Clinical Neurophysiology xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Clinical Neurophysiology journal homepage: www.elsevier.com/locate/clinph
Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy Howan Leung a,⇑, Cannon X.L. Zhu b, Danny T.M. Chan b, Wai S. Poon b, Lin Shi a, Vincent C.T. Mok a, Lawrence K.S. Wong a a b
Division of Neurology, Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Hong Kong Special Administrative Region Division of Neurosurgery, Department of Surgery, The Chinese University of Hong Kong, Hong Kong Special Administrative Region
a r t i c l e
i n f o
Article history: Accepted 1 January 2015 Available online xxxx Keywords: High-frequency oscillations Hyperexcitability Refractory epilepsy
h i g h l i g h t s High-frequency oscillations captured at the onset of seizure may help determine surgical outcome in
patients with refractory epilepsy. Cortical areas demonstrating hyperexcitability may be associated with ictal high-frequency oscillations. By examining hyperexcitability and ictal high-frequency oscillations, the decision-making process for
surgical resection to treat refractory epilepsy may be improved with values of sensitivity and specificity that are more optimal.
a b s t r a c t Objective: High-frequency oscillations (HFOs, 80–500 Hz) from intracranial electroencephalography (EEG) may represent a biomarker of epileptogenicity for epilepsy. We explored the relationship between ictal HFOs and hyperexcitability with a view to improving surgical outcome. Methods: We evaluated 262 patients with refractory epilepsy. Fifteen patients underwent electrode implantation, and surgical resection was performed in 12 patients using a semi-prospective design. Ictal intracranial EEGs were examined by continuous wavelet transform (CWT). Significant ictal HFOs were denoted by normalized wavelet power above the 50th percentile across all channels. Each patient underwent functional mapping with cortical electrical stimulation. Hyperexcitability was defined as the appearance of afterdischarges or clinical seizures after electrical stimulation (50 Hz, biphasic, pulse width = 0.5 ms, 5 s, 5 mA). Results: Among the group of patients achieving Engel Class I/II outcome at 1+ year, the mean proportion of significant ictal HFOs among resected channels for any given patient was 69% (33.3–100%). The respective figures for conventional frequency ictal patterns (CFIPs), hyperexcitability, and radiological lesion were 68.3% (26.3–100%), 39.6% (0–100%), and 52.8% (0–100%). Statistical significance was only achieved with ictal HFOs when comparing patients with Engel Class I/II outcomes versus III/IV outcomes (12.6% vs. 4.2%, the number of channels as the denominator, p = 0.005). Results: Further analysis from all patients irrespective of the surgical outcome showed that ictal HFOs co-occurred with CFIP (p < 0.001), hyperexcitability (p < 0.001), and radiological lesion (p < 0.001). The combination of ictal HFOs/hyperexcitability improved the sensitivity from 66.7% to 100%, and the specificity from 66.7% to 75% when compared with ictal HFOs or hyperexcitability alone. Conclusions: We confirmed the utility of ictal HFOs in determining surgical outcome. Ictal HFOs are affiliated to cortical hyperexcitability, which may represent a pathological manifestation of epileptogenicity. Significance: Presurgical evaluation of refractory epilepsy may incorporate both ictal HFOs and cortical stimulation in determining epileptogenic foci. Ó 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
⇑ Corresponding author at: 10/F, Clinical Science Building, Prince of Wales Hospital, Shatin NT, Hong Kong Special Administrative Region. Tel.: +852 2632 1855; fax: +852 2637 3852. E-mail address: [email protected] (H. Leung). http://dx.doi.org/10.1016/j.clinph.2015.01.009 1388-2457/Ó 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
2
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
1. Introduction The challenge of treating refractory epilepsy with surgery begins with the exploration of the epileptogenic zone, defined as ‘‘the minimal area of cortex that must be resected to produce seizure freedom’’ (Luders et al., 1993). This apparent basis for presurgical evaluation has fueled research on multimodal investigations that identify or define the electrographic details to which epileptogenic zones conform. In 1954, Penfield and Jasper had already used the concepts of interictal corticography to establish the limits of the epileptogenic zone. By 1962, Talairach and Bancaud introduced the technique of stereo-electroencephalography to study ictal events and advance knowledge of the epileptogenic zone. Today, high-frequency oscillations (HFOs) are added to the armamentarium and hailed as a new biomarker to signify epileptogenesis. We have a far greater number of tools with which to investigate patients with refractory epilepsy compared to what was available in the era of Penfield and Jasper, who meticulously adopted cortical stimulation to reproduce aura and other ictal manifestations. The concept of hyperexcitability was a hallmark of epilepsy from the days of the forerunners in epilepsy research, but it can still be applied today in presurgical evaluation in the setting of intracranial electrode implantation. HFOs in the range of 80–500 Hz have been identified and classified (Buzsaki et al., 1992) for operational reasons, and their utility in seizure-onset identification has been demonstrated in animal models of epilepsy (Bragin et al., 1999, 2003), and later in intracranial microelectrodes from hippocampal and entorhinal cortices of patients with epilepsy (Bragin et al., 2005; Staba et al., 2002, 2004) and those with neocortical epilepsy (Weiss et al., 2013). These results were replicable with intracranial macroelectrodes (Jirsch et al., 2006; Urrestarazu et al., 2007). We are interested in the analysis of ictal HFOs, as the literature on this aspect of HFO is relatively limited. The traditional method of determining an intracranial seizure-onset zone using conventional frequency ictal patterns (CFIP) suffers from a number of drawbacks: (1) difficulty in deciphering which of the onset patterns may assume superior significance, (2) proximity in time sequence between similar onset patterns of adjacent channels, and (3) problems encountered with rapid spread of discharges. Fujiwara and researchers also pointed out the following: (4) the surface findings of grid electrodes may not indicate what may be happening in deeper cortical sources near the bottom of the sulci, (5) the depth electrode is often impractical and only feasible for a small area of enquiry, (6) epileptogenic areas with a tangential orientation either have undetectable amplitudes or are too widespread, and (7) electrodes may have been placed in such a way that the visual analysis of ictal patterns could be misleading (Fujiwara et al., 2012). In a study of patients with temporal lobe epilepsy using intracranial EEG, the spatial and time-related properties of HFOs were explored. A spatial correlation of HFO with a seizure-onset zone was observed, and the researchers found the presence of 100–200 Hz HFOs at least 8 s before seizure (Khosravani et al., 2009). In a pediatric study of nine children with neocortical epilepsy, both wide-band (250 Hz) and narrow-band (68–164 Hz) HFOs were registered. In those patients achieving seizure freedom, more electrodes recorded HFOs inside the resection margin than outside, both before and after clinical seizure onset (Ochi et al., 2007). A study of six patients with neocortical epilepsy showed that ictal HFOs, manifested as frequencies P70 Hz with sustained evolution, had a higher peak frequency and more spatially restricted appearance, and they were 10 times more likely to be resected than ictal HFOs without evolution (Modur et al., 2011). In a study that examined intracranial EEG patterns and radiological lesions, HFOs were significantly associated with the developing patterns at seizure onset, regardless of the patterns themselves (Perucca et al., 2014).
