Endovascular Treatment plus Medical Treatment versus Medical Treatment Alone in Ischemic Stroke: A Systematic Review and Meta-Analysis

Ischemic stroke is associated with atherosclerosis where the brain tissue encounters hypoxia and hypoperfusion leading to neurological damage. Thus, stroke is a high burden devastating disease that increases the possibility of limb paralysis and consciousness disruptions [1, 2].

The initial treatment of large vessel occlusion ischemic stroke begins with the delivery of intra-arterial thrombolytic drugs within 4.5 h from the symptoms onset [3]. Alteplase has the ability to bind with fibrin and convert plasminogen to plasmin; thus, improving thrombolysis and vessel restoration. As a result, blood supply is replenished in brain tissue, preventing necrosis [4, 5]. However, intravenous alteplase displays much less efficacy with recanalization of proximal vessel occlusions [6, 7]; this is evident from previous studies which showed that 60–80% of patients failed to regain functional independence or died within 90 days after the onset of stroke [8, 9].

Endovascular therapy can be divided into multiple approaches; including pharmacologic thrombolysis, guidewire, or microcatheter manipulation of the clot, mechanical and aspiration thrombectomy, and stent-retriever technology [10]. Presently, one of the most used treatments for bridging endovascular occlusion is solitaire stent mechanical thrombectomy. One of the main benefits of this approach is its rapid, simplistic, and high recanalization rate [11]. On the other hand, the main drawback of endovascular therapy is the duration of time before initiation of treatment is increased [12, 13]. Furthermore, it needs specialized intervention teams and centers, due to the difficulty in getting the catheter to the site of occlusion. In addition, other limitations to this approach include arterial wall damage, fragmentation, and embolization of the thrombus [12‒14].

Several studies compared between adding endovascular therapy to medical therapy and medical therapy alone [15]. A previous meta-analysis done in 2015 reviewed these articles; however, the aforementioned meta-analysis included only 8 trials and did not compare the two interventions in terms of safety [15]. This necessitates conducting more updated systematic review and meta-analysis to include the recent trials about the topic to account for the development of newer endovascular treatment methods. The aim of this systematic review and meta-analysis was to compare the efficacy and safety of endovascular and medical therapy against medical therapy alone in treating ischemic stroke patients.

The data that support the findings of this study are available from the corresponding author upon reasonable request. In this meta-analysis, we followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). This study was prospectively registered in PROSPERO (CRD42022345306). The Institutional Review Board (IRB) at our institution reviewed and exempted this study from the need of IRB approval.

The search was conducted on July 10, 2022 and updated on August 10, 2022 by AAT and TNA, independently using the following databases: PubMed, Scopus, CENTRAL, Google Scholar, Clinicaltrials.gov. The following keywords and their related MeSH terms were used: (Endovascular OR Thrombectomy OR Embolectomy OR Aspiration Thrombectomy OR Percutaneous Aspiration Thrombectomy) AND (Stroke OR Apoplexy OR CVA OR Cerebrovascular Accident OR Cerebral Stroke OR Cerebrovascular Accident OR Cerebrovascular Apoplexy OR Cerebrovascular Stroke) AND (Clinical Trial). Afterward, the search results were cross matched and any discrepancy was solved by discussion. The search results, after cross matching, were imported to Rayyan (www.rayyan.ai) and duplicates were removed.

