The clinical spectrum of narcolepsy and idiopathic hypersomnia

Although narcolepsy has been recognized as a syndrome for more than 100 years, its clinical spectrum is not well defined. Reports by Fischer ^[1] and Westphal ^[2] preceded that of Gelineau in 1880, ^[3] who used the French word ``narcolepsie'' to describe a patient with short sleep episodes of sudden onset and spells of weakness triggered by emotion. Gowers <sup>[4]</sup> in 1893 distinguished narcolepsy from trance, characterized by a prolonged unarousable state due to hysteria, and from African sleeping sickness characterized by increasing somnolence leading to coma and death. Although Westphal, Fischer, and Gelineau had described cataplexy, Lowenfeld <sup>[5]</sup> in 1902 was apparently the first to insist that the term ``narcolepsy'' be applied only when sleepiness is accompanied by episodic weakness.

Adie, ^[6] like Lowenfeld, believed narcolepsy-cataplexy to be a single disease, whereas Wilson, ^[7] Notkin and Jelliffe, ^[8] and others referred to ``the Narcolepsies''--a symptom complex that could be seen in its entirety, or in parts, and could occur with encephalitis, multiple sclerosis, head trauma, cerebral neoplasm, psychiatric disease, or with no apparent antecedent. In the early 20th century, diagnosis was based solely on clinical features, and many patients with other sleep disorders were misdiagnosed as narcoleptic. For example, obstructive sleep apnea syndrome was almost certainly the diagnosis in an obese patient without cataplexy who snored loudly and ``almost choked'' during sleep, ^[9] and in the patients with ``familial narcolepsy'' characterized by somnolence, obesity, and cyanosis during sleep. ^[10]

The concept of Yoss and Daly ^[11] of the narcoleptic tetrad of symptoms--sleepiness, cataplexy, hallucinations, and sleep paralysis-as a specific clinical syndrome, the discovery of sleep-onset rapid eye movement (REM) periods associated with narcolepsy, and the discoveries that obstructive sleep apnea syndrome and periodic limb movement syndrome were causes of daytime somnolence that were usually not associated with sleep-onset REM periods supported the view of narcolepsy as a single disease. Guilleminault and Dement ^[12] differentiated ``REM narcolepsy'' (characterized by sleep-onset REM periods and irresistible sleep attacks of short duration with or without auxiliary symptoms) from ``non-REM narcolepsy'' (characterized by increased slow-wave sleep and longer less-refreshing naps) and from hypersomnia (characterized by excessive, long-lasting, but not necessarily irresistible sleep without auxiliary symptoms). The emphasis on sleep-onset REM periods as a key feature that could be used to differentiate narcolepsy from other disorders gained favor and was reflected in the definition of narcolepsy adopted by participants in the First International Symposium on Narcolepsy in 1975 ^[13] and in subsequent nosologies ^[14,15] Table 1.

Current views of the syndrome of idiopathic hypersomnia are derived largely from the work of Bedrich Roth and colleagues ^[16-18] who distinguished narcolepsy, characterized by irresistible sleep attacks of short duration, from other hypersomniac states characterized by less imperative sleep episodes of longer duration lasting hours to weeks. Roth subdivided non-narcoleptic functional hypersomnias into three types: (1) monosymptomatic idiopathic hypersomnia, characterized by diurnal hypersomnia and long nocturnal sleep periods; (2) polysymptomatic hypersomnia characterized by the same symptoms as well as by prolonged confusion and disorientation upon awakening (sleep drunkenness); and (3) neurotic hypersomnias. ^[17] The International Classification of Sleep Disorders' (ICSD) ^[15] definition of ``idiopathic hypersomnia'' Table 1 includes clinical features virtually identical to the ``functional hypersomnia'' described by Roth, but series of such patients are rare and the frequency of key clinical features is often not reported. ^[18,19]

