INTRODUCTION
Neuropsychology and epilepsy have enjoyed a long and mutually beneficial relationship (Loring, Reference Loring2010; Novelly, Reference Novelly1992). The epilepsies have provided a window to advance understanding of brain function, producing insights into the effects of focal epilepsies and their surgical intervention, leading to a greater understanding of the brain mechanisms underlying a wide variety of cognitive and behavioral constructs (Elger et al., Reference Elger, Helmstaedter and Kurthen2004; Jones-Gotman et al., Reference Jones-Gotman, Smith, Risse, Westerveld, Swanson, Giovagnoli and Piazzini2010). Neuropsychology has also played an important clinical role by characterizing the impact of epilepsy through its relationship with factors such as age of onset of epilepsy, etiology, seizure type and syndrome, medications, duration of epilepsy, and electroencephalographic (EEG) abnormalities (Dodrill & Matthews, Reference Dodrill and Matthews1992; Folsom, Reference Folsom1952; Helmstaedter & Witt, Reference Helmstaedter and Witt2012; Tarter, Reference Tarter1972; Trimble & Thompson, Reference Trimble and Thompson1986; Vermeulen & Aldenkamp, Reference Vermeulen and Aldenkamp1995).
In this review, we focus on important advances that have occurred in the field of neuropsychology and epilepsy from the time of the formation of the International Neuropsychological Society in 1967 to the present, or some 50 years. The advances to be described include a brief synopsis of the state of the field antecedent to these developments and their impact going forward, along with examples of key studies that changed the paradigms of the time. We conclude with a brief discussion of what we think the future holds for clinical and research directions in the neuropsychology of epilepsy.
NEUROBIOLOGY OF COGNITIVE DISORDERS IN EPILEPSY: EXPANDING THE LANDSCAPE OF SYNDROME-SPECIFIC NEUROPSYCHOLOGICAL IMPAIRMENT
The characterization and nosology of seizures and epilepsy syndromes has evolved through the decades, with initiation of formal efforts to construct an internationally agreed classification system in the 1960s, with ongoing modifications based on new knowledge (Berg et al., Reference Berg, Berkovic, Brodie, Buchhalter, Cross, van Emde Boas and Scheffer2010; Scheffer et al., Reference Scheffer, Berkovic, Capovilla, Connolly, French, Guilhoto and Zuberi2017; Wolf, Reference Wolf2009). Focal or localization-related epilepsies are commonly characterized in terms of their lobar site of origin, and neuropsychology has historically focused on structure–function relationships such as executive function in frontal lobe epilepsy (FLE) and memory in temporal lobe epilepsy (TLE).
The critical distinctions between declarative and procedural memory systems were demonstrated by Milner through careful study of patient H.M. (Milner, Reference Milner1965), with the later development of the material-specific model of anterograde memory (Milner, Reference Milner1972). This model posited that, if seizure onset originated from the language dominant temporal lobe, then verbal learning and memory would be adversely affected. Although somewhat less robust, if seizure onset occurred in the nondominant temporal lobe, learning and memory for non-verbal material such as designs or faces would be affected with largely spared verbal memory (Barr, Reference Barr1997). Other cognitive abilities were expected to be relatively unaffected because both seizure onset and focal epileptiform abnormalities were restricted to the temporal lobe structures that encoded new memories.
Studies began to accumulate, however, demonstrating that both cognitive and brain abnormalities associated with the focal epilepsies could extend beyond the seizure onset zone. Although memory impairments are a common feature of TLE related to hippocampal neuronal loss and sclerosis, they occurred against the backdrop of more distributed cognitive abnormality affecting not only memory, but also IQ, executive functions, language, sensorimotor, and other abilities (e.g., Agah et al., Reference Agah, Asgari-Rad, Ahmadi, Tafakhori and Aghamollaii2017; Baxendale & Thompson, Reference Baxendale and Thompson2010; Glosser et al., Reference Glosser, Cole, French, Saykin and Sperling1997; Guimaraes et al., Reference Guimaraes, Li, Rzezak, Fuentes, Franzon, Augusta Montenegro and Guerreiro2007; Oyegbile, Dow, et al., Reference Oyegbile, Hansen, Magnotta, O’Leary, Bell, Seidenberg and Hermann2004; Rzezak et al., Reference Rzezak, Fuentes, Guimarães, Thome-Souza, Kuczynski, Li and Valente2007; Stretton & Thompson, Reference Stretton and Thompson2012; Wang et al., Reference Wang, Liou, Chen, Chiu, Chen, Cheng and Hua2009).
Similarly, genetic generalized epilepsies (GGE) were thought to have neuropsychological deficits restricted to attention and executive function, but a recent meta-analysis of patients with GGE, including juvenile myoclonic epilepsy, demonstrated impairment in all cognitive domains except for visuospatial abilities and, of note, without disproportionate impairment of executive function (Loughman, Bowden, & D’Souza, Reference Loughman, Bowden and D’Souza2014). Likewise, neuropsychological impairments may be broader than anticipated in FLE and related to distributed brain anomalies (Braakman et al., Reference Braakman, Ijff, Vaessen, Debeij-van Hall, Hofman, Backes and Aldenkamp2012, Reference Braakman, Vaessen, Jansen, Debeij-van Hall, de Louw, Hofman and Backes2014, Reference Braakman, Vaessen, Jansen, Debeij-van Hall, de Louw, Hofman and Backes2015), with distinctions between focal epilepsy syndromes in children with epilepsy (e.g., temporal versus frontal) often less crisp than believed, this due to several factors including the impact of epilepsy on neurodevelopmental processes (Smith, Reference Smith2016).
Neuroimaging studies have documented a similar pattern of distributed abnormalities in brain networks associated with focal epilepsies. For example, distributed morphometric abnormalities in “normal” clinical MRIs began to be reported in surgical patients with TLE with hippocampal sclerosis which were found to be associated not only with poorer surgical outcome (Sisodiya et al., Reference Sisodiya, Moran, Free, Kitchen, Stevens, Harkness and Shorvon1997) but also with adverse cognitive (memory) outcomes (Baxendale et al., Reference Baxendale, Sisodiya, Thompson, Free, Kitchen, Stevens and Shorvon1999). Neuroimaging research using diverse techniques documented the range of abnormalities extending beyond the epileptogenic temporal lobe to ipsilateral extra-temporal regions, the contralateral hemisphere, and subcortical structures and cerebellum (Keller & Roberts, Reference Keller and Roberts2008).
