Clinical Neurophysiology xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Clinical Neurophysiology journal homepage: www.elsevier.com/locate/clinph
Relationships between interictal epileptic spikes and ripples in surface EEG Nicole van Klink a,b,⇑, Birgit Frauscher a, Maeike Zijlmans b,c, Jean Gotman a a
Montreal Neurological Institute, McGill University, Montreal, Canada Brain Center Rudolf Magnus, Department of Neurology and Neurosurgery, University Medical Center Utrecht, The Netherlands c SEIN – Stichting Epilepsie Instellingen Nederland, Heemstede, The Netherlands b
a r t i c l e
i n f o
Article history: Accepted 16 April 2015 Available online xxxx Keywords: Electroencephalography Epilepsy Non-invasive EEG Ripples Spikes High frequency oscillations
h i g h l i g h t s 64% of the ripples start before the spike starts. Spikes with ripples are on average shorter, and have higher amplitude and slope. Ripples in surface EEG have a smaller spatial spread than spikes.
a b s t r a c t Objective: Ripples (80–250 Hz) have been shown to be a more specific biomarker for the epileptogenic zone than epileptic spikes in intracranial EEG and even surface EEG. Ripples often co-occur with spikes. We investigated the spatiotemporal relation between spikes and ripples, and differences between spikes that do and do not co-occur with ripples. Methods: We marked 50 time points with spikes in bipolar surface EEG during NREM sleep in patients with focal or multifocal epilepsy. We marked ripples that occurred with spikes and calculated parameters relating spikes and ripples: the duration, amplitude and slope of spikes, the timing of the start of ripples and spikes and the proportion of overlap. Results: In total 219 ripples and 5995 individual spikes were marked in 31 patients. Spikes with ripples were on average shorter, had higher amplitude and higher slope than spikes without ripples. 64% of ripples started before spikes started. Spikes occurred on 13 (5–26) channels per patient, and ripples on 3 (0–14) channels, which were also spike channels. Conclusion: Ripples precede rather than follow spikes, so ripples are unlikely to result from spikes. Significance: Ripples and spikes seem not one-on-one coupled, but certain states of the brain can accommodate both. Ó 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
1. Introduction High frequency oscillations (HFOs, >80 Hz) in invasive intracranial recordings are newly proposed biomarkers for the brain area that generates seizures (Zijlmans et al., 2012; Jacobs et al., 2012; Bragin et al., 1999). HFOs are more specific for defining the epileptogenic zone than spikes (Jacobs et al., 2009). In line with this finding, they seem to be more related to disease activity (Zijlmans
⇑ Corresponding author at: University Medical Center Utrecht, Department of Neurology and Neurosurgery, HP C03.131, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands. Tel.: +31 887557735. E-mail address: [email protected] (N. van Klink).
et al., 2009, 2011) and possibly predict outcome after epilepsy surgery (Jacobs et al., 2010; Van’t Klooster et al., 2015). HFOs between 80 and 250 Hz, called ripples, can also be found in the non-invasive surface electroencephalogram (EEG). This increases the population for whom HFO analysis is available, beyond the small group of drug-resistant focal epilepsy patients who are candidates for surgery with intracerebral electrodes. Reports describe 50–100 Hz activity in ictal recordings (Kobayashi et al., 2004, 2009), ripples in interictal EEG in adults with focal epilepsy (Andrade-Valenca et al., 2011; Melani et al., 2013; Zelmann et al., 2013), in children with continuous spike and wave in slow-wave sleep (Kobayashi et al., 2010), and idiopathic partial epilepsy (Kobayashi et al., 2011). A recent study showed that interictal surface ripples decrease after treatment
