Emotional Experience During Rapid-eye-movement Sleep in Narcolepsy

NARCOLEPSY

Emotional Experience During Rapid-eye-movement Sleep in Narcolepsy Roar Fosse, PhD1,2; Robert Stickgold, PhD2; J. Allan Hobson, MD2 1Institute

of Psychology, University in Oslo, Oslo; 2Laboratory of Neurophysiology, Harvard Medical School, Boston

Study Objectives: To describe emotional experience during sleep-onset rapid-eye-movement [(REM) SOREM] sleep and nighttime REM in narcoleptic patients and to relate any differences in REM emotion to the more general abnormalities of this disorder. Design: Awakenings were performed from SOREM (REM at the onset of daytime naps and nighttime sleep) and nighttime emergent (ascending) REM in 15 patients with narcolepsy and from nighttime REM in 9 normal healthy participants. Subjects rated the occurrence and intensity of discrete emotion types for each line in their REM-mentation reports. Fragmentation of REM was measured and related to emotion. Setting: Subjects were studied in their own homes over 2 consecutive days and nights (3 nights for normals) and were monitored by ambulatory polysomnography. Participants: Fifteen patients with narcolepsy, aged 17 to 70 years (mean = 45.3) and 9 normal healthy subjects, aged 31 to 60 years (mean = 43.0). Results: Emotions were found more often and were more intense in narINTRODUCTION THE ABNORMAL REGULATION OF RAPID-EYE-MOVEMENT (REM) SLEEP IS A CENTRAL COMPONENT IN THE SYMPTOMATOLOGY OF NARCOLEPSY.1-3 The REM sleep often occurs at sleep onset, and many patients experience intrusive features of REM, such as muscle paralysis and hallucinations during wake-sleep transitions. Among the most striking behavior symptoms of narcolepsy is the triggering of the muscle paralysis of REM during otherwise alert and aroused waking (cataplexy).1-3 This atypical expression of REM features is thought to result from genetically mediated abnormalities in hypothalamic, brain stem, and basal forebrain function.1,2,4 Cataplexy represents the clearest indication that emotional processes are involved in the REM abnormalities of narcolepsy. Emotion is a potent trigger of cataplexy in most patients, with positive emotions such as joy/elation, excitement, and surprise being particularly likely to elicit cataplectic attacks.5,6 Animal studies have indicated that when cataplexy is induced by positive (or reward) emotions, cholinergic and dopaminergic activity in regions of the basal forebrain and extended amygdala, which normally participate in the regulation of positive emotion,7-12 are dramatically altered.2,13,14 Whereas positive emotion can trigger cataplexy in narcolepsy, negative emotion appears to be increased during sleep. Increased negative emotion is seen particularly when REM-like hallucinations or muscle paralysis appear at wake-sleep transitions,1,15-21 and have been reported also for nocturnal sleep.15 As with positive emotion, this increase in neg-

coleptic SOREM than in nighttime REM of either narcoleptic or normal subjects, with anxiety/fear exhibiting the strongest increase, followed by joy/elation. Comparing nighttime REM in narcoleptic and normal subjects, narcoleptics were found to have more intense feelings of anxiety/fear and of joy/elation but to have a less frequent experience of surprise and anger. Positive and negative emotions occurred in a balanced fashion in SOREM and nighttime REM in narcoleptic subjects. In the SOREM of narcoleptic patients, high levels of positive emotions, in particular of joy/elation, were associated with a less fragmented (more stable) REM sleep. Conclusion: The REM sleep of patients with narcolepsy affords a unique opportunity to study emotion and to analyze its psychophysiology. Narcolepsy intensifies REM-dream emotion, especially anxiety/fear and joy/elation, and this is most clearly seen during SOREM sleep. The changes in REM emotion of narcoleptic patients could reflect the effect of the fundamental pathology of this disorder upon neurobiologic systems that support cognitive-emotional functions. ative emotion is likely to be paralleled by abnormalities in neural processes that modulate these emotions. The function of noradrenaline and serotonin, which are implicated in the regulation of anxiety, fear, and depression,22-24 are altered at several levels of the neuraxis in narcolepsy.25,26 Tricyclic antidepressants, which inhibit the reuptake of norepinephrine and serotonin and are potent REM inhibitors, reduce the occurrence of hypnagogic hallucinations and sleep paralysis and diminish the sleep-associated negative emotion in narcoleptic patients.1 Changes in negative emotion within the REM sleep of narcoleptic patients would be predicted based on the monoaminergic-cholinergic imbalance of this disorder. Not much is known about changes in emotion within the REM sleep of patients with narcolepsy, although feelings of anxiety and fear have been reported more frequently than in REM reports from normal subjects.15 To investigate emotion in narcoleptic REM, 15 patients with narcolepsy and 9 healthy control subjects were asked to rate their dream emotions upon being instrumentally awakened from this stage (for details of emotions in the normal group, see Miller et al26). All subjects slept in their own homes and were monitored with ambulatory polysomnography by one of the authors (RF). Since the pathophysiology of narcolepsy might alter not only emotion, but also the stability of REM, the latter was measured and related to emotion. Results are discussed with reference to the altered neurobiology and the general alterations in REM mentation in narcoleptic patients. Other findings from the present study, and associated methodologic and neurocognitive issues, have been presented and discussed elsewhere.27-30 METHODS

Disclosure Statement This research was supported by the Norwegian Research Counsel and the National Institutes of Health (National Institute of Mental Health MH48,832). Submitted for publication October 2001 Accepted for publication May 2002 Address correspondence to: Roar Fosse, PhD, Laboratory of Neurophysiology, Department of Psychiatry, Harvard Medical School, Massachusetts Mental Health Center, 74 Fenwood Road, Boston, MA 02115; E-mail: [email protected]. SLEEP, Vol. 25, No. 7, 2002 724

