Eating and its Relationships with Subjective Alertness and Sleep in Narcoleptic SUbjects Living without Temporal Cues

Transcription

1 Sleep 13(6): , Raven Press, Ltd., New York 1990 Association of Professional Sleep Societies Eating and its Relationships with Subjective Alertness and Sleep in Narcoleptic SUbjects Living without Temporal Cues *Charles P. Pollak and tjudith Green *Department of Psychiatry, Cornell University Medical College, New York, New York; tpsychology Department, The William Paterson College of New Jersey, Wayne, New Jersey; and the Institute of Chronobiology, New York Hospital-Cornell Medical Center, Westchester Division, White Plains, New York, U.S.A. Summary: The sleep and meal patterns of normal subjects appear to be governed by a common timing mechanism. To investigate whether the postulated mechanism may be disordered in narcolepsy, a disorder of sleep timing, we analyzed the sleep, eating, and subjective alertness of six narcoleptic subjects and seven normal controls while they lived in a temporal isolation laboratory. When subjects were free to eat and sleep whenever they chose ("freerunning"), the meal patterns and nutrient intakes of the free-running narcoleptic subjects and controls were similar; no evidence of an eating disorder intrinsic to narcolepsy was found. When meals were offered on a 24-h schedule, narcoleptic subjects ate more frequently than did the controls. In two of three narcoleptic subjects whose free-running biological days markedly lengthened, intermeal intervals lengthened proportionately. This was evidence that the timing of sleep (bed-dark) periods and meals was indeed governed by a common mechanism. Meal onsets of narcoleptic subjects were preceded by a 90-min period of decreased napping and, when meals were scheduled, by increased subjective alertness. They were followed by a I50-min period of increased napping and decreased subjective alertness. Postprandial deactivation was not found in controls. The deactivation could not be explained by a postabsorptive effect of food, since the contents of meals that were followed by naps did not differ from those that were not followed by naps. While a preabsorptive effect of meals has not been ruled out, we suggest that postmeal deactivation may be evidence that the mechanism that times sleep periods and meals also modulates the temporal pattern of narcoleptic naps. Key Words: N arcolepsy-temporal isolation-human eating pattern-circadian Alertness-Sleepiness-Napping. Previous studies of normal subjects living in isolation from all temporal cues suggested that free-running sleep and meal patterns are governed by a common timing Accepted for publication May Address correspondence and reprint requests to Dr. C. P. Pollak at Institute of Chronobiology, New York Hospital-Cornell Medical Center, Westchester Division, 21 Bloomingdale Road, White Plains, NY 10605, U.S.A. 467

2 468 C. P. POLLAK AND J. GREEN mechanism (1,2). In those studies, intermeal intervals increased nearly in proportion to the sleep-wake peiiod in subjects whose sleep--\vake cycle lengthened and desynchronized from other circadian rhythms. We have recently found that narcoleptic subjects living without environmental temporal cues were unable to limit sleep to circadian time intervals, but did not spend more time sleeping than controls (3). Narcolepsy is thus a disorder of the timing, not the duration, of sleep. To investigate whether the postulated common timing mechanism for sleep and meals is disordered in narcolepsy, we analyzed the temporal organization of meals of the same time-isolated narcoleptic subjects. We also examined their nutrient intake, subjective alertness, and tendency to nap around mealtimes, since it has been proposed that narcolepsy is associated with an eating disorder that includes increased intake, snacking, overweight, and postprandial drowsiness (4). METHODS Six narcoleptic subjects were compared with seven normal controls. The narcoleptic group consisted of two men and four women; the controls were two men and five women. The mean ages of the groups were 55.2 ± 9.9 (SD) and 57.3 ± 13.4 years, respectively, and did not differ significantly. The narcoleptic subjects were initially heavier than the controls, but the-mean body mass indices did not differ significantly (Table 1). I The diagnosis of narcolepsy was based on a clinical history of long-standing, daily,' excessive daytime sleepiness plus occurrences of cataplexy. All narcoleptic subjects had an abnormal, five-nap Multiple Sleep Latency Test (5,6) and were drugfree for 14 TABLE 1. Starting weights and weight changes during temporal isolation Starting weight Weight change Code names kg kg/cm2a kg % Narcoleptic subjects ZZOI ZZ ZZ ZZ ZZ ZZ Mean 80.5t ** -2.0 SD Controls FR FR FR FR FR FR FR Mean SD t Differs from controls by t test, tp < ** By paired t test, p < a Body mass index. Sleep, Vol. /3, No.6, 1990

