- Email: [email protected]
Hyperdense Cerebral Artery Computed Tomography Sign Is Associated with Stroke Severity Rather than Stroke Subtype
Hyperdense Cerebral Artery Computed Tomography Sign Is Associated with Stroke Severity rather than Stroke Subtype Jana Novotna, MD,*† Pavla Kadlecov a, MSc,‡ Anna Czlonkowska, MD, PhD,xjj Miroslav Brozman, MD, PhD,{ Viktor Svigelj, MD,# Laszlo Csiba, MD, PhD,** Janika K~ orv, MD, PhD,†† Vida Demarin, MD, PhD,‡‡ Aleksandras Vilionskis, MD,xx Robert Mikulık, MD, PhD,* and for the SITS-EAST Investigators
Background: The hyperdense cerebral artery sign (HCAS) on unenhanced computed tomography (CT) in acute ischemic stroke is a valuable clinical marker, but it remains unclear if HCAS reflects clot composition or stroke etiology. Therefore, variables independently associated with HCAS were identified from a large international data set of patients treated with intravenous thrombolysis. Methods: All stroke patients undergoing intravenous thrombolysis from the Safe Implementation of Treatments in Stroke-EAST (SITS-EAST) database between February 2003 and December 2011 were analyzed. A general estimating equation model accounting for within-center clustering was used to identify factors independently associated with HCAS. Results: Of all 8878 consecutive patients, 8375 patients (94%) with available information about HCAS were included in our analysis. CT revealed HCAS in 19% of patients. Median baseline National Institutes of Health Stroke Scale (NIHSS) score was 12, mean age was 67 6 12 years, and 3592 (43%) patients were females. HCAS was independently associated with baseline NIHSS (odds ratio [OR], 1.11; 95% confidence interval [CI], 1.10-1.12), vessel occlusion (OR, 5.02; 95% CI, 3.317.63), early ischemic CT changes (OR, 1.63; 95% CI, 1.31-2.04), year (OR, 1.07; 95% CI, 1.02-1.12), and age (10-year increments; OR, .90; 95% CI, .84-.96). Cardioembolic stroke was not associated with HCAS independently of baseline NIHSS. In different centers, HCAS was reported in 0%-50% of patients. Conclusions: This study illustrates significant variation in detection of HCAS among stroke centers in routine
From the *International Clinical Research Center, Department of Neurology, St. Anne’s Hospital, Brno, Czech Republic; †Masaryk University, Brno, Czech Republic; ‡International Clinical Research Center, St. Anne’s Hospital, Brno, Czech Republic; xSecond Department of Neurology, Institute of Psychiatry and Neurology, Warsaw, Poland; jjDepartment of Experimental and Clinical Pharmacology, Medical University of Warsaw, Warsaw, Poland; {Department of Neurology, Faculty Hospital Nitra and Constantine the Philosopher University Nitra, Nitra, Slovakia; #Department of Vascular Neurology and Neurological Intensive Care, University Medical Centre Ljubljana and Zdravstveni Nasveti, Ljubljana, Slovenia; **Department of Neurology, Medical and Health Science Center, University of Debrecen, Debrecen, Hungary; ††Department of Neurology and Neurosurgery, University of Tartu, Tartu, Estonia; ‡‡Department of Neurology, Sestre Milosrdnice University Hospital Center, Zagreb, Croatia; and xxDepartment of Neurology and Neurosurgery, Vilnius University and Republican Vilnius University Hospital, Vilnius, Lithuania. Received September 24, 2013; revision received February 9, 2014; accepted April 28, 2014. Jana Novotn a, Pavla Kadlecova, and Robert Mikulık received research support from the European Regional Development
Fund-FNUSA-ICRC Project (No.CZ.1.05/1.1.00/02.0123). Robert Mikulık, Anna Czlonkowska, Viktor Svigelj, Janika K~ orv, and Aleksandras Vilionskis received research support through a grant from the European Union Executive Agency for Health and Consumers (EAHC). Anna Czlonkowska received support until 2008 from the Polish National Program for the Prevention and Treatment of Cardiovascular Diseases. Robert Mikulık, Anna Czlonkowska, Janika K~ orv, and Aleksandras Vilionskis received honoraria payments and travel support from Boehringer Ingelheim. Laszlo Csiba received honoraria payment from Bayer, Boehringer Ingelheim, MSD, Sanofi-Aventis, and Egis and was an advisory board member in the MATCH and ROCKET trials and in the Bayer company. Address correspondence to Jana Novotn a, MD, International Clinical Research Center, Department of Neurology, St. Anne’s University Hospital, Pekarsk a 53, 656 91, Brno, Czech Republic. E-mail: novotna.j @centrum.cz. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.04.034
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2014: pp 1-7
1
ET AL. J. NOVOTNA
2
clinical practice. Accounting for within-center data clustering, stroke subtype was not independently associated with HCAS; HCAS was associated with the severity of neurologic deficit. Key Words: Dense artery sign—acute stroke— thrombolysis—brain computed tomography. Ó 2014 by National Stroke Association
Introduction Hyperdense and dot artery signs on unenhanced computed tomography (CT) scan of patients with acute ischemic stroke correspond with the presence of thrombus in the proximal or distal cerebral artery.1-5 Hyperdense cerebral artery sign (HCAS) is a valuable clinical marker because it is associated with a more severe neurologic deficit, and it predicts larger stroke volume and poorer functional outcome6-8 following thrombolytic therapy with intravenous recombinant tissue plasminogen activator. A clot composition analysis from a mechanical embolectomy procedure further sparked interest in HCAS findings on CT, because it showed that clots composed mostly from red blood cells were more often visible as HCAS on CT scans than fibrin-dominant clots.9 This analysis is in line with a previous in vitro study that concluded that Hounsfield units are lower in platelet-rich thrombi than in erythrocyte-rich thrombi.10 If unenhanced CT can truly reflect clot composition, then detection of HCAS could lead to some very important clinical implications, such as predicting the efficacy of thrombolytic treatment. Therefore, the goal of the present study was to identify factors associated with HCAS on admission CT using a large international data set of acute stroke patients treated with intravenous thrombolysis. The second goal was to provide data about variability in HCAS detection amongst stroke centers. We hypothesized that stroke subtype or etiology, laboratory parameters, time from symptom onset to imaging, and several other factors would potentially influence the presence of HCAS through clot composition. Our second hypothesis was that stroke centers would differ in reporting of HCAS.
Methods Data on all the patients from the Safe Implementation of Treatment in Stroke-EAST (SITS-EAST) Register between February 28, 2003, and December 31, 2011, were analyzed. The SITS-International Stroke Thrombolysis Registry11,12 and SITS-EAST methodology13 were described in previous publications. Patients were included into the register if they had acute ischemic stroke and had received intravenous alteplase at a dose of .9 mg/kg. The time window from symptom onset to treatment was initially 3 hours and was extended to 4.5 hours after October 2008.11,14
Unenhanced CT scan in stroke thrombolysis is required on admission and at follow-up 24-36 hours after thrombolytic treatment. CT angiography was optional. Only the cases with confirmed baseline data and admission CT results were included in the analysis. All scans were evaluated locally, and the presence of HCAS on plain CT (implying, but not limited to dense middle cerebral artery signs) and arterial occlusion on CT angiography (reported in the SITS database without specification of occlusion site) was recorded separately. Neurologic deficit was assessed by National Institutes of Health Stroke Scale (NIHSS) score. NIHSS of 0 means no neurologic deficit; the higher the NIHSS score, the more severe the neurologic deficit (severe strokes have NIHSS score around 20). Cardioembolic stroke was defined as an atrial fibrillation or International Classification of Diseases, version 10 (ICD 10), diagnosis I63.4 in accordance with the Trial of Org 10172 in Acute Stroke Treatment definition.15 For sensitivity analysis, cardioembolic stroke was alternatively defined as an atrial fibrillation or as ICD 10, diagnosis I63.4. Other demographic and clinical characteristics used in our analysis included sex, age, systolic blood pressure, current smoking status, baseline blood glucose, past medical history (arterial hypertension, diabetes mellitus, hyperlipidemia, congestive heart failure, and use of antiplatelets), presence of early ischemic changes on CT, onset to imaging time, number of recombinant tissue plasminogen activator treatments in the relevant center per year, and year of treatment. Three CT-related variables (CT slice thickness, CT scanner producer, and year of CT scanner production) were obtained from an additional online questionnaire distributed among participating stroke centers. The need for ethical approval or patient consent for participation in the SITS-EAST register varied among the participating countries, but approvals were obtained in those countries in which it was a requirement. The Ethics Committee of the Karolinska Institute in Stockholm approved the SITS-MOnitoring STudy.
