- Email: [email protected]
Diffusion-weighted Imaging at b1000 for Identifying Intracerebral Hemorrhage: Preliminary Sensitivity, Specificity, and Inter-rater Variability
Diffusion-weighted Imaging at b1000 for Identifying Intracerebral Hemorrhage: Preliminary Sensitivity, Specificity, and Inter-rater Variability Galina Keigler, MD,* Ilan Goldberg, MD,* Roni Eichel, MD,* John M. Gomori, MD,† Jose E. Cohen, MD,‡ and Ronen R. Leker, MD, FAHA*
Background: Noncontrast computed tomography (NCCT) is the gold standard to detect intracerebral hemorrhage (ICH) and ischemic stroke (IS) in patients presenting with acute focal syndromes. Diffusion-weighted magnetic resonance imaging (DW-MRI) obtained at b1000 is highly sensitive to identify acute IS but its sensitivity and specificity to detect ICH has not been systematically studied. Methods: Patients with a diagnosis of ICH on NCCT were prospectively enrolled and underwent DW-MRI at b1000. Patients with suspected ischemia and a negative NCCT served as controls. All diffusion-weighted imaging (DWI) scans were evaluated blindly by 4 experienced raters. Sensitivity, specificity, and inter-rater variability of the DWI b1000 scans for detection of ICH were determined. Results: In this preliminary pilot study, 15 patients with ICH and 17 patients with IS were included. All ICH lesions seen on NCCT showed a typical pattern on DW-MRI at b1000 with a hypointense core surrounded by a hyperintense rim. ICH volumes and size were similar on NCCT and MRI. All cases of IS were identified on the DWI scans but none were apparent on NCCT. The mean sensitivity and specificity of DW-MRI at b1000 for ICH were 94% and 93.5%, respectively, and the inter-rater variability for ICH detection on DWI was excellent (k 5 .84). Conclusions: DW-MRI at b1000 has a diagnostic yield similar to NCCT for detecting ICH and superior to NCCT for detecting IS. Therefore, DW-MRI may be considered as the initial screening tool for imaging patients presenting with focal neurologic symptoms suggestive of stroke. Key Words: Stroke—diffusion-weighted MRI—cerebrovascular disease— intracerebral hemorrhage. Ó 2014 by National Stroke Association
From the *Department of Neurology, Hadassah-Hebrew University Medical Center, Jerusalem; †Department of Radiology, HadassahHebrew University Medical Center, Jerusalem; and ‡Department of Neurosurgery, Hadassah-Hebrew University Medical Center, Jerusalem, Israel. Received November 28, 2013; revision received January 6, 2014; accepted February 1, 2014. This study was supported by the Peritz and Chantal Scheinberg Cerebrovascular Research Fund and the Sol Irwin Juni Trust Fund. All authors have nothing to disclose. Address correspondence to Ronen R. Leker, MD, FAHA, Department of Neurology, Hebrew University-Hadassah Medical Center, PO Box 12000, Jerusalem 91120, Israel. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.02.005
Noncontrast computed tomography (NCCT) is usually used as the initial imaging modality in patients presenting to the Emergency Department (ED) with acute focal neurologic symptoms and signs suggestive of stroke.1 It is primarily useful in identifying patients with intracerebral hemorrhage (ICH) with sensitivity and specificity rates that approach 100% and may also be useful for identifying signs of early ischemic stroke (IS), although with much lower sensitivity and specificity rates.1-4 Magnetic resonance imaging (MRI) and especially diffusionweighted imaging (DWI) is much more sensitive and specific than NCCT for the detection of early ischemia,1-4 and it is currently recommended as a screening tool for patients with IS. Over the past decade, several studies evaluated different MRI protocols to estimate their ability to detect
