XVII. European Stroke Conference
Nice, France

Oral Session:

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 10:30 - 10:40 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

01
Cerebral Amyloid Angiopathy predisposes to tPA related haemorrhage in patients presenting with acute ischaemic stroke: A PiB PET study
J.V.Ly    C.C.Rowe    J.A.Zavala    H. Ma    V.L.Villemagne    G. Okeefe    U. Ackerman    J. Sachinidis    H. Tochon-Danguy    G.A.Donnan
 

National Stroke Research Institute

AUSTRALIA

Background: There is increasing evidence that Cerebral Amyloid Angiopathy (CAA) may be an important predisposing factor for the haemorrhagic complications of tPA therapy in patients presenting with acute ischaemic stroke. N-methyl-[11C]2-(4’-methylaminophenyl)-6-hydroxybenzothiazole (11C PiB) is a PET amyloid ligand which also binds to beta-amyloid in cerebrovascular walls. We hypothesized that patients who developed parenchymal haemorrhage (PH) after tPA may have increased 11C PiB PET retention compared to those with no haemorrhagic complications. Methods: Patients treated within 3 hours of onset of ischaemic stroke with tPA were studied using PET to compare PiB retention in those with and without parenchymal haematoma formation. Both were compared to age matched controls. Distribution Volume Ratio (DVR) parametric images were created using Logan graphical analysis with the cerebellar cortex as reference region. Differences between medians were assessed using Wilcoxon test. Results: Fifteen tPA treated stroke patients (mean age 73.80 ± SD 11) and fifteen normal controls (mean age 73.10 ± SD 6) were studied. Neocortical PiB retention was significantly higher among the seven patients with PH (median DVR 1.47, IQR 1.44-1.53) compared to eight patients without PH (median DVR 1.33, IQR 1.29-1.38) (p =0.015), and to normal controls (median DVR 1.31, IQR 1.20-1.39) (p = 0.012). ROC curve analyses showed a neocortical PiB DVR of 1.41 or greater to be the optimal threshold for predicting parenchymal haemorrhage with a sensitivity of 0.86 and 1-specificity of 0.13. Conclusion: Ischaemic stroke patients treated with tPA who develop PH have higher neocortical PiB retention compared to those without PH, suggesting underlying CAA as a predisposing factor for tPA related haemorrhage. The high sensitivity and specificity for this finding may provide an impetus for the development of a more practical rapid pre treatment screening technique

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 10:40 - 10:50 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

02
Improving inter-observer agreement about brain microbleeds: development of the Brain Observer MicroBleed Scale (BOMBS)
C. Cordonnier    G. Potter    C. Jackson    F. Doubal    C.L.M.Sudlow    J.M.Wardlaw    R. Al-Shahi Salman                     
 

Division of Clinical Neurosciences, University of Edinburgh

FRANCE

Background and purpose: Gradient echo (GRE) T2* magnetic resonance imaging (MRI) of the brains of adults with stroke frequently detects brain microbleeds (BMB). If the diagnostic and prognostic significance of BMBs are to be investigated and used for these purposes in clinical practice, observer variation in BMB assessment must be minimized. Methods: Two doctors used a pilot rating scale to describe the number and distribution of BMBs (round, low signal lesions, <10 mm diameter on GRE T2* MRI) on MRI of 264 adults with stroke or TIA, blinded to clinical data and their counterpart’s ratings. Disagreements were adjudicated by a third observer, common sources of disagreement identified, and these data used to develop a revised rating scale (BOMBS), which was tested by the same observers in a separate cohort of 156 adults with stroke. Results: In the pilot study, agreement about the presence of BMBs at any location was moderate (kappa [Қ] 0.44, 95% confidence interval [CI] 0.32 to 0.56). Agreement about the presence of BMB(s) was worse in lobar (Қ 0.44, 95%CI 0.30 to 0.58) than in deep (Қ 0.62, 95%CI 0.48 to 0.76) or posterior fossa locations (Қ 0.66, 95% CI 0.47 to 0.84). Disagreements usually occurred in assessing whether none or one BMB was present and was largely due to the possible presence of BMB mimics. Assessment of BMBs using our new rating scale, BOMBS, improved agreement on the presence of BMB(s) in any location (Қ 0.68, 95% CI 0.49 to 0.86) and in lobar locations (Қ 0.78, 95% CI 0.60 to 0.97). Conclusion: We found a variety of reasons for observer disagreement about the presence of BMBs. When these reasons were incorporated into a rating scale that also accounted for observer uncertainty, interobserver reliability improved.

