Effects of endovascular therapy on acute ischemic stroke:An updated meta-analysis of randomized controlled trials
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.193792
Source of Support: None, Conflict of Interest: None
Objects: To clarify the effects of endovascular therapy (ET) for acute ischemic stroke (AIS) patients, we conducted an updated meta-analysis using data from randomized controlled trials (RCTs).
Keywords: Cerebral infarction; clinical trials; stroke; thrombolytic therapy
Acute ischemic stroke (AIS) has remained a prevalent disease throughout the world over the past few decades, characterized by its high mortality and morbidity rates. The key to managing AIS lies in recanalization of the occluded vessels in time to rescue the ischemic penumbra surrounding the ischemic core. Intravenous thrombolysis (IVT) with tissue plasminogen activator (TPA) has been found to be beneficial in AIS patients. However, a rather narrow thrombolytic time window and low recanalization rates have limited the widespread adoption of IVT to more populations, especially when confronted with larger or distal vessel occlusion. This has kept researchers searching for alternative therapies that benefit more patients. Based on the results of two multicenter randomized controlled trials (RCTs), endovascular therapy (ET) was expected to be a promising remedy that avoids the disadvantages of IVT., However, in 2013, three large-sample multicenter RCTs (IMS3, MR RESCUE, SYNTHESIS EXPANSION) and a preliminary meta-analysis failed to show any superiority of ET over IVT or other standard treatment in terms of improving functional outcome, which placed the actual effectiveness of ET in doubt.,,,
In 2015, researchers were encouraged by the significant efficacy of ET shown in newly reported RCTs (ESCAPE, EXTEND IA, MR CLEAN, SWIFT PRIME, REVASCAT, THERAPY, and THRACE).,,,,,, It seems that ET has now achieved an overwhelming superiority compared to IVT or other standard management. Thus, this paper presents an updated meta-analysis evaluating the overall effects of ET for AIS patients based on the available evidence.
We developed this research in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement. PubMed, EMBASE, and the Cochrane Library databases were systematically searched from inception until April 2015 for English-language articles. The search terms were as follows: “endovascular therapy,” “endovascular treatment,” “thrombectomy,” “intra-arterial,” “stroke,” and “brain infarct” in different combinations. Reference lists of relevant publications were checked manually to supplement the electronic search and collect any important research that may have been missed. Retrieved entries were then gathered and any duplicate items were removed.
Studies were included for further analysis if the following inclusion criteria were met: (1) RCTs on AIS treatment; (2) patients diagnosed with AIS; (3) combining ET with IVT, or ET alone versus IVT or other standard treatment; ET referred to intra-arterial (IA) pharmacologic thrombolysis, mechanical thrombectomy or a combination therapy; and, (4) reported outcome measures included: Modified Rankin Scale (mRS) at 90 days, mortality, and symptomatic intracerebral hemorrhage (sICH) at 90 days.
Studies were excluded if they were: (1) Observational studies or nonrandomized controlled studies or quasi - RCTs; (2) low quality studies; (3) studies that lacked sufficient data on the relevant outcomes; and, (4) repeated studies derived from the same datasets which had multiple publications (the largest trial was included and the rest excluded).
Two researchers (Chao P and Na L) independently screened each retrieved article from the databases by reviewing its title and abstract. Then, potentially eligible studies were further evaluated by reading the full-text. Any disagreements in the process were resolved through discussion or by consulting a third reviewer (Zhou-ping Tang, an experienced expert in neurology) until a consensus was reached.
Quality assessment and data extraction
Articles were critically appraised separately by the two researchers (Chao P and Na L) who referred to the bias assessment tool provided by the Cochrane Handbook for Systematic Reviews of Interventions version 5.1.0. (http://community.cochrane.org/handbook) The checklists included in this tool indicate seven key aspects: (1) Random sequence generation; (2) allocation concealment; (3) blinding of participants and personnel; (4) blinding of outcome assessment; (5) incomplete outcome data; (6) selective reporting; and, (7) other bias. The risk of biases concerning the studies was described as low risk, high risk, or unclear risk for each item. Conflicts were solved by discussion and consultation. Data collection and extraction were independently conducted by the two researchers (Chao P and Na L). Differences in opinions were resolved through discussion and consultation. Extracted data were verified again to ensure transcription accuracy.
