For patients who have suffered acute ischemic stroke (AIS) in the anterior circulation, the key to saving ischemic penumbra is to open the occluded vessels responsible as soon as possible (1). The cerebral collateral circulation (CC) is an essential factor for the early existence of ischemic penumbra (2).

The CC can be divided into three levels. The first level is the circle of Willis. The second level is ophthalmic collateral arteries, pia-mater anastomosis, and other relatively small collateral anastomoses. The third level is neovascularization, which is time-consuming and cannot be evaluated appropriately by digital subtraction angiography (DSA) (3, 4). Commonly used methods for detecting cerebral CC include DSA, multimodal computed tomography angiography (CTA), magnetic resonance perfusion (MRP), and arterial spin labeling (ASL). DSA is the “gold standard” for evaluating a grade-I and -II CC (5). However, it is a traumatic and expensive procedure that necessitates the support of a professional neurointerventional team. Multimodal CTA (6, 7) may be less sensitive for assessment of the leptomeningeal CC compared with that using DSA. DSA and CTA require injection of a contrast agent and carry the risk of contrast-related injuries. MRP (8) or ASL (9) can be used for perfusion assessment rather than direct evaluation for CC. Presence of good CC may enhance the success rate of reperfusion and recanalization in endovascular treatment for AIS (10). CC is a predictor of maintaining brain tissue and good clinical prognosis after AIS, and poorer CC may result in poorer perfusion.

Susceptibility-weighted imaging (SWI) is a magnetic resonance imaging (MRI) sequence. It has high spatial resolution and sensitivity to identify differences in magnetic susceptibility between tissues. Besides its capacity to detect cerebral microbleeds, SWI is considered to reflect oxygen saturation in brain tissues and cerebral vessels. Asymmetrical cortical vein sign (ACVS) denotes more and/or larger venous signals in the cerebral cortex of one side of SWI than that in the contralateral side (11, 12). Several studies show that ACVS is more common in patients with large-vessel occlusion or severe stenosis and reflects cerebral hypoperfusion. Park et al. (13) found that total mismatch of diffusion-weighted imaging (DWI)–SWI may be a predictor of good response to treatment in patients with AIS and is associated with cerebral perfusion. In patients with moyamoya disease, ACVS on SWI might be considered as a neuroimaging marker for the evaluation of hemodynamics (14). Moreover, ACVS was observed to be significantly lessened after revascularization (15). ACVS has also been studied in the outcome of AIS. Our recent study (16) found that ACVS might be a useful predictor of early neurological deficits in AIS patients with symptomatic large artery stenosis or occlusion after r-tPA treatment. As perfusion status was significantly affected by CC, it will be interesting to know the relationship between ACVS and CC. To date, very few studies have investigated the relationship between ACVS and CC. Xu et al. (17) found that ACVS was associated with good leptomeningeal collateral flow assessed by hyperintense vessel sign (HVS) on T2-FLAIR images rather than DSA. We wished to assess if ACVS could predict the cerebral CC assessed by DSA in AIS patients with ipsilateral severe stenosis/occlusion of the anterior circulation (ISSACS).

Sixty-two patients (mean age, 59.2 ± 10.6 years; 69.4% male) were evaluated. Thirty-one patients (50%) had occlusion of the internal carotid artery or M1 segment of the MCA. According to the ASITN scale score, 19 patients (30.65%) had a good CC (grade 3–4), and 43 (69.35%) patients had a poor-to-moderate CC (grade 0–2). Among the 30 patients with moderate-to-severe ACVS, only three (10%) patients had a good CC, and 27 (90%) patients had a poor-to-moderate CC; among the 32 patients with none or mild ACVS, 16 (50%) of them had a good CC and the other 50% had a moderate-to-severe CC. Patients with a good CC did not differ significantly from those with a poor-to-moderate CC in terms of age, sex, vascular risk factors, or previous stroke (P > 0.05 for all). Patients with a good CC were more likely to have large-artery occlusion (73.7 vs. 39.5%, P = 0.013) (Table 1).

We constructed two logistic regression models with ACVS grading and none or mild ACVS entered into the models, respectively, together with age and large-artery occlusion. In model 1, no ACVS (compared with severe ACVS, odds ratio (OR) = 40.329, 95% confidence interval (CI) = 2.817–577.422, P = 0.006), mild ACVS (compared with severe ACVS, OR = 17.186, 95%CI = 1.735–170.224, P = 0.015), and large-artery occlusion (OR = 45.645, 95%CI = 4.603–452.592, P = 0.001) correlated with a good CC. In model 2, none or mild ACVS (OR = 36.848, 95%CI = 5.516–246.171, P < 0.001) was significantly associated with a good CC judged by DSA, adjusted by age and large-artery occlusion (Table 2). We also conducted a logistic regression model with poor CC serving as a dependent variable. Moderate-to-severe ACVS did not significantly correlate with poor CC (Supplementary Table 1).

The ROC curve showed the predictive ability of ACVS grading and none or mild ACVS for CC. The AUC for predicting a good CC for none or mild ACVS and ACVS grading was 0.735 and 0.706, respectively (Figure 3).