http://dx.doi.org/10.1016/j.clinph.2015.04.059 1388-2457/Ó 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: van Klink N et al. Relationships between interictal epileptic spikes and ripples in surface EEG. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.04.059
2
N. van Klink et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
with ACTH in children with West Syndrome (Kobayashi et al., 2015). Co-occurrence of ripples with epileptic spikes is reported in 44–64% of all ripples in intracranial and surface recordings (Melani et al., 2013; Andrade-Valenca et al., 2011; Urrestarazu et al., 2007; Jacobs et al., 2009; Wang et al., 2013). Ripples co-occurring with a spike are more related to the seizure onset zone than ripples without a spike (Wang et al., 2013). There seems to be a relation between ripples and spikes, but so far no study addressed this important aspect in detail. Knowledge about this relation could help to get insight into the mechanisms of ripple generation and the clinical relevance of the coupling between ripples and spikes. Our goal was to further investigate the relationship between ripples and epileptic spikes in non-invasive surface EEG. The two main research questions were: Are there differences between spikes that show and those that do not show a ripple at the same time? What is the temporal and spatial relation between spikes and ripples? 2. Methods 2.1. Patients We included adult patients with focal or multifocal epilepsy who were admitted to the EEG-telemetry unit of the Montreal Neurological Hospital, and showed more than one epileptic spike or sharp wave per minute during N2 or N3 sleep. Patients who only showed polyspikes were excluded because of the difficulty in assessing the relationship between the ripple and the spike. We consecutively included patients from September 2014 back in time until more than a total number of 200 ripples were marked, a number which we thought was sufficient to assess spike-ripple relationship. All patients gave informed consent in agreement with the Research Ethics Board of the Montreal Neurological Institute and Hospital. 2.2. EEG recording, data selection and event marking Recordings were performed with the Harmonie system (Stellate, Montreal, Canada) with electrodes placed according to the 10–20 EEG system, with additional zygomatic and F9/F10, T9/T10 and P9/P10 electrodes and FCz as reference. The low pass filter was set at 300 Hz and the sampling frequency at 1000 Hz. We used the recording of the second night after admission. Typically the whole night (9 pm–9 am) was recorded. We selected epochs of N2 and N3 sleep. Sleep stages were marked according to the American Academy of Sleep Medicine (AASM) criteria (Berry et al., 2012) by one reviewer and checked and discussed with a second reviewer. We manually marked 50 consecutive time points with spikes in each patient in N2 or N3 sleep, starting at the first sleep cycle of that night. Epochs within two hours before or after a seizure were excluded. The first 50 time points with spikes were marked (Fig. 1, vertical lines); they usually occurred in the first sleep cycle, but could extend beyond. The length of the epochs therefore differed per patient, because it depended on the spike rate. The spikes were marked in a bipolar montage with a 15 s/page time scale and a 10 lV/mm amplitude scale. Spikes were marked on the individual channels as well, to allow for comparison with ripples on specific channels (Fig. 1, horizontal lines). Only isolated spikes were marked: spikes occurring within 1 s from a marked spike were excluded, as were spikes including artefacts. Subsequently, ripples were manually marked in a 400 ms time window around each spike in a bipolar montage with an 80 Hz finite impulse response (FIR) high pass filter, at a 1.5 s/page time scale and a 1 lV/mm amplitude scale. Ripples were marked by one reviewer and checked and discussed with a second reviewer. We defined a ripple as at least four oscillations above 80 Hz, which
