Participating Sites

This research protocol was carried out simultaneously at the Beth Israel Deaconess Medical Center in Boston, MA and also at the Leiden University Medical Centre in Leiden, Netherlands. The BIDMC and Leiden Committees on Clinical Investigations approved the research protocols.

Recruitment

In Boston, subjects with N+C and N-C were recruited from sleep clinics in the Boston area and through the Narcolepsy Network. Healthy controls were recruited through the internet. In Leiden, subjects with narcolepsy were recruited from the sleep clinic at Leiden University Medical Center, and healthy controls were recruited through newspaper advertisements.

Subject Characteristics

We tested a total of 77 subjects: 47 women and 30 men (N+C, N = 30, 10 male; N-C, N = 16, 9 male; and healthy controls, N = 32, 11 male). These were matched across groups for sex, age, and smoking behavior (Table 1). All subjects took their regular medications during this study, including medications for narcolepsy. Dosages used at the time of participation were not included in the analyses.

Table 1 Subject characteristics (more ...)

Inclusion/Exclusion Criteria

Participants 18-64 years of age were included. The diagnoses of narcolepsy with and without cataplexy were confirmed using ICSD-2 criteria,13 including review of each individual’s overnight PSG and MSLT by a board-certified sleep neurologist. To assure that patients were clinically stable, narcolepsy patients were selected for the study only if they had a definite diagnosis for > 6 months and if they were free of medication changes for over a month. Informed consent was obtained prior to medical record review.

Subjects were excluded if they met current DSM-IV14 diagnostic criteria for substance or alcohol abuse or dependence,15 gambling addiction, binge-eating disorder, generalized anxiety disorder, depression, or other psychiatric illness, as these conditions have characteristic neurobehavioral profiles that would have skewed the results.16–18 While the protocol allowed for recreational drug use in the past (defined as drug use without diagnosis of abuse, dependence, or participation in a detoxification program), individuals were excluded for recreational drug use within the past week. Smoking was not considered an exclusion criterion.

Procedures

Statistics

Groups were compared using the independent samples t-test. Dichotomous values were analyzed using the χ² test. Possible medication effects on BART and questionnaire scores were investigated in a multivariate analysis. P-values lower than 0.05 were considered to be significant. For the categorical analysis of the results on the EIS, BES, BDI-II, and the BAI, validated cut-off points were used.15,20,27,31

Sleepiness and Vigilance

Even though subjects with N+C and N-C were taking their regular medications, they had higher ESS scores than the control group (p < 0.001) (Table 2). Similarly, subjects with N+C or N-C had longer reaction times on the PVT than controls (p < 0.001).

Table 2 Test results (more ...)

Personality

The responses on the Zuckerman Sensation Scale were comparable among the 3 groups in all 5 categories—general sensation seeking, thrill and adventure seeking, experience seeking, disinhibition, and boredom susceptibility (Table 2). On the Eysenck Impulsiveness Scale (EIS), subjects in the N+C group had significantly higher scores than controls (p < 0.05), while there was a trend for the N-C group to score higher than controls (p = 0.07). After applying the age and sex-specific cutoffs for the EIS, we found that 10% of the N+C subjects were categorized as impulsive, compared to 6.7% of the N-C and none of the controls (Figure 1D). Upon closer analysis of the EIS responses, we found that the N+C and N-C groups differed greatly from controls in the way in which they answered 7 of the 19 EIS Questions (Table 3).

Categorical analysis of the results on the Beck Depression Inventory II (BDI-II), the Beck Anxiety Inventory (BAI), the Gormally Binge Eating Scale (BES), and the Eysenck Impulsiveness Scale (EIS) for the groups Narcolepsy with Cataplexy (N+C), Narcolepsy without Cataplexy (N-C), and Controls (Ctrl). Validated cut-off points were used.15,20,27,31

Figure 1 Categorical analysis of the results on the Beck Depression Inventory II (BDI-II), the Beck Anxiety Inventory (BAI), the Gormally Binge Eating Scale (BES), and the Eysenck Impulsiveness Scale (EIS) for the groups Narcolepsy with Cataplexy (N+C), Narcolepsy without Cataplexy (N-C), and Controls (Ctrl). Validated cut-off points were used.15,20,27,31 (more ...)

Table 3 Significant differences (more ...)

Depression and Anxiety

On both depression scales—the BDI-II and HADS Depression Scales—the N+C and the N-C groups scored significantly higher than the control group, with responses in the N+C group indicating higher levels of depression (Table 2, Figure 1A).

