G Model
YSEIZ-2306; No. of Pages 3 Seizure xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Seizure journal homepage: www.elsevier.com/locate/yseiz
The stability of spike counts in children with interictal epileptiform activity Mark H. Libenson *, Amit Haldar, Anna L. Pinto Division of Epilepsy and Clinical Neurophysiology, Department of Neurology, Boston Children’s Hospital and Harvard Medical School, Boston, MA, USA
A R T I C L E I N F O
A B S T R A C T
Article history: Received 4 January 2014 Received in revised form 2 March 2014 Accepted 6 March 2014
Purpose: Little is known about the stability of serial measures of spike counts in children or whether spike counts are an inherently stable or unstable measure. We investigated the variation in first- and second-night spike counts in children undergoing 48-h ambulatory EEG recording. Methods: We analyzed 40 consecutive 48-h ambulatory EEGs performed at Boston Children’s Hospital that manifested spikes but no seizures. Distinct spike foci in the same child were counted separately. We visually counted all spikes in the first 20 min after the first sleep spindle during nighttime sleep, comparing the first and second nights. Results: Fifty-five unique spike foci were counted in 40 children (age range: 9 months to 19 years; median: 8.4 years). Considerable variation was seen when comparing Night 1 and Night 2 spike counts: for all foci, Night 1 mean and median spike counts were 304.5 and 126 and Night 2 counts were 309.5 and 148, respectively. For each focus, the mean change in spike frequency between Night 1 and Night 2 was 42.1% (median = 28.3%, IQR 19.0–50.0%). The coefficient of variation of 0.94 suggested a large amount of variation. The percentage change weighted according to high or low spike frequency was 25.1%. Conclusion: In 40 children with 55 unique spike foci, significant variability in spike frequency was seen between consecutive nights of sleep, suggesting significant natural variation in spike frequency. A quarter of spike foci varied by 50% or more. Spike counts separated by longer intervals may show even more dramatic natural variation. ß 2014 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.
Keywords: Epilepsy Spike counts Spike frequency Spike stability Interictal epileptiform activity EEG
1. Introduction Multiple controversies surround the clinical significance of interictal epileptiform activity (IEA) in the EEG. The identification of epileptiform activity has been generally accepted as a useful, though imperfect tool in identifying zones of electrical irritability and possible seizure onset in cerebral cortex. Whether IEA is simply a useful marker for identifying abnormally irritable cortex or whether ongoing spike activity itself also causes some type of disruption in brain function has become the focus of debate.1–3 Those who hold the view that ongoing spike activity causes harm and that its potential reduction or eradication may confer a benefit to the patient have proposed various interventions (usually
* Corresponding author at: Division of Epilepsy and Clinical Neurophysiology, Fegan 9, Boston Children’s Hospital, 300 Longwood Avenue, Boston, MA 02115, USA. Tel.: +1 617 355 8071; fax: +1 617 730 0463. E-mail addresses: [email protected] (M.H. Libenson), [email protected] (A. Haldar), [email protected] (A.L. Pinto).
pharmacologic) to reduce spike activity with the hope of improving cognitive function. There is an additional question regarding the relationship between spike frequency and the epileptogenicity of a specific focus: does a focus that spikes more frequently have a stronger tendency to give rise to a seizure compared to a ‘‘quieter’’ focus? Thus, there are many reasons that have lead clinicians and researchers to count spikes in the EEG. In order to study these questions related to spike frequency in a meaningful way, it is important to have an idea of the natural behavior of spike counts in the EEG. Before studying the impact of various maneuvers (e.g., drug therapies) on spike counts or investigating the relationship of spike counts to epileptogenicity, it is important to understand the stability of spike frequency in the absence of intervention (in its ‘‘natural state’’), though little is known about this subject. For clinicians who already obtain spike counts while managing certain patients with epilepsy the question of natural variation in spike frequency is important. Is it enough to make a single measurement of spike counts to get a good assessment of a spike locus’s inherent frequency or does the count vary widely from day to day? If the natural variation in spike
http://dx.doi.org/10.1016/j.seizure.2014.03.005 1059-1311/ß 2014 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.
Please cite this article in press as: Libenson MH, et al. The stability of spike counts in children with interictal epileptiform activity. Seizure: Eur J Epilepsy (2014), http://dx.doi.org/10.1016/j.seizure.2014.03.005
G Model
YSEIZ-2306; No. of Pages 3
20 0
10 0
0
counts from day to day is small, then moderate or large changes in observed spike counts after introduction of a medication are more likely to be related to the introduction of that drug. If natural daily variation is high, then it becomes more important to make multiple observations on different days in order to describe the spike frequency in an individual patient. Because little is known regarding the natural day-to-day variation in spike counts in the absence of intervention, we set out to quantify the extent of natural variation in spike counts during sleep between Night 1 and Night 2 in ambulatory EEGs (aEEGs) recorded in children on two consecutive nights.