Statistics Patients were stratified according to the presence of HCAS on admission CT scan. Demographic and baseline characteristics were compared between groups of HCAS patients and non-HCAS patients by means of descriptive statistics and the t test or Wilcoxon test (if normality was seriously violated) for continuous parameters and the
DENSE ARTERY SIGN DEPENDS ON STROKE SEVERITY
chi-square test for binomial or categorical variables. Potential nonlinear relationships were explored by plotting each continuous/ordinal variable against the presence of HCAS. Variables independently associated with HCAS were identified by general estimating equations (GEEs). To identify variables independently associated with HCAS, a GEE was performed. This method, compared with, for example, logistic regression, accounts for within-center clustering of the data, limiting the effect of intercenter variability on the analysis of factors associated with HCAS. The variables entered in the model were either associated with HCAS at P less than .10 or clinically important variables (age and systolic blood pressure). Only cases with confirmed baseline data and admission CT results were included in the GEE (the numbers of available cases for each variable are listed in Table 1). Because CT angiography was optional, CT angiography data were treated as a 3-category variable: (1) CT angiography not done; (2) occlusion documented on CT angiography; or (3) absence of occlusion on CT angiography.
3
The analysis was performed on all patients who had available data for the presence of HCAS on CT scan. Next, a subgroup analysis was performed on only those patients who had occlusion confirmed on CT angiography. Commonly used goodness-of-fit statistics (quasi-likelihood under independence model criterion)16 were used to fit the final model among the competing models. The statistical significance level was set at less than .05.
Results Of all 8878 consecutive patients, 8375 patients (94%) with available information about HCAS were included in our analysis. Of these, HCAS was present in 1553 patients (19%), median baseline NIHSS score was 12, mean age was 67 6 12 years, and 3592 patients (43%) were females. CT angiography was available in 2540 patients (30%). The baseline characteristics of patients with and without HCAS on admission CT are summarized in Table 1. Several differences were noticed between the 2
Table 1. Characteristics of all patients and patients with presence and absence of hyperdense cerebral artery sign (HCAS) All Characteristics
*N
Sex, female, n (%) Age, mean (SD), y Baseline NIHSS score, median (Q1-Q3) Systolic blood pressure, mean (SD), mm Hg Glucose, median (Q1-Q3), mmol/L Early ischemic changes on CT, n (%) Congestive heart failure, n (%) Hypertension, n (%) Hyperlipidemia, n (%) Diabetes mellitus, n (%) Current smoking, n (%) Use of aspirin, dipyridamole, or clopidogrel before stroke, n (%) Onset to imaging time, median (Q1-Q3), min Cardioembolic stroke,z n (%) Vessel occlusion on CT angiography,x n (%) Absence of vessel occlusion on CT angiography,x n (%) Number of rt-PA treatments per center per y, median (Q1-Q3) Year of treatment, median
8375 8375 8100 8199 8105 7942 8173 8234 7291 8248 8375 8375
8375 (100%) 3592 (43) 67 (12) 12 (8-17) 153 (21) 7 (6-8) 869 (11) 1037 (13) 6107 (74) 2583 (35) 1841 (22) 1814 (22) 2442 (29)
HCAS absent *N 6822 6822 6646 6699 6607 6521 6668 6719 5951 6731 6822 6822
6822 (82%) 2876 (42) 67 (12) 11 (7-16) 153 (21) 7 (6-8) 643 (10) 862 (13) 4991 (74) 2095 (35) 1520 (23) 1510 (22) 1999 (29)
HCAS present 1553 (19%)
P valuey
1553 716 (46) 1553 67 (12) 1454 16 (12-19) 1500 153 (20) 1498 7 (6-8) 1421 226 (16) 1505 175 (12) 1515 1116 (74) 1340 488 (36) 1517 321 (21) 1553 304 (20) 1553 443 (29)
.01 .995 ,.001# 0.1 .92# ,.001 .17 .62 .40 .23 .03 .54
*N
8001 93 (70-125) 8375 3054 (37) 8375 1467 (18) 8375 1073 (13)
6521 94 (70-127) 6822 2409 (36) 6822 969 (14) 6822 984 (14)
1480 1553 1553 1553
90 (67-120) 645 (42) 498 (32) 89 (6)
.001# ,.001 ,.001 ,.001
8375
25 (13-52)
6822
25 (13-48)
1553
27 (14-56)
.001
8375
2009
6822
2009
1553
2009
,.001#
Abbreviations: CT, computed tomography; ICD, International Classification of Diseases; NIHSS, National Institutes of Health Stroke Scale; rt-PA, recombinant tissue plasminogen activator; SD, standard deviation; TOAST, Trial of Org 10172 in Acute Stroke Treatment. *Number of available values are presented in column N. As data are considered as randomly missing, the estimation of percentages and the chi-square test were based on reported cases only. yP value of test comparing characteristics of patients with HCAS present versus absent: the t test or Wilcoxon test (marked by #) was used for continuous parameters and the chi-square test for binomial or categorical parameters. zCardioembolic stroke was defined as an atrial fibrillation or ICD 10, diagnosis I63.4 in accordance with the TOAST definition. xCT angiography was optional. CT angiography was done for 2540 patients (ie, 30% of 8735 patients). CT angiography not done was treated as a category (not as a missing value).
