This protocol is drafted based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Protocols (PRISMA-P) guidelines (see online Additional file 1. PRISMA-P Checklist) . The registration of this systematic review and meta-analysis is shown on the International Prospective Register of Systematic Reviews 'PROSPERO' database (CRD42020199093). If any modification is made to this protocol, the record will be updated in PROSPERO.
A literature search will be performed by two independent reviewers. We will search the following databases, including MEDLINE, EMBASE, Web of Science, and the Cochrane Library, for eligible studies. We will also search WHO trail register (https://www.who.int/ictrp/en/), for potential studies. The preliminary search terms, including stroke, large-vessel occlusion, acute ischemic stroke, tandem occlusion, internal carotid artery occlusion, and mechanical thrombectomy, will be utilized to capture all studies eligible for inclusion. The search strategy will be presented detailly on supplementary material (see online Additional file 2. Search strategy).
The study selection will strictly follow criteria from the Population, Intervention, Comparison, Outcome (PICO) Model.
Inclusion criteria: aged 18 years or older with AIS with ≥70% stenosis or occlusion of the extracranial internal carotid artery (ICA) and concomitant occlusion of the intracranial ICA and/or proximal middle cerebral artery (MCA). Arterial stenosis or occlusion will be confirmed by imaging examinations such as computed tomographic angiography (CTA) or digital subtraction angiography (DSA). The grade of arterial stenosis will be measured according to NASCET (North American Symptomatic Carotid Endarterectomy) criteria.
The intervention will be anterograde approach which is defined as proximal-to-distal recanalization, with angioplasty and/or stenting of the extracranial lesion first, followed by intracranial thrombectomy [15, 18]. For this technique, a microwire followed by a microcatheter will pass the proximal occlusion at first. If difficult to cross the extracranial occlusion site, intraarterial thrombolysis or balloon dilation will be performed to relieve occlusion and establish a narrow passage so that the microwire or the microcatheter could pass through the occlusion when necessary. Then angioplasty and/or stenting will be performed at the extracranial ICA lesion. The choice of stenting will be depended on the neurointerventionists. After extracranial occlusion is recanalized, intracranial thrombectomy will be conducted in terms of stent retrieval, intra-arterial thrombolysis, angioplasty and aspiration.
The comparator will be retrograde approach, which is defined as distal-to-proximal recanalization, with treatment of the intracranial occlusion with mechanical thrombectomy first, followed by treatment of the extracranial ICA occlusion [15, 18]. Once the proximal occlusion site was crossed with a microwire, intracranial occlusion thrombectomy was applied subsequently. Stent placement or balloon angioplasty of the extracranial ICA lesion was then performed after intracranial thrombectomy.
At least one of the following items should be reported:
- Successful recanalization defined as modified Thrombolysis in Cerebral Infarction scale (mTICI) 2b and 3, determined by post-interventional DSA
- Favorable outcome defined as the mRS (modified Rankin Score) ≤ 2 or equal to pre-stroke score at 3-month follow-up
- Mean times of onset to arrival, onset to puncture, and onset to recanalization
- Intracerebral hemorrhage (ICH) defined as intracranial hemorrhage on imaging and a minimum increase of 4 points on the National Institutes of Health Stroke Scale (NIHSS) within 24 hours post-intervention, in accordance with the second European Australasian Acute Stroke Study classification (ECASS II) 
- Procedure and device related complications: arterial dissection, arterial perforation, stent fracture or stent deployment failure
- Mortality at 3-month follow-up
RCTs and observational studies (cohort studies, case-control studies) will be included. The inclusion of observational studies is to gather sufficient data for outcome evaluation and to minimize the type-II errors that can result from the lack of statistical power found in sole RCTs .
- Patients’ age less than 18 years old
- Patients with intracranial hemorrhage, significant cerebellar mass effect, and acute hydrocephalus on computed tomography or magnetic resonance imaging before stroke.
- No report about aforementioned outcomes or an inability to extract the exact number of complications.
- Unsuitable study types, such as case series report, and studies with unavailable full text.
Data selection and analysis
Selection of studies
Two reviewers (RX and XQ) will screen and select the eligible studies, independently. First, the reviewers will screen titles, keywords, and abstracts among the initial searched studies, and the irrelevant studies will be excluded. We will then acquire the full texts of the rest of studies. Next, the reviewers will check the full-texted studies to evaluate their included eligibility and document the reasons for study exclusion. When data from the same trial are reported in more than one article, we will select the most recent study. A third reviewer (TW) will help to resolve any disagreements between the former two reviewers.
