- Email: [email protected]
Residual cognitive effects of uncomplicated idiopathic and cryptogenic epilepsy
Epilepsy & Behavior 13 (2008) 614–619
Contents lists available at ScienceDirect
Epilepsy & Behavior journal homepage: www.elsevier.com/locate/yebeh
Residual cognitive effects of uncomplicated idiopathic and cryptogenic epilepsy q Anne T. Berg a,*, John T. Langfitt b, Francine M. Testa c,d, Susan R. Levy c,d, Francis DiMario e, Michael Westerveld f, Joseph Kulas e a
Department of Biology, Northern Illinois University, DeKalb, IL 60115, USA Department of Neurology, University of Rochester, Rochester, NY 14642, USA c Department of Pediatrics, Yale Medical School, New Haven, CT 06520, USA d Department of Neurology, Yale Medical School, New Haven, CT 06520, USA e Departments of Pediatrics and Neurology, University of Connecticut, and Connecticut Children’s Medical Center, Hartford, CT 06106, USA f Department of Neurosurgery, Yale Medical School, New Haven, CT 06520, USA b
a r t i c l e
i n f o
Article history: Received 18 May 2008 Revised 16 July 2008 Accepted 21 July 2008 Available online 19 August 2008 Keywords: Epilepsy Neuropsychology Cognitive deficit Children Epidemiology Prognosis
a b s t r a c t We assessed residual cognitive deficits in young people with idiopathic and cryptogenic epilepsy. In the setting of an ongoing prospective study, we invited participants initially diagnosed and enrolled in the cohort 8–9 years earlier to undergo standardized neuropsychological assessment. Sibling controls were invited when available. We analyzed 143 pairs in which cases had idiopathic or cryptogenic epilepsy and both case and control had normal intelligence. Compared with that for siblings, the Full Scale IQ for cases was 3.3 points lower (P = 0.01) mainly due to slower processing speed, which was 5.6 points lower (P = 0.0004). Word reading (P = 0.04) and spelling (P = 0.01), but not other scores, were also lower in cases. Remission status and drug use did not influence findings. In young people of normal intelligence with idiopathic or cryptogenic childhood-onset epilepsy, substantial residual effects of epilepsy appear to be confined largely to slower processing speed. Ó 2008 Elsevier Inc. All rights reserved.
1. Introduction Epilepsy is associated with cognitive difficulties for a variety of reasons including the underlying symptomatic causes of epilepsy, the transient effects of seizures and interictal discharges, the cumulative effects of repeated seizures on brain function and structure, and the drugs used to suppress seizures [1–3]. The idiopathic and cryptogenic epilepsies may themselves be associated with some degree of behavioral and cognitive impairment independent of seizures and antiepileptic drugs (AEDs) and even prior to the onset of seizures [4,5]. It is not known whether the functional disturbances that cause epilepsy also cause cognitive impairment, just by themselves. To determine this, it is necessary to take into account the other known causes of cognitive disturbance that occur with varying frequency in persons with epilepsy (e.g., perinatal stroke, other brain lesions, mental retardation, active seizures, and AEDs). Many prior studies that performed detailed neuropsychological evaluations reported significant case–control differences across broad cognitive domains. These studies often focused on small groups [6–12]. Pa-
q This work was funded by a grant from the National Institutes of Health, National Institute of Neurological Disorders and Stroke RO1/R37-NS31146. * Corresponding author. E-mail address: [email protected] (A.T. Berg).
1525-5050/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.yebeh.2008.07.007
tients typically were studied during the active phase of their disorder, many while on AEDs [6,7,9–17]. Several studies enrolled prevalent cases with epilepsy, often from specialty centers, raising concerns that more severe or intractable cases were overrepresented [8,12,13,18], and some studies included individuals who would be considered to have mild mental retardation (i.e., Full Scale IQ < 70) [13,14] or did not state whether they were excluded [15]. In contrast, the National Collaborative Perinatal Project (NCPP) found little difference between cases and sibling controls tested at 7 years, provided both case and control were judged neurologically normal [19]. It is worth noting that in small clinical studies, even syndromes such as benign rolandic epilepsy, a form of epilepsy with a near100% chance of complete seizure remission [20], are associated with evidence of subtle but significant cognitive difficulties [8–11,16]. Current literature does not adequately address whether there are substantial differences among specific forms of idiopathic epilepsy with respect to the nature and severity of cognitive difficulties. It remains unclear whether there is an association between epilepsy and cognitive difficulties that is independent of the effects of other associated causes of cognitive impairment, seizures, and medications. To test whether such an association exists, we examined cognitive function in a sample of neurologically normal individuals with epilepsy and generally normal intelligence, who were participating in a large prospective cohort study of individuals with
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619
615
childhood and adolescent-onset epilepsy. We used tests of general intelligence, attention, and memory, cognitive domains that are often affected by epilepsy and medications.
In secondary analyses, we included those pairs with uncomplicated epilepsy in which either the case or control had a FSIQ of 60– 79, to compare our results with studies that included such patients or did not specify IQ inclusion/exclusion criteria.
2. Methods
2.4. Analytic approach and methods
2.1. Recruitment
For both primary and secondary analyses, we performed a series of matched tests to determine the presence and magnitude of case–control difference for overall intelligence, index scores, other cognitive functions, and school achievement. We also determined whether the case’s medication, remission status, or syndrome type might explain any case–control differences observed. Matched t tests were used for bivariate analyses. Multiple linear regression techniques were used with the difference score as the outcome variable to determine whether there were case–sibling differences, as represented by the intercept term in the model, and the extent to which remission and treatment status might explain those differences. All tests had 90% power to detect an approximately one-third standard deviation difference between cases and controls with an a error of P = 0.05. Because the cognitive measures are intercorrelated, two-tailed P values unadjusted for multiple testing are reported in the tables and text. They are conservatively interpreted, however, in light of the multiple comparisons and with reference to the threshold needed with a Bonferroni correction.
The Connecticut Study of Epilepsy prospectively enrolled children (ages 1 month up to 16 years) during 1993–1997. Intensive follow-up has been ongoing since then. Etiology, seizure types, and epilepsy syndromes were classified according to internationally recommended criteria at the beginning of the study [21,22] and reclassified over the years as new information accrued. Details of the methods, including the initial response rate for recruitment into the cohort (80%), were reported previously [23,24]. Clinical neuroimaging, particularly with MRI, was common in this cohort at the time of initial diagnosis [25]. Over the years more imaging was done, and during the 9-year assessment phase, we offered a research imaging study to cohort members. In all, 85% of the entire cohort had at least one MRI study. 2.2. Standardized cognitive assessments During the period 2002–2006, at approximately 8–9 years after initial study entry for each child, all families were invited to participate in a standardized assessment protocol that included a neuropsychological test battery. When available, we also enrolled a neurologically normal full-sibling control without epilepsy or unprovoked seizures. Tests were administered by a licensed psychologist or trained psychometrician. Either Wechsler’s child (Wechsler Intelligence Scale for Children III, WISC-III [26]) or adult (Wechsler Adult Intelligence Scales III, WAIS-III [27]) scale, as appropriate for the subject’s age, was used to measure general cognitive function. We used an abbreviated form of the WISC-III that preserves the four-factor structure of the full battery [28] and, to facilitate comparability, a modified form of the WAIS-III consistent with the WISC-III short form to estimate IQ and index scores. Full Scale IQ (FSIQ) scores provide a global measure of intelligence. The index scores assess four key components of intelligence, derived by factor analysis performed by the test developers. Verbal memory was assessed with the California Verbal Learning Test (child [29] or adult [30] version, CVLT); visual memory with the Rey Complex Figure test (CFT) [31]; and attention with the Continuous Performance Test (CPT) [32]. The Wide Range Achievement Test, Revision 3 (WRAT3) [33], was used to measure academic achievement. Age-adjusted, standardized scores obtained from the test manuals were used in all analyses. 2.3. Eligibility for analysis Primary analyses involved pairs in which the case had uncomplicated epilepsy (i.e., idiopathic or cryptogenic epilepsy, no epileptic encephalopathy) and both case and control had intelligence in the low-average range or higher (i.e., FSIQ P 80). Pairs were excluded from these analyses if the case had an epileptic encephalopathy or remote symptomatic epilepsy or if either the case or control had a FSIQ < 80. Cases with structural abnormalities on MRI (other than incidental findings) were by definition excluded from the group with idiopathic and cryptogenic epilepsy. The FSIQ exclusion reflected our goal of assessing the effect of epilepsy independent of other neurological conditions including coexisting mental retardation. Because we used a short form of the IQ tests and to be conservative, we excluded anyone whose FSIQ fell in the borderline range (70–79) and below.