ET AL. J. NOVOTNA
4
groups: initial stroke severity (median NIHSS score) was higher in HCAS patients than in those without HCAS. An infarct was seen more on admission CT scan in patients with HCAS than in those without. Patients with HCAS had a higher frequency of vessel occlusion on CT angiography and were more frequently diagnosed with cardioembolic stroke. CT scanners used in participating centers (data available from 89 of 145 centers) were from 4 producers: Siemens (41%), General Electric (22%), Philips (21%), and Toshiba (15%). The median CT scanner year of production was 2007 (interquartile range, 2004-2009; range from 1997 to 2011). The median of CT slice thickness for unenhanced CT was 3 mm (interquartile range, 2-5). Table 2 lists the results of the multivariate analysis to identify variables independently associated with HCAS. Of 8375 patients, 794 (9%) were eliminated because of incomplete information about baseline NIHSS score (n 5 635), systolic blood pressure (n 5 536), or early ischemic CT changes (n 5 433), leaving 7581 patients (91%) for the final complete case analysis. HCAS was
independently associated with baseline NIHSS (odds ratio [OR], 1.11; 95% confidence interval [CI], 1.10-1.12), vessel occlusion (OR, 5.02; 95% CI, 3.31-7.63), early ischemic CT changes (OR, 1.63; 95% CI, 1.31-2.04), year of treatment (OR, 1.07; 95% CI, 1.02-1.12), and age (OR, .90; 95% CI, .84-.96). Stroke due to cardioembolism was not independently associated with HCAS on admission. The relationship between baseline NIHSS and the presence of HCAS and occlusion on CT angiography is shown in Figure 1. Figure 1 demonstrates that in patients with higher NIHSS score, both HCAS and occlusion on CT angiography were detected more frequently than in patients with lower NIHSS score. In patients with NIHSS score lesser than or equal to 5, the occlusion on CT angiography was still detected in 32% of cases. HCAS was diagnosed in 0%-50% of patients in centers with greater than or equal to 10 treated patients. HCAS was detected in 89 patients (8%) who had no vessel occlusion on baseline CT angiography (n 5 1073), compared to 498 patients (34%) who had occlusion (n 5 1467). In patients with occlusion on baseline CT angiography, the prevalence of
Table 2. Results of univariate and multivariate analysis to identify predictors of hyperdense cerebral artery sign (HCAS; all patients with available data about baseline HCAS on CT, complete case analysis, n 5 8375) Univariate analysis
Multivariate analysisy
Variable
*OR (95% CI)
P valuey
*OR (95% CI)
P valuez
Sex, female Age, by 10-year increments, y Baseline NIHSS score, by 1-point increments Systolic blood pressure, mm Hg Early ischemic changes on CT Glucose, by 5-mmol/L increments Hypertension Diabetes mellitus Hyperlipidemia Congestive heart failure Current smoking Use of aspirin, dipyridamole, or clopidogrel before stroke Onset to imaging time, min Cardioembolic strokex Vessel occlusion on CT angiography Number of rt-PA treatments per center per y Number of rt-PA treatments per center, by 100-patient increments Year of treatment, by 1-year increments Center CT slice thickness, mm CT scanner producer Year of CT scanner production
1.17 (1.05-1.31) 1.00 (.96-1.05) 1.12 (1.11-1.14) 1.00 (1.00-1.00) 1.73 (1.47-2.04) .99 (.96-1.01) .97 (.85-1.10) .92 (.80-1.05) 1.05 (.93-1.19) .89 (.75-1.05) .86 (.75-.98) .96 (.85-1.10) 1.00 (1.00-1.00) 1.30 (1.16-1.46) 5.68 (4.5-7.2) 1.00 (1.00-1.01) 1.11 (1.00-1.23) 1.05 (1.02-1.08) NA 1.06 (.97-1.17) NA 1.05 (.98-1.14)
.01 1.00 ,.001 .11 ,.001 .21 .62 .23 .40 .17 .03 .55 .56 ,.001 ,.001 ,.001 .05 ,.001 ,.001 .21 .35 .18
1.08 (.96-1.21) .90 (.84-.96) 1.11 (1.10-1.12) 1.00 (1.00-1.00) 1.63 (1.31-2.04)
.21 .001 ,.001 .49 ,.001
1.04 (.88-1.23)
.66
1.08 (.91-1.28) 5.0 (3.31-7.63)
.37 ,.001
1.03 (.94-1.13) 1.07 (1.02-1.12)
.56 ,.001
Abbreviations: CI, confidence interval, CT, computed tomography; ICD, International Classification of Diseases; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio; rt-PA, recombinant tissue plasminogen activator; TOAST, Trial of Org 10172 in Acute Stroke Treatment. *OR: probability modeled is for the presence of HCAS on CT. yUnivariate analysis was performed using logistic regression; CT slice thickness, CT scanner producer, and year of CT scanner production were analyzed with generalized estimating equations. Multivariate analysis was performed with generalized estimating equations. zP value of the Wald chi-square test of parameter significance. xCardioembolic stroke was defined as an atrial fibrillation or ICD 10, diagnosis I63.4 in accordance with the TOAST definition.
DENSE ARTERY SIGN DEPENDS ON STROKE SEVERITY
Figure 1. The relationship between baseline National Institutes of Health Stroke Scale (NIHSS; y axis) and the presence of HCAS and occlusion on computed tomography (CT) angiography (x axis). In patients with higher NIHSS score, both hyperdense cerebral artery sign (HCAS) and occlusion on CT angiography were detected more frequently. Occlusion on CT angiography was detected in 91 cases (32%) of 283 patients with NIHSS score less than or equal to 5. Estimation of percentages and confidence intervals was based on reported cases only (8100 patients with known data about baseline NIHSS and HCAS and 2376 patients with known data about baseline NIHSS and CT angiography).
HCAS in centers with greater than or equal to 10 treated patients was 34%, and ranged from 10% to 86%. According to our additional sensitivity analysis, when baseline NIHSS score (but no other variable) is excluded from the final model, variable cardioembolic stroke (defined as I63.4 or atrial fibrillation) becomes significantly associated with the presence of HCAS (OR, 1.23; 95% CI, 1.03-1.47). Using alternative definitions of cardioembolic stroke did not change the results—cardioembolic stroke was not associated with the presence of HCAS. A subgroup analysis on patients with the presence of occlusion on CT angiography (complete case analysis, n 5 1212) showed that HCAS on CT was independently associated with baseline NIHSS score (OR, 1.08; 95% CI, 1.06-1.10), early ischemic CT changes (OR, 1.67; 95% CI, 1.12-2.49), baseline serum glucose (OR, .78; 95% CI, .62.98), and age (OR, .99; 95% CI, .98-1.00). As in overall analysis, stroke due to cardioembolism was not independently associated with HCAS on admission. The results of a subgroup univariate and multivariate regression analysis are available in Table S1 in Appendix.
Discussion We found that HCAS was present in 19% of all patients in the SITS-EAST registry. Previous studies reported a similar or slightly higher prevalence of HCAS (15%32%).6,17-19 In our study, a higher prevalence of HCAS was observed in some centers, whereas other centers never diagnosed HCAS (range, 0%-50%). When considering only the patients with occlusion on CT angiography, HCAS was diagnosed in up to 89% of cases in some centers. We suggest that a large intercenter variation in HCAS detection on CT is not
5
related to the fact that patients in some centers differ in their biological characteristics, but rather to the fact that quality of care, imaging expertise, or scanning protocols differ among stroke centers. Our data reflect the limitations of diagnosing HCAS in routine clinical practice, based on qualitative criteria with low interobserver agreement (Kappa statistic, .36-.55).20-24 In our study, the important finding was that patients with HCAS had higher baseline stroke severity and the presence of clot on CT angiography. As shown in Figure 1, with increasing baseline NIHSS score, there was a linear increase in the chance of diagnosing HCAS. Because a higher NIHSS score mostly reflects larger thrombus size,25,26 the simplest interpretation could be that a clot must be present and must be big enough to be visible as HCAS. Although our data did not contain any information about several well-established predictors of stroke severity, such as core or penumbra size, clot location, and collateral status, our interpretation about the importance of clot size is strongly supported by 4 previous smaller studies on patients with CT angiography confirmed intracranial occlusion. In one of these studies (n 5 78),27 thrombus volumes were significantly larger in patients with HCAS than in those without this sign. In this and in the 3 other studies (n 5 45,28 n 5 58,29 and n 5 5430), slice thickness of less than or equal to 1.25 mm on CT scans had 100% sensitivity to diagnose HCAS. Consequently, the detection of smaller clots on unenhanced CT might be overcome by decreasing slice thickness and increasing expertise. Our data, however, did not show an association between CT slice thickness and HCAS detection. The value of expertise is supported by our next finding; with every year, there were better odds of detecting HCAS. In our study, the next variable associated with HCAS was early ischemic changes on CT. This finding can have 2 interpretations. First, the presence of early ischemic changes (eg, focal swelling or parenchymal hypoattenuation) has been reported to be determined by the duration and the level of cerebral hypoperfusion.31 Therefore, it is possible that early ischemic changes are, in addition to NIHSS scores, markers of clot size. Alternatively, some physicians may have regarded the presence of HCAS as one of the manifestations of ‘‘early ischemic changes,’’ which would create a false positive association between these 2 variables. Another finding in our study is that older patients had a lower probability of being diagnosed with HCAS. We hypothesize that this association between age and HCAS can be mediated by processes related to the aging of the brain, such as cerebral atrophy, leukoaraiosis, or vessel tortuosity. This finding was also reported in a previous SITS analysis (n 5 10,023).6 However, another smaller, single-center study (n 5 1,010)19 showed an opposite relationship between age and the presence of
ET AL. J. NOVOTNA
6
HCAS, probably because of different study populations (an unselected general acute ischemic stroke cohort vs. patients treated with thrombolysis). One of the most interesting results of our study, contradicting the findings of previous studies, is that stroke mechanism is not independently associated with the presence of HCAS. Specifically, cardioembolic strokes are not more likely to have HCAS on admission CT, independently from other baseline variables. According to our analysis, although patients with cardioembolic stroke are more likely to have HCAS, this is rather a consequence of their higher baseline stroke severity, and we suggest that it is not due to differences in clot composition. However, from this study alone, we cannot determine the relationship between thrombus density and stroke subtypes.32 A secondary important result of our analysis is that 6% of the patients with low NIHSS score (,5) had HCAS, and 36% had occlusion on baseline CT angiography. Such findings support the evidence that low neurologic deficit by itself may not be sufficient justification for excluding patients from thrombolysis.33-35 The main limitation of our study is that our data, like other data in the SITS registry, were not externally validated. For the future, it would be very useful to collect imaging in the registries, to allow for the validation of imaging data. The strength of our analysis is that the predictors of HCAS remained basically the same, regardless of whether the study population was limited to the subgroup with CT angiography confirmed occlusion. The only minor disparity (because of different patient populations) was that the association between glucose and HCAS was present only in a subgroup analysis, and the association between year of treatment and HCAS was present only in the overall analysis. In conclusion, our study underscores the differences in the detection of HCAS among stroke centers in routine clinical practice. Given the importance of the HCAS finding, it could be useful to standardize diagnosis of HCAS and possibly to implement quantitative detection methods. Accounting for within-center data clustering, we found that the presence of HCAS depends on stroke severity rather than stroke subtype. Our results indirectly challenge the hypothesis that clot composition affects the visibility of the clot on CT. Our study provides further evidence, using a large thrombolytic registry, that unenhanced CT should be explored as a noninvasive, fast, relatively inexpensive, and widely available tool for clot imaging.
