Assessment of Collaterals Using Multiphasic CT Angiography in Acute Stroke: Its Correlation with Clinical Outcomes
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.333473
Source of Support: None, Conflict of Interest: None
Keywords: Acute stroke, clinical outcomes, hemorrhage, multi-phase CT angiography.
Cerebral neuronal damage is not uniform in the initial hours after the onset of stroke. In some patients, the infarct gets completely developed within a few hours, whereas in others the ischemic tissue remains viable for several hours to a few days depending on the degree of intracranial collaterals, which plays a significant protective role., The collaterals were studied by various angiographic modalities, including computed tomography (CT),, magnetic resonance imaging (MRI), and catheter angiography. Plain CT followed by CT angiography has been the preferred modality to investigate acute stroke in patients with any major vessel occlusion. Furthermore, a triage method of treatment, such as intravenous thrombolysis or mechanical thrombectomy, is favored for such individuals. The majority of the published studies have employed sCT angiography as the imaging method to assess the intracranial collateral status. However, the concept of multiphase CT (mCT) angiography is recently gaining prominence, and a few stroke centers have started to adopt this technique as the initial imaging modality for patients with acute stroke. The mCT angiography is a modification of the single-phase protocol and involves the addition of a few more angiographic phases for an enhanced evaluation of the intracranial collaterals. Menon et al., in their research, applied the mCT angiography protocol for collateral grading and showed that the radiation dose and acquisition time were lower than that in CT perfusion. In addition, unlike the latter method, complicated postprocessing steps and the use of additional intravenous contrast agents are not required in the former.
We hypothesized that the collateral score obtained using mCT angiography would predict the clinical outcomes of the patients in a better way than the sCT angiography. Hence, this study attempted to compare the multiphase CTA collateral score with the single-phase score and correlate the obtained results with the clinical outcomes (mRS [modified Rankin scale] 90 days) and ASPECTS (Alberta Stroke Program Early CT score) at 24 hours in acute ischemic stroke patients.
This was a prospective study conducted between October 2016 and May 2018 with approval from the institutional ethical committee. All acute stroke patients who had undergone plain CT followed by mCT angiography within 8 hours of stroke symptom onset were included in the research. Only those with pretreatment NIHSS (National Institutes of Health Stroke Scale) score ≥5 (i.e., minor stroke was excluded) and anterior circulation major vessel occlusion of middle cerebral artery (MCA; M1, M2), internal carotid artery (ICA), or combined MCA–ICA were included. Patients with baseline CT intracranial hemorrhage were excluded.
Examinations were performed using the Philips 256-slice CT scanner. All patients who fulfilled the inclusion criteria underwent standard unenhanced plain CT with 0.625-mm section thickness, which was reconstructed into 3 mm-thick images to evaluate the CT ASPECTS score. Scanning was triggered by bolus tracking, with the region of interest being placed in the posterior aortic arch and the trigger threshold being set at 150 HU. Triple-phase CT angiograms of the brain vasculature were acquired. The region extending from the arch of the aorta to the vertex was covered during the peak arterial phase. The skull base and the vertex were the regions included in the peak and late venous phases, respectively. The angiograms were acquired at an interval of 8 seconds. A total of 50 mL of contrast material (iodixanol, Visipaque) was injected at the rate of 5 mL/s, followed by 30 mL of the normal saline chase. The source images were reformatted into 3-mm thick axial, coronal, and sagittal projections. The first phase of the mCT angiography was performed as per the standard protocol used for traditional sCT angiography. In the multiphase technique, major vessel occlusion was noted and standard ALBERTA mCT collateral grading from 0 to 5 was done, with good collaterals being defined as Grade 4 or above. Later, single-phase TAN collateral grading (0–3) was assessed. Grades 2 and 3 were recorded as good collaterals. Baseline NIHSS was evaluated and recorded by an experienced neurologist. Age, gender, risk factors, and type of treatment received (conservative management, intravenous thrombolysis, or mechanical thrombectomy) were documented. Two independent radiologists with 7 and 6 years of experience in the field checked the collateral grading for interobserver reliability.
