Endovascular Treatment for Acute Isolated Internal Carotid Artery Occlusion

We conducted a retrospective multicenter observational study with propensity score matching and pooled data from the university hospitals of Dresden and Heidelberg, both Germany; Boston, MA, USA; and Lausanne, Switzerland. Between 1 January 2016 and 31 December 2020, we included consecutive patients with AIS due to symptomatic isolated occlusion of the ICA and no evidence of intracranial occlusion other than the ICA on noninvasive baseline imaging with computed tomography angiography (CTA) or magnetic resonance angiography (MRA). Eligible patients were adults with a NIHSS score ≥ 1 attributable to the vascular territory of the occluded ICA and treatment initiation within 24 h of stroke onset or last seen well. Additional inclusion parameters were prestroke functional independence with a modified Rankin scale (mRS) score of ≤ 2, no extensive infarction with a baseline Alberta stroke program early CT score (ASPECTS) ≥ 6, and no intracranial hemorrhage on baseline imaging.

This study was approved by the Ethics Committee of the TU Dresden, Germany (EK 272072017) with waiver of informed consent. The study complied with the Declaration of Helsinki and the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.

We collected the following patient characteristics: age, sex, comorbidities and vascular risk factors, prior antithrombotic and statin medications, prestroke mRS, and NIHSS on admission. We recorded stroke onset or the time the patient was last seen well and whether patients were admitted directly or transferred from another hospital. We collected data on baseline imaging with CT/CTA or MRI/MRA. An isolated ICA occlusion was diagnosed using standard CTA or MRA and confirmed by angiography in case the patient received EVT. Perfusion imaging was optional. We evaluated infarct size (ASPECTS), the side and location of occlusion, the Tan collateral score (0–3) with scores of 2 and 3 indicating good collaterals [14].

After qualifying imaging, patients were treated with EVT or BMT alone including intravenous thrombolysis (IVT) and antiplatelet therapy, according to international guidelines. The criteria for selecting patients for EVT included clinical radiological mismatch with neurological symptoms and small infarct size on noncontrast CT. The treating physicians chose the treatment strategy and devices at their discretion, including aspiration and/or stent retriever thrombectomy for EVT. Carotid stent with or without angioplasty was utilized for treatment of an underlying high-grade ICA stenosis. We did not use embolic protection devices. Prior to angioplasty and/or stenting procedures, patients received an acetylsalicylic acid (ASA) loading dose. If no intracranial hemorrhage was detected on follow-up imaging, clopidogrel was additionally administered. Tirofiban was used as an alternative to ASA.

We recorded the quality of reperfusion with the expanded thrombolysis in cerebral ischemia (eTICI) scale. The blood pressure was managed according to international and local guidelines during and after the procedure. The patients were monitored on a stroke unit or neurological intensive care unit. First follow-up imaging was performed within 24 h.

The primary endpoint was favorable functional outcome (mRS ≤ 2) 90 days after treatment. We also analyzed NIHSS score at discharge and delta NIHSS (admission–discharge NIHSS scores).

Safety outcome measures were symptomatic intracranial hemorrhage, death, and serious complications of EVT. In detail, we recorded any complication of EVT that could be or was associated with less favorable neurological outcome and any type of intracranial hemorrhage at follow-up imaging. Distal thrombus migration was defined as evidence of secondarily downstream migration of the thrombus with occlusion of the M1 segment of the middle cerebral artery during EVT [15]. Symptomatic intracranial hemorrhage was defined as any bleeding at 24 h on follow-up imaging associated with neurological deterioration of ≥ 4 points NIHSS [16].

