Reperfusion measurements, treatment time, and outcomes in patients receiving endovascular treatment within 24 hours of last known well

Abstract Aims The aim of this study was to explore the interaction between reperfusion and treatment time on the outcomes of patients undergoing endovascular treatment presenting within 24 h of last known well, and to compare the predictive ability of different reperfusion measurements on outcomes. Methods Eligible patients from a single‐center cohort were enrolled in this study. Reperfusion was assessed using reperfusion index (decreased volume of hypoperfusion lesion compared with baseline) measured by repeated perfusion imaging, and modified treatment in cerebral ischemia score measured by digital subtraction angiography, respectively. The interactions between reperfusion measurements and treatment time on outcomes were explored using multivariate‐adjusted logistic and linear regression models. The predictive abilities of reperfusion measurements on outcomes were compared using area under the receiver operating characteristic curve (ROC‐AUC) and values of R‐square. Results Reperfusion index and treatment time had significant interactions on 3‐month modified Rankin Scale (mRS) 0–2 and infarct growth (p for interaction <0.05). Although the AUCs were statistically similar (AUCs of mRS 0–2 prediction, mTICI≥2b:0.63, mTICI≥2c:0.59, reperfusion index≥0.5:0.66, reperfusion index ≥0.9:0.73, P value of any of the two AUCs >0.05), reperfusion index≥0.9 showed the highest R‐square values in outcome prediction (R‐square values of 3‐month mRS 0–2 and infarct growth = 0.21) among all the reperfusion measurements. Conclusion Treatment time mitigated the effect of reperfusion on outcomes of patients receiving endovascular treatment within 24 h of last known well. Reperfusion index≥0.9 might serve as a better proxy of good outcomes compared with other reperfusion measurements.


| INTRODUC TI ON
Emergent endovascular treatment (EVT) has been a routine practice for acute large vessel occlusion (LVO) patients presenting within 24 hours of last known well (LKN). 1,2 Reperfusion is the key determinant of EVT efficacy, and is time-dependent with greater benefit of earlier treatment in patients punctured within 12 h of LKN. [3][4][5][6][7][8] However, a post hoc analysis of the Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution 2 (DEFUSE 2) study has shown that the benefit of EVT would not be undermined by delayed time in patients with imaging selection, 9 where the number of patients with good collaterals, also known as the "slow progressors" increases with time. 3,10 Whereas, there are odds where collateral flow can fail over time even for patients showing benign perfusion profiles. 11 Additionally, the stringent target mismatch criteria in DEFUSE serious studies 12,13 combining mismatch ratio, volume of core, penumbra, and areas with severe delay are hard to strictly follow under real-world clinical practice, with which a considerable number of patients will be ruled out. Therefore, we hypothesized that reperfusion and treatment time had interactions on outcomes in patients undergoing EVT within 24 h of LKN in real-world practice.
Reperfusion restoration measured by repeated perfusion imaging and modified treatment in cerebral ischemia (mTICI) score measured by digital subtraction angiography (DSA) imaging are the two major measurements of reperfusion status post-EVT. 12,14,15 Nevertheless, few studies directly compare the predictive ability of these two measurements in outcome prediction. 16 Imaging and clinical analysis of real-world EVT patient data who follow a more flexible mismatch standard of imaging selection can provide a unique prospective of associations between treatment time and outcomes, as well as the prediction ability of different reperfusion measurements. Therefore, using the data of a singlecenter cohort of EVT patients who presented within 24 h of LKN, the aim of this study was to (1)   Healthcare) at 4.5 ml/s, followed by 20 ml saline. CTP scan was initiated at 7 s after the contrast agent bolus. CTA with acquisition from aorta arc to vertex was performed immediately after perfusion CT.

| Imaging protocol and imaging analysis
Brain standard reconstruction was then performed. To avoid radiation exposure to the lens, the gantry angle was parallel to and above the orbital roof. Whether CTP and CTA were acquired before or after intravenous rtPA depended on the judgment of stroke neurologists. Reperfusion index≥0.5 was considered as major reperfusion, 12 and reperfusion index≥0.9 was considered as complete reperfusion. 18 The final infarct volume (FIV) was also calculated using MIStar with planimetric techniques by semi-automatically drawing regions of interests (ROIs) in DWI/NCCT images. Infarct growth was calculated by subtracting baseline core volume from the final infarct volume. Imaging of DSA was centrally reviewed by a neurointerventionist (Dr. YF Ling) with 2-year experience of neurointervention who was blinded to the reperfusion analysis and clinical outcome. Partial reperfusion was considered as mTICI score ≥ 2a.
Major reperfusion was defined as mTICI score ≥ 2b, and complete reperfusion was defined as mTICI score ≥ 2c. 15

