Anatomical versus functional integrity

Patients with medically refractory epilepsy are at risk for gradual neuropsychological deterioration, poor academic performance, decreased occupational success, pain disorders, neurobehavioral conditions, and reduced quality of life [10–13]. Temporal lobe epilepsy (TLE) is the most common epilepsy form amenable for surgical resection [14]. Patients undergoing anterior temporal lobectomy (ATL) have a 1-year postsurgical seizure freedom rate in the order of 65% to 77% [15, 16]. Seizure freedom rates at 10-year postsurgical follow-up have been reported to be 27% and 66% for frontal and TLE, respectively [15–17]. Wide variations in seizure freedom rate have been reported for nonlesional MRI patients, ranging from 18% to 63% [18, 19]. A positive MRI is a consistent predictor of favorable seizure outcome after ATL.

High-resolution MRI is used presurgically to identify structural brain abnormalities. MRI lesions are often in the area of seizure onset, and generally correlate with the dysfunctional region. However, around one-third of patients undergoing ATL have negative MRI findings [20, 21]. As clinical lesions are known to alter normal brain functions, negative MRI findings suggest greater functional brain integrity, and are associated with a poorer prognosis after resective surgery [18]. The presurgical functional integrity of the tissue to be resected is also a strong predictor of postsurgical memory deficits following ATL [22]. Besides MRI-negative results, other indicators of presurgical functional integrity of the temporal region to be resected include greater activation on fMRI during memory tasks, a lack of significant asymmetry in TL activation on FDG-PET [23], and intact memory on presurgical neuropsychological assessment or Wada test [24].

Memory decline is the most frequently reported deficit in TLE patients after resective surgery. Patients with a language-dominant TL seizure focus are at risk for auditory/verbal memory deficits, whereas those with a nondominant TL seizure focus may display visual memory deficits [25, 26]. Deficits in naming ability (word-finding, naming to description, confrontational naming) and semantic fluency are also frequent in patients with dominant TL seizures [27, 28]. Recent research suggests these deficits are much more pronounced when sampling a broader range of object type (e.g., names of persons and landmarks are more impacted than man-made objects) [29, 30]. Patients with average or higher memory and language scores present higher risk of postsurgical memory/language declines compared to those with lower scores in these domains [31], although functionally appreciable declines can continue to occur even in patients with impaired performance on standardized cognitive measures.

Neuropsychological deficit patterns as an aid to confirm seizure localization

Due to the lack of abnormality, MRI-negative patients tend to be more challenging for seizure localization than lesional patients. In such cases, neuropsychological evaluation can sometimes be useful for confirming the seizure onset zone. The vast majority of neuropsychological studies have focused on TLE; some have studied frontal lobe epilepsies (FLE) and very few parietal (PLE) or occipital epilepsies (OLE). Nevertheless, some characteristic neuropsychological deficit profiles have emerged over time, and can be used to confirm lateralization or localization of seizure onset in many epilepsy patients. Tables 20.1A and B summarize some of these findings, and include specific neuropsychological tests that can be used to determine the presence of cognitive deficits associated with each brain region (tables adapted from [33, 34]).

Table 20.1A Neuropsychological profiles of presurgical deficits in temporal lobe epilepsy

L: Left, R: Right.

Table 20.1B Neuropsychological profiles of presurgical deficits in extratemporal epilepsies

L: Left, R: Right, PCE: Posterior cortical epilepsies, OLE: Occipital lobe epilepsies, PLE: Parietal lobe epilepsies