Participants Fifteen individuals with narcolepsy, aged 17 to 70 years (mean 45.3, SD = 13.2) participated in the study after providing informed consent. The patients responded to a recruitment letter sent through the Norwegian Narcolepsy Association to a group of 40 persons suggested by the Association to be potential participants in this study. From a set of 20 respondents, 5 patients withdrew before the study started either because of severe narcolepsy or because they could not remain off medication for REM Emotions in Narcolepsy—Fosse et al

the required time. Eight of the remaining 15 patients held or had recently held leading positions in the Norwegian Narcolepsy Association. The group of narcoleptic patients did not exhibit an elevated level of depression on the Beck Depression Inventory when compared to the normal control group. All 15 narcoleptic subjects were HLA-DR2 type positive, reported a history of cataplexy, and revealed both increased sleepiness and the occurrence of SOREM episodes on the Multiple Sleep Latency Test. Four patients used no medication, 11 used stimulants (amphetamine or methylphenidate), 1 used paroxetine, and 2 used fluoxetine. Medications used to treat narcolepsy or that were known or assumed to affect sleep were discontinued at least 1 week before the study onset. Nine healthy adults, aged 31 to 60 years (mean, 43.0; SD = 8.7) constituted the normal control group. (Additional information about this control population has been previously published.27,28) These participants were either acquaintances of patients with narcolepsy (n = 4) or were recruited through advertisements (n = 5). None of the controls used any medication known or suspected to affect sleep. No subject in either group suffered from sleep apnea, restless legs syndrome, or any other identified disorders known to alter sleep, and they had no history of psychiatric disorders.

The emotion types that were reported by the subjects using this instrument, and the categories into which these emotions were sorted, are shown in Table 1. A single entry of emotion on a report form was defined as an emotion incident. The emotion (type) incidence rate for a report was defined as the total number of different emotions in that report. The report prevalence of emotion for each subject was defined as the percentage of reports from that subject with at least one incident of emotion. The prevalence of a particular emotion or class of emotions for a subject was defined as the percentage of reports from that subject containing at least one incident of that emotion or class. The overall prevalence of a particular emotion or class of emotions was defined as the average subject report prevalence for that emotion or class. When analyzing the intensity of emotion, only those reports (and subjects) were included that had at least one entry of the emotion or emotion class in question. When more than one incidence of a particular emotion type was present in a report, the highest intensity score for this type was used. The general intensity level of a REM report was defined as the highest intensity score for any emotion type in that report. Some analyses were performed using a total score for emotion for each report. This total emotion score was defined as the sum of the highest intensity scores for the different emotion types noted in that report. For example, a report with the following emotion entries would be given a total emotion score of 5: anxiety/fear 1, joy/elation 2, surprise 2, joy/elation 1.

Apparatus Six channels on an Oxford Medilog ambulatory system (9000) were used, of which two recorded electroencephalogram (C4-A1 and C3-A2), two recorded electrooculogram (left and right outer canthus with referrals to A1), and two recorded submental electromyogram (EMG). Signals were stored on cassette tape and simultaneously displayed on a portable Toshiba 3200 SXC personal computer. The REM periods were identified using Rechtschaffen and Kales’ criteria,31 modified to allow for tonically enhanced EMG amplitude during sleep-onset REM (SOREM) in narcoleptic patients. Participants slept in their own beds in their own homes. The experimenter was situated in a room separated from the bedroom by at least one closed door and performed awakenings through a portable, 2-way intercom system. In the narcolepsy group, windows were covered during daytime sessions.

Fragmentation of REM The relation between emotion and the stability or fragmentation of REM sleep was investigated by use of a fragmentation index. This index reflected the total number of brief arousals from sleep, as well as brief stage non-REM (NREM) intrusions within a REM period.28 Arousals and NREM intrusions were identified as the conjunct occurrence of increased EMG potential lasting at least 1.5 seconds, together with electroencephalographic changes indicative of waking or NREM.33 For each REM period, a fragmentation density score was calculated by dividing the sum of brief arousals and intermittent NREM intrusions by the number of minutes spent in REM.

Scoring of Emotions

Time Since Sleep Onset

For each line in a written mentation report form, eight columns were used to indicate any experience of the following emotions or groups of emotions: anger, anxiety/fear, sadness, shame, joy/elation, love/eroticism, surprise, and “other” emotions.27,32 For the “other” category, subjects described the type of emotion at the bottom of the form. The numbers 1 (low), 2 (medium), and 3 (high) were used to indicate the intensity of emotions. No attempt was made to distinguish between emotions and moods; the subjects were instructed to register any experience of emotion regardless of its duration or the degree of relatedness to formal dream elements.

For SOREM sleep, we investigated changes in emotion as a function of total time spent in sleep before awakening. The SOREM reports from each subject were divided into an early-report and late-report subset. The sleep time since sleep onset was defined as the time from sleep onset to the start of SOREM plus the time spent within SOREM. Reports with a sleep time since sleep onset less than the median were combined into the early subset, while those with times greater than the median made up the late subset. For subjects with an uneven number of reports, the median report was grouped with the subset having a mean sleep time from sleep onset closest to the median value. Emotion scores used for each report were the total scores. Procedures Due to the need to keep the narcoleptic patients off medication, all of the recording sessions for these subjects had to be performed intensively during the available time window. A pilot study showed that 2 consecutive days and nights were generally manageable for most subjects. The daytime sessions were scheduled 3 to 5 hours after habitual morning awakening. The order of the nighttime and daytime sessions was counterbalanced across subjects. The control group was studied for 3 consecutive nights, which gave both an additional nighttime session but one less experimental session than in the narcolepsy group. On each night, all subjects slept through at least one full REM period without being awakened. Awakenings were performed between 5 and 15 minutes into nighttime REM periods and between 3 and 15 minutes into SOREM.

Table 1—Emotion types and emotion categories. Emotions were classified as positive, negative, or neutral. The last two types, frustration/irritation and stress/despair, were defined post-hoc and were indicated by subjects under the category of "other." Emotion type

Positive emotions

Anxiety/fear Anger Joy/elation Love-eroticism Sadness Shame Surprise Frustration/irritation Stress/despair

SLEEP, Vol. 25, No. 7, 2002

Negative emotions X X

X X X X Neutral X X

725

REM Emotions in Narcolepsy—Fosse et al

The frequency and timing of mentation reports from the narcoleptic and normal subjects are shown in Table 2.