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2014: pp 1-5
1
G. KEIGLER ET AL.
2
early products of hemoglobin and distinguish hyperacute hemorrhage from ischemia.5-12 These studies found that several MRI sequences and, in particular, susceptibilityweighted imaging (SWI) including T2* may be superior to NCCT for the detection of ICH and hemorrhagic transformation of IS.5-13 In addition, MRI may contribute to determining the cause of ICH by distinguishing among cerebral amyloid angiopathy, hypertensive bleeds, and vascular malformations.5 Other MRI sequences including DWI obtained at b0 have also been reported to detect ICH but with a slightly reduced sensitivity and specificity compared with susceptibility images.11 Therefore, it may be claimed that a full MRI protocol may be needed in the ED for stroke imaging. However, a full multiparametric stroke MRI takes significantly longer to complete and interpret than an NCCT and may thus delay therapy in patients with IS who are eligible for thrombolysis or endovascular recanalization. In contrast, DWI can be acquired rapidly,4 is associated with outcome, and has an excellent interrater variability.14,15 Furthermore, we have previously shown that using a combination of clinical and DWI parameters can rapidly identify patients with acute ischemia and distinguish them from stroke mimics.16
Patients and Methods All patients with acute IS admitted to our institution are enrolled in a prospective database. The Institutional Review Board has granted a general permission to collect routine research data on all stroke patients. As per the institutional protocol, all patients with suspected stroke/transient ischemic attack are seen by a neurologist and undergo NCCT of the brain. Patients with ICH were prospectively identified on NCCT. All included patients underwent a screening DWI-only MRI protocol that included DWI and apparent diffusion coefficient (ADC) maps. Scans were mostly acquired on a 3 T Siemens MRI scanner (Siemens, Erlangen, Germany) with the following characteristics: field of view 230 3 230 mm, acquisition matrix 128 3 100, slice/gap 5 5/0 (in millimeters), repetition time/echo time 5 4600/93 (in milliseconds), b values 5 0, 500, 1000. In a few instances, scans were acquired on a General Electric 1.5 T MRI scanner (GE healthcare, Milwaukee, WI) with the following characteristics: field of view 240 3 217 mm, acquisition matrix 128 3 100, slice/gap 5 5/1 (in millimeters), repetition time/echo time 5 3800/96 (in milliseconds), b values 5 0, 500, 1000. MRI scans were obtained as soon as possible after admission. Consecutive patients presenting to the ED with acute neurologic symptoms and signs who had an NCCT scan negative for ICH also underwent DWI-only protocol and were included as controls for this study. Scans were obtained as soon as possible and took less than 5 minutes to complete in all cases. Patients with evidence for bright hyperintense lesions on DWI that were hypointense on the ADC map were classified
as having an acute IS on DWI. Patients with evidence of a hypointense core surrounded by a hyperintense bright rim on DWI were classified as having an ICH on DWI. ICH volumes and locations were calculated on the NCCT and MRI scans using the ABC/2 formula, which takes into account the largest diameters of the hematoma multiplied by the number of slices it appears on, and ICH locations were determined using the Alberta Stroke Program Early CT Scale (ASPECTS) system subfields.17 For evaluation of sensitivity, specificity, and inter-rater variability, DWI studies were interpreted by senior neuroradiologist (J.M.G.) or vascular neurosurgeon (J.E.C.) and stroke neurologists (R.E. or R.R.L.) in all cases. Evaluators were blinded to the patient’s neurologic status, the NCCT scan, and the final discharge diagnosis from the Department of Neurology. Baseline demographics, risk factor profiles, clinical characteristics, and final discharge diagnoses and outcome were accrued. Neurologic deficits were graded with the National Institutes of Health Stroke Scale. Statistical evaluations were performed with the SigmaStat package (SPSS, IBM, Armonk, NY). For univariate analysis, patients with positive and negative DWI results were compared using the Student t test or chi-square test. Sensitivity and specificity rates were obtained for each rater, and the data presented represent the mean 6 standard deviation and ranges. Kappa (k) statistics for multiple evaluators were used for the evaluation of inter-rater variability. k values greater than .7 were considered good and greater than .8 excellent.
Results For this pilot study, we included 15 patients with ICH on the admission based on NCCT scan and 17 patients with similar stroke severity that had evidence for IS on the DWI scan. These controls were specifically used to determine if hemorrhagic transformations of ISs from true ICH could be easily discerned and to help determine the sensitivity and specificity of DWI at b1000 in small lesions. The baseline characteristics of the included patients are shown in Table 1. The time from symptom onset to DWI data acquisition in ICH patients ranged from 3-144 hours with a median of 25 hours and did not differ from IS patients (median 28 hours; range .18-168 hours). All hemorrhages were clearly identified on NCCT scans. In all patients with ICH, a similar pattern was seen on the MRI b1000 images: a hypointense core that measured similar to ICH boundaries on NCCT surrounded by a bright hyperintense rim not seen on the initial NCCT (Fig 1). The boundaries of this lesion coregistered with the hematoma on NCCT. In contrast, acute IS on DWI scans showed the typical bright hyperintense DWI core, which was hypointense on the ADC maps.