The inclusion criteria of selecting the studies were if they were controlled randomized trials in design and compared between endovascular treatment along with medical treatment and medical treatment alone in terms of disability, revascularization, mortality, dependence, quality of life, symptomatic intracerebral hemorrhage (SICH), asymptomatic intracerebral hemorrhage (ICH), recurrent stroke, and any serious adverse effects. The exposure of interest was using endovascular treatment along with medical treatment and the outcomes of interest were disability, revascularization, mortality, dependence, quality of life, SICH, ICH, recurrent stroke, and any serious adverse effects. Disability was measured using Modified Rankin Scale (mRS) where a score of 0–2 indicated low or no disability and National Institute of Health Stroke Scale (NIHSS) where a score of 15 or less indicated moderate or minimal stroke symptoms. Revascularization was assessed using Thrombolysis and Cerebral Infarction (TICI) scale as TICI 2B and TICI 3 demonstrate at least 50% and complete revascularization, respectively. Mortality was defined as death within 90 days of stroke onset. Dependence was assessed using Barthel Index (BI) with scores higher than 90 indicating slight or no dependence. Quality of life was measured using EuroQoL Scale (EQ5D) with higher scores illustrating better quality of life. SICH, ICH, and recurrent stroke were considered as outcomes due to their importance and significant impact on morbidity and mortality among stroke patients [9]. Serious adverse effects included pneumonia, cardiac events, extracranial hemorrhage, infections, cerebral edema, and vascular dissection and perforation. Additionally, all the mentioned outcomes were assessed as binary variables and continuous variables.

The variables of interest were extracted by AAT and TNA independently and then checked by FHA and LMH and any discrepancy was solved by discussion. After the data were extracted, the quality of the included studies was assessed using the Revised Cochrane risk of bias tool for randomized trials (RoB2) which was also done by AAT and TNA independently and then checked by FHA and LMH and any difference in the scoring was solved by discussion.

After the data were extracted, the odds ratio (OR) and its corresponding 95% Confidence Intervals (95% CI) were calculated for binary outcomes using Altman et al.’s [16] equation and if any zero had been encountered in the outcomes, 0.5 was added to all cells. The OR and its related 95% CIs were used as the effect size for the binary variables. For continuous outcomes, the mean and standard deviation were used as the measure of effect in data analysis. Whenever median and interquartile range were encountered in the extracted data, they were converted to mean and standard deviation using the method described by Hozo et al. [55]. Finally, the mean and standard deviation for the continuous outcomes was used to measure the differences in the outcomes between the intervention groups using the weighted mean difference (WMD) and its related confidence intervals. The analysis was done by creating a model for each outcome by pooling the studies that assessed the same outcomes. Additionally, we performed subanalysis for the studies that did not consider Large Vessel Obstruction (LVO) criteria for patients’ inclusion. LVO was defined as the occlusion of proximal middle cerebral arteries (M1 and M2) and the internal carotid artery [17]. The studies were pooled using the random-effect model when I² was >50% while they were pooled using the fixed effect models when I² was ≤50%. We used Cochran’s Q heterogeneity test and I² statistic to assess statistical heterogeneity. Meta XL, version 5.3 (EpiGear International, Queensland, Australia) was used in the data analysis.

The search yielded 2,570 articles, of which 520 were duplicates. The remaining 2,050 articles were screened using their title/abstract, and 1,931 articles were excluded because they were reviews, cadaveric studies, protocols, editorials, non-interventional design, and wrong population. The remaining 119 studies were tested against the inclusion criteria using their full-text form, and 97 were excluded due to wrong intervention and not including data regarding the outcome of interest. Finally, 22 clinical trials were included in this systematic review and meta-analysis [8, 9, 18‒37]. Online supplementary Figure 1 (for all online suppl. material, see https://doi.org/10.1159/000531285) describes the detailed process of study selection.

The total number of patients was 5,049 from 22 randomized clinical trials. The majority of the included trials were conducted in multiple countries, closely followed by those conducted in the USA. The mean age of the intervention group was 66.22 ± 8.66 whereas that of the control group was 66.41 ± 9.05. The percentage of males in the intervention group was 55.5% whereas the percentage of males in the control group was 55.2%. In addition, 20 trials considered the LVO criteria in patient inclusion while only 2 trials did not consider it. Online supplementary Table 1 shows the characteristics of the included studies. Moreover, 10 trials evaluated the Alberta Stroke Program Early CT Score (ASPECTS) among their participants at baseline. The model that pooled these studies showed that there was no significant difference in the ASPECTS between the intervention and control groups at baseline (online suppl. Fig. 2: WMD = −0.21, 95% CI: −0.57 to 0.15).