-1 00
M.H. Libenson et al. / Seizure xxx (2014) xxx–xxx
2
2. Methods
Percentage Change We retrospectively analyzed consecutive 48-h aEEGs performed at Children’s Hospital Boston between August 2009 and June 2010. We excluded aEEGs with (1) no spikes on Night 1 of the recording, (2) recordings with electrographic or electroclinical seizures, (3) excessive artifact or other technical problems that precluded accurate spike counting, (4) AED adjustments during the recording period, (5) the absence of distinct sleep spindles during sleep, and (6) more than three independent spike foci. In studies with more than one independent spike focus, each independent spike focus was counted and reported separately. Recordings were made using the standard international 10–20 system electrode set. In order to eliminate the impact of wakefulness, drowsiness or depth of sleep on spike frequency, spikes were counted during sleep. Spikes were visually counted for each night for 20 consecutive minutes starting from the occurrence of the first sleep spindle after onset of night-time sleep. If an arousal occurred in the first 20 min, the count was obtained for the first contiguous 20-min period of sleep (beginning with a sleep spindle). Spike counts for Night 1 and Night 2 from the same aEEG were always performed sequentially by the same reader (AH or AP). All spike counts are reported ‘‘per 20-min interval.’’ 2.1. Statistics Statistical analyses were done on the absolute value of the percentage change from Night 1. The coefficient of variation (CV) of the differences, a measure of the dispersion of a distribution (in this case a measure of the variability in spike counts, or the unpredictability of the spike count on Night 2 based on the spike count observed on Night 1), is calculated by dividing the standard deviation of the distribution of the percentage differences by its mean. 2.1.1. Weighted analysis In order to avoid the bias of foci with lower spike counts of generating higher apparent percentage changes between Night 1 and Night 2 because of low numbers, we repeated the statistical analysis by weighting the percentage change of each focus according to the absolute number of spikes occurring at that focus, weighting each percentage change according to the number of spikes counted on Night 1 for that focus divided by the total number of spikes counted for Night 1 among all 55 foci. In this way, a focus with a 33% decrease in spike count from 600 to 400 would have 100 times more impact on the weighted mean percentage change compared to a focus whose spike count decreased from 6 to 4. 3. Results Forty consecutive ambulatory EEGs that passed exclusion criteria from 40 patients were analyzed, totaling 55 unique spike foci (34 studies with a single spike focus, six with two independent
Fig. 1. Percentage change in spike count between Night 1 and Night 2 for 55 individual spike foci in 40 unique patients.
foci, and three with three foci). The age range of the patients was 9 months to 19.9 years (median age: 8.4 years). All of the patients had focal epilepsy except for one each with ultimate diagnoses of syncope, night terrors, acute disseminated encephalomyelitis (ADEM) and motor tics. In the majority of cases, the referring physician obtained the recording either to ascertain whether a given behavior was an epileptic seizure, or to assess the patient for unrecognized seizure activity. Of the patients with epilepsy, four had benign rolandic epilepsy and one had the Landau–Kleffner syndrome. Twenty-nine of the 40 patients were on antiepileptic drug treatment at the time of the recordings. The counted spikes arose from all brain regions without a particular predilection for one area. Considerable variation in absolute spike counts was seen among the 40 patients (range: 0–1440 spikes per 20-min interval). For all 110 observations (55 from Night 1 and 55 from Night 2) the mean spike count was 307 (S.D. = 33.7) per 20-min period and the median spike count was 137. Mean and median spike counts for Night 1 (304.5 and 126) and Night 2 (309.5 and 148), respectively, were comparable (p = 0.77, paired t-test, two-tailed). This implied that there was no appreciable ‘‘first-’’ or ‘‘second-night effect’’ (i.e., no systematic upward or downward trend from Night 1 to Night 2). For each focus, the mean change of spike frequency between Night 1 and Night 2 was 42.1% (S.D. = 39.6). The median change was 28.3%, the range was 0.0–172.5%, and the interquartile range was 19.0–50.0% (see Fig. 1). The coefficient of variation of the differences between Night 1 and Night 2 was 0.94, suggesting a large amount of variation. Among the 55 spike foci there was a wide variation in absolute spike frequency. For instance, 12 foci had fewer than 10 spikes counted on Night 1. In order to avoid the problem of higher apparent percentage changes in spike foci with low spike counts, we repeated the analysis using a weighting scheme based on absolute spike counts (as described in Section 2.1 above). As expected, the weighted mean percentage change of spike counts was lower than the unweighted mean at 25.1% (S.D. = 11.08). The weighted CV of the differences was 0.47. 