Does epileptic activity impair sleep-related memory consolidation in epilepsy? A critical and systematic review

Abstract

Study Objectives:

People with epilepsy often complain about disturbed sleep and cognitive impairment. Beyond seizures, the occurrence of interictal epileptic activity during sleep is also increasingly recognized to negatively impact cognitive functioning, including memory processes. The aim of this study was to critically review the effect of interictal epileptic activity on sleep-related memory consolidation.

Methods:

PubMed and PsychINFO databases were systematically searched to identify experimental studies that investigated sleep-related memory consolidation and the relationships between sleep-related epileptic activity and memory in adults and children with epilepsy. This review also highlights hypotheses regarding the potential pathophysiological mechanisms.

Results:

A total of 261 studies were identified; 27 of these met selection criteria. Only 13 studies prospectively assessed the effect of sleep on memory in epilepsy. Most studies reported no alteration of sleep-related memory consolidation in patients, with either similar retention levels following a period containing sleep (n = 5) or improved memory performance postsleep (n = 4). Two studies in children with epilepsy found impaired sleep-related memory consolidation. Ten studies, of which 6 were in childhood epilepsy syndromes, reported a debilitating effect of sleep-related epileptic activity on memory functioning.

Conclusions:

Conclusions from existing studies were hampered by small sample sizes, heterogeneous patient groups, and variations in memory assessment techniques. Overall, results to date preclude any definitive conclusions on the alteration of sleep-related memory consolidation in epilepsy. We discuss methodological considerations specific to people with epilepsy and provide suggestions on how to best investigate the relationship between epileptic activity, sleep, and memory consolidation in future studies.

Citation:

Latreille V, Schiller K, Peter-Derex L, Frauscher B. Does epilepticimpair sleep-related memory consolidation in epilepsy? A critical and systematic review. J Clin Sleep Med. 2022;18(10):2481–2495.

INTRODUCTION

Seizures are the hallmark of epilepsy, yet individuals with epilepsy experience many additional symptoms of which sleep disturbances¹ and memory impairment² are among the most frequent and debilitating complaints. The mechanisms by which seizures influence memory functioning are poorly understood and are likely to be multifactorial (ie, neuropathology, mood disorders, antiepileptic drugs^3,4). Beyond seizures, the occurrence of interictal epileptiform discharges (IEDs) is also increasingly recognized to negatively impact memory processes. Indeed, in adults with drug-resistant focal epilepsy, some studies have shown that the presence of IEDs during the learning, consolidation, and/or retrieval phases of a memory task results in impaired performances.^5–8

While the learning and retrieval phases of memory exclusively occur during wakefulness, the consolidation of newly acquired memories into long-term memory is thought to preferentially occur during offline states, including unoccupied quiet rest and sleep.⁹ Indeed, in healthy young adults multiple studies have shown the benefit of sleep on learning and memory processes.¹⁰ It was thus proposed that epilepsy patients may benefit less from a night of sleep relative to a control group, especially if this sleep period contains epileptic activity. By disrupting sleep-related memory consolidation processes, occurrence of epileptic activity during sleep would disrupt the physiological mechanisms responsible for strengthening memories. However, the evidence supporting this hypothesis in humans remains scarce. We provide an extensive and critical review to highlight the main findings of: (1) the effect of epilepsy on sleep-related memory consolidation, (2) the interactions between interictal epileptic activity during sleep and memory, and (3) the potential pathophysiological mechanisms underlying impaired sleep-related memory consolidation in epilepsy.

We conclude by discussing the methodological considerations specific to people with epilepsy and provide suggestions on how to best investigate the relationship between epileptic activity, sleep microstructure, and sleep-related memory consolidation in future studies.

METHODS

We performed a systematic search in November 2021 of the electronic databases of PubMed and PsychINFO for original articles, following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines. Articles were screened for eligibility by 2 reviewers (V.L., K.S.). Search terms are detailed in Figure 1 and the study selection criteria are presented in Table 1. Inclusion criteria were: (1) clinical diagnosis of epilepsy, (2) assessment of epileptic activity by electroencephalography (EEG), (3) assessment of sleep either by formal polysomnography or EEG, (4) objective memory testing, and (5) articles written in the English, French, or German language. Articles were excluded if they investigated the effects of pharmacological or surgical interventions. Additional articles were identified from citation lists. A total of 27 studies fulfilled the selection criteria. Last, animal studies were reviewed specifically for discussion on the neurophysiological mechanisms supporting sleep-related memory consolidation. This study is a systematic review of the published literature and is therefore exempt from institutional review board approval.

Figure 1. Literature search flowchart following PRISMA guidelines.

PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analysis

Table 1.

Study selection criteria.

Inclusion Criteria	Clinical diagnosis of epilepsy
	Assessment of epileptic activity by electroencephalography (EEG)
	Assessment of sleep either by formal polysomnography (PSG) or EEG
	Assessment of memory by objective cognitive testing
	For sleep-related memory consolidation section: Memory was assessed before and after an interval comprising a sleep period.
	For correlation and association analyses section: Memory was independently assessed from the sleep recording and can be separated by several months.
Exclusion Criteria	Reviews or case studies
	Papers not written in the English, French, or German language
	Papers investigating the effects of pharmacological, nonpharmacological, or surgical interventions on sleep or memory functions

RESULTS

Sleep-related memory consolidation in epilepsy

From 2011 to 2021, 13 studies specifically investigated the effect of sleep on memory consolidation in epilepsy. As highlighted in Table 2, these studies used different methodological designs and heterogeneous epilepsy populations, which limits direct study comparisons. Nonetheless, the methodological design is important when studying sleep-related memory consolidation. Notably, to infer the benefits of sleep over the mere passage of time (wakefulness), the interval periods between learning and retrieval should be of similar duration, one containing wakefulness and the other containing sleep. Half of the studies (7/13 studies) used a sleep-wake memory paradigm, and the remaining studies compared memory performances using an interval condition comprising either only a sleep period (3/13 studies) or a mix of sleep and wake periods (3/13 studies). As an estimate of effect sizes, we provide Cohen’s d values along with the main statistically significant results in Table 2. If not provided in the published article, effect sizes were calculated whenever possible using reported means and standard deviations or using results of independent t tests. Consistent with the literature in healthy adults, all but 1 study¹¹ could demonstrate a benefit of sleep over wakefulness on memory retention in their healthy comparison sample, with medium to large effect sizes (0.40–4). However, in epilepsy patients, especially in the adult population, results are less straightforward and can be summarized as below.

