One of the aims of neuropsychological evaluation that is particularly relevant to the preoperative evaluation of the seizure patient is to provide information that may be helpful in lateralizing or localizing cerebral dysfunction. Although the role of neuropsychology in this process may appear diminished in the face of improved functional and structural neuroimaging procedures, neuropsychologists are often called upon to provide convergent data or to assist when these medical technologies provide equivocal results. This role is well illustrated in the case of temporal lobe epilepsy (TLE), a realm where the measurement of cognitive skills has been used to infer the integrity of medial temporal lobe (MTL) structures. Of particular
interest in this realm has been the examination of material-specific memory deficits and the degree to which such deficits may be helpful in lateralizing dysfunction to the right or left cerebral hemisphere. Specifically, as discussed subsequently, evidence has linked left MTL structures to the mediation of memory for verbally encodeable material; although evidence has been more mixed in the case of right MTL structures, they have been considered to possibly have a unique role in mediating memory for nonverbal material.

The iterations of the Wechsler Memory Scale (WMS) have been among the most widely used batteries in both clinical and research contexts with respect to the lateralization of MTL dysfunction. Unfortunately, they have not been consistently successful in assessing material-specific memory as the construct is currently conceptualized. Factor analytical work on the latest version of the WMS (WMS-III; [1]) has suggested that a three factor model comprising working memory, auditory memory, and visual memory best characterizes its structure (2,3). However, these studies have cast doubt on the cohesiveness of the visual factor, as the primary visual subtests (Faces and Family Pictures) demonstrated weak relationships with one another. That these subtests are not strongly correlated is also reported in the Wechsler Adult Intelligence Scale (WAIS-III)/WMS-III Technical Manual (4). Within patients with TLE, a group in which one would expect material specificity of memory to be particularly observable due to the frequent unilateral nature of their MTL pathology, the WMS-III was found to be best characterized by a two-factor model comprising working memory and general memory (5).

Correlation of WMS-III components with neuroimaging data has also provided mixed evidence with regard to its ability to detect lateralized dysfunction. For example, whereas the Immediate and Delayed Auditory (verbal) Memory Indices of the WMS-III were reported to be significantly correlated with left hippocampal volume, the Immediate and Delayed Visual (nonverbal) Indices were correlated with both right and left hippocampal volume (6). At the subtest level, findings have been mixed even with regard to the primary measures of Auditory memory (Logical Memory and Verbal Paired Associates [VPA]). For example, it was reported that the left, right, and combined bilateral MRI hippocampal volumes were positively correlated with percent retention on the VPA subtest, and the bilateral MRI hippocampal volume was positively correlated with percent retention on the Logical Memory subtest (7). Neither faces nor family pictures were significantly related to hippocampal volumes in this investigation. Another study reported that the metabolic functioning of the left hippocampal formation in patients with TLE was associated with the percentage of information recalled on the Logical Memory subtest, although a supplementary visual subtest, Visual Reproduction (VR), was not associated with right hippocampal functioning (8). In summary, although certain aspects of the WMS-III appear closely linked to neural substrates of memory, the ability of this measure to lateralize MTL pathology, particularly with regard to right MTL structures, has not been consistently supported.

In addition to examining the correlation of WMS-III components with structural and metabolic variables, its utility in detecting lateralized cerebral dysfunction has been approached by comparing performances within and across populations with known unilateral seizure foci. Given the theoretical assumptions on the material specificity of memory, one would expect to see a double-dissociation where the left temporal lobe epilepsy (LTLE) group both performs more poorly on measures of auditory memory compared with the right temporal lobe epilepsy (RTLE) group and demonstrates a relative weakness on these measures, although the RTLE group exhibits the opposite pattern of depressed visual memory on both between- and within-group comparisons. The WAIS-III/WMS-III Technical Manual (4) presents data on a small group of patients with unilateral TLE, 15 with LTLE and 12 with RTLE. Between
groups, the LTLE group performed more poorly on the WMS-III Auditory Indices, but the two groups had equivalent performances on the Visual Indices. However, the within-group comparisons demonstrated the expected pattern, with the LTLE group performing better on Visual measures compared with Auditory measures and the RTLE group performing relatively worse on Visual measures compared with Auditory measures.

