Comparison of Time Usage and Success Rate Between CT Angiography- and CT Perfusion- Based Imaging Assessment Strategy for the Patients with Acute Ischemic Stroke

Background: Our study aimed to compare the time usage and success rate between CTA- and CTP- based assessment strategy, and to clarify the risk factors associated with CTP scan failure. Methods: Clinical and radiological data of 437 consecutive AIS patients who underwent multiphase CTA or CTP for pre-treatment evaluation were retrospectively enrolled (CTA group, n=302; CTP group, n=135). Time usage and success rate of CTA and CTP were compared using Mann-Whitney U test and Chi-Squared Test. Univariate analysis and receiver operating curve analysis were used to clarify the risk factors and their values for predicting CTP scan failure. Results: Time usage of CTP scan and reconstruction was signicantly longer than that of CTA [775sec vs 263.5sec, P<0.001]. CTP scan showed signicantly higher failure rate than CTA (11% vs 1%, P<0.001). Severe motion was the most common cause of CTP failure (n=12, 80%). Baseline National Institute of Health Stroke Scale (NIHSS) score in CTP failure group was signicantly higher than that in CTP success group [17 vs 13, P=0.007]. Baseline NIHSS score of 11 was the optimal threshold to predict CTP failure with an area under the curve of 0.771, sensitivity of 86.7% and specicity of 45.0%. Conclusions: CTP-based strategy showed longer time consumption and higher failure rate than CTA-based strategy. High baseline NIHSS score was signicantly associated with CTP scan failure in AIS patients.

Severe motion was the most common cause of CTP failure (n=12, 80%). Baseline National Institute of Health Stroke Scale (NIHSS) score in CTP failure group was signi cantly higher than that in CTP success group [17 vs 13, P=0.007]. Baseline NIHSS score of 11 was the optimal threshold to predict CTP failure with an area under the curve of 0.771, sensitivity of 86.7% and speci city of 45.0%.
Conclusions: CTP-based strategy showed longer time consumption and higher failure rate than CTAbased strategy. High baseline NIHSS score was signi cantly associated with CTP scan failure in AIS patients.

Background
Stroke is a leading cause of disability and mortality in the worldwide [1]. Multimodal CT, including nonenhanced computed tomography (NCCT), CT angiography (CTA), CT perfusion (CTP) imaging, are important imaging techniques for the accurate pre-treatment assessment of acute ischemic stroke (AIS) patients [2]. In clinical practice, NCCT is mainly used to exclude intracerebral hemorrhage, and estimate the extent of early ischemic injury via the Alberta Stroke Program Early CT Score (ASPECTS) [3]. CTA can help to identify whether large vessel occlusion (LVO) occurred or not, and grade the collateral ow status [4,5]. CTP can accurately quantify the "tissue window" including the volume of infarct core and ischemic penumbra, so as to establish a reasonable treatment strategy for the AIS patients, especially in an extended time window [6,7]. However, how to establish the most reasonable imaging assessment strategy for the AIS patients in different time window is still in debate.
The 2019 AHA/ASA guideline recommended CTP as the pre-treatment imaging assessment for AIS patients with an onset time over 6 hours according to DEFUSE 3 trial [2,7]. While for AIS patients within 6 hours of onset with LVO, since the time to reperfusion has a signi cant effect on clinical outcome, the 2019 AHA/ASA guideline recommended angiography assessment only for selecting mechanical thrombectomy (MT) eligibility, and treatment should not be delayed by additional perfusion imaging [2].
Previous studies reported that CTP needed more time to acquire, process and interpret and hold high rate of post-processing failure. These disadvantages might contribute to the recommendation not to use CTP for imaging assessment for patients within the time window [8,9]. However, Olivot et al recently reported that 29 percent of AIS patients with onset time less than 6 hours had no target mismatch. Using CTP to evaluate target mismatch for patients within the time window was also important for identifying patients had potential to respond favorably to reperfusion therapy [10]. Therefore, there was still no consensus on how to establish the imaging assessing strategy for the AIS patients in different time window. In our opinion, full understanding of the time usage and failure rate of CTA and CTP-based imaging assessment strategies was of great importance for making the most reasonable imaging protocol for AIS patients.
Therefore, the purpose of our study was to compare the time usage and success rate between CTA-and CTP-based imaging assessment strategy for the AIS patients, and to clarify the risk factors potentially associated with CTP scan failure.