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 10:50 - 11:00 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

03
Radiological investigation of non-traumatic, supratentorial intracerebral haemorrhage: tri-national survey
C. Cordonnier    C.J.M.Klijn    J. van Beijnum    R. Al-Shahi Salman                                          
 

University of Edinburgh

UNITED KINGDOM

Background: Knowledge about the prevalence of underlying causes of non-traumatic intracerebral haemorrhage (ICH) is limited, as is information about the diagnostic utilities of radiological investigations to detect them. EUSI guidelines do not recommend the further investigation of “ICH involving the putamen, globus pallidus, thalamus, internal capsule, periventricular white matter, pons or cerebellum, particularly in patients with known hypertension.” Methods: We sent a structured postal survey to assess the patterns of supratentorial ICH investigation to members of the professional organisations for stroke physicians/neurologists, neurologists, neurosurgeons, and neuroradiologists in France, the United Kingdom, and the Netherlands. We sought opinions on whether, how, and when to investigate ICH scenarios, accompanied by illustrations of brain computed tomography, in eight categories determined by ICH location (lobar versus deep), pre-stroke hypertension (present versus absent), and age (<45 versus >45 years). Results: We received 617 questionnaires (50% response rate), with 99% data completeness. Further investigation would have been undertaken by 99% of respondents in normotensive adults aged <45 with lobar or deep ICH, 76% of normotensive adults aged >45 with deep ICH, and 31% of adults aged >45 with deep ICH and pre-stroke hypertension. Age <45 years was the strongest influence on the decision to further investigate ICH (OR 16, 95%CI 13 to 20), followed by no history of pre-stroke hypertension (OR 5, 95%CI 4 to 6), and lobar ICH location (OR 2, 95%CI 1 to 2). Conclusions: Youth is the strongest influence on the decision to investigate patients with supratentorial ICH. Current clinical practice does not correspond to EUSI guidelines. Studies are required to assess the diagnostic utilities of brain imaging techniques in determining ICH cause, and which patient subgroups to investigate, when, and how.

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 11:00 - 11:10 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

04
Comparison of perfusion weighted imaging techniques to assess tissue at risk and reperfusion in acute ischemic stroke. A DEFUSE substudy.
V. Thijs    R. Dobosi    M. Lansberg    J. Olivot    M. Mlynash    R. Bammer    L.  Wechsler    G.W.Albers              
 

University Hospitals Leuven

BELGIUM

Objective: To compare commonly used analysis methods for perfusion weighted imaging (PWI) in acute ischemic stroke patients. Methods: PWI volumes from a prospective, multicenter study (Diffusion Weighted Imaging Evaluation for Understanding Stroke Evolution, DEFUSE) were recalculated using non arterial input function (AIF) based Time-to-peak (TTP) and Mean Transit Time (MTT) maps and compared with the original AIF-based Tmax volumes used in DEFUSE. Lesion volumes were determined by manual delineation of thresholded TTP and MTT maps with a 4 or a 6 seconds delay compared to a contralateral non-ischemic brain region. Bland-Altman plots were created and diagnostic measures of mismatch and reperfusion between MTT, TTP and Tmax were calculated. The rate of the primary endpoint of DEFUSE, favourable clinical response among mismatch patients with and without reperfusion, was recalculated based on MTT or TTP volumes. Results: 74 patients underwent PWI and DWI within 3 to 6 hours after symptom onset and were treated with intravenous tPA regardless of the baseline magnetic resonance profile. Bland-Altman plots demonstrated large limits of agreement indicating that volumes determined by thresholded TTP or MTT differed substantially from Tmax volumes. Misclassification rates with Tmax indices of mismatch and reperfusion used as the reference standard were between 18% and 39% for mismatch and between 16 and 31% for reperfusion. If these alternative analysis techniques had been used as the primary analysis method for PWI volume determination in DEFUSE, a relationship between reperfusion and good clinical outcome in mismatch patients would not have been found. Discussion: Commonly used analysis methods for PWI may produce misleading results regarding the presence of a mismatch or reperfusion. A relationship between good clinical response and reperfusion in mismatch patients would not have been demonstrated had thresholded MTT or TTP maps been used as the primary analysis method in the DEFUSE study.

