Decision Making in Narcolepsy with Cataplexy

Abstract

Objectives:

To investigate decision-making and addictive behaviors in narcolepsy-cataplexy (NC). NC is caused by the loss of hypothalamic neurons that produce hypocretins. The hypocretin system plays a crucial role in sleep, wakefulness, and energy homeostasis, and is also involved in emotion regulation, reward processing, and addiction.

Setting:

Academic sleep center.

Patients:

23 subject with NC and 23 matched healthy controls.

Design:

We used the Iowa Gambling Task (IGT) to assess decision making under ambiguity condition based on emotional feedback processing and the Game of Dice Task (GDT) to assess decision making under risk condition. All participants underwent a semi-structured psychiatric interview and completed the Beck Depression Inventory-II and the UPPS Impulsive Behavior Scale. Patients underwent one night of polysomnography followed by an MSLT, with neuropsychological evaluation performed between MSLT sessions.

Measurements and Results:

NC patients had higher depressive symptoms and showed a significant lack of perseverance. One NC patient had a past history of drug dependence. NC patients also exhibited selective reduced IGT performance and normal performance on the GDT. No clinical or polysomnographic characteristics were associated with increased sensitivity to reward and/or decreased sensitivity to punishment. However, lack of perseverance in NC patients was associated with disadvantageous decision making on the IGT.

Conclusion:

We demonstrated a lack of perseverance and a selective reduced performance on decision making under ambiguity in NC in contrast to normal decision making under explicit conditions. Patients with narcolepsy-cataplexy may opt for choices with higher immediate emotional valence, regardless of higher future punishment, to compensate for their reduced reactivity to emotional stimuli.

Citation:

Bayard S; Abril B; Yu H; Scholz S; Carlander B; Dauvilliers Y. Decision making in narcolepsy with cataplexy. SLEEP 2011;34(1):99-104.

NARCOLEPSY WITH CATAPLEXY (NC) IS A PRIMARY SLEEP-WAKE DISORDER CHARACTERIZED BY EXCESSIVE DAYTIME SLEEPINESS (EDS) AND CLINICAL manifestations of dissociated REM sleep, including cataplexy, hypnagogic hallucinations, sleep paralysis, and REM sleep behavior disorder (RBD). The pathognomonic symptom of NC is cataplexy, a condition in which short episodes of muscle tone loss are triggered by strong emotions.¹ Recent advances in its pathophysiology show that deficient hypocretin transmission causes NC, evidenced by a marked decrease found in hypocretin-1 levels in the cerebrospinal fluid (CSF) and in the number of hypocretin neurons in postmortem brain tissues.^2,3

The hypocretin neurons are exclusively located in the lateral hypothalamus and project widely throughout the central nervous system.^4–6 They are important in regulating arousal, motivation, and stress states.⁷ The hypocretin neurons also send direct and indirect projections to dopaminergic regions, i.e., the ventral tegmental area (VTA), the nucleus accumbens, and the amygdala, which also suggests an important role in reward processing and drug abuse behavior.^8,9 In rodents, intra-VTA infusions of hypocretin have been shown to reinstate drug-seeking behavior.¹⁰ More recently, cocaine-induced plasticity in the VTA dopaminergic neurons was shown to depend on hypocretin efferents.¹¹ In vivo administration of a hypocretin-1 receptor antagonist blocked cocaine-induced potentiation of excitatory currents in VTA dopamine neurons. In addition, hypocretin/orexin-deficient mice narcoleptic models (i.e. orexin knock-out and orexin/ataxin-3 transgenic mice) were less responsive in developing amphetamine-induced locomotor sensitization test, a behavioral paradigm that normally involved a progressive increase in the locomotor-activating effects of addictive drugs.¹² All these studies underline the important role of hypocretin neurons in reward processing and addictive behaviors.