These abnormalities included diffusely distributed cortical thinning (Keller et al., Reference Keller, Richardson, O’Muircheartaigh, Schoene-Bake, Elger and Weber2015; Lin et al., Reference Lin, Salamon, Lee, Dutton, Geaga, Hayashi and Thompson2007; McDonald et al., Reference McDonald, Hagler, Ahmadi, Tecoma, Iragui, Gharapetian and Halgren2008), other bilateral anomalies in cortical surface features (Oyegbile, Magnotta, et al., Reference Oyegbile, Dow, Jones, Bell, Rutecki, Sheth and Hermann2004), as well as distributed abnormalities in white matter volumes and connectivity (Alhusaini et al., Reference Alhusaini, Doherty, Scanlon, Ronan, Maguire, Borgulya and Cavalleri2012; Hermann et al., Reference Hermann, Seidenberg, Bell, Rutecki, Sheth, Wendt and Magnotta2003; Mueller et al., Reference Mueller, Laxer, Cashdollar, Buckley, Paul and Weiner2006; Otte et al., Reference Otte, van Eijsden, Sander, Duncan, Dijkhuizen and Braun2012; Ronan et al., Reference Ronan, Murphy, Delanty, Doherty, Maguire, Scanlon and Fitzsimons2007; Seidenberg et al., Reference Seidenberg, Kelly, Parrish, Geary, Dow, Rutecki and Hermann2005; Slinger, Sinke, Braun, & Otte, Reference Slinger, Sinke, Braun and Otte2016), these effects often more pronounced in patients with the syndrome of mesial temporal lobe epilepsy. Postmortem neuropathological investigations of TLE patients are pertinent and consistent in this regard as well (Blanc et al., Reference Blanc, Martinian, Liagkouras, Catarino, Sisodiya and Thom2011; Margerison & Corsellis, Reference Margerison and Corsellis1966; Sinjab, Martinian, Sisodiya, & Thom, Reference Sinjab, Martinian, Sisodiya and Thom2013).
While early structure-function investigations in TLE focused on relationships between hippocampal atrophy and memory (Lencz et al., Reference Lencz, McCarthy, Bronen, Scott, Inserni, Sass and Spencer1992; Trenerry et al., Reference Trenerry, Jack, Ivnik, Sharbrough, Cascino, Hirschorn and Meyer1993), subsequent studies reported a wider range of brain-behavior links such as associations between frontal abnormality with executive function (Keller, Baker, Downes, & Roberts, Reference Keller, Baker, Downes and Roberts2009), the numerous studies of this type cumulatively informing an evolving neurobiological architecture of cognitive impairment in TLE (Hermann, Lin, Jones, & Seidenberg, Reference Hermann, Lin, Jones and Seidenberg2009).
This paradigm shift recognized that the landscape of cognitive abnormality within a localization related epilepsy was not driven solely by the location of the epilepsy, but by the nature, range and severity of underlying distributed neurobiological abnormalities (Bell, Lin, Seidenberg, & Hermann, Reference Bell, Lin, Seidenberg and Hermann2011), the etiology and course of which continues to intensely investigated (Rayner, Jackson, & Wilson, Reference Rayner, Jackson and Wilson2016a). Cognitive abilities depend on distributed neural networks and the degree to which these networks are characterized and how they are impacted by epilepsy will be a major path of investigation going forward (McDonald et al., Reference McDonald, Leyden, Hagler, Kucukboyaci, Kemmotsu, Tecoma and Iragui2014).
PATHWAYS TO COMORBIDITIES: BIDIRECTIONAL RELATIONSHIPS AND THEIR CLINICAL IMPLICATIONS
Epilepsy patients carry a significantly elevated burden of medical/somatic, psychiatric, and social disorders in addition to cognitive difficulties (Gaitatzis, Sisodiya, & Sander, Reference Gaitatzis, Sisodiya and Sander2012; Keezer, Sisodiya, & Sander, Reference Keezer, Sisodiya and Sander2016; Tellez-Zenteno, Matijevic, & Wiebe, Reference Tellez-Zenteno, Matijevic and Wiebe2005; Tellez-Zenteno, Patten, Jette, Williams, & Wiebe, Reference Tellez-Zenteno, Patten, Jette, Williams and Wiebe2007). Although the co-occurrence of these disorders with epilepsy is well documented, their causal pathways including temporal sequence are often not clear (Keezer et al., Reference Keezer, Sisodiya and Sander2016). A common view has been that epilepsy including recurrent seizures, persistent interictal EEG abnormalities, treatment effects, and psychosocial factors all contribute to the development and progression of comorbid cognitive and psychiatric conditions after initial seizure onset.
A major disruption to this paradigm was the demonstration of a bidirectional relationship between epilepsy and cognitive, psychiatric, and social comorbidities in both pediatric and adult epilepsies (Forsgren & Nystrom, Reference Forsgren and Nystrom1990; Hesdorffer, Reference Hesdorffer2016; Hesdorffer, Hauser, Annegers, & Cascino, Reference Hesdorffer, Hauser, Annegers and Cascino2000; Keezer et al., Reference Keezer, Sisodiya and Sander2016). It began to be appreciated that psychiatric comorbidities frequently appeared before seizure onset and could represent a risk factor for subsequent epilepsy, or both could be due to a common underlying mechanism. This bidirectional relationship exists for multiple comorbid conditions including depression, suicidality, autism spectrum disorder, attention deficit hyperactivity disorder (ADHD), psychosis, and schizophrenia (Lin, Mula, & Hermann, 2012). Much remains to be learned about the potential mechanisms underlying bidirectional relationships. Kanner (2012) suggested that pertinent to the epilepsy-depression relationship may be endocrine abnormalities, structural and functional abnormalities of cortical and subcortical structures, neurotransmitter abnormalities, and/or immunological abnormalities.
