Article Text

See original article:

Download PDFPDF

The probability of middle cerebral artery MRA flow signal abnormality with quantified CT ischaemic change: targets for future therapeutic studies
  1. P A Barber1,
  2. A M Demchuk1,
  3. M D Hill1,2,
  4. J H Warwick Pexman1,
  5. M E Hudon1,3,
  6. R Frayne1,4,
  7. A M Buchan1
  1. 1Calgary Stroke Program, Department of Clinical Neuroscience, Seaman Family Magnetic Research Centre, 1403-29 St. NW, Calgary, AB, Canada
  2. 2Department of Medicine and Department of Community Health Sciences, University of Calgary, Calgary, AB, Canada
  3. 3Department of Clinical Radiology, University of Calgary, Calgary, AB, Canada
  4. 4Department of Radiology, University of Calgary, Calgary, AB, Canada
  1. Correspondence to:
 P A Barber Assistant Professor
 Department of Clinical Neurosciences, University of Calgary, Institute for Biodiagnostics (West), Room 153, 3330 Hospital Drive, Calgary, Canada AB T2N 4N1;


Objectives: In this study we define the probability of vascular abnormality in the middle cerebral artery (MCA) territory according to the extent of ischaemic change seen using computed tomography (CT). We assessed the sensitivity and specificity of the hyperdense middle cerebral artery (HMCA) and the “dot” sign using magnetic resonance angiography (MRA).

Methods: Patients presenting with ischaemic stroke had a CT scan (<6 h) prior to MRI (<7 h). A quantitative CT scoring system (ASPECTS) was applied to CT and diffusion weighted images (DWI) at baseline and follow up (24 h) by five independent observers. The presence of HMCA and the MCA “dot” sign was also evaluated. An expert reader assessed the 3D time of flight (TOF) MRA in the anterior circulation for areas of decreased vascular signal in the MCA territory, with an absent signal taken to represent severely reduced or absent flow.

Results: A total of 100 consecutive patients had baseline CT and MR scans. The median NIHSS was 9. The median CT ASPECTS was 8 and equalled the median DWI ASPECTS. There were a total of 10 HMCA and 19 MCA “dot” signs, with four patients having both HMCA and “dot” signs. A total of 47 MRA flow signal abnormalities were observed in the anterior circulation.

Conclusions: In the absence of accessible neurovascular imaging, the extent of CT ischaemia (ASPECTS) is a strong predictor of vascular occlusion. The CT hyperdense artery signs have a high positive predictive value but low negative predictive value.

  • ACAs, anterior cerebral artery
  • ASPECTS, Alberta Stroke Program Early CT Score
  • BA, basilar artery
  • CT, computed tomography
  • DWI, diffusion weighted imaging
  • FLAIR, fluid attenuated inversion recovery
  • HMCA, hyperdense MCA
  • ICA, internal carotid artery
  • MCA, middle cerebral artery
  • MRA, magnetic resonance angiography
  • mRS, modified Rankin scale
  • NIHSS, National Institute of Health Stroke Scale
  • PCA, posterior cerebral artery
  • PROACT-II, ProLyse for Acute Cerebral Thromboembolism trial
  • tPA, tissue plasminogen activator
  • acute stroke
  • computed tomography
  • magnetic resonance angiography
  • vascular occlusion

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Ischaemic stroke is currently the third leading cause of death in the Western world and the leading cause of acquired adult disability.1 In Caucasians, occlusion of extra- or intracranial cerebral arteries by thromboembolic material can be demonstrated early after stroke onset in 80% of stroke patients2 making thrombolysis a rational therapeutic choice. Intravenous tissue plasminogen activator (tPA) remains the only drug and route of administration approved in North America and Europe for acute stroke treatment based on the results of the NINDS tPA trials and is recommended in all disabling ischaemic stroke patients fulfilling the clinical criteria of the NINDS tPA study.3 Other thrombolytic trials have produced neutral results.4–6 These trials had a common methodological design, and enrolled patients with ischaemic stroke by excluding intracranial haemorrhage using computed tomography but did not use any form of neurovascular imaging to identify a large artery occlusion (either extra- or intracranial). In contrast, the ProLyse for Acute Cerebral Thromboembolism trial (PROACT-II)7 randomised a more homogenous population of acute stroke patients with angiographically proven middle cerebral artery occlusions and demonstrated a positive result.

