J Neurol (2007) 254:296–305 DOI 10.1007/s00415-006-0355-0
Franca Tecchio Patrizio Pasqualetti Filippo Zappasodi Mario Tombini Domenico Lupoi Fabrizio Vernieri Paolo Maria Rossini
Received: 30 January 2006 Received in revised form: 6 July 2006 Accepted: 25 July 2006 Published online: 7 March 2007
F. Tecchio Æ F. Zappasodi Istituto di Scienze e Tecnologie della Cognizione (ISTC), CNR Rome, Italy P. Pasqualetti Æ D. Lupoi Æ F. Vernieri P.M. Rossini (&) Dipartimenti di Neuroscienze e di Diagnostica per immagini Ospedale ‘‘Fatebenefratelli’’ AFaR Isola Tiberina (Rome), Italy Tel.: + 39 06 6837382 Fax: + 39 06 6837360 E-Mail:
[email protected],
[email protected] P. Pasqualetti Æ P.M. Rossini IRCCS ‘‘Centro S. Giovanni di Dio-Fatebenefratelli’’ Brescia, M. Tombini Æ F. Vernieri Æ P.M. Rossini Clinica Neurologica Universita` Campus Biomedico Rome, Italy
ORIGINAL COMMUNICATION
Outcome prediction in acute monohemispheric stroke via magnetoencephalography
j Abstract Background Follow-
ing an ischemic stroke a highly variable clinical outcome is commonly evident despite similar onset symptoms as well as lesion characteristics. The aim of this study was to identify indexes providing early prediction of functional recovery, in addition to clinical severity and lesion dimension at onset of stroke. Methods In 32 patients, magnetoencephalographic (MEG) parameters collected in the acute phase (< 10 days from symptoms onset, T0) from affected (AH) and unaffected (UH) hemispheres at rest and evoked by sensory stimuli were evaluated in association with the clinical outcome in a stabilized phase (T1, median 7.8 months) classified with three levels: worsening, partial and full recovery. Results Multiple multinomial logistic regression indicated AH gamma and UH delta band powers able to prognosticate clinical out-
Introduction It is common experience in following up stroke patients who survive the acute stage, to observe a widespread range of long-term outcome despite a nearly identical early clinical picture. Among prognostic indicators, the most considered ones include lesion site and volume as evaluated via Magnetic
come at T1. After inclusion in this analysis, lesion volume had the strongest predictive ability, and UH delta band power remained as a predictive factor with a measurable cut-off, maximizing both sensitivity and specificity of the prediction: a patient with UH delta below cut-off would recover to some extent; a patient with UH delta above cut-off would have a probability of about 70% to worsen. Conclusions MEG UH delta and AH gamma band powers were found to provide useful information about long-term outcome prognosis. Only the increase of delta band activity in the unaffected hemisphere contains information about the outcome in addition to the lesion volume. j Key words acute stroke Æ middle cerebral artery (MCA) Æ MEG Æ clinical outcome Æ prognostic indications
Resonance Imaging [4, 21, 34]. In addition to the lesion characteristics, new prognostic indicators are needed to better predict final outcome, which besides helping to better understand pathophysiology of poststroke brain recovery, allow a more precise planning of pharmacological and rehabilitative treatments in individual patients. Recent findings suggest that there is some derangement of basic neuro-vascular engagement in brain
297
areas previously affected by stroke, i.e. some uncoupling of cerebral vasomotor reactivity from neuronal firing normally stirring it up [61]: this underlines the usefulness of considering neuronal activity features ‘‘per se’’ in greater detail, together with vascular and metabolic indexes, when planning therapeutic strategies in stroke. Increasing evidence is accumulating that adequate procedures modulating neuronal activity, such as neuroprotective drugs [56, 13, 48] and rehabilitative procedures should start as early as possible [51, 25, 28]. This supports research efforts in identifying biological and neurophysiological markers in the acute phase mostly associated with the long term clinical outcome, in order to better define treatments. With this goal, previous studies have proposed various neurophysiological parameters as prognostic tools, such as amplitude abnormalities of Somatosensory and Motor Evoked Potentials [20, 23, 28]; spontaneous electric brain activity evaluated via quantitative EEG (qEEG) collected within the first 72 hours of brain infarct [19]; and modifications of the delta index, a measure of the rate of average scalp delta power change between 6 and 15 hours after stroke [27]. However, other studies reported only limited prognostic value of parameters based on EEG signal [54, 74, 12]. Magnetoencephalography (MEG) represents a noninvasive technique able to identify the synchronous firing of neurons from restricted cortical areas, in relation to either spontaneous cerebral activity or in response to external stimuli [22]. At a given time during the cortical processing, distinguished from the spatial coordinates, the strength can be measured which roughly reflects the number of neurons firing synchronously, as well as the orientation [55] of the generator sources (Equivalent Current Dipoles = ECD). The brain response morphology provides indirect information on the underlying neural circuitry connectivity [68, 72]. MEG is particularly suitable for stroke studies, as the presence of morbid tissue near the cerebral generators has minimal effects on the scalp distribution of the magnetic field [32, 53]. MEG provided clinically relevant information in stabilized post-stroke stages on ‘plastic’ reorganization of the brain areas around or connected to the lesion [28, 58, 59]. However, to the best of our knowledge, systematic studies in the acute post-stroke stages [81, 52, 73] that aim to define whether MEG provides prognostic information on long-term outcome, have not been carried out.
Patients and Methods We performed our study in strict accordance with all the external, internal and statistical criteria of quality assessment of predictive models reported in previously documented scientific literature [18].
