The baboons were sedated with an intramuscular injection of s-ketamine 5-6 mg/kg (KetaVed, Phoenix Scientific, St. Joseph, Missouri). They were removed from their squeeze cages and their scalp, shoulders and one leg were shaved. A catheter was placed in the saphenous vein. All animals received atropine 0.5 mg intramuscularly to reduce oropharyngeal secretions. EEG electrodes were affixed and the placement of soft restraints and facemask or Velcro straps fitted to limit head movement in the scanner, a continuous ketamine infusion (initially at 6 mg/kg/hr) was then started. The animals’ EEG, respiration and movement were monitored continuously. The ketamine infusion rate was adjusted by increments of 0.25-0.5 mg/kg/hr in order optimize sedation. The ketamine dose was lowered if a baboon demonstrated tonic motor activity or a suppressed background on EEG, or increased if the baboon demonstrated voluntary movements. The ketamine dose required for sedation was significantly higher in CTL than PS baboons.

Dynamic PET images were acquired on a CTI/Siemens HR+ scanner (Siemens, Munich, Germany; 63 contiguous slices, 2.4 mm thickness, 12 frames of 10 seconds each) in 3D mode. Transmission scans were performed for attenuation correction. Five CBF measurements were acquired in each animal. The first, third and fifth scans measured resting CBF, while ILS was delivered at 25 Hz during the second and fourth scans. The baboons received a bolus dose of 20 mCi ¹⁵O-labelled water (H₂¹⁵O) every 8-10 minutes. The total acquisition time for each scan was 120 seconds. Thirty-two activation and 48 activation scans were acquired for PS and CTL groups, over 80% being included in the analysis. The remaining scans were excluded either due to motion artifact or oversedation.

At the end of each PET session, the baboons were intubated and connected to a ventilator (Datex-Ohmeda Aestiva 5, Madison, Wisconsin, U.S.A.). The ketamine drip was discontinued, and the baboons received isoflurane at 1-1.5% inspiratory minimum alveolar concentration (MAC). The baboons were placed supine into a Siemens 3T Trio MRI scanner (Siemens, Munich, Germany) using an eight-channel receive-only primate head coil. The baboons were immobilized using soft restraints. Heart rate, pulse oximetry and respiratory rates were monitored. High-resolution (500µm isotropic), high gray-matter/white-matter contrast (~25%) and T1-weighted (~25 signal-to-noise) images were used to acquire six full-resolution volumes were acquired using a 3D TurboFlash sequence with an adiabatic inversion contrast pulse (TE/TR/TI = 3.04/2100/785 msec, flip angle=13 degrees). Motion correction was performed after the six volumes were aligned to control for head movements between acquisitions.

PET images were reconstructed to a matrix size of 128 x 128 x 75, using a 5 mm Hahn filter resulting in images with a spatial resolution of about 7 mm full-width at half-maximum (FWHM). For each injection of H₂¹⁵O, the time of bolus arrival in the brain was confirmed to be in the third frame (20-30 seconds). Therefore, the first two frames (20 seconds) of the data were excluded from further analysis. Frames 3-12 were corrected for intra-scan motion using MCFLIRT tool in FSL software (FMRIB, Oxford, UK) and averaged. All resting H₂¹⁵O scans were also corrected for inter-scan motion using MCFLIRT. The PET images were value-normalized to a whole-brain mean count of 1000. MRI scans of all baboons were aligned to a baboon MRI template. Each animal’s PET scan was co-registered with its own MRI using the Convex Hull algorithm [33]. Further analysis of PET and MRI data were performed once the images were processed to a common space (Fig. 1). Bonferroni correction was used to control for multiple comparisons.

Fig. (1). Coregistration of PET and MRI in the Baboon. Legend: Averaged H215O-PET images co-registered with a MRI template.

EEG recordings were performed during the PET scans to monitor the level of sedation, ictal and interictal epileptic discharges, occurring in resting state or during photoparoxysmal and photoconvulsive responses. EEG techniques have been described elsewhere [35]. The EEG recording was reviewed over the 90-second PET acquisition period for each animal, and the presence of IEDs and brief myoclonic seizures, predominantly consisting of eyelid myoclonus, were documented on the basis if muscle or blink artifacts. All of the EEG samples were reviewed by a single investigator (CAS) who was blinded to identity and prior EEG diagnosis of the baboons.

