Epilepsy & Behavior 16 (2009) 113–119
Contents lists available at ScienceDirect
Epilepsy & Behavior journal homepage: www.elsevier.com/locate/yebeh
Transient News Events Test: Feasibility in assessment of post-temporal lobectomy remote memory deficits Beth A. Leeman a,b,c,*, Eric A. Macklin c,d, Donald L. Schomer a,c, Margaret G. O’Connor a,c a
Department of Neurology, Beth Israel Deaconess Medical Center, Boston, MA, USA Department of Neurology, Massachusetts General Hospital, Boston, MA, USA c Harvard Medical School, Boston, MA, USA d Massachusetts General Hospital Biostatistics Center, Boston, MA, USA b
a r t i c l e
i n f o
Article history: Received 13 May 2009 Revised 4 June 2009 Accepted 7 June 2009 Available online 29 July 2009 Keywords: Epilepsy Seizure Memory Retrograde Remote Cognition Temporal lobe resection
a b s t r a c t Although anterograde memory deficits are well documented in patients with epilepsy, the extent to which remote memory deficits occur is less clear. This is due in part to a lack of reliable methods for assessment. The present study examined the feasibility of using the Transient News Events Test (TNET) to assess remote memory in subjects status post anterior temporal lobectomy (ATL) for the treatment of refractory seizures. Results indicated significantly poorer performance of the patient group compared to healthy controls. The decrement in performance within the patient group was evident only for items from more recent time periods. Reasons for an apparent stability of the most remote memories with ATL and implications regarding hippocampal function are reviewed. In conclusion, the TNET provides a feasible method for assessment of remote memory function in patients with epilepsy, with decrements in performance noted in comparison to a healthy control group in this retrospective study. Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction Memory is not a unitary construct; rather, it is mediated by a variety of component processes that depend on different neural systems. More specifically, memory may be characterized as nondeclarative (implicit) or declarative (explicit) [1]. Nondeclarative memory includes items that are not dependent upon awareness, as in conditioning, priming, or motor skills (i.e., how to ride a bicycle). Declarative memories, in contrast, are those for facts (i.e., the bicycle is red) or events (i.e., receiving the bicycle as a birthday gift), associated with awareness of the prior learned episodes. In addition, memory comprises anterograde and retrograde processes, reflecting the formation of new memories and the recollection of remote memories, respectively. Although retrograde memory often refers specifically to memory for events occurring prior to a given time point (i.e., an injury), the timing of hippocampal damage in epilepsy is often unclear. Hence, we use the terms retrograde memory and the more general remote memory interchangeably.
* Corresponding author. Address: Behavioral Neurology Unit, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, KS-2, Boston, MA 02215, USA. Tel.: +1 617 667 4074; fax: +1 617 667 7981. E-mail address:
[email protected] (B.A. Leeman). 1525-5050/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.yebeh.2009.06.012
Anterograde declarative memory deficits are well documented in patients with temporal lobe epilepsy (TLE), including those who are newly diagnosed as well as those with long-standing seizures [2,3]. Anterograde memory function may be further compromised by ongoing interictal epileptiform discharges [4], anticonvulsant medication use [5], or anterior temporal lobectomy (ATL) for the treatment of medically refractory epilepsy [6,7]. Less well studied are retrograde declarative memory deficits in the setting of epilepsy. Several authors have demonstrated remote memory dysfunction in patients with TLE, both pre- and postresection [8–11]. The routine neuropsychological batteries used in pre- and post-operative examinations, however, do not typically evaluate remote memory abilities. This is due in part to a lack of valid methods for assessment [12]. Studies have examined remote memory for items such as autobiographical facts and events, famous names and faces, and public news events. Use of autobiographical stimuli, however, has notable limitations. The only commercially available battery, the Autobiographical Memory Interview, has received criticism for its lack of sensitivity [9,13]. Perhaps more concerning, however, is the inability to document responses as correct or incorrect when studying autobiographical material. In addition, it may be difficult to precisely date the learning episodes or control for the frequency of rehearsal. The tests of memory for public facts and events used in the literature had limitations as well, in that the periods assessed were
114
B.A. Leeman et al. / Epilepsy & Behavior 16 (2009) 113–119
