- Email: [email protected]
Spectral analysis of EEG gamma rhythms associated with tonic seizures in Lennox–Gastaut syndrome
Epilepsy Research (2009) 86, 15—22
journal homepage: www.elsevier.com/locate/epilepsyres
Spectral analysis of EEG gamma rhythms associated with tonic seizures in Lennox—Gastaut syndrome Katsuhiro Kobayashi ∗, Takushi Inoue, Yoshiaki Watanabe, Makio Oka, Fumika Endoh, Harumi Yoshinaga, Yoko Ohtsuka Department of Child Neurology, Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, and Okayama University Hospital, Shikatacho 2-chome 5-1, Okayama 700-8558, Japan Received 7 September 2008; received in revised form 10 March 2009; accepted 15 March 2009 Available online 16 April 2009
KEYWORDS Lennox—Gastaut syndrome; Tonic seizure; Ictal EEG; Time—frequency analysis; High-frequency oscillation; Gamma rhythm
Summary Purpose: EEG gamma rhythms, which are found in association with epileptic spasms in infants with West syndrome, were explored in the ictal EEGs of tonic seizures in older patients with Lennox—Gastaut syndrome (LGS) to investigate the pathophysiology of the disease. Methods: The subjects were 20 patients with LGS (11 males, 9 females; age range: 3 years 1 month to 29 years 3 months) who had at least one digitally recorded tonic seizure with minimal artifacts. A time—frequency analysis was applied to each patient’s ictal EEG data. Results: A total of 54 seizures were analyzed, excluding spasms in clusters. The ictal EEGs of the tonic seizures showed only diffuse desynchronization in 10 seizures, and desynchronization followed by rhythmic activity in 21. The ictal discharges started as rhythmic activity of varying amplitude without initial desynchronization in 23 seizures. In a total of 25 seizures from 13 patients, gamma rhythms with frequencies ranging from 43 to 101.6 Hz were detected by temporal expansion of the ictal EEG traces and spectral analysis. In 24 (96%) of these seizures, gamma rhythms were observed at seizure onset corresponding to visually identified desynchronization. In the remaining seizure, gamma rhythms were found in association with transient suppression of high-amplitude rapid discharges. Conclusion: The detection of gamma rhythms in the ictal EEGs of tonic seizures indicated that some tonic seizures might have generative mechanisms in common with epileptic spasms, and that these mechanisms are possibly related to desynchronization at seizure onset. © 2009 Elsevier B.V. All rights reserved.
Introduction
∗ Corresponding author. Tel.: +81 86 235 7372; fax: +81 86 235 7377. E-mail address: k [email protected] (K. Kobayashi).
Tonic seizures are the cardinal seizure type in Lennox—Gastaut syndrome (LGS) and in related symptomatic generalized epilepsy (SGE) (Beaumanoir and Blume, 2005; Ohtahara, 1988). Their resistance to treat-
0920-1211/$ — see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.eplepsyres.2009.03.011
4 (20.0) 5 (25.0) 20 [11/9] Total
Percentages are shown in parentheses. WS, West syndrome; sz, seizure.
2 (10.0)
7 [4/3] Patients without ictal gamma rhythms
2 months to 7 years 3 months [1 year 6 months]/3 years 1 months to 29 years 3 months [11 years 7 months]
13 (65.0)
3 (15.0)
1 (14.3) 0 (0) 0 (0) 2 (28.6) 3 (42.9)
3 (23.1) 5 (38.5) 3 (23.1) 0 (0) 10 (76.9) 13 [7/6] Patients with ictal gamma rhythms
2 months to 4 years 8 months [1 year 5 months]/3 years 1 months to 25 years [9 years 8 months] 2 months to 7 years 3 months [1 year 10 months]/4 years 11 months to 29 years 3 months [15 years]
Focal sz Myoclonic sz Atypical absence Atonic sz Epileptic spasms in series
Seizure type (other than tonic) Age at sz onset/examination (mean) Number of patients (M/F)
Table 1
Patient characteristics.
Patients
Patient group
Patients and methods
The subjects of the present study were 20 patients with a diagnosis of LGS (11 males, 9 females; age range: 3 years 1 month to 29 years 3 months) who had at least one tonic seizure with minimal artifacts digitally recorded between November 2003 and January 2008 (Table 1). During this period, isolated tonic seizures were recorded in a total of 30 patients with LGS who were at least 3 years of age at the time of examination, but the ictal EEG data of the remaining 10 patients could not be used for the present analysis because of contaminating massive muscle activity and/or artifacts. We used the following diagnostic criteria of LGS: tonic seizures (especially tonic spasms or brief tonic seizures) and/or multiple seizure types consisting of atypical absences and others, along with diffuse slow spike-waves (DSSW) in the interictal EEG (Ohtahara, 1988). All patients were mentally retarded. They all showed highamplitude 8—14 Hz rapid rhythms in sleep EEG. The underlying disorders were tuberous sclerosis in three patients, cortical malformation including lissencephaly in three, sequel of acute encephalopathy or meningitis in three, sequel of premature birth with intraventricular hemorrhage or periventricular leukomalacia in two, cerebral atrophy of unknown origin in two, sequel of severe neonatal asphyxia in one, multiple malformations in one, an arach-
9 (69.2)
History of WS
ment poses a serious clinical problem in epileptology. LGS is regarded as the most mature form of age-dependent epileptic encephalopathy (Ohtahara, 1978; Ohtahara and Yamatogi, 1990), and the evolution of LGS from West syndrome (WS) suggests a common pathophysiology shared by the two syndromes. High-frequency activity of more than 60—80 Hz can be invasively recorded from epileptogenic brain regions, and the study of this activity type is now attracting attention (Fisher et al., 1992; Traub et al., 2001; Rampp and Stefan, 2006). Gamma rhythms in the frequency range of 50—100 Hz are found in the scalp ictal EEGs of epileptic spasms in patients with WS (Kobayashi et al., 2004) and in early infantile epileptic encephalopathies such as Ohtahara syndrome and early myoclonic encephalopathy (Kobayashi et al., 2007). These EEG gamma rhythms are regular and morphologically distinct from irregular muscle activity in temporally expanded EEG traces. The ictal gamma rhythms appear to be closely related to the generative mechanisms of spasms (Inoue et al., 2008). Observation of high-frequency rhythms in the gamma and faster bands in ictal electrocorticography of epileptic spasms, and successful suppression of spasms by the resection of cortical regions showing these high-frequency rhythms, strengthen the possibility of a relationship between cortical high-frequency rhythms and the generative mechanisms of spasms (Akiyama et al., 2005; Ochi et al., 2007; RamachandranNair et al., 2008). Based on the age-dependent transition of WS to LGS, we speculated that tonic seizures in older patients, like spasms in WS, may also have generative mechanisms linked to gamma rhythms. Therefore, we set out to investigate the ictal EEG patterns associated with tonic seizures using time—frequency spectral analysis, as in our previous study on spasms (Kobayashi et al., 2004), to explore gamma rhythms in patients with LGS.
13 (65.0)
K. Kobayashi et al.
4 (57.1)
16
Spectral analysis of EEG gamma rhythms
17
noid cyst in one, and unknown with no detectable brain lesion by MRI in the remaining four. Informed consent for this study was obtained from the parents of the patients.