The relationship between hyperexcitability and HFOs was explored in one study, which examined interictal HFO and electrical stimulation of areas using intracranial macroelectrodes. The EEG segments consisted of 5 min of slow-wave sleep in patients with intracranial implantation. Interictal HFO rates were negatively correlated with thresholds for response to electrical stimulation, lending support to the notion that HFO and hyperexcitability may share a similar mechanistic platform (Jacobs et al., 2010a,b). We hypothesize the following points in our current study: (1) Among patients who attain satisfactory surgical outcomes, a higher proportion of the resected channels demonstrate significant ictal HFOs. A test of significance will be carried out with CFIP, hyperexcitability, and radiological lesion using the surgical outcome. (2) Ictal HFOs are associated significantly with CFIP, hyperexcitability, and radiological lesion using all available channels. (3) The sensitivity and specificity of using ictal HFOs together with hyperexcitability in determining surgical outcome may be superior to ictal HFOs, CFIP, hyperexcitability, or radiological lesion alone.
2. Methods We prospectively evaluated 262 patients with refractory epilepsy at a university-affiliated hospital during the period between 2007 and 2012. Eighty patients were eligible for direct resective surgery, and another 55 patients had sufficient information to localize or lateralize a potential area of resection based on MRI, surface video EEG, clinical psychological testing, positron emission tomography (PET), single photon emission computed tomography (SPECT), or Wada test, i.e. a testable hypothesis formulated for implantation of an intracranial electrode. The inclusion criteria were as follows: (1) age P18 and (2) refractory epilepsy with an implantation hypothesis. A typical implantation schedule would consist of grid electrodes to sample the neocortical areas and depth electrodes to sample the mesial temporal regions. The exclusion criteria were as follows: (1) inability to provide written informed consent and (2) lack of ictal episodes being captured during intracranial monitoring. Fifteen patients underwent intracranial EEG when the technical placement of electrodes could be resolved and financed. At the time of recruitment for the study, intracranial EEG was not a fundable item under the Hospital Authority of Hong Kong, so patients required additional financial support to undergo this part of their treatment. The patient characteristics are given in Table 1. We used the NicoletÒ machine (Viasys, Santa Ana, CA, USA) for video electroencephalographic monitoring with a minimum sampling rate of 1024 Hz. The intracranial electrodes used met the standard for impedance and stability with platinum contacts of 4.0-mm diameter and 2.3-mm exposure (Ad-Tech, Irving, TX, USA). The ictal recordings from intracranial EEG were analyzed by visual analysis of CFIP, and the final plan for resection was made during a joint epilepsy surgery meeting (HL, XZ, DC, WP, and KW), after consideration of both electrographic information and the results of functional mapping. Inpatient intracranial EEG monitoring may last between 7 and 14 days. Off-line analysis of ictal recordings was carried out using a Matlab-based platform by an independent assessor blinded to the clinical information (SL). In the wake of 2011, we considered ictal HFOs as part of the evidence in the discussion leading up to resective surgery. A minimum follow-up period of 1 year is required for all patients to determine their surgical outcome. Surgical outcome was determined using the Engel classification for epilepsy surgery (Engel, 1993).
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
Patient
Age
Sex
1. IK
35
M
2. CFK
26
F
3. LIF
25
M
4 WKM
38
F
5. LMW
34
6. KYH
Seizure type
MRI
Ictal surface EEG
PET
SPECT
Implantation
CPS
Normal
Left temporal
NA
Left temporal grid and depth electrode
CPS, GTCS
Right MTS
Bilateral temporal onset
NA
Bilateral depth electrode
SPS, CPS
Normal
Bi-frontal
CPS
Bi-lateral MTS
Right temporal
Left mid frontal hyper-perfusion NA
Left frontal grids and strips
26
F
15
CPS, GTCS
Left temporal cavernoma
Left temporal
NA
Left temporal grid + functional mapping
33
M
10
CPS, GTCS
Right frontal
NA
Right frontal grid
7. LHK 8. TYM
37 31
M F
20 15
CPS, GTCS CPS, GTCS
Left frontal Left lateralization
NA NA
NA NA
9. TWY 10. UCS 11. YKY 12. YHY
41 25 33 29
F M F M
12 10 18 10
CPS CPS, GTCS CPS, GTCS CPS, GTCS
Left temporal Bilateral temporal Left temporal Right frontal
Left temporal Left temporal Left temporal NA
NA Left temporal NA Right frontal
Left frontal grid Left occipital, parietal grids and inter-hemispheric strips Bilateral depth electrodes Left temporal grids and strips Left temporal grid and strips Right frontal grid and inter-hemispheric strips
13. LWS 14. LKY 15. WKK Patient
38 F 27 F 39 M Resection
22 4 34
CPS CPS, GTCS CPS, GTCS
Right frontal focal cortical dysplasia Left frontal encephalomalacia Left occipital encephalomalacia Bilateral MTS Normal Normal Right frontal encephalomalacia Normal Bilateral MTS Bilateral MTS Surgical outcome
Left temporal hypo-metabolism Right temporal hypo-metabolism Left mid-frontal hypo-metabolism Right temporal hypo-metabolism Left temporal hypometabolism NA
Right frontal Bilateral temporal Right temporal
NA Bilateral temporal Normal Pathology
Right frontal NA Normal
Right frontal grid Bilateral depth electrode Bilateral depth electrode
1. IK 2. CFK
Left anterior temporal lobectomy and topectomy Right anterior temporal lobectomy and hippocampectomy Left frontal topectomy Right anterior temporal lobectomy and hippocampectomy Left temporal lesionectomy Right frontal lesionectomy Left frontal lesionectomy Left occipital lesionectomy Right temporal lobectomy and hippocampectomy (based on CFIP only) Left temporal lobectomy and hippocampectomy Left temporal lobectomy and hippocampectomy right frontal lesionectomy NA NA NA