Patient selection
The study was approved by our center's ethical committee for studies in humans. The requirement for a written informed consent was waived due to the retrospective nature of the study. We included consecutive patients admitted to our emergency department with a diagnosis of AIS from March 2018 to October 2019. We selected patients who (1) were older than 18 years; (2) presented to the emergency department with symptoms consistent with AIS; (3) underwent NCCT + CTA or NCCT + CTP for pretreatment evaluation; (4) were treated by endovascular treatment (EVT) or intravenous thrombolysis (IVT).

Image protocol
Baseline CT was performed using a 128-section CT scanner (Optima CT 660, GE Medical Systems). For AIS patients with an onset time less than 5 hours, imaging protocol included NCCT (120 kVp, 100-350 auto-mAs, and contiguous 5-mm axial sections ranging from the vertex to the skull base) and multiphase CTA. The multiphase CTA generated time-resolved cerebral angiograms of brain vasculature from the skull base to the vertex in three phases after contrast agent injection. Aortic arch vertex CT angiography made up the rst phase. Image acquisition was timed to occur during the peak arterial phase in the healthy brain and was triggered by bolus monitoring. The remaining two phases are from the skull base to the vertex in the equilibrium/peak venous and late venous phases in the healthy brain. Images were acquired with a 0.625-mm section thickness. The second phase was acquired after a delay of 8 seconds that allows for table repositioning to the skull base. Scanning duration for the rst phase was 5.1 seconds and for each additional phase was 2.6 seconds. A total of 50-mL nonionic iodinated contrast (iopromide, Ultravist 370, Bayer Schering Pharma) was administered at a ow rate of 4 mL/s, followed by a 40-mL saline chaser at a rate of 4 mL/s. For AIS patients in whom the onset time ≥ 5 hours or with unclear time, imaging protocol included NCCT and CTP. CTP scan was conducted using the cradle table technique, allowing a coverage of whole brain.
The imaging parameters for CTP scan were as follows: 4-dimensional adaptive spiral mode, periodic spiral approach, 80 mm in z-axis, 1.7 s temporal resolution, 30 consecutive spiral scans, 100 kVp, 200 mAs, rotation time of 0.4 s and 0.984 maximum pitch. A scan delay of 2 s was applied after injecting 50-mL ( ow rate 5mL/s) nonionic iodinated contrast (iopromide, Ultravist 370, Bayer Schering Pharma, Germany), followed by a 30-mL saline at the same rate. CTP data were analyzed retrospectively in a commercial software (Advantage Workstation 4.7; GE Healthcare, Milwaukee, WI, USA), using singular value deconvolution algorithms. Simulated CTA source image was reconstructed from the peak arterial phase in the normal distal ICA of the source CTP images for detecting vessel occlusion, based on the arterial input function (AIF). CTP data were reconstructed with a thickness of 5mm for perfusion analysis, and with a slice thickness of 0.625 mm for CTA analysis.

Data analysis
The patients were divided into CTA group and CTP group based on the imaging modality used. We collected and recorded the time usage of NCCT scan (de ned as time interval from the rst location image to the last NCCT image), CTA or CTP scan and reconstruction (de ned as time interval from the last NCCT image to the termination of CTA or CTP reconstruction), and the total imaging duration (NCCT + CTA or CTP time usage). The potential causes of CTP or CTA scan failure included motion, streak artifact, and poor arrival of contrast bolus due to poor basic condition of vessel. Two raters (both with more than 5 years experience in neuroradiology) who were blinded to the clinical data assessed whether failures occurred or not. In case of disagreement between two raters, consensus was reached by discussion with a senior radiologist (with more than 20 years experience in neuroradiology).
Demographic, clinical and imaging information such as age, gender, baseline National Institute of Health Stroke Scale (NIHSS) score, onset time and occlusion site were collected for univariate analysis between CTP success group and failure group.