often ill-defined. Studies have examined memory for the early 1990s, for example, using the stimulus of Bill Clinton. This does not truly test memory from that period, however, as he has been continuously in the public eye since that time. The Transient News Events Test (TNET) addresses these issues by using stimuli that assess free recall and recognition of popular news events that were prominent in the media only for limited periods of time. Items were drawn from world almanacs and the New York Times Index. Each event appeared briefly in the news between the 1950s and 1990s, then declined in frequency of reporting over the subsequent 3 years. Items were then grouped by hemidecade, such that remote memory would be tested for distinct periods (i.e., 1950–1954). O’Connor et al. [14] administered the TNET to 200 healthy control subjects, who demonstrated better recall for more recent events than for more remote events (Appendix A: see Supplementary material). The present study assesses the feasibility of using this novel paradigm to examine remote memory in subjects status-post temporal lobectomy for the treatment of refractory seizures. The expectation was to find overall poorer task performance in the patients with epilepsy compared to a healthy control group when controlling for age, sex, and educational level. 2. Methods 2.1. Subjects The present study is a retrospective analysis of data collected from patients with refractory TLE who had undergone standard ATL at Beth Israel Deaconess Medical Center (BIDMC) from December 1987–September 2000. The study was conducted with approval by the BIDMC institutional review board, and all participants provided informed consent prior to testing. All subjects with epilepsy received neuropsychological evaluations, including the TNET, at a mean of 5 years post-surgery (range: 2.3–9 years). Sodium amytal testing revealed left-lateralized language functions in all participants. TNET data had also been obtained from a healthy control group, similar to the patients with respect to age, educational level, and sex distribution. To ensure that controls and subjects with epilepsy had comparable levels of interest in news events, a self-rating score (0–2, with 2 reflecting the most interest) was obtained from each subject. In addition, to screen for adequate exposure to the news or discussion regarding current events, place of residence and work history were reviewed for each participant with epilepsy. 2.2. Transient News Events Test The TNET is a verbally administered battery that consists of a series of questions regarding public events from various domains (e.g., politics, crimes, and entertainment). The items were derived from world almanacs and the New York Times Index, including events from 1950–1989 in controls and 1950–1994 in subjects with epilepsy. Events of a similar nature, the details of which might be confusing (e.g., plane crashes or terrorist attacks), were
excluded. The New York Times Index was used to determine the number of days each item appeared in the news over a 4-year period. The average initial exposure for each item was 69 days per year. Frequency of reporting decreased by 84% in the first year and by 99% in the third year after initial exposure. Items were grouped into 5-year periods, with no significant differences in fade-out rates between the hemidecades. Three items were tested per hemidecade. For each item, there were associated free recall and recognition questions. Recall questions were worded such that the most salient aspect of the event was mentioned. This wording decreased potential interference caused by momentary word-finding difficulties. Recognition testing was administered in a forced-choice format, with each correct answer paired with one distracter. The recognition questions were structured such that the first question tested general knowledge of the event, and the second question required more detailed information. Recall and recognition scores were computed separately. Full 2point credit was given if the participant identified two predetermined critical pieces of information for each recall question. One point was given if one critical fact was recalled. If full credit was received for the recall question, the recognition testing was not administered and full credit for recognition was given automatically. Two recognition questions were paired with each recall item. Each correct answer to a recognition question was worth 1 point. This resulted in a maximum possible score of 6 points per hemidecade for recall and 6 points per hemidecade for recognition. A sample test item is shown in Fig. 1. 2.3. Statistical analysis The primary aim was to assess the feasibility of using the TNET to test remote memory function in patients with epilepsy post-ATL. The hypothesis was that subjects with epilepsy would have lower recognition and recall TNET scores than healthy controls. The mean recognition and recall TNET scores per hemidecade were calculated separately for each group. To assess statistical significance of any possible differences, univariate mixed model linear regression analyses were conducted, with group (subjects with epilepsy vs controls) as the single predictor of interest and TNET recall and recognition scores as the outcome measures. These were followed by full models, which also included age at the time of neuropsychological testing, sex, and educational level as covariates. Univariate regression models were calculated for all variables. Each covariate was kept in the full model regardless of univariate significance, however, as they were clinically important factors for which to adjust. In addition, prior data suggested beneficial effects of greater age and male sex on remote memory function [14]. From the primary aim, four secondary questions followed. First, if the subjects with epilepsy showed a decrement in performance, was this true across all hemidecades? To address this issue, mixed model linear regression was conducted using a subject group by hemidecade interaction term. Predictors in the models included group, hemidecade, and their interaction, with the outcome measures of TNET recall and recognition score. In addition, full models
Fig. 1. Sample TNET item. Adapted, with permission, from O’Connor et al. [14].