We conducted data analysis using Review Manager version 5.1.2 software (http://ims.cochrane.org/revman). Statistical heterogeneities among trials were described using the I2 test and P value. A fixed-effects model was employed in the absence of significant heterogeneity (I2 < 50%); otherwise, a random-effects model was used as an alternative. However, it was not appropriate to carry out an overall meta-analysis when there existed a substantial heterogeneity (I2 > 75%) among the included studies. Sensitivity analysis could then be used to inspect the sources of heterogeneity. A subgroup analysis could be considered to measure different pooled effects. Outcomes of these dichotomous variables were presented as odds ratios (OR), using a 95% confidence interval (95% CI) and P value.
When the number of included studies was >10, a funnel plot was used to explore the publication bias.
The study flow diagram of literature search and selection is presented in [Figure 1]. Briefly, we initially identified a total of 4769 potentially relevant articles after removal of duplicate items from the electronic database. No additional articles were included after the assessment of reference lists. After reading titles and abstracts, 2202 of these initial articles were then excluded for not fulfilling the inclusion criteria. The full-texts of the remaining 159 articles were read and 146 articles were excluded for the following reasons: (1) Single arm trial (n = 23); (2) retrospective study (n = 122); and, (3) lack of relevant data (n = 1). The remaining 13 studies went into the quality assessment stage. All 13 trials were assessed as high quality and, therefore, eligible for further pooling analysis.,,,,,,,,,,,,
Description of studies
All included studies were multicenter enrolled. In total, 3269 patients were included, and the sample size ranged from 16 to 656. Four studies compared ET alone with IVT or standard treatment ,,, while the other nine compared ET plus IVT with IVT alone or standard treatment.S,,,,,, Three studies used urokinase (UK) as a thrombolytic agent when intra-arterial thrombolysis (IAT) was done,, 10 used tissue plasmnogen activator (TPA),,,,,, and 1 allowed either TPA or UK. The enrollment time window ranged from 3 h to 12 h after the symptom onset. Eight studies enrolled patients in a time window of not <6 h after the symptom onset ,,,,,,, and 5 were within <6 h.,,,, The characteristics of included studies are illustrated in [Table 1].
All studies described the randomization procedure and only two trials failed to demonstrate its allocation concealment procedure., Since blinding participants and patients are infeasible in studies involving a surgical procedure, we focused more on the blinding of outcome assessment. Only 1 of the 13 studies did not do this. The risk of bias for the included studies was considered low [Figure 2].
Good functional outcomes (modified Rankin Scale 0–2 at 90 days)
As shown in [Figure 3]a, 13 studies involving 3225 patients reported clinical outcomes measured by mRS at 90 days. Patients with AIS receiving ET were more likely to have a good clinical outcome (mRS scoring 0–2 at 90 days) from random-effects analysis (OR = 1.70; 95% CI, 1.32–2.19; P < 0.0001). However, a high heterogeneity was detected (P = 0.003, I2 = 60%).
Therefore, we performed a sensitivity analysis to explore the potential sources of heterogeneity using the leave-one-out approach, which excludes each study in turn to find out its impact on the overall estimates. Regardless of which study was excluded, a dramatic benefit was observed in favor of ET, with the pooled OR ranging from 1.61 (95% CI, 1.25–2.07, P = 0.0002) to 1.79 (95% CI, 1.38–2.34, P < 0.0001). Notably after excluding the SYNTHESIS Expansion study, the heterogeneity decreased with the I2 dropping significantly from 60% to 42%.