EEG recording and frequency analysis EEGs were recorded at a digital sampling frequency of 500 Hz with a Nihon Kohden Neurofax (Tokyo, Japan), which used a bandpass filter ranging from 0.016 to 300 Hz before digital sampling. The international 10—20 electrode placement system was used. We simultaneously recorded video and/or electromyograms (EMGs) from the bilateral deltoid muscles in most patients. The conventional EEG traces (10—20 s per page) were initially reviewed to identify the ictal events using a time constant of 0.1 s. The traces were then temporally expanded (2 s per page) with a time constant of 0.03 s (corresponding to a high-pass filter of 5.3 Hz) in order to study the details of activity faster than 40 Hz (for example, see Fig. 1B and C). We investigated the time evolution of the high-frequency power spectrum of the ictal activity by applying the Gabor transform (a Fourier transform with a 200 ms wide Gaussian window) to the original EEG data, as in our previous study regarding the derivation pairs of F3-C3, F4-C4, P3-O1 and P4-O2 (Kobayashi et al., 2004; see Fig. 2). Gamma peaks were identified as clearly visible spectral spots with frequencies faster than 40 Hz that were surrounded by an area of low power and that corresponded to rhythmic EEG activity. The width of each spectral segment was 5 s, and the frequency range was 10—120 Hz. Computation was performed using a program written in-house for MATLAB (version 6.5.1; MathWorks Inc., Natick, MA). In each time—frequency spectrum, peak frequency and peak power of the gamma rhythms were measured for statistical analysis. Differences in peak frequencies among the derivations of F3-C3, F4-C4, P3-O1 and P4-O2 were evaluated by one-way analysis of variance (ANOVA). Because peak power varied greatly across patients, the pattern of distribution of the gamma rhythms over the scalp was evaluated separately for each individual. The distribution pattern was somewhat arbitrarily defined as fronto-central dominant in each seizure wherein average peak power in the bilateral frontocentral regions (F3-C3 and F4-C4) was more than twice that in the bilateral parieto-occipital regions (P3-O1 and P4-O2), and it was defined as parieto-occipital dominant when the above situation was reversed. When the average peak power was not more than twice as great in one region than in the other, the pattern of distribution of the gamma rhythms was regarded as showing no clear dominance. When there were multiple gamma peaks in an individual spectrum, the peak with the greatest power was selected for statistical analysis (Inoue et al., 2008). We used an unpaired t-test to look for relationships between the presence of ictal gamma rhythms in at least one seizure and (1) patient age at seizure onset, and (2) patient age at examination (the first examination in cases of multiple examinations). We used Fisher’s exact test to look for a relationship between the presence of ictal gamma rhythms and (1) the presence of various other types of seizures, and (2) a history of WS. Relationships were considered statistically significant if p < 0.05.
Results A total of 121 tonic seizures were recorded from 30 patients; of these, 54 seizures with minimum artifacts from 20 patients were analyzed as described above. Epileptic spasms in clusters were excluded from the study. The observed ictal EEG patterns of tonic seizures were as follows (see Table 2). Diffuse amplitude attenuation
Figure 1 Ictal EEG of a brief tonic seizure (the corresponding time—frequency spectra are shown in Fig. 2). In (A) and (B) ‘‘Delt.’’ refers to the deltoid muscle. (A) A conventional trace showing diffuse desynchronization associated with a brief tonic seizure (arrow). (B) A temporally expanded trace showing diffuse gamma rhythms with regular sinusoidal morphology in the frequency range of 60—80 Hz at seizure onset (horizontal lines). Muscle activity contamination in the temporal and frontopolar channels is very irregular and morphologically different from the ictal gamma rhythms. (C) A magnification of two traces indicated by horizontal lines in part (B). The point of the arrow in the conventional trace corresponds to that of the arrowhead in the expanded trace.
(desynchronization) was found in 10 seizures from four patients (Fig. 1). Desynchronization was succeeded by rhythmic beta activity partly mixed with alpha and/or theta activity in 21 seizures from 11 patients (Fig. 3). Desynchronization had a leading spike in 13 seizures from eight
18
Table 2
Ictal EEG observations of tonic seizures and their associated gamma rhythms.
Visual ictal EEG pattern
Diffuse desynchronization Diffuse desynchronization and subsequent rhythmic activity Rhythmic activity with low-amplitude onset Rhythmic activity with medium-amplitude onset High-amplitude rapid discharges Rapid discharges evolving to desynchronization and subsequent rhythmic activity Total
Number of seizures analyzed (waking/sleep)
Mean seizure duration (mean ± SD; s)
Gamma rhythms Peak frequency of gamma rhythms detected (mean ± SD; Hz) in F3-C3/F4-C4/P3-O1/P4-O2
10 [8/2]
3.7 ± 1.3
10 (100)
21 [12/9]
9.4 ± 5.8
14 (66.7)
66.0 ± 14.9/63.3 ± 14.4/71.8 ± 13.0//74.5 ± 15.8 77.3 ± 9.9/77.4 ± 11.7/74.4 ± 10.8/75.6 ± 10.4
Distribution of gamma rhythms No dominance Fronto-central Parieto-occipital detected dominant dominant 7
1
2
10
3
1
4 [0/4]
10.3 ± 4.0
0 (0)
N/A
N/A
N/A
N/A
8 [1/7]
6.4 ± 2.5
0 (0)
N/A
N/A
N/A
N/A
10 [0/10]
6.6 ± 3.3
0 (0)
N/A
N/A
N/A
N/A
1 (100)
43.0/43.0/46.9/43.0
1
0
0
25 (46.3)
71.4 ± 14.2/70.4 ± 15.3/72.3 ± 12.5/73.8 ± 13.7
18 (72.0)
4 (16.0)
3 (12.0)
1 [0/1]
54 [21/33]
40
8.0 ± 6.4
Percentages are shown in parentheses. SD, standard deviation.
K. Kobayashi et al.
Spectral analysis of EEG gamma rhythms
19
Figure 2 Time—frequency spectra of ictal EEG activity (the corresponding EEG is shown in Fig. 1). Spectral peaks of the ictal gamma rhythms stand out as isolated spots (arrows). The gamma rhythms slow down gradually from 60 to 80 Hz at onset to 40—60 Hz at the end. The spectra of muscle activity recorded from the bilateral deltoid muscles (Delt) are noisy and include a wide range of frequency components—–spectral patterns that are different from those of the EEG gamma rhythms. The onset of spectral peaks from EEG gamma rhythms (vertical red lines) precedes the onset of deltoid muscle activity. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
patients. Rhythmic discharges started insidiously as lowamplitude (≤50 V) beta activity without desynchronization and evolved to include alpha/beta activity in four seizures from four patients, and rhythmic discharges began as medium-amplitude (50—150 V) activity in eight seizures from five patients. Bursts of high-amplitude (≥200 V) 8—14 Hz rapid discharges including some beta activity were observed in ten seizures from three patients. Rapid discharges evolved to show diffuse amplitude attenuation and subsequent rhythmic beta/alpha activity in the remaining seizure. Of all 54 seizures, 21 (38.9%) occurred during wakefulness, and the remaining 33 (61.1%) during sleep. The majority (8/10, 80.0%) of seizures with only desynchronization occurred during wakefulness. In contrast, most (7/8, 87.5%) of the seizures with medium-amplitude onset activity and all (10/10) of the seizures with high-amplitude rapid discharges occurred during sleep. Gamma rhythms with peak frequencies ranging from 43.0 to 101.6 Hz were detected in 25 of the 54 seizures studied (46.3%); these seizures occurred in 13 out of the 20 patients (see Tables 1 and 2). In 24 (96.0%) of these seizures, ictal gamma rhythms were observed at seizure onset, corresponding to the visually identified desynchronization in EEG. Gamma rhythms were found in all of the ten seizures showing only desynchroniza-
tion (Figs. 1 and 2). They were observed in 14 (66.7%) of the 21 seizures with onset desynchronization and subsequent rhythmic activity (Figs. 3 and 4). Frequency of the ictal activity tended to run down from the gamma band to the beta band with the progression of the seizure (Figs. 2 and 4). In the single seizure that started with high-amplitude rapid discharges and subsequently evolved to amplitude attenuation and then to rhythmic beta/alpha activity, gamma rhythms were detected at the time of transient suppression of rapid discharges. In this exceptional seizure, however, the peak frequencies of the gamma rhythms were relatively low (43.0—46.9 Hz) and close to the beta band. In the remaining 29 seizures, the ictal activity lacked clear gamma peaks in the temporally expanded EEG traces and the time—frequency spectra. In all cases where ictal gamma rhythms were detected, they were regular in morphology, as illustrated in the magnified traces (Figs. 1C and 3C), and thus discernible from muscle activity contamination with irregular morphology in the EEG. The spectra of the ictal gamma rhythms showed corresponding well-defined peaks in the gamma band, in contrast to the broad and noisy spectra of muscle activity. This finding agrees well with previous studies (Kobayashi et al., 2004, 2007).
