Can temporal lobe epilepsy surgery ameliorate accelerated long-term forgetting?

Neuropsychologia 53 (2014) 64–74

Contents lists available at ScienceDirect

Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia

Can temporal lobe epilepsy surgery ameliorate accelerated long-term forgetting? Stephen J. Evans a, Gemma Elliott a, Hazel Reynders b, Claire L. Isaac a,b,n a b

Clinical Psychology Unit, Department of Psychology, The University of Sheffield, Sheffield S10 2TP, UK Clinical Neuropsychology Services, Royal Hallamshire Hospital, 12 Claremont Crescent, Sheffield S10 2JF, UK

art ic l e i nf o

a b s t r a c t

Article history: Received 24 March 2013 Received in revised form 29 October 2013 Accepted 12 November 2013 Available online 20 November 2013

Accelerated long-term forgetting (ALF) is a relatively newly identified phenomenon in neuropsychology which has been associated with temporal lobe epilepsy (TLE). ALF is characterised by intact acquisition and retention of memories over delays of minutes and hours, but abnormally fast forgetting over delays of 24 h or more. The causes of ALF are unknown; however disruption of “slow” consolidation processes through seizure activity in the temporal lobes is proposed as a possible explanation. We looked to establish whether seizure control following epilepsy surgery ameliorated ALF in patients with TLE. Parallel sets of verbal and visual stimuli were administered comparing seven TLE patients and 25 healthy controls, matched on key demographic characteristics. Free recall and recognition were assessed at both pre-surgery/time 1 and post-surgery/time 2 at delays of 25 or 45 s, 30 min and one week. The TLE group retained significantly less verbally and visually learned material between 30 min and one week at the pre-surgery assessment than the control group. Comparison of the groups at post-surgery assessment indicated evidence of improved retention in the TLE group for both visual and verbal material, despite reduced initial registration on the verbal sub-tests. Exploratory analysis of individuals indicated heterogeneity in the patient group with regards to the presence/absence of ALF and post-surgical improvement in ALF. The findings offer some support to the theory that ALF is associated with uncontrolled seizures and that elimination of seizures via epilepsy surgery may improve retention by providing a stable environment for “slow” consolidation to occur. However, our results suggest that this is unlikely to be the sole cause and that “slow” consolidation may normally depend also on the integrity of structures within the neocortex or medial temporal lobes. Further investigation of these apparent heterogeneous groups may be informative in further defining the nature and causes of ALF. & 2013 Elsevier Ltd. All rights reserved.

Keywords: Accelerated long-term forgetting Temporal lobe epilepsy Seizures Epilepsy surgery Memory

1. Introduction Recent research has found that some individuals with Temporal Lobe Epilepsy (TLE) show a distinct pattern of forgetting, whereby information is acquired normally and retained over delays of minutes or hours, but is forgotten abnormally fast over delays of weeks or months (Bell & Giovagnoli, 2007; Butler & Zeman, 2008). This relatively newly described phenomenon is termed accelerated long-term forgetting (ALF) by most researchers, although it has also been called long term amnesia by some (Kapur et al., 1996). The former term will be used throughout this paper for consistency. ALF is assumed to be caused by a disruption to the process of consolidation. Squire and Alvarez (2005) draw a distinction between the roles of “fast” and “slow” consolidation in the n Corresponding author at: Clinical Neuropsychology Services, Royal Hallamshire Hospital, 12 Claremont Crescent, Sheffield S10 2JF, UK. Tel.: þ44 114 271 3770; fax: þ44 114 226 8944. E-mail address: c.l.isaac@sheffield.ac.uk (C.L. Isaac).

0028-3932/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.neuropsychologia.2013.11.007

learning of new material. The former is thought to be reliant on medial temporal structures, including the hippocampus, to maintain representations in neocortex and accounts for memory retention over shorter intervals. The latter is thought to involve the gradual withdrawal of dependence on medial temporal lobe structures as representations become established within the neocortex over longer periods of time. One explanation for ALF is that during this period memories that have been subjected to the “fast” consolidation process are vulnerable to disruption as it is thought that stability in the neocortical environment is required for successful “slow” consolidation. Mayes et al. (2003) postulate three explanations for ALF in TLE in the context of the consolidation model; the first two posit that pathology causes damage to either medial temporal lobe structures or the neocortex, preventing the slow consolidation process due to damage in either of these systems. Previous research has examined the role of structural pathology in ALF (see Wilkinson et al., 2012; Muhlert et al., 2011) with both papers highlighting the possibility that ALF is related to temporal damage outside medial

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

temporal lobe (MTL) structures and that MTL damage may be a marker for this. The third explanation is that seizure activity disrupts the neocortical environment, thus preventing memories from becoming independently established within neocortex (Kapur et al., 1997; Squire & Alvarez, 2005). Few studies have systematically studied the role of seizures in ALF. Mameniskiene, Jatuzis, Kaubrys and Budrys (1996) assessed verbal and non-verbal memory in patients with TLE and control participants at immediate, 30 min and 4 week delays. Results indicated that patients performed significantly worse than controls at all delays. However, comparison of subgroups of patients who either experienced or did not experience seizures during the 4 week delay indicated that the seizure group demonstrated significantly less percent retention on a test of story recall (results on list recall and complex figure recall failed to achieve significance). They noted that complex partial and secondary generalised seizures appeared to be more detrimental to memory than simple partial seizures and that a group of patients with more frequent seizures 4/¼4 showed poorer memory than those with less frequent seizures (o4). Regression analysis indicated that number of complex partial seizures was a predictor for memory at the 4 week delay. These effects, however, were not specific to ALF as participants with TLE demonstrated more general deficits in memory. An earlier study by Jokeit, Daamen, Zang, Janszky, and Ebner (2001) examined ten patients with TLE using a word-position associative learning test whilst being monitored using EEG. Participants were presented with 12 words randomly positioned on a computer screen in four possible positions (left, right, down or up) and were asked to remember the positions. The process was repeated three times with each presentation followed by cued recall to establish initial learning. The participants were then retested at 30 min and 24 h delays, with ALF calculated by subtracting the 24 h delay scores from the 30 min scores. The authors found no association between seizures and retention performance in patients with right-sided TLE; however patients with left-sided TLE showed ALF on the memory task if a seizure was experienced during the 24 h delay. Two recent studies have considered the relationship between seizures and ALF, Muhlert et al. (2011) reported no correlation between seizures and ALF in a sample of seven TLE participants on verbal and visual memory tasks tested at three delays: immediate, 30 min and 1 week. However, the number of seizures experienced by the group over the week was very small; therefore the reliability of the correlates in this study can be called into question. Wilkinson et al. (2012) provided more substantive evidence of the role of seizures in ALF, with 27 TLE patients completing tests of verbal and visual recall at three delays: immediate, one hour and six weeks. Participants' seizure frequency over the six week delay was found to be positively associated with ALF. One of the drawbacks of the studies reported so far is that none adopted an AB experimental design to compare forgetting rates during a period of seizures (A) and then again when seizures are controlled (B). Few studies examining ALF have used this method, however the ones that have use two approaches: the first is to test for ALF pre- and again post-AED intervention, with the goal of comparing the seizure-free (post-AED) period with the preintervention period. This has achieved some success, with Tramoni et al. (2011) and Jansari, Davis, McGibbon, Firminger and Kapur (2010) reporting improvements post-seizure remission. The second is to study ALF pre- and post-epilepsy surgery. To our knowledge there is only one published pre- and post-surgery study in the ALF literature, a single case study by Gallassi et al. (2011), who looked at ALF pre- and post-left temporal polectomy in a patient MT. MT was a 58 year old man who experienced daily seizures and left-frontotemporal pulsating headaches before receiving surgery. He had experienced subjective memory deficits just under a year