In patients who had undergone mechanical thrombectomy, the time to groin puncture, onset to reperfusion, and mTICI (modified treatment in cerebral infarction) grading were also recorded. All patients underwent 24 hour-plain CT as per our institutional protocol. The details of CT ASPECTS (0–10) and hemorrhagic transformation by ECASS (European Cooperative Acute Stroke Study) grading were assessed by another independent radiologist with 7 years of experience, who was blinded to the CT angiography results. NIHSS at discharge and mRS (modified Rankin scale) at 90 days were also recorded. Good and poor functional outcomes were defined by mRS scores of 0–2 and 3–6, respectively.
Statistical differences between the variables in the sCT and mCT collateral groups were assessed using Student's t test (for continuous variables) or Fisher's exact test (for categorical variables). Multivariate logistic regression analysis was performed for the variables exhibiting significant differences in the two tests to assess their odds ratio and P values. Receiver operator characteristic (ROC) curve analysis of sCT and mCT angiography collaterals for predicting the functional independency (mRS at 90 days), the area under the curve (AUC), Youden's J point, and test efficiency were calculated individually for both the collateral gradings. Moreover, sensitivity, specificity, likelihood ratios, positive and negative predictive values, and accuracy of both the collateral scoring methods in predicting the 3-month functional outcome (mRS) were also assessed.
Demographics: A total of 56 consecutive patients who had experienced acute anterior circulation stroke with major vessel occlusion were involved in this prospective study. Twenty-one females and 35 males (male-to-female sex ratio of 5:3) were included, and their mean age was 63 years range: (37- 86 years) [Table 1]. Forty-two cases of MCA occlusion and 14 cases of ICA/combined ICA–MCA occlusion were present.
Clinical condition at admission and follow-up: For the 56 patients, the mean NIHSS score and mean ASPECTS at admission were 16 ± 6 and 6 ± 2, respectively. The mean time from symptom onset to baseline CT was 3 hours 20 minutes. Twenty-nine patients (52%) underwent mechanical thrombectomy, and stent retrievers were used in all cases. The mean door-to-groin puncture time was 51 ± 22 minutes. Nine patients (16%) received both intravenous thrombolysis and mechanical thrombectomy, whereas 4 (7%) received only intravenous thrombolysis. Mechanical thrombectomy was not done in few individuals as they were not eligible (>6 hours of stroke symptoms, ASPECTS <6) based on the guidelines at the time of study period (2016–2018). In 76% (22 out of 29) of the patients, TICI Grade 2B or 3 recanalization was achieved. The mean time from symptom onset to arterial recanalization was 263 ± 80 minutes. The 24-hour follow-up of the mean CT ASPECTS score was 4. Of the 56 patients, six exhibited PH 2 (parenchymal hematoma type 2) ECASS2 type of hemorrhagic transformation, whereas three displayed PH 1 bleed. The mean NIHSS at discharge was 8. The patients were followed up after 90 days, and 48% of them attained a good functional outcome of mRS ≤2.
Variables based on the collateral status: Patients with coronary heart disease [Table 1] had poor mCT collaterals (25% vs. 8%, P = 0.002), which was statistically significant. Besides, those with hypertension, atrial fibrillation, diabetes, and hyperlipidemia also had poor multiphasic collaterals but not at a statistically significant level. Among the patients with good collaterals in mCT, 62% underwent thrombectomy, whereas only 43% of those with poor collaterals were subjected to the procedure (P = 0.01). There was no difference in the time-to-groin puncture between the patient groups in both sCT and mCT.
Multivariate logistic regression model: [Table 2] Among the variables with significant differences in the Student's t test and Fisher's exact test, baseline ASPECTS, NIHSS at admission, mechanical thrombectomy, multiphase collateral score, and 24-hour admission were found to be statistically significant in terms of the independent predictors of favorable outcomes. The odds ratio of the multiphase collateral score was much more efficient in predicting a good functional outcome than that of the single-phase collaterals (the odds ratio was 15.1; P = 0.001, 95% CI (2.8, 81) for mCT vs. 2.2; P = 0.2 for sCT).