Table 2—General characteristics of REM mentation reports REM condition

Number of reports with content (% of all reports)]

Mean time in REM (min)

62 (98) 71 (87) 88 (89) 221 (91)

6.7 8.9 8.8 8.2

SOREM narcolepsy1 Nighttime REM narcolepsy Nighttime REM normals Total

Statistical Analysis Analyses were performed on subject averages. Differences were tested using ANOVA when appropriate and possible. Other analyses were performed using paired and unpaired t tests or nonparametric equivalents. In some cases, narcoleptic subjects were excluded from comparisons due to their failure to enter SOREM. Additional analyses were performed using the mentation report as the unit of statistical analysis to control for type II errors. The analyses were performed using SPSS (SPSS, Inc, Chicago, IL) and StatView statistical software (SAS Institute Inc, Cary, NC).

1Only

13 of the 15 narcoleptic subjects had reports from daytime SOREM sleep; REM, rapid eye movement; SOREM, sleep-onset REM

In order to minimize the possible influence of fragmented sleep on mental activity, awakenings were performed immediately upon spontaneous arousal or transition to NREM if the awakening criteria were otherwise met. Awakenings were not performed if the last minutes of the REM record had dissociated features from NREM or waking, such as spindles, alpha activity, or a combination of tonically active EMG and a lack of REMs (for an example, see Broughton et al’s previously published work34). Upon being awakened, participants wrote down their preawakening conscious experiences in as much detail as possible; they then indicated the type and intensity of emotion on all appropriate report lines. In addition, participants rated their level of self-reflective awareness and of experienced control over the course of the mental activity and chose a photograph from a photo board that best matched the vividness of their most vivid visual hallucination. (We have previously described these procedures in greater detail.28)

RESULTS The current study investigates three questions. First, what are the differences, if any, in the REM emotion of narcoleptic and normal subjects? Second, how does the emotion profile differ between SOREM and nighttime REM within the narcolepsy group? Third, do REM emotions vary in conjunction with the fragmentation level of REM within and between the subject groups? As shown below, the analysis revealed significant findings for each of these questions. Prevalence of Emotions Two-hundred and eleven instances of emotion were noted in 113 (85%) of the 133 REM reports from the 15 narcoleptic subjects. Of these, 137 were reported from SOREM (mean incidence rate = 2.2 dif-

Table 3—Emotions in nighttime rapid-eye-movement (REM) sleep and sleep-onset REM (SOREM) in narcoleptics Emotion type and intensity*

REM dream element 1

I was sleeping in my bedroom when H came in. I was irritated because I couldn't sleep and I considered getting up. Then I flew out of the door on a cookie plate or something. I flew around near the house. O and H were watching and they envied me.

Irritation 1 Joy/elation 3

2

I had to take this "test" and walk from one house to another between each "part." The second time I was walking took a long time because a woman who passed by was quite interested in my stuff, which was on a table outside. I went back and gave her a slap on her fingers and said that if she touched my stuff again, I would call the police. When I came down to the house where the experimenter was, a kind of hotel, I had to do a little fashion show with the clothes I was wearing.

Irritation 3

3

There were two people in the room that I knew, both of them cared for me and helped me. It felt good. One of them laid down in my bed next to me - and we talked about me perhaps being pregnant. At this point it became quite "cosy" (erotic).

Shame 2 Love/ Eroticism 2

4

I'm lying on my back, dipping in a lake in Italy or something. I have a large bottle of wine in my arm pit, the kind of bottle with a tap on it. I can see my toes pointing boldly up towards a very blue sky.

Joy/elation 3 Surprise 3

5

I was playing with a crow. It was sitting high up on a ledge on a mountain side and it was hitting a ball down to me with its head and I hit it back with mine, and we went on like that. Then I was over in the bed. A fox came and lay down besides me. I covered it with blankets. Then I suddenly got anxious and turned around. My mother and father were not lying there anymore. Where had they gone? (Dad has been dead for 15 years, mother is still alive).

Joy/elation 2

6

I thought I was awake and talked about the dreams and the schemas. First, a tiny worm ate its way through the electrodes and then ate my veins. Then a very tiny whale did the same. Then I "woke up." Both the blankets rose vertically in the room, the top blanket landed beneath the other. I kept shouting "no."

Anxiety/fear 3

7

An Italian man, with the look of an Arab, took over the entire research project. He tried to seduce most of the girls he tested. He laid down in a bed beside mine and clearly wanted something more than to sleep. His friend came in with a large black vibrator. I shouted NO to him and he respected that.

Surprise 2 Anger 2 Anxiety/fear 3 Sadness 3

8

I was at my sister's place. For some reason we needed to borrow a skirt. I found a piece of chocolate under the bed and we ate it. We had to break off the piece that had already been eaten.

Joy/elation 3

9

I was sitting on the subway and two guys were going to wash the lamp holders. I was furious because they didn't do the job properly, and I scolded them for having a poor work ethic.

Anger 3

10

A large hole was dug in the ground outside my house, close to the wall. Nothing about the surroundings was like in real life. Four girls/women were arriving in a car. K, who I knew from before and who is a member of the Narcolepsy Association, was the driver. Everybody in the car were wearing motorcycle "caps" in leather and they were happy and talkative.

Joy/elation 3

The first six report segments are from SOREM and the last four from nighttime (emergent) REM during the night. The reports are translated from Norwegian. *Intensity levels: 1 = low, 2 = medium, and 3 = high.

SLEEP, Vol. 25, No. 7, 2002

726


Joy/elation 3

Anxiety/fear 2

Positive and negative emotion within narcoleptic SOREM. Fiftythree (39%) of the 137 discrete incidents of emotion in SOREM were classed as positive and 69 (50%) were classed as negative, with the remaining 15 (11%) being instances of surprise. This difference in incident rates between negative and positive emotion was not significant (binominal test, n = 122, P = .17). In terms of report prevalence, positive emotion was found in 65% of the reports and negative emotion in 53% (Table 4), and mean report prevalences for positive and negative emotion did not differ significantly (paired t test, t(12) = .8; P = .47).