65

prior to the surgery, which he reported had been worsening over time. Neuropsychological examination revealed ALF at the oneweek delay pre-surgery, which was calculated by averaging forgetting scores (30 min minus 1 week) across three tests of verbal and visual memory. The authors retested MT 15 months after surgery and found that ALF had improved on measures of verbal memory but not visual memory. Despite these positive findings, the study was methodologically limited due to the authors using a control group which was not IQ-matched and only using tests of recall and not recognition. They also retested the participant with identical materials pre- and post-surgery which may have confounded the results due to repeated exposure to the testing materials. The present study is the first pre and post-surgery group study in ALF, looking to replicate the findings of Gallassi et al. (2011). The aim was to explore whether seizure reduction established through epilepsy surgery had an ameliorative effect on ALF in the TLE group. We looked to investigate one of the possible causes of ALF outlined by Mayes et al. (2003), namely that seizure activity disrupts the stable environment required for the process of “slow” consolidation, therefore preventing the retention of newly acquired memories over long-delays. Our second aim was to improve the previous methodology by using specially constructed parallel sets of stimuli with difficult-to-rehearse material.

2. Material and methods 2.1. Design overview This study comprised a longitudinal quasi-experimental design, using a parallel battery of measures to assess ALF pre- and up to one year post-epilepsy surgery. The ALF materials comprised two parallel sets of visual and verbal testing materials. One set presented and tested pre-surgery and the other presented and tested postsurgery. Each set comprised recall and recognition paradigms which were used to test retention of visual scenes and verbal stories. The presentation of the sets of stimuli was counterbalanced between the participants and controls such that half the participants received set A first, and half the participants set B. 2.2. Setting This study was conducted in clinics at the Royal Hallamshire Hospital (RHH) in conjunction with the Clinical Psychology Unit at the University of Sheffield. The project was ethically approved by the South Yorkshire Research Ethics Committee. Participants gave informed consent before participating in the project. 2.3. Participants 2.3.1. TLE patients Patients who fulfilled the inclusion criteria were identified by a consultant clinical neuropsychologist. The inclusion criteria required that individuals: (a) had a formal diagnosis of TLE, (b) were due to undergo epilepsy surgery, (c) spoke English as their first language, (d) were aged between 18 and 75, (e) were assessed as having a Full Scale IQ above 80 on the Wechsler Adult Intelligence Scale-third edition (WAIS-III; Wechsler, 1997a, 1997b), and (d) were not diagnosed with comorbid neurological conditions or severe psychiatric illness. A total of seven patients (3 males, 4 females) were recruited, (see Table 1 for demographic data). TLE participants permitted us to access additional information about seizure activity through their medical records. This information included age of onset of epilepsy, seizure frequency, MRI data and current medication use. All seven TLE patients had epilepsy surgery during the study comprising a left (n ¼3) or right (n ¼4) amygdalo-hippocampectomy, depending on the lateralisation of the patients epileptic activity (established through EEG and MRI data). 2.3.2. Control group Control participants were recruited via email or poster, either from Sheffield Teaching Hospitals or the University of Sheffield email systems. The 25 participants most similar to the patients with TLE were selected from a pool of 60. Background measures collected from the controls at the initial testing appointment included reading derived IQ scores (Wechsler Test of Adult Reading (WTAR; Wechsler, 2001), handedness, mood measures and medical screening questions pertaining to the inclusion criteria.

66

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

Table 1 Characteristics of TLE participants TLE-1

TLE-2

TLE-3

TLE-4

TLE-5

TLE-6

TLE-7

Sex Age (years) Education (years) FSIQ Age of onset Duration (years) Seizure frequency Seizure types MRI

M 41 11 110 2 39 4–6/month CPS, GTCS, Aura Left MTS

F 41 13 82 9 months 40 10–15/month SPS, CPS, GTCS Left MTS

M 54 10 87 7 months 53 7/month Nocturnal Right HC volume loss

F 42 16 108 24 18 Every 10–28 days SPS, CPS Right HCS

M 20 11 85 19 2 1/month CPS Right MTS

F 42 13 113 36 6 4–6/month CPS Right MTS

Seizure onset-EEG No. AEDs

Left 2

Left 3

F 57 10 99 47 10 6–7/day Aura, GTCS Left amygdala abnormality & CD Left 1

Right 3

Right 2

Right 1

Right 2

MTS¼ mesial temporal sclerosis; HC ¼ hippocampus; HSC ¼hippocampal sclerosis; CD¼ cortical dysplasia; SPS ¼simple partial seizure; CPS ¼complex partial seizure; GTCS ¼ generalised tonic clonic seizure; AEDs ¼ antiepileptic drugs; and NC¼nocturnal.

Table 2 TLE and control group demographics: means, standard deviations and t-test.

N Gender (M/F) Age IQ Anxiety (HADS) Depression (HADS)

Control

TLE

25 12M, 13F 38.10 (14.57) 99.40 (4.73) 6.44 (4.92) 3.12 (2.88)

7 3M, 4F 39.71 (15.77) 94 (8.2) 10.00 (4.72) 5.14 (2.11)

Table 2 indicates that controls and participants with TLE were matched on key variables known to affect memory including age, IQ and mood. 2.4. Materials 2.4.1. Visual scenes test overview The visual scenes test comprised 618 colour photographs, 309 of which were randomly allocated to set A and 309 to set B. Each set contained nine visual scenes to be used for visual recall and 150 targets and 150 foils to be used for visual recognition. At presentation each scene was displayed individually to participants using Microsoft PowerPoint presentation software on a laptop computer. To ensure a reasonable level of recall each recall scene was presented twice whereas each recognition scene was presented only once. The nine recall scenes were evenly spaced throughout the first half of the presentation and were repeated in the second half in the same order. Thus, throughout the presentation every ninth scene presented was a recall scene. The other scenes were recognition targets. The presentation finished with the final six remaining recognition targets. To ensure memory for the visual scenes was adequately matched between patients with TLE and the controls, a multiple presentation procedure was used. This involved repeating administration of the visual scenes test once, immediately after the first presentation for the TLE group. As the visual task used 309 photographs/set, the material was considered difficult to rehearse. 2.4.2. Visual scenes recall Visual recall was assessed using nine scenes featuring prominent, easily identifiable environments (e.g., “The Bathroom Scene”). Each scene included six foreground items, some of which were not natural to the picture to prevent participants guessing items based on the name of the scene. Recall of three of the nine scenes was tested at each delay (Immediate, 30 min and one week). At presentation each recall scene was preceded with the name of the scene appearing in large black text on the centre of the computer screen. This was followed by the corresponding scene (see Fig. 1). As the scene appeared, the top left quadrant was highlighted in yellow for one second, followed by the top right, then bottom left, then bottom right. After all four quadrants had been highlighted the picture of the scene remained on the screen for three additional seconds to allow participants to view the scene without any highlighted quadrants. At test, recall performance was assessed using three measures: item recall tested how many individual items the participant was able to recall from the scene, with a maximum score of six per scene (eighteen per delay); spatial recall required the participants to identify the correct location of each remembered item using a recall grid which was split into four numbered quadrants. This also had a maximum score of six per scene (eighteen per delay); descriptive recall required the participants to describe what any recalled items looked like or what they/it was

