Evaluation of early reperfusion and IV tPA therapy using diffusion- and perfusion-weighted MRI

Methods.

Twelve patients were evaluated prospectively in the hyperacute period of acute ischemic stroke in a protocol that was approved by the human subjects committee. These were patients admitted to the Stanford Stroke Service between November 1996 and March 1998. All patients in this study presented within 6 hours of symptom onset. Six of these patients were seen within 3 hours of stroke onset and received thrombolytic treatment with rt-PA, which was administered in compliance with National Institute of Neurological Disorders and Stroke guidelines. rt-PA was always administered before MRI was performed and after head CT was performed to rule out intracranial hemorrhage. Six patients were ineligible for rt-PA treatment either because of contraindications to tPA therapy or because they were seen in a 3 to 6-hour time window, and these patients were frequently enrolled in placebo-controlled trials of potential neuroprotective agents. The six patients receiving rt-PA were imaged 3 to 5 hours postictus onset (mean, 4 hours 20 minutes), whereas the six patients not treated with rt-PA were imaged 4 to 7 hours postictus (mean, 5 hours 25 minutes).

MRI was performed using echo planar imaging (EPI) on a 1.5-T General Electric Signa magnet (Milwaukee, WI). Multislice whole-brain DWI was performed using the following parameters: 16 slices; repetition time (TR), 8,100 msec; echo time (TE), 110 msec; inversion recovery time, 2,342 msec; slice thickness, 5 mm; gap, 2.5 mm; matrix, 128 × 128; and field of view, 24 cm. B values were 0 and 741 sec/mm2. DW images were acquired in the x, y, and z directions with and without an inversion pulse (fluid-attenuated inversion recovery [FLAIR] sequence). The x-, y-, and z-direction DW scans were averaged to minimize hyperintensities due to anisotropic water diffusion. EPI diffusion images were processed to generate average (trace) apparent diffusion coefficient (ADC) maps. PWI was performed using dynamic susceptibility contrast-enhanced MRI. Gradient echo, single-shot EPI was used during injection of 20 mL gadolinium (0.2 mmol/kg). PWI acquisition values were TR, 2,000 msec; TE, 60 msec; with 35 time points obtained over 12 slices. Other parameters were the same as for DWI. PW images were processed to generate maps of time-to-bolus peak (TTP), relative mean transit time, and relative cerebral blood volume. Regions of interest (ROIs) were first determined by a neuroradiologist (M.M.) and then hand traced using special computer software (MRVision Company, Menlo Park, CA). All measurements were performed by a single observer (C.B.) to minimize interobserver error. Areas of abnormality were defined as hyperintensity on DWI, or hyperintensity (i.e., a delay of contrast bolus) on TTP maps. The ROIs from the DW images were transferred to the ADC maps obtained with FLAIR for measurement of the ADC within the lesion. The FLAIR ADC maps were used to minimize hyperintense areas on ADC due to CSF.9,10 In addition, the ROIs from the DW images were also transferred to the T2-weighted FLAIR images for assessment of serial signal intensity changes.

DW and PW images were obtained at five time points. The initial scan was performed as soon as possible after presentation. Serial repeat scans were then performed at +3 to 6 hours (T2), +24 to 36 hours (T3), and +5 to 7 days (T4) after the initial scan. A final scan was performed as a chronic time point at approximately 30 days (T5).

A ≥10% difference between PWI and DWI volumes was considered to indicate a PWI/DWI mismatch. ADC maps were assessed qualitatively for regions of hypointensity (low ADC) and/or hyperintensity (high ADC) compared with surrounding and contralateral normal brain. In addition, ADC maps were assessed quantitatively in two ways. First, average ADC values for the entire region of DWI abnormality were compared with contralateral control values to yield relative ADC values for the entire lesion. In addition, areas identified visually as being hyperintense on ADC maps were measured and compared with contralateral control areas to assess the maximum degree of ADC change in a specific ROI.

Neurologic assessments were performed using the NIH Stroke Scale (NIHSS). All patients were examined by a neurologist certified in performing the NIHSS.

The DWI and PWI changes seen in the patients receiving rt-PA versus non-rt-PA–treated patients were analyzed and compared. Patients were separated into two groups based on the time to resolution of perfusion abnormalities. Early reperfusers were defined as patients experiencing a complete or near-complete resolution of PWI abnormalities by T3. In some cases this was seen by T2. Patients without early reperfusion who had persistent PWI changes at T3 constituted the other group. ADC changes and T2-weighted FLAIR changes were assessed for the two groups.

