and measurement errors, the first and
99th percentile of EOAI data were excluded resulting in 62,125 patients available for analysis
(Figure 1 and On-line Figure 1). In-hospital outcomes were collected from the TVT registry, and
both stroke and re-hospitalization were adjudicated by a board-certified cardiologist using Valve
Academic Research Consortium definitions (4). This process involved review of specific site
queries and deidentified source records as needed.
For clinical events after hospital discharge, data from CMS administrative claims were
used. Death following hospital discharge was identified using the Medicare Denominator file.
Re-hospitalization events were determined from CMS administrative claims data using the
International Classification of Diseases, Ninth and Tenth Revisions, Clinical Modification codes
for rehospitalization for heart failure and stroke. Linkage was achieved for 37,470 (68% of
eligible) Medicare patients (on-line figure 1).
6
Data Analyses.
Baseline demographics, comorbidities, past cardiac history, cardiac anatomy and function, and
procedural factors were analyzed to identify predictors of PPM. These characteristics are
reported as frequencies for discrete factors and medians with quartiles for continuous measures.
Pearson chi-square tests are reported for the categorical factors and the chi-square rank test is
reported for continuous characteristics. A multivariate logistic regression model of severe PPM
was generated. Variables analyzed included: age, gender, non-White/Hispanic, BSA, left
ventricular ejection fraction, number of days from 2011 to procedure date, aortic valve mean
gradient, severe tricuspid valve insufficiency, BMI, peripheral arterial disease, atrial
fibrillation/flutter, severe mitral regurgitation, diabetes, severe post procedure valvular
insufficiency, prior implantable cardioverter-defibrillator, hemoglobin, glomerular filtration rate,
proximal left anterior descending stenosis, porcelain aorta, degenerative aortic valve disease
etiology, current/recent smoker, prior stroke/transient ischemic attack, prior non-aortic valve
procedure, New York Heart Association class IV, pacemaker, chronic liver disease, prior
myocardial infarction, dialysis, prior CABG, valve prosthesis size, valve-in-valve procedure.
The reported model includes all pre-specified factors regardless of statistical significance. The
forest plot displays specific key factors of the overall model which were of clinical interest.
The primary outcomes of interest for this study were death, heart failure hospitalization,
death or HF, stroke, and the overall KCCQ score 1 year after TAVR. Unadjusted and adjusted
Cox proportional models were generated for the binary endpoints of interest. The Generalized
Estimating Equation (GEE) method with exchangeable working correlation structure was used to
account for within-hospital clustering. The hazard ratios and 95% confidence interval Chi-square
and p-value are reported. The adjusted model includes characteristics found to be predictive of
7
30-day mortality in the TAVR registry and additional risk factors considered relevant by the
authors. The stroke model also included discharge antiplatelet therapy. Correlation between risk
factors was assessed. Cumulative incidence plots were generated. For endpoints with the
competing risk of death (HF hospitalizaiton and stroke), the Fine-Gray test statistic is reported.
For all other endpoints the Chi-square test is reported. These models were run an additional time
on patients who were alive at 30 days. Additional sub-group mortality models were run to assess
interactions between severe PPM and age (dichotomized by the median), gender, LVEF <40% or
>=40%, BMI <30 or >=30 kg/m2, aortic valve mean gradient (<40 or >=40 mmHg), and
baseline atrial fibrillation/flutter.
We assessed quality of life (QOL) at 30 days and 1 year after TAVR using the overall
KCCQ score in 9285 patients. To avoid the bias of missing non-random KCCQ measurements
due to worse baseline health status, sites reporting <50% completeness of measurements were
excluded. To ensure that the cohort of patients represented the overall TAVR population, we
used inverse probability weighting to increase the weight of patients who were most like those
with missing KCCQ measurements. These weights were attained from a multivariable logistic
regression model which predicted the probability of having KCCQ data and used in the
multivariate linear regression model of 1-year KCCQ score. This model was generated to assess
the relationship between 1-year KCCQ measurement and severe PPM after adjusting for other
known factors. To address within site bias, results from the GEE method were implimented.
Modelings assumptions of linearity and normality were tested. An additional model was
generated to assess KCCQ at 1 year in the context of a favorable outcome. This endpoint is
defined as being alive at 1 year, reporting a 1-year KCCQ score of > 60 and having a <10-point
decrease in KCCQ score from baseline (20). The same approach stated above was used in this
8
analysis. The resulting multivariate model was a logistic regression model. All modeling
assumptions were assessed.
SAS statistical software, version 9.3 (SAS Institute Inc., Cary, North Carolina), was used
for all calculations. Analyses were performed at the TVT Registry Analysis Center at the Duke
Clinical Research Institute.
Results
Patients and Incidence of PPM.