doing (if applicable). The maximum score for descriptive recall was twelve points per scene, two per correct item (maximum 36 points per delay). 2.4.3. Visual scene recognition For presentation of the recognition target scenes each visual scene appeared for one second on the computer screen followed by a blank screen for one second. For the recognition test, three sets of stimuli each containing 50 targets and 50 foils were constructed. One set was presented at each delay (immediate, 30 min, one-week). 2.4.4. Stories test overview The story test comprised three stories matched for difficulty and length, each containing twenty information units (Isaac & Mayes, 1999). The stories were recorded by one of the authors (GE) onto a Windows Media Audio File (WMF) and were played to the participants through the laptop computer. The order of presentation of the stories was counterbalanced to minimise the possibility of order effects. A multiple presentation procedure was used whereby patients received two presentations of each story to ensure as far as possible that initial learning was matched. Recall of all three stories was assessed at immediate delay. At subsequent delays two stories were recalled. Thus one story was recalled at all three delays, one story was recalled at immediate and 30 min delays and one story was recalled at immediate and one week delays. This procedure allowed us to minimise rehearsal of the two stories tested twice as participants were instructed that the target story was the one recalled at all three delays. We hypothesised that any rehearsal by the participants would be focussed on this story. This also allowed us to examine the effects of repeated rehearsal on ALF. 2.4.5. Story recall Participants were asked to recall as much about the stories as possible by being asked to recall the “first”, “second” or “third” story that they had heard. They were provided with a prompt if they could not remember any information from a particular story. A point was awarded if they remembered the actual words for each unit or paraphrased the exact meaning of the words. 2.4.6. Story recognition The story recognition task comprised a twelve-question forced choice assessment procedure, with four possible answers to each of the twelve questions (e.g. what was the name of the boy, Wesley Manningham, Wesley Massingham, Warren Massingham, Warren Manningham). The twelve questions were asked after each recall trial at each delay and the answer was positioned randomly for each question to prevent the participant from correctly guessing the same position each time. The questions were presented in chronological order so that earlier answers did not cue later responses. 2.4.7. Additional measures 1. Seizure diary: Participants in the pre-surgery TLE group were asked to complete a seizure diary over the one-week delay to explore the relationship between frequency of seizures during the delay and ALF. Participants recorded the type of seizure they experienced and the date and time it occurred. 2. Perceived memory questionnaire: To explore the relationship between perceived memory and ALF, participants were asked to complete a memory questionnaire. Both controls and TLE groups were asked to fill in whether they felt they had a memory problem (yes/no) and, if so, whether this occurred over the first few minutes, first few hours, first few days or over a number of weeks. Patients with TLE were asked to report when their TLE was diagnosed.

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

67

Fig. 1. Note, full recognition phase comprises nine photographs and nine blank screens, figure provides illustrative sequence with five blanks and five photographs. Each full presentation comprises nine sequences of recognition phases, each followed by a recall scene (different pictures and scenes are used for each phase). 3. Assessment of mood: Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983). Participants were asked to complete the HADS at pre- and postsurgery assessments. 2.5. Procedure Testing took place over four sessions, the first session lasting around 1.5 h and the one week follow up session lasting around 20 min. The process was then repeated approximately six months after the initial testing sessions for the controls or at least six months post-surgery for the TLE group. This was to allow a sufficient post-surgical recovery for the TLE group. The first session comprised the initial presentation of the stimuli followed by testing the participants' recall and recognition at immediate and thirty minute delays. The HADS, WTAR and perceived memory questionnaire were completed and participants in the TLE group were instructed in how to fill in the seizure diary over the following week. At both presurgery/time 1 and post-surgery/time 2 assessments participants were forewarned that their memories for the material would be tested the following week. This instruction was included to minimise any differences in the tendency to rehearse material across the two assessment periods during the long delay. 2.5.1. Visual scenes test The Administration of the visual scenes task was preceded by a practice trial, comprising one recall and eight recognition scenes, which followed the same testing procedure as the subsequent experimental trial. Participants were shown the following instructions before the practice presentation and again before the full initial presentation: “You are about to see lots of pictures; your recognition for which will be tested. Each picture will appear for one second. During this time, you should name an object in the picture. So, if the picture has a car in it, just say “car.” Some pictures appear five times in a row. One section will be outlined at a time – please name something in each of the outlined parts. These scenes have names. Read these aloud and remember them. You will later be asked to recall the parts of these pictures in detail.”

Following the initial presentation the participants completed a 45 s distraction task comprising a number appearing in large font on the computer screen which they were required to identify as either odd or even. The tests were then administered, the order of which was dependent on the counterbalancing procedure assigned to each participant number. Recall data were collected by asking participants to remember as much as they could about one of the nine visual scenes, for example: “Can you tell me what was in the Car Boot Scene” (item recall question)

Participants were then asked to indicate where each previously recalled item was on the spatial recall grid (spatial recall). Following this they were asked to describe what any previously recalled items looked like (descriptive recall). For the

visual scenes recognition test participants were shown another PowerPoint presentation which was preceded by the following instruction on screen: “You will now see a series of pictures. You need to decide whether or not you have seen the picture before. Answer yes or no.”

Answers were recorded by the researcher as hits or false positives..

2.5.2. Stories test Participants were given the following instructions: “I am going to play you three stories, one at a time. I want you to listen to each story and remember what happens in it. Try to remember the main points. After each story ends you will be asked to tell me as much as you can remember about the story. Pay special attention to the first story as I will be most interested in your memory for this one and will ask you about it again later today and next week.”

Following the initial presentation participants were asked to think about the story for twenty seconds (or after the second presentation for the TLE group). Free recall assessed the participants' ability to recount as much of each stories as possible without additional cueing or prompting. Following each immediate recall task, the recognition task was administered, which required participants to provide a verbal answer to twelve multiple choice questions read aloud by the researcher about the story they has just heard. Participants were asked to wait until all of the potential answers were given before answering to ensure similitude between the presentation procedures.