4. Discussion In children undergoing ambulatory EEG monitoring, spike counts obtained for consecutive nights’ sleep were relatively unstable, showing an average 42.1% difference from night to night. Half of all foci had a percentage change in spike count over 28.3% and a quarter of all foci varied by 50% or more. When each percentage change was weighted according to spike frequency in order to reduce the impact of large percentage changes related to low spike counts, a weighted mean percentage change of 25.1%
Please cite this article in press as: Libenson MH, et al. The stability of spike counts in children with interictal epileptiform activity. Seizure: Eur J Epilepsy (2014), http://dx.doi.org/10.1016/j.seizure.2014.03.005
G Model
YSEIZ-2306; No. of Pages 3 M.H. Libenson et al. / Seizure xxx (2014) xxx–xxx
was still calculated. Some of the changes we observed in individual patients were relatively large, and in a few the changes were greater than an order of magnitude. We are not aware of others who have described the natural stability of spike frequency in children in the absence of intervention. Sanchez-Fernandez et al. studied the evolution of spike percentage and spike-wave index in children with continuous spike-wave during slow wave sleep and electrical status epilepticus of sleep but over a period of years, seeking to describe the natural history of those disorders over long periods rather than the stability of those indices in the short term.4 Sundaram et al. found that spikes were most likely to be present on the EEG within 48 h of a seizure but intrapatient changes in spike frequencies were not reported.5 Lower pre-operative spike frequency has been found to be a positive prognostic factor for success after temporal lobectomy.6 Many past studies that have examined spike frequency have analyzed correlations between spike counts and cognitive function,3,7–10 including in benign rolandic epilepsy11,12 but none specifically studied the stability or reproducibility of spike counts with serial observations in the short term. Other past studies have often concentrated on assessing the impact of specific AEDs on spike frequency.13–18 Clemens et al. counted spikes by sleep state, and eyes-open and eyes-closed during wakefulness.19 Others have studied spike frequency as a function of sleep state, but did not specifically study stability of spike counts in the same patient in the same state.20,21 Although they did not specifically study the question of reproducibility of serial measurements of spike frequency as we did, Camfield et al. asked a philosophically similar question: how often were the findings of a first and second EEG obtained 6 months or less apart in children with newly diagnosed epilepsy concordant? They found rates of discordance between 40% and 70% for type of abnormality (e.g., focal versus generalized findings) and concluded that ‘‘The interictal EEG in childhood epilepsy appears to be an unstable test.’’22 We measured changes in spike frequency over an interval of approximately 24 h. It is interesting to speculate how the counts might vary with longer inter-observation intervals in the absence of other intervention. The possibility that spike counts could vary even more if measured a week apart, or even months apart, seems likely but needs to be explored in future studies, as it is conceivable that high or low spike counts could cluster in an individual patient over certain periods just as seizures are known to cluster during certain time periods in some patients. Limitations of our study design suggest possible aims for future studies. We combined the results of children with a wide variety of epilepsy subtypes in our analysis. Our study does not tell us whether spike stability differs between epilepsy subtypes (e.g., focal cortical dysplasia versus benign rolandic epilepsy) as numbers in each diagnostic subgroup were not high enough to make these distinctions. Our study also did not analyze spike frequencies during wakefulness – some may argue that daytime spikes may have a greater cognitive impact than spikes occurring during sleep, implying another area for future study. Finally, some portion of the observed variation is likely related to the inherent variation in any measurement technique (as opposed to true patient variation), though we believe that visual spike counting by the same observer has helped to minimize this effect. 5. Conclusions The natural variation in nocturnal spike counts in children is considerable, with over half of subjects having an absolute percentage change in spike counts greater than 28%. The fact that