More recently, some authors have suggested that the presence of cataplexy should be a requirement for a diagnosis of narcolepsy. Honda ^[20] proposed that cataplexy and recurrent daytime sleep episodes of at least 6 months' duration must both be present to make a diagnosis of narcolepsy. In a series of 190 Japanese subjects diagnosed by his criteria, 100% were positive for the HLA-DR2 antigen. ^[21] Furthermore, a definition of narcolepsy that requires cataplexy yields a more homogeneous group than a definition that relies on the presence of REM sleep abnormalities. ^[22] On the other hand, the frequency of HLA-DR2 in patients without cataplexy who are diagnosed with narcolepsy based on ICSD criteria appears to be increased compared with the general population (reviewed in reference 23), suggesting that at least some patients with narcolepsy without cataplexy have the genotype associated with narcolepsy-cataplexy.

In our clinical practice, we have seen a number of patients with excessive daytime sleepiness who do not fit existing ICSD diagnostic categories or who do not have the combination of clinical symptoms and polygraphic findings said to be typical of narcolepsy or idiopathic hypersomnia. We therefore reviewed clinical findings, questionnaire results, and sleep studies in patients with narcolepsy, idiopathic hypersomnia, and related disorders to determine the frequency of symptoms included in the ICSD definitions of these disorders and the relation of these symptoms to sleep-onset REM periods.

Methods.

From a database of 3,618 patients evaluated between 1985 and 1993 at our Sleep Disorders Center, we retrospectively reviewed clinical evaluations, questionnaires, and polygraphic data on patients who (1) had completed a detailed sleep questionnaire; (2) had undergone diagnostic polysomnography with a subsequent Multiple Sleep Latency Test (MSLT); (3) were aged more than equals to 18 years; (4) had a mean sleep latency less than 6 minutes on the MSLT; (5) had an apnea-hypopnea index less than 10 and a periodic leg movement (PLM) index less than 30; (6) had fewer than 10 entries to stage 1 sleep from a deeper stage of sleep per hour of sleep (a measure that correlates to some degree with arousal frequency); (7) were free of sedating or REM-suppressing medications at the time of polygraphic evaluation; (8) had no medical or psychiatric condition thought to contribute to sleepiness; and (9) had only one sleep disorder diagnosis. As nocturnal sleep disruption caused by sleep apnea, PLMs, or other factors is associated with an increased frequency of sleep-onset REM periods, we included criteria 6 and 9 above to minimize the impact of disrupted sleep on the data analysis. We used a more liberal cutoff for PLMs than for respiratory events because PLMs are less often associated with arousals.

All patients were diagnosed based on ICSD criteria with the exception that the subjects with narcolepsy were divided into those with definite cataplexy based on clinical assessment and those without definite cataplexy. In this analysis, we included all subjects who met the above criteria and were diagnosed with narcolepsy-cataplexy (NC) (25 women; 14 men), narcolepsy without cataplexy (NwC) (13 women; 15 men), and idiopathic hypersomnia (IH) (17 women; 9 men). We also included a group of subjects meeting the above criteria whom we ``diagnosed'' with excessive daytime sleepiness not otherwise specified (EDS-NOS) (10 women; 9 men). Most of these subjects met criteria for IH but it was our judgment that other diagnoses had not been fully excluded. For comparison, we included subjects meeting the above criteria who were diagnosed with insufficient sleep syndrome (ISS) (13 women; 9 men) or with mild sleep disordered breathing (SDB) (7 women; 14 men). The group with SDB was limited to subjects with an apnea-hypopnea index less than 10 to minimize the impact of nocturnal sleep disruption on the results and to determine whether subjects who fit the above criteria and were diagnosed with mild sleep apnea differed from the other patient groups in clinical characteristics and questionnaire responses.

During polysomnography, the following were recorded using standard techniques: EEG (C3, C4, O1, O2 by the International 10-20 System), chin EMG, electro-oculogram, ECG, respiratory effort (mercury strain gauges or piezoelectric belts), airflow at the nose and mouth (thermistors), and bilateral anterior tibialis EMG with surface electrodes. ^[24-26] Oxygen saturation was monitored by pulse oximetry (Biox 3700, Ohmeda Corporation, Madison, WI). All studies were performed in the Sleep Laboratory with continuous monitoring by polysomnographic technologists and were recorded on paper at 10 mm/sec paper speed using 15-, 17-, or 21-channel polygraphs (Nihon-Kohden America, Inc., Irvine, CA; Grass Instruments, Quincy, MA).

