Article Text
PDF
Abstract
Objective There is limited evidence to support decision making on antiplatelet therapy following transcatheter aortic valve implantation (TAVI). Our aim was to assess the efficacy and safety of aspirin-only (ASA) versus dual antiplatelet therapy (DAPT) following TAVI.
Methods We performed a systematic review and pooled analysis of individual patient data from 672 participants comparing single versus DAPT following TAVI. Primary endpoint was defined as the composite of net adverse clinical and cerebral events (NACE) at 1 month, including all-cause mortality, acute coronary syndrome (ACS), stroke, life-threatening and major bleeding.
Results At 30 days a NACE rate of 13% was observed in the ASA-only and in 15% of the DAPT group (OR 0.83, 95% CI 0.48 to 1.43, p=0.50). A tendency towards less life-threatening and major bleeding was observed in patients treated with ASA (OR 0.56, 95% CI 0.28 to 1.11, p=0.09). Also, ASA was not associated with an increased all-cause mortality (OR 0.91, 95% CI 0.36 to 2.27, p=0.83), ACS (OR 0.5, 95% CI 0.05 to 5.51, p=0.57) or stroke (OR 1.21; 95% CI 0.36 to 4.03, p=0.75).
Conclusions No difference in 30-day NACE rate was observed between ASA-only or DAPT following TAVI. Moreover, a trend towards less life-threatening and major bleeding was observed in favour of ASA. Consequently the additive value of clopidogrel warrants further investigation.
Statistics from Altmetric.com
Introduction
Transcatheter aortic valve implantation (TAVI) is an alternative treatment for patients with severe symptomatic aortic stenosis considered to be at high risk for conventional surgical valve replacement. Despite its clinical benefits, TAVI is associated with the risk of both haemorrhagic and ischaemic cerebrovascular events periprocedural and postprocedural. The overall incidence of bleeding following TAVI has been reported in 41% of the patients in the meta-analysis of Généreux et al,1 including 16% life-threatening bleeding events. The incidence of stroke following TAVI averages about 3% (range 0%–6%) and has been reported both early and late postprocedural.2 In order to limit the risk of cerebral embolisation, the current European guidelines recommend a combination of low-dose aspirin (ASA) and a thienopyridine early after TAVI, followed by ASA or a thienopyridine alone. The recently published American Heart Association (AHA) guidelines recommend clopidogrel 75 mg daily for the first 6 months after TAVI in addition to life-long ASA 75–100 mg daily. These recommendations are now based on expert opinion (level of evidence: C). Due to the limited studies and subsequent evidence on antithrombotic therapy following TAVI, a heterogeneity in treatment has been reported among centres.3
Our aim was to systematically review the efficacy and safety of antiplatelet therapy, specifically ASA versus dual antiplatelet therapy (DAPT), in TAVI patients not treated with vitamin K antagonists (VKAs) in a patient level pooled analysis.
Methods
Data sources and study selection
We searched PubMed, Medline, EMBASE and the Current Controlled Trials for English-language studies on the antithrombotic treatment following TAVI published between 1 January 2002 and April 2014. We chose 2002 as the lower boundary of the search given that the first human case description of TAVI was published in that year.
The following key words were used: ‘transcatheter aortic valve implantation’, ‘percutaneous aortic valve’, ‘transfemoral aortic valve implantation’, ‘transapical aortic valve implantation’, ‘percutaneous aortic valve implantation’, ‘endovascular aortic valve implantation’, ‘Core Valve’, ‘Edwards Sapien valve’, ‘JenaValve’, ‘heart valve prosthesis implantation’, ‘transcatheter aortic valve replacement’, ‘trans subclavian aortic valve implantation’, ‘platelet aggregation inhibitors’, ‘antiplatelet’, ‘dual antiplatelet therapy’, ‘clopidogrel’, ‘aspirin’, ‘acetylsalicylic acid’, ‘single antiplatelet therapy’, ‘thienopyridine’. Additionally, we manually searched the abstracts of the following conferences: Transcatheter Cardiovascular Therapeutics, European Society of Cardiology and American Heart Association. Moreover, we reviewed the references of reviews and identified studies for additional articles.
Studies were included if they complied with the following requirements: inclusion of patients with aortic stenosis treated with TAVI, clear description of postprocedural antithrombotic treatment including one group treated with single antiplatelet therapy (SAT) and another treated with DAPT, and minimum follow-up of 1 month. Studies were excluded if only one antithrombotic treatment was considered or when the treatment groups merely included DAPT versus SAT in combination with VKA treatment. Study selection was done by two independent reviewers (MH and RD).
