Stroke Volume Evolution Following Endovascular Therapy on DWI and FLAIR

Christian Federau¹, Soren Christensen¹, Michael Mlynash¹, Jenny Tsai¹, Sun Kim¹, Greg Zaharchuk¹, Matus Straka¹, Nishant Mishra¹, Maarten Lansberg¹, and Greg Albers¹

¹Stanford University, Stanford, CA, United States

Synopsis

We studied the evolution of the infarct volume between an early post-revascularization scan (within 24 h of symptom onset) and day 5 in patients of the CRISP and DEFUSE 2 cohort studies. On the early post-revascularization scan, FLAIR lesions were smaller compared to DWI, but were larger at day 5. The early post-revascularization stroke volume on DWI, compared to FLAIR, was closer, and correlated better with the day 5 DWI and FLAIR lesion volumes. Together, our findings suggest that DWI is a better early surrogate marker of stroke volume.

Target Audience

Neurologists, neuroradiologists, and scientists with an interest in clinical stroke trials.

Introduction

The best time-point and imaging modality for assessment of final infarct volume in patients with stroke is unclear. Follow-up scans within 24h are easy to obtain, because the patients are most of the time still hospitalized, but may underestimate final infarct size, particularly in patients who do not reperfuse. Scans at a later time point might be more exact, but are more difficult to obtain, as patients might already be discharged home. We studied the evolution of the infarct volume assessed within 24 h of symptom onset, to the volume of the infarct on day 5, as assessed on both DWI and FLAIR MRI.

Methods

The CTP to predict Response to recanalization in Ischemic Stroke Project (CRISP1), and the Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution 22 (DEFUSE 2), were two NIH funded multicenter cohort studies of consecutive acute stroke patients who underwent intraarterial thrombectomy. The subset of patients of those two studies, who had DWI and FLAIR imaging both early post-revascularization and at day 5, were included in this study. Exclusion criteria were parenchymal bleeding (ECASS3 PH1 or PH2), malignant profile (infarct core >70ml), and technically inadequate studies. Regions of interest of the volume of the stroke were outlined by an experienced neuroradiologist. Lesion volumes were evaluated using median, interquartile range, and correlation plots. Statistical significance was assessed using two-tailed Wilcoxon signed rank and rank sum tests. We compared correlated and independent correlations using Steiger’s and Fisher’s Z-tests respectively. Significance level was set to α<0.05. Further, probabilistic stroke lesion maps, to visualize the anatomical extend of the lesions as function of time (early post-revascularization and day 5) and modality (DWI and FLAIR), were generated as follows: images with lesions on the left side were flipped, and all images, together with their regions of interest, were coregistered to the MNI template brain4 using MINC5. Finally, in the subset of patients with known stroke onset time, we evaluated the time dependence of the lesion growth on both modalities. We dichotomized the patients into groups according to the timing of their early post-revascularization scan: ≥18h vs <18 h after stroke onset.

Results

146/339 patients from the CRISP (91/201) and DEFUSE 2 (55/138) studies had DWI and FLAIR images acquired both early post-revascularization (median 18.0 [12.4-23.9] hours after symptom onset) and at day 5. Of these, 31 were excluded due to parenchymal bleeding, 2 due to malignant profile, and 4 due to technically inadequate images. Therefore 109 patients were included (Table 1). Early post-revascularization stroke lesions were centered on the basal ganglia (Fig. 1). Median infarct volume was smaller on FLAIR (19.4ml; [IQR] 7.0-42.3), compared to DWI (22.5ml; 11.3-63.0; p<0.0001). On day 5, median infarct volume was larger on FLAIR (52.1ml; 20.7-117.2) compared to DWI (35.5ml; 15.9-90.6; p<0.0001), and median volume growth was greater on FLAIR (28.9ml; 10.8-68.5) compared to DWI (10.0ml; 1.9-36.8; p<0.0001). Correlation between DWI and FLAIR lesion volumes was better at day 5 (r = 0.968) than early post-revascularization (r=0.923; p<0.001). Lesion volumes correlated strongly between early post-revascularization and day 5, but the correlation was significantly stronger for DWI than FLAIR (r=0.883 vs r=0.782; p=0.013) (Fig. 2). Further, early post-revascularization DWI volumes correlated better than FLAIR volumes with both day 5 FLAIR volumes (r=0.880 vs. 0.782; p<0.001) and day 5 DWI volumes (r=0.883 vs. 0.774; p<0.001) (Fig. 2). Finally regarding the timing, median volume growth on DWI was significantly smaller when the early post-revascularization scan was obtained ≥18h post stroke onset (4.6ml; -0.7-12.8), compared to <18h (13.3ml; 1.6-46.7; p=0.03), but were not significantly different on FLAIR (≥18h: 26.1ml; 11.8-56.9; <18h: 21.2ml; 5.4-56.8; p=0.65) (Fig. 3).

Conclusion

This study suggests that early post-revascularization DWI is better than FLAIR for predicting day 5 stroke lesion volume.

Acknowledgements

CF is supported by the Swiss National Science Foundation.

References

1. Lansberg et al, Int Stroke Conf, Feb 2016, Los Angeles 2. Lansberg et al, The Lancet Neurology 2012;11(10):860-867. 3. Larrue et al, Stroke 2001;32(2):438-441. 4. Fonov et al, NeuroImage 2011;54(1):313-327. 5. http://www.bic.mni.mcgill.ca/ServicesSoftware/MINC.

Figures

Table 1: Baseline characteristics. Data are n (%) or median (interquartile range), unless otherwise specified. NIHSS = NIH stroke scale; Onset to therapy = time between symptom onset to beginning of endovascular therapy; ICA = internal carotid artery; MCA = middle cerebral artery.

Fig. 1: Probabilistic stroke lesions maps, showing the extension of stroke lesion as drawn on DWI (b=1000 s/mm2) and FLAIR, immediately after endovascular thrombectomy and at day 5 post symptoms onset. The lesion extensions are centered on the basal ganglia, and are more extended on DWI (b=1000 s/mm2) than FLAIR after revascularization, but slightly more extended on FLAIR at day 5.

Fig. 2: Correlation plots between DWI and FLAIR, obtained early post-revascularization and at day 5. The dotted line represents the equality line. The straight line represents the trend line. Early post-revascularization, infarct volumes were generally larger on DWI than on FLAIR, and correlated better with infarct volumes measured at day 5, on both DWI and FLAIR.

Fig. 3: Time dependence of the volume growth between early post-revascularization and day 5, as measured on DWI and on FLAIR respectively. The vertical line has been drawn at 18h post symptom onset. This data suggests that stroke lesion growth occur before 18h on DWI, but continues for a more prolonged time on FLAIR.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)

0602