Statistical analysis
Continuous variables were reported as median with interquartile range (IQR). Categorical variables were reported as proportions. The comparison of time duration between CTA group and CTP group was used Mann-Whitney U test. Chi-Squared test was used to compare the failure rates between CTA group and CTP group. Demographic, clinical and imaging information between CTP success and failure group were compared using Chi-Squared Test or Mann-Whitney U test. Receiver operating characteristic (ROC) curves were used to assess the ability of the signi cant associated factors to predict the CTP scan failures. Area under the ROC curve (AUC), sensitivity and speci city were calculated and reported. All statistical analyses were performed using MedCalc (version 12.3.0) or SPSS (version 23.0). A two-sided P value less than 0.05 was considered signi cant.

Patients
Among 437 patients enrolled in our study, 302 (69%) patients underwent CTA-based, while 135 (31%) patents underwent CTP-based imaging assessment strategy. The mean age of the patients in CTA group was 70 (IQR 61-77) years, and 60% were male. The median age of the patients was 67 (IQR 60-75) in CTP group, and 61% were male. The median baseline NIHSS score was 12 (IQR 7-19) for CTA group, and 13 (IQR 9-17) for CTP group. The median time interval from stroke onset to CT scan was 2.5 (IQR 1.5-3.5) hours for CTA group, and 6 (IQR 5-8.5) hours for patients in CTP group with clear onset time. Among them 32 (24%) patients in CTP group had awakened from sleep with symptoms of a stroke and with unclear onset time, and the onset time was the time they were last seen well. The number of patients with occlusion in internal carotid artery or MCA, basilar and other sites was 206 (68%), 21(7%) and 75(25%) respectively for the CTA group, while 97(72%), 10(7%) and 28(21%) for the CTP group. In the CTA group, 61 (20%) patients were treated with EVT after intravenous tPA, 152 (50%) patients were treated with intravenous tPA alone, and 89 (30%) patients were treated with EVT alone. While in the CTP group, 16 (12%) patients received EVT after intravenous tPA, 32 (24%) patients received intravenous tPA alone, and 87 (64%) patients received EVT alone. Baseline demographic and clinical characteristics of CTA and CTP group are summarized in Table 1.

Comparison of time usage
Comparison of time usage between CTA and CTP group was shown in Table 2 and Fig. 1  Comparison of scan failure rate and the factors leading to scan failure The scan failure rate of CTP group was 11% (15/135), while that of CTA group was 1% (3/302). The scan failure rate of CTP group was signi cantly higher than CTA group (P < 0.001). The comparison of failure rate between CTA and CTP group was shown in Fig. 2. Causes of CTP scan failure included severe motion (n = 12, 80%), streak artifact (n = 1, 7%) and poor arrival of contrast (n = 2, 13%), while causes of CTA scan failure were severe motion (n = 2, 67%) and streak artifact (n = 1, 33%). The comparisons of demographic, clinical and imaging variables between CTP failure and CTP success group were shown in Table 3. According to the univariate analysis results, baseline NIHSS score was the only variable associated with CTP scan failure [IQR, 17 (13-20) vs 13 (8-16), P = 0.007]. ROC curves of using baseline NIHSS score to predict CTP scan failure were shown in Fig. 3. Using a baseline NIHSS score of 11 as the threshold value, optimal performance could be achieved for predicting CTP scan failure (AUC, 0.771; sensitivity, 86.7%; speci city, 45.0%).

