Association of symptoms of psychiatric disease and electroencephalographic patterns in idiopathic generalized epilepsy

Objective: Idiopathic generalized epilepsies (IGE) are genetic epilepsies with alterations of thalamo-frontocortical circuits that play a major role in seizure generation and propagation. Psychiatric diseases and drug resistance are strongly associated, but it remains unknown if they are symptoms of the same pathophysiological process. Hypothesizing that the same network alterations are associated with the frequency of epileptic discharges (ED) and psychiatric symptoms, we here tested the association of self-reported psychiatric symptoms and IGE severity estimated by electroencephalographic (EEG) biomarkers. Methods: Idiopathic generalized epilepsies patients were asked to ﬁll out four validated psychiatric screening tools assessing symptoms of personality disorders (Standard Assessment of Personality-Abbreviated Scale), depression (Major Depression Inventory), impulsiveness (Barratt Impulsiveness Scale), and anxiety (brief Epilepsy Anxiety Survey Instrument). Blinded to results and clinical data on the patients, we analyzed the patients’ EEGs, assessed, and quantiﬁed ED. The number and duration of ED divided by the duration of the EEG served as a proxy for the severity of IGE that was correlated with the results of the psychiatric screening. Results: Paired data from 64 patients were available for analysis. The duration of EDs per minute EEG was inversely associated with the time since the last seizure. The number of patients with generalized poly-spike trains ( n = 2), generalized paroxysmal fast activity ( n = 3), and prolonged epileptiform discharges ( n = 10) were too low for statistically meaningful analyses. Self-reported symptoms of depression, personality disorder, and impulsivity were not associated with EDs. In contrast, the duration of EDs per minute EEG was associated with self-reported symptoms of anxiety in univariate analyses, not signiﬁcant, however, following adjustment for time since the last seizure in regression models. Signiﬁcance: Self-reported symptoms of psychiatric diseases were not strongly associated with EDs as the best available quantiﬁable biomarker of IGE severity. As expected, the duration of EDs per minute and anxiety was inversely associated with time since the last seizure. Our data argue against a direct link be-tween the frequency of EDs – as an objective proxy of IGE severity – and psychiatric symptoms. (cid:1) 2023 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Idiopathic generalized epilepsies comprise three subsyndromes in adults; Juvenile Myoclonic Epilepsy (JME), Juvenile Absence Epilepsy (JAE), and Epilepsy with Generalized Tonic-Clonic Seizures alone (GTCS) [1,2]. Together they account for 15-20% of all epilepsies [3] and it is still disputed whether they should be seen as three different disease entities, or as a spectrum of a single dis-ease with several phenotypes [1,[4][5][6]. Onset is usually in adolescence and the etiology is presumably genetic, although the exact mode of inheritance is not completely understood [1]. IGE isper definition -not associated with severe cognitive decline or deficits [7], but recent studies showed a substantial subjective burden of disease [8] in affected patients and poorer social status and more frequent psychiatric comorbidity as compared to healthy controls [9].
Multiple studies have found an association between IGE and psychiatric disorders with depression and anxiety being the most prevalent [9][10][11][12][13]. A detailed psychiatric assessment of JME patients diagnosed 35% with one or more psychiatric disorders and 23% with a personality disorder [13]. A systematic review found a point prevalence of any psychiatric disorder in IGE patients of up to 51% [10]. There is, however, no consensus on the crucial question if IGEassociated psychiatric symptoms are part of the endophenotype of IGE, secondary to having a chronic disease, or a combination of both [8,14].
The association between psychiatric diagnosis and drug resistance in IGE is well established [15]. In JME patients, anxiety was linked to frequent seizures, drug resistance, and poorer financial situation [12]. A survey from Portugal described social impairment and stigma in IGE and confirmed the established association of psychiatric disease with drug-resistant IGE [16]. In our recent study, IGE patients with recent seizures had higher scores for depression and anxiety compared to seizure-free patients. Scores for depression, personality disorder, and impulsivity exceeded scores in the general Danish population irrespective of recent seizures [8] thereby supporting the hypothesis that psychiatric disease is a part of the endophenotype of IGE, and not just a consequence of poor seizure control.
The pathophysiological link between psychiatric comorbidities and IGE is unknown. A plausible hypothesis is that alterations in cerebral networks, which are established in the frontocorticothalamic region for IGE patients [17,18], also contribute to the psychiatric comorbidity seen in IGE patients. If correct, the same network alterations that cause IGE would also cause psychiatric comorbidity in IGE patients and more severe network alterations would result in more severe IGE and more severe psychiatric symptoms.
To test this hypothesis, markers (or scores) for IGE severity are necessary. Given the large differences in response to different antiseizure medication (ASM) and the problem of pseudoresistance, the number of ASM only partially reflects the severity of the underlying IGE [19][20][21]. Imaging markers of network changes [22] are neither established nor available for larger screening of patients. Thus, the number and duration of epileptic discharges (ED) on EEG are the most objective indicators for IGE severity but are influenced by timing and treatment [23].
We hypothesized that more severe frontocortical network alterations increase disease severity, increase the frequency of EDs on EEGs, and increase the risk for interictal psychiatric symptoms. We tested this hypothesis in an IGE cohort with available EEGs and completed surveys on psychiatric symptoms of personality disorder, anxiety, depression, and impulsivity.