Table 4—Prevalence of positive and negative emotions Narcoleptic SOREM Narcoleptic nighttime REM Normal nighttime REM *Percentage

Positive emotions (%) Negative emotions (%) 53 65* 49 31 49 42

of all reports

ferent emotions per report) and 74 were reported from nighttime REM (mean incidence rate = 1.0 per report). Examples from reports with emotions are given in Table 3. Although emotion was found to occur in most REM dreams in both the normal and the narcoleptic group, SOREM in narcoleptic subjects contained emotion significantly more often (mean subject report prevalence = 92%) than did nighttime REM in either narcoleptic (81%, paired t test, t(12) = 2.3, P = .044) or normal subjects (76%, unpaired t test, t(20) = 2.2, P = .041). The overall prevalence of emotion was not significantly different for nighttime REM in the two subject groups (unpaired t test, t(22) = 0.7, P = .49). An enhancement of emotion was thus found in the SOREM dreams of narcoleptics. As can be seen in Table 4, the higher emotion prevalence in SOREM reflected higher levels of both positive and negative emotion. In the control group, positive and negative emotions occurred in a balanced manner, as previously described.27 Similar results were seen for the narcoleptics, and these findings are described below.

Positive and negative emotion within nighttime narcoleptic REM. Of the 74 incidents of emotion in nighttime REM reports from narcoleptic subjects, 33 (45%) were classified as positive and 31 (42%) were classified as negative, rates that were not significantly different (binominal test, n = 74, P = .90). In terms of prevalence, positive emotion was found in 49% of reports and negative emotion in 31% (Table 3), which again did not differ significantly (paired t test, t(14) = .96; P = .96). Based on these general prevalence findings for positive and negative emotions, we next analyzed the prevalence of discrete emotions, focusing on the relative distribution of discrete emotions within each REM condition in narcoleptic subjects and on the differences in the prevalence of the most common emotions between narcoleptic and normal subjects. Prevalence of Discrete Emotions The discrete emotions that were most commonly seen in narcoleptic REM were the same as those seen in normal REM (Figure 1). In both REM conditions, joy/elation, anxiety/fear, anger, and surprise showed the highest prevalence rates in narcoleptics. We first analyzed the relative prevalence of these four discrete emotions within SOREM and nighttime REM in narcoleptic patients.

Figure 1—Prevalence of discrete emotions

Distribution of discrete emotions in narcoleptic nighttime REM. In the nighttime REM of narcoleptic patients, joy/elation had close to three times the mean report prevalence rate per subject as did anxiety/fear, anger, and surprise (Figure 1). This primacy of joy/elation was significant compared to both anxiety/fear (Wilcoxon signed rank test, z(14) = 2.0; P = .047) and anger (z(14) = 2.1; P = .036), with a similar tendency compared to surprise (z(14) = 1.8; P = .077). No difference was found between anxiety/fear, anger, and surprise. Thus, joy/elation was the leading emotion in nighttime narcoleptic REM, paralleling findings previously reported for the control group.27 Distribution of discrete emotions in narcoleptic SOREM. In contrast to nighttime REM, narcoleptic SOREM showed no significant difference between the report prevalence of joy/elation (45% of reports) and anxiety/fear (36%) with paired t tests on subject means, P > .25). However, joy/elation was still significantly more common than surprise (Wilcoxon signed rank test, z(12) = 2.3; P = .021), and occurred twice as often as anger (45% vs 24% of reports). The latter difference was significant only when tested on a per-report basis (z(41) = 12.3; P = .024) and not when tested on a per-subject basis (P> .5). In addition, and again in contrast to nighttime REM, anxiety/fear tended to be more prevalent than surprise in SOREM (Wilcoxon signed rank test, z(11) = 1.7; P = .086). These findings indicate, first, that joy/elation and anxiety/fear are the two most common emotions in narcoleptic SOREM and, second, that the prevalence of anxiety/fear in SOREM was enhanced compared to both control and narcoleptic nighttime REM. Differences in emotion types between REM conditions and subject groups. The above findings indicate that at least anxiety/fear should be more prevalent in narcoleptic SOREM than in nighttime REM in either subject group, which is in line with the results shown below. In addition, surprise also differed between narcoleptic and normal subjects, although in a direction opposite to that seen for anxiety/fear.

Percentage of reports (report prevalence) containing each discrete emotion. Error bars are SEM. The “stress-despair” category included disappointment, indecisiveness, discomfort, and hopelessness in addition to stress and despair. The emotion data for the normal group are presented in more detail in Fosse et al.27 REM, rapid eye movement; SOREM, sleep-onset REM

SLEEP, Vol. 25, No. 7, 2002

727


Figure 2—Emotion intensity levels (A) Emotion intensity level in mentation reports. (B) Intensity of emotion incidents

(A) Percentage of reports (prevalence) with no emotions and with a maximum intensity rating of low, medium, and high in the three rapid-eye-movement (REM) categories. (B) Percentage of all incidents of emotions with low, medium, and high intensity scores. REM, rapid-eye-movement; SOREM, sleep-onset REM

Anxiety/fear. The report prevalence of anxiety/fear was 170% higher in narcoleptic SOREM than in nighttime REM in these patients (paired t test on subject averages, t(12) = 2.1, P = .052), and 220% higher than in nighttime REM in the normal group (unpaired t test, t(20) = 1.9, P = .074; when tested on a per-report basis using the Mann-Whitney test, z = 3.1; n = 150; P = .002). No difference in the prevalence of anxiety/fear was found for nighttime REM in the two subject groups (P = .92). Joy/elation. In contrast to anxiety/fear, the prevalence of joy/elation was only 20% to 25% higher in SOREM in narcoleptic patients than in nighttime REM in either subject group, a difference which was not significant (P > .60). Surprise. In nighttime REM, surprise occurred twice as often in controls (25% of the reports) as in narcoleptics (13%). This difference tended toward significance [non-parametric Mann-Whitney U-test, z(23) = 1.6, n = 24, P = .097; on a per-report basis, t(157) = 1.9, P = .061]. Other emotions. No significant differences between the REM conditions or subject groups were found for anger or for the less frequent emotions of love/eroticism, shame, and sadness. However, the prevalence of anger was twice as high in SOREM (mean subject prevalence of 24%) as in nighttime REM (12%) in narcolepsy, with intermediary values for the normal group (17%).