The study population was comprised of 62,125 patients after exclusion of the first and 99th
percentile of EOAI of all enrolled patients in the STS/ACC TVT registry for commercial TAVR
between January 2014 and March 2017 (on-line Figure 1). The mean + SD for the EOAI was 1.0
+ 0.3 cm2/m2 (range 0.4 to 2.1 cm2/m2) (Figure 1). Severe and moderate PPM were observed in
12.1% (n=7514) and 24.6% (n=15271) of patients, respectively (Central Illustration), and did
not change significantly between 2014 and 2017. Baseline factors for all patients and those with
severe, moderate, and no PPM are compared in Table 1. The mean age for all patients was 81
years, 46% were female, and 94% White. Cardiac co-morbidities included prior CABG (26%),
prior stroke (12%), diabetes (38%), moderate/severe chronic lung disease (26%), > stage 3
chronic kidney disease (48%), class III/IV HF (80%), atrial fibrillation/flutter (40%). Patients
with PPM were younger, more likely White, and had more cardiac and non-cardiac co-morbid
conditions (Table 1). These patients also had smaller annulus diameters and were more likely to
have undergone a VIV procedure. The percentage of patients receiving a valve prosthesis <23
mm diameter was 27.9% (40.0%, 32.1%, and 24.0% for severe, moderate, and no PPM,
respectively, p<0.0001). The percentage of patients undergoing TAVR VIV was 5.6% (14.7%,
6.1%, and 3.6% for severe, moderate, and no PPM, respectively, p<0.0001).
9
Predictors of PPM
Multivariate logistic regression was utilized to identify predictors of severe PPM (Table 2 and
Figure 2). Important predictors (OR [95%CI]) included a valve prosthesis < 23 mm diameter
(2.8 [2.6-3.0], p<0.001), VIV procedure (2.8 [2.5-3.0], p<0.001), larger BSA (1.7 per 0.2 unit
increase [1.7-1.8], p<0.001), lower left ventricular ejection fraction (1.1 per 5% decrease [1.08-
1.11], p<0.001), Non-White/Hispanic (1.2 [1.1-1.3], p<0.001), female (1.5 [1.4-1.6], p<0.001),
younger age, atrial fibrillation, larger BMI, higher aortic valve mean gradient, prior CABG, and
severe mitral or tricuspid regurgitation.
Outcomes
Linkage of registry patients to CMS administrative claims data was possible for 37,470 Medicare
patients (68% of eligible patients). These patients had important differences from the patients
who were not linked, including older age, more likely female, white, and more likely to have
cardiac and other co-morbid conditions (on-line Table 1). The incidence of severe and moderate
PPM in the linked population was similar to the overall population at 11.4% and 24.4%,
respectively.
At 30 days of follow-up, patients with severe PPM had higher rates of HF hospitalization,
stroke, and death. After 1 year, patients with severe PPM also had a higher mortality in addition
to the other endpoints (Table 3, on-line figure 2 and on-line Table 2). At 1 year, mortality was
17.2%, 15.6%, and 15.9% in severe, moderate, and no PPM patients, respectively (p=0.02).
Heart failure hospitalization had occurred in 14.7%, 12.8%, and 11.9% of patients with severe,
moderate, and no PPM, respectively (p<0.0001). After multivariate adjustment, only severe
PPM was associated with the adverse outcomes of death (Central Illustration), HF
hospitalization, and combined death or HF hospitalization (Figure 3). The adjusted HRs
10
(95%CI) for severe PPM as compared to moderate and none (combined) for death, HF
hospitalization, and stroke at 1 year were 1.19 (1.09-1.431, p<0.001), 1.12 (1.02-1.24, p=0.017),
and 0.98 (0.82-1.15, p=0.8) (Figure 3 and on-line Table 2).
There were no significant interactions identified between severe PPM and mortality in
sub-groups of patients dichotomized by median age, gender, BMI, LV EF, BMI, aortic valve
mean gradient, or atrial fibrillation/flutter (Table 3). Similarly, the major findings were
unchanged after exclusion of patients who died by 30 days (on-line Table 3).
Quality of Life.
Patients with severe PPM had higher baseline mean KCCQ scores (44.6 + 24.3) compared with
patients with moderate (41.8 + 24.3) or no (39.5 + 23.7) PPM (p<0.0001). The change in KCCQ
score 30 days after TAVR (reported in 74% of patients) improved less in patients with severe
PPM (27.4 + 26.8) compared with moderate (29.2 + 27.0) and no (31.1 + 27.4) PPM such that
mean scores were similar in all 3 groups at 30 days. Mean KCCQ score (75.6 + 21.7) was
available in 67% of patients at 1 year. In multivariate linear regression models, there was no
difference between severe and not severe (moderate or no) PPM in KCCQ score at 1 year (effect
estimate 0.722, 95% CI 0.064-8.122, p=0.792) or in favorable outcome (effect estimate 0.986,
95% CI 0.807-1.203, p=0.886) (On-line Table 4).