20
K. Kobayashi et al. ber of electrode channels used in this study, we could not detect a relationship between peak frequency or power and spatial distribution of gamma rhythms associated with tonic seizures. Finally, the patients were divided into two groups based on whether gamma rhythms were detected in at least one seizure (see Table 1). The presence of ictal gamma rhythms was not statistically related to age at seizure onset (p = 0.630), age at examination (p = 0.152), or the history of WS (p = 0.651). In addition, no statistical relationship was found between the presence of ictal gamma rhythms and the presence of other types of seizures, including epileptic spasms in clusters (p = 0.174), atonic seizures (p = 0.111), atypical absences (p = 0.521), myoclonic seizures (p = 0.114), or focal seizures (p = 1.000).
Discussion
Figure 3 Ictal EEG of a tonic seizure (the corresponding time—frequency spectra are shown in Fig. 4). (A) A conventional trace showing diffuse desynchronization (arrow) and subsequent rhythmic activity associated with a tonic seizure. (B) A temporally expanded trace showing diffuse gamma rhythms in the frequency range of 60—100 Hz at seizure onset (horizontal lines). Muscle activity contamination in the bilateral temporal channels is very irregular and morphologically different from the ictal gamma rhythms. (C) A magnification of two traces indicated by horizontal lines in part (B). The point of the arrow in the conventional trace corresponds to that of the arrowhead in the expanded trace.
In addition to identifying gamma rhythms in the EEG spectra, we evaluated their characteristics across different regions of the scalp. The peak frequencies of the gamma rhythms did not significantly differ between the bilateral fronto-central and parieto-occipital regions (p = 0.857). In terms of power, the gamma rhythms were fronto-central dominant in 16.0% of cases and parieto-occipital dominant in 12.0%; the remaining 72.0% of seizures with gamma rhythms showed no clear dominance. Thus, with the limited num-
Gamma rhythms were detected in scalp EEGs in association with at least 46.3% of tonic seizures, and were particularly associated with tonic seizures exhibiting onset desynchronization. The presence of small amount of gamma rhythms in the remaining tonic seizures cannot be precluded because they exhibited spectral spreading from beta bands with occasional noise contamination, and it is possible that some gamma activity occurred but failed to form recognizable peaks. The seizures analyzed in the present study were axial or axiorhizomelic tonic (Gastaut and Broughton, 1972), as spectral analysis was difficult to perform on global tonic seizures that were accompanied by violent movements with massive muscle activity and/or artifacts in the EEGs. We identified ictal gamma rhythms as seizure-associated gamma rhythms only when they exhibited both clear EEG morphology and corresponding spectral patterns. Muscle activity contamination generally has irregular morphology and is recorded at electrodes close to the muscles (Kobayashi et al., 2004). In accordance with our prior studies, Otsubo et al. (2008) reported that high-frequency activity of muscular origin that contaminated intracranial EEGs was randomly scattered without a specific frequency band in spectral analysis. Therefore, ictal gamma rhythms can be differentiated from muscle activity contamination. They are also different from 60 Hz alternating current (AC) artifacts, because AC artifacts have monotonous morphology with a frequency exactly at 60 Hz (Kobayashi et al., 2004). It is hard to detect cortical activity of very high frequency by scalp EEG because of attenuation of higher frequencies between cortex and scalp (Nunez and Srinivasan, 2006). The frequency of cortical oscillations associated with epileptic spasms reaches 250 Hz (Ochi et al., 2007), and the ictal gamma rhythms recorded over the scalp may be only the relatively slower components of the ictal cortical activity. Intracranial EEG is ideal for the investigation of highfrequency activity, but it would have been unethical to use on the patients of the present study, who lacked findings suggesting that their tonic seizures were actually secondary generalizations from a cortical focus. The ictal gamma rhythm is posterior dominant in epileptic spasms (Kobayashi et al., 2004), corresponding to the
Spectral analysis of EEG gamma rhythms
21
Figure 4 Time—frequency spectra of ictal EEG activity (the corresponding EEG is shown in Fig. 3). Spectral peaks of the ictal gamma rhythms are indicated by arrows. The gamma rhythms slow down gradually from 60 to 100 Hz at seizure onset to 40—60 Hz with progression of seizure discharges.
posterior predominant distribution of abnormalities that is seen in West syndrome (Oka et al., 2004). In LGS, the distribution of EEG abnormalities, such as DSSW, is generally diffuse (Beaumanoir and Blume, 2005). We could not detect regional difference of the ictal gamma rhythms in the tonic seizures analyzed with the limited number of electrode channels used in this study. The desynchronization that was observed alone or at the onset of longer tonic seizures in the EEGs resembled the ictal activity of epileptic spasms in WS in terms of both morphology and association with gamma rhythms (Kobayashi et al., 2005). The desynchronization pattern actually included slow potential changes with superposition of low-amplitude activity (Figs. 1 and 3), and it mimicked the ictal slow waves of epileptic spasms. Seizures resembling epileptic spasms have been reported in LGS, and these seizures cause the patients to drop (‘‘axial spasms’’ in Egli et al., 1985; ‘‘flexor spasms’’ in Ikeno et al., 1985). We could not tell if the seizures analyzed in the present study caused the patients to drop because they were supine during the EEG recording; an additional study would be necessary to resolve this question. The three well-known ictal EEG patterns of tonic seizures are (1) desynchronization, (2) beta activity at 20 ± 5 Hz that is initially of low voltage and progressively increases in amplitude, and (3) a slower high-amplitude recruiting rhythm at about 10 Hz (Gastaut and Broughton, 1972; Yamatogi and Ohtahara, 1981). In this study, the ictal EEG patterns were subdivided for the purpose of analysis (Table 2); each seizure displayed one or more of the patterns listed above. There appeared to be a spectrum of ictal EEG patterns. At one end is desynchronization that tends to occur diurnally more often than nocturnally. In these seizures, gamma rhythms are often superimposed on desynchronization. At the other end of the spectrum lies a pattern of high-amplitude rapid discharges with consistently nocturnal occurrence. Rhythmic beta activity with varying amplitude
falls somewhere in between. In intracranial EEGs of epileptic patients, the background gamma activity is greater in the waking state than in sleep (Gross and Gotman, 1999). Occurrence of gamma rhythms in diurnal seizures might be facilitated by a general tendency of the neuronal system to generate more gamma activity during the waking state, though its precise mechanisms remain to be elucidated. It is known that seizures with intense tonicity are often associated with desynchronization in EEG, while seizures with weaker tonicity tend to show high-amplitude activity (Gastaut et al., 1963; Miyakoshi et al., 1977). The decrease in spectral frequency of the ictal rhythms from the gamma band to the beta band was found to be associated with the progression of seizure discharges from desynchronization to rhythmic beta/alpha/theta activity. The frequency of ictal activity is known to decrease from the beta band to about 10 Hz in some tonic seizures (Gastaut and Broughton, 1972). In an intracerebral EEG study, fast seizure activity in the 60—90 Hz band was associated with very low spatial correlation among cerebral regions, indicating a functional decoupling of brain sites at seizure onset (Wendling et al., 2003). This functional decoupling was followed by an abnormally high re-coupling when the seizure developed into activity of decreasing frequency and increasing amplitude. A phenomenon similar to this functional decoupling and re-coupling among brain regions may take place during the progression of tonic seizures with activity changes from the gamma band to the slower bands. In generalized corticoreticular epilepsy, a complex and abnormal interaction is thought to exist among the cortex, the thalamus, and the brainstem (Gloor, 1979). Tonic seizures are generally believed to originate from the brainstem (Beaumanoir and Blume, 2005). We have previously suggested that ictal gamma rhythms in EEG represent the cortical component of the abnormal process of seizure generation (Kobayashi et al., 2004). The detection of gamma rhythms in association with tonic seizures, and especially
22 in association with onset desynchronization, might indicate that the cortex is involved in the initial phase of the generation process of at least some tonic seizures, and that this process involves the functional decoupling of cerebral regions.
