Seizure: European Journal of Epilepsy
Volume 19, Issue 6 , Pages 347-351, July 2010

Bimodal ultradian seizure periodicity in human mesial temporal lobe epilepsy

  • Matthew Karafin

      Affiliations

    • Iowa Comprehensive Epilepsy Program, Department of Neurology, University of Iowa Carver College of Medicine, Iowa City, IA, United States
    • Current address: Johns Hopkins University School of Medicine, Baltimore, MD, United States.
    • ,
  • Erik K. St. Louis

      Affiliations

    • Iowa Comprehensive Epilepsy Program, Department of Neurology, University of Iowa Carver College of Medicine, Iowa City, IA, United States
    • Corresponding Author InformationCorresponding author. Current address: Mayo Clinic, Rochester, MN, United States. Tel.: +1 507 266 7456; fax: +1 507 266 7772.
    • ,
  • M. Bridget Zimmerman

      Affiliations

    • Department of Biostatistics, University of Iowa College of Public Health, Iowa City, IA, United States
    • ,
  • Jon David Sparks

      Affiliations

    • Department of Biostatistics, University of Iowa College of Public Health, Iowa City, IA, United States
    • ,
  • Mark A. Granner

      Affiliations

    • Iowa Comprehensive Epilepsy Program, Department of Neurology, University of Iowa Carver College of Medicine, Iowa City, IA, United States

Received 21 January 2010; accepted 18 May 2010. published online 28 June 2010.

Article Outline

Abstract 

Background

Mesial temporal lobe epilepsy (mTLE) has been suggested to follow a circadian rhythm. Previous research found an afternoon peak in mTLE seizure occurrence. We evaluated the pattern of seizure occurrence in patients with well-localized mTLE and hypothesized that peak seizure frequency would occur in the afternoon, and that this pattern would not be altered by age, gender, or seizure focus.

Methods

We retrospectively identified consecutive mTLE patients with a seizure-free outcome following anterior temporal lobectomy from 1993 to 2004 with video-EEG captured seizures. We recorded and plotted the 24-h clock time for each seizure and performed cosinor analysis. SAS Proc GLIMMIX was used to fit the linearized transform of the cosinor model. Negative binomial regression fitted by the generalized estimating equations (GEE) method was also performed to estimate and compare the mean seizure rates over a 24-h day.

Results

Sixty mTLE patients monitored between 2 and 16 days were analyzed. Mean (standard deviation), median number of seizures per subject were 10.47(7.86), 9.00. Cosinor plots indicated that the function had two modes: 7–8a.m. and 4–5p.m. GEE analysis was consistent with peak seizure frequency occurrence at 6–8a.m. (p<0.0001) and 3–5p.m. (p<0.01).

Conclusions

We found a bimodal pattern of seizure occurrence in human mTLE, with peak seizure frequencies occurring between 6–8a.m. and 3–5p.m. confirming an afternoon peak, as well as a previously unsuspected morning peak in seizure occurrence that provides rationale for future investigations of antiepileptic drug chronopharmacology and informs patient counseling regarding patterns of seizure occurrence.

Keywords: Epilepsy, Mesial temporal lobe, Circadian, Periodicity

 

Back to Article Outline

1. Introduction 

For over one hundred years, physicians have observed temporal periodicity for seizure occurrence that can be classified into diurnal, nocturnal, and diffuse patterns.1 Specific patterns of seizure occurrence exist within these three classifications. For example, historical references indicated that diurnal seizures occurred most frequently on awakening and during the late afternoon.2, 3, 4

While early studies were limited by lack of objective verification of seizure occurrence such as electroencephalography (EEG), more recent research has supported the notion that seizures occur in a predictable non-random fashion.5, 6, 7 Location of the seizure focus in partial epilepsy may play a key role in determining the predominant temporal pattern. Frontal and parietal lobe seizures occur mainly during early morning hours, especially during sleep,5, 8, 9 occipital seizures occur most frequently during late afternoon hours,9 while temporal lobe seizures occur most frequently during morning and peak during late afternoon hours.9, 10