Of the 191 potentially eligible studies, 180 were excluded at the title/abstract level. The remaining 11 studies were subjected to a more detailed review using the predefined inclusion and exclusion criteria. An additional six studies were excluded for the following reasons: DAPT was compared with SAT combined with VKA in TAVI patients with known atrial fibrillation (AF; n=3), ischaemic and bleeding events were only reported in patients treated with DAPT (n=2), bleeding events were the only reported outcome (n=1) and exclusion of 30-day events from the analysis (n=1; see figure 1).
Flow diagram of studies included in this meta-analysis. DAPT, dual antiplatelet therapy; VKA, vitamin K antagonist; SAT, single antiplatelet therapy.
In total, four studies were identified with a total of 640 patients, including two randomised controlled trials (RCTs).4 ,5 The principal investigator of each trial was approached and all consented to collaborate in a patient level pooled analysis. They provided the requested data, and vouched for the correctness of the data. Institution review board (IRB) approval was obtained by three studies and informed consent was obtained in all studies. Poliacikova et al6 did not request IRB approval considering that this was a retrospective analysis of prospective gathered data from the National British Cardiovascular Interventions Society TAVI database for which informed consent was previously obtained. For the current patient pooled analysis we obtained approval from the medical ethical committee of the Academic Medical Center, Amsterdam, The Netherlands.
Study population
We included patient-level data from four studies comparing single versus DAPT following TAVI. In all of these studies, clinical endpoints included all-cause mortality, acute coronary syndrome (ACS), stroke, life-threatening and major bleeding as defined by the Valve Academic Research Consortium (VARC) criteria.7
The recently published SAT-TAVI study is a RCT comparing 81 mg ASA lifelong with DAPT (75 mg clopidogrel and 81 mg ASA) for 6 months in 120 patients.5 Patients randomised to DAPT were provided a preloading dose of 300 mg clopidogrel 1 day preprocedural.
In the RCT by Ussia et al,4 79 patients were randomised to 100 mg ASA lifelong or 3 months DAPT (75 mg clopidogrel and 100 mg ASA). Patients randomised to DAPT were provided a preloading dose of 300 mg clopidogrel 1 day preprocedural.
Durand et al8 compared two antiplatelet regimens in 292 patients enrolled in three centres participating in the FRANCE 2 TAVI registry.8 In two centres the adopted antiplatelet treatment included SAT 1 day preprocedural and the third centre adopted DAPT as the standard treatment. In the DAPT group, a loading dose of 300 mg clopidogrel was provided 1 day preprocedural when treated transfemoral. In the transapical cases no loading dose of clopidogrel was administered but clopidogrel 75 mg was started 1 day postprocedural in combination with ASA 75 mg lifelong.
Lastly, Poliacikova et al6 prospectively compared two antiplatelet regimens in a patient registry treated between December 2007 and June 2012. Before May 2010 the hospital policy included DAPT for 6 months, including preloading with ASA 300 mg and clopidogrel 300 mg 2–6 h preprocedural, followed by 6 months 75 mg of both ASA and clopidogrel. After May 2010, the policy changed to only ASA lifelong. For the current analysis additional data on patients treated between May 2010 and March 2014 were provided (N=343).
Data collection
The data requested for each study patient included the following at baseline: date of inclusion, age, gender, body mass index, diabetes mellitus, AF, renal disease, prior cerebrovascular event, peripheral artery disease, prior myocardial infarction (MI), prior coronary artery bypass grafting, prior percutaneous coronary intervention, left ventricular ejection fraction (LVEF), New York Heart Association functional classification, aortic valve peak gradient (mm Hg), aortic valve area (cm2), European System for Cardiac Operative Risk Evaluation score, Society of Thoracic Surgeons (STS) score (%); antithrombotic treatment: loading dose, maintenance dose, treatment duration; procedural characteristics: date TAVI, approach (transfemoral, transapical, subclavian, transaortic), valve type, valve size, device success and during follow-up: 30-day occurrence of all-cause mortality (all-cause or cardiovascular death), acute coronary syndrome (ACS), cerebrovascular event, cerebrovascular event classification (major/minor or transient ischaemic attack) and bleeding (life-threatening or disabling, major or minor bleeding). Data were checked for consistency against the original publications.
Definition of outcomes
The primary endpoint was defined as the net adverse clinical and cerebral events (NACE) at 1 month, a composite of all-cause mortality, ACS, stroke (major and minor), life-threatening and disabling bleeding or major bleeding as defined by VARC criteria.10 Moreover, the individual endpoints of the composite were compared between patients treated with ASA and DAPT.