Ethics and approval
The study was approved and registered by the regional office for data security of the Region of Southern Denmark (approval: 20-23429 and 18/25648). All patients had previously given written informed consent to direct contact and the use of their electronic medical files. All participants gave informed consent to study participation and the use of medical files and data using the personal digital mailing system e-boks (e-boks.dk).

Patient cohort and data collection
For this study, we contacted patients with confirmed IGE from a well-characterized cohort (e.g., [9]). At inclusion, patients were 18 years old and fulfilled the recently published diagnostic criteria of the International League against Epilepsy [2]. Patients with childhood absence epilepsy were excluded unless they transformed into JME/JAE/GTCS during puberty. In brief, all experienced one or a combination of the following generalized seizure types after the age of ten: absence, myoclonic, tonic-clonic, and myoclonic-tonic-clonic seizures. Inclusion criteria required also typical syndrome-specific EEG changes, however, normal EEG were accepted in patients with their first seizure before the age of 20 and normal magnet resonance imaging (MRI) and classical description of generalized seizures consistent with IGE. EEG was without consistent focal activity or focal slowing, EEG background was normal [2]. MRI with potentially epileptogenic structural abnormalities, other seizure types (e.g., atonic seizures) or intellectual disability defined as an inability to attend the Danish primary school were exclusion criteria.

Clinical data
The time of the last seizure was self-reported as part of the survey. All other basic clinical characteristics were extracted from the patient's electronic medical files and complemented by direct contact with the patients as described in [7].

Survey and questionnaire
The survey was previously validated and published in [8]. It was created, stored, and distributed via REDCap [24] hosted by the Open Patient Data Explorative Network (OPEN). In this study, we used the Danish version of four standardized and validated questionnaires: the Barratt Impulsiveness Scale-Brief (BIS-brief) [25][26], Standardized Assessment of Personality Abbreviated Scale -Adolescent version (SAPAS-AV) [27], Major Depression Inventory (MDI) [28], the Epilepsy Anxiety Survey Instrument brief version (br-EASI) [8,29]. Further, patients were asked about epilepsy characteristics that were used to validate the survey and the time since the last seizure. On 7th December 2020, a link with the survey was sent directly to patients through the personal digital mailing system e-boks (e-boks.dk), which is linked to the personal registration number. The survey was open from December 2020 to March 2021, and the responses were collected during this period.

Definition of cut-off values in psychiatric screening tools
We used the scores from validated psychiatric screening tools as a measurement for self-reported interictal psychiatric symptoms and used the following cut-offs to divide patients into two groups. No established cut-off value exists for BIS-brief [25]. The Danish version/translation of all screening tools was validated previously.
Patients with 7 points in br-EASI were defined as having ''anxiety". Patients with <7 in br-EASI were defined as having ''no anxiety" as proposed in [29].
Patients with 20 points in MDI were defined as ''depressive". Patients with <20 points in MDI were defined as ''non-depressed" similar to [30].
Patients with 3 points in SAPAS-AV were defined as ''no personality disorder". Patients with >3 points in SAPAS-AV were defined as having a ''personality disorder". This cut-off was chosen because it is recommended for general population research [31].