Supplementary Data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014. 04.034.
References 1. Gacs G, Fox AJ, Barnett HJ, et al. CT visualization of intracranial arterial thromboembolism. Stroke 1983;14:756-762. 2. Goldmakher GV, Camargo EC, Furie KL, et al. Hyperdense basilar artery sign on unenhanced CT predicts thrombus and outcome in acute posterior circulation stroke. Stroke 2009;40:134-139. 3. Jensen UR, Weiss M, Zimmermann P, et al. The hyperdense anterior cerebral artery sign (HACAS) as a computed tomography marker for acute ischemia in the anterior cerebral artery territory. Cerebrovasc Dis 2010;29:62-67. 4. Barber PA, Demchuk AM, Hudon ME, et al. Hyperdense sylvian fissure MCA ‘‘dot’’ sign: a CT marker of acute ischemia. Stroke 2001;32:84-88. 5. Shetty SK. The MCA dot sign. Radiology 2006;241:315-318. 6. Kharitonova T, Ahmed N, Thoren M, et al. Hyperdense middle cerebral artery sign on admission CT scan–prognostic significance for ischaemic stroke patients treated with intravenous thrombolysis in the safe implementation of thrombolysis in Stroke International Stroke Thrombolysis Register. Cerebrovasc Dis 2009;27:51-59. 7. Manelfe C, Larrue V, von Kummer R, et al. Association of hyperdense middle cerebral artery sign with clinical outcome in patients treated with tissue plasminogen activator. Stroke 1999;30:769-772. 8. Tomsick T, Brott T, Barsan W, et al. Prognostic value of the hyperdense middle cerebral artery sign and stroke scale score before ultraearly thrombolytic therapy. AJNR Am J Neuroradiol 1996;17:79-85. 9. Liebeskind DS, Sanossian N, Yong WH, et al. CT and MRI early vessel signs reflect clot composition in acute stroke. Stroke 2011;42:1237-1243. 10. Kirchhof K, Welzel T, Mecke C, et al. Differentiation of white, mixed, and red thrombi: value of CT in estimation of the prognosis of thrombolysis phantom study. Radiology 2003;228:126-130. 11. Wahlgren N, Ahmed N, Eriksson N, et al. Multivariable analysis of outcome predictors and adjustment of main outcome results to baseline data profile in randomized controlled trials: Safe Implementation of Thrombolysis in Stroke-MOnitoring STudy (SITS-MOST). Stroke 2008; 39:3316-3322. 12. Wahlgren N, Ahmed N, Davalos A, et al. Thrombolysis with alteplase for acute ischaemic stroke in the Safe Implementation of Thrombolysis in Stroke-Monitoring Study (SITS-MOST): an observational study. Lancet 2007;369:275-282. 13. Mikulik R, Kadlecova P, Czlonkowska A, et al. Factors influencing in-hospital delay in treatment with intravenous thrombolysis. Stroke 2012;43:1578-1583. 14. Hacke W, Kaste M, Bluhmki E, et al. Thrombolysis with alteplase 3 to 4.5 hours after acute ischemic stroke. N Engl J Med 2008;359:1317-1329. 15. Adams HP Jr, Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 1993;24:35-41. 16. Pan W. Model selection in estimating equations. Biometrics 2001;57:529-534. 17. Abul-Kasim K, Brizzi M, Petersson J. Hyperdense middle cerebral artery sign is an ominous prognostic marker despite optimal workflow. Acta Neurol Scand 2010; 122:132-139. 18. Qureshi AI, Ezzeddine MA, Nasar A, et al. Is IV tissue plasminogen activator beneficial in patients with hyperdense artery sign? Neurology 2006;66:1171-1174.
DENSE ARTERY SIGN DEPENDS ON STROKE SEVERITY 19. Paciaroni M, Agnelli G, Floridi P, et al. Hyperdense middle cerebral and/or internal carotid arteries in acute ischemic stroke: rate, predictive factors and influence on clinical outcome. Cerebrovasc Dis 2011;32:239-245. 20. Barber PA, Darby DG, Desmond PM, et al. Identification of major ischemic change. Diffusion-weighted imaging versus computed tomography. Stroke 1999;30: 2059-2065. 21. Fiebach JB, Schellinger PD, Jansen O, et al. CT and diffusion-weighted MR imaging in randomized order: diffusion-weighted imaging results in higher accuracy and lower interrater variability in the diagnosis of hyperacute ischemic stroke. Stroke 2002;33:2206-2210. 22. Marks MP, Holmgren EB, Fox AJ, et al. Evaluation of early computed tomographic findings in acute ischemic stroke. Stroke 1999;30:389-392. 23. Schriger DL, Kalafut M, Starkman S, et al. Cranial computed tomography interpretation in acute stroke: physician accuracy in determining eligibility for thrombolytic therapy. JAMA 1998;279:1293-1297. 24. Tomsick TA, Brott TG, Chambers AA, et al. Hyperdense middle cerebral artery sign on CT: efficacy in detecting middle cerebral artery thrombosis. AJNR Am J Neuroradiol 1990;11:473-477. 25. Lewandowski CA, Frankel M, Tomsick TA, et al. Combined intravenous and intra-arterial r-TPA versus intraarterial therapy of acute ischemic stroke: Emergency Management of Stroke (EMS) Bridging Trial. Stroke 1999;30:2598-2605. 26. Fischer U, Arnold M, Nedeltchev K, et al. NIHSS score and arteriographic findings in acute ischemic stroke. Stroke 2005;36:2121-2125.
7 27. Kim EY, Yoo E, Choi HY, et al. Thrombus volume comparison between patients with and without hyperattenuated artery sign on CT. AJNR Am J Neuroradiol 2008; 29:359-362. 28. Kim EY, Lee SK, Kim DJ, et al. Detection of thrombus in acute ischemic stroke: value of thin-section noncontrastcomputed tomography. Stroke 2005;36:2745-2747. 29. Riedel CH, Jensen U, Rohr A, et al. Assessment of thrombus in acute middle cerebral artery occlusion using thin-slice nonenhanced computed tomography reconstructions. Stroke 2010;41:1659-1664. 30. Riedel CH, Zoubie J, Ulmer S, et al. Thin-slice reconstructions of nonenhanced CT images allow for detection of thrombus in acute stroke. Stroke 2012;43:2319-2323. 31. Hirano T, Yonehara T, Inatomi Y, et al. Presence of early ischemic changes on computed tomography depends on severity and the duration of hypoperfusion: a single photon emission-computed tomographic study. Stroke 2005;36:2601-2608. 32. Niesten JM, van der Schaaf IC, Biessels GJ, et al. Relationship between thrombus attenuation and different stroke subtypes. Neuroradiology 2013;55:1071-1079. 33. Meretoja A, Putaala J, Tatlisumak T, et al. Off-label thrombolysis is not associated with poor outcome in patients with stroke. Stroke 2010;41:1450-1458. 34. Kruetzelmann A, Siemonsen S, Gerloff C, et al. Thrombolysis targeting MRI defined tissue at risk in minor stroke. J Neurol Neurosurg Psychiatry 2009;80:1156-1158. 35. Mikulik R, Goldemund D, Reif M, et al. Outcome of patients with negative CT angiography results for arterial occlusion treated with intravenous thrombolysis. Stroke 2009;40:868-872.