Statistical analysis
Continuous variables are presented as either mean±SD or median with IQRs and compared with the unpaired Student's t test if they followed a normal distribution, and with the independent Mann–Whitney U test when they did not. Categorical variables are expressed as frequency (percentages) and were compared using Pearson's χ2 test.
For the non-randomised studies we applied propensity score matching to account for differences in baseline characteristics between patients treated with ASA or DAPT. In each observational study separately, a multivariable logistical regression model was used to estimate propensity scores, with the two groups (ie, DAPT vs ASA) as dependent variable and potential confounders as covariates. In the study by Durand et al,7 we included prior MI, age, peripheral artery disease and LVEF, and in the study by Poliacikova et al, we included New York Heart Association functional classification and valve type in the multivariable logistical regression models.6 For variables STS and EuroSCORE, there were some missing data, but were both not predictive for antiplatelet treatment and therefore not included in the propensity model. The propensity matched cohort was established by matching each individual case treated with DAPT in random order to the patient from the control group with the closest propensity score. The difference between the propensity scores of the patients treated with DAPT and their matched counterparts could be no more than 0.25 times the SD of the propensity for DAPT. The propensity score derivation and matching process were performed in each dataset separately. The degree of balance in covariates between patients with ASA and DAPT after matching was assessed by calculating the standardised difference and assessment of global imbalance.
Matched patients in the two cohorts were pooled with the patient data of the RCTs for further analysis. The OR with 95% CI was estimated for each individual study with logistical regression for the RCTs and the adapted Mantel–Haenszel method for the matched cohorts. Considering that both unmatched and matched studies were included, a pooled OR of NACE and individual endpoints was obtained by adaption of the Mantel–Haenszel method. In this method the unmatched studies and the matched pairs within the matched studies are treated as strata.11 Forest plots for each endpoint were generated by inserting the calculated ORs in Review manager using the generic inverse variance methods.
Heterogeneity between studies was assessed by visually comparing ORs for each study with the overall estimate and by calculating an I2 value for every outcome. A statistical significance for hypothesis testing was set at a p value <0.05. All statistical analyses were performed using SPSS software (V.20.0; SPSS, Chicago, Illinois, USA), R (R Foundation for Statistical Computing, Vienna, Austria) and Review Manager software (V.5.2; The Nordic Cochrane Centre, Copenhagen, Denmark).
Results
A total of 672 TAVI patients treated with ASA (N=415) or DAPT (N=257) were included in the initial pooled database. We excluded patients in the SAT group who were treated with VKA in addition to SAT (N=108) or DAPT(N=2), VKA-only (N=28) and clopidogrel-only(N=25). Table 1 describes the baseline characteristics according to the included studies. After pooling patient data from the four studies, patients were moderately matched between the two treatment groups (table 2). However, after propensity score matching in the two observational cohorts, patients included in the final analysis were well matched (table 3 and see online supplementary table S1). The final pooled database comprised 434 patients, including 235 patients from the matched cohorts and 199 patients from the RCTs.
Description of patient characteristics in individual studies
Baseline patient characteristics pooled non-matched cohorts with RCTs
Baseline patient characteristics pooled matched cohorts with RCTs
NACE was reported in 13% of the patients treated with only ASA and in 15% of the patients treated with DAPT (pooled OR 0.83, 95% CI 0.48 to 1.43, p=0.50; figure 2). A tendency towards less life-threatening and major bleeding was observed in the ASA compared with the DAPT treatment group (pooled OR 0.56, 95% CI 0.28 to 1.11, p=0.09). With regards to the remaining individual endpoints, no significant difference was observed between all-cause mortality (pooled OR 0.91, 95% CI 0.36 to 2.27, p=0.83), ACS (pooled OR 0.5, 95% CI 0.05 to 5.51, p=0.57) and stroke (pooled OR 1.21; 95% CI 0.36 to 4.03, p=0.75).
Clinical endpoints in pooled analysis of matched cohorts and RCTs. ASA, aspirin; DAPT, dual antiplatelet therapy; NACE, net adverse clinical and cerebral events; ACS, acute coronary syndrome.
Furthermore, we assessed the clinical outcomes separately for the pooled RCTs and matched cohort (table 4). All-cause mortality and ACS events were not increased in the ASA treatment group in both the RCTs and the matched cohorts.