EEG analysis
We extracted medical files and EEG recordings from all patients who consented (N = 115). From February 2022 to April 2022 author SM extracted and analyzed the patients' EEG recordings with most EDs under the supervision of TK. Both were blinded for the clinical data assessed by JG. As a first step, the descriptions from the available EEG recordings in the patients' medical files were assessed and used to identify the patients' most abnormal EEG. Only standard EEG recordings and sleep EEG recordings recorded from 2007-2022 were used.
From the most abnormal EEG, date and time, duration of EEG recording, number of brief/abortive ED < 1 s, the number of long ED > 1 s, and the cumulative duration of all EDs > 1 s were determined.
We calculated the ''total duration of EDs" by adding the number of brief/abortive EDs with the cumulative duration of all EDs > 1 s and of all brief/abortive EDs presuming that they had an approximate duration of 1 s. By dividing the ''total duration of EDs" by the duration of EEG recording, we adjusted for different durations of EEG recordings.
We calculated the ''total number of EDs" by adding the number of brief/abortive EDs with the number of long EDs. We adjusted for different durations of EEG recordings by dividing them by the duration of EEG recordings.

Statistics and protocol
The statistical analyses were predefined in a protocol before data processing. Statistical analyses were performed using IBM SPSS version 28. We used the Mann-Whitney U test for ordinal data and scatterplots with a linear trend line for continuous data. For correlation analysis, one-sided Spearman's rho correlation was used. After testing for collinearity, we used a logistic regression model to adjust for confounders (age, age at first seizure, number of ASMs tried, seizures last year). For all analyses, a p-value < 0.05 was considered significant. Missing data were excluded from analyses.

Patient demographics
The survey and patient population were validated in a previous study [8]. In this study, we only included patients from the 'Funen cohort' with available EEG data and electronic patient files. For all analyses, a single outlier with a status epilepticus-like EEG and ED during almost 40% of the recording was removed. Importantly, results with and without the outlier did not differ (data not shown). An overview of patient demographics is given in Table 1. The mean age at first seizure was 16.9 years, 64% had no seizures in the last year. The mean duration of EEG recordings was 39.2 min, and the average duration of ED per minute was 0.55 s (0.92% of the recordings). The quantified EEG variables showed good inter-test reliability (Fig. 1A, B), and we chose to use the duration of EDs (>1 s) per minute EEG recording as the main outcome variable for further analysis. Table 2 shows the association between self-reported interictal symptoms of anxiety, depression, and personality disorder and EDs. No association was found between self-reported symptoms of depression (MDI score), and symptoms of personality disorder (SAPAS-AV score). BIS-brief did not correlate with EDs as illustrated in Fig. 2A. Self-reported symptoms of anxiety (br-EASI score) were significantly associated with the number of brief EDs per minute, the number of long EDs per minute, and the duration (s) of long EDs per minute ( Table 2).

Association between anxiety, time since last seizure and epileptic discharges
The time since the last seizure was inversely associated with the average duration of EDs per minute and the br-EASI score (Fig. 2B, p = 0.04, one-sided Spearsman's rho and Fig. 2C, p = 0.004, one-sided Spearsman's rho). In line with this association, the average duration of EDs per minute and br-EASI score were associated, too (Fig. 2D, p = 0.02, one-sided Spearsman's rho). In binary logistic regression analysis using br-EASI > 7 as the binary outcome and correcting for possible confounders (age, age at first seizure, number of ASM tried, seizures last year), the association between br-EASI score and the average duration of EDs per minute disappeared, null-associations were also found in additional exploratory analyses using different models and endpoints (data not shown). The variance inflation factor was below 10 for all variables introduced into the model indicating low collinearity of the variables used.