Table 2.

Overview of prospective studies assessing sleep-related memory consolidation in children and adults with epilepsy.

Study	Sample Analyzed	Measurements	Target Variables	Main Results and Effect Sizes (Cohen’s d)†	Impaired Consolidation during Sleep
Deak et al 201111	TLE, n = 7 Mage = 44.0 y (range N/A) Controls, n = 9 Mage = 44.7 y (range N/A)	a) Scalp + PSG + Actigraphy: no IED detection b) Verbal memory: Learning of 12 words, followed by immediate free recall and cued recall for omitted words (selective reminding procedure) until 2 consecutive trials with 100% correct responses, and delayed free recall at 30 min. Postinterval delayed free recall and recognition trials c) Motor sequence learning Learning of a finger tapping motor sequence task over 12 training trials, followed by 12 retest trials 24 h later	Interval sleep and wake conditions for verbal memory: ∼12 h Interval (sleep + wake) condition for motor learning: 24 h N3 % Verbal memory: recall at last learning trial, 30-min delayed recall and postinterval delayed recall, memory retention Motor learning: performance speed change from the last 6 trials at training to the last 6 trials at retest	1) In the sleep condition, no differences between patients and controls on verbal memory performance at immediate or delayed recalls 2) In the wake condition, patients had poorer memory retention and recalled 30% fewer words than controls after 12 h of wakefulness 3) Better verbal memory retention associated with greater N3% 4) No group differences on motor learning trials at training or retest	No
Urbain et al 201116	Idiopathic epilepsy syndromes of childhood, n = 4 Mage = 8.3 (range 7–10) y Controls (sleep group), n = 12 Mage = 9.3 y (range N/A) Controls (wake group), n = 12 Mage = 9.9 y (range N/A)	a) Scalp EEG: visual IED detection b) Verbal memory Learning of 22–32 word-pairs (age-adjusted) until the criterion of 60% correct responses was met, with immediate cued recall and postinterval delayed cued recall	Interval sleep condition for patients: ∼11 h Interval sleep and wake conditions for control groups: ∼11 h Verbal memory: % correct responses at immediate and delayed recalls, memory retention	1) Better memory retention in controls who slept than those who remained awake (d = 0.88) 2) Patients had poorer memory retention following sleep relative to controls	Yes
Bjørnaes et al 201323	Electrical status epilepticus during sleep, n = 23 Mage N/A (range 5–10 y)	a) 24-h ambulatory scalp EEG: semiautomatic IED detection b) Verbal memory Learning of 10 words until the criterion of 100% correct responses was met, followed by postsleep delayed free recall c) Visual recognition i) Presentation of 16 faces (5 s each), followed by immediate and postsleep delayed multiple-choice recognition trials ii) Presentation of 16 designs (5 s each), followed by immediate and postsleep delayed multiple-choice recognition trials	Interval (sleep + wake) condition: unknown Spike index: % of time with < 3 s between IED in 10-min epochs during NREM, REM, and wake Verbal memory: best learning trial, postsleep delayed free recall and memory retention Visual memory: immediate and postsleep recognition scores	1) No correlation between spike indexes during sleep and postsleep memory retention 2) Higher spike index during NREM and REM sleep associated with poorer verbal learning	Not examined‡
Fitzgerald et al 201318	Focal and generalized epilepsy (diagnosed or probable), n = 39 Mage = 38.0 y (range N/A) Controls, n = 15 Mage = 40.3 y (range N/A)	a) 24-h ambulatory scalp EEG (epilepsy only): visual IED detection b) Verbal memory Learning of 13 abstract words until the criterion of 85% correct responses was met, followed by delayed free recalls at 30-min, 24-h, and 4 days c) Visual memory (n = 32) Learning of 13 abstract designs until the criterion of 75% correct responses was met, followed by delayed free recalls at 30-min, 24-h, and 4 days	Interval (sleep + wake) condition: 24 h Total sleep time, N3 and REM duration, presence of naps Verbal and visual memory: % of information lost between 30-min and 24-h delayed recalls	1) No correlation between nighttime sleep variables and % of information lost on memory tasks 2) Presence of daytime napping associated with greater retention on the visual memory task	No
Atherton et al 201413	Focal epilepsy with impaired awareness seizures (transient epileptic amnesia), n = 11 Mage = 67.7 (range 60–76) y Controls, n = 12 Mage = 63.5 y (range N/A)	a) Sleep EEG data N/A b) Verbal memory (n = 15) A-B, A-C interference learning paradigm: Learning of 30 word-pairs (A-B) until the criterion of 60% correct responses was met, with immediate cued recall and 30 min delayed cued recall. Next-morning single-exposure learning of word-pairs A-B, followed by interference pairs (A-C) and 10 min delayed (postinterval) cued recall of A-B and A-C c) Video events task Presentation of a short film (3 min) with identification of plot points, followed by postinterval delayed free recall of plots points from the film	Interval sleep and wake conditions: 12 h Verbal memory: # correct A-B word-pairs at immediate recall, 30 min delayed recall, and 12 h delayed recall; # correct interference A-C pairs Video events task: % of correct plot points recalled	1) Greater memory recall following sleep than wake condition on the verbal task’s target word-pairs (A-B), but similarly in patients (d = 3.9) and controls (d = 1.1) 2) No beneficial effect of sleep on the verbal task’s interference pairs (A-C) 3) No beneficial effect of sleep on the video events task	No
Moroni et al 201419	Drug-resistant focal epilepsy, n = 6 Mage = 26.8 (range 18–38) y TLE, n = 4 FLE, n = 1 PLE, n = 1	a) Scalp + intracranial EEG + PSG: visual IED detection (rejection) b) Spatial memory i) Learning phase: learning of location of 6 landmarks in a 3D virtual environment to form a cognitive map and then navigate within the virtual environment, with recalls (mapping) of landmarks locations every 2 min until 100% correct responses; ii) Immediate and postsleep delayed retrievals: navigation route to reach target landmarks locations over 9 trials	Interval sleep condition: ∼19 h Relative spectral power in low/high delta (0.5–2 Hz, 2.1–4 Hz), theta (4.1–8 Hz), alpha (8.1–12 Hz), sigma (12.1–16 Hz), and beta (16.1–30 Hz) over hippocampal and neocortical areas during first NREM sleep cycle Spatial memory: median time delay to reach target landmarks between immediate and postsleep delayed recall	1) No difference in task performance following sleep, but 4/6 patients improved their performance postsleep 2) Increased hippocampal high-delta power during postlearning sleep relative to baseline sleep in all patients, which was associated with better postsleep spatial performance	No