To our knowledge, no studies on human narcolepsy have considered changes in sensitivity to reward and punishment. Given that subjects with altered feedback sensitivity showed decision-making problems, and that impulsive decision making was a condition of intolerance to delay-of-reward,¹³ we decided to analyze the responses to decision-making paradigms in NC patients free of psychostimulants.

The aims of the present study were to (1) assess the presence of substance use disorder, pathological gambling, depression, and impulsive behavior using a standard evaluation instrument; (2) measure performance on decision-making paradigms under ambiguity and risk conditions in patients with NC compared to sex- and age- matched normal controls; and (3) study the impact of clinical and polysomnographic variables on decision-making performance.

SUBJECTS AND METHODS

Subjects

Twenty-three adult patients with NC (9 females and 14 males, aged 18-76 years) were included in this study. Ten patients were drug-naïve, and 13 had stopped medication to participate. Medication included modafinil (n = 12), tricyclic antidepressant (n = 2), selective serotonin reuptake inhibitor (n = 1), selective serotonin norepinephrine reuptake inhibitor (n = 1), and sodium oxybate (n = 1). They were examined at the Sleep Disorders Clinic of the University of Montpellier. The NC diagnosis was made based on ICSD-2 criteria,¹⁴ including the presence of EDS and clear-cut cataplexy, HLA DQB1*0602, and ≥ 2 sleep onset REM periods (SOREMP) during the multiple sleep latency test (MSLT).

None of the patients had any current psychiatric disorder (based on DSM-IV criteria) or other neurological disorders. Subjects with a respiratory event index (apnea index + hypopnea index) > 10 were excluded from the study. Narcoleptic patients did not take psychostimulants, anticataplectic medications, or any other medication known to influence sleep or decision making for at least one month prior to the evaluation.

We recruited 23 sex- and age- matched normal subjects (9 females and 14 males, aged 18-77 years) for the control group. All healthy controls were community-dwelling adults living in Montpellier-France recruited from local associative networks. 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 intake known to influence sleep or decision making. No control had any complaint of excessive sleepiness (Epworth Sleepiness Scale < 10).

All patients and control subjects gave their informed written consent to take part in the study, which was approved by the local ethics committee (Comité de Protection des Personnes Sud Méditerranée IV, Montpellier, France).

Clinical and Neuropsychological Evaluation

All subjects participated in a standardized face-to-face clinical interview and were asked to complete questionnaires and neuropsychological tests. Participants were tested individually in 40-min session. Patients with NC underwent one night of polysomnographic (PSG) recording in the sleep laboratory followed by a MSLT with neuropsychological evaluation performed from 09:00 to 12:00 in between MSLT sessions.

Past and present substance (alcohol and drug) use disorders and pathological gambling were diagnosed according to criteria of the Diagnostic and Statistical Manual of Mental Disorders, 4th edition, Text Revision (DMS-IV-TR) Axis I Disorders (psychiatric disorders), using the Structured Clinical Interview. Self-reported depression severity was assessed with the 21-item Beck Depression Inventory-II (BDI-II).¹⁵ Premorbid IQ was estimated using the French adaptation of the National Adult Reading Test (fNART).¹⁶

All subjects also completed a French version of the UPPS Impulsive Behavior Scale.^17,18 The scale consists of 45 items evaluating 4 different facets of impulsivity, labelled urgency (12 items), (lack of) premeditation (11 items), (lack of) perseverance (10 items), and sensation seeking (12 items).

Iowa Gambling Task (IGT)

We used the IGT to assess decision making under ambiguity.¹⁹ Participants were told that the goal was to win as much fictitious money as possible. The task entailed selecting a series of 100 cards from 4 decks (A, B, C, and D), but participants were not informed of the number of trials. Although they were told that some card decks might be better than others, they did not know which card decks these were. Decks A and B were classified as disadvantageous because the final balance was negative, with high immediate monetary gains but even higher future losses. In contrast, selecting a card from decks C and D produced small gains but smaller unpredictable losses. These selections were therefore classified as advantageous. Before starting the task, participants were instructed to choose cards from any deck and were allowed to switch decks at any time. We calculated an overall net score to analyze IGT performance (advantageous minus disadvantageous card selections). A positive net score indicates more frequent selection from advantageous decks, whereas a negative net score indicates more frequent selection from disadvantageous decks. Additionally, task performance was divided into 5 blocks of 20 card selections. Net scores were calculated for each block to determine the overall task performance profile.