3
such large variations in spike counts are seen from night to night suggests that changes of this magnitude when an intervention has been attempted should be interpreted with caution, as they may be due to natural variation. On the other hand, extremely large changes in spike frequency, such as those greater than an order of magnitude, are uncommon on consecutive nights. Conflict of interest None declared. Acknowledgment We would like to thank Matthew Gregas, Ph.D. for his assistance with the statistical analysis. References 1. Van Bogaert P, Urbain C, Galer S, Ligot N, Peigneux P, De Tiege X. Impact of focal interictal epileptiform discharges on behaviour and cognition in children. Neurophysiol Clin 2012;42(1–2):53–8. 2. Binnie CD. Cognitive impairment during epileptiform discharges: is it ever justifiable to treat the EEG? Lancet Neurol 2003;2(12):725–30. 3. Holmes GL, Lenck-Santini PP. Role of interictal epileptiform abnormalities in cognitive impairment. Epilepsy Behav 2006;8(3):504–15. 4. Fernandez IS, Peters JM, Hadjiloizou S, Prabhu SP, Zarowski M, Stannard KM, et al. Clinical staging and electroencephalographic evolution of continuous spikes and waves during sleep. Epilepsia 2012;53(7):1185–95. 5. Sundaram M, Hogan T, Hiscock M, Pillay N. Factors affecting interictal spike discharges in adults with epilepsy. Electroencephalogr Clin Neurophysiol 1990;75(4):358–60. 6. Krendl R, Lurger S, Baumgartner C. Absolute spike frequency predicts surgical outcome in TLE with unilateral hippocampal atrophy. Neurology 2008;71(6):413–8. 7. Ebus S, Arends J, Hendriksen J, van der Horst E, de la Parra N, Hendriksen R, et al. Cognitive effects of interictal epileptiform discharges in children. Eur J Paediatr Neurol 2012;16(6):697–706. 8. Wang SB, Weng WC, Fan PC, Lee WT. Levetiracetam in continuous spike waves during slow-wave sleep syndrome. Pediatr Neurol 2008;39(2):85–90. 9. Mintz M, Legoff D, Scornaienchi J, Brown M, Levin-Allen S, Mintz P, et al. The underrecognized epilepsy spectrum: the effects of levetiracetam on neuropsychological functioning in relation to subclinical spike production. J Child Neurol 2009;24(7):807–15. 10. Aarts JH, Binnie CD, Smit AM, Wilkins AJ. Selective cognitive impairment during focal and generalized epileptiform EEG activity. Brain 1984;107(Pt 1):293–308. 11. Sarco DP, Boyer K, Lundy-Krigbaum SM, Takeoka M, Jensen F, Gregas M, et al. Benign rolandic epileptiform discharges are associated with mood and behavior problems. Epilepsy Behav 2011;22(2):298–303. 12. Kanemura H, Sano F, Aoyagi K, Sugita K, Aihara M. Do sequential EEG changes predict atypical clinical features in rolandic epilepsy? Dev Med Child Neurol 2012;54(10):912–7. 13. Schmidt D. The influence of antiepileptic drugs on the electroencephalogram: a review of controlled clinical studies. Electroencephalogr Clin Neurophysiol Suppl 1982;36:453–66. 14. Libenson MH, Caravale B. Do antiepileptic drugs differ in suppressing interictal epileptiform activity in children? Pediatr Neurol 2001;24(3):214–8. 15. Stodieck S, Steinhoff BJ, Kolmsee S, van Rijckevorsel K. Effect of levetiracetam in patients with epilepsy and interictal epileptiform discharges. Seizure 2001;10(8):583–7. 16. Gotman J, Marciani MG. Electroencephalographic spiking activity, drug levels, and seizure occurrence in epileptic patients. Ann Neurol 1985;17(6):597–603. 17. Gotman J, Koffler DJ. Interictal spiking increases after seizures but does not after decrease in medication. Electroencephalogr Clin Neurophysiol 1989;72(1):7–15. 18. Cavitt J, Privitera M. Levetiracetam induces a rapid and sustained reduction of generalized spike-wave and clinical absence. Arch Neurol 2004;61(10):1604–7. 19. Clemens Z, Janszky J, Clemens B, Szucs A, Halasz P. Factors affecting spiking related to sleep and wake states in temporal lobe epilepsy (TLE). Seizure 2005;14(1):52–7. 20. Rossi GF, Colicchio G, Pola P. Interictal epileptic activity during sleep: a stereoEEG study in patients with partial epilepsy. Electroencephalogr Clin Neurophysiol 1984;58(2):97–106. 21. Malow BA, Kushwaha R, Lin X, Morton KJ, Aldrich MS. Relationship of interictal epileptiform discharges to sleep depth in partial epilepsy. Electroencephalogr Clin Neurophysiol 1997;102(1):20–6. 22. Camfield P, Gordon K, Camfield C, Tibbles J, Dooley J, Smith B. EEG results are rarely the same if repeated within six months in childhood epilepsy. Can J Neurol Sci 1995;22(4):297–300.