Discussion
This is the largest study to date of PPM following TAVR, including >60,000 patients
treated with commercial devices in the United States between 2014 and 2017. The major
findings are: 1. Severe and moderate PPM are common after TAVR occurring in 12% and 24%,
respectively of patients; 2. Severe PPM is related to prosthesis and patient factors and can be
predicted by small (<23 mm diameter) valve prosthesis, valve-in-valve procedure, larger BSA,
11
female sex, younger age, non-White/Hispanic, lower ejection fraction, atrial fibrillation, and
severe mitral or tricuspid regurgitation; 3. Severe PPM, though not moderate PPM, is associated
with an increased 1-year risk for mortality and heart failure rehospitalization when compared
with patients with moderate or no PPM (Central Illustration); 4. Finally, quality of life as
measured by KCCQ score is initially higher in patients with severe PPM (as compared with
moderate and none), but there is less improvement at 30 days with no difference in QOL
outcome at 1 year. These findings regarding the frequency and association of severe PPM with
worse outcomes have important implications for further improving outcomes in patients
undergoing TAVR.
Previous studies of PPM
Prior studies of PPM following SAVR have demonstrated important, but variable, adverse
associations with survival, left ventricular mass regression, functional status and quality of life
(5). In a meta-analysis, Head and colleagues demonstrated that both severe and moderate PPM
were associated with an increase in all-cause mortality of 34% and 19%, respectively, with
follow-up periods up to 10 years (6). However, other single center or smaller studies have failed
to confirm the effect of PPM on outcomes (21-23). Methodologic differences in studies may
account for the varying results. Using predicted EOAI based on manufacturer’s data, literature-
derived valve performance, or measured prosthesis diameter has limited accuracy for predicting
in vivo valve area (24). Other important factors include use of gradient alone, use of categorical
versus continuous EOAI values, and failure to adequately adjust for co-variate factors (5, 25). In
a recent study of 59,779 patients included in the STS database who had surgery between 2004
and 2014, Fallon and colleagues used literature-derived projected EOAI and observed severe and
moderate PPM in 11% and 54%, respectively, of patients (7). In Fallon’s study, severe and
12
moderate PPM were both associated with an increased risk for mortality, readmission for heart
failure, and redo AVR (7). However, the incidence of moderate PPM in this study was
particularly high, likely due to the use of overly optimistic manufacturer’s estimates of EOA (2).
Several studies have examined the incidence and short term outcomes in patients with
PPM after TAVR (Table 4). In these studies, the incidence of severe PPM has varied from 1 to
28%. The associations with outcomes have included a short-term increase in mortality, less
symptomatic improvement, an increased risk for acute kidney injury, and less left ventricular
mass regression, and not all studies demonstrated an association with mortality (10, 11, 13, 14,
16, 17). In the studies that included a surgical comparison group, all showed a reduced incidence
of PPM with TAVR as compared to surgical AVR (10, 11, 13). The different outcomes in these
studies are likely due to small numbers, different TAVR prostheses, and differences in the
patient populations.
The present investigation
The present study is the largest one to date to examine the incidence, predictors, and outcomes of
PPM in TAVR patients. We utilized the STS/ACC TVT Registry of >60,000 patients
undergoing commercial TAVR procedures as well as linkage to CMS to examine outcomes in
Medicare patients. We utilized individual patient level measured EOAI, a strength of our
analysis, as predicted EOA specifically in TAVR patients may be inappropriate as the final
geometric expansion of the TAVR prosthesis is unknown and may not be symmetrical (26). Our
findings on the incidence of severe and moderate PPM are consistent with most prior studies
which utilized measured EOAI. We demonstrate that even in short term (1-year) follow-up,
severe (but not moderate) PPM is associated with higher mortality and HF rehospitalization after
adjustment for co-morbid risk factors.
13
Our findings differ from the recent OCEAN-TAVI study in Japanese patients which
found a much lower incidence of severe PPM (<1%) and no association with survival at 1 year
(14). This difference likely relates to a combination of different prostheses and the smaller size
of the Japanese population (BSA 1.41 m2) as compared with the median BSA in our study of US
patients (BSA 1.87 m2). Following surgical AVR, Mohty et al demonstrated an increased effect
of severe PPM on mortality in patients <75 years of age, those with lower EF, and lower BMI
(27), prompting the VARC-2 guideline to recommend a lower cutoff for severe PPM of <0.60
cm2/m2 (instead of 0.65 cm2/m2) in patients with BMI >= 30 kg/m2 (4). These results conflict
with those of Fallon et al. who reported a greater effect of severe PPM in patients with BMI >30
kg/m2 (7). We did not observe significant interactions for severe PPM and mortality following
TAVR in these or other subgroups (Table 3).