The reality is that a simpler approach relying on clinical parameters and non invasive technology may facilitate more appropriate and widespread use of thrombolysis. The three objectives of this study were: (1) to describe the frequency of magnetic resonance angiographic (MRA) flow signal abnormalities of the internal carotid artery and/or middle cerebral artery in an acute ischaemic stroke population with disabling deficits; (2) to define the probability of intracranial abnormality in the anterior circulation by a quantitative topographical scoring system, the Alberta Stroke Program Early CT Score (ASPECTS)8,9; and (3) to determine the sensitivity and specificity of the hyperdense middle cerebral artery (HMCA)and “dot” signs using MRA as the gold standard.10,11

We speculated that MRA would provide similar anatomical localisation to formal cerebral angiography, that the HMCA sign represented middle cerebral artery occlusion in the main stem of the artery, and the MCA “dot” sign would correlate with occlusion either in the distal portion of the middle cerebral artery or in the M2 branches.12,13


The period of patient recruitment was from October 1999 to July 2001. The principal inclusion criterion for the study was acute disabling ischaemic stroke, measurable by the National Institute of Health Stroke Scale (NIHSS) score, within 6 h of onset. All patients sequentially had a CT brain scan within 6 h followed by MRI within 7 h of onset. Some patients who met accepted criteria were treated with intravenous thrombolysis and the ensuing MR study was performed while the tPA infusion was running. Information collected on subjects included: demographic details; an NIHSS score on initial presentation; time of symptom onset and time at each imaging procedure; and functional outcome at 3 months assessed by the modified Rankin scale (mRS). The mRS was evaluated by a nurse practitioner not involved with the initial individual patient care. Favourable outcome was defined as independence (mRS of 0–2). The study was approved by the local ethics committee, and was part of a study to compare CT with diffusion weighted imaging (DWI) in acute ischaemic stroke.14

All CT scans were performed on fourth generation scanners and were considered to be optimal quality. The standard CT scan protocol was 5 mm slice thickness without contrast enhancement, 120 kV, 170 mA at 2 s, contrast favoured algorithm, inferior orbitomeatal baseline, filmed at appropriate window width and level setting of 80/40 Hounsfield units.9

MR imaging was performed on a 3 T scanner (General Electric Medical Systems, Waukesha, WI) equipped with a standard head quadrature imaging coil and high-speed gradients. The rationale for using 3 T MRI rather than 1.5 T MRI is that the former has better signal to noise ratio.15 The imaging sequences included sagittal T1-weighted, axial diffusion-weighted, and perfusion-weighted imaging, fluid attenuated inversion recovery (FLAIR) imaging, and axial multi-slab 3D time of flight MR angiography (MRA). The isotropic diffusion-weighted images were obtained using a single shot EPI (b = 1000 s/mm2, 19–5 mm thick slices). The imaging parameters for the conventional two-slab 3D TOF sequence were 24/3.3/15° (TR/TE/flip angle), acquisition band with of 12.5 Hz, and a 240×192×46 mm acquired volume with a 256×192×42 matrix, which was then reconstructed to a 512×512×84 matrix. Axial slabs were prescribed from approximately the level of the skull base to the circle of Willis.

Each axial imaging technique (that is, CT and MRI) was read independently by each of five readers, blinded to all clinical and patient information apart from side of stroke symptoms. The rationale for this was to make the imaging interpretation more clinically relevant. The images were read in the following order: baseline CT, FLAIR, and DWI followed by the follow up CT, FLAIR, and DWI. Each imaging modality/technique was interpreted in isolation of other modalities, with a minimum period of 1 week between readings of scans by the same individual.

