Increased Corticospinal Tract Fractional Anisotropy Can Discriminate Stroke Onset Within the First 4.5 Hours
Background and Purpose—The role of diffusion tensor imaging in determining stroke age remains unclear. We tested the ability of diffusion tensor imaging metrics to discriminate ischemic stroke <4.5 hours of onset.
Methods—We enrolled 60 consecutive patients for multimodal 1.5 T MRI within 12 hours of middle cerebral artery ischemic stroke onset. We measured fractional anisotropy (FA), mean diffusivity (MD), apparent diffusion coefficient (ADC), and T2-weighted signal intensity in affected ipsilateral and unaffected contralateral deep gray matter, cortical gray matter, deep white matter in the corticospinal tract (CST), and subcortical white matter and calculated ipsilateral-to-contralateral ratios (r). Hyperintensity in infarcted tissue was considered fluid-attenuated inversion recovery-positive.
Results—We analyzed the 48 patients (17 women; mean age, 68±14 years) with known onset. In 25 (52.1%) patients, onset was ≤4.5 hours (mean, 182.3±65.6 minutes). Variables differing significantly between infarcts <4.5 hours and >4.5 hours were rFA CST (P=0.001), rMD cortical gray matter (P=0.036), rADC cortical gray matter (P=0.009), rT2 CST (P=0.006), and fluid-attenuated inversion recovery (P<0.001). rFA at CST was the most reliable to discriminate infarcts <4.5 hours (Goodman-Kruskal=0.76). The sensitivity, specificity, and positive and negative predictive values for infarct <4.5 hours of onset by rFA at CST >0.970 were 93.8%, 84.6%, 88.2%, and 91.7%, respectively.
Conclusions—These preliminary results suggest rFA at CST may be a surrogate marker of acute stroke age.
Knowing the onset of acute stroke is a prerequisite for intravenous tissue plasminogen activator, because this treatment is approved only within 4.5 hours of onset.1 Reliable MRI (MRI) surrogate markers of lesion age are needed, because 14% to 28% of patients discover stroke on waking up.2 Fluid-attenuated inversion recovery (FLAIR) can identify patients within 4.5 hours of onset with moderate accuracy.3 Diffusion tensor imaging (DTI) measures anisotropic water diffusion as fractional anisotropy (FA).4 FA can detect microstructural changes attributable to ischemia, but the value of DTI-metrics as a biological tissue clock remains unclear. We tested the utility of DTI-metrics for differentiating ischemic strokes with onset of <4.5 hours.
Sixty consecutive patients with first-ever middle cerebral artery territory infarction underwent multimodal MRI on a 1.5 T scanner (Gyroscan Intera; Philips Medical Systems, Best, the Netherlands) within 12 hours of symptom onset. Forty-eight patients with known onset were analyzed (17 women; mean age, 68±14 years). Our ethics committee approved the study, and all patients provided written informed consent.
DTI were acquired by using single-shot echo-planar imaging sequences with the sensitivity encoding parallel-imaging scheme. Diffusion-sensitized gradients were applied along 15 noncollinear directions with a b-value of 1000 s/mm2. Other DTI acquisition parameters were TR/TE, 6795/72 ms; 23×23-cm FOV; 112×112 matrix size; and 2.05×2.05×3 mm voxel size. DTI acquisition took about 3 minutes.
Diffusion tensor images were coregistered; 2 neuroradiologists used NeuroScape 2.0 MR Stroke Edition (Olea Medical, La Ciotat, France) to place free-hand regions of interest (ROIs) on deep and cortical gray matter, deep white matter at the level of the corticospinal tract (CST), and subcortical white matter in the slice, where the infarct had the largest diameters on diffusion-weighted sequences. We measured FA, mean diffusivity (MD), apparent diffusion coefficient (ADC), and T2-weighted signal intensity in the ipsilateral affected side and in the homologous contralateral regions, and then calculated the ipsilateral-to-contralateral ratios (r). Mean ROI area was 47±10 mm2. Hyperintensity in infarcted tissue was considered FLAIR-positive; discordant FLAIR ratings were resolved by consensus. Interobserver agreements were calculated. The two measurements were averaged for statistical analysis by paired t tests (Minitab version 184.108.40.206; Minitab Inc, State College, Pennsylvania).
