Address correspondence and reprint requests to Dr. Harvey S. Levin, 1709 Dryden Road, Suite 725, Houston, TX 77030, U.S.A. E-mail: email@example.com
Summary: Purpose: To assess the pre- and postsurgical frequency of memory, emotional, and vocational impairments in patients who underwent anterior temporal lobectomy (ATL), and to assess the relationship between emotional disturbance and memory abilities after ATL.
Methods: Retrospective analysis of data was performed on 90 patients with medically intractable complex partial seizures who underwent ATL between 1981 and 2003. Patients were evaluated an average of 5 months before surgery and 11.3 months after surgery.
Results: A moderate to high frequency of memory impairment (44.4%; verbal or nonverbal), emotional disturbance (38.9%) and unemployment (27.8%) existed in the same individuals both before and after surgery. There were small to moderate rates of new onset memory (18.9%), emotional (11.1%), and vocational (7.8%) difficulties after surgery often regardless of seizure control outcome. Patients who underwent left-ATL and had emotional disturbance after surgery had the lowest verbal memory test scores.
Conclusions: Results highlight the importance of taking into account emotional status when assessing memory abilities after ATL. Results replicate the finding of moderate to high frequencies of memory impairment, emotional disturbance, and unemployment both before and after ATL. Results provide support for the rationale that cognitive, psychiatric and vocational interventions are indicated to mitigate the problems that exist before and persist after ATL.
A related goal of this study was to assess the relationship between emotional disturbance and memory impairment following ATL. Several studies have reported a relationship between depression and memory impairment in patients with medically intractable epilepsy (Dulay et al., 2004; Helmstaedter, 2004; Helmstaedter et al., 2004). These studies have shown that the association of memory deficit with depression was stronger in patients with a left-sided seizure focus than in right-sided patients. However, no previous study has characterized the relationship between postsurgical emotional disturbance and memory function after ATL.
A retrospective analysis of data was performed on a sample of patients who underwent ATL to relieve medically intractable complex partial seizures. Between July 1981 and December 2003, 164 patients underwent ATL and received neuropsychological testing at the Baylor Comprehensive Epilepsy Center at The Methodist Hospital in Houston. The final sample consisted of 90 patients who met inclusion criteria; (a) postsurgical Verbal Intellectual Quotient (IQ) or Performance IQ >70 (mean presurgery Full Scale IQ = 92.6; six patients excluded for an IQ below 70), (b) seizure outcome and neuropsychological functioning evaluated at least 3 months postATL (average of 11.3 months after ATL, range 3–45 months; an additional six patients removed who were tested within 3 months of surgery), (c) age 17 years or older at posttesting (an additional eight removed who were below age 17 years), (d) had pre- and postsurgical memory and IQ data (an additional 54 excluded for missing data). The mean age of the final sample at ATL was 31 years (range 17–57 years; 59% female). Eighty-eight percent of the sample was right-handed (Oldfield, 1971), 10% were left handed, and 2% were ambidextrous. The Baylor College of Medicine Institutional Review Board approved the study.
Lateralization and localization
Seizure localization and lateralization were confirmed upon concordance of 24-h video/EEG monitoring and neuroimaging data. Seizure monitoring involved scalp electrodes often with follow-up bilateral sphenoidal, depth, or subdural electrodes. At least three seizures were typically recorded over a period lasting about a week. Comprehensive reviews of video/EEG monitoring used in this study are described elsewhere (Mizrahi et al., 1990; Binne and Mizrahi, 1997).
Pre- and postsurgical seizure frequency was categorized (Spencer et al., 1982) into the following: [1a] seizure free with no auras; [1b] rare seizure with cause, or seizure free with 1–10 auras per month; [1c] 95% or greater reduction in seizures (excludes 1a and 1b) or greater than 10 auras per month;  75–94% reduction in seizures;  50–74% reduction in seizures;  less than 50% reduction in seizures; and  increased seizures. A good seizure outcome was defined as a 95% or greater seizure reduction after ATL (categories 1a, 1b, and 1c).
Retrospective chart review was used to determine the presence or absence of depression and/or anxiety. We defined emotional disturbance based on the examining neuropsychologist's recommendation for psychiatric treatment before and after ATL. The neuropsychologist's recommendation for psychiatric treatment was based on clinical interview, behavioral observations, review of history of psychiatric treatment and self-report (based on the Minnesota Multiphasic Personality Inventory [MMPI], Beck Depression Inventory [BDI], and/or Washington Psychosocial Seizure Inventory) (Dodrill et al., 1980; Beck et al., 1988; Butcher et al., 1989). In the case where no mention was made regarding emotional status by the neuropsychologist (within a neuropsychological report), the patient's self-report (based on a MMPI scale 2 or scale 7 T score >70 or a BDI score >12) was used to define emotional disturbance. There was no difference (Fisher's exact p value is ns) in the frequency of emotional disturbance when the diagnosis was based on the neuropsychological report writer's recommendations (addressed in 73.3% of the sample) versus the use of self-report only (addressed in 27.7% of the sample). Further, BDI and MMPI scores did not differ between the 73.3% that received recommendations versus the 27.7% that did not receive recommendations in the neuropsychological report.
