Introduction

Even though computerized tomography (CT) is still considered to be the standard brain imaging modality in acute ischemic stroke [1], recent studies have suggested that magnetic resonance imaging (MRI) may help to identify patients with the most potential to benefit from thrombolytic therapy [2–8]. MRI using diffusion-weighted images (DWI) improves the ability to diagnose acute cerebral infarction (CI) in terms of both quantification of actual CI volume and accurate CI localization [9–18]. MRI can also detect significant cerebral arterial occlusion or severe stenosis [19, 20]. MRI allows the evaluation of cerebral collateral flow and also imaging of reversible ischemic changes [21–24]. Nevertheless, to date, few studies have evaluated the impact of these pretreatment MRI parameters on clinical outcome [6–8, 15, 25–27].

The aim of our study was to assess the influence of initial CI volume measured on MRI-DWI sequences on clinical outcome in acute stroke patients with middle cerebral artery (MCA) (M1–2 segment) occlusion detected on MR angiography, who subsequently underwent intravenous/intraarterial thrombolysis (IVT/IAT). Initial CI volume (VDWI-I) was retrospectively compared with neurological deficit evaluated using the NIH stroke scale (NIHSS) on admission and 24 h later, and with the 90-day clinical outcome using the modified Rankin scale (mRS) in a series of 25 consecutive patients.

The MRI protocol used was based on actual data in the literature and the published results of several recent studies [1, 3–5, 7, 14, 15, 28–33].

Subjects and methods

A group of 25 consecutive acute ischemic stroke patients with MCA occlusion (M1-2 segment) detected on MR angiography, who underwent IVT or IAT according to recent guidelines [34, 35] between January and September 2005, were retrospectively analyzed. The demographic and clinical characteristics of the patients are shown in Table 1.

Table 1 Patient demographic and clinical characteristics
Full size table

On admission, blood pressure was measured, electrocardiogram was recorded, and blood samples were taken. Clinical status was evaluated using the NIHSS by a certified neurologist. An MRI examination followed immediately. After admission, patients were treated according to published guidelines [34, 35].

MRI was performed on a Magnetom Symphony 1.5-T Maestro Class (Siemens, Erlangen, Germany) with quantum gradients (syngo2004A) and a standard head coil (CP head array coil).

The MRI protocol included the following sequences: (1) localizer, (2) T2-turbo spin echo (TSE), (3) fluid-attenuated inversion recovery (FLAIR), (4) DWI, (5) 3D time of flight magnetic resonance angiography (TOF MRA). The total acquisition time (AT) was 11 min 28 s. Sequences 2–4 were applied to acquire data from the same set of slices (standard number of slices 19, slice thickness 5 mm, distance factor 30%). The standard slice orientation was oblique axial, approximately parallel to skull base in order to minimize susceptibility artifacts in echo-planar imaging (EPI) sequences.

The sequence parameters were as follows: T2-TSE TR/TE/ETL 4,000/99/9 ms, FOV 230×173 mm, matrix 256×256, AT 1 min 34 s; FLAIR 8,050/112/ETL 21/2 concatenation, FOV 230 mm, FOV phase 76.6%, matrix 256×151, AT 2 min 26 s. These sequences were used to assess hemorrhage and detect local demyelination changes including sites of ischemic demyelination.

The EPI-DWI sequence parameters were as follows: 3,200/94/EPI factor 128/3 averages, FOV 230×230 mm, matrix 128×128 with interpolation, TA 1 min 20 s. MRI data were acquired with three diffusion weightings: b=0, DWI b=500, and DWI b=1,000. The fourth type of image was an automatically created apparent diffusion coefficient (ADC) map. DWI traces show average local diffusivity in the brain tissue examined when b is 500 and 1,000. This sequence was applied to assess hemorrhage (b=0: T2*-EPI, susceptibility-sensitive sequence) and detect sites of reduced diffusion (DWI, b=500 and 1,000). The 3D-TOF MRA sequence parameters were as follows: 43/7.15, 3 slabs, 32 partitions/slab, slice thickness 1 mm, FOV 200×150 mm, matrix 512×192, AT 5:59 min. The images obtained—maximum intensity projection (MIP) and sublayers—would illustrate closure of the main arterial trunk of the circle of Willis or its branches.