2.5. Ethics approval All procedures were approved at each phase of the study by the institutional review boards of all participating institutions and conform to all state and federal laws and international standards. Written informed consent parental permission and child assent or subject consent were obtained. 3. Results A total of 613 children were originally enrolled in the cohort. Fig. 1 provides the derivation of the analytic sample for these analyses. During the time of recruitment for the testing, 530 cohort members were actively followed. Of those, 389 (73%) had idiopathic or cryptogenic epilepsy and intellectual function known or estimated to be within or above the mildly retarded range (IQ > 60), based on prior review of all information in their medical and education records to date [34]. Of these, 272 (70%) had neuropsychological testing. We compared the 272 eligible tested cases with the 117 who were not tested (Table 1). These comparisons suggest that those who were tested were reasonably comparable to those who were not with respect to potentially important factors that we could reliably ascertain, although there was a tendency for participating compared with nonparticipating cases to have parents who had attained higher levels of education. Matched sibling controls were tested for 172 of 272 (63%) cases. Controls for 100 cases were not obtained because there was no full sibling (N = 38), the only sibling was too young (N = 5), or the only sibling had epilepsy or another severe neurological condition (N = 8). In the remaining 49 cases, there were only 7 absolute refusals. In the other instances, controls lived out of state, worked, or were in school (including college). Scheduling difficulties and travel concerns ultimately precluded their participation in neuropsychological testing. Of the 172 cases, all but 3 had undergone at least one MRI study. In 154 instances (89.5%) this was a research MRI scan obtained under a standardized seizure protocol. Of the 172 matched pairs, 29 pairs had either a case (N = 23, 13%) or control (N = 9, 5%) with a FSIQ < 80. In three pairs both
616
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619
83 Lost to followup (includes 13 deaths)
Complicated N=31 (27%)
Uncomplicated N=52 (63%)
613 originally recruited 1993-97
Complicated N=140 (26%)
Not tested N=80 Tested N=60
530 actively followed in 2002-06 Uncomplicated N=390 (74%)
Not tested N=118 (30%) Tested N=272 (70%)
Sib-control N=172 (63%) No control N=100 (37%)
Case&control FSIQ>=80 N=143
Case or control FSIQ<80 N=29
Fig. 1. Derivation of analytic sample.
Table 1 Comparisons of potential differences between eligible cases who were tested and those who were not tested Not tested (N = 118)
Tested (N = 272)
P value (df for v2)
7 (6%) 66 (56%) 44 (37%)
5 (2%) 144 (53%) 123 (45%)
0.05 (2)
100 (85%) 8 (7%) 8 (7%) 1 (1%)
232 (85%) 25 (9%) 13 (5%) 2 (1%)
0.76 (3)
56 (47%) 61 (53%)
137 (50%) 135 (50%)
0.65 (1)
Age at study entry <5 years (N = 124) 5–9 years (N = 179) P10 years (N = 87)
33 (28%) 62 (53%) 22 (19%)
90 (33%) 117 (43%) 65 (24%)
0.19 (2)
Epilepsy syndrome group Idiopathic localization related Cryptogenic focal Idiopathic generalized Undetermined
20 50 37 10
(17%) (42%) (31%) (8%)
43 (16%) 133 (49%) 79 (29%) 17 (29%)
0.67 (3)
61 (52%) 56 (48%)
134 (49%) 138 (51%)
0.60 (1)
31 (26%) 86 (74%)
96 (35%) 176 (65%)
0.09 (1)
88 (75%) 29 (25%)
187 (69%) 85 (31%)
0.20 (1)
Highest level of parental education
Any special education No (N = 195) Yes (N = 195) Remission status* <5 years seizure free (N = 128) P5 years seizure free (N = 262) Taking AEDsa No (N = 275) Yes (N = 115) a
Assessed at the time of testing for those tested, or at the time of the global cognitive assessment if study neuropsychological testing was not performed.
were borderline. Thus, there were 143 pairs for our primary analysis of case–control differences in uncomplicated epilepsy in individuals with FSIQs P 80. In these 143 pairs, the mean age at testing was 15.0 years for cases and 15.6 years for controls. Females constituted 49% of the cases and 57% of the controls. Ninety-five (66%) cases were at least 5 years seizure free; 99 (69%) were no longer taking AEDs; 88 (61.5%) were both seizure free and off AEDs. Only
9 (6.2%) cases were taking more than one AED. In the case group, 20 (14%) had idiopathic focal, 71 (50%) cryptogenic focal, 40 (28%) idiopathic generalized, and 12 (8%) unclassified epilepsy. In 49 pairs, both case and control completed the WAIS-III; in 68 pairs, both completed the WISC-III; in 26 pairs, they were discordant for test version (in 17 pairs, case WISC-III/control WAIS-III; in 9 pairs the converse).
617
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619
3.1. Primary analyses: Case–control differences—uncomplicated epilepsy and FSIQ P80 (N = 143 pairs) Cases scored slightly worse than controls on average on all of the 16 scales from five different tests that were compared. This difference was substantial for processing speed only. Cases scored an average of 5.6 points lower than sibling controls (P = 0.0004) (Table 2a). The critical value for a Bonferroni correction is 0.003. The case, control, and differences scores were all consistent with a normal distribution and the average difference was not due to a few outliers. In all, 29% of cases had scores that fell > 1 SD beneath control values, whereas only 10% of cases had scores > 1 SD above control values. In a series of multiple regression analyses, we tested whether the case–control differences were substantially influenced by adjustment for remission status, AED status of the case, and the case’s syndrome. None was. We also considered whether test version might influence the findings. We first adjusted for whether the case and control within each pair underwent the same or different IQ tests. This did not influence results. We limited the analyses to pairs in which the case and control were concordant for the IQ test version. None of the findings was substantially altered. Most importantly, the case–control difference for processing speed did not vary significantly as a function of test version ( 5.0, P = 0.04, for pairs tested with the WISC-III, and 6.3, P = 0.009, for pairs tested with the WAIS-III). In the subgroup in which cases were 5 years seizure free and off AEDs (N = 90 pairs), this processing speed difference persisted ( 6.0, P = 0.004). 3.2. Secondary analyses: Inclusion of cases and controls with borderline intelligence or mild mental retardation (FSIQ = 60–79) When we included the 29 pairs in which either the case or control or both had a FSIQ of 60–79, we found much stronger evidence of case–control differences (Table 2b). This is not surprising as more cases (13%) than controls (5%) fell in that range. Cases scored
worse than their sibling controls on most cognitive indices, memory tasks, and school achievement, but not on attention scores (CPT). Even with a corrected critical value of 0.003, several of the differences would still be considered significant. In a series of t tests and multiple linear regression analyses, we tested whether remission status (<5 vs > 5 years seizure free) or current AED use explained case–control differences, represented by the intercept in the regression model. Remission status and AED use did contribute to some case–control differences. Cases still scored significantly worse than their controls even after adjustment for these factors. 4. Discussion No prior published study to our knowledge has determined whether epilepsy alone has an impact on cognition, independent of associated factors and causes of epilepsy that can by themselves adversely affect cognitive function. In young people of normal intelligence (FSIQ P80) and normal neurological status, we found little evidence indicating substantial residual impairment in either general or specific cognitive functioning. Although overall intelligence, measured by the FSIQ, was significantly lower in cases than controls, almost all of this effect was due to processing speed, which remained substantially lower in cases despite overall intelligence in the ‘‘normal range.” The cases’ remission and medication status did not explain this finding. References to slowed mental processing have been noted since antiquity in descriptions of persons with epilepsy. Formal studies have documented slowed reaction time relative to controls, but have been confounded by the effects of medications, which also slow processing [35]. One study of children with recently diagnosed epilepsy reported that psychomotor function, measured in part by the coding subtest from the WISC, also used in processing speed, was most affected [15]. Slowed processing speed is a nonspecific finding associated with diffuse brain dysfunction and one of the earliest presenting neurocognitive deficits in disorders affecting primarily white matter early in the course of the disease
Table 2 Case–control differences when (A) both case and control have a FSIQ > 80 and (B) either case or control may have a FSIQ of 60–79 Test score
FSIQ
b
A. Case and control FSIQ > 80 (N = 143 pairs)
B. Case or control FSIQ < 80 (N = 172 pairs) a
Case mean (SD)
Control mean (SD)
P value
Case mean (SD)
Control mean (SD)
P valuea
104.3 (13.9)
107.6 (12.7)
0.01
99.4 (17.6)
104.9 (14.3)
<0.0001
Index scoresb Verbal comprehension Perceptual organization Working memory Processing speed
108.2 (15.2) 106.4 (14.5) 101.9 (14.5) 99.1 (15.2)
110.5 106.8 104.2 104.7
0.09 0.78 0.11 0.0004
103.6 (17.8) 101.8 (17.9) 98.2 (16.4) 96.4 (16.3)
108.0 104.4 102.3 103.4
0.001 0.09 0.004 <0.0001
CVLT List A 1–5c Short-delay free recalld Long-delay free recalld
52.6 (10.9) 0.10 (1.00) 0.12 (0.98)
53.5 (11.2) 0.18 (0.98) 0.27 (0.91)
0.47 0.45 0.20
51.0 (11.5) 0.08 (1.13) 0.01 (1.03)
53.1 (11.1) 0.15 (0.94) 0.23 (0.88)
0.07 0.03 0.03
Rey Complex Figure Immediate recall Delayed recall
49.4 (14.4) 48.7 (14.3)
51.1 (14.1) 50.0 (13.8)
0.30 0.43
46.7 (15.4) 45.8 (15.4)
50.7 (14.1) 49.6 (13.7)
0.01 0.008
CPT Omissions Commissions Reaction time
49.7 (15.4) 50.6 (11.8) 46.9 (10.4)
48.2 (9.2) 49.3 (11.6) 46.6 (10.8)
0.28 0.33 0.79
50.3 (15.0) 50.9 (11.7) 47.5 (10.6)
48.6 (9.8) 49.9 (11.6) 46.9 (10.9)
0.14 0.39 0.52
WRAT Reading Spelling Arithmetic
103.4 (12.9) 102.9 (13.1) 98.2 (18.0)
105.7 (9.0) 105.9 (10.5) 100.9 (13.3)
0.04 0.01 0.09
100.1 (14.8) 99.7 (14.9) 94.7 (19.0)
104.0 (10.1) 104.3 (11.1) 99.8 (13.3)
0.0005 0.0001 0.0009
a b c d
Based on matched t test. Scaled to a mean of 100 and SD of 15. T scores, mean = 50, SD = 10. Z score, mean = 0, SD = 1.