were clearly distinct from the background EEG. We did not use a maximum number of oscillations or maximum duration because surface EEG ripples are generally short. Ripples with irregular frequency or amplitude morphology were suspicious to be artefacts and were therefore not marked. 2.3. Data analysis All statistics were calculated with IBM SPSS Statistics 22 (IBM Corp., Armonk, NY, USA). A p-value R Diffuse volume loss
1 9 0
319 115 160
8 9 10 11
46/M 76/F 45/M 26/F
36 56 38 14
CPS + GTCS GTCS GTCS CPS
L FT NA LT NA
Bilateral F (18) L T (11) Bilateral T + F (19) L FT (7) L FT (18) L T, R FT (19) L C (12)
L TPO (7) Bilateral F + O (14) L F (1) L F (6) N N N N N
0 0 0 0
185 298 155 138
12
53/M
3
L FT
L T (13)
N
0
166
13 14
23/F 18/M
9 17
SPS + CPS + rare GTCS CPS and GTCS CPS, rare GTCS
No abnormalities L hippocampal atrophy Bilateral hippocampal malrotation + L T atrophy Parry-Romberg syndrome. L cerebral atrophy. History: 2 L F resections. L hypothalamic tumor lesion
LT RF
L FT, rare R T (8) R T, few L T (17)
L T (5) N
10 0
188 121
15 16 17 18
39/F 53/M 60/M 50/M
15 37 58 2
CPS, rare GTCS CPS, rare GTCS CPS CPS and GTCS
L FT RT NA LT
L T (9) R T (7) L FT (9) L T, few R T (21)
L F (2) N L FT (3) L FC (3)
2 0 5 4
149 170 198 190
19 20
40/M 35/F
11 30
CPS CPS
R FT RT
R T, few L T (13) R T (8)
R T + L P (4) R T (3)
5 8
217 163
21 22 23
30/F 24/F 26/M
13 15 0
CPS and GTCS CPS and GTCS CPS
L F, rare R F (13) R FT (10) Bilateral F (17)
158 222 268
39/M
26
CPS, rare GTCS
L FC (4) R FT (5) Bilateral F (6) N
7 14 23
24
0
180
25 26
60/F 57/F
26 13
CPS, rare GTCS GTCS + absences
L FC NA Bilateral F Bilateral T RT L FT
L O DNET. History: partial resection DNET R F middle sulcus suspect FCD + R thamalus sequelae ischemia L PH gyrus lesion + pituitary cyst R T anterior cavernoma L MTL lesion R lateral ventricle mass + ventriculomegaly + diffuse white matter disturbances No abnormalities R P and T encephalomalacia. History: R TP astrocytoma lesionectomy, no recurrence. L F precentral gyrus suspect FCD R F anterior FCD + L T nodular heterotopia L MCA sequelae encephalitis
5 19
157 228
27
42/M
8
CPS, rare GTCS
L FT
L F, R T (12)
R FT (4) Bilateral FT (11) N
0
123
28 29 30
21/F 21/M 25/F
2 10 9
CPS, rare GTCS CPS CPS and GTCS
R T (20) R P, R T (20) Bilateral T (17)
N R T > L T (6) L T > R T (6)
0 11 23
176 262 217
31
52/M
NA
NA
RT R T>L T Bilateral T NA
Bilateral T (17)
Bilateral F (4)
6
174
Bilateral T (15) R T (5) Bilateral T(26)
R hippocampal atrophy R T atrophy + R hippocampal sclerosis No abnormalities. History: L F resection and callosotomy R F precentral gyrus hyperintensity. History: 3 L T, F and OF resections of FCD Bilateral hippocampal atrophy, R > L R PH, TO and cingulate white matter blurring R T and hippocampal atrophy + R T middle gyrus suspect FCD Bilateral F gyri and gyri recti contusions
Spike focus EEG (#Chann) = location of most of the interictal epileptiform discharges, with the total number of channels involved. Ripple (#Chann) = location where most marked ripples occurred, with the total number of channels involved. MRI = MRI findings. #Ripple = number of marked ripples. #Spikes = number of individual marked spike. F = female, M = male, CPS = complex partial seizures, GTCS = generalized tonic clonic seizures, SPS = simple partial seizures, L = left, R = right, F = frontal, T = temporal, C = central, P = parietal, O = occipital, FT = fronto-temporal, FC = fronto-central, TPO = temporo-parieto-occipital, MTL = mesial temporal lobe, DNET = dysembryoplastic neuroepithelial tumor, FCD = focal cortical dysplasia, PH = parahippocampal, NA = not available.
from ripples with spike (median with spike 56 ms (48–70), without spike 57 ms (49–67); Z = 0.20, p = 0.844).
