The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed in conducting this systematic review and meta-analysis [23]. Given that this study incorporated a comprehensive review and analysis of existing literature, ethical clearance was not necessary. The protocol was registered on PROSPERO with ID: CRD420251181199. The authors confirm that all data supporting the findings of this study are accessible within the article and its supplementary materials.

Two independent reviewers (H. Y. G and X. X. S) systematically searched four databases: PubMed (MEDLINE), Embase, Web of Science (WOS), and Cochrane Central Register of Controlled Trials (CENTRAL) until 20th November 2025. The search strategy utilized a blend of controlled vocabulary, such as Medical Subject Headings (MeSH), along with free-text terms. The primary search terms encompassed “intracranial arterial diseases”, “intracranial arteriosclerosis”, “intracranial atherosclerotic stenosis”, “angioplasty”, “stent”, and “randomized controlled trial”. No restrictions on language or publication status were applied. All search strategies, including the complete ones, are made available in Table S1.

The selection criteria for this research adhered to the PICOS framework (population, intervention, comparator, outcome, and study design): (P) Adult patients with ischemic stroke or TIA caused by atherosclerotic stenosis (70% - 99%) of a major intracranial artery (internal carotid artery, the middle cerebral artery, vertebral artery, or basilar artery may be affected. (I) Any form of endovascular intervention (such as balloon angioplasty and/or stent placement) in conjunction with medication therapy; (C) Routine drug therapy. CMT usually comprises antiplatelet medications, statins, and rigorous control of blood pressure, blood sugar levels, and lifestyle factors. (O) Our primary outcomes included any stroke or death, ischemic stroke, death, intracranial hemorrhage (ICH), and transient ischemic attack (TIA) within 30 days. Secondary outcomes encompassed any stroke or death, ischemic stroke, and death occurring between 31 days and 1 year post-randomization. For this landmark analysis, patients who experienced the primary outcome event within the first 30 days were excluded from the denominator for the corresponding secondary outcome because they were no longer at risk. These events, along with ICH and TIA, were also assessed over the entire follow-up period; S, RCTs. The following are exclusion criteria: 1) Non-ICAS, which includes conditions such as aortic dissection, vasculitis, and moyamoya disease. 2) Comprehensive assessments, descriptive analyses, observational investigations, case reports, and commentaries; There were studies that did not have any extractable data available.

Two researchers (H. Y. G and X. X. S) independently reviewed the titles and abstracts of the references in the search results to eliminate obviously irrelevant literature. We thoroughly examined the content of the remaining references; the same two researchers (H. Y. G and X. X. S) independently reviewed the studies, identified which to include, and documented the reasons for excluding those that didn’t meet the criteria. During the screening process, any disagreements were resolved by discussing or consulting with a third researcher named J. T. Z. For the meta-analysis, exclusively publicly published studies were utilized. ENDNOTE 20 was utilized for managing all references.

Two independent researchers extracted data utilizing a pre-designed standardized form. The extracted content encompasses fundamental research data such as the name of the first author, the year of publication, the country associated with the study, and the study design employed. The key attributes to consider when studying research subjects include the sample size, age, gender, and medical history. Intervention and control measures: a tailored treatment plan and an appropriate sample size. Outcome data: the number of primary and secondary outcomes, the total number of participants, and the duration of follow-up.

To evaluate the methodological quality of the RCTs that were included, the Cochrane Risk of Bias Assessment Tool was employed [24]. This instrument evaluates the potential for bias across seven distinct areas: Random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and researchers (implementation bias)There are several factors that can introduce bias into research studies, including the blinding of outcome assessors (also known as measurement bias), incomplete outcome data (which is referred to as follow-up bias), selective reporting (a type of reporting bias), and other potential sources of bias. The risk level for each domain was categorized as “low risk” and “high risk”. The evaluation process was concluded independently by two researchers, utilizing RevMan 5.4 software for data management and visualization. Two researchers (H. Y. G and X. X. S) independently carried out quality control and bias assessment. All conflicts were addressed through discussion with the corresponding author (J. T. Z).

The statistical analyses were conducted utilizing RevMan 5.4 software. For binary variables, effect sizes were calculated using the risk ratio (RR) and its 95% confidence interval (CI). In meta-analyses involving zero-event studies, the Mantel-Haenszel method with a continuity correction of 0.5 was applied to both arms of the affected trials to enable calculation of the risk ratio, as implemented in RevMan 5.4. Sensitivity analyses excluding zero-event studies were also considered to assess the robustness of the findings. To evaluate the presence of heterogeneity across studies, both the P-value from the Q test and the I² statistic were employed. In the event of substantial variability (P < 0.05 or I² > 50%), a random-effects model is employed to consolidate the data. Alternatively, when the fixed-effects model is employed [25]. Whenever data permitted, we conducted subgroup analyses to investigate possible sources of heterogeneity. To explore the potential impact of different endovascular techniques, a pre-specified subgroup analysis was planned to compare studies using stent implantation versus those using submaximal balloon angioplasty (SBA) alone. The interaction test (P for interaction) was used to assess whether the treatment effect differed between these subgroups.