Sud et al 201420	Focal to bilateral or focal drug-resistant epilepsy, n = 10 Mage = 11.4 (range 8–17) y TLE, n = 7 Extra-TLE, n = 3	a) Scalp EEG + actigraphy: visual IED detection b) Verbal memory Learning of 15 words over 5 trials (Rey Auditory Verbal Learning Test), followed by immediate and postinterval delayed free recalls	Interval sleep and wake conditions: ∼11 h Average IED rate during NREM and wake (25 min epochs) Verbal memory: correct recall at the last learning trial, memory retention	1) Better memory retention following sleep than wake condition; however, exclusion of an outlier made this difference nonsignificant 2) Increased IED rate during sleep compared to wake	No
Galer et al 201517	Idiopathic focal epilepsy, n = 15 Mage = 9.0 (range 6–12) y Centrotemporal, n = 11 Occipital, n = 4 Controls, n = 8 Mage = 9.5 (range 7–12) y	a) Scalp EEG + PSG: automatic or visual IED detection b) Visuospatial memory (n = 9) Learning of localization of 12 object-pairs (2D object localization) until the criterion of 60% correct responses was met, with immediate cued recall and postsleep delayed cued recall c) Verbal memory (n = 15) Learning of 22–32 word-pairs (age-adjusted) until the criterion of 60% correct responses was met, with immediate cued recall and postsleep delayed cued recall	Interval sleep condition: ∼11 h Spike-wave index during first 30 min of NREM sleep Visuospatial and verbal memory: memory retention	1) Poorer memory retention following sleep in patients relative to controls on the visuospatial (d = 1.39) and verbal (d = 0.91) tasks 2) Higher spike–wave index during NREM sleep associated with poorer memory retention on the visuospatial task	Yes
Atherton et al 201615	Focal epilepsy with impaired awareness seizures (transient epileptic amnesia), n = 11 Mage = 67.7 (range 60–76) y Controls, n = 12 Mage = 63.5 y (range N/A)	a) Sleep EEG + PSG: visual IED detection, automatic spindle detection b) Verbal memory Same as in Atherton et al 2014 (A-B, A-C interference learning paradigm), with an additional recall after 1 week	Interval sleep and wake conditions: 12-h N3% Spindle rate in N3 sleep Sleep benefit: memory retention difference after 12 h and 1 week between sleep and wake conditions	1) Greater memory recall following sleep than wake condition on the verbal task over 12 h, but similarly in patients (d = 3.9) and controls (d = 1.1) as reported in Atherton et al, 2014 2) In patients, greater sleep benefits over 12 h were associated with lower N3% 3) In controls, greater sleep benefits over 1 week were associated with greater N3% 4) No relationship between sleep benefit and spindle rates	No
Sarkis et al 201612	Focal epilepsy, n = 11 Mage = 39.7 (range 21–56) y TLE, n = 2 FLE, n = 1 FTLE, n = 8	a) Scalp EEG + PSG: no IED detection, automatic spindle detection b) Visuospatial memory Learning of localization of 15 object-pairs (2D object localization) until the criterion of 40% correct responses was met, with immediate cued recall and postinterval delayed cued recall	Interval sleep and wake conditions: ∼12 h Presence of seizures Duration of N2 and N3 sleep Sleep spindles in N2 and N3: count, density, duration Visuospatial memory: memory retention	1) Better memory retention following sleep than wake condition on visuospatial task (d = 1.09) 2) Patients with nocturnal seizures (n = 3) tended to have lower memory retention 3) Longer N3 duration associated with better memory retention 4) No relationship between sleep benefit and spindle rates	No
Chan et al 201714	Focal epilepsy, n = 22 Mage = 11.5 (range 6–16) y TLE, n = 6 Extra-TLE, n = 10 Unknown, n = 6 Controls, n = 21 Mage = 10.6 (range 6–16) y	a) Scalp EEG + PSG (intracranial EEG in 1 patient): visual IED detection b) Visuospatial memory Learning of localization of 12 object-pairs (2D object localization) until the criterion of 40% correct responses was met, with immediate cued recall and postinterval delayed cued recall c) Verbal memory Learning of 20–40 word-pairs (age-adjusted) until the criterion of 60% correct responses was met, with immediate cued recall and postinterval delayed cued recall	Interval sleep condition: ∼15 h Interval wake condition: ∼8 h IED rate during first 2 sleep cycles Visuospatial and verbal memory: memory retention Sleep benefit: memory retention difference between sleep and wake conditions	1) Better memory retention following sleep than wake condition on verbal task (d = 0.67) and visuospatial (d = 0.40) tasks, but similarly in patients and controls 2) Higher IED rate during sleep associated with lower sleep benefit on verbal task	No
Van Schalkwijk et al 201822	Focal epilepsy, n = 25 Mage = 39.4 y (range 24–59) TLE, n = 16 Extra-TLE, n = 9	a) 24-h ambulatory scalp EEG: visual seizure and IED detection b) Autobiographical event List of events and actions to be recalled after delays of 30 min and 24 h (without forewarning) c) Story memory Story learning followed by immediate recall (repeated until 80% correct recall), and 30 min and 24-h delayed recalls d) Visuospatial memory Copy of Rey Complex Figure followed by delayed recalls at 30 min and 24 h e) Verbal learning Learning of 15 words over 5 trials (Rey Auditory Verbal Learning Test) the next morning, no delayed recalls	Interval (sleep + wake) condition: 24 h Presence of seizures or IEDs during 24 h REM latency and N3% Memory retention on tasks b, c, and d: recall % at 24 h relative to 30 min Next-morning verbal learning: total learning score (word list)	1) Longer REM latency associated with poorer autobiographical overnight retention 2) Presence of seizures associated with poorer overnight story retention, while presence of IEDs associated with poorer overnight visuospatial retention 3) Higher N3% overnight associated with better verbal learning the next day	Not examined‡
Storz et al 202021	Self-limited focal epilepsies of childhood, n = 14 Mage = 8.7 (range 5–11) y Controls, n = 15 Mage = 7.5 (range 6–9) y	a) Scalp EEG + PSG: visual IED detection b) Visuospatial memory Learning of the localization of 15 object-pairs (2D object localization) until the criterion of 40% correct responses was met, with immediate cued recall and postsleep delayed cued recall c) Verbal memory Learning of 20 word-pairs until criteria of 60% correct responses, with immediate cued recall and postsleep delayed cued recall	Interval sleep condition: ∼12.5 h Spike-wave index during NREM sleep (first 10 min and last 10 min of sleep) IED lateralization: left, right or bilateral Visuospatial and verbal memory: immediate recall, memory retention, postsleep delayed recalls	1) No overnight change in memory retention on visuospatial or verbal tasks in patients, but controls had better verbal memory retention postsleep (d = 2.3) 2) No difference between patients and controls on immediate or postsleep delayed recalls 3) Higher IED rate across NREM sleep cycles associated with poorer postsleep delayed recall on the verbal task, but not memory retention 4) No effect of IED localization on any memory measure	No

