Predictive factors of quality of life in temporal and extratemporal lobe epilepsy: association with affective temperament profiles and psychiatric comorbidities

ABSTRACT Background: Investigating predictive factors of reduced quality of life (QOL) of patients and their association with focal epilepsy can improve management and treatment strategies. Objective: This study aimed to investigate the relationship between affective temperaments, depression, anxiety, disease characteristics, and QOL and to explore the predictors of QOL in patients with temporal lobe epilepsy (TLE) and extratemporal lobe epilepsy (ETLE). Methods: A total of 50 patients with TLE, 51 patients with ETLE, and 70 controls were enrolled. Affective temperaments were evaluated using the Temperament Evaluation of Memphis, Pisa, Paris, and San Diego-Autoquestionnaire (TEMPS-A). QOL was assessed by the Short-Form Health Survey (SF-36). Beck Depression Inventory and Beck Anxiety Inventory were used to explore depression and anxiety severity, respectively. Results: Compared with the controls, patients with TLE and ETLE had significantly higher scores on all TEMPS-A scales, except on hyperthymic temperament. All the SF-36 subscale scores were lower in the TLE and ETLE groups. Linear regression analysis revealed that depressive symptoms, anxiety, depressive and irritable temperament, and seizure frequency were significant predictors of QOL in TLE. Patients with ETLE with temperamental disturbances, depressive symptoms, and polytherapy had a poorer QOL. Conclusions: Affective temperaments, psychiatric disorders, and clinical factors may predict impaired QOL in patients with TLE and ETLE. Further studies are needed to identify predictors of QOL in various epilepsy subtypes.


iNtrODUctiON
Epilepsy is a chronic neurological disorder of the brain characterized by recurrent seizures that affects individuals of all ages 1 . Epidemiological studies have revealed that patients with epilepsy (PWE) have a higher incidence of psychiatric comorbidities than the general population 2 . Although there is no clear evidence, pathogenic mechanisms underlying neurobiological, iatrogenic, and psychosocial factors may cause mood disorders (MDs) in the course of epilepsy 3 .
Depending on the chronic and episodic nature of epilepsy, occurrence of unexpected seizures and adverse effects of antiseizure medicines (ASMs) negatively influence quality of life (QOL). In addition, stigmatization, low self-esteem, and social isolation are potential factors for poorer QOL 4 . Although decreased QOL in PWE is well established, its association with MDs in this group is often overlooked 5 . Previous studies have suggested that patients with temporal lobe epilepsy (TLE) are more likely to develop psychiatric disorders due to the role of the limbic system in regulating mood and behavior 6,7 . On the contrary, several authors failed to demonstrate the superiority of the temporal corticosubcortical network in the development of psychiatric comorbidities 4,8 . Thus, the predominance of psychiatric symptoms in TLE remains controversial.
Temperament is defined as an automatic emotional response to events based on the underlying biological and genetic tendencies of personality 9 . Akiskal et al. described five temperaments that can be measured by the Temperament Evaluation of Memphis, Pisa, Paris, and San Diego-Autoquestionnaire (TEMPS-A). According to this tool, affective temperaments can provide clues to the subclinical manifestations of MDs 10 .
To date, two studies have assessed affective temperaments in a population with mixed types of epilepsy using TEMPS-A 11,12 . Therefore, this study aimed to evaluate 1) affective temperaments, 2) the relationship between temperaments, depressive symptoms, anxiety, disease characteristics, and QOL, and 3) possible predictors of QOL in patients with TLE and extratemporal lobe epilepsy (ETLE).

Participants
We screened 101 adult patients with a diagnosis of focal epilepsy and follow-up duration of at least 2 years in our epilepsy outpatient clinic. Patients were evaluated by their medical records including seizure semiology, video electroencephalography (EEG) and/or ictal and interictal EEG analysis, and high-resolution 1.5 Tesla cranial magnetic resonance imaging (MRI). Patients with TLE and ETLE were classified according to the anatomical origin (epileptogenic focus originated from temporal or outside the temporal lobe, respectively), based on semiologic, electroencephalographic, and neuroimaging findings with the International League Against Epilepsy classification of focal epilepsies (1989) 13 . All the valid and reliable scales were applied by an experienced psychiatrist. Patients with high psychopathological scores were referred to the liaison psychiatry unit of our hospital.
The control group consisted of seventy age-and sex-matched healthy volunteers. Individuals with a known psychiatric comorbidity, cognitive disorder and acute/chronic medical disease that could affect test performance were excluded.

Questionnaires
Sociodemographic and clinical data were collected by face to face interview. TEMPS-A, Short Form Health Survey (SF-36), Beck Depression Inventory (BDI) and Beck Anxiety Inventory (BAI) were applied to all participants.