Each axial CT image (CT) was scored using the Alberta Stroke Program Early CT Score (ASPECTS).8,9 A normal CT scan received an ASPECTS of 10 points; a single point was subtracted for an area of ischaemic change as defined by von Kummer,16,17 which included any or all parenchymal hypoattenuation, loss of grey white differentiation, and focal brain swelling. Parenchymal hypoattenuation was defined as a region of abnormally decreased attenuation of brain structures relative to attenuation of other parts of the same structures or of the contralateral hemisphere. Focal brain swelling was defined as any focal narrowing of the cerebrospinal fluid space due to compression by adjacent brain structures producing effacement of the cortical sulci or ventricular compression. A score of 0 indicated diffuse ischaemic involvement throughout the MCA territory. The 24 h CT scan was read under the same conditions.

The presence of an HMCA sign10 and a sylvian fissure MCA “dot” sign11 was assessed. The HMCA was defined as an MCA denser than its counterpart and any other vascular structure excluding obvious calcification. The MCA “dot” sign was defined as the hyperdensity of an arterial structure (seen as a dot) in the sylvian fissure relative to the contralateral side or to other vessels within the sylvian fissure.

The MR angiogram was assessed by a single expert reader blind to all information apart from stroke symptom side, firstly at baseline, and then at the follow up at 24 h. The reasons for using MRA as the preferred imaging modality were its non invasiveness, the avoidance of radiation contrast agents, and its superior accuracy compared to other non invasive vascular techniques such as transcranial Doppler. For each non contrast examination the collapsed maximum intensity projection and source images were reviewed. The reader detailed the vascular abnormalities by the vascular signal intensity in the following extra- and intracranial vessels: the left and right internal carotid artery (ICA), the basilar artery (BA), the anterior cerebral arteries (ACAs), the left and right main stem of the middle cerebral artery segments (M1) and the MCA branch division (M2) regions of the MCA, and the left and right main posterior cerebral artery (PCA) segments.


Data are reported using descriptive statistics. Baseline and follow up CT and DWI ASPECTS values were derived as the median of the results of the five observers. Comparisons of proportions, sensitivity and specificity, relative risks, and confidence intervals were assessed using 2×2 tables and exact methods. The presence of either M1 or M2 MRA flow signal abnormality was used to perform the sensitivity calculation. Logistic regression was used to adjust for baseline stroke severity in the assessment of the association between ASPECTS and arterial occlusion. Functional outcome was dichotomised into independence (mRS 0–2) and dependence or death (mRS 3–6).


A total of 100 consecutive patients were enrolled into the study, of whom 69% were male, and the mean (SD) age was 68 (13.9) years. The median NIHSS was 9 (interquartile range 3–16). A total of 39 patients received tPA: 33 patients received intravenous tPA alone, and six received a combined approach of intravenous tPA followed by intra-arterial thrombolysis into the angiographically defined thrombus.18 Of the patients 60% were independent (modified Rankin score 0–2) at 3 months (two patients were lost to follow up at 3 months). There were nine deaths, of which two were due to tPA-related fatal intracerebral haemorrhages.

The median baseline CT ASPECTS was 8, which was the same as the median baseline DWI ASPECTS. A total of 67% (95% confidence interval (CI) 0.59 to 0.78) of the patients had CT ischaemic change (ASPECTS⩽9), while 79% (95% CI 0.70 to 0.87) of the DWI scans identified areas of hyperintense signal (DWI ASPECTS⩽9). Six patients (6%) had evidence of posterior circulation ischaemia on baseline DWI with or without coincident anterior circulation stroke. The mean (SD) time from symptom onset to CT was 117 (70) min compared to 219 (80) min to MR imaging (a mean (SD) difference of 102 (51) min, p = 0.0002). On CT, there were 10 hyperdense MCA signs and 19 MCA “dot” signs (four patients had both hyperdense MCA and “dot” signs) (fig 1). There were a total of 47 MRA flow signal abnormalities in the anterior circulation, of which 14 involved the ICA, either with M1 (five) or M2 (seven) alone, or both M1 and M2 (two). A total of 18 MRA abnormalities involved M1 (without ICA involvement), and 15 M2 MRA abnormalities (without involvement of either ICA or M1). The sensitivity and specificity of the hyperdense MCA and MCA “dot” sign are presented in table 1.