We found important predictors of PPM, with the highest ORs in larger patients and those
receiving smaller prostheses. Other studies (both surgical and transcatheter) have confirmed that
predictors of PPM include larger BSA (7, 10, 11, 13, 14, 17, 22), smaller prosthesis size (7, 10,
14, 21), more severe baseline aortic stenosis (16, 17), younger age (7, 10, 13, 22), female sex (7,
22), left ventricular dysfunction (7, 22), and severe mitral or tricuspid regurgitation (7).
This is the first study to assess the association between PPM after TAVR and quality of
life utilizing a quantitative measure (KCCQ score), although some prior studies have assessed
symptomatic status. Bleiziffer demonstrated a decrease in exercise capacity 6 months after
surgery in 312 patients (28), while most other surgical series have demonstrated no effect of
PPM on short term functional status (5). After TAVR, studies of PPM have shown either no
association with a change in NYHA class at 6 months to 2 years (11, 13, 16) or less improvement
at 6 months (17). We also assessed the association of severe PPM with a favorable outcome that
14
combines quality of life and mortality as previously described by Arnold (20). We did not find
an association between severe PPM and QOL (either overall KCCQ score or favorable outcome)
at 1 year. Potential factors that could explain the discrepancy between HF rehospitalization and
QOL include missing data in QOL measurements, patients’ self-reported perception of QOL
which improved in all groups, changes in therapy after HF hospitalization, or survival bias.
Limitations
This is an observational registry study and has the inherent limitations associated with
retrospective analyses including residual measured and unmeasured confounding. However, this
is a very large study with all commercial TAVR procedures performed in the US in a recent time
frame. It is possible that procedural complications affected our outcomes analysis, however a
separate analysis of 30-day survivors (on-line table 3) did not suggest an effect on our
conclusions. Linkage to CMS Medicare claims was obtained in a subset of the entire study
population. This subset remains large and similar to previous TVT to Medicare linkage efforts
with a similar incidence of severe and moderate PPM. Finally, EOAI was calculated from
measured echocardiographic hemodynamics at hospital discharge. It is possible that these
measurements could be influenced by peri-procedural issues and might be more accurate if
obtained at a later time point. Nonetheless, our measured values for EOAI are consistent with
prior studies and more accurate than those obtained by either projection or geometric
measurement.
Clinical Implications and Summary
Our findings suggest that efforts should be made to identify and limit the risk for PPM after
TAVR. Surgeons have employed a variety of techniques to reduce the risk for PPM, including
aortic root enlargement and the use of supra-annular prostheses and those with thinner sewing
15
rings. Most importantly, PPM can be recognized before or predicted at the time of surgery (24,
25). Awareness of this problem and the use of techniques to minimize its occurrence have
resulted in a reduction of 55% in the incidence of severe PPM from 13.8% in 2004 to 6.2% in
2014 (7). We did not observe a similar trend in our study of TAVR patients over a shorter time
period and confounded by device iterations and evolving patient indications.
The TAVR community should follow this lead by identifying patients at risk for severe
PPM and considering techniques to reduce the risk. Jilaihawi demonstrated that optimal
(reduced LV depth) positioning of a self-expanding prosthesis was associated with a reduction in
moderate and severe PPM from 48% to 16% (15). In a non-randomized comparison of devices
utilized for valve-in-valve TAVR, several studies have demonstrated lower gradients and less
PPM with the use of self-expanding as compared to balloon-expandable prostheses (29, 30). A
recent hemodynamic study also demonstrated that self-expanding prostheses have larger EOA
for the same labelled size device (31). Finally, fracture of a previous surgical prosthesis prior to
TAVR can allow for placement of larger TAVR prostheses for VIV implants (32). A future
study that compares devices and techniques to limit PPM in patients at risk for severe PPM
would be of interest to guide decision making in this population.
16
Clinical Perspectives
Competency in Medical Knowledge: Prosthesis-patient mismatch is common after
transcatheter aortic valve replacement, occurring in 12% (severe) and 25% (moderate) of
patients.
Competency in Patient Care and Procedural Skills: Severe prosthesis-patient mismatch can
be identified based on echocardiographic assessment of valve hemodynamics and is associated
with a number of patient factors.
Translational Outlook: A future study that compares devices and techniques to identify and
limit the risk of prosthesis-patient mismatch after TAVR would be of interest to guide decision
making in this population.
