Lennox-Gastaut syndrome (LGS) is a severe epileptic disorder with characteristic electroclinical features but diverse etiologies. The shared electroclinical characteristics suggest that common cerebral networks are involved in generating seizures. We sought to reveal these networks by comparing ictal and interictal single-photon emission computed tomography (SPECT).
We identified 10 ictal–interictal SPECT pairs from seven patients with LGS (median age 11 years; range 1–38) who were studied during video electroencephalography (EEG)–confirmed tonic seizures. We performed a voxel-wise comparison of ictal and interictal SPECT studies across the group. The evolution of blood flow changes was explored by examining early and late injection groups.
Median duration of tonic seizures was 10 s (range 6–29 s), and injection latency from seizure offset was −8 to 48 s. In the early injection group (<10 s; three studies), there was hyperperfusion over pons and cerebellar hemispheres (p<0.05 cluster corrected family wise error), and hypoperfusion bilaterally over the pericentral region, with a trend toward hyperperfusion over bilateral superior and middle frontal gyri, and lateral parietal cortex. In the late injection group, there was hyperperfusion over midline and lateral cerebellar regions, with hypoperfusion widely over bilateral frontal regions.
This study suggests that the tonic seizures of LGS result from activity in a network, containing bilateral frontal and parietal association areas and the pons. We postulate that tonic seizures recruit the corticoreticular system, which connects frontal attentional areas to the pontine reticular formation, and is normally responsible for postural tone and orienting behavior.
Lennox-Gastaut syndrome (LGS) is a severe epilepsy syndrome, usually beginning in early childhood (Gastaut et al., 1966; Berg et al., 2010). Interictally, patients show frequent runs of diffuse 1.5–2.5 Hz slow spike-and-waves (SSWs), and bursts of low-voltage fast activity (paroxysmal fast activity [PFA]) predominantly during non–rapid eye movement (REM) sleep (Beaumanoir, 1985; Dulac & N'Guyen, 1993; Arzimanoglou et al., 2009). Cognitive impairment typically occurs in LGS, presumed to reflect disruption of brain development by frequent seizures and interictal discharges (Blume, 2004). Various cortical lesions are observed in 10–20% of patients with LGS (Bladin, 1985; Li et al., 1995; Goldsmith et al., 2000; Quarato et al., 2002; Freeman et al., 2003; Kondo et al., 2005; You et al., 2007), whereas a variety of chromosomal and genetic aberrations are also reported (Orrico et al., 2009; Yeung et al., 2009).
Tonic seizures are characteristic of LGS (Arzimanoglou et al., 2009), and they typically involve sudden neck and trunk flexion, facial grimace, proximal limb stiffening, and arm abduction (Dulac & N'Guyen, 1993). Distal limb muscles are relatively spared. However, tonic seizures can at times be clinically subtle, with only facial grimace or brief exhalation. The typical electroencephalography (EEG) pattern of tonic seizures is a high-amplitude slow wave followed by diffuse attenuation evolving to low-voltage fast activity (LVFA; 10–20 Hz), usually lasting 5–20 s (Fig. 1). This is typically followed by generalized, high-amplitude notched delta activity, similar to SSWs, lasting tens of seconds to minutes (Dulac & N'Guyen, 1993). Patients may be confused or agitated during the SSW phase.
Although the pathogenesis of LGS is not well understood, the shared electrical and clinical features provide insights into potential underlying mechanisms: (1) There are bilateral, diffuse EEG patterns seen during interictal and ictal discharges, suggesting widespread cerebral recruitment into the epileptic process of LGS; (2) LVFA is a feature of tonic seizures in LGS, and the filtering effects of scalp and meninges on scalp EEG mean that neural generators of these rhythms must be located near recording electrodes, implying significant involvement of widespread cortical regions (Tao et al., 2007); (3) the electroclinical presentation of LGS is similar despite a wide variety of underlying causes, including lesions in varying locations, suggesting that a common cerebral network becomes recruited; (4) some patients with LGS show resolution of seizures and interictal activity after resection of a causative cortical lesion (Fohlen et al., 2003; Lee et al., 2010), suggesting that cortex is playing an important role in initiating and maintaining these abnormal epileptic circuits; and (5) seizures involve axial muscle stiffening, tending to spare distal limb muscles, arguing against recruitment of corticospinal (pyramidal) pathways (Lemon et al., 2012).