3. LIF 4. WKM 5. 6. 7. 8. 9.
LMW KYH LHK TYM TWY
10. 11. 12. 13. 14. 15.
UCS YKY YHY LWS LKY WKK
Seizure duration 5 15 9
Engel Class I/II Engel Class I/II
FCD Mesial temporal sclerosis
Engel Class I/II Engel Class I/II
FCD Mesial temporal sclerosis
Engel Engel Engel Engel Engel
Cavernous haemangioma FCD Gliosis Gliosis No definite pathological diagnosis
Class Class Class Class Class
I/II I/II III/IV III/IV III/IV
Engel Class III/IV Engel Class III/IV Engel Class I/II NA NA NA
Right temporal depth electrode
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
Mesial temporal sclerosis No pathological diagnosis Gliosis and arteriovenous malformation NA NA NA
3
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
Table 1 Characteristics of patients with refractory epilepsy undergoing intracranial implantation. CPS = complex partial seizure, FCD = focal cortical dysplasia, GTCS = generalized tonic–clonic seizures, MRI = magnetic resonance imaging, EEG = electroencephalogram, PET = interictal positron emission tomography, SPECT = ictal single photon emission computed tomography, MTS = mesial temporal sclerosis, NA = not applicable or available.
4
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
2.1. HFO analysis
2.2. Conventional frequency ictal patterns
Each ictal EEG epoch was examined during the joint epilepsy surgery meeting, and the entire ictal recording was exported in ‘‘edf’’ format with an addition of 5 s prior to and 5 s following the episode. One EEG segment without overt artifacts or breaks was analyzed per patient. The EEG was recorded using the contralateral mastoid process as a reference. Thereafter, we created bipolar montages between consecutive electrode points. For example, an 8 4 grid electrode will generate a montage based on electrodes 1–2, 2–3, 3–4, 4–5, 5–6, 6–7, 7–8, 9–10, 10–11, 11–12, etc. This configuration may facilitate direct comparison with CFIP, which is most often read in bipolar montages. Less artifact may be anticipated with bipolar montage, as the reference may display interference due to equipment that is operated on alternate currents (AC) in the vicinity. On the other hand, the use of bipolar montage may entail the following drawbacks: (1) the channels by bipolar montage have to be aligned in a specific manner, and in the current study a horizontal configuration is used rather than the vertical or diagonal configuration. (2) The number of channels would always be less than the number of electrode points (for example, if the number of electrodes is 10, then the total number of channels will only be nine). Wavelet analysis is used for the determination of the frequency of the intracranial signals. The EEG signal, F(t), is subjected to continuous wavelet transform (CWT) with a mother wavelet w. The CWT compares the EEG signal to shifted, compressed, or stretched versions of the wavelet w, the gradation of which may be termed scale (=a). The EEG signals may be reconstructed into CWT coefficients C (a, b), where a is the scale and b the position parameter:
The ictal epochs were reviewed by epileptologists experienced in the interpretation of intracranial EEG. This may be defined as the first registrable change in EEG pattern from the ongoing background prior to, or at the same time as, any demonstrable semiology recorded on video. Its recognition may be facilitated by applying the appropriate low-frequency filter to make the tracing readable with visual analysis. The gain may be set at 100– 500 lV/cm. The following types of patterns are considered typical CFIPs: low-voltage fast activity, low-frequency/high-amplitude periodic spikes, sharp activity 50th percentile) among the resected channels within an individual patient gives a mean of 69% (33.3–100%). The P0 for patients achieving Engel Class III/IV outcome is also depicted in Table 2 (second part). Ictal HFOs were not included in the decision process for any of these patients. By contrast, the proportion of significant ictal HFOs among the resected channels gives a mean of 17.5% (0–50%) for any individual patient. A significantly higher number of resected channels with ictal HFOs was found in the group achieving Engel I/II outcome compared to those achieving Engel III/IV outcome (12.6% vs. 4.2%, p = 0.005) (Table 3, denominator designated to the total number of channels = 370). Multivariate analysis confirmed that resected channels with ictal HFOs were significantly associated with Engel I/II outcome (p = 0.05, relative risk (RR) = 8.1) (Table 3). P0 was not available for patient #12 as there was no recordable ictal event. P0 for patients #13–15 is shown in Table 2 (third part). In these patients, resection was not performed for reasons explained above. The proportion of CFIPs identified among the resected channels for the two groups of patients may be exemplified in Table 2. For patients achieving Engel Class I/II outcome, the mean proportion for any given patient is 68.3% (26.3–100%). This figure may be explained by the fact that some cortical areas that fell within the standard margins of lobectomy may be resected even without CFIP.