DWI B1000 IN ICH
3
Table 1. Univariate analysis comparing patients with ICH and IS Variable/group
ICH (n 5 15)
IS (n 5 17)
P value
Age 6 SD (median) Gender (male %) Hypertension [n (%)] Diabetes mellitus [n (%)] Hyperlipidemia [n (%)] Smoking [n (%)] Admission NIHSS score (median) Lesion location (%) Subcortical [n (%)] Cortical [n (%)] Brain stem [n (%)] ASPECTS $7 [n (%)] Lesion volume Time taken for MRI scan (h)
66.6 6 17.1 (69) 7 (47) 11 (73) 3 (20) 2 (13) 1 (7) 12.3 6 6.4 (11)
64.4 6 16.2 (65) 9 (53) 10 (59) 5 (29) 6 (35) 6 (35) 4.6 6 3.5 (4)
.712 1 .472 .691 .229 .088 .001 .741
5 (33) 9 (60) 1 (7) 13 (87) 21.3 6 15.9 34.2 6 34
7 (41) 8 (47) 2 (12) 15 (88) 10.9 6 20.9 38.9 6 43.2
1 .128 .737
Abbreviations: ASPECTS, Alberta Stroke Program Early Computed Tomography Scale; ICH, intracerebral hemorrhage; IS, ischemic stroke; NIHSS, National Institutes of Health Stroke Scale.
ICH involved subcortical areas (thalamus 6 basal ganglia) in 60%, cortical areas in 33%, and pons in the remaining 7% of the patients. IS lesion locations did not significantly differ from ICH locations (Table 1). Although ICH volumes tended to be larger than IS lesion volumes on DWI, the differences were not statistically significant (Table 1). ASPECTS scores were also similar between the groups with 87% and 88% of patients having ASPECTS scores of 7 or more. ICH volumes and ASPECTS ICH scores were similar on the NCCT and DWI scans. When interpreted individually by the reviewers, the mean sensitivity and specificity of DWI at b1000 for detection of ICH were 94.2 6 8.0% (range, 83%-100%) and 93.5 6 7.7% (range, 85%-100%), respectively. The observed inter-rater agreement was excellent with a free marginal k of .79 and a fixed marginal k of .84. In patients with IS, NCCT showed signs of early ischemia in none of the patients, whereas DWI images were positive in all patients with IS.
Discussion The main findings of this pilot study are that DW-MRI obtained at b1000 is highly sensitive and specific for the diagnosis of ICH with an excellent rate of inter-rater agreement. Furthermore, as expected the sensitivity and specificity of DWI at b1000 was higher than NCCT for detection of IS, although all hemorrhages were also clearly seen on NCCT. The locations of ICH in the present study are typical of the distribution observed in clinical practice. These findings suggest that the diagnostic yield of b1000 DW-MRI in the case of acute stroke may be higher than the currently used gold standard of NCCT. Therefore, our findings raise the hypothesis that a short 5-minute test of DW-MRI at b1000 may be all that is needed for
the accurate diagnosis of IS and ICH in the ED. This hypothesis prompts us to suggest a randomized study comparing NCCT with DWI-only MRI in patients presenting to the ED with suspected stroke to evaluate the impacts on correct diagnosis and outcome in such patients. The present study differs from earlier studies; in this the evaluated DWI images were obtained at b1000 rather than at b0. Images obtained at b0 are more T2 weighted and may be useful in detecting deoxygenated blood. However, many fresh bleeds are not enough deoxygenated yet and therefore b0 imaging may fall short in diagnosing these bleeds in comparison with the results seen with b1000.11 This may also explain why in previous studies SWI appeared to be more sensitive than DWI for hematoma detection and especially for hemorrhagic transformations. Clearly, the present results showing very high sensitivity and specificity of DWI at b1000 in detecting hemorrhage and the higher resolution afforded by newer MR equipment should prompt future studies to compare DWI at b1000 with SWI. The present pilot study also raises questions about the significance of some of the DWI findings at b1000. Thus, the significance of the presence of the hyperintense bright rim surrounding the core of hemorrhage is not clear. Although some studies using positron emission tomography have suggested that this area may represent abnormal metabolism, others have attributed it to a blooming artifact.9,11,12,18-20 Future studies should determine whether the characteristics of this zone (eg, width and intensity) have any impact on outcome or any relevance to ICH etiology (eg, hypertensive versus amyloid related versus malformations). In this context, it should also be mentioned that microhemorrhages on SWI sometimes correspond to bright lesions on DWI. Thus, including patients with microhemorrhages into a study comparing DWI at b1000 with SWI would possibly allow
G. KEIGLER ET AL.
4
Figure 1. DWI in ICH. Appearance of a right subcortical ICH on noncontrast computed tomography (A) and DWI at b1000 (B). Note the similar appearance of the hematoma in all projections on both modalities. Abbreviations: DWI, diffusion-weighted imaging; ICH, intracerebral hemorrhage.
