Correspondence Sophie Bayard, PhD, Service de Neurologie, Hôpital Gui-de-Chauliac, 80 avenue Augustin Fliche, 34295 Montpellier cedex 5, France. Tel.: (33) 4 67 33 90 85; fax: (33) 4 67 33 72 85; e-mail: firstname.lastname@example.org
Cataplexy is pathognomonic of narcolepsy with cataplexy, and defined by a transient loss of muscle tone triggered by strong emotions. Recent researches suggest abnormal amygdala function in narcolepsy with cataplexy. Emotion treatment and emotional regulation strategies are complex functions involving cortical and limbic structures, like the amygdala. As the amygdala has been shown to play a role in facial emotion recognition, we tested the hypothesis that patients with narcolepsy with cataplexy would have impaired recognition of facial emotional expressions compared with patients affected with central hypersomnia without cataplexy and healthy controls. We also aimed to determine whether cataplexy modulates emotional regulation strategies. Emotional intensity, arousal and valence ratings on Ekman faces displaying happiness, surprise, fear, anger, disgust, sadness and neutral expressions of 21 drug-free patients with narcolepsy with cataplexy were compared with 23 drug-free sex-, age- and intellectual level-matched adult patients with hypersomnia without cataplexy and 21 healthy controls. All participants underwent polysomnography recording and multiple sleep latency tests, and completed depression, anxiety and emotional regulation questionnaires. Performance of patients with narcolepsy with cataplexy did not differ from patients with hypersomnia without cataplexy or healthy controls on both intensity rating of each emotion on its prototypical label and mean ratings for valence and arousal. Moreover, patients with narcolepsy with cataplexy did not use different emotional regulation strategies. The level of depressive and anxious symptoms in narcolepsy with cataplexy did not differ from the other groups. Our results demonstrate that narcolepsy with cataplexy accurately perceives and discriminates facial emotions, and regulates emotions normally. The absence of alteration of perceived affective valence remains a major clinical interest in narcolepsy with cataplexy, and it supports the argument for optimal behaviour and social functioning in narcolepsy with cataplexy.
The pathognomonic symptom of narcolepsy with cataplexy (NC) is cataplexy, characterized by a sudden drop in muscle tone triggered by emotions suggesting a close interaction between emotions and motor control (Dauvilliers et al., 2007). The emotional triggers of cataplexy had more often a positive valence (e.g. joking, laughter) than negative (e.g. anger, stress) or neutral.
Recent insights into the causes of NC point to the early loss of hypothalamic neurons producing hypocretin/orexin, with a striking decrease in hypocretin-1 concentration in cerebrospinal fluid (CSF) (Dauvilliers et al., 2007; Peyron et al., 2000). Hypocretins are produced exclusively by neurons localized in the lateral hypothalamus (De Lecea et al., 1998; Peyron et al., 1998), a region involved in emotion and sleep regulation (Sakurai, 2005).
A wide range of neuroanatomical structures and neurophysiological systems has been implicated in the phenomenon of emotion and emotion regulation strategies that influence its perception and expression. Among these cerebral structures, the limbic system and more specifically the amygdala have crucial implications (Olson et al., 2007).
The amygdala also plays a major role in the interpretation of emotionally significant stimuli, and has strong projections to the hypocretin area and brainstem regions regulating muscle tone and the sleep/wake process (Sakurai, 2005). Recent studies have explored the structural and functional integrity of the amygdala in human NC. Based on proton resonance spectroscopy results, one study suggested amygdala involvement in NC (Poryazova et al., 2009). As previously documented in patients with amygdala lesions, patients with NC failed to exhibit startle potentiation during unpleasant stimuli (Khatami et al., 2007). A psychophysiological investigation in NC also revealed an attenuated reaction to unpleasant pictures (Tucci et al., 2003). Recent functional magnetic resonance imaging (fMRI) studies examined brain activation patterns in patients with NC when shown humorous material. One of these studies revealed an enhanced ventral striatum and hypothalamus response in patients with NC while shown humorous cartoons (Reiss et al., 2008). In contrast, positive humorous pictures elicited reduced hypothalamic response together with pronounced activity in amygdala in NC (Schwartz et al., 2008). Recently, the same group reported reduced amygdala activity during aversive conditioning together with increased activity in the bilateral amygdala and the dorsal striatum during positive emotional stimuli in NC (Ponz et al., 2010a,b). Taken together, these findings provided convincing evidence to an abnormal amygdala function in patients with NC, which may result in abnormal emotional processing under both pleasant and unpleasant conditions.