(14.9) (14.3) (13.2) (13.6)
(16.7) (15.4) (13.8) (14.0)
618
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619
(e.g., multiple sclerosis, multi-infarct dementia) [36–38]. Potentially, the disturbances themselves that lead to idiopathic and cryptogenic epilepsy persist in the form of relative inefficiency of processing, even in cognitively and neurologically normal individuals whose epilepsy is in remission. These findings differ from those in the current literature that suggest substantial cognitive impairments in young people with idiopathic and cryptogenic epilepsy. There are several potential reasons for this. First, by using full siblings as controls we were able to assess the impact associated with epilepsy after adjustment for the two greatest determinants of cognitive function, genetic and environmental factors [39]. This was not the case in other studies with the exception of one with several methodological similarities to our own. In that study, after excluding those with IQ scores <80, no difference was found in FSIQ when comparing cases with matched sibling controls [19]. Second, most of the literature focuses on patients with active epilepsy on treatment. In one often-cited study of 251 prevalent cases, 94% of patients were on treatment, with 58% taking two or more drugs [13]. In a study of 53 children with recent-onset epilepsy, 77% of cases were taking AEDs when tested (15). Antiepileptic drugs have a substantial impact on cognitive function [3,40–43]. Additionally, interictal abnormalities, even without seizures, may impair cognitive function transiently [1]. This suggests that the active stage of epilepsy may be the most vulnerable period for experiencing cognitive deficits. Third, some studies included subjects with FSIQs < 80. We initially excluded these to isolate the effects of epilepsy itself from other factors that might cause both seizures and subnormal intellectual function. When we later included them, our results were more consistent with published reports. This suggests that some findings in the literature may reflect inclusion of individuals whose overall level of cognitive function falls below the normal range. It is possible that some studies included children with structural brain abnormalities, the presence of which is often not suspected prior to the onset of seizures and which cannot be determined without appropriate neuroimaging. For example, Byars et al. recently demonstrated that children with such previously unsuspected lesions performed less well on a battery of neuropsychological tests than did children with negative imaging studies [44]. Epidemiologists consider mental retardation alone as sufficient grounds for classifying the cause of epilepsy as ‘‘remote symptomatic” (symptomatic) [22]. Technically, this includes anyone with a FSIQ < 70. The basis for this is an older retrospective, chart review study of patients diagnosed from 1935 to 1974 in which mental retardation was defined as a FSIQ < 70. No information was available, however, about whether cognitive testing was performed or if this designation was simply based on clinical impressions at the time of initial diagnosis. Infants and young children with mild mental retardation would not have been tested and detected. The study, for the most part, considered only severe mental retardation, often in association with cerebral palsy [45]. Because IQ can reflect environmental factors as well as the effects of seizures and medications, we required the FSIQ to be < 60 to use IQ alone (absent abnormal motor examination, abnormal imaging, etc.) as the basis for designating the underlying cause as ‘‘symptomatic” [34]. The prospective identification and intensive follow-up of our community-based cohort generally diminish concerns over selection bias that can be caused by prevalence sampling methods or by studies that do not report their response rates at all, a failing common in the clinical literature in this field. We were not able to test all study subjects or recruit controls for those that we did. Unlike most studies, however, we have been able to describe the derivation of our analytic sample, demonstrate high response rates,
and also provide an analysis of factors that might raise concerns about selection bias. Although individuals could self-select to participate in the cognitive assessments, there is little evidence of major differences between participating and nonparticipating cases. Despite the prospective nature of this study, we do not have longitudinal data with standardized neurocognitive testing done at initial diagnosis and then years later. We do not know if cognitive function changed from what it was 8–9 years earlier; for some, this would have been during infancy and very early childhood. Studies of children with new-onset epilepsy report that they score significantly worse then controls [14,15]. These studies examine children on medication, with recent seizures, and in the active stages of their disorder. Had our study subjects been tested at onset, a similar pattern of findings might have been observed. We think this would be unlikely. If the cases’ current levels of function reflected substantial loss over 8–9 years that would mean that they would have had to be functioning at a substantially higher level than their sibling controls prior to onset, because the current difference between cases and controls is so modest. On the other hand, cases whose FSIQ was in the borderline to very mild mental retardation range, the ones we excluded from the primary analyses, may represent a subgroup that experienced declines since onset. In that regard, some studies considered changes in cognitive function from before or at onset [19,46] to about 3 years later or during the course of chronic epilepsy [47] and found no evidence of change. Nonetheless, declines in cognitive function cannot be fully ruled out. They have been seen in studies that focus on severe epilepsy and institutionalized patients [48]. Determining the exact determinants of such changes may not be a simple matter. These findings should be reassuring for the more than half of individuals with uncomplicated childhood-onset epilepsy and normal intelligence. If the findings for processing speed reflect true, residual cognitive comorbidity associated with epilepsy, this may have implications for schooling and employment. We caution, however, that these findings should not be used to minimize the importance of cognitive comorbidity in young people with epilepsy. In a separate analysis we reported that a quarter of the entire cohort had borderline to severe intellectual impairment [34]. Further, even if the effects of epilepsy appear relatively mild 8 to 9 years later in individuals with overall normal cognitive function, the impact may have been much greater during the active phase of the disorder; a large proportion of the idiopathic and cryptogenic cases received special education services [49]. Our study is not designed to test whether such interventions are effective; however, we strongly caution that our findings should not be used to argue that they are unnecessary. Older studies with prolonged follow-up of social and employment outcomes in adulthood have suggested that young adults with childhood-onset epilepsy are relatively impaired in these domains compared with their peers [50,51]. Long-term follow-up of our cohort has the potential to bridge the gap between manifestations of the disorder in childhood and consequences in adulthood. Acknowledgments We thank all the physicians in Connecticut who have allowed us to recruit and follow their patients. Eugene Shapiro provided essential administrative help throughout. Shlomo Shinnar participated in earlier phases of this study. Above all, the study would not have been possible without the generous help of the many families who have participated over the years. References [1] Binnie CD. Cognitive impairment during epileptiform discharges: is it ever justifiable to treat the EEG? Lancet Neurol 2003;2:725–30.