References Akiyama, T., Otsubo, H., Ochi, A., Ishiguro, T., Kadokura, G., RamachandranNair, R., Weiss, K., Rutka, J.T., Snead III, O.C., 2005. Focal cortical high-frequency oscillations trigger epileptic spasms: confirmation by digital video subdural EEG. Clin. Neurolophysiol. 116, 2819—2825. Beaumanoir, A., Blume, W., 2005. The Lennox—Gastaut syndrome. In: Roger, J., Bureau, M., Dravet, C., Genton, P., Tassinari, C.A., Wolf, P. (Eds.), Epileptic Syndromes in Infancy, Childhood and Adolescence, 4th ed. John Libbey Eurotext, Montrouge, pp. 125—148. Egli, M., Mothersill, I., O’Kane, M., O’Kane, F., 1985. The axial spasm—–the predominant type of drop seizure in patients with secondary generalized epilepsy. Epilepsia 26, 401—415. Fisher, R.S., Webber, W.R.S., Lesser, R.P., Arroyo, S., Uematsu, S., 1992. High-frequency EEG activity at the start of seizures. J. Clin. Neurophysiol. 9, 441—448. Gastaut, H., Roger, J., Ouahchi, S., Timsit, M., Broughton, R., 1963. An electro-clinical study of generalized epileptic seizures of tonic expression. Epilepsia 4, 15—44. Gastaut, H., Broughton, R., 1972. Epileptic Seizures. Clinical and Electroencephalographic Features, Diagnosis and Treatment. Charles Thomas, Springfield, IL. Gloor, P., 1979. Generalized epilepsy with spike-and-wave discharge: a reinterpretation of its electrographic and clinical manifestations. Epilepsia 20, 571—588. Gross, D.W., Gotman, J., 1999. Correlation of high-frequency oscillations with the sleep—wake cycle and cognitive activity in humans. Neuroscience 94, 1005—1018. Ikeno, T., Shigematsu, H., Miyakoshi, M., Ohba, A., Yagi, K., Seino, M., 1985. An analytic study of epileptic falls. Epilepsia 26, 612—621. Inoue, T., Kobayashi, K., Oka, M., Yoshinaga, H., Ohtsuka, Y., 2008. Spectral characteristics of EEG gamma rhythm associated with epileptic spasms. Brain Dev. 30, 321—328. Kobayashi, K., Oka, M., Akiyama, T., Inoue, T., Abiru, K., Ogino, T., Yoshinaga, H., Ohtsuka, Y., Oka, E., 2004. Very fast rhythmic activity on scalp EEG associated with epileptic spasms. Epilepsia 45, 488—496. Kobayashi, K., Oka, M., Inoue, T., Ogino, T., Yoshinaga, H., Ohtsuka, Y., 2005. Characteristics of slow waves on EEG associated with epileptic spasms. Epilepsia 46, 1098—1105.
K. Kobayashi et al. Kobayashi, K., Inoue, T., Kikumoto, K., Endoh, F., Miya, K., Oka, M., Yoshinaga, H., Ohtsuka, Y., 2007. Relation of spasms and myoclonus to suppression-burst on EEG in epileptic encephalopathies in early infancy. Neuropediatrics 38, 244—250. Miyakoshi, M., Yagi, K., Osawa, T., Seino, M., Wada, T., 1977. Correlative study on symptomatology of epileptic seizures—–behavioral manifestation and electroencephalographic modalities. Folia. Psychiatr. Neurol. Jpn. 31, 451—461. Nunez, P.L., Srinivasan, R., 2006. Electric Fields of the Brain. The Neurophysics of EEG, 2nd ed. Oxford University Press, Oxford. Ochi, A., Otsubo, H., Donner, E.J., Elliott, I., Iwata, R., Funaki, T., Akizuki, Y., Akiyama, T., Imai, K., Rutka, J.T., Snead III, O.C., 2007. Dynamic changes of ictal high-frequency oscillations in neocortical epilepsy: using multiple band frequency analysis. Epilepsia 48, 286—296. Ohtahara, S., 1978. Clinico-electrical delineation of epileptic encephalopathies in childhood. Asian Med. J. 21, 499—509. Ohtahara, S., 1988. Lennox—Gastaut syndrome. Considerations in its concept and categorization. Jpn. J. Psychiatry Neurol. 42, 535—542. Ohtahara, S., Yamatogi, Y., 1990. Evolution of seizures and EEG abnormalities in childhood epilepsy. In: Wada, J.A., Ellingson, R.J. (Eds.), Clinical Neurophysiology of Epilepsy. EEG Handbook (Revised Series), vol. 4. Elsevier Science, Amsterdam. Oka, M., Kobayashi, K., Akiyama, T., Ogino, T., Oka, E., 2004. A study of spike-density on EEG in West syndrome. Brain Dev. 26, 105—112. Otsubo, H., Ochi, A., Imai, K., Akiyama, T., Fujimoto, A., Go, C., Dirks, P., Donner, E.J., 2008. High-frequency oscillations of ictal muscle activity and epileptogenic discharges on intracranial EEG in a temporal lobe epilepsy patient. Clin. Neurophysiol. 119, 862—868. RamachandranNair, R., Ochi, A., Imai, K., Benifla, M., Akiyama, T., Holowka, S., Rutka, J.T., Sneed III, O.C., Otsubo, H., 2008. Epileptic spasms in older pediatric patients: MEG and ictal highfrequency oscillations suggest focal-onset seizures in a subset of epileptic spasms. Epilepsy Res. 78, 216—224. Rampp, S., Stefan, H., 2006. Fast activity as a surrogate marker of epileptic network function? Clin. Neurophysiol. 117, 2111—2117. Traub, R.D., Whittington, M.A., Buhl, E.H., LeBeau, F.E.N., Bibbig, A., Boyd, S., Cross, H., Baldweg, T., 2001. A possible role for gap junctions in generation of very fast EEG oscillations preceeding the onset of, and perhaps initiating, seizures. Epilepsia 42, 153—170. Wendling, F., Bartolomei, F., Bellanger, J.J., Bourient, J., Chauvel, P., 2003. Epileptic fast intracerebral EEG activity: evidence for spatial decorrelated at seizure onset. Brain 126, 1449—1459. Yamatogi, Y., Ohtahara, S., 1981. Age-dependent epileptic encephalopathy: a longitudinal study. Folia. Psychiatr. Neurol. Jpn. 35, 321—332.
