Cardiovascular Revascularization Medicine, 2018-07-01, Volume 19, Issue 5, Pages 545-552, Copyright © 2017 Elsevier Inc. Abstract Objective Proximal embolic protection devices (P-EPD) and distal filters (DF) are used to prevent distal cerebral embolizations during carotid artery stenting (CAS). We compared their comparative effectiveness in regards to prevention of intraprocedural and periprocedural adverse events, including ischemic lesions (ipsilateral and contralateral), stroke, transient ischemic attacks (TIA) and death. We also compared the combination of the two neuroprotection strategies vs. a single strategy in regards to ischemic lesions and stroke. Materials & methods This study was performed according to the PRISMA and MOOSE guidelines and eligible studies were identified through search of PubMed, Scopus and Cochrane Central. A meta-analysis was conducted with the use of a random effects model. The I-square statistic was used to assess for heterogeneity. Results Twenty-nine studies involving 16,307 patients were included. There was a significant reduction in ischemic lesions with the use of P-EPD among observational studies (RR: 0.66 [0.45–0.97]). There were no statistically significant differences for the other outcomes between the two treatment groups. Conclusions There is a number of studies reporting outcomes on the comparison between P-EPD and DF for CAS. P-EDP can reduce distal embolization phenomena resulting into ischemic lesions when compared to DF based on the results from real-world studies. P-EPD was not superior however, in regards to periprocedural stroke, TIA and death. Further studies are anticipated to provide a clear answer to this debate. Highlights Embolic protection devices are used in carotid artery stenting Proximal balloon occlusion and distal filter protection are the most commonly used Proximal protection is associated with less new ischemic lesions Randomized trials are needed to identify the best neuroprotection device 1 Introduction Cerebrovascular accidents (stroke and transient ischemic attack (TIA) are the third leading cause of death in the United States. A large percentage of stroke patients have concurrent extracranial carotid artery disease, which usually manifests as carotid artery stenosis . Open surgical methods such as carotid endarterectomy (CEA) as well as endovascular interventions such as carotid artery angioplasty ± carotid artery stenting (CAS) can be used for the treatment of carotid artery stenosis . CAS is a minimally invasive procedure, avoiding the risks associated with open surgery . However, CAS right now is limited by the increased risk of periprocedural stroke, mainly attributed to distal embolization phenomena . A variety of embolic protection devices (EPD) have been developed to improve the safety of CAS, including distal filters (DF) and proximal protection (P-EPD) devices, notably represented by proximal balloon occlusion (PBO) and flow reversal (FRS) devices . The two neuroprotection strategies (proximal vs. distal) prevent embolizations with different mechanisms . Even if results are inconsistent across the literature, P-EPD seems to be associated with reduced risk for distal debris migration into the anterior cerebral circulation . Recently, a number of small observational studies, investigated the combination of both DF and P-EPD techniques during the same procedure . Most studies assess the effectiveness of EPDs by measuring the incidence of intra- and periprocedural stroke and by detecting silent ischemic lesions with diffuse-weighted magnetic resonance imaging (DW-MRI) and microembolization signals (MES) with transcranial Doppler (TCD). Interestingly, distal embolizations during CAS have the potential to cause cognitive dysfunction. An association between DW-MRI ischemic lesions and cognitive dysfunction was reported by Maggio et al., while Zhou et al. reported that microembolizations during CAS could be associated with memory decline after the procedure. Reaching to conclusions for the optimal neuroprotection strategies is thus very important not only for improving the procedural outcomes but also to improve the long-term outcomes of patients with carotid artery stenosis undergoing endovascular interventions. Our aim in this study was to systematically review the literature for studies comparing proximal vs. distal neuroprotection strategies and compare their effectiveness with a meta-analysis among randomized and real-world studies. 