Clinical endpoints of the pooled matched cohorts with the RCTs and separately for RCTs and matched cohorts
In addition, the outcomes and OR (95% CI) for each study (RCTs and matched cohorts) are presented in figure 3. The effect of ASA on NACE and life-threatening and major bleeding was similar across studies. Due to the low number of ACS events it was not possible to determine the 95% CI of the OR per study. Consequently, we did not include a forest plot for ACS endpoint and could only estimate the pooled OR.
Relationship of antiplatelet therapy (aspirin vs dual antiplatelet therapy (DAPT)) with outcomes in individual studies. Forest plot for net adverse clinical and cerebral events (NACEs), all-cause mortality, stroke and life-threatening bleeding as outcomes. Log OR, values >1.0 indicate in favour of DAPT. The ORs with its 95% CI for each study are denoted by red squares and black lines. The size of the red square is proportional to the weight assigned to the study in the pooled estimate. The combined OR estimate for each endpoint is represented by a black diamond, where diamond width corresponds to 95% CI bounds. ASA, aspirin; LTB, life-threatening bleeding.
With the exemption of the low heterogeneity observed for NACE (I2=8%), no heterogeneity was observed for any of the individual endpoints among the RCTs and matched cohorts (I2=0%).
Discussion
In the current collaborative patient-level pooled analysis of all available studies comparing two antiplatelet regimes, we found no difference in 30-day NACE between ASA-only or DAPT following TAVI. However, a trend towards less life-threatening and major bleeding was observed in the ASA treatment group.
When considering the individual endpoints, it is noted that the composite is largely dominated by all-cause mortality and life-threatening or major bleeding. All-cause mortality was not significantly different between the two groups, but was less reported in the ASA treatment group. Moreover, no clear benefit was observed of DAPT in comparison with ASA in preventing stroke occurrence 30 days post-TAVI (2.8% vs 2.4%, p=0.75).
In order to determine the most effective antiplatelet strategy for stroke prevention, a clear understanding of the time occurrence and the pathophysiological mechanisms of stroke following TAVI is of importance. Following the procedure, a high-risk period exists within the first 24 h, with 50% of the stroke events reported in this short time window. This early postprocedural stroke (<24 h) is often ischaemic in nature and more likely to result from embolisation of calcified material during valve implantation or dislodgement of atheromatous debris during various catheter and wire manipulations.12 ,13 Also, small cerebral emboli that result in asymptomatic cerebral infarcts have frequently been observed on postprocedural cerebral MRI, with a prevalence ranging from 62% to 93%.14 There is accumulating evidence that these asymptomatic (‘silent’) cerebral infarcts are related to risk of future stroke, dementia, cognitive decline and depression. Currently, there is increased interest in embolic protection devices aimed at preventing these iatrogenic cerebral emboli by filtering cerebral blood flow from the aortic arch during the TAVI procedure. Temporary implantation of such filter-based devices could provide additional protection for early postprocedural stroke.15 ,16
In addition to embolisation, cerebral hypoperfusion resulting from hypotension during rapid right ventricular pacing and during aortic balloon valvuloplasty has also been postulated as a mechanism for periprocedural stroke.
Following the initial 24 h, an increased stroke risk remains for up to 2 months after the procedure with declining incidence and transitioning to a late phase with a similar hazard as patients with aortic valve disease. The aetiology of postprocedural stroke is less well understood and is thought to result from thromboembolism. Proposed mechanisms include a prothrombotic state of the implanted aortic valve prior to complete endothelialisation of the valve prosthesis, thromboembolisms resulting from modifications in blood flow rheology due to paravalvular leakage and native leaflet compression against the aortic wall and new-onset AF.
However, it should be borne in mind that antithrombotic treatment is a double-edged sword with the therapeutic potential of reducing the ischaemic complications and simultaneously increasing the bleeding risk. Patients who are treated with TAVI are elderly patients with frequent comorbidities including anaemia, hypertension, peripheral vascular disease and renal failure, making these patients highly susceptible to bleeding.16–18 In the current pooled analysis, ASA was associated with a tendency to a decreased number of life-threatening and major bleedings. Subsequently, it is conceivable that in an RCT study powered for life-threatening and major bleeding, ASA would be associated with significant less events compared with DAPT.
The rational for DAPT duration was adopted from the notion that the time for endothelialisation and integration process of the valve stent takes about 3 months, when the risk of stroke shows a corresponding reduction. Histopathological studies of the CoreValve explants have demonstrated this process in a porcine pericardium bioprosthesis.3 Subsequently, the Canadian Cardiovascular Society recommends the use of ASA indefinitely and a combination with clopidogrel for 1–3 months. In the recently published AHA/American College of Cardiology guidelines for the management of patients with valvular heart disease, 75 mg clopidogrel daily is recommended for the initial 6 months after TAVI in addition to life-long ASA (75–100 mg daily), level of evidence class C. In the European guidelines a combination of low-dose ASA and a thienopyridine is recommended early after TAVI, followed by ASA or a thienopyridine alone. The dosage and duration of DAPT are not further specified. Limited studies have been performed on this topic and subsequent support for the guidelines on post-TAVI antithrombotic treatment is currently insufficient.
