Abstract
Background and Purpose—The aim of this study was to assess the frequency and extent of early diffusion-weighted imaging (DWI) lesion reversal after endovascular therapy and to determine whether early reversal is sustained or transient.
Methods—MRI with DWI perfusion imaging was performed before (DWI 1) and within 12 hours after (DWI 2) endovascular treatment; follow-up MRI was obtained on day 5. Both DWIs were coregistered to follow-up MRI. Early DWI reversal was defined as the volume of the DWI 1 lesion that was not superimposed on the DWI 2 lesion. Permanent reversal was the volume of the DWI 1 lesion not superimposed on the day 5 infarct volume. Associations between early DWI reversal and clinical outcomes in patients with and without reperfusion were assessed.
Results—A total of 110 patients had technically adequate DWI before endovascular therapy (performed median [interquartile range], 4.5 [2.8–6.2] hours after onset); 60 were eligible for this study. Thirty-two percent had early DWI reversal >10 mL; 17% had sustained reversal. The median volume of permanent reversal at 5 days was 3 mL (interquartile range, 1.7–7.0). Only 2 patients (3%) had a final infarct volume that was smaller than their baseline DWI lesion. Early DWI reversal was not an independent predictor of clinical outcome and was not associated with early reperfusion.
Conclusions—Early DWI reversal occurred in about one third of patients after endovascular therapy; however, reversal was often transient and was not associated with a significant volume of tissue salvage or favorable clinical outcome.
Introduction
Diffusion-weighted imaging (DWI) can identify early infarction and estimate ischemic core lesions.1–7 The random Brownian motion of water protons determines DWI signal intensity, and the apparent diffusion coefficient (ADC) values are a quantitative measure of diffusion.8 In both animals and human models, ADC values decrease after brain tissue ischemia and are related to metabolism failure and cytotoxic edema.9
Numerous reports have documented that early DWI lesions in patients with ischemic stroke can be reversible; especially after reperfusion therapy.10–15 The incidence, permanence, and clinical significance of DWI reversibility have not been well characterized. Conflicting reports about the incidence and prognostic significance of DWI reversal have been published.12–16 Some authors think that DWI reversal has clinical significance, whereas others have concluded that DWI reversal is uncommon and unlikely to be clinically meaningful.13,14 The aim of this study was to assess the frequency and the extent of early DWI reversal after endovascular reperfusion therapy and to determine whether early reversal is sustained or transient.
Methods
This is a substudy of Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution (DEFUSE) 2. The design, methodology, and primary results of DEFUSE 217 have been reported. This study was approved by the local institutional review boards, and all patients provided informed consent. MRI with DWI (DWI 1) and perfusion-weighted imaging was performed at baseline, before endovascular treatment, and within 12 hours after the procedure (DWI 2). Fluid attenuated inversion recovery (FLAIR) imaging was also performed at 5-day follow-up. Patients with baseline DWI lesions >10 mL were included in this analysis.
The volume of hypoperfusion on perfusion-weighted imaging (Tmax >6 s) was obtained at the time of patient enrollment in DEFUSE 2 using an automated software program: RApid processing of Perfusion and Diffusion (RAPID).18 Reperfusion was defined as per the primary DEFUSE 2 study: a >50% of reduction in the perfusion lesion volume (Tmax >6 s) between the baseline and early post-treatment MRI. Angiographic analysis, based on thrombolysis in cerebral infarction reperfusion scores, was used to determine reperfusion if the baseline or early follow-up perfusion study was not performed or technically inadequate. Thrombolysis in cerebral infarction 2B was defined as partial reperfusion of 50% to 99% of the vascular distribution of the occluded artery and thrombolysis in cerebral infarction 3 as complete reperfusion (a thrombolysis in cerebral infarction 2B or 3 qualified as successful reperfusion). Favorable clinical response was defined as an improvement in the National Institutes of Health Stroke Scale of ≥8 points between baseline and day 30 or a National Institutes of Health Stroke Scale score ≤1 at day 30. Good functional outcome was defined as a modified Rankin Scale score ≤2 at day 90. Associations between reperfusion and early DWI reversal were also assessed in this study.