A large body of evidence involves the amygdala structure in the perception and judgement of fear (Adolphs et al., 1994; Morris et al., 1996) as well as happiness (Fusar-Poli et al., 2009), and in the processing of negative emotions, especially sadness (Fusar-Poli et al., 2009). Furthermore, this limbic structure is implicated in the modulation of the reappraisal emotional regulation strategy that regulates the experience of emotion and its behavioural and physiological components at an early stage (Goldin et al., 2008; Gross and Oliver, 2008).
This study aimed to measure the performance of patients with NC in a behavioural task known to be impaired in the context of extensive amygdala lesions, i.e. facial emotion recognition task. Most of the human studies support the key role of amygdala, in emotional processing, especially in fear facial emotion recognition. Some human data also suggest abnormal amygdala function in NC. We therefore investigated the hypothesis that NC would show impaired recognition of fear, happiness and sadness in facial expressions compared with patients affected with central hypersomnia without cataplexy (HwoC) and healthy controls. Furthermore, we study the emotional regulation strategies in these patients (using a self-rated questionnaire) by hypothesizing that patients with NC would score higher than patients affected with HwoC and healthy controls on the reappraisal suppression dimension.
Materials and methods
Twenty-one adult patients with NC (12 females, age 18–74 years) were included in this study. Nine patients were treatment-naïve and 12 were drug-free (psychostimulant and/or anti-cataplectic medication) for at least 1 month prior to evaluation. The diagnosis of NC met the standard criteria (American Academy of Sleep Medicine, 2005), with the presence of EDS, clear-cut cataplexy, HLA DQB1*0602, and a multiple sleep latency test (MSLT) score of <8 min with at least two sleep-onset rapid eye movement periods (SOREMPs). Patients with NC with a lumbar puncture (n =4) had undetectable CSF hypocretin-1 levels. Patients with NC were compared with 23 drug-free sex-, age- and intellectual level-matched adult patients with HwoC. Eleven patients (seven females, age 17–67 years) diagnosed with narcolepsy without cataplexy (i.e. mean sleep latency <8 min, and two or more SOREMPs; ISCD-II) were recruited, and 12 patients with idiopathic hypersomnia (eight females, age 16–60 years) with or without long sleep time (i.e. mean sleep latency <8 min, and less than two SOREMPs; ISCD-II). All patients affected with HwoC were treatment-naïve. A lumbar puncture was performed in two patients with narcolepsy without cataplexy and with normal CSF hypocretin-1 levels (i.e. above 200 pg∙mL−1). Note that this study was conducted between 2007 and 2010. After being included, patients had a medical follow-up to monitor treatment tolerance and efficacy, and none of them has developed cataplexy to date. We also recruited 21 sex-, age- and intellectual level-matched control subjects (14 females, age 18–60 years). Inclusion criteria for controls were the ability to understand and give informed consent, no history of neurological or psychiatric disease, and the absence of any medication known to influence sleep or cognition.
All participants (n = 65) were recorded for at least one night followed by the MSLT the next day. The MSLT consisted of five naps scheduled at 2-h intervals starting at 09:00 hours. EDS was self-assessed using the Epworth Sleepiness Scale (ESS). Intellectual capacity was estimated with the French version of the National Adult Reading Test (f-NART). None of the patients or healthy controls had any current psychiatric disorders (including major depressive and anxiety disorders) based on DSM-IV criteria (American Psychiatric Association., 2004) or other neurological disorder.
All subjects gave their written informed consent to participate in the study, which was approved by the Montpellier University Hospital's ethics committee.
Assessment of psychological status and cognitive control of emotion
All subjects participated in a standardized face-to-face clinical interview, and were asked to complete questionnaires [21-item Beck Depression Inventory-II (BDI-II; Beck et al., 1998) and State Trait Anxiety Inventory form Y (STAI)] producing state anxiety and trait anxiety scores (Spielberger, 1983), and neuropsychological tests the day after polysomnography from 09:00 to 12:00 hours between MSLT sessions. All patients were tested just after the MSLT scheduled at 09:00 or 11:00 hours.