Later work with a larger sample had similar findings as patients with LTLE performed more poorly than patients with RTLE on the Auditory Delayed Index, particularly on the VPA subtest, with no differences emerging for Visual domains (9). The within-group comparisons again revealed relatively lower Visual performances for patients with RTLE; however, the LTLE group exhibited a general reduction across both visual and auditory domains. The presence of unusually large differences between the indices (occurring in <5% of the standardization sample) was more helpful in correctly lateralizing, but such differences were rarely observed in this sample, limiting the clinical usefulness of such an approach. In summary, a variety of research methodologies including factor analytic, neuroimaging, and psychometrically based approaches have indicated that the WMS-III has shortcomings in its ability to discriminate between patients with LTLE and RTLE on the group level.

The WMS-III has demonstrated greater ability to differentiate LTLE and RTLE groups following surgical intervention (10), though this information is obviously less relevant to the aims of preoperative neuropsychological evaluation. Although difficulties in differentiating these groups preoperatively may reflect problems in current conceptualizations of memory, problems with the test structure itself may also be contributory. For example, within the WMS-III Visual Memory Index, the Family Pictures subtest is of particular concern due to questions over what abilities it actually measures. From a theoretical perspective, one of the strengths of this subtest in its sensitivity to right temporal localization would appear to be its assessment of recall for spatial location (11). Unfortunately, this aspect of the subtest is not quantitatively differentiated from memory for character names and activities, with these latter aspects requiring significant verbal information processing. In addition, memory for verbally presented stories has been found to be the strongest predictor of Family Pictures performance in patients with seizures, even when considered in the context of measures of working memory and nonverbal memory (11). In this study, no between-group differences were observed on the Family Pictures subtest between patients with LTLE and RTLE. The other primary Visual Index subtest, Faces, has also been reported to not differentiate between LTLE and RTLE groups (9) and has not consistently demonstrated strong loading on Visual Memory factors in either nonclinical (2) or TLE populations (5).

Difficulties in detecting lateralized dysfunction of nondominant temporal lobe structures are certainly not unique to the WMS-III. For example, although patients with LTLE have been found to perform more poorly than patients with RTLE on other measures of verbal memory including list learning and story recall tasks (12,13), the groups have been found to not differ on recall for purported measures of nonverbal memory, including memory for geometric designs and figures (13). In addition, although verbal memory measures, particularly list learning, have been found to significantly predict Wada performance during right hemisphere anesthetization, the aforementioned nonverbal memory performances were not predictive of performance with anesthetization of the dominant left hemisphere. In fact the converse was observed, as they were predictive of performance during right hemisphere injection (13).

In addition to measures of memory, measures of confrontation naming have been examined as potential contributors to the lateralization of MTL dysfunction, particularly as hippocampal functioning has been linked to naming and word-retrieval abilities (8,14,15). The Boston Naming Test
(BNT; [16]) has been reported to significantly predict lateralization of the site of later surgery above and beyond other commonly used presurgical measures, including the WMS-III Auditory and Delayed Memory Indices and the WAIS-III (17) Verbal Comprehension and Perceptual Organization Indices (18). The BNT’s predictive ability was moderated by WAIS-III full-scale IQ (FSIQ), age at seizure onset, and duration of epilepsy. Specifically, the ability of this measure to predict ultimate left-sided surgery was strongest when patients exhibited relatively low BNT score, high FSIQ, later age of onset, and shorter duration of epilepsy. The likelihood of right temporal resection was best predicted in the face of high BNT score, low FSIQ, early onset of seizures, and shorter duration of epilepsy.