Of particular interest is co-aggregation of cognitive, behavioral, and even brain structural abnormalities in the nonaffected siblings of persons with epilepsy (Alhusaini et al., Reference Alhusaini, Whelan, Doherty, Delanty, Fitzsimons and Cavalleri2016; Aronu & Iloeje, Reference Aronu and Iloeje2011; Badawy, Vogrin, Lai, & Cook, Reference Badawy, Vogrin, Lai and Cook2013; Chowdhury et al., Reference Chowdhury, Elwes, Koutroumanidis, Morris, Nashef and Richardson2014; Hesdorffer, Caplan, & Berg, Reference Hesdorffer, Caplan and Berg2012; Iqbal et al., Reference Iqbal, Caswell, Muir, Cadden, Ferguson, Mackenzie and Duncan2015, Reference Iqbal, Caswell, Hare, Pilkington, Mercer and Duncan2009; Keller et al., Reference Keller, O’Muircheartaigh, Traynor, Towgood, Barker and Richardson2014; Levav et al., Reference Levav, Mirsky, Herault, Xiong, Amir and Andermann2002; Singhi, Bansal, Singhi, & Pershad, Reference Singhi, Bansal, Singhi and Pershad1992; Smith et al., Reference Smith, Kavros, Clarke, Dorta, Tremont and Pal2012; Tsai et al., Reference Tsai, Pardoe, Perchyonok, Fitt, Scheffer, Jackson and Berkovic2013; Verrotti et al., Reference Verrotti, Matricardi, Di Giacomo, Rapino, Chiarelli and Coppola2013; Wandschneider et al., Reference Wandschneider, Centeno, Vollmar, Symms, Thompson, Duncan and Koepp2014, Reference Wandschneider, Kopp, Kliegel, Stephani, Kurlemann, Janz and Schmitz2010). These reports suggest the presence of distinct endophenotypes, raising the possibility of shared genetic mechanisms, although the influence of environmental and other factors remain to be clarified.
Bidirectionality and conceptualizing epilepsy as a disease of brain networks raised the possibility that behavioral comorbidities may be present at the time of diagnosis. Indeed, adults and elders with new onset epilepsies display a pattern of mild diffuse cognitive impairment at diagnosis and before treatment initiation (Taylor et al., Reference Taylor, Kolamunnage-Dona, Marson, Smith, Aldenkamp and Baker2010; Witt & Helmstaedter, Reference Witt and Helmstaedter2012; Witt et al., Reference Witt, Werhahn, Kramer, Ruckes, Trinka and Helmstaedter2014). These patients have abnormal 1-year and 5-year cognitive courses that predominantly affect memory, executive function, and psychomotor speed (Baker, Taylor, Aldenkamp, & SANAD group, Reference Baker, Taylor and Aldenkamp2011; Taylor & Baker, Reference Taylor and Baker2010) (see Witt and Helmstaedter Reference Witt and Helmstaedter2015 for review).
In children with new onset epilepsies, elevated are rates of behavioral problems including ADHD, depression and anxiety, early life histories of academic struggles and provision of supportive academic services, and a pattern of distributed cognitive anomalies that may even precede the first recognized seizure (Austin et al., Reference Austin, Harezlak, Dunn, Huster, Rose and Ambrosius2001; Berg et al., Reference Berg, Smith, Frobish, Levy, Testa, Beckerman and Shinnar2005; Fastenau et al., Reference Fastenau, Johnson, Perkins, Byars, deGrauw, Austin and Dunn2009; Hoare, Reference Hoare1984; Oostrom, Smeets-Schouten, Kruitwagen, Peters, & Jennekens-Schinkel, Reference Oostrom, Smeets-Schouten, Kruitwagen, Peters and Jennekens-Schinkel2003; Overvliet, Aldenkamp, Klinkenberg, Vles, & Hendriksen, Reference Overvliet, Aldenkamp, Klinkenberg, Vles and Hendriksen2011; Pohlmann-Eden et al., Reference Pohlmann-Eden, Aldenkamp, Baker, Brandt, Cendes, Coras and Hermann2015). These findings demonstrate the need for neuropsychology to be involved in the routine clinical care of epilepsy patients, screening new onset cases for pertinent comorbidities to initiate timely treatment, and following patients to monitor problematic trajectories (Wilson et al., Reference Wilson, Baxendale, Barr, Hamed, Langfitt, Samson and Smith2015).
DISCOVERING QUALITY OF LIFE: THE CONCEPT, ITS QUANTIFICATION, AND APPLICABILITY
Quality of life (QoL), reflected in the general wellbeing of individuals and societies, represents a core paradigm shift in the quality of care of people with epilepsy. This came about in stages. Before the availability of formal health-related quality of life (HRQOL) measures, patient status was characterized by traditional clinical interview or the use of tests that had been developed for much broader application (e.g., personality and IQ assessment). Thus, “QoL” was characterized by a mix of diverse measures representing a battery-type approach to characterize disease, treatment effects, and patient status.
Disease-specific measures were later developed to characterize unique issues influencing life with epilepsy. The most influential early epilepsy-specific measure was the Washington Psychosocial Seizure Inventory (WPSI) developed by Carl Dodrill and colleagues at the University of Washington, which assessed family background, emotional adjustment, interpersonal adjustment, vocational adjustment, financial status, adjustment to seizures, medicine and medical management, with a summary evaluation of overall psychosocial functioning (Dodrill, Batzel, Queisser, & Temkin, Reference Dodrill, Batzel, Queisser and Temkin1980). The WPSI was translated and used internationally, contributing to a “common language” for characterizing psychosocial complications.
The most recent stage has been the adoption of formal HRQOL logic and methods. This approach uses a general measure (e.g., the Rand 36-Item Health Survey, aka SF-36) for comparison with other conditions, supplemented by epilepsy specific items. Work in epilepsy surgery was again prominent in leading the way, with Vickrey and colleagues first developing the Epilepsy Surgery Inventory-55 (Vickrey et al., Reference Vickrey, Hays, Rausch, Engel, Visscher, Ary and Brook1995). This mixed generic and disease-specific instrument was expanded into the Quality of Life in Epilepsy Inventory-89 items (QOLIE-89) (Devinsky et al., Reference Devinsky, Vickrey, Cramer, Perrine, Hermann, Meador and Hays1995) to be applicable to settings outside of epilepsy surgery with less severely affected patients.