2.6. Statistical analyses Data were analysed using IBM SPSS 20.0 for Windows. Descriptive statistics were produced and data were checked for normality, skewedness, floor and ceiling effects and outliers. Data were transformed using logarithm transforms in SPSS if data were not normally distributed. An alpha of 0.05 was adopted throughout unless specified otherwise stated

2.6.1. Corrected measures Two correction measures were used for calculating scores for the visual scenes test. scores For spatial discrimination the correction described by Hunkin, Parkin, and Longmore (1994) was used to account for the number items recalled. For descriptive free recall a correction described by Muhlert et al. (2011) was used (s) which takes into consideration the number of items recalled. Visual scene recognition scores were analysed using signal detection theory (Macmillan & Creelman, 1991) which uses the number of hits (correctly identified items) and false positives (falsely identified items) to calculate an index of accuracy based on (d′) scores.

68

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

2.6.2. Group analysis Independent samples t-tests were used to compare immediate memory performance between the control and TLE groups to investigate whether initial learning was equated. This was performed at both pre- and post-surgery assessments. Split-plot analyses of variance were then used to look for significant group differences in forgetting rates pre-surgery/time 1 and post-surgery/time 2 between the TLE and control group for each of the individual experimental measures. These were followed by appropriate contrast analyses to investigate forgetting between short and long delays. To directly compare forgetting rates pre-surgery/time 1 and post-surgery/time 2 it was necessary to use 2-way ANOVA of retention scores due to groups not always being matched for initial learning at the shortest delay. Pearson's bivariate correlations were used to measure the strength of the association between clinical measures deemed pertinent to forgetting. Due to multiple correlations being made a more stringent alpha level of.01 was applied.

2.6.3. Individual analyses Individual analysis was carried out by calculating the percentage of TLE and control participants who evidenced impaired retention over the 30 min and oneweek delay (calculation from Bell, Fine, Dow, Seidenberg, & Hermann, 2005). Participants were considered impaired if they evidenced forgetting greater than 1.96 standard deviations from the mean of the control group (equivalent to 0.05 alpha level). To explore individual differences in more detail we analysed the forgetting of individual patients pre- and post-surgery by comparing percent forgotten with that of controls using the method described by Crawford and Howell (1998) for comparing individual's scores with those of a small n control group.

3. Results 3.1. Data checking Tests of normality revealed non-normally distributed data for all of the ANOVA analyses as tested by the Shapiro–Wilk test. Data violating this testing assumption were transformed using the SPSS data transform function. The specific transform function applied was dependent on the skew of the data. Where necessary the Greenhouse Geisser correction was applied to identify and correct sphericity in the data. 3.2. Group analyses 3.2.1. Performance at initial delays To ensure initial learning was adequately matched between the TLE and control participants, independent sample t-tests were used to compare performance on the eight test variables at the immediate delay (see Table 3). Pre-surgery/time 1, analyses indicated no significant differences between the groups on six of the eight variables pre-surgery, with the exception of visual recognition and repeated story recognition. For these two subtests

additional independent samples t-tests were run to examine performance at the 30 min delay and indicated no significant difference between the TLE and control group t(30) ¼ 0.85, p4 0.05 and t(30) ¼  1.79, p4 0.05. On these grounds we decided to proceed with the analysis of these results. Post-surgery/time 2, independent sample t-tests examining performance of the two groups at the short delay indicated that the scores of the two groups were matched on the visual scenes tests but that there were significant differences between the groups on story recall, story recognition, repeated story recall and repeated story recognition. Additional analysis of these results has been included for exploratory purposes, although the data are treated cautiously due to the possibility of a scaling effect. We then proceeded with the analyses of forgetting. Fig. 2 illustrates forgetting of the TLE and control groups on the individual subtests. 3.3. Visual scenes tests 3.3.1. Item free recall Split-plot ANOVAs were used to compare item free recall between the groups (TLE and Control) at the three delays (immediate, 30 min and one week) pre-surgery/time 1 and post-surgery/ time 2. Pre-surgery/time 1 the ANOVA indicated a main effect of group that approached significance F(1,30)¼4.077, p¼0.052, η2p ¼ 0.120, a significant main effect of delay F(2,60)¼43.288, po.001 η2p ¼0.591, but no significant delay-by-group interaction F (2,60)¼ 0.831, p40.05, η2p ¼0.027. This indicates that groups forgot similar amounts of information over the delays. Post-surgery/time 2 the ANOVA indicated significant main effects of group F(1,30)¼ 7.193, po0.05 η2p ¼0.193, and delay F(1.483,44.498)¼ 16.681, po0.001, η2p ¼ 0.357, but no significant delay-by-group interaction F(1.483,44.498)¼0.144, p40.05, η2p ¼0.005. 3.3.2. Spatial free recall The Split-plot ANOVA conducted on the pre-surgery/time 1 data found a significant main effect of group F(1,30)¼4.384, po0.05 η2p ¼ 0.127, a significant main effect of delay F(1.353,40.599)¼ 30.235, po0.001 η2p ¼0.502, and a significant delay-by-group interaction F(1.353,40.599)¼6.890, po0.05 η2p ¼ 0.187. Analysis of paired contrasts revealed no significant interaction between the immediate and 30-minute delay F(1, 30)¼0.912, p40.05, η2p ¼ 0.029. However, a significant interaction was found between the 30 min and one-week delays F(1, 30)¼11.265, po0.05, η2p ¼ 0.273. This indicates that forgetting was accelerated in the TLE group between 30 min and one week. When the ANOVA was

Table 3 Immediate visual and verbal recall and recognition scores at pre-surgery and post-surgery. Variable

Pre-surgery

Post-surgery

TLE M (SD)

Control M (SD)

TLE M (SD)

Control M (SD)

Visual scenes Item recall Visual recall (z′) Descriptive recall (s) Visual recognition (d′)

13.86 (4.30) 6.37 (1.04) 71.66 (18.60) 2.91* (0.96)

15.40 (2.45) 6.23 (1.68) 78.34 (10.47) 4.02* (0.53)

11.43 (6.50) 5.56 (2.23) 69.21 (18.53) 3.44 (0.96)

15.28 (2.49) 6.59 (0.77) 80.50 (10.55) 3.44 (0.92)

Verbal stories Story recall Story recognition Repeated story recall Repeated story recognition

11.57 (5.16) 9.57 (1.13) 10.71 (3.03) 8.14* (2.12)

11.28 (4.47) 9.84 (1.60) 12.36 (3.33) 9.92* (1.73)

7.86* (4.77) 8.14* (2.19) 8.14* (3.62) 7.85* (2.27)

12.04* 10.20* 13.04* 9.64*

(3.10) (2.08) (3.42) (1.97)

Note—Pre¼ pre-surgery testing period; post ¼ six months post-surgery testing period; TLE ¼ temporal lobe epilepsy group (n ¼7); CON¼control group (N ¼25); and Recog ¼ recognition. n

p¼ o 0.05.