Please cite this article in press as: Libenson MH, et al. The stability of spike counts in children with interictal epileptiform activity. Seizure: Eur J Epilepsy (2014), http://dx.doi.org/10.1016/j.seizure.2014.03.005
YSEIZ-2306; No. of Pages 3 Seizure xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Seizure journal homepage: www.elsevier.com/locate/yseiz
The stability of spike counts in children with interictal epileptiform activity Mark H. Libenson *, Amit Haldar, Anna L. Pinto Division of Epilepsy and Clinical Neurophysiology, Department of Neurology, Boston Children’s Hospital and Harvard Medical School, Boston, MA, USA
A R T I C L E I N F O
A B S T R A C T
Article history: Received 4 January 2014 Received in revised form 2 March 2014 Accepted 6 March 2014
Purpose: Little is known about the stability of serial measures of spike counts in children or whether spike counts are an inherently stable or unstable measure. We investigated the variation in first- and second-night spike counts in children undergoing 48-h ambulatory EEG recording. Methods: We analyzed 40 consecutive 48-h ambulatory EEGs performed at Boston Children’s Hospital that manifested spikes but no seizures. Distinct spike foci in the same child were counted separately. We visually counted all spikes in the first 20 min after the first sleep spindle during nighttime sleep, comparing the first and second nights. Results: Fifty-five unique spike foci were counted in 40 children (age range: 9 months to 19 years; median: 8.4 years). Considerable variation was seen when comparing Night 1 and Night 2 spike counts: for all foci, Night 1 mean and median spike counts were 304.5 and 126 and Night 2 counts were 309.5 and 148, respectively. For each focus, the mean change in spike frequency between Night 1 and Night 2 was 42.1% (median = 28.3%, IQR 19.0–50.0%). The coefficient of variation of 0.94 suggested a large amount of variation. The percentage change weighted according to high or low spike frequency was 25.1%. Conclusion: In 40 children with 55 unique spike foci, significant variability in spike frequency was seen between consecutive nights of sleep, suggesting significant natural variation in spike frequency. A quarter of spike foci varied by 50% or more. Spike counts separated by longer intervals may show even more dramatic natural variation. ß 2014 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.
Keywords: Epilepsy Spike counts Spike frequency Spike stability Interictal epileptiform activity EEG
1. Introduction Multiple controversies surround the clinical significance of interictal epileptiform activity (IEA) in the EEG. The identification of epileptiform activity has been generally accepted as a useful, though imperfect tool in identifying zones of electrical irritability and possible seizure onset in cerebral cortex. Whether IEA is simply a useful marker for identifying abnormally irritable cortex or whether ongoing spike activity itself also causes some type of disruption in brain function has become the focus of debate.1–3 Those who hold the view that ongoing spike activity causes harm and that its potential reduction or eradication may confer a benefit to the patient have proposed various interventions (usually
* Corresponding author at: Division of Epilepsy and Clinical Neurophysiology, Fegan 9, Boston Children’s Hospital, 300 Longwood Avenue, Boston, MA 02115, USA. Tel.: +1 617 355 8071; fax: +1 617 730 0463. E-mail addresses: [email protected] (M.H. Libenson), [email protected] (A. Haldar), [email protected] (A.L. Pinto).
pharmacologic) to reduce spike activity with the hope of improving cognitive function. There is an additional question regarding the relationship between spike frequency and the epileptogenicity of a specific focus: does a focus that spikes more frequently have a stronger tendency to give rise to a seizure compared to a ‘‘quieter’’ focus? Thus, there are many reasons that have lead clinicians and researchers to count spikes in the EEG. In order to study these questions related to spike frequency in a meaningful way, it is important to have an idea of the natural behavior of spike counts in the EEG. Before studying the impact of various maneuvers (e.g., drug therapies) on spike counts or investigating the relationship of spike counts to epileptogenicity, it is important to understand the stability of spike frequency in the absence of intervention (in its ‘‘natural state’’), though little is known about this subject. For clinicians who already obtain spike counts while managing certain patients with epilepsy the question of natural variation in spike frequency is important. Is it enough to make a single measurement of spike counts to get a good assessment of a spike locus’s inherent frequency or does the count vary widely from day to day? If the natural variation in spike
http://dx.doi.org/10.1016/j.seizure.2014.03.005 1059-1311/ß 2014 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.
Please cite this article in press as: Libenson MH, et al. The stability of spike counts in children with interictal epileptiform activity. Seizure: Eur J Epilepsy (2014), http://dx.doi.org/10.1016/j.seizure.2014.03.005
G Model
YSEIZ-2306; No. of Pages 3
20 0
10 0
0
counts from day to day is small, then moderate or large changes in observed spike counts after introduction of a medication are more likely to be related to the introduction of that drug. If natural daily variation is high, then it becomes more important to make multiple observations on different days in order to describe the spike frequency in an individual patient. Because little is known regarding the natural day-to-day variation in spike counts in the absence of intervention, we set out to quantify the extent of natural variation in spike counts during sleep between Night 1 and Night 2 in ambulatory EEGs (aEEGs) recorded in children on two consecutive nights.