All subjects completed the Emotional Regulation Questionnaire (ERQ), being a 10-item self-reported measure of an individual's tendency to use reappraisal and expressive suppression to regulate emotion (Gross and Oliver, 2008). Each item consists of the point Likert scale (1 = strongly disagree; 7 = strongly agree). The ERQ consists of two subscales: six items for reappraisal [e.g. ‘When I want to feel more positive emotion (such as joy or amusement), I change what I'm thinking about’]; and four items for expressive suppression (e.g. ‘When I am feeling positive emotions, I am careful not to express them’), with subscales scored as the mean for the items. We also proposed a modified version of this questionnaire (‘cataplexy-adapted ERQ version’), in adding the phrase ‘to avoid cataplexy’ at the end of each sentence in the expressive suppression scale. Finally, we asked patients with NC to rate the amount of social/professional inconvenience induced by their cataplexy condition on a four-point scale.
Facial emotion recognition task
Participants were shown slides of the faces of six different individuals, each displaying six different basic emotions (happiness, surprise, fear, anger, disgust and sadness) and six neutral faces for a total of 42 stimuli. We used a range of photographs from the Ekman and Friesen series of Pictures of Facial Affect (1976) and a procedure similar to those previously published (Adolphs et al., 1994).
The 42 stimuli were presented eight times across eight separate blocks (each corresponding to one judgement task). In a first step, participants were required to judge in a counterbalanced order each stimulus on two distinct dimensions: arousal and valence. For the arousal dimension, participants rated whether the faces appeared to be relaxing or stimulating on a five-point scale (0 = most relaxing; 5 = most stimulating). For valence, participants rated on a five-point scale whether the faces appeared to be pleasant or not (0 = most unpleasant; 5 = most pleasant). In a second step, participants had to judge the emotion expressed by the face (i.e. emotional intensity). Thus, for a particular facial expression, participants rated the intensity of happiness, surprise, fear, anger, disgust and sadness on a scale from 0 (least) to 5 (most). For each participant, the five blocks (for the judgement of each specific facial expression) were presented in a random order. No feedback was given to indicate the correct emotion displayed. In each of the eight experimental blocks the stimuli were presented in a random order.
Mean ratings for valence and arousal as a function of the six basic emotions expressed by faces and neutral condition were computed for each participant. Participants were tested individually in 45-min sessions.
Data were examined for normal distribution and homogeneity of variance. Group differences in demographic and clinical variables, and rating scale scores were analysed with one-way between-groups analysis of variance (anova) for continuous variables, the Kruskal–Wallis test for ordinal data, and the χ² test for categorical variables. Two 3 × 7 multivariate analysis of variance (manova) were performed with group as a between-subjects factor and the basic emotions expressed by the stimuli and neutral condition as a within-subjects factor, with valence and arousal ratings as dependent variables. Finally, a 3 × 6 manova was performed with group as a between-subjects factor and the basic emotions expressed by the stimuli as a within-subjects factor, with the intensity rating of each emotion on its prototypical labels as dependent variables. Statistical analyses were performed with spss version 12.0 for Windows (SPSS, Chicago, USA). The level of significance was α< 0.05.
Demographic, clinical and polysomnographic data
No group difference was observed for age, estimated intellectual level or gender (Table 1). No difference was found between patients with NC and HwoC for age at onset or duration of illness. Patients with NC had more frequent sleep paralysis (76%) and hypnagogic hallucinations (76%) than HwoC (respectively, 21and 23%), but patients with NC and HwoC were matched on ESS scores. Nineteen percent of patients with NC reported at least one generalized cataplexy per day and 26% one per week. Over half the patients judged the social/professional inconvenience induced by cataplexy as ‘considerable’ to ‘very considerable’.
Table 1. Demographic data, clinical characteristics and sleep parameters in NC, HwoC, and control groups
NC (n = 21)
HwoC (n = 23)
Controls (N = 21)
HwoC, hypersomnia without cataplexy; MSLT, multiple sleep latency tests; NC, narcolepsy with cataplexy; REM, rapid eye movements.
