Hemodynamic and metabolic responses to activation, deactivation and epileptic discharges Bojana Stefanovic, * Jan M. Warnking, Eliane Kobayashi, Andrew P. Bagshaw, Colin Hawco, Franc ¸ois Dubeau, Jean Gotman, and G. Bruce Pike Montreal Neurological Institute, 3801 University Street, Montreal, QC, Canada H3A 2B4 Received 2 February 2005; revised 24 March 2005; accepted 19 May 2005 Available online 5 July 2005 To investigate the coupling between the hemodynamic and metabolic changes following functional brain activation as well as interictal epileptiform discharges (IEDs), blood oxygenation level dependent (BOLD), perfusion and oxygen consumption responses to a unilateral distal motor task and interictal epileptiform discharges (IEDs) were examined via continuous EEG-fMRI. Seven epilepsy patients per- formed a periodic (1 Hz) right-hand pinch grip using ¨8% of their maximum voluntary contraction, a paradigm previously shown to produce contralateral M1 neuronal excitation and ipsilateral M1 neuronal inhibition. A multi-slice interleaved pulsed arterial spin labeling and T 2 *-weighted gradient echo sequence was employed to quantify cerebral blood flow (CBF) and BOLD changes. EEG was recorded throughout the imaging session and reviewed to identify the IEDs. During the motor task, BOLD, CBF and cerebral metabolic rate of oxygen consumption (CMR O 2 ) signals increased in the contra- and decreased in the ipsilateral primary motor cortex. The relative changes in CMR O 2 and CBF were linearly related, with a slope of 0.46 T 0.05. The ratio of contra- to ipsilateral CBF changes was smaller in the present group of epilepsy patients than in the healthy subjects examined previously. IEDs produced both increases and decreases in BOLD and CBF signals. In the two case studies for which the estimation criteria were met, the coupling ratio between IED-induced CMR O 2 and CBF changes was estimated at 0.48 T 0.17. These findings provide evidence for a preserved coupling between hemodynamic and metabolic changes in response to both functional activation and, for the two case studies available, in response to interictal epileptiform activity. D 2005 Elsevier Inc. All rights reserved. Keywords: EEG-fMRI; Epilepsy; Negative BOLD; Perfusion; Oxygen consumption Introduction Although the relationship between ictal and interictal epileptic activity is not entirely understood (Badier and Chauvel, 1985, 1995; Alarcon et al., 1994; de Curtis and Avanzini, 2001; Avoli, 2001; Janszky et al., 2001), interictal epileptiform discharges (IEDs) represent a very specific marker of epilepsy, the delineation of the irritative zone (Rosenow and Luders, 2001) being of particular interest for presurgical evaluations of epileptic patients (Penfield and Jasper, 1954; Kanner et al., 1995; McKhann et al., 2000). Interictal activity has traditionally been studied with electroencephalography (EEG), IEDs producing pronounced and stereotyped electroencephalographic trace deviations. Although ictal activity is generally associated with increased metabolism and perfusion (Duncan, 1997), no consistent changes in cerebral metabolic rate of glucose consumption (CMR Glc ) or cerebral blood flow (CBF) in response to interictal epileptiform activity have been demonstrated (Theodore et al., 1985; Ochs et al., 1987). This has often been ascribed to the poor sensitivity of the methods employed, e.g., low statistical power and poor temporal resolution of positron emission tomography (PET) studies, the latter leading to amalgamation of different states within each measurement (Duncan, 1997; Sperling and Skolnick, 1995). In the last decade, functional magnetic resonance imaging (fMRI) has been deployed in conjunction with EEG (Ives et al., 1993; Huang-Hellinger et al., 1995; Warach et al., 1996) to improve the EEG-based localization of the irritative zone and enable investigation of the hemodynamic and metabolic correlates of IEDs with high spatial and temporal resolution (Seeck et al., 1998; Krakow et al., 1999; Patel et al., 1999; Lazeyras et al., 2000; Jager et al., 2002; Al-Asmi et al., 