The model that investigated the difference between the intervention and the control group in terms of disability as a binary variable provided by mRS included 17 trials. This model showed a significant association between the intervention and mRS score (0–2) (Fig. 1: OR = 1.61; 95% CI: 1.27–2.06); the heterogeneity of this model was significant (p value = 0.00, I² = 77%). The difference in mRS scores as continuous variables was assessed by 9 trials. The model that pooled these trials showed that the intervention group had significantly lower mRS mean compared to the control group (online suppl. Fig. 3; WMD = −0.59; 95% CI: −1.15 to −0.02); the heterogeneity of this model was significant (p value = 0.00, I² = 99%). Furthermore, the model that studied the difference between the intervention and control groups in disability as a binary variable measured by the NIHSS (score of 15 or less) included 6 trials. It revealed a significant association between the intervention and NIHSS score 15 or less (Fig. 2: OR = 2.13; 95% CI: 1.04–4.34); the heterogeneity of this model was significant (p value = 0.00, I² = 77%). On the other hand, the model investigating the difference in NIHSS scores as continuous variables included 5 trials. This model displayed that the intervention group had significantly lower NIHSS mean compared to the control group (online suppl. Fig. 4: WMD = −4.52; 95% CI: −6.32 to −2.72); the heterogeneity was significant (p value = 0.00, I² = 97%).

The difference in revascularization between the intervention and control groups was assessed by a model using the TICI score in 5 trials. It demonstrated a significant association between the intervention and at least 50% to complete revascularization (Fig. 3: OR = 4.9; 95% CI: 2.26–10.61); the heterogeneity of this model was significant (p value = 0.00, I² = 89%). The rate of revascularization in the endovascular treatment group was evaluated in 12 trials and found to be 77% (online suppl. Fig. 5: 95% CI: 0.71–0.83); the heterogeneity of this model was significant (p value = 0.00, I² = 83%).

The model that examined the difference in mortality between the intervention and control groups included 16 trials. Moreover, it showed a significant reduction in mortality in the endovascular treatment group (Fig. 4: OR = 0.79; 95% CI: 0.68–0.92); the heterogeneity of this model was not significant (p value = 0.25, I² = 18%).

The difference in terms of dependence between the intervention and control groups was appraised by the Barthel Index and included 4 trials. A significant association was found between the intervention and BI score (90 or higher) (online suppl. Fig. 6: OR = 2.18; 95% CI: 1.61–2.94); the heterogeneity was not significant (p value = 0.10, I² = 52%).

The model investigating the difference in quality of life provided by EQ5D between the intervention and control groups included 2 trials. No significant association was found between the intervention and better quality of life estimated by higher EQ5D scores (online suppl. Fig. 7: WMD = 0.17, 95% CI: −0.14 to 0.49); the heterogeneity of this model was significant (p value = 0.00, I² = 98%).

Studying the difference between the intervention and control groups was carried out by another model which involved 17 trials. There was no significant association between the intervention and symptomatic intracranial hemorrhage (online suppl. Fig. 8: OR = 0.98, 95% CI: 0.58–1.68); the heterogeneity of this model was significant (p value = 0.00, I² = 85%).

The difference between the intervention and control groups in terms of ICH was assessed by a model which included 6 trials. A significant association was found between the intervention and ICH (online suppl. Fig. 9: OR = 1.69, 95% CI: 1.3–2.21); with insignificant heterogeneity (p value = 0.09, I² = 47%).

The model which examined the difference in recurrent stroke between the intervention and control groups included 7 trials. This model exhibited no significant association between the intervention and stroke occurrence (online suppl. Fig. 10: OR = 0.65, 95% CI: 0.32–1.33); the heterogeneity of this model was significant (p value = 0.01, I² = 67%).

The difference between the intervention and control groups in terms of serious adverse effects was investigated by a model which included 14 trials. No significant association was shown by this model between the intervention and serious adverse effects (online suppl. Fig. 11: OR = 1.11, 95% CI: 0.63–1.97); the heterogeneity of this model was significant (p value = 0.00, I² = 90%).