Limitations
Several limitations should be addressed appropriately. NACE was chosen as the primary endpoint for pragmatic purposes of covering both the efficacy and the safety of ASA versus DAPT and increasing statistical efficiency. However, stroke and bleeding are influenced in opposite directions by these treatment arms and may have resulted in the absence of difference observed in the composite endpoint. However, in the current analysis, we have also borne in mind the separate endpoints. Large RCTs powered to address the risk/benefit effect of ASA are needed before any vast conclusions can be made.
Future studies on this topic should include longer follow-up, loading dose, in particular with regards to approach (transfemoral or transapical), and treatment duration. We only address the ischaemic and bleeding endpoint at 30 days, but it seems conceivable that the majority of stroke and life-threatening and major bleeding occur early post-TAVI.
With regards to the bleeding endpoint, we could not provide additional information on the specific type of bleeds since this was not provided for all studies.
Propensity score matching was applied to eliminate differences in observed confounders among patients with and without dual antiplatelet. This statistical method however is unable to account for differences in unobserved confounders. Therefore, we cannot rule out residual confounding. These results are based on limited a number of four studies. We did not adjust for all sources of heterogeneity between studies such as differences related to study sites or design type. However, with the adjusted Mantel–Haenszel method, we did account for the study level and matched pairs within study.
Finally, the current meta-analysis does not address the antithrombotic treatment in patients previously treated with VKA, including AF. In addition to pre-existing AF, reported in >25% of patients undergoing TAVI, new-onset AF following TAVI occurs in 14%–32% of patients and is an independent determinant of stroke following TAVI.12 ,19 The risk/benefit of additional antithrombotic treatment in these patients is challenging and even more limited evidence is available to support decision making in these patients. We would encourage future studies to assess the antithrombotic treatment in TAVI patients with an additional indication for VKA.
Conclusions
In the current first patient-level pooled analysis of all available studies on antiplatelet therapy following TAVI, no difference in 30-day NACE rate was observed between ASA only or DAPT. Moreover, ASA was associated with a tendency to a decreased number of life-threatening and major bleedings. Consequently, the additive value of clopidogrel on top of ASA may be questioned and warrants further investigation.
Key messages
What is already known on this subject?
Following the implementation of transcatheter aortic valve implantation (TAVI), an increasing amount of patients who are considered high risk for conventional surgical valve replacement are provided an alternative by undergoing TAVI. Although survival has greatly been improved in these patients, TAVI is associated with both haemorrhagic and cerebrovascular events periprocedural and postprocedural. Unfortunately limited studies have been performed to support the current guidelines on the optimal antiplatelet treatment following TAVI.
What might this study add?
By performing a collaborative patient-level pooled meta-analysis on all available studies comparing two antiplatelet treatment regimens following TAVI, we provide an overview of the current evidence on this topic. In the current study, no difference in net adverse clinical and cerebral events was observed between patients treated with aspirin (ASA)-only or dual antiplatelet treatment (DAPT), including ASA and clopidogrel. However, ASA was associated with a decreased tendency of life-threatening and major bleeding.
How might this impact on clinical practise?
Based on the results of the current study, the additive value of clopidogrel on top of ASA may be questioned but warrants further investigation. However, a large randomised trial with both ischaemic and bleeding clinical endpoints is awaited regarding the efficacy and safety of ASA-only versus DAPT.
Acknowledgments
We gratefully acknowledge the contribution of Jaqueline Limpens, MSc, a clinical librarian at the Academical Medical Center, University of Amsterdam for providing support in our literature search. Also, we thank Dr Nan van Geloven, a statistician at the Clinical Research Unit of the Academical Medical Center, University of Amsterdam for providing support in the statistical analysis.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Footnotes
Contributors All authors have provided substantial contributions to the conception, acquisition for the work; insights in the drafting/revising critically of the intellectual content of the manuscript; final approval of the version to be published; agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Competing interests None declared.
Patient consent Obtained.
Ethics approval Various IRB institutions considering it is a patient-level pooled meta-analysis.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Considering that this is a patient-level pooled analysis of four different studies, the data of the current collaborative meta-analysis are not available for data sharing.