Game of Dice Task (GDT)

We used the GDT to assess decision making under risky conditions. Participants were asked to maximize their fictive starting capital of 1000 € within 18 dice throws, according to the standard protocol.²⁰ Before each throw, subjects had to choose between the different single numbers or a combination of 2, 3, or 4 numbers. Each choice was associated with a gain or loss, depending on the probability of occurrence: 1000 € gain/loss for the choice of a single number (winning probability 1:6), 500 € for 2 numbers (winning probability 2:6), 200 € for 3 numbers (winning probability 3:6), and 100 € for 4 numbers (winning probability 4:6). Participants received visual and oral feedback (gain or loss) on their previous decision, and the change in capital was displayed. To determine the decision-making risk, we classified the choice of 3 or 4 number combinations as “non-risky” (≥ 50% winning probability) and the choice of 1 or 2 numbers as “risky” (< 50% winning probability). We calculated a net score by subtracting the number of risky choices from the number of non-risky choices.

Statistical Analysis

Prior to analysis, all data were examined for normality with the Kolmogorov-Smirnov statistic and for homogeneity of variance with the Levene test. Nonparametric statistics were used to compare GDT variables, which were not normally distributed. Log10 could not normalize scores for which skewness values still exceeded 3.²¹ Parametric statistics were applied to IGT variables for which skewness values fell within the acceptable range (i.e., < 1.0). Groups were compared with the t-test or the Mann-Whitney U test on independent samples. Performance on the IGT was analyzed using a repeated measures analysis of variance with group as the between-subject factor and blocks as the within-subject factor. Bravais-Pearson R and Spearman rho correlations were used to explore the relationships between clinical and polysomnographic variables and the decision-making tasks. The level of significance was α < 0.05. All statistical analyses were performed with SPSS version 16 for Windows.

RESULTS

Results of the demographic, clinical and polysomnographic variables, and HLA typing are summarized in Table 1. NC patients who had lumbar puncture (n = 3) had undetectable CSF hypocretin-1 levels.

Table 1.

Clinical, polysomnographic, and HLA typing data of the 23 unrelated patients with narcolepsy-cataplexy

Demographics			Narcolepsy Characteristics								MSLT Findings
Patient no.	Age	Gender	HLA DR2 (DQB1*0602)	Age at onset	Disease duration	Cataplexy	ESS	SP	HH	RBD	Mean sleep latency, min	SOREM /5
1	21	M	+	20	1	+	22	-	+	+	4	5
2	23	F	+	17	6	+	18	-	-	-	8	3
3	25	M	+	11	14	+	22	+	+	+	7	4
4	31	M	+	18	13	+	18	+	-	+	5	4
5	31	M	+	14	17	+	13	+	+	+	8	3
6	37	M	+	32	5	+	18	+	-	-	7	5
7	39	M	+	15	24	+	16	-	-	-	5	2
8	57	F	+	29	28	+	19	-	-	-	9	2
9	57	M	+	18	39	+	21	+	+	+	6	4
10	60	M	+	18	42	+	20	-	+	+	na	2
11	24	F	+	17	7	+	18	+	+	-	4	5
12	39	M	+	35	4	+	21	-	+	-	8	5
13	32	M	+	6	18	+	19	+	+	-	8	3
14	18	F	+	22	4	+	19	-	-	-	2	2
15	27	M	+	24	3	+	18	+	+	-	3	4
16	76	M	+	18	58	+	19	+	+	+	3	5
17	43	F	+	17	26	+	20	+	+	-	5	3
18	60	F	+	16	44	+	21	-	+	-	9	2
19	29	F	+	12	17	+	14	+	+	+	8	2
20	42	F	+	9	33	+	18	+	+	-	na	3
21	42	M	+	28	14	+	14	-	-	-	2	4
22	43	M	+	30	13	+	na	+	+	+	7	2
23	29	F	+	19	10	+	22	+	-	-	1	5