The model that pooled the studies that considered LVO criteria and assessed disability as a binary variable provided by mRS included 15 studies. This model showed that the intervention was significantly associated with mRS (0–2) (online suppl. Fig. 12; OR = 1.78; 95% CI: 1.38–2.31). Four trials considered LVO criteria and assessed disability as a binary variable provided by NIHSS and the model that included these trials revealed that the intervention was significantly associated with NIHSS score (0–15) (online suppl. Fig. 13; OR = 2.75; 95% CI: 1.47–5.14). Furthermore, all the studies that assessed disability as continuous variable provided by mRS considered LVO criteria. The model investigating mortality among the studies that considered LVO criteria included 14 trials and showed that there was significant association between the intervention and mortality (online suppl. Fig. 14; OR = 0.78; 95% CI: 0.66–0.92). All the studies that assessed revascularization provided by TICI scores, dependence, and quality of life considered the LVO criteria. Moreover, the pooled rate of revascularization among the studies that considered LVO criteria was 78% (online suppl. Fig. 15; 95% CI: 70–84%). Regarding SICH, 15 trials considered LVO criteria and assessed SICH and the model that included these trials showed that there was no significant difference between the intervention group and the control group in SICH occurrence (online suppl. Fig. 16; OR = 0.98; 95% CI: 0.53–1.80). In addition, the model that pooled the trials that considered LVO criteria and assessed asymptomatic ICH included 5 trials. This model showed that the intervention was significantly associated with ICH (online suppl. Fig. 17; OR = 1.88; 95% CI: 1.28–2.76). The model that considered LVO criteria and evaluated stroke recurrence included 5 trials. This model showed that there was no significant association between the intervention and the control in recurrent stroke occurrence (online suppl. Fig. 18; OR = 0.76; 95% CI: 0.31–1.90). Additionally, 12 trials considered LVO criteria and evaluated serious adverse effects. The model that included these studies revealed that there was no significant difference between the intervention group and the control in the occurrence of serious adverse effects (online suppl. Fig. 19: OR = 1.31; 95% CI: 0.79–2.19).

The model that included trials that assessed disability as a binary variable provided by mRS and did not consider LVO criteria included 2 trials. This model showed that there was no significant association between the intervention group and mRS (0–2) (online suppl. Fig. 20; OR = 0.91; 95% CI: 0.60–1.38). Only Ciccone et al. [27] did not consider LVO criteria and assessed disability as a binary variable provided by NIHSS. This study showed that there was no significant association between the intervention and NIHSS score (0–15). None of the trials in this subanalysis group evaluated disability as a continuous variable provided by mRS and NIHSS. Moreover, only 2 trials assessed mortality and did not consider LVO criteria. The model that included these studies showed that there was no significant association between the intervention and mortality (online suppl. Fig. 21; OR = 0.84; 95% CI: 0.59–1.20). None of the trials included in this subanalysis assessed revascularization provided by TICI scores, dependence and quality of life. Only Broderick et al. [8] did not consider LVO criteria and assessed revascularization rate among the intervention group. This study showed that the rate of revascularization was 77%. Two trials evaluated SICH and did not consider LVO criteria. The model that pooled these trials showed that there was no significant difference in the occurrence of SICH (online suppl. Fig. 22; OR = 1.04; 95% CI: 0.61–1.79). Moreover, only Broderick et al. [8] investigated asymptomatic ICH and did not consider LVO criteria. The study showed that there was significant association between the intervention and asymptomatic ICH. In addition, only 2 trials assessed recurrent stroke and did not consider LVO criteria. The model that pooled these studies showed that there was no significant association between the intervention and recurrent stroke (online suppl. Fig. 23; OR = 0.47; 95% CI: 0.15–1.46). The model that assessed studies that did not consider LVO criteria and evaluated serious adverse effects pooled 2 studies. This model showed that there was no significant difference between the intervention and the control in the occurrence of serious adverse effects (online suppl. Fig. 24; OR = 0.54; 95% CI: 0.08–3.50).