Infarct volumes were measured on DWI trace images (b=1,000) and calculated as total hyperintense area in single slices multiplied by effective slice thickness [(actual slice thickness + distance factor)/interslice gap]. Follow-up MRI was performed in all patients after 24 h. Clinical outcome was evaluated 3 months after CI using the mRS. The relationship between the initial infarct volume and neurological deficit severity was assessed, defining a cut-off point for maximum VDWI-I still associated with a good clinical outcome. Several cut-off points of VDWI-I were subsequently tested with the aim of maximizing both sensitivity and specificity for a good clinical outcome (mRS 0–2).

Pearson’s correlation analysis, an independent sample test, and the non-parametric Mann-Whitney test were used to assess the significance of differences.

Results

The analysis included 25 patients with acute stroke. The mean delay between symptom onset and MRI was 149 min. The demographic and clinical baseline data are shown in Table 1. VDWI-I ranged from 1.3 ml to 321 ml, quantified from hyperintense lesions on DWI sequences.

The relationship between VDWI-I and neurological deficit on admission is demonstrated in Fig. 1, and clinical evolution after 24 h and 90 days is shown in Figs. 2 and 3, respectively. The correlation analysis demonstrated a strong positive correlation between VDWI-I and 24-h clinical outcome (r=0.74) and a moderate positive correlation between VDWI-I and 90-day clinical outcome (r=0.67; Figs. 4 and 5).

Fig. 1
figure 1

Relationship between initial infarct volume and neurological deficit on admission

Full size image
Fig. 2
figure 2

Relationship between initial infarct volume and 24-h clinical outcome

Full size image
Fig. 3
figure 3

Relationship between initial infarct volume and 90-day clinical outcome

Full size image
Fig. 4
figure 4

Pearson’s correlation analysis of the correlation between initial infarct volume (VDWI-I) and 24-h clinical outcome

Full size image
Fig. 5
figure 5

Pearson’s correlation analysis of the correlation between initial infarct volume (VDWI-I) and 90-day clinical outcome

Full size image

The cut-off point of 70 ml was found to be the maximum VDWI-I still associated with a good clinical outcome and corresponding to the maximum sensitivity and specificity (80%) achieved. Figure 6 demonstrates this resulting cut-off point, on the basis of which the patients were divided into two groups (VDWI-I ≤70 ml and >70 ml; Table 2).

Fig. 6
figure 6

Sensitivity and specificity for good clinical outcome (mRS 0–2) in relation to initial infarct volume

Full size image
Table 2 Demographic and clinical parameters of patient subgroups
Full size table

At 24 h (Fig. 7), patients with VDWI-I >70 ml had a significantly higher NIHSS score (P=0.0018).

Fig. 7
figure 7

Correlation between initial infarct volume and 24-h clinical outcome

Full size image

Intracranial hemorrhage was present at 24 h in one patient with VDWI-I ≤70 ml and in three patients with VDWI-I >70 ml (Table 2). There was a statistically significant difference in mRS score between patients with VDWI-I ≤70 ml and those with VDWI-I >70 ml (P=0.003, Fisher’s exact test). The median mRS scores in the two groups are shown in Table 2. The independent samples test showed significantly higher mRS (3–5) in those with VDWI-I >70 ml than in those with VDWI-I ≤70 ml (P=0.026; Fig. 8). The significance of this difference was confirmed by the non-parametric Mann-Whitney test (P=0.003).

Fig. 8
figure 8

Correlation between initial infarct volume (VDWI-I) and 90-day clinical outcome

Full size image

Mortality at 90 days was 71.5% in those with VDWI-I >70 ml and 0% in those with VDWI-I ≤70 ml. All these patients died during the first 7 days after thrombolysis; brain edema was the cause of death in all cases. Intracranial hemorrhage also occurred in three of these patients.

