Transcatheter aortic valve implantation (TAVI) is a rapidly evolving safe method with decreasing incidence of perioperative stroke. There is a void in literature concerning the impact of stroke after TAVI in predicting 30-day stroke-related mortality. The primary aim of this meta-analysis was to determine whether perioperative stroke increases risk of stroke-related mortality after TAVI. Online databases, using relevant keywords, and additional related records were searched to retrieve articles involving TAVI and stroke after TAVI. Data were extracted from the finalized studies and analyzed to generate a summary odds ratio (OR) of stroke-related mortality after TAVI. The stroke rate and stroke-related mortality rate in the total patient population were 3.07% (893 of 29,043) and 12.27% (252 of 2,053), respectively. The all-cause mortality rate was 7.07% (2,053 of 29,043). Summary OR of stroke-related mortality after TAVI was estimated to be 6.45 (95% confidence interval 3.90 to 10.66, p <0.0001). Subgroup analyses were performed among age, approach, and valve type. Only 1 subgroup, transapical TAVI, was not significantly associated with stroke-related mortality (OR 1.97, 95% confidence interval, 0.43 to 7.43, p = 0.42). A metaregression was conducted among females, New York Heart Association class III/IV status, previous stroke, valve type, and implantation route. All failed to exhibit any significant associations with the OR. In conclusion, perioperative strokes after TAVI are associated with >6 times greater risk of 30-day stroke-related mortality. Transapical TAVI is not associated with increased stroke-related mortality in patients who suffer from perioperative stroke. Preventative measures need to be taken to alleviate the elevated rates of stroke after TAVI and subsequent direct mortality.
Periprocedural stroke, new neurologic deficit within 30 days after the procedure, has been decreasing significantly after transcatheter aortic valve implantation (TAVI). A recent analysis found that perioperative stroke was similar between transfemoral and transapical approach and different valve designs. A possible intraoperative cause for stroke is commonly from particulate embolization due to manipulation of the catheter in the diseased vasculature. Further hypoperfusion during the valve placement in the setting of emboli might increase the risk of perioperative stroke. Periprocedural stroke after coronary artery bypass grafting and carotid endarterectomy has been associated with increased mortality. Thus, there is a need in the literature to accurately describe the association between perioperative stroke and stroke-related mortality. A clearer determination of the association after TAVI could aid in development of periprocedural risk assessment, capturing periprocedural emboli, detecting periprocedural hypoperfusion with the use of neurophysiological monitoring, and intensive medical management for stroke to reduce the mortality. Our primary aim of this study was to determine whether perioperative stroke is associated with increased mortality during TAVI. Our secondary aim was to investigate whether there is a significant difference in stroke-related mortality with various approaches.
Methods
This systematic review complies with the Preferred Reporting Items for Systematic Reviews and Meta-analyses statement and focuses on perioperative strokes and operative mortality after TAVI. The meta-analysis encompasses the sequential steps listed here and described in detail in the following sections: study search, study screening, data extraction, outcomes, assessment of risk bias, and statistical analysis of data.
Studies were included if they were either randomized controlled trials (RCT) or prospective or retrospective cohort studies that reported data on the operative mortality in patients with and without perioperative strokes after TAVI. The study participants are adult patients of either gender undergoing TAVI. No distinction was made between patients experiencing first-time and repeat procedures.
PubMed, EMBASE, and Web of Science were used as the search databases. Search terms were altered to effectively index each database (i.e., MeSH terms for PubMed) but included all terms relevant to a comprehensive collection of the literature. A complete list of search terms used and the results are provided Supplementary Table 1 . In addition to these databases, 3 other relevant meta-analyses were cross-referenced to ensure comprehensiveness.
Two researchers (R.V.H. and A.M.) completed the study screening, and a third researcher (P.D.T.) settled all disagreements to ensure consistency of the screening algorithm. The inclusion criteria were respectively (1) published in English, (2) inclusive of an abstract, (3) RCT or cohort study design, (4) conducted in patients undergoing TAVI only, (5) inclusive of postoperative neurologic assessment and 30-day follow-up data, (6) total sample size ≥50 patients, and (7) conducted in adults, defined as the pooled population aged ≥18 years. The full texts of potentially relevant articles were then screened for inclusion in the final analysis.