j Patients Thirty-two male and female patients ages 30 to 86 years, (mean 68 ± 12,15 females, 17 males) were enrolled in the study after a firstever acute ischemic stroke, diagnosed on the basis of clinical history and examination and confirmed by brain magnetic resonance imaging (MRI). Inclusion criteria were: clinical evidence of a motor and/or a sensory deficit of the upper limb and a neuroradiological diagnosis of ischemic brain damage. Exclusion criteria were: neuroradiological evidence of involvement of both hemispheres; brain haemorrhage; a previous stroke; neuroradiological evidence of multifocal hypoxic-ischemic encephalopathy; peripheral neuropathies that would cause sensory testing and nerves stimulation to be ineffective; dementia; severe aphasia and other conditions mitigating compliance. None of the enrolled patients was treated by intravenous or intra-arterial thrombolysis. The best clinical management of medical problems (cardiac and respiratory care, fluid and metabolic maintenance, blood pressure, glucose and body temperature control) was provided for all patients, according to ad hoc guidelines [1]. Anticoagulant therapy, if indicated (i.e. cardioembolism), or antiplatelet therapy was administered as well. The experimental protocol was approved by the Hospital Ethical Committee and all patients signed written forms of consent. j Clinical evaluation (T0, T1) The NIHSS and the Barthel Index (BI) scores were used for neurological assessment of stroke severity and autonomy in daily life activities. Clinical scores and MEG recordings were collected between day 2 and day 10 (T0, mean 5.2 ± 2.6 days) following the stroke, both from the affected (AH) and unaffected hemisphere (UH). All patients completed the MEG examination without problems. The timing of recordings fits quite well the adequate inception time indicated as within one week of onset [18]. Clinical follow-up was repeated in a stabilized condition (T1, median 7.8 months). Four patients were unavailable for clinical evaluation at T1. Clinical outcome was evaluated by the effective recovery (ER), defined as (NIHSS score at T0 - NIHSS score at T1) / (NIHSS score at T0 – NIHSS score in healthy condition(author is my insertion correct – the meaning was obscure). The denominator represents the total possible recovery, and equals NHISS score at T0 since in healthy condition NIHSS score equals 0. Because the distribution of ER was strongly asymmetric (36% of patients obtained the maximum ER = 1) and no transformation would have been able to obtain a Gaussian distribution of residuals in the regression model, the patient’s outcome was classified as worsening (ER < 0), partial recovery (0 < ER < 100%), full recovery (ER = 100%). Three patients died before the second evaluation for stroke-related reasons, and were included in the worsening group. The clinical assessment in T1 was not performed at a fixed time after stroke onset, but at ‘appropriate times’ [18]; all patients were evaluated at least 45 days after stroke onset and 50% of them were examined after 8 months, i.e. in a similar stage in the disease process, when the long-term outcomes are more meaningful. j MEG recordings (T0) Brain magnetic fields were recorded by means of a 28-channel system [67] covering a scalp area of about 180 cm2, located inside a magnetically shielded room (Vacuumschmelze GMBH). Cerebral activity was picked-up (band-pass filtering 0.48–250 Hz, sampling rate 1000 Hz) from the parieto-frontal region of each hemisphere with the sensors array centred –during two successive trials– on C3 (left rolandic) and C4 (right rolandic) of the International 10–20 electroencephalographic system. Previous studies have shown that
298
Table 1 Acute phase - Rest activity CONTROLS
ABSOLUTE POWER (log10 fT)
Delta Theta Alfa Beta1 Beta2 Gamma Total Power Spectral entropy IAF
Left lesion
Right lesion
Left
Right
Left (AH)
Right (UH)
Left (UH)
Right (AH)
Correlation with NIHSST0 p
3.09 (.11) 3.07 (.16) 3.36 (.17) 3.38 (.19) 3.17 (.14) 2.96 (.09) 3.64 (0.15) 5.79 (0.31) 9.9 (1.2)
3.08 (.15) 3.09 (.15) 3.39 (.19) 3.44 (.17) 3.19 (.15) 2.98 (.11) 3.68 (0.15) 5.75 (0.35) 9.7 (1.2)
› 3.17 * (.13) › 3.21 * (.14) 3.41 (.19) 3.36 (.18) 3.09 (.15) 2.93 (.13) 3.67 (0.14) 5.58 (0.43) fl 9.0 * (1.3)
3.13 (.14) 3.15 (.15) 3.38 (.21) 3.35 (.18) fl 3.11 * (.12) 2.93 (.10) 3.63 (0.16) 5.76 (0.30) 9.4 (1.0)
› 3.20 * (.12) ›3.23 * (.15) 3.41 (.16) 3.34 (.11) 3.04 (.08) 2.88 (.11) 3.65 (0.14) 5.53 (0.48) 9.3 (0.8)
› 3.33 ** (.18) › 3.29 * (.18) 3.36 (.23) fl 3.25 * (.19) fl 2.95** (.09) fl 2.82 * (.05) 3.71 (0.15) fl 5.22 * (0.43) fl 8.3 * (1.0)
n.s. n.s. n.s. n.s. n.s. 0.001 n.s. 0.028 0.031
Absolute powers in classical frequency bands (Delta: 2–3.5 Hz, Theta: 4–7.5 Hz, Alpha: 8–12.5 Hz, Beta1: 13–23 Hz, Beta2: 23.5–33 Hz, Gamma: 33.5–44 Hz, [36]) in the two hemispheres (after logarithmic transformation), in healthy subjects and in patients classified according with the lesions side. Asterisks indicate significant difference with respect to the corresponding hemispheric value in controls (* < .050, ** < .001), arrows indicate increase or reduction. Bold values indicate significant interhemispheric AH-UH asymmetries (normal < .050, italic < .001). In the last column statistical correlation with the NIHSS score at symptoms onset is reported only for the AH, as no value in UH was associated with the clinical condition such sensor location entirely covers the sensorimotor brain areas for hand control as determined via anatomical MRI [69, 59]. The subjects were comfortably lying on a non-magnetic hospital bed, with their eyes open to reduce the spread of the parieto-occipital spontaneous activity toward the rolandic region. Spontaneous activity was recorded for three minutes in each hemisphere. The 28channel system does not allow simultaneous recording of spontaneous activity from both hemispheres, but this limitation does not hamper spectral characteristics’ estimate. In fact, in the literature the test-retest reliability of rest activity spectral characteristics have been reported. They show that evaluations based on more than 1 min of non-artifactual tracts is quite stable and also confirm our research on healthy subjects spontaneous activity recorded in different sessions (3–5 times) changing the order of left-right recordings. Recording characteristics and analysis procedures, as well as complete description of the features of the neuronal activity evoked by bilateral median nerve stimulation (SEF) [52] and at rest [73] at T0 have been previously described. In Appendix 1, the experimental and analysis procedures are fully provided. Average features of rest activities in patients are reported in Table 1. Values are compared with reference ones in a control population (21 healthy subjects, aged between 32 and 95 years (mean 67 ± 18 years, 10 females, 11 males).
j Magnetic Resonance Imaging (T0) Brain MRI at T0 was taken at 1.5 T, using Turbo Spin-Echo (TSE) and Spin-Echo (SE) T1 and T2 weighted sequences. T1 weighted images were also taken after intravenous administration of gadolinium-DTPA. All sequences provided contiguous 5 mm thick slices in sagittal, coronal and axial planes. Lesion characterization was performed on axial slices; the image where the lesion appeared with maximal dimension was selected and the size of the lesion was classified according to a three-degree scale: 1 for small (£ 5 cm3), 2 for medium (30 cm3). A lesion was further classified as ‘‘cortical’’ (C) when mainly cortical areas were involved or the lesion extended to subcortical white matter excluding basal ganglia through internal capsule; it was classified ‘‘subcortical’’ (S) when there was no visible cortical involvement and basal ganglia, thalamus, caudate nucleus, nucleus lenticularis or internal capsule were affected (Dromerick and Reding 1995); finally it was classified ‘‘cortico-subcortical’’ (SC) when both cortical and subcortical structures were involved.