The natural spread of epileptiform activity through neuronal networks can cloud these patterns, with neurocognitive deficit patterns often suggesting more widespread dysfunction than the focal seizure onset zone. Confounding patterns are more likely to be observed if patients are tested following secondarily generalized seizure spread, and these deficits can persist beyond the postictal period if they occur with regular frequency and intensity. In fact, widespread structural and functional changes have been observed in such patients using PET scans [35], diffusion tensor imaging (DTI) [36], and volumetric studies [37], with some of these abnormalities improving or resolving if seizure control is achieved. In such cases, the neurocognitive results will sometimes implicate widespread regions of dysfunction, yet there are still frequently patterns across tests that can be useful for sorting out the main area of abnormality. For example, while memory deficits are common in patients experiencing either TLE or FLE, the pattern of performance across subcomponents of memory measures can suggest greater involvement of specific brain regions. Patients with FLE often exhibit more difficulty with organizational aspects of learning, which can lead to problems with both encoding and retrieval information, while recognition recall may remain intact [38, 39]. In contrast, TLE patients often do not improve with recognition cueing. The FLE patients have also been shown to exhibit problems with release from proactive interference [40], recalling the temporal order of events [41], and recalling the context of learning [42], errors which can contribute to heightened “false positive” identifications and confusion of information across tests. While both TLE and FLE patients frequently exhibit verbal fluency problems [28], FLE patients have been shown to exhibit greater improvement when provided with structured cueing on such tasks [43]. Traditionally, this pattern was thought to demonstrate that their deficits resulted from organizational aspects of searching memory (access problems) rather than a primary loss of information (degradation). More recent work is suggesting that the TLE patients may also have access problems, which results from structural “decoupling” of network regions rather than access problems resulting from executive dysfunction seen in FLE [44]. The TLE patients often exhibit improvement in verbal fluency tasks that are more dependent on frontal lobe regions if they become seizure-free (e.g., letter and action “verb” fluencies), while functions more dependent on TL networks tend to persist or worsen with surgery (e.g., semantic/category fluency) [45].

Some deficit patterns are more associated with specific regions, such as the tendency for left TLE patients to exhibit auditory/verbal memory problems and naming deficits and for right TLE patients to sometimes exhibit visual memory deficits [25, 26]. However, as noted, many epilepsy patients will experience broader structural and functional deficits, and will not produce these focal neurocognitive profiles [12]. Left TLE patients with more mesial dysfunction will frequently exhibit deficits in associational learning [46] and binding materials that cut across sensory, motor, and cognitive domains (e.g., relating a name or broader linguistic information with a face or voice). Some evidence suggests that contextual learning, such as story recall, may be more dependent on lateral TL regions [47]. Naming has also been more associated with lateral and temporal polar regions [48, 49], although indirect evidence has led some to conclude hippocampal involvement in naming [50]. Recent studies have found that right TLE patients frequently exhibit recognition deficits that appear more akin to a limited “prosopagnosic” condition (e.g., not recognizing familiar or famous faces, animals, or landmarks) [44, 51]. In addition to the memory and verbal fluency patterns noted above, FLE patients will also exhibit deficits in attention, motor processing [52], behavioral problems [53], and broader executive skills (e.g., metacognition, sequencing, set shifting, response inhibition) [54–56].

At present, we lack definitive profiles for preoperative functioning in the posterior cortical epilepsies and have no prospective postsurgical outcome studies available for this patient group [33]. Data from other neurological groups with parietal lobe damage suggest that this group might exhibit deficits involving visuoperception, sensory processing, visual–spatial processing, constructional praxis, and attention at baseline. There are a few isolated research reports suggesting presurgical deficits in some of these areas (e.g., visual–spatial processing, sensory functioning) [57]. Presurgical visual field defects appear to occur more often in OLE patients with mesial rather than lateral OL seizure onset zones [58]. One very small, retrospective study examining the neurocognitive status of children with occipital lobe (OL) seizure onset suggests that such patients experience an elevated rate of scholastic difficulty, psychiatric disorders (i.e., primarily depression), and cognitive dysfunction involving problems with face processing and making spatial judgments [59].

Neuropsychological outcomes in MRI-negative patients

Neuropsychological outcomes, in MRI-negative patients have recently gained increasing attention, although only a few studies have directly addressed this issue. Many older outcome studies focused exclusively on patients with MTS and excluded more complex cases (e.g., post-traumatic or postencephalitic epilepsies) in an effort to produce the cleanest characterization patterns. More recently, the number of MTS cases at most epilepsy surgical programs has seemed to decrease with a proportional rise in the number of MRI-negative cases and those with widespread brain changes (e.g., frontal lobe encephalomalacia in a post-traumatic epilepsy patient with TL seizure onset). From the limited number of available studies, it appears that MRI-negative TL patients are more susceptible to cognitive decline following resective surgery than lesional TL patients [19, 31, 60, 61], although some studies have failed to find a significant difference between these patient groups [62]. These latter studies often involved small sample sizes, differed in their selection criteria for offering resective surgery to nonlesional patients, and varied in the neuropsychological measures used.

Based on available data, some general trends in well-localized, nonlesional cases include: (a) greater declines in verbal memory following left ATL resection [19, 31, 60], (b) greater declines in visual memory following right ATL procedures [60], and (c) greater declines in visual confrontation naming performance and semantic fluency in left TL cases [19].