-1 00
M.H. Libenson et al. / Seizure xxx (2014) xxx–xxx
2
2. Methods
Percentage Change We retrospectively analyzed consecutive 48-h aEEGs performed at Children’s Hospital Boston between August 2009 and June 2010. We excluded aEEGs with (1) no spikes on Night 1 of the recording, (2) recordings with electrographic or electroclinical seizures, (3) excessive artifact or other technical problems that precluded accurate spike counting, (4) AED adjustments during the recording period, (5) the absence of distinct sleep spindles during sleep, and (6) more than three independent spike foci. In studies with more than one independent spike focus, each independent spike focus was counted and reported separately. Recordings were made using the standard international 10–20 system electrode set. In order to eliminate the impact of wakefulness, drowsiness or depth of sleep on spike frequency, spikes were counted during sleep. Spikes were visually counted for each night for 20 consecutive minutes starting from the occurrence of the first sleep spindle after onset of night-time sleep. If an arousal occurred in the first 20 min, the count was obtained for the first contiguous 20-min period of sleep (beginning with a sleep spindle). Spike counts for Night 1 and Night 2 from the same aEEG were always performed sequentially by the same reader (AH or AP). All spike counts are reported ‘‘per 20-min interval.’’ 2.1. Statistics Statistical analyses were done on the absolute value of the percentage change from Night 1. The coefficient of variation (CV) of the differences, a measure of the dispersion of a distribution (in this case a measure of the variability in spike counts, or the unpredictability of the spike count on Night 2 based on the spike count observed on Night 1), is calculated by dividing the standard deviation of the distribution of the percentage differences by its mean. 2.1.1. Weighted analysis In order to avoid the bias of foci with lower spike counts of generating higher apparent percentage changes between Night 1 and Night 2 because of low numbers, we repeated the statistical analysis by weighting the percentage change of each focus according to the absolute number of spikes occurring at that focus, weighting each percentage change according to the number of spikes counted on Night 1 for that focus divided by the total number of spikes counted for Night 1 among all 55 foci. In this way, a focus with a 33% decrease in spike count from 600 to 400 would have 100 times more impact on the weighted mean percentage change compared to a focus whose spike count decreased from 6 to 4. 3. Results Forty consecutive ambulatory EEGs that passed exclusion criteria from 40 patients were analyzed, totaling 55 unique spike foci (34 studies with a single spike focus, six with two independent
Fig. 1. Percentage change in spike count between Night 1 and Night 2 for 55 individual spike foci in 40 unique patients.
foci, and three with three foci). The age range of the patients was 9 months to 19.9 years (median age: 8.4 years). All of the patients had focal epilepsy except for one each with ultimate diagnoses of syncope, night terrors, acute disseminated encephalomyelitis (ADEM) and motor tics. In the majority of cases, the referring physician obtained the recording either to ascertain whether a given behavior was an epileptic seizure, or to assess the patient for unrecognized seizure activity. Of the patients with epilepsy, four had benign rolandic epilepsy and one had the Landau–Kleffner syndrome. Twenty-nine of the 40 patients were on antiepileptic drug treatment at the time of the recordings. The counted spikes arose from all brain regions without a particular predilection for one area. Considerable variation in absolute spike counts was seen among the 40 patients (range: 0–1440 spikes per 20-min interval). For all 110 observations (55 from Night 1 and 55 from Night 2) the mean spike count was 307 (S.D. = 33.7) per 20-min period and the median spike count was 137. Mean and median spike counts for Night 1 (304.5 and 126) and Night 2 (309.5 and 148), respectively, were comparable (p = 0.77, paired t-test, two-tailed). This implied that there was no appreciable ‘‘first-’’ or ‘‘second-night effect’’ (i.e., no systematic upward or downward trend from Night 1 to Night 2). For each focus, the mean change of spike frequency between Night 1 and Night 2 was 42.1% (S.D. = 39.6). The median change was 28.3%, the range was 0.0–172.5%, and the interquartile range was 19.0–50.0% (see Fig. 1). The coefficient of variation of the differences between Night 1 and Night 2 was 0.94, suggesting a large amount of variation. Among the 55 spike foci there was a wide variation in absolute spike frequency. For instance, 12 foci had fewer than 10 spikes counted on Night 1. In order to avoid the problem of higher apparent percentage changes in spike foci with low spike counts, we repeated the analysis using a weighting scheme based on absolute spike counts (as described in Section 2.1 above). As expected, the weighted mean percentage change of spike counts was lower than the unweighted mean at 25.1% (S.D. = 11.08). The weighted CV of the differences was 0.47. 