17
References
1. Rahimtoola SH. The problem of valve prosthesis-patient mismatch. Circulation 1978;58:20-4.
2. Pibarot P, Dumesnil JG. Valve prosthesis-patient mismatch, 1978 to 2011: from original concept
to compelling evidence. J Am Coll Cardiol. 2012;60(13):1136-9.
3. Zoghbi WA, Chambers JB, Dumesnil JG, et al. Recommendations for Evaluation of Prosthetic
Valves With Echocardiography and Doppler Ultrasound. J Am Soc Echo 2009;22:975-1014.
4. Kappetein AP, Head SJ, Genereux P, et al. Updated standardized endpoint definitions for
transcatheter aortic valve implantation: The Valve Academic Research Consortium-2 consensus
document. J Thorac Cardiovasc Surg 2013;145:6-23.
5. Daneshvar SA, Rahimtoola SH. Valve prosthesis-patient mismatch (VP-PM): a long-term
perspective. J Am Coll Cardiol. 2012;60(13):1123-35.
6. Head SJ, Mokhles MM, Osnabrugge RL, et al. The impact of prosthesis-patient mismatch on
long-term survival after aortic valve replacement: a systematic review and meta-analysis of 34
observational studies comprising 27 186 patients with 133 141 patient-years. Eur Heart J.
2012;33(12):1518-29.
7. Fallon JM, DeSimone JP, Brennan JM, et al. The Incidence and Consequence of Prosthesis-
Patient Mismatch After Surgical Aortic Valve Replacement. Ann Thorac Surg. 2018;106(1):14-
22
8. Flameng W, Herregods MC, Vercalsteren M, Herijgers P, Bogaerts K, Meuris B. Prosthesis-
patient mismatch predicts structural valve degeneration in bioprosthetic heart valves. Circulation.
2010;121(19):2123-9.
9. Popma JJ, Khabbaz K. Prosthesis-patient mismatch after "high-risk" aortic valve replacement. J
Am Coll Cardiol. 2014;64(13):1335-8.
18
10. Pibarot P, Weissman NJ, Stewart WJ, et all. Incidence and sequelae of prosthesis-patient
mismatch in transcatheter versus surgical valve replacement in high-risk patients with severe
aortic stenosis: a PARTNER trial cohort--analysis. J Am Coll Cardiol. 2014;64(13):1323-34.
11. Zorn GL 3rd, Little SH, Tadros P, et al. Prosthesis-patient mismatch in high-risk patients with
severe aortic stenosis: A randomized trial of a self-expanding prosthesis. J Thorac Cardiovasc
Surg. 2016;151(4):1014-22, 1023.e1-3.
12. Little SH, Oh JK, Gillam L, et al. Self-Expanding Transcatheter Aortic Valve Replacement
Versus Surgical Valve Replacement in Patients at High Risk for Surgery: A Study of
Echocardiographic Change and Risk Prediction. Circ Cardiovasc Interv. 2016;9(6). pii: e003426.
13. Thyregod HG, Steinbrüchel DA, Ihlemann N, et al. No clinical effect of prosthesis-patient
mismatch after transcatheter versus surgical aortic valve replacement in intermediate- and low-
risk patients with severe aortic valve stenosis at mid-term follow-up: an analysis from the
NOTION trial. Eur J Cardiothorac Surg. 2016 Oct;50(4):721-728. Epub 2016 Mar 22.
14. Miyasaka M, Tada N, Taguri M, e tal. Incidence, Predictors, and Clinical Impact of Prosthesis-
Patient Mismatch Following Transcatheter Aortic Valve Replacement in Asian Patients: The
OCEAN-TAVI Registry.; OCEAN-TAVI Investigators. JACC Cardiovasc Interv.
2018;11(8):771-780.
15. Jilaihawi H, Chin D, Spyt T, et al. Prosthesis-patient mismatch after transcatheter aortic valve
implantation with the Medtronic-Corevalve bioprosthesis. Eur Heart J. 2010;31(7):857-64.
16. Tzikas A, Piazza N, Geleijnse ML, et al. Prosthesis-patient mismatch after transcatheter aortic
valve implantation with the medtronic CoreValve system in patients with aortic stenosis. Am J
Cardiol. 2010;106(2):255-60.
19
17. Ewe SH, Muratori M, Delgado V, et al. Hemodynamic and clinical impact of prosthesis-patient
mismatch after transcatheter aortic valve implantation. J Am Coll Cardiol. 2011;58(18):1910-8
18. Carroll JD, Edwards FH, Marinac-Dabic D, et al. The STS-ACC Transcatheter Valve Therapy
National Registry: a new partnership and infrastructure for the introduction and surveillance of
medical devices and therapies. J Am Coll Cardiol. 2013;62(11):1026-1034.