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

69

Fig. 2. Pre- and post-surgery comparison of the eight experimental subtests.

conducted using post-surgery/time 2 data it indicated a significant main effect of delay F(1.631, 48.938)¼43.913, po0.001 η2p ¼ 0.594, but no main effect of group F(1, 30)¼2.770, p40.05, η2p ¼0.085, and no significant delay-by-group interaction F(1.631, 48.938)¼.268, p40.05 η2p ¼0.009. This indicates that the TLE group did not evidence ALF post-surgery. The 2-way ANOVA to directly compare percent retention between the 30 min and one-week delay at pre-/ time 1 and post-surgery/time 2 indicated a group by condition interaction approaching significance F(1,63)¼3.144, p¼ 0.081,

η2p ¼ 0.050. This suggests that there was a trend towards improved retention post-surgery/time 2 in the TLE group compared to the controls. 3.3.3. Descriptive free recall Split plot repeated measures ANOVAs were used to compare the groups at the three delays (immediate, 30 min and one week) pre/time 1 and post-surgery/time 2. The ANOVA comparing the

70

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

groups pre-surgery found significant main effects of group F (1,30)¼12.783, p o0.05, η2p ¼ 0.299, and delay F(1.528, 45.826) ¼ 23.448, p o0.001, η2p ¼0.439, and a significant delay-by-group interaction F(1.528, 45.826) ¼4.688, po 0.05, η2p ¼0.135. Analysis of the contrasts between the pairs of delays found no significant interaction between the immediate and 30 min delay F(1,30)¼ 1.540, p4 0.05, η2p ¼ 0.049. However an interaction approaching significance was found between the 30 min and one-week delay F (1,30)¼3.479, p ¼0.072, η2p ¼ 0.104. A similar ANOVA conducted on the post-surgery/time 2 data indicated significant main effects of delay F(1.630, 48.912) ¼16.746, po 0.001, η2p ¼0.358, and group F(1, 30) ¼7.469, p o0.05, η2p ¼0.199, but no significant delay-by-group interaction F(1.630, 48.912) ¼ 0.702, p4 0.05 η2p ¼0.023. This shows that TLE group was not evidencing ALF post-surgery/time 2. The 2way ANOVA to directly compare percent retention between the 30 min and one-week delay at pre-/time 1 and post-surgery/time 2 indicated a group-by-condition interaction approaching significance F(1,63)¼3.721, p ¼0.058, η2p ¼0.058, indicating a trend for improved forgetting in the patient group post-surgery. 3.3.4. Visual recognition Using d′ scores, two split-plot ANOVAs were run over the three delays pre/time 1 and post-surgery/time 2. The pre-surgery ANOVA revealed significant main effects of group F(1,30) ¼6.144, p o0.05 η2p ¼ 0.170, delay F(2,60) ¼60.332, po 0.001 η2p ¼0.668, and a delay-by-group interaction approaching significance F(2, 60) ¼3.066, p ¼0.054 η2p ¼ 0.093. A similar ANOVA carried out with the post-surgery/time 2 data indicated a significant main effect of delay F(1.629, 46.754) ¼23.805, po 0.001 η2p ¼ 0.442, but no significant main effect of group F(1, 30) ¼ 0.122, p 40.05, η2p ¼0.004, and no significant delay-by-group interaction F(1.629, 48.856) ¼ 0.858, p4 0.05 η2p ¼0.028. This indicates that the TLE group were not experiencing ALF post-surgery. The 2-way ANOVA to directly compare percent retention between the 30 min and one-week delay at pre-/time 1 and post-surgery/time 2 indicated a group-bycondition interaction approaching significance F(1,63)¼3.665, p ¼0.060, η2p ¼ 0.058, indicating a trend for improved forgetting in the patient group post-surgery 3.4. Stories tests 3.4.1. Story recall A split plot repeated measures ANOVA comparing story recall between immediate and one-week delay at the pre-surgery/time 1 assessment found no significant main effect of group F(1,30)¼ 1.401, p 40.05, η2p ¼0.045, a significant main effect of delay F (1,30)¼91.433, p o0.001, η2p ¼ 0.753, and a significant delay-bygroup interaction F(1,30)¼ 9.379, po .05, η2p ¼ 0.238, indicating that the TLE group forgot at a faster rate between the immediate and one week delay. A similar analysis carried out with the postsurgery/time 2 data indicated significant main effects of group F (1,30)¼6.576, p o0.05, η2p ¼0.180, and delay F(1,30) ¼62.506, p o0.001, η2p ¼0.676, but no significant delay-by-group interaction F(1,30)¼ 0.802, p 40.05, η2p ¼ 0.026. The 2-way ANOVA to directly compare percent retention between the 30 min and one-week delay at pre-/time 1 and post-surgery/time 2 indicated no groupby-condition interaction F(1,63)¼ 1.566, p ¼ 4 0.05, η2p ¼0.025. Similar analyses were carried out using data from the story assessed at the immediate and 30 min delays. A split plot repeated measures ANOVA carried out on the data from the pre-surgery/ time 1 assessment found no significant main effect of group F (1,30)¼3.912, p 40.05, η2p ¼0.48, but a significant main effect of delay F(1,30)¼20.905, p o0.001, η2p ¼ 0.993, and a significant delay-by-group interaction F(1,30) ¼11.040, p o0.05, η2p ¼0.895, indicating that the TLE group forgot at a faster rate between the

immediate and 30 min delay. A similar analysis carried out with the post-surgery/time 2 data indicated significant main effects of group F(1,30)¼ 13.848, p o0.05, η2p ¼0.9490, and delay F(1,30) ¼ 17.424, p o.001, η2p ¼.981, but no significant delay-by-group interaction F(1,30) ¼0.021, p 40.05, η2p ¼ 0.052, indicating improved retention of the story between immediate and 30 min delays post-surgery. 3.4.2. Repeated story recall Split plot ANOVAs comparing story recall over the three delays were used to explore the effect of repeated recall between the 30 min and 1 week delays. The ANOVA run with pre-surgery/time 1 data found a significant main effects of group F(1,30)¼6.488, po 0.05, η2p ¼0.178, delay F(1.757, 52.717) ¼38.692, p o0.001, η2p ¼ 0.563, and a delay-by-group interaction approaching significance F(1.757, 52.717) ¼2.753, p ¼0.079, η2p ¼0.084. Post-surgery/ time 2 the ANOVA indicated significant main effects of group F (1,30) ¼8.755, po0.05, η2p ¼0.226, and delay F(2, 60)¼ 23.612, po 0.001, η2p ¼0.440, but no significant delay-by-group interaction F(2, 60) ¼1.993, p 40.05, η2p ¼0.062. The 2-way ANOVA to directly compare percent retention between the 30 min and one-week delay at pre-/time 1 and post-surgery/time 2 indicated a significant group-by-condition interaction F(1,63) ¼5.956, po0.05, η2p ¼ 0.090 indicating improved retention in the patient group post-surgery. 3.4.3. Story recognition Split plot ANOVAS comparing TLE and control groups' story recognition between the immediate and one-week delays presurgery/time 1 found a main effect of group approaching significance F(1, 30)¼ 3.473, p ¼0.072, η2p ¼0.104, a significant main effect of delay F(1, 30) ¼50.273, p o0.001, η2p ¼0.626, and a significant delay-by-group interaction F(1, 30) ¼8.178, po 0.05, η2p ¼0.214. This indicates that the TLE participants experienced abnormal forgetting between the immediate and 1 week delays. When the ANOVA was rerun with data from the post-surgery/time 2 assessment it found a significant main effect of group F(1, 30) ¼5.180, po 0.05, η2p ¼0.147, a significant main effect of delay F(1, 30) ¼ 28.285, po .001, η2p ¼0.485 but no delay-by-group interaction F(1, 30)¼ 0.007, p 40.05, η2p ¼0.000, indicating no difference in the forgetting rates of the two groups post-surgery/time 2. The 2-way ANOVA to directly compare percent retention between the immediate and one-week delay at pre-/time 1 and post-surgery/ time 2 indicated no significant group-by-condition interaction F (1,63) ¼2.033, p 40.05, η2p ¼0.033., indicating no significant change in retention in the patient group post surgery. Similar analyses were used to analyse the data from the story tested at the immediate and 30 min delays. The ANOVA run on the pre-surgery/time 1 data indicated no significant main effects of group F(1, 30) ¼0.746, p 40.05, η2p ¼0.133, or delay F(1, 30) ¼3.709, p4 .05, η2p ¼0.462, and no delay-by-group interaction F(1, 30)¼ .295, p 4 0.05, η2p ¼0.082, indicating no group difference in forgetting rates between the immediate and 30 min delays. When the ANOVA was rerun with data from the post-surgery/time 2 assessment it found no significant main effect of group F(1, 30)¼ 0.746, p 40.05, η2p ¼0.133 a significant main effect of delay F (1, 30) ¼5.375, p o0.05, η2p ¼0.612 but no delay-by-group interaction F(1, 30) ¼0.299, p 40.05, η2p ¼ 0.176, again indicating no difference in forgetting rates between the groups over the short delay. 3.4.4. Repeated story recognition Split plot ANOVAs were used to compare repeated story recognition performance of the TLE and control group across the three delays. The ANOVA using the pre-surgery/time 1 data found