4. Discussion In children undergoing ambulatory EEG monitoring, spike counts obtained for consecutive nights’ sleep were relatively unstable, showing an average 42.1% difference from night to night. Half of all foci had a percentage change in spike count over 28.3% and a quarter of all foci varied by 50% or more. When each percentage change was weighted according to spike frequency in order to reduce the impact of large percentage changes related to low spike counts, a weighted mean percentage change of 25.1%
Please cite this article in press as: Libenson MH, et al. The stability of spike counts in children with interictal epileptiform activity. Seizure: Eur J Epilepsy (2014), http://dx.doi.org/10.1016/j.seizure.2014.03.005
G Model
YSEIZ-2306; No. of Pages 3 M.H. Libenson et al. / Seizure xxx (2014) xxx–xxx
was still calculated. Some of the changes we observed in individual patients were relatively large, and in a few the changes were greater than an order of magnitude. We are not aware of others who have described the natural stability of spike frequency in children in the absence of intervention. Sanchez-Fernandez et al. studied the evolution of spike percentage and spike-wave index in children with continuous spike-wave during slow wave sleep and electrical status epilepticus of sleep but over a period of years, seeking to describe the natural history of those disorders over long periods rather than the stability of those indices in the short term.4 Sundaram et al. found that spikes were most likely to be present on the EEG within 48 h of a seizure but intrapatient changes in spike frequencies were not reported.5 Lower pre-operative spike frequency has been found to be a positive prognostic factor for success after temporal lobectomy.6 Many past studies that have examined spike frequency have analyzed correlations between spike counts and cognitive function,3,7–10 including in benign rolandic epilepsy11,12 but none specifically studied the stability or reproducibility of spike counts with serial observations in the short term. Other past studies have often concentrated on assessing the impact of specific AEDs on spike frequency.13–18 Clemens et al. counted spikes by sleep state, and eyes-open and eyes-closed during wakefulness.19 Others have studied spike frequency as a function of sleep state, but did not specifically study stability of spike counts in the same patient in the same state.20,21 Although they did not specifically study the question of reproducibility of serial measurements of spike frequency as we did, Camfield et al. asked a philosophically similar question: how often were the findings of a first and second EEG obtained 6 months or less apart in children with newly diagnosed epilepsy concordant? They found rates of discordance between 40% and 70% for type of abnormality (e.g., focal versus generalized findings) and concluded that ‘‘The interictal EEG in childhood epilepsy appears to be an unstable test.’’22 We measured changes in spike frequency over an interval of approximately 24 h. It is interesting to speculate how the counts might vary with longer inter-observation intervals in the absence of other intervention. The possibility that spike counts could vary even more if measured a week apart, or even months apart, seems likely but needs to be explored in future studies, as it is conceivable that high or low spike counts could cluster in an individual patient over certain periods just as seizures are known to cluster during certain time periods in some patients. Limitations of our study design suggest possible aims for future studies. We combined the results of children with a wide variety of epilepsy subtypes in our analysis. Our study does not tell us whether spike stability differs between epilepsy subtypes (e.g., focal cortical dysplasia versus benign rolandic epilepsy) as numbers in each diagnostic subgroup were not high enough to make these distinctions. Our study also did not analyze spike frequencies during wakefulness – some may argue that daytime spikes may have a greater cognitive impact than spikes occurring during sleep, implying another area for future study. Finally, some portion of the observed variation is likely related to the inherent variation in any measurement technique (as opposed to true patient variation), though we believe that visual spike counting by the same observer has helped to minimize this effect. 5. Conclusions The natural variation in nocturnal spike counts in children is considerable, with over half of subjects having an absolute percentage change in spike counts greater than 28%. The fact that