Discussion
Our study demonstrated three major ndings. First, there was no signi cant difference in time consumption for NCCT scan between CTA and CTP-based imaging assessment strategy. However, CTPbased imaging assessment strategy needed signi cant longer time for imaging scan and reconstruction than CTA-based strategy. The time delayed for CTP-based strategy was approximately 10 minutes. Second, we found that CTP scan showed a higher failure rate than CTA scan. The failure rate of CTP scan was 11%, and motion was the most common cause. Third, we found the admission NIHSS were signi cantly associated with CTP scan failure.
In our study, CTP-based imaging assessment strategy showed relatively longer time consumption than CTA-based strategy. Multiphase CTA included a three-time phases scan, while CTP scan contained 30 consecutive spiral scans of the brain. More images raw data would be obtained from CTP scan, and subsequent transform of raw data from scan workstation to reconstruction workstation might be one reason of time delay in CTP-based strategy. Besides that, compared with reconstruction of vascular image from CTA, reconstruction of CTP parametric maps seemed to be a more complicated process. This situation might also be associated the time delay of CTP-based strategy. Currently, automated perfusion analysis software has been widely used in large comprehensive stroke centers. It may dramatically shorten the time usage of CTP reconstruction and subsequent analysis [11][12][13]. While in primary stroke center, the automated reconstruction software was still not popularized. The decision to open up an advanced CTP examination to assess the AIS patients should be taken into careful consideration for the potential longer time consumption of CTP imaging strategy.
Besides longer time consumption, CTP also showed higher failure rate than CTA in our study. Kauw et al reported a CTP failure rate of 11%, and they found that motion was the leading cause, followed by streak artifacts and poor contrast bolus arrival, which was in line with our study [9]. The main reason for the higher failure rate of CTP might be that it was di cult for the patients to remain stationary during the longer scan time than CTA. In consistent with the prior study, we found that baseline NIHSS was the risk factor that signi cantly associated with CTP failure [9]. Patients with higher baseline NIHSS would be neurologically more severe, more restless, and more likely to be mobile during the scanning process. We recommended to take caution on the patients with high NIHSS score, and to take effective measures such as xed band and sedation drugs to keep them static during CTP scan.
When CTP scan failure occurred, usually repeated CTP scan was not suggested because of that waiting the clearance of the contrast retention within the brain would increase the time consumption. Therefore, some previous studies tried to nd fungible imaging biomarkers. Nannoni et al reported that NCCT ASPECTS correlated signi cantly with the CTP infarct core, especially in the AIS patients with late window and large-vessel occlusion. Therefore, they concluded that the use of ASPECTS could be a surrogate maker of CTP core in late-arriving AIS patients with large-vessel occlusion [14]. In addition, simulated multiphase CTA reconstructed from CTP raw data was also reported to be useful in re ecting the perfusion information to a certain extent. Better collateral circulation was tightly correlated with small ischemic core volumes. A multiphase CTA collateral score more than 3 could optimally predict a target mismatch on CTP and a good clinical outcome in AIS patients [15].
Our study had some limitations. First, our study was a retrospective study conducted in single center. As different imaging modalities and different post-processing software were used in different centers, further multi-center trials were needed to be performed. Second, the sample size was relatively small and may limit the reliability of results. Third, the time duration of CTA and CTP post-processing could be in uenced by the experience of different radiologists, so a uni ed post-processing train standard should be taken in further studies.

Conclusion
In summary, our study found that CTP-based imaging assessment strategy needed more time usage, and showed higher failure rate than CTA-based strategy. High baseline NIHSS score might be a risk factor associated with the CTP scan failure in AIS patients.

Declarations
Funding: Comparison of time consumption between CTA and CTP group. CTA indicates computed tomography angiography; CTP, computed tomography perfusion; IQR, interquartile range.

Figure 2
Comparison of failure rate between CTA and CTP group. CTA indicates computed tomography angiography; CTP, computed tomography perfusion; IQR, interquartile range.

Figure 3
Receiver-operating characteristic curves of using baseline National Institutes of Health Stroke Scale to predict the failure of CTP.