(paired t test, t(12) = 2.5; P = .031) or control subjects (unpaired t test, t(16) = 2.7; P = .017) (Fig. 2A and 2B). No difference in overall emotion intensity level was seen for the nighttime REM of the two groups (P = .91). Intensity of Discrete Emotion Types As shown below, the higher intensity level of emotion in narcoleptic SOREM was found to be reflected in higher intensity ratings for each of anxiety/fear and joy/elation. Moreover, a similar although not equally strong difference in the intensity of these emotions was seen for nighttime REM in the two groups. Compared to normal nighttime REM, joy/elation was more intense in both narcoleptic SOREM (unpaired t test, t(17) = 2.7; P = .015) and narcoleptic nighttime REM (t(19) = 1.7; P = .094) (Figure 3). A difference in the intensity of joy/elation between the nighttime REM of the two groups was evident also when tested on a per-report level (t(56) = 2.5; P = .014). Anxiety/fear was found to be more intense in narcoleptic SOREM than in the nighttime REM of the control group (t(13) = 3.6; P = .003). When comparing anxiety/fear in nighttime REM in the two groups, a limiting factor for the statistical analysis was that only 7 narcoleptic and 6 normal subjects reported any incidence of this emotion. However, for

SUMMARY Positive and negative emotions were found to occur in a balanced manner in both SOREM and nighttime REM in narcoleptics, which is in line with that seen for the control group. On the other hand, narcoleptic SOREM had higher prevalence rates for both positive and negative emotions compared to nighttime REM in either group. Among the discrete emotion types, joy/elation was the most common in both SOREM and nighttime REM in narcoleptic patients, paralleling that seen in the normal group. However, specific for SOREM in narcoleptic patients was a relatively strong increase in the prevalence of anxiety/fear. In contrast, surprise was indicated to be less common in the narcoleptic group when nighttime REM was compared to normal REM.

Figure 3— Intensity of discrete emotions

Intensity of Emotions The difference in the prevalence of emotion in narcoleptic and normal REM was complemented by differences in the intensity level of emotion. We first compared the overall emotion intensity level in the reports from the different REM conditions and subject groups by use of the highest intensity score given to any emotion type in each report. Thus measured, reports from narcoleptic SOREM had higher maximum intensity levels than did reports from the nighttime REM of either narcoleptic SLEEP, Vol. 25, No. 7, 2002

Average intensity scores per report for each emotion. Only those reports that contained the respective emotion are included. Intensity scores: 1 = low, 2 = medium, and 3 = high. Error bars are SEM. Intensity data for the normal group have been previously described.27 REM, rapid-eye-movement; SOREM, sleep-onset REM

728


these subjects, the mean intensity level of anxiety/fear was 92% above baseline (score of 1 = mild intensity) in narcoleptic subjects and only 31% above baseline in the normal group. This difference was not significant when tested on a per-subject level (unequal variances, t(8.5) = 1.5, p = .16) but approached significance when tested on a per-report basis (t(17) = 1.8, p = .096). The data did not allow for tests of differences in the intensity of discrete emotion types between the SOREM and nighttime REM of narcoleptic subjects. Overall, REM emotions were the most intense in narcoleptic SOREM, followed by nighttime REM in narcoleptic patients, with the lowest intensity levels—in particular of anxiety/fear and joy/elation–seen in the REM sleep of controls.

to SOREM and REM time prior to awakening. Thus, the elevated level of emotion in narcoleptic SOREM was not specifically linked to the closeness of the reports to the preceding wake period. Emotion and REM Sleep Fragmentation Our next question was whether the difference in emotion that was seen between the subject groups reflected differences in the stability of the REM state. However, as previously reported,28 no difference in REM fragmentation (the inverse of REM stability) was seen between groups or between SOREM and nighttime REM in narcoleptic subjects, whether analyzed by subject (unpaired and paired t tests, P > .4) or by reports (1way ANOVA across the three REM categories, P > .7). These findings are likely to reflect the particular awakening protocol of the current study, with subjects being awakened immediately when spontaneous arousal was detected and the awakening criteria were otherwise met, and are not likely to reflect a true lack of difference in the stability of REM between narcoleptic and normal subjects. Although REM fragmentation did not appear to underlie the variation in emotion between subject groups and REM conditions, fragmentation could still be differentially associated with REM mentation in individual narcoleptic and normal subjects. We thus asked whether subjects with high emotion scores had particularly high (or low) degrees of REM fragmentation. The emotion measure used in these analyses was the total emotion score for each report, with analysis performed on the subject level using average report scores. For several analyses, this emotion measure did not exhibit a normal distribution across subjects. Since the fragmentation measure instead showed a normal distribution, in the analysis below, fragmentation was consistently used as the dependent variable and emotion was used as the independent variable. As shown below, a relation between REM fragmentation and emotion was evident in the narcoleptic but not the control group.

SOREM Emotions: The Effect of Time Spent in Sleep Since the hypnagogic hallucinations of the wake/sleep transition period in narcoleptic patients have been reported to be highly emotional and often unpleasant, we asked whether the elevated emotions seen in narcoleptic SOREM could be a function of the proximity to the wake state. If this were indeed the case, a lower level of emotion within SOREM might be expected for reports derived after longer periods of sleep following sleep onset. As shown below, this hypothesis was not supported by the data. Eleven narcoleptic patients were included who had at least two SOREM reports obtained from different times into sleep since sleep onset. Nine of the 11 narcoleptic patients had a higher total emotion score in the late-sleep-onset condition than in the early condition (binomial test assuming equal probability for increase and decrease, n = 11, P = .065). The increase from early to late sleep onset was significant for negative emotion (paired t test, t(1, 10) = 2.9; p = .017) but not for positive emotion (p = .57). As can be seen in Figure 4, when all 62 SOREM reports from the 13 subjects were included, positive and negative emotions, as well as each of the most frequent emotion types belonging to these categories, tended to increase with elapsed time in sleep before awakening. Similar results were seen separately for both sleep time prior