3
such large variations in spike counts are seen from night to night suggests that changes of this magnitude when an intervention has been attempted should be interpreted with caution, as they may be due to natural variation. On the other hand, extremely large changes in spike frequency, such as those greater than an order of magnitude, are uncommon on consecutive nights. Conflict of interest None declared. Acknowledgment We would like to thank Matthew Gregas, Ph.D. for his assistance with the statistical analysis. References 1. Van Bogaert P, Urbain C, Galer S, Ligot N, Peigneux P, De Tiege X. Impact of focal interictal epileptiform discharges on behaviour and cognition in children. Neurophysiol Clin 2012;42(1–2):53–8. 2. Binnie CD. Cognitive impairment during epileptiform discharges: is it ever justifiable to treat the EEG? Lancet Neurol 2003;2(12):725–30. 3. Holmes GL, Lenck-Santini PP. Role of interictal epileptiform abnormalities in cognitive impairment. Epilepsy Behav 2006;8(3):504–15. 4. Fernandez IS, Peters JM, Hadjiloizou S, Prabhu SP, Zarowski M, Stannard KM, et al. Clinical staging and electroencephalographic evolution of continuous spikes and waves during sleep. Epilepsia 2012;53(7):1185–95. 5. Sundaram M, Hogan T, Hiscock M, Pillay N. Factors affecting interictal spike discharges in adults with epilepsy. Electroencephalogr Clin Neurophysiol 1990;75(4):358–60. 6. Krendl R, Lurger S, Baumgartner C. Absolute spike frequency predicts surgical outcome in TLE with unilateral hippocampal atrophy. Neurology 2008;71(6):413–8. 7. Ebus S, Arends J, Hendriksen J, van der Horst E, de la Parra N, Hendriksen R, et al. Cognitive effects of interictal epileptiform discharges in children. Eur J Paediatr Neurol 2012;16(6):697–706. 8. Wang SB, Weng WC, Fan PC, Lee WT. Levetiracetam in continuous spike waves during slow-wave sleep syndrome. Pediatr Neurol 2008;39(2):85–90. 9. Mintz M, Legoff D, Scornaienchi J, Brown M, Levin-Allen S, Mintz P, et al. The underrecognized epilepsy spectrum: the effects of levetiracetam on neuropsychological functioning in relation to subclinical spike production. J Child Neurol 2009;24(7):807–15. 10. Aarts JH, Binnie CD, Smit AM, Wilkins AJ. Selective cognitive impairment during focal and generalized epileptiform EEG activity. Brain 1984;107(Pt 1):293–308. 11. Sarco DP, Boyer K, Lundy-Krigbaum SM, Takeoka M, Jensen F, Gregas M, et al. Benign rolandic epileptiform discharges are associated with mood and behavior problems. Epilepsy Behav 2011;22(2):298–303. 12. Kanemura H, Sano F, Aoyagi K, Sugita K, Aihara M. Do sequential EEG changes predict atypical clinical features in rolandic epilepsy? Dev Med Child Neurol 2012;54(10):912–7. 13. Schmidt D. The influence of antiepileptic drugs on the electroencephalogram: a review of controlled clinical studies. Electroencephalogr Clin Neurophysiol Suppl 1982;36:453–66. 14. Libenson MH, Caravale B. Do antiepileptic drugs differ in suppressing interictal epileptiform activity in children? Pediatr Neurol 2001;24(3):214–8. 15. Stodieck S, Steinhoff BJ, Kolmsee S, van Rijckevorsel K. Effect of levetiracetam in patients with epilepsy and interictal epileptiform discharges. Seizure 2001;10(8):583–7. 16. Gotman J, Marciani MG. Electroencephalographic spiking activity, drug levels, and seizure occurrence in epileptic patients. Ann Neurol 1985;17(6):597–603. 17. Gotman J, Koffler DJ. Interictal spiking increases after seizures but does not after decrease in medication. Electroencephalogr Clin Neurophysiol 1989;72(1):7–15. 18. Cavitt J, Privitera M. Levetiracetam induces a rapid and sustained reduction of generalized spike-wave and clinical absence. Arch Neurol 2004;61(10):1604–7. 19. Clemens Z, Janszky J, Clemens B, Szucs A, Halasz P. Factors affecting spiking related to sleep and wake states in temporal lobe epilepsy (TLE). Seizure 2005;14(1):52–7. 20. Rossi GF, Colicchio G, Pola P. Interictal epileptic activity during sleep: a stereoEEG study in patients with partial epilepsy. Electroencephalogr Clin Neurophysiol 1984;58(2):97–106. 21. Malow BA, Kushwaha R, Lin X, Morton KJ, Aldrich MS. Relationship of interictal epileptiform discharges to sleep depth in partial epilepsy. Electroencephalogr Clin Neurophysiol 1997;102(1):20–6. 22. Camfield P, Gordon K, Camfield C, Tibbles J, Dooley J, Smith B. EEG results are rarely the same if repeated within six months in childhood epilepsy. Can J Neurol Sci 1995;22(4):297–300.
Please cite this article in press as: Libenson MH, et al. The stability of spike counts in children with interictal epileptiform activity. Seizure: Eur J Epilepsy (2014), http://dx.doi.org/10.1016/j.seizure.2014.03.005