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619 [2] Dulac O. Epileptic encephalopathy. Epilepsia 2001;42(Suppl. 3):23–6. [3] Meador KJ. Cognitive and memory effects of the new antiepileptic drugs. Epilepsy Res 2006;68:63–7. [4] Austin JK, Harezlak J, Dunn DW, Huster GA, Rose DF, Ambrosius WT. Behavior problems in children before first recognized seizures. Pediatrics 2001;107:115–22. [5] McAfee AT, Chilcott KE, Johannes CB, Hornbuckle K, Hauser WA, Walker AM. The incidence of first provoked and unprovoked seizure in pediatric patients with and without psychiatric diagnoses. Epilepsia 2007;48:1075–82. [6] Sturniolo MG, Galletti F. Idiopathic epilepsy and school achievement. Arch Dis Child 1994;70:424–8. [7] Neyens LGJ, Aldenkamp AP, Meinardi HM. Prospective follow-up of intellectual development in children with recent onset of epilepsy. Epilepsy Res 1999;1999:85–90. [8] Piccirilli M, D’Alessandro P, Tiacci C, Ferroni A. Language lateralization in children with benign partial epilepsy. Epilepsia 1988;29:19–25. [9] Metz-Lutz MN, Kleitz C, t Martin A, Massa R, Hirsch E, Marescaux C. Cognitive development in benign focal epilepsies of childhood. Dev Neurosci 1999;21:182–90. [10] Staden U, Isaacs E, Boyd SG, Brandl U, Neville BG. Language dysfunction in children with rolandic epilepsy. Neuropaediatrics 1998;29:242–8. [11] Lundberg S, Frylmark A, Eeg-Olofsson O. Children with rolandic epilepsy have abnormalities of oromotor and dichotic listening performance. Dev Med Child Neurol 2005;47:603–8. [12] Henkin Y, Sadeh M, Kivity S, Shabtai E, Kishon-Rabin L, Gadoth N. Cognitive function in idiopathic generalized epilepsy of childhood. Dev Med Child Neurol 2005;47:126–32. [13] Bulteau C, Jambaque I, Viguier D, Kieffer V, Dellatolas G, Dulac O. Epileptic syndromes, cognitive assessment and school placement: a study of 251 children. Dev Med Child Neurol 2000;42:319–27. [14] Oostrom K, Schouten A, Kruitwagen C, Peters A, Jennekens-Schinkel A. Behavioral problems in children with newly diagnosed idiopathic or cryptogenic epilepsy attending normal schools are in majority not persistent. Epilepsia 2003;44:97–106. [15] Hermann B, Jones J, Sheth R, Dow C, Koehn M, Seidenberg M. Children with new-onset epilepsy: neuropsychological status and brain structure. Brain 2006;129:2609–19. [16] Giordani B, Caveney AF, Laughrin D, et al. Cognition and behavior in children with benign epilepsy with centrotemporal spikes (BECTS). Epilepsy Res 2006;70:89–94. [17] Parrish J, Geary E, Jones J, Seth R, Hermann B, Seidenberg M. Executive functioning in childhood epilepsy: parent-report and cognitive assessment. Dev Med Child Neurol 2007;49:412–6. [18] Bailet LL, Turk WR. The impact of childhood epilepsy on neurocognitive and behavioral performance. a prospective longitudinal study. Epilepsia 2000;41:426–31. [19] Ellenberg JH, Hirtz DG, Nelson KB. Do seizures in children cause intellectual deterioration? N Engl J Med 1986;314:1085–8. [20] Bouma PA, Bovenkerk AC, Westendorp RG, Brouwer OF. The course of benign partial epilepsy of childhood with centrotemporal spikes: a meta-analysis. Neurology 1997;48:430–7. [21] Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia 1989;30:389–99. [22] Commission on Epidemiology and Prognosis, International League Against Epilepsy. Guidelines for epidemiologic studies on epilepsy. Epilepsia 1993;34:592–6. [23] Berg AT, Shinnar S, Levy SR, Testa FM. Newly diagnosed epilepsy in children: presentation at diagnosis. Epilepsia 1999;40:445–52. [24] Berg AT, Vickrey BG, Testa FM, et al. How long does it take epilepsy to become intractable? A prospective investigation. Ann Neurol 2006;60:73–9. [25] Berg AT, Testa FM, Levy SR, Shinnar S. Neuroimaging in children with newly diagnosed epilepsy: a community-based study. Pediatrics 2000;106:527–32.
619
[26] Wechsler D. Wechsler intelligence scale for children. third ed. San Antonio, TX: Psychological Corp.; 1991. [27] Wechsler D. Wechsler adult intelligence scale. revised ed. San Antonio, TX: Psychological Corp; 1981. [28] Donders J. A short form of the WISC-III for clinical use. Psychol Assess 1997;9:15–20. [29] Delis DC, Kramer JH, Kaplan E, Ober BA. CVLT-C: California Verbal Learning Test, children’s version. San Antonio, TX: Psychological Corp; 1994. [30] Kramer JH, Kaplan E, Ober BA. California verbal learning test. research ed. San Antonio, TX: Psychological Corp; 1987. [31] Meyers JE, Meyers KR. Rey Complex figure test and recognition trial. Lutz, FL: Psychological Assessment Resources; 1995. [32] Conners CK. Conners’ continuous performance test (CPT-II). North Tonawanda, NY: Multi-Health Systems; 2000. [33] Wilkinson GJ. The wide range achievement test. third ed. Wilmington, DE: Jastak; 1993. [34] Berg AT, Langfitt JT, Testa FM, et al. Global cognitive function in children with epilepsy: a community-based study. Epilepsia 2008;49:608–14. [35] Trimble MR, Thompson PJ. Neuropsychological aspects of epilepsy. In: Grant I, Adams KM, editors. Neuropsychological assessment of neuropsychiatric disorders. New York: Oxford University Press; 1986. p. 321–46. [36] Schulz D, Kopp B, Kunkel A, Faiss JH. Cognition in the early stage of multiple sclerosis. J Neurol Neurosurg Psychiatry 2006;253:1002–10. [37] Sanfilipo MP, Benedict RH, Weinstock-Guttman B, Bakshi R. Gray and white matter brain atrophy and neuropsychological impairment in multiple sclerosis. Neurology 2006;66:685–92. [38] Peters N, Opherk C, Danek A, Ballard C, Herzog J, Dichgans M. The pattern of cognitive performance in CADASIL: a monogenic condition leading to subcortical ischemic vascular dementia. Am J Psychiatry 2005;162:2078–85. [39] Gray JR, Thompson PM. Neurobiology of intelligence: science and ethics. Nat Rev Neurosci 2004;5:471–82. [40] Meador KJ, Loring DW, Moore EE, et al. Comparative cognitive effects of phenobarbital, phenytoin, and valproate in healthy adults. Neurology 1995;45:1494–9. [41] Meador KJ. Cognitive side effects of antiepileptic drugs. Can J Neurol Sci 1994;21(Suppl. 3):S12–6. [42] Meador KJ, Loring DW, Vahle VJ, et al. Cognitive and behavioral effects of lamotrigine and topiramate in healthy volunteers. Neurology 2005;64:2108–14. [43] Meador KJ, Loring DW, Ray PG, et al. Differential cognitive and behavioral effects of carbamazepine and lamotrigine. Neurology 2001;56:1177–82. [44] Byars AW, de Grauw TJ, HJohnson CS, et al. The association of MRI findings and neuropsychological functioning after the first recognized seizure. Epilepsia 2007;48:1067–74. [45] Annegers JF, Hauser WA, Elveback LR. Remission of seizures and relapse in patients with epilepsy. Epilepsia 1979;20:729–37. [46] Oostom KJ, van Teeseling H, Smeets-Schouten A, Peters ACB, JennekensSchinkel A. Three to four years after diagnosis: cognition and behavior in children with ‘epilepsy only’: a prospective, controlled study. Brain 2005;128:1546–55. [47] Aldenkamp AP, Alpherts WCJ, De Bruine-Seeder D, Dekker MJA. Test–retest variability in children with epilepsy: a comparison of WISC-R profiles. Epilepsy Res 1990;7:165–72. [48] Dodrill C, Ojemann GA. Do recent seizures and recent changes in antiepileptic drugs impact perfomance on neuropsychological tests in subtle ways that might easily be missed? Epilepsia 2007;48:1833–41. [49] Berg AT, Smith SN, Frobish D, et al. Special education needs in children with newly diagnosed epilepsy. Dev Med Child Neurol 2005;47:749–53. [50] Camfield C, Camfield P, Smith B, Gordon K, Dooley J. Biologic factors as predictors of social outcome of epilepsy in intellectually normal children: a population-based study. J Pediatr 1993;122:869–73. [51] Shackleton DP, Kasteleijn-Nolst Trenite DGA, de Craen AJM, Vandenbroucke JP, Westendorp RGJ. Living with epilepsy: long-term prognosis and psychosocial outcomes. Neurology 2003;61:64–70.