Image Analysis
DWI 1, DWI 2, and day 5 FLAIR lesions were outlined using MIPAV software (http://mipav.cit.nih.gov/) by 2 raters (H.M.W., M.I.). DWI scans were windowed by the readers to determine the visually optimal level to outline the lesions. Regions of interest were manually drawn using the power paint function and levelset regions of interest tool to outline lesions. Lesions on the DWI 2 b1000 maps were identified with the benefit of reference to the DWI 1 b1000 maps and ADC values to avoid regions of T2 shine-through. Wilcoxon matched pair signed-rank test and Hodges–Lehman estimate of the median difference were performed to determine consistency in semimanual volume measurements among raters. Coregistration of DWI 1 and DWI 2 to the day 5 FLAIR images was performed by MINC tools (Montreal Neurological Institute, McGill University, Montreal, Canada) and verified by 2 reviewers (S.C., M.I.). Overlap versus nonoverlap voxels in the coregistered images were compared, and corresponding volumes were calculated by Matlab (R2013a; Mathworks, Natick, MA).
Early DWI reversal was assessed using the coregistered images, and each lesion was superimposed into day 5 FLAIR space. We assessed whether the lesions that exhibited early DWI reversal had transient versus sustained reversal. Early DWI reversal was considered present if a portion of the DWI 1 lesion not superimposed by DWI 2 was >10 mL, and permanent DWI reversal was defined as DWI 1 not superimposed by day 5 FLAIR >10 mL. These definitions were adopted based on previous studies.13,14
The changes in mean ADC values between baseline (ADC 1) and early follow-up after reperfusion therapy (ADC 2) were also determined for the following volumes: (1) regions that demonstrated transient DWI reversal, (2) regions with permanent DWI reversal; and (3) regions with no DWI reversal.
The prognostic significance of early reversal was assessed in a regression model to determine whether transient reversal was an independent predictor of good outcome. Besides the baseline independent predictors (age, DWI volume, and National Institutes of Health Stroke Scale) that had previously been identified in the DEFUSE studies,17,19 time from onset to treatment was also included in this multivariate analysis. Statistical analysis was performed with IBM SPSS Statistics (version 20).
Results
All 110 patients enrolled in the DEFUSE 2 study had a DWI performed before endovascular therapy. Of these, 10 cases had no early follow-up DWI after endovascular therapy and 10 had no FLAIR images at day 5, and 30 patients had baseline DWI lesions ≤10 mL. Therefore, 60 patients were eligible for this substudy, and their clinical characteristics are summarized in Table 1. The baseline MRI was performed 4.5 hours (interquartile range [IQR], 2.8–6.2) after symptom onset, and early follow-up MRI was performed 3.0 hours (IQR, 1.3–5.9) after completion of endovascular therapy. Onset to start of endovascular treatment (femoral puncture) was 5.9 hours (IQR, 4.0–7.9). The median difference was 0.5 mL in DWI volumes independently assessed by 2 readers (95% confidence interval, 0.0–2.0; P=0.201). Coregistration did not alter the DWI volumes; the original volumes and the volumes obtained after coregistration did not differ (P=0.99 for DWI 1 and P=0.95 for DWI 2).
Baseline Characteristics, Clinical Outcomes, and Lesion Volumes
The coregistered DWI 1 median (IQR) lesion volume was 26.8 mL (13.8–56.9) and DWI 2 median (IQR) volume was 41.3 mL (18.1–92.8). The median (IQR) volume of early DWI reversal was 7.3 mL (4.1–13.0 mL; median [IQR], 33.3% [15.4%–46.2%] of the DWI 1 lesion). At day 5, the median volume of permanent DWI reversal was 3 mL (IQR, 1.7–7.0), representing a median of 12.8% of the DWI 1 volume. Lesion growth at day 5 was considerably larger than the volume of permanent DWI reversal volume, resulting in day 5 infarct volumes that were a median of 61.5 mL larger than the DWI 1 volume (Table 2). Only 2 patients had day 5 FLAIR lesion volumes that were smaller than their DWI 1 volume (6.2 and 7.0 mL smaller; Figure, right).