journal homepage: www.elsevier.com/locate/epilepsyres
Spectral analysis of EEG gamma rhythms associated with tonic seizures in Lennox—Gastaut syndrome Katsuhiro Kobayashi ∗, Takushi Inoue, Yoshiaki Watanabe, Makio Oka, Fumika Endoh, Harumi Yoshinaga, Yoko Ohtsuka Department of Child Neurology, Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, and Okayama University Hospital, Shikatacho 2-chome 5-1, Okayama 700-8558, Japan Received 7 September 2008; received in revised form 10 March 2009; accepted 15 March 2009 Available online 16 April 2009
KEYWORDS Lennox—Gastaut syndrome; Tonic seizure; Ictal EEG; Time—frequency analysis; High-frequency oscillation; Gamma rhythm
Summary Purpose: EEG gamma rhythms, which are found in association with epileptic spasms in infants with West syndrome, were explored in the ictal EEGs of tonic seizures in older patients with Lennox—Gastaut syndrome (LGS) to investigate the pathophysiology of the disease. Methods: The subjects were 20 patients with LGS (11 males, 9 females; age range: 3 years 1 month to 29 years 3 months) who had at least one digitally recorded tonic seizure with minimal artifacts. A time—frequency analysis was applied to each patient’s ictal EEG data. Results: A total of 54 seizures were analyzed, excluding spasms in clusters. The ictal EEGs of the tonic seizures showed only diffuse desynchronization in 10 seizures, and desynchronization followed by rhythmic activity in 21. The ictal discharges started as rhythmic activity of varying amplitude without initial desynchronization in 23 seizures. In a total of 25 seizures from 13 patients, gamma rhythms with frequencies ranging from 43 to 101.6 Hz were detected by temporal expansion of the ictal EEG traces and spectral analysis. In 24 (96%) of these seizures, gamma rhythms were observed at seizure onset corresponding to visually identified desynchronization. In the remaining seizure, gamma rhythms were found in association with transient suppression of high-amplitude rapid discharges. Conclusion: The detection of gamma rhythms in the ictal EEGs of tonic seizures indicated that some tonic seizures might have generative mechanisms in common with epileptic spasms, and that these mechanisms are possibly related to desynchronization at seizure onset. © 2009 Elsevier B.V. All rights reserved.
Introduction
∗ Corresponding author. Tel.: +81 86 235 7372; fax: +81 86 235 7377. E-mail address: k [email protected] (K. Kobayashi).
Tonic seizures are the cardinal seizure type in Lennox—Gastaut syndrome (LGS) and in related symptomatic generalized epilepsy (SGE) (Beaumanoir and Blume, 2005; Ohtahara, 1988). Their resistance to treat-
0920-1211/$ — see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.eplepsyres.2009.03.011
4 (20.0) 5 (25.0) 20 [11/9] Total
Percentages are shown in parentheses. WS, West syndrome; sz, seizure.
2 (10.0)
7 [4/3] Patients without ictal gamma rhythms
2 months to 7 years 3 months [1 year 6 months]/3 years 1 months to 29 years 3 months [11 years 7 months]
13 (65.0)
3 (15.0)
1 (14.3) 0 (0) 0 (0) 2 (28.6) 3 (42.9)
3 (23.1) 5 (38.5) 3 (23.1) 0 (0) 10 (76.9) 13 [7/6] Patients with ictal gamma rhythms
2 months to 4 years 8 months [1 year 5 months]/3 years 1 months to 25 years [9 years 8 months] 2 months to 7 years 3 months [1 year 10 months]/4 years 11 months to 29 years 3 months [15 years]
Focal sz Myoclonic sz Atypical absence Atonic sz Epileptic spasms in series
Seizure type (other than tonic) Age at sz onset/examination (mean) Number of patients (M/F)
Table 1
Patient characteristics.
Patients
Patient group
Patients and methods
The subjects of the present study were 20 patients with a diagnosis of LGS (11 males, 9 females; age range: 3 years 1 month to 29 years 3 months) who had at least one tonic seizure with minimal artifacts digitally recorded between November 2003 and January 2008 (Table 1). During this period, isolated tonic seizures were recorded in a total of 30 patients with LGS who were at least 3 years of age at the time of examination, but the ictal EEG data of the remaining 10 patients could not be used for the present analysis because of contaminating massive muscle activity and/or artifacts. We used the following diagnostic criteria of LGS: tonic seizures (especially tonic spasms or brief tonic seizures) and/or multiple seizure types consisting of atypical absences and others, along with diffuse slow spike-waves (DSSW) in the interictal EEG (Ohtahara, 1988). All patients were mentally retarded. They all showed highamplitude 8—14 Hz rapid rhythms in sleep EEG. The underlying disorders were tuberous sclerosis in three patients, cortical malformation including lissencephaly in three, sequel of acute encephalopathy or meningitis in three, sequel of premature birth with intraventricular hemorrhage or periventricular leukomalacia in two, cerebral atrophy of unknown origin in two, sequel of severe neonatal asphyxia in one, multiple malformations in one, an arach-
9 (69.2)
History of WS
ment poses a serious clinical problem in epileptology. LGS is regarded as the most mature form of age-dependent epileptic encephalopathy (Ohtahara, 1978; Ohtahara and Yamatogi, 1990), and the evolution of LGS from West syndrome (WS) suggests a common pathophysiology shared by the two syndromes. High-frequency activity of more than 60—80 Hz can be invasively recorded from epileptogenic brain regions, and the study of this activity type is now attracting attention (Fisher et al., 1992; Traub et al., 2001; Rampp and Stefan, 2006). Gamma rhythms in the frequency range of 50—100 Hz are found in the scalp ictal EEGs of epileptic spasms in patients with WS (Kobayashi et al., 2004) and in early infantile epileptic encephalopathies such as Ohtahara syndrome and early myoclonic encephalopathy (Kobayashi et al., 2007). These EEG gamma rhythms are regular and morphologically distinct from irregular muscle activity in temporally expanded EEG traces. The ictal gamma rhythms appear to be closely related to the generative mechanisms of spasms (Inoue et al., 2008). Observation of high-frequency rhythms in the gamma and faster bands in ictal electrocorticography of epileptic spasms, and successful suppression of spasms by the resection of cortical regions showing these high-frequency rhythms, strengthen the possibility of a relationship between cortical high-frequency rhythms and the generative mechanisms of spasms (Akiyama et al., 2005; Ochi et al., 2007; RamachandranNair et al., 2008). Based on the age-dependent transition of WS to LGS, we speculated that tonic seizures in older patients, like spasms in WS, may also have generative mechanisms linked to gamma rhythms. Therefore, we set out to investigate the ictal EEG patterns associated with tonic seizures using time—frequency spectral analysis, as in our previous study on spasms (Kobayashi et al., 2004), to explore gamma rhythms in patients with LGS.
13 (65.0)
K. Kobayashi et al.
4 (57.1)
16
Spectral analysis of EEG gamma rhythms
17
noid cyst in one, and unknown with no detectable brain lesion by MRI in the remaining four. Informed consent for this study was obtained from the parents of the patients.