F refers to females; M, males; ESS, Epworth Sleepiness Scale; SP, sleep paralysis; HH, hypnagogic hallucinations; RBD, REM sleep behavior disorder; MSLT, multiple sleep latency test; SOREM, sleep onset REM; na, not available.

The t-tests revealed no difference between patients with NC and controls in either years of education or f-NART (P values > 0.1). NC patients showed significantly more EDS on the ESS and had higher scores on the BDI-II and the UPPS Perseverance sub-scale (Table 2). None of the NC patients or controls had a recent or past history of pathological gambling. Only one NC patient (No. 5) had a past history of drug (cannabis) dependence, stopped for at least 3 years.

Table 2.

Demographic characteristics and scale scores of patients with narcolepsy and healthy control subjects

	Patients n = 23	Controls n = 23	t or χ2	P
Age	38.4 ± 14.9	38.5 ± 14.6	0.02	0.98
Sex (F/M)	10/13	11/12	< 0.001	1
Years of education	12.2 ± 2.1	12.1 ± 2.1	-0.20	0.84
f-NART	104 ± 20.9	110 ± 6.1	1.35	0.18
ESS	18.7 ± 2.6	5.7 ± 2.6	16.37	< 0.0001
BDI-II	12.4 ± 7.3	6.2 ± 4.6	3.36	0.003
UPPS
Urgency	29.9 ± 3.8	28.7 ± 7	0.67	0.48
Premeditation	21.6 ± 3.9	20.3 ± 5.7	0.89	0.37
Perseverance	20 ± 4.2	16.6 ± 3.1	2.96	0.005
Sensationseeking	30.4 ± 8.2	29.7 ± 10.1	0.26	0.78
Total	102.1 ± 13.2	95.3 ± 19.8	1.27	0.21

f-NART refers to French version of the National Reading Test; ESS, Epworth Sleepiness Scale; BDI-II, 21-item Beck Depression Inventory-II; UPPS, Impulsive Behavior Scale: Urgency, Premeditation, Perseverance, Sensation seeking.

Ambiguous Decisions on the IGT

As shown in Figure 1, the total IGT net score of NC patients was significantly lower than that of controls (P = 0.011). The percentages of NC patients having a total net score ≥ 0 (43%) and a total net score < 0 (57%) differed from those of controls (net score < 0 = 13%; net score ≥ 0 = 87%; χ² = 3.86, P = 0.049), indicating a preference for disadvantageous alternatives in NC.

Figure 1.

IGT total net score for NC patients and controls. Means (± SEM) are given. *P < 0.05. IGT refers to Iowa Gambling Task.