Contents lists available at ScienceDirect
Epilepsy & Behavior journal homepage: www.elsevier.com/locate/yebeh
Residual cognitive effects of uncomplicated idiopathic and cryptogenic epilepsy q Anne T. Berg a,*, John T. Langfitt b, Francine M. Testa c,d, Susan R. Levy c,d, Francis DiMario e, Michael Westerveld f, Joseph Kulas e a
Department of Biology, Northern Illinois University, DeKalb, IL 60115, USA Department of Neurology, University of Rochester, Rochester, NY 14642, USA c Department of Pediatrics, Yale Medical School, New Haven, CT 06520, USA d Department of Neurology, Yale Medical School, New Haven, CT 06520, USA e Departments of Pediatrics and Neurology, University of Connecticut, and Connecticut Children’s Medical Center, Hartford, CT 06106, USA f Department of Neurosurgery, Yale Medical School, New Haven, CT 06520, USA b
a r t i c l e
i n f o
Article history: Received 18 May 2008 Revised 16 July 2008 Accepted 21 July 2008 Available online 19 August 2008 Keywords: Epilepsy Neuropsychology Cognitive deficit Children Epidemiology Prognosis
a b s t r a c t We assessed residual cognitive deficits in young people with idiopathic and cryptogenic epilepsy. In the setting of an ongoing prospective study, we invited participants initially diagnosed and enrolled in the cohort 8–9 years earlier to undergo standardized neuropsychological assessment. Sibling controls were invited when available. We analyzed 143 pairs in which cases had idiopathic or cryptogenic epilepsy and both case and control had normal intelligence. Compared with that for siblings, the Full Scale IQ for cases was 3.3 points lower (P = 0.01) mainly due to slower processing speed, which was 5.6 points lower (P = 0.0004). Word reading (P = 0.04) and spelling (P = 0.01), but not other scores, were also lower in cases. Remission status and drug use did not influence findings. In young people of normal intelligence with idiopathic or cryptogenic childhood-onset epilepsy, substantial residual effects of epilepsy appear to be confined largely to slower processing speed. Ó 2008 Elsevier Inc. All rights reserved.
1. Introduction Epilepsy is associated with cognitive difficulties for a variety of reasons including the underlying symptomatic causes of epilepsy, the transient effects of seizures and interictal discharges, the cumulative effects of repeated seizures on brain function and structure, and the drugs used to suppress seizures [1–3]. The idiopathic and cryptogenic epilepsies may themselves be associated with some degree of behavioral and cognitive impairment independent of seizures and antiepileptic drugs (AEDs) and even prior to the onset of seizures [4,5]. It is not known whether the functional disturbances that cause epilepsy also cause cognitive impairment, just by themselves. To determine this, it is necessary to take into account the other known causes of cognitive disturbance that occur with varying frequency in persons with epilepsy (e.g., perinatal stroke, other brain lesions, mental retardation, active seizures, and AEDs). Many prior studies that performed detailed neuropsychological evaluations reported significant case–control differences across broad cognitive domains. These studies often focused on small groups [6–12]. Pa-
q This work was funded by a grant from the National Institutes of Health, National Institute of Neurological Disorders and Stroke RO1/R37-NS31146. * Corresponding author. E-mail address: [email protected] (A.T. Berg).
1525-5050/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.yebeh.2008.07.007
tients typically were studied during the active phase of their disorder, many while on AEDs [6,7,9–17]. Several studies enrolled prevalent cases with epilepsy, often from specialty centers, raising concerns that more severe or intractable cases were overrepresented [8,12,13,18], and some studies included individuals who would be considered to have mild mental retardation (i.e., Full Scale IQ < 70) [13,14] or did not state whether they were excluded [15]. In contrast, the National Collaborative Perinatal Project (NCPP) found little difference between cases and sibling controls tested at 7 years, provided both case and control were judged neurologically normal [19]. It is worth noting that in small clinical studies, even syndromes such as benign rolandic epilepsy, a form of epilepsy with a near100% chance of complete seizure remission [20], are associated with evidence of subtle but significant cognitive difficulties [8–11,16]. Current literature does not adequately address whether there are substantial differences among specific forms of idiopathic epilepsy with respect to the nature and severity of cognitive difficulties. It remains unclear whether there is an association between epilepsy and cognitive difficulties that is independent of the effects of other associated causes of cognitive impairment, seizures, and medications. To test whether such an association exists, we examined cognitive function in a sample of neurologically normal individuals with epilepsy and generally normal intelligence, who were participating in a large prospective cohort study of individuals with
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619
615
childhood and adolescent-onset epilepsy. We used tests of general intelligence, attention, and memory, cognitive domains that are often affected by epilepsy and medications.
In secondary analyses, we included those pairs with uncomplicated epilepsy in which either the case or control had a FSIQ of 60– 79, to compare our results with studies that included such patients or did not specify IQ inclusion/exclusion criteria.
2. Methods
2.4. Analytic approach and methods
2.1. Recruitment
For both primary and secondary analyses, we performed a series of matched tests to determine the presence and magnitude of case–control difference for overall intelligence, index scores, other cognitive functions, and school achievement. We also determined whether the case’s medication, remission status, or syndrome type might explain any case–control differences observed. Matched t tests were used for bivariate analyses. Multiple linear regression techniques were used with the difference score as the outcome variable to determine whether there were case–sibling differences, as represented by the intercept term in the model, and the extent to which remission and treatment status might explain those differences. All tests had 90% power to detect an approximately one-third standard deviation difference between cases and controls with an a error of P = 0.05. Because the cognitive measures are intercorrelated, two-tailed P values unadjusted for multiple testing are reported in the tables and text. They are conservatively interpreted, however, in light of the multiple comparisons and with reference to the threshold needed with a Bonferroni correction.
The Connecticut Study of Epilepsy prospectively enrolled children (ages 1 month up to 16 years) during 1993–1997. Intensive follow-up has been ongoing since then. Etiology, seizure types, and epilepsy syndromes were classified according to internationally recommended criteria at the beginning of the study [21,22] and reclassified over the years as new information accrued. Details of the methods, including the initial response rate for recruitment into the cohort (80%), were reported previously [23,24]. Clinical neuroimaging, particularly with MRI, was common in this cohort at the time of initial diagnosis [25]. Over the years more imaging was done, and during the 9-year assessment phase, we offered a research imaging study to cohort members. In all, 85% of the entire cohort had at least one MRI study. 2.2. Standardized cognitive assessments During the period 2002–2006, at approximately 8–9 years after initial study entry for each child, all families were invited to participate in a standardized assessment protocol that included a neuropsychological test battery. When available, we also enrolled a neurologically normal full-sibling control without epilepsy or unprovoked seizures. Tests were administered by a licensed psychologist or trained psychometrician. Either Wechsler’s child (Wechsler Intelligence Scale for Children III, WISC-III [26]) or adult (Wechsler Adult Intelligence Scales III, WAIS-III [27]) scale, as appropriate for the subject’s age, was used to measure general cognitive function. We used an abbreviated form of the WISC-III that preserves the four-factor structure of the full battery [28] and, to facilitate comparability, a modified form of the WAIS-III consistent with the WISC-III short form to estimate IQ and index scores. Full Scale IQ (FSIQ) scores provide a global measure of intelligence. The index scores assess four key components of intelligence, derived by factor analysis performed by the test developers. Verbal memory was assessed with the California Verbal Learning Test (child [29] or adult [30] version, CVLT); visual memory with the Rey Complex Figure test (CFT) [31]; and attention with the Continuous Performance Test (CPT) [32]. The Wide Range Achievement Test, Revision 3 (WRAT3) [33], was used to measure academic achievement. Age-adjusted, standardized scores obtained from the test manuals were used in all analyses. 2.3. Eligibility for analysis Primary analyses involved pairs in which the case had uncomplicated epilepsy (i.e., idiopathic or cryptogenic epilepsy, no epileptic encephalopathy) and both case and control had intelligence in the low-average range or higher (i.e., FSIQ P 80). Pairs were excluded from these analyses if the case had an epileptic encephalopathy or remote symptomatic epilepsy or if either the case or control had a FSIQ < 80. Cases with structural abnormalities on MRI (other than incidental findings) were by definition excluded from the group with idiopathic and cryptogenic epilepsy. The FSIQ exclusion reflected our goal of assessing the effect of epilepsy independent of other neurological conditions including coexisting mental retardation. Because we used a short form of the IQ tests and to be conservative, we excluded anyone whose FSIQ fell in the borderline range (70–79) and below.