j Statistical analysis The aim of the first statistical analysis was to identify which of the neurophysiological, neuroradiological and clinical variables observed in acute post-stroke, were significantly associated with the clinical outcome at T1 (worsening, partial and total recovery). Age and stroke severity were also entered in the model as potentially important predictors. Predictive variables were those MEG parameters previously shown to be sensitive to clinical status in stroke [58, 59, 81, 82, 52, 73]. To reduce the number of possible predictors and achieve a higher cases-to-variables ratio in the multiple regression model, we made a selection among the MEG parameters on the base of significance of bivariate correlations (assessed by means of Spearman’s rho) between outcome and each parameter. To be maximally sensitive in this first phase, we choose to use a relatively large type-I error (.10 bilateral) without any attempt to control ‘‘alpha-inflation’’. A more conservative approach was applied in the second step of analysis. In the second step, we tried to face the problem that the prognostic value of some of the previously analysed variables could be accounted for by the concomitant variation of some others. Thus, to understand which subset of parameters could predict the clinical outcome, we applied a multiple multinomial logistic regression with the three-level outcome entered as the dependent variable. Once a continuous neurophysiological index relationship to the outcome was significant and after controlling for confounding factors, an attempt was made to identify a potentially useful cut-off point. For this purpose, we used the Receiver-Operating-Characteristic (ROC) curve, obtained by connecting, on a Cartesian diagram, the points (xi,yi) where x denote the false-positive rates (1-specificity), y the true-positive rates (sensitivity) and i ranges from 1 to k, k being the number of all possible cut-offs. The Area-Under-Curve could be considered a measure of accuracy (predictive accuracy, in our case) and ranges from 0.5 (accuracy corresponding to chance) to 1 (perfect accuracy). The point (xm,ym) corresponds to the minimal distance from the point of maximal specificity and sensitivity (x = 0, y = 1) and was chosen as the ‘‘optimal’’ cut-off.
Results As shown in Table 2, the clinical outcome was inversely related with the following items at T0: clinical
299
status, lesion size, absolute delta power in UH (in AH a similar association was found, but below the significance threshold). It was positively associated with the following items: M20 strength, beta2 and gamma activity in AH. In addition, a better clinical recovery was found in patients with left hemisphere lesion with respect to those with a right hemisphere damage, and in patients with inter-hemispheric difference in strength of the two M20 and M30 components not exceeding normative limits (Table 2). After this selection, a first regression model considering exclusively MEG variables was applied. The only independent MEG variables resulted in absolute UH delta (Chi-square of the likelihood ratio = 12.74, p = .002) and AH gamma (Chi-square of the likelihood ratio = 11.29, p = .004) band powers. Due to the small sample size, the standard errors of the estimated parameters did not allow to compute reliable 95% confidence intervals. However, higher UH delta dramatically reduced the probability of a partial (p = 0.038) and complete (p = 0.016) recovery with respect to worsening. On the other hand, higher amount of AH gamma was associated with a higher probability of complete recovery vs. worsening (p = .017). As shown in Fig. 1, UH delta was specifically high in patients with worsening outcome and AH gamma was specifically normal in those undergoing complete recovery. In other words, partial and complete recovery were characterised by overlapping UH delta, whereas worsening and partial recovery by overlapping AH gamma (Fig. 1). Alterations in patients were increased for UH delta and decreased for AH gamma with respect to control values. After adding the lesion characteristics (volume and site) as well as the clinical status at T0 and age among the potentially predictive factors, we found that the lesion volume had the strongest predictive ability (Chi-square of the likelihood ratio = 17.20, p < .001), and UH delta absolute band power maintained a significant predictive value (Chi-square of the likelihood ratio = 15.16, p = .001). Area Under Curve resulted in 0.898 (p = 0.004) and the optimal cut-off was equal to 3.2. The corresponding sensitivity was 100% and specificity 79%. On the basis of the incidence of worsening in our cohort the Positive Predictive Value, obtained by means of Bayes’ formula, was 69% and the Negative Predictive Value was 100%. This means that a patient with UH delta below 3.2 (a negative value for the test) would recover to some extent, whilst a patient with UH delta above 3.2 (a positive value for the test) would have a high probability of clinical worsening (about 70%). When the lesion volume was taken into account, the following was observed: all 5 patients with a small volume recovered and no added prognostic contri-
Table 2 Bivariate association with the clinical outcome Spearman’s rho vol Les. side d_uh (*) g_ah (*) NIHSST0 b2_ah (*) Sasy20a Sasy30a s20ah
Mann-Whitney U
)0.505 )0.487 0.449 )0.402 0.399
39
42 42 0.352
p 0.003 0.003 0.008 0.014 0.017 0.027 0.028 0.028 0.033
(*) 24 cases Significant bivariate associations between selected MEG, MRI and clinical parameters with respect to the clinical outcome scores. Variables are sorted by statistical significance. Mann-Whitney test was used instead of Spearman’s correlation for dichotomous variables Lesion volume (vol); lesion side (Les. side); absolute power of rhythmic band activity in UH delta (d_uh), AH gamma (g_ah) and AH beta2 (b2_ah); NIHSS score in acute phase (NIHSST0); dichotomous parameters (1 altered, 0 normal) resulting from inter-hemispheric asymmetry of first (M20) and second (M30) brain components intensities above the reference value (0.40 for both, Tecchio et al 1997, Sasy20a, Sasy30a); amplitude of M20 in the AH, estimated by the strength of a single equivalent dipole (s20ah)
bution of UH delta was found; among the 6 patients with medium volume, the only worsening was observed in a patient with UH delta above the cut-off; finally, among the 17 patients with large lesion volume, all 7 below the cut-off, recovered partially or totally, and among the 10 above the cut-off , 8 worsened. In particular, when considering only patients with large lesion volume, UH delta was inversely associated with the clinical outcome (p = .006). To clarify the findings obtained by multiple stepwise regression, the correlation matrix between variables associated with outcome in the bivariate analysis was investigated. UH delta and AH gamma presented a clearly different pattern (Table 3): UH delta did not correlate with any other parameter (pvalues were consistently above .350) whilst AH gamma nicely correlated with nearly all clinical, MRI and MEG features. Consistently, lesion size was strongly associated with all MEG variables but UH delta activity. Since the prognostic value of lesion volume resulted higher than AH gamma in the bivariate analysis and these two parameters were strongly associated (Spearman’s correlation -.490, p = .010, Table 3), the principle of parsimony led AH gamma to be excluded from the predictive factors. In other words, AH gamma was overcome by lesion volume. Alternatively, UH delta seems to play an independent role with respect to both the neurophysiological picture and the lesion volume. Its prognostic value was confirmed even after partialling out the size of the lesion. As shown in Table 4, the association between band powers in the two hemispheres, consistently present in the healthy control group, lowered as a
300
Table 3 Bivariate association between relevant variables Spearman’s rho UH delta
AH gamma
lesion size
Fig. 1 Means and the corresponding standard errors of UH delta and AH gamma absolute band powers in the acute post-stroke stage, according to the clinical outcome during follow-up at T1. Control values are also showed for comparison
result of the lesion in all bands but delta. This indicated that in all bands but delta, the power increase/ decrease was strictly higher in the AH than in the UH, resulting in a inter-hemispheric correlation decrease; in the delta band, instead, the increase was similar in both the hemispheres, leaving the inter-hemispheric correlation similar to the level in controls. In addition to the correlation of UH delta power with all other neuroradiological, clinical and neurophysiological relevant indexes, further investigation was devoted to the relationship between contra-lesional delta power and the damage to any of the sensorimotor pathway regions, i.e. white matter thalamo-cortical sensory projections and corticospinal tract, thalamus, internal capsule, corona radiate, parietal and frontal central cortical areas. No association was observed (Spearman’s rho 0.650 consistently for the involvement of central frontal or parietal lobes, thalamus, internal capsule, corona radiata).