115
B.A. Leeman et al. / Epilepsy & Behavior 16 (2009) 113–119
were derived, which also included age at testing, sex, and educational level as covariates. Second, a possible effect of laterality (right vs left temporal lobectomy) was addressed. To examine this issue, mixed model linear regression was performed. Right- or left-sided resections were the predictors of interest, and TNET recall and recognition scores served as the outcome measures. Third, the relationship between performance on the TNET and preoperative anterograde verbal memory was assessed, using the Rey Auditory Verbal Learning Test (RAVLT). The RAVLT involves verbal list learning, from which a ‘‘total learning score” (TLS) is derived. The TLS represents the total number of words correctly recalled, summed over five encoding trials. Mixed model linear regression was conducted, with preoperative TLS as the predictor of interest and postoperative TNET recall and recognition scores as the outcome measures. If remote memory and anterograde memory represent separate processes with distinct underlying neural circuits, a significant association would not be expected. Finally, the relationship between memory performance during Wada testing and TNET score was examined. The Wada memory test score reflects memory for objects shown after amobarbital injection. The percentage of objects recalled after left injection was subtracted from the percentage of objects recalled after right injection (Wada R – L). The percentage, rather than raw score, was calculated, as subjects may have been shown different numbers of objects (range: 6–8 per hemisphere tested). Mixed model linear regression was performed, with Wada R – L as the predictor of interest and TNET recall and recognition scores as the outcome measures.
3.2. Transient News Events Test Mean TNET scores, plotted by subject group for each hemidecade, reveal poorer performance of the patient group compared to the controls for recognition (Fig. 2) and recall (Fig. 3). The pattern of scores over time for both subject groups is similar to that seen in prior controls [14]. The figures demonstrate a shallow slope of forgetting for the most recent events, then a more precipitous decline followed by a plateau for the most remote items, particularly when assessing recall. The univariate regression analyses indicated that the observed differences between patients and controls were significant for recognition (b = 0.6140, p = 0.0295), but not for recall scores (b = 0.6530, p = 0.1149) (Table 2). Univariate regression analyses for age at testing, sex, and educational level were not statistically significant (Table 2). When these covariates were added to form the full models, the following regression equations were derived (Table 3):
TNET recognition score ¼ 1:7609 þ 0:6955 ðgroupÞ þ 0:0266 ðageÞ þ 0:1050 ðeducationÞ þ 0:0897 ðsexÞ; in which group (subjects with epilepsy vs controls) was a significant predictor (p = 0.0120), and
3. Results 3.1. Subjects Participants included 15 subjects with epilepsy post-temporal lobectomy and 19 healthy controls. Mean age at the time of neuropsychological testing was 41.8 years (range: 29–59) for the subjects with epilepsy and 41.5 years (range: 30–58) for the controls. Educational level and interest in news events were similar between the patients and controls. All subjects with epilepsy lived in the New England area according to current demographic information, and had a history of employment and/or higher education. Each of the 10 patients for whom childhood history was available had been raised in the United States, primarily in Massachusetts and surrounding states. Additional demographic information is summarized in Table 1.
Fig. 2. Mean TNET recognition scores by hemidecade for patients and controls. Bars indicate ±1 SD. Years represent the median year of each hemidecade (i.e., 1952 represents the period 1950–1954).
Table 1 Demographic information.
N Age at testing (years; mean, range) Sex Education (years; mean, range) Interest rating (mean, range)* Hemisphere Seizure frequency (number per month at time of surgery, range) Number of AEDs at time of surgery (mean, range) Age at recurrent seizure onset (years; mean, range) Seizure duration prior to surgery (years; mean, range) Engel outcome score (mean, range)** *
TLE
Controls
15 41.8 (29–59) 5M, 10F 15.3 (12–20) 0.46 (0–2) 6R, 9L 65.4 (2–345)
19 41.5 (30–58) 5M, 14F 14.6 (12–18) 0.67 (0–2)
2 (1–4) 13.6 (4.5 months–39 years) 13.0 (4–32) 2 (1–4)
Data available in 13 subjects with epilepsy, 18 controls. At time of neuropsychological testing. Data available in 13 subjects with epilepsy.
**
Fig. 3. Mean TNET recall scores by hemidecade for patients and controls. Bars indicate ±1 SD. Years represent the median year of each hemidecade (i.e., 1952 represents the period 1950–1954).