3 EATING IN NARCOLEPSY 469 days or more at the time of the study. The controls denied all symptoms of narcolepsy. In the course of the study, continuous observation of the daytime behavior, subjective alertness ratings, and polygraphic sleep recordings of the control subjects confirmed that they did not have narcolepsy. All subjects were in good general health, as shown by physical examination and routine laboratory tests performed just before the start of the study. All gave informed consent before entering the laboratory. The subjects lived for consecutive days in temporal isolation apartments of the Institute of Chronobiology. The apartments are comfortably furnished but have no clocks, windows, telephone, radio, or television. Subjects were always able to communicate with technicians in an adjacent control area, either by calling on a two-way intercom or by pressing a call button. During periods when subjects were out of bed and lights were on, technicians entered the apartment about three times an hour to draw blood samples from an indwelling venous catheter, serve food, and take away dishes. (The blood samples were used for hormone determinations, which will be reported in another paper.) The technicians' work schedules were randomized so subjects could not infer the time of day from the person on duty. The investigators paid calls to the subjects at irregular times. During most of their waking time, subjects were free to read, listen to music, occupy themselves with hobbies, exercise, or converse with a technician. During the first 4 days (controls) or 6 days (narcoleptic subjects), the subjects followed a 24-h sleep schedule that was tailored to resemble the pattern recorded by them at home for 2 weeks before entering the laboratory. Room lights were cycled by the experimenters between -300 lux during waking periods and 0 lux during sleeping periods. Meals were also scheduled during this period: breakfast was served 1-2 h after rise time; an evening snack was served 2 h before scheduled bedtime; lunch and dinner were served at equal intervals during the remaining time spent out of bed. Each meal was announced by a technician. The subject could either decline the meal or request it by selecting the appropriate meal name ("breakfast," "lunch," "dinner," or "snack") from a list displayed on a computer monitor in the apartment. Subjects were free to choose what they ate from a range of preferences recorded in intake interviews. After the period of scheduled sleep and meal opportunities, subjects were allowed to retire to bed, arise, and eat whenever they chose ("free-running"). The narcoleptic subjects free ran for 8-14 (mean of 9.7) biological days. Some control subjects were studied for longer periods of time, but to maintain comparability, data from only the first 10 consecutive biological days were used. Room lights were lowered to 0 lux whenever a subject retired to bed, and they were increased to -300 lux when a subject signaled that it was time to get up. To request a meal while free-running, the subjects selected "meal request" from the computer menu. Any meal could be requested at any time with the provision that subjects name the meal as described above. Subjects selected the contents of the meals and ate at their own pace. All foods were recorded by a technician at the time they were served. The total caloric content of meals as well as the quantity of each macronutrient were computed using the Nutritionist III diet analysis program (N-Squared Computing, Silverton, OR, 1985). Narcoleptic subjects wore sleep recording electrodes, and a continuous polygraphic recording of sleep and wakefulness was made for the entire duration of the study. Together with continuous observation of behavior by closed-circuit television, this made it possible to detect any episodes of sleep that occurred when subjects were out Sleep, Vol. 13, No.6, 1990

4 470 c. P. POLLAK AND 1. GREEN of bed. Subjects were instructed to confine sleep to the periods spent in bed and darkness, but any naps detected by observation or polygraphic recording were not interrupted. Using a visual analog scale displayed on the monitor of a personal computer, each narcoleptic subject rated the subjective level of alertness at randomized intervals of (mean of 20) min. A technician used the intercom to inform the subject when a rating was due. If a narcoleptic subject happened to be napping at the time, the subject was not awakened and the rating was omitted. In many subjects, the frequency distribution of alertness ratings did not approximate a normal distribution. The ratings