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 11:10 - 11:20 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

05
Optimal mismatch definitions for detecting treatment response in acute stroke
S. Christensen    M. Parsons    D. De Silva    M. Ebinger    K. Butcher    J.  Fink    S. Davis                     
 

Royal Melbourne hospital / University of Melbourne

AUSTRALIA

Background: The use of PWI/DWI (perfusion/diffusion weighted imaging) mismatch (MM) may have the potential to identify patients most likely to respond to thrombolysis. Current PWI methodology has high sensitivity but poor specificity to infarct growth. This may dilute treatment effects in a penumbral selection trial due to inclusion of patients with modest probability of growth. We hypothesized that more precise MM definitions would increase the differences in growth between patients with reperfusion (RP) and without reperfusion (non-RP), with the assumption that RP is a surrogate marker for thrombolytic treatment response. Methods: We retrospectively analyzed MRI data from the double blind randomized placebo-control EPITHET trial of tPA to identify the MM definition that best identified potential treatment responders. We included patients with acute PWI deficits, day 3-5 PWI/DWI and follow up imaging. We defined RP as PWI volume decrease of >90% on day 3-5 and MM as PWI-DWI volume > 10 ml. The MM definition was successively tightened by increasing the Tmax threshold (the threshold used to define the ischemic penumbra) from 2 to 12 s. For each threshold the difference in median infarct growth between RP and non-RP was assessed. Results: Reperfusion occurred in 34 out of 81 patients. The difference in median growth between RP and non-RP was significant (p<0.001) at each Tmax threshold. The magnitude of this difference increased as a function of the Tmax threshold (Spearmans rho=0.94, p=0.017). The median growth difference between groups increased from 45 to 88 ml when the Tmax threshold was raised from 2 to 12 s. Discussion: A more restrictive mismatch definition using a higher perfusion threshold is more likely to demonstrate treatment effect by selecting patients with higher probability of infarct growth. The actual selection of threshold will depend on a specificity/sensitivity trade off analysis. The details of how this optimized MM definition affects growth differences in the tPA vs. Placebo groups will be also presented.

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 11:20 - 11:30 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

06
Do MR perfusion thresholds reliably identify core and penumbra of ischaemic tissue in acute ischaemic stroke?
T. Carpenter    M. Lee    CS.Rivers    JM.Wardlaw                                          
 

University of Edinburgh, *University of Leeds

UNITED KINGDOM

Background: MR perfusion imaging (PWI) thresholds might distinguish infarcted from at risk tissue in acute ischaemic stroke and so guide treatment decisions. We used receiver operator characteristic (ROC) curves to identify PWI thresholds in the acute stroke DWI lesion, final infarct on T2-weighted imaging and “penumbra” (tissue in the final T2 but not the DWI lesion). Methods: Patients with acute ischaemic stroke had acute PWI, DWI, and T2-weighted imaging (FT2) at 1 month or more. We derived quantitative parametric (q) maps of cerebral blood flow (CBF), cerebral blood volume (CBV) and mean transit time (MTT). We used ROC curves to test the ability of different acute qPWI thresholds to identify tissue in the acute DWI, penumbra and FT2 lesions, per perfusion parameter and patient. We then used “figures of merit” to identify the optimum threshold, per perfusion parameter, for infarcted vs. salvaged tissue, for the whole cohort. Results: 32 patients were imaged at median 7.3 hours. Median qMTT differed between DWI (7.3s, range 3.2-14.5), penumbra (5.96s, 2.9-12.4) and FT2 (6.5s, 3.0-13.4), both p<0.01. Median qCBF differed between DWI and penumbra (16.5 vs 23.0 ml/100g/min, range 6.8-58.2 vs 11.5-71.3, p=0.002) but not FT2 (19.3, 6.7-58.9, p=0.15). Median qCBV did not differ between DWI, penumbra or FT2. ROC curves and figures of merit showed that qMTT and qCBF were better than qCBV at identifying DWI and FT2 tissue (p<0.002), but because the range of PWI values in each tissue region was very wide, there was no difference in the qMTT and qCBF figures of merit for DWI, penumbra or FT2 lesions (p>0.222). Time of scan had no effect on ROC results. Conclusions: Although median qMTT and qCBF values differ between tissue that survives or infarcts, there is too much heterogeneity for a single PWI threshold at any acute time point to be used to identify acute core or penumbra reliably. This indicates that PWI threshold values should not be used to guide treatment decisions.