According to ROB2, all of the included trials (100.0%) had high overall risk of bias as all of the them had high risk of bias arising from the randomization process (online suppl. Fig. 25). The detailed ROB2 quality assessment of the included trials is illustrated in online supplementary Figure 26. The publication bias plots showed asymmetric funnel plots (online suppl. Fig. 27, 28).

Stroke is a major cause of long-term disability in adults [38] and the second cause of death worldwide [38] with an estimated 30-days mortality of 15% in high-income countries [39]. This systematic review and meta-analysis of randomized clinical trials that included 5,049 patients from 22 randomized clinical trials aimed to compare between endovascular treatment along with medical treatment and medical treatment alone in terms of safety and efficacy.

Our study demonstrated that endovascular therapy had better efficacy than medical treatment alone in the treatment of ischemic strokes. This was evident through a significantly greater proportion of patients who received endovascular treatment, achieving an important outcome of an mRS score ≤2 or an NIHSS score <15. Moreover, the majority of patients who received endovascular treatment had a 4 times higher chance of more than 50% rate of revascularization than patients who received medical treatment alone. These results coincide with findings in the literature that emphasize good functional outcomes and successful revascularization being more frequent in endovascular treatment [15, 34, 40]. This can also be explained by the reperfusion hypothesis which posits that cerebral ischemia is divided into the core which is dead tissue that is unsalvageable and the penumbra which is salvageable tissue that can be restored once the circulation is restored [41]. Early attempts of revascularization and restoration of blood flow result in rescue of the penumbra from the irreversible damage and extension of the core which in return leads to better outcomes [41]. Our results also showed a significant decrease in mortality associated with the endovascular treatment which is consistent with the previous literature [32]. On the other hand, the previous meta-analysis done in 2015 which included only 8 trials did not show any significant improvement in mortality in the endovascular treatment group [15]. The difference between our study and the previous meta-analysis can be explained by the fact that our study included a larger number of patients and clinical trials that used more developed endovascular treatment methods compared to the previous meta-analysis; hence, our results are considered valid. Furthermore, the efficacy of endovascular treatment was demonstrated by a higher rate of independence shown on the Barthel Index. On the other hand, there was no significant improvement in the quality of life associated with adding endovascular treatment to medical therapy. These findings in the context of the demonstrated improvement in terms of disability and functional dependence can be explained by the low number of studies that evaluated quality of life as an outcome, thus the analysis might be underpowered to reveal any significant difference.

Despite the fact that endovascular treatment might offer benefit to patients with ischemic stroke which is concordant with the perfusion theory, the safety was of major concern [15]. The use of microcatheters with or without retriever devices inside the brain vessels might cause several complications to patients with ischemic stroke including: intracerebral hemorrhage, recurrent stroke, arterial dissection, and arterial perforation [42]. Our findings demonstrated that endovascular treatment as an addition to medical therapy was not significantly different in the rate of SICH, recurrent stroke, and serious adverse effects compared to medical therapy alone. However, endovascular treatment addition was significantly associated with a higher rate of ICH. The latest meta-analysis that was done in 2015 did not investigate the safety of endovascular treatment as an addition to medical treatment [15]. On the other hand, a meta-analysis of 5 randomized clinical trials of individualized data showed that endovascular treatment was not associated with higher risk of intracerebral hemorrhage [42]. The difference between symptomatic and asymptomatic ICH is in the occurrence of clinical neurological deterioration [43]. Symptomatic ICH as a complication of stroke was associated with poor prognosis and patient deterioration [44]. Also, this complication was identified to be most probably due to thrombolytic therapy [45]. Since thrombolytic therapy was administered in both study treatment arms, it was expected to be of similar rates in both arms. On the other hand, asymptomatic ICH was hypothesized to be part of the hemorrhagic transformation of stroke which is considered a part of the natural progression of stroke [46]. Hemorrhagic transformation of stroke occurs when peripheral blood extravasates across a disrupted blood-brain barrier into the brain following ischemic stroke [47]. Accordingly, since endovascular treatment involves instrumentation of brain vessels, it is expected to increase the incidence of such complications. Although the impact of asymptomatic ICH on the prognosis is not completely clear, studies were not able to detect any effect of asymptomatic ICH on patient prognosis as of yet [48, 49].