2 Methods This review protocol has been registered in the PROSPERO International Prospective Register of systematic reviews ( http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42017059931 ). This systematic review and meta-analysis was performed according to the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) and the MOOSE (Meta-analyses of Observational Studies in Epidemiology) guidelines . 2.1 Search strategy and selection criteria Systematic literature searches were conducted in PubMed, Scopus and Cochrane Central. The algorithm used for PubMed was the following: (“carotid artery stenting” OR CAS OR “carotid artery stenosis”) AND (“distal embolic protection device” OR filter OR “distal cerebral protection” OR “proximal embolic protection device” or “flow reversal” OR “proximal cerebral protection”). The search was conducted by two independent investigators (AL, PT). Disagreements were resolved by a third investigator (DGK). References lists of the included studies were also manually reviewed in order to identify further eligible articles. A study was considered eligible for this meta-analysis if fulfilled one or more of the predefined inclusion criteria: i) randomized controlled trials (RCTs) or prospective and retrospective observational studies comparing proximal vs. distal embolic protection devices or the combination of these two methods; ii) studies that reported quantitative data on clinical outcomes of interest; iii) studies published in English, up to April 1st, 2017. Studies with high risk of bias or studies reporting on irrelevant outcomes were excluded. 2.2 Data extraction and outcomes Two reviewers, blind to each other (AL, PT), independently extracted the relevant data from the eligible studies. All disagreements were resolved following discussion and final decision was reached by consensus with the addition of a third reviewer (DGK). Data extraction was performed for the following predefined variables: first author and year of publication, study design (RCT vs. observational), inclusion and exclusion criteria, sample size (total and in each individual arm), baseline patient characteristics (age, age > 80, gender, body mass index (BMI), smoking history, chronic obstructive lung disease (COPD), coronary artery disease, diabetes, dyslipidemia, hypertension, target carotid artery stenosis %, stroke history, myocardial infarction (MI) history, stent type, embolic protection type, PBO type (with or without arteriovenous shunt), DF type (eccentric or concentric), timing of imaging (< vs. > 24), sensitivity of imaging (1.5 tesla scanner or 3-tesla scanner), percentage of symptomatic patients, number of events in each group and in total for the outcomes of interest, definitions of the outcomes of interest. The primary outcome was incidence of new ischemic lesions/patient during a CAS procedure. Secondary outcomes included ipsilateral and contralateral ischemic lesions/patient, stroke, TIA, and death. 2.3 Risk of bias assessment Risk of bias was assessed independently by two investigators (AL, PT) with the Cochrane tool for randomized studies and with the Robins tool for non-randomized studies. Discrepancies in quality assessment were resolved via consensus. 2.4 Statistical synthesis and analysis When duplicates were identified, the most recent study was included unless the earliest version reported more relevant outcomes. The random effects model was used to account for heterogeneity among studies. Heterogeneity was assessed with the Higgins I-square. For assessment of publication bias, funnel plots and Egger's regression test were used. A forest plot was used to graphically display the effect size in each study and the pooled estimate. A P value < 0.05 was considered significant. STATA 14.1 (StataCorp, College Station, Texas) was used as statistical software. 