Background: The hyperdense cerebral artery sign (HCAS) on unenhanced computed tomography (CT) in acute ischemic stroke is a valuable clinical marker, but it remains unclear if HCAS reflects clot composition or stroke etiology. Therefore, variables independently associated with HCAS were identified from a large international data set of patients treated with intravenous thrombolysis. Methods: All stroke patients undergoing intravenous thrombolysis from the Safe Implementation of Treatments in Stroke-EAST (SITS-EAST) database between February 2003 and December 2011 were analyzed. A general estimating equation model accounting for within-center clustering was used to identify factors independently associated with HCAS. Results: Of all 8878 consecutive patients, 8375 patients (94%) with available information about HCAS were included in our analysis. CT revealed HCAS in 19% of patients. Median baseline National Institutes of Health Stroke Scale (NIHSS) score was 12, mean age was 67 6 12 years, and 3592 (43%) patients were females. HCAS was independently associated with baseline NIHSS (odds ratio [OR], 1.11; 95% confidence interval [CI], 1.10-1.12), vessel occlusion (OR, 5.02; 95% CI, 3.317.63), early ischemic CT changes (OR, 1.63; 95% CI, 1.31-2.04), year (OR, 1.07; 95% CI, 1.02-1.12), and age (10-year increments; OR, .90; 95% CI, .84-.96). Cardioembolic stroke was not associated with HCAS independently of baseline NIHSS. In different centers, HCAS was reported in 0%-50% of patients. Conclusions: This study illustrates significant variation in detection of HCAS among stroke centers in routine
From the *International Clinical Research Center, Department of Neurology, St. Anne’s Hospital, Brno, Czech Republic; †Masaryk University, Brno, Czech Republic; ‡International Clinical Research Center, St. Anne’s Hospital, Brno, Czech Republic; xSecond Department of Neurology, Institute of Psychiatry and Neurology, Warsaw, Poland; jjDepartment of Experimental and Clinical Pharmacology, Medical University of Warsaw, Warsaw, Poland; {Department of Neurology, Faculty Hospital Nitra and Constantine the Philosopher University Nitra, Nitra, Slovakia; #Department of Vascular Neurology and Neurological Intensive Care, University Medical Centre Ljubljana and Zdravstveni Nasveti, Ljubljana, Slovenia; **Department of Neurology, Medical and Health Science Center, University of Debrecen, Debrecen, Hungary; ††Department of Neurology and Neurosurgery, University of Tartu, Tartu, Estonia; ‡‡Department of Neurology, Sestre Milosrdnice University Hospital Center, Zagreb, Croatia; and xxDepartment of Neurology and Neurosurgery, Vilnius University and Republican Vilnius University Hospital, Vilnius, Lithuania. Received September 24, 2013; revision received February 9, 2014; accepted April 28, 2014. Jana Novotn a, Pavla Kadlecova, and Robert Mikulık received research support from the European Regional Development
Fund-FNUSA-ICRC Project (No.CZ.1.05/1.1.00/02.0123). Robert Mikulık, Anna Czlonkowska, Viktor Svigelj, Janika K~ orv, and Aleksandras Vilionskis received research support through a grant from the European Union Executive Agency for Health and Consumers (EAHC). Anna Czlonkowska received support until 2008 from the Polish National Program for the Prevention and Treatment of Cardiovascular Diseases. Robert Mikulık, Anna Czlonkowska, Janika K~ orv, and Aleksandras Vilionskis received honoraria payments and travel support from Boehringer Ingelheim. Laszlo Csiba received honoraria payment from Bayer, Boehringer Ingelheim, MSD, Sanofi-Aventis, and Egis and was an advisory board member in the MATCH and ROCKET trials and in the Bayer company. Address correspondence to Jana Novotn a, MD, International Clinical Research Center, Department of Neurology, St. Anne’s University Hospital, Pekarsk a 53, 656 91, Brno, Czech Republic. E-mail: novotna.j @centrum.cz. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.04.034
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2014: pp 1-7
1
ET AL. J. NOVOTNA
2
clinical practice. Accounting for within-center data clustering, stroke subtype was not independently associated with HCAS; HCAS was associated with the severity of neurologic deficit. Key Words: Dense artery sign—acute stroke— thrombolysis—brain computed tomography. Ó 2014 by National Stroke Association
Introduction Hyperdense and dot artery signs on unenhanced computed tomography (CT) scan of patients with acute ischemic stroke correspond with the presence of thrombus in the proximal or distal cerebral artery.1-5 Hyperdense cerebral artery sign (HCAS) is a valuable clinical marker because it is associated with a more severe neurologic deficit, and it predicts larger stroke volume and poorer functional outcome6-8 following thrombolytic therapy with intravenous recombinant tissue plasminogen activator. A clot composition analysis from a mechanical embolectomy procedure further sparked interest in HCAS findings on CT, because it showed that clots composed mostly from red blood cells were more often visible as HCAS on CT scans than fibrin-dominant clots.9 This analysis is in line with a previous in vitro study that concluded that Hounsfield units are lower in platelet-rich thrombi than in erythrocyte-rich thrombi.10 If unenhanced CT can truly reflect clot composition, then detection of HCAS could lead to some very important clinical implications, such as predicting the efficacy of thrombolytic treatment. Therefore, the goal of the present study was to identify factors associated with HCAS on admission CT using a large international data set of acute stroke patients treated with intravenous thrombolysis. The second goal was to provide data about variability in HCAS detection amongst stroke centers. We hypothesized that stroke subtype or etiology, laboratory parameters, time from symptom onset to imaging, and several other factors would potentially influence the presence of HCAS through clot composition. Our second hypothesis was that stroke centers would differ in reporting of HCAS.
Methods Data on all the patients from the Safe Implementation of Treatment in Stroke-EAST (SITS-EAST) Register between February 28, 2003, and December 31, 2011, were analyzed. The SITS-International Stroke Thrombolysis Registry11,12 and SITS-EAST methodology13 were described in previous publications. Patients were included into the register if they had acute ischemic stroke and had received intravenous alteplase at a dose of .9 mg/kg. The time window from symptom onset to treatment was initially 3 hours and was extended to 4.5 hours after October 2008.11,14
Unenhanced CT scan in stroke thrombolysis is required on admission and at follow-up 24-36 hours after thrombolytic treatment. CT angiography was optional. Only the cases with confirmed baseline data and admission CT results were included in the analysis. All scans were evaluated locally, and the presence of HCAS on plain CT (implying, but not limited to dense middle cerebral artery signs) and arterial occlusion on CT angiography (reported in the SITS database without specification of occlusion site) was recorded separately. Neurologic deficit was assessed by National Institutes of Health Stroke Scale (NIHSS) score. NIHSS of 0 means no neurologic deficit; the higher the NIHSS score, the more severe the neurologic deficit (severe strokes have NIHSS score around 20). Cardioembolic stroke was defined as an atrial fibrillation or International Classification of Diseases, version 10 (ICD 10), diagnosis I63.4 in accordance with the Trial of Org 10172 in Acute Stroke Treatment definition.15 For sensitivity analysis, cardioembolic stroke was alternatively defined as an atrial fibrillation or as ICD 10, diagnosis I63.4. Other demographic and clinical characteristics used in our analysis included sex, age, systolic blood pressure, current smoking status, baseline blood glucose, past medical history (arterial hypertension, diabetes mellitus, hyperlipidemia, congestive heart failure, and use of antiplatelets), presence of early ischemic changes on CT, onset to imaging time, number of recombinant tissue plasminogen activator treatments in the relevant center per year, and year of treatment. Three CT-related variables (CT slice thickness, CT scanner producer, and year of CT scanner production) were obtained from an additional online questionnaire distributed among participating stroke centers. The need for ethical approval or patient consent for participation in the SITS-EAST register varied among the participating countries, but approvals were obtained in those countries in which it was a requirement. The Ethics Committee of the Karolinska Institute in Stockholm approved the SITS-MOnitoring STudy.