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 11:30 - 11:40 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

07
Inverse mismatch and lesion growth in small subcortical ischemic stroke
J.B.Fiebach    A. Hopt    T. Vucic    P. Brunecker    C.H.Nolte    C.A.Doege    G.J.Jungehulsing    C. Kunze    S. Wegener    A. Villringer
 

Charité Universitätsmedizin Berlin Dep of Neurology

GERMANY

Background: In territorial ischemic stroke, infarction typically develops within the borders of the initial perfusion disturbance. Motivated by preliminary findings indicating a different pathophysiology in small subcortical stroke we investigated whether in these patients, lesion growth towards infarction can occur beyond the margins of the affected vascular territories of the perforating arteries. Method: In 19 consecutive patients, stroke MRI was performed within 14 hours after ictus, and at days 2 and 6 (±1). Sizes of diffusion and perfusion lesions were determined on days 1 and 2. “Final lesion volume” was measured on T2-weighted images on day 6. Findings: At the initial examination, the mean diffusion lesion (ADC lesion size: 1.82±1.2 ml) was larger (p=0.0002) than the perfusion lesion (MTT lesion size: 0.72±0.69ml). Such an “Inverse Mismatch“ (ADC-lesion > MTT-lesion) was present in 14/19 patients at baseline and in all patients on day 2. Final lesion volume at day 6 was 3.2±1.6ml which was larger than the initial perfusion deficit (p=0.02), a finding which was present in each patient. Conclusion: In small subcortical ischemic stroke, “Inverse Mismatch” is frequent, and infarction develops beyond the initial perfusion lesion during the first week after symptom onset, which is strikingly different from territorial infarcts. Development of infarction in areas with initially normal perfusion indicates that cytotoxic processes probably triggered by the infarct core are a dominant mechanism for lesion growth. Areas with normal perfusion but threatened by cytotoxic damage developing over several days seem prime targets for neuroprotective therapy.

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 11:40 - 11:50 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

08
Penumbral mismatch is underestimated using standard volumetric methods and this is exacerbated with time.
H. Ma    J.A.Zavala    H. Teoh    J.V.Ly    T. Phan    S. Arakawa    S. Davis    L. Churilov    P. Wright    G.A.Donnan
 

National Stroke Research Institute, Australia

AUSTRALIA

Background In acute ischemic stroke perfusion weighted image (PWI) and diffusion weighted image (DWI) mismatch on magnetic resonance imaging (MR) is used as an indicator of the presence of the penumbra and the selection of patients for clinical trials of therapy. The approximate volumetric method and the more precise co-registration method are two used for mismatch volume (MV) calculation. We hypothesized that volumetric and co-registration methods provide different estimates of the MV and its salvage rate and that these differences depend upon the time since stroke onset. Methods Ischemic stroke patients presenting within 48 hours of onset were imaged with MR. Volumetric MV was defined as PWI defect (Tmax 2 seconds plus) minus DWI lesion. After co-registration of images the co-registration MV was defined as the region of PWI defect which was not overlapped by the DWI lesion. The mismatch salvage rate for each method was defined as the percentage of MV not proceeding to infarction verified by repeat MR T2 sequence at about 3 months. Results 72 patients were recruited with median age of 74.0 years. The median time to MR from stroke onset was 5.9 hours (IQR: 3.0-20.4). There was a consistent underestimation of the MV using the volumetric method (median 9.3ml, IQR: 0-63) when compared to the co-registration method (median 20.1ml, IQR: 3.2-69.8, p<0.0001). The proportion of penumbral tissue salvaged at 3 months was underestimated using the former (32.2%, IQR: 0-66.6 for the volumetric and 83.3%, IQR: 66.1-98.9 for the co-registration method, p<0.0001). The percentage differences in MV increased with time post stroke (p =0.006). Conclusions In acute ischemic stroke the use of a volumetric analysis of MR volumes results in a consistent underestimation of both the MV and the percentage of mismatch salvaged when compared to the more precise co-registration method. The degree of the former underestimation increased with time. These findings have implications for patient selection in MR based therapeutic trials using longer time windows.