Learning and memory were tested an average of 5 months prior to surgery (SD = 4.6) and an average of 11.3 months (SD = 8.6) after surgery using the consistent long-term retrieval (CLTR) and delayed free recall scores of the Verbal Selective Reminding Test (VSRT; Buschke and Fuld, 1974) and the nonverbal analogue to the VSRT, the Nonverbal Selective Reminding Test (NVSRT; Fletcher, 1985). The CLTR measures consistent recall of items over trials (sum of words consecutively recalled beginning with the first successful trial and ending with the 12th trial for the VSRT and 8th trial for the NVSRT) and delayed recall measures the number of items on the word list recalled 30 min after the 12 learning trials. The VSRT and NVSRT have been shown to be sensitive to lateralized memory impairment (Lee et al., 1989; Ribbler and Rausch, 1990; Plenger et al., 1996; Loring et al., 2000). Full Scale, Verbal, and Performance IQs were assessed using the Wechsler Adult Intelligence Scale—Revised or 3rd edition (Wechsler, 1981; Wechsler, 1997). Verbal memory z-scores were compared to Verbal IQ z-scores, and verbal memory impairment was defined as a delayed recall memory score 1.5 standard deviations below IQ level. Nonverbal memory impairment was evaluated similarly using the discrepancy between the nonverbal memory z-scores and Performance IQ z-scores. Defining memory impairment in this way (a) helped to take into account the fact that individuals with high IQs and average to below average memory functioning still have a relative and noticeable memory impairment and (b) helped to control for the overestimation of memory impairment in patients with lower IQs given that general cognitive impairment may emulate or exacerbate the appearance of memory losses (Kupke and Lewis, 1985; Oresick and Broder, 1988; Waldman et al., 1991; Rapport et al., 1997). The percentages of verbal or nonverbal memory impairment were found to be similar when a 1.0 standard deviation discrepancy was used. The memory and intelligence tests were part of a neuropsychological test battery administered in a 6-h session with a 1-h lunch break (York et al., 2003).
Employment information was obtained through structured interviews in the preoperative and postoperative assessments. Patients were questioned about current employment status, student status, and type of employment. If information was missing about employment, questions regarding employment status on the Washington Psychosocial Seizure Inventory were used when available (Dodrill et al., 1980). Employment was defined as having a part-time or full-time job, or being a homemaker or full time student.
A detailed description of the surgical procedure has been published (Mizrahi et al., 1990). Briefly, the anterior 3.5–4.0 cm of the lateral temporal lobe was resected en bloc from the superior temporal gyrus to the collateral fissure. Then, the anterior 3 cm of the hippocampal formation, the subiculum, and the parahippocampal gyrus were removed en bloc using an operating microscope, and the amygdala was removed with suction. No instances of wound or intracerebral infection, hemiparesis or death occurred following ATL. Surgeries were performed by one surgeon (RGG) between 1981 and 1999 and by another neurosurgeon (trained by RGG) between 2000 and 2003 (DY). The final sample consisted of 39 individuals with a left-sided ATL and 51 with a right-sided ATL.
Resected tissue was analyzed based on criteria established at the 1987 International Conference on the Surgical Treatment of Epilepsy (Armstrong and Bruton, 1987; Armstrong, 1993). The four pathology groups were classic Ammon's horn sclerosis (AHS), atypical mesial temporal lobe sclerosis, low-grade tumoral tissue (primarily ganglioglioma, with several dysembroplastic neuroepithelial tumor), or other (inconclusive, infarct, arteriovenous malformation, microdysgenesis). One neuropathologist (DDA) conducted all pathology examinations. Detailed descriptions of the neuropathological analyses are described elsewhere (Zhu et al., 1990; Zhu et al., 1997; Armstrong and Mizrahi, 1998; Erdamar et al., 2000).