116
B.A. Leeman et al. / Epilepsy & Behavior 16 (2009) 113–119
Table 2 Univariate regression analyses. TNET recall
TNET recognition
p = 0.1149 b = 0.6530
p = 0.0295 b = 0.6140
Hemi-decade
p < 0.0001 b = 0.1179
p < 0.0001 b = 0.0845
Age
p = 0.1178 b = 0.0387
p = 0.1141 b = 0.0274
Sex
p = 0.8797 b = 0.0709
p = 0.9178 b = 0.0339
Education
p = 0.0858 b = 0.1738
p = 0.3281 b = 0.0714
Group
Univariate regression results for each covariate as a single predictor of interest, with TNET recall and recognition scores as the outcome measures. Statistically significant results (p 6 0.05) are highlighted in bold.
Table 4 Predicted differences in TNET recognition scores for the patient group compared to controls by hemidecade. Hemidecade
Effect estimate
P-value
95% C.I.
1952 1957 1962 1967 1972 1977 1982 1987
0.6207 0.1364 0.1956 1.0131 1.5324 1.3640 1.4588 0.8412
0.1599 0.7570 0.6464 0.0182 0.0004 0.0015 0.0007 0.0494
1.4881, 0.2467 0.7310, 1.0038 1.0346, 0.6434 1.8521, 0.1742 2.3714, 0.6935 2.2030, 0.5250 2.2978, 0.6198 1.6802, 0.0022
Effect estimates for the subject group by hemidecade interaction term on TNET recognition score, adjusted for age, sex and educational level. Estimates reflect the predicted difference in TNET scores for the patient group compared to controls for each hemidecade, with negative numbers indicating poorer performance in the patient group. Subjects with epilepsy demonstrated poorer performance compared to controls for more recent time periods (1965–1989) as opposed to more remote time periods (1950–1964). Years indicate the median year of each hemidecade (i.e. 1952 represents the period of 1950–1954). Statistically significant results (p 6 0.05) are highlighted in bold.
Table 3 Mulitvariate regression analyses. TNET recall
TNET recognition
Group
p = 0.0389 b = 0.8124
p = 0.0120 b = 0.6955
Age
p = 0.1219 b = 0.0363
p = 0.1069 b = 0.0266
Sex
p = 0.7640 b = 0.1404
p = 0.7851 b = 0.0897
Education
p = 0.0426 b = 0.2148
p = 0.1575 b = 0.1050
Multivariate regression results for the full model including group as the predictor of interest; age, sex and educational level as covariates; and TNET recall and recognition scores as the outcome measures. Statistically significant results (p 6 0.05) are highlighted in bold.
TNET recall score ¼ 2:3271 þ 0:8124 ðgroupÞ þ 0:0363 ðageÞ þ 0:2148 ðeducationÞ þ 0:1404 ðsexÞ;
To explore the effect of lateralization, a linear regression analysis examined the predictive value of left or right ATL. Although nonsignificant, the results suggested a trend for those who underwent left ATL to have poorer recognition (left: b = 0.5920, p = 0.0777; right: b = 0.6457, p = 0.0929) and recall (left: b = 0.9033, p = 0.0631; right: b = 0.2848, p = 0.6085) scores when compared to all other subjects. Results demonstrated no significant relationship between recognition or recall TNET score and preoperative RAVLT-TLS in the 11 subjects with available RAVLT data. In the eight subjects with documented Wada memory test scores, the Wada R – L measure significantly predicted TNET recognition (b = 1.7778, p = 0.0348) and recall (b = 2.4928, p = 0.0249). Specifically, greater right-lateralized memory function on Wada testing predicted better post-operative remote memory function, whereas left lateralization was associated with poorer remote memory function as assessed by the TNET.
in which education (p = 0.0426) and group (p = 0.0389) were significant predictors. The models indicated that a subject in the patient group would have a recognition score 0.6955 points lower and a recall score 0.8124 points lower than a healthy control of the same age, sex, and educational level. 3.3. Secondary analyses To assess the possible interaction between subject group and hemidecade, mixed model linear regression was performed, indicating significant effects of the interaction term on recognition (p = 0.0070) and recall (p = 0.0098) scores. Full models, including group, hemidecade, a group hemidecade interaction term, age, gender, and education, were also calculated. In the full model, the group hemidecade interaction was a significant predictor (p = 0.0079) of the TNET recognition outcome measure. Age, sex, and educational level were not significantly predictive of the recognition score. In the full model for the TNET recall outcome measure, significant predictors included education (b = 0.2252, p = 0.0315) and the group hemidecade interaction (p = 0.0107). Age and sex were not significant predictors of recall score. The interactions indicated poorer patient performance compared to that of controls for more recent items (1965–1989), but equivalent performance for more remote items (1950–1964), for recognition (Table 4 and Fig. 4) and recall (Table 5 and Fig. 5).