In summary, the commonly used neuropsychological measures have shown some promise in detecting dysfunction of the usually dominant left MTL structures, although their ability to infer right MTL dysfunction has been less successful. In terms of memory, current neuropsychological measures in isolation appear most sensitive to material specificity only after anterior temporal lobectomy (ATL) (10). A number of issues including the dually encodeable nature of currently used stimuli, the possibility of a generalized left hemisphere dominance for memory, and the involvement of extrahippocampal structures in memory have been suggested as potential contributors to current difficulties in lateralizing dysfunction (13,19,20). It is important to note, however, that the often discouraging findings reviewed in the preceding text are on the basis of group-level analyses. These present a different scenario than that approached by the practitioner, who attempts to draw conclusions with consideration to both group-level and intraindividual comparisons. This practice is best approached within a multidisciplinary context that allows integration with Wada testing and appropriate neuroimaging and electroencephalographic data. Of additional importance for consideration are seizure-related variables such as the chronicity of an individual’s epilepsy. For example, an earlier age of onset may be more likely associated with cerebral reorganization of memory processes (13), less material-specific memory impairment (21), and less postsurgical decline (22).

Another important aim of the neuropsychological evaluation in the context of epilepsy surgery is to anticipate potential postoperative declines in cognitive abilities. This task is complicated by the fact that many variables other than baseline test performance are related to postsurgical cognitive outcome including age of onset and duration of epilepsy, degree of hippocampal sclerosis, age at the time of surgery, seizure etiology, localization and lateralization of seizure focus, side of resection, degree of seizure control achieved, medication issues, behavioral factors, and gender (14,23,24,25,26). Further complicating this area of work, the measurement of cognitive change is influenced by such factors as regression to the mean, practice effects, and test-retest reliability (27,28). Although there have been numerous investigations of optimal approaches to predicting postoperative decline, drawing firm conclusions from this literature is difficult because of the use of different neuropsychological assessment instruments, the evolution of different approaches to surgery (e.g., en bloc vs. tailored resections), and inconsistency in time to follow-up assessments across studies.

Although chronic epilepsy has been associated with the progressive decline of cognitive functioning (29,30), ATL does not appear to cause further decline in most general intellectual abilities and may even be associated with improvement in some areas (23,31). However, surgical intervention has been often associated with further decline in memory functioning, particularly in terms of verbal memory. This is observable in both early and late postoperative stages, with left anterior temporal
lobectomy (LATL) and higher preoperative verbal memory abilities being particularly consistent predictors of postsurgical verbal memory decline (26,32,33,34,35,36). Although higher preoperative abilities likely reflect greater integrity of associated cerebral structures, “(a) low preoperative performance level often indicates damage of the corresponding structure and is usually associated with a relatively stable pre- to postoperative performance because patients cannot lose much more, when the damaged structure is removed” ([33], p. 93).

In addition to higher baseline verbal memory performances and LATL being related to greater postsurgical declines in verbal memory, these variables in the context of poor right hemisphere mediation of memory as assessed by the Wada have been reported to be predictive of postoperative decline in WMS Logical Memory performance (34). Similarly, higher baseline verbal memory performance in the context of lower baseline nonverbal memory scores were reported to be significant predictors of verbal memory decline on a word list learning task 3 months following left selective amygdalohippocampectomy (SAH; [33]). Illustrating the complicated picture created by other variables, older age at surgery and later age of epilepsy onset were also significant predictors of the level of decline. In addition, another study reported that the predictive ability of preoperative level of performance might differ on the basis of the specific neuropsychological measure being considered and the side of resection. For example, higher levels of verbal memory abilities may be related to greater decline for patients who underwent LATL only, although higher nonverbal memory abilities may actually be related to less decline and be independent of the side of the resection (37).