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

71

Percentage of participants impaired

100 TLE pre

90

Control pre

80

TLE post

70

Control post

60 50 40 30 20 10 0

Item recall 30 min -1 wk

Spatial recall 30 min -1 wk

Visual recognition

Descriptive recall 30 min -1 wk

story recall imm -1 wk

story recognition imm -1 wk

Fig. 3. Pre- and post-surgery comparison of percentage of participants impaired over the one-week delay.

significant main effects of group F(1, 30) ¼10.557, p o0.05, η2p ¼0.260, delay F(1.409, 42.274) ¼16.768, p o0.001, η2p ¼ 0.359, and a significant delay-by-group interaction F(1.409, 42.274) ¼ 5.480, p o0.05, η2p ¼0.154. Planned contrasts performed between the pairs of delays (immediate to 30 min and 30 min to one week) found no significant delay-by-group interaction over the short delay F(1, 30) ¼ 0.204, po 0.05, η2p ¼0.007, but a significant interaction between the 30 min and one-week delay F(1, 30) ¼7.449, p o0.05, η2p ¼0.199. This indicates that the TLE participants experienced ALF of verbal material over the long delay despite repeated testing of the material at 30 min. When the ANOVA was rerun with the post-surgery/time 2 data it found a significant main effects of group F(1, 30) ¼4.532, p o0.05, η2p ¼0.131, and delay F(11.396,41.868)¼6.497, p o0.05, η2p ¼0.178 but no delay-by-group interaction F(1.396, 41.868)¼0.084, p4 .05, η2p ¼0.003. The 2-way ANOVA to directly compare percent retention between the 30 min and one-week delay at pre-/time 1 and post-surgery/time 2 indicated no significant group-by-condition interaction F(1,63) ¼2.199, p 40.05, η2p ¼ 0.035 indicating no significant change in retention in the patient group post surgery. 3.5. The effect of seizures on forgetting Six out of the seven patients with TLE experienced seizures either one or two seizures during the delay, therefore meaningful correlation analysis was not possible. 3.6. Correlations Pearson's bivariate correlations used to investigate the relationship between retention (30 min to 1 week) on item-recall, spatialrecall, descriptive-recall, story-recall and story-recognition subtests with variables considered relevant to forgetting in TLE: full Scale IQ, perceived long-term memory, age of onset of epilepsy, duration of epilepsy, frequency of seizures, AED use, and HADS derived anxiety and depression scores. None of the variables significantly correlated with forgetting at a.01 level.

forgetting between 30 min and 1 week delays that was 1.96 or more standard deviations from the mean of the control. Fig. 3 indicates that a greater number of TLE participants evidenced impaired retention at pre-surgery/time 1 test than the control group. A second analysis used the method described by Crawford and Howell (1998) to compare each patient's percent forgetting score on each test with that of the controls both pre- and post-surgery. This method uses the t distribution rather than the z distribution and is less likely to result in a Type I error. As this analysis was exploratory we adopted a 1-tailed alpha of 0.05. Table 4 provides a summary of the forgetting rate of each patient across each of the tests in comparison with the control group. Based on visual analysis of the resulting patterns it is clear that two patients (TLE1 and TLE 4) show no evidence of ALF on any of the experimental measures either pre- or post-surgery. The other five patients show some evidence of ALF, although only one patient (TLE 3) shows clear-cut ALF with no significant forgetting over the short delays pre-surgery. Post surgery two patients (TLE2 and TLE5) show no evidence of improved ALF. Patients TLE 3, TLE6 and TLE7 show clear improvement in ALF, though there is no evidence that forgetting over the short delay has improved. 3.8. Psychosocial measures and perceived memory Perceived memory was not found to be significantly correlated with ALF over the 30 min or one-week delay on any of the subtests. The same results were found when the correlations were repeated post-surgery. A paired sample t-test was used to detect any statistically significant within group differences in the TLE group HADS score pre/time 1 and post-surgery/time 2. TLE participants scored lower on the HADS anxiety subscale postsurgery/time 2 (M ¼5.858, SD ¼5.727) compared to pre-surgery/ time 1 (M¼ 10.00, SD ¼ 4.726). This constitutes a statistically significant decrease of 4.143 points, 95% CI [ 0.176, 8.462], t (6)¼2.347, p o0.05, d ¼0.87.

3.7. Individual analysis

4. Discussion

3.7.1. Forgetting rates Individual analyses were used to investigate the possibility that our group of patients was heterogeneous for evidence of ALF. First, using the method described by Bell et al. (2005) we calculated the number of participants from TLE and control groups who showed

This study explored the relationship between seizures and ALF by investigating whether seizure elimination through resective surgery would improve ALF in patients with AED- refractory TLE. ALF assessments were conducted pre-surgery and again between six months and one year post-surgery. Care was taken in the

 þ þ  þ   

 þ     þ þ        

 þ þ  þ     þ þ  þ   

þ þ þ  þ     þ   þ þ þ þ

þ þ þ  þ         þ þ þ

    þ þ          

     þ þ þ        

            þ   

     þ          

 þ þ  þ þ þ þ þ þ þ  þ þ  þ

                þ        Post  surgery Item free recal Spatial free recall Descriptive recall Visual recognition Verbal recall Repeated verbal recall Verbal recognition Repeated verbal recognition

þ ¼ Forgetting significantly greater than that of controls (p o0.05 1-tailed).  ¼ Forgetting not significantly different from controls.