Table 1

 Sensitivity and specificity of the hyperdense middle cerebral artery (HMCA) and MCA “dot” sign using 3D TOF MRA as the gold standard

Figure 1

 (A) A baseline CT scan performed on a 64 year old male patient who presented with acute dysphasia and right sided weakness, 2 h into the symptoms. A hyperdense MCA “dot” sign (arrow) suggests a thromboembolic occlusion of the M2 branch artery. There are also early CT ischaemic changes in the posterior insula and posterior temporal lobe. Incidentally there is an old right PCA stroke. (B) MRA confirming a flow signal abnormality suggesting either occlusion or slow flow in the M2 branch. The patient received intravenous tPA, but unfortunately was functionally dependent at 3 months.

Both ASPECTS value (fig 2) and baseline NIHSS score predicted the presence of an intra- or extracranial occlusion. The relative risk of an occlusion on 3D TOF with a baseline NIHSS⩾10 was 2.22 (95% CI 1.39 to 3.56). After adjusting for baseline stroke severity, for each decrement of one ASPECTS point, the odds of an occlusion involving the internal carotid artery, M1 or M2 branches rose 2.7 fold (95% CI 1.8 to 4.1). Alternately, when the baseline CT ASPECTS was less than or equal to 7, the probability of occlusion on subsequent MR angiography was 0.88 (95% CI 0.72 to 0.97) for a relative risk of occlusion of 3.27 (95% CI 2.2 to 5.0). No interaction was detected between ASPECTS value and baseline NIHSS score (likelihood ratio test, p = 0.70), meaning the ability of ASPECTS to predict the MRA flow signal abnormality is not dependent on the NIHSS score.

Figure 2

 A histogram of baseline CT ASPECTS (horizontal axis) values and the probability of MRA flow signal abnormality (%, vertical axis) in the anterior circulation. The numbers in parentheses refer to the number in each ASPECTS group. ICA, internal carotid artery; M1 branch, middle cerebral artery main stem; M2 branch, main division of M1 branch.


Neurovascular imaging can assist in the appropriate selection of stroke patients for thrombolysis.19 Patients without occlusions receiving treatment with thrombolysis are possibly exposed to risk without obvious potential benefit. Results from the PROACT-II study suggest that, among patients with confirmed intracranial occlusion, the time window for useful intervention is longer than 3 h following the onset of symptoms.

In this cohort of consecutive patients stroke severity (median NIHSS 9) was not disimilar to that of patients enrolled in either the ECASS II or ATLANTIS B thrombolysis clinical trials.4,5 Yet only 47% of cases in this study had demonstrable MRA flow signal abnormalities in the MCA vascular territory. The most striking observation from this study was that ASPECTS predicts MRA flow signal abnormality independently of the NIHSS score. As the ASPECTS value decreased the probability of an MRA flow signal abnormality increased such that all cases with an ASPECTS of 5 and below had an MRA flow abnormality. When the CT scan was normal (ASPECTS 10) the prevalence of MRA flow abnormality in the anterior circulation was small (11%). Both ASPECTS and NIHSS were independent predictors of occlusion, and therefore, may be important clinical indices in predicting vascular occlusion. The relationship between NIHSS and MRA flow signal abnormalities has been previously reported.20

The HMCA sign has been associated with severe neurological deficit, extensive infarction in the MCA territory, and poor outcome21 and correlates well with MCA occlusion based on angiography.12,13 The recently described MCA “dot” sign11 correlates with distal MCA occlusion identified by conventional angiography. We now report the sensitivity and specificity of these signs using 3D TOF MRA as the gold standard. In accordance with previous angiographic studies,22 the sensitivity of both these signs is in the poor to moderate range, but both have very high specificity, confirming that when present they have a high positive predictive value for thromboembolic occlusion.