TEMPS-A
Affective temperaments were evaluated by the TEMPS-A questionnaire: a self-administered, Likert-type scale consisting of 110 items that measures five subdimensions as suggested by Akiskal et al.: depressive, cyclothymic, hyperthymic, irritable, and anxious temperaments 10 .

BDI
BDI is a 21-item self-report instrument that assesses the presence and severity of depressive symptoms in the past two weeks. Each item is rated on a 4-point Likert-type scale ranging from 0 to 3. The total score ranges from 0 to 63; subjects with ≥16 points are considered to have depression 14 .

BAI
BAI is used to assess anxiety levels and consists of 21 questions with a Likert scale ranging from 0 to 3 and raw scores ranging from 0 to 63 15 . Scores are categorized as minimal (0 to 7), mild (8 to 15), moderate (16 to 25), and severe anxiety (30 to 63).

SF-36
SF-36 consists of 8 subscales that assess QOL: physical functioning (PF), role limitation due to physical problems (RP), mental health (MH), role limitation due to emotional problems (RE), social functioning (SF), general health (GH), vitality (VT), and bodily pain (BP). All scores range from 0 to 100 and higher scores reflect better QOL 16 .

Statistical analysis
SPSS 17.0 (SPSS Inc., Chicago, IL, USA) was used for statistical analysis. Descriptive analysis was presented using mean ± standard deviation or percentages and numbers. Categorical data were analyzed using the chi-square test. Clinical data were compared between groups using the independent sample t-test and Mann-Whitney U test for normally and non-normally distributed variables, respectively. All scores were compared among the three groups with one-way analysis of variance. For non-normally distributed data, Spearman correlation analysis was used to assess the relationship between TEMPS-A, BDI, BAI, clinical variables, and SF-36 subscales. Multiple linear regression analysis was performed to explore the predictive effect of temperaments, depressive symptoms, anxiety, and clinical variables on QOL. A p value less than 0.05 was considered statistically significant.

Demographic and clinical characteristics
There were no statistical difference between sociodemographic parameters among patients and controls. ETLE patients had significantly fewer seizures than TLE (p= 0.028) ( Table 1).

comparison of teMPS-A, BDi, BAi, and SF-36 scores
Patients with TLE and ETLE had no significant differences in TEMPS-A categories. Depressive, cyclothymic, anxious, and irritable scores in TLE and ETLE were statistically higher than controls (p<0.001). However, there were no significant difference in hyperthymic scores among the groups (p=0.628).
Compared to the control group, PWE had significantly higher BDI and BAI scores. No differences were found in BDI and BAI scores between the two patient groups. In addition, SF-36 subscales in TLE and ETLE patients were significantly lower than controls. There were no significant differences in SF-36 subscales among patient groups ( Table 2).

Correlation analysis of patients with TLE
SF-36 subscales scores, except for MH, were negatively correlated with seizure frequency in TLE patients. Number of ASMs was negatively correlated with GH. Depressive, cyclothymic, and anxious temperament and BDI and BAI scores were negatively correlated with all SF-36 subscales. There was a positive correlation with hyperthymic temperament and PF, GH, and RE. Irritable temperament was negatively correlated with all subscales except SF (Table 3).

Correlation analysis of patients with ETLE
Number of ASMs were negatively correlated with SF-36 subscales except PF and MH. Depressive and anxious temperament and BDI scores were negatively correlated with all SF-36 subscale. Cyclothymic temperament, irritable temperament, and BAI scores, were negatively correlated with all SF-36 subscales, except MH, RE, and VT, respectively. Hyperthymic temperament scores were positively correlated with PF, GH, MH, and VT ( Table 4).   correlated with depressive temperament, BAI and irritable temperament. RP (44.7% of the variance, R=0.685) was significantly correlated with seizure frequency and BAI. SF (35.1% of the variance, R=0.604) was significantly correlated with BAI. RE (32.5% of the variance, R=0.593) was significantly correlated with anxious temperament and seizure frequency. These results suggest that TLE patients with higher scores on depressive temperament, irritable temperament, BDI and BAI and increased seizure frequency had a poorer QOL.

Patients with ETLE
The MH subscale of the SF-36 was strongly correlated with BDI, hyperthymic and irritable temperaments with 52.9% of the variance explained (R=0.747). The SF subscale of the SF-36 test (45.5% of the variance, R=0.682) was significantly correlated with anxious temperament. In addition, PF (37.8% of the variance, R=0.416) was significantly correlated with BDI and anxious and hyperthymic temperaments. In addition, the BP subscale (34.3% of the variance, R=0.586) was significantly correlated with depressive temperament. The RP subscale was significantly correlated with anxious temperament and number of ASMs, with 34% of the variance explained (R=0.583). VT (31.8% of the variance, R=0.588) was significantly correlated with anxious and hyperthymic temperaments. RE was significantly correlated with anxious temperament (29.8% of the variance, R=0.558). GH (24.2% of the variance, R=0.522) was negatively correlated with depressive and hyperthymic temperaments. These results indicated that receiving polytherapy, temperamental disturbances and comorbid depressive symptoms predict QOL in patients with ETLE. Table 5 outlines the statistically significant results.