In addition, we performed a subgroup analysis based on the type of intervention, comparing ET alone with IVT or standard treatment, and comparing ET plus IVT with IVT or standard treatment [Figure 3]b. The summary estimate for trials comparing ET alone with IVT or standard treatment was 1.45 (95% CI, 0.84–2.50, P = 0.18). Nine trials compared ET plus IVT with IVT or standard treatment, and the overall estimate was 1.82 (95% CI, 1.38–2.40, P < 0.0001).
[Figure 3]c demonstrates the results of subgroup analysis based on whether patients were enrolled with evidence of baseline angiography. The pooled ORs in the subgroups of evidence of baseline angiography and no evidence of angiography were 1.98 (95% CI, 1.63–2.41; P < 0.00001; I2 = 8%, P = 0.37) and 1.11 (95% CI, 0.72–1.71; P = 0.63; I2 = 54%, P = 0.62) respectively.
Mortality at 90 days
Data on mortality at 90 days were available in seven studies [Figure 4]a. Heterogeneity among studies was low (I2 = 31%, P = 0.19) and a fixed-effects model was used. The overall OR value was 0.64 (95% CI, 0.49–0.83, P = 0.0007) demonstrating that mortality at 90 days was significantly less frequent in patients in the ET group than that of the IVT group.
Symptomatic intracerebral hemorrhage at 90 days
No heterogeneity was found through the I2 test (I2 = 0, P = 0.69). The pooled results from the fixed-effects model are presented in [Figure 4]b and revealed that there was no difference in the rate of sICH between the ET group and the IVT group (OR = 1.15, 95% CI, 0.66–2.01, P = 0.62).
We found little evidence of publication bias. The funnel plot was nearly symmetrical with an even distribution of the included studies [Figure 5].
This meta-analysis was conducted on the basis of comprehensive literature searching. All included RCTs were multicenter enrolled and well-designed. The strength of evidence was consolidated. Although previous meta-analyses have been conducted before on this topic, the numbers of previously included studies were few and most of the included trials involved small sample sizes.,, Accordingly, the results from previous meta-analyses were affected heavily by one certain study.
Our findings were consistent with that of the positive RCTs recently released. We found that: (1) The OR of the good functional outcome (mRS 0–2 at 90 days) was significantly enhanced by ET than that using IVT or other standard treatment; (2) ET was superior in decreasing the 90-day mortality rate; and, (3) the rate of sICH was equally frequent in the two groups. The subgroup analysis revealed that the baseline computed tomography (CT) or magnetic resonance (MR) angiography was necessary for ensuring a valid reperfusion and a better functional outcome. The subgroup analysis also revealed that a fixed time window of 6 h in the patients' enrollment probably excluded some of those patients who could potentially benefit from ET.
Heterogeneity among studies for the primary outcome was obviously affected by the SYNTHESIS EXPANSION study. Patients in this study were not routinely inspected by baseline MR or CT angiography for enrollment. Moreover, no lower limit of National Institutes of Health Stroke Scale (NIHSS) score was imposed in the criteria and the median NIHSS score of enrolled patients (ET group: 13, interquartile range [IQR], 9 to 17; IV TPA group: 13, IQR 9 to 18) were lower than that of other trials. Patients with mild neurological defects were more inclined to have a good functional outcome and the beneficial effects of treatment may not have been observed after 3 months had passed. Patients with posterior circulation stroke were also enrolled in this study. The proportion of posterior circulation patients was 10% in the ET group (18 of 181) and 6% in the control group (11 of 181). Whether the outcomes of these patients differ from patients presenting with anterior circulation occlusion is unclear. Besides, as we know, new generation thrombectomy devices have shown superiority over the early retriever devices resulting in a higher reperfusion and recanalization rates, a lower mortality rate, and better functional results (TREVO 2, SWIFT)., While in this trial, 14 patients receiving ET were treated with the first generation of mechanical thrombectomy devices (Merci or Penumbra) and a good portion of the patients were treated with IA TPA. Solitaire and Penumbra were used in only 23 subjects.