4. Discussion We studied the differences between surface EEG spikes that do and do not co-occur with ripples, and found in a group comparison that spikes with ripples have a shorter duration, higher amplitude and higher slope than spikes without ripples. There is, however, no clear cut-off value for these characteristics which can separate spikes with from spikes without ripples. We also studied the temporal relation between spikes and ripples, and found that the majority of ripples start before the onset of the spike and that ripples rarely extend beyond the spike. Ripples occur on fewer channels than spikes, and the ripple channels are among the spike channels. The relation between ripples and spikes on one channel are not tightly time-locked, since ripples disappeared when spikes
were averaged. Our findings indicate that ripples on a spike are not ripples evoked by a spike. The difference between spikes with or without ripples has not been studied so far, although it has been suggested that ripples might discriminate between clinically important ‘red’ spikes and nonspecific, propagated ‘green’ spikes (Engel et al., 2009). The common thought that the sharper, higher amplitude spikes are more specific for the epileptogenic zone and co-occur with ripples more often than sharp waves is generally confirmed in this study, but not all sharp spikes co-occurred with ripples, and non-sharp spikes with ripples also existed. The possibility to detect ripples in non-invasive EEG is an important finding for the use of ripples in clinical practice. One of the critical issues for reliable detection is how to identify if oscillations arise from the brain or are related to filtering of sharp transients (Bénar et al., 2010). When a ripple is a filter artefact of a spike, the ripple is expected to be centered on the peak of the spike. Most of the ripples in this study occur for the main part
Please cite this article in press as: van Klink N et al. Relationships between interictal epileptic spikes and ripples in surface EEG. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.04.059
5
N. van Klink et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
Fig. 3. Examples of spikes with (A and C) and without (B and D) ripples. Both spikes with short and long duration can co-occur with a ripple.
Table 2 Spike characteristics for spikes with and without ripple. Spike with ripple
Spike without ripple
p-value
Duration (ms) -FHW -SHW
86.1 (52.5 30.4 (17.5 51.6 (27.4
94.0 (71.9 35.9 (25.8 54.4 (41.5
Contents lists available at ScienceDirect
Clinical Neurophysiology journal homepage: www.elsevier.com/locate/clinph
Relationships between interictal epileptic spikes and ripples in surface EEG Nicole van Klink a,b,⇑, Birgit Frauscher a, Maeike Zijlmans b,c, Jean Gotman a a
Montreal Neurological Institute, McGill University, Montreal, Canada Brain Center Rudolf Magnus, Department of Neurology and Neurosurgery, University Medical Center Utrecht, The Netherlands c SEIN – Stichting Epilepsie Instellingen Nederland, Heemstede, The Netherlands b
a r t i c l e
i n f o
Article history: Accepted 16 April 2015 Available online xxxx Keywords: Electroencephalography Epilepsy Non-invasive EEG Ripples Spikes High frequency oscillations
h i g h l i g h t s 64% of the ripples start before the spike starts. Spikes with ripples are on average shorter, and have higher amplitude and slope. Ripples in surface EEG have a smaller spatial spread than spikes.
a b s t r a c t Objective: Ripples (80–250 Hz) have been shown to be a more specific biomarker for the epileptogenic zone than epileptic spikes in intracranial EEG and even surface EEG. Ripples often co-occur with spikes. We investigated the spatiotemporal relation between spikes and ripples, and differences between spikes that do and do not co-occur with ripples. Methods: We marked 50 time points with spikes in bipolar surface EEG during NREM sleep in patients with focal or multifocal epilepsy. We marked ripples that occurred with spikes and calculated parameters relating spikes and ripples: the duration, amplitude and slope of spikes, the timing of the start of ripples and spikes and the proportion of overlap. Results: In total 219 ripples and 5995 individual spikes were marked in 31 patients. Spikes with ripples were on average shorter, had higher amplitude and higher slope than spikes without ripples. 64% of ripples started before spikes started. Spikes occurred on 13 (5–26) channels per patient, and ripples on 3 (0–14) channels, which were also spike channels. Conclusion: Ripples precede rather than follow spikes, so ripples are unlikely to result from spikes. Significance: Ripples and spikes seem not one-on-one coupled, but certain states of the brain can accommodate both. Ó 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
1. Introduction High frequency oscillations (HFOs, >80 Hz) in invasive intracranial recordings are newly proposed biomarkers for the brain area that generates seizures (Zijlmans et al., 2012; Jacobs et al., 2012; Bragin et al., 1999). HFOs are more specific for defining the epileptogenic zone than spikes (Jacobs et al., 2009). In line with this finding, they seem to be more related to disease activity (Zijlmans
⇑ Corresponding author at: University Medical Center Utrecht, Department of Neurology and Neurosurgery, HP C03.131, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands. Tel.: +31 887557735. E-mail address: [email protected] (N. van Klink).