Discussion In the search for a post-stroke recovery indicator, the strongest prognostic value, in addition to the lesion
NIHSST0 sasy20a b2_ah g_ah sasy30a vol s20ah b2_ah NIHSST0 vol sasy20a s20ah sasy30a d_uh s20ah b2_ah sasy30a g_ah sasy20a NIHSST0 d_uh
Mann-Whitney U
0.186 56 0.096 0.063 )0.049 0.012 0.833 )0.624 )0.490
58
28 0.370 44 0.063 )0.496 )0.529 )0.49
45 45
0.397 )0.049
p 0.352 0.464 0.632 0.755 0.798 0.809 0.952 0.000 0.001 0.010 0.042 0.058 0.162 0.755 0.005 0.005 0.010 0.010 0.018 0.027 0.809
Section of association matrix between MEG and MRI data for those variables significantly associated to the outcome by the bivariate analysis. Variables are sorted by statistical significance. Mann-Whitney test was used instead of Spearman’s correlation for dichotomous variables Variables labels as in Table 3
volume, was provided by a high amount of ‘slow’ rhythmic oscillations within the delta band in MEG recordings from the unaffected hemisphere in the early days after stroke. In this way, our model provided additional predictions with respect to the baseline clinical evaluations. Delta rhythm in the adult brain during wakeful rest state is frequently found in damaged areas; the increase of their absolute power in UH in the acute stroke phase probably points toward a transcallosal diaschisis. The term diaschisis refers to a more or less transient alteration of brain function remote from the lesion, likely involving cerebral blood flow (CBF) [37] and brain metabolism reduction [80, 16] as well as spontaneous or stimulus-evoked variations of neuronal firing properties [24, 3, 45]. Von Monakow, who coined the term ‘diaschisis’ in a pre-EEG epoch [46, 47], suggested the loss of excitatory input from damaged areas to cause a lower metabolism in remote connected areas, structurally healthy. Both cross-sectional and longitudinal studies employing neurophysiological and functional neuroimaging techniques in stroke patients have shown that clinical recovery is paired with shifts of excitability balance between the affected and unaffected hemispheres, with recovery of activation levels in the affected side, and a reduction of both sides overactivations [30, 14, 41, 60]. This suggests that recovery is maximal when the brain regions that normally
301
Table 4 Inter-hemispheric band powers correlation
delta theta alpha beta gamma
r p r p r p r p r p
Controls (n = 17)
Patients (n = 27)
Comparison of correlations (p)
0.579 0.015 0.885 0.000 0.929 0.000 0.942 0.000 0.890 0.000
0.409 0.034 0.490 0.009 0.735 0.000 0.750 0.000 0.540 0.004
0.501 0.010 0.034 0.020 0.015
Inter-hemispheric band powers correlations in Controls and Patients and their comparisons
execute the ‘lost or damaged’ function are reintegrated into the original network devoted to the previously lost, and now recovered, functions [60]. EEG recordings in both experimental animal models of stroke [63, 35, 39] and in human patients [30, 2] showed that an acute ischemic insult is often associated with abnormality of background electrical brain activity with a decrease of EEG amplitude and an increase in delta band power not only, as it is usually due to the lesion in the affected hemisphere, but also in the contralateral hemisphere. In particular, it has been reported that these contralateral abnormalities of background rhythm correlate with impairment of consciousness and poor outcome [2]. However, there is no general consensus about the genesis of abnormalities on contra-lesional brain regions. Some authors correlate the bilateral EEG alterations with the presence of brain oedema [63] or perfusion abnormalities [39]. On the other hand, other authors proved no correlation in the contralesional hemisphere between regional cerebral blood flow (rCBF) and slower EEG frequencies [43]. In our study, the increase of delta band absolute power in the UH emerged as the most sensible indicator of poor long-term outcome. This was not associated to any of the MEG indexes in the AH but delta band power, suggesting that UH delta activity is not mediated by the neuronal dysfunction severity present in the AH. Moreover, both AH and UH delta powers were not associated with the lesion volume, but they were associated with each other. This suggests that the amount of perilesional functionally silent, but still alive neurons with their silencing and heavy damage, contribute to slow band AH activity [77, 8,12] and might lead to transcallosal spread of delta activity in the UH. There is much scientific evidence supporting the notion that neuronal reorganization and plastic phenomena following deafferentation begin in the very early stages, and continue for several weeks after
stroke. In adult rats after ischemic cortical lesions [15] two sequential patterns of low-frequency (day 1, 0.2–2 Hz; day 2–3, 0.1–0.4 Hz) synchronized neuronal activity from the non-affected cortex were observed, paired with a robust axonal sprouting of contralateral corticostriatal neurons into the denervated striatum. In humans after stroke the neuronal activity is characterized by distant areas’ hyper-excitability [65, 11, 52] and reduced intracortical inhibition [17]. Experimental findings from animal studies, using photothrombotically induced cortical lesion, showed bilateral alterations of excitability, providing evidence that cortical GABAergic inhibition decreases [10], in parallel with a down-regulation of GABAA receptor binding and GABAA receptor subunits [64, 49]. Changes in the calcium current following transient middle cerebral arterial occlusion have been demonstrated in the contralateral non-infarcted tissue [9]. Extra-cellular calcium concentration fluctuations have been observed related to delta rhythm generation during slow sleep [40, 66]. On the basis of these findings, we can hypothesize that the increased delta activity observed in the contralateral hemisphere can be an expression of early neuronal reorganization. In agreement with previous research findings, the lesion volume has been found as a strong individual predictive index of clinical outcome (p < 0.001). In fact, many studies demonstrated that in supratentorial stroke the acute lesion volume, evaluated with different techniques, predicts clinical recovery. The larger visible lesion on standard T2 weighted MRI [62] and diffusion-weighted imaging (DWI) were associated with worse clinical outcome at T1 [78, 38, 75, 6, 7, 49, 4, 50]. We found that some MEG parameters correlated with clinical outcome: AH SEF absence or significantly reduced and excessive interhemispheric strength asymmetries of generator sources among those related to brain reactivity; AH beta2 and gamma and UH delta band powers among those describing spontaneous rhythm characteristics. Multiple multinomial logistic regression on MEG parameters indicated that the only AH independent prognostic variable was the gamma band power, showing preserved levels in patients with good recovery. This finding underlines the relevant role of gamma band activity for the functionality of cortical areas devoted to hand control, in addition to the known contribution in cognitive motor tasks [31, 42], as an indicator of selective neural recruitment [44], and as coding feature for functional prevalence in hand sensory areas [70, 71]. A relationship between gamma band activity preservation and gap junction (GJs) functional integrity could be hypothesized, since interneuron dendrites GJs have been reported to enhance syn-