Temporal lobe seizures have a particularly robust temporal pattern, and may follow a circadian rhythm.11 A circadian rhythm can be defined as a self-sustained biological activity that spontaneously oscillates with a periodicity near 24h.10 Temporal lobe seizure occurrence in both humans and nonhuman mammals can be accurately modeled by a cosine function with a periodicity of 24h, and a peak frequency occurring during the afternoon.5, 12 Limbic seizures in rats continue to occur in a reliably entrained pattern even when the rats were deprived of all circadian cues such as daylight.11

The nature of the relationship between temporal lobe epilepsy and the circadian clock day remains unclear. While a specific circadian cycle for human temporal lobe epilepsy has been suggested, it is unclear whether these patterns are influenced by specific biological and clinical variables such as patient demographics, mesial temporal lobe seizure focus localization, or lateralization of the seizure focus.

We sought to analyze the temporal patterns of seizures in patients with well-localized mesial temporal lobe epilepsy, a homogeneous patient group distinct from previous studies, which analyzed a broader group of partial epilepsies including neocortical foci.5, 9, 12, 13 We hypothesized that peak periodicity of seizure onset was in the later afternoon hours, and that patient's clinical demographic characteristics and seizure focus localization and lateralization would have little effect on periodicity for mesial temporal lobe seizure occurrence.

Back to Article Outline

2. Methods 

2.1. Subjects 

We reviewed the UIHC Epilepsy Surgery Database from 1993 to 2004 for consecutive refractory mesial temporal lobe epilepsy patients who presented to the hospital for video-EEG monitoring for preoperative evaluation. Patients (n=60) were included if the preoperative video-EEG performed at our institution demonstrated epilepsy with a mesial temporal lobe focus, and subsequent demonstration of Engel I (seizure free) outcome after ATL (anterior temporal lobe) surgery. EEG data was recorded using Standard 10–20 system electrode placements from Grass Telefactor analog or Nihon–Kohden digital video-EEG equipment, and reviewed by an epileptologist with board certification by the American Board of Clinical Neurophysiology (E.K.S., M.G.) and American Board of Psychiatry and Neurology in Clinical Neurophysiology (E.K.S.). We analyzed each seizure for the onset time, seizure focus, and the International League Against Epilepsy (ILAE) seizure type over the patient's entire stay at the university hospital14 (specific seizure events were not included if there was an unclear onset or offset by ictal, clinical, or EEG features). Other factors considered during analysis included the patient's age (at the time of hospitalization), gender, ictal EEG lateralization (left, right, bilateral, indeterminate, none), seizure focus by lateralization of lobectomy, and type of seizure (simple partial, complex partial, secondary generalized, or subclinical electrographic seizure). Factors which could not be controlled included seizure medication use and tapering approaches or activity level. Treatment regimens and changes made to a person's activity level or medication were tailored to the patient's individual needs for diagnostic purposes, are thus were not consistent. Patients were not selected based on number of recorded seizures or length of hospital stay.

The majority of the subjects had one hospitalization where they were monitored during a period of 2–16 days. There were 60 subjects with 67 individual visits in the data set. Seven subjects were monitored for two distinct and separate visits. Subjects with two visits either had their seizure counts combined to form one single visit, or each visit was treated as a separate observation. Decisions on whether to combine or keep separate observation periods were based on the duration of the observation periods. Two subjects had their visits combined, and four subjects were treated as separate observations in the data set.

2.2. The cosinor analysis 

A cosinor analysis was performed to model seizure occurrence over a 24-h interval. Plots of seizure occurrences were created for all subjects to see if the cosinor model would fit the data. Plots were also created for seizure occurrences using all subjects simultaneously. Psychogenic seizures and poorly classified seizures were excluded from the analysis. Due to small counts of seizure occurrences for many individuals, three different time intervals were analyzed. The first is the standard 1-h interval with 24 time points (i.e. 0, 1, 2, 3, etc.). The second is a 2-h interval with 12 time points (0–1, 2–3, 3–4, etc.). The third is a 3-h interval with eight time points (0–2, 3–5, 4–6, etc.).