EEG recording and frequency analysis EEGs were recorded at a digital sampling frequency of 500 Hz with a Nihon Kohden Neurofax (Tokyo, Japan), which used a bandpass filter ranging from 0.016 to 300 Hz before digital sampling. The international 10—20 electrode placement system was used. We simultaneously recorded video and/or electromyograms (EMGs) from the bilateral deltoid muscles in most patients. The conventional EEG traces (10—20 s per page) were initially reviewed to identify the ictal events using a time constant of 0.1 s. The traces were then temporally expanded (2 s per page) with a time constant of 0.03 s (corresponding to a high-pass filter of 5.3 Hz) in order to study the details of activity faster than 40 Hz (for example, see Fig. 1B and C). We investigated the time evolution of the high-frequency power spectrum of the ictal activity by applying the Gabor transform (a Fourier transform with a 200 ms wide Gaussian window) to the original EEG data, as in our previous study regarding the derivation pairs of F3-C3, F4-C4, P3-O1 and P4-O2 (Kobayashi et al., 2004; see Fig. 2). Gamma peaks were identified as clearly visible spectral spots with frequencies faster than 40 Hz that were surrounded by an area of low power and that corresponded to rhythmic EEG activity. The width of each spectral segment was 5 s, and the frequency range was 10—120 Hz. Computation was performed using a program written in-house for MATLAB (version 6.5.1; MathWorks Inc., Natick, MA). In each time—frequency spectrum, peak frequency and peak power of the gamma rhythms were measured for statistical analysis. Differences in peak frequencies among the derivations of F3-C3, F4-C4, P3-O1 and P4-O2 were evaluated by one-way analysis of variance (ANOVA). Because peak power varied greatly across patients, the pattern of distribution of the gamma rhythms over the scalp was evaluated separately for each individual. The distribution pattern was somewhat arbitrarily defined as fronto-central dominant in each seizure wherein average peak power in the bilateral frontocentral regions (F3-C3 and F4-C4) was more than twice that in the bilateral parieto-occipital regions (P3-O1 and P4-O2), and it was defined as parieto-occipital dominant when the above situation was reversed. When the average peak power was not more than twice as great in one region than in the other, the pattern of distribution of the gamma rhythms was regarded as showing no clear dominance. When there were multiple gamma peaks in an individual spectrum, the peak with the greatest power was selected for statistical analysis (Inoue et al., 2008). We used an unpaired t-test to look for relationships between the presence of ictal gamma rhythms in at least one seizure and (1) patient age at seizure onset, and (2) patient age at examination (the first examination in cases of multiple examinations). We used Fisher’s exact test to look for a relationship between the presence of ictal gamma rhythms and (1) the presence of various other types of seizures, and (2) a history of WS. Relationships were considered statistically significant if p < 0.05.
Results A total of 121 tonic seizures were recorded from 30 patients; of these, 54 seizures with minimum artifacts from 20 patients were analyzed as described above. Epileptic spasms in clusters were excluded from the study. The observed ictal EEG patterns of tonic seizures were as follows (see Table 2). Diffuse amplitude attenuation
Figure 1 Ictal EEG of a brief tonic seizure (the corresponding time—frequency spectra are shown in Fig. 2). In (A) and (B) ‘‘Delt.’’ refers to the deltoid muscle. (A) A conventional trace showing diffuse desynchronization associated with a brief tonic seizure (arrow). (B) A temporally expanded trace showing diffuse gamma rhythms with regular sinusoidal morphology in the frequency range of 60—80 Hz at seizure onset (horizontal lines). Muscle activity contamination in the temporal and frontopolar channels is very irregular and morphologically different from the ictal gamma rhythms. (C) A magnification of two traces indicated by horizontal lines in part (B). The point of the arrow in the conventional trace corresponds to that of the arrowhead in the expanded trace.
(desynchronization) was found in 10 seizures from four patients (Fig. 1). Desynchronization was succeeded by rhythmic beta activity partly mixed with alpha and/or theta activity in 21 seizures from 11 patients (Fig. 3). Desynchronization had a leading spike in 13 seizures from eight
18
Table 2
Ictal EEG observations of tonic seizures and their associated gamma rhythms.
Visual ictal EEG pattern
Diffuse desynchronization Diffuse desynchronization and subsequent rhythmic activity Rhythmic activity with low-amplitude onset Rhythmic activity with medium-amplitude onset High-amplitude rapid discharges Rapid discharges evolving to desynchronization and subsequent rhythmic activity Total
Number of seizures analyzed (waking/sleep)
Mean seizure duration (mean ± SD; s)
Gamma rhythms Peak frequency of gamma rhythms detected (mean ± SD; Hz) in F3-C3/F4-C4/P3-O1/P4-O2
10 [8/2]
3.7 ± 1.3
10 (100)
21 [12/9]
9.4 ± 5.8
14 (66.7)
66.0 ± 14.9/63.3 ± 14.4/71.8 ± 13.0//74.5 ± 15.8 77.3 ± 9.9/77.4 ± 11.7/74.4 ± 10.8/75.6 ± 10.4
Distribution of gamma rhythms No dominance Fronto-central Parieto-occipital detected dominant dominant 7
1
2
10
3
1
4 [0/4]
10.3 ± 4.0
0 (0)
N/A
N/A
N/A
N/A
8 [1/7]
6.4 ± 2.5
0 (0)
N/A
N/A
N/A
N/A
10 [0/10]
6.6 ± 3.3
0 (0)
N/A
N/A
N/A
N/A
1 (100)
43.0/43.0/46.9/43.0
1
0
0
25 (46.3)
71.4 ± 14.2/70.4 ± 15.3/72.3 ± 12.5/73.8 ± 13.7
18 (72.0)
4 (16.0)
3 (12.0)
1 [0/1]
54 [21/33]
40
8.0 ± 6.4
Percentages are shown in parentheses. SD, standard deviation.
K. Kobayashi et al.
Spectral analysis of EEG gamma rhythms
19
Figure 2 Time—frequency spectra of ictal EEG activity (the corresponding EEG is shown in Fig. 1). Spectral peaks of the ictal gamma rhythms stand out as isolated spots (arrows). The gamma rhythms slow down gradually from 60 to 80 Hz at onset to 40—60 Hz at the end. The spectra of muscle activity recorded from the bilateral deltoid muscles (Delt) are noisy and include a wide range of frequency components—–spectral patterns that are different from those of the EEG gamma rhythms. The onset of spectral peaks from EEG gamma rhythms (vertical red lines) precedes the onset of deltoid muscle activity. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
patients. Rhythmic discharges started insidiously as lowamplitude (≤50 V) beta activity without desynchronization and evolved to include alpha/beta activity in four seizures from four patients, and rhythmic discharges began as medium-amplitude (50—150 V) activity in eight seizures from five patients. Bursts of high-amplitude (≥200 V) 8—14 Hz rapid discharges including some beta activity were observed in ten seizures from three patients. Rapid discharges evolved to show diffuse amplitude attenuation and subsequent rhythmic beta/alpha activity in the remaining seizure. Of all 54 seizures, 21 (38.9%) occurred during wakefulness, and the remaining 33 (61.1%) during sleep. The majority (8/10, 80.0%) of seizures with only desynchronization occurred during wakefulness. In contrast, most (7/8, 87.5%) of the seizures with medium-amplitude onset activity and all (10/10) of the seizures with high-amplitude rapid discharges occurred during sleep. Gamma rhythms with peak frequencies ranging from 43.0 to 101.6 Hz were detected in 25 of the 54 seizures studied (46.3%); these seizures occurred in 13 out of the 20 patients (see Tables 1 and 2). In 24 (96.0%) of these seizures, ictal gamma rhythms were observed at seizure onset, corresponding to the visually identified desynchronization in EEG. Gamma rhythms were found in all of the ten seizures showing only desynchroniza-
tion (Figs. 1 and 2). They were observed in 14 (66.7%) of the 21 seizures with onset desynchronization and subsequent rhythmic activity (Figs. 3 and 4). Frequency of the ictal activity tended to run down from the gamma band to the beta band with the progression of the seizure (Figs. 2 and 4). In the single seizure that started with high-amplitude rapid discharges and subsequently evolved to amplitude attenuation and then to rhythmic beta/alpha activity, gamma rhythms were detected at the time of transient suppression of rapid discharges. In this exceptional seizure, however, the peak frequencies of the gamma rhythms were relatively low (43.0—46.9 Hz) and close to the beta band. In the remaining 29 seizures, the ictal activity lacked clear gamma peaks in the temporally expanded EEG traces and the time—frequency spectra. In all cases where ictal gamma rhythms were detected, they were regular in morphology, as illustrated in the magnified traces (Figs. 1C and 3C), and thus discernible from muscle activity contamination with irregular morphology in the EEG. The spectra of the ictal gamma rhythms showed corresponding well-defined peaks in the gamma band, in contrast to the broad and noisy spectra of muscle activity. This finding agrees well with previous studies (Kobayashi et al., 2004, 2007).