DWI Reversal Volumes and Lesion Growth in Patients With and Without Reperfusion
Left, Example of a patient with transient diffusion-weighted imaging (DWI) reversal. The time between symptom onset and MRI was 2.4 hours, the endovascular procedure began at 3 hours. This patient had complete reperfusion (not shown) and complete reversal of the DWI lesion on the early follow-up scan, which was obtained 1.8 hours after the end of the procedure. The late follow-up scan was performed on day 3 and demonstrates that the DWI lesion reappeared and a faint lesion of similar size and location on the fluid attenuated inversion recovery (FLAIR) sequence was now present. Right, Example of a patient with sustained DWI reversal. The symptom onset to MRI time was 5 hours and the endovascular procedure began at 6.5 hours. The DWI lesion was larger on the baseline scan than it was on the early follow-up scan (obtained 1.8 hours after the procedure). The infarct volume on the day 5 FLAIR was also smaller than the initial DWI lesion. Areas of sustained reversal are highlighted with arrows.
A total of 32% (19/60) of the patients had >10 mL of early DWI reversal, representing a median (IQR) of 38.5% (24.1%–57.0%) of the DWI 1 lesion (median [IQR] volume of reversal, 17.5 [13.5–34.4] mL). Of these 19 patients with transient reversal, 10 patients had persistent reversal >10 mL and the median (IQR) volume of permanent reversal was 18.3 mL (13.7–26.0 mL; Figure, left).
Reperfusion status was determined by perfusion-weighted imaging in 53 patients and angiography in 7. Of 60 patients, 41 patients experienced early reperfusion. Infarct growth was significantly less in patients with reperfusion (Table 2; P=0.002). There was no significant association between reperfusion and the volume of early or permanent reversal (Table 2).
For the binary logistic regression analysis, National Institutes of Health Stroke Scale and onset time to treatment were not significant independent predictors; only age and DWI volume at baseline were retained as adjusting factors. Early DWI reversal was not an independent predictor of favorable clinical response (P=0.569) or good functional outcome (P=0.311).
At the time of the early follow-up scan, the mean ADC values increased significantly by 111±30×10−6 mm2/s in regions of transient DWI reversal (P=0.003). Regions of permanent reversal had nonsignificant elevations in ADC (58±106×10−6 mm2/s), P=0.28. In regions with no DWI reversal, the ADC values decreased between the baseline and early follow-up scan by −39±56×10-6 mm2/s, but this difference was not significant (P=0.17).
Discussion
The key finding of our study is that early DWI reversibility is not uncommon, but it is typically transient. In addition, permanent DWI reversal usually involves only a small volume of tissue; the median permanent reversal volume of 3 mL noted in this study is similar to that identified in other recent studies.13,14 Furthermore, even among our patients who achieved reperfusion after the endovascular procedure, infarct growth into adjacent tissue was typically much greater than the volume of permanent reversal. Therefore, the final infarct size was almost invariably at least as large as the baseline DWI volume. Furthermore, we were not able to document an association between transient DWI reversal and favorable clinical outcomes.
Changes in DWI lesion evolution in this study were analyzed by coregistration of the baseline DWI and early follow-up DWI into day 5 FLAIR space. The rationale for the coregistration approach is that growth in one region and reversal in another region may result in no net volume change, so a purely volumetric approach may mask reversal. However, it should be noted that coregistration-based measurement of reversal and growth are sensitive to small inaccuracies in the registration because of minor differences between lesion outlines that inevitably occur, even for static lesions. For example, a small displacement of a large lesion that is truly static over time will give rise to an apparent growth and reversal volume of similar magnitude. Therefore, in general, the coregistration approach tends to overestimate DWI reversal, whereas a purely volumetric approach tends to underestimate reversal.