Proc GLIMMIX was used to fit linear models using established transformations.15 In total, 18 different models were fit. Three types of correlation structures were considered: unstructured, Toeplitz, and unstructured correlations. The models were fit using the three different time intervals: 1h, 2h, and 3h. The response was considered to have a Poisson or negative binomial distribution with an offset equal to the log of the number of days under observation. Of the 18 models, only three converged. The 1-h, 2-h, and 3-h interval models converged using Poisson regression and the Toeplitz correlation structures.

Back to Article Outline

3. Results 

The summary of the demographics of the patient population is reported in Table 1. We reviewed the epilepsy data of 60 patients. The male:female ratio was 1:2, and there were more left temporal lobe seizure foci than right (34 left; 26 right). Ages were well matched between female and male patients (age (sd): female=35.2 (9.12), male=37.3 (9.26)).

Table 1. Demographics of mesial temporal lobe epilepsy subjects (n=60).
GenderNumber of subjectsAge (sd)Seizure focus (L/R)
Male2037.3 (9.26)11/9
Female4035.2 (9.12)23/17

A total of 694 seizures were recorded across all patients. On average, patients had just over 10 (mean (sd), median=10.47 (7.86), 9.00) seizures during a mean of approximately 7.5 hospital days (mean (sd), median=7.69 (3.33), 7.00). Patients had about 1.5 seizures per day (mean (sd), median=1.54 (1.17), 1.33). Seizure characteristics are shown in Table 2. The majority of seizures occurred during the awake state, accounting for about 78% of all seizures recorded. Seizures arising from stages 1 and 2 sleep accounted for the majority of the remainder (19.7%). 64.9% of seizures were complex partial, 20% were simple partial, and the majority of the remaining seizures were secondary generalized seizures (6.5%). EEG data demonstrated that 36% of seizures were of right temporal lobe onset, and 31.3% were of left temporal onset. Another 31% of EEGs recorded had an indeterminate focus for the seizure event. Approximately 56% of all seizures occurred during day light hours (0700–1900).

Table 2. Summary of video-EEG findings from all seizures recorded (n=694).
Type of seizure recorded
Complex partial seizure451 (64.9)
Simple partial seizure139 (20.0)
Secondary generalized seizure45 (6.5)
Electrographic seizure42 (6.1)
Other11 (1.6)
Unknown4 (.58)

EEG determined seizure focus
Right250 (36.0)
Left217 (31.3)
Bilateral11 (1.6)
Indeterminate78 (11.2)
None62 (8.9)
No data76 (10.9)

Of the three different time intervals used in the analysis, all three of the plots indicated that seizures became more frequent at two periods during the 24-h day, 6–8a.m. and 3–5p.m. (Fig. 1). This pattern was similarly observed when looking at the seizure frequency for each patient, averaged across all patients (Fig. 2). GEE analysis revealed that these peaks were statistically significant (p<0.0001 and p<0.01 respectively) when compared to the lowest seizure frequency during the 24-h period (12:00–2:59a.m.). Moreover, a cosine function could fit the data well. The plots indicated that the function should have two modes, one occurring around 7–8 in the morning and the other occurring around 3–5p.m., mirroring the patterns seen in the raw seizure frequency data (Fig. 3). Demographic variables such as age, sex, and seizure lateralization by ATL did not significantly alter seizure patterns (p=.42, .51, and .55 respectively (Fig. 4, Fig. 5, Fig. 6). Interestingly, the greatest contribution to the afternoon peak appears to be from female subjects (Fig. 4) and patients aged 30–39 (Fig. 5).

  • View full-size image.
  • Fig. 2. 

    Seizure frequency for all mTLE subjects (3-h intervals). Peak seizure frequency occurred between 0600–0800 (0.242, 95% CI 0.182, 0.322) and 1500–1700 (0.200, 95% CI 0.144, 0.252).

  • View full-size image.
  • Fig. 4. 

    Seizure frequency by gender. Females had a higher mean seizure frequency during the afternoon peak hours of 1500–1700 (0.22, 95% CI 0.17, 0.34) than males (0.16, 95% CI 0.08, 0.29), but the overlapping confidence intervals during this timeframe indicate the difference was not statistically significant.

  • View full-size image.
  • Fig. 5. 

    Seizure frequency by age. Age range of 30–39 had the most robust bimodal ultradian seizure frequency peaks, but the patterns of this group were not statistically different from those under 30 or over 40 given overlapping confidence intervals.