In this study, a subanalysis was conducted based on the type of vessel occluded in an acute ischemic stroke. It was observed that LVO acute ischemic strokes had outcomes that mimicked those of the main results in terms of efficacy (disability, revascularization, mortality, and dependence) and safety (SICH, asymptomatic ICH, recurrent strokes, and serious adverse effects). On the other hand, other vessel occlusions had contradictory results compared to the main analysis in terms of efficacy but similar effects in terms of safety. The difference between LVO and other vessel occlusion in terms of outcomes that we observed in our study was consistent with the existing literature [50, 51]. This may be due to the presence of multiple collaterals and the slow progression of a large vessel stroke allowing an adequate development of collateral circulation to that area before the onset of acute ischemic stroke [52, 53]. Consequently, this can improve the flow to the ischemic areas thus improving the outcome of endovascular treatment in these subgroups.

It is important to mention that time is an important aspect in using endovascular treatment among ischemic stroke patients. Although the time between symptoms onset and management was different across the included studies, a previous meta-analysis investigated the issue and demonstrated that performing the procedure within 7.3 h of the symptom’s onset was associated with improved outcomes [54]. In addition, the study showed that the sooner the reperfusion was achieved, the better the functional outcomes [54] which highlight the importance of patients awareness programs, out-of-hospital care, and in-hospital management to shorten symptom onset-to-treatment times.

This is the most updated and the largest systematic review and meta-analysis that compared endovascular treatment along with medical treatment and medical treatment alone in the management of ischemic stroke in terms of safety and efficacy. In addition, this is the first systematic review and meta-analysis to demonstrate that adding endovascular treatment to medical treatment in the management of stroke does not only improve neurological outcomes but also improve mortality rates. Also, this study was conducted based on the PRISMA and Cochrane Collaboration guidelines. However, this study has several limitations. First of all, we only included studies that were published in the English language which might limit the generalizability of our results. Second, a large proportion of the included studies showed moderate concern of bias according to RoB2. Furthermore, there was high heterogeneity across some of our outcomes which might be attributed to the difference in the retriever devices used, the time to reperfusion, and characteristics of patients between the included trials. The difference in inclusion criteria between the trials was accounted for by conducting subanalysis across all our outcomes while the time to perfusion was already done by several individual data meta-analyses. However, we were not able to conduct subanalysis according to retriever device used due to the fact that each trial used multiple retriever devices and data regarding each retriever device were not available. Finally, the publication bias funnel plot showed asymmetry, suggesting a risk for publication bias.

Our study suggests evidence derived from randomized clinical trials that endovascular treatment combined with medical therapy is superior to medical therapy alone in the management of stroke in terms of mortality, disability, reperfusion rate, and functional dependence with no difference in adverse effects between the 2 interventions except for ICH. However, these findings should be interpreted cautiously due to the high heterogeneity in the analysis. The main analysis findings were reproducible in the subanalysis of the trials that enrolled patients with LVO but not with the trials that did not account for LVO criteria; further trials are needed to assess the safety and efficacy of endovascular treatment among stroke patients without LVO.

The University of Jordan IRB reviewed and exempted this study from the need of IRB approval.

The authors have no conflicts of interest to declare.

The author(s) received no financial support for the research, authorship, and/or publication of this article.

A.A.T. and T.N.A. were involved in conceptualization; A.A.T., L.M.A., F.H.A., and T.N.A. were involved in data curation, formal analysis, investigation, methodology, project administration, resources, software, validation, visualization, and writing the original draft; A.A.T. and T.N.A. were involved in supervision and reviewing and editing the manuscript.

Ahmad A. Toubasi and Thuraya N. Al-Sayegh contributed equally to this manuscript.

All data generated or analyzed during this study are included in this article and its online supplementary material files. Further inquiries can be directed to the corresponding author.