Artery and venous sinus occlusion image score (AVOIS): A novel method to evaluate occlusive cerebral arteries and venous diseases

Abstract Aim To establish an artery and venous sinus occlusion image score (AVOIS) which is compatible in both cerebral arteries and venous system diseases. Methods A total of 188 consecutive patients with the final diagnosis of anterior circulation infarct (ACI) and 56 consecutive patients with cerebral venous and sinus thrombosis (CVST) were retrospectively studied. The AVOIS was developed based on the severity of occlusive changes of main intracranial arteries and venous sinuses (present = 0, partial occlusion = 1, absent = 2), and divided into four groups (CVST group: 0, 1‐5, 6‐10, >10. ACI group: 0, 1‐5, 6‐10, >10) arbitrarily. A receiver operating characteristic (ROC) curve was applied to discover the sensitivity and specificity of AVOIS. The National Institutes of Health Stroke Scale (NIHSS), Clot Burden Score (CBS) were set as the reference. Logistic regression models were developed to adjust for baseline clinical variables and AVOIS. Length of hospital stay (LOS) was also evaluated using the Kaplan‐Meier estimator. Results For the CVST group, a positive correlation between AVOIS and NIHSS was discovered (Spearman's ρ = 0.54, p < 0.001). For the ACI group, ROC showed relatively high sensitivity (84.8%) and specificity (81.8%). Besides, the probability of time to discharge was significantly different among the AVOIS subgroups as well (p < 0.001). Conclusion The AVOIS can be used to evaluate the treatment of patients with acute stroke caused by cerebral venous sinus thrombosis and anterior circulation large vessel occlusion. It is a reliable and convenient method that may help prompt prognosis and guide the treatment of individual patients.


| INTRODUC TI ON
Acute ischemic stroke (AIS) is an emergency with a critical time window of treatment. [1][2][3] A classification tool for AIS must be reproducible, reliable, and suitable for the assessment of disease severity levels. 4,5 Previously, clinical rating instruments, such as the National Institutes of Health Stroke Scale (NIHSS), 6 Boston Acute Stroke Imaging Scale (BASIS), 4 Alberta Stroke Program Early CT Score (ASPECTS), 7,8 and Clot Burden Score (CBS) 9,10 have been used to evaluate the early ischemic changes of the middle cerebral artery (MCA) and the brain parenchyma supplied by MCA in patients with stroke. However, the current methodologies have some shortcomings. 11 For instance, NIHSS is limited in the ability to evaluate the effects of arterial occlusion treatment, and it provides little guidance on how to improve prompt and specific treatment. 4,12 CBS has been considered as an efficient evaluation tool for AIS of the anterior circulation, 13 but its utilization is limited in the evaluation of vessels with partial filling defects. 4 Furthermore, the above classification instruments focused specifically on patients with thrombotic stroke in cerebral arteries rather than thrombus formation in large veins (cerebral venous sinus). As far as we know, there is no relevant report on the scoring system which is based on the intracranial venous images and focuses on the evaluation of the patency of the intracranial veins. Although the intracranial venous system is as important as the arteries system, occlusion of cerebral veins, such as cerebral venous and sinus thrombosis (CVST), receives far less attention as compared with arterial stroke like acute anterior circulation infarct (ACI).
In this context, a novel classification instrument-Artery and Venous Occlusion Image Score (AVOIS)-was designed by quantification of cerebral arteries and venous sinuses changes on CTA/ MRA images. To test the value of AVOIS in the evaluation of cerebral venous occlusion and artery disease, patients diagnosed with CVST and ACI were included in this study.