differentiating between ischemic and nonischemic lesions at the periphery of the microhematoma. Our study has several limitations. First, this is a small pilot study and as such the findings need to be corroborated in larger future studies and be regarded as hypothesis generating. Second, the time from symptom onset to imaging varied significantly in our patients but they were mostly done within the first 36 hours from admission. Although DWI at b1000 was able to detect all ICHs with a similar pattern in all our patients, it is possible that different patterns will emerge at different time points after onset and this needs to be further characterized in future studies. Third, it needs to be emphasized that performing an MRI as the screening test for all patients with possible stroke is not without possible drawbacks. First, MRI equipment and imaging cost more than CT and therefore it may be argued that MRI imaging in the setting of acute stroke is not cost effective. However, the difference in costs of equipment is not very large with new CT equipment averaging around US $1,000,000 and MRI equipment about US $1,200,000. Second, NCCT may be
advantageous to MRI in terms of relatively fast acquisition time (,2 minutes), and high safety profile including lack of contraindications, such as the presence of cardiac pacemaker, metal implants, claustrophobia, renal failure, or allergy to the contrast. The disadvantages of performing an NCCT are exposure to radiation and the low sensitivity to early brain ischemia. Third, it may be argued that MRI is not widely available in all hospitals compared with CT. We argue that this does not mean that the current state needs to continue unchanged and that similar to not using inaccurate methods such as pneumoencephalography or electroencephalography for the diagnosis of stroke, we should stop using NCCT when possible and instead use the tool that gives the most accurate diagnosis to date, for example, DW-MRI.
Conclusions The findings suggest that a brief DWI-only protocol obtained at b1000 may have higher sensitivity and specificity rates for detection of IS and ICH compared with
DWI B1000 IN ICH
NCCT. The cost effectiveness of this approach needs to be evaluated in future randomized studies.
References 1. Jauch EC, Saver JL, Adams HP Jr, et al. Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2013;44:870-947. 2. Fiebach J, Jansen O, Schellinger P, et al. Comparison of CT with diffusion-weighted MRI in patients with hyperacute stroke. Neuroradiology 2001;43:628-632. 3. Masdeu JC, Irimia P, Asenbaum S, et al. EFNS guideline on neuroimaging in acute stroke. Report of an EFNS task force. Eur J Neurol 2006;13:1271-1283. 4. Schellinger PD, Bryan RN, Caplan LR, et al. Evidencebased guideline: the role of diffusion and perfusion MRI for the diagnosis of acute ischemic stroke: report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Neurology 2010;75:177-185. 5. Wijman CA, Venkatasubramanian C, Bruins S, et al. Utility of early MRI in the diagnosis and management of acute spontaneous intracerebral hemorrhage. Cerebrovasc Dis 2010;30:456-463. 6. Kidwell CS, Chalela JA, Saver JL, et al. Comparison of MRI and CT for detection of acute intracerebral hemorrhage. JAMA 2004;292:1823-1830. 7. Kidwell CS, Wintermark M. Imaging of intracranial haemorrhage. Lancet Neurol 2008;7:256-267. 8. Kidwell CS, Saver JL, Mattiello J, et al. Diffusion-perfusion MR evaluation of perihematomal injury in hyperacute intracerebral hemorrhage. Neurology 2001;57: 1611-1617. 9. Schellinger PD, Fiebach JB. Intracranial hemorrhage: the role of magnetic resonance imaging. Neurocrit Care 2004;1:31-45. 10. Fiebach JB, Schellinger PD, Gass A, et al. Stroke magnetic resonance imaging is accurate in hyperacute intracere-
5
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
bral hemorrhage: a multicenter study on the validity of stroke imaging. Stroke 2004;35:502-506. Linfante I, Llinas RH, Caplan LR, et al. MRI features of intracerebral hemorrhage within 2 hours from symptom onset. Stroke 1999;30:2263-2267. Morita N, Harada M, Yoneda K, et al. A characteristic feature of acute haematomas in the brain on echoplanar diffusion-weighted imaging. Neuroradiology 2002;44:907-911. Lin DD, Filippi CG, Steever AB, et al. Detection of intracranial hemorrhage: comparison between gradient-echo images and b(0) images obtained from diffusion-weighted echo-planar sequences. AJNR Am J Neuroradiol 2001; 22:1275-1281. 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. Luby M, Bykowski JL, Schellinger PD, et al. Intra- and interrater reliability of ischemic lesion volume measurements on diffusion-weighted, mean transit time and fluid-attenuated inversion recovery MRI. Stroke 2006; 37:2951-2956. Eichel R, Hur TB, Gomori JM, et al. Use of DWI-only MR protocol for screening stroke mimics. J Neurol Sci 2013; 328:37-40. Pexman JH, Barber PA, Hill MD, et al. Use of the Alberta Stroke Program Early CT Score (ASPECTS) for assessing CT scans in patients with acute stroke. AJNR Am J Neuroradiol 2001;22:1534-1542. Schellinger PD, Fiebach JB, Hoffmann K, et al. Stroke MRI in intracerebral hemorrhage: is there a perihemorrhagic penumbra? Stroke 2003;34:1674-1679. Zazulia AR, Diringer MN, Videen TO, et al. Hypoperfusion without ischemia surrounding acute intracerebral hemorrhage. J Cereb Blood Flow Metab 2001;21:804-810. Zazulia AR, Videen TO, Powers WJ. Transient focal increase in perihematomal glucose metabolism after acute human intracerebral hemorrhage. Stroke 2009;40: 1638-1643.
