Hyperdense Basilar Artery Sign on Unenhanced CT Predicts Thrombus and Outcome in Acute Posterior Circulation Stroke
Background and Purpose— In acute stroke patients, the presence of a hyperdense middle cerebral artery sign on unenhanced CT is a specific but insensitive indicator of acute thrombosis. Our purpose was to determine whether the hyperdense basilar artery (HDBA) sign has utility in detecting thrombosis and predicting outcome in patients presenting with signs and symptoms of posterior circulation stroke.
Methods— Unenhanced CT scans obtained within 24 hours of symptom onset in 95 patients with suspected posterior circulation stroke were reviewed. Three neuroimagers blinded to clinical outcome and results of the concurrent CT angiography (which served as the reference standard) rated presence of HDBA sign on a 5-point scale for level of certainty (1=definitely absent; 5=definitely present). Receiver operating characteristic curve analysis was performed. Short-term outcome was measured by discharge National Institute of Health Stroke Scale (NIHSS) scores; long-term outcome was measured by 6-month modified Rankin score (dichotomized, poor outcome defined as modified Rankin score >2). The following variables were correlated with short-term and long-term outcome by univariate analysis: HDBA sign, age, sex, time from stroke onset to imaging, admission NIHSS, history of stroke/TIA, atrial fibrillation, coronary artery disease, hypertension, diabetes, hypercholesterolemia, tobacco use, and thrombolysis. Variables showing correlation with P<0.1 were included in multiple regression analysis.
Results— Using a level of certainty cutoff score of ≥4 (probable, definite), HDBA sign had 71% sensitivity, 98% specificity, 94% accuracy, 83% positive predictive value, and 95% negative predictive value for basilar artery occlusion. In univariate analysis, factors significantly correlated with discharge NIHSS were: admission NIHSS (P<0.0001; r=0.77), HDBA sign (P=0.01), and diabetes (P=0.02). Factors showing significant correlation or association with poor long-term outcome were age (P=0.02), admission NIHSS (P=0.007), HDBA sign (P=0.02), and history of stroke or TIA (P=0.007). The odds ratio of HDBA sign for predicting poor long-term outcome was 5.3 (95% CI, 1.1–33.3). In multiple regression analysis, the only independent predictors of discharge NIHSS were admission NIHSS (P<0.0001) and HDBA sign (P=0.004). Significant independent predictors of poor long-term outcome were age (P=0.02), admission NIHSS (P=0.008), history of stroke/TIA (P=0.03), and HDBA sign (P=0.05).
Conclusion— In patients presenting with a high pretest probability of posterior circulation stroke based on clinical symptoms, the presence of the HDBA sign on unenhanced CT is a strong predictor of basilar artery thrombosis, and both short- and long-term outcome.
Acute basilar artery occlusion, whether from primary atheromatous disease or embolism, is a devastating event, likely to result in significant morbidity and mortality. Although high-level evidence regarding the role of thrombolytic therapy in such cases is lacking, it is clear that early diagnosis is essential if direct or supportive measures are to be of benefit.1 Computed tomography angiography (CTA) has been shown to be highly accurate in detecting acute basilar artery occlusion,2,3 and other forms of advanced imaging, including MRI with perfusion imaging, have been used to assess the viability of tissue in patients with basilar artery occlusion.4 Despite this, most community medical centers do not yet routinely perform advanced imaging in stroke patients, but rather continue to rely on unenhanced CT for their clinical decision-making. Because assessment of the posterior fossa by CT is notoriously difficult, knowledge of specific imaging signs that facilitate detection of posterior fossa ischemic lesions could be of great clinical value.