Treatment with other investigational neuroprotective agents was permitted in all patients. In the non-rt-PA–treated group, five of six patients participated in double-blind, placebo-controlled neuroprotective studies (nalmefene, n = 2; lubeluzole, n = 2; aptiganel chloride, n = 1). In the rt-PA–treated group two of six patients were participants in neuroprotective studies (nalmefene). Individual treatment assignments were unknown at the time of data analysis.

Statistical analysis was performed using standard computerized software (SigmaStat 2.0, Jandel Scientific, San Rafael, CA). For comparisons of neurologic outcomes, Student’s t-test was used. Fisher’s exact test and the Mann–Whitney U test were used to compare proportional outcomes between different groups as appropriate.

Results.

The mean age of the 12 patients studied was 70 ± 10 years (range, 49 to 84 years). Data on time to treatment and initial NIHSS scores on all patients are shown in the table. The initial (T1) DW and PW images were performed 3 to 7 hours after symptom onset (mean, 5 ± 1 hours). Subsequent MR images were performed at a mean of 9 ± 1 hours (T2), 34 ± 9 hours (T3), 6 ± 1 days (T4), and 40 ± 23 days (T5) after symptom onset. In the rt-PA–treated patients, thrombolytic infusion was initiated a mean of 100 ± 40 minutes after symptom onset (range, 51 to 165 minutes). There was no significant difference in time to scan between the rt-PA- (mean, 4.3 ± 0.8 hours) and non-rt-PA- (mean, 5.3 ± 1.1 hours) treated groups.

Early reperfusers versus nonearly reperfusers.

PWI was performed successfully at 24 to 36 hours (T3) in 11 of 12 patients, and PWI normalization by this time (i.e., early reperfusion) was observed in 6 of 11 patients. The overall ADC value for the entire abnormal ROI as seen on the DW image in the early reperfusion group was significantly greater than in the nonreperfusion group by T2 (unpaired t-test, p = 0.04). The difference remained significant at T3 (p = 0.002) and T4 (p = 0.0005). A relative difference in ADC values between the two groups increased from T2 to T4, converging only by T5 (figure 1). Moreover, frankly elevated ADC regions were seen within the abnormal area on ADC maps (ipsilateral ADC/contralateral ADC, 1.46 ± 0.19) by T3 in five of six early reperfusers (83%) versus zero of five nonearly reperfusers (0%; p = 0.015, Fisher’s exact test). In addition, signal intensities of the ischemic region on heavily T2-weighted FLAIR images increased more rapidly in the early reperfusion patients (figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Comparison of the temporal evolution of the mean relative apparent diffusion coefficient (ADC) values (rADC = ADC in lesion/ADC in contralateral region of interest) in early reperfusers (n = 6) versus nonearly reperfusers (n = 5). Early reperfusion is defined as regaining normal perfusion by 24 to 36 hours. The rADC values are reduced in both groups immediately after stroke onset (approximately 5 hours postictus), but become significantly different by the second time point (approximately 10 hours). rADC is higher in the early reperfusers throughout the first week and then it converges to elevated values at the chronic time point that are similar to the nonearly reperfusers. The rADC values given are mean ± SD throughout the entire lesion identified on diffusion-weighted imaging. *p < 0.05. rADC = relative apparent diffusion coefficient.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Comparison of the temporal evolution of the mean relative fluid-attenuated inversion recovery (FLAIR) signal intensity (intensity in lesion/intensity in contralateral region of interest) in early reperfusers (n = 6) versus nonearly reperfusers (n = 5). The relative FLAIR intensities are near normal (i.e., value close to one) in both groups immediately after stroke onset (approximately 5 hours postictus), but there is a much more rapid increase in signal intensity in the group of early reperfusers. The later time points demonstrate similar hyperintensities on FLAIR for both groups. The relative signal intensity values given are mean ± SE throughout the entire lesion identified on diffusion-weighted imaging. *p < 0.05. FLAIR = fluid-attenuated inversion recovery.

rt-PA versus non-rt-PA–treated patients.

The rt-PA–treated patients did not differ significantly from the non-rt-PA–treated patients in initial NIHSS score, age, or time to initial MRI. The average NIHSS score for all subjects was 11 ± 9 (see the table).