19. Sorajja P, Vemulapalli S, Feldman T, et al. Outcomes With Transcatheter Mitral Valve Repair in
the United States: An STS/ACC TVT Registry Report. J Am Coll Cardiol. 2017;70(19):2315-
2327.
20. Arnold SV, Spertus JA, Vemulapalli S, et al. Quality of Life Outcomes After Transcatheter
Aortic Valve Replacement in an Unselected Population. A Report from the STS/ACC TVT
Registry. JAMA Cardiology 2017;2(4):409-416.
21. Hanayama N, Christakis GT, Mallidi HR, et al. Patient prosthesis mismatch is rare after aortic
valve replacement: valve size may be irrelevant. Ann Thorac Surg. 2002;73(6):1822-9;
discussion 1829.
22. Tully PJ, Aty W, Rice GD, Bennetts JS, Knight JL, Baker RA. Aortic valve prosthesis-patient
mismatch and long-term outcomes: 19-year single-center experience. Ann Thorac Surg.
2013;96(3):844-50.
23. Monin JL, Monchi M, Kirsch ME, et al. Low-gradient aortic stenosis: impact of prosthesis-
patient mismatch on survival. Eur Heart J. 2007;28(21):2620-6. Epub 2007 Sep 27.
24. Bleiziffer S, Eichinger WB, Hettich I, et al. Prediction of valve prosthesis-patient mismatch prior
to aortic valve replacement: which is the best method? Heart. 2007;93(5):615-20.
25. Pibarot P, Dumesnil JG, Cartier PC, Métras J, Lemieux MD. Patient-prosthesis mismatch can be
predicted at the time of operation. Ann Thorac Surg. 2001;71(5 Suppl):S265-8
20
26. Mangels DR, Siki M, Menon R, et al. Hemodynamic Effects of Valve Asymmetry in Sapien 3
Transcatheter Aortic Valves. J Invasive Cardiol. 2018;30(4):138-143.
27. Mohty D, Dumesnil JG, Echahidi N, et al. Impact of Prosthesis-Patient Mismatch on Long-Term
Survival After Aortic Valve Replacement. J Am Coll Cardiol 2009;53:39-47.
28. Bleiziffer S, Eichinger WB, Hettich I, et al. Impact of patient-prosthesis mismatch on exercise
capacity in patients after bioprosthetic aortic valve replacement. Heart 2008;94:637-641.
29. Dvir D, Webb JG, Bleiziffer S, et al. Transcatheter aortic valve implantation in failed
bioprosthetic surgical valves.; Valve-in-Valve International Data Registry Investigators. JAMA.
2014;312(2):162-70.
30. Pibarot P, Simonato M, Barbanti M, et al. Impact of Pre-Existing Prosthesis-Patient Mismatch on
Survival Following Aortic Valve-in-Valve Procedures. JACC Cardiovasc Interv.
2018;11(2):133-141.
31. Hahn RT, Leipsic J, Douglas PS, et al. Comprehensive Echocardiographic Assessment of
Normal Transcatheter Valve Function. JACC Cardiovasc Imaging. 2018 Jun 8. pii: S1936-878X
(18)30358-9.
32. Chhatriwalla AK, Allen KB, Saxon JT, et al. Bioprosthetic Valve Fracture Improves the
Hemodynamic Results of Valve-in-Valve Transcatheter Aortic Valve Replacement. Circ
Cardiovasc Interv. 2017;10(7). pii: e005216
21
Figure Legends
Central Illustration: Incidence and Effect on Survival of Severe PPM After TAVR.
This figure shows the incidence of PPM in the entire study population (62,125 patients) and the
adjusted 1-year mortality for 37,470 patients with CMS Medicare claims linkage. It demonstrates
that severe PPM is common after TAVR and is associated with greater 1-year mortality (HR
1.19). Further investigation is warranted into prevention of severe PPM in patients undergoing
TAVR.
Figure 1: EOAI Histogram of Effective Orifice Area Index.
The distribution of effective orifice area indexed to body surface area (EOAI) is shown for the
study population (N = 62,125) after exclusion of the 1st and 99th percentile In order to account for
data entry and measurement errors. The mean + SD for the EOAI was 1.0 + 0.3 cm2/m2 (range
0.4-2.1 cm2/m2)
Figure 2: Forest plot of predictors of severe PPM: Predictors of Severe PPM.
Significant predictors of severe PPM in a multivariate logistic regression model are shown in a
Forest plot (values are odds ratios with 95% CI and p values).
Figure 3: Adjusted Event Curves for the Effects of Severe PPM on 1-year Outcomes.