Statistics Patients were stratified according to the presence of HCAS on admission CT scan. Demographic and baseline characteristics were compared between groups of HCAS patients and non-HCAS patients by means of descriptive statistics and the t test or Wilcoxon test (if normality was seriously violated) for continuous parameters and the
DENSE ARTERY SIGN DEPENDS ON STROKE SEVERITY
chi-square test for binomial or categorical variables. Potential nonlinear relationships were explored by plotting each continuous/ordinal variable against the presence of HCAS. Variables independently associated with HCAS were identified by general estimating equations (GEEs). To identify variables independently associated with HCAS, a GEE was performed. This method, compared with, for example, logistic regression, accounts for within-center clustering of the data, limiting the effect of intercenter variability on the analysis of factors associated with HCAS. The variables entered in the model were either associated with HCAS at P less than .10 or clinically important variables (age and systolic blood pressure). Only cases with confirmed baseline data and admission CT results were included in the GEE (the numbers of available cases for each variable are listed in Table 1). Because CT angiography was optional, CT angiography data were treated as a 3-category variable: (1) CT angiography not done; (2) occlusion documented on CT angiography; or (3) absence of occlusion on CT angiography.
3
The analysis was performed on all patients who had available data for the presence of HCAS on CT scan. Next, a subgroup analysis was performed on only those patients who had occlusion confirmed on CT angiography. Commonly used goodness-of-fit statistics (quasi-likelihood under independence model criterion)16 were used to fit the final model among the competing models. The statistical significance level was set at less than .05.
Results Of all 8878 consecutive patients, 8375 patients (94%) with available information about HCAS were included in our analysis. Of these, HCAS was present in 1553 patients (19%), median baseline NIHSS score was 12, mean age was 67 6 12 years, and 3592 patients (43%) were females. CT angiography was available in 2540 patients (30%). The baseline characteristics of patients with and without HCAS on admission CT are summarized in Table 1. Several differences were noticed between the 2
Table 1. Characteristics of all patients and patients with presence and absence of hyperdense cerebral artery sign (HCAS) All Characteristics
*N
Sex, female, n (%) Age, mean (SD), y Baseline NIHSS score, median (Q1-Q3) Systolic blood pressure, mean (SD), mm Hg Glucose, median (Q1-Q3), mmol/L Early ischemic changes on CT, n (%) Congestive heart failure, n (%) Hypertension, n (%) Hyperlipidemia, n (%) Diabetes mellitus, n (%) Current smoking, n (%) Use of aspirin, dipyridamole, or clopidogrel before stroke, n (%) Onset to imaging time, median (Q1-Q3), min Cardioembolic stroke,z n (%) Vessel occlusion on CT angiography,x n (%) Absence of vessel occlusion on CT angiography,x n (%) Number of rt-PA treatments per center per y, median (Q1-Q3) Year of treatment, median
8375 8375 8100 8199 8105 7942 8173 8234 7291 8248 8375 8375
8375 (100%) 3592 (43) 67 (12) 12 (8-17) 153 (21) 7 (6-8) 869 (11) 1037 (13) 6107 (74) 2583 (35) 1841 (22) 1814 (22) 2442 (29)
HCAS absent *N 6822 6822 6646 6699 6607 6521 6668 6719 5951 6731 6822 6822
6822 (82%) 2876 (42) 67 (12) 11 (7-16) 153 (21) 7 (6-8) 643 (10) 862 (13) 4991 (74) 2095 (35) 1520 (23) 1510 (22) 1999 (29)
HCAS present 1553 (19%)
P valuey
1553 716 (46) 1553 67 (12) 1454 16 (12-19) 1500 153 (20) 1498 7 (6-8) 1421 226 (16) 1505 175 (12) 1515 1116 (74) 1340 488 (36) 1517 321 (21) 1553 304 (20) 1553 443 (29)
.01 .995 ,.001# 0.1 .92# ,.001 .17 .62 .40 .23 .03 .54
*N
8001 93 (70-125) 8375 3054 (37) 8375 1467 (18) 8375 1073 (13)
6521 94 (70-127) 6822 2409 (36) 6822 969 (14) 6822 984 (14)
1480 1553 1553 1553
90 (67-120) 645 (42) 498 (32) 89 (6)
.001# ,.001 ,.001 ,.001
8375
25 (13-52)
6822
25 (13-48)
1553
27 (14-56)
.001
8375
2009
6822
2009
1553
2009
,.001#
Abbreviations: CT, computed tomography; ICD, International Classification of Diseases; NIHSS, National Institutes of Health Stroke Scale; rt-PA, recombinant tissue plasminogen activator; SD, standard deviation; TOAST, Trial of Org 10172 in Acute Stroke Treatment. *Number of available values are presented in column N. As data are considered as randomly missing, the estimation of percentages and the chi-square test were based on reported cases only. yP value of test comparing characteristics of patients with HCAS present versus absent: the t test or Wilcoxon test (marked by #) was used for continuous parameters and the chi-square test for binomial or categorical parameters. zCardioembolic stroke was defined as an atrial fibrillation or ICD 10, diagnosis I63.4 in accordance with the TOAST definition. xCT angiography was optional. CT angiography was done for 2540 patients (ie, 30% of 8735 patients). CT angiography not done was treated as a category (not as a missing value).
ET AL. J. NOVOTNA
4
groups: initial stroke severity (median NIHSS score) was higher in HCAS patients than in those without HCAS. An infarct was seen more on admission CT scan in patients with HCAS than in those without. Patients with HCAS had a higher frequency of vessel occlusion on CT angiography and were more frequently diagnosed with cardioembolic stroke. CT scanners used in participating centers (data available from 89 of 145 centers) were from 4 producers: Siemens (41%), General Electric (22%), Philips (21%), and Toshiba (15%). The median CT scanner year of production was 2007 (interquartile range, 2004-2009; range from 1997 to 2011). The median of CT slice thickness for unenhanced CT was 3 mm (interquartile range, 2-5). Table 2 lists the results of the multivariate analysis to identify variables independently associated with HCAS. Of 8375 patients, 794 (9%) were eliminated because of incomplete information about baseline NIHSS score (n 5 635), systolic blood pressure (n 5 536), or early ischemic CT changes (n 5 433), leaving 7581 patients (91%) for the final complete case analysis. HCAS was
independently associated with baseline NIHSS (odds ratio [OR], 1.11; 95% confidence interval [CI], 1.10-1.12), vessel occlusion (OR, 5.02; 95% CI, 3.31-7.63), early ischemic CT changes (OR, 1.63; 95% CI, 1.31-2.04), year of treatment (OR, 1.07; 95% CI, 1.02-1.12), and age (OR, .90; 95% CI, .84-.96). Stroke due to cardioembolism was not independently associated with HCAS on admission. The relationship between baseline NIHSS and the presence of HCAS and occlusion on CT angiography is shown in Figure 1. Figure 1 demonstrates that in patients with higher NIHSS score, both HCAS and occlusion on CT angiography were detected more frequently than in patients with lower NIHSS score. In patients with NIHSS score lesser than or equal to 5, the occlusion on CT angiography was still detected in 32% of cases. HCAS was diagnosed in 0%-50% of patients in centers with greater than or equal to 10 treated patients. HCAS was detected in 89 patients (8%) who had no vessel occlusion on baseline CT angiography (n 5 1073), compared to 498 patients (34%) who had occlusion (n 5 1467). In patients with occlusion on baseline CT angiography, the prevalence of
Table 2. Results of univariate and multivariate analysis to identify predictors of hyperdense cerebral artery sign (HCAS; all patients with available data about baseline HCAS on CT, complete case analysis, n 5 8375) Univariate analysis
Multivariate analysisy
Variable
*OR (95% CI)
P valuey
*OR (95% CI)
P valuez
Sex, female Age, by 10-year increments, y Baseline NIHSS score, by 1-point increments Systolic blood pressure, mm Hg Early ischemic changes on CT Glucose, by 5-mmol/L increments Hypertension Diabetes mellitus Hyperlipidemia Congestive heart failure Current smoking Use of aspirin, dipyridamole, or clopidogrel before stroke Onset to imaging time, min Cardioembolic strokex Vessel occlusion on CT angiography Number of rt-PA treatments per center per y Number of rt-PA treatments per center, by 100-patient increments Year of treatment, by 1-year increments Center CT slice thickness, mm CT scanner producer Year of CT scanner production
1.17 (1.05-1.31) 1.00 (.96-1.05) 1.12 (1.11-1.14) 1.00 (1.00-1.00) 1.73 (1.47-2.04) .99 (.96-1.01) .97 (.85-1.10) .92 (.80-1.05) 1.05 (.93-1.19) .89 (.75-1.05) .86 (.75-.98) .96 (.85-1.10) 1.00 (1.00-1.00) 1.30 (1.16-1.46) 5.68 (4.5-7.2) 1.00 (1.00-1.01) 1.11 (1.00-1.23) 1.05 (1.02-1.08) NA 1.06 (.97-1.17) NA 1.05 (.98-1.14)
.01 1.00 ,.001 .11 ,.001 .21 .62 .23 .40 .17 .03 .55 .56 ,.001 ,.001 ,.001 .05 ,.001 ,.001 .21 .35 .18
1.08 (.96-1.21) .90 (.84-.96) 1.11 (1.10-1.12) 1.00 (1.00-1.00) 1.63 (1.31-2.04)
.21 .001 ,.001 .49 ,.001
1.04 (.88-1.23)
.66
1.08 (.91-1.28) 5.0 (3.31-7.63)
.37 ,.001
1.03 (.94-1.13) 1.07 (1.02-1.12)
.56 ,.001
Abbreviations: CI, confidence interval, CT, computed tomography; ICD, International Classification of Diseases; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio; rt-PA, recombinant tissue plasminogen activator; TOAST, Trial of Org 10172 in Acute Stroke Treatment. *OR: probability modeled is for the presence of HCAS on CT. yUnivariate analysis was performed using logistic regression; CT slice thickness, CT scanner producer, and year of CT scanner production were analyzed with generalized estimating equations. Multivariate analysis was performed with generalized estimating equations. zP value of the Wald chi-square test of parameter significance. xCardioembolic stroke was defined as an atrial fibrillation or ICD 10, diagnosis I63.4 in accordance with the TOAST definition.