DiScUSSiON
This study explored the predictors of QOL and examined the relationship among QOL, affective temperaments, depression, anxiety, and clinical characteristics in patients with TLE and ETLE. Previous reports have demonstrated the crucial role of affective temperaments in the prodromal phases and MDs 9,10 . Higher scores on depressive, cyclothymic, and anxious temperaments were associated with a diagnosis of major depressive disorder 10 . Additionally, individuals with irritable, cyclothymic, and hyperthymic temperaments have a tendency to develop bipolar disease 9,17 . In our study, TLE and ETLE groups had significantly higher scores on the depressive, cyclothymic, irritable, and anxious subscales of TEMPS-A than healthy controls. Similar to our results, Yazıcı et al. have shown that all temperament scores were significantly increased, except the hyperthymic temperaments, in PWE 11 . In addition, Taycan et al. found that PWE had higher scores on all TEMPS-A subscales 12 . These reports claimed that PWE present temperamental disturbances that can be a potential precursor for the development of MDs. However, no significant differences in TEMPS-A, depression, and anxiety scores were found between the TLE and ETLE groups. Consistent with our results, Sanchez-Gistau et al. compared the prevalence of interictal psychiatric disorders with SCID-I in patients with refractory TLE and ETLE. In that study, the hypothesis that patients with TLE have more psychiatric comorbidities than ETLE was not accepted 18 .
In our study, affective temperaments, except hyperthymic, were negatively correlated with almost all SF-36 subscales. Interestingly, we detected a positive correlation with hyperthymic temperament and some SF-36 subscales in both patient groups. This situation may explain why excessively confident, grandiose, energetic, and expansive patterns of people with hyperthymic temperament have a paradoxical tendency to mark optimistic choices of the SF-36 questionnaire. More importantly, in our regression analysis, depressive, anxious and irritable temperaments showed a predictive role for low QOL of patients with TLE. In addition, in our regression model, predictors of low QOL in patients with ETLE were depressive, anxious, irritable, and hyperthymic temperaments. These results indicate the determinative role of subclinical MDs on QOL.
To our knowledge, no published study has compared QOL and affective temperament profiles among patients with focal epilepsy. Our findings showed that patients with TLE and ETLE had higher scores on the depression and anxiety questionnaire than the healthy group. Similarly, recent epidemiological studies have identified a higher prevalence of psychiatric disorders in PWE than in the general population 2,19 . Approximately, onethird of PWE experience depression or anxiety in their lifetime 20 . PWE have fear of having seizures and difficulties in participating in social groups and feel alone and rejected, which lead to an increase in psychiatric comorbidities. However, our data does not support the prevalent thought about the superiority of TLE on impaired depression and anxiety scales. Swinkels et al. emphasized that lateralization and localization do not determine the risk of depression, anxiety, and personality in patients with TLE and ETLE, which is consistent with our results 8 . In a prospective multicenter epilepsy surgery study, Devinsky et al. did not find a relationship between depression and anxiety levels and the side or lobe of the seizure focus prior to surgery 21 . In addition, Sanchez-Gistau et al. studied interictal psychiatric disorders in patients with refractory TLE and ETLE and did not support the hypothesis of temporal lobe's priority in psychiatric disorders 18 . In contrast, previous studies have claimed that patients with TLE are more susceptible to affective disorders than those with ETLE or generalized epilepsy 6,22 . Nevertheless, the effect of temporolimbic hyperconnection in TLE on the development of MDs has been controversial among studies, due to the probable effect of the heterogeneity of seizure classification and methodological approaches.
In our sample, SF-36 subscale scores were lower in patients than in controls, without statistical differences between the TLE and ETLE groups. More importantly, in the linear regression analysis, patients with TLE who had comorbid anxiety, increased seizure frequency, depressive temperament, and depressive symptoms had poorer QOL. Additionally, in patients with ETLE, affective temperamental disturbances, depressive symptoms, and polytherapy were predictive factors of lower QOL. Our results confirmed the important role of psychiatric status, frequent seizures, and ASMs in QOL. In patients with TLE, all SF-36 subscale scores were significantly correlated with depression and anxiety scores. All subscales, except MH, were negatively correlated with seizure frequency. In our study, depressive symptoms and comorbid anxiety were the strongest predictors of QOL in the TLE group. Kutlu et al. compared the SF-36 scores of 60 PWE and 33 healthy people and found a correlation between vitality and increased seizure frequency and variables such as educational status, age at seizure onset, disease duration, ASMs, and types of seizures 23 . Similar to this study, Mutluay et al. revealed that PWE with good treatment response had significantly better scores on the SF-36 24 . A hospital-based study conducted in the Czech Republic confirmed that seizure frequency, employability, and psychiatric comorbidities are strong predictors of QOL in PWE 25 . A systematic review from the UK summarized 93 QOL studies found in Medline, Embase, and Cochrane Library databases that were published before July 2010. This review pointed out the associations among seizure frequency, seizure severity, and levels of depression and anxiety with reduced QOL in PWE 26 . Similar to our report, many studies have shown that comorbid depression and anxiety have stronger effects than demographic and disease-related factors 27,28 . On the contrary, some studies have demonstrated the superiority of clinical factors on QOL.