For patients achieving Engel Class III/IV outcome, the mean proportion is 90% (50–100%). This figure is anticipated as the majority of the decision-making process was based on CFIP. The statistical test of significance, however, did not reveal any positive result for the comparison between the two groups (11.4% vs. 10%, p = 0.73) (Table 3, the denominator referred to all channels used). The proportion of channels demonstrating the condition of hyperexcitability among the resected channels is shown in Table 2. Among patients achieving Class I/II outcome, the mean proportion is 39.6% (0–100%) for any given patient, and among patients achieving Class III/IV outcome, this value is 12.1% (0–35.7%). The comparison for the two groups is not statistically significant (5.7% vs. 2.8%, p = 0.17, number of channels as denominator). The presence of radiological lesion on MRI is found on average in 52.8% (0–100%) of resected channels for any given patient achieving Engel Class I/II outcome. By comparison, the figure is 43.1% (0–100%) for patients achieving Engel Class III/IV outcome. The comparison is not statistically significant for the two groups (2.5% vs. 8%, p = 0.38, number of channels as denominator).
3.2. The association analysis among ictal HFOs, CFIP, hyperexcitability, and radiological lesion Mathematically, it would be feasible to carry out an analysis of association for all the variables, but from a logical standpoint, we may limit our findings to only that between ictal HFOs and each of CFIP, hyperexcitability, and radiological lesion. Using all available channels (n = 462), we may demonstrate that the number of channels with ictal HFOs co-occurring with CFIP is significantly higher (71.9%, 41/57 vs. 13.5%, 55/405) than those channels whose ictal HFOs do not occur with CFIP (p < 0.001). In a similar fashion, using all available channels, we may show that the proportion of channels with ictal HFOs is higher among those meeting the condition of hyperexcitability (22/57, 38.5%) than those not meeting the condition of hyperexcitability (7/405, 1.7%) (p < 0.001). Third, the proportion of channels with ictal HFOs is higher among those with
Fig. 3. An example of the implantation schedule (patient #1) demonstrating areas with conventional frequency ictal patterns, ictal high-frequency oscillations, hyperexcitability, and radiological lesions.
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
7
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
Fig. 4. An example of the implantation schedule (patient #7) demonstrating areas with conventional frequency ictal patterns, ictal high-frequency oscillations, hyperexcitability, and radiological lesions.
Table 2 Summary table for analysis of ictal HFOs. Patient
Range of P0
Maximal channels
Significant HFOs among resected channels
CFIP among resected channels
Hyperexcitability among resected channels
Radio-logical lesion
Involvement of ictal HFOs in clinical decision making
Analysis 1 2 3 4 5 6
of wavelet power spectrum densities, P0 for patients achieving Engel Class I/II outcome 1-2103.1 PT1-PT2 9/19(47.4%) 5/19(26.3%) 1-13.0 L3-4 4/6(66.7%) 3/6(50%) 1-686.9 53–54 2/3(66.7%) 2/3(66.7%) 1-6.37 A2–A3 1/3(33.3%) 2/3(66.7%) 1-30.4 14–15 2/2(100%) 2/2(100%) 1-142.5 6–7 2/2(100%) 2/2(100%)
4/19(21%) 0/6(0%) 2/3(66.7%) 0/3(0%) 2/2(100%) 1/2(50%)
0/19(0%) 3/6(50%) 0/3(0%) 2/3(66.7%) 2/2(100%) 2/2(100%)
Yes Yes No No No No
Analysis 7 8 9 10 11
of wavelet power spectrum densities, P0 for patients achieving Engel Class III/IV outcome 1-139.9 45–46 7/14(50%) 7/14(50%) 1-265.8 52–53 1/8 (12.5%) 8/8(100%) 1-11.4 LD6-7 0/5(0%) 5/5(100%) 1-82.6 C1-2 1/4 (25%) 4/4(100%) 1-58.3 TP1-2 0/6 (0%) 6/6(100%)
5/14(35.7%) 0/8(0%) 0/5(0%) 1/4(25%) 0/6(0%)
5/14(35.7%) 8/8(100%) 4/5(80%) 0/4(0%) 0/6(0%)
No No No No No
Analysis 13 14 15
of wavelet power spectrum densities, P’ for patients who were not operated 1-134.0 A7-8 NA NA 1-17.7 LD2-3 NA NA 1-44.8 LD4-5 NA NA
NA NA NA
radiological lesions (20/57, 35.1%) than those without radiological lesions (26/405, 6.4%) (p < 0.001) (Table 2). 3.3. Sensitivity and specificity of a combined modality On the utility of any given modality of investigation, using the intention-to-treat method, ictal HFOs may achieve the highest sensitivity while hyperexcitability may attain the highest specificity (Table 4). Combining the two modalities would be a logical option for improving the selection of surgical candidacy. Our results
NA NA NA
suggested that, by using ictal HFOs together with hyperexcitability, the sensitivity can be maintained at 100% (CI 0.52–1), while the specificity may be increased from 66.7% (CI 0.31–0.91) to 75% (CI 0.36–0.96) when compared to using ictal HFOs alone.
4. Discussion In the current study, we have demonstrated that the number of channels containing ictal HFOs within the resection margins were
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
8
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
Table 3 Summary table for statistical analysis. HFO = high frequency oscillations, CFIP = conventional frequency ictal patterns. Factor
Engel Class I/II
Engel Class III/IV
p-Value
Analysis of resected channels from patients achieving Engel Class I/II outcome versus those achieving Engel Class III/IV outcome Ictal HFOs 20/158 (12.6%) 9/212 (4.2%) 0.005 CFIP 18/158 (11.4%) 21/212 (10%) 0.733 Hyperexcitability 9/158 (5.7%) 6/212 (2.8%) 0.17 Radiological 9/158 (2.5%) 17/212 (8%) 0.38 lesion Co-occurring with ictal HFOs Analysis of channels EEG CFIP Hyperexcitability Radiological lesion
Not occurring with ictal HFOs
p-Value
from all patients with ictal HFOs captured on intracranial 41/57 (71.9%) 22/57 (38.5%) 20/57 (35.1%)
55/405 (13.5%) 7/405 (1.7%) 26/405 (6.4%)
Contents lists available at ScienceDirect
Clinical Neurophysiology journal homepage: www.elsevier.com/locate/clinph
Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy Howan Leung a,⇑, Cannon X.L. Zhu b, Danny T.M. Chan b, Wai S. Poon b, Lin Shi a, Vincent C.T. Mok a, Lawrence K.S. Wong a a b
Division of Neurology, Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Hong Kong Special Administrative Region Division of Neurosurgery, Department of Surgery, The Chinese University of Hong Kong, Hong Kong Special Administrative Region
a r t i c l e
i n f o
Article history: Accepted 1 January 2015 Available online xxxx Keywords: High-frequency oscillations Hyperexcitability Refractory epilepsy
h i g h l i g h t s High-frequency oscillations captured at the onset of seizure may help determine surgical outcome in
patients with refractory epilepsy. Cortical areas demonstrating hyperexcitability may be associated with ictal high-frequency oscillations. By examining hyperexcitability and ictal high-frequency oscillations, the decision-making process for
surgical resection to treat refractory epilepsy may be improved with values of sensitivity and specificity that are more optimal.