A box plot shows the relationship between ACVS grading (abscissa) and NIHSS stroke score (ordinate) (Figure 4). ACVS grading correlated with the severity of stroke (Spearman r = 0.275, P = 0.031).

The main finding of our study was that, in AIS patients with ISSACS, ACVS grading (especially none or mild ACVS on SWI) strongly predicted a good CC on DSA. This finding is practical for physicians to acquire, in a prompt and non-traumatic way, CC status using only MRI without the need for contrast-agent injection.

The principle of SWI is dependent mainly on deoxyhemoglobin content in venous blood (21). In patients with AIS, the vasodilation caused by ischemia in the cerebral cortex on the affected side (as well as an imbalance in the supply and demand of oxygen in hypoperfused tissue after cerebral-artery occlusion) leads to a relative increase in the deoxyhemoglobin level in the corresponding draining veins and a relative decrease in the oxyhemoglobin level. This action leads to ACVS, which, as a new imaging marker of CC status, is discussed in a few studies (8, 13). A study on the mismatch between DWI and SWI indicates that the increased oxygen demand of hypoperfused brain tissue leads to an increase in the deoxyhemoglobin:oxyhemoglobin ratio. If the insufficiency of cerebral perfusion can be improved and blood vessels can be recanalized, the risk of irreversible ischemic injury can be reduced (13, 22).

The CC is the basis of ischemic penumbra. A good CC can provide sufficient perfusion for the affected hemisphere (23) and reduce the need of ischemic tissue for a metabolic reserve. Early and rapid identification of ischemic penumbra can help to decide whether intravenous thrombolysis or mechanical thrombectomy is indicated. We found that, in comparison with people with severe ACVS, those without or with only mild ACVS were more likely to have a good CC even if they had symptomatic ISSACS. In addition, none or mild ACVS seems to be a strong indicator for a good CC. These results are in accordance with data from a study by Verma et al. (8), which also found that the better the state of the CC, the smaller was the range of ACVS. However, that study had only 33 cases and conducted univariate comparisons only. Determination of ACVS is difficult if bilateral large arteries have severe stenosis/occlusion. Thus, we recruited only AIS patients with ISSACS. Furthermore, we also adjusted these results by age and large-artery occlusion and confirmed this strong association. However, we did not find that severity of ACVS could predict a poor CC. Moderate-to-severe ACVS commonly implicates hypoperfusion, which does not always mean the presence of poor CC. ACVS on MRI-SWI is a non-invasive technique without radiation and contrast allergy risk. Not every AIS patient with large arterial stenosis or occlusion needs a DSA evaluation. It can partially replace DSA assessment of CC in patients with contraindications or unwilling to have DSA for collateral circulation evaluation. However, ACVS is not suitable for patients with bilateral severe large arterial stenosis or occlusion.

In a study of dynamic changes of SWI signals in AIS, Baik et al. (15) found that ACVS was significantly lessened after revascularization. Also, the prognosis of these patients was, in general, not poor but, for patients with hypoperfusion who did not have brain improvement in a timely and complete manner, ACVS on SWI indicated that these patients would be more prone to early neurological deterioration. Previously, we also found that ACVS can predict early neurological deterioration effectively in AIS patients with symptomatic large-artery stenosis/occlusion after treatment with recombinant tissue plasminogen activator (24).

ACVS was difficult to quantify accurately. We used the ASPECT scoring to evaluate the range of ACVS. ASPECT scoring was initially used for the measures of the early CT ischemic changes range in predicting benefit with intravenous thrombolysis. A lower ASPECT score implicates a larger range of involvement. ACVS reflects the change of vein shape and number and is difficult to quantify. SWI-ASPECT scoring can be used to evaluate the range of ACVS as recommended by some studies. Only one study (25) attempted to define ACVS quantitatively using susceptibility percentage change. However, to the best of our knowledge, no studies have successfully quantified the extent of ACVS using objective methods.

The strengths of our study were that the CC was confirmed using DSA (the gold standard) and that we adjusted for confounders in our analyses. However, our study had four main limitations. First, the sample size of our study was small, and the regression model might have been not very stable. Second, we recruited only AIS patients with ipsilateral arterial lesions. Thus, this finding may be valuable only for unilateral severe stenosis/occlusion in the anterior circulation and not applicable for bilateral lesions. Third, as mentioned before, ACVS is difficult to quantify, and we assessed it via a semi-quantitative method. Last, we did not have dynamic evaluation of ACVS as no patients in our stroke center would have SWI examination before thrombolysis or thrombectomy according to the guideline of our stroke center.

Our study provides evidence for the reliability of ACVS in judging the CC in treatment of AIS in the anterior circulation. Further longitudinal studies using larger patient cohorts and more objective ACVS quantification, e.g., using machine-learning methods, are warranted to justify this finding.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of Affiliated Dongguan Hospital, Southern Medical University. The ethics committee waived the requirement of written informed consent for participation.

Y-HZ and Y-KC conceptualized and designed the study, acquired and analyzed data, and drafted the manuscript for intellectual content. Y-HZ and G-PL took measurements. Y-LL and C-QX measured SWI data. R-XL, Z-QW, W-MX, and W-DH selected and recruited the patients. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.