Relation between REM fragmentation and emotion within each subject group. In the control group, no emotion category (total, positive, negative) and no discrete emotion type showed any significant relationship to fragmentation (df = 7, P > .30). In the narcolepsy group, subjects’ scores on the category of positive emotion was inversely related to their REM-fragmentation level (Pearson correlation coefficient, r(12)= -.51; P = .091). Moreover, the 6 narcoleptic subjects with the highest scores on positive emotion had a significantly lower fragmentation level than did the remaining 6 subjects (unpaired t test, t(11) = 2.9; P = .015). A similar difference was seen separately for SOREM (t(12) = 2.6; P = .027) and for nighttime REM (t(13) = 1.9, P = .087).Thus, the narcoleptic patients with the highest scores on positive emotion had the most stable (least fragmented) REM sleep. In contrast, no relation to fragmentation level was seen for the overall negative emotion category in the narcoleptic subjects (P > .3). As with the general category of positive emotion, joy/elation was inversely related to REM fragmentation across the narcoleptic subjects (r(12) = -.55, P = .064). The 6 patients with the highest scores on joy/elation had a significantly lower fragmentation level than those with low scores on joy/elation (t(10) = 2.6, P = .027). This inverse relation was evident also for SOREM considered separately (r(12) = -.55, p = .062; t(10) = 2.9, P = .015) but not for nighttime REM (P > .70). In contrast to the inverse relation to fragmentation that was seen for joy/elation, a positive relation to fragmentation was indicated for anxiety/fear. The 7 narcoleptic subjects with the highest scores on anxiety/fear had twice the fragmentation level as compared to the average of the remaining 5 subjects, although this difference failed to reach adequate statistical significance (t(11) = 2.0, P = .074). No other discrete emotion type was significantly related to the fragmentation of REM within the narcolepsy group. Thus, a significant inverse relation was found for positive emotion and, in particular, joy/elation to the fragmentation of REM in narcolepsy being the most evident in SOREM. In contrast, subjects with more anxious and fearful dreams tended to have a higher level of REM fragmentation.

Figure 4—SOREM emotion in narcolepsy and time spent in sleep since sleep onset

Changes in emotion with time in sleep are based on the total emotion scores for the sleep-onset rapid eye movement (SOREM) reports (n = 62). The three conditions represent total elapsed time since sleep onset, including both latency from SOREM (mean of 2.0 min) and time spent in SOREM sleep. Each of the three conditions contained approximately 20 SOREM reports. On a per-subject level, negative emotions showed a significant increase with elapsed time spent in sleep (P = .017).

SLEEP, Vol. 25, No. 7, 2002

729


in narcolepsy.7-12,38,39 Animal studies have revealed a basic hypersensitivity of forebrain cholinergic activity in narcolepsy, accompanied by alterations in dopaminergic function in the ventral tegmental area, the amygdala, the brainstem, and the caudate.2,14 These changes appear to be directly involved in the triggering of cataplexy by positive and exciting emotion.2,13 Since dopamine as well as interactions between dopamine and acetylcholine are known to support positive emotion,7-12,38,39 alterations in these systems might contribute to the increase in positive emotions seen during narcoleptic REM and might explain the specific relation between REM-sleep instability and positive emotions that we report here. In contrast, abnormality in noradrenergic activity in narcolepsy is likely to modulate emotions such as anxiety and fear. The noradrenergic locus coeruleus, in addition to contributing to arousal and focused attention, is known to be a key structure for alarm reactions, exhibiting high levels of engagement in threatening situations associated with anxiety and fear.40 The constraints that appear to be placed upon anxiety and fear during normal REM27 are plausible consequences of the normally strong noradrenergic blockade during this stage. The higher fragmentation of REM in narcoleptic patients is proposed to reflect an ineffective locus coeruleus blockade caused by the more basic deficiencies in hypothalamic hypocretin activity,41 perhaps in combination with a less stable cholinergic neuromodulation.42 As a result, one would expect the observed increase in locus coeruleus-mediated emotions of fear and anxiety, driven by the action of noradrenaline at several levels of the neuraxis—ranging from the brainstem to the prefrontal cortex.

DISCUSSION Emotional experience was found to be more prevalent and intense during SOREM in narcoleptic patients than during nighttime REM in either narcoleptic or normal subjects. A higher prevalence rate and intensity of positive as well as negative emotions were found in SOREM, with the most pronounced difference seen for anxiety/fear. For nighttime REM, narcoleptic subjects had a trend toward a lower prevalence rate for surprise but more intense feelings of joy/elation and anxiety/fear. Narcoleptic patients with high levels of joy/elation and of positive emotion in general had a less fragmented SOREM sleep. We know of no other controlled studies in the literature that measured the prevalence or intensity of discrete emotions in narcoleptic REM or compared these features with those of normal subjects. It is nonetheless noteworthy that single-case and clinical-study reports have asserted that dreaming during daytime naps and shortly after nighttime sleep onset in narcolepsy is particularly unpleasant and often frightening.15-18 These reports are in line with our finding that dreaming is more emotional and anxious during narcoleptic SOREM than during normal REM. Comparison of the New Results to Previous Findings Two instrumental awakening studies of REM mentation in narcolepsy were found in the literature, but these studies reported only general characteristics of the emotional experience. Vogel20 found that unpleasant and pleasant dreams occurred in a balanced fashion during narcoleptic SOREM, and Hishikawa et al19 measured only unpleasant emotions and found these to be experienced more often during SOREM than during early nighttime (emergent) REM. Both findings are consistent with the present results. Hishikawa et al19 found SOREM dreams to be more often unpleasant when accompanied by an enhanced level of reflective awareness and a sensation of motor paralysis. This combination of features was seen most often early in SOREM periods that developed immediately after sleep onset. Based on these observations, Hishikawa et al suggested that when unpleasant hallucinatory mentation takes place during the hypnagogic sleep-onset period in narcolepsy, altered biologic aspects of the REM state itself are likely to be involved.19 In the current study, awakenings were performed later into SOREM sleep than in Hishikawa et al’s study and were less likely to lead to recall of mental activity from the transition period between waking and sleep. Nonetheless, we found that emotion was reported at high levels throughout SOREM, with an increase seen in negative emotion as time spent in sleep since sleep onset increased. Enhanced emotion thus appears to characterize not only the hypnagogic wake-sleep (REM) transition period, but also the entire REM period when occurring shortly after sleep onset in narcoleptic patients.