| Patients
This is a retrospective study, and the inclusion met ethical standards approved by ethical committees of the First Affiliated Hospital of Chongqing Medical University. For patients who were conscious and cooperative, written informed consent was obtained from both patients and their legal guardians. For comatose patients, written informed consent was obtained from their legal guardians or their health care surrogates, and then from the patients themselves when they regained decision-making capacity.
From 01/2019 to 01/2020, patients who had a final diagnosis of the CVST and ACI were included in the study. Non-contrast CT (NCCT) or MRI was the default imaging modality in patients with suspected cerebral vascular disorder. The treating stroke neurologist decided to whether proceed to CTA/MRA and decided who was selected for endovascular therapy (EVT) based on CT/MRI characteristics. According to the chief cause of acute stroke, patients with ACI or CVST were naturally grouped. CTA or MRA was performed before and after treatment, and the initial NIHSS score and the follow-up NIHSS were recorded in the electronic medical records.
Other factors (ie, age, gender, and serum glucose) were extracted from medical records.
For the CVST group, inclusion criteria were given as: 1) age ranged 20-80 years; 2) with acute disabling neurological deficits; 3) with significant symptoms (eg, headache and vomiting); 4) with the presence of an intraluminal filling defect on CTA. Exclusion criteria were given as: 1) with primary cerebral diseases (eg, intracranial space-occupying lesions, encephalitis, psychosis, dementia, or psychosis, etc.); 2) with severe dysfunction of other organs or systems (such as serious respiratory and circulatory dysfunction, electrolyte disorders, etc.) For the ACI group, inclusion criteria were given as: 1) age ranged 20-80 years; 2) with acute disabling neurological deficits;

| Assessment of thrombus burden
A novel score system called the Artery and Venous Sinus Occlusion Image Score (AVOIS) was developed based on quantification analysis of the intracranial thrombus in both venous sinus and artery (Table 1, Figures 1-4). According to the CTA or MRA, anterior circulation and venous sinuses were allotted different scores (present = 0, partial occlusion = 1, absent = 2) by two individual researchers through the double-blind method. Except for the symptom side, they were blind to the patient's name, birth, and examination date. If there was a conflict between the two researchers, a third well-trained researcher would review the images and give his interpretation. Then the final consensus was reached after a discussion. The total score on AVOIS in venous sinuses was accumulated to 14 points and was divided into four groups (0, 1-5, 6-10, >10) arbitrarily. Similar to the venous sinuses, the cumulative score in anterior circulation was 14 and divided into four groups (0, 1-5, 6-10 and >10) as well.

| Primary outcome measurement
The modified Rankin scale (mRS) score at 90 days was evaluated by a stroke neurologist in the stroke center in our hospital, who did not know the results of baseline NIHSS score, CT/CTA, MRI/ MRA, or acute clinical events. Functional independence (favorable outcome) was defined as mRS score ≤2, unfavorable outcome as >2.

| Statistical analysis
Statistical analyses were accomplished by IBM SPSS version 26.0 software (IBM SPSS Statistics). Data were presented using standard descriptive statistics. All data have been subject to tests for normality and data that do not exhibit a normal distribution was analyzed via a non-parametric rank-sum test (Kruskal-Wallis Test).
Multivariable logistic regression models were developed to adjust for baseline clinical variables and AVOIS in predicting clinical outcomes. Length of hospital stay (LOS) data was also evaluated using the Kaplan-Meier estimator. p-value <0.05 was required for significance. A receiver operating characteristic (ROC) curve was applied to discover the sensitivity and specificity of AVOIS in the diagnosis.

| RE SULTS
We identified 267 consecutive patients who performed CTA or MRA for clinically suspected acute ischemic stroke within 24 h from symptom onset. Among these, 56 patients had a final diagnosis of venous sinus system thrombosis, 188 patients had a final diagnosis of anterior circulation ischemic stroke, 13 patients had posterior circulation ischemic stroke. Another 10 patients were excluded as they were not accordant with the inclusion criteria for the unqualified pre-mRS and NIHSS. TA B L E 1 Summary of the artery and venous sinus occlusion image score (AVOIS)

Present Partial occlusion Absent
Venous sinuses

| Cerebral venous and sinus thrombosis group
The clinical characteristics of 56 CVST patients are summarized in Interestingly, the higher clot burden was accompanied by the higher OR and the poorer clinic outcome in both favorable and unfavorable groups. When the AVOIS score increased, the risk of acquiring an unfavorable outcome (mRS > 2) increased accordingly (Table 4).
In the CVST group, patients with higher AVOIS (ie, higher thrombus burden) had higher baseline NIHSS scores ( Figure 5A). There was a highly significant positive correlation of AVOIS with NIHSS scores (Spearman's ρ = 0.56, p < 0.001). We applied the Kaplan-Meier estimator to evaluate the LOS data ( Figure 5B). This is a kind of analysis commonly used to test the curative effect such as survival and had been applied to estimate the probability of time to discharge from   (Table S2). In various subgroups, AVOIS in the ACI group showed highly significant differences. (Table S3).

| Anterior circulation infarct group
To assess whether AVOIS was superior to CBS in the diagnosis of ACI, ROC was utilized for testing the sensitivity and specificity ( Figure 6). In the ACI group, AVOIS showed relatively high sensitivity (84.8%) and specificity (81.8%), and the best cutoff point was 5.5.