In attempt to compare the CBF changes between photosensitive (PS) and asymptomatic, control (CTL) baboons averaged resting scans were subtracted from averaged activation scans (during ILS at 25 Hz) in each individual, the before the subtraction scans were averaged for each group. The subtraction analyses demonstrated multiregional activations and deactivations in both PS and CTL groups, but with more frontotemporal cortical and subcortical activation in the PS group (Fig. 2). Frontally, the medial orbitofrontal and anterior cingulate gyri were activated in the PS group, as well as the medial and anterolateral temporal cortices. The activations were asymmetric, usually more prominent over the right hemisphere. Subcortically, the activations were noted in the striatum, particularly the globus pallidum, bilaterally, the pineal gland and in the left lateral geniculate nucleus. Deactivations were also widespread, engulfing the frontal, central and parietal cortices, particularly the left SMA. The cortical activations in the CTL group were localized to the occipital lobes, particularly superior to the external calcarine sulci bilaterally, the left posterior cingulate gyrus, angular and principal gyri, and subcortically to the right lateral geniculate nucleus and putamen. Deactivations were noted in the temporal lobes, with persistent activation of the amygdalae, orbitofrontal and medial frontoparietal regions, particularly in the right SMA. The inverted CBF in some brain regions, even the activation or deactivation of homologous cortical regions, suggest that corticocortical networks respond differently in PS and CTL baboons, even in the lateralization of CBF changes.

Absent activations of the occipital cortices in PS baboons contrasts BOLD fMRI findings and H₂¹⁵O-PET in photosensitive humans [29, 31]. Quantification of normalized CBF changes reflected increases in occipital CBF in PS baboons at rest compared to CTL animals, while CBF increases during ILS were similar. The net CBF change consequently did not achieve statistical significance. The regional activation of orbitofrontal cortices and diffuse frontoparietal hypoperfusion in PS baboons during ILS were also unexpected findings. Early electrophysiological studies using intracranial electrodes demonstrated bilateral activation during interictal and ictal epileptic discharges in the rolandic and prerolandic frontal cortices during the photoparoxysmal response, with a delayed involvement of the orbitofrontal cortices, only with sustained ictal discharges [7]. The absence of activation of the frontoparietal cortices during ILS was also noted in previous studies, and was thought to reflect cortical inhibition or CBF undershoot following the IEDs [29, 30]. The fact that CBF tends to decrease between generalized spike-and-wave discharges or absence seizures, reduces the sensitivity of detecting CBF increases associated with ictal discharges [38]. The activation of the orbitofrontal cortices in PS baboons, on the other hand, may reflect a lower baseline or resting CBF. Alternatively, fMRI studies of absence seizures in humans also demonstrated a sustained increase of orbitofrontal CBF before and after brief IEDs, lasting longer than in the lateral or medial frontal regions [39]. Whether this would be the case during the brief discharges associated with the photoparoxysmal response is unknown.

Subcortically, only the right thalamus was activated, ipsilateral to the larger occipital, orbitofrontal and temporal activations. While the thalami were activated during sustained ictal discharges associated with absence seizures [40-42], photoparoxysmal responses result in predominantly cortical activations as they propagate predominantly by cortico-cortical pathways not involving the thalamus, particularly in the setting of the brief generalized polyspike or spike-and-wave discharges associated with myoclonic seizures [31]. Furthermore, CBF was increased bilaterally in the internal globus pallidum and putamen in PS baboons, possibly reflecting the activation of nigrostriatal pathways in response to generalized seizures [43].

In order to better evaluate the differences in resting CBF between the PS and CTL baboons, we searched for ketamine or IED-related effects on CBF. Subtraction scans, comparing resting states of CTL and PS baboons, did not demonstrate any statistically significant differences in any brain regions. As CBF changes differed between PS and CTL baboons in a dose-dependent fashion, CBF was compared in a subset of the animals receiving the same, weight-adjusted, ketamine dose during the resting scans. This comparison demonstrated increased occipital CBF and decreased frontal CBF in the PS baboons. In healthy human subjects receiving increasing, subtherapeutic doses of ketamine, global CBF is increased, particularly in the frontal regions, with little to no change in the occipital regions [44]. This baseline activation of the occipital lobes which increases with ketamine dose increases has no obvious explanation. It is not likely to be driven by IEDs, as these are expressed mainly in the frontal regions, but rather by ketamine-induced cortical disinhibition. Nonetheless, while the CBF patterns correlating ketamine dose and IED rate differed substantially in the frontal lobes of PS baboons, occipitally the correlation analyses appeared to be concordant for both variables [37].

Furthermore, correlation analyses with resting IED rate demonstrated a symmetrical pattern of activation of the primary occipital cortices, lateral parieto-occipital regions, angular gyri and of the primary somatosensory cortices (Fig. 3). Subcortical structures are also activated, though less symmetrically, with the left thalamus showing increasing activation with IEDs. It is possible that the duration of the interictal or ictal discharges would be better correlated with bilateral thalamic CBF increases, but scalp EEG may be too unreliable to precisely quantify discharge duration. Nonetheless, correlation of resting CBF with IED rate may have reproduced at least part of the network responsible for the generation and propagation IEDs. Hence, in contrast to previous observations [19], it is likely that the parieto-occipital regions contain either the electrophysiological or biochemical framework to elicit and synchronize IEDs, even in the absence of ILS.