REM Sleep Fragmentation and Emotion The propensity for positive emotions to trigger the REM paralysis of cataplexy implies a connection between the positive emotion systems of the brain and the underlying REM abnormalities of narcolepsy. It was, therefore, noteworthy that high levels of positive emotions such as joy and elation were associated with less-fragmented REM sleep in these patients, a relation that was particularly evident during SOREM. No such relation was seen in the control group, and no similar inverse relation to fragmentation was indicated for negative emotion in the narcoleptic subjects. Increased REM-sleep fragmentation is a typical reflection of the instability of REM in narcolepsy34 and can be conceived of as the mirror image of the REM paralysis of cataplexy. Whereas cataplexy is marked by a decreased muscle tone, fragmentation of REM sleep is reflected in an increased muscle tone and accompanying electroencephalographic changes of arousal or NREM. If the relation between positive emotion and motor paralysis that is seen during cataplexy were retained during SOREM, more-stable muscle paralysis would correlate with increased positive dream emotions in narcoleptic patients, as in fact was found here. The narcoleptic patients with the most-fragmented REM sleep had not only lower levels of positive emotions but also higher levels of anxiety/fear. This increase in fragmentation for subjects with high levels of anxiety/fear did not reach statistical significance in our data set (P = .074), but if confirmed by subsequent research, it would indicate the existence of a functional relationship between the basic processes of REM fragmentation and the appearance of anxiety in REM mentation.

REM Sleep in Narcolepsy: A Context for Altered Emotion According to the activation-synthesis hypothesis,35-37 sleep mentation is controlled at the level of basic neurobiologic processes (activation) as well as by higher cognitive and emotional function (synthesis), which are presumably modified by these neurobiologic processes. In narcoleptic REM, each of these fundamental levels has been altered, and we propose that this could explain much of the overall difference in emotion seen between narcoleptic and normal subjects.

Thoughts, Hallucinations, and Emotions In addition to the contributions from bottom-up biologic processes, top-down synthetic activity36 is likely to influence emotion in REM. The REM-dream emotions are embedded in the hallucinations and thoughts of the ongoing dream27 (Table 3), similar to the percepts and thoughts that contextualize much waking emotion.43 Alterations in hallucination and thought in narcoleptic REM might thus contribute to the changes observed in emotion. The perceptual scenario of REM dreams is changed in distinctive ways in narcoleptic patients. In the current reports from SOREM and nighttime REM, narcoleptic patients less often reported the type of

REM Emotion and Monoaminergic/Cholinergic Imbalance We propose that bottom-up biologic processes contribute to both the normative profile of emotion that is seen during REM27,35,36 and to the altered emotion profile of narcolepsy. Central to the symptoms of narcolepsy is a monoaminergic-cholinergic imbalance that reflects abnormalities of hypothalamic hypocretin function.2 Abnormalities in cholinergic and dopaminergic neuronal processes in the extended amygdala–basal-forebrain region, known to constitute important components of the reward system of the brain, become altered SLEEP, Vol. 25, No. 7, 2002

730


bizarre improbabilities in actions, characters, and objects that are most typical for normal REM.28 Moreover, in nighttime narcoleptic REM, visual scenarios were less vivid and the dream reports were only half as long as in the normal group.28 Such a decreased level of bizarreness and visual vividness could be expected to lead to a decrease in emotion, especially a lower prevalence rate of surprise, which was found in nighttime narcoleptic REM. Changes in the nature of thoughts might likewise affect emotion in REM. The enhanced level of reflective awareness in narcoleptic REM19,20,28 might permit the dreamer to better identify hallucinated events as strange, bizarre, and even frightening, resulting in a higher propensity to initiate emotion based on reasoned expectations.

12. 13.

14. 15.

CONCLUSION 16.

The sleep disorder narcolepsy intensifies emotion during REM, with the most pronounced change seen for anxiety/fear, followed by joy/elation. We propose, first, that the propensity to feel anxiety/fear is increased during narcoleptic REM due to inadequate blockade of noradrenergic activity in the locus coeruleus and, second, that the propensity to experience positive emotion is increased due to cholinergic hypersensitivity and dopaminergic dysregulation. At the end of this cascade of events, REM emotion in narcoleptic, as well as normal, subjects is likely determined by cortical regions that support thought, perception, and emotion. Neither neurobiology nor psychology alone can provide an adequate explanation of the alterations seen in REM dream emotion in narcolepsy. Instead, an integrated cognitive neuroscience perspective needs to be adopted to test hypotheses of conjunct changes in cognitive-emotional function, regional brain activation, and neurochemical brain modulation in narcoleptic REM.

17. 18. 19. 20. 21. 22.

ACKNOWLEDGEMENT The authors want to acknowledge Khalil Kayed for his general contributions to this research.

23.

REFERENCES

24.

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

Overeem S, Mignot E, Gert van Dijk J, Lammers GJ. Narcolepsy: clinical features, new pathophysiologic insights, and future perspectives. J Clin Neurophysiol 2001;18:78-105 Mignot E. Pathophysiology of narcolepsy. In: Kryger MH, Roth T, Dement WC, eds. Principles and practice of sleep medicine. Philadelphia: WB Saunders Co; 2000:663-75 Hishikawa Y, Shimizu T. Physiology of REM sleep, cataplexy, and sleep paralysis. Adv Neurol 1995;67:245-71 Mignot E, Thorsby E. Narcolepsy and the HLA system. N Engl J Med 2001;344:692 Guilleminault C. Cataplexy. In: Guilleminault C, Dement WC, Passouant P, eds. Narcolepsy. New York: Spectrum; 1976:125-44 Lammers GJ, Overeem S, Tijssen MA, van Dijk JG. Effects of startle and laughter in cataplectic subjects: a neurophysiological study between attacks. Clin Neurophysiol 2000;111:1276-81 Drevets WC, Gautier C, Price JC, et al. Amphetamine-induced dopamine release in human ventral striatum correlates with euphoria. Biol Psychiatry 2001;49:81-96 Mirenowicz J, Schultz W. Preferential activation of midbrain dopamine neurons by appetitive rather than aversive stimuli. Nature 1996;379:449-51 Kapp BS, Supple WF Jr, Whalen PJ. Effects of electrical stimulation of the amygdaloid central nucleus on neocortical arousal in the rabbit. Behav Neurosci 1994;108:81-93 Holland PC, Han JS, Gallagher M. Lesions of the amygdala central nucleus alter performance on a selective attention task. J Neurosci 2000;20:6701-6 Han JS, Holland PC, Gallagher M. Disconnection of the amygdala

SLEEP, Vol. 25, No. 7, 2002

25. 26. 27. 28. 29. 30. 31.