et al., 2009, 2011) and possibly predict outcome after epilepsy surgery (Jacobs et al., 2010; Van’t Klooster et al., 2015). HFOs between 80 and 250 Hz, called ripples, can also be found in the non-invasive surface electroencephalogram (EEG). This increases the population for whom HFO analysis is available, beyond the small group of drug-resistant focal epilepsy patients who are candidates for surgery with intracerebral electrodes. Reports describe 50–100 Hz activity in ictal recordings (Kobayashi et al., 2004, 2009), ripples in interictal EEG in adults with focal epilepsy (Andrade-Valenca et al., 2011; Melani et al., 2013; Zelmann et al., 2013), in children with continuous spike and wave in slow-wave sleep (Kobayashi et al., 2010), and idiopathic partial epilepsy (Kobayashi et al., 2011). A recent study showed that interictal surface ripples decrease after treatment
http://dx.doi.org/10.1016/j.clinph.2015.04.059 1388-2457/Ó 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: van Klink N et al. Relationships between interictal epileptic spikes and ripples in surface EEG. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.04.059
2
N. van Klink et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
with ACTH in children with West Syndrome (Kobayashi et al., 2015). Co-occurrence of ripples with epileptic spikes is reported in 44–64% of all ripples in intracranial and surface recordings (Melani et al., 2013; Andrade-Valenca et al., 2011; Urrestarazu et al., 2007; Jacobs et al., 2009; Wang et al., 2013). Ripples co-occurring with a spike are more related to the seizure onset zone than ripples without a spike (Wang et al., 2013). There seems to be a relation between ripples and spikes, but so far no study addressed this important aspect in detail. Knowledge about this relation could help to get insight into the mechanisms of ripple generation and the clinical relevance of the coupling between ripples and spikes. Our goal was to further investigate the relationship between ripples and epileptic spikes in non-invasive surface EEG. The two main research questions were: Are there differences between spikes that show and those that do not show a ripple at the same time? What is the temporal and spatial relation between spikes and ripples? 2. Methods 2.1. Patients We included adult patients with focal or multifocal epilepsy who were admitted to the EEG-telemetry unit of the Montreal Neurological Hospital, and showed more than one epileptic spike or sharp wave per minute during N2 or N3 sleep. Patients who only showed polyspikes were excluded because of the difficulty in assessing the relationship between the ripple and the spike. We consecutively included patients from September 2014 back in time until more than a total number of 200 ripples were marked, a number which we thought was sufficient to assess spike-ripple relationship. All patients gave informed consent in agreement with the Research Ethics Board of the Montreal Neurological Institute and Hospital. 2.2. EEG recording, data selection and event marking Recordings were performed with the Harmonie system (Stellate, Montreal, Canada) with electrodes placed according to the 10–20 EEG system, with additional zygomatic and F9/F10, T9/T10 and P9/P10 electrodes and FCz as reference. The low pass filter was set at 300 Hz and the sampling frequency at 1000 Hz. We used the recording of the second night after admission. Typically the whole night (9 pm–9 am) was recorded. We selected epochs of N2 and N3 sleep. Sleep stages were marked according to the American Academy of Sleep Medicine (AASM) criteria (Berry et al., 2012) by one reviewer and checked and discussed with a second reviewer. We manually marked 50 