Search result and study selection

The initial database search using the aforementioned keywords yielded 3695 articles, which decreased to 2915 articles after removing duplicates. After completing the title and abstract screening, 29 studies were included in the full-text screening. Eight articles were included in this study, with an additional seven identified by screening the citations of the included articles. Finally, 15 RCTs were included in the meta-analysis (Fig. 1).

Fig. 1Study baseline characteristics

Fifteen RCTs with 3897 patients met the inclusion criteria and were included in the meta-analysis. Of the patient cohort, 1939 were allocated to the intervention group (EVT + MT) and 1958 to the control group (MT alone). Thirteen RCTs were multicenter studies, whereas that by Khoury was a single-center study [4]. Table 1, Supplementary Table 2, and Supplementary Table 3 show the demographic characteristics, reported comorbidities, definition of the infarct core for each study, and baseline characteristics for all 16 included RCTs. A summary of the results of risk-of-bias assessment is shown in Supplementary Fig. 1.

Table 1 Baseline patient characteristics of included studiesmRS ordinal shift distribution at 90 days

Thirteen RCTs reported an ordinal shift analysis of the mRS distribution. After pooling the results of the ordinal shift distribution, EVT plus MT was associated with a significant reduction in disability at 90 days over MT alone (OR = 1.91, 95% CI [1.61–2.26]), P < 0.00001, I2 = 54%) (Fig. 2A). In sensitivity analysis, removing the study of Goyal et al. [11] resulted in a significant decrease in heterogeneity (OR = 1.79, 95% CI [1.53–2.10]), P < 0.00001, I2 = 42%) (Supplementary Fig. 2). A sub-group analysis of four RCTs based on age, both with groups patients less than 70 years old and 70 years or older, demonstrated significant difference favoring intervention with homogenous effects (OR = 1.72, 95% CI [1.40–2.11], P < 0.00001, I2 = 0%) and (OR = 1.40, 95% CI [1.11–31.77], P = 0.004, I2 = 0%), respectively (Supplementary Fig. 3).

Fig. 2

Efficacy outcomes. A mRS ordinal shift distribution at 90 days. B mRS 0–1. C mRS 0–2. D mRS 0–3. E ENI. F Barthel Index score

Excellent functional outcome (mRS 0–1) at 90 days

Four RCTs reported excellent functional outcomes between the two groups. EVT plus MT was found to be significantly associated with excellent functional outcome with homogenous effects (OR = 2.37, 95% CI [1.45–3.87], P = 0.0006, I2 = 0%) (Fig. 2B).

Functional independence (mRS 0–2) at 90 days

All RCTs investigated the differences between EVT plus MT and EVT alone. The pooled analysis showed that EVT plus MT was significantly associated with functional independence (OR = 2.19, 95% CI [1.81–2.64], P < 0.00001, I2 = 46%) (Fig. 2C). Visual inspection revealed that the studies were skewed to the right, resulting in an asymmetrical distribution in the funnel plot (Supplementary Fig. 4). Egger’s regression test for funnel plot asymmetry yielded a t-statistic of 2.3625, with a p-value of 0.0359, presenting evidence of asymmetry. Begg’s rank correlation test revealed a z-statistic of 1.92 with a P-value of 0.0554, indicating borderline evidence of funnel plot asymmetry (Supplementary Fig. 5). Infarct core-based analysis was further performed. Patients with both large and small infarct cores treated with EVT plus MT showed significant improvement in functional independence at 90 days (OR = 2.50, 95% CI [1.76–3.54], P < 0.00001, I2 = 26%) and (OR = 2.27, 95% CI [1.67–3.08], P < 0.00001, I2 = 57%), respectively (Fig. 3).

Fig. 3

Forest plot of sub-group analysis based on infarct core size showing the odds of functional independence (mRS 0–2)

Independent ambulation (mRS 0–3) at 90 days

Five RCTs reported differences in independent ambulation at 90 days between the two groups. EVT plus MT was significantly associated with independent ambulation with a homogenous effect (OR = 2.17, 95% CI [1.75–2.69], P < 0.00001, I2 = 6%) (Fig. 2D).

Early neurological improvement (ENI)

A pooled analysis of nine RCTs comparing the two groups showed that EVT plus MT was significantly associated with ENI compared with MT alone (OR = 3.28, 95% CI [2.47–4.34], P < 0.00001, I2 = 39%) (Fig. 2E).

A pooled analysis of three RCTs comparing the two groups showed that EVT plus MT was significantly associated with a Barthel Index score of 95–100 at 90 days compared with MT alone, with a homogeneous effect (OR = 2.53, 95% CI [1.83–3.52], P < 0.00001, I2 = 29%) (Fig. 2F).

Barthel Index score of 95–100 at 90 days

A pooled analysis of three RCTs comparing the two groups showed that EVT plus MT was significantly associated with a Barthel Index score of 95–100 at 90 days compared with MT alone, with a homogeneous effect (OR = 2.53, 95% CI [1.83–3.52], P < 0.00001, I2 = 29%) (Fig. 2F).