†If not provided in the published article, effect sizes (Cohen’s d) were calculated where possible using reported means and standard deviations or using independent t tests study results. ‡No conclusion can be drawn because the study did not assess the effect of the interval period (comprised of both sleep and wake) on next-day memory retention. EEG = electroencephalography, FLE = frontal lobe epilepsy, FTLE = fronto-temporal lobe epilepsy, IED = interictal epileptiform discharges, M_age = mean age, NREM = nonrapid eye movement, OLE = occipital lobe epilepsy, PLE = parietal lobe epilepsy, PSG = polysomnography, TLE = temporal lobe epilepsy.

Sleep benefit

In 1 study (without a control group), a benefit of sleep over wakefulness on nonverbal memory retention was found in a small sample of adults with focal epilepsy.¹² Comparing epilepsy patients to an age-matched control group, 3 studies (from 2 distinct study samples) reported a benefit of sleep on verbal and/or nonverbal memory tasks, although this was similar between patients and the control group,^13–15 with medium to large effect sizes. In other words, they did not find impaired sleep-related memory consolidation but rather a sleep benefit in their patient population.

Impaired sleep-related memory consolidation

By contrast, 2 studies performed in children with childhood epilepsies reported lower memory retention following sleep relative to the control group,^16,17 with large effect sizes.

No sleep effects

Finally, 5 studies found no statistically significant change in memory retention scores postsleep (or combined sleep and wakefulness¹⁸) in adults^11,19 and children with epilepsy,^20,21 suggesting similar forgetting rates following a period of sleep or wakefulness.

Of note, 2 prospective studies^22,23 included in Table 2 could not be grouped into any of these 3 category findings because they did not assess the effect of the interval period (comprised of both sleep and wakefulness) on next-day memory retention; rather, they investigated the effect of sleep-related IEDs on memory function (described in the section “Correlation and association analyses between interictal epileptic activity during sleep and memory”). Methodological designs also differ regarding (1) inclusion of a healthy control group (8/13 studies), (2) assessment of both verbal and nonverbal memory (7/13 studies), and (3) target memory variables as a proxy of sleep-related memory consolidation (eg, correct items at delayed recall, memory retention, sleep benefit). Earlier studies used more varied memory testing paradigms, while more recent studies employed a more systematic approach, involving either a word-pair task or the 2-dimensional (2D) object localization task as a proxy for verbal and nonverbal memory. Additionally, in half of these studies (6/13), patients did not undergo full polysomnography (PSG) and thus sleep staging was performed based on the EEG. Overall, the level of training of the sleep scorers is rarely stated explicitly, and training in both sleep and epilepsy neurophysiology is required for correct interpretation of the sleep EEG of individuals with epilepsy.²⁴ Although the standard American Academy of Sleep Medicine criteria²⁵ requires PSG (EEG and additional electrooculography and chin electromyography) for sleep stage scoring, it is still feasible to score sleep on the EEG alone, especially for non–rapid eye movement (NREM) sleep stages N2 and N3. For scoring of rapid eye movement sleep, the addition of chin electromyography is a prerequisite. Finally, sleep scoring can be challenging in epilepsy²⁴ and thus should be performed by trained neurophysiologists experienced in both sleep and epilepsy.

Six studies assessed EEG macrostructure (eg, sleep stage duration) or microstructure (eg, spectral power) for correlates of memory retention in epilepsy. They found that more time spent in N3 sleep was associated with better memory retention,^11,12 and that increased delta power (2–4 Hz) in the hippocampus during postlearning sleep was associated with improved performance on a spatial memory task.¹⁹ Relatedly, van Schalkwijk and colleagues²² found that more time spent in N3 sleep overnight was associated with better verbal learning abilities (encoding) the next morning. By contrast, an opposite relationship between sleep benefits and N3 sleep was found in older adults with focal epilepsy with impaired awareness seizures (transient epileptic amnesia), suggesting a deleterious effect of N3 sleep on memory retention in this older patient group.¹⁵ Finally, one study including 39 patients with focal or generalized epilepsy failed to find statistically significant correlations between nighttime sleep microarchitecture variables and memory retention on verbal and nonverbal memory tasks at a 24-hour delayed recall, but the presence of daytime napping following learning was associated with less forgetting of information on the nonverbal memory task 24 hours later.¹⁸

In general, sample sizes were too small to perform epilepsy subgroup comparisons, for example, in temporal lobe epilepsy (TLE) vs extra-TLE, or left- vs right-hemispheric epilepsy, and therefore no conclusion can be made with regard to localization or lateralization effects on sleep-related memory consolidation. Future studies using larger sample sizes should investigate whether lateralization of IEDs during sleep has a material-specific effect (eg, verbal vs nonverbal) on memory consolidation. Moreover, the current sample sizes precluded any detailed analysis of other clinical and sociodemographic influences (eg, antiepileptic drugs [AED], sex) on sleep-related memory consolidation in epilepsy patients.²⁹ This is of particular interest since some AED and benzodiazepines do modify sleep macroarchitecture (eg, N3 sleep²⁹) and microarchitecture (eg, sleep spindles^30,31). These variables should be considered in future work.