In the anterior circulation, the presence of a “hyperdense middle cerebral artery sign” on unenhanced CT has been established as a predictor of vessel occlusion with nearly 100% specificity5 and sensitivity estimates that range from 5% to 50%.6,7 The diagnostic and prognostic value of a hyperdense basilar artery (HDBA) sign has been less well-established, in part because of the heterogeneity of previous study cohorts.1,8 Reversal of hyperdense artery sign (both in middle cerebral and basilar arteries) has been associated with improved outcomes.9 In this study, we assessed the value of the HDBA sign in a homogeneous sample of patients presenting with an acute posterior circulation syndrome. Our purpose was to determine whether the HDBA sign has utility in detecting thrombosis and predicting outcome in patients presenting with signs and symptoms of posterior circulation stroke.
Materials and Methods
We analyzed data from 742 consecutive patients enrolled in a prospective cohort study at 2 university-based hospitals, the Screening Technology and Outcomes Project in Stroke (STOPStroke), in which emergency CTA for patients suspected of having ischemic stroke (stroke, TIA, or stroke mimics) was performed within 24 hours of symptoms onset. In STOPStroke, admission unenhanced CT scans were obtained in all subjects, followed immediately by CTA. Patients were excluded if iodinated contrast agent administration was contraindicated (ie, history of contrast agent allergy, pregnancy, congestive heart failure, increased baseline creatinine level) or if there was evidence of intracranial hemorrhage on unenhanced CT scans.
Acute stroke history, demographic data, medical history, and admission NIHSS scores were obtained as part of patient admission work-up. NIHSS scores were obtained at discharge (patients who died during admission were assigned a discharge NIHSS score of 42). Follow-up modified Rankin scale (mRS) scores were obtained at 6 months. This interval was chosen to maximize functional recovery (which may occur between 3 and 6 months) while minimizing loss to follow-up and non–stroke-related deaths. Patients were classified as having a favorable outcome if mRS was 0 to 2, or a poor outcome if mRS was >2.
The prospective STOPStroke study and our current analysis received institutional review board approval at both participating institutions and were compliant with the Health Insurance Portability and Accountability Act. At enrollment for the prospective study, all subjects gave informed consent for participation in both the prospective study and any future retrospective studies using STOPStroke data.
For this analysis, patients were excluded if no clinical follow-up information could be obtained either by phone interview or review of medical records, if NIHSS scores at admission or discharge were not available, or if the time from symptom onset to imaging could not be reliably established (eg, “wake-up” strokes). The remaining patients were classified by clinical syndrome as defined by the Oxfordshire classification, and only patients who presented with a posterior circulation syndrome were selected.10 Of the 742 patients examined, 97 were excluded from analysis because there was no usable follow-up information. Thirty-one patients were excluded because no discharge NIHSS score was available. Two patients were excluded because no admission NIHSS score was available. Three patients were excluded because the time from symptom onset to imaging could not be reliably established. Of the 609 remaining patients, 95 had a documented posterior circulation syndrome, comprising our study patient population (in 173, information on the clinical syndrome was not available, and 341 had another clinical syndrome). The cohort analyzed thus consists of prospectively recruited patients whose clinical presentation suggested posterior circulation etiology.
Unenhanced CT and CT angiographic acquisitions were performed according to standard departmental protocols with 16- or 64-section multidetector CT scanners (LightSpeed; GE Healthcare).11 Unenhanced CT was performed, with the patient in a head holder, in the transverse plane. Representative sample parameters, with minimal variations between scanners and sites shown as ranges, were as follows: 120 to140 kVp, 170 mA, 2-second scan time, and 5-mm section thickness. Imaging with these parameters was immediately followed by biphasic helical scanning, performed at the same head tilt as was unenhanced CT. CTA was performed after a 25-second delay (40 seconds for patients in atrial fibrillation) and administration of 100 to 140 mL of a nonionic contrast agent (Isovue; Bracco Diagnostics) at an injection rate of 3 mL/s by using a power injector (Medrad Power Injector; Medrad) via an 18-gauge intravenous catheter. Parameters were 140 kVp, 220 to 250 mA, 0.8 to 1.0-second rotation time, 2.5-mm section thickness, 1.25-mm reconstruction interval, 3.75 mm per rotation table speed, and 0.75:1 pitch. Images were obtained from the C6 vertebral body level through the circle of Willis. Immediately afterward, a second set of images was obtained from the aortic arch to the skull base. Afterward, source images were reconstructed into standardized maximum intensity projection views of the intracranial and extracranial vasculature.