DENSE ARTERY SIGN DEPENDS ON STROKE SEVERITY
Figure 1. The relationship between baseline National Institutes of Health Stroke Scale (NIHSS; y axis) and the presence of HCAS and occlusion on computed tomography (CT) angiography (x axis). In patients with higher NIHSS score, both hyperdense cerebral artery sign (HCAS) and occlusion on CT angiography were detected more frequently. Occlusion on CT angiography was detected in 91 cases (32%) of 283 patients with NIHSS score less than or equal to 5. Estimation of percentages and confidence intervals was based on reported cases only (8100 patients with known data about baseline NIHSS and HCAS and 2376 patients with known data about baseline NIHSS and CT angiography).
HCAS in centers with greater than or equal to 10 treated patients was 34%, and ranged from 10% to 86%. According to our additional sensitivity analysis, when baseline NIHSS score (but no other variable) is excluded from the final model, variable cardioembolic stroke (defined as I63.4 or atrial fibrillation) becomes significantly associated with the presence of HCAS (OR, 1.23; 95% CI, 1.03-1.47). Using alternative definitions of cardioembolic stroke did not change the results—cardioembolic stroke was not associated with the presence of HCAS. A subgroup analysis on patients with the presence of occlusion on CT angiography (complete case analysis, n 5 1212) showed that HCAS on CT was independently associated with baseline NIHSS score (OR, 1.08; 95% CI, 1.06-1.10), early ischemic CT changes (OR, 1.67; 95% CI, 1.12-2.49), baseline serum glucose (OR, .78; 95% CI, .62.98), and age (OR, .99; 95% CI, .98-1.00). As in overall analysis, stroke due to cardioembolism was not independently associated with HCAS on admission. The results of a subgroup univariate and multivariate regression analysis are available in Table S1 in Appendix.
Discussion We found that HCAS was present in 19% of all patients in the SITS-EAST registry. Previous studies reported a similar or slightly higher prevalence of HCAS (15%32%).6,17-19 In our study, a higher prevalence of HCAS was observed in some centers, whereas other centers never diagnosed HCAS (range, 0%-50%). When considering only the patients with occlusion on CT angiography, HCAS was diagnosed in up to 89% of cases in some centers. We suggest that a large intercenter variation in HCAS detection on CT is not
5
related to the fact that patients in some centers differ in their biological characteristics, but rather to the fact that quality of care, imaging expertise, or scanning protocols differ among stroke centers. Our data reflect the limitations of diagnosing HCAS in routine clinical practice, based on qualitative criteria with low interobserver agreement (Kappa statistic, .36-.55).20-24 In our study, the important finding was that patients with HCAS had higher baseline stroke severity and the presence of clot on CT angiography. As shown in Figure 1, with increasing baseline NIHSS score, there was a linear increase in the chance of diagnosing HCAS. Because a higher NIHSS score mostly reflects larger thrombus size,25,26 the simplest interpretation could be that a clot must be present and must be big enough to be visible as HCAS. Although our data did not contain any information about several well-established predictors of stroke severity, such as core or penumbra size, clot location, and collateral status, our interpretation about the importance of clot size is strongly supported by 4 previous smaller studies on patients with CT angiography confirmed intracranial occlusion. In one of these studies (n 5 78),27 thrombus volumes were significantly larger in patients with HCAS than in those without this sign. In this and in the 3 other studies (n 5 45,28 n 5 58,29 and n 5 5430), slice thickness of less than or equal to 1.25 mm on CT scans had 100% sensitivity to diagnose HCAS. Consequently, the detection of smaller clots on unenhanced CT might be overcome by decreasing slice thickness and increasing expertise. Our data, however, did not show an association between CT slice thickness and HCAS detection. The value of expertise is supported by our next finding; with every year, there were better odds of detecting HCAS. In our study, the next variable associated with HCAS was early ischemic changes on CT. This finding can have 2 interpretations. First, the presence of early ischemic changes (eg, focal swelling or parenchymal hypoattenuation) has been reported to be determined by the duration and the level of cerebral hypoperfusion.31 Therefore, it is possible that early ischemic changes are, in addition to NIHSS scores, markers of clot size. Alternatively, some physicians may have regarded the presence of HCAS as one of the manifestations of ‘‘early ischemic changes,’’ which would create a false positive association between these 2 variables. Another finding in our study is that older patients had a lower probability of being diagnosed with HCAS. We hypothesize that this association between age and HCAS can be mediated by processes related to the aging of the brain, such as cerebral atrophy, leukoaraiosis, or vessel tortuosity. This finding was also reported in a previous SITS analysis (n 5 10,023).6 However, another smaller, single-center study (n 5 1,010)19 showed an opposite relationship between age and the presence of
ET AL. J. NOVOTNA
6
HCAS, probably because of different study populations (an unselected general acute ischemic stroke cohort vs. patients treated with thrombolysis). One of the most interesting results of our study, contradicting the findings of previous studies, is that stroke mechanism is not independently associated with the presence of HCAS. Specifically, cardioembolic strokes are not more likely to have HCAS on admission CT, independently from other baseline variables. According to our analysis, although patients with cardioembolic stroke are more likely to have HCAS, this is rather a consequence of their higher baseline stroke severity, and we suggest that it is not due to differences in clot composition. However, from this study alone, we cannot determine the relationship between thrombus density and stroke subtypes.32 A secondary important result of our analysis is that 6% of the patients with low NIHSS score (,5) had HCAS, and 36% had occlusion on baseline CT angiography. Such findings support the evidence that low neurologic deficit by itself may not be sufficient justification for excluding patients from thrombolysis.33-35 The main limitation of our study is that our data, like other data in the SITS registry, were not externally validated. For the future, it would be very useful to collect imaging in the registries, to allow for the validation of imaging data. The strength of our analysis is that the predictors of HCAS remained basically the same, regardless of whether the study population was limited to the subgroup with CT angiography confirmed occlusion. The only minor disparity (because of different patient populations) was that the association between glucose and HCAS was present only in a subgroup analysis, and the association between year of treatment and HCAS was present only in the overall analysis. In conclusion, our study underscores the differences in the detection of HCAS among stroke centers in routine clinical practice. Given the importance of the HCAS finding, it could be useful to standardize diagnosis of HCAS and possibly to implement quantitative detection methods. Accounting for within-center data clustering, we found that the presence of HCAS depends on stroke severity rather than stroke subtype. Our results indirectly challenge the hypothesis that clot composition affects the visibility of the clot on CT. Our study provides further evidence, using a large thrombolytic registry, that unenhanced CT should be explored as a noninvasive, fast, relatively inexpensive, and widely available tool for clot imaging.