Partial/complete recanalization

Six RCTs encompassing 1401 patients reported differences in partial and complete recanalization. EVT plus MT demonstrated statistically significant higher rates in achieving partial/complete recanalization of occluded vessels (OR = 2.18, 95% CI [1.66–2.87], P < 0.00001, I2 = 84%) (Supplementary Fig. 6). In the sensitivity analysis, removing Brekhemer et al. [13] resulted in the most significant change in heterogeneity, which was deemed as moderate heterogeneity (OR = 2.37, 95% CI [1.92–2.93], P < 0.00001, I2 = 54%) (Supplementary Fig. 7). Infarct core-based analysis was further performed. Large infarct core and small infarct core patients who treated with EVT plus MT showed higher rates of partial/complete recanalization (OR = 2.47, 95% [1.99–3.06], P < 0.00001) and (OR = 2.37, 95% [1.76–3.21], P < 0.00001, I2 = 64%), respectively (Fig. 4).

Fig. 4

Forest plot of sub-group analysis based on infarct core size showing the rates of partial/complete recanalization

Any intracerebral hemorrhage

Five RCTs enrolling a total of 1305 patients investigated the incidence of any ICH. The pooled analysis showed a statistically significant effect favoring MT alone (OR = 1.77, 95% CI [1.30–2.42], P = 0.007, I2 = 72%) (Fig. 5A). In the sensitivity analysis, removing Khoury et al. [7] yielded statistically significant and homogenous effect (OR = 2.06, 95% CI [1.46–2.92], P < 0.0001, I2 = 47%) (Supplementary Fig. 8). Infarct core-based analysis was subsequently performed, with analysis showing that patients with both large and small infarct cores had insignificant differences between the two groups, although MT alone has lower rates of any ICH (OR = 2.93, 95% [1.08–7.95], P = 0.03, I2 = 79%) and (OR = 2.27, 95% [0.86–5.98], P = 0.10, I2 = 37%), respectively (Supplementary Fig. 9).

Fig. 5

Safety outcomes. A Any intracerebral hemorrhage. B Symptomatic intracranial hemorrhage (sICH). C Mortality at 90 days. D Early neurological worsening (ENW)

Symptomatic intracranial hemorrhage (sICH)

Fourteen RCTs enrolling 3806 patients assessed the differences in sICH between the two groups. There was no significant difference between the two arms, with a homogeneous effect (OR = 1.31, 95% [0.94–1.83], P = 0.11, I2 = 0%) (Fig. 5B). Infarct core-based analysis was subsequently performed. Analysis of patients with both large and small infarct cores showed insignificant difference between the two groups (OR = 1.67, 95% [0.93–3.00], P = 0.09, I2 = 0%) and (OR = 1.04, 95% [0.59–1.84], P = 0.90, I2 = 0%), respectively (Supplementary Fig. 10).

Mortality at 90 days

All RCTs assessed the association between mortality at 90 days between the two groups. The pooled estimate revealed that there was no significant difference between the two groups (OR = 0.86, 95% [0.74–1.00], P = 0.05, I2 = 23%) (Fig. 5C). Visual inspection revealed that the studies were skewed to the right, showing an asymmetrical distribution in the funnel plot (Supplementary Fig. 11). Egger’s regression test for funnel plot asymmetry yielded a t-statistic of − 0.9419 with a P-value of 0.3634, indicating no significant evidence of asymmetry. Begg’s rank correlation test showed a z-statistic of 0.15 with a P-value of 0.8820, further suggesting no significant evidence of funnel plot asymmetry (Supplementary Fig. 12). Infarct core-based analysis was performed. Large infarct core and small infarct core patients showed insignificant difference between the two groups (OR = 0.85, 95% [0.71–1.01], P = 0.07, I2 = 0%) and (OR = 0.82, 95% [0.57–1.17], P = 0.26, I2 = 53%), respectively (Supplementary Fig. 13).

Early neurological worsening (ENW)

Five RCTs with 1011 patients assessed ENW. The pooled analysis showed insignificant difference between the two groups (OR = 1.07, 95% CI [0.76–1.49], P = 0.71, I2 = 62%) (Fig. 5D). In a sensitivity analysis, the largest change in heterogeneity occurred when the study by Nogueira et al. [16] was removed, resulting in low heterogeneity (OR = 1.40, 95% CI [0.95–2.08], P = 0.09, I2 = 7%) (Supplementary Fig. 14).

Parenchymal hematoma

Four RCTs comprising 1297 patients investigated the differences in parenchymal hematoma type 1 between the two arms. There was no statistically significant difference between the two groups, with a homogeneous effect (OR = 1.03, 95% [0.55–1.93], P = 0.93, I2 = 0%) (Supplementary Fig. 15). Seven RCTs comprising 1789 patients assessed the rates of parenchymal hematoma type 2. There was no statistically significant difference between the two groups, with a homogeneous effect (OR = 1.31, 95% [0.87–1.97], P = 0.20, I2 = 0%) (Supplementary Fig. 16).

Subarachnoid hemorrhage

Four RCTs with a total of 1297 patients assessed the SAH rates of subarachnoid hemorrhage between the two groups. The pooled analysis further demonstrated that MT alone was significantly associated with lower rates of subarachnoid hemorrhage (OR = 3.75, 95% [1.44–9.73], P = 0.007, I2 = 0%) (Supplementary Fig. 17).