Discussion
In this study, we hypothesized that the same severe network alterations cause both more severe IGE -as quantified by EDs on EEG -and increased risk for psychiatric symptoms. We tested this hypothesis in a large cohort with available EEGs and completed screening for self-reported interictal symptoms of personality disorder, anxiety, depression, and impulsivity. Our analysis revealed consistent null associations for all psychiatric symptoms apart from anxiety likely explained by the established association between recent seizures and symptoms of anxiety in epilepsy patients [34]. In summary, we conclude that the psychiatric endophenotype is not strongly associated with the frequency of the EDs on the patient's most abnormal EEG. Despite some limitations due to retrospective design and the limited number of participants discussed below, we think that our data allows some pathophysiological conclusions.
The interaction of psychiatric symptoms and IGE is complex. In epilepsy patients, psychiatric disease and symptoms may be due to a common pathophysiological process or secondary due to the challenges associated with a chronic brain disease, or both. The third mechanism, forced normalization (i.e., induction of psychosis by improvement of EDs) is extremely rare in IGE and mainly seen in temporal lobe epilepsy [35,36]. However, the impact of having a chronic disease differs between patients and during life. Having unpredictable seizures are well-established triggers for affective symptoms like anxiety and depression in IGE patients [37] and reported symptoms may be secondary to the disease. This conclusion is supported by the strong association of mainly affective disorders with drug resistance [37][38][39].
Conversely, other symptoms like impulsivity and symptoms of personality disorders are less affected by seizures and are therefore   a postulated part of the endophenotype of IGE. In line, increased impulsivity is observed in JME patients [8,14] and the development of psychiatric symptoms starts before epilepsy onset [40]. Accepting the paradigm that interictal psychiatric symptoms are part of the endophenotype of IGE and the fact that IGE is associated with neurodevelopment disorders [41], it is surprising that interictal psychiatric symptoms, especially impulsivity, and the frequency of ED are not directly linked to each other since we would assume that more impaired neurodevelopment would be associated with more severe epilepsy. Therefore, the frequency of ED and interictal psychiatric symptoms may be caused by two different processes, which is the main conclusion of our study. This interpretation is in line with our previous studies showing that other features of the endophenotype, like magnetic evoked potentials or neuropsychological features, were not associated with treatment response [42][43].
Thus, the strong association between drug-resistant or more difficult-to-treat IGE and psychiatric comorbidity remains unexplained. The simplistic concept that one process, e.g., more severe network changes, results in both more severe epilepsy and more frequent psychiatric comorbidity is, however, probably incorrect.
Our conclusions are based on several assumptions. Most importantly, we assume that the quantity of epileptic discharges in the patients' most abnormal standard EEG (usually at diagnosis) is a proxy for the severity of IGE. While plausible and supported by our own data, falsification of this assumption is difficult given the lack of better and established quantifiable markers of ''IGE severity". We chose to use the most abnormal EEG, typically the EEG at diagnosis, to reduce treatment effects [23]. A disadvantage of this approach was that EEG and psychiatric symptoms were assessed at different time points, sometimes years apart. This may certainly influence affective symptoms (depression, anxiety) that vary substantially during life. However, impulsivity and personality traits are more stable during life [44] and are likely more closely associated with the endophenotype [8,14]. Importantly and despite the temporal distance between EEG and assessment of the psychiatric scores, we found the expected association between time since the last seizure, anxiety and EDs on the patient's most abnormal standard EEG [34].
A limitation of this study is the use of scores from self-reported psychiatric validated screening tools and not psychiatric diagnoses. It allowed us to include a greater number of participants in the analyses and it allowed us to screen for multiple psychiatric diseases. Given the screening tools used are validated and widely used, we do not assume that a detailed psychiatric assessment would have changed our overall conclusion. The sample size of this study was another limitation. It was sufficient to exclude moderate to strong associations between ED and self-reported interictal psychiatric symptoms. However, it does not allow analyses of IGE subsyndromes (e.g., JME vs. GTCS) and identifying more subtle differences between the groups would have required a much higher number of participants.
The EEG data findings are in line with previous studies. The association of ED [45] and prolonged ED > 3 s [32,46] was described previously, and the rate of GPTs and GPFAs was similar to other studies using routine EEGs. The reason why we do not find many patients with GPT in this study is probably that our mean duration of EEG recording was 39.2 min, too short to capture one GPT [47].

Conclusion
Self-reported interictal symptoms of psychiatric diseases were not strongly associated with the studied EEG features on the most abnormal EEG. As expected, the duration of EDs per minute and anxiety were inversely associated with time since the last seizures.
Our data argue against a direct link between the quantity of EDsas an objective proxy of IGE severity -and interictal psychiatric symptoms.
Contribution SM: analyzing EEG, statistical analysis, writing and approval of the manuscript.
TK: design of the study, analyzing EEG, supervision of the study, revision and approval of the manuscript.
JG: Collection of clinical data, approval of the manuscript. CB: design of the study, supervising, writing and approval of the manuscript.

Declaration of Competing Interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: CPB has received honoraria from UCB Nordic A/S, Eisai, and Angelini Pharma and travel support from Angelini Pharma.