2.5. Ethics approval All procedures were approved at each phase of the study by the institutional review boards of all participating institutions and conform to all state and federal laws and international standards. Written informed consent parental permission and child assent or subject consent were obtained. 3. Results A total of 613 children were originally enrolled in the cohort. Fig. 1 provides the derivation of the analytic sample for these analyses. During the time of recruitment for the testing, 530 cohort members were actively followed. Of those, 389 (73%) had idiopathic or cryptogenic epilepsy and intellectual function known or estimated to be within or above the mildly retarded range (IQ > 60), based on prior review of all information in their medical and education records to date [34]. Of these, 272 (70%) had neuropsychological testing. We compared the 272 eligible tested cases with the 117 who were not tested (Table 1). These comparisons suggest that those who were tested were reasonably comparable to those who were not with respect to potentially important factors that we could reliably ascertain, although there was a tendency for participating compared with nonparticipating cases to have parents who had attained higher levels of education. Matched sibling controls were tested for 172 of 272 (63%) cases. Controls for 100 cases were not obtained because there was no full sibling (N = 38), the only sibling was too young (N = 5), or the only sibling had epilepsy or another severe neurological condition (N = 8). In the remaining 49 cases, there were only 7 absolute refusals. In the other instances, controls lived out of state, worked, or were in school (including college). Scheduling difficulties and travel concerns ultimately precluded their participation in neuropsychological testing. Of the 172 cases, all but 3 had undergone at least one MRI study. In 154 instances (89.5%) this was a research MRI scan obtained under a standardized seizure protocol. Of the 172 matched pairs, 29 pairs had either a case (N = 23, 13%) or control (N = 9, 5%) with a FSIQ < 80. In three pairs both
616
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619
83 Lost to followup (includes 13 deaths)
Complicated N=31 (27%)
Uncomplicated N=52 (63%)
613 originally recruited 1993-97
Complicated N=140 (26%)
Not tested N=80 Tested N=60
530 actively followed in 2002-06 Uncomplicated N=390 (74%)
Not tested N=118 (30%) Tested N=272 (70%)
Sib-control N=172 (63%) No control N=100 (37%)
Case&control FSIQ>=80 N=143
Case or control FSIQ<80 N=29
Fig. 1. Derivation of analytic sample.
Table 1 Comparisons of potential differences between eligible cases who were tested and those who were not tested Not tested (N = 118)
Tested (N = 272)
P value (df for v2)
7 (6%) 66 (56%) 44 (37%)
5 (2%) 144 (53%) 123 (45%)
0.05 (2)
100 (85%) 8 (7%) 8 (7%) 1 (1%)
232 (85%) 25 (9%) 13 (5%) 2 (1%)
0.76 (3)
56 (47%) 61 (53%)
137 (50%) 135 (50%)
0.65 (1)
Age at study entry <5 years (N = 124) 5–9 years (N = 179) P10 years (N = 87)
33 (28%) 62 (53%) 22 (19%)
90 (33%) 117 (43%) 65 (24%)
0.19 (2)
Epilepsy syndrome group Idiopathic localization related Cryptogenic focal Idiopathic generalized Undetermined
20 50 37 10
(17%) (42%) (31%) (8%)
43 (16%) 133 (49%) 79 (29%) 17 (29%)
0.67 (3)
61 (52%) 56 (48%)
134 (49%) 138 (51%)
0.60 (1)
31 (26%) 86 (74%)
96 (35%) 176 (65%)
0.09 (1)
88 (75%) 29 (25%)
187 (69%) 85 (31%)
0.20 (1)
Highest level of parental education
Any special education No (N = 195) Yes (N = 195) Remission status* <5 years seizure free (N = 128) P5 years seizure free (N = 262) Taking AEDsa No (N = 275) Yes (N = 115) a
Assessed at the time of testing for those tested, or at the time of the global cognitive assessment if study neuropsychological testing was not performed.
were borderline. Thus, there were 143 pairs for our primary analysis of case–control differences in uncomplicated epilepsy in individuals with FSIQs P 80. In these 143 pairs, the mean age at testing was 15.0 years for cases and 15.6 years for controls. Females constituted 49% of the cases and 57% of the controls. Ninety-five (66%) cases were at least 5 years seizure free; 99 (69%) were no longer taking AEDs; 88 (61.5%) were both seizure free and off AEDs. Only
9 (6.2%) cases were taking more than one AED. In the case group, 20 (14%) had idiopathic focal, 71 (50%) cryptogenic focal, 40 (28%) idiopathic generalized, and 12 (8%) unclassified epilepsy. In 49 pairs, both case and control completed the WAIS-III; in 68 pairs, both completed the WISC-III; in 26 pairs, they were discordant for test version (in 17 pairs, case WISC-III/control WAIS-III; in 9 pairs the converse).
617
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619
3.1. Primary analyses: Case–control differences—uncomplicated epilepsy and FSIQ P80 (N = 143 pairs) Cases scored slightly worse than controls on average on all of the 16 scales from five different tests that were compared. This difference was substantial for processing speed only. Cases scored an average of 5.6 points lower than sibling controls (P = 0.0004) (Table 2a). The critical value for a Bonferroni correction is 0.003. The case, control, and differences scores were all consistent with a normal distribution and the average difference was not due to a few outliers. In all, 29% of cases had scores that fell > 1 SD beneath control values, whereas only 10% of cases had scores > 1 SD above control values. In a series of multiple regression analyses, we tested whether the case–control differences were substantially influenced by adjustment for remission status, AED status of the case, and the case’s syndrome. None was. We also considered whether test version might influence the findings. We first adjusted for whether the case and control within each pair underwent the same or different IQ tests. This did not influence results. We limited the analyses to pairs in which the case and control were concordant for the IQ test version. None of the findings was substantially altered. Most importantly, the case–control difference for processing speed did not vary significantly as a function of test version ( 5.0, P = 0.04, for pairs tested with the WISC-III, and 6.3, P = 0.009, for pairs tested with the WAIS-III). In the subgroup in which cases were 5 years seizure free and off AEDs (N = 90 pairs), this processing speed difference persisted ( 6.0, P = 0.004). 3.2. Secondary analyses: Inclusion of cases and controls with borderline intelligence or mild mental retardation (FSIQ = 60–79) When we included the 29 pairs in which either the case or control or both had a FSIQ of 60–79, we found much stronger evidence of case–control differences (Table 2b). This is not surprising as more cases (13%) than controls (5%) fell in that range. Cases scored
worse than their sibling controls on most cognitive indices, memory tasks, and school achievement, but not on attention scores (CPT). Even with a corrected critical value of 0.003, several of the differences would still be considered significant. In a series of t tests and multiple linear regression analyses, we tested whether remission status (<5 vs > 5 years seizure free) or current AED use explained case–control differences, represented by the intercept in the regression model. Remission status and AED use did contribute to some case–control differences. Cases still scored significantly worse than their controls even after adjustment for these factors. 4. Discussion No prior published study to our knowledge has determined whether epilepsy alone has an impact on cognition, independent of associated factors and causes of epilepsy that can by themselves adversely affect cognitive function. In young people of normal intelligence (FSIQ P80) and normal neurological status, we found little evidence indicating substantial residual impairment in either general or specific cognitive functioning. Although overall intelligence, measured by the FSIQ, was significantly lower in cases than controls, almost all of this effect was due to processing speed, which remained substantially lower in cases despite overall intelligence in the ‘‘normal range.” The cases’ remission and medication status did not explain this finding. References to slowed mental processing have been noted since antiquity in descriptions of persons with epilepsy. Formal studies have documented slowed reaction time relative to controls, but have been confounded by the effects of medications, which also slow processing [35]. One study of children with recently diagnosed epilepsy reported that psychomotor function, measured in part by the coding subtest from the WISC, also used in processing speed, was most affected [15]. Slowed processing speed is a nonspecific finding associated with diffuse brain dysfunction and one of the earliest presenting neurocognitive deficits in disorders affecting primarily white matter early in the course of the disease
Table 2 Case–control differences when (A) both case and control have a FSIQ > 80 and (B) either case or control may have a FSIQ of 60–79 Test score
FSIQ
b
A. Case and control FSIQ > 80 (N = 143 pairs)
B. Case or control FSIQ < 80 (N = 172 pairs) a
Case mean (SD)
Control mean (SD)
P value
Case mean (SD)
Control mean (SD)
P valuea
104.3 (13.9)
107.6 (12.7)
0.01
99.4 (17.6)
104.9 (14.3)
<0.0001
Index scoresb Verbal comprehension Perceptual organization Working memory Processing speed
108.2 (15.2) 106.4 (14.5) 101.9 (14.5) 99.1 (15.2)
110.5 106.8 104.2 104.7
0.09 0.78 0.11 0.0004
103.6 (17.8) 101.8 (17.9) 98.2 (16.4) 96.4 (16.3)
108.0 104.4 102.3 103.4
0.001 0.09 0.004 <0.0001
CVLT List A 1–5c Short-delay free recalld Long-delay free recalld
52.6 (10.9) 0.10 (1.00) 0.12 (0.98)
53.5 (11.2) 0.18 (0.98) 0.27 (0.91)
0.47 0.45 0.20
51.0 (11.5) 0.08 (1.13) 0.01 (1.03)
53.1 (11.1) 0.15 (0.94) 0.23 (0.88)
0.07 0.03 0.03
Rey Complex Figure Immediate recall Delayed recall
49.4 (14.4) 48.7 (14.3)
51.1 (14.1) 50.0 (13.8)
0.30 0.43
46.7 (15.4) 45.8 (15.4)
50.7 (14.1) 49.6 (13.7)
0.01 0.008
CPT Omissions Commissions Reaction time
49.7 (15.4) 50.6 (11.8) 46.9 (10.4)
48.2 (9.2) 49.3 (11.6) 46.6 (10.8)
0.28 0.33 0.79
50.3 (15.0) 50.9 (11.7) 47.5 (10.6)
48.6 (9.8) 49.9 (11.6) 46.9 (10.9)
0.14 0.39 0.52
WRAT Reading Spelling Arithmetic
103.4 (12.9) 102.9 (13.1) 98.2 (18.0)
105.7 (9.0) 105.9 (10.5) 100.9 (13.3)
0.04 0.01 0.09
100.1 (14.8) 99.7 (14.9) 94.7 (19.0)
104.0 (10.1) 104.3 (11.1) 99.8 (13.3)
0.0005 0.0001 0.0009
a b c d
Based on matched t test. Scaled to a mean of 100 and SD of 15. T scores, mean = 50, SD = 10. Z score, mean = 0, SD = 1.