A hospital-based Chinese study claimed that adverse effects of ASMs had a stronger influence than number of ASMs, depression, and anxiety on poorer QOL 29 . A hospital-based study from Russia reported that increased seizure frequency was the strongest predictor of QOL 30 . In the MEPSY study, QOL was assessed using the Quality of Life in Epilepsy Inventory (QOLIE)-10 in 702 people with epilepsy from Korea, and the strongest predictor was adverse effects of ASMs, followed by depression, seizure control, anxiety, and household income 31 . However, these studies have compared the determinants of QOL in mixed groups of PWE. Previous reports support our findings that psychiatric status has a stronger effect than disease-related variables on determining QOL 32,33 . In addition, a number of studies have examined the QOL of patients with refractory TLE and documented a significant association between frequent seizures on poorer QOL 34,35 . Our results agreed with those of previous studies where frequent seizures predicted poorer QOL in TLE. Dourado et al. used the QOLIE-31 to assess the correlation between seizure-related variables and QOL in patients with TLE, ETLE, and idiopathic generalized epilepsy 36 . They found similar scores among patient groups, and increased seizure frequency was a predictor for worse QOL. Consistent with these results, we found no significant differences in SF-36 scores between patient groups. However, in our regression model, seizure frequency was not a predictive factor of poorer QOL in ETLE. Ertem et al. assessed the effects of psychiatric comorbidity, demographic factors, and clinical features on QOL in patients with mesial temporal lobe epilepsy, juvenile myoclonic epilepsy, and healthy controls using the QOLIE-89. In their report, only psychiatric comorbidity in PWE affected QOL regardless of seizure syndrome 37 . Similar to our study, Chen et al. explored the predictive factors of QOL in patients with TLE using the QOLIE-31 inventory 38 . In their study, depression, anxiety, long epilepsy duration, and adverse effects of ASMs were significant predictors for lower QOL. Previous studies that investigated QOL in TLE have controversial results. This difference may be related to the variability in sample size, psychiatric evaluation, methodological process, and heterogeneity of QOL tests.
In psychiatric practice, ASMs have mood-stabilizer and anxiolytic potentials in addition to their antiseizure effect. The beneficial effect of ASMs on QOL and MDs has been well established in PWE 39 . Nevertheless, many studies have reported that polytherapy and side effects of ASMs can lead to impaired QOL 29,38 . In our sample, the number of ASMs was correlated with worse general health in patients with TLE, whereas a significant correlation was detected in all SF-36 subscales, except physical functioning in ETLE. In the linear regression analysis, the number of ASMs in the ETLE group had a negative predictive role on the physical subscale of SF-36. Further studies are needed to elucidate the effect of ASMs on QOL in patients with focal epilepsy.
This study had several limitations. First, our study was carried out in a single center with a relatively small sample from a tertiary care hospital. Second, we classified seizures based on the International League Against Epilepsy 1989 criteria, which is a noninvasive tool for the localization and lateralization of epilepsy 13 . Third, we did not consider the type of ASMs and time interval between seizures, which might interfere with behavioral and cognitive changes. Fourth, participants were not assessed using the SCID 40 . Finally, our study included nonrandomized patients, which might have biased the results due to the selection of individuals who possibly have better cognition and QOL. Thus, further studies with larger groups are needed to assess predictive factors of QOL in different epilepsy types.
Our study has several strengths. To our knowledge, this is the first study that investigated affective temperaments, disease characteristics, and psychiatric status as possible predictors of QOL in patients with TLE and ETLE. Previous studies have assessed affective temperaments only in patients with mixed types of epilepsy. Second, our study provides a new insight into subclinical manifestations of MDs in focal epilepsy. Third, all questionnaires were validated and applied by an experienced psychiatrist.
In conclusion, affective temperamental disturbances, comorbid psychiatric disorders, and disease-related variables can affect the QOL of patients with TLE and ETLE. An important but overlooked topic is the negative effect of affective temperaments. From a clinical perspective, our findings suggest that evaluating factors that may impair QOL in patients with different epilepsy types can improve treatment strategies for subclinical MDs and comorbid psychiatric disorders.