Back to Article Outline

4. Discussion 

Our study found that temporal lobe seizures occurred in a predictable pattern peaking in the early morning from 6 to 8a.m. and in the mid-afternoon from 3 to 5p.m. These patterns occurred consistently whether combining the seizure data for all patients or when looking at patient's seizure frequency individually. Demographic features including age, seizure focus, and gender did not significantly influence the seizure patterns seen.

Our finding of a pattern of seizure periodicity in temporal lobe epilepsy is consistent with previous human studies and animal models. Two prior studies of human temporal lobe seizures revealed a periodicity of 24h, with one peak in the mid-afternoon, but no second morning peak was identified.5, 12 More recently, two smaller studies of invasively recorded mesial temporal and neocortical temporal and extratemporal seizures identified a pattern of peak seizure occurrence in both morning and late afternoon hours as seen in our large series of surgically localized mesial temporal lobe onset seizures.9, 13 Additionally, a previous comparison of human mTLE with a rat model of mTLE found that the highest seizure frequency was in the afternoon, similar to invasive intracranial human data from the two more recent published works by Duckrow and Tcheng13 and Durazzo et al.9 (1648 in the rat model and 1500 in human patients).5

Taken together, our results support the possibility that human arousal mechanisms may play a substantial role in seizure occurrence in patients with mesial temporal lobe epilepsy similar to the pattern seen in some idiopathic generalized epilepsy syndromes, particularly juvenile myoclonic epilepsy and generalized tonic–clonic seizures on awakening.16, 17 An intimate relationship of seizure occurrence directly following arousal in mesial temporal lobe epilepsy has been previously reported. It was thought to be an exceptional rather than commonplace mechanism of activation18 prior to the bimodal pattern of periodicity in mTLE being more clearly elucidated by our own and other recent case series.9, 13

While the two other recent series of intracranially monitored patients also confirm this distinctive bimodal pattern of seizure periodicity in mTLE, our methodology offers a departure from these and other prior studies by limiting analysis only to those patients with well-localized mTLE, specifically analyzing patients rendered seizure free by anterior temporal lobectomy. The biological basis of seizure activation likely following arousal in our mTLE study population is somewhat puzzling. In primary generalized epilepsy, alternating periods of drowsiness and arousal seen during the initial waking hours following arousal for the day may escalate interictal EEG bursts that culminate in clinical ictal seizure events.19 Previous research in human partial epilepsy suggests that melatonin could play a role. Melatonin appears to have antiepileptic properties.20 Melatonin is depressed in patients with mTLE in comparison to normal control subjects, and during seizures, melatonin secretion increases threefold.21 Melatonin reaches its concentration nadir during early morning hours, and thus its relative depletion during this timeframe could potentiate a morning seizure peak. In contrast, histamine, a key neurochemical modulating wakefulness in the mammalian brain, appears to mediate largely anticonvulsant rather than pro-epileptogenic effects in vitro.22 Clearly, given the complex and potentially opposing influences mediated by these neuromodulators upon seizure occurrence, further research of the neurobiological basis for morning seizure facilitation in mTLE is needed.

Whether melatonin or fluctuating drowsiness could explain the morning-peak seizure frequency seen in our retrospective study of mTLE cannot be determined. Future prospective studies should analyze melatonin and the patient's state of arousal and sleep staging preceding seizures in mTLE and other epilepsy syndromes.

The second afternoon peak of seizure occurrence in mTLE confirmed in our study has been well documented. The biologic underpinning of this afternoon peak is also poorly understood. One hypothesis is that the suprachiasmatic nucleus (SCN) plays a key role in this afternoon seizure pattern. The SCN of the anterior hypothalamus has been well studied in the mammalian central nervous system. Neurons of the SCN fire with a self-sustained periodicity having highest activity during daylight hours, which in turn influences many endogenous biological cycles, including temperature regulation, sleep–wake, and release of several critical hormones. Unique to other cycles, the firing pattern of the SCN is conserved across species, regardless of whether the organism is diurnal or nocturnal.5 Interestingly both nocturnal and diurnal animals show the same day dominated cycle of peak seizure occurrence with temporal lobe onset seizures.11 Although SCN neurons project widely throughout the brain, the SCN and the amygdala connect via the paraventricular nucleus of the thalamus.23, 24 Consequently, the afternoon peak in temporal lobe seizures may be related to the intimate connectivity between SCN and the mesial temporal region, consequently rendering the temporal lobe more sensitive to the circadian firing patterns of the SCN than are regions that are more distant and isolated from the SCN.