Adjusted event curves for HF rehospitalization, mortality and HF rehospitalization, and stroke
for severe, moderate, and no PPM are shown. At 1 year, HR (95% CI) for adverse outcome with
severe PPM are 1.12 (1.02-1.24, p=0.017), 1.13 (1.06-1.22, p<0.001), and 0.98 (0.82-1.16,
p=NS) for HF rehospitalization, mortality and HF rehospitalization, and stroke, respectively.
The outcome for mortality is shown in the Central Illustration.
22
Table 1: Baseline and Procedural Factors by Prosthesis-Patient Mismatch
Baseline Variable
(% or median
with 25th, 75th
quartiles)
All
(N =
62,125)
Severe PPM
(N = 7,514)
Moderate
PPM (N =
15,271)
None
(N = 39,340)
p
Age 82 (76, 87) 79 (72, 85) 81 (75, 86) 83 (77, 87) <0.0001
Gender (%male) 53.7 53.7 55.0 53.2 0.0007
Race (%African-
American)
3.8 5.2 4.3 3.3 <0.0001
Prior pacemaker
(%)
14.9 15.3 15.1 14.8 NS
Prior CABG (%) 25.5 29.4 26.3 24.5 <0.0001
Prior Stroke (%) 11.9 11.2 11.7 12.1 NS
PAD (%) 29.7 28.2 29.4 30.2 0.0017
HTN (%) 90.2 90.4 91.3 89.7 <0.0001
DM (%) 38.3 46.5 41.8 35.4 <0.0001
CLD (mod/severe) 26.1 30.4 27.6 24.7 <0.0001
CKD (Stage
3,GFR <60) (%)
48.3 50.3 49.7 47.4 <0.0001
STS PROM 6.0 (3.9,
9.3)
5.9 (3.7, 9.4) 5.8 (3.8,
9.2)
6.1 (4.0, 9.3) <0.0001
LV EF 58 (47, 63) 55 (43, 62) 57 (45, 63) 58 (50, 65) <0.0001
Prior MI (%) 24.1 25.5 24.9 23.5 <0.0001
NYHA III/IV (%) 79.6 82.4 80.2 78.9 <0.0001
AF/Fl (%) 40.0 42.6 41.2 39.0 <0.0001
Baseline KCCQ
Score
41 (24, 60) 36 (21, 56) 39 (22, 59) 43 (26, 63) <0.0001
BSA (M2) 1.87 (1.69,
2.04)
1.98 (1.80,
2.17)
1.93 (1.76,
2.10)
1.83 (1.66,
1.99)
<0.0001
Mean aortic
gradient (mmHg)
42 (34, 50) 42 (33, 51) 42 (34, 50) 42 (34, 50) NS
Procedural
Variable
VIV procedure (%) 5.6 14.7 6.1 3.6 <0.0001
Prosthesis <23mm
diameter (%)
27.9 40.0 32.1 24.0 <0.0001
Post AVA (cm2) 1.76 (1.40,
2.14)
1.10 (1.00,
1.23)
1.45 (1.30,
1.60)
2.00 (1.78,
2.40)
<0.0001
Post mean gradient
(mmHg)
9 (7, 13) 13 (9, 18) 11 (8, 14) 8 (6, 11) <0.0001
Post AR
(mod/severe, %)
2.8 2.1 2.7 2.9 <0.0001
LOS (days,
mean+SD)
5.9+9.4 6.6+17.0 5.8+8.2 5.7+7.6 <0.0001
23
Abbreviations: CABG (Coronary Artery Bypass Graft), PAD (Peripheral Artery Disease), HTN
(Hypertension), DM (Diabetes Mellitus), CLD (Chronic Liver Disease), CKD (Chronic Kidney
Disease), GFR (Glomerular Filtration Rate), STS PROM (Society Thoracic Surgeons Predicted
Risk of Mortality), LVEF (Left Ventricular Ejection Fraction), MI (Myocardial Infarction),
NYHA (New York Heart Association), PPM (Prosthesis-Patient Mismatch), AF/Fl (Atrial
Fibrillation and Flutter), KCCQ (Kansas City Cardiomyopathy Questionnaire), BSA (Body
Surface Area), VIV (Valve-in-Valve), AVA (Aortic Valve Area), AR (Aortic Regurgitation),
LOS (Length of Stay).