a b s t r a c t Objective: High-frequency oscillations (HFOs, 80–500 Hz) from intracranial electroencephalography (EEG) may represent a biomarker of epileptogenicity for epilepsy. We explored the relationship between ictal HFOs and hyperexcitability with a view to improving surgical outcome. Methods: We evaluated 262 patients with refractory epilepsy. Fifteen patients underwent electrode implantation, and surgical resection was performed in 12 patients using a semi-prospective design. Ictal intracranial EEGs were examined by continuous wavelet transform (CWT). Significant ictal HFOs were denoted by normalized wavelet power above the 50th percentile across all channels. Each patient underwent functional mapping with cortical electrical stimulation. Hyperexcitability was defined as the appearance of afterdischarges or clinical seizures after electrical stimulation (50 Hz, biphasic, pulse width = 0.5 ms, 5 s, 5 mA). Results: Among the group of patients achieving Engel Class I/II outcome at 1+ year, the mean proportion of significant ictal HFOs among resected channels for any given patient was 69% (33.3–100%). The respective figures for conventional frequency ictal patterns (CFIPs), hyperexcitability, and radiological lesion were 68.3% (26.3–100%), 39.6% (0–100%), and 52.8% (0–100%). Statistical significance was only achieved with ictal HFOs when comparing patients with Engel Class I/II outcomes versus III/IV outcomes (12.6% vs. 4.2%, the number of channels as the denominator, p = 0.005). Results: Further analysis from all patients irrespective of the surgical outcome showed that ictal HFOs co-occurred with CFIP (p < 0.001), hyperexcitability (p < 0.001), and radiological lesion (p < 0.001). The combination of ictal HFOs/hyperexcitability improved the sensitivity from 66.7% to 100%, and the specificity from 66.7% to 75% when compared with ictal HFOs or hyperexcitability alone. Conclusions: We confirmed the utility of ictal HFOs in determining surgical outcome. Ictal HFOs are affiliated to cortical hyperexcitability, which may represent a pathological manifestation of epileptogenicity. Significance: Presurgical evaluation of refractory epilepsy may incorporate both ictal HFOs and cortical stimulation in determining epileptogenic foci. Ó 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
⇑ Corresponding author at: 10/F, Clinical Science Building, Prince of Wales Hospital, Shatin NT, Hong Kong Special Administrative Region. Tel.: +852 2632 1855; fax: +852 2637 3852. E-mail address: [email protected] (H. Leung). http://dx.doi.org/10.1016/j.clinph.2015.01.009 1388-2457/Ó 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
2
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
1. Introduction The challenge of treating refractory epilepsy with surgery begins with the exploration of the epileptogenic zone, defined as ‘‘the minimal area of cortex that must be resected to produce seizure freedom’’ (Luders et al., 1993). This apparent basis for presurgical evaluation has fueled research on multimodal investigations that identify or define the electrographic details to which epileptogenic zones conform. In 1954, Penfield and Jasper had already used the concepts of interictal corticography to establish the limits of the epileptogenic zone. By 1962, Talairach and Bancaud introduced the technique of stereo-electroencephalography to study ictal events and advance knowledge of the epileptogenic zone. Today, high-frequency oscillations (HFOs) are added to the armamentarium and hailed as a new biomarker to signify epileptogenesis. We have a far greater number of tools with which to investigate patients with refractory epilepsy compared to what was available in the era of Penfield and Jasper, who meticulously adopted cortical stimulation to reproduce aura and other ictal manifestations. The concept of hyperexcitability was a hallmark of epilepsy from the days of the forerunners in epilepsy research, but it can still be applied today in presurgical evaluation in the setting of intracranial electrode implantation. HFOs in the range of 80–500 Hz have been identified and classified (Buzsaki et al., 1992) for operational reasons, and their utility in seizure-onset identification has been demonstrated in animal models of epilepsy (Bragin et al., 1999, 2003), and later in intracranial microelectrodes from hippocampal and entorhinal cortices of patients with epilepsy (Bragin et al., 2005; Staba et al., 2002, 2004) and those with neocortical epilepsy (Weiss et al., 2013). These results were replicable with intracranial macroelectrodes (Jirsch et al., 2006; Urrestarazu et al., 2007). We are interested in the analysis of ictal HFOs, as the literature on this aspect of HFO is relatively limited. The traditional method of determining an intracranial seizure-onset zone using conventional frequency ictal patterns (CFIP) suffers from a number of drawbacks: (1) difficulty in deciphering which of the onset patterns may assume superior significance, (2) proximity in time sequence between similar onset patterns of adjacent channels, and (3) problems encountered with rapid spread of discharges. Fujiwara and researchers also pointed out the following: (4) the surface findings of grid electrodes may not indicate what may be happening in deeper cortical sources near the bottom of the sulci, (5) the depth electrode is often impractical and only feasible for a small area of enquiry, (6) epileptogenic areas with a tangential orientation either have undetectable amplitudes or are too widespread, and (7) electrodes may have been placed in such a way that the visual analysis of ictal patterns could be misleading (Fujiwara et al., 2012). In a study of patients with temporal lobe epilepsy using intracranial EEG, the spatial and time-related properties of HFOs were explored. A spatial correlation of HFO with a seizure-onset zone was observed, and the researchers found the presence of 100–200 Hz HFOs at least 8 s before seizure (Khosravani et al., 2009). In a pediatric study of nine children with neocortical epilepsy, both wide-band (250 Hz) and narrow-band (68–164 Hz) HFOs were registered. In those patients achieving seizure freedom, more electrodes recorded HFOs inside the resection margin than outside, both before and after clinical seizure onset (Ochi et al., 2007). A study of six patients with neocortical epilepsy showed that ictal HFOs, manifested as frequencies P70 Hz with sustained evolution, had a higher peak frequency and more spatially restricted appearance, and they were 10 times more likely to be resected than ictal HFOs without evolution (Modur et al., 2011). In a study that examined intracranial EEG patterns and radiological lesions, HFOs were significantly associated with the developing patterns at seizure onset, regardless of the patterns themselves (Perucca et al., 2014).