On the Beck Anxiety Scale (BAI), the N+C and the N-C group scored more anxious than the control group. On the HADS anxiety scale, the N+C group again scored significantly higher than controls or the N-C group (Table 2, Figure 1B).

Addiction and Substance Abuse

The surveys of substance and alcohol abuse and compulsive gambling showed no significant differences between groups. However, on the Binge Eating Scale (BES) the N+C subjects scored significantly higher than controls (p < 0.05), with 16.7% Of the N+C subjects scored as moderate binge eaters and 6.7% as severe binge eaters (Figure 1C). All of the N-C subjects scored in the non-binge-eating range. Among the control subjects, only 3.1% were found to be moderate binge eaters and none were severe binge eaters.

Balloon Analogue Risk Task

There were no statistically significant differences among the groups on either BART trial, including the total number of pumps, the average number of pumps per balloon, and the number of exploded balloons (Table 2). All 3 groups scored higher (more risk-taking) on the second BART trial when the stakes were higher, with similar increases in all groups.

Effects of Medications

As the N+C and N-C subjects were maintained on their regular medications, we examined the effects on medication using a multivariate analysis. There were no significant effects of age, sex, modafinil, amphetamines, GHB, antidepressants, or benzodiazepines/opioids/muscle relaxants on BART, PVT, or questionnaire scores.

Differences Between Sites

Comparing the Boston and Leiden populations, no variables differed significantly across all 3 groups. However, the following differences were highly significant between 2 groups (all p < 0.001): Narcolepsy with cataplexy subjects in Boston were more likely to use modafinil (53% vs. 12%); narcolepsy without cataplexy subjects in Boston were more likely to have used LSD/PCP/ketamine (50% vs. 0%); and controls in Leiden had higher scores on the AUDIT (7.1 ± 0.5 vs. 2.3 ± 1.0). Supplemental Table S1 presents a complete overview of differences between the 2 research sites.

Hypocretin Deficiency and Addiction

While narcoleptic patients are often treated with highly addictive drugs (including amphetamines), the clinical picture is that they rarely become addicted.9,10 However, our results do not support the idea that hypocretin-deficient individuals are less prone to addiction. It is possible that hypocretin-deficient patients may process reward via a different mechanism, not solely involving the mesolimbic pathway. Recent fMRI findings by Ponz et al.39 suggest that in contrast to controls, in non-medicated narcoleptics with cataplexy, activity in the ventral tegmental area is not modulated by high reward expectancy, and activity in the ventral striatum is reduced during winning. Interestingly, in the 12 hypocretin-deficient patients studied, there was reward-associated increased activity in the amygdala and dorsal striatum.

Risk-taking, Impulsiveness, and Sleepiness

In the current study, all narcolepsy patients reported excessive daytime sleepiness, yet they had normal levels of risk taking. In contrast, healthy sleep deprived subjects have shown increased risk-taking on various gambling tasks and questionnaires, such as the Brief Sensation Seeking Scale (BSSS), the Evaluation of Risk (EVAR) scale, and the BART.38,40,41 Possibly, sleepiness also increases risk-taking behavior in narcolepsy with cataplexy, but this effect is masked by their hypocretin deficiency. However, this would not explain the findings in the non-hypocretin deficient narcolepsy without cataplexy group. More likely, the sleepiness experienced by narcoleptic subjects is of a different nature than the sleepiness caused by sleep deprivation in otherwise healthy persons.

Though risk taking appeared normal, N+C subjects had significantly higher Eysenck Impulsiveness Scale (EIS) scores than controls, and the N-C subjects had a trend towards higher EIS scores. Specifically, N+C subjects differed from controls on questions such as Do you generally do and say things without stopping to think? and Do you mostly speak without thinking things out? Differently, N-C subjects more often answered in a positive way to questions such as Do you need to use a lot of self-control to keep out of trouble? and Do you think an evening out is more successful if it is unplanned or arranged at the last moment? These observations could shed light on the nature of narcolepsy with and without cataplexy. Perhaps those with cataplexy speak or act impulsively because they appreciate that their sleepiness can impair their ability to sustain attention. In their fight with chronic sleepiness, it may work best for them to grab the moment. The different responses on the EIS between narcoleptics with and without cataplexy suggest that some quality of sleepiness may differ between the two groups.