While initial studies using HRQOL measures demonstrated that improvements after surgery were linearly related to a reduction in seizure frequency (Vickrey et al., Reference Vickrey, Hays, Rausch, Engel, Visscher, Ary and Brook1995), the importance of other predictors was soon recognized, including mood, cognition, employment, driving, and antiepileptic drug (AED) cessation (for a review, see Wilson and Engel, Reference Wilson and Engel2010). Gilliam et al. (Reference Gilliam, Kuzniecky, Meador, Martin, Sawrie, Viikinsalo and Faught1999) reported a striking negative linear relationship between HRQOL and symptoms of depression; and Kanner, Barry, Gilliam, Hermann, and Meador (Reference Kanner, Barry, Gilliam, Hermann and Meador2010) subsequently noted that co-morbid anxiety and depression had the greatest impact on HRQOL. Also very influential was the “Liverpool school of HRQOL” (Baker, Reference Baker1998; Baker, Jacoby, Buck, Stalgis, & Monnet, Reference Baker, Jacoby, Buck, Stalgis and Monnet1997; Baker, Smith, Dewey, Jacoby, & Chadwick, Reference Baker, Smith, Dewey, Jacoby and Chadwick1993; Jacoby, Snape, & Baker, Reference Jacoby, Snape and Baker2009), demonstrating that, although seizure frequency was a common outcome measure in clinical trials and surgical series, seizure severity had been ignored. Seizure severity, but not seizure frequency, was related to self-esteem, locus of control, and anxiety.
While HRQOL research has principally focused on improvements in well-being associated with epilepsy treatments, the Melbourne group demonstrated that the diagnosis of a first seizure gives rise to a process of losing and restoring perceived control, characterized by two psychological adjustment trajectories that hinged on the experience of a limited or pervasive loss of control, leading to different medical and HRQOL outcomes (Velissaris, Saling, Newton, Berkovic, & Wilson, Reference Velissaris, Saling, Newton, Berkovic and Wilson2012). The latter trajectory involves a more extensive process of personal re-evaluation and psychosocial change, akin to what occurs after trauma (Velissaris, Wilson, Saling, Newton, & Berkovic, Reference Velissaris, Wilson, Saling, Newton and Berkovic2007).
OUTCOMES OF EPILEPSY SURGERY: CHALLENGING CONVENTIONAL WISDOM
The neuropsychology of epilepsy surgery has its roots in the early pioneering programs in Montreal (Penfield and Milner), London (Falconer and Meyer), and Chicago (Bailey, Gibbs, and Halstead). Epilepsy surgery was initially considered a final treatment option following many years of failed medication management and devastating psychosocial complications. The 1980s witnessed the beginning of an explosion of interest in epilepsy surgery during which time two influential international symposia occurred, the Palm Desert Workshops on Epilepsy Surgery held in 1986 and 1992 (Engel, Reference Engel1987, Reference Engel1993). These were transformative in fueling critical thinking about patient evaluation and selection, surgical treatments, outcome assessments, and surgery as a mainstream option to consider before decades of failed medical therapy.
Cognitive Outcomes
A primary concern regarding anterior temporal lobectomy (ATL) is the risk of significant post-operative memory decline/amnesia (Milner, Reference Milner1958), and neuropsychological practice has traditionally addressed three issues: (1) interest in the congruence of neuropsychological deficits with the area of ictal onset, (2) reliance on the material-specific model of memory function to assess whether the contralateral temporal lobe had sufficient functional capacity to sustain memory function post-operatively, and (3) assessment of post-operative memory risk using the intracarotid amobarbital (aka Wada) test.
Three developments altered this classic paradigm. First, some surgeons excised hippocampus en bloc, which facilitated quantitative assessment of neuronal loss and gliosis in hippocampal subfields which could then be related to preoperative memory and Wada test performance (O’Rourke et al., Reference O’Rourke, Saykin, Gilhool, Harley, O’Connor and Sperling1993; Rausch & Babb, Reference Rausch and Babb1987; Sass et al., Reference Sass, Lencz, Westerveld, Novelly, Spencer and Kim1991, Reference Sass, Sass, Westerveld, Lencz, Novelly, Kim and Spencer1992, Reference Sass, Spencer, Kim, Westerveld, Novelly and Lencz1990). Second, advances in MR imaging allowed noninvasive quantification of hippocampal volumes which could then be correlated with preoperative memory results (Lencz et al., Reference Lencz, McCarthy, Bronen, Scott, Inserni, Sass and Spencer1992; Trenerry et al., Reference Trenerry, Jack, Ivnik, Sharbrough, Cascino, Hirschorn and Meyer1993). MRI also allowed examination of hippocampal-memory relationships both ipsilateral and contralateral to the side of eventual surgery. Third, a logical extension of this work was examination of postoperative memory outcomes as a function of markers of ipsilateral hippocampal integrity reflected in cell counts, quantitative MRI, or adequacy of preoperative memory function. Resection of a less diseased/more functional hippocampus was found to be a primary driver of memory decline (Hermann, Wyler, Somes, Berry, & Dohan, Reference Hermann, Wyler, Somes, Berry and Dohan1992; Rausch and Babb, Reference Rausch and Babb1993; Sass et al., Reference Sass, Westerveld, Buchanan, Spencer, Kim and Spencer1994), findings that continue to be reported and elaborated upon (Coras et al., Reference Coras, Pauli, Li, Schwarz, Rossler, Buchfelder and Blumcke2014; Witt et al., Reference Witt, Coras, Schramm, Becker, Elger, Blumcke and Helmstaedter2015).
This changing paradigm was crystallized by Chelune (Reference Chelune1995) in which he suggested that the functional integrity of the to-be-resected hippocampus, as reflected by less hippocampal atrophy, higher hippocampal cell counts, or better preoperative memory performance, was a primary mediator of pre- to postoperative memory change, rather than the functional reserve of the contralateral temporal lobe as initially proposed (Milner, Branch, & Rassmussen, Reference Milner, Branch and Rassmussen1962). This reformulation emphasizing ipsilateral rather than contralateral hippocampal function has influenced assessment techniques including preoperative memory assessment via functional MRI (fMRI), in which a strong predictor of memory decline following ATL is ipsilateral posterior hippocampus activation (Bonelli et al., Reference Bonelli, Powell, Yogarajah, Samson, Symms, Thompson and Duncan2010).