MRA provides anatomical and physiological evidence of flow through the major collateral pathways, and allowed non invasive correlation of the CT vascular signs. Conventional angiography remains the gold standard for assessing the collateral pathways and is consistent with anatomical studies.23 But the risk of inducing stroke in this particular patient group may be similar to or higher than the complication rate in the Asymptomatic Carotid Surgery Trial of 1%.24 Time of flight angiography allows the arteries of the circle of Willis to be examined by the endogenous contrast of flow-related spin enhancement and a flow related abnormality can indicate either an absence of flow or slow flow. The rationale for the use of MRA as the gold standard was its safety (cf formal cerebral angiography), and avoidance of radiation and contrast bolus as with CTA. In addition its accuracy is probably superior to other techniques such as transcranial Doppler.25 In a study comparing TOF MR angiography of carotid stenosis, flow void artefacts represent severe carotid stenosis when compared to formal angiography.26 The major drawback of this method is the depiction of the distal intracranial vessels which may account for some of the under-reporting of distal MCA branch abnormalities in this study.27

Previous data support the concept that stroke patients can be selected on the basis of stroke severity and quantified CT ischaemia.4,6–8,28,29 There is mounting evidence that the extent of CT ischaemia is very important in determining response to therapy. In a post hoc analysis of ECASS I,29 patients with signs of ischaemia involving ⩽1/3 MCA benefited from tPA treatment when compared to placebo (OR 3.43, 95% CI 1.61 to 7.33). The benefit was less clear for patients with normal CT scans or parenchymal hypodensity involving >1/3 MCA, and did not exclude the possibility of harm.

These data combined with the current work support the idea of a therapeutic trichotomy in acute stroke patients: (1) the normal CT scan (low prevalence of middle cerebral artery occlusion) may indicate patients who will improve spontaneously; (2) the CT scan with early but not extensive ischaemic changes (ASPECTS 7–9; high prevalence of MCA occlusion) may indicate the ideal thrombolysis candidate; and (3) the CT scan with extensive CT ischaemic change (high prevalence of MCA occlusion) may indicate the patient least likely to benefit from thrombolysis. This stratification may change with time from stroke onset. Therefore, we propose that successful tPA adminstration for acute stroke depends on careful clinical assessment, combined with selection of patients with a CT scan showing moderate rather than extensive ischaemic change. This group of patients have a high probability of large artery occlusion with associated “tissue at risk”, and therefore may benefit most from thrombolysis. This concept would require further validation before it impacts patient care but a systematic approach to the assessment of non contrast CT with ASPECTS may provide a useful surrogate in future thrombolysis trials.

New CT techniques such as CT angiography, CT perfusion, and post contrast and blood pool analysis are being increasingly evaluated in clinical practice: the latter has been shown to increase the visualisation of ischaemic areas by 40%.30 However, the functional outcome following tPA administration is time dependent,31 and additional contrast enhanced CT techniques, as with MRI, may delay initiation of treatment with no additional benefit.


The authors thank Kathryn Werdal and Andrea Cole-Haskaynes for administrative aid in organising films for review.



  • This study was supported by the Alberta Foundation for Health Research. Dr Barber was supported by the Canadian Institute of Health Research, the Heart and Stroke Foundation of Canada, and the Alberta Heritage Foundation for Medical Research. Dr Hill was supported by the Heart and Stroke Foundation of Alberta, NWT, Nunavut and the Canadian Institutes for Health Research. Dr Demchuk was supported by the Alberta Heritage Foundation for Medical Research and the Canadian Institute of Health Research. Dr Frayne was supported by Heart and Stroke Foundation of Canada and the Alberta Heritage Foundation for Medical Research. Dr Buchan was supported by the Heart and Stroke Foundation of Canada, and the Alberta Heritage Foundation for Medical Research.

  • Competing interests: none declared

Linked Articles

  • Correction
    BMJ Publishing Group Ltd