3 Results 3.1 Characteristics of the eligible studies Literature search yielded 1391 potentially relevant articles. After screening titles and abstracts, 47 articles were retrieved for full-text evaluation and 29 studies satisfied the predetermined search criteria and were included in this meta-analysis ( Fig. 1 ). From them, 11 were RCTs and 18 were observational studies including in total 16,307 patients. Fig. 1 PRISMA search flow diagram (last search: 04/01/2017). Table 1 Important baseline characteristics of patients enrolled in the included studies. a StudyCountryProximal protection (N)Distal filter (N)Male total, %Age mean ± SDCAD total, %Diabetes total %Dyslipidemia total %Hypertension total %Previous stroke or previous TIA total %Smoking (past or current, or combined) total %Symptomatic patients before the procedure total % Vuruskan 2017 Turkey 40 41 NR NR NR NR NR NR NR NR NR Varbella 2016 Italy 259 35(combo) 75.5 72 ± 7 50.7 23.5 44.5 72.8 15.6 59.2 46.9 Hernandez 2014 Spain 79 208 78 NR 27.9 42.8 60.3 80.8 NR NR 80.8 Yang2013 China 18 14 NR NR NR NR NR NR NR NR NR Schmidt 2004 Germany 21 21 80 NR 52.5 42.8 47.6 90.5 NR NR 30.9 Powell 2006 Lebanon 99 42 84.4 NR 24.8 33.4 71.6 79.4 NR NR NR El-Koussy2007 Switzerland 25 19 70.4 NR NR NR NR NR NR NR 56.8 Flach 2007 Netherlands 10 23 84.8 66 45.5 12.1 66.7 54.5 NR 60.6 NR Iyer 2007 International 192 2812 66.3 NR NR 27.2 65.6 75.2 NR 39.3 43.1 Taha 2009 Japan 15 4 NR NR NR NR NR NR NR NR Brewster 2011 USA 53 216 61.3 NR NR 40.1 83.6 73.9 40.9 42.4 Gupta 2011 USA 5 14 76.2 68 30.1 47.6 76.2 85.7 NR 28.5 NR Plessers 2016 Belgium 16 8 75 67.6 NR 37.5 87.5 58.4 NR NR 50 Aytac 2016 Turkey 19 26 73.3 66.9 ± 10 NR 48.9 NR NR 73.4 66.7 100 Plessers 2015 Belgium 10 10 65 NR NR 10 80 40 NR NR NR Kajihara 2015 Japan 54(combo) 24 NR NR NR NR NR NR NR NR NR Giri 2015 USA 590 9656 62.4 NR NR 38.3 89 91.7 14.9 73.9 NR Bastug 2015 Turkey 45 45 54.4 NR 81.1 57.8 43.3 51.1 27.8 60 63.3 Mokin 2014 USA 70 70 72.8 NR 30 32.8 65 67.1 NR 60 35.7 Akkaya 2014 Turkey 50 50 57 NR NR 59 41 51 NR 59 NR Tatli 2013 Turkey 48 40 68.2 NR 36.4 38.6 NR 68.2 NR 79.5 68.1 Castro 2013 Brazil 21 19 62.5 69.1 ± 8 NR 40 70 97.5 22.5 32.5 82.5 Cano 2013 Brazil 30 30 66.7 NR 70 40 78.3 93.4 25 NR NR Leal 2012 Spain 31 33 90.6 NR NR 45.3 50 68.7 25 37.5 68.7 Bijuklic 2012 Germany 31 31 77.4 NR 56.5 29 83.9 98.4 NR 14.5 40.3 Pavlik 2017 Czech Republic 19 37 NR NR NR NR NR NR NR NR NR Khripun 2016 France 675 109 NR NR NR NR NR NR NR NR NR Pacchioni 2014 Italy 30 30 NR 72 ± 7 NR NR NR NR NR NR NR Caputi 2010 Italy 26 27 79.2 NR 60.4 25.5 94.3 71.7 NR 33.9 11.3 a CAD: coronary artery disease; combo: combination arm; N: number of patients; NR: not reported; TIA: transient ischemic attack. Risk of bias for RCTs is presented separately in Supplementary Table 1 . Overall, twelve observational studies were assessed as having low and six as moderate risk of bias ( Supplementary Table 2 ). Detailed study characteristics are presented in Table 1 . In total, 14 of the included studies were conducted in Europe. The dual neuroprotection strategy was implemented by two studies. The most commonly used stent types in CAS interventions were the PRECISE stent (Cordis, Miami Lakes, FL) and the carotid WALLSTENT (Boston Scientific, Natick, MA). The MO.MA system (Invatec, Roncadelle, Brescia, Italy) for P-EPD was utilized by 15 studies. Among DF devices used, eight studies used the FilterWire EZ (Boston Scientific, Santa Clara, CA) and seven studies used ANGIOGUARD (Cordis, Bridgewater, New Jersey). 3.2 DW-MRI detected ischemic lesions There was no difference in ischemic lesions between the two groups among RCTs (RR:0.75 [0.53–1.06]; p = 0.01; I 2 = 59.5%) ( Fig. 2 ). Among observational studies, P-EDP was associated with a statistically significant decrease in ischemic lesions, with moderate heterogeneity (RR: 0.66 [0.45–0.97]; I 2 = 65%) ( Fig. 2 ). Among studies that separately reported ipsilateral and contralateral new DW-MRI ischemic lesions, no difference was found between the two neuroprotection strategies ( Supplementary Figs. 1 & 2 ). Two studies included an arm with a combination neuroprotection strategy with both P-EPD and DF. We proceeded to a comparison among DF only neuroprotection strategy vs. combination strategy. Patients who were treated with a combination strategy had a lower risk of ischemic lesions (RR: 0.50; 95% CI:0.33–0.76; I 2 = 0%) ( Fig. 3 ). 3.3 Stroke/transient ischemic attack (TIA) There were no differences in regards to stroke rates among observational (RR: 0.77; 95% CI: 0.34–1.78; I 2 = 0) and randomized studies (RR: 0.49; 95% CI: 0.34–1.78; I 2 = 0) ( Fig. 4 ). When we compared the combination of the two neuroprotection strategies vs. a single strategy (either P-EPD or DF), again there was no difference in stroke rates (RR: 1.27; 95% CI: 0.21–7.84; I 2 = 0) ( Fig. 5 ). We also proceeded to a meta-analysis among studies that specifically reported outcomes for TIA rates. The two strategies were similarly effective in preventing TIA (RR: 1.33; 95% CI: 0.61–2.85 among observational studies and RR: 0.63; 95% CI: 0.16–2.46 among RCTs) ( Supplementary Fig. 3 ). 