Author + information
- Received April 26, 2016
- Accepted May 7, 2016
- Published online August 9, 2016.
- Eric Van Belle, MD, PhDa,
- Christian Hengstenberg, MDb,c,
- Thierry Lefevre, MDd,
- Christian Kupatt, MDe,
- Nicolas Debry, MDa,
- Oliver Husser, MD, PhDb,
- François Pontana, MD, PhDf,
- Grégory Kuchcinski, MDf,
- Efthymios N. Deliargyris, MDg,
- Roxana Mehran, MDh,
- Debra Bernstein, PhDg,
- Prodromos Anthopoulos, MDi,
- George D. Dangas, MD, PhDh,∗ ( )(, )
- for the BRAVO-3 MRI Study Investigators
- aDepartment of Cardiology, Centre Hospitalier Régional Universitaire (CHRU) Lille and Unité Mixte de Recherche (UMR1011), Lille, France
- bDeutsches Herzzentrum München, Technische Universität München, Munich, Germany
- cDeutsches Zentrum für Herz-Kreislauf-Forschung E.V. (DZHK), partner site Munich Heart Alliance, Munich, Germany
- dInstitut Cardio Vasculaire Paris Sud, Paris, France
- eLudwig Maximilian University of Munich (LMU) Munich, Munich, Germany
- fDepartment of Radiology, CHRU Lille, Lille, France
- gThe Medicines Company, Parsippany, New Jersey
- hThe Zena and Michael A. Wiener Cardiovascular Institute, Icahn School of Medicine at Mount Sinai, New York, New York
- iThe Medicines Company, Zurich, Switzerland
- ↵∗Reprint requests and correspondence:
Dr. George D. Dangas, The Zena and Michael A. Wiener Cardiovascular Institute, Icahn School of Medicine at Mount Sinai, One Gustave L. Levy Place, KCC-6th Floor (#82), Box 1030, New York City, New York, 10029.
Background Cerebral embolization is a frequent complication after transcatheter aortic valve replacement (TAVR). We hypothesized that cerebral embolization may be reduced by anticoagulation with bivalirudin during TAVR.
Objectives This study sought to determine the proportion of patients with new cerebral embolus after TAVR and to investigate whether parenteral procedural anticoagulation strategies affect cerebral embolization.
Methods The BRAVO (Effect of Bivalirudin on Aortic Valve Intervention Outcomes)-3 randomized trial compared bivalirudin with unfractionated heparin in patients undergoing transfemoral TAVR. A prospective cerebral magnetic resonance imaging (MRI) substudy was conducted in 4 sites; 60 patients were imaged with brain MRI after TAVR. Primary endpoint was proportion of patients with new cerebral emboli on MRI. Secondary endpoints included quantitative MRI analyses of cerebral lesions and neurological outcomes at 48 h and 30 days.
Results Patients were randomized to bivalirudin (n = 29) versus heparin (n = 31). The proportion of patients with new cerebral emboli on MRI did not differ between bivalirudin and heparin groups (65.5% vs. 58.1%; p = 0.55). Groups were similar for median number of emboli per patient (1 [interquartile range (IQR): 0 to 3] vs. 1 [IQR: 0 to 1]; p = 0.08), total volume of emboli (45 [IQR: 0 to 175] mm3 vs. 33 [IQR: 0 to 133] mm3; p = 0.86), or proportion of patients with a clinical neurological deficit at 48 h or 30 days. All patients who presented clinically with stroke had evidence of new emboli on MRI.
Conclusions This study documented cerebral embolization in nearly two-thirds of patients during contemporary TAVR. There were no significant differences in cerebral embolization for bivalirudin versus heparin anticoagulation during TAVR. (Open-Label, Randomized Trial in Patients Undergoing TAVR to Determine Safety and Efficacy of Bivalrudin vs. UFH [BRAVO-2/3]; NCT01651780)
Transcatheter aortic valve replacement (TAVR) has become a nearly routine treatment option for elderly patients with severe, symptomatic aortic stenosis who are at high surgical risk. Despite the clear clinical benefits of TAVR, its outcomes are limited by complications such as paravalvular aortic regurgitation, vascular complications, bleeding events, and neurological embolizations (1).