Software of EndNote X9 (Clarivate Analytics, Philadelphia, USA) will be used for the management of included studies. A standardized form will be used for data extraction by two reviewers (YW and YF), independently. The extracted data will include the following information:
- Study characteristics, such as type of study, authors, year of publication, sample size, and number of patients.
- Patient characteristics, such as mean age, gender, medical history, site of occlusion, baseline NIHSS (National Institutes of Health Stroke Scale) score and the numbers of patients treated with IV thrombolysis.
- Intervention characteristics, such as the time of onset to puncture or recanalization, numbers of patients treated with angioplasty or stent alone, numbers of patients treated anterograde or retrograde approach first.
- The aforementioned primary and secondary outcomes.
Discrepancy in data extraction between two reviewers will be resolved by a third reviewer (TW) when it cannot be settled by discussion. For unclear or missing data, we will contact the corresponding authors via email. If no responses after two emails, we will exclude this study for meta-analysis and record this case in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) flow chart .
Risk of bias assessment
Two review authors (KY and YM) will independently assess risk of bias in each included study. Reviewers will use risk of bias 2.0 tool of the Cochrane Collaboration criteria for RCTs and quasi-experimental studies, the Newcastle-Ottawa Scale (NOS, see Additional file 3) for observational studies including cohort study and case control study, and the Methodological Quality and Synthesis of Case Series and Case Reports for case series, respectively [24-26]. The following seven domains will be used as criteria to assess the risk of bias for RCTs and quasi-experimental studies:
- Random sequence generation
- Allocation concealment
- Blinding of participants and personnel
- Blinding of outcomes assessments
- Incomplete outcome data
- Selective outcome report
- Other possible bias
The risk of bias for observational studies will be assessed by three domains including selection, compatibility and exposure. Each domain of included studies will be given a score based on the risk of bias. We will rank the level of risk of bias as low, unclear or high, and provide information from the study report together with a justification for our judgment in the ‘Risk of bias’ tables . If there is opinion disagreement between the two reviewers, initial face-to-face discussion will be performed. Also, further team group discussion will be held if necessary.
Measures of treatment effect and data synthesis
Data analysis for outcomes of antegrade or retrograde approach for the management of tandem lesions in AIS is practical when there are two or more studies accessible for a specific variable. The data analysis will be performed using Stata statistical software (version 15.0, Stata Corp, College station, Texas, USA). Treatment effect will be reported as relative risk (RR) with 95% CIs (confidence intervals) because most of the outcomes are dichotomous data. For continuous variables such as mean time of onset to recanalization, we will use mean differences (MDs) with 95% CIs. A narrative presentation of the study results will be provided under circumstance of lacking data.
Assessment of clinical and methodological heterogeneity
Assessment of heterogeneity
Heterogeneity will be measured with the I2 statistic before any outcome is pooled. The heterogeneity of mild (<40%), moderate (40–60%), or substantial (>60%) will be graded depending on the pooled results. On condition that the results with substantial heterogeneity and sufficient number of included trials, we will use subgroup analysis to examine the reasonable origins of heterogeneity, which may include different study locations, patient characteristics, and endovascular treatments. Sensitivity analysis is also utilized to appreciate studies with high risk of bias through step-wise exclusion of studies and observation of combined bias in remaining studies.
Assessment of reporting biases
We will conduct a thorough protocol review of the included studies to evaluate reporting biases. Funnel plot and Egger’s regression tests are suitable for assessing publication bias if the included studies exceed 10 . If the included studies are less than 10, we will solely assess reporting bias qualitatively based on the characteristics of the included studies.
Assessment of pooled effect estimates
When the heterogeneity of results is high (>60%) in included studies, the DerSimonian and Laird method for random model will be applied; otherwise, the Mantel-Haenszel method for mixed model will be performed . Statistical significance is defined at p < 0.05.
We will evaluate the quality of evidence contributing to pooled effect estimates of the main outcomes with Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for observational studies and the principle of the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system for RCTs [30, 31]. We will use methods and recommendations described in the Cochrane Handbook for Systematic Reviews of Intervention and the GRADEpro GDT software . Eventually, we will construct a table summarizing the overall study results.