All studies were scored manually by experienced polysomnographic technologists for sleep stages, leg movements, and respiration with standard techniques. ^[24-26] An apnea-hypopnea index was calculated as the number of apneas plus hypopneas per hour of sleep. Hypopneas were defined as clear reductions of respiratory airflow or effort lasting 10 seconds or longer associated with either an arousal or a drop in oxygen saturation of at least 4%. A PLM index was calculated as the number of minutes during which a PLM occurred divided by the number of hours of sleep. MSLTs were performed and scored according to published guidelines. ^[27] Four nap opportunities were recorded unless one sleep-onset REM period occurred, in which case a fifth nap was added. The mean sleep latency on the MSLT was defined as the mean time from lights out to the first 30-second epoch scored as sleep. A sleep-onset REM period was defined as one or more epochs of REM sleep occurring within 15 minutes of the first epoch scored as sleep.

Some patients underwent more than one diagnostic polysomnogram or MSLT. For each patient, we determined the total number of sleep-onset REM periods, defined as REM sleep occurring within 15 minutes of sleep onset, and the total number of opportunities for sleep-onset REM periods, defined as the number of episodes of falling asleep either at the beginning of the polysomnogram or at the beginning of a nap test period on the MSLT. We calculated a measure of propensity for REM sleep to occur at sleep onset (REM sleep propensity) as the number of sleep-onset REM periods divided by the number of opportunities for sleep-onset REM periods.

For the 157 subjects, clinical notes were reviewed to determine whether subjects had cataplexy, sleep paralysis, or hypnagogic or hypnopompic hallucinations based on clinical assessment. HLA test results were reviewed. Subject responses to 13 items of a 66-item sleep questionnaire, composed in part from items derived from the Stanford Sleep Questionnaire and Assessment of Wakefulness, ^[28] were tabulated to further assess sleep habits and symptoms. Subjects were instructed to answer the questions as they applied to the preceding 6 months. The questions used included the following:

1. Do you feel that you are excessively sleepy during the day?

2. What times do you usually go to bed on weekdays and weekends?

3. What times do you usually get up on weekdays and weekends?

4. What is the total amount of sleep that you usually get at night?

5. How long does it usually take you to ``get going'' after you get out of bed?

6. How many naps do you take on purpose in a usual weekday?

7. What is the total amount of sleep that you get during naps in a usual weekday?

8a,b. How often do you have vivid dream-like images or hallucinations (even though you are still awake) when falling asleep? when awakening?

9a,b. How often do you feel paralyzed (unable to move) when falling asleep? when awakening?

10. How often do you find that you have performed a complex act, such as driving a car to the wrong destination, and not remember how you did it?

11. How often do you find yourself doing things that make no sense, such as writing nonsense or mixing chocolate with gravy?

12. How often do you have a feeling of ``weak knees'' when you laugh?

13. How often do you have episodes of sudden muscular weakness (paralysis or inability to move) when laughing, angry, or in other emotional or exciting situations?

Questions 8 to 13 were answered with a five-point scale as follows: 1 equals never or almost never; 2 equals seldom: not more than once per month; 3 equals occasionally: 1 to 3 times per month; 4 equals often: more than 1 to 2 times per week; 5 equals Always or almost always, or almost every day.

We combined answers to questions 8a and 8b to determine the proportion of subjects who reported hypnagogic or hypnopompic hallucinations occasionally or more frequently. Answers to questions 9a and 9b were combined in similar fashion for sleep paralysis, and answers to questions 10 and 11 were used to assess automatic behavior.

The presence of definite cataplexy, defined as sudden bilateral weakness precipitated by emotion, was determined solely by clinical assessment. Answers to questions 12 and 13 were assessed to determine if any hint of cataplexy--defined as an answer other than ``never or almost never'' to either question--was present in subjects who did not have definite cataplexy based on clinical assessment. For questions 8a, 8b, 9a, 9b, 12, and 13, we also compared the responses of the subject groups with those of a group of 108 control subjects from the general population (68 women, 40 men; age range, 18 to 69 years).

