Unfavorable venous outflow correlates with poor prognosis in acute ischemic stroke due to large vessel occlusion (AIS-LVO) patients assessed dynamically and quantitatively based on four-dimensional computed tomography angiography/perfusion (4D-CTA/CTP)

Background: The condition of cerebral venous outflow (VO) and tissue level collaterals (TLC) can indirectly reflect cerebral microvascular perfusion in patients with acute ischemic stroke due to large vessel occlusion (AIS-LVO). This study aimed to investigate the association among VO, TLC, and long-term functional prognosis based on four-dimensional computed tomography angiography/perfusion (4D-CTA/CTP); and to compare the predictive value of different VO indicators for poor outcome. Methods: Enrolled AIS-LVO patients were categorized into good and poor outcome groups based on 90-day modified Rankin Scale (mRS) score, and they all underwent 4D-CTA/CTP examination at baseline. Veins were evaluated by three indicators: (I) four-dimensional cortical venous collaterals score (4D-VCS); (II) combined indicators of velocity and extent. Nine combined indicators were obtained according to the combination of velocity (Fast or Slow), and extent (Good or Poor), which were divided into three categories: favorable, moderate, and unfavorable; (III) cortical vein opacification score (COVES) at the venous phase. TLC was evaluated according to the hypoperfusion intensity ratio (HIR). Mann-Whitney U test or Chi-squared test was used to compare indicators between groups. Binary logistic regression analysis was used to explore the relationship between vein indicators and prognosis, and combined the vein indicators, TLC, and clinical indicators to establish 11 multivariable models. At last, the receiver operating characteristic (ROC) curve performance was compared by DeLong test. Results: A total of 172 patients were included. Compared with good outcome group, patients of poor outcome group showed advanced age {73.137±11.020 years old; 76 [interquartile range (IQR), 66–81] years old}, higher National Institutes of Health Stroke Scale (NIHSS) [12.34±6.308; 12 (IQR, 8–16)], lower 4D-VCS [5.67±3.988; 5 (IQR, 3–8)], lower COVES [3.088±1.904; 3 (IQR, 2–5)], Slow1 + Poor and Slow2 + Poor. The results of binary logistic regression analysis showed that the advanced age and high NIHSS were independent predictors for poor outcome in all 11 models (P<0.05). Besides, low 4D-VCS [odds ratio (OR) =0.863; 95% CI: 0.791–0.941; P<0.05], low COVES (OR =0.737; 95% CI: 0.591–0.919; P<0.05), Slow1 + Poor (8.878; 95% CI: 1.063–74.150; P<0.05), and Slow2 + Poor (OR =8.878; 95% CI: 1.063–74.150; P<0.05) can also independently predict poor outcome. Among them, 4D-VCS had the largest area under the curve (AUC) of 0.744 (95% CI: 0.668–0.820, P<0.001) and got an optimal cutoff value of 8.5. Moreover, through DeLong test, it can be found that Model 8 (AUC =0.827, 95% CI: 0.765–0.889; P<0.05), which included 4D-VCS, TLC, and clinical information, had statistical difference with Model 1 (P>0.05), which only took the clinical information into account. Conclusions: The three evaluation indicators of VO based on 4D-CTA/CTP can independently predict the long-term functional prognosis of AIS-LVO patients, and 4D-VCS had relatively higher predictive value. In the prognosis prediction of AIS-LVO patients, the combined model including 4D-VCS, clinical indicators, and TLC was superior to the model which only included clinical indicators.