Background: Noncontrast computed tomography (NCCT) is the gold standard to detect intracerebral hemorrhage (ICH) and ischemic stroke (IS) in patients presenting with acute focal syndromes. Diffusion-weighted magnetic resonance imaging (DW-MRI) obtained at b1000 is highly sensitive to identify acute IS but its sensitivity and specificity to detect ICH has not been systematically studied. Methods: Patients with a diagnosis of ICH on NCCT were prospectively enrolled and underwent DW-MRI at b1000. Patients with suspected ischemia and a negative NCCT served as controls. All diffusion-weighted imaging (DWI) scans were evaluated blindly by 4 experienced raters. Sensitivity, specificity, and inter-rater variability of the DWI b1000 scans for detection of ICH were determined. Results: In this preliminary pilot study, 15 patients with ICH and 17 patients with IS were included. All ICH lesions seen on NCCT showed a typical pattern on DW-MRI at b1000 with a hypointense core surrounded by a hyperintense rim. ICH volumes and size were similar on NCCT and MRI. All cases of IS were identified on the DWI scans but none were apparent on NCCT. The mean sensitivity and specificity of DW-MRI at b1000 for ICH were 94% and 93.5%, respectively, and the inter-rater variability for ICH detection on DWI was excellent (k 5 .84). Conclusions: DW-MRI at b1000 has a diagnostic yield similar to NCCT for detecting ICH and superior to NCCT for detecting IS. Therefore, DW-MRI may be considered as the initial screening tool for imaging patients presenting with focal neurologic symptoms suggestive of stroke. Key Words: Stroke—diffusion-weighted MRI—cerebrovascular disease— intracerebral hemorrhage. Ó 2014 by National Stroke Association
From the *Department of Neurology, Hadassah-Hebrew University Medical Center, Jerusalem; †Department of Radiology, HadassahHebrew University Medical Center, Jerusalem; and ‡Department of Neurosurgery, Hadassah-Hebrew University Medical Center, Jerusalem, Israel. Received November 28, 2013; revision received January 6, 2014; accepted February 1, 2014. This study was supported by the Peritz and Chantal Scheinberg Cerebrovascular Research Fund and the Sol Irwin Juni Trust Fund. All authors have nothing to disclose. Address correspondence to Ronen R. Leker, MD, FAHA, Department of Neurology, Hebrew University-Hadassah Medical Center, PO Box 12000, Jerusalem 91120, Israel. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.02.005
Noncontrast computed tomography (NCCT) is usually used as the initial imaging modality in patients presenting to the Emergency Department (ED) with acute focal neurologic symptoms and signs suggestive of stroke.1 It is primarily useful in identifying patients with intracerebral hemorrhage (ICH) with sensitivity and specificity rates that approach 100% and may also be useful for identifying signs of early ischemic stroke (IS), although with much lower sensitivity and specificity rates.1-4 Magnetic resonance imaging (MRI) and especially diffusionweighted imaging (DWI) is much more sensitive and specific than NCCT for the detection of early ischemia,1-4 and it is currently recommended as a screening tool for patients with IS. Over the past decade, several studies evaluated different MRI protocols to estimate their ability to detect
Journal of Stroke and Cerebrovascular Diseases, Vol. -, No. - (---), 2014: pp 1-5