Fig. 4. Effect estimates for the subject group hemidecade interaction term on TNET recognition score. The ‘‘0” line indicates no difference in performance between the patient group and controls, and points below this line indicate how much lower a patient’s score would be compared to a control of the same age, sex, and educational level for each hemidecade. Significant differences were noted between groups as seen in Table 4. Subjects with epilepsy demonstrated poorer performance than controls for more recent periods (1965–1989) as opposed to more remote periods (1950–1964). Effect estimates are plotted along the median year of each hemidecade (i.e., 1952 represents the period 1950–1954).
B.A. Leeman et al. / Epilepsy & Behavior 16 (2009) 113–119 Table 5 Predicted differences in TNET recall scores for the patient group compared to controls by hemidecade. Hemidecade
Effect estimate
P-value
95% C.I.
1952 1957 1962 1967 1972 1977 1982 1987
0.4837 0.4189 0.1198 1.0953 1.4286 1.9900 2.0110 1.1760
0.4002 0.4661 0.8301 0.0507 0.0110 0.0004 0.0004 0.0360
1.6142, 1.5494, 1.2184, 2.1939, 2.5272, 3.0886, 3.1097, 2.2746,
0.6469 0.7116 0.9788 0.0034 0.3300 0.8914 0.9124 0.0773
Effect estimates for the subject group x hemidecade interaction term on TNET recall score, adjusted for age, sex and educational level. Estimates reflect the predicted difference in TNET scores for the patient group compared to controls for each hemidecade, with negative numbers indicating poorer performance in the patient group. Subjects with epilepsy demonstrated poorer performance compared to controls for more recent time periods (1965–1989) as opposed to more remote time periods (1950–1964). Years indicate the median year of each hemidecade (i.e. 1952 represents the period of 1950–1954). Statistically significant results (p 6 0.05) are highlighted in bold.
4. Discussion The present study examined the feasibility of using the Transient News Events Test to assess remote memory deficits in patients who had undergone standard ATL for the treatment of refractory seizures. The findings indicated that this tool can be used to assess retrograde declarative memory for discrete time periods in this population. The data revealed better memory for more recent than more remote events in both patient and control groups, but overall significantly poorer long-term memory in subjects post-temporal lobectomy. This pattern was true for recognition, a measure of memory storage, in a univariate regression analysis as well as in a model adjusted for age, sex, and educational level. Significantly poorer recall was also noted in the patient group, using a multivariate regression model adjusted for age, sex, and education.
Fig. 5. Effect estimates for the subject group hemidecade interaction term on TNET recall score. The ‘‘0” line indicates no difference in performance between the patient group and controls, and points below this line indicate how much lower a patient’s score would be compared to a control of the same age, sex, and educational level for each hemidecade. Significant differences, or differences approaching significance, were noted between groups as seen in Table 5. Subjects with epilepsy demonstrated poorer performance than controls for more recent periods (1965–1989) as opposed to more remote periods (1950–1964). Effect estimates are plotted along the median year of each hemidecade (i.e., 1952 represents the period 1950–1954).
117
The data support the hypothesis that memory deficits in the setting of epilepsy occur not just with anterograde tasks, but also with tests for remote memories dating 31–36 years prior to testing. Prior studies indicate that post-operative memory dysfunction negatively affects quality of life [7]. How remote memory dysfunction specifically impacts job or school performance and quality of life, however, is unknown and should be explored in future studies. Whether cognitive rehabilitation strategies would be helpful also merits investigation. Although our results demonstrate remote memory dysfunction in patients with TLE status-post resection, it is unclear whether this deficit was sustained with surgery, as preoperative data were unavailable. Although one study documented mild retrograde memory loss pre- to post-resection [15], others found no significant post-resection decline [16]. Furthermore, Viskontas et al. [10] noted equivalent remote memory task performance in subjects who had received epilepsy surgery and those who had not undergone resection. This suggests that, in some subjects, remote memory deficits may be incurred not just with resection, but with lesser degrees of hippocampal injury. It is unknown, however, whether the critical injury is sustained at the time of the first seizure or later in the evolution of epilepsy. Limited data suggest no effect of age at seizure onset on remote memory function [10]. Nevertheless, repeated testing over time, specifically pre- and post-resection, with assessments of the timing of seizure onset or surgery in larger numbers of subjects, is necessary to further explore these issues. Limitations of the present data include the lack of available information and lack of consistency regarding the number of years prior to surgery that each test item occurred. An event from 1980 may have happened 10 years prior to surgery for one subject, but 20 years prior to surgery for another. For a subset of six subjects who underwent surgery within the same year, the graphs of TNET scores over time look quite similar to the above data (Figs. 6 and 7). Future studies with larger cohorts may be designed to examine effects of the timing of hippocampal injury in more detail. An exploratory analysis suggested an interaction between group and hemidecade, that is, the effect of subject group depended on the time period tested. Data revealed poorer performance for the patient group than for the controls with respect to retrograde recall and recognition for more recent events, since
Fig. 6. Mean TNET recognition scores for subjects with epilepsy and controls. Bars indicate ±1 SD. The X axis reflects the number of years prior to surgery that the tested items occurred. The final time point also contained questions regarding items that occurred 0–2 years post-resection. Data are from six patients who underwent resection in 1992 and six controls matched for age, sex, and education.