Analysis of the raw data was performed for categorical variables with the χ²-test and with Fisher’s exact test as appropriate. Continuous variables were analyzed with the Mann-Whitney U-test. Continuous variables were presented as median and interquartile range (IQR), categorical variables as absolute and relative frequencies. Propensity score matching (PSM) method was 1:1 nearest neighbor matching without replacement. The caliper was 0.2, the propensity score was estimated with logistic regression, the target estimand was the average treatment effect on the treated (ATT). Covariates with pronounced imbalance at baseline, significant association to outcome, and clinical relevance were included as variables to be balanced in the PSM model. The following covariates were selected for propensity score matching: NIHSS on admission, ipsilateral extracranial occlusion, age, hypercholesterolemia, history of stroke, prestroke mRS, sex, and atrial fibrillation. The degree of equal distribution of individual covariates before and after PSM was measured by the absolute standardized mean difference (SMD). A SMD < 0.1 was taken as the limit for a balanced matching. To account for matching, univariate conditional logistic regression was chosen as primary logistic model to estimate the treatment effect in the matched sample. In order not to miss any confounders of the analysis, the distribution of all parameters in the raw group and in the patients excluded by matching was analyzed for unequal distribution. The detailed method of PSM and sensitivity analysis is described in the supplemental material. For all statistical analyses, a p-value of < 0.05 was considered significant. Analyses were performed using STATA 17.0 (Stata Corp, College Station, TX, USA), StatXact 12.0 (Cytel, Cambridge, MA, USA) and R Statistical Software (version 4.2.1; R Foundation for Statistical Computing, Vienna, Austria).

Of 4635 patients treated with anterior circulation large vessel occlusion during the study period, 164 (4%) had an isolated symptomatic ICA occlusion. We excluded 2 patients in whom treatment initiation was beyond 24 h after symptom onset and 13 patients with a prestroke mRS > 2. Of the 149 included patients, 74 (50%) received EVT and BMT and 75 (50%) BMT only. Table 1 shows the baseline characteristics of the raw sample.

The number of observations was 149 in the original and 90 (45 in EVT and 45 in BMT) in the matched sample (Fig. 1). After PSM, a sufficient adjustment for the imbalanced covariates could be achieved (Table S2, Figure S1a–c). No covariate with a SMD > 0.1 after matching was significantly related to outcome. Detailed matching results and sensitivity analyses are described in the supplemental material.

After matching, the median age was 65 years (IQR 56–78) and 54 (60%) patients were male. The median NIHSS score on admission was 9 (IQR 5–15) in the EVT and 9 (IQR 4–15) in the BMT group (p = 0.65). In the EVT group 31 (69%) patients were directly admitted to the tertiary hospital versus 39 (87%) patients in the BMT group (p = 0.07). The baseline characteristics of the matched sample are detailed in Table 2.

In the analysis of matched patients, the rate of favorable outcome (mRS ≤ 2 at 90 days) after treatment was 56% (n = 25) in the EVT and 38% (n = 17) in the BMT groups. In univariate conditional logistic regression, the likelihood of favorable outcome was not significantly higher for patients treated with EVT (OR 1.89; 95% CI 0.84–4.24; p = 0.12). Multivariate conditional logistic regression including all covariates that were unbalanced after matching (except collateral score due to partial lack of values) also showed no significant advantage of additional EVT treatment (OR 2.35; 95% CI 0.88–6.29; p = 0.09; Table 3) as well as the mRS levels of the EVT compared to the BMT group at day 90 after stroke (p = 0.09, Fig. 2). Median NIHSS at discharge was significantly lower in the EVT group (4, IQR 1–7) than in the BMT group (7, IQR 3–20; OR 0.95; 95% CI 0.90–1.00; p = 0.04). Patients treated with EVT also showed a better reduction in NIHSS score when comparing admission and discharge scores (median reduction of 6 vs. 1 NIHSS points; p = 0.02). The ASPECTS score after treatment was higher in the EVT group (median 8, IQR 7–10) compared to BMT (8, IQR 5–8; p = 0.05; Table S4).

In the matched sample, the rate of symptomatic ICH was not significantly higher in the EVT compared with the BMT group (7%, n = 3 vs. 4%, n = 2; OR: 1.50; 95% CI 0.25–8.98; p = 0.66). The mortality rate at 90 days was comparable between both groups (EVT 11%, n = 5 versus BMT 13%, n = 6; OR 0.80; 95% CI 0.21–2.98; p = 0.74; Table S4).

In the raw sample, periprocedural complications occurred in 5 (11%) patients treated with EVT, of whom 3 (7%) had early neurological deterioration with an increased NIHSS of ≥ 4 points within 24 h. In detail, the complications were one distal thrombus migration, one arterial dissection, one extensive parenchymal hemorrhage, and one inadvertent stent retriever detachment. Multiple complications were reported in one patient.