consecutive time points with spikes in each patient in N2 or N3 sleep, starting at the first sleep cycle of that night. Epochs within two hours before or after a seizure were excluded. The first 50 time points with spikes were marked (Fig. 1, vertical lines); they usually occurred in the first sleep cycle, but could extend beyond. The length of the epochs therefore differed per patient, because it depended on the spike rate. The spikes were marked in a bipolar montage with a 15 s/page time scale and a 10 lV/mm amplitude scale. Spikes were marked on the individual channels as well, to allow for comparison with ripples on specific channels (Fig. 1, horizontal lines). Only isolated spikes were marked: spikes occurring within 1 s from a marked spike were excluded, as were spikes including artefacts. Subsequently, ripples were manually marked in a 400 ms time window around each spike in a bipolar montage with an 80 Hz finite impulse response (FIR) high pass filter, at a 1.5 s/page time scale and a 1 lV/mm amplitude scale. Ripples were marked by one reviewer and checked and discussed with a second reviewer. We defined a ripple as at least four oscillations above 80 Hz, which
were clearly distinct from the background EEG. We did not use a maximum number of oscillations or maximum duration because surface EEG ripples are generally short. Ripples with irregular frequency or amplitude morphology were suspicious to be artefacts and were therefore not marked. 2.3. Data analysis All statistics were calculated with IBM SPSS Statistics 22 (IBM Corp., Armonk, NY, USA). A p-value R Diffuse volume loss
1 9 0
319 115 160
8 9 10 11
46/M 76/F 45/M 26/F
36 56 38 14
CPS + GTCS GTCS GTCS CPS
L FT NA LT NA
Bilateral F (18) L T (11) Bilateral T + F (19) L FT (7) L FT (18) L T, R FT (19) L C (12)
L TPO (7) Bilateral F + O (14) L F (1) L F (6) N N N N N
0 0 0 0
185 298 155 138
12
53/M
3
L FT
L T (13)
N
0
166
13 14
23/F 18/M
9 17
SPS + CPS + rare GTCS CPS and GTCS CPS, rare GTCS
No abnormalities L hippocampal atrophy Bilateral hippocampal malrotation + L T atrophy Parry-Romberg syndrome. L cerebral atrophy. History: 2 L F resections. L hypothalamic tumor lesion
LT RF
L FT, rare R T (8) R T, few L T (17)
L T (5) N
10 0
188 121
15 16 17 18
39/F 53/M 60/M 50/M
15 37 58 2
CPS, rare GTCS CPS, rare GTCS CPS CPS and GTCS
L FT RT NA LT
L T (9) R T (7) L FT (9) L T, few R T (21)
L F (2) N L FT (3) L FC (3)
2 0 5 4
149 170 198 190
19 20
40/M 35/F
11 30
CPS CPS
R FT RT
R T, few L T (13) R T (8)
R T + L P (4) R T (3)
5 8
217 163
21 22 23
30/F 24/F 26/M
13 15 0
CPS and GTCS CPS and GTCS CPS
L F, rare R F (13) R FT (10) Bilateral F (17)
158 222 268
39/M
26
CPS, rare GTCS
L FC (4) R FT (5) Bilateral F (6) N
7 14 23
24
0
180
25 26
60/F 57/F
26 13
CPS, rare GTCS GTCS + absences
L FC NA Bilateral F Bilateral T RT L FT
L O DNET. History: partial resection DNET R F middle sulcus suspect FCD + R thamalus sequelae ischemia L PH gyrus lesion + pituitary cyst R T anterior cavernoma L MTL lesion R lateral ventricle mass + ventriculomegaly + diffuse white matter disturbances No abnormalities R P and T encephalomalacia. History: R TP astrocytoma lesionectomy, no recurrence. L F precentral gyrus suspect FCD R F anterior FCD + L T nodular heterotopia L MCA sequelae encephalitis
5 19
157 228
27
42/M
8
CPS, rare GTCS
L FT
L F, R T (12)
R FT (4) Bilateral FT (11) N
0
123
28 29 30
21/F 21/M 25/F
2 10 9
CPS, rare GTCS CPS CPS and GTCS
R T (20) R P, R T (20) Bilateral T (17)
N R T > L T (6) L T > R T (6)
0 11 23
176 262 217
31
52/M
NA
NA
RT R T>L T Bilateral T NA
Bilateral T (17)
Bilateral F (4)
6
174
Bilateral T (15) R T (5) Bilateral T(26)