302
chrony of gamma oscillations in distributed networks [76], by virtue of their ability to induce synchronous firing in principal neurons. A possible neuroprotective role of modulating gap junctional intercellular communication has been investigated in cerebral ischemia: GJ inhibitors, when not limited by toxicity, have been found to have a significant therapeutic potential in the functional recovery from acute stroke [56, 48]. As a corollary observation, the side of stroke resulted significantly correlated with clinical outcome (p = .002). Papers focusing on stroke side as a potential prognostic tool of clinical outcome after stroke reported contradictory findings. Some studies showed no differences in outcome relating to it, while others found that patients affected by right hemispheric damage recovered less than those with left [57, 79]. Nevertheless, in our patient cohort, right hemispheric damage showed a larger lesion volume than left, therefore biasing the discrimination between the respective roles of lesion side and volume in longterm recovery. Before investigating the potential prognostic role of neurophysiological parameters in stroke patients, a detailed complete analysis of the neuronal activity of the areas devoted to the hand control in the acute state was performed both in response to a simple sensory stimulus [52] and at rest [73]. In addressing these aspects, MEG technique is proven to be suitable to obtain reliable indexes, especially for evoked activity [67, 55, 68, 69, 72, 81]. In the present study involving a 32 patient cohort, only rest activity characteristics contributed to predicting the outcome. We performed our study in strict accordance to published procedural guidelines and obtained a strong reduction of the variables of interest that had an effect on final outcome among the several neurophysiological parameters related to rest and evoked brain activity. This is the basis upon which we plan to reach a suitable event-per-variable (EPV, around 10 [18]) ratio in future studies. The unexpected finding of predictive information coming only from rest neural activity characteristics, after confirmation in larger samples, makes it mandatory to verify if EEG could provide the same information. Since it is well known that whole scalp, high spatial resolution EEG is highly stressful for the patient, our MEG investigation could identify sub-regions of interest to limit the EEG montage. In conclusion, MEG indicators in the early poststroke stages were found to provide useful information about the final outcome: the reduction of AH gamma and the increase of UH delta band powers. A cut-off was obtained for UH delta power, so that a patient with UH delta below 3.2 would partially or totally recover, whilst a patient with UH delta above
3.2 would have a probability of worsening of about 70%. Present results indicate that the increase of delta band activity in the unaffected hemisphere contains information about the outcome in addition to the lesion volume. j Acknowledgements This work has been partially supported by the 2003 - protocol 2003060892 of the Italian Department of University and Research (MIUR) and by the RBNE01AZ92_003, PNR 2001–2003, Fondi di Investimento per la Ricerca di Base (FIRB) and by the IST/FET Integrated Project NEUROBOTICS - The fusion of NEUROscience and roBOTICS, Project no. 001917 under the 6th Framework Programme. The Authors thank Professors Vittorio Pizzella and GianLuca Romani for their continued support, Doctors Claudia Altamura, Maria Filippi, Antonio Oliviero, Francesco Tibuzzi, Giancarlo Zito for patient testing and TNFP Matilde Ercolani for her excellent technical support.
Appendix 1 In all the text, the term ‘‘absolute’’ is used to indicate the absolute value of a variable with no reference to the value of the same variable in the other hemisphere; for the inter-hemispheric difference the term ‘‘asymmetry’’ is employed.
j Evoked activity recording and analysis Subjects underwent separate electrical stimulation of their left and right median nerves at the wrist. Stimuli consisted of 0.2 ms electric pulses (cathode proximal), with an inter-stimulus interval of 641 ms, and delivered through surface disks. Stimulus intensities were adjusted just above the threshold inducing a painless thumb twitch. Data were collected from the contralateral hemisphere, focusing on two early waves (M20, M30, i.e., respectively around 20 and 30 ms from the stimulus). These components are selectively generated in the brain’s sensorimotor areas contralateral to the stimulated hand, are quite stable, repeatable and independent of the subject’s attention. Averaged M20 and M30 latencies, strengths and locations of their equivalent current dipoles (ECD), and their inter-hemispheric differences from stroke patients sample were compared with a normative data set recorded from a population of sex-age matched healthy subjects. The SEF wave morphologies in the two hemispheres were also compared via a correlation based technique (see [66]). This morphologic analysis provides an evaluation of the inter-hemispheric correlations between sensory processes activated within primary sensorimotor areas, i.e. in the time interval starting at the M20 onset and lasting 30 ms; due to
303
strong inter-hemispheric similarity in the same subject, a normative set was defined (sim_SI, 0.76 [66]) and the wave shape analysis was carried out taking the unaffected hemisphere’s wave shape as a template. Since a measurable onset of the M20 component is needed to define the two epochs, patients with missing SEF are not included in the inter-hemispheric morphology evaluation, unless being considered out of normative ranges.