(14.9) (14.3) (13.2) (13.6)
(16.7) (15.4) (13.8) (14.0)
618
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619
(e.g., multiple sclerosis, multi-infarct dementia) [36–38]. Potentially, the disturbances themselves that lead to idiopathic and cryptogenic epilepsy persist in the form of relative inefficiency of processing, even in cognitively and neurologically normal individuals whose epilepsy is in remission. These findings differ from those in the current literature that suggest substantial cognitive impairments in young people with idiopathic and cryptogenic epilepsy. There are several potential reasons for this. First, by using full siblings as controls we were able to assess the impact associated with epilepsy after adjustment for the two greatest determinants of cognitive function, genetic and environmental factors [39]. This was not the case in other studies with the exception of one with several methodological similarities to our own. In that study, after excluding those with IQ scores <80, no difference was found in FSIQ when comparing cases with matched sibling controls [19]. Second, most of the literature focuses on patients with active epilepsy on treatment. In one often-cited study of 251 prevalent cases, 94% of patients were on treatment, with 58% taking two or more drugs [13]. In a study of 53 children with recent-onset epilepsy, 77% of cases were taking AEDs when tested (15). Antiepileptic drugs have a substantial impact on cognitive function [3,40–43]. Additionally, interictal abnormalities, even without seizures, may impair cognitive function transiently [1]. This suggests that the active stage of epilepsy may be the most vulnerable period for experiencing cognitive deficits. Third, some studies included subjects with FSIQs < 80. We initially excluded these to isolate the effects of epilepsy itself from other factors that might cause both seizures and subnormal intellectual function. When we later included them, our results were more consistent with published reports. This suggests that some findings in the literature may reflect inclusion of individuals whose overall level of cognitive function falls below the normal range. It is possible that some studies included children with structural brain abnormalities, the presence of which is often not suspected prior to the onset of seizures and which cannot be determined without appropriate neuroimaging. For example, Byars et al. recently demonstrated that children with such previously unsuspected lesions performed less well on a battery of neuropsychological tests than did children with negative imaging studies [44]. Epidemiologists consider mental retardation alone as sufficient grounds for classifying the cause of epilepsy as ‘‘remote symptomatic” (symptomatic) [22]. Technically, this includes anyone with a FSIQ < 70. The basis for this is an older retrospective, chart review study of patients diagnosed from 1935 to 1974 in which mental retardation was defined as a FSIQ < 70. No information was available, however, about whether cognitive testing was performed or if this designation was simply based on clinical impressions at the time of initial diagnosis. Infants and young children with mild mental retardation would not have been tested and detected. The study, for the most part, considered only severe mental retardation, often in association with cerebral palsy [45]. Because IQ can reflect environmental factors as well as the effects of seizures and medications, we required the FSIQ to be < 60 to use IQ alone (absent abnormal motor examination, abnormal imaging, etc.) as the basis for designating the underlying cause as ‘‘symptomatic” [34]. The prospective identification and intensive follow-up of our community-based cohort generally diminish concerns over selection bias that can be caused by prevalence sampling methods or by studies that do not report their response rates at all, a failing common in the clinical literature in this field. We were not able to test all study subjects or recruit controls for those that we did. Unlike most studies, however, we have been able to describe the derivation of our analytic sample, demonstrate high response rates,
and also provide an analysis of factors that might raise concerns about selection bias. Although individuals could self-select to participate in the cognitive assessments, there is little evidence of major differences between participating and nonparticipating cases. Despite the prospective nature of this study, we do not have longitudinal data with standardized neurocognitive testing done at initial diagnosis and then years later. We do not know if cognitive function changed from what it was 8–9 years earlier; for some, this would have been during infancy and very early childhood. Studies of children with new-onset epilepsy report that they score significantly worse then controls [14,15]. These studies examine children on medication, with recent seizures, and in the active stages of their disorder. Had our study subjects been tested at onset, a similar pattern of findings might have been observed. We think this would be unlikely. If the cases’ current levels of function reflected substantial loss over 8–9 years that would mean that they would have had to be functioning at a substantially higher level than their sibling controls prior to onset, because the current difference between cases and controls is so modest. On the other hand, cases whose FSIQ was in the borderline to very mild mental retardation range, the ones we excluded from the primary analyses, may represent a subgroup that experienced declines since onset. In that regard, some studies considered changes in cognitive function from before or at onset [19,46] to about 3 years later or during the course of chronic epilepsy [47] and found no evidence of change. Nonetheless, declines in cognitive function cannot be fully ruled out. They have been seen in studies that focus on severe epilepsy and institutionalized patients [48]. Determining the exact determinants of such changes may not be a simple matter. These findings should be reassuring for the more than half of individuals with uncomplicated childhood-onset epilepsy and normal intelligence. If the findings for processing speed reflect true, residual cognitive comorbidity associated with epilepsy, this may have implications for schooling and employment. We caution, however, that these findings should not be used to minimize the importance of cognitive comorbidity in young people with epilepsy. In a separate analysis we reported that a quarter of the entire cohort had borderline to severe intellectual impairment [34]. Further, even if the effects of epilepsy appear relatively mild 8 to 9 years later in individuals with overall normal cognitive function, the impact may have been much greater during the active phase of the disorder; a large proportion of the idiopathic and cryptogenic cases received special education services [49]. Our study is not designed to test whether such interventions are effective; however, we strongly caution that our findings should not be used to argue that they are unnecessary. Older studies with prolonged follow-up of social and employment outcomes in adulthood have suggested that young adults with childhood-onset epilepsy are relatively impaired in these domains compared with their peers [50,51]. Long-term follow-up of our cohort has the potential to bridge the gap between manifestations of the disorder in childhood and consequences in adulthood. Acknowledgments We thank all the physicians in Connecticut who have allowed us to recruit and follow their patients. Eugene Shapiro provided essential administrative help throughout. Shlomo Shinnar participated in earlier phases of this study. Above all, the study would not have been possible without the generous help of the many families who have participated over the years. References [1] Binnie CD. Cognitive impairment during epileptiform discharges: is it ever justifiable to treat the EEG? Lancet Neurol 2003;2:725–30.