Correlation and association analyses between interictal epileptic activity during sleep and memory

One way to assess whether sleep-related IEDs disrupt memory processes is to prospectively examine how memory performances change after a night of sleep and whether these changes correlate with ongoing epileptic activity during sleep. From the 13 prospective studies presented in Table 2, only 6 analyzed IED rates during sleep or wakefulness periods in their samples.^{14,17,20–23} Among these, 3 found that a higher IED rate during sleep^14,17 or the 24-hour interval period comprising both sleep and wake²² was associated with lower overnight memory retention. These associations are observed more consistently in children with epilepsy, who are generally more active spikers than adults with focal epilepsy. It therefore raises the question whether a certain level of IEDs is needed to disrupt memory consolidation processes during sleep. Indeed, it was suggested that occasional focal IEDs during sleep may be less disruptive than very frequent and widespread IEDs during sleep, as is commonly observed in children with epilepsy.³² In most studies, no seizures occurred during the experimentation, but similarly to IEDs, their presence during the interval period²² or during sleep¹² was associated with lower overnight memory retention scores. Finally, in 1 study, a higher IED rate during NREM sleep was associated with lower verbal learning abilities but not overnight memory retention.²³ One recent study by Lambert and colleagues³³ is worth mentioning, although it did not directly test sleep-related memory consolidation as recall was performed 1 week after learning and not the following morning (not included in Table 2). The authors prospectively studied 20 patients (18 with mesiotemporal epilepsy) using intracranial EEG recordings along with multiple memory testing sessions (immediate, 30-min delayed, and 1-week delayed recalls), and found that the amount of hippocampal IEDs during NREM sleep was inversely correlated with verbal memory performance after 1 week. Interestingly, this association was specific to sleep, such that there was no association between hippocampal IEDs during wakefulness and memory performance at 1-week delay. Moreover, the number of seizures was the main predictive factor for impaired verbal memory impairment after 1 week.³³ Altogether, these studies highlight a deleterious effect not only of seizures but also of sleep-related IEDs on memory consolidation. To better measure the extent of this effect and its underlying mechanism, studies using invasive methods such as intracranial EEG combined with a sleep-wake memory paradigm are warranted.

In Table 3, we review the existing studies which cross-correlated EEG-detected epileptic activity during sleep and memory function in adults and children with various epilepsy syndromes. Specifically, we included in Table 3 studies which investigated (1) sleep as objectively assessed with EEG (with or without PSG), (2) epileptic activity in the EEG, and (3) objective memory performance based on the clinical neuropsychological evaluation. For these studies, the EEG and/or PSG recordings were not systematically performed on the same day of the neuropsychological exam, and the time interval between the recordings and memory testing likely varied from weeks to months, although it was only rarely provided in the methods.

Table 3.

Overview of studies assessing the effect of epileptic activity during sleep on memory in children and adults.