PWI was performed successfully at 3 to 7 hours (T1) in 11 of 12 patients. At this initial time point, the volume of the PWI abnormality was less than the volume of the DWI abnormality (PWI < DWI) in five of six rt-PA–treated patients (83%). In contrast, PWI < DWI was present in only one of five of the non-rt-PA–treated patients (20%; p = 0.08, Fisher’s exact test). In addition, early reperfusion (by T2 or T3) was seen in five of six rt-PA patients (83%) compared with one of five non-rt-PA patients (20%; p = 0.08, Fisher’s exact test). A portion of the ADC map in the region of the acute ischemic lesion was found to rise above the normal contralateral values by T3 in four of six rt-PA–treated patients and two of six non-rt-PA–treated patients. In all patients these regions of early elevated ADC ultimately became incorporated into the region of final infarction on the 30-day T2-weighted MR images. A representative example is shown in figure 3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. The heavily T2-weighted, CSF-suppressed fluid-attenuated inversion recovery (FLAIR) images (A), diffusion-weighted images (DWI; B), apparent diffusion coefficient (ADC) maps obtained with CSF suppression (C), and perfusion time-to-peak maps (D) of four slices in a patient treated with recombinant tissue plasminogen activator are presented for 5 time points. (A, B) Hyperintensity on the DWI in the right middle cerebral artery (MCA) territory is seen as early as 3 hours after symptom onset, whereas the lesion is not evident on FLAIR at 3 hours. (C) As expected, the ADC is low throughout the lesion at the 3-hour time point. However, early normalization of the ADC is seen at 8 hours with subsequent regions of abnormally high ADC at 37 hours. The ADC is elevated throughout the lesion at 21 days. (D) The initial perfusion abnormality, denoted by the hyperintense regions on the time-to-peak maps, was rather extensive throughout the right MCA territory at 3 hours, and had almost completely corrected at 8 hours. The early ADC increases are thought to be a result of this early reperfusion. There are spurious regions of hyperintensity in both hemispheres at 37 hours due to difficulty fitting the bolus dynamics because there was a poor bolus and some patient motion at this time point.

Clinical outcomes.

Clinical outcomes are summarized in the table. The mean improvement for all 12 patients in NIHSS score by T5 was 5 ± 4 points. Although the rt-PA–treated patients had a greater average improvement in NIHSS score compared with the non-rt-PA–treated group (8 ± 8 versus 3 ± 4), this did not reach statistical significance (p = 0.28, Student’s t-test).

The relationship between DWI, PWI, and clinical outcome was also examined. Overall, PWI > DWI was observed in 5 of 11 patients at T1. However, there was no significant difference in outcome between patients with PWI > DWI versus patients with PWI < DWI. Of interest, PWI > DWI was seen in only one patient in the rt-PA–treated group at T1. This patient had a large PWI abnormality with only a small amount of DWI change. In this patient, there was a marked improvement in neurologic symptoms 3 hours after treatment. By 1 week, her deficit had resolved almost completely. Serial MR images revealed rapid resolution of her PWI changes and only a very small DWI lesion that slightly enlarged over time (figure 4). This patient had the greatest improvement in NIHSS score (22 points) of any patient studied (see the table, Patient 5).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Diffusion-weighted images and perfusion time-to-peak maps for three slices in a patient treated with recombinant tissue plasminogen activator who demonstrated a marked neurologic improvement (NIH Stroke Scale score improvement of 18 points by 24 hours). The time-to-peak maps at 4.5 hours reveal a large region of perfusion abnormality (arrows) but very little, if any, diffusion abnormality. The perfusion abnormality persists at 10 hours and then is resolved permanently by 32 hours. Therefore, serial scans reveal an early resolution of the perfusion-weighted imaging (PWI) abnormality and only a slight increase in the ischemic region by diffusion-weighted imaging (DWI), the final infarct volume being much smaller than the original area of PWI abnormality. The patient made a full recovery. Note that the distortions in the time-to-peak maps are due to susceptibility-induced signal dropout in the original gradient-echo images.

Discussion.