The relationship between hyperexcitability and HFOs was explored in one study, which examined interictal HFO and electrical stimulation of areas using intracranial macroelectrodes. The EEG segments consisted of 5 min of slow-wave sleep in patients with intracranial implantation. Interictal HFO rates were negatively correlated with thresholds for response to electrical stimulation, lending support to the notion that HFO and hyperexcitability may share a similar mechanistic platform (Jacobs et al., 2010a,b). We hypothesize the following points in our current study: (1) Among patients who attain satisfactory surgical outcomes, a higher proportion of the resected channels demonstrate significant ictal HFOs. A test of significance will be carried out with CFIP, hyperexcitability, and radiological lesion using the surgical outcome. (2) Ictal HFOs are associated significantly with CFIP, hyperexcitability, and radiological lesion using all available channels. (3) The sensitivity and specificity of using ictal HFOs together with hyperexcitability in determining surgical outcome may be superior to ictal HFOs, CFIP, hyperexcitability, or radiological lesion alone.
2. Methods We prospectively evaluated 262 patients with refractory epilepsy at a university-affiliated hospital during the period between 2007 and 2012. Eighty patients were eligible for direct resective surgery, and another 55 patients had sufficient information to localize or lateralize a potential area of resection based on MRI, surface video EEG, clinical psychological testing, positron emission tomography (PET), single photon emission computed tomography (SPECT), or Wada test, i.e. a testable hypothesis formulated for implantation of an intracranial electrode. The inclusion criteria were as follows: (1) age P18 and (2) refractory epilepsy with an implantation hypothesis. A typical implantation schedule would consist of grid electrodes to sample the neocortical areas and depth electrodes to sample the mesial temporal regions. The exclusion criteria were as follows: (1) inability to provide written informed consent and (2) lack of ictal episodes being captured during intracranial monitoring. Fifteen patients underwent intracranial EEG when the technical placement of electrodes could be resolved and financed. At the time of recruitment for the study, intracranial EEG was not a fundable item under the Hospital Authority of Hong Kong, so patients required additional financial support to undergo this part of their treatment. The patient characteristics are given in Table 1. We used the NicoletÒ machine (Viasys, Santa Ana, CA, USA) for video electroencephalographic monitoring with a minimum sampling rate of 1024 Hz. The intracranial electrodes used met the standard for impedance and stability with platinum contacts of 4.0-mm diameter and 2.3-mm exposure (Ad-Tech, Irving, TX, USA). The ictal recordings from intracranial EEG were analyzed by visual analysis of CFIP, and the final plan for resection was made during a joint epilepsy surgery meeting (HL, XZ, DC, WP, and KW), after consideration of both electrographic information and the results of functional mapping. Inpatient intracranial EEG monitoring may last between 7 and 14 days. Off-line analysis of ictal recordings was carried out using a Matlab-based platform by an independent assessor blinded to the clinical information (SL). In the wake of 2011, we considered ictal HFOs as part of the evidence in the discussion leading up to resective surgery. A minimum follow-up period of 1 year is required for all patients to determine their surgical outcome. Surgical outcome was determined using the Engel classification for epilepsy surgery (Engel, 1993).
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
Patient
Age
Sex
1. IK
35
M
2. CFK
26
F
3. LIF
25
M
4 WKM
38
F
5. LMW
34
6. KYH
Seizure type
MRI
Ictal surface EEG
PET
SPECT
Implantation
CPS
Normal
Left temporal
NA
Left temporal grid and depth electrode
CPS, GTCS
Right MTS
Bilateral temporal onset
NA
Bilateral depth electrode
SPS, CPS
Normal
Bi-frontal
CPS
Bi-lateral MTS
Right temporal
Left mid frontal hyper-perfusion NA
Left frontal grids and strips
26
F
15
CPS, GTCS
Left temporal cavernoma
Left temporal
NA
Left temporal grid + functional mapping
33
M
10
CPS, GTCS
Right frontal
NA
Right frontal grid
7. LHK 8. TYM
37 31
M F
20 15
CPS, GTCS CPS, GTCS
Left frontal Left lateralization
NA NA
NA NA
9. TWY 10. UCS 11. YKY 12. YHY
41 25 33 29
F M F M
12 10 18 10
CPS CPS, GTCS CPS, GTCS CPS, GTCS
Left temporal Bilateral temporal Left temporal Right frontal
Left temporal Left temporal Left temporal NA
NA Left temporal NA Right frontal
Left frontal grid Left occipital, parietal grids and inter-hemispheric strips Bilateral depth electrodes Left temporal grids and strips Left temporal grid and strips Right frontal grid and inter-hemispheric strips
13. LWS 14. LKY 15. WKK Patient
38 F 27 F 39 M Resection
22 4 34
CPS CPS, GTCS CPS, GTCS
Right frontal focal cortical dysplasia Left frontal encephalomalacia Left occipital encephalomalacia Bilateral MTS Normal Normal Right frontal encephalomalacia Normal Bilateral MTS Bilateral MTS Surgical outcome
Left temporal hypo-metabolism Right temporal hypo-metabolism Left mid-frontal hypo-metabolism Right temporal hypo-metabolism Left temporal hypometabolism NA
Right frontal Bilateral temporal Right temporal
NA Bilateral temporal Normal Pathology
Right frontal NA Normal
Right frontal grid Bilateral depth electrode Bilateral depth electrode
1. IK 2. CFK
Left anterior temporal lobectomy and topectomy Right anterior temporal lobectomy and hippocampectomy Left frontal topectomy Right anterior temporal lobectomy and hippocampectomy Left temporal lesionectomy Right frontal lesionectomy Left frontal lesionectomy Left occipital lesionectomy Right temporal lobectomy and hippocampectomy (based on CFIP only) Left temporal lobectomy and hippocampectomy Left temporal lobectomy and hippocampectomy right frontal lesionectomy NA NA NA