Although neurological complications were identified early as a potentially major clinical issue (1), recent data have indicated no increased risk of stroke for patients undergoing TAVR versus a surgical approach (2,3). Systematic magnetic resonance imaging (MRI) studies (4–8) have also shown that new cerebral emboli were observed in most patients within days after TAVR. Such silent events may accelerate cognitive decline over time (9,10). With the possibility of extending TAVR to intermediate-risk patients with aortic stenosis using new TAVR devices (11), the incidence of neurological events must be reduced as much as possible.
Cerebral emboli may originate from several sources, such as material related to the manipulation of the aortic valve, and/or the aortic arch. A recent randomized study in patients undergoing TAVR demonstrated a reduction of cerebral emboli by the use of a protection device (8). Nevertheless, the majority of patients even in the “embolic protected” group still had evidence of new cerebral emboli.
A pharmacologic approach to prevent cerebral emboli during TAVR has not been tested. Previous studies have shown in the majority of cases that cerebral emboli are composed of thrombotic material (12,13). Furthermore, native aortic valve tissue is disrupted at the time of TAVR and contains very high levels of thrombin and tissue factor (14). When performing both balloon valvuloplasty and TAVR, this tissue factor is exposed to the circulation, promoting coagulation. It is therefore important to investigate whether different procedural anticoagulation strategies have an effect on rates of cerebral embolization, and to identify any relationship between “silent emboli” as seen on MRI and clinical stroke (8).
In the BRAVO (Effect of Bivalirudin on Aortic Valve Intervention Outcomes)-3 randomized trial on TAVR procedural pharmacotherapy, bivalirudin was noninferior to heparin in terms of net adverse cardiovascular events within 30 months. There were no significant differences in overall major adverse cardiac/cerebrovascular or bleeding events; only the individual rate of periprocedure myocardial infarction was lower with bivalirudin (15). A pre-specified nested MRI substudy was included in the BRAVO-3 trial (15,16).
BRAVO-3 was an open-label, randomized, controlled trial comparing bivalirudin with unfractionated heparin (UFH) in 802 high-risk or inoperable patients undergoing TAVR in 31 European and North American sites (15). The main exclusion criteria were the presence of a previous mechanical or mitral bioprosthetic valve and severe left ventricular dysfunction with ejection fraction of <15%. The results have been reported previously (15).
The present nested prospective BRAVO-3 MRI study (16) was conducted at 4 European centers of the BRAVO-3 trial. The institutional review board at each site approved the study protocol and activities as part of the BRAVO-3 protocol. The centers were selected because their patients were routinely imaged post-TAVR with diffusion-weighted (DW)-MRI of the brain, and the investigators expressed relevant interest. A total of 178 patients were enrolled in the BRAVO-3 trial in these 4 centers; 69 patients (38.8%) provided additional informed consent to participate in MRI imaging and were included in the substudy at the time of randomization.
DW-MRI was performed after TAVR and before hospital discharge. Cerebral MRI 1.5-T (Philips, Amsterdam, the Netherlands; GE, Little Chalfont, United Kingdom; or Siemens, Berlin, Germany) was performed during the hospital stay after TAVR. Recent ischemic lesions were defined as a hypersignal in DW-MRI sequences. As pre-procedural MRI was not performed, and because pre-existing lesions are rare before TAVR, all post-procedural recent lesions were considered new. The detailed imaging protocol included the following sequences, performed with a total acquisition time of 21.4 s: 1) transversal fluid attenuated inversion recovery, repetition time 6,000 ms, echo time 120 ms, slice thickness 5 mm, matrix 128 × 256, which allowed identification of chronic lesions; 2) transversal apparent diffusion coefficient maps, repetition time 4,800 ms, echo time 100 ms, slice thickness 5 mm, diffusion gradient b values of 0, 500, and 1,000 s/mm2, matrix 128 × 256, which allowed identification of recent ischemic lesions defined as a hypersignal; and 3) T2-weighted turbo spin echo, repetition time 2,921 ms, echo time 78 ms, slice thickness 5 mm, matrix 128 × 256, which allowed the identification of infarct demarcation. Data were evaluated by core lab analysis by 2 independent physicians (1 radiologist [F.P.] and 1 neuroradiologist [G.K.]) blinded to treatment and patient clinical outcomes. Validated qualitative and quantitative methods were applied (Syngo.via version VA30, Siemens). The neurological status of each patient was evaluated in hospital and at 30 days.