1
G. KEIGLER ET AL.
2
early products of hemoglobin and distinguish hyperacute hemorrhage from ischemia.5-12 These studies found that several MRI sequences and, in particular, susceptibilityweighted imaging (SWI) including T2* may be superior to NCCT for the detection of ICH and hemorrhagic transformation of IS.5-13 In addition, MRI may contribute to determining the cause of ICH by distinguishing among cerebral amyloid angiopathy, hypertensive bleeds, and vascular malformations.5 Other MRI sequences including DWI obtained at b0 have also been reported to detect ICH but with a slightly reduced sensitivity and specificity compared with susceptibility images.11 Therefore, it may be claimed that a full MRI protocol may be needed in the ED for stroke imaging. However, a full multiparametric stroke MRI takes significantly longer to complete and interpret than an NCCT and may thus delay therapy in patients with IS who are eligible for thrombolysis or endovascular recanalization. In contrast, DWI can be acquired rapidly,4 is associated with outcome, and has an excellent interrater variability.14,15 Furthermore, we have previously shown that using a combination of clinical and DWI parameters can rapidly identify patients with acute ischemia and distinguish them from stroke mimics.16
Patients and Methods All patients with acute IS admitted to our institution are enrolled in a prospective database. The Institutional Review Board has granted a general permission to collect routine research data on all stroke patients. As per the institutional protocol, all patients with suspected stroke/transient ischemic attack are seen by a neurologist and undergo NCCT of the brain. Patients with ICH were prospectively identified on NCCT. All included patients underwent a screening DWI-only MRI protocol that included DWI and apparent diffusion coefficient (ADC) maps. Scans were mostly acquired on a 3 T Siemens MRI scanner (Siemens, Erlangen, Germany) with the following characteristics: field of view 230 3 230 mm, acquisition matrix 128 3 100, slice/gap 5 5/0 (in millimeters), repetition time/echo time 5 4600/93 (in milliseconds), b values 5 0, 500, 1000. In a few instances, scans were acquired on a General Electric 1.5 T MRI scanner (GE healthcare, Milwaukee, WI) with the following characteristics: field of view 240 3 217 mm, acquisition matrix 128 3 100, slice/gap 5 5/1 (in millimeters), repetition time/echo time 5 3800/96 (in milliseconds), b values 5 0, 500, 1000. MRI scans were obtained as soon as possible after admission. Consecutive patients presenting to the ED with acute neurologic symptoms and signs who had an NCCT scan negative for ICH also underwent DWI-only protocol and were included as controls for this study. Scans were obtained as soon as possible and took less than 5 minutes to complete in all cases. Patients with evidence for bright hyperintense lesions on DWI that were hypointense on the ADC map were classified
as having an acute IS on DWI. Patients with evidence of a hypointense core surrounded by a hyperintense bright rim on DWI were classified as having an ICH on DWI. ICH volumes and locations were calculated on the NCCT and MRI scans using the ABC/2 formula, which takes into account the largest diameters of the hematoma multiplied by the number of slices it appears on, and ICH locations were determined using the Alberta Stroke Program Early CT Scale (ASPECTS) system subfields.17 For evaluation of sensitivity, specificity, and inter-rater variability, DWI studies were interpreted by senior neuroradiologist (J.M.G.) or vascular neurosurgeon (J.E.C.) and stroke neurologists (R.E. or R.R.L.) in all cases. Evaluators were blinded to the patient’s neurologic status, the NCCT scan, and the final discharge diagnosis from the Department of Neurology. Baseline demographics, risk factor profiles, clinical characteristics, and final discharge diagnoses and outcome were accrued. Neurologic deficits were graded with the National Institutes of Health Stroke Scale. Statistical evaluations were performed with the SigmaStat package (SPSS, IBM, Armonk, NY). For univariate analysis, patients with positive and negative DWI results were compared using the Student t test or chi-square test. Sensitivity and specificity rates were obtained for each rater, and the data presented represent the mean 6 standard deviation and ranges. Kappa (k) statistics for multiple evaluators were used for the evaluation of inter-rater variability. k values greater than .7 were considered good and greater than .8 excellent.
Results For this pilot study, we included 15 patients with ICH on the admission based on NCCT scan and 17 patients with similar stroke severity that had evidence for IS on the DWI scan. These controls were specifically used to determine if hemorrhagic transformations of ISs from true ICH could be easily discerned and to help determine the sensitivity and specificity of DWI at b1000 in small lesions. The baseline characteristics of the included patients are shown in Table 1. The time from symptom onset to DWI data acquisition in ICH patients ranged from 3-144 hours with a median of 25 hours and did not differ from IS patients (median 28 hours; range .18-168 hours). All hemorrhages were clearly identified on NCCT scans. In all patients with ICH, a similar pattern was seen on the MRI b1000 images: a hypointense core that measured similar to ICH boundaries on NCCT surrounded by a bright hyperintense rim not seen on the initial NCCT (Fig 1). The boundaries of this lesion coregistered with the hematoma on NCCT. In contrast, acute IS on DWI scans showed the typical bright hyperintense DWI core, which was hypointense on the ADC maps.
DWI B1000 IN ICH
3
Table 1. Univariate analysis comparing patients with ICH and IS Variable/group
ICH (n 5 15)
IS (n 5 17)
P value
Age 6 SD (median) Gender (male %) Hypertension [n (%)] Diabetes mellitus [n (%)] Hyperlipidemia [n (%)] Smoking [n (%)] Admission NIHSS score (median) Lesion location (%) Subcortical [n (%)] Cortical [n (%)] Brain stem [n (%)] ASPECTS $7 [n (%)] Lesion volume Time taken for MRI scan (h)
66.6 6 17.1 (69) 7 (47) 11 (73) 3 (20) 2 (13) 1 (7) 12.3 6 6.4 (11)
64.4 6 16.2 (65) 9 (53) 10 (59) 5 (29) 6 (35) 6 (35) 4.6 6 3.5 (4)
.712 1 .472 .691 .229 .088 .001 .741
5 (33) 9 (60) 1 (7) 13 (87) 21.3 6 15.9 34.2 6 34
7 (41) 8 (47) 2 (12) 15 (88) 10.9 6 20.9 38.9 6 43.2
1 .128 .737
Abbreviations: ASPECTS, Alberta Stroke Program Early Computed Tomography Scale; ICH, intracerebral hemorrhage; IS, ischemic stroke; NIHSS, National Institutes of Health Stroke Scale.