Discussion

No direct relationship was found between the initial infarct volume measured on DWI and the neurological deficit on admission (Fig. 1). Two patients with the same severity of deficit and the same type of arterial occlusion (MCA, M1–2 segment) could have VDWI-I differing by as much as tens of milliliters (Fig. 9). This difference was mainly caused by the actual individual state of cerebral collateral flow.

Fig. 9
figure 9

Different initial infarct volumes in two patients with MCA occlusion on admission. a Right MCA occlusion (3D TOF MRA); b left MCA occlusion (3D TOF MRA); c right MCA infarct on DWI (volume 136 ml); d left MCA infarct on DWI (volume 11.3 ml)

Full size image

The actual infarct volume is considered to be an important factor in the decision to perform thrombolysis. Initial infarct volume is considered to be an independent predictor of subsequent spontaneous CI hemorrhagic transformation and also for intracerebral hemorrhage growth after thrombolysis [36]. Based on the results of this study, one may also predict eventual CI progression and the resulting clinical outcome from the initial infarct volume.

Davalos et al. retrospectively found infarct volume progression in patients with VDWI-I ≤25 ml and initial NIHSS score >8, on average about 68 ml in a group of 166 acute CI patients, examined within 12 h of stroke onset. This progression occurred during the first 6 h in one of three of these patients—mainly in those who did not receive thrombolysis [8].

In the group of patients with VDWI-I ≤70 ml and initial NIHSS >8, infarct volume progression was found in nine patients with a mean volume of progression of 27.5 ml.

Derex et al. observed a significantly worse clinical outcome in association with a larger pretreatment DWI infarct volume in 49 stroke patients treated by IVT. They consider the initial DWI lesion volume and the recanalization to be independent factors in determining the final infarct size [27].

Oppenheim et al. consider quantification of the initial CI volume to be an accurate method for assessing the risk of malignant infarct progression in patients with MCA occlusion. They identified a threshold volume of 145 ml in their group of 28 patients. No patient with a CI volume below this value had a malignant MCA infarction in spite of a high NIHSS score on admission (16.5±4) [15]. In our study, 71.5% of the patients with MCA occlusion and with VDWI-I >70 ml died within 7 days from CI onset (Fig. 4).

Although perfusion-weighted imaging (PWI) is performed as a standard part of the MRI protocol in many stroke centers and has been used in previously reported studies, we did not use it in our MRI stroke protocol and therefore also did not subsequently interpret the ischemic penumbra according to the PWI/DWI mismatch for several reasons. We believe that PWI/DWI mismatch is an inaccurate approximation of the real ischemic penumbra [37–39]. Furthermore, quantification of PWI lesion volume generally uses only qualitative indices and quantitative calculations are very time-consuming and use controversial mathematical models [40]. Definitive infarct volume is usually far smaller than the initial PWI area [37–39]. For example, Heiss et al. compared the size of the PWI/DWI mismatch area with the area of reduced cerebral blood flow and increased oxygen extraction (over 150%) using positron emission tomography (PET) in acute stroke patients. Despite a good correlation between PWI and PET in the detection of perfusion defect, mismatch volume did not correspond to the volume of the area with increased oxygen extraction on PET scans. Thus PWI/DWI mismatch may not be a reliable correlate of ischemic penumbra [41].

Several methodological limitations of the presented study should be mentioned. The first one is the relatively small number of patients examined. This was a result of the generally low number of patients fulfilling the strict time criteria for thrombolytic therapy, event with MCA occlusion. Secondly, quantification of infarct volume was performed manually, because no semiautomatic quantification software was available. Therefore the quantification could have been affected by subjective operator error. Finally, interpretation of the results may have been limited by nonhomogeneity of the group: the patients had received two types of recanalization therapy, but both with the same effect on the occluded artery.

Conclusion

Quantification of initial CI volume can be considered important for an accurate and safe evaluation of the indications for thrombolytic therapy in acute stroke patients.

The MRI-DWI quantification of initial CI volume was able to predict clinical outcome in patients with acute CI with MCA occlusion treated by IVT or IAT.