Statistical analysis was performed using contingency tables with chi-square analysis and one-way analysis of variance (Statgraphics Plus, Manugistics, Rockville, MD).

Results.

The mean age of onset of sleepiness was between 23 and 27 years in all groups. Compared with the subjects with IH and NwC, those with NC had significantly greater nocturnal sleep disturbance with less stage 3-4 sleep, lower sleep efficiency, more than 60% more awakenings, and almost 50% more stage 1 sleep. They also had a higher incidence of sleep paralysis and sleep-related hallucinations Figure 1. Comparison of the NC group with the group with EDS-NOS yielded similar results. (A Table summarizingclinical features, selected polysomnographic and MSLT variables, and questionnaire results is on file with the National Auxiliary Publications Service [NAPS]. See Note at end of text.)

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Proportion of subjects with sleep-related hallucinations, sleep paralysis, and automatisms (question 10) based on questionnaire assessment in each of six diagnostic groups. NC equals narcolepsy-cataplexy; N equals narcolepsy without cataplexy; IH equals idiopathic hypersomnia; NOS equals excessive daytime sleepiness not otherwise specified; ISS equals insufficient sleep syndrome; SDB equals sleep disordered breathing.

In contrast, there were few differences between the IH group and the NwC group apart from nocturnal REM sleep latency (92 versus 63 min; p less than 0.05) and the number of sleep-onset REM periods (0.2 versus 2.3; p less than 0.01) on the MSLT. Sleep efficiencies (90% for IH versus 93% for NwC) and amounts of stage 3-4 sleep (42 min for IH versus 39 min for NwC) were similar. Subjects with IH were more likely than those with NwC to report sleep-related hallucinations by questionnaire (41% versus 14%; p less than 0.05) but not by clinical assessment (47% versus 30%); the incidences of sleep paralysis and automatic behavior were not different statistically. Questionnaire assessments of week-day hours in bed, number of intentional naps, hours of sleep at night, and hours to ``get going'' in the morning were almost identical in the two groups. Subjects with idiopathic hypersomnia were more likely to describe themselves as excessively sleepy (95% versus 72%; p less than 0.05), but subjects with NwC had shorter mean sleep latency on the MSLT (2.8 versus 3.6 min; p less than 0.05).

As expected, subjects with ISS spent the least amount of time in bed (6.0 versus 7.6-7.9 hours for the other groups; p less than 0.01) and had the highest sleep efficiencies (95% versus 86-93%; p less than 0.01). When compared with the other five groups combined, they were less likely to have sleep paralysis and sleep-related hallucinations, reported fewer hours of sleep per night, spent fewer hours in bed on weekdays, and weekends, and took longer to ``get going'' in the morning (1.1 versus 0.5-0.8 hours). When compared with all other groups, subjects in the SDB group were less likely to have sleep-related hallucinations by clinical assessment and were less likely to describe themselves as excessively sleepy (58% versus 67-97%; p less than 0.05). They also showed trends toward greater amounts of stage 1 sleep and lesser amounts of stage REM sleep and stage 3-4 some of the sleep.

HLA testing was performed in only some of the subjects. The highest proportion of positivity for HLA-DR2 occurred in the NC group (9 of 11), with a low-frequency of DR2 positivity in non-narcoleptic groups (4 of 16) and an intermediate incidence for the NwC group (4 of 7). Tests for HLA-DQ1 were positive in 9 of 10 in the NC group, 7 of 7 in the NwC group, and 10 of 15 in the non-narcoleptic group. One Asian-American man with NC was negative for both HLA-DR2 and HLA-DQ1. We did not perform statistical analysis of HLA results because of the small number of subjects tested.