Although it is clear that early DWI lesions are not equivalent to the ischemic core,20,21 clinical and laboratory22 studies have demonstrated that despite early reperfusion, only limited tissue salvage occurs in coregistered regions that harbored acute DWI lesions. Early DWI reversal may result from the transient rise in ADC values that occurs after reperfusion. This transient rise seems to be related to vasogenic edema rather than tissue salvage.23 Therefore, follow-up imaging obtained within several hours after reperfusion may erroneously suggest tissue salvage because DWI changes can disappear as a result of the abrupt rise in ADC values. These previous clinical observations are supported by our results. Similar changes in ADC values have also been reported in animal models10,22 and subsequent follow-up imaging or pathological examination frequently confirms infarction in these regions. In addition, DWI reversal is not an all or none phenomenon; regions of DWI reversal often subsequently develop subtle abnormalities on delayed T2 or FLAIR imaging. Data from animal models suggest that these subtle lesions may involve selective neuronal necrosis rather than infarction.24
The reported incidence of DWI reversal varies from 6.7%14 to 50%.15 Reasons for this wide variation include the fact that determination of the final infarct volume is challenging, and methodologies vary considerably between studies. One variable relates to the window and level settings used when viewing FLAIR or T2 lesions. Changes in window settings can have a meaningful effect on measured lesion volumes. The timing of the follow-up scan may also have an effect on final infarct volume. For example, follow-up scans performed within a week after symptom onset may lead to underestimation of DWI reversal because the actual infarct volume is overestimated because of vasogenic edema. In contrast, obtaining final infarct volumes from late scans (30–90 days) may overestimate reversal because of infarct atrophy. Early edema or late atrophy can also cause tissue shifts (such as ventricular displacement or enlargement) that make accurate coregistration challenging.
This study has several limitations. The small sample size limits our ability to determine the incidence of DWI reversal precisely, as well as associations between reversal and reperfusion or clinical outcomes. Previous studies have documented an association between reperfusion and DWI reversal; our study noted only a nonsignificant trend, possibly because of inadequate power. Because the DEFUSE 2 study was not a randomized study, we are unable to determine whether the incidence of DWI reversal is different in patients treated with endovascular therapy when compared with an untreated control group. In addition, the median baseline DWI lesion volume was relatively small in DEFUSE 2; therefore, we are unable to address the rates of DWI reversal accurately in patients with large baseline DWI lesions. The median (IQR) time between symptom onset and endovascular therapy was 5.9 (4.4–7.9) hours in this study. Consequently, our results do not apply to patients treated at earlier time points, especially in light of the fact that animal studies have found DWI reversal to be more likely to occur in association with early reperfusion.10,25 We also do not have data about the exact time reperfusion/recanalization occurred (DEFUSE 2 used the time of completion of the endovascular procedure as a surrogate for the time of endovascular reperfusion).
Conclusions
Early DWI reversal occurs in ≈30% of patients after endovascular therapy and can involve a substantial percentage of the initial DWI volume; however, permanent DWI reversal is uncommon and typically involves only a small volume of tissue. Early DWI lesions seem to be a promising indicator for tissue that is likely to be unsalvageable in patients with acute stroke.
Sources of Funding
The Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution (DEFUSE) 2 study was funded by grants from the National Institute for Neurological Disorders and Stroke (R01 NS03932505 to Dr Albers and K23 NS051372 to Dr Lansberg).
Disclosures
Dr Albers has served as a consultant for Covidien, Codman, Lundbeck and Genentech and was the Chair of the DSMB for TREVO 2 (Concentric). Drs Bammer and Albers have equity interest in iSchemaView. The other authors report no conflicts.
Footnotes
Guest Editor for this article was Steven C. Cramer, MD, MMSc.
- Received May 13, 2013.
- Accepted January 16, 2014.
- © 2014 American Heart Association, Inc.
References
Jump to
This Issue
Article Tools
Share this Article
- Early Diffusion-Weighted Imaging Reversal After Endovascular Reperfusion Is Typically Transient in Patients Imaged 3 to 6 Hours After Onset