3. LIF 4. WKM 5. 6. 7. 8. 9.
LMW KYH LHK TYM TWY
10. 11. 12. 13. 14. 15.
UCS YKY YHY LWS LKY WKK
Seizure duration 5 15 9
Engel Class I/II Engel Class I/II
FCD Mesial temporal sclerosis
Engel Class I/II Engel Class I/II
FCD Mesial temporal sclerosis
Engel Engel Engel Engel Engel
Cavernous haemangioma FCD Gliosis Gliosis No definite pathological diagnosis
Class Class Class Class Class
I/II I/II III/IV III/IV III/IV
Engel Class III/IV Engel Class III/IV Engel Class I/II NA NA NA
Right temporal depth electrode
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
Mesial temporal sclerosis No pathological diagnosis Gliosis and arteriovenous malformation NA NA NA
3
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
Table 1 Characteristics of patients with refractory epilepsy undergoing intracranial implantation. CPS = complex partial seizure, FCD = focal cortical dysplasia, GTCS = generalized tonic–clonic seizures, MRI = magnetic resonance imaging, EEG = electroencephalogram, PET = interictal positron emission tomography, SPECT = ictal single photon emission computed tomography, MTS = mesial temporal sclerosis, NA = not applicable or available.
4
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
2.1. HFO analysis
2.2. Conventional frequency ictal patterns
Each ictal EEG epoch was examined during the joint epilepsy surgery meeting, and the entire ictal recording was exported in ‘‘edf’’ format with an addition of 5 s prior to and 5 s following the episode. One EEG segment without overt artifacts or breaks was analyzed per patient. The EEG was recorded using the contralateral mastoid process as a reference. Thereafter, we created bipolar montages between consecutive electrode points. For example, an 8 4 grid electrode will generate a montage based on electrodes 1–2, 2–3, 3–4, 4–5, 5–6, 6–7, 7–8, 9–10, 10–11, 11–12, etc. This configuration may facilitate direct comparison with CFIP, which is most often read in bipolar montages. Less artifact may be anticipated with bipolar montage, as the reference may display interference due to equipment that is operated on alternate currents (AC) in the vicinity. On the other hand, the use of bipolar montage may entail the following drawbacks: (1) the channels by bipolar montage have to be aligned in a specific manner, and in the current study a horizontal configuration is used rather than the vertical or diagonal configuration. (2) The number of channels would always be less than the number of electrode points (for example, if the number of electrodes is 10, then the total number of channels will only be nine). Wavelet analysis is used for the determination of the frequency of the intracranial signals. The EEG signal, F(t), is subjected to continuous wavelet transform (CWT) with a mother wavelet w. The CWT compares the EEG signal to shifted, compressed, or stretched versions of the wavelet w, the gradation of which may be termed scale (=a). The EEG signals may be reconstructed into CWT coefficients C (a, b), where a is the scale and b the position parameter:
The ictal epochs were reviewed by epileptologists experienced in the interpretation of intracranial EEG. This may be defined as the first registrable change in EEG pattern from the ongoing background prior to, or at the same time as, any demonstrable semiology recorded on video. Its recognition may be facilitated by applying the appropriate low-frequency filter to make the tracing readable with visual analysis. The gain may be set at 100– 500 lV/cm. The following types of patterns are considered typical CFIPs: low-voltage fast activity, low-frequency/high-amplitude periodic spikes, sharp activity 50th percentile) among the resected channels within an individual patient gives a mean of 69% (33.3–100%). The P0 for patients achieving Engel Class III/IV outcome is also depicted in Table 2 (second part). Ictal HFOs were not included in the decision process for any of these patients. By contrast, the proportion of significant ictal HFOs among the resected channels gives a mean of 17.5% (0–50%) for any individual patient. A significantly higher number of resected channels with ictal HFOs was found in the group achieving Engel I/II outcome compared to those achieving Engel III/IV outcome (12.6% vs. 4.2%, p = 0.005) (Table 3, denominator designated to the total number of channels = 370). Multivariate analysis confirmed that resected channels with ictal HFOs were significantly associated with Engel I/II outcome (p = 0.05, relative risk (RR) = 8.1) (Table 3). P0 was not available for patient #12 as there was no recordable ictal event. P0 for patients #13–15 is shown in Table 2 (third part). In these patients, resection was not performed for reasons explained above. The proportion of CFIPs identified among the resected channels for the two groups of patients may be exemplified in Table 2. For patients achieving Engel Class I/II outcome, the mean proportion for any given patient is 68.3% (26.3–100%). This figure may be explained by the fact that some cortical areas that fell within the standard margins of lobectomy may be resected even without CFIP.