Bivalirudin (Angiomax/Angiox, The Medicines Company, Parsippany, New Jersey) was administered in an initial bolus dose (0.75 mg/kg), immediately followed by continuous intravenous infusion (1.75 mg/kg/h in patients with normal renal function). The infusion rate was adjusted to 1.4 or 1.0 mg/kg/h, respectively, in patients with moderate or severe renal impairment (glomerular filtration rate 30 to 59 or <30 ml/min/1.73 m2, respectively). The infusion was continued until the TAVR valve was successfully implanted. The dosing of UFH (repeated weight-adjusted boluses to target the recommended activated clotting time of >250 s), and the decision to reverse its action with protamine at the end of the procedure were done according to local standards. The TAVR procedure was carried out according to the local practices with either a balloon-expandable or a self-expanding heart valve system. After the procedure, patients were recommended to receive aspirin 75 to 100 mg daily and clopidogrel 75 mg daily for a period defined by institutional standard practices.
The primary outcomes of the BRAVO-3 MRI study were the proportion of patients with at least 1 new cerebral embolus diagnosed on DW-MRI and whether different procedural anticoagulation strategies have an effect on rates of cerebral embolization after TAVR. Secondary endpoints were the total volume of new cerebral emboli per patient, the total number of new cerebral emboli per patient, the total number of new cerebral emboli in each cerebral hemisphere and patient, and the neurological outcomes at 48 h and 30 days. The occurrence of major bleeding (defined as Bleeding Academic Research Consortium type ≥3b [overt bleeding plus hemoglobin drop ≥5 g/dl (where hemoglobin drop is related to bleed), cardiac tamponade, bleeding requiring surgical intervention for control (excluding dental/nasal/skin/hemorrhoid), or bleeding requiring intravenous vasoactive drugs] ) in the bivalirudin and heparin groups was also measured.
The assumptions for sample size calculations for the BRAVO-3 MRI study included an estimated rate of patients with at least 1 cerebral embolus of 80% of the control group (4,5) and an estimated relative risk reduction of 50% in the experimental group, 90% power, and an alpha error level of 0.05. These assumptions led to a sample size of 58 patients (29 patients per arm, arcsin approximation). This MRI imaging substudy of BRAVO-3 has a largely exploratory nature, conforming to the general rule of such subsidiary studies of clinical trials.
Continuous variables are reported as mean ± SD or medians with interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. The logistic EuroSCORE (European System for Cardiac Operative Risk Evaluation) was calculated by means of a logistic-regression equation, on a scale from 0% to 100%, with higher scores indicating greater surgical risk and a score of >20% indicating very high risk (18). Event rates were tested using the chi-square test. The primary data analysis was performed according to the intention-to-treat principle. The Kaplan–Meier method was used for the time-to-event analysis based on all available follow-up data and the log-rank test was used for the comparison. Multivariable Cox proportional hazards models, adjusting for baseline covariates, were conducted. Statistical analyses were performed using SAS software (version 9.2, SAS Institute, Cary, North Carolina).
Of the 69 patients of the BRAVO-3 MRI study, 33 were randomized to bivalirudin and 36 to UFH. Post-procedural MRI could not be performed in 9 patients (13%) (Figure 1). DW-MRI was performed 4.2 ± 3.9 days after TAVR.
Most of the baseline characteristics of the patients were well matched between the bivalirudin (n = 29) and heparin (n = 31) groups (Table 1). Overall, the population was representative of a contemporary TAVR population (55.0% female patients, mean age 81 ± 6 years, mean EuroSCORE 13.0 ± 8.0). Seven patients (11.7%) had a history of stroke/transient ischemic attack (including 2 with a modified Rankin score ≥2). History of or documented atrial fibrillation was more frequent in the heparin group (25.8% vs. 6.9%; p = 0.05).