Information pertaining to the study design, the population characteristics, the intraoperative details, and postoperative outcomes was gathered from the chosen studies. In addition, information on TAVI approach was collected for each population if available.
For this study, perioperative strokes were defined as any onset of new neurologic symptoms lasting longer than 24 hours and occurring within 30 days of TAVI. The main outcome of interest was 30-day or in-hospital mortality among patients with perioperative strokes. Within the 30-day mortality, death listed as a direct cause of stroke was considered to be significant.
The risk of bias of the included studies was determined as per the recommendations of the Cochrane Collaboration. Thirty-three cohort designs and case–control studies were quality assessed using the Newcastle–Ottawa Scale (NOS). The quality of these studies was assessed across 3 domains: selection, comparability, and outcome for the 31 cohort studies. In 2 case–control studies, the final domain of quality assessment was exposure as opposed to outcome. According to the NOS, a maximum of 4 stars maybe awarded for selection, 2 stars for comparability, and 3 stars for outcome/exposure. Studies that receive 9 stars are of the highest quality. The quality of RCT was assessed using the Cochrane Risk of Bias assessment tool and conducted using Review Manager 5.3. Only 1 study among those included in the final analysis was an RCT. The results of the quality assessment are presented in the Supplementary Table 2 and Supplementary Figure 1 .
Statistical analysis was executed using version 3.1.2 of the METAFOR package for R. Data extracted from each of the finalized studies were used to calculate the summary odds ratio (OR) for 30-day stroke-related mortality outcomes in patients with perioperative strokes versus patients without perioperative strokes after TAVI. A forest plot of the summary log ORs was constructed along with the calculation of the I 2 statistic to both visualize and quantify heterogeneity among the studies. For the purposes of this study, I 2 values of 25%, 50%, and 75% corresponded to low, medium, and high heterogeneity, respectively. An overarching estimate of OR was calculated after relevant details were pooled from the individual studies using a random-effects model. This type of model was used because this analysis involved a random sampling of patients who underwent TAVI. Heterogeneity was explored using a sensitivity analysis and metaregression. The metaregression was carried out to look for an association between the following covariates and the pooled OR estimates: (1) female gender, (2) New York Heart Association (NYHA) class III/IV heart failure status, (3) history of previous stroke, (4) valve type, and (5) valve implantation route. Subgroup analyses were performed on the basis of the following parameters: (1) mean age <81.37 years, (2) mean age >81.37 years, (3) exclusively transfemoral approach, (4) exclusively transapical approach, (5) exclusively Edwards valve, and (6) exclusively CoreValve. Publication bias was assessed using a funnel plot and the Egger regression test ( Supplementary Figure 2 ).
Results
A search of Web of Science, PubMed, EMBASE, and 3 other relevant meta-analyses resulted in 1,526 studies based on keyword search. After removing 646 duplicates, 880 studies were screened based on their abstracts. This screening process left 92 studies for full-text analysis. Of these 92 articles, 34 qualified for the data extraction and were included in the final analysis ( Figure 1 ). Of the 34 finalized studies, 31 were cohort studies, 2 were case-control studies, and 1 was an RCT.
The combined study cohort comprised 29,043 patients who underwent TAVI. Of this number, 49.34% (14,330 of 29,043) were male patients. The mean age across all studies was 81.37 years. NYHA class III/IV information was available for 27 of the 34 finalized studies, which corresponded to 75.19% (21,835 of 29,043) of patients. In this 75.19% of patients, 82.15% (17,939 of 21,835) were reported to have heart failure systems classifying them within class III or IV ( Table 1 ). Only 11 of the studies provided information on previous stoke before undergoing TAVI, which equated to 5.87% (1,705 of 29,043) of the overall patient population. Of this subset, 13.31% (227 of 1,705) patients were reported to have suffered a previous stroke ( Table 1 ).