Single-photon emission computed tomography (SPECT) is a nuclear medicine technique that images regional cerebral blood flow (CBF) to identify brain regions that are active during a seizure (Newton et al., 1992a,b). A technetium-labelled radiotracer is injected intravenously at the time of a seizure, capturing a snapshot of brain perfusion at the time (Catafau, 2001), with actual imaging performed 60–90 min later when the patient is stabilized. Comparison of interictal and periictal images across groups of subjects with the same epilepsy syndrome (Knowlton et al., 2004) can reveal involved epilepsy networks and patterns of seizure propagation, for example, in spontaneous and electroconvulsive therapy–induced generalized tonic–clonic seizures (Blumenfeld et al., 2003, 2009).
Aims and Hypotheses
The aim of this study was to reveal the cerebral networks involved during the tonic seizures of LGS. We hypothesized that SPECT perfusion changes during tonic seizures involve a network of frontal and parietal association cortices, similar to the areas observed with our EEG–functional magnetic resonance imaging (fMRI) studies of PFA (Pillay et al., 2013), an interictal discharge that shares a number of EEG features with tonic seizures in patients with LGS.
We identified patients with LGS who had ictal SPECT between 2001 and 2011 at Austin Health and The Royal Children's Hospital (RCH), Melbourne. The inclusion criteria were the following: (1) typical electroclinical features of LGS with confirmed tonic seizures on video-EEG monitoring (VEM; e.g., Fig. 1); (2) interictal EEG with SSW and generalized PFA; and (3) SPECT radiotracer injected during or immediately after a tonic seizure (Table 1).
Table 1. Electroclinical characteristics
Tonic seizure topography
EEG in tonic phase
Seizure duration (s)
Injection latency (s)
EEG during SPECT injection
Lt/Rt, left/right; F, frontal; P, parietal; T, temporal; O, occipital; CD, cortical dysplasia; sym/asym, symmetric/asymmetric; UL, upper limbs; HA slow, high amplitude slow wave; LVFA, low voltage fast activity or electrodecrement; GSSW, generalized slow sharp/spike-and-wave.
Ictal SPECT studies (n = 10 scans in seven subjects, apatient had two ictal SPECT studies). Injection latency = time (in seconds) from the offset of low voltage fast activity to injection of radiotracer (negative value = number of seconds before offset).
Radiotracer injection times were determined by review of VEM by two readers blinded to SPECT imaging results. Seizure onset was defined as the earliest EEG change or earliest clinical evidence of seizure activity, whereas seizure offset was defined as the end of LVFA, which usually correlated with the end of stiffening. SPECT injection time was defined as the moment when the plunger of the radiotracer syringe was fully depressed. Injection latency was defined as the time between seizure offset and injection. A negative value for injection latency indicates the radiotracer was injected prior to the offset of LVFA.
SPECT data acquisition and reconstruction
Periictal SPECT scans were performed as soon as practical after the injection, with some patients requiring general anesthesia. Interictal scans were performed when the patient had been seizure free for at least several hours. The SPECT scans at Austin Health (subjects 1, 2, and 3) were obtained with a triple head camera, whereas RCH scans were obtained with a single or dual head camera (see details Table S1).
Statistical parametric mapping was used to localize statistically significant periictal SPECT increases or decreases compared to interictal SPECT image on a voxel-by-voxel basis. Preprocessing and analyses were performed using Statistical Parametric Mapping (SPM8; http://www.fil.ion.ucl.ac.uk/spm/). Images were realigned, spatially normalized to Montreal Neurological Institute (MNI) space (specifically the standard SPECT template distributed with SPM8), saved with voxel size of 2 mm × 2 mm × 2 mm, and smoothed using a Gaussian kernel of 8 mm × 8 mm × 8 mm. Within-brain masking was undertaken using a robust two-step approach that involved application of an intensity-threshold determined iteratively such that the image is thresholded at half the mean intensity of the remaining voxels, as implemented in iBrain (Abbott & Jackson, 2001), followed by application of the Brain Extraction Tool (BET; Smith, 2002). Global intensity normalization was performed to correct for differences in total brain counts among scan pairs.
Group analyses of subjects' ictal–interictal subtractions were performed using two-tailed, unpaired Student's t-tests in SPM. We displayed the data at two thresholds. The first was an empiric threshold of p<0.02 uncorrected, with extent threshold (k) of 125 voxels (equivalent to a volume of 1 cm3), as published previously (Blumenfeld et al., 2003), to display trends in the data. The second, more rigorous threshold employed cluster-level family wise error (FWE) correction to display p<0.05 corrected for multiple comparisons. Group analyses were performed in all ictal–interictal SPECT data (pooled data analysis) and in subgroups using different injection latency as a cutoff point to explore the spatial evolution of blood flow changes over time (subgroup analysis).