Relatedly, the material-specific memory model has been unable to account for multiple failures to link right hippocampal resection to “non-verbal” or “visual memory” declines as well as reports of mildly impaired story recall in patients with left or right mesial temporal foci (Saling et al., Reference Saling, Berkovic, O’Shea, Kalnins, Darby and Bladin1993). In a major contribution and arguably paradigm shifting proposal, Saling overviewed the historical development of the material-specific model of memory, the findings inconsistent with the model, and the logic leading to his major reformulation (Saling, Reference Saling2009) that task-specificity, rather than material-specificity, is more relevant in predicting verbal memory outcomes as it reflects the intra-temporal organization of memory. In particular, Saling noted that arbitrary relational forms of learning (i.e., unrelated word pairs) appear mediated by specific potions of the left medial temporal lobe (perirhinal cortex) which can be distinguished from semantically rich forms of verbal learning (i.e., semantically loaded associations) mediated by the lateral temporal cortex (anterior inferior temporal gyrus), with potential corresponding similarities in task specific findings in the right medial and lateral temporal lobes.
Lastly, important has been the shift from characterizing cognitive outcome following surgery in groups to characterizing (and predicting) outcome at the individual level (e.g., development of reliable change indices and standardized regression-based measures). In addition to providing a more rigorous approach to assess meaningful cognitive change attributable to surgery, these methods highlighted the variability in cognitive performance observed in individuals with epilepsy over time (with or without intervening surgery) (Martin et al., Reference Martin, Sawrie, Roth, Gilliam, Faught, Morawetz and Kuzniecky1998; Sawrie et al., Reference Sawrie, Chelune, Naugle and Lüders1996; Sherman et al., Reference Sherman, Wiebe, Fay-McClymont, Tellez-Zenteno, Metcalfe, Hernandez-Ronquillo and Jetté2011).
From Anterograde to Semantic Memory Systems
Also changing over this time was a greater appreciation of the impact of focal epilepsy, particularly left temporal lobe epilepsy, on semantic memory (Bell et al., Reference Bell, Hermann, Woodard, Jones, Rutecki, Sheth and Seidenberg2001; Seidenberg et al., Reference Seidenberg, Griffith, Sabsevitz, Moran, Haltiner, Bell and Hermann2002; Tippett, Glosser, & Farah, Reference Tippett, Glosser and Farah1996). Multiple reports suggested a robust relationship between decreased confrontation naming ability and left temporal seizure onset, which oftentimes exceeded the relationship between left TLE and verbal learning and memory (Busch, Frazier, Iampietro, Chapin, & Kubu, Reference Busch, Frazier, Iampietro, Chapin and Kubu2009). More than 40% of left ATL patients compared to only 5% of right ATL patients exhibit naming decline postoperatively (Busch et al., Reference Busch, Floden, Prayson, Chapin, Kim, Ferguson and Najm2016).
TLE patients are also impaired at naming famous faces (Benke, Kuen, Schwarz, & Walser, Reference Benke, Kuen, Schwarz and Walser2013; Drane et al., Reference Drane, Ojemann, Phatak, Loring, Gross, Hebb and Tranel2013) and, furthermore, the naming deficits can be category dependent (Tippett et al., Reference Tippett, Glosser and Farah1996). Interestingly, left TLE patients that display impaired confrontation naming are generally able to recognize famous faces and objects (Drane et al., Reference Drane, Ojemann, Aylward, Ojemann, Johnson, Silbergeld and Tranel2008), whereas right TLE can be associated with recognition impairment that is thought to result from impaired processing in the ventral visual stream, reflecting a mild visual agnosia. This suggests that assessment of visual naming fails to capture the full impact of epilepsy and epilepsy surgery on the semantic memory network.
Plasticity of the Epileptic Brain
The gold standard for establishing cerebral language dominance before epilepsy surgery has been the Wada test, which pharmacologically inactivates the region of the internal carotid artery while the patient performs multiple cognitive tasks including language and memory. This technique not only was critical for identification of language before surgery, in which language mapping or less extensive resection would typically be performed for language dominant resections, but was critical in establishing the relationship between handedness and cerebral dominance, including pathological left handedness in which early left brain lesions cause a shift in language and dominance from the predisposed left hemisphere to the right hemisphere (Rasmussen & Milner, Reference Rasmussen and Milner1977).
However, as Wada techniques became increasingly refined, they also demonstrated that exclusive right hemisphere representation was rare and that language reorganization typically existed on a continuum (Loring et al., Reference Loring, Meador, Lee, Murro, Smith, Flanigin and King1990) which has now been well-described in the fMRI literature (Abbott, Waites, Lillywhite, & Jackson, Reference Abbott, Waites, Lillywhite and Jackson2010; Binder et al., Reference Binder, Swanson, Hammeke, Morris, Mueller, Fischer and Haughton1996). When a shift in the biological predisposition for left cerebral language dominance occurs, it is often associated with age of injury, region of onset, and magnitude of injury (Stewart et al., Reference Stewart, Swanson, Sabsevitz, Rozman, Janecek and Binder2014). A shift of language dominance, however, may occur in the absence of a shift of verbal learning and memory (Loring et al., Reference Loring, Meador, Lee, Murro, Smith, Flanigin and King1990; Wood, Saling, O’Shea, Jackson, & Berkovic, Reference Wood, Saling, O’Shea, Jackson and Berkovic1999).
Several activation paradigms reliably identify language non-invasively (Binder et al., Reference Binder, Swanson, Hammeke, Morris, Mueller, Fischer and Haughton1996; Gaillard et al., Reference Gaillard, Balsamo, Xu, McKinney, Papero, Weinstein and Theodore2004). Reliable fMRI identification of language laterality has substantially altered the practice of the preoperative neuropsychological evaluation for epilepsy by decreasing the need for evaluation of language with the Wada test (Baxendale, Reference Baxendale2009; Baxendale, Thompson, & Duncan, Reference Baxendale, Thompson and Duncan2008). What has not yet been satisfactorily developed, however, is a valid fMRI memory task to replace the need for Wada memory testing in selected patients, but important is evidence indicating that patterns of language activation predict the risk of postsurgical verbal memory outcomes (see Szaflarski et al., Reference Szaflarski, Gloss, Binder, Gaillard, Golby, Holland and Theodore2017; Binder et al., Reference Binder, Swanson, Sabsevitz, Hammeke, Raghavan and Mueller2010).