| Author, Year | Sample Size | Valve Type | NYHA Class III+IV | %NHYA Class III+IV | Age (mean) | Number Male | Log Euroscore | Previous Stroke |
|---|---|---|---|---|---|---|---|---|
| Alassar, 2013 | 119 | Medtronic CoreValve: 110, Edwards SAPIEN: 9 | 75 | 0.63 | 81.00 | 71 | – | – |
| Al-Attar, 2009 | 50 | Edwards SAPIEN: 50 | 47 | 0.94 | 83.00 | 27 | 28.00 | 11 |
| Amat-Santos, 2013 | 136 | Edward XT: 8, Edward SAPIEN: 128 | – | – | 79.00 | 54 | – | 6 |
| Biner, 2014 | 82 | Medtronic CoreValve | – | – | 83.30 | 35 | 24.00 | – |
| Buellesfeld, 2012 | 207 | Medtronic CoreValve: 180, Edwards SAPIEN 27 | 146 | 0.71 | 82.63 | 80 | 22.77 | – |
| Conradi, 2014 | 50 | Edwards SAPIEN XT | 42 | 0.84 | 78.00 | 27 | 21.00 | 8 |
| Dewey, 2010 | 171 | SAPIEN: 171 | 150 | 0.88 | 83.90 | 84 | 35.80 | 31 |
| Ewe, 2010 | 50 | Edwards SAPIEN | 45 | 0.90 | 79.80 | 28 | 24.00 | – |
| Fearon, 2014 | 1023 | – | 981 | 0.96 | 84.78 | 579 | 24.25 | – |
| Grubitzsch, 2014 | 804 | Freedom SOLO | 405 | 0.50 | 74.90 | 441 | 10.70 | – |
| Hahn, 2014 | 1869 | – | 1781 | 0.95 | 84.43 | 931 | 26.74 | – |
| Himbert, 2009 | 75 | – | 71 | 0.95 | 82.00 | 41 | 26.00 | – |
| Holmes, 2015 | 12182 | – | 9879 | 0.81 | 84.00 | 5866 | – | – |
| Kapadia, 2014 | 255 | Edwards SAPIEN | 238 | 0.93 | 80.71 | 149 | 26.12 | – |
| Kasel, 2014 | 100 | SAPIEN XT | – | – | 80.00 | 41 | 11.50 | 7 |
| Litzler, 2012 | 61 | SAPIEN | 44 | 0.72 | 81.00 | 36 | 27.50 | 6 |
| Ludman, 2015 | 3933 | SAPIEN: 2036, CoreValve: 1897 | – | – | 81.30 | 1883 | – | – |
| Makkar, 2013 | 2554 | Edwards SAPIEN | – | – | 84.46 | 1337 | 26.49 | |
| Mendiz, 2013 | 51 | Medtronic CoreValve | 38 | 0.75 | 79.00 | 18 | 20.00 | 5 |
| Meredith, 2014 | 119 | – | 91 | 0.76 | 84.40 | 52 | 6.90 | – |
| Munoz-Garcia, 2012 | 162 | CoreValve | 117 | 0.72 | 79.00 | 50 | 21.00 | – |
| Nombela-Franco, 2012 | 1061 | SAPEIN: 388, CoreValve: 349 | 886 | 0.84 | 81.00 | 538 | – | – |
| Nuis, 2012 | 214 | CoerValve | 175 | 0.82 | 80.00 | 107 | 13.80 | 49 |
| Rodes-Cabeau, 2010 | 339 | SAPIEN, SAPIEN XT | 308 | 0.91 | 81.00 | 152 | – | – |
| Smith, 2011 | 348 | Edwards SAPIEN | 328 | 0.94 | 83.60 | 201 | 29.30 | – |
| Stahli, 2013 | 350 | Medtronic CoreValve: 189 Edwards SAPIEN: 158, Medtronic Engager: 3 | 253 | 0.72 | 82.40 | 171 | 22.10 | – |
| Tay, 2011 | 253 | – | – | – | 85.00 | 129 | 28.00 | 43 |
| Ussia, 2010 | 110 | CoreValve | 69 | 0.63 | 81.00 | 50 | 26.70 | – |
| van der Boon, 2012 | 230 | – | 184 | 0.80 | 80.20 | 118 | 16.40 | 38 |
| Walther, 2013 | 150 | SAPIEN XT | – | – | 81.60 | 136 | 24.30 | – |
| Wenaweser, 2013 | 389 | Medtronic CoreValve: 224, Edwards SAPIEN: 165 | 255 | 0.66 | 82.50 | 165 | 24.30 | 23 |
| Wendler, 2013 | 120 | SAPIEN XT | 101 | 0.84 | 80.30 | 117 | 23.40 | – |
| Yousef, 2015 | 108 | Edwards SAPIEN: 61, CoreValve: 47 | 74 | 0.69 | 75.50 | 69 | 17.20 | – |
| Zahn, 2013 | 1318 | SAPIEN: 246, CoreValve: 1074 | 1156 | 0.88 | 81.70 | 547 | 19.00 | – |
| Total | 29043 | – | 17939 | 81.37 | 227 |
The 30-day perioperative stroke rate and stroke-related mortality rate for the study cohort were 3.07% (893 of 29,043) and 12.27% (252 of 2,053), respectively. The 30-day all-cause mortality rate was 7.07% (2,053 of 29,043). In patients with perioperative strokes, the 30-day stroke-related postoperative mortality rate was found to be 28.22% (252 of 893). Among patients without perioperative strokes, the 30-day mortality rate was much lower at 6.40% (1,801 of 28,150). The approach of TAVI was also documented in studies where information on the various routes taken was provided ( Table 2 ).