Supplementary Data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014. 04.034.
References 1. Gacs G, Fox AJ, Barnett HJ, et al. CT visualization of intracranial arterial thromboembolism. Stroke 1983;14:756-762. 2. Goldmakher GV, Camargo EC, Furie KL, et al. Hyperdense basilar artery sign on unenhanced CT predicts thrombus and outcome in acute posterior circulation stroke. Stroke 2009;40:134-139. 3. Jensen UR, Weiss M, Zimmermann P, et al. The hyperdense anterior cerebral artery sign (HACAS) as a computed tomography marker for acute ischemia in the anterior cerebral artery territory. Cerebrovasc Dis 2010;29:62-67. 4. Barber PA, Demchuk AM, Hudon ME, et al. Hyperdense sylvian fissure MCA ‘‘dot’’ sign: a CT marker of acute ischemia. Stroke 2001;32:84-88. 5. Shetty SK. The MCA dot sign. Radiology 2006;241:315-318. 6. Kharitonova T, Ahmed N, Thoren M, et al. Hyperdense middle cerebral artery sign on admission CT scan–prognostic significance for ischaemic stroke patients treated with intravenous thrombolysis in the safe implementation of thrombolysis in Stroke International Stroke Thrombolysis Register. Cerebrovasc Dis 2009;27:51-59. 7. Manelfe C, Larrue V, von Kummer R, et al. Association of hyperdense middle cerebral artery sign with clinical outcome in patients treated with tissue plasminogen activator. Stroke 1999;30:769-772. 8. Tomsick T, Brott T, Barsan W, et al. Prognostic value of the hyperdense middle cerebral artery sign and stroke scale score before ultraearly thrombolytic therapy. AJNR Am J Neuroradiol 1996;17:79-85. 9. Liebeskind DS, Sanossian N, Yong WH, et al. CT and MRI early vessel signs reflect clot composition in acute stroke. Stroke 2011;42:1237-1243. 10. Kirchhof K, Welzel T, Mecke C, et al. Differentiation of white, mixed, and red thrombi: value of CT in estimation of the prognosis of thrombolysis phantom study. Radiology 2003;228:126-130. 11. Wahlgren N, Ahmed N, Eriksson N, et al. Multivariable analysis of outcome predictors and adjustment of main outcome results to baseline data profile in randomized controlled trials: Safe Implementation of Thrombolysis in Stroke-MOnitoring STudy (SITS-MOST). Stroke 2008; 39:3316-3322. 12. Wahlgren N, Ahmed N, Davalos A, et al. Thrombolysis with alteplase for acute ischaemic stroke in the Safe Implementation of Thrombolysis in Stroke-Monitoring Study (SITS-MOST): an observational study. Lancet 2007;369:275-282. 13. Mikulik R, Kadlecova P, Czlonkowska A, et al. Factors influencing in-hospital delay in treatment with intravenous thrombolysis. Stroke 2012;43:1578-1583. 14. Hacke W, Kaste M, Bluhmki E, et al. Thrombolysis with alteplase 3 to 4.5 hours after acute ischemic stroke. N Engl J Med 2008;359:1317-1329. 15. Adams HP Jr, Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 1993;24:35-41. 16. Pan W. Model selection in estimating equations. Biometrics 2001;57:529-534. 17. Abul-Kasim K, Brizzi M, Petersson J. Hyperdense middle cerebral artery sign is an ominous prognostic marker despite optimal workflow. Acta Neurol Scand 2010; 122:132-139. 18. Qureshi AI, Ezzeddine MA, Nasar A, et al. Is IV tissue plasminogen activator beneficial in patients with hyperdense artery sign? Neurology 2006;66:1171-1174.
DENSE ARTERY SIGN DEPENDS ON STROKE SEVERITY 19. Paciaroni M, Agnelli G, Floridi P, et al. Hyperdense middle cerebral and/or internal carotid arteries in acute ischemic stroke: rate, predictive factors and influence on clinical outcome. Cerebrovasc Dis 2011;32:239-245. 20. Barber PA, Darby DG, Desmond PM, et al. Identification of major ischemic change. Diffusion-weighted imaging versus computed tomography. Stroke 1999;30: 2059-2065. 21. Fiebach JB, Schellinger PD, Jansen O, et al. CT and diffusion-weighted MR imaging in randomized order: diffusion-weighted imaging results in higher accuracy and lower interrater variability in the diagnosis of hyperacute ischemic stroke. Stroke 2002;33:2206-2210. 22. Marks MP, Holmgren EB, Fox AJ, et al. Evaluation of early computed tomographic findings in acute ischemic stroke. Stroke 1999;30:389-392. 23. Schriger DL, Kalafut M, Starkman S, et al. Cranial computed tomography interpretation in acute stroke: physician accuracy in determining eligibility for thrombolytic therapy. JAMA 1998;279:1293-1297. 24. Tomsick TA, Brott TG, Chambers AA, et al. Hyperdense middle cerebral artery sign on CT: efficacy in detecting middle cerebral artery thrombosis. AJNR Am J Neuroradiol 1990;11:473-477. 25. Lewandowski CA, Frankel M, Tomsick TA, et al. Combined intravenous and intra-arterial r-TPA versus intraarterial therapy of acute ischemic stroke: Emergency Management of Stroke (EMS) Bridging Trial. Stroke 1999;30:2598-2605. 26. Fischer U, Arnold M, Nedeltchev K, et al. NIHSS score and arteriographic findings in acute ischemic stroke. Stroke 2005;36:2121-2125.
7 27. Kim EY, Yoo E, Choi HY, et al. Thrombus volume comparison between patients with and without hyperattenuated artery sign on CT. AJNR Am J Neuroradiol 2008; 29:359-362. 28. Kim EY, Lee SK, Kim DJ, et al. Detection of thrombus in acute ischemic stroke: value of thin-section noncontrastcomputed tomography. Stroke 2005;36:2745-2747. 29. Riedel CH, Jensen U, Rohr A, et al. Assessment of thrombus in acute middle cerebral artery occlusion using thin-slice nonenhanced computed tomography reconstructions. Stroke 2010;41:1659-1664. 30. Riedel CH, Zoubie J, Ulmer S, et al. Thin-slice reconstructions of nonenhanced CT images allow for detection of thrombus in acute stroke. Stroke 2012;43:2319-2323. 31. Hirano T, Yonehara T, Inatomi Y, et al. Presence of early ischemic changes on computed tomography depends on severity and the duration of hypoperfusion: a single photon emission-computed tomographic study. Stroke 2005;36:2601-2608. 32. Niesten JM, van der Schaaf IC, Biessels GJ, et al. Relationship between thrombus attenuation and different stroke subtypes. Neuroradiology 2013;55:1071-1079. 33. Meretoja A, Putaala J, Tatlisumak T, et al. Off-label thrombolysis is not associated with poor outcome in patients with stroke. Stroke 2010;41:1450-1458. 34. Kruetzelmann A, Siemonsen S, Gerloff C, et al. Thrombolysis targeting MRI defined tissue at risk in minor stroke. J Neurol Neurosurg Psychiatry 2009;80:1156-1158. 35. Mikulik R, Goldemund D, Reif M, et al. Outcome of patients with negative CT angiography results for arterial occlusion treated with intravenous thrombolysis. Stroke 2009;40:868-872.