Life Outcomes
A goal of epilepsy surgery has always been to improve the postoperative life of people with epilepsy (Taylor, Reference Taylor1993; Jones, Blocher, & Jackson, Reference Jones, Blocher and Jackson2013). In fact, in the early days of the University of Illinois surgical program, a pioneering electroencephalographer (Frederic Gibbs) convinced a reluctant neurosurgeon (Percival Bailey) to operate solely on the basis of EEG since, without surgery, the patients’ compromised quality of life would persist (Hermann & Stone, Reference Hermann and Stone1989). However, it may be a somewhat naïve assumption to believe that a patient with long-standing compromised cognitive, psychiatric, or social status will suddenly improve their life situation following successful surgery without broader rehabilitation efforts.
The Melbourne group identified a complex set of psychological and social changes that need to occur after surgery for a patient to maximally benefit from seizure freedom (Bladin, Reference Bladin1992; Wilson, Bladin, & Saling, Reference Wilson, Bladin and Saling2001). This process, driven by a change in patient identity as they learn to “become well,” gives rise to a set of psychological and social features that constitute a syndrome termed the “burden of normality.” Although seizure freedom may be necessary for improved HRQOL after surgery, it is not sufficient, with patients and families also requiring post-operative rehabilitation and psychosocial support as they adapt to the new challenges of a seizure free lifestyle. This post-operative support is also critical for approximately one third of patients who experience the significant comorbidity of depression after epilepsy surgery, including suicidality and de novo presentations. A range of factors, including a pre-operative psychiatric history and smaller contralateral hippocampal volume, have been identified as risk factors (Wrench, Matsumoto, Inoue, & Wilson, Reference Wrench, Matsumoto, Inoue and Wilson2011).
IATROGENIC EFFECTS OF TREATMENT: COGNITIVE AND BEHAVIORAL EFFECTS OF ANTIEPILEPSY DRUGS
Because most epilepsy drugs have comparable efficacy in controlling seizures for appropriate epilepsy indications, differential neuropsychological effects are now commonly used by prescribing physicians to inform initial AED selection. The first systematic characterization of different AED profiles was by Dodrill and Troupin (Reference Dodrill and Troupin1977), although this study was confounded by trial design issues since comparable drug levels were not contrasted.
The most influential paper reporting cognitive AED effects was the demonstration that phenobarbital used for seizure prophylaxis after a febrile seizure was associated with FSIQs that were approximately ½ SD lower than placebo treated children (Farwell et al., Reference Farwell, Lee, Hirtz, Sulzbacher, Ellenberg and Nelson1990). These differences persisted following phenobarbital discontinuation indicating cumulative treatment effects from which the children were unable to catch up (Sulzbacher, Farwell, Temkin, Lu, & Hirtz, Reference Sulzbacher, Farwell, Temkin, Lu and Hirtz1999). These studies, combined with the absence of benefit in decreasing subsequent seizures, contributed to current practice in which AED prophylaxis is not routine, and phenobarbital is generally avoided as a first-line epilepsy treatment.
One criticism of neuropsychological AED studies is that group differences may be difficult to apply on an individual patient basis, and studies are typically not dose-ranging. With increasing application of reliable change indices, however, epidemiological statistics such as the number needed to harm (NNH) will likely better characterize risks of cognitive impairment at different levels of AED doses (Loring, Williamson, Meador, Wiegand, & Hulihan, Reference Loring, Williamson, Meador, Wiegand and Hulihan2011). Several reviews of the cognitive side effects of AEDs have been published, although information on the more recently introduced medications are not included (Loring et al., Reference Loring, Meador, Lee, Murro, Smith, Flanigin and King1990; Vermeulen & Aldenkamp, Reference Vermeulen and Aldenkamp1995).
Even when neuropsychological outcomes are not the primary endpoint in clinical trials, they form a component of overall treatment “effectiveness” beyond treatment “efficacy.” AED differences on the Continuous Performance Task contributed to specific initial treatment recommendations for ethosuximide over valproate in treating childhood absence epilepsy since both medicines were equally effective in treating seizures (Glauser et al., Reference Glauser, Cnaan, Shinnar, Hirtz, Dlugos and Masur2010). Neuropsychology has also demonstrated that children exposed to valproate during their mothers’ pregnancy have lower IQ than children exposed to carbamazepine, lamotrigine, or phenytoin, which has led to changes in the FDA warning label (Meador et al., Reference Meador, Baker, Browning, Clayton-Smith, Combs-Cantrell, Cohen and Loring2009). Neuropsychology is now routinely used to characterize safety of new procedures and new devices, and has demonstrated a surprising finding that not only is long-term responsive neurostimulation to control epilepsy not associated with cognitive side effects, but that there may actually be a therapeutic benefit (Loring, Kapur, Meador, & Morrell, Reference Loring, Kapur, Meador and Morrell2015).
WHERE WILL (SHOULD) NEUROPSYCHOLOGY HEAD DURING THE NEXT DECADE?
Neuropsychology will continue to play an important role in epilepsy research and care. Although the specifics of neuropsychology’s clinical and research applications will likely differ considerably from how neuropsychology is currently practiced, epilepsy neuropsychology will likely change along a common continuum shared by changes in the cognitive assessments and research trends in other diseases.
Changes in Assessment Paradigms
We anticipate that it may become more difficult to obtain approval to conduct comprehensive neuropsychological evaluations outside the context of surgical evaluation, although changes in insurance coverage are difficult to predict. On the positive side, however, we also anticipate shorter but more routine testing of all epilepsy patients due to the high rates of cognitive and behavioral issues, including in new onset cases, with screening becoming a routine part of efforts designed to maximize and maintain cognitive and behavioral health. As part of routine screening, there will likely be greater reliance on computerized assessment such as the NIH Cognitive Toolbox (Weintraub et al., Reference Weintraub, Dikmen, Heaton, Tulsky, Zelazo, Bauer and Gershon2013) or the NIH PROMIS (Nowinski et al., Reference Nowinski, Siderowf, Simuni, Wortman, Moy and Cella2016), and hopefully innovative approaches such as virtual reality tests designed to maximize ecological validity of constructs.