Helmstaedter and colleagues [60] reported in a small series of individuals with TLE that MRI-negative patients exhibited better memory functioning at baseline than MRI-positive patients, and declined more significantly following resective surgery. At follow-up, both groups were found to be comparable in performance level, and it was concluded that the MRI-negative group had more to lose with surgery. Overall, findings were strongest for verbal memory decline in left MRI-negative patients, but significant declines were also observed in visual memory for right MRI-negative patients. In contrast, a research group in Finland [62] reported that declines in memory were comparable for lesional and nonlesional TLE patients, although seizure outcome was poorer in nonlesional patients (40%). This study, however, employed a limited range of cognitive measures, and it did not include tasks that have been proven to be especially sensitive to hippocampal function (e.g., associative learning tasks). Using a more sensitive list learning task, visual confrontation naming and semantic fluency, Bell and colleagues [19] found greater declines in verbal memory in a small group (n = 40) of MRI-negative TLE patients. They reported seizure improvement in 60% of their patients, although their study was retrospective and the choice of candidates may have been more selective than other studies. Using regression-based measures to identify individual patient declines, Seidenberg and colleagues [31] found that following left ATL, MRI-negative TLE patients declined on multiple verbal memory measures, memory for simple geometric designs, confrontation naming, and verbal conceptualization; whereas declines in MRI-positive left TLE patients were restricted to the ability to learn hard word-pair associations. Patients undergoing right ATL without structural MRI abnormalities demonstrated decline in memory for simple geometric designs, whereas right ATL patients with MRI abnormalities demonstrated no significant postoperative change.

Knowledge of cognitive outcome following surgery in extratemporal lobe epilepsies is primarily restricted to FLE. Moreover, nonlesional MRI status has not been systematically studied as a predictor of outcome in any of the extratemporal epilepsies.

Although there is not a great deal of postsurgical data available for FLE, there is evidence that both motor and neurocognitive functions decline with FL surgeries [38, 63, 64]. Outcome seems dependent on location of the surgical resection, and there are no definitive studies clearly indicating whether this differs on the basis of MRI status. It seems probable there will be some interaction between lesion location and MRI status, but larger, prospective studies will be required to establish such patterns. Deficits observed in FLE patients have included a variety of deficits in executive control processes (e.g., response inhibition, generative fluency, complex problem solving), aspects of memory performance, and motor functions [38, 63, 64].

There are few studies examining postsurgical outcome for the posterior cortical epilepsies, with most representing small, retrospective case series that blend a number of heterogeneous patients. There has been no examination of MRI status in these studies, although many of the posterior cortical epilepsies are lesional in nature. For OLE, a few small case series studies have highlighted declines in performance-based intellectual functioning, aspects of visuoperceptual processing, and a variety of visual abnormalities (e.g., visual-field cuts, alterations in color vision) [65–67]. A couple of studies have indicated that larger OL resections result in greater declines in intellectual functioning. Verbal intellectual functioning often improves in these patients with improved seizure outcome. With parietal resections, there have been reports of sensory deficits with inclusion of the postcentral gyrus, visuoperceptual, and constructional deficits following right-sided resections, and language deficits after left parietal resections.

Neuropsychological testing as an aid to predict seizure freedom

Blind assessment of neuropsychological data has not proven very useful for predicting seizure freedom following surgery, likely due to many patients having structural and functional deficits that greatly exceed the seizure onset zone. However, when combined with other available data, neuropsychological test scores often add incremental value for predicting seizure freedom. Studies in this area, which have generally been few in number, have often examined more selective groups of epilepsy patients (e.g., TLE cases only, excluding post-traumatic epilepsy); and have typically analyzed the multivariate prediction of seizure freedom based on available neurological, disease-related, and neuropsychological factors [31, 68, 69]. The presence of an MRI lesion tends to be one of the stronger predictors of seizure freedom regardless of location of seizure onset [70]. Nevertheless, neuropsychological test scores and Wada memory lateralization scores both contribute unique variance to the prediction of seizure freedom in both lesional and nonlesional epilepsy. In TLE cases, for example, Wada memory asymmetries that strongly favor the functioning of the contralateral (unresected) hemisphere, significantly predict a better postoperative seizure outcome than cases where this asymmetry is lacking [71]. Hennessy et al. [72] reported that patients having neurocognitive profiles that lined up with the side of surgery based on expected patterns of function tended to have better seizure outcomes than those that did not, although this was not always the case, especially for right ATL resection candidates. Potter et al. [68] observed that an index of language function that included confrontation naming and nonverbal memory added unique variance to predicting seizure freedom beyond that provided by demographic variables (side of surgery, age of epilepsy onset) and MRI findings of MTS. There have also been some studies demonstrating that patients with lower baseline general intellectual functioning experience worse postoperative seizure control [72–74]. This again seems consistent with the idea that a focal resection in a patient with essentially global brain dysfunction is less likely to be successful than in a patient with well-localized, focal deficits consistent with the presumed seizure onset zone.