Study	Sample Analyzed	Measurements	Target Variables	Main Results
Baglietto et al 200128	Childhood epilepsy with centrotemporal spikes, n = 9 Mage = 8.9 (range 6–11) y Controls, n = 9 (range N/A)	a) Scalp EEG + PSG (T0 = IED activation during sleep; T1 = IED remission): visual IED detection b) Verbal working memory Digit span: repetition of series of numbers forward and backward c) Visuospatial working memory Corsi’s block-tapping: reproduction of sequences of movements by tapping blocks	IED index during NREM sleep at T0 and T1 (interval range, 6–24 months) Working memory: total scores at T0 and exam closest to T1 (interval range, 6–12 months)	1) At T0 (IED activation during NREM sleep), patients had poorer visuospatial working memory than controls 2) At T1 (IED remission), patients’ performances on working memory tasks normalized to the controls’ levels
Ebus et al 201241	Children with various epilepsy syndromes, n = 182 Mage = 9.5 (range 6–12) y	a) 24-h ambulatory scalp EEG: visual IED detection b) Verbal memory Learning of 15 words over 5 trials (Rey Auditory Verbal Learning Test), followed by immediate and delayed recalls c) Visuospatial memory Copy of Rey Complex Figure followed by delayed recall at 20 min Memory and EEG recordings were performed within a period of 2 weeks	IED index during first hour of sleep (< 1% vs > 1%, (< 10% vs > 10%, (< 10% vs > 50% of sleep time) IED laterality: dominant vs nondominant Verbal and visuospatial memory: delayed recall	1) No difference between patients with high or low IED indices during sleep on memory tasks 2) Occurrence of IED in the nondominant hemisphere during sleep associated with poorer visuospatial memory delayed recall scores
Hwang et al 201339	Childhood epilepsy with centrotemporal spikes, n = 33 Mage = 8.2 (range 5–15) y	a) Scalp EEG: visual IED detection b) Verbal memory Learning of 15 words over 5 trials (Rey Auditory Verbal Learning Test), followed by immediate and delayed recalls c) Visuospatial memory Copy of Rey Complex Figure followed by delayed recall at 20 min	Spike-wave index during sleep: high (> 10 IED/min) vs low (< 10 IED/min) IED localization: left, right or bilateral Verbal memory: learning, immediate and delayed recalls Visuospatial memory: delayed recall	1) No difference between patients with high or low spike-wave indices during sleep on memory tasks 2) Occurrence of IED in the left hemisphere during sleep associated with poorer verbal memory delayed recall
Lv et al 201342	Focal epilepsy, n = 67 Mage = 27.0 (18–48) y FLE, n = 17 TLE, n = 30 OLE, n = 8 Other, n = 1	a) 24-h ambulatory scalp EEG: visual IED detection b) Memory scale Subtests assessing working memory, learning, short-term memory, long-term memory, and verbal and nonverbal memory	IED index during sleep (< 1% vs 1%–10% vs 10%–50% vs > 50% sleep time) IED distribution: focal vs nonfocal Focal IED localization: frontal, temporal Memory scale score	1) No difference between patients with various IED indices during sleep on memory scores 2) No difference between patients with focal and nonfocal IEDs during sleep on memory scores 3) Patients with temporal IEDs during sleep had poorer memory scores
Glennon et al 201640	Drug-resistant focal epilepsy, n = 103 Mage = 12.7 (range N/A) y FTLE, n = 20 TLE, n = 38 FTLE, n = 12 Other, n = 24	a) Scalp EEG: visual IED detection (sleep in n = 87) b) Memory scale Subtests assessing verbal (n = 54) and nonverbal memory (n = 57)	IED frequency: frequent (> 1 IED/10 sec) vs infrequent/none (< 1 IED/10 sec) Presence of sleep-related IED enhancement (increase in IED frequency from wake to sleep) Verbal and nonverbal memory scores	1) No correlation between IED frequency during sleep and visual or verbal memory scores 2) No correlation between sleep-enhanced IEDs and visual or verbal memory scores
Liu et al 201635	Controlled epilepsy, n = 164 Mage = 37.7 (range N/A) y FTLE (Left, n = 123; Right, n = 32) Other, n = 9	a) 24-h ambulatory scalp EEG: visual IED detection b) Verbal memory Learning of 15 words over 5 trials (Rey Auditory Verbal Learning Test), followed by immediate and delayed recalls c) Memory scale 3-subtests (Addenbrooke’s Cognitive Examination-Revised, score 0–26) assessing working memory, semantic memory, and short-term verbal memory Memory was tested on the day of the EEG recordings	Presence of IEDs during sleep (n = 61) Verbal memory: immediate and delayed recalls Memory scale total score	1) No difference between patients with and without IEDs during sleep on verbal memory task 2) Patients with IEDs during sleep performed more poorly on the memory scale than those without IEDs 3) Higher IED rate during NREM sleep associated with poorer scores on the memory scale
Maltoni et al 201636	Epileptic encephalopathy with continuous spike-and-wave during sleep, n = 61 Mage = 6.1 (range N/A) y	a) Diurnal sleep-deprived scalp EEG: detection method N/A b) Verbal short-term memory Presentation of 24 words, followed by immediate multiple-choice recognition trials c) Visual short-term memory Presentation of 24 figures, followed by immediate multiple–choice recognition trials	Spike-wave rate during NREM sleep (≥ 25/min) Verbal and visual memory recognition scores	1) No correlation between spike-wave rate during NREM sleep and verbal or visual memory performances
Nissenkorn et al 201738	Childhood epilepsy with centrotemporal spikes, n = 34 Mage = 6.2 (range 4–10) y	a) Scalp EEG: visual IED detection b) Short-term verbal memory Repetition of meaningful sentences	Spike-wave index during 1 h of NREM sleep: high (≥ 50%) vs low index (≤ 50% sleep time) Verbal memory score	1) No difference between patients with high and low spike-wave indices during sleep on short-term memory scores
Chakravarty et al 201934	Drug-resistant focal epilepsy, n = 37 Mage = 24.7 (range N/A) y TLE, n = 19 Controlled focal epilepsy, n = 37 Mage = 24.0 (range N/A) y TLE, n = 26 Controls, n = 40 Mage = 25.1 (range N/A) y	a) Scalp EEG + PSG: visual IED detection b) Memory scale 10-subtests assessing remote memory, recent memory, working memory, attention, verbal memory retention, visual memory retention Memory was tested the day before or after sleep recordings	Seizure refractoriness: comparison of drug-resistant vs controlled epilepsy IED rate during NREM sleep Memory scale score	1) Patients with drug-resistant epilepsy had higher IED rates during N2 sleep than those with controlled epilepsy 2) Higher IED rate during N2 sleep associated with poorer memory scores
Novak et al 201937	Drug-resistant focal epilepsy with generalized quasiperiodic epileptiform activity in sleep (“hurdles”), n = 24 Mage = 8.9 (range 2–17) y Drug-resistant focal epilepsy without “hurdles”, n = 24 Mage = 10.5 (range 3–18) y	a) 24-h ambulatory scalp EEG: semiautomatic IED detection, visual marking of “hurdles” b) Verbal memory Subtests including learning of stories and word-pairs, followed by immediate recall c) Visuospatial memory Subtests including learning of dot locations and faces, followed by immediate recall	Presence of “hurdles”: EEG pattern consisting of generalized quasiperiodic IED in NREM sleep IED index during NREM sleep (first 30–45 min of the first sleep cycle) and restful wake (10–15 min) Verbal and visuospatial memory scores	1) No difference between patients with and without “hurdles” on verbal and visuospatial memory scores
Munckhof et al 202043	Epileptic encephalopathy with electrical status epilepticus in sleep, n = 29 Mage = 6.7 (range N/A) y	a) Scalp EEG: automatic spike detection b) Verbal working memory Digit span: repetition of series of numbers forward and backward Memory and EEG recordings were performed within a period of 4 months	Spike-wave index during first and last hour of NREM sleep Digit span forward score	1) No correlation between spike-wave index and verbal working memory scores
Zhang et al 202027	Childhood epilepsy with centrotemporal spikes, n = 61 Mage = 10.8 (range 7–17) y Controls, n = 60 Mage = 10.6 (range 7–16) y	a) Sleep-deprived scalp EEG: visual IED detection b) Verbal memory Paired association learning test (details N/A) c) Verbal working memory Digit span: repetition of series of numbers forward and backward d) Spatial memory (details N/A)	Spike-wave index during NREM sleep (first sleep cycle) Digit span forward and backward scores	1) Patients with a high (≥ 55%) spike-wave index during NREM sleep had poorer verbal working memory and spatial memory scores than those with a lower index (< 55%)
Zhu et al 202126	TLE, n = 34 Mage = 30.7 (range N/A) y Controls, n = 28 Mage = 32.9 (range N/A) y	a) Scalp EEG (epilepsy only): visual IED detection b) Verbal working memory Digit span: repetition of series of numbers forward and backward c) Short-term visual memory Presentation of 8 cards including graphics, Chinese characters, and symbols, followed by immediate multiple-choice recognition trials d) Eye-tracking Short-term Memory Game Presentation of targets (colored dots) over a 4 × 3 array with increasing difficulty (1‒4 targets), followed by multiple-choice recognition trial (4 trials)	Total number of bilateral IEDs during NREM sleep Digit span forward and backward scores Short-term visual memory total score Eye-tracking short-term memory: visit counts, total visit time, average visit time, and time of first fixation (during presentation and recognition of targets)	1) Higher number of IEDs during NREM sleep associated with poorer Digit span backward scores 2) No correlation between the number of IEDs and eye-tracking short-term memory variables

EEG = electroencephalography, FLE = frontal lobe epilepsy, FTLE = fronto-temporal lobe epilepsy, IED = interictal epileptiform discharges, M_age = mean age, NREM = nonrapid eye movement, OLE = occipital lobe epilepsy, REM = rapid eye movement, TLE = temporal lobe epilepsy.