Management of antiplatelet medication did not differ significantly between the treatment groups (Table 1). On admission, 73.3% (n = 44) of the population was treated with at least 1 antiplatelet agent and 20.0% (n = 12) with dual antiplatelet therapy. Most patients (91.7%) did not undergo P2Y12 pre-procedural loading. After the procedure, 98.3% of patients received at least 1 antiplatelet agent; 68.3% received dual antiplatelet therapy.
Management of post-procedural anticoagulant therapy differed between the 2 groups. Mimicking the difference in the atrial fibrillation rate, patients randomized to bivalirudin were less likely to receive anticoagulant therapy after the procedure versus patients randomized to heparin (75.9% vs. 96.8%, p = 0.017) (Table 1).
All TAVR procedures were successful. A balloon-expandable valve device was implanted in 81.7% of patients (Table 1). Balloon post-dilation was performed in 45.0% of patients and an embolic protection device was used in 2 patients (3.3%).
Neurological imaging and outcomes
The proportion of patients with ≥1 new post-procedural cerebral emboli on MRI was 61.7% and did not differ between the bivalirudin and heparin groups (65.5% vs. 58.1%, respectively; p = 0.55) (Table 2). The proportion of patients with a large embolus volume (total volume ≥1,000 mm3) did not differ between groups (10.3% vs. 12.9%, respectively, p = 0.76). The mean number, the total volume, the median and maximal volume of individual emboli per patient, and the proportion of patients with a clinical stroke at 48 h did not differ significantly between groups (Table 2). Cerebral events on MRI in patients with or without a clinical stroke or transient ischemic attack are summarized in Table 3. All patients who presented clinically with stroke showed evidence of new emboli on MRI. The total volume of emboli, the volume of single embolus per patient, and the volume of the largest embolus per patient were higher in patients presenting with versus without stroke at 30 days.
The median number of lesions was 1 (IQR: 0 to 3) in the bivalirudin group and 1 (IQR: 0 to 1) in the UFH group (Table 2). When analyzing embolization per hemisphere, no difference in “anatomical brain distribution” was seen between the 2 groups. The median number of lesions in the right (p = 0.10) or the left hemisphere (p = 0.12) was not statistically different between groups. Figure 2 shows new cerebral lesions on MRI scans in 4 patients.
Subgroup analyses did not demonstrate any significant interaction relating to a differential effect between the 2 study treatments (Central illustration). Predictors of new cerebral lesions are listed in Table 4. No deaths or myocardial infarctions were observed. When compared with UFH, the use of bivalirudin was not associated with significant differences in major bleeding (Bleeding Academic Research Consortium type ≥3b) (relative risk: 0.53; 95% confidence interval [CI]: 0.05 to 5.58; p = 1.00) at 48 h or before discharge, or with net adverse clinical events at 30 days (relative risk: 0.46; 95% CI: 0.13 to 1.61; p = 0.30) (Table 5).
The BRAVO-3 MRI study was the first investigation of the potential benefit of a parenteral intraprocedural pharmacological strategy in reducing risk of cerebral emboli in patients undergoing TAVR. It is also among the largest studies to date to investigate the occurrence of cerebral emboli after TAVR without embolic protection device assessment. The primary results were that cerebral lesions as detected with DW-MRI are frequently observed (61.7% among all patients) after contemporary TAVR, and choice of parenteral anticoagulant agent during TAVR did not significantly affect these rates (Central Illustration).
Similar to our results, intraprocedural intra-aortic embolic protection with a transaortic device in TAVR showed new cerebral lesions in 69% of control subjects versus 58% of patients in the device group (p = 0.70) (19). The incidence of new embolic lesions was higher in other studies: a randomized device trial (8) showed new cerebral lesions in 88.5% of control subjects and in 78.8% of patients in the active (intention-to-treat) arm. A single-center randomized study of a dual-filter cerebral protection system reported ischemic brain lesions in 99% of patients regardless of protection (20).