Finally, it is important for the neuropsychologist to provide input to the epilepsy surgery program regarding potential neuropsychological risks with surgery. Although a patient may have neurological findings supportive of having a good seizure-free outcome, the possible risk of significant neuropsychological deficits could outweigh the potential benefits of becoming seizure-free [75].

Neuropsychological assessment

One of the best predictors of postoperative neuropsychological decline is the cognitive test performance obtained from presurgical neuropsychological baseline assessment, although laterality of seizure focus is also an important factor. Neuropsychological scores are robust predictors with a unique variance to predict memory outcome [34].

Patients with higher cognitive performance at baseline are at higher risk of postsurgical decline [32, 69], whereas those with baseline deficits are at less risk of decline. As MRI-negative patients tend to have higher cognitive scores at baseline, it is not surprising that they would have an elevated risk of decline. One mediating variable appears to be age of seizure onset, with an earlier age of onset associated with increased likelihood of reorganization of brain functions and less risk of decline [76]. Patients whose neuropsychological assessment deficits lateralize ipsilaterally to the side of their planned surgical resection tend to have better outcomes than those whose results are less clearly lateralizing. Nevertheless, patients with good presurgical lateralization to the hemisphere to be resected can still experience a decline following surgery.

Wada test

The regions of the brain that control the functional capacity of language and memory vary widely across individuals. Their lateralization has been historically done using the Wada test to assess function of each cerebral hemisphere in independently supporting memory and language prior to surgery [77].

Many comprehensive epilepsy centers have replaced the Wada test with fMRI or MEG, or both, and there continues to be debate about the role of Wada because of the variability of Wada protocols and their lack of standardization [77]. However, the Wada test remains to be an important clinical tool for establishing language and memory hemispheric dominance for predicting neuropsychological outcome of epilepsy surgery [78]. Research suggests that Wada data are somehow independent from fMRI or MEG data [77]. More specifically, while the Wada test assesses behavior with inactivation, fMRI and MEG assess behavior with activation. Thus, the Wada test approximates the effects of surgery, but it is being replaced in part by fMRI due to its invasive nature.

As indicated in Table 20.1, when memory is adequate with ipsilateral amobarbital injection and poor with contralateral injection, then the Wada test results suggest a lesser risk of memory deficits after resection. Conversely, when memory is poor with ipsilateral injection and adequate with contralateral injection, then there is a higher risk of postoperative memory loss.

Increasing evidence suggests that more selective surgical procedures as opposed to extended standard resections can reduce the cognitive deficits associated with surgery [48, 60], providing support to the functional adequacy model.

CASE 1: Left TLE; MRI-negative

The patient is a right-handed male in his early 40s presenting with an approximate 15-year history of epilepsy. His first seizure was characterized by loss of memory followed by secondary generalization of the seizure. He reported having two to three seizures per year, described as “spacing out” or “staring,” but these events did not generalize when he was compliant with his medications. He was previously a heavy alcohol drinker, and he described multiple convulsive events in the context of alcohol withdrawal, but he had been abstinent for over 10 years. His reported seizure frequency was likely underestimated based upon the discrepancy of self-reported events vs. recorded seizures during video-EEG monitoring. Although interictal EEG was normal, there were five seizures with left temporal EEG seizure onset recorded. He had one cousin with epilepsy. At the time of evaluation, he was taking 500 mg of phenytoin and 400 mg of topiramate.

He obtained a high school diploma, and although he was never diagnosed with a specific learning disability, he participated in special educational classes beginning in middle school. As seen in Table 20.2, preoperative neuropsychological assessment revealed average general functioning, with a slight advantage for nonverbal compared to verbal tasks. His neuropsychological profile suggested mild cognitive difficulty consistent with left TLE, including poor naming, decreased generative fluency to letter prompts, and poorer verbal than visual memory reflected by an 11-point discrepancy on the Wechsler Memory Scale-III. Verbal list learning and recall were in the impaired range. (Poor generative letter fluency often occurs in patients with left TLE if there is disruption of frontal lobe regions from seizure spread, but this deficit could also reflect cognitive side effects of topiramate.)