Our study has a number of limitations. Patients analyzed in our study were evaluated during a routine hospital stay. Therefore, light exposure, sleep schedule, activity level, medication modification and schedule, and events experienced were tailored uniquely to each patient and could not be controlled for in the current analysis. Interestingly, despite lack of ability to control for variables such as medication type, time of administration or dosage modification, or activity level changes, our study demonstrated a 24-hour pattern closely mirroring the historical observational results reported by Langdon-Down and Brain in 1929,2 a more naturalistic study that was not confounded by these variables, again underscoring that temporal lobe seizures do not have a simple periodicity of 24 hours, but rather a bimodal or dual ultradian set of seizure frequency peaks in the morning and late afternoon hours.

Sleep deprivation was not uniformly applied to the study population. Studies regarding the influence of sleep deprivation on epileptic seizure occurrence have shown conflicting results on its precise influence on seizure provocation. While sleep deprivation may increase the incidence of interictal epileptiform discharges and lower seizure threshold,25, 26 a recent randomized prospective study based in the epilepsy monitoring unit showed no difference in seizure activation between varying sleep deprivation schedules.27

Given our retrospective design, our study was not demographically balanced. We had more females than males, and more left than right seizure foci after ATL. While these differences were not statistically significant, we did find that females had a higher mean seizure frequency during the afternoon peak of 1500–1700 than did our male subjects (Fig. 4). Additional future prospective studies controlling for these demographic imbalances are necessary to determine whether any potential biological differences in seizure periodicity between genders may exist.

In summary, this study analyzed the periodicity of well-localized mesial temporal lobe seizure occurrence. We found a bimodal pattern of seizure periodicity similar to other recently published studies analyzing temporal lobe epilepsy. A strength of our study was its focus upon surgically localized mTLE, allowing us to conclude that hemispheric lateralization of the seizure focus did not significantly affect the temporal seizure pattern, nor did age or gender appear to impact seizure occurrence. A contribution to the growing body of research concerning seizure periodicity is that a bimodal peak seizure pattern exists when limiting analysis to rigorously localized temporal lobe seizure foci. Future prospective studies with larger numbers of patients are necessary to elucidate seizure periodicity in mesial temporal lobe epilepsy and other distinct epilepsy syndromes, and to limit potential confounds by controlling for the variables of daylight exposure, season, sleep deprivation, and drug withdrawal.

Back to Article Outline

Conflicts of interest statement 

None of the authors has any conflict of interest to disclose.

Back to Article Outline

Acknowledgements 

Dr. Erik K. St. Louis received research grant support from NIH K12 #510 17 3220 04000 11812015 during the timeframe of completion of the study.