In general, comparative analysis of the occurrence of new emboli during TAVR is relatively difficult among different studies. Whereas rates of emboli are generally high, substantial variations can be observed (Table 6). Individual study design parameters and MRI imaging/timing may have influenced their results. Studies in which <70% of patients had a post-procedural MRI (i.e., due to >30% dropout rate) had the highest rates of emboli (>80%) (5,7,8), compared with rates of <65% in a study in which >80% of patients underwent MRI (6); in the present study, 87% of patients underwent post-procedural MRI (only 13% dropout rate).
A minority of patients with new embolic lesions after TAVR shows overt neurological deficits (3). However, the incidence of such deficits is variable, due to the clinical endpoint definition used and the lack of standardized post-procedural imaging and routine neurocognitive assessment. For overall embolic events, the total lesion volume is currently considered the most informative brain imaging measure, with excellent intrarater and inter-rater concordance for DW-MRI (21,22). Also, the BRAVO-3 MRI study measured the total volume of emboli (median: 45 mm3), with no significant difference between bivalirudin and heparin groups. The median volume of the largest embolus per patient was 44.5 mm3, which is greater than the 35.5 mm3 documented in the embolic protection arm of a recent randomized study (8).
Regarding clinical neurological deficits, overt stroke is an extreme on a spectrum of possible adverse neuroembolic outcomes after TAVR. Over time, the occurrence of stroke after TAVR has declined to 1.7% to 3.4% (3,11,23). In the present MRI study, the rate of overt stroke in the UFH group was elevated (6.5% at 48 h and 12.9% at 30 days) and may reflect the small sample size, as the overall stroke rates in the main BRAVO-3 study were 2.0% at 48 h and 3.1% at 30 days without any significant difference between bivalirudin versus heparin groups (15).
Importance of cerebral embolization
Although stroke rates after TAVR are low, the high incidence of neuroembolic events remains to be addressed, as the occurrence of “silent” embolic lesions can lead to acceleration of cognitive decline (10,24). It is also very important to clarify the potential “temporal” relationship between the presence of “silent emboli” detected by MRI and the occurrence of clinical strokes and then whether MRI “hits” can be seen as appropriate surrogates of “clinical stroke.” In that context, the results of the present study are important because they suggest that silent embolic lesions are the “hidden iceberg” of clinical stroke. The observation that all 6 patients with a clinical stroke, including the 4 with a post-procedural stroke (between 2 and 30 days), already had new cerebral emboli visible on the MRI is of major clinical importance and should be investigated in larger trials. The parallel decrease in the rate of clinical stroke observed since the early days of TAVR, and the parallel decrease of the frequency of silent cerebral emboli since 2010, is consistent with this hypothesis (Table 6).
A transcranial Doppler study during TAVR demonstrated that most procedural embolic events occur during balloon valvuloplasty, manipulation of catheters across the aortic valve, and valve implantation (25). The mechanical interactions between the transcatheter heart valve, the calcified native valve, and the aortic wall appear to play an important role. Balloon post-dilation of the valve prosthesis for treatment of significant paravalvular leaks, repeated device implantation attempts, and valve prosthesis dislodgment/embolization are also associated with higher rates of neuroembolic events, mainly ≤24 h after TAVR (26). Native aortic valve tissue is disrupted at the time of TAVR. It contains high levels of thrombin and tissue factor, which is exposed to the circulation during the procedure (14). Therefore, the BRAVO-3 MRI study hypothesized that the use of the direct thrombin inhibitor bivalirudin would achieve predictable procedural anticoagulation and would therefore reduce the risk of cerebral emboli during TAVR versus UFH. The main BRAVO-3 trial results indicated a lower periprocedural myocardial infarction rate with bivalirudin versus heparin (15). However, a prothrombotic state may endure after the procedure. Both balloon-expandable and self-expanding transcatheter valves are stent-mounted systems. The metallic struts exposed to the circulation may theoretically trigger platelet activation and initiation of the coagulation cascade (26). This enduring prothrombotic state is not covered by either parenteral procedural anticoagulant and may be a reason for the failure to show a reduction in cerebral emboli on DW-MRI at day 4 in the BRAVO-3 MRI study. Data from the PARTNER (Placement of Aortic Transcatheter Valve Trial) (27) displayed an early peaking high hazard phase for neurological events in the first post-operative week, followed by a constant lower hazard phase throughout follow-up.