While some may see this as a threat to traditional clinical neuropsychological practice, we view this as an opportunity to characterize cognitive and behavioral comorbidities on a broader basis rather than through selective referrals, with screened patients meeting various thresholds of cognitive impairment continuing to require more comprehensive evaluation for effective clinical management. In our Internet age with growing interest in “big data” approaches, the time would appear right to develop large interactive databases for multicenter collaboration and research purposes. Facilitative movements in this direction have begun in Europe (Vogt et al., Reference Vogt, Aikia, Del Barrio, Boon, Borbely and Bran2017) and within the International League Against Epilepsy (Wilson et al., Reference Wilson, Baxendale, Barr, Hamed, Langfitt, Samson and Smith2015).
Moving from Characterization to Intervention
The neuropsychology of epilepsy literature has been one largely of characterization of disease and treatment impacts. Although there has been increasing emphasis on screening and early identification of comorbidities, there is a clear need for systematic intervention trials to improve practice and patient outcomes. Appearing are reports of interventions designed to improve memory and cognition (Adams et al., Reference Adams, Alipio-Jocson, Inoyama, Bartlett, Sandhu, Oso and Meador2017; Caller et al., Reference Caller, Ferguson, Roth, Secore, Alexandre, Zhao and Jobst2016; Del Felice et al., Reference Del Felice, Alderighi, Martinato, Grisafi, Bosco, Thompson and Masiero2017; Farina, Raglio, & Giovagnoli, Reference Farina, Raglio and Giovagnoli2015; Jackson, Makin, & Baker, Reference Jackson, Makin and Baker2015), mood disorders (Gandy, Sharpe, & Perry, Reference Gandy, Sharpe and Perry2013), and epilepsy self-management (Jobst & MEW network, Reference Jobst2017; Shegog et al., Reference Shegog, Bamps, Patel, Kakacek, Escoffery, Johnson and Ilozumba2013; Wagner et al., Reference Wagner, Modi, Johnson, Shegog, Escoffery, Bamps and Smith2017). Both pharmacologic and behavioral interventions will be critical to improving patient life outcomes in patients with chronic epilepsy (Del Felice et al., Reference Del Felice, Alderighi, Martinato, Grisafi, Bosco, Thompson and Masiero2017) and those who have undergone epilepsy surgery (Mazur-Mosiewicz et al., Reference Mazur-Mosiewicz, Carlson, Hartwick, Dykeman, Lenders, Brooks and Wiebe2015). The reviews to date have noted the numerous weaknesses and challenges inherent in these attempts, but the time appears right to move forward on major intervention efforts.
Increasing Development of New Measures, Terminology, and Classification
Colleagues in neurology, EEG, surgery, radiology, and other fields have been rapidly developing and advancing impressive novel technologies to apply to epilepsy and its comorbidities. Neuropsychology, in contrast, has the dubious distinction of advancing via new versions of established general tests (Loring & Bauer, Reference Loring and Bauer2010). That said, there have been innovative inroads into new approaches to assess language function (Hamberger & Cole, Reference Hamberger and Cole2011), category specific naming (Drane et al., Reference Drane, Ojemann, Aylward, Ojemann, Johnson, Silbergeld and Tranel2008), and social function (theory of mind) (Giovagnoli, Reference Giovagnoli2014), which are important for epilepsy care and will influence practice and research. We also continue to lack a common language of assessment and cognitive diagnosis internationally, few common international epilepsy specific measures, and lack a cognitive taxonomy such as that used in aging and dementia, which would facilitate international communication and research. These issues will develop going forward.
Especially critical will be efforts to increase the cultural sensitivity of measures given the changing demographics across countries. Much of the existing literature on the neuropsychology of epilepsy is based on studies performed in English-speaking countries such that the assessment and research approaches available to the world are geared to that population. Initial steps to address these limitations and develop a more global approach to understanding epilepsy as it is manifested internationally has motivated neuropsychologists to address this issue with the aim to develop methodology with more universal applications (Djordjevic, Reference Djordjevic2011; Jones-Gotman et al., Reference Jones-Gotman, Smith, Risse, Westerveld, Swanson, Giovagnoli and Piazzini2010; Ho & Lee, Reference Ho and Lee2011; Saez et al., Reference Saez, Bender, Barr, Rivera Mindt, Morrison, Hassenstab and Vazquez2014; Witt & Helmstaedter, Reference Witt and Helmstaedter2011).
Increasing Interest in Neurodegenerative Proteins
Before the development of effective AEDs, “dementia” was considered an inevitable end state of chronic epilepsy, and the early epilepsy colonies frequently contained specific buildings for patients with brittle psychiatric disorders and dementia. Concern regarding disease progression, cognitive decline and development of frank disorders associated with cognitive aging is increasing (Breuer et al., Reference Breuer, Grevers, Boon, Bernas, Bergmans, Besseling and Aldenkamp2017; Subota et al., Reference Subota, Pham, Jette, Sauro, Lorenzetti and Holroyd-Leduc2017), especially in the context of the aging of the world population. Interest will increase to identify neuroimaging (e.g., Pardoe et al., Reference Pardoe, Cole, Blackmon, Thesen and Kuzniecky2017) and cognitive biomarkers of “progressive disease” and “accelerated aging.” Assessing the proteins of neurodegeneration, including tau (Puvenna et al., Reference Puvenna, Engleler, Banjara, Brennan, Schreiber, Dadas and Janigro2016; Sen et al., Reference Sen, Thom, Martinian, Harding, Cross, Nikolic and Sisodiya2007; Tai et al., Reference Tai, Koepp, Duncan, Fox, Thompson, Baxendale and Thom2016; Thom et al., Reference Thom, Liu, Thompson, Phadke, Narkiewicz, Martinian and Sisodiya2011) and beta-amyloid (Joutsa et al., Reference Joutsa, Rinne, Hermann, Karrasch, Anttinen, Shinnar and Sillanpaa2017; Mackenzie & Miller, Reference Mackenzie and Miller1994) are under way, and we anticipate more work examining the presence and distribution of neurodegenerative proteins and their relationship to cognition, cognitive change, and cognitive aging in epilepsy.