Binge Eating

Narcoleptics with cataplexy were more likely to report moderate and severe binge eating. This may reflect more impulsive behavior or a strategy for staving off sleepiness, but based on work in rodents, it seems unlikely that people with narcolepsy would find food more rewarding.6 The higher prevalence of eating disorders and possibly increased food intake in narcolepsy is clinically important, as people with narcolepsy are more obese than healthy controls.42 Prior studies of obesity in narcolepsy have yielded mixed results regarding eating habits, total food intake, food choice, basal metabolic rate, and physical activity. In a study of narcoleptics with and without cataplexy, both groups had increased scores on a questionnaire developed for detecting eating disorders (EAT-40).43 Similarly, narcoleptics with cataplexy scored significantly higher than age-and sex-matched controls on measures of eating disorders contained in the Schedules for Clinical Assessment in Neuropsychiatry (SCAN 2.1) and on DSM-IV criteria for bulimia and Eating Disorder NOS.44 Conversely, in a study of 116 patients with narcolepsy with and without cataplexy, there was no increased incidence of eating disorders (anorexia nervosa, bulimia, and binge eating disorder), as defined by the DSM-IV criteria.45 Furthermore, self-reported total food intake and physical activity as measured by actigraphy appeared normal in narcoleptic subjects.46,47 Many questions remain about the causes of obesity in narcolepsy, but our results indicate that binge eating is common in narcolepsy with cataplexy.

We chose to include subjects with self-reported past overeating who had never been formally diagnosed or treated for a binge-eating disorder. On close review of the results, only one of the four N+C subjects who reported past overeating scored high on the BES. The remaining moderate and severe BES scores came from subjects, who did not report a problem on initial screening. This may indicate a larger scale binge-eating problem associated with orexin deficiency than captured by our testing and necessitates further investigation.

Anxiety and Depression

As in previous studies, we found higher levels of depression and anxiety in patients with narcolepsy using the BDI-II, the BAI, and the HADS questionnaires.48–50 One limitation is that the BDI-II includes measures of physical aspects of depression that overlap with the symptoms of narcolepsy, such as fatigue, excessive daytime sleepiness, and decreased energy levels. However, the HADS scale does not contain these questions, and it also revealed higher levels of depression and anxiety in both N+C and N-C. This suggests that hypocretin deficiency or the life impact of narcolepsy affects mood independent of daytime sleepiness.

A closer look reveals that subjects in the N+C group were more often diagnosed with depression than those in the N-C and control groups. This finding and the higher depression scores in the N+C group raise the possibility that severe hypocretin deficiency may cause or aggravate depression. In fact, prior studies showed that low hypocretin-1 levels correlated with the severity of depressive symptoms and the prevalence of attempted suicide.51,52 Perhaps decreased hypocretin signaling reduces signaling in reward and motivation pathways that then contributes to depression.

Limitations

Our study provides new perspectives on reward- and risk-seeking behaviors in narcolepsy, but a few limitations should be mentioned. One is that the scales we used for the assessment of depression, anxiety, risk taking, impulsiveness, and binge eating are not validated for use in these populations. Another is that all participants continued their regular medications, which could have influenced mood or have affected impulsiveness. For example, amphetamines or antidepressants could have improved mood in the narcolepsy subjects, masking the true prevalence of depression and influencing impulsivity. However, it is unlikely that the final results were influenced by this, as a multivariate analysis did not show any effects of medications.

Hypocretin-1 concentrations in the cerebrospinal fluid were not known for the subjects included in this study. As such, it is not certain that all N+C subjects were really hypocretin deficient. However, 90% of N+C patients have no detectable hypocretin-1 in their cerebrospinal fluid (CSF). In contrast, nearly all individuals with N-C have normal CSF concentrations of hypocretin.1,11,12

We excluded subjects with a prior history of addiction, substance abuse, or psychiatric illness. This may seem counterintuitive, as these exclusion criteria are directly related to our main hypothesis. However, to truly study the prevalence of addiction and substance abuse, one would need a substantially larger study population and a more epidemiological approach, which is very difficult in a relatively rare disease such as narcolepsy. Inclusion of subjects with ongoing substance abuse or psychiatric problems would have skewed the results.

Smoking was not considered an exclusion criterion. To rule out any influence of smoking, subjects were matched for smoking behavior. In future studies, it would be of interest to consider smoking as addictive behavior. However, in our clinical experience, patients often smoke because nicotine has an alerting effect. This would make it difficult to distinguish between addictive behavior and behavior to counterbalance sleepiness.

The study included populations from Boston and Leiden, and unique cultural characteristics could have influenced risk-taking behavior. The main differences between sites were that in Boston, more N+C subjects used modafinil, more N-C subjects used LSD/PCP/ketamine, and controls had higher alcohol consumption (Supplemental Table S1) However, post hoc analysis showed that these factors did not significantly influence the final results, and the main findings of our study were present in both populations.