First, potential selection bias was assessed by comparing demographic and disease characteristics between patients included in the study versus patients excluded from analyses using analysis of variance (ANOVA) and χ2. Next, independent-samples t-test and χ2 analyses were conducted to assess between-group (side of surgery) differences on the descriptive and outcome data (demographic, memory, emotional, employment, seizure outcome, and pathology data). Then, pre- to postsurgical changes in the primary outcome measures (any memory impairment, emotional disturbance, and unemployment) were assessed using the McNemar test. Next, χ2 analyses were conducted to assess between-group (good versus poor seizure frequency outcome) differences in the primary outcome measures. Finally, repeated measures ANOVA was used to test for interactions and main effects between emotional disturbance (based on postsurgical status), side of surgery, and verbal/nonverbal memory abilities. Raw memory scores (as opposed to the memory impairment categorical variable previously discussed) served as the dependent variable for this final analysis.
To assess selection bias, analyses were conducted comparing demographic and seizure data from patients 17 years and older excluded from the study (N = 66) to data from patients included in the study. ANOVA indicated no significant difference in age at surgery, years of education, years with seizures, age at onset of seizures, and presurgery seizure frequency between the 90 patients who were included in the study and the patients who were excluded. Similarly, χ2 analyses indicated no significant between group difference in seizure frequency outcome or type of neuropathology following surgery.
Table 1 displays the means, standard deviations, and percentages for demographic and disease features, IQ, seizure outcome, and pathology data by side of surgery and for the total sample. Independent samples t-test and χ2 analyses indicated no significant differences based on side of surgery for any of the demographic or seizure-related variables. Fig. 1 shows the frequency of any memory impairment (verbal or nonverbal), emotional disturbance and unemployment before and after surgery. There was a significant reduction in the percentage of patients who were unemployed after surgery (35.5%) compared to before surgery (46.7%) regardless of side of seizure focus (McNemar test p value <0.05). There was no significant change in the frequency of any memory impairment or emotional disturbance from before to after surgery.
Table 1. Demographic, IQ and seizure-related information divided by side of surgery
Right ATL (n = 51)
Left ATL (n = 39)
Total (N = 90)
ATL, anterior temporal lobectomy; M, mean; SD, standard deviation; 1a, seizure free with no auras; 1b, rare seizure with cause, or seizure free with 1–10 auras per month; 1c, 95% or greater reduction in seizures (excludes 1a and 1b), or greater than 10 auras per month; 2, 75–94% reduction in seizures; 3, 50–74% reduction in seizures; 4, less than 50% reduction in seizures; 5, increased seizure; AHS, Ammon's horn sclerosis; MTS, mesial temporal sclerosis; Independent-samples t-tests indicated no significant between-group differences for side of surgery on any of the demographic and seizure-related information. Preoperative assessment was performed an average of 5 months prior to surgery (SD, 4.6) and the postoperative findings were obtained after an average of 11.3 months (SD, 8.6).
Age at onset (yr)
Duration of illness (yr)
PreATL seizure freq. (month)
PostATL seizure freq. (month)
PreATL Full Scale IQ
Table 2 shows the percentages of individuals with memory (any memory impairment, verbal memory impairment, nonverbal memory impairment), emotional or vocational impairments present both before and after surgery, present only before surgery, new onset after surgery, and absent both before and after surgery. There were large percentages of patients with impairments before surgery that continued to exist after surgery. Of note, there were significantly more patients after left ATL with verbal memory impairment than right-ATL patients with nonverbal memory impairment (χ2= 4.2, p = 0.041; not shown in the table). Based on a concern that our classification system for memory impairment (e.g., verbal memory z score 1.5 SD below VIQ z-score) is not consistent with the traditional system of defining memory impairment (at a percentile cutoff based on published normative data), additional frequencies were computed to show how many patients in our sample performed below the 5th percentile rank (labeled as “any memory score <5%” in Table 2). Note the similar percentages between the two memory classification systems.
Table 2. The percentage of individuals with an impairment present both before and after surgery, present only before surgery, present only after surgery, and absent both before and after surgery
Present before and after surgery N (%)
Present only before surgery N (%)
New onset after surgery N (%)
Absent before and after surgery N (%)
Total N (%)
aIndicates the percentage of patients with a verbal or nonverbal memory impairment or both based on memory performance—1.5 standard deviations below IQ level.
bIndicates the percentage of patients with a verbal or nonverbal memory impairment or both based on memory performance below the 5th percentile rank.
Any memory impairmenta
Any memory impairment <5%b
Seizure outcome and impairment
Seventy percent of the sample (N = 63) had a 95% or better reduction in seizure frequency after surgery. The frequency of memory, emotional, and vocational impairments after ATL was compared between good and poor seizure outcome groups using χ2 analyses. There were no significant differences in rate of memory or vocational impairments between the good and poor seizure outcome groups. Also, Full Scale IQ level was not related to seizure outcome. There was a trend where the good seizure outcome group was more likely to be employed after surgery (Fisher's exact test p = 0.10).