20
K. Kobayashi et al. ber of electrode channels used in this study, we could not detect a relationship between peak frequency or power and spatial distribution of gamma rhythms associated with tonic seizures. Finally, the patients were divided into two groups based on whether gamma rhythms were detected in at least one seizure (see Table 1). The presence of ictal gamma rhythms was not statistically related to age at seizure onset (p = 0.630), age at examination (p = 0.152), or the history of WS (p = 0.651). In addition, no statistical relationship was found between the presence of ictal gamma rhythms and the presence of other types of seizures, including epileptic spasms in clusters (p = 0.174), atonic seizures (p = 0.111), atypical absences (p = 0.521), myoclonic seizures (p = 0.114), or focal seizures (p = 1.000).
Discussion
Figure 3 Ictal EEG of a tonic seizure (the corresponding time—frequency spectra are shown in Fig. 4). (A) A conventional trace showing diffuse desynchronization (arrow) and subsequent rhythmic activity associated with a tonic seizure. (B) A temporally expanded trace showing diffuse gamma rhythms in the frequency range of 60—100 Hz at seizure onset (horizontal lines). Muscle activity contamination in the bilateral temporal channels is very irregular and morphologically different from the ictal gamma rhythms. (C) A magnification of two traces indicated by horizontal lines in part (B). The point of the arrow in the conventional trace corresponds to that of the arrowhead in the expanded trace.
In addition to identifying gamma rhythms in the EEG spectra, we evaluated their characteristics across different regions of the scalp. The peak frequencies of the gamma rhythms did not significantly differ between the bilateral fronto-central and parieto-occipital regions (p = 0.857). In terms of power, the gamma rhythms were fronto-central dominant in 16.0% of cases and parieto-occipital dominant in 12.0%; the remaining 72.0% of seizures with gamma rhythms showed no clear dominance. Thus, with the limited num-
Gamma rhythms were detected in scalp EEGs in association with at least 46.3% of tonic seizures, and were particularly associated with tonic seizures exhibiting onset desynchronization. The presence of small amount of gamma rhythms in the remaining tonic seizures cannot be precluded because they exhibited spectral spreading from beta bands with occasional noise contamination, and it is possible that some gamma activity occurred but failed to form recognizable peaks. The seizures analyzed in the present study were axial or axiorhizomelic tonic (Gastaut and Broughton, 1972), as spectral analysis was difficult to perform on global tonic seizures that were accompanied by violent movements with massive muscle activity and/or artifacts in the EEGs. We identified ictal gamma rhythms as seizure-associated gamma rhythms only when they exhibited both clear EEG morphology and corresponding spectral patterns. Muscle activity contamination generally has irregular morphology and is recorded at electrodes close to the muscles (Kobayashi et al., 2004). In accordance with our prior studies, Otsubo et al. (2008) reported that high-frequency activity of muscular origin that contaminated intracranial EEGs was randomly scattered without a specific frequency band in spectral analysis. Therefore, ictal gamma rhythms can be differentiated from muscle activity contamination. They are also different from 60 Hz alternating current (AC) artifacts, because AC artifacts have monotonous morphology with a frequency exactly at 60 Hz (Kobayashi et al., 2004). It is hard to detect cortical activity of very high frequency by scalp EEG because of attenuation of higher frequencies between cortex and scalp (Nunez and Srinivasan, 2006). The frequency of cortical oscillations associated with epileptic spasms reaches 250 Hz (Ochi et al., 2007), and the ictal gamma rhythms recorded over the scalp may be only the relatively slower components of the ictal cortical activity. Intracranial EEG is ideal for the investigation of highfrequency activity, but it would have been unethical to use on the patients of the present study, who lacked findings suggesting that their tonic seizures were actually secondary generalizations from a cortical focus. The ictal gamma rhythm is posterior dominant in epileptic spasms (Kobayashi et al., 2004), corresponding to the
Spectral analysis of EEG gamma rhythms
21
Figure 4 Time—frequency spectra of ictal EEG activity (the corresponding EEG is shown in Fig. 3). Spectral peaks of the ictal gamma rhythms are indicated by arrows. The gamma rhythms slow down gradually from 60 to 100 Hz at seizure onset to 40—60 Hz with progression of seizure discharges.
posterior predominant distribution of abnormalities that is seen in West syndrome (Oka et al., 2004). In LGS, the distribution of EEG abnormalities, such as DSSW, is generally diffuse (Beaumanoir and Blume, 2005). We could not detect regional difference of the ictal gamma rhythms in the tonic seizures analyzed with the limited number of electrode channels used in this study. The desynchronization that was observed alone or at the onset of longer tonic seizures in the EEGs resembled the ictal activity of epileptic spasms in WS in terms of both morphology and association with gamma rhythms (Kobayashi et al., 2005). The desynchronization pattern actually included slow potential changes with superposition of low-amplitude activity (Figs. 1 and 3), and it mimicked the ictal slow waves of epileptic spasms. Seizures resembling epileptic spasms have been reported in LGS, and these seizures cause the patients to drop (‘‘axial spasms’’ in Egli et al., 1985; ‘‘flexor spasms’’ in Ikeno et al., 1985). We could not tell if the seizures analyzed in the present study caused the patients to drop because they were supine during the EEG recording; an additional study would be necessary to resolve this question. The three well-known ictal EEG patterns of tonic seizures are (1) desynchronization, (2) beta activity at 20 ± 5 Hz that is initially of low voltage and progressively increases in amplitude, and (3) a slower high-amplitude recruiting rhythm at about 10 Hz (Gastaut and Broughton, 1972; Yamatogi and Ohtahara, 1981). In this study, the ictal EEG patterns were subdivided for the purpose of analysis (Table 2); each seizure displayed one or more of the patterns listed above. There appeared to be a spectrum of ictal EEG patterns. At one end is desynchronization that tends to occur diurnally more often than nocturnally. In these seizures, gamma rhythms are often superimposed on desynchronization. At the other end of the spectrum lies a pattern of high-amplitude rapid discharges with consistently nocturnal occurrence. Rhythmic beta activity with varying amplitude
falls somewhere in between. In intracranial EEGs of epileptic patients, the background gamma activity is greater in the waking state than in sleep (Gross and Gotman, 1999). Occurrence of gamma rhythms in diurnal seizures might be facilitated by a general tendency of the neuronal system to generate more gamma activity during the waking state, though its precise mechanisms remain to be elucidated. It is known that seizures with intense tonicity are often associated with desynchronization in EEG, while seizures with weaker tonicity tend to show high-amplitude activity (Gastaut et al., 1963; Miyakoshi et al., 1977). The decrease in spectral frequency of the ictal rhythms from the gamma band to the beta band was found to be associated with the progression of seizure discharges from desynchronization to rhythmic beta/alpha/theta activity. The frequency of ictal activity is known to decrease from the beta band to about 10 Hz in some tonic seizures (Gastaut and Broughton, 1972). In an intracerebral EEG study, fast seizure activity in the 60—90 Hz band was associated with very low spatial correlation among cerebral regions, indicating a functional decoupling of brain sites at seizure onset (Wendling et al., 2003). This functional decoupling was followed by an abnormally high re-coupling when the seizure developed into activity of decreasing frequency and increasing amplitude. A phenomenon similar to this functional decoupling and re-coupling among brain regions may take place during the progression of tonic seizures with activity changes from the gamma band to the slower bands. In generalized corticoreticular epilepsy, a complex and abnormal interaction is thought to exist among the cortex, the thalamus, and the brainstem (Gloor, 1979). Tonic seizures are generally believed to originate from the brainstem (Beaumanoir and Blume, 2005). We have previously suggested that ictal gamma rhythms in EEG represent the cortical component of the abnormal process of seizure generation (Kobayashi et al., 2004). The detection of gamma rhythms in association with tonic seizures, and especially
22 in association with onset desynchronization, might indicate that the cortex is involved in the initial phase of the generation process of at least some tonic seizures, and that this process involves the functional decoupling of cerebral regions.