Although dual antiplatelet therapy after TAVR is recommended, the duration varies (28), and the dosage and timing of a loading dose of clopidogrel remain unspecified. The optimal antiplatelet therapy, with special attention to the early period after TAVR, needs to be addressed in large randomized, controlled trials (29). This analysis from the BRAVO-3 MRI study and the findings from recent MRI studies (Table 6) demonstrate the importance of dedicated MRI studies with high rates (>80%) of post-procedural MRI. Patients presenting clinically with stroke showed evidence of new emboli on MRI, emphasizing the importance, and the potential prognostic value, of these silent events, which need to be further investigated in dedicated large-scale trials.
Even though the BRAVO-3 MRI study is among the largest post-TAVR MRI studies to date, the analysis is based on a small number of patients and is underpowered for clinical event assessment; consequently, the results should be interpreted with caution. Other limitations include lack of information on the use of post-procedural protamine, and no MRI data at baseline or at 30 days. Although the data come from 4 different centers, the between-center variance component was small enough to be ignored.
The BRAVO-3 MRI study confirms the high rate of clinically silent cerebral embolization after TAVR, as reported in other trials, without demonstrating any benefit regarding procedural anticoagulation with bivalirudin versus heparin in reducing these events. The results introduce a potential link between silent emboli on MRI and post-procedural clinical stroke.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: Cerebral embolism, often clinically silent, is a common complication of TAVR, regardless of anticoagulation with bivalirudin or unfractionated heparin.
TRANSLATIONAL OUTLOOK: Larger prospective studies are needed to compare the safety and efficacy of various antithrombotic drug regimens to reduce the risk of ischemic stroke in patients undergoing TAVR.
Ping Gao (The Medicines Company) contributed to the statistical analysis. Sophie Rushton-Smith, PhD (MedLink Healthcare Communications), provided editorial assistance, which was funded by The Medicines Company.
The BRAVO-3 MRI study is funded by The Medicines Company. Dr. Kupatt has received lecture fees from Edwards Lifesciences. Drs. Deliargyris and Bernstein are employees of The Medicines Company. Dr. Mehran received grants from Eli Lilly, Daichi-Sankyo, Bristol-Myers Squibb, AstraZeneca, and The Medicines Company; has received consulting fees from Janssen Pharmaceuticals, Inc., Osprey Medical Inc., and Watermark Research Partners; and has served on the scientific advisory boards for Abbott Laboratories. Drs. Anthopoulos is a stockholder and employee of The Medicines Company. Dr. Dangas has served on the advisory board of and received grants from The Medicines Company. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Van Belle and Hengstenberg contributed equally to this work. Deepak L. Bhatt, MD, MPH, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- confidence interval
- interquartile range
- magnetic resonance imaging
- transcatheter aortic valve replacement
- unfractionated heparin
- Received April 26, 2016.
- Accepted May 7, 2016.
- 2016 American College of Cardiology Foundation
- Ghanem A.,
- Muller A.,
- Nahle C.P.,
- et al.
- Kahlert P.,
- Knipp S.C.,
- Schlamann M.,
- et al.
- Arnold M.,
- Schulz-Heise S.,
- Achenbach S.,
- et al.
- Van Mieghem N.M.,
- Schipper M.E.,
- Ladich E.,
- et al.
- Van Mieghem N.M.,
- El Faquir N.,
- Rahhab Z.,
- et al.
- Dangas G.D.,
- Lefevre T.,
- Kupatt C.,
- et al.,
- for the BRAVO-3 Investigators
- Mehran R.,
- Rao S.V.,
- Bhatt D.L.,
- et al.
- Haussig S.,
- Mangner N.,
- Dwyer M.G.,
- et al.
- Vogt G.,
- Laage R.,
- Shuaib A.,
- Schneider A.,
- VISTA Collaboration
- Luby M.,
- Bykowski J.L.,
- Schellinger P.D.,
- Merino J.G.,
- Warach S.
- Miller D.C.,
- Blackstone E.H.,
- Mack M.J.,
- et al.,
- for the PARTNER Stroke Substudy Writing Group and Executive Committee
- Rodes-Cabau J.,
- Kahlert P.,
- Neumann F.J.,
- et al.