Transitioning From Clinical Seizure Features to Modifiable Risk Factors
In the general neuropsychology literature, and especially in the Alzheimer’s disease and related preclinical disease literature, a wide range of lifestyle, health, and social factors have been shown to be associated with accelerated cognitive aging and dementia. Increasing interest has focused on potentially modifiable risk factors (e.g., diabetes mellitus, hypertension, obesity, depression, physical inactivity, smoking, low education) (Barnes & Yaffe, Reference Barnes and Yaffe2011; Daviglus et al., Reference Daviglus, Bell, Berrettini, Bowen, Connolly, Cox and Trevisan2010; Norton, Matthews, Barnes, Yaffe, & Brayne, Reference Norton, Matthews, Barnes, Yaffe and Brayne2014). These and other potentially modifiable factors are rarely examined in relation to cognition in epilepsy, although this is likely to change given the demonstrated increased prevalence of these modifiable risk factors in people with epilepsy reported in population based surveys (e.g., Kobau et al., Reference Kobau, Zahran, Thurman, Zack, Henry and Schachter2008).
Work in epilepsy is beginning to appear on factors such as exercise, obesity (Baxendale et al., Reference Baxendale, McGrath, Donnachie, Wintle, Thompson and Heaney2015; Hamed, Reference Hamed2015), vascular risk (Hamed, Reference Hamed2014), and metabolic abnormalities in epilepsy, sometimes examined in relation to cognitive status (Baxendale et al., Reference Baxendale, McGrath, Donnachie, Wintle, Thompson and Heaney2015). These are important developments as such risk factors may affect brain as well as cognition. For example, a recent population-based investigation demonstrated association of atrophy of cerebral white matter volume with overweight and obese individuals, with maximal effects in middle-age corresponding to an estimated increase of brain age of 10 years (Ronan et al., Reference Ronan, Alexander-Bloch, Wagstyl, Farooqi, Brayne, Tyler and Fletcher2016). Other population-based investigations have demonstrated elevated vascular risk factors that may explain the high risk of cerebral and cardiac vascular disease in epilepsy patients (e.g., hyperlipidemia) (Harnod, Chen, Li, Sung, & Kao, Reference Harnod, Chen, Li, Sung and Kao2014), which may also be related to untoward antiepileptic medication effects (Brodie et al., Reference Brodie, Mintzer, Pack, Gidal, Vecht and Schmidt2013).
Neurobehavioral Phenotypes
A clinical reality is that patients with any specific epilepsy syndrome exhibit significant variability in cognitive and behavioral presentations. Our notions of what to expect (memory in TLE, executive dysfunction in FLE) are useful, but often imprecise for reasons discussed previously. There may be discernible latent cognitive and behavioral groups within any epilepsy syndrome, or even latent groups that manifest across diverse epilepsy syndromes. For example, there may be patients with essentially normal cognitive profiles, severely impaired profiles, or profiles characterized by impairments in specific cognitive abilities represented within or across syndrome groups.
Identifying latent groups or phenotypes has been underused in the neuropsychology of epilepsy and would seem to be of heuristic value. Distinct latent neurobehavioral groups have been identified within JME (Valente et al., Reference Valente, Rzezak, Moschetta, de Vincentiis, Coan and Guerreiro2016), Rolandic epilepsy (Smith et al., Reference Smith, Kavros, Clarke, Dorta, Tremont and Pal2012), and children with mixed seizure types (Hermann et al., Reference Hermann, Zhao, Jackson, Jones, Dabbs, Almane and Rathouz2016). Latent developmental trajectories have been identified in children with TLE (Wilson et al., Reference Wilson, Micallef, Henderson, Rayner, Wrennall, Spooner and Harvey2012), in patterns of fMRI language activation (Berl et al., Reference Berl, Zimmaro, Khan, Dustin, Ritzl, Duke and Gaillard2014), and presentations of MR abnormality in TLE (Bernhardt, Hong, Bernasconi, & Bernasconi, Reference Bernhardt, Hong, Bernasconi and Bernasconi2015). This approach may identify groups of patients with similar neurobehavioral profiles, provide a more accurate representation of the variability inherent within syndromic groups, and be related to clinical or neuroimaging data to clarify underlying neurobiology (Dabbs, Jones, Seidenberg, & Hermann, Reference Dabbs, Jones, Seidenberg and Hermann2009).
Latent groups can be observed in regard to the behavioral comorbidities as well. Rayner, Jackson, and Wilson (Reference Rayner, Jackson and Wilson2016b) identified two clinically distinct, symptom-based phenotypes of depression in people with focal epilepsy, namely a cognitive and somatic phenotype reflecting (Wilson & Baxendale, Reference Wilson and Baxendale2014) different clusters of cognitive and vegetative symptoms associated with differential disruption of the underlying network in which the epilepsy principally resides (e.g., the cognitive control or affective network and their interactions (Wilson, Reference Wilson2011). In the short-term, this conceptual advance promises to improve the recognition and management of depression in people with epilepsy, while in the longer-term it may lead to insights into its pathogenesis and the development of individually-tailored treatments (Rayner et al., Reference Rayner, Jackson and Wilson2016a).
CONCLUSION
The 50 years since the founding of the INS have seen many contributions to epilepsy care and research by clinical and experimental neuropsychology, just a few of which were touched on in this review. Progress can be marked by major shifts in thinking, or paradigm shifts altering clinical care, and we have seen our share in this field. Our progress has been substantial and often linked to advances in neurology, radiology, EEG, psychiatry, and other fields caring for people with epilepsy, so the collaborative nature of progress will continue to be key. Fifty years is the current lifespan of the INS, but it is but a brief epoch in the history of the epilepsies, and much remains to be done. With the continuing application of diverse sophisticated multimodal neuroimaging techniques (Chang et al., Reference Chang, Kemmotsu, Leyden, Kucukboyaci, Iragui, Tecoma and McDonald2017) and increasingly innovative analytic approaches (Doucet et al., Reference Doucet, Rider, Taylor, Skidmore, Sharan, Sperling and Tracy2015; Vaessen et al., Reference Vaessen, Jansen, Vlooswijk, Hofman, Majoie, Aldenkamp and Backes2012), the relationship of neurobiological markers of cerebral integrity to cognition and cognitive outcomes will continue to strengthen our understanding of the intersection of cognition, behavior, brain structure, and epilepsy and its treatment.