Onset was <4.5 hours in 25 patients (52.1%; 182.3±65.6 minutes). CST was affected in 35 patients (72.91%). Significant differences between infarcts <4.5 and >4.5 hours were found for rFA at CST (P=0.001; Figure in the online-only Data Supplement), rMD cortical gray matter (P=0.036), rADC cortical gray matter (P=0.009), rT2 at CST (P=0.006), and FLAIR (P<0.001; Table). Logistic binary regression models and receiver-operating characteristic curves demonstrated rFA at CST most reliably discriminated infarcts <4.5 hours (Goodman-Kruskal=0.76; Kendall=0.39; Figure 1), with sensitivity, specificity, and positive and negative predictive values for an rFA at CST cutoff of >0.970 being 93.8%, 84.6%, 88.2%, and 91.7%, respectively. Interobserver reliability for ROI measurements and FLAIR rating were good (interclass correlation coefficient 0.89 and 0.75, respectively).
This prospective study in patients with territorial middle cerebral artery stroke found rFA at CST increased in the first 4.5 hours, and rFA>0.970 reliably predicted whether <4.5 hours had elapsed since onset.
FA is a DTI-metric of the relative difference in water diffusivities along multiple axes expressed by 3 eigenvalues: lambda 1 (principal axis or axial diffusivity), representing water motion along the length of axons, and lambda 2 and lambda 3 (shorter perpendicular axes or radial diffusivity), indicating perpendicular water diffusion across the axon.5 Oligodendrocyte swelling from cytotoxic edema, one of the earliest morphological changes after stroke, results in extra-axonal water and compression of the axoplasm by swollen myelin sheaths, which translates to greater decline in radial than axial water diffusivity in infarcted white matter and increased rFA.5,6 Although the cellular basis of this phenomenon is not completely clear, it could explain the increase in rFA in the first 4.5 hours (Figure 2A). Our findings of elevated FA in hyperacute stroke are consistent with previous studies.4,6–11 As the infarct evolves over time, the breakdown of the axons eventually results in massive water accumulation in axon tracts that translates to hyperintensity in FLAIR and T2-weighted images.10 In the hyperacute stage, large decreases in FA suggest loss of cellular integrity with irreversible axonal injury not reflected on conventional MRI (Figure 2B). We detected significant FA alterations only in the CST, probably attributable to the greater anisotropy of the tightly packed single-directed fibers.
Certain limitations merit comment. Manual ROI placement is subject to operator bias; automated ROI analysis or voxel-based analysis may resolve this issue. Our small sample limits the power of our findings. The rFA cutoff value was not applicable to other regions of territorial infarction, although most patients (72.91%) had CST damage. The expected rate of CST involvement in real-world practice is, therefore, critical to the applicability of the technique, and an independent validation cohort would be necessary. Finally, analyzing radial diffusivities might provide additional information about stroke time.4
In conclusion, our preliminary results suggest that FA at CST can reliably discriminate ischemic strokes with onset <4.5 hours.
Sources of Funding
This work was partially supported by a grant from the Spanish Ministry of Health (Fondo Investigaciones Sanitarias, reference PI060745).
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.111.678110/-/DC1.
- Received November 2, 2012.
- Revision received December 17, 2012.
- Accepted December 19, 2012.
- © 2013 American Heart Association, Inc.
- Sakai K,
- Yamada K,
- Nagakane Y,
- Mori S,
- Nakagawa M,
- Nishimura T
- Liu Y,
- D’Arceuil HE,
- Westmoreland S,
- He J,
- Duggan M,
- Gonzalez RG,
- et al
- Ozsunar Y,
- Grant PE,
- Huisman TA,
- Schaefer PW,
- Wu O,
- Sorensen AG,
- et al
- Yang Q,
- Tress BM,
- Barber PA,
- Desmond PM,
- Darby DG,
- Gerraty RP,
- et al