It was observed that 82% of the patients [Table 3] with a good mCT collateral score had mRS ≤2, whereas only 54% of those with a good sCT collateral score exhibited similar functional independence.
The ASPECTS at 24 hours (5.2 in mCT vs. 3.0 in sCT, P = 0.002) was also significantly better in patients with a good mCT collateral score than in those with a good sCT collateral score. None of the patients with a good mCT collateral score [Table 3] displayed significant hemorrhagic transformation (PH 2, P < 0.001).
Sensitivity analysis: Good mCT collateral scores [Table 4] were shown to have 78% sensitivity, 81% specificity, and 80% accuracy in predicting the 3-month functional outcome. In contrast, sCT had only 54% accuracy. The positive and negative predictive values were 0.82 and 0.76, respectively, for the mCT scoring. The mCT collateral scoring method had a higher positive likelihood ratio of 4.2 when compared with the ratio of 1.15 for the single-phase system.
Correlation analysis: The mCT scoring had a statistically better Spearman rank correlation (P < 0.05) “R” value than the single-phase system, with baseline ASPECTS, NIHSS at admission, and mRS at 90 days.
ROC curve analysis: ROC for predicting a good functional outcome was done separately for the mCT and sCT collateral scoring. The AUC was 0.853 (95% CI [0.73, 0.97]) for the former, indicating that it is a good indicator (0.8–0.9) for predicting long-term functional independence. By using Youden's J point analysis, an mCT collateral score of 4 and above was found to be the optimum cutoff point for predicting a good long-term functional outcome, with a high sensitivity of 82% and a specificity of 83%. Meanwhile, sCT scoring had an AUC value of only 0.609 (95% CI [0.43, 0.78]), implying that it is a poor predictor of functional independence. Youden's J point for sCT collateral scoring revealed that a score of 2 was the optimum cutoff point, with a sensitivity of 41% and a specificity of 76%. A representative case is depicted in [Figure 1].
Interobserver reliability: The collateral gradings were assessed by two independent radiologists with 7 and 6 years of experience, showing an excellent interrater reliability and a kappa value of 0.9 for both sCT and mCT collateral evaluations.
In this study, the collaterals were found to be more effectively assessed by mCT angiography than by sCT. The multiphase grading correlated better than the single-phase grading with the baseline NIHSS, baseline and 24-hour CT ASPECTS, and 90-day mRS. Furthermore, the former grading was found to have a prognostic value in predicting the risk of significant hemorrhagic transformation. Our results demonstrated the good interobserver reliability of both sCT and mCT in collateral grading.
Patients with good collateral scores in either single or multiphase scoring depicted a favorable NIHSS score initially, which agrees with the earlier studies involving single-phase collateral scoring. Those with good collaterals in both methods had decent baseline CT ASPECTS, which might be due to the sustained perfusion of the penumbra by the satisfactory leptomeningeal collaterals. The mCT collateral scoring had better correlation with the 24-hour final CT infarct volume measured using CT ASPECTS. Some researchers have shown that patients with good intracranial collaterals were less likely to have symptomatic intracerebral hemorrhagic transformation.,
In our investigation, none of the good mCT collateral group patients had PH2 type of hemorrhagic transformation, whereas 18% in the poor collateral group had bleeding. Thus, mCT served as a good predictor of hemorrhagic transformation. The group with a good mCT collateral score exhibited superior functional independence as per the mRS at 90 days. This result was statistically significant, with 86% being functionally independent as against the mere 54% for the good sCT collateral score. A cutoff value of 4 or above in the currently followed ALBERTA mCT scoring was shown to predict a good functional independence with favorable sensitivity and specificity. In the sCT TAN scoring method of 0 to 3, a grade of 2 or above was observed to predict a positive clinical outcome even though the sensitivity was less than 50%. Among the patients who underwent mechanical thrombectomy, most (97%) had mCT collaterals of 3 to 5. We presume that some of the false-positive and false-negative estimates of collaterals on outcome were possibly due to the influence of other risk factors of the patients. As the procedure was done according to universally accepted guidelines, it can be assumed that mCT collateral scoring is one of the best imaging tools to predict the patients who might benefit from mechanical thrombectomy.