ICH involved subcortical areas (thalamus 6 basal ganglia) in 60%, cortical areas in 33%, and pons in the remaining 7% of the patients. IS lesion locations did not significantly differ from ICH locations (Table 1). Although ICH volumes tended to be larger than IS lesion volumes on DWI, the differences were not statistically significant (Table 1). ASPECTS scores were also similar between the groups with 87% and 88% of patients having ASPECTS scores of 7 or more. ICH volumes and ASPECTS ICH scores were similar on the NCCT and DWI scans. When interpreted individually by the reviewers, the mean sensitivity and specificity of DWI at b1000 for detection of ICH were 94.2 6 8.0% (range, 83%-100%) and 93.5 6 7.7% (range, 85%-100%), respectively. The observed inter-rater agreement was excellent with a free marginal k of .79 and a fixed marginal k of .84. In patients with IS, NCCT showed signs of early ischemia in none of the patients, whereas DWI images were positive in all patients with IS.
Discussion The main findings of this pilot study are that DW-MRI obtained at b1000 is highly sensitive and specific for the diagnosis of ICH with an excellent rate of inter-rater agreement. Furthermore, as expected the sensitivity and specificity of DWI at b1000 was higher than NCCT for detection of IS, although all hemorrhages were also clearly seen on NCCT. The locations of ICH in the present study are typical of the distribution observed in clinical practice. These findings suggest that the diagnostic yield of b1000 DW-MRI in the case of acute stroke may be higher than the currently used gold standard of NCCT. Therefore, our findings raise the hypothesis that a short 5-minute test of DW-MRI at b1000 may be all that is needed for
the accurate diagnosis of IS and ICH in the ED. This hypothesis prompts us to suggest a randomized study comparing NCCT with DWI-only MRI in patients presenting to the ED with suspected stroke to evaluate the impacts on correct diagnosis and outcome in such patients. The present study differs from earlier studies; in this the evaluated DWI images were obtained at b1000 rather than at b0. Images obtained at b0 are more T2 weighted and may be useful in detecting deoxygenated blood. However, many fresh bleeds are not enough deoxygenated yet and therefore b0 imaging may fall short in diagnosing these bleeds in comparison with the results seen with b1000.11 This may also explain why in previous studies SWI appeared to be more sensitive than DWI for hematoma detection and especially for hemorrhagic transformations. Clearly, the present results showing very high sensitivity and specificity of DWI at b1000 in detecting hemorrhage and the higher resolution afforded by newer MR equipment should prompt future studies to compare DWI at b1000 with SWI. The present pilot study also raises questions about the significance of some of the DWI findings at b1000. Thus, the significance of the presence of the hyperintense bright rim surrounding the core of hemorrhage is not clear. Although some studies using positron emission tomography have suggested that this area may represent abnormal metabolism, others have attributed it to a blooming artifact.9,11,12,18-20 Future studies should determine whether the characteristics of this zone (eg, width and intensity) have any impact on outcome or any relevance to ICH etiology (eg, hypertensive versus amyloid related versus malformations). In this context, it should also be mentioned that microhemorrhages on SWI sometimes correspond to bright lesions on DWI. Thus, including patients with microhemorrhages into a study comparing DWI at b1000 with SWI would possibly allow
G. KEIGLER ET AL.
4
Figure 1. DWI in ICH. Appearance of a right subcortical ICH on noncontrast computed tomography (A) and DWI at b1000 (B). Note the similar appearance of the hematoma in all projections on both modalities. Abbreviations: DWI, diffusion-weighted imaging; ICH, intracerebral hemorrhage.