24
Table 2. Event Rates and Association of PPM with One-Year Endpoints
Endpoints (1-year)
Event Rate
(% vs %)
Unadjusted
Hazard Ratio
(95% CI) P-value
Adjusted
Hazard Ratio
(95% CI) P-value
Death
PPM Severe vs not
Severe
17.2 vs 15.8 1.12 (1.03 -
1.22)
0.011 1.19 (1.09 -
1.31)
<0.001
PPM Overall 0.027 <0.001
Moderate vs None 15.6 vs 15.9 0.98 (0.92 -
1.04)
0.441 1.00 (0.93 -
1.07)
0.999
Severe vs None 17.2 vs 15.9 1.11 (1.02 -
1.22)
0.019 1.19 (1.09 -
1.31)
<0.001
HF Hospitalization
PPM Severe vs not
Severe
14.7 vs 12.2 1.22 (1.11 -
1.33)
<0.001 1.12 (1.02 -
1.24)
0.017
PPM Overall <0.001 0.049
Moderate vs None 12.8 vs 11.9 1.08 (1.00 -
1.15)
0.036 1.02 (0.95 -
1.10)
0.567
Severe vs None 14.7 vs 11.9 1.24 (1.13 -
1.37)
<0.001 1.13 (1.03 -
1.25)
0.014
Death or HF
Hospitalization
PPM Severe vs not
Severe
26.8 vs 24.2 1.13 (1.06 -
1.21)
<0.001 1.13 (1.06 -
1.22)
<0.001
PPM Overall 0.001 0.002
Moderate vs None 24.6 vs 24.1 1.02 (0.97 -
1.07)
0.463 1.00 (0.95 -
1.06)
0.861
Severe vs None 26.8 vs 24.1 1.14 (1.06 -
1.22)
<0.001 1.13 (1.05 -
1.22)
<0.001
Stroke
PPM Severe vs not
Severe
3.8 vs 4.2 0.90 (0.77 -
1.05)
0.168 0.98 (0.82 -
1.16)
0.798
25
Endpoints (1-year)
Event Rate
(% vs %)
Unadjusted
Hazard Ratio
(95% CI) P-value
Adjusted
Hazard Ratio
(95% CI) P-value
PPM Overall 0.012 0.836
Moderate vs None 3.8 vs 4.4 0.86 (0.76 -
0.96)
0.011 0.96 (0.84 -
1.10)
0.587
Severe vs None 3.8 vs 4.4 0.86 (0.74 -
1.01)
0.059 0.97 (0.81 -
1.15)
0.701
26
Table 3.
Subgroup Analyses (Adjusted Models) of Association of Severe PPM and All-Cause
Mortality at 1 Year.
Subgroup Analyses
Mortality
Effect estimate
(95% CI) Chi-Square
P-value
Overall Age by Severe PPM Interaction 2.506 0.113
Severe PPM vs No Severe PPM Age <=83
years
1.123 (0.999,
1.261)
Severe PPM vs No Severe PPM Age >83
years
1.285 (1.129,
1.463)
Overall Gender by Severe PPM Interaction 0.866 0.352
Severe PPM vs No Severe PPM Male 1.153 (1.020,
1.303)
Severe PPM vs No Severe PPM Female 1.252 (1.104,
1.420)
Overall LVEF by Severe PPM Interaction 1.877 0.171
Severe PPM vs No Severe PPM LVEF <40% 1.082 (0.904,
1.294)
Severe PPM vs No Severe PPM LVEF
>=40%
1.250 (1.127,
1.385)
Overall BMI by Severe PPM Interaction 1.611 0.204
Severe PPM vs No Severe PPM BMI <30
kg/m2
1.149 (1.031,
1.281)
Severe PPM vs No Severe PPM BMI >=30
kg/m2
1.277 (1.115,
1.464)
Overall Mean AV Gradient by Severe PPM
Interaction
0.681 0.409
Severe PPM vs No Severe PPM AV
Gradient <40 mmHg
1.227 (1.084,
1.387)
Severe PPM vs No Severe PPM AV
Gradient >=40 mmHg
1.147 (1.022,
1.288)
Overall Afib/Flutter by Severe PPM
Interaction
0.000 0.995
Severe PPM vs No Severe PPM with A
Fib/Flutter
1.193 (1.065,
1.337)
27
Subgroup Analyses
Mortality
Effect estimate
(95% CI) Chi-Square
P-value
Severe PPM vs No Severe PPM No
Fib/Flutter
1.193 (1.048,
1.358)
28
Table 4. Prior Studies of PPM After TAVR
Reference N Severe/Moderate
(%)
Follow-
up
(months)
Effects of PPM
Tzikas, 2010
(16)
74 16/23 6 No effect on mortality, functional
status
Ewe, 2011 (17) 165 18 6 Less left ventricular mass regression
and symptomatic improvement
Pibarot, 2014
(10)
2211 28/32 12-24 Reduced survival, less left
ventricular mass regression
Thyregod, 2016
(13)
145 14/36 24 Trends to more major adverse
cardiovascular events,
hospitalizations and reduced
functional status
Zorn, 2016 (11) 389 7/19 12 Reduced survival, less left
ventricular mass regression, higher
acute kidney injury
Miyasaka, 2018
(14)
1558 1/9 12 No effect on mortality