j Spontaneous brain activity recording and analysis Spontaneous activity was recorded for three minutes. After a visual inspection data and the application of an artefact rejection procedures [5], the Power Spectral Density (PSD) was estimated for each MEG channel via the Welch procedure (2048 ms duration, Hanning window, 60% overlap, about 180 artefact free trials used). The total PSD was calculated as the mean of the PSDs obtained by the 16 inner gradiometer channels which covered a circular area of about 12 cm diameter. Total signal power was obtained by integrating the PSD value in the 2–44 Hz frequency interval. Spectral properties were investigated in the classical frequency bands [36], instead of being settled on the basis of individual spectral characteristics [33], as spectral properties are known to be affected by stroke. The investigated frequency bands were: 2–3.5 Hz (Delta), 4–7.5 Hz (Theta), 8–12.5 Hz (Alpha), 13–23 Hz (Beta1), 23.5–33 Hz (Beta2), 33.5–44 Hz (Gamma). Relative power spectral density
(rPSD) was obtained as the ratio of PSD to total power in the 2–44 Hz frequency range. The relative power within each frequency band was calculated in the same way. All power values, absolute and relative, were log transformed in order to better fit normal distribution for statistical analysis. Spectral entropy quantifies the complexity of the frequency content, i.e. the spectral shape. It gives a measure of how much a rPSD is fragmented (minimal entropy) or flat (maximal entropy), independently of the total power. For example, a sinusoid is characterized by only one spectral component and has minimal entropy; on the opposite extreme white noise, whose PSD contains all the frequencies with the same weight, has maximal entropy. Entropy is calculated in the 2–44 Hz frequency interval:
44 X
rPSDðf Þ log2 rPSDðf Þ
f ¼2
Since our frequency resolution was 0.49 Hz in a total range 2–44 Hz, the total number of points is 86 (= 42/0.49), resulting in a maximal entropy value of 6.476 (flat spectrum); for the lower limit, we considered one peak whose amplitude was 99% above the basal level, resulting in entropy value of 4.037 (peaked spectrum). We also defined a hemispheric individual alpha frequency (IAF) as the frequency with maximal PSD in the 6–13 Hz band in each hemisphere [33].
References 1. Adams HP Jr, Adams JR, Brott T, del Zoppo GJ, Furlan A, Goldstein LB, et al. (2003) Guidelines for the early management of patients with ischemic stroke. A Scientific Statement from the Stroke Council of the American Stroke Association. Stroke 34:1056 –1083 2. Ahmed I (1988) Predictive value of the electroencephalogram in acute hemispheric lesions. Clin Electroencephalogr 19:205–9 3. Andrews RJ (1991) Transhemispheric diaschisis. A review and comment. [Review] Stroke 22:943–9 4. Baird AE, Dambrosia J, Janket S-J, Eichbaum Q, Chaves C, Silver B et al. (2001) A three-item scale for the early prediction of stroke recovery. Lancet 357:2095–99
5. Barbati G, Porcaro C, Zappasodi F, Rossini PM, Tecchio F (2004) Optimization of an independent component analysis approach for artifact identification and removal in magnetoencephalographic signals. Clin Neurophysiol 115:1220–32 6. Barber PA, Darby DG, Desmond PM, Yang Q, Gerraty RP, Jolley D (1998) Prediction of stroke outcome with echoplanar perfusionand diffusionweighted MRI. Neurology 51:418–26 7. Beaulieu C, de Crespigny A, Tong DC, Moseley ME, Albers GW, Marks MP (1999) Longitudinal magnetic resonance imaging study of perfusion and diffusion in stroke: evolution of lesion volume and correlation with clinical outcome. Ann Neurol 46:568–78
8. Benzel EC, Lewine JD, Bucholz RD, Orrison WW (1993) Magnetic source imaging: a review of the Magnes system of biomagnetic technologies incorporated. Neurosurgery 33:252–9 9. Bruehl C, Neumann-Haefelin T, Witte OW (2000) Enhancement of whole cell calcium currents following transient MCAO. Brain Res 884:129–38 10. Buchkremer-Ratzmann I, August M, Hagemann G, Witte OW (1996) Electrophysiological transcortical diaschisis after cortical photothrombosis in rat brain. Stroke 27:1105–11 11. Butefisch CM, Netz J, Wessling M, Seitz RJ, Homberg V (2003) Remote changes in cortical excitability after stroke. Brain 126:470–81
304
12. Butz M, Gross J, Timmermann L, Moll M, Freund H-J, Witte OW, et al. (2004) Perilesional pathological oscillatory activity in the magnetoencephalogram of patients with cortical brain lesions. Neurosci Letters 355:93–96 13. Calabresi P, Cupini LM, Centonze D, Pisani F, Bernardi G (2003) Antiepileptic drugs as a possible neuroprotective strategy in brain ischemia. Ann Neurol 53:693–702 14. Calautti C, Leroy F, Guincestre JY, Marie RM, Baron JC (2001) Sequential activation brain mapping after subcortical stroke: changes in hemispheric balance and recovery. Neuroreport 12:3883–6 15. Carmichael ST, Chesselet MF (2002) Synchronous neuronal activity is a signal for axonal sprouting after cortical lesions in the adult. J Neurosci 22:6062–70 16. Chu WJ, Mason GF, Pan JW, Hetherington HP, Liu HG, San Pedro EC, et al. (2002) Regional cerebral blood flow and magnetic resonance spectroscopic imaging findings in diaschisis from stroke. Stroke 33:1243–8 17. Cicinelli P, Pasqualetti P, Zaccagnini M, Traversa R, Oliveri M, Rossini PM (2003) Interhemispheric asymmetries of motor cortex excitability in the postacute stroke stage: a paired-pulse transcranial magnetic stimulation study. Stroke 34:2653–8 18. Counsell C, Dennis M (2001) Systematic Review of Prognostic Models in Patients with Acute Stroke. Cerebrovasc Dis 12:159–70 19. Cuspineda E, Machado C, Aubert E, Galan L, Llopis F, Avila Y (2003) Predicting outcome in acute stroke: a comparison between QEEG and the Canadian Neurological Scale. Clin Electroencephalogr 34:1–4 20. Dachy B, Biltiau E, Bouillot E, Dan B, Deltenre P (2003) Facilitation of motor evoked potentials in ischemic stroke patients: prognostic value and neurophysiologic correlations. Clin Neurophysiol 114:2370–2375 21. De Reuck J, Paemeleire K, Van Maele G, Goethals M (2004) The prognostic significance of changes in lesion size of established cerebral infarcts on computed tomography of the brain. Cerebrovasc Dis 17:320–5 22. Del Gratta C, Pizzella V, Tecchio F, Romani G-L (1986) Magnetoencephalography - a non invasive brain imaging method with 1 ms time resolution. Rep Prog Phys 2001 64:1759–1814