TA B L E 2 Baseline characteristics of CVST group
The area under the curve (AUC) of AVOIS was 0.925, which was larger than the area of the CBS (AUC CBS = 0.813).

| DISCUSS ION
In the present study, a novel neuroimaging-based stroke score system called AVOIS has been developed. After being tested in CVST and ACI patients and compared with other image scoring tools, we found that AVOIS was an effective method that was compatible in both the cerebral venous system and the arteries. The LOS and prognosis could be also predicted by AVOIS in both CVST and ACI groups. For CVST patients, AVOIS provided an additional novel evaluation method for thrombus formation in large cerebral veins (venous sinus) and was highly effective and accurate in predicting poor outcomes. In the ACI group, AVOIS was also shown to be of important practical value as it is simple to use and can yield reliable information for the evaluation of thrombotic stroke in cerebral arteries.
One of the main purposes of this study was therefore to find a reliable method to evaluate the intracranial venous system. By testing in CVST patients, we found the availability of AVOIS in the assess-  For example, the LOS of patients with AVOIS > 10 was almost three times as high as that of AVOIS = 0. Also, the logistic regression analysis showed that AVOIS was one of the risk factors for unfavorable outcomes at 3 months (mRS > 2), and a higher AVOIS was strongly associated with a poor prognosis. All these findings suggested that AVOIS might have potential clinical application in the diagnosis and follow-up of CVST. Based on this analysis, we speculated that patients with lower AVOIS scores could more easily achieve a better prognosis.
Another purpose of this study was to test the availability of AVOIS in the assessment of unilateral occlusive changes in the major cerebral arteries. As we know, ASPECTS is based on brain parenchymal damage on NCCT, 7 and NIHSS is mainly based on a standard neurological examination. 18 CBS and BATMAN were designed for anterior circulation and posterior circulation respectively and were both angiography-based scoring systems. 13,19 However, CBS has a minor flaw in that partial filling defects in anterior circulation were rated as patent when evaluating, 13 which may be unreasonable because that may lead to biases in the evaluation of partial occlusion of the blood vessel. Likewise, we noticed similar potential problems in the scoring tool of the vertebra basilar system (BATMAN). In AVOIS, partial occlusion (ie, partially present, or congenital stenosis) was allotted one point, while complete occlusion and no occlusion were allotted two and zero separately. This is one of the main differences between CBS and AVOIS. To gain more insight in the diagnosis of ACI, we performed a ROC analysis and found AVOIS had a higher AUC-ROC (0.925) as compared to CBS (0.813), demonstrating this novel instrument was highly sensitive, specific, and accurate in diagnosing.
In conclusion, we believe that the above findings on AVOIS are novel and might have clinical significance. First, as a classification tool of stroke, AVOIS has been proven to be compatible in both cerebral venous system and arteries. Second, in cases in which cerebral venous sinus thrombosis and infraction were suspected, AVOIS may aid in diagnosis and decisions about disposition due to its accuracy, convenience, and simplicity. Finally, our findings in this study suggested that AVOIS has potential clinical application in the follow-up of occlusive cerebrovascular diseases.
However, we must acknowledge there are several methodological shortcomings of the AVOIS in the assessment. First, the thrombosis in deep cerebral veins, such as cavernous sinus, inferior and superior petrosal sinus, cannot be evaluated according to this version of the image score. Second, known contrast agent allergy has limited the clinical application of AVOIS in some patients. Third, the LOS in this report could be affected by various factors such as pneumonia, urinary tract infection, sepsis, and complication with other system diseases, which may influence the total hospital stay time.
What's more, this is a single-center and small sample research and needs further study to confirm the validity of AVOIS.

| CON CLUS ION
The AVOIS provides a novel accurate semi-quantity evaluation method to evaluate occlusive cerebral arteries and venous diseases and may help guide the treatment of individual patients and predict prognosis.

ACK N OWLED EG M ENTS
We acknowledge Dr. Jiang for the design and management of this study, Prof. Xie for the data collection and processing, Prof. Dan and F I G U R E 6 The receiver operating characteristic curve (ROC) of AVOIS and CBS. The area under the curve (AUC) of AVOIS was 0.925, which was larger than the area of the CBS (AUC CBS = 0.813).