To better understand what relates to employment after surgery, we assessed the relationship between IQ level and vocational status. Fig. 2 shows the relationship between IQ and unemployment. A Mantel-Haenszel test for linear association analysis indicated a significant difference in the frequency of unemployment as a function of IQ after surgery (p = 0.008). Specifically, there was a higher frequency of unemployment in patients with lower IQ levels, and all of the patients with an IQ <80 that were unemployed before surgery were still unemployed after surgery. An additional correlation analysis indicated that lower IQ level was correlated with earlier age at onset of epileptic seizures (r = 0.32, p < 0.05) suggesting that IQ is a proxy for other variables that lead to difficulty with employment (e.g., more severe epilepsy).
Emotional disturbance and memory
Repeated measure ANOVAs identified a significant Time (pre-, post-) × Side of Surgery (left, right) × Emotional Disturbance (present, absent) interaction for VSRT delayed recall raw score (Hotelling's Trace p values = 0.02) and a trend for VSRT CLTR raw score (Hotelling's Trace p = 0.06). Fig. 3 shows the three-way interaction for VSRT delayed recall demonstrating that patients with emotional disturbance and a left-sided surgery performed poorer on the verbal memory test compared to all other groups including left-sided surgery patients without emotional disturbance. A main effect was also found after surgery where patients with left-sided ATL had significantly poorer verbal memory abilities (using VSRT CLTR and delayed recall raw scores) compared to patients who underwent right ATL (Hotelling's Trace p values = 0.001). There were no significant interactions or main effects using nonverbal memory raw scores.
The present study assessed (a) the frequency of memory, emotional and vocational impairments before and after ATL, (b) the relationship between seizure frequency outcome with the primary outcome measures, and (c) the relationship between emotional disturbance and memory impairment after ATL. We extended the findings from previous research by finding a relationship between emotional status and level of verbal memory impairment after ATL. Consistent with previous research, we found that despite a good seizure outcome in more than two-thirds of the patients, memory impairment (63.3%), emotional disturbance (50%), and unemployment (35.6%) (Fig. 1 and Table 2) were frequently present at follow-up. Our findings provide support for the rationale that cognitive, emotional and vocational interventions are indicated to mitigate the problems that persist and interact after ATL.
A significant interaction among the presence of emotional disturbance, side of surgery and poorer performance on the verbal memory test was found after surgery (Fig. 3). Specifically, patients who underwent left ATL and had emotional disturbance after surgery had the lowest verbal memory test scores compared to all other groups. There was also a drop in the number of words recalled by patients without emotional disturbance after left ATL, but the reduction was significantly greater for patients with emotional disturbance. On the other hand, right-ATL patients with and without emotional disturbance maintained or improved their verbal recall from before to after surgery. No relationship was found between nonverbal memory and emotional disturbance.
Several reasons may explain why a relationship exists between emotional disturbance and verbal memory impairment after ATL. Verbal memory and learning deficits may contribute to frustration that exacerbates or induces chronic emotional disturbance (Giovagnoli et al., 1997), or emotional disturbance may induce cognitive deficits (Sun and Alkon, 2004). The relationship may also reflect a disruption in the limbic system's role in the regulation of both emotion and memory (Kanner, 2005). The lateralized findings may be associated with the hypothesis that emotional disturbance has a lateralized influence on cognitive functions only in patients with a left-sided seizure focus (Dulay et al., 2004; Helmstaedter et al., 2004). Similarly, others have reported the phenomenon of an interaction between depressive symptoms and cognitive impairment after a left hemisphere stroke, but not right hemisphere stroke (Robinson et al., 1986; Bolla-Wilson et al. 1989; Downhill and Robinson, 1994; Spalletta et al., 2002). The absence of a relationship between nonverbal memory and emotional disturbance after right-sided ATL may reflect this lateralized phenomenon. Covert verbalization of depressive thoughts and rumination may also interfere more with encoding words and paragraphs than visual designs and spatial loci used in nonverbal memory tests. The absence of a relationship between nonverbal memory and emotional disturbance after left ATL may reflect the unique association that emotional disturbance has with left-hemisphere mediated functions. It remains to be determined if there is a relationship between emotional disturbance and language functions such as confrontation naming and word fluency after ATL.