    þ þ  þ

þ þ þ  þ þ þ þ     þ þ  þ                 Pre-surgery Item free recall Spatial free recall Descriptive recall Visual recognition Verbal recall Repeated verbal recall Verbal recognition Repeated verbal recognition

Short Short Long Short

Long

Short

Long

Short

Long

Short

Long

Short

Long

TLE 7 TLE 6 TLE 5 TLE 4 TLE 3 TLE 2 TLE 1

Table 4 Table demonstrating the presence/absence of significant forgetting of individual TLE patients in comparison with control group.

       

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

Long

72

design of the study to address some methodological issues that occur in studies of ALF. First we constructed parallel sets of stimuli that had been matched for difficulty so as to reduce practice effects. In selecting the complex scene and multiple story stimuli we aimed to minimise the potential for rehearsal of the stimuli. In addition, participants were forewarned that their memories would be tested at the longer delays in both pre-surgery and post-surgery conditions to minimise differences in tendency to rehearse or adopt other strategies across time points. Patient participants were matched on key demographic characteristics with a group of healthy controls whose memories were assessed at similar time points. A repeated exposure matching procedure was successful in equating initial learning across patient and control groups in six out of the eight sub-tests pre-surgery/ time 1. Results from the remaining sub-tests (visual recognition and repeated story recognition) were interpreted cautiously. The TLE group exhibited ALF compared to the control group presurgery/time 1, showing significantly greater forgetting between the 30 min and one-week delay on tests of spatial recall, descriptive recall, story recall, story recognition and repeated story recognition. Forgetting of visual scene recognition and repeated story recall approached significance pre-surgery/time 1. Forgetting between the immediate and 30 min delays was not significantly different between the patient and control groups with the exception of story recall. The accelerated forgetting was not correlated with measures of anxiety and depression or with patients' perceptions of their memories. Findings of ALF in the TLE group pre-surgery are consistent with those of previous studies, (Martin et al., 1991; Blake, Wroe, Breen, & McCarthy 2000 and Muhlert et al., 2011). For example Muhlert et al. (2011) assessed 14 patients with TLE, testing memory over three weeks. Comparing individual analyses between the present study and those of Muhlert et al. reveals a similar percentage of patients with TLE evidencing ALF, with approximately half of the participants in each study showing impaired retention on those subtests where ALF was demonstrated. One difference between our findings and Muhlert et al. (2011) is the lack of impairment found in our item recall sub-test. This may be due to the fact that the long delay period was longer in the Muhlert et al. study. However, this does not explain why our participants evidenced similar impairments across the other subtests. At post-surgery/time 2 assessment initial learning was matched on four out of the eight sub-tests, with initial performance on all four story-related measures significantly different to that of the controls. It seems likely that this poorer initial performance was related to resection of the hippocampus, as previous research has identified memory problems following epilepsy surgery (Sherman et al., 2011). Results from the postsurgery story sub-tests were therefore interpreted cautiously, as they may have been subject to a scaling effect although it has also been argued that poorer initial learning may be more likely to lead to ALF (e.g., Hoefeijzers, Dewar, Della-Sala, Zeman, & Butler, 2013). Post-surgery/time 2, patients with TLE evidenced unimpaired retention between the 30 min and one-week delays on all eight sub-tests. Given the possibility of a scaling effect these results need to be interpreted cautiously, however, findings are supportive of the hypothesis that elimination of seizure activity would improve ALF post-surgery. This change from significant ALF presurgery, to no significant ALF post-surgery was demonstrated on tests of spatial recall, descriptive recall, story recognition and repeated story recognition. The TLE patients were borderline impaired on spatial recognition and repeated story recall presurgery/time 1, and were unimpaired post-surgery/time 2. On the story recall subtest, forgetting pre-surgery was accelerated over

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

both short and long delays and so it was not possible to identify ALF. The change cannot be attributed to a change in AED as no participants experienced a change in medication post-surgery. It is possible that despite our attempts to prevent an ordering effect this did nonetheless affect the results. If this were the case then it would be necessary to argue that the ordering effect whereby participants adopted more effective retention strategies postsurgery affected the performance of the patient participants but not the control participants. This possibility is difficult to rule out, however. The findings of improved retention in patients post-surgery were not robust enough to be detected in a direct comparison of retention scores between pre-surgery/time 1 and post-surgery/ time 2. Analysis of pre- and post-surgery retention at an individual level indicated that the TLE group was heterogeneous with regard to ALF and this could explain the weak finding at the group level. Visual analysis suggested that two of the seven patients (TLE1 and TLE 4) patients did not demonstrate ALF pre-surgery. Of the five who did show ALF only one (TLE 3) showed a pure ALF pattern with normal retention over the short delays on all subtests and accelerated forgetting on six of the eight subtests. Two patients (TLE2 and TLE5) showed no evidence of improved retention postsurgery, whereas the other three (TLE3, TLE6, TLE7) showed improved retention. Our findings support those of Gallassi et al. (2011) who found that their participant (MT) improved on tests of visual and verbal recall between 30 min and one week following a temporal polectomy. Like two of our patients (TLE6 and TLE7) MT also evidenced accelerated forgetting over the short delay both pre- and postsurgery and so he did not demonstrate a pure form of ALF. Our findings demonstrate, however, that MT may not be representative of the population of people with TLE who undergo resective surgery and that as with our patients (TLE2 and TLE5), surgery does not necessarily improve ALF. Consideration of the demographic and seizure related information about our patients indicated no systematic variation that could account for the differences in performance. Consistent with Mameniskiene, Jatuzis, Kaubrys and Budrys (1996), there was some evidence that the presence of complex partial seizures may be associated with accelerated forgetting. Our patients TLE6 and TLE 7 were the only patients who experienced solely complex partial seizures and both showed ALF with improvement postsurgery. Interestingly, retention over the short delay did not improve post-surgery. The relationship between complex partial seizures is clearly not straightforward, however, as patient TLE3 showed clearest evidence of ALF but did not experience complex partial seizures. A recent study by Wilkinson et al. (2012) and found that seizure activity between a one-hour and six-week delay was associated with ALF. The current study cannot make similar inferences about seizure frequency, although the improvement in ALF post-surgery in some of our patients and the contemporaneous remission of seizure activity is supportive of the hypothesis that seizure activity in some way contributes to ALF. Our findings lend some support to the hypothesis that seizure activity has a causal role in ALF. However, the heterogeneity in forgetting rates in our patient group pre-surgery and the finding that ALF did not improve in all patients following surgery, even when they were seizure-free, suggests that seizure activity is not the sole cause. Further evidence that overt seizures are not necessary for ALF is demonstrated by frequent reports of ALF in patients with Transient Epileptic Amnesia (TEA). It is possible that sub-clinical seizure activity may cause ALF in this patient group (e. g. Butler et al., 2013). It is also possible, however, that ALF is caused by more than one kind of functional deficit that disrupts the slow consolidation of memory representations in neocortex (Mayes

73

et al., 2003). Mayes et al. postulated three possibilities: that ALF is a milder form of anterograde amnesia related to subtle structural damage to the hippocampus, that it is dependent on the location and type of seizures or that it is related to an interaction between structural damage and seizure characteristics. Our results add further support to the hypothesis that memories that undergo rapid initial consolidation remain labile and undergo a further period of slow consolidation. It seems likely that as well as requiring a stable environment free from seizure activity, the process of slow consolidation is also dependent on structural integrity within the medial temporal lobes or neocortex. Precisely which structures are important remains unclear. Future work might look to repeat this study with a larger clinical sample with the possibility of identifying subgroups of patients. This study suggests that three groups may be identifiable; those who do not show ALF, those who show ALF pre-surgery but not post-surgery, and those whose pre-surgery ALF does not improve. It may be that careful study of small groups with pure deficits may be more informative than studies of large groups that may be heterogeneous.