Several between-group differences were noted for sleep and REM sleep latencies, total sleep time and sleep efficiency (Table 1). Patients with NC and HwoC were matched for sleep latency on the MSLT, but differed from the control group. Finally, at least two SOREMPs were present in all cases of narcolepsy with or without cataplexy, with a maximum of one SOREMP in idiopathic hypersomnia and none in controls.
Assessment of psychological status
The presence of a current psychiatric disorder (including major depressive and anxiety disorders) was an exclusion criteria for both patients with central hypersomnia and controls. No difference was observed between groups on level of anxiety (trait and state) or depressive symptoms (Table 2). In addition, the proportion of individuals with moderate to severe level of depressive symptoms (i.e. BDI > 19) did not differ between groups. Because the BDI-II refers to areas that could be affected by central hypersomnia per se, we decided to proceed with a further analysis without considering items 15 ‘Loss of energy’, 16 ‘Changes in sleeping pattern’ or 20 ‘Tiredness or fatigue’. However, results remained unchanged.
Table 2. Emotional status and cognitive control of emotion in NC, HwoC, and control groups
NC (n = 21)
HwoC (n = 23)
Controls (n = 21)
HwoC, hypersomnia without cataplexy; NC, narcolepsy with cataplexy
No difference was noted between groups on ERQ. More specifically, the three groups (NC, HwoC and controls) did not differ in how they used expressive suppression, a strategy of inhibiting behaviours associated with emotional response. The same pattern of results was observed for cognitive reappraisal, which alters the trajectory of emotional response by reformulating the meaning of a situation. Item per item analysis revealed a similar overlap between patients with NC, HwoC and controls. A large intra-group variability was observed on the ERQ version adapted to cataplexy in NC (Table 2). However, we found no relationship between scores on the cataplexy-adapted ERQ version, demographic and clinical variables (i.e. age at onset, duration of illness, cataplexy severity and frequency, and severity of social/professional inconvenience induced by cataplexies).
Facial emotion recognition task
A minority of patients was tested after the MSLT session scheduled at 11:00 hours (n =10). Their performances were similar to those observed in participants evaluated after the MSLT session scheduled at 09:00 hours. Furthermore, we want to underline that all patients produce sleep episodes at all MSLT sessions (09:00, 11:00; 13:00, 15:00, 17:00 hours).
Valence and arousal ratings
Fig. 1a shows highly significant effects of emotion type on valence rating (F2,6 = 212, P <0.001). There was no group effect, and interaction did not reach significance. Happy faces were judged more pleasant than other facial emotions (all P <0.001). Faces expressing surprise or neutral condition were rated significantly higher (i.e. more pleasant) than angry, disgusted and fearful faces (all P <0.001). No difference was noted between angry, disgusted and fearful faces (all P >0.08). Disgusted faces were rated more unpleasant than sad ones (P <0.001).
An effect of emotion type was observed on arousal rating (F2,6 = 31.70, P <0.001), with no group or interaction effects (Fig. 1b). No difference was noted between faces judged angry, disgusted and fearful (all P >0.9), which were rated significantly more aroused than other facial emotions overall (all P <0.05). Happy faces were rated less aroused than surprised (P =0.015) and sad faces (P <0.001). No difference was noted between sad and surprised faces or between happy and neutral faces, which were systematically rated less aroused than all other emotions (all P <0.001).
The three groups (NC, HwoC and controls) showed similar performance in rating facial emotions (Fig. 2). Based on this performance pattern, and to avoid multiple statistical analyses, a manova was performed only on the intensity rating of each emotion on its prototypical label (e.g. rating the intensity of disgust, fear, etc.). No group effect was observed across the emotions sadness, disgust, anger, fear, surprise or happiness (Fig. 3). Patients with NC did not differ from those with HwoC or controls in rating an emotion on its prototypical label. No correlation was found between the modified ERQ version and the intensity rating of each emotion on its prototypical label. Finally, no cataplectic attack was observed during the entire experiment, and none fell asleep during the testing session.