This study was retrospective, with SPECT and MRI scans already acquired for clinical indications. The project had approval by the Human Research Ethics Committees of both Austin Health and the RCH.
Ten ictal–interictal SPECT pairs from seven LGS patients were analyzed. Three patients had two periictal SPECT scans (see Table 1). Because of the small number of available studies, we included all scans in a fixed-effects analysis, restricting inference to the group that we studied. The median age was 11 years (range 1–38 years; four patients were male). All had symmetric or asymmetric axial tonic seizures. Five subjects had focal structural abnormalities on brain MRI, probably reflecting their consideration for epilepsy surgery. Mean tonic seizure duration was 10 s (range 6–29 s) and injection latency was −8 to +48 s.
Across the group (n = 10 scans), tonic seizures were associated with increases in CBF over the lateral parietal lobe and cerebellum, and reduced CBF bilaterally in frontal and occipital regions (Fig. S1), with cerebellar increases and mesial frontal and occipital decreases in perfusion surviving the more stringent statistical threshold (p<0.05 cluster corrected FWE). Subgroup analysis by injection latency suggested evolution of CBF between early (<10 s) and late (>10 s postictal) injections.
In the early injection group (n = 3), ictal hyperperfusion was observed over bilateral superior and middle frontal gyri and lateral parietal cortices, with hypoperfusion seen over primary cortical areas, including bilateral pericentral and mesial occipital regions (Fig. 2A). Ictal hyperperfusion in pons and cerebellum, and hypoperfusion in the pericentral and occipital regions remained significant after correction for multiple comparisons (p<0.05 cluster corrected FWE).
In the late injection group (n = 7), hyperperfusion was observed bilaterally over the lateral parietal cortex and diffusely in the cerebellum, whereas the pons no longer showed change (Fig. 2B). Late ictal hyperperfusion of upper midline cerebellar structures, and cerebellar hemispheres survived significance testing at the more rigorous threshold. There was significantly reduced perfusion seen widely over both frontal lobes. Statistical comparison of the two groups revealed that CBF was greater in the pons early, and greater in anterior medial cerebellum later (p<0.05 FWE; Fig. 3).
This SPECT study suggests that there is a common pattern of CBF change during tonic seizures of LGS. There is early increased perfusion in the pons, cerebellum, and bilateral frontoparietal association cortices, switching to a pattern of hypoperfusion of the bilateral frontal association cortex postictally. Despite prominent muscle contraction, there appears to be reduced CBF in primary motor cortex. The early ictal SPECT changes mirror those we have seen in EEG-fMRI of PFA in LGS (Pillay et al., 2013). This study suggests that the distinctive electroclinical phenomenology of tonic seizures reflects activity in a corticopontine pathway.
This is a small study, and larger numbers are needed to confirm these initial observations. Recruiting large numbers is difficult, because tonic seizures tend to be brief, and vigorous movement in the early phase can make early injection difficult. Furthermore, patients with LGS are often not considered as surgical candidates, meaning they are unlikely to be referred for ictal SPECT. From a statistical standpoint, our inferences are limited to the small group that we have studied. The group includes a high proportion of subjects with LGS associated with a lesion, which may have influenced results, and SPECT images were not all on the same scanner. However, the striking similarity between these ictal SPECT changes in tonic seizures, and the EEG-fMRI activation maps of interictal epileptiform activity in LGS (Pillay et al., 2013) lead us to believe these observations likely reflect underlying neuronal network behavior in LGS patients more generally.
Although the corticospinal pathway is the main pathway for voluntary movement, there is good evidence of an additional pathway contributing to axial motor control, responsible for maintenance of posture and orienting behaviors. Retrograde tracer studies in cats (Matsuyama et al., 2004), and lesional studies in cats (Rossi & Brodal, 1956) and primates (Kuypers et al., 1962), have revealed a descending corticoreticular pathway from premotor regions (Brodmann area 6), to the reticular formation of the brainstem, located in the pons. This pathway descends through corona radiata and the posterior limb of internal capsule, and then passes through midbrain tegmentum and terminates at the pontomedullary reticular formation. MRI tractography studies have identified this pathway in humans (Yeo et al., 2012). The ictal SPECT hyperperfusion in premotor cortices and pons we observed would be consistent with activity in the corticoreticular pathway. Outflow from the pons is likely to be by way of the reticulospinal pathway, which originates from the pontomedullary reticular formation and runs through ventromedial subcortical spinal system (Kuypers et al., 1962; Lemon et al., 2012). The reticulospinal pathway innervates axial and proximal limb muscles, and stimulation of the brainstem reticular formation in animals produces motor attacks that show similarity to the tonic seizures of LGS (Burnham, 1987).