23. Escudero JV, Sancho J, Bautista D, Escudero M, Lopez-Trigo J (1998) Prognostic value of motor evoked potential obtained by transcranial magnetic brain stimulation in motor function recovery in patients with acute ischemic stroke. Stroke 29:1854–1859 24. Feeney DM, Baron JC. Diaschisis. [Review]. Stroke 17:817–30 25. Feeney DM (1997) From laboratory to clinic: noradrenergic enhancement of physical therapy for stroke or trauma patients. Adv Neurol 73:383–94 26. Fierro B, La Bua V, Oliveri M, Daniele O, Brighina F (1999) Prognostic value of somatosensory evoked potentials in stroke. Electromyograph Clin Neurophysiol 39:155–60 27. Finnigan SP, Rose SE, Walsh M, Griffin M, Janke AL, McMahon KL, et al. (2004) Correlation of quantitative EEG in acute ischemic stroke with 30-day NIHSS score: comparison with diffusion and perfusion MRI. Stroke 35:899– 903 28. Forss N, Hietanen M, Salonen O, Hari R (1999) Modified activation of somatosensory cortical network in patients with right-hemisphere stroke. Brain 122:1889–99 29. Friel KM, Nudo RJ (1998) Recovery of motor functions after focal cortical injury in primates: compensatory movement patterns used during rehabilitative training. Somatosens Mot Res 15:173–89 30. Giaquinto S, Cobianchi A, Macera F, Nolfe G (1994) EEG recordings in the course of recovery from stroke. Stroke 25:2204–9 31. Gross DW, Gotman J (1999) Correlation of high-frequency oscillations with the sleep–wake cycle and cognitive activity in humans. Neuroscience 94:1005–18 32. Huang JC, Nicholson C, Okada YC (1990) Distortion of magnetic evoked fields and surface potentials by conductivity differences and boundaries in brain tissue. Biophysical J 57:1155–667 33. Klimesch W (1999) EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. [Review]. Brain Res Rev 29:169–195 34. Kluytmans M, Van Everdingen KJ, Kappelle LJ, Ramos LM, Viergever MA, van der Grond J (2000) Prognostic value of perfusion- and diffusion-weighted MR imaging in first 3 days of stroke. Eur Radiol 10:1434–41 35. Komatsumoto S, Greenberg JH, Hickey WF, Reivich M (1995) Clinicopathological observations in middle cerebral artery occlusion in the cat. Neurol Res 17:120–8
36. IFSECN (1974) A glossary of terms commonly used by clinical electroencephalographers. Electroencephalogr Clin Neurophysiol 37:538–548 37. Lagreze HL, Levine RL, Pedula KL, Nickles RJ, Sunderland JS, Rowe BR (1987) Contralateral flow reduction in unilateral stroke: evidence for transhemispheric diaschisis. Stroke 18:882–6 38. Lovblad KO, Baird AE, Schlaug G, Benfield A, Siewert B, Voetsch B et al. (1997) Ischemic lesion volumes in acute stroke by diffusion-weighted magnetic resonance imaging correlate with clinical outcome. Ann Neurol 42:164–170 39. Lu XC, Williams AJ, Tortella FC (2001) Quantitative electroencephalography spectral analysis and topographic mapping in a rat model of middle cerebral artery occlusion. Neuropathol Appl Neurobiol 27:481–95 40. Massimini M, Amzica F (2001) Extracellular calcium fluctuations and intracellular potentials in the cortex during the slow sleep oscillation. J Neurophysiol 85:1346–50 41. Matteis M, Vernieri F, Troisi E, Pasqualetti P, Tibuzzi F, Caltagirone C, et al. (2003) Early cerebral hemodynamic changes during passive movements and motor recovery after stroke. J Neurol 250:810–7 42. Meador KJ, Ray PG, Echauz JR, Loring DW, Vachtsevanos GJ (2002) Gamma coherence and conscious perception. Neurology 59:847–54 43. Melamed E, Lavy S, Portnoy Z, Sadan S, Carmon A (1975) Correlation between regional cerebral blood flow and EEG frequency of the contralateral hemisphere in acute cerebral infarction. J Neurol Sci 26:21–27 44. Menon V, Freeman WJ, Cutillo BA, Desmond JE, Ward MF, Bressler SL, et al. (1996) Spatio-temporal correlations in human gamma band electrocorticograms. Electroencephalogr Clin Neurophysiol 98:89–102 45. Meyer JS, Obara K, Muramatsu K (1993) Diaschisis [Review]. Neurol Res 15:362–6 46. von Monakow C (1895) Experimentelle und pathologisch-anatomische Untersuchungen uber die Haubenregion, Sehhugel und die Regio subthalamica, nebst Beitragen zur Kenntnis fruh erworbener Grob- und Kleihirndefekten. Arch Psychtrie Nervenkrankheiten 27:1–128 47. von Monakow C (1914 ) Die Lokalisation im Großhirn und der Abbau der Funktion durch Kortikale Herde. Bergmann JF, Wiesbaden
305
48. Nakase T, Fushiki S, Naus CC (2003) Astrocytic gap junctions composed of connexin 43 reduce apoptotic neuronal damage in cerebral ischemia. Stroke 34:1987–93 49. Neumann-Haefelin T, Wittsack HJ, Wenserski F, Siebler M, Seitz RJ, Modder U et al. (1999) Diffusion- and perfusion-weighted MRI. The DWI/ PWI mismatch region in acute stroke. Stroke 30:1591–97 50. Nighoghossian N, Hermier M, Adeleine P, Derex L, Dugor JF, Philippeau F, et al. (2003) Baseline magnetic resonance imaging arameters and stroke outcome in patients treated by intravenous tissue plasminogen activator. Stroke 34:458–63 51. Nudo RJ, Wise BM, SiFuentes F, Milliken GW (1996) Neural substrates for the effects of rehabilitative training on motor recovery after ischemic infarct. Science 272:1791–94 52. Oliviero A, Tecchio F, Zappasodi F, Pasqualetti P, Salustri C, Lupoi D, et al. (2004) Brain sensorimotor hand area functionality in acute stroke: insights from magnetoencephalography. Neuroimage 23:542–550 53. Orrison WW, Lewine JD (1993) Magnetic source imaging in neurosurgical practice. Perspect Neurol Surg 4:141–8 54. Pe´re´on Y, Aubertin P, Guihe´neuc P (1995) Prognostic significance of electrophysiological investigations in stroke patients: somatosensory and motor evoked potentials and sympathetic skin response. Neurophysiol Clin 25:146–57 55. Pizzella V, Tecchio F, Romani G-L, Rossini PM (1999) Functional localization of the sensory hand area with respect to the motor central gyrus knob. Neuroreport 10:1–6 56. Rawanduzy A, Hansen A, Hansen TW, Nedergaard M (1997) Effective reduction of infarct volume by gap junction blockade in a rodent model of stroke. J Neurosurg 87:916–20 57. Robbins J, Levine RL, Maser A, Rosenbek JC, Kempster GB (1993) Swallowing after unilateral stroke of the cerebral cortex. Arch Phys Med Rehabil 74:1295–300 58. Rossini PM, Tecchio F, Pizzella V, Lupoi D, Cassetta E, Pasqualetti P, et al. (1998) On the reorganization of sensory hand areas after mono-hemispheric lesion: a functional (MEG)/ anatomical (MRI) integrative study. Brain Res 782:153–166 59. Rossini PM, Tecchio F, Pizzella V, Lupoi D, Cassetta E., Pasqualetti P (2001) Interhemispheric differences of sensory hand areas after monohemispheric stroke: MEG/MRI integrative study. Neuroimage 14:474–485
60. Rossini PM, Calautti C, Pauri F, Baron JC (2003) Post-stroke plastic reorganisation in the adult brain. [Review]. Lancet Neurol 2:493–502 61. Rossini PM, Altamura C, Ferretti A, Vernieri F, Zappasodi F, Caulo M, et al. (2004) Does cerebrovascular disease affect the coupling between neuronal activity and local hemodinamics? Brain 127:99–110 62. Saunders DE, Clifton AG, Brown MM (1996) Measurement of infarct size using MRI predicts prognosis in middle cerebral artery infarction. Stroke 26:2272–6 63. Scavini C, Rozza A, Bo P, Lanza E, Favalli L, Savoldi F et al. (1990) Kappaopioid receptor changes and neurophysiological alterations during cerebral ischemia in rabbits. Stroke 21:943– 7 64. Schiene K, Bruehl C, Zilles K, Qu M, Hagemann G, Kraemer M et al. (1996) Neuronal hyperexcitability and reduction of GABAAreceptor expression in the surround of cerebral photothrombosis. J Cereb Blood Flow Metab 16:906–14 65. Shimizu T, Hosaki A, Hino T, Sato M, Komori T, Hirai S et al. (2002) Motor cortical disinhibition in the unaffected hemisphere after unilateral cortical stroke. Brain 125:1896–1907 66. Steriade M (2004) Neocortical cell classes are flexible entities. Nat Rev Neurosci 5:121 67. Tecchio F, Rossini PM, Pizzella V, Cassetta E, Romani GL (1997) Spatial properties and interhemispheric differences of the sensory hand cortical representation: a neuromagnetic study. Brain Res 767:100–8 68. Tecchio F, Pasqualetti P, Pizzella V, Romani GL, Rossini PM (2000) Morphology of somatosensory evoked fields: inter-hemispheric similarity as a index for physiological and pathological neural connectivity. Neurosci Lett 287:203–6 69. Tecchio F, Padua L, Aprile I, Rossini PM (2002) Carpal Tunnel Syndrome modifies sensory hand cortical somatotopy: a MEG study. Hum Brain Mapping 17:28–36 70. Tecchio F, Babiloni C, Zappasodi F, Vecchio F, Pizzella V, Romani G-L, et al. (2003) Gamma synchronization in human primary somatosensory cortex as revealed by somatosensory evoked neuromagnetic fields. Brain Res 986:63–70
71. Tecchio F, De Lucia M, Salustri C, Montuori M, Bottaccio M, Babiloni C, et al. (2004) District-related frequency specificity in hand cortical representation: dynamics of regional activation and intra-regional functional connectivity. Brain Res 1014:80–86 72. Tecchio F, Zappasodi F, Pasqualetti P, Rossini PM (2005) Neural connectivity in hand sensorimotorbrain areas: an evaluation by evoked fields morphology. Hum Brain Mapp 24:99–108 73. Tecchio F, Zappasodi F, Pasqualetti P, Tombini M, Salustri C, Oliviero A, et al. (2005) Rhythmic brain activity at rest from rolandic areas in acute monohemispheric stroke: a magnetoencephalographic study. Neuroimage 28:72–78 74. Timmerhuis TPJ, Hageman G, Oosterloo SJ, Rozeboom AR (1996) The prognostic value of cortical magnetic stimulation in acute middle cerebral artery infarction compared to other parameters. Clin Neurol Neurosurg 98:231–6 75. Tong DC, Yenari MA, Albers GW, O’Brien M, Marks MP, Moseley ME (1998) Correlation of perfusion- and diffusion weighted MRI with NIHSS score in acute (6.5 hour) ischemic stroke. Neurology 50:864–870 76. Traub RD, Kopell N, Bibbig A, Buhl EH, LeBeau FE, Whittington MA (2001) Gap junctions between interneuron dendrites can enhance synchrony of gamma oscillations in distributed networks. J Neurosci 21:9478–86 77. Vieth JB (1990) Magnetoencephalography in the study of stroke cerebrovascular accident. Adv Neurol 54:261–269 78. Warach S, Dashe JF, Edelman RR (1996) Clinical outcome in ischemic stroke predicted by early diffusionweighted and perfusion magnetic resonance imaging: a preliminary analysis. J Cereb Blood Flow Metab 16:53–59 79. Ween JE, Alexander MP, D’Esposito M, Roberts M (1996) Factors predictive of stroke outcome in a rehabilitation setting. Neurology 47:388–92 80. Wise R, Gibbs J, Frackowiak R, Marshall J, Jones T (1986 ) No evidence for transhemispheric diaschisis after human cerebral infarction. Stroke 17:853– 61 81. Wikstrom H, Roine RO, Salonen O, Lund KB, Salli E, Ilmoniemi RJ, et al. (1999) Somatosensory evoked magnetic fields from the primary somatosensory cortex (SI) in acute stroke. Clin Neurophysiol 110:916–23 82. Wikstrom H, Roine RO, Aronen HJ, Salonen O, Sinkkonen J, Ilmoniemi RJ, et al. (2000) Specific changes in somatosensory evoked magnetic fields during recovery from sensorimotor stroke. Ann Neurol 47:353–60