Fig. (3). Correlation of CBF with Rate of Interictal Epileptic Discharges in the Baboon.

One approach to identify connectivity of brain regions in PS and CTL baboons, is to evaluate CBF changes associated with altered parameters of stimulation. Variability of occipital CBF change was reported in healthy humans exposed to ILS, with CBF maximized at 8 Hz frequency, resonating at the same frequency of the awake posterior background [25]. Outside of the occipital lobes, CBF changes also appeared in the lateral and medial frontal lobes, temporal lobes and cingulate gyrus, which tended to decrease at increasing ILS frequencies, while CBF tended to increase in the visual association cortices [26]. A biphasic CBF change similar to the occipital lobes was only noted in the anterior cingulate region, but was maximal at 4 Hz.

In an ongoing study evaluating regional CBF changes due to ILS-stimulus frequency, preliminary findings in five PS and 3 CTL baboons demonstrated marked differences between them [45]. Averaged subtractions scans for three PS and CTL animals were compared at 5, 10, 15 and 25 Hz frequencies for particular regions of interest, including the occipital, posterior cingulate, orbitofrontal and motor cortices. Similar to humans, biphasic changes were noted in the occipital lobes, posterior cingulate cortex and motor cortices in the CTL baboons. CBF values, on the other hand, were maximal at higher frequencies in baboons than humans, namely 10-15 Hz (Fig. 4). While CBF maximized at ILS frequencies of 15-25 Hz in the PS baboons, the overall biphasic changes were maintained in the same brain regions. The only exceptions were the occipital lobes, where CBF increased progressively to 25 Hz. While these regions may also adhere to a biphasic response if ILS frequencies were to be increased further, progressive activation at increasing ILS frequencies may reflect decreased inhibition in PS baboons, particularly at frequencies that are most likely to trigger photoparoxysmal discharges. As this finding may be in part an artifact of ketamine’s pharmacodynamic effect, it is necessary to validate this finding with anesthetics with a different mode of action. It will be also important to test the CBF responses at higher ILS frequencies in order to confirm that 25 Hz is associated in maximal excitation in the PS baboon.

Fig. (4). ROI Analysis of CBF change with Variable ILS Frequency in the Baboon. Legend: L left, R right, V1 primary visual area, OF orbitofrontal cortex, PC posterior cingulate gyrus, M1 primary motor cortex. hatched columns pertain to photosensitive baboons.

Covariance analyses to correlate CBF changes with respect to a single region of interest. As the occipital lobe is the entrance point for visual stimulation, which reliably produced CBF increases in both CTL and PS, it provides an excellent region of reference for covariance analyses. The CBF changes that are reported outside of the occipital lobe in the setting of a covariance analysis, are secondary to occipital CBF increases. Surprisingly, the CTL baboons demonstrated asymmetrical, and mainly unilateral, activation patterns in brain regions outside of the occipital lobes, implying connectivity of the occipital lobes with brain regions in the same hemisphere. However, there was increased activation of brain regions with respect to left than right occipital lobe activation. Whether this asymmetry represents a particular ocular dominance is unclear, but it may merely imply a preferential processing of gross, high-frequency visual stimuli in the left hemisphere. The PS baboons, on the other hand, demonstrated widespread activation of both occipital lobes, and symmetrical activation of the parietal lobes and the sensorimotor cortices, regardless of the lateralization of the region of interest. From this analysis, and similar to the results of the correlations analysis of IEDs, it was apparent that synchronization of the photoparoxysmal response occurred at the level of the occipital lobes, with symmetrical coactivation of parietal association areas and sensorimotor cortices. While these regional CBF coactivations may not determine the areas where spontaneous or ILS-induced ictal or interictal epileptic discharges are generated, they do outline the brain regions that are likely nodes in a network subserving the propagation of the photoparoxysmal response along known anatomical pathways [20, 21]. This finding is in stark contrast with earlier electrophysiological studies demonstrating the generation of IEDs only at the level of the perirolandic (predominantly motor) cortices, with propagation to the medial and lateral frontal cortices [7]. These studies reported that ictal or interictal epileptic discharges appear occipitally only after their expression in the frontal lobes. It is unclear, whether the emphasis on the frontal lobes by these earlier studies was based upon a sampling bias limited by the use of depth electrodes or recording equipment, but the CBF changes identified by our group need to be verified electrophysiologically.