A more in-depth analysis of IGT performance revealed significant main effects for group (F = 7.09, P = 0.011) and blocks (F = 20.27, P < 0.001). The blocks × group interaction also reached significance (F = 3.49, P = 0.009), indicating that patients and controls displayed different decision-making patterns during the task (Figure 2). Comparisons of group performance on each IGT block revealed net score differences on block 4 (P = 0.028) and block 5 (P = 0.003) and a similar tendency on block 3 (P = 0.080). After controlling for age in the NC group, no correlation was observed between IGT variables, disease characteristics (i.e., age at onset, disease duration, presence of RBD, sleep paralysis, and hypnagogic hallucinations), and depression inventory. In addition, no association was found between objective (MSLT) and subjective (ESS) daytime sleepiness, and performances on decision-making tasks. To avoid ceiling effects and to explore further the potential influence of sleepiness on decision making, IGT performances of narcoleptic patients with MSLT above (n = 11) and under (n = 12) the median of their group were compared. The 2 subgroups of patients did not differ on IGT block 4 (P = 0.92) or block 5 (P = 0.31). However, NC patients with IGT net score < 0 (n = 10; 22.4 ± 3.6) scored higher on the UPPS Perseverance sub-scale than NC patients with IGT net score ≥ 0 (n = 13; 17.3 ± 3.2; Z = −2.66; P = 0.006). Patients who demonstrated a preference for disadvantageous alternatives thus showed a lack of perseverance. Finally, we may add that NC patient No. 5, with a past history of drug dependence, did not differ from the other NC patients on IGT variables.

Figure 2.

IGT net scores on each of the 5 blocks, consisting of 20 card selections each, for NC patients and controls. Means (± SEM) are given. *P < 0.05; **P < 0.01. IGT refers to Iowa Gambling Task.

Risky Decisions on the GDT

We did not find any difference between GDT net scores of patients with NC and controls (P = 0.56). Note also that Patient No. 5 did not differ from the other NC patients. The positive net score for both groups implied that neither controls nor patients preferred risky options. As illustrated in Figure 3, a more detailed evaluation of GDT performance revealed no difference between groups in the mean frequency of choices of alternatives (P's > 0.22).

Figure 3.

Mean frequency (± SEM) of selection of each alternative GDT category by NC patients and controls, revealing similar decision-making patterns in both groups. All P's > 0.22. GDT refers to Game of Dice Task.

We analyzed the feedback effect on strategy shifts in 19 controls and 17 NC patients, but failed to find any between-group difference in risky and/or non-risky choices after receiving positive or negative feedback on a risky alternative. Finally, no correlations were observed in the NC group between GDT performance, disease characteristics (i.e., age at onset, disease duration, the presence of RBD, sleep paralysis and hypnagogic hallucinations, and MSLT data), and scales (ESS, BDI-II, UPPS).

DISCUSSION

The present study reports for the first time on selective reduced performance in decision making in NC patients under ambiguous conditions compared to normal decision making under risky conditions. Patients with NC opted more frequently than controls for decks with high immediate reward regardless of higher future punishment.

Decision making is a key function in real-life situations, and disturbances of this ability can lead to problems in life. When a choice must be made, information about expected outcomes has to be implicitly or explicitly memorized and integrated with information on emotions and behavior. Abnormal decision-making characterized by increased sensitivity to reward and/or decreased sensitivity to punishment leads to disadvantageous choices, implying an alteration in the evaluation of the outcomes of decisions. Patients with NC made disadvantageous choices involving immediate high reward (monetary gain), whereas future high punishment (monetary loss) had little influence.

The performance discrepancies found between two well-validated ambiguous (IGT) and risky (GDT) decision-making paradigms in NC may be due to different cerebral networks involved under these conditions. Hence, results from previous studies show that GDT performance is strongly associated with tasks measuring executive functions.^19,22 In contrast, IGT performance, which is associated with implicit rules, is not related to executive functions, but rather depends on emotional and cognitive feedback processing to avoid disadvantageous and prefer advantageous choices.^19,23 More precisely, brain lesion studies have confirmed that the fronto-cortico-striatal loop is crucially involved in reward processing as measured by the GDT,²⁰ whereas the limbic loop is preferentially involved in reward processing as measured by the IGT.^19,22 These two paradigms have been recognized as sensitive to psychiatric conditions such as substance dependence and abuse,^24–26 pathological gambling,^27,28 and in Parkinson disease with mesolimbic and mesocortical circuit alterations.^28,29

IGT decision making also requires the integrity of the amygdala.¹⁹ The amygdala is involved in regulating the emotional state, and neuroimaging studies have reported that the amygdala was involved in the emotional processing of monetary outcomes in simulated gambling tasks.³⁰ Recent studies in human NC have reported abnormal activity in the emotional network including the amygdala,^31–33 suggesting its possible involvement in abnormal sensitivity to reward and punishment.