These results suggest that besides side of excision, the presence of depression and anxiety should be taken into account when evaluating and providing feedback to patients about verbal memory deficits after ATL. Questions remain regarding whether or not verbal memory deficits will improve after ATL if emotional disturbance is treated with psychotherapeutic or psychopharmaceutic interventions. Of note, studies in other patient populations have found that treatment of emotional disturbance is associated with improved cognitive functioning (Stoudemire et al., 1995; Fann et al., 2001; Doraiswamy et al., 2003; Rocca et al., 2005).
The frequency of any memory impairment (using both the criteria of memory impairment 1.5 SD below IQ level and <5% criteria) is within the range found in previous ATL follow-up studies (e.g., 11), and is consistent with previous findings indicating that left-sided ATL produces material-specific raw score verbal memory deficits (Ivnik et al., 1987; Chelune et al., 1991; Hermann et al., 1992; Martin et al., 1998; Helmstaedter et al., 2003). On the other hand, raw score nonverbal memory reductions were relatively infrequent following right ATL when nonverbal memory was measured using a recall procedure analogous to verbal selective reminding (Naugle et al., 1993).
There was significant improvement in our samples' unemployment frequency from 35.6% after surgery compared with 46.7% before surgery. Further, there was a trend where employment after surgery was associated with a good seizure outcome, which is consistent with previous research (Augustine et al., 1984; Kellett et al., 1997; Lendt et al., 1997; Jones et al., 2002; Maganti et al., 2003). Sperling et al. (1996) showed that complete seizure freedom and not just a reduction in seizure frequency after ATL might be the most important issue in gaining employment. We found that the frequency of unemployment was directly related to IQ given that there was a higher frequency of unemployment in individuals with an IQ lower than 80. In fact, no one with an IQ <80 (N = 17) became employed after surgery. Conversely, there was an association between having a higher IQ and being employed after surgery (Fig. 3). The finding that lower IQ is related to earlier age at onset of epileptic seizures is consistent with previous research (Strauss et al., 1995; Glosser et al., 1997, however, see Bjornaes et al., 2001; Thompson and Duncan, 2005) and suggests that a lower IQ level results from other factors (e.g., more severe epilepsy) with unemployment as the end result. The relationship between seizure control and employment may also be related to resumption of driving. For example, Reeves et al. (1997) found that employment after surgery was predicted by medical clearance to drive as a result of being seizure free.
Rehabilitation of secondary conditions appears warranted given the frequency of problems that exist before surgery and persist or arise after ATL. Although the need for interventions to mitigate emotional distress (Davis et al., 1984; Tan and Bruni, 1986; Gillham, 1990; Taube and Calman, 1992; Au et al., 2003; Goldstein et al., 2003), compensate for memory deficits (Aldenkamp and Vermeulen, 1991; Gupta and Naorem, 2003, see Schulman and Barr, 2002), and prepare patients for vocational reentry (Fraser et al., 1983; Fraser and Chaplin, 2001) has been noted for individuals with epileptic seizures, there is a dearth of research on specialized programs for post-ATL patients (Fraser, 1988). Further, there are no published randomized clinical trials that evaluated the efficacy of interventions to ameliorate the problems that occur after ATL. Prospective research is needed to study the efficacy of structured, comprehensive rehabilitation programs that aim to mitigate or compensate for secondary conditions after ATL. The similar frequencies of memory impairment, emotional disturbance, and unemployment in patients with and without a good seizure outcome found in the present study suggest that vocational, emotional, and cognitive interventions should target patients across the spectrum of seizure outcome.
Limitations of this study include its use of archival data, which potentially introduces selection bias. Thus, replication in a prospective cohort would strengthen extrapolation to other operated epilepsy patients. Other limitations include the lack of “Diagnostic and Statistical Manual of Mental Disorders” (DSM) diagnostic criteria for mood or anxiety disorders. It may be that the pattern of findings is related to the criterion measure of emotional disturbance and/or the specific DSM diagnostic category of mood or anxiety disorders. Another issue necessitating caution is the generalizability of our results to patients with new onset epilepsy, as well as to patients who are not good candidates for surgical intervention. Future studies should also assess other cognition domains besides memory. We focused on memory deficit, depression, and vocational disability because these secondary conditions were frequently present in our cohort and all three are potentially amenable to postsurgical rehabilitative and psychiatric interventions. The small to moderate sample size necessitates the need for future research to replicate the findings before final conclusions can be made.
In conclusion, moderate to high frequencies of memory, emotional and vocational impairments occur before ATL, and these impairments persist after ATL often regardless of good seizure control. Further, the presence of an emotional disturbance after left ATL may exacerbate verbal memory problems beyond the well-documented levels typically found at follow-up in these patients. Results support the rationale that cognitive, emotional and vocational interventions are needed to mitigate the problems that persist after ATL.