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Back to Article Outline

References 

  1. Gowers W. Course of epilepsy. In:  Gowers W editors. Epilepsy and other chronic convulsive diseases: their causes, symptoms and treatment. New York: William Wood; 1885;p. 157–164
  2. Langdon-Down M, Brain WR. Time of day in relation to convulsion in epilepsy. Lancet. 1929;1:1029–1032
  3. Patry F. The relationship of time of day, sleep, and other factors to the incidence of epileptic seizures. American Journal of Psychiatry. 1931;10:789–813
  4. Griffiths GM, Fox JT. Rhythm in epilepsy. Lancet. 1938;2:409–416
  5. Quigg M, Clayburn H, Straume M, Menaker M, Bertran EH. Temporal distribution of partial seizures: comparison of an animal model with human partial epilepsy. Annals of Neurology. 1998;43:748–755
  6. Kellaway P. Sleep and epilepsy. Epilepsia. 1985;26:15–30
  7. Stevens JR, Kodoma H, Lonsbury B, Mills L. Ultradian characteristics of spontaneous seizure discharges recorded by radio telemetry in man. Electroencephalography and Clinical Neurophysiology. 1971;31:313–325
  8. Bazil CW, Walczak TS. Effects of sleep and sleep stage on epileptic and nonepileptic seizures. Epilepsia. 1997;38:56–62
  9. Durazzo TS, Spencer SS, Duckrow RB, Novotny EJ, Spencer DD, Zaveri HP. Temporal distributions of seizure occurrence from various epileptogenic regions. Neurology. 2008;70:1265–1271
  10. Quigg M. Seizures and circadian rhythms. In: Sleep and epilepsy: the clinical spectrum. New York: Elsevier Science; 2002;p. 127–142
  11. Quigg M, Clayburn H, Straume M, Menaker M, Bertram EH. Effects of circadian regulation and rest-activity state on spontaneous seizures in a rat model of limbic epilepsy. Epilepsia. 2000;41:505–509
  12. Pavlova MK, Shea SA, Bromfield EB. Day/night patterns of focal seizures. Epilepsy & Behavior. 2004;5:44–49
  13. Duckrow RB, Tcheng TK. Daily variation in an intracranial EEG feature in humans detected by a responsive neurostimulator system. Epilepsia. 2007;48:614–1620
  14. Commission on Classification and Terminology of the International League Against Epilespy . Proposal for revised clinical and electroencephalographic classification of epileptic seizures. Epilepsia. 2004;22:489–501
  15. Mikulich SK, Zerbe GO, Jones RH, Crowley TJ. Comparing linear and nonlinear mixed model approaches to cosinor analysis. Statistics in Medicine. 2003;22:3195–3211
  16. Zifkin B, Andermann E, Andermann F. Mechanisms, genetics, and pathogenesis of juvenile myoclonic epilepsy. Current Opinion in Neurology. 2005;18:147–153
  17. Wolff H, St. Louis EK, Zimmerman MB, Granner MA. Seizure occurrence in primary generalized epilepsies is predominantly diurnal. Annals of Neurology. 2008;64:S25
  18. Malow BA, Bowes RJ, Ross D. Relationship of temporal lobe seizures to sleep and arousal: a combined scalp-intracranial electrode study. Sleep. 2000;23:231–234
  19. Niedermeyer E. Primary (idiopathic) generalized epilepsy and underlying mechanisms. Clinical Electroencephalography. 1996;27:1–21
  20. Fauteck JD, Bockmann J, Böckers TM, et al. Melatonin reduces low-Mg2+ epileptiform activity in human temporal slices. Experimental Brain Research. 1995;107:321–325
  21. Bazil CW, Short D, Crispin D, Zheng W. Patients with intractable epilepsy have low melatonin, which increases following seizures. Neurology. 2000;55:1746–1748
  22. Okuma C, Hirai T, Kamei C. Mechanism of the inhibitory effect of histamine on amygdaloid-kindled seizures in rats. Epilepsia. 2001;42:1494–1500
  23. Peng Z, Grassi-Zucconi G, Bentivoglio M. Fos-related protein expression in the midline paraventricular nucleus of the rat hypothalamus: basil oscillation and relationship with limbic efferents. Experimental Brain Research. 1995;104:21–29
  24. Peng Z, Bentivaglio M. The thalamic paraventricular nucleus relays information from the suprachiasmatic nucleus to the amygdala: a combined anterograde and retrograde tracing study in the rat at the light and electron microscopic levels. Neurocytology. 2004;33:101–116
  25. Sammaritano M, Gigli G, Gotman J. Interictal spiking during wakefulness and sleep and the localization of foci in temporal lobe epilepsy. Neurology. 1991;41:290–297
  26. Rajna P, Veres J. Correlations between night sleep duration and seizure frequency in temporal lobe epilepsy. Epilepsia. 1993;34:574–579
  27. Malow BA, Passaro E, Milling C, Minecan DN, Levy K. Sleep deprivation does not affect seizure frequency during inpatient video-EEG monitoring. Neurology. 2002;59:1371–1374

PII: S1059-1311(10)00109-3

doi:10.1016/j.seizure.2010.05.005

Seizure: European Journal of Epilepsy
Volume 19, Issue 6 , Pages 347-351, July 2010

Article Tools

* Image Usage & Resolution