R hippocampal atrophy R T atrophy + R hippocampal sclerosis No abnormalities. History: L F resection and callosotomy R F precentral gyrus hyperintensity. History: 3 L T, F and OF resections of FCD Bilateral hippocampal atrophy, R > L R PH, TO and cingulate white matter blurring R T and hippocampal atrophy + R T middle gyrus suspect FCD Bilateral F gyri and gyri recti contusions
Spike focus EEG (#Chann) = location of most of the interictal epileptiform discharges, with the total number of channels involved. Ripple (#Chann) = location where most marked ripples occurred, with the total number of channels involved. MRI = MRI findings. #Ripple = number of marked ripples. #Spikes = number of individual marked spike. F = female, M = male, CPS = complex partial seizures, GTCS = generalized tonic clonic seizures, SPS = simple partial seizures, L = left, R = right, F = frontal, T = temporal, C = central, P = parietal, O = occipital, FT = fronto-temporal, FC = fronto-central, TPO = temporo-parieto-occipital, MTL = mesial temporal lobe, DNET = dysembryoplastic neuroepithelial tumor, FCD = focal cortical dysplasia, PH = parahippocampal, NA = not available.
from ripples with spike (median with spike 56 ms (48–70), without spike 57 ms (49–67); Z = 0.20, p = 0.844).
4. Discussion We studied the differences between surface EEG spikes that do and do not co-occur with ripples, and found in a group comparison that spikes with ripples have a shorter duration, higher amplitude and higher slope than spikes without ripples. There is, however, no clear cut-off value for these characteristics which can separate spikes with from spikes without ripples. We also studied the temporal relation between spikes and ripples, and found that the majority of ripples start before the onset of the spike and that ripples rarely extend beyond the spike. Ripples occur on fewer channels than spikes, and the ripple channels are among the spike channels. The relation between ripples and spikes on one channel are not tightly time-locked, since ripples disappeared when spikes
were averaged. Our findings indicate that ripples on a spike are not ripples evoked by a spike. The difference between spikes with or without ripples has not been studied so far, although it has been suggested that ripples might discriminate between clinically important ‘red’ spikes and nonspecific, propagated ‘green’ spikes (Engel et al., 2009). The common thought that the sharper, higher amplitude spikes are more specific for the epileptogenic zone and co-occur with ripples more often than sharp waves is generally confirmed in this study, but not all sharp spikes co-occurred with ripples, and non-sharp spikes with ripples also existed. The possibility to detect ripples in non-invasive EEG is an important finding for the use of ripples in clinical practice. One of the critical issues for reliable detection is how to identify if oscillations arise from the brain or are related to filtering of sharp transients (Bénar et al., 2010). When a ripple is a filter artefact of a spike, the ripple is expected to be centered on the peak of the spike. Most of the ripples in this study occur for the main part
Please cite this article in press as: van Klink N et al. Relationships between interictal epileptic spikes and ripples in surface EEG. Clin Neurophysiol (2015), http://dx.doi.org/10.1016/j.clinph.2015.04.059
5
N. van Klink et al. / Clinical Neurophysiology xxx (2015) xxx–xxx
Fig. 3. Examples of spikes with (A and C) and without (B and D) ripples. Both spikes with short and long duration can co-occur with a ripple.
Table 2 Spike characteristics for spikes with and without ripple. Spike with ripple
Spike without ripple
p-value
Duration (ms) -FHW -SHW
86.1 (52.5 30.4 (17.5 51.6 (27.4
94.0 (71.9 35.9 (25.8 54.4 (41.5