| Statistical analysis
Statistical analysis was performed using Stata v15.1 (StataCorp, College Station). Graphs were drawn using Stata v15.1 (StataCorp, College Station) and Prism v8.0 (GraphPad Software). A two-tailed p < 0.05 was considered as significant. Mean and standard deviation were used to describe continuous variables if normally distributed, or median and interquartile range (IQR) if skewedly distributed.
Categorical variables were described using percentage. For continuous variables, normality was tested using Shapiro-Wilk test.
Differences of demographic and clinical and imaging data were compared using Student's t-test, Wilcoxon rank-sum test or Kruskal-Wallis test for continuous variables, and χ2 test or Fisher's exact test for categorical variables. The correlation between any two continuous variables was tested using Spearman correlation coefficient if not normally distributed, or Pearson correlation coefficient if normally distributed. The interaction between reperfusion index / mTICI score and each time interval was tested using multivariateadjusted linear/logistic regression model adjusted by any baseline demographic, imaging and clinical variables that were significant in the univariate analysis or variables that were considered clinically relevant.
The predictive ability of reperfusion measurements, including major/complete reperfusion assessed by CTP/DSA on the functional outcome was calculated using the area under the receiver operating characteristic curve (AUC-ROC) and compared using a method developed by Pepe et al. and Janes et al. 19,20 adjusted for other baseline demographic, imaging and clinical variables that were significant in the univariate analysis or variables that were considered clinically relevant. Additionally, R-square values were also calculated using multivariate logistic regression models. The one with the highest AUC and R-square value was considered the best predictor of functional independency. For infarct growth, multivariate simple linear regression models were used to assess reperfusion measurements in infarct growth prediction. R-square values of the linear regression models were also provided to compare their predictive value.
In this study, any multivariate analysis concerning mRS 0-2 were adjusted by age and baseline National Institutes of Health Stroke Scale (NIHSS) score as statistically significant or clinically relevant baseline variables, and any multivariate analysis concerning infarct growth were adjusted by baseline core volume, history of diabetes mellitus and MR/NCCT scan as statistically significant or clinically relevant variables.  Table 1.

| Interaction between time intervals and major/ complete reperfusion on outcomes
Among all the baseline demographic, clinical and imaging data, as well as reperfusion index≥0.9 (Table S1). Therefore, age, together with baseline NIHSS, was entered in the multivariateadjusted models of 3-month mRS 0-2 as statistically significant or clinically relevant baseline variables. As for the univariate analysis concerning infarct growth, only history of diabetes mellitus and baseline infarct core showed trend of significance among all the baseline variables (history of diabetes mellitus p = 0.06, baseline infarct core p = 0.05, Table S2). Therefore, baseline core volume, history of diabetes mellitus, and MR/NCCT scan were entered in the multivariate-adjusted model of infarct growth as statistically significant or clinically relevant baseline variables.
For all the time intervals, the interaction between time from LKN to groin puncture, time from LKN to multimodal imaging and reperfusion index on 3-month mRS 0-2 was statistically significant, respectively (adjusted for age and baseline NIHSS, Time from LKN to multimodal imaging: p for interaction = 0.046, Figure 1A; time from LKN to groin puncture: p for interaction =0.035, Figure 1B). As for the prediction of infarct growth, the interaction between time from LKN to groin puncture, time from LKN to ER arrival, time from LKN to reperfusion and reperfusion index were statistically significant (adjusted for baseline core volume, history of diabetes mellitus and MR/NCCT scan, Time from LKN to ER arrival: P for interaction =0.04, Figure 1C; time from LKN to groin puncture: p for interaction = 0.02, Figure 1D; time from LKN to reperfusion: p for interaction =0.003, Figure 1E). No statistical significance was found between any time interval and mTICI score (Table S3).

| Comparisons of the predictive ability of major/complete reperfusion of outcomes
The AUCs of reperfusion index was not statistically different from This is also one of the few studies comparing the predictive performance of different reperfusion measurements obtained by repeated perfusion imaging and DSA on functional and imaging outcomes. In our study, reperfusion index≥0.9 was a better predictor of 3-month functional dependency and infarct growth compared with reperfusion index≥0.5, mTICI≥2b and mTICI≥2c. Similar to our study using DT >3 s, a degree of 90% reperfusion measured by Tmax>6 s was suggested as an appropriate reperfusion threshold in outcome prediction, outplaying mTICI score. 16 Recently, mTICI2c, defined as near complete perfusion except slow flow or few distal cortical emboli 15 or 90%-99% filling of the occluded vascular territory, 15

AUTH O R CO NTR I B UTI O N S
All author participated in data collection, critical review and revision of this manuscript. XC, QD designed the study. LH, YL analyzed the data. LH drafted this manuscript, prepared tables and figures.

CO N FLI C T O F I NTE R E S T
The authors have nothing to disclose.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.