Acknowledgements We thank all the participants, both TLE and controls who participated in this study. We also thank Dr. Richard Grunewald, Professor Markus Reuber and Dr. Stephen Howell from whose clinics patients were recruited. Thanks also to Miss Debra Ford and Dr. Emily Mayberry for help with recruitment. Additional thanks to Miss Eleanor Joyce for proof reading earlier drafts of this project. None of the authors has any conflict of interest to disclose

References Bell, B. D., Fine, J., Dow, C., Seidenberg, M., & Hermann, B. P. (2005). Temporal lobe epilepsy and the selective reminding test: The conventional 30-minute delay suffices. Psychological Assessment, 17(1), 103–109. Bell, B. D., & Giovagnoli, A. R. (2007). Recent innovative studies of memory in temporal lobe epilepsy. Neuropsychology Review, 17, 455–476. Blake, R. V., Wroe, S. J., Breen, E. K., & McCarthy, R. A. (2000). Accelerated forgetting in patients with epilepsy: evidence for an impairment in memory consolidation. Brain, 123(3), 472–483. Butler, C., van Erp, W., Bhaduri, A., Hammes, A., Heckemann, R., & Zeman, A. (2013). Magnetic resonance volumetry reveals focal brain atrophy in transient epileptic amnesia. Epilepsy and Behavior, 28(3), 363–369. Butler, C. R., & Zeman, A. Z. (2008). Recent insights into the impairment of memory in epilepsy: Transient epileptic amnesia, accelerated long-term forgetting and remote memory impairment. Brain, 131(9), 2243–2263. Crawford, J. R., & Howell, D. C. (1998). Comparing an individual’s test score against norms derived from small samples. The Clinical Neuropsychologist, 12(4), 482–486. Gallassi, R., Sambati, L., Poda, R., Maserati, M. S., Oppi, F., Giulioni, M., & Tinuper, P. (2011). Accelerated long-term forgetting in temporal lobe epilepsy: Evidence of improvement after temporal pole lobectomy. Epilepsy and Behaviour, 22(4), 793–795. Hoefeijzers, S., Dewar, M., Della-Sala, S., Zeman, A., & Butler, C. (2013). Accelerated long-term forgetting in transient epileptic amnesia: An acquisition or consolidation deficit? Neuropsychologia, 51(8), 1549–1555. Hunkin, N. M., Parkin, A. J., & Longmore, B. E. (1994). Aetiological variation in the amnesic syndrome: Comparisons using the list discrimination task. Neuropsychologia, 32(7), 819–825. Isaac, C. L., & Mayes, A. R. (1999). Rate of forgetting in amnesia: I. Recall and recognition of prose. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25(4), 942–962. Jansari, A. S., Davis, K., McGibbon, T., Firminger, S., & Kapur, N. (2010). When “longterm memory” no longer means “forever”: Analysis of accelerated long-term forgetting in a patient with temporal lobe epilepsy. Neuropsychologia, 48(6), 1707–1715. Jokeit, H., Daamen, M., Zang, H., Janszky, J., & Ebner, A. (2001). Seizures accelerate forgetting in left-sided temporal lobe epilepsy. Neurology, 57(1), 125–126. Kapur, N., Miller, J., Colbourn, C., Abbott, P., Kennedy, P., & Docherty, T. (1997). Very long-term amnesia in association with temporal lobe epilepsy: Evidence for multiple-stage consolidation processes. Brain and Cognition, 35(1), 58–70.

74

S.J. Evans et al. / Neuropsychologia 53 (2014) 64–74

Kapur, N., Scholey, K., Moore, E., Barker, S., Brice, J., & Thompson, S. (1996). Longterm retention deficits in two cases of disproportionate retrograde amnesia. Journal of Cognitive Neuroscience, 8(5), 416–434. Macmillan, N. A., & Creelman, C. D. (1991). Detection theory: A user's guide. New York: Cambridge University Press. Mameniskiene, R., Jatuzis, S., Kaubrys, G., & Budrys, V. (1996). The decay of memory between delayed and long-term recall in patients with temporal lobe epilepsy. Epilepsy and Behaviour, 8(1), 278–288. Martin, R. C., Loring, D. W., Meador, K. J., Lee, G. P., Thrash, N., & Arena, J. G. (1991). Impaired long-term retention despite normal verbal learning in patients with temporal lobe dysfunction. Neuropsychology, 5(1), 3–12. Mayes, A. R., Isaac, C. L., Holdstock, J. S., Cariga, P., Gummer, A., & Roberts, N. (2003). Long-term amnesia: A review and detailed illustrative case study. Cortex, 39(4– 5), 567–603. Muhlert, N., Grunewald, R. A., Hunkin, N. M., Reuber, M., Howell, S., Reynders, H., & Isaac, C. L. (2011). Accelerated long-term forgetting in temporal lobe and idiopathic generalised epilepsies. Neuropsychologia, 49(9), 2417–2426. Sherman, E. M. S., Wiebe, S., Fay-McClymont, T. B., Tellez-Zenteno, J., Metcalf, A., Hernandez-Ronquillo, L., Hader, W. J., & Jette, N. (2011). Neuropsychological

outcomes after epilepsy surgery: Systematic review and pooled estimates. Epilepsia, 52(5), 857–869. Squire, L. R., & Alvarez, P. (2005). Retrograde amnesia and memory consolidation: A neurobiological perspective [Review]. Current Opinion in Neurobiology, 5(2), 169–177. Tramoni, E., Felician, O., Barbeau, E. J., Guedj, E., Guye, M., Bartolomei, F., & Ceccaldi, M. (2011). Long-term consolidation of declarative memory: Insight from temporal lobe epilepsy. Brain, 134(3), 816–831. Wilkinson, H., Holdstock, J. S., Baker, G., Herbert, A., Clague, F., & Downes, J. J. (2012). Long-term accelerated forgetting of verbal and non-verbal information in temporal lobe epilepsy. Cortex, 48(3), 317–332. Wechsler, D. (1997a). Wechsler memory scale (3rd ed.). San Antonio, TX: The Psychological Corporation. Wechsler, D. (1997b). Wechsler adult intelligence scale (3rd ed.). San Antonio, TX: The Psychological Corporation. Wechsler, D. (2001). Wechsler test of adult reading: examiner's manual. San Antonio, Texas: The Psychological Corporation. Zigmond, A. S., & Snaith, R. P. (1983). The hospital anxiety and depression scale. Acta Psychiatrica Scandinavica, 67, 361–370.