This study constitutes the first clinical investigation of a classical emotional treatment (i.e. facial affect recognition) and the reappraisal emotional regulation strategies (using a self-rated questionnaire) in patients with NC compared with HwoC and healthy controls. The evidence suggests that facial expressions are the most biologically and socially significant visual stimuli in the human environment. We found that patients with NC did not differ from patients with HwoC or healthy controls on intensity rating of each emotion on its prototypical label, and mean ratings for valence and arousal. Moreover, patients with NC did not use different emotional regulation strategies.
Emotional experiences can be described by two factors: valence (how negative or positive); and arousal (how calming or exciting). The two dimensions are regulated by different brain structures: the amygdala plays a role in modulating arousal; whereas the non-amygdala networks, involving mainly the prefrontal cortex, are involved in valence attribution regardless of intensity (Kensinger, 2004). Clinical evidence has shown that sleep and emotion interact in complex ways. Recent findings suggest that the hypothalamus, including the hypocretin region in close anatomical connection with the amygdala, plays a role in normal human emotional processing (Sakurai, 2005).
Cataplexy is closely related with hypocretin deficiency and is frequently triggered by strong emotions (Dauvilliers et al., 2007). The emotional valence of cataplectic attack triggers is more often positive, but can also be negative or neutral (Mattarozzi et al., 2008). In the present study, we decided to present a large number of unfamiliar faces expressing the six basic emotions as well as neutral faces. In addition, subjects were asked to rate each of these faces across eight distinct subjective emotional dimensions, with fear being traditionally the most difficult emotion to recognize and happiness the easiest.
Sleep deprivation and sleepiness per se may impair overall cognitive performance, especially the recognition of human emotions. A recent study reported the negative impact of one night of sleep deprivation on the ability to recognize human facial emotion intensity in normal young subjects (van der Helm et al., 2010). We therefore controlled for the potential influence of sleepiness on cognitive performance when comparing the results of patients with NC with those with HwoC, matched for objective and subjective level of sleepiness. Patients and controls were also matched for other potentially confounding factors, such as age, estimated intellectual level, gender, duration of illness, and levels of depressive and anxious symptoms between patient groups. Hence, recognition performance on facial emotional expression was similar for patients with NC and HwoC and healthy controls, for intensity rating of each emotion and mean ratings for valence and arousal.
The amygdala has a well-documented role in the processing of emotionally salient information, particularly for aversive stimuli (Sotres-Bayon et al., 2004). The results of psychophysiological, spectroscopy and fMRI imaging studies argue for amygdala dysfunction in NC (Ponz et al., 2010a,b; Poryazova et al., 2009; Schwartz et al., 2008; Tucci et al., 2003). However, no change in amygdala metabolism using 18FDG positron emission tomography scans (Dauvilliers et al., 2010; Joo et al., 2004) or when using voxel-based morphometry analysis (Overeem et al., 2003) have been reported in human NC. One fMRI study reported no change in amygdala activity in patients with NC when shown positive humorous material (Reiss et al., 2008). One SPECT study during status cataplecticus failed to report any change on amygdala perfusion (Chabas et al., 2007); however, another SPECT study reported hyperperfusion in right amygdala during episodes of cataplexy (Hong et al., 2006). Behavioural data from physiological and functional brain imaging studies are inconsistent on subjective emotional experience in NC. Some studies indicate that emotional perception was similar between NC and controls (Khatami et al., 2007; Schwartz et al., 2008), while others did not (Reiss et al., 2008; Tucci et al., 2003). Note that these studies used small samples, with no control group matched for hypersomnia, and the emotional materials and rating scales were very heterogeneous. In our study, we found no between-group change in subjective emotional judgement in the recognition of facial expressions of six basic emotions. We must emphasize that all patients with NC in the present study showed a normal pattern of behavioural emotional response at the individual level. Moreover, patients with NC presented similar self-rated emotional regulation strategies to drug-free patients with HwoC and healthy controls. As in our clinical experience, some patients reported avoiding specific emotional situations that trigger cataplexy, but without significant difference from controls and patients with HwoC. However, we observed considerable response variability in our cataplexy-adapted ERQ version, but no association with either cataplexy frequency or social/professional inconvenience induced by cataplexy.