The certainty of the evidence for each primary and secondary outcome was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach [25][26]. The evidence from randomized controlled trials starts as high certainty but can be downgraded based on five domains: risk of bias (study limitations), inconsistency (unexplained heterogeneity), indirectness (population, intervention, comparator, outcome), imprecision (wide confidence intervals or sparse data), and publication bias. The certainty was categorized into four levels: high, moderate, low, and very low. Two review authors (H. Y. G and X. X. S) independently performed the GRADE assessments. Any disagreements were resolved by discussion or consultation with a third author (J. T. Z). The GRADE evidence profile or “Summary of Findings” table was generated using GRADEpro GDT software (https://gradepro.org).

The initial database search produced 452 records. Following the removal of 43 duplicate entries, the titles and abstracts of 423 distinct records underwent screening. After the title and abstract screening process, a total of 400 records were eliminated. The texts of the remaining 23 articles were obtained and evaluated for eligibility. In the end, six RCTs satisfied the inclusion criteria and were incorporated into both the qualitative synthesis and quantitative meta-analysis. The PRISMA flow diagram (Figure 1) provides a summary of the study selection process.

We incorporated a total of six RCTs [17]-[22]. The main features of the study are outlined in Table 1. A total of 1641 individuals with sICAS were included in the study. Of the participants, 817 were allocated to the EVT combined with CMT group, while 824 were assigned to the CMT group. The baseline characteristics are depicted in Table 2.

The methodological quality of the included RCTs was assessed using the Cochrane Risk of Bias tool. As summarized in Figure 2 and Figure 3, all six trials were judged to have a low risk of bias for random sequence generation and allocation concealment. However, due to the nature of the interventions, blinding of participants and personnel was not feasible in any of the included trials, leading to a high risk of performance bias. The risk of detection bias was considered low in most studies as outcome assessors were blinded. Overall, the evidence base is derived from well-conducted RCTs, but the open-label design represents a notable methodological limitation. Figure 2 and Figure 3 present an overview of the bias found in the included studies.

Figure 1. PRISMA flow chart of the screening process.

Table 1. Summary characteristics of the included RCTs [17]-[22].

PTAS: Percutaneous Transluminal Angioplasty and Stenting; SBA: Submaximal Balloon Angioplasty; MT: Medical Therapy; CMT: Conventional Medical Therapy; TIA: Transient Ischemic Attack.

Table 2. Baseline characteristics of the participants.

EVT: Endovascular Treatment; CMT: Conventional Medical Therapy; HTN: Hypertension; HLD: Hyperlipidemia; DM: Diabetes Mellitus; CAD: Coronary Artery Disease; TIA: Transient Ischemic Attack; NR: Not Report.

Figure 2. Risk of bias graph.

Figure 3. Risk of bias summary.

3.4.1. Primary Outcome

Within 30 days: six studies reported the outcome of any stroke or death (Figure 4(a)), with outcome data available for 1540 of the 1641 randomized patients

Figure 4. Primary endpoints within 30 days.

(EVT: 717; CMT: 823). The risk of any stroke or death was significantly higher in the EVT group than in the CMT group (RR = 2.31; 95% CI: 1.57 - 3.39; P < 0.0001), with no heterogeneity observed among studies (P = 0.46; I² = 0%). Five studies reported any ischemic stroke (Figure 4(b)), with outcome data available for 1490 of the 1641 randomized patients. EVT significantly increased the risk of ischemic stroke (RR = 1.97; 95% CI: 1.22 - 3.19; P = 0.006), with no heterogeneity among studies (P = 0.91; I² = 0%). Six studies reported all-cause death (Figure 4(c)), with outcome data available for 1640 of the 1641 randomized patients. The risk of death was significantly higher in the EVT group than in the CMT group (RR = 4.96; 95% CI: 1.28 - 19.21; P = 0.02), with no heterogeneity among studies (P = 0.99; I² = 0%). Six studies reported ICH (Figure 4(d)), with outcome data available for 1640 of the 1641 randomized patients. The risk of ICH was significantly higher in the EVT group than in the CMT group (RR = 10.55; 95% CI: 2.90 - 38.38; P = 0.0004), with no heterogeneity among studies (P = 0.92; I² = 0%). Three studies reported TIA (Figure 4(e)), with a sample size of 538. No significant difference was observed between the two groups (RR = 0.87; 95% CI: 0.33 - 2.27; P = 0.78), and there was no heterogeneity among studies (P = 0.50; I² = 0%). The summary of these results is presented in Table 3.