A general anesthetic (GA) was induced in 22 patients among 165 patients (13.3%) receiving ET in the SYNTHESIS EXPANSION study. The outcome of these patients was unclear and whether GA neutralized the beneficial effects remains unknown. Data from some retrospective studies and a meta-analysis revealed that patients receiving ET who were operated by conscious sedation or local anesthesia had preferable outcomes than in those in whom GA was induced.,, However, a fair proportion of the patients who received ET were under GA in several positive studies (EXTEND IA, 36%; MR CLEAN, 37.8%; SWIFT PRIME, 37.0%). It seemed that the beneficial effects of ET might not be neutralized by GA or may only be affected mildly. Comparisons of the effects of anesthesia type on the outcome have not yet been analyzed in the RCTs.
Baseline angiographic confirmation, advanced imaging evaluation, and new generation mechanical thrombectomy devices were most likely to be associated with favorable results. Among positive trials, the result of the EXTEND IA study was the most obviously beneficial. The absolute difference between the groups in the 90 day functionally independent outcome (mRS, 0–2) was 31% in favor of ET with an OR of 3.75 (95% CI, 1.38–10.17). Nearly 71.4% of the patients achieved functional independence at 90 days. This proportion was the highest in the studies with positive results (PROACT2, 40.0%; SYNTHESIS PILOT, 56%; MR CLEAN, 32.6%; ESCAPE, 53.0%; SWIFT PRIME, 60.2%; REVASCAT, 43.7%; THRACE, 54.2%). However, this trial was terminated early when efficacy was achieved, so it was criticized that the effects might have been amplified due to the small sample size. Whether or not this is the case, a few instructive experiences can be learnt from this trial: (1) Rapid management (time to groin puncture after symptom onset: Median, 210 min; interquartile range [IQR] 166 to 251); and, (2) confirmation of baseline angiography was needed to inspect the occlusion vessel, collateral circulation and reversible injured brain tissue to avoid invalid reperfusion. Multi-model imaging also excluded large volume ischemic cores and consequently, no patients suffered from sICH in the ET group. Perfusion CT or MR imaging was useful in identifying patients with salvageable tissue. Taking the SWIFT PRIME study as another example, a subgroup analysis found that patients with a target-mismatch penumbral profile and excellent collateral circulation were more likely to benefit from ET. (3) New generation mechanical thrombectomy devices were widely employed (solitaire flow restoration stent retriever, 77.1%) and a high recanalization rate was observed (86.2%).
RCTs on posterior circulation ischemic stroke were scarce. Although IMS3, SYTHESIS Pilot, and SYNTHESIS EXPANSION studies claimed that patients with acute posterior circulation stroke were included, the detailed information of these patients were not described, or the patients numbers were small. A systematic analysis comparing 344 patients receiving IAT with 76 receiving IVT on basilar artery occlusion found that death or dependency were equal: 78% (59 of 76) versus 76% (260 of 344), respectively (P = 0.82). The recanalization rate was higher in the IAT group (65%; 225 of 344) than in the IVT group (53%; 40 of 76 patients; P = 0.05), but the survival rates of these two groups showed no difference (IVT 50%, 38 of 76; IAT 45%, 154 of 344 patients; P = 0.48).
However, data were included from nonrandomized trials or retrospective studies. The efficacy of ET for acute posterior circulation stroke is not yet well established.
Certain limitations existed. A sampling bias may arise as we imposed an English-language restriction on the literature search. A substantial number of studies published in other languages, such as Chinese, were, therefore, not identified or included. In fact, the overwhelming majority of them may fail to meet the methodological protocal based on the Cochrane review criteria, which would reduce the credibility of evidence.
ET is effective in improving the functional outcome of patients and in decreasing the mortality rates of AIS, while not increasing the risk of sICH.
This research was supported by grants from the National Natural Science Foundation of China (No. 81171089; 81471201) and the Key Clinical Program of the Ministry of Health of China (2010).
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]