A.T. Berg et al. / Epilepsy & Behavior 13 (2008) 614–619 [2] Dulac O. Epileptic encephalopathy. Epilepsia 2001;42(Suppl. 3):23–6. [3] Meador KJ. Cognitive and memory effects of the new antiepileptic drugs. Epilepsy Res 2006;68:63–7. [4] Austin JK, Harezlak J, Dunn DW, Huster GA, Rose DF, Ambrosius WT. Behavior problems in children before first recognized seizures. Pediatrics 2001;107:115–22. [5] McAfee AT, Chilcott KE, Johannes CB, Hornbuckle K, Hauser WA, Walker AM. The incidence of first provoked and unprovoked seizure in pediatric patients with and without psychiatric diagnoses. Epilepsia 2007;48:1075–82. [6] Sturniolo MG, Galletti F. Idiopathic epilepsy and school achievement. Arch Dis Child 1994;70:424–8. [7] Neyens LGJ, Aldenkamp AP, Meinardi HM. Prospective follow-up of intellectual development in children with recent onset of epilepsy. Epilepsy Res 1999;1999:85–90. [8] Piccirilli M, D’Alessandro P, Tiacci C, Ferroni A. Language lateralization in children with benign partial epilepsy. Epilepsia 1988;29:19–25. [9] Metz-Lutz MN, Kleitz C, t Martin A, Massa R, Hirsch E, Marescaux C. Cognitive development in benign focal epilepsies of childhood. Dev Neurosci 1999;21:182–90. [10] Staden U, Isaacs E, Boyd SG, Brandl U, Neville BG. Language dysfunction in children with rolandic epilepsy. Neuropaediatrics 1998;29:242–8. [11] Lundberg S, Frylmark A, Eeg-Olofsson O. Children with rolandic epilepsy have abnormalities of oromotor and dichotic listening performance. Dev Med Child Neurol 2005;47:603–8. [12] Henkin Y, Sadeh M, Kivity S, Shabtai E, Kishon-Rabin L, Gadoth N. Cognitive function in idiopathic generalized epilepsy of childhood. Dev Med Child Neurol 2005;47:126–32. [13] Bulteau C, Jambaque I, Viguier D, Kieffer V, Dellatolas G, Dulac O. Epileptic syndromes, cognitive assessment and school placement: a study of 251 children. Dev Med Child Neurol 2000;42:319–27. [14] Oostrom K, Schouten A, Kruitwagen C, Peters A, Jennekens-Schinkel A. Behavioral problems in children with newly diagnosed idiopathic or cryptogenic epilepsy attending normal schools are in majority not persistent. Epilepsia 2003;44:97–106. [15] Hermann B, Jones J, Sheth R, Dow C, Koehn M, Seidenberg M. Children with new-onset epilepsy: neuropsychological status and brain structure. Brain 2006;129:2609–19. [16] Giordani B, Caveney AF, Laughrin D, et al. Cognition and behavior in children with benign epilepsy with centrotemporal spikes (BECTS). Epilepsy Res 2006;70:89–94. [17] Parrish J, Geary E, Jones J, Seth R, Hermann B, Seidenberg M. Executive functioning in childhood epilepsy: parent-report and cognitive assessment. Dev Med Child Neurol 2007;49:412–6. [18] Bailet LL, Turk WR. The impact of childhood epilepsy on neurocognitive and behavioral performance. a prospective longitudinal study. Epilepsia 2000;41:426–31. [19] Ellenberg JH, Hirtz DG, Nelson KB. Do seizures in children cause intellectual deterioration? N Engl J Med 1986;314:1085–8. [20] Bouma PA, Bovenkerk AC, Westendorp RG, Brouwer OF. The course of benign partial epilepsy of childhood with centrotemporal spikes: a meta-analysis. Neurology 1997;48:430–7. [21] Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia 1989;30:389–99. [22] Commission on Epidemiology and Prognosis, International League Against Epilepsy. Guidelines for epidemiologic studies on epilepsy. Epilepsia 1993;34:592–6. [23] Berg AT, Shinnar S, Levy SR, Testa FM. Newly diagnosed epilepsy in children: presentation at diagnosis. Epilepsia 1999;40:445–52. [24] Berg AT, Vickrey BG, Testa FM, et al. How long does it take epilepsy to become intractable? A prospective investigation. Ann Neurol 2006;60:73–9. [25] Berg AT, Testa FM, Levy SR, Shinnar S. Neuroimaging in children with newly diagnosed epilepsy: a community-based study. Pediatrics 2000;106:527–32.
619
[26] Wechsler D. Wechsler intelligence scale for children. third ed. San Antonio, TX: Psychological Corp.; 1991. [27] Wechsler D. Wechsler adult intelligence scale. revised ed. San Antonio, TX: Psychological Corp; 1981. [28] Donders J. A short form of the WISC-III for clinical use. Psychol Assess 1997;9:15–20. [29] Delis DC, Kramer JH, Kaplan E, Ober BA. CVLT-C: California Verbal Learning Test, children’s version. San Antonio, TX: Psychological Corp; 1994. [30] Kramer JH, Kaplan E, Ober BA. California verbal learning test. research ed. San Antonio, TX: Psychological Corp; 1987. [31] Meyers JE, Meyers KR. Rey Complex figure test and recognition trial. Lutz, FL: Psychological Assessment Resources; 1995. [32] Conners CK. Conners’ continuous performance test (CPT-II). North Tonawanda, NY: Multi-Health Systems; 2000. [33] Wilkinson GJ. The wide range achievement test. third ed. Wilmington, DE: Jastak; 1993. [34] Berg AT, Langfitt JT, Testa FM, et al. Global cognitive function in children with epilepsy: a community-based study. Epilepsia 2008;49:608–14. [35] Trimble MR, Thompson PJ. Neuropsychological aspects of epilepsy. In: Grant I, Adams KM, editors. Neuropsychological assessment of neuropsychiatric disorders. New York: Oxford University Press; 1986. p. 321–46. [36] Schulz D, Kopp B, Kunkel A, Faiss JH. Cognition in the early stage of multiple sclerosis. J Neurol Neurosurg Psychiatry 2006;253:1002–10. [37] Sanfilipo MP, Benedict RH, Weinstock-Guttman B, Bakshi R. Gray and white matter brain atrophy and neuropsychological impairment in multiple sclerosis. Neurology 2006;66:685–92. [38] Peters N, Opherk C, Danek A, Ballard C, Herzog J, Dichgans M. The pattern of cognitive performance in CADASIL: a monogenic condition leading to subcortical ischemic vascular dementia. Am J Psychiatry 2005;162:2078–85. [39] Gray JR, Thompson PM. Neurobiology of intelligence: science and ethics. Nat Rev Neurosci 2004;5:471–82. [40] Meador KJ, Loring DW, Moore EE, et al. Comparative cognitive effects of phenobarbital, phenytoin, and valproate in healthy adults. Neurology 1995;45:1494–9. [41] Meador KJ. Cognitive side effects of antiepileptic drugs. Can J Neurol Sci 1994;21(Suppl. 3):S12–6. [42] Meador KJ, Loring DW, Vahle VJ, et al. Cognitive and behavioral effects of lamotrigine and topiramate in healthy volunteers. Neurology 2005;64:2108–14. [43] Meador KJ, Loring DW, Ray PG, et al. Differential cognitive and behavioral effects of carbamazepine and lamotrigine. Neurology 2001;56:1177–82. [44] Byars AW, de Grauw TJ, HJohnson CS, et al. The association of MRI findings and neuropsychological functioning after the first recognized seizure. Epilepsia 2007;48:1067–74. [45] Annegers JF, Hauser WA, Elveback LR. Remission of seizures and relapse in patients with epilepsy. Epilepsia 1979;20:729–37. [46] Oostom KJ, van Teeseling H, Smeets-Schouten A, Peters ACB, JennekensSchinkel A. Three to four years after diagnosis: cognition and behavior in children with ‘epilepsy only’: a prospective, controlled study. Brain 2005;128:1546–55. [47] Aldenkamp AP, Alpherts WCJ, De Bruine-Seeder D, Dekker MJA. Test–retest variability in children with epilepsy: a comparison of WISC-R profiles. Epilepsy Res 1990;7:165–72. [48] Dodrill C, Ojemann GA. Do recent seizures and recent changes in antiepileptic drugs impact perfomance on neuropsychological tests in subtle ways that might easily be missed? Epilepsia 2007;48:1833–41. [49] Berg AT, Smith SN, Frobish D, et al. Special education needs in children with newly diagnosed epilepsy. Dev Med Child Neurol 2005;47:749–53. [50] Camfield C, Camfield P, Smith B, Gordon K, Dooley J. Biologic factors as predictors of social outcome of epilepsy in intellectually normal children: a population-based study. J Pediatr 1993;122:869–73. [51] Shackleton DP, Kasteleijn-Nolst Trenite DGA, de Craen AJM, Vandenbroucke JP, Westendorp RGJ. Living with epilepsy: long-term prognosis and psychosocial outcomes. Neurology 2003;61:64–70.