118
B.A. Leeman et al. / Epilepsy & Behavior 16 (2009) 113–119
Fig. 7. Mean TNET recall scores for subjects with epilepsy and controls. Bars indicate ±1 SD. The X axis reflects the number of years prior to surgery that the tested items occurred. The final time point also contained questions regarding items that occurred 0–2 years post-resection. Data are from six patients who underwent resection in 1992 and six controls matched for age, sex, and education.
the mid-1960s. Performance was equivalent, however, for events that occurred prior to that time. This might suggest that the most remote memories are more stable, resistant to injury. Alternative explanations, however, must be considered. First, the earlier hemidecades predate the onset of recurrent seizures in 8 of the 14 subjects born in or prior to 1965. Perhaps a lack of difference in TNET scores between the patient group and controls indicates only that significant injury to memory circuits had not yet occurred, which may be more evident if data were coded based upon timing of seizure onset. Second, these results may be limited by the power to detect a difference in group performance for the most remote hemidecades. The confidence intervals for the estimates of group differences over these time periods were often wide, indicating that a potentially clinically meaningful difference may have been missed. Third, a floor effect may be present, in that controls performed so poorly that demonstrating worse performance in patients would be quite difficult. For the recognition task, a mean score of 3 (50% correct) represents chance. The patient and control scores were at or below chance for the most remote hemidecades. This may reflect a general instability of remote memory. On average, however, subjects were preadolescent in 1965. Hence, the hemidecades prior to the mid-1960s may predate an interest in news events. Furthermore, it is unclear whether the ‘‘forgetting curves” represent loss of more remote memories or a failure of initial encoding. As O’Connor et al. [17] and Kapur et al. [18] suggest, gaps in long-term memory may reflect anterograde memory dysfunction caused by continued seizures or ‘‘subclinical” discharges. The group hemidecade interaction may reflect that more remote epochs occurred prior to the evolution of epilepsy. Memories may be less well encoded during later periods as seizures developed or progressed. Limited data suggesting minimal, if any, progressive anterograde memory deficit in epilepsy, however, argue against this hypothesis [2,19]. In addition, Lah et al. [9] concluded: ‘‘A seizure disorder interfered with stored memories (i.e., for events and people that were encountered years before the onset of seizures) and did not just interfere with new memory consolidation.” Moreover, the present data indicate no significant relationship between the pre-operative RAVLT-TLS, a measure of anterograde verbal memory, and the post-operative TNET scores. This finding suggests that different mechanisms underlie antero-
grade list learning tasks and remote memory for public events. The possible relationship between new memory formation and remote memory function should be explored in prospective studies, however, with a larger number of subjects. Perhaps a long-term study examining anterograde memory over time, coupled with later tests of remote memory for those time periods, would help to address this issue. If a group hemidecade interaction exists, what would that suggest regarding hippocampal function? Two competing claims regarding hippocampal involvement in memory have been posed: the consolidation and multiple trace theories. The standard model, consolidation theory, asserts that the hippocampus is required for memory retrieval only for a limited time [20]. Although initially necessary for memory storage and retrieval, once consolidation is complete, the hippocampus is no longer needed. Hence, if one sustains hippocampal damage, only unconsolidated memories should be lost. The present data are at odds with this theory, however, in that the relative deficit in remote memory retrieval extends over 30 years. Although some authors propose that the processes involved in consolidation may continue for months or years [18], to require 30 years for consolidation is intuitively problematic. The slope of the recall curve presents another concern. One might expect that if lack of consolidation is the key feature underlying remote memory loss, there would be a more abrupt change in the curve reflecting the point at which consolidation is complete. Hence the present data do not fully support the consolidation model. The multiple trace theory suggests that the hippocampus is always involved in memory retrieval [21]. According to this theory, feature information is stored in the neocortex, whereas information regarding spatial and temporal context is stored in the hippocampus. The hippocampus initially encodes the event, then serves as an index or pointer to the neocortical storage sites, binding all