differentiating between ischemic and nonischemic lesions at the periphery of the microhematoma. Our study has several limitations. First, this is a small pilot study and as such the findings need to be corroborated in larger future studies and be regarded as hypothesis generating. Second, the time from symptom onset to imaging varied significantly in our patients but they were mostly done within the first 36 hours from admission. Although DWI at b1000 was able to detect all ICHs with a similar pattern in all our patients, it is possible that different patterns will emerge at different time points after onset and this needs to be further characterized in future studies. Third, it needs to be emphasized that performing an MRI as the screening test for all patients with possible stroke is not without possible drawbacks. First, MRI equipment and imaging cost more than CT and therefore it may be argued that MRI imaging in the setting of acute stroke is not cost effective. However, the difference in costs of equipment is not very large with new CT equipment averaging around US $1,000,000 and MRI equipment about US $1,200,000. Second, NCCT may be
advantageous to MRI in terms of relatively fast acquisition time (,2 minutes), and high safety profile including lack of contraindications, such as the presence of cardiac pacemaker, metal implants, claustrophobia, renal failure, or allergy to the contrast. The disadvantages of performing an NCCT are exposure to radiation and the low sensitivity to early brain ischemia. Third, it may be argued that MRI is not widely available in all hospitals compared with CT. We argue that this does not mean that the current state needs to continue unchanged and that similar to not using inaccurate methods such as pneumoencephalography or electroencephalography for the diagnosis of stroke, we should stop using NCCT when possible and instead use the tool that gives the most accurate diagnosis to date, for example, DW-MRI.
Conclusions The findings suggest that a brief DWI-only protocol obtained at b1000 may have higher sensitivity and specificity rates for detection of IS and ICH compared with
DWI B1000 IN ICH
NCCT. The cost effectiveness of this approach needs to be evaluated in future randomized studies.
References 1. Jauch EC, Saver JL, Adams HP Jr, et al. Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2013;44:870-947. 2. Fiebach J, Jansen O, Schellinger P, et al. Comparison of CT with diffusion-weighted MRI in patients with hyperacute stroke. Neuroradiology 2001;43:628-632. 3. Masdeu JC, Irimia P, Asenbaum S, et al. EFNS guideline on neuroimaging in acute stroke. Report of an EFNS task force. Eur J Neurol 2006;13:1271-1283. 4. Schellinger PD, Bryan RN, Caplan LR, et al. Evidencebased guideline: the role of diffusion and perfusion MRI for the diagnosis of acute ischemic stroke: report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Neurology 2010;75:177-185. 5. Wijman CA, Venkatasubramanian C, Bruins S, et al. Utility of early MRI in the diagnosis and management of acute spontaneous intracerebral hemorrhage. Cerebrovasc Dis 2010;30:456-463. 6. Kidwell CS, Chalela JA, Saver JL, et al. Comparison of MRI and CT for detection of acute intracerebral hemorrhage. JAMA 2004;292:1823-1830. 7. Kidwell CS, Wintermark M. Imaging of intracranial haemorrhage. Lancet Neurol 2008;7:256-267. 8. Kidwell CS, Saver JL, Mattiello J, et al. Diffusion-perfusion MR evaluation of perihematomal injury in hyperacute intracerebral hemorrhage. Neurology 2001;57: 1611-1617. 9. Schellinger PD, Fiebach JB. Intracranial hemorrhage: the role of magnetic resonance imaging. Neurocrit Care 2004;1:31-45. 10. Fiebach JB, Schellinger PD, Gass A, et al. Stroke magnetic resonance imaging is accurate in hyperacute intracere-
5
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
bral hemorrhage: a multicenter study on the validity of stroke imaging. Stroke 2004;35:502-506. Linfante I, Llinas RH, Caplan LR, et al. MRI features of intracerebral hemorrhage within 2 hours from symptom onset. Stroke 1999;30:2263-2267. Morita N, Harada M, Yoneda K, et al. A characteristic feature of acute haematomas in the brain on echoplanar diffusion-weighted imaging. Neuroradiology 2002;44:907-911. Lin DD, Filippi CG, Steever AB, et al. Detection of intracranial hemorrhage: comparison between gradient-echo images and b(0) images obtained from diffusion-weighted echo-planar sequences. AJNR Am J Neuroradiol 2001; 22:1275-1281. 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. Luby M, Bykowski JL, Schellinger PD, et al. Intra- and interrater reliability of ischemic lesion volume measurements on diffusion-weighted, mean transit time and fluid-attenuated inversion recovery MRI. Stroke 2006; 37:2951-2956. Eichel R, Hur TB, Gomori JM, et al. Use of DWI-only MR protocol for screening stroke mimics. J Neurol Sci 2013; 328:37-40. Pexman JH, Barber PA, Hill MD, et al. Use of the Alberta Stroke Program Early CT Score (ASPECTS) for assessing CT scans in patients with acute stroke. AJNR Am J Neuroradiol 2001;22:1534-1542. Schellinger PD, Fiebach JB, Hoffmann K, et al. Stroke MRI in intracerebral hemorrhage: is there a perihemorrhagic penumbra? Stroke 2003;34:1674-1679. Zazulia AR, Diringer MN, Videen TO, et al. Hypoperfusion without ischemia surrounding acute intracerebral hemorrhage. J Cereb Blood Flow Metab 2001;21:804-810. Zazulia AR, Videen TO, Powers WJ. Transient focal increase in perihematomal glucose metabolism after acute human intracerebral hemorrhage. Stroke 2009;40: 1638-1643.