We may nevertheless hypothesize that amygdala is involved in NC, although its activation is not essential for normal performance on facial expression recognition. Even if some studies in patients with bilateral amygdala damage showed no systematic impairment in the recognition of emotional facial expressions, most of them showed severe deficits (Adolphs et al., 1994). Most functional imaging studies have consistently reported amygdala involvement in healthy controls in the perception and judgement of fear (Adolphs et al., 1994; Morris et al., 1996) and happiness (Fusar-Poli et al., 2009), and in the processing of other negative emotions, especially sadness (Fusar-Poli et al., 2009; Schmolck and Squire, 2001). The metabolic changes inconstantly reported within the amygdala structure in NC may be, however, localized in only one hemisphere (i.e. the right hemisphere), being without any consequence on facial expression recognition (Poryazova et al., 2009). Another potential explanation for our results may relate on the normal amygdala activation after processing facial expressions of fear or happiness with normal recognition performance on emotional facial expressions due to the recruitment of alternate brain circuitries in excluding the amygdala pathway. Future researches may focus on experimental tasks designs that force the subject to be engaged in specific pathways.
Several limitations in our study need to be addressed. Unfortunately, CSF hypocretin-1 levels were not available for all patients, but we could assume undetectable levels in almost all NC subjects, as cataplexy, MSLT results and HLA DQB1*0602 positive were clear-cut in all cases. Similarly, we would expect normal CSF hypocretin-1 levels in controls, and in most patients with HwoC (Dauvilliers et al., 2007). Even if hypocretin cells are activated in emotional and sensorimotor conditions similar to those that trigger cataplexy in NC, our results do not support the direct role of the hypocretin system in recognition of emotional facial expressions, but without direct assessment of modulation of any emotional response. Our current results may be in line with a disconnection between the amygdala and the hypocretin region. Hence, hypocretin-mediated amygdala dysfunction may be involved in the strong emotional triggering of cataplexy. We may acknowledge that the emotional triggers related to the presented facial expressions may lack the strength for activation of the hypothalamic amygdala pathways in NC, and especially to trigger cataplexy attacks. In addition, our results rather address recognition of emotion in facial displays being a perceptual process per se that may take place without concomitant emotional experience. Amygdala is classically activated when processing facial expressions of fear; however, its normal functioning may be unnecessary to obtain normal performance on facial expression recognition tasks. Hence, subjects affected with chronic disease including narcolepsy may develop alternate strategies by recruiting different networks to produce normal recognition performances.
Finally, we are aware that small sample size is often a limitation especially when reporting negative results. However, NC is an orphan disease (0.02%), and the prospective recruitment of 21 drug-free patients with NC corresponds to a relatively sizeable group. We have computed an a posteriori power statistical analysis between the groups based on the smallest intensity score difference for fear expression criteria. This smallest difference was observed between patients with NC and patients with HwoC. In the hypothesis of an alpha risk of 0.016 (0.05/3) and a beta risk of 0.20 with 21 patients in the NC group and 23 patients in the HwoC group, a mean intensity of, respectively, 3.83 and 3.95, and a common standard deviation of 0.94, we obtain a power of 2.5% (bilateral test). To show a significant increase between NC group and HwoC group of 0.12 (3.95–3.83) with a power of 0.80 and a type I error α = 0.016, more than 1300 subjects per group would be necessary, which is totally impossible for an orphan disease. Note that if this a posteriori power statistical analysis was computed based on the difference for fear expression criteria between controls (mean = 3.64) and patients with NC (mean = 3.83), more than 1000 subjects per group would be necessary. In sum, patients with NC did not significantly differ from controls in their emotional judgement ability in a classical emotional processing task (i.e. facial affect recognition) as well as in their explicit emotion regulation strategies. The absence of difference in perceived affective valence remains of major clinical interest in NC, and it supports the argument for optimal behaviour and social functioning in NC.
Conceived and designed the experiments: SB, YD. Performed the experiments: SB, MLC. Analysed the data: SB, YD. Contributed reagents/materials/analysis tools: SB, YD. Wrote the paper: SB, YD.
This study was supported by La Fondation Julienne Dumestre. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Professor Dauvilliers has consulted for UCB Pharma, Cephalon and Bioprojet. The other authors have indicated no financial conflicts of interest.