Table 20.2 Known predictors of neuropsychological outcome in MRI-negative surgical cases

Wada memory results indicated no undue risk of decline following left ATL. Language was lateralized to the left hemisphere and memory score asymmetries suggested that the left TL (right hemisphere injection) was functioning at a lower level (Wada recognition = 5/10) than the right TL (left hemisphere injection, Wada recognition = 9/10). Risk factors associated with cognitive decline include later age of seizure onset in his mid-20s and normal MRI.

The patient experienced significant postoperative declines following left ATL in both naming and verbal memory, with a smaller decline noted in his visual memory performance (this decline was limited to a single test that required naming ability to perform). Although list learning ability was in the impaired range preoperatively and estimated to be at or below the 1st percentile, the 18-point raw score decline reflects meaningful postsurgical change. In addition, although the patient could still recognize familiar and famous persons, he experienced a severe naming impairment for these individuals. Likewise, he was severely impaired at generating any proper nouns. For example, he could not produce the names of stores, TV shows, famous persons or places when prompted. He was reportedly seizure-free, but was still experiencing auras. In contrast to the cognitive declines stated above, the patient experienced an improvement in generative fluency to letter prompts. As he was still on the same AED regimen, this improvement is thought to be due to an absence of seizure impact upon frontal lobe regions [45].

CASE 2: Right TLE; MRI-negative

The patient is a right-handed male in his 40s and presents with an approximately 5-year history of epilepsy. His seizures rapidly generalized but these were initially controlled with medication. At the time of evaluation, he was experiencing two to three seizures per year despite optimal medical management. During his video-EEG monitoring, there were three events indicating right anterior temporal seizure onset. He was taking 700 mg of carbamazepine.

The patient was a college graduate and had worked at a high-level job that required seizure freedom for performance. At the time of surgery, he was working in a management role for the same company but desired to return to his prior job. Preoperative neuropsychological evaluation demonstrated normal naming, normal verbal learning and memory, but a large 26-point discrepancy in favor of verbal memory on the Wechsler Memory Scale delayed indices, and his visual memory performance was in the low end of the average range.

Wada results indicated that the patient was left hemisphere dominant for language and that memory functioning was normal for both temporal lobes (Wada recognition: left injection=9/10; right injection = 10/10). The patient was considered to be at increased risk for postoperative memory decline due to his intact baseline cognitive functioning, normal Wada memory scores bilaterally, late age of seizure onset, and normal MRI. However, because the surgery planned was a nondominant right ATL, verbal memory and language were not at risk. The patient chose to go ahead with surgery, in the hopes that he could return to his prior level of employment.

Following right anterior temporal lobectomy, he was seizure-free, but he experienced large declines in his visual memory (32 points) with additional deficits involving object recognition and attention. Although list learning did not change, the patient experienced important declines in working memory and generative verbal fluencies. He also experienced decreased ability to recognize familiar or famous individuals and landmarks, and selective declines in his ability to recognize animals (i.e., he was much worse at recognizing some categories of animals, such as birds). For example, when presented with the picture of a pink flamingo standing on one leg, he responded, “that’s a turkey…the bird we eat at Thanksgiving.” He was also unable to perform a spatial memory task postsurgically. He described an inability to recognize familiar persons, difficulty with route learning, and increased difficulty to correctly associate names with faces. These deficits compromised his ability to work in sales since surgery, and also contributed to new problems with mood. He never returned to his previous level of employment, and his wife divorced him after several years of marriage.

Neuropsychological domains of research

Future neuropsychological research should include prospective evaluation of MRI-negative epilepsy patients, with the goals of systematically employing appropriate cognitive measures, directly relating these measures to functional connectivity and neuroimaging results, and exploring algorithms for outcome prediction following surgery. Such research should also examine broader neurocognitive domains of function, as emerging evidence suggests that important ability areas have not been adequately investigated. Extratemporal lobe epilepsies have not been well studied with or without regards to lesional status. Even in the better-studied TLE population, many important functions have not been adequately evaluated, such as object recognition and category-related naming and semantic fluencies, semantic memory, face-name learning, and aspects of emotional and social/theory of mind processing. It will also be important to study the differential impact upon cognition and function of the various treatment interventions available for both nonlesional and lesional cases (e.g., ablations using radiofrequency or ultrasound, selective laser amygdalohippocampotomy, gamma knife).