| Author, Year | Sample Size | Transfemoral | %transfemoral | Subclavian | Transapical | %transapical | 30 Day Outcomes | |||
|---|---|---|---|---|---|---|---|---|---|---|
| Stroke | Mortality | |||||||||
| Positive | Negative | With stroke | Without stroke | |||||||
| Alassar, 2013 | 119 | 119 | 100 | 0 | 0 | 0 | 5 | 114 | 0 | 5 |
| Al-Attar, 2009 | 50 | 35 | 70 | 0 | 15 | 30 | 2 | 48 | 0 | 7 |
| Amat-Santos, 2013 | 136 | 0 | 0 | 0 | 135 | 100 | 7 | 129 | 0 | 16 |
| Biner, 2014 | 82 | 82 | 100 | 0 | 0 | 0 | 4 | 78 | 0 | 2 |
| Buellesfeld, 2012 | 207 | 207 | 100 | 0 | 0 | 0 | 8 | 199 | 4 | 10 |
| Conradi, 2014 | 50 | 0 | 0 | 0 | 50 | 100 | 1 | 49 | 0 | 2 |
| Dewey, 2010 | 171 | 136 | 80 | 0 | 35 | 20 | 12 | 159 | 8 | 3 |
| Ewe, 2010 | 50 | 27 | 54 | 0 | 23 | 46 | 3 | 47 | 1 | 4 |
| Fearon, 2014 | 1023 | 1023 | 100 | 0 | 0 | 0 | 47 | 976 | 27 | 17 |
| Grubitzsch, 2014 | 804 | – | – | – | – | – | 15 | 789 | 1 | 13 |
| Hahn, 2014 | 1869 | – | – | – | – | – | 63 | 1806 | 22 | 39 |
| Himbert, 2009 | 75 | 51 | 68 | 0 | 24 | 32 | 3 | 72 | 0 | 8 |
| Holmes, 2015 | 12182 | 6807 | 56 | – | – | – | 298 | 11884 | 67 | 780 |
| Kapadia, 2014 | 255 | 255 | 100 | 0 | 0 | 0 | 4 | 251 | 0 | 1 |
| Kasel, 2014 | 100 | 100 | 100 | 0 | 0 | 0 | 7 | 93 | 0 | 2 |
| Litzler, 2012 | 61 | 0 | 0 | 0 | 61 | 100 | 3 | 58 | 0 | 8 |
| Ludman, 2015 | 3933 | 2967 | 75 | 190 | 761 | 19 | 112 | 3821 | 26 | 224 |
| Makkar, 2013 | 2554 | 1474 | 58 | 0 | 1080 | 42 | 84 | 2470 | 29 | 125 |
| Mendiz, 2013 | 51 | 50 | 98 | – | – | – | 1 | 50 | 0 | 2 |
| Meredith, 2014 | 119 | 119 | 100 | – | – | – | 7 | 112 | 0 | 5 |
| Munoz-Garcia, 2012 | 162 | 150 | 93 | 12 | 0 | 0 | 2 | 160 | 2 | 5 |
| Nombela-Franco, 2012 | 1061 | 726 | 68 | 9 | 332 | 31 | 54 | 1007 | 20 | 72 |
| Nuis, 2012 | 214 | 208 | 97 | 6 | 0 | 0 | 19 | 195 | 3 | 15 |
| Rodes-Cabeau, 2010 | 339 | 167 | 49 | 0 | 177 | 52 | 8 | 331 | 2 | 34 |
| Smith, 2011 | 348 | 244 | 70 | 0 | 104 | 30 | 26 | 322 | 1 | 11 |
| Stahli, 2013 | 350 | 290 | 83 | 0 | 60 | 17 | 10 | 340 | 1 | 31 |
| Tay, 2011 | 253 | 168 | 66 | 0 | 85 | 34 | 23 | 230 | 5 | 18 |
| Ussia, 2010 | 110 | 107 | 97 | 3 | 0 | 0 | 5 | 105 | 2 | 9 |
| van der Boon, 2012 | 230 | 223 | 97 | 7 | 0 | 0 | 18 | 212 | 2 | 18 |
| Walther, 2013 | 150 | 0 | 0 | 0 | 150 | 100 | 4 | 146 | 1 | 12 |