Out of the 13 identified studies,^{26–28,34–43} 7 found statistically significant relationships between IEDs during sleep and memory performance, with higher IED rates being associated with poorer verbal or nonverbal memory recall in childhood epilepsy syndromes^27,28,39,41 and adults with epilepsy.^34,35,42 These studies used either working/short-term memory tasks^26–28 or global measures of memory function with borderline to weak results.^34,35 When more specific tests of learning and memory were used, even in a large cohort of 164 adults with epilepsy,³⁵ studies failed to demonstrate significant negative correlations between sleep-related IEDs and memory performances.^{36–38,40,41} These inconsistencies might be explained in part by the methodological design and retrospective nature of these studies, where the EEG analyzed for epileptic activity was independently recorded from the memory testing sessions and sometimes separated by several months. Another confounding factor relates to the heterogeneity of the epilepsy population investigated, both from the epilepsy syndrome point of view (eg, localization/lateralization of IEDs, IED frequency, presence of lesions) and neuropsychological functioning (eg, verbal or nonverbal memory deficits). The potential insensitivity of conventional neuropsychological tests to detect cognitive deficits in specific patient populations was also raised by 1 study.⁴³ Moreover, although the use of scalp EEG is convenient for clinical studies, this technique limits the assessment of the “real” epilepsy burden, especially in mesiotemporal epilepsy cases, in which measurement of hippocampal activity is important yet not feasible using scalp EEG. Finally, it should be noted that the method chosen for IED detection—visual or automatic—may contribute to the heterogeneity of the results.

Given the focus of this review on a better understanding of how epilepsy might impair sleep-related memory consolidation processes, we included in the present article only studies that specifically assessed memory functioning. As such, this work does not incorporate studies that investigated the relationships between general cognitive functioning (eg, intelligence quotient) and sleep measures in children with various epilepsy syndromes. The reader is referred to 2 recent reviews on the topic.^32,44

Potential physiological mechanisms underlying impairment of sleep-related memory consolidation in epilepsy

The presence of ictal and interictal epileptic activity during sleep has been found in some studies to negatively impact memory consolidation processes in epilepsy patients.^{12,14,17,21,22,33} The precise mechanisms are still under study, but recent evidence highlights a potential disrupting effect of IEDs on sleep microstructure, and provides a pathophysiological perspective supporting chronic memory impairment in epilepsy.

According to the active system consolidation theory,⁴⁵ the freshly encoded memories are reactivated during a subsequent sleep period. This sleep-related consolidation process is believed to be mediated by a precise temporal coupling between sleep oscillations during NREM sleep,⁴⁶ namely (1) cortical slow waves (SWs, high amplitude oscillations with a frequency < 4 Hz), (2) thalamocortical spindles (waxing-waning waves with a frequency of 10–16 Hz), and (3) hippocampal sharp-wave ripples (high-frequency transients at a frequency of ∼80–250 Hz). Based on this theory, the temporal occurrence of these sleep oscillations follows a specific orchestration of top-down and bottom-up influences in a loop-like fashion.^46,47 The up-state (depolarizing or “active” phase) of the SW drives the generation of thalamocortical spindles which, via their excitable troughs, will entrain hippocampal networks to generate ripples and thereby reactivate memory representations at targeted neocortical sites. This SW-spindle-ripple dialogue is believed to facilitate the transfer of newly acquired and labile memory traces from the hippocampus to the neocortex for more permanent long-term storage.⁴⁵ However, only a few animal studies have directly demonstrated the role of this triple-phase locking of sleep oscillations for memory consolidation processes.^48,49 Whether the same mechanism supports sleep-related memory consolidation in humans is undetermined. One major reason relates to the fact that direct access to the hippocampus is fundamental for studying sleep-related memory consolidation, as sharp-wave ripples are generated within this key structure, yet noninvasive EEG monitoring as currently performed does not allow this access.

Using intracranial recordings from patients with drug-resistant epilepsy, a few studies assessed the impact of epilepsy or epileptic activity on the occurrence of spindles and SWs. Findings in drug-resistant temporal lobe epilepsy patients suggest that hippocampal spindles are less frequent in the affected hemisphere,^50,51 and that higher hippocampal spiking activity is associated with a reduced spindle rate.⁵¹ Regarding SWs, our group found that IEDs were more frequent at the time of high amplitude SWs, with a particularly high occurrence during the transition of the up- to the down-state of the SW,⁵² thereby differing from the physiological sleep oscillations (spindles and sharp-wave ripples) that increase during the up-states.⁵³ Subsequent studies found similar results, with IEDs occurring early during the down-state of the SWs.^54,55

The occurrence of IEDs seems to not only impact sleep oscillations individually but also how they interact. In drug-resistant epilepsy patients implanted with foramen ovale electrodes, the presence of mesiotemporal damage alters the temporal coupling between parahippocampal sharp-wave ripples and scalp SWs, with an inconsistent or even absent temporal coupling.⁵³ The authors suggested that a lack of synchrony between SWs and sharp-wave ripples in cases of mesiotemporal lesions may contribute to impaired sleep-related memory consolidation in epilepsy patients. This hypothesis is tempting although ground evidence is still lacking in humans. Relatedly, recent animal work has investigated the coupling dynamics between hippocampal IEDs, sharp-wave ripples, and prefrontal spindles and their association with memory using a spatial navigation task.⁵⁶ They found a pathological coupling between hippocampal IEDs and spindles, which was negatively associated with spatial memory performance.⁵⁶ These findings suggest that the pathologic IED-spindle coupling seems to surpass the “normal” physiological ripple-spindle interaction necessary for optimal memory consolidation. In subsequent work,⁵⁷ the same group replicated the pathological IED-spindle coupling in patients with drug-resistant epilepsy undergoing intracranial EEG monitoring. This pathologically coordinated brain activity was not restrained locally and extended to medium-to-long range connections outside the ictal network, which could contribute to cognitive impairments in patients as suggested by the authors.⁵⁷ Future work will be needed to understand how this pathologic coupling between IEDs and sleep oscillations (SWs, spindles, and sharp-wave ripples) may disrupt memory consolidation processes during sleep in epilepsy patients.