Image review was independently performed on a picture archiving and communication system workstation (Impax; Agfa Technical Imaging Systems) by a board-certified neuroradiologist and clinical neurologists experienced in stroke neuroimaging interpretation (M.H.L. E.C.S.C., and W.J.K.). Reviewers were blinded to follow-up clinical and imaging findings but had information on the patients’ age, sex, and presenting clinical symptoms. None of the reviewers had participated in patient recruitment. Variable window width and center level settings were used for optimal ischemic hypoattenuation detection with unenhanced CT.12 On unenhanced CT images, all major vessels in the circle of Willis were assessed for the presence of a dense vessel sign. The presence of a HDBA sign was rated according to a 5-point level of certainty scale (5, definitely present; 4, probably present; 3, equivocal; 2, probably absent; 1, definitely absent). After the unenhanced CT, readers assessed the presence of partial or complete basilar artery occlusion by thrombus on CTA according to the same certainty scale. For this analysis, we dichotomized their readings as indicating the presence of basilar artery thrombosis if they rated it 4 or 5 on the CTA. Disagreements in readings were resolved in consensus.
To evaluate the test performance characteristics of the HDBA sign in identifying basilar artery occlusion, with occlusion on CTA as the gold standard, we used standard receiver operating characteristic curve analysis methods (Rockit 0.9B, beta version, 1998; University of Chicago).13 Sensitivity, specificity, accuracy, positive predictive value, and negative predictive value for different certainty levels were calculated.
Univariate and multiple regression analyses were performed using commercial software (SAS 9.1, 2002–2003; SAS Institute). Continuous input variables included in the analyses were age, admission NIHSS score, and time from symptom onset to imaging. Categorical input variables included sex, the presence of a HDBA on unenhanced CT, thrombolysis (intravenous, intra-arterial, or both), a history of stroke or TIA, atrial fibrillation, coronary artery disease, hypertension, diabetes mellitus, hypercholesterolemia, and tobacco use (current or past). Output variables were discharge NIHSS and the 6-month mRS (dichotomized into favorable and unfavorable, as described). Variables with P<0.1 on univariate analysis were included in the multiple regression analysis.
For continuous input variables, effectiveness in predicting unfavorable outcome (6-month mRS >2) was assessed with a t test or the Wilcoxon rank sum test, and correlation with discharge NIHSS was assessed using either Pearson or Spearman correlation, as appropriate. For categorical input variables, univariate analysis was performed using either the χ2 test or the Fisher exact test for prediction of 6-month mRS >2, and t test or Wilcoxon rank sum test for prediction of discharge NIHSS, as appropriate. In the multiple regression analyses, correlation was assessed with linear regression for continuous outcome variables, and logistic regression was assessed for categorical outcome variables.
Mean age was 65.5±16.3 years. Fifty-seven (60%) patients were male. Mean time from onset to imaging was 10 hours and 34 minutes (SD, 10 hours 53 minutes). Fourteen (15%) of the patients had partial or complete occlusion of the basilar artery demonstrated by CTA. All but 1 of these patients had posterior circulation infarcts confirmed by parenchymal imaging (MR or CT); the patient who did not have a posterior parenchymal infarct had partial occlusion on CTA. Twelve patients had a HDBA sign. Seven (7%) received thrombolytic therapy (2 intra-arterial, 3 intravenous, and 2 a combination of intravenous followed by intra-arterial thrombolysis). Table 1 summarizes the characteristics of the patients. Patient age showed a nearly normal distribution, whereas admission NIHSS and the time to imaging showed nonparametric distributions. Three patients died before discharge. Catheter angiography was performed in only 6 of the 95 patients, and within 6 hours of the CTA in 3 of these (the remainder had angiography between 21 hours and 4 days after the CTA). In all 3 patients, the early angiographic findings confirmed the CTA results.