| Wenaweser, 2013 | 389 | 308 | 79 | 5 | 76 | 20 | 12 | 377 | 6 | 11 |
| Wendler, 2013 | 120 | 120 | 100 | 0 | 0 | 0 | 2 | 118 | 1 | 5 |
| Yousef, 2015 | 108 | 90 | 83 | 5 | 8 | 7 | 3 | 105 | 0 | 9 |
| Zahn, 2013 | 1318 | 1160 | 88 | 35 | 113 | 9 | 21 | 1297 | 21 | 278 |
| Total | 29043 | 17413 | 272 | 3289 | 893 | 28150 | 252 | 1801 | ||
The 30-day stroke-related mortality rate was >4 times higher for patients with perioperative stroke than the 30-day mortality rate for patients without perioperative strokes. This corresponded to a summary OR across all finalized studies of 6.45 (95% confidence interval [CI] 3.90 to 10.66, p <0.0001). The individual ORs from each of the included studies ranged from 0.46 to 157.41 ( Table 3 ). The summary and individual ORs were used in the making of forest plot ( Figure 2 ). The weight of each study toward the summary OR was also included.
| Author | Year | Odds Ratio (95 % CI) |
|---|---|---|
| Al-Attar | 2009 | 1.11 (0.05-25.43) |
| Himbert | 2009 | 1.08 (0.05-22.85) |
| Dewey | 2010 | 104.00 (19.83-545.35) |
| Ewe | 2010 | 5.37 (0.39-73.09) |
| Rodes-Cabeau | 2010 | 2.91 (0.56-14.99) |
| Ussia | 2010 | 7.11 (1.05-48.27) |
| Smith | 2011 | 1.13 (0.14-9.12) |
| Tay | 2011 | 3.27 (1.087-9.84) |
| Buellesfeld | 2012 | 18.90 (4.11-86.84) |
| Litzler | 2012 | 0.85 (0.04-17.94) |
| Munoz-Garcia | 2012 | 141.36 (6.04-3309.63) |
| Nombela-Franco | 2012 | 7.64 (4.18-13.95) |
| Nuis | 2012 | 2.25 (0.59-8.60) |
| van der Boon | 2012 | 1.35 (0.29-6.33) |
| Alassar | 2013 | 1.81 (0.09-37.06) |
| Amat-Santos | 2013 | 0.46 (0.02-8.41) |
| Makkar | 2013 | 9.89 (6.09-16.06) |
| Mendiz | 2013 | 6.47 (0.21-202.45) |
| Stahli | 2013 | 1.11 (0.13-9.03) |
| Walther | 2013 | 3.72 (0.36-38.60) |
| Wenaweser | 2013 | 33.27 (9.24-119.76) |
| Wendler | 2013 | 22.60 (1.23-416.11) |
| Zahn | 2013 | 157.41 (9.50-2606.81) |
| Biner | 2014 | 3.40 (0.14-81.9) |
| Conradi | 2014 | 6.33 (0.20-198.33) |
| Fearon | 2014 | 76.15 (35.93-161.42) |
| Grubitzsch | 2014 | 4.26 (0.52-34.87) |
| Hahn | 2014 | 24.31 (13.24-44.63) |
| Kapadia | 2014 | 18.55 (0.66-520.31) |
| Kasel | 2014 | 2.44 (0.11-55.61) |
| Meredith | 2014 | 1.30 (0.06-25.87) |
| Holmes | 2015 | 4.13 (3.11-5.47) |
| Ludman | 2015 | 4.85 (3.07-7.68) |
| Yousef | 2015 | 1.45 (0.07-30.25) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