In an exploratory subgroup analysis stratified by endovascular device type, the increased risk of any stroke or death within 30 days appeared to be consistent across trials using stenting (RR = 2.57; 95% CI: 1.68 - 3.93) and the single trial

Table 3. Overview of primary and secondary outcomes in patients with symptomatic intracranial stenosis treated with EVT vs. CMT.

RR: Risk Ratio; CI: Confidence Interval; EVT: Endovascular Treatment; CMT: Conventional Medical Therapy; ICH: Intracranial Hemorrhage; TIA: Transient Ischemic Attack.

using SBA (RR = 1.55; 95% CI: 0.62 - 3.88), with no significant interaction between subgroups (P for interaction = 0.31). These findings should be interpreted with caution, given the limited number of studies in each subgroup.

3.4.2. Secondary Outcomes

Beyond 30 days through 1 year: Three studies reported any stroke or death between 31 days and 1 year (Figure S1(a)), with a total of 571 patients who were event-free at 30 days included in this landmark analysis. The risk of CMT outcome was higher in the CMT group than in the EVT group (RR = 0.10, 95% CI: 0.03 - 0.37; P = 0.0006), with no heterogeneity among studies (P = 0.47; I² = 0%). Five studies reported any ischemic stroke (Figure S1(b)), with a sample size of 1529. There was no statistically significant difference in this outcome between the two groups (RR = 0.49, 95% CI: 0.21 - 1.11; P = 0.09), with moderate heterogeneity among studies (P = 0.08; I² = 52%). We conducted a subgroup analysis to explore the impact of stent type on the outcome (Figure S3). Five studies reported deaths (Figure S1(c)), with a sample size of 1490. There was no significant difference in the incidence of the outcome event between the two groups (RR = 0.34, 95% CI: 0.08 - 1.40; P = 0.13). There was no heterogeneity among the studies (P = 1.00; I² = 0%).

In the entire follow-up, five studies reported any stroke or death outcome throughout the follow-up endpoint (Figure S2(a)), with a sample size of 970. EVT was significantly associated with an increased risk of any stroke or death (RR = 1.49; 95% CI: 1.12 - 1.99; P = 0.007), with low heterogeneity among studies (P = 0.14; I² = 45%). Five studies reported any ischemic stroke (Figure S2(b)), with a sample size of 839. The risk of CMT was higher in the CMT group than in the EVT group (RR = 0.52; 95% CI: 0.33 - 0.83; P = 0.006), with low heterogeneity among studies (P = 0.24; I² = 30%). Five studies reported deaths (Figure S2(c)), with a sample size of 1452. The results were not significantly different (RR = 1.57; 95% CI: 0.77 - 3.19; P = 0.22), and there was no heterogeneity among the studies (P = 0.61; I² = 0%). Four studies reported ICH (Figure S2(d)), with a sample size of 1379. The ICH risk in the EVT group was significantly higher than that in the CMT group (RR = 5.89; 95% CI: 1.55 - 22.44; P = 0.009), with no heterogeneity among studies (P = 0.68; I² = 0%). Four studies reported TIA (Figure S2(e)), with a sample size of 836. The results showed no significant difference (RR = 0.68; 95% CI: 0.36 - 1.27; P = 0.22), with low heterogeneity among studies (P = 0.14; I² = 46%).

The GRADE assessment demonstrates that EVT is significantly associated with a substantial increase in the risks of death within 30 days, stroke recurrence, ischemic stroke, and intracranial hemorrhage. The certainty of evidence for the 30-day outcomes is mostly high. In contrast, the certainty of evidence for secondary outcomes is generally moderate, and there remains a certain degree of uncertainty regarding the risks of any stroke or death. Although certain outcomes suggest potential trends of benefit or harm, the imprecision and inconsistency of specific results necessitate a cautious interpretation of the long-term effects (Table 4).

Table 4. GRADE summary of findings table: certainty of evidence for primary and secondary outcomes.

CI: Confidence Interval; RR: Risk ratio; RCTs: Randomized Controlled Trials; EVT: Endovascular Treatment; CMT: Conventional Medical Therapy; ICH: Intracranial Hemorrhage; TIA: Transient Ischemic Attack. Explanations: ^aLow number of events (<300 events); ^bA wide confidence interval that does not exclude the appreciable harm/benefit (downgraded for imprecision);^cI² > 50%.