Lee et al. (35) recently conducted a meta-analytic review of the literature on surgery-related changes in ATL. They selected only studies that assessed memory performance with WMS Logical Memory and VR subtests in the left language-dominant patients who underwent ATL before and within 24 months after surgery. The findings were reported to indicate that in patients who underwent LATL, verbal memory performances were generally lower than that in patients who underwent right anterior temporal lobectomy (RATL) both before and after surgery. Consistent with the findings reported in the preceding text (10), this finding was particularly pronounced after surgery. Examining within-group changes, the LATL group demonstrated a trend toward significantly declined verbal memory following surgery, but did not demonstrate the hypothesized postsurgical improvements in nonverbal memory. Although some trends emerged in the expected directions for patients who underwent RATL, overall they did not demonstrate significantly poorer relative performance on measures of nonverbal memory before or after surgery, and also did not show any statistically significant decline in this domain following surgery. Overall then, this review indicated that the risk of postsurgical decline is greatest for verbal memory functions following LATL, with relatively little change being consistently reported following RATL.

However, caution should be used in applying these findings to the individual case, as they are based on group-level comparisons that may not fully appreciate the number of individual patients who demonstrate significant changes in each domain. For example, one investigation reported no clinically meaningful group-level declines on verbal or nonverbal memory tasks, although examination of the base rates indicated that 50% of a LATL group and over 30% of a RATL group demonstrated clinically significant decline on Immediate and Delayed Logical Memory (37). In addition, approximately 20% from each group demonstrated meaningful declines on Immediate and Delayed VR. A later investigation had similar findings, with up to 51% of patients undergoing left SAH and up to 32% of those undergoing right SAH demonstrating significant losses on a measure of list learning (33). Also consistent with the previous investigation, approximately 27% of each group demonstrated a decline on a measure of memory for abstract designs. Beyond side of
resection, these rates of change may differ based on preoperative ability levels and the specific ability under study. For example, one group reported that 75% of patients with an average or above-average preoperative language/verbal memory abilities who underwent LATL demonstrated significant declines in verbal memory, although such declines were seen in only 30% of those with low-average baseline abilities and 40% of those with borderline or impaired abilities (32).

As indicated in the discussion of approaches to lateralization of cognitive dysfunction, language abilities are also closely related to MTL functioning and therefore represent an area of potential decline following ATL. Although declines in overall verbal abilities as captured by Verbal IQ (VIQ) have been reported following ATL, a recent investigation suggested that these abilities largely return to baseline or even improve by 6 years postsurgery (23). Perhaps of greater risk are confrontation naming abilities, a skill-set with reported ties to MTL functioning that is not specifically reflected in the VIQ. Declines in confrontation naming abilities have been reported in up to 40% of LATL cases, but do not appear to be a consequence of RATL (37). It has been reported that such declines after LATL are more likely to be observed in patients with later onsets of their seizure disorder (38) and mild to no baseline hippocampal sclerosis (14). Complicating the ability to predict such decline from a neuropsychological standpoint, the latter study reported that preoperative naming abilities did not significantly predict the level of postoperative decline.

In summary, the ability of the neuropsychological examination to predict postsurgical changes in the cognitive functioning is complicated by a number of factors, including the influence of multiple demographic, medical, and psychological factors. Most consistently reported are findings that LATL carries a relatively elevated risk of postsurgical decline in verbal memory and confrontation naming abilities. Also, in the case of verbal memory in particular, higher levels of preoperative performance further elevate the risk of significant declines. However, such general findings often break down at the individual level, as illustrated by the fact that verbal memory declines may also be seen in a significant number of patients undergoing RATL. Declines in nonverbal memory appear largely independent of the side of resection, and have been reported to occur in as many as 30% of the cases. Newly emerging data suggest that the advent of tailored approaches to ATL, however, may produce much lower rates of postsurgical cognitive decline in these domains than previously reported, although still yielding equivalent levels of seizure control (39). As in approaching the task of preoperative lateralization, attempts to predict the potential cognitive morbidity of epilepsy surgery should not be carried out using baseline psychometric data alone, as such data must be considered within the broader context of the aforementioned moderating variables. Finally, it is important to remain sensitive to the influence that cognitive abilities have on functional skills, emotional well-being, and overall quality of life (36).