In the cerebral cortex, tonic seizures showed increased CBF in premotor, prefrontal, and parietal regions, in a distribution similar to that we have observed with EEG-fMRI of generalized PFA (Pillay et al., 2013). The distribution of this is consistent with involvement of the frontal and parietal attentional network. A range of functional imaging studies have revealed there is an “attentional network” of frontal (anterior cingulate, bilateral dorsolateral prefrontal cortices) and parietal (intraparietal sulcus and lateral parietal lobe) areas whose levels of neural activity rise and fall together (Fox et al., 2005; Toro et al., 2008). This distributed network is recruited during a range of cognitive tasks, and shows reduced activity during the resting state. It appears that tonic seizures are associated with increased activity in this attentional network. Increased SPECT radiotracer uptake over these association cortices has been observed in generalized tonic–clonic seizures induced by electroconvulsive therapy (Blumenfeld et al., 2003). In contrast, frontoparietal hypoperfusion was seen during the ictal and postictal phase of spontaneous secondarily generalized tonic–clonic seizures (Blumenfeld et al., 2009).
Tonic seizures tend to be brief, making it difficult to get true ictal injections. We were primarily interested in understanding the cerebral networks responsible for the tonic phase of seizures in LGS, when the EEG shows LVFA. Because seizure duration is variable, we chose to compare injections during or shortly after the offset of fast activity to later injections, to ensure injections captured CBF changes associated with LVFA. Detailed electrophysiologic and imaging studies, undertaken to better understand the basis of the fMRI signal, show blood flow remains elevated for around 10 s after brief neural events (Logothetis & Wandell, 2004). Therefore, <10 s postictal injections are likely to be capturing the neural networks of LVFA, whereas the later injections are largely capturing the early “postictal” phase, dominated by SSWs.
Our findings suggest that slow spike and wave is associated with reduced blood flow in frontal association cortex, mirroring our EEG-fMRI findings of SSWs in LGS (Pillay et al., 2013). However, we cannot exclude the possibility that the early “postictal” hypoperfusion in the frontal lobe we observe is the result of differences in neurovascular coupling in the frontal lobe, rather than true differences in neuronal behavior across the brain. We note that prior SPECT studies of focal seizures have shown rapid switching in the frontal lobe from ictal to postictal perfusion patterns (Harvey et al., 1993).
We observed cerebellar hyperperfusion in the early and late injections. Cerebellar hemispheres showed hyperperfusion early, with upper midline cerebellar structures developing hyperperfusion later. Cerebellar hyperperfusion has been reported previously in spontaneous secondarily generalized seizures (Blumenfeld et al., 2009), with cerebellar hemisphere involvement becoming more prominent later in the seizure, associated with reduced perfusion in frontoparietal association cortices. Cerebellar hyperperfusion has also been reported in focal frontal lobe epilepsy, with associated hyperperfusion in the motor cortical regions (Harvey et al., 1993), as opposed to the pericentral hypoperfusion we observe. We cannot know whether the increased cerebellar perfusion we observe in tonic seizures is primary or secondary to the cortical, pontine, and other changes we observe. However, given that the electroclinical features of LGS suggest an important role for the cortex in seizure generation, we suspect the cerebellar perfusion changes are in response to ictal activity in the cortex.
We postulate that the tonic seizures in LGS arise in bilateral frontal and parietal association cortices, which are linked as an intrinsic part of the attentional system. Epileptic activity is rapidly amplified by this system, and then projects via corticoreticular pathways to the pontine reticular formation, and from there via reticulospinal pathways to spinal motor neurons. This would explain why tonic seizures of LGS share similar, axial motor predominant clinical features, despite etiologies that include cortical lesions of varied type and location. Corticoreticular involvement would also be consistent with the curious phenomenon observed in some LGS patients of startle-induced tonic seizures (Manford et al., 1996), as startle would heavily drive these attentional pathways. We believe these concepts provide a framework to understand the otherwise confusing electroclinical features of LGS. More specifically, these findings help with interpretation of bilateral or apparently diffuse SPECT changes in tonic seizures.
The nuclear medicine departments of Austin Health and Royal Children's Hospital Melbourne provided access to the imaging data. Mr Aaron Warren assisted with manuscript preparation. Professor Philip Thompson provided neuroanatomic insights. Funding support was provided by the National Health and Medical Research Council of Australia (NHMRC – project grant ID no. 628725), and the Operational Infrastructure Support Program of the State Government of Victoria, Australia.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
Utcharee Intusoma is Assistant Professor of Pediatric Neurology, Prince of Songkla University, Thailand.