The receiver operating characteristic curve for basilar artery occlusion based on dense vessel sign is shown in Figure 1. A level of certainty cutoff score of 4 (probably present) corresponded to the optimal operating point of the receiver operating characteristic curve, with sensitivity of 71.4%, specificity of 97.5%, accuracy of 93.7%, positive predictive value of 83.3%, and negative predictive value of 95.2%. This cutoff value was used to define the presence of the HDBA sign for the univariate and multivariate analyses. Figure 2 demonstrates a sample case of a patient with a basilar artery occlusion. Figure 3 demonstrates a sample case of a patient with a false-positive HDBA sign.
Results of univariate analyses for correlation with short-term outcome (discharge NIHSS score) and poor long-term outcome (mRS >2 at 6 months) are shown in Tables 2 and 3⇓. The presence of a HDBA showed significant correlations with both short-term outcome (P=0.01) and poor long-term outcome (OR, 5.6; 95% CI, 1.1–29.3; P=0.02). For short-term outcome, significant correlations were also seen with admission NIHSS (P<0.0001) and a history of diabetes mellitus (P=0.02); thrombolysis showed a trend which approached significance (P=0.06). For long-term outcome, significant correlations were also seen with age (P=0.02), admission NIHSS (P=0.007), and previous stroke or TIA (P=0.007).
In the multiple regression analysis (Tables 4 and 5⇓), statistically significant independent correlations with short-term outcome were seen for the HDBA sign (P=0.004) and admission NIHSS (P<0.0001). For long-term outcome, HDBA, patient age, previous stroke or TIA, and admission NIHSS all showed significant independent predictive value. A history of diabetes and whether a patient received thrombolysis were not independent predictors of either short- or long-term outcomes.
Our data show that the HDBA sign can be useful both diagnostically by predicting the presence of basilar artery thrombosis, and prognostically by predicting both short- and long-term outcomes in patients with a clinical picture suggestive of posterior circulation stroke. Our other findings are clinically intuitive and are in agreement with earlier studies: admission NIHSS scores predicted both short- and long term outcomes, as previously reported.14,15 Patient age showed a correlation with worse long-term outcomes, but not with short-term outcomes. A history of stroke or TIA correlated with long-term outcome, but not with short-term outcome. Other cardiovascular risk factors we examined did not independently predict long- or short-term outcome.
Previous work1 has suggested that the presence of the HDBA sign in patients who received thrombolysis for a known basilar artery occlusion predicted a greater chance of successful treatment, thought to be attributable to the correlation between clot density and acuity, which would presumably correlate with a positive outcome. Because our cohort did not include a substantial number of patients who were scanned within 3 hours of onset, and because we did not examine the rate of recanalization, we cannot address this issue directly. Regardless, the correlation between thrombolysis and short-term outcome approached statistical significance in univariate analysis, but this did not persist in the multiple regression.
Patients in this study were recruited prospectively without regard to anterior or posterior circulation symptoms, and imaging interpretation was performed without knowledge of CTA findings or outcomes. Our analysis focused on those who were suspected of having a posterior circulation stroke syndrome clinically. Because the physicians interpreting the scans had information about the presenting syndrome, this elevated pretest probability probably increased the accuracy of radiological interpretation. However, this does not represent a weakness of our analysis, because in the clinical setting those who interpret the scan should be, and usually are, aware of the clinical presentation,16 and the analyzed cohort represents a realistic clinical scenario, to which the results of our analysis are directly applicable.
One limitation of our analysis is that not every factor that may be relevant for prediction of outcome was included in our models (such as admission blood glucose, for which a diagnosis of diabetes is an imperfect surrogate). Another limitation is that the majority of our CTA results could not be verified by reference to another vascular imaging modality and thus may not reflect a true “gold standard.” Unfortunately, because of the small number of patients (14) who had basilar artery thrombosis, our study did not have enough power to yield information on the predictive value of the HDBA sign in patients who have CTA-proven basilar artery occlusion. Thus, in institutions where CTA is performed routinely during a stroke workup, our findings cannot be used to support any additional role for this finding on unenhanced CT.