References Akiyama, T., Otsubo, H., Ochi, A., Ishiguro, T., Kadokura, G., RamachandranNair, R., Weiss, K., Rutka, J.T., Snead III, O.C., 2005. Focal cortical high-frequency oscillations trigger epileptic spasms: confirmation by digital video subdural EEG. Clin. Neurolophysiol. 116, 2819—2825. Beaumanoir, A., Blume, W., 2005. The Lennox—Gastaut syndrome. In: Roger, J., Bureau, M., Dravet, C., Genton, P., Tassinari, C.A., Wolf, P. (Eds.), Epileptic Syndromes in Infancy, Childhood and Adolescence, 4th ed. John Libbey Eurotext, Montrouge, pp. 125—148. Egli, M., Mothersill, I., O’Kane, M., O’Kane, F., 1985. The axial spasm—–the predominant type of drop seizure in patients with secondary generalized epilepsy. Epilepsia 26, 401—415. Fisher, R.S., Webber, W.R.S., Lesser, R.P., Arroyo, S., Uematsu, S., 1992. High-frequency EEG activity at the start of seizures. J. Clin. Neurophysiol. 9, 441—448. Gastaut, H., Roger, J., Ouahchi, S., Timsit, M., Broughton, R., 1963. An electro-clinical study of generalized epileptic seizures of tonic expression. Epilepsia 4, 15—44. Gastaut, H., Broughton, R., 1972. Epileptic Seizures. Clinical and Electroencephalographic Features, Diagnosis and Treatment. Charles Thomas, Springfield, IL. Gloor, P., 1979. Generalized epilepsy with spike-and-wave discharge: a reinterpretation of its electrographic and clinical manifestations. Epilepsia 20, 571—588. Gross, D.W., Gotman, J., 1999. Correlation of high-frequency oscillations with the sleep—wake cycle and cognitive activity in humans. Neuroscience 94, 1005—1018. Ikeno, T., Shigematsu, H., Miyakoshi, M., Ohba, A., Yagi, K., Seino, M., 1985. An analytic study of epileptic falls. Epilepsia 26, 612—621. Inoue, T., Kobayashi, K., Oka, M., Yoshinaga, H., Ohtsuka, Y., 2008. Spectral characteristics of EEG gamma rhythm associated with epileptic spasms. Brain Dev. 30, 321—328. Kobayashi, K., Oka, M., Akiyama, T., Inoue, T., Abiru, K., Ogino, T., Yoshinaga, H., Ohtsuka, Y., Oka, E., 2004. Very fast rhythmic activity on scalp EEG associated with epileptic spasms. Epilepsia 45, 488—496. Kobayashi, K., Oka, M., Inoue, T., Ogino, T., Yoshinaga, H., Ohtsuka, Y., 2005. Characteristics of slow waves on EEG associated with epileptic spasms. Epilepsia 46, 1098—1105.
K. Kobayashi et al. Kobayashi, K., Inoue, T., Kikumoto, K., Endoh, F., Miya, K., Oka, M., Yoshinaga, H., Ohtsuka, Y., 2007. Relation of spasms and myoclonus to suppression-burst on EEG in epileptic encephalopathies in early infancy. Neuropediatrics 38, 244—250. Miyakoshi, M., Yagi, K., Osawa, T., Seino, M., Wada, T., 1977. Correlative study on symptomatology of epileptic seizures—–behavioral manifestation and electroencephalographic modalities. Folia. Psychiatr. Neurol. Jpn. 31, 451—461. Nunez, P.L., Srinivasan, R., 2006. Electric Fields of the Brain. The Neurophysics of EEG, 2nd ed. Oxford University Press, Oxford. Ochi, A., Otsubo, H., Donner, E.J., Elliott, I., Iwata, R., Funaki, T., Akizuki, Y., Akiyama, T., Imai, K., Rutka, J.T., Snead III, O.C., 2007. Dynamic changes of ictal high-frequency oscillations in neocortical epilepsy: using multiple band frequency analysis. Epilepsia 48, 286—296. Ohtahara, S., 1978. Clinico-electrical delineation of epileptic encephalopathies in childhood. Asian Med. J. 21, 499—509. Ohtahara, S., 1988. Lennox—Gastaut syndrome. Considerations in its concept and categorization. Jpn. J. Psychiatry Neurol. 42, 535—542. Ohtahara, S., Yamatogi, Y., 1990. Evolution of seizures and EEG abnormalities in childhood epilepsy. In: Wada, J.A., Ellingson, R.J. (Eds.), Clinical Neurophysiology of Epilepsy. EEG Handbook (Revised Series), vol. 4. Elsevier Science, Amsterdam. Oka, M., Kobayashi, K., Akiyama, T., Ogino, T., Oka, E., 2004. A study of spike-density on EEG in West syndrome. Brain Dev. 26, 105—112. Otsubo, H., Ochi, A., Imai, K., Akiyama, T., Fujimoto, A., Go, C., Dirks, P., Donner, E.J., 2008. High-frequency oscillations of ictal muscle activity and epileptogenic discharges on intracranial EEG in a temporal lobe epilepsy patient. Clin. Neurophysiol. 119, 862—868. RamachandranNair, R., Ochi, A., Imai, K., Benifla, M., Akiyama, T., Holowka, S., Rutka, J.T., Sneed III, O.C., Otsubo, H., 2008. Epileptic spasms in older pediatric patients: MEG and ictal highfrequency oscillations suggest focal-onset seizures in a subset of epileptic spasms. Epilepsy Res. 78, 216—224. Rampp, S., Stefan, H., 2006. Fast activity as a surrogate marker of epileptic network function? Clin. Neurophysiol. 117, 2111—2117. Traub, R.D., Whittington, M.A., Buhl, E.H., LeBeau, F.E.N., Bibbig, A., Boyd, S., Cross, H., Baldweg, T., 2001. A possible role for gap junctions in generation of very fast EEG oscillations preceeding the onset of, and perhaps initiating, seizures. Epilepsia 42, 153—170. Wendling, F., Bartolomei, F., Bellanger, J.J., Bourient, J., Chauvel, P., 2003. Epileptic fast intracerebral EEG activity: evidence for spatial decorrelated at seizure onset. Brain 126, 1449—1459. Yamatogi, Y., Ohtahara, S., 1981. Age-dependent epileptic encephalopathy: a longitudinal study. Folia. Psychiatr. Neurol. Jpn. 35, 321—332.