As the initial analysis did not show clear distinctions apart from REM sleep measures among the groups of subjects diagnosed with NwC, IH, and EDS-NOS, we combined these subjects into a single group that we called ``hypersomnia without cataplexy'' (n equals 73). Compared with the NC group, subjects with hypersomnia without cataplexy were less likely to have sleep-related hallucinations and sleep paralysis, but questions related to automatic behavior did not differentiate the two groups Table 2. When compared with control subjects and with a combined group of subjects with nonessential hypersomnia (ISS and SDB), subjects with hypersomnia without cataplexy were more likely to have sleep-related hallucinations, sleep paralysis, and questionnaire responses suggesting the possibility of cataplexy.

To further assess the significance of REM sleep propensity in patients without cataplexy, we then excluded the 10 subjects in the hypersomnia-without-cataplexy group who had any hint of cataplexy as defined in Methods. We divided the remaining 63 subjects into three groups: (a) 16 subjects with REM-sleep propensity of 40% or more (mean equals 63%); (b) 20 with REM-sleep propensity from 1 to 39% (mean equals 22%); and (c) 27 subjects with no sleep-onset REM periods on polysomnography or the MSLT (REM-sleep propensity equals 0%). Sleep habits and symptoms of sleep paralysis, hallucinations, and automatic behavior were not significantly different in the three groups, and there were no statistically significant differences in polygraphic measures apart from REM-sleep-related measures. Table onfile with NAPS. See Note at end of text.)

Discussion.

In this retrospective study, we assessed clinical symptoms, questionnaire responses, and polygraphic findings in six groups of subjects with excessive daytime sleepiness to try to define the relation between sleep-onset REM periods and symptoms. We tried to minimize the impact on our results of disrupted nighttime sleep, which can affect the frequency of sleep-onset REM periods, by selecting subjects for the study who did not have frequent arousals associated with SDB or PLMs. Our findings of significant clinical and polygraphic differences between subjects with NC and all other groups is consistent with previous reports of more sleep disturbance and a higher frequency of sleep paralysis and sleep-related hallucinations in subjects with cataplexy compared with those with NwC and tend to confirm the findings of Moscovitch et al ^[22] and others that NC is a distinct disorder that can be differentiated from other hypersomnia syndromes based on symptoms and polygraphic findings. Our findings that habitually short sleep hours, increased sleep on weekends, and high sleep efficiency differentiate subjects with ISS from other syndromes of excessive sleepiness confirm previous reports and suggest that diagnostic criteria for this disorder are valid.^[29,30] The trends toward fewer hallucinations and sleep paralysis, more awakenings, less REM sleep, and more stage 1 sleep in the group diagnosed with SDB support our impression that sleepiness in these subjects was related to SDB and not to brain dysfunction.

On the other hand, apart from REM sleep measures, we did not find major differences in clinical and polygraphic features between subjects with NwC and those with IH. Symptoms of sleep paralysis and hypnagogic or hypnopompic hallucinations--presumed to be linked to increased propensity for early onset of REM sleep--were not associated with REM sleep propensity in these subjects and were not significantly more common in subjects with NwC than in subjects with IH. Subjects meeting criteria for IH did not sleep longer, take longer naps, or take longer to ``get going'' in the morning. Automatic behavior, included in the ICSD diagnostic criteria for narcolepsy but not for IH Table 1, was useful diagnostically only by being more common in subjects with NC than in a combined group of subjects with mild SDB or ISS; it did not help to differentiate subjects with narcolepsy from those with IH.

Acknowledgments

We thank Ronald D. Chervin and Claudio Bassetti for critical reviews of the manuscript.

Note.

Readers can obtain 3 pages of supplementary material from the National Auxiliary Publications Service, c/o Microfiche Publications, PO Box 3513, Grand Central Station, New York, NY 10163-3513. Request document no. 05238. Remit with your order (not under separate cover), in US funds only, $7.75 for photocopies or $4.00 for microfiche. Outside the United States and Canada, add postage of $4.50 for the first 20 pages and $1.00 for each 10 pages of material thereafter, or $1.75 for the first microfiche and $.50 for each fiche thereafter. There is $15.00 invoicing charge on all orders filled before payment.