of the information into a coherent memory trace. Each time the memory is retrieved, reactivation of the trace occurs, and a new, often slightly different trace is established. This results, over time, in multiple, related traces that may facilitate retrieval. The theory supposes that older memories will have more traces, such that their retrieval is easier. If multiple traces make older events more resistant to damage, this may account for equivalent memory task performance between subjects with epilepsy and controls for more remote periods. By definition, however, the TNET assesses only memory for transient events. Hence, there should be no difference in the number of traces between older and newer memories, calling the multiple trace explanation into question. Standard ATL is not a selective surgery, however, in that other mesial and lateral temporal structures are resected. Hence, findings may relate to damage of surrounding areas in addition to the hippocampus. Furthermore, participants in the present study had variable pathology, some indicative of cortical as well as hippocampal abnormalities. Perhaps some subjects had more widespread disease contributing to remote memory difficulties. Study of a more restricted population (i.e., mesial temporal sclerosis only) would be necessary to further delineate this issue. Although the current data are not consistent with consolidation theory, and cannot be explained by the multiple trace theory, hippocampal damage is only part of the story in these subjects. The contribution of laterality was also examined, with a trend suggestive of remote memory deficits predicted by left-sided surgeries. This is consistent with prior reports indicating greater impairment in remote memory for public events in those with left TLE or left ATL [9,11,12]. The present data also indicate that greater right-lateralized memory task performance on pre-operative Wada testing predicts better post-operative TNET scores, consistent with the notion that resections of the dominant temporal lobe compromise remote memory function. Nevertheless, an effect of laterality
B.A. Leeman et al. / Epilepsy & Behavior 16 (2009) 113–119
may reflect the method of assessment rather than the site of remote memory storage itself, in that the TNET is administered verbally. The present effect of hemisphere (right vs left ATL) was not robust, however, and may indicate use of some nonverbal retrieval strategies as well (i.e., ‘‘picturing” the event). Several limitations of the present study should be mentioned, notably the use of retrospective data. Hence, some desired information was unavailable (e.g., pre-operative TNET scores, data regarding the time of last seizure prior to testing). It may also be questioned whether healthy subjects provided an appropriate control group. Future studies may be designed using controls with extratemporal lobe epilepsy, such that other factors would be comparable (e.g., use of anticonvulsant medications, time spent out of work or in the hospital due to seizures). A prior study suggested that those with extratemporal lobe epilepsy performed similarly to healthy subjects on tests of remote memory, however, with deficits present only in a TLE group [22]. On the basis of these prior data, healthy subjects appear to provide an adequate control. A limitation inherent to all studies of memory is the inability to control how a given subject will recall a specified event. Although the TNET proposes to assess retrograde, semantic memory, an episodic component to retrieval cannot be excluded. A subject may identify the current president, for example, because she or he recalls that fact (semantic memory), or because she or he remembers watching the inauguration ceremony (episodic memory). It is perhaps more accurate to state that the TNET assesses remote, declarative memory. Finally, readers are cautioned that results reflect group data, and individual measures may be quite variable. Future studies may be directed at prediction of individual remote memory abilities. In summary, the present data suggest the feasibility of using the Transient News Events Test for assessment of remote memory deficits in patients status-post temporal lobectomy for the treatment of refractory seizures. This method probes memory for items from discrete time periods, events that occurred transiently in the news. The findings suggest poorer remote memory task performance in the patient group than in healthy controls. The effects of time period tested, date of surgery or seizure onset, occurrence of recent seizure activity, and laterality of resection should be explored in future studies, as should the impact on quality of life and efficacy of possible interventions. Acknowledgments This work was supported by the Harvard Medical School Scholars in Clinical Science Program (K30 RR022292-07). The authors thank Dr. Steven Schachter, Dr. Blaise Bourgeois, Dr. Daniel Hoch, and Dr. Steven Greenberg for their invaluable comments regarding the protocol and data analysis.