3.4 Death The neuroprotection strategy did not affect the mortality risk among observational studies (RR: 0.75; 95% CI: 0.28–1.99; I 2 = 0%) and RCTs (RR: 1.01; 95% CI: 0.20–5.07; I2 = 0%) ( Fig. 6 ). Fig. 2 This forest plot presents the comparison between P-EPD and DF in terms of new ischemic lesions/patients. We found a significant difference between the two strategies among observational studies. Fig. 3 This forest plot presents the comparison between single and dual protection in terms of new ischemic lesions/patients. We did not find a significant difference between the two strategies. Fig. 4 This forest plot presents the comparison between P-EPD and DF in terms of stroke. We did not find a significant difference between the two strategies. Fig. 5 This forest plot presents the comparison between single and dual protection in terms of stroke. We did not find a significant difference between the two strategies. Fig. 6 This forest plot presents the comparison between P-EPD and DF in terms of death. We did not find a significant difference between the two strategies. 4 Discussion In this systematic review and meta-analysis we compared two different neuroprotection strategies (P-EPD vs. DF) in regards to important intraprocedural or periprocedural endpoints. P-EPD (with PBO or FRS) can minimize the risk of DW-MRI detected ischemic lesions after CAS. Interestingly, the use of a combination neuroprotection strategy decreased the incidence of ischemic lesions compared to DF alone. However, the type of neuroprotection used was not associated with other outcomes, including stroke, TIA or death. The occlusive symptoms and underlying pathology of carotid artery atherosclerosis were primarily described by C Miller Fischer in 1951. CAS is a less invasive treatment modality compared to CEA and is generally preferred in poor surgical candidates . Sardar et al. compared CAS with endarterectomy and found similar rates of ipsilateral periprocedural stroke, whereas CAS was associated with lower rates of periprocedural MI and cranial nerve palsy . However, in the same meta-analysis minor periprocedural stroke rates were higher in CAS group. Considering the increased risk for periprocedural stroke for patients who undergo endovascular interventions for carotid artery stenosis, CAS should be always performed with the adjunctive use of a neuroprotection device to reduce thromboembolic complications. The proximal balloon technique consists of a long sheath catheter with a central working lumen connected to two balloons that are inflated to occlude the common carotid artery (CCA) and external carotid artery (ECA) without crossing the lesion, thus blood flow to the ipsilateral internal carotid artery (ICA) is blocked. Neuroprotection with proximal flow reversal system consists of three main components; a balloon sheath, a balloon wire and an external filter. The balloon sheath has an occlusion balloon that is placed in the distal common carotid artery to transiently cause cessation of antegrade blood flow. The balloon wire is placed in the ECA to prevent collateral ECA to ICA flow. The external filter which is the key component, is placed ex vivo between the sheath and the femoral vein, in order to create an arteriovenous shunt . When the CCA and ECA balloons are inflated, the pressure gradient between femoral artery and vein reverses the normal blood flow and blood goes from ICA through the sheath and external filter .DF's primary difference with the two P-EPD techniques is that DF placement requires crossing of the carotid plaque without having secured first embolic protection. A filter is subsequently deployed in the ICA distal to the lesion, in order to protect the brain from intraprocedural distal debris migration . One of the at least theoretical but important advantages associated with DF compared to P-EPD, is that DF can preserve intraprocedural blood flow within the internal carotid artery (ICA). DW-MRI has emerged as a sensitive tool for the detection of distal embolism during CAS. However, whether or not these detected asymptomatic lesions after CAS are clinically relevant, has yet to be elucidated . Bijuklic et al. reported that asymptomatic distal thromboembolisms during CAS had no prognostic impact in a large real-world study