DISCUSSION

As highlighted by this review, whether sleep-related memory consolidation is impaired in epilepsy remains unclear. A total of 261 studies were identified, 27 of which met the selection criteria. Only 13 studies prospectively assessed the effect of sleep on memory consolidation in epilepsy, with inconsistent methodology and findings: Most studies reported no alteration of sleep-related memory consolidation in patients, with either similar retention levels following a period containing sleep (n = 5) or improved memory performance postsleep (n = 4). Two studies in children with epilepsy found impaired sleep-related memory consolidation in these individuals compared to the control group. Finally, based on both prospective and retrospective studies (n = 19), findings support a potential debilitating effect of nocturnal epileptic activity on memory functioning: Around half of the studies (10/19) reported statistically significant relationships. This effect was more consistently observed in children with childhood epilepsies, who are known to have more frequent and widespread IEDs relative to adults with focal epilepsy. It is therefore likely that the epileptic load influences the extent of memory impairment by sleep-related IEDs. These findings stimulate the age-old question of whether it is clinically beneficial to systematically assess, quantify, and even treat IEDs. Future interventional studies targeting sleep-related IEDs and memory functioning will hopefully aid treatment decisions in clinical practice and thereby offer some relief for patients with epilepsy dealing with sleep and memory problems.

This critical literature review also highlighted some methodological issues in the field that will need to be addressed in future studies. We propose the following recommendations for studying sleep-related memory consolidation and its potential underlying mechanism in epilepsy:

Confounding factors

One way to overcome the interindividual patient variability and the influence of clinical variables such as the underlying pathology, pharmacotherapy, and mood disorders—all known to interfere with memory consolidation—is to use an intrapatient design. Moreover, whenever possible, the patient population should be homogeneous with regard to epilepsy subtype, or the sample size sufficiently large to assess IED lateralization effects (left vs right hemisphere) on memory.

Epileptic activity monitoring

Continuous EEG monitoring with high spatio-temporal resolution for the presence of epileptic activity during the entire experimental protocol is warranted to assess any potential correlation with memory function. EEG should be combined with PSG for sleep scoring, and accurate IED detection methods should be used. On the one hand, visual IED detection by expert epileptologists provides a high specificity, but there is a high intra- and interrater variability,⁵⁸ and it can be a laborious task for prolonged EEG recordings. On the other hand, automatic IED detection avoids these pitfalls and may be more relevant for long recordings such as those including overnight sleep, although a final plausibility check is nevertheless necessary. Finally, presurgical patients admitted for invasive EEG monitoring with electrodes exploring the hippocampus constitute an ideal target population to study the effect of hippocampal IEDs during sleep on memory function. This is also a unique opportunity to investigate in humans the potential role of IEDs on sleep oscillations’ characteristics (rates of spindles, SWs, and sharp-wave ripples), their temporal relationships (eg, SW-spindle coupling, SW-ripple coupling), and how it might affect sleep-related memory consolidation processes. Certainly, they should be kept in mind that the use of intracranial EEG monitoring has its own inherent limitations, namely, it involves the investigation of only specific brain areas, it is performed in an in-patient setting, and pain/discomfort as well as pain management can also impact the quality of sleep. Moreover, patients requiring intracranial EEG are patients with drug-resistant focal epilepsy who usually have more complex epilepsy syndromes and are typically on more than 2 antiepileptic medications, which may hinder conclusions.

Interictal epileptic activity load

It is not currently known how high the epileptic activity load needs to be to interfere with sleep-related memory consolidation. An intrapatient approach with multiple testing conditions (low vs high epileptic activity load) could help to elucidate this question.

Study design

Standard research protocols for testing sleep-related memory consolidation comprise wake and sleep conditions, where learning takes place either in the morning or in the evening. It should be kept in mind that this protocol is not free of bias: For example, it does not account for circadian influences, which are known to influence cognitive functioning.⁵⁹ Study designs using daytime napping will help to circumvent this circadian effect as learning and recall are performed at the same time of day.

Use of appropriate and sensitive memory tests

For research purposes, an important consideration in selecting tests relates to the appropriateness of the instrument for the population under investigation, ie, how effective will the test be at demonstrating the hypothesis or phenomenon in question. In the context of epilepsy, one needs to consider whether the assessment technique effectively targets episodic verbal or nonverbal memory with adequate sensitivity to detect deficits in patients, a population known to have heterogenous performance levels. While the presence of ceiling effects on a task can mask any sleep benefits—leaving only little room for change (enhancement) in memory performance on subsequent recalls—one should avoid the opposite, namely, the floor effect, which is observed when a task is too difficult. This can be managed using learning-to-criterion (eg, 60% correct responses during immediate recall), which ensures a minimal amount of learning at baseline across all participants. Most instruments developed for research have not been validated on clinical populations and, unfortunately, they generally have low ecological validity and personal relevance, while it is known that sleep preferentially supports memories that are relevant for future behavior.¹⁰ Moreover, normative data from control participants are usually not available. The inclusion of an age- and sex-matched healthy control group is therefore important.

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. This project was supported by start-up funding of the Montreal Neurological Institute-Hospital to B.F. V.L. is supported by a Banting Postdoctoral Fellowship from the Canadian Institutes of Health Research. K.S. is supported by a Postdoctoral Fellowship from the Savoy Foundation. B.F. is supported by salary awards (“Chercheur‐boursier clinicien Junior 2 and Senior”) of the Fonds de Recherche du Québec–Santé 2018-2025. The authors report no conflicts of interest.

ABBREVIATIONS

AED

antiepileptic drugs

EEG

electroencephalogram

IED

interictal epileptiform discharges

NREM

non–rapid eye movement

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analysis

PSG

polysomnography

REM

rapid eye movement

SW

slow waves

2D

2-dimensional

TLE

temporal lobe epilepsy