Various factors have been identified to influence the intracranial collaterals. In our study, only coronary heart disease (CHD) was recognized as a statistically significant variable leading to poor collaterals (mCT). This observation might be due to the lower ejection fraction in CHD patients, which secondarily leads to poor intracranial blood flow and collateral formation. Hypertensive and hyperglycemic patients were also noted to have poor mCT collaterals, but the values were not statistically significant. Prior studies involving angiography also indicated that patients with hyperglycemia had poor collaterals, but they did not achieve independent predictor status in multivariate analysis. Atrial fibrillation and smoking had no significant influence on intracranial collaterals in both single and multiphase angiography in our study, as noted in MRCLEAN (Multicenter Randomized Clinical Trial of Endovascular Treatment of Acute Ischemic Stroke in the Netherlands). The overall good outcome predictors in our study were baseline ASPECTS, NIHSS at admission, mechanical thrombectomy procedure, and multiphase collaterals. Among these, mCT collateral was found to be the most significant predictor, followed by mechanical thrombectomy.
A study by Menon et al. employed data from the PRoveIT (Precise and Rapid assessment of collaterals using multiphase CT angiography in the triage of patients with acute ischemic stroke for interarterial therapy) study encompassing 147 patients. Unlike this work that included not only acute stroke patients admitted within 12 hours but also distal and nonocclusion cases, our research encompassed only those who presented within 8 hours and exhibited major vessel occlusion. Hence, our work is likely to be more informative in terms of routine clinical practice involving patients in the therapeutic window. In this research, ROC analysis showed a C-statistic value of 0.6 for the mCT collateral score in predicting mRS at 90 days, and it was superior to the sCT scoring. In our study, the C-statistic for mCT collateral scoring was much higher at 0.85 and was inferred to considerably surpass the sCT scoring.
In the investigation by Flores et al., poor collaterals in mCT angiography predicted progression to malignant MCA infarction. On multivariate analysis, mCT collateral was the only independent predictor of malignant stroke, which agrees with our results in correlating well with the 24-hour infarct volume measured using CT ASPECTS. Recently, a study by García Tornel et al evaluated the use of mCT angiography in acute stroke patients treated with endovascular reperfusion. In this research, only age and mCT angiography were shown to be the independent predictors of functional outcome on performing logistic regression analysis. In our study, ASPECTS and NIHSS scores were also identified to be the independent predictors of functional outcome. In the investigation by García Tornel et al, even though symptomatic hemorrhagic transformation was less in patients with good mCT collaterals, it did not attain a level of statistical significance. However, in our study, none of the patients with good mCT collaterals developed PH2 type of bleeding, whereas 18% of those with poor collaterals experienced the bleeding.
Ours was a prospective study that included consecutive cases of acute anterior circulation ischemic stroke. However, there are a few limitations. Although ours was one of the largest prospective studies that assessed intracranial collaterals by using the multiphasic collateral system, the number of patients was limited. Besides, we did not include the posterior circulation and isolated ACA circulation strokes in which the role of mCT collaterals must be analyzed. The strength of our study lies in the meticulous readings of our experienced neuroradiologists and the assessment of interobserver kappa.
The results of our research establish that mCT angiography is significantly more efficient than the traditional single-phase method in predicting the early radiological and functional outcomes as well as the long-term functional independence. The presence of good mCT collateral scoring was one of the strongest predictors of long-term functional independence, irrespective of the endovascular management. The mCT collateral scoring system was highly accurate and specific in predicting the hemorrhagic transformation of acute infarcts. A cutoff value of 4 or above was shown to predict a good functional independence.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Table 1], [Table 2], [Table 3], [Table 4]