For patients achieving Engel Class III/IV outcome, the mean proportion is 90% (50–100%). This figure is anticipated as the majority of the decision-making process was based on CFIP. The statistical test of significance, however, did not reveal any positive result for the comparison between the two groups (11.4% vs. 10%, p = 0.73) (Table 3, the denominator referred to all channels used). The proportion of channels demonstrating the condition of hyperexcitability among the resected channels is shown in Table 2. Among patients achieving Class I/II outcome, the mean proportion is 39.6% (0–100%) for any given patient, and among patients achieving Class III/IV outcome, this value is 12.1% (0–35.7%). The comparison for the two groups is not statistically significant (5.7% vs. 2.8%, p = 0.17, number of channels as denominator). The presence of radiological lesion on MRI is found on average in 52.8% (0–100%) of resected channels for any given patient achieving Engel Class I/II outcome. By comparison, the figure is 43.1% (0–100%) for patients achieving Engel Class III/IV outcome. The comparison is not statistically significant for the two groups (2.5% vs. 8%, p = 0.38, number of channels as denominator).
3.2. The association analysis among ictal HFOs, CFIP, hyperexcitability, and radiological lesion Mathematically, it would be feasible to carry out an analysis of association for all the variables, but from a logical standpoint, we may limit our findings to only that between ictal HFOs and each of CFIP, hyperexcitability, and radiological lesion. Using all available channels (n = 462), we may demonstrate that the number of channels with ictal HFOs co-occurring with CFIP is significantly higher (71.9%, 41/57 vs. 13.5%, 55/405) than those channels whose ictal HFOs do not occur with CFIP (p < 0.001). In a similar fashion, using all available channels, we may show that the proportion of channels with ictal HFOs is higher among those meeting the condition of hyperexcitability (22/57, 38.5%) than those not meeting the condition of hyperexcitability (7/405, 1.7%) (p < 0.001). Third, the proportion of channels with ictal HFOs is higher among those with
Fig. 3. An example of the implantation schedule (patient #1) demonstrating areas with conventional frequency ictal patterns, ictal high-frequency oscillations, hyperexcitability, and radiological lesions.
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
7
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
Fig. 4. An example of the implantation schedule (patient #7) demonstrating areas with conventional frequency ictal patterns, ictal high-frequency oscillations, hyperexcitability, and radiological lesions.
Table 2 Summary table for analysis of ictal HFOs. Patient
Range of P0
Maximal channels
Significant HFOs among resected channels
CFIP among resected channels
Hyperexcitability among resected channels
Radio-logical lesion
Involvement of ictal HFOs in clinical decision making
Analysis 1 2 3 4 5 6
of wavelet power spectrum densities, P0 for patients achieving Engel Class I/II outcome 1-2103.1 PT1-PT2 9/19(47.4%) 5/19(26.3%) 1-13.0 L3-4 4/6(66.7%) 3/6(50%) 1-686.9 53–54 2/3(66.7%) 2/3(66.7%) 1-6.37 A2–A3 1/3(33.3%) 2/3(66.7%) 1-30.4 14–15 2/2(100%) 2/2(100%) 1-142.5 6–7 2/2(100%) 2/2(100%)
4/19(21%) 0/6(0%) 2/3(66.7%) 0/3(0%) 2/2(100%) 1/2(50%)
0/19(0%) 3/6(50%) 0/3(0%) 2/3(66.7%) 2/2(100%) 2/2(100%)
Yes Yes No No No No
Analysis 7 8 9 10 11
of wavelet power spectrum densities, P0 for patients achieving Engel Class III/IV outcome 1-139.9 45–46 7/14(50%) 7/14(50%) 1-265.8 52–53 1/8 (12.5%) 8/8(100%) 1-11.4 LD6-7 0/5(0%) 5/5(100%) 1-82.6 C1-2 1/4 (25%) 4/4(100%) 1-58.3 TP1-2 0/6 (0%) 6/6(100%)
5/14(35.7%) 0/8(0%) 0/5(0%) 1/4(25%) 0/6(0%)
5/14(35.7%) 8/8(100%) 4/5(80%) 0/4(0%) 0/6(0%)
No No No No No
Analysis 13 14 15
of wavelet power spectrum densities, P’ for patients who were not operated 1-134.0 A7-8 NA NA 1-17.7 LD2-3 NA NA 1-44.8 LD4-5 NA NA
NA NA NA
radiological lesions (20/57, 35.1%) than those without radiological lesions (26/405, 6.4%) (p < 0.001) (Table 2). 3.3. Sensitivity and specificity of a combined modality On the utility of any given modality of investigation, using the intention-to-treat method, ictal HFOs may achieve the highest sensitivity while hyperexcitability may attain the highest specificity (Table 4). Combining the two modalities would be a logical option for improving the selection of surgical candidacy. Our results
NA NA NA
suggested that, by using ictal HFOs together with hyperexcitability, the sensitivity can be maintained at 100% (CI 0.52–1), while the specificity may be increased from 66.7% (CI 0.31–0.91) to 75% (CI 0.36–0.96) when compared to using ictal HFOs alone.
4. Discussion In the current study, we have demonstrated that the number of channels containing ictal HFOs within the resection margins were
Please cite this article in press as: Leung H et al. Ictal high-frequency oscillations and hyperexcitability in refractory epilepsy. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.01.009
8
H. Leung et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
Table 3 Summary table for statistical analysis. HFO = high frequency oscillations, CFIP = conventional frequency ictal patterns. Factor
Engel Class I/II
Engel Class III/IV
p-Value
Analysis of resected channels from patients achieving Engel Class I/II outcome versus those achieving Engel Class III/IV outcome Ictal HFOs 20/158 (12.6%) 9/212 (4.2%) 0.005 CFIP 18/158 (11.4%) 21/212 (10%) 0.733 Hyperexcitability 9/158 (5.7%) 6/212 (2.8%) 0.17 Radiological 9/158 (2.5%) 17/212 (8%) 0.38 lesion Co-occurring with ictal HFOs Analysis of channels EEG CFIP Hyperexcitability Radiological lesion
Not occurring with ictal HFOs
p-Value
from all patients with ictal HFOs captured on intracranial 41/57 (71.9%) 22/57 (38.5%) 20/57 (35.1%)
55/405 (13.5%) 7/405 (1.7%) 26/405 (6.4%)