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 11:50 - 12:00 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

09
Expediting MRI-based proof of concept stroke trials using an earlier primary endpoint
M. Ebinger    S. Christensen    D.A.DeSilva    M.W.Parsons    C.R.Levi    A. Peeters    P.A.Barber    C.F.Bladin    G.A.Donnan    S.M.Davis
for the EPITHET investigators

Royal Melbourne Hospital, Melbourne

AUSTRALIA

Background Prior to pivotal phase III trials of acute stroke therapies, ‘proof of concept’ MRI trials are used to assess efficacy with surrogate imaging endpoints. We hypothesized that subacute DWI lesion volume could replace late T2-weighted lesion volume as the primary endpoint of such trials. Methods In the Echoplanar Imaging Thrombolytic Evaluation Trial (EPITHET), patients with acute ischemic stroke presenting within 3-6 hours, were randomised to tPA or placebo. Lesion volumes on three different time points (baseline, day 3-5, and day 90) as well as clinical scores were assessed. Only those patients who had imaging at all three time points were included in this blinded substudy. Results Seventy-two patients (mean age 71 years, 34 men) of 100 participants had imaging at all three time points. Median baseline DWI lesion volume was 19.6 mL (IQR 8.5-41.4), median DWI lesion volume on day 3-5 was 45.2 mL (IQR 15.9-125.1), and median T2-weighted lesion volume on day 90 was 23.0 mL (IQR 5.4-81.9). Correlation was moderate between baseline and day 90 lesion volumes (Pearson’s rho 0.77, P<0.01) and strong between day 3-5 and day 90 lesion volumes (Pearson’s rho 0.95, P<0.01). The correlation coefficient between day 90 NIHSS score and lesion volumes was 0.64 (P<0.01) at baseline, 0.81 (P<0.01) at day 3-5 and 0.77 (P<0.01) at day 90. Conclusion Subacute lesion volume correlated strongly with late lesion volume, and subacute lesion volume at day 3-5 predicted neurological deficit at day 90. These results support an early primary imaging end point for proof of concept stroke trials, expediting and economizing future research.

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 12:00 - 12:10 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

10
Diffusion Tensor Imaging in Acute Pontine Stroke: Reduction of Fractional Anisotropy of the Pontine Corticospinal Tract Correlates with a modified NIHSS
C. Ottomeyer    K. Szabo    C.  Rossmanith    R.  Kern    J.  Binder    J.G.Hirsch    M.G.Hennerici    A. Gass              
 

Universitaetsklinikum Mannheim University of Heidelberg

GERMANY

Introduction: Diffusion tensor imaging (DTI) can visualize the main fiber bundles of the brain (eg, corticospinal tract, CST) and provides quantitative information on tissue integrity. We compared the DTI derived parameter fractional anisotropy (FA) of the CST of acute pontine stroke lesions with the clinical deficit. Methods: 15 acute pontine stroke patients with motor system deficits admitted to the stroke unit were studied with standardized MRI at 1.5 T including T2-weighted, diffusion-weighted images (DWI), and DTI (EPI, isotropic distribution along 30 directions, b-value 900 s/mm2, 2.5 mm³ isotropic spatial resolution, TA 7min). In each patient the CST was identified on color-coded DTI maps and FA values of the infarcted CST was compared to the contralateral non-affected CST with nordicICE software. The percentage of FA reduction was compared to the worst clinical deficit in the first 48 hours after symptom onset as measured by a modified NIHSS (motor/sensory/articulation) Results: The mean FA of the non affected CST was 0.51, while the mean FA of the ischemic CST was 0.41. Reduction of the FA in the affected CST ranged from 28 to 40%. The mean modified NIHSS was 4.4 (SD 2.3; range 1-9). There was a significant inverse correlation of the maximal reduction of the FA value with the degree of motor paresis as evaluated by the NIHSS (beta = -0.53, t = -2.67, p < 0.05). Conclusions: Acute pontine stroke is a well defined anatomical and pathophysiological model for corticospinal tract ischemia. In its acute phase the reduction of the FA of pontine corticospinal fibers correlates closely with a modified NIHSS. Whether DTI derived measures also show prognostic value in regard to the short term or long term clinical outcome will be subject of follow-up studies.

http://www.eurostroke.org/ni_graphics/g_aid586.htm
 

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 12:10 - 12:20 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

11
Performance of MRI based cerebral blood flow maps in early stroke compared to 15O-water positron emission tomography: threshold analysis and influencing factors
J. Sobesky    O. Zaro Weber    W. Moeller Hartmann    W.D.Heiss                                          
 