of 837 patients . Although DW-MRI is the technique utilized by the majority of these studies, TCD monitoring to measure MES can be also used to compare the efficacy of different neuroprotection strategies Montorsi et al. measured MES at different time intervals during CAS. They concluded that the majority of MES in the DF group were detected at the exact same time with crossing of the lesion with the DF. The mechanical disruption of the plaque can be a potential explanation of this phenomenon. On the other hand, the highest MES detection time point in the proximal balloon occlusion group was reported during balloon retrieval or deflation. Possible predisposing factors for MES in the proximal balloon group include poor aspiration before the balloon deflation, atherosclerosis of CCA at the site of balloon inflation and plaque debris prolapsing through the stent when blood flow is restored. Our findings for the superiority of P-EPD strategy in regards to lower rates of DW-MRI detected ischemic lesions are also supported by the study from Stabile et al. Previous meta-analyses have failed to prove P-EPD or DF superiority for stroke, TIA and death protection. Despite our robust methodology and the fact that we present the largest sample up to date, we were not able to find a difference between the two strategies for these endpoints. Recently, the combined neuroprotection strategy during CAS has gained interventionalists' attention. We are the first meta-analysis to include the combination arm in our results. Even if our data for the combination strategy is based on low level of evidence studies, our results indicate that the combination strategy is superior to DF alone for ischemic lesions prevention. Whether this superiority is driven by the fact that P-EPD is part of the combination strategy or because of a synergistic effect between the two techniques, remains unclear. We believe that this combined strategy warrants more prospective investigation in order safer conclusions to be reached. 4.1 Study limitations To our knowledge, this meta-analysis includes the largest patient sample published on this topic. However, our results should be interpreted in the context of several limitations. The different definitions used among the included studies, led to heterogeneity in our results. We were unable to adjust for patient specific characteristics, since we did not have access to patient specific data. This was a meta-analysis of 29 studies, including many different centers and operators. In addition, there was heterogeneity among P-EPD groups, since both FRS and PBO were used and devices of different technologies were implemented. Furthermore, DW-MRI was implemented at different time intervals across different centers and that could result in under- or over reporting of new ischemic lesions after CAS. 5 Conclusions This meta-analysis systematically evaluated all the current neuroprotection data and compared P-EPD vs. DF among patient undergoing CAS. Our results show that the two treatment strategies did not differ significantly in stroke, TIA or death rates. Interestingly, P-EPD use was associated with a lower risk of ischemic lesions among observational studies, but not among RCTs. The combination method of both P-EPD and DF can be a promising future approach but the current level of evidence is low. Future RCTs or well-designed prospective real-world studies should compare P-EPD vs. DF vs. the combination method in order to reach safer conclusions for the optimal practice for the interventionalists. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The following are the supplementary data related to this article. Supplementary Fig. 1 This forest plot presents the comparison between P-EPD and DF in terms of new ischemic lesions at the ipsilateral site. We did not find a significant difference between the two strategies. Supplementary Fig. 2 This forest plot presents the comparison between P-EPD and DF in terms of new ischemic lesions at the contralateral site. We did not find a significant difference between the two strategies. Supplementary Fig. 3 This forest plot presents the comparison between P-EPD and DF in terms of new transient ischemic attacks. We did not find a significant difference between the two strategies. Supplementary Table 1 Risk of Bias Assessment for RCTs (Cochrane Tool) Supplementary Table 2 Risk of Bias Assessment for Observational Studies (Robins-I Tool)