Decision making under implicit conditions may depend on other brain structures.³⁴ Reward processing in probabilistic tasks recruits dopaminergic regions that project to the striatum and the medial prefrontal cortex. Like dopamine, hypocretin is an essential neural substrate for many types of motivated behavior. Hence, hypocretin potentiates glutamate-mediated responses of VTA dopaminergic neurons,¹² and through the amygdala, the hypocretin system increases emotion-related behavioral responses.^31–33 As hypocretin neurons project to reward-associated brain regions, we may hypothesize their implication in feedback-based decision making. However, patients with NC are not insensitive to reward per se; they selected more disadvantageous alternatives than controls, but they were not as severe as those with ventromedial prefrontal lesions.¹⁹

We did not find that higher disease severity, age, or disease duration were associated with increased sensitivity to reward and/or decreased sensitivity to punishment. As patients with NC presented higher self-reported depression symptoms and altered IGT performance has been reported in patients with major depressive disorder,³⁵ the absence of correlation between IGT performance and BDI-II scores suggests that our results cannot be explained by depression per se. Similarly, because impaired IGT responses have been reported in controls following two nights of sleep loss,³⁶ the absence of any correlation between IGT performance and sleepiness, and the absence of differences between IGT performances of narcoleptic patients with MSLT above (n = 11) and under (n = 12) the median of their group point to the involvement of a more stable neurobiological state in decision-making problems in NC. As IGT is essentially a learning task, we may also hypothesize that NC patients failed to gradually learn to choose advantageous alternatives. One limitation of this study is that NC patients were not evaluated for memory functioning. However, it is documented in the literature that untreated NC patients do not have organic memory deficit.³⁷

Using the UPPS Impulsive Behavior Scale, we found for the first time a significant lack of perseverance in NC. This abnormal impulsive personality trait may lead to a general blunting of emotional reactivity, and therefore abnormal decision making on the IGT. Hence, patients with NC may opt for choices with higher immediate emotional valence, regardless of the amount of reward or punishment, in an attempt to compensate for their reduced reactivity to emotional stimuli. This assumption is corroborated by the lack of perseverance observed in NC patients along with their disadvantageous decision-making pattern on the IGT. Despite the absence of literature, clinical experience suggests that patients with NC rarely require increasing therapeutic doses of psychostimulants (even when treated for years).

Several limitations in our study need to be addressed. First, the sample size is low although sufficient to demonstrate significant reduced performance on decision making under ambiguity. Second, we cannot formally exclude the potential influence of sleepiness on decision making performances in narcolepsy. Finally, cerebrospinal fluid hypocretin-1 measurement was available in only three patients, all undetectable CSF hypocretin-1 levels. However, we may expect low levels in almost all patients included, as they were all sporadic cases, HLA DQB1*0602 positive, with clear-cut cataplexy, and with at least two sleep onset REM periods during the MSLT. Further studies on decision-making paradigms are needed, including central hypersomnia patients with and without (in the absence of cataplexy) hypocretin deficiency and with and without psychostimulant medication.

In summary, the present study finds a lack of perseverance and a selective reduced performance on decision making under ambiguity in narcolepsy-cataplexy in contrast to normal decision making under explicit conditions. Patients with narcolepsy-cataplexy may opt for choices with higher immediate emotional valence, regardless of higher future punishment, to compensate for their reduced reactivity to emotional stimuli.

DISCLOSURE STATEMENT

This was not an industry supported study. Dr. Dauvilliers has consulted for UCB Pharma, Cephalon, Bioprojet, GlaxoSmithKline, and Boehringer Ingelheim. The other authors have indicated no financial conflicts of interest.