32. 33. 34.

731

central nucleus and substantia innominata/nucleus basalis disrupts increments in conditioned stimulus processing in rats. Behav Neurosci 1999;113:143-51 Chiba AA, Bucci DJ, Holland PC, Gallagher M. Basal forebrain cholinergic lesions disrupt increments but not decrements in conditioned stimulus processing. J Neurosci 1995;15:7315-22 Reid MS, Tafti M, Nishino S, Sampathkumaran R, Siegel JM, Mignot E. Local administration of dopaminergic drugs into the ventral tegmental area modulates cataplexy in the narcoleptic canine. Brain Res 1996;733:83-100 Nishino S, Mignot E. Pharmacological aspects of human and canine narcolepsy. Prog Neurobiol 1997;52:27-78 Broughton RJ. Neurology and dreaming. Psychiatr J University Ottawa 1982;7:101-10 Montplaisir J. Disturbed nighttime sleep. In: Guilleminault C, Dement WC, Passouant P, eds. Narcolepsy. New York: Spectrum; 1976:43-56. Parkes JD, Baraister M, Marsden CD, Asselman P. Natural history, symptoms and treatment of the narcolepsy syndrome. Acta Neurol Scand 1975;52 Roth B, Bruhova S: Dreams in narcolepsy, hypersomnia and dissociated sleep disorders. Exp Med Surg 1969;27:187-209.. Hishikawa Y, Koida H, Yoshino K, et al: Characteristics of REM sleep accompanied by sleep paralysis and hypnagogic hallucinations in narcoleptic patients. Waking Sleeping 1978;2:113-23 Vogel G. Mentation reported from naps in narcoleptics. In: Guilleminault C, Dement WC, Passouant P, eds. Narcolepsy. New York: Spectrum; 1976: 161-8 Schwartz BA. Electroencephalographic periodicity in sleeping sickness: modulation by stages of sleep. Rev Electroencephalogr Neurophysiol Clin 1971;1:70-2 Bremner JD, Krystal JH, Southwick SM, Charney DS. Noradrenergic mechanisms in stress and anxiety. Synapse 1996;23:39-51 Salzman C, Miyawaki EK, le Bars P, Kerrihard TN. Neurobiologic basis of anxiety and its treatment. Harvard Rev Psychiatr 1993;1:197-206. Nelson JC. A review of the efficacy of serotonergic and noradrenergic reuptake inhibitors for treatment of major depression. Biol Psychiatry 1999;46:1301-8 Faull KF, Zeller-DeAmicis LC, Radde L, et al. Biogenic amine concentrations in the brains of normal and narcoleptic canines: current status. Sleep 1986;9:107-10 Miller JD, Faull KF, Bowersox SS, Dement WC. CNS monoamines and their metabolites in canine narcolepsy: a replication study. Brain Res 1990;509:169-71 Fosse R, Stickgold R, Hobson JA. The mind in REM sleep: reports of emotional experience. Sleep 2001;24:947-55 Fosse R. REM mentation in narcoleptics and normals: an empirical test of two neurocognitive theories. Conscious Cogn 2000;9:488-509 Fosse R. REM mentation in narcoleptics and normals. Conscious Cogn 2000;9:514-5 Fosse R. REM mentation in narcoleptics and normal controls: an empirical test of two neurocognitive theories. Sleep 2000;23:A176-7 Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Los Angeles: UCLA Brain Information Service Brain Research Institute; 1968 Merritt JM, Stickgold R, Pace-Schott E, Williams J, Hobson JA. Emotion profiles in the dreams of men and women. Conscious Cogn 1994;3:46-60. EEG arousals: scoring rules and examples: a preliminary report from the Sleep Disorders Atlas Task Force of the American Sleep Disorders Association. Sleep 1992;15:173-84 Broughton R, Valley V, Aguirre M, Roberts J, Suwalski W, Dunham W. Excessive daytime sleepiness and the pathophysiology of REM Emotions in Narcolepsy—Fosse et al

35. 36. 37. 38.

39. 40.

41. 42. 43.

narcolepsy-cataplexy: a laboratory perspective. Sleep 1986;9:20515 Hobson JA, McCarley RW. The brain as a dream-state generator: an activation-synthesis hypothesis of the dream process. Am J Psychiatry 1977;134:1335-48 Hobson JA: The dreaming brain. New York: Basic Books; 1988 Hobson JA, Pace-Schott EF, Stickgold R. Dreaming and the brain: toward a cognitive neuroscience of conscious states. Behav Brain Sci 2000;23:793-842 Kapp BS, Whalen PJ, Supple WF, Pascoe JP. Amygdaloid contributions to conditioned arousal and sensory information processing. In: Aggleton JP, eds. The amygdala: neurobiological aspects of emotion, memory, and mental dysfunction. New York: Wiley-Liss; 1992:229-54 Han JS, McMahan RW, Holland P, Gallagher M. The role of an amygdalo-nigrostriatal pathway in associative learning. J Neurosci 1997;17:3913-9 Tanaka M, Yoshida M, Emoto H, Ishii H. Noradrenaline systems in the hypothalamus, amygdala and locus coeruleus are involved in the provocation of anxiety: basic studies. Eur J Pharmacol 2000;405:397-406 Dauvilliers Y, Neidhart E, Lecendreux M, Billiard M, Tafti M. MAO-A and COMT polymorphisms and gene effects in narcolepsy. Mol Psychiatry 2001;6:367-72 Kilduff TS, Peyron C. The hypocretin/orexin ligand-receptor system: implications for sleep and sleep disorders. Trends Neurosci 2000;23:359-65 Izard CE. Four systems for emotion activation: cognitive and noncognitive processes. Psychol Rev1993;100:68-90

SLEEP, Vol. 25, No. 7, 2002

732