119
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.yebeh.2009.06.012. References [1] Squire LR, Zola-Morgan S. The medial temporal lobe memory system. Science 1991;253:1380–6. [2] Aikia M, Salmenpera T, Partanen K, Kalviainen R. Verbal memory in newly diagnosed patients and patients with chronic left temporal lobe epilepsy. Epilepsy Behav 2001;2:20–7. [3] Pulliainen V, Kuikka P, Jokelainen M. Motor and cognitive functions in newly diagnosed adult seizure patients before antiepileptic medication. Acta Neurol Scand 2000;101:73–8. [4] Aarts JH, Binnie CD, Smit AM, Wilkins AJ. Selective cognitive impairment during focal and generalized epileptiform EEG activity. Brain 1984;107(Pt. 1):293–308. [5] Ortinski P, Meador KJ. Cognitive side effects of antiepileptic drugs. Epilepsy Behav 2004;5:S60–5. [6] Ivnik RJ, Sharbrough FW, Laws Jr ER. Anterior temporal lobectomy for the control of partial complex seizures: information for counseling patients. Mayo Clin Proc 1988;63:783–93. [7] Rausch R, Kraemer S, Pietras CJ, Le M, Vickrey BG, Passaro EA. Early and late cognitive changes following temporal lobe surgery for epilepsy. Neurology 2003;60:951–9. [8] Lah S, Grayson S, Lee T, Miller L. Memory for the past after temporal lobectomy: impact of epilepsy and cognitive variables. Neuropsychologia 2004;42:1666–79. [9] Lah S, Lee T, Grayson S, Miller L. Effects of temporal lobe epilepsy on retrograde memory. Epilepsia 2006;47:615–25. [10] Viskontas IV, McAndrews MP, Moscovitch M. Remote episodic memory deficits in patients with unilateral temporal lobe epilepsy and excisions. J Neurosci 2000;20:5853–7. [11] Voltzenlogel V, Despres O, Vignal JP, Steinhoff BJ, Kehrli P, Manning L. Remote memory in temporal lobe epilepsy. Epilepsia 2006;47:1329–36. [12] Barr WB, Goldberg E, Wasserstein J, Novelly RA. Retrograde amnesia following unilateral temporal lobectomy. Neuropsychologia 1990;28:243–55. [13] Kopelman MD, Wilson BA, Baddeley AD. The autobiographical memory interview: a new assessment of autobiographical and personal semantic memory in amnesic patients. J Clin Exp Neuropsychol 1989;11:724–44. [14] O’Connor MG, Sieggreen MA, Bachna K, Kaplan B, Cermak LS, Ransil BJ. Longterm retention of transient news events. J Int Neuropsychol Soc 2000;6:44–51. [15] Lah S, Lee T, Grayson S, Miller L. Changes in retrograde memory following temporal lobectomy. Epilepsy Behav 2008;13:391–6. [16] Kemp S, Coughlan AK, Goulding PJ, Abercrombie K. Measurement of remote memory pre- and post-temporal lobectomy: a longitudinal case study. Epilepsy Behav 2007;10:195–202. [17] O’Connor M, Sieggreen MA, Ahern G, Schomer D, Mesulam M. Accelerated forgetting in association with temporal lobe epilepsy and paraneoplastic encephalitis. Brain Cogn 1997;35:71–84. [18] Kapur N, Millar J, Colbourn C, Abbott P, Kennedy P, Docherty T. Very long-term amnesia in association with temporal lobe epilepsy: evidence for multiplestage consolidation processes. Brain Cogn 1997;35:58–70. [19] Holmes MD, Dodrill CB, Wilkus RJ, Ojemann LM, Ojemann GA. Is partial epilepsy progressive? Ten-year follow-up of EEG and neuropsychological changes in adults with partial seizures. Epilepsia 1998;39:1189–93. [20] Squire LR, Cohen NJ, Nadel L. The medial temporal region and memory consolidation: a new hypothesis. In: Weingartner H, Parker ES, editors. Memory consolidation: psychobiology of cognition. Hillsdale, NJ: Lawrence Erlbaum; 1984. p. 185–210. [21] Nadel L, Moscovitch M. Memory consolidation, retrograde amnesia and the hippocampal complex. Curr Opin Neurobiol 1997;7:217–27. [22] Bergin PS, Thompson PJ, Baxendale SA, Fish DR, Shorvon SD. Remote memory in epilepsy. Epilepsia 2000;41:231–9.