2003). However, full use of the information afforded from fMRI BOLD data in the combined EEG-fMRI investigations is predicated on the understanding of the physiological changes determining the BOLD response, which are incompletely understood even in normal functional brain activa- tion. Detailed investigation of the BOLD response to IEDs has only recently begun (Lemieux et al., 2001; Benar et al., 2002; Bagshaw et al., 2004; Aghakhani et al., 2004). Regional negative BOLD responses to IEDs have been observed (Salek-Haddadi et al., 2003b; Archer et al., 2003a,b; Bagshaw et al., 2004; Aghakhani et al., 2004), but their origins are presently unknown. In contrast to focal epilepsies where an uncoupling between CBF and CMR Glc has been suggested (Gaillard et al., 1995; Fink et al., 1996; Breier et al., 1997; Bruehl et al., 1998) (and disputed 1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2005.05.038 * Corresponding author. Fax: +1 514 398 2975. E-mail address: [email protected] (B. Stefanovic). Available online on ScienceDirect (www.sciencedirect.com). www.elsevier.com/locate/ynimg NeuroImage 28 (2005) 205 – 215
11
Embed
Hemodynamic and metabolic responses to activation, deactivation and epileptic discharges
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
www.elsevier.com/locate/ynimg
NeuroImage 28 (2005) 205 – 215
Hemodynamic and metabolic responses to activation, deactivation
and epileptic discharges
Bojana Stefanovic,* Jan M. Warnking, Eliane Kobayashi, Andrew P. Bagshaw, Colin Hawco,
Francois Dubeau, Jean Gotman, and G. Bruce Pike
Montreal Neurological Institute, 3801 University Street, Montreal, QC, Canada H3A 2B4
Received 2 February 2005; revised 24 March 2005; accepted 19 May 2005
Available online 5 July 2005
To investigate the coupling between the hemodynamic and metabolic
changes following functional brain activation as well as interictal
central; Pc: posterior cingulate; Mc: mid-cingulate; A: scattered throughout the brain. Note that the number of events comprises both isolated events and bursts.
B. Stefanovic et al. / NeuroImage 28 (2005) 205–215 207
Experiment
The scanning protocol consisted of a high-resolution 3 D RF-
B. Stefanovic et al. / NeuroImage 28 (2005) 205–215 209
ipsilateral BOLD, (�37, �22, 53) for contralateral CBF, and (35,
�24, 54) for ipsilateral CBF.
A typical set of BOLD signal and CBF time courses, in both
contra- and ipsilateral M1-ROIs of a subject, is shown in Fig. 2.
Fig. 3 displays all measured BOLD and CBF data pairs, for
hypercapnic perturbation and motor task, as well as the calculated
iso-CMRO2contours. In 5 out of 7 subjects, the magnitude of CBF
and BOLD signal changes were larger in the contra- than in the
ipsilateral ROI.
The maximum achievable BOLD signal increase (M), obtained
by linear fitting of the average hypercapnia data across all subjects,
was 0.046 T 0.013, corresponding to aDR2* of�0.9 T 0.2 s1. The v2
analysis indicated a good fit ( q = 0.39) (Press et al., 1992). The
calculated CMRO2and the corresponding measured CBF changes,
for each subject, are displayed in Fig. 4. The slope of the straight line
fit to these data yielded a CMRO2/CBF coupling ratio of 0.46 T 0.05
(with q of 0.92 indicating an excellent fit (Press et al., 1992)).
Comparison with controls
The data of Fig. 3 has been replotted in Fig. 5, in conjunction
with the corresponding data from our previous study, which
Fig. 2. Time courses of contralateral (positive) BOLD (a) and CBF (b), as well as ip
standard errors are shown as dashed lines. All time course data have been low pas
24 sessions.
employed the same motor paradigm in a group of healthy subjects
(Stefanovic et al., 2004). Notably, the ratio of contra- to ipsilateral
CBF responses is significantly (P ¨ 0.017) smaller in epileptic
patients (2.2 T 1.3) than in healthy subjects (4.2 T 2.3). Average
contra- and ipsilateral CBF responses across subjects are shown in
Fig. 6. A very similar level of deactivation-induced percent CBF
decrease is seen in the two groups, in contrast to a smaller
excitation-induced percent CBF increase in the epilepsy patients.