This study is the first to evaluate systematically patients treated with thrombolytic therapy in acute ischemic stroke with serial DW and PW images. The most striking finding was an unexpected, rapid rise in the aggregate ADC in patients with early reperfusion. It should be pointed out that there still was marked heterogeneity within the ischemic region. Some areas became hyperintense very early (T2 or T3) whereas other portions remained with low ADC values. In previous studies,11-14 it has been shown that aggregate ADC values typically remain low for 4 to 10 days after acute ischemia. This low ADC has been hypothesized to be due to the more hindered movement of water molecules in ischemic tissue compared with normal tissue.15 A likely explanation for this phenomenon of early ADC rise is that regions of vasogenic edema develop earlier in those patients who reperfuse more rapidly. Our finding of a steep, early rise in the signal intensity of T2-weighted FLAIR images in the early reperfusers compared with the nonearly reperfusers supports the idea that this change is due to water influx from vasogenic edema. Another possible explanation may be that early reperfusion promotes more rapid cell lysis, which also alters the existing environment in which the water molecules exist. Both vasogenic edema and cell lysis have been shown histopathologically to cause elevated ADC values in animal models of infarction.16 It is important to recognize that in this study even those areas showing early increased ADC went on to infarction on T2-weighted MRI at chronic time points in all observed patients, suggesting that the early reperfusion we are observing largely does not preserve these regions from subsequent necrosis.

The study also shows that rapid imaging with PWI is sensitive to early change in cerebral perfusion caused by rt-PA treatment. The observation of PWI volumes being less than DWI in 83% (five of six) of thrombolytic-treated patients compared with 20% (one of five) of untreated patients at 3 to 7 hours postictus (T1) suggests successful, rapid thrombolysis-induced reperfusion. A recent report17 of 17 patients imaged within the first 24 hours of stroke (mean, 12.2 hours postictus) has also shown that PWI and DWI demonstrate differences between the volumes of tissue involved when comparing the DWI to PWI volumes. Only six of the patients in that study were imaged in the first 6 hours following stroke onset. However, these authors found that none of the six patients studied in the first 6 hours showed PWI < DWI. This underscores our finding that five of six patients imaged at T1 in the rt-PA treatment group had PWI < DWI, again suggesting that a thrombolytic affect is being observed. In addition, our study demonstrated earlier normalization (i.e., before T3) of the PW image at 24 to 36 hours (T3), which was seen in 83% (five of six) of rt-PA–treated subjects, compared with 20% (one of five) of the non-rt-PA–treated patients, suggesting a more complete reperfusion with rt-PA therapy.

These results also confirm that major differences in the volume of DWI and PWI lesions can be present in acute stroke patients. These mismatches of diffusion and perfusion may have a substantial impact on patient management. The greatest neurologic improvement in this patient group occurred in the patient with a substantial PWI > DWI lesion mismatch treated with rt-PA. This would be consistent with the hypothesis that a large PWI lesion identifies tissue at risk for infarction, whereas the early DWI lesion identifies tissue already severely injured and unlikely to respond to thrombolytic treatment. Such PWI > DWI patients may be the best candidates for thrombolytic therapy. As important, the finding of early PWI normalization in non-rt-PA–treated patients (seen in one patient in this series) is consistent with the concept of spontaneous reperfusion—a phenomenon that has also been associated with good outcome. These results are suggestive of the potential power of DWI and PWI in guiding thrombolytic treatment. It should be pointed out, however, that pretreatment scans were not obtained in this series and the potential utility of this technique remains to be confirmed.

Asymptomatic hemorrhagic conversion was observed in 66% of our study patients, consistent with previous observational studies.18,19 We observed that hemorrhage was always associated with areas of ischemia on DWI and abnormal perfusion by PWI. Of importance, none of these areas of hemorrhage were symptomatic. Therefore the clinical relevance of this bleeding, if any, is uncertain. DWI and PWI may allow further detailed study of the risks associated with hemorrhage, and these imaging sequences may help to monitor its progression.

We did not perform DWI and PWI before the initiation of thrombolysis. Therefore, we do not know the state of the brain before the initiation of treatment. Future studies will require that DWI and PWI be performed both before and after treatment to document accurately the effects of therapy on MRI-detected changes. We cannot discount the possibility that some of the effects seen were related to neuroprotective treatment. However, to date, no study of neuroprotective therapy has shown an unequivocal benefit to treatment or definitive effect on DWI or PWI. In addition, the major changes observed in this study were perfusion related, and neuroprotective agents are not necessarily expected to affect perfusion dynamics.