In addition to prediction of outcome, the greatest utility of our results is in the situation in which a patient presents with a posterior stroke syndrome to a facility that does not routinely perform CTA for all stroke patients. In such circumstances, assessment of the posterior fossa should include careful scanning for a dense basilar sign. The presence of this sign, in the clinically relevant setting, would indicate the possible need for urgent CTA or transfer to a facility with more extensive diagnostic and treatment resources.
In conclusion, in patients presenting with a high pretest probability of posterior circulation stroke, the presence of a hyperdense basilar artery sign on unenhanced CT is an accurate predictor of basilar artery thrombosis and both short- and long-term outcome.
The authors thank Vadim Frenkel, Hong Yu, Jean-Fresnel Josaphat, and Anthony Ranson for their invaluable assistance in database management.
Sources of Funding
This work was supported by National Institutes of Health grant AHRQ RO1 HS11392 (PI–Karen Furie, MD).
G.V.G. and E.C.S. contributed equally to this work.
- Received April 2, 2008.
- Revision received May 27, 2008.
- Accepted June 12, 2008.
Arnold M, Nedeltchev K, Schroth G, Baumgartner RW, Remonda L, Loher TJ, Stepper F, Sturzenegger M, Schuknecht B, Mattle HP. Clinical and radiological predictors of recanalisation and outcome of 40 patients with acute basilar artery occlusion treated with intra-arterial thrombolysis. J Neurol Neurosurg Psychiatry. 2004; 75: 857–862.
Brandt T, Knauth M, Wildermuth S, Winter R, von Kummer R, Sartor K, Hacke W. Ct angiography and doppler sonography for emergency assessment in acute basilar artery ischemia. Stroke. 1999; 30: 606–612.
Lev MH, Farkas J, Rodriguez VR, Schwamm LH, Hunter GJ, Putman CM, Rordorf GA, Buonanno FS, Budzik R, Koroshetz WJ, Gonzalez RG. Ct angiography in the rapid triage of patients with hyperacute stroke to intraarterial thrombolysis: Accuracy in the detection of large vessel thrombus. J Comput Assist Tomogr. 2001; 25: 520–528.
Ostrem JL, Saver JL, Alger JR, Starkman S, Leary MC, Duckwiler G, Jahan R, Vespa P, Villablanca JP, Gobin YP, Vinuela F, Kidwell CS. Acute basilar artery occlusion: Diffusion-perfusion mri characterization of tissue salvage in patients receiving intra-arterial stroke therapies. Stroke. 2004; 35: e30–e34.
Leys D, Pruvo JP, Godefroy O, Rondepierre P, Leclerc X. Prevalence and significance of hyperdense middle cerebral artery in acute stroke. Stroke. 1992; 23: 317–324.
Mandava P, Kent TA. Reversal of dense signs predicts recovery in acute ischemic stroke. Stroke. 2005; 36: 2490–2492.
Lev MH, Gonzalez RG. Ct angiography and ct perfusion imaging. In: Toga AW, Mazziotta JC, eds. Brain mapping: The methods. San Diego: Academic Press; 2002: 427–484.
Adams HP Jr, Davis PH, Leira EC, Chang KC, Bendixen BH, Clarke WR, Woolson RF, Hansen MD. Baseline nih stroke scale score strongly predicts outcome after stroke: A report of the trial of org 10172 in acute stroke treatment (toast). Neurology. 1999; 53: 126–131.
Frankel MR, Morgenstern LB, Kwiatkowski T, Lu M, Tilley BC, Broderick JP, Libman R, Levine SR, Brott T. Predicting prognosis after stroke: A placebo group analysis from the national institute of neurological disorders and stroke rt-pa stroke trial. Neurology. 2000; 55: 952–959.