Unversity of Cologne

GERMANY

Background: The accuracy of perfusion weighted magnet resonance imaging (PW-MRI) based maps of cerebral blood flow (mriCBF) remains a matter of debate. We validated mriCBF on quantitative CBF measurement by 15O-water positron emission tomography (petCBF) with respect to penumbral flow (<20 ml/100g/min). Methods: In acute and subacute stroke patients, mriCBF and petCBF maps were compared. In a volumetric analysis, the performance of predefined mriCBF thresholds (<40,<30,<20,<10 ml/100g/min; quantitative analysis with arterial input function) was assessed using the volume of penumbral flow on PET as the target volume. The degree of congruence was expressed as the ratio C (C= Volume mriCBF / Volume petCBF). The influence of vessel pathology, hypoperfusion size and time point of imaging was described. Results: In 24 stroke patients (median time MRI to PET: 68 minutes; 16 patients imaged within 24 hours after stroke) the median volume of penumbral flow (petCBF) was 78.5 ccm. On visual inspection, an excellent qualitative congruence was found.for mriCBF. In the pooled analysis, mriCBF <20ml/100g/min best identified penumbral flow (median C-ratio: 1.0) but showed a wide interindividual range (C-ratio 0.3 to 3.5). Ipsilateral vessel pathology, time point of imaging and size of penumbral hypoperfusion did not significantly influence the C-ratio. Discussion: CBF maps using a threshold of <20 ml/100g/min well identified penumbral flow. However, a considerable interindividual variance was found and could not be explained by routine clinical data. Our results strongly support the validity of MRI based CBF measurement in clinical routine but they also underline the need of a further specification of the MRI based mismatch concept.

Brain Imaging: new developments  
Date: Thursday 15 May 2008  
Time: 12:20 - 12:30 -  Room: Calliope
Chair: J.-C. Baron, United Kingdom and N. Nighoghossian, France

12
Sensitivity of fluid attenuated inversion recovery (FLAIR) imaging for ischemic lesions increases with time from symptom onset to >90% at the end of the 6 hour time window in acute stroke
G. Thomalla    P. Rossbach    M. Rosenkranz    S. Siemonsen    A. Krützelmann    J. Fiehler    C. Gerloff                     
 

University Hospital Hamburg-Eppendorf

GERMANY

Background: Ischemic lesions become visible on fluid attenuated inversion recovery (FLAIR) imaging within the first hours after stroke. However, the sensitivity of FLAIR imaging within the first 6 hours of ischemia and it’s relation to time from symptom onset is not well defined. We studied the sensitivity of FLAIR imaging for acute ischemic lesions as compared to diffusion weighted imaging (DWI) during the first 6 hours of stroke. Methods: We analyzed data of consecutive acute ischemic stroke patients studied by stroke MRI within 6 hours. MR images were rated by four experienced raters blinded to clinical information. The presence of ischemic lesions were judged in three steps: 1) on the FLAIR image blinded to DWI, 2) on the DWI, 3) on the FLAIR image with knowledge of DWI. Ischemic lesions were judged as visible if identified by at least three of four raters. Results: Data of 104 patients (43% females, mean age 65 y.) were analyzed. Mean (SD) time between symptom onset to imaging was 150 (+/- 65) min, mean National Institutes of Health Stroke Scale (NIHSS) Score was 14 +/- 5.6. Overall sensitivity of FLAIR alone was 36%, compared to 97% for DWI. Knowledge of DWI information increased the visibility of ischemic lesion on FLAIR images to 63%. The sensitivity of FLAIR increased over time (<1.5/1.5-3/3-4.5/4.5-6 h) for FLAIR alone (8%/31%/62%/44%, p=0.009) and FLAIR with knowledge of DWI (39%/53%/95%/89%, p<0.001), while sensitivity of DWI remained stable (92%/97%/100%/100%, p=0.573). A negative FLAIR image in the case of a visible lesion on DWI allowed the allocation to <3 hours with a high specificity (93%) positive predicate value (95%). Discussion: We demonstrate a clear time dependency of the sensitivity of FLAIR imaging for acute ischemia in hyperacute stroke. Together with information from DWI, ischemic lesions can be identified on FLAIR images in >90% of patients after 3-6 hours. An ischemic lesion not visible on FLAIR images is very likely to be less than 3 hours old. Together with DWI, FLAIR imaging may help identify patients eligible for thrombolysis in cases of unknown symptom onset.