The same trend of lower contra-to ipsilateral BOLD responses in
epileptic patients relative to controls is observed (2.7 T 2.5 vs. 3.7 T2.1), though it does not reach significance (P ¨ 0.21).
IED-induced responses
Six out of seven patients exhibited epileptiform activity in the
course of the scanning session (cf. Table 1 for the summary of EEG
findings). In each of these subjects, the interictal epileptiform
discharges induced both increases and decreases in BOLD and
CBF signals, as summarized in Table 1. Only two of the six
subjects exhibited sufficiently co-localized (i.e., at most 5 mm
separation between their respective ROI centers of mass, as
described in Methods) statistically significant changes in both
silateral (negative) BOLD (c) and CBF (d) percent changes in subject 1. The
s filtered with a Hanning window (FWHM = 20 s) prior to averaging across
Fig. 3. The percent changes in BOLD and CBF signals in the ipsilateral
ROIs (green circles) and contralateral ROIs (red triangles) for each subject.
The average hypercapnia data (black squares) are displayed along with the
corresponding fit (indicated by crosses), representing the baseline iso-
CMRO2contour. The estimate of the maximum achievable BOLD signal
change was substituted into the equation [13] of the deoxyhemoglobin
dilution model (Hoge et al., 1999) to generate non-baseline iso-CMRO2
contours (shown as solid black curves), at 10% intervals. The shaded area
corresponds to the shaded region of Fig. 4.
Fig. 5. The percent changes in BOLD and CBF signals induced by the
motor task in the ipsilateral ROIs (green circles: epilepsy patients; blue
circles: healthy subjects) and contralateral ROIs (red triangles: epilepsy
patients; magenta triangles: healthy subjects).
B. Stefanovic et al. / NeuroImage 28 (2005) 205–215210
BOLD and CBF. These included right parietal and right cuneus
regions in subject 1; and bilateral frontal, left occipital, bilateral
precentral, left precuneus and right cuneus regions in subject 7.
The two BOLD/CBF ROI pairs in subject 1 and 13 BOLD/CBF
ROI pairs in subject 7 met the above criteria. Sample t value maps
from subject 1 are shown in Fig. 7. As described in Methods,
BOLD and CBF data were averaged across the ROIs of each
subject prior to the calculation of CMRO2changes. The optimal
linear fit between the resulting CMRO2estimates and CBF data in
the ROIs of these two subjects is displayed in Fig. 8. The slope of
the straight line fit to these data yielded a CMRO2/CBF coupling
Fig. 4. The oxygen consumption changes corresponding to each subject’s
perfusion changes induced by the motor task in the ipsilateral ROIs (green
circles) and contralateral ROIs (red triangles). The optimal straight line fit
(q = 0.92) to these data is shown superimposed, yielding a coupling ratio of
0.46 T 0.05. The shaded region represents the standard error in the linear fit.
ratio of 0.48 T 0.17 (with q of 0.80 indicating a very good fit (Press
et al., 1992)).
Discussion
The present experiments provide, for the first time, a complete
set of BOLD, CBF and CMRO2measurements following functional
activation, deactivation and IEDs in epilepsy patients. They
demonstrate a preserved coupling between perfusion and oxygen
consumption changes in epilepsy patients. As was the case for the
healthy volunteers, the CMRO2/CBF relationship was consistent
between regions of positive and negative BOLD responses to a
motor task (with DCMRO2/DCBF of 0.46 T 0.05). For the regions
that showed statistically significant IED-induced changes in both
BOLD and CBF (thus allowing for CMRO2estimation), a similar
CMRO2/CBF coupling ratio, of 0.48 T 0.17, was estimated. Overall,
these findings are consistent with the general notion of epilepsy as
Fig. 6. The average, motor task induced, percent changes in CBF signal in
ipsi- and contralateral ROIs of healthy subjects (C) and epileptic patients (E).
Fig. 7. Sample BOLD (left) and CBF (right) t value maps in a subject (Subject Id. 1), overlaid on the corresponding anatomical slices. The regions of positive
responses are shown in the top row; the regions of negative responses, in the bottom row. The centers of mass for the overlapping regions are shown with a
cross hair.
Fig. 8. The oxygen consumption changes corresponding to across ROI
average IED-induced perfusion and BOLD changes in subjects 1 and 7. The
averages for ROIs showing IED-induced signal decreases are shown as
green circles; the averages for ROIs showing IED-induced signal increases,
as red triangles. The optimal straight line fit (q = 0.80) to these data is
shown superimposed, yielding a coupling ratio of 0.48 T 0.17. The shaded
region represents the standard error in the linear fit.
B. Stefanovic et al. / NeuroImage 28 (2005) 205–215 211
a disorder of neuronal excitability, involving neuronal disinhibition
and hyperexcitability.
While neuronal hyperexcitability is thought to characterize
most epilepsy syndromes, the pathophysiology of these diseases is
still incompletely understood. The role of genetics in idiopathic
generalized epilepsy has long been suspected (Metrakos and
Metrakos, 1961) and a number of different IGE subsyndromes
have recently been associated with distinct mutations in GABAA
receptor sub-units (Macdonald et al., 2004; Gutierrez-Delicado and
Serratosa, 2004). Although IGE patients have normal structural
MRI, regional decreases in their N-acetyl aspartate levels have
been reported (Savic et al., 2004), suggesting a heterogeneous,
diffuse neuronal abnormality. The average resting metabolism and
flow in IGE patients are largely unremarkable (Duncan, 1997;
Theodore et al., 1985; Ochs et al., 1987; Kapucu et al., 2003;
Devous et al., 1990), in sharp contrast to the hypometabolism and
hypoperfusion frequently observed in the area of the epileptogenic
focus, its immediate surround or elsewhere in the brain of patients
with localization-related epilepsies (Kuhl et al., 1980; Engel et al.,
1982; Lee et al., 1986; Franck et al., 1986; Kim et al., 2001). While
initial seizures in serial seizure animal models were accompanied
by the expected increases in cerebral blood volume, arterial blood
pressure, cortical oxygen tension and cytochrome oxidase pressure,
one or more of these variables failed to rise in response to
B. Stefanovic et al. / NeuroImage 28 (2005) 205–215212
subsequent seizures, testifying to a gradual breakdown of neuro-
vascular coupling in these patients (Kreisman et al., 1981, 1983).
Similarly, in a near-infrared spectroscopy study of pediatric
epileptic seizures, an early CBV increase gradually changed to a
CBV decrease in the course of the seizure in a patient with tonic
status epilepticus (Haginoya et al., 2002). In line with the present
findings, these data suggest a preserved interictal neurovascular
coupling that is progressively compromised in the course of either
sustained or highly repetitive ictal events.
Each patient in this study was taking one or more antiepileptic
drugs (AEDs), which were reported to reduce baseline CMRGlc and
CBF (Theodore et al., 1989; Leiderman et al., 1991; Spanaki et al.,
1999; Gaillard et al., 1996). A reason for these reductions may lie
in the decreased metabolic requirements following the enhance-
ment of cerebral inhibitory neurotransmission (Theodore, 1988). In
view of the effect of valproate on brain and CSF GABA levels
(Loscher, 1979, 1981), it is important to note that increased CSF
GABA (which is linearly related to brain GABA (Palfreyman et al.,
1983; Petroff et al., 1996)), following administration of the GABA
agonist muscimol, was found to affect both blood flow and glucose
consumption as to maintain a normal relationship between the two
(Kelly and McCulloch, 1983). While the medications might have
shifted the absolute global flow and metabolism in these subjects,
there is no evidence in support of their influence on the relationship
between metabolic and hemodynamic responses to changes in
neuronal activity. Furthermore, we do not expect any spatial
variation of their effects across homologous brain regions—
namely, primary motor cortices—of interest for our functional
paradigm.
While fMRI BOLD studies of physiological brain activation are
often done in surgical epileptic candidates to map the eloquent
cortex (Deblieck et al., 2003; Diehl et al., 2003; Huettel et al.,
2004; Szaflarski et al., 2004), there are no studies making
simultaneous measurements of activation-induced BOLD, CBF
and CMRO2changes in epilepsy patients, likely due to the
complexity and limited SNR of such measurements. The slope of
0.46 T 0.05 of the best line fit to both contra- and ipsilateral M1
CMRO2vs. CBF percent signal changes found here is in excellent
agreement with the value of 0.44 T 0.04 we reported in an earlier
study of healthy volunteers performing the same motor task. It is
also in reasonable agreement (given the paradigm differences and
the expected intersubject variability) with the ratios reported by our
and other groups for the contralateral primary motor cortex
activation in studies of BOLD signal increases, with the average
of 0.35 T 0.03 found in this lab (Atkinson et al., 2000) and 0.33 T0.06 reported by Kastrup (Kastrup et al., 2002).
The ratio of the changes in perfusion—between the contrala-
teral region of neuronal activation and homologous ipsilateral
region of neuronal deactivation—was decreased in the epileptic
patients compared to the corresponding value in healthy subjects
studied previously. This significant difference in the CBF ratios
resulted from a decrease in the activation-induced CBF increases,
with a preserved range of deactivation-induced CBF decreases. In
view of the diffuse cortical hyperexcitability presumed to exist in
these patients and the suggested dominant contribution of
presynaptic potentials to the total metabolic demands of neuronal
activity (Logothetis et al., 2001), it is tempting to speculate that the
relative metabolic cost of neuronal activation with respect to
neuronal deactivation may be diminished in these patients when
compared to healthy volunteers, though the neuronal excitability
(and hence the energetic cost of the activation) may well be
influenced by AEDs (Tassinari et al., 2003). The dissociation of the
effects of the underlying pathology from those of the medications
is, however, presently unavailable.
In contrast to the sustained after-depolarizations and multiple
spike discharges characteristic of the ictal state (Matsumoto and
Ajmone Marsan, 1964a), IEDs are associated with a paroxysmal
depolarization shift of the resting neuronal membrane potential,
bursts of action potentials and ensuing hyperpolarization and hence
inhibition (Matsumoto and Ajmone Marsan, 1964b). In view of
this pronounced difference in the electrophysiological signature of
the two states, the CBF and CMRO2changes induced by interictal
discharges are expected to be far less conspicuous than their ictal
counterparts (Prevett et al., 1995; Engel et al., 1985; Theodore et
al., 1985). Likely due to the limited sensitivity of the methods in
combination with the sparcity of the interictal discharges in most
patients, there are few reports of metabolic and hemodynamic
changes induced by IEDs in epilepsy patients. No effect of the
spike and wave activity on the CMRGlc was observed in a PET
study of a group of generalized epilepsy patients, though there was
a slight trend toward CMRGlc increases in IGE patients (Ochs et al.,
1987). In a PET study of a reflex epilepsy patient, a 34.6% increase
in CBF and 13% increase in CMRGlc were measured in a region
concordant with the site of maximal ictal EEG abnormality, as
determined by implanted electrodes (Bittar et al., 1999). In a group
of patients with photosensitive epilepsy, a significant blood flow
increase was measured in the hypothalamus during a photo-
paroxysmal response (da Silva et al., 1999). In the caudate nucleus,
the CBF increase instigated by the intermittent photic stimulation
was abolished during the photoparoxysmal response (da Silva et
al., 1999).
Both widespread positive and negative BOLD responses to
IEDs have been reported in a number of EEG-fMRI studies (Salek-
Haddadi et al., 2003b; Archer et al., 2003a,b; Bagshaw et al., 2004;
Aghakhani et al., 2004). In the present study, we also observed
regions of both BOLD and CBF increases and decreases. Since
BOLD signal has a complex dependence on a number of
physiological parameters (Davis et al., 1998; Hoge et al., 1999),
we also estimated the corresponding oxygen consumption changes,
to obtain a more direct marker of the underlying metabolic costs.
Due to a combination of factors—the limited PASL contrast-to-
noise ratio at 1.5 T; few, transient IED events (compared to e.g.,
numerous repetitions of the block motor paradigm employed when
estimating the DCMRO2following neuronal activation/deactiva-
tion); and the stringent requirement for the overlap of BOLD/CBF
ROIs (to ensure robust, colocalized BOLD and CBF measure-
ments), the set of data available for quantification of oxygen
consumption changes in response to IEDs was severely curtailed.
A total of only 15 ROIs (showing statistically significant signal
changes)—from 2 patients—were sufficiently overlapped to allow
the estimation of the corresponding oxygen consumption. In these
2 case studies, BOLD changes were invariably accompanied by
CBF changes of the same sign, with the estimated DCMRO2/DCBF
coupling ratio of 0.48 T 0.17, thus very close to the one observed
for the functional activation in these patients as well as the one
obtained in healthy volunteers (Stefanovic et al., 2004). Nonethe-
less, the paucity of data available for this estimation, in com-
bination with the large variability of both epileptic syndromes and
the nature and dosage of medications customarily prescribed in its
treatment preclude any general conclusions about the hemody-
namic and metabolic responses to IEDs to be made from the
present results.
B. Stefanovic et al. / NeuroImage 28 (2005) 205–215 213
While these data suggest that IED-induced negative BOLD
responses may arise from the larger flow relative to oxygen
consumption decreases, as observed for motor task-induced
negative BOLD responses in healthy volunteers (Stefanovic et
al., 2004), other explanations of negative BOLD phenomena are
still possible. This is particularly true of epileptogenic zones in
focal epilepsies and responses to ictal activity, where neurovascular
coupling may well be compromised, as suggested earlier. Never-
theless, it is tempting to apply Gloor’s account of the spike and
wave phenomenon (Gloor, 1978), thus hypothesizing that the
presently measured negative CBF and BOLD responses result from
the net deactivation of the region due to a locally predominant
cortical inhibition relative to excitation. This also allows for the
existence of regions showing no BOLD response, due to a balance
between changes in local excitation and inhibition, integrated over
the interval determined by the effective BOLD temporal resolution,
as proposed earlier (Archer et al., 2003a). Finally, the metabolic
costs of IEDs and hence the ensuing CMRO2response and,
indirectly, BOLD response are likely affected by the relative
contributions of changes in synchronicity vs. synaptic activity to
the generation of IEDs (Salek-Haddadi et al., 2003b).
Conclusion
We observed normal hemodynamic responses to hypercapnic
perturbation in a group of epilepsy patients with generalized IEDs.
A consistent linear relationship between oxygen consumption and
perfusion changes during motor task performance in regions of
sustained positive as well as negative BOLD response was found.
The slope of the linear fit to CMRO2vs. CBF changes from both
ipsi- and contralateral ROIs was 0.46 T 0.05, in close agreement
with the coupling ratio found in an earlier study of healthy
volunteers. On the other hand, a decreased ratio of the magnitude
of contra- to ipsilateral flow changes was observed in the patient
group. Interictal epileptiform discharges produced a similar
coupling, with DCMRO2/DCBF of 0.48 T 0.17. The current
findings suggest a preserved coupling between metabolic and
hemodynamic processes underlying BOLD increases and
decreases in epileptic patients, in response to both normal
functional activation and IEDs and provide no evidence for a
disturbance in the interictal cerebral vascular responses in this
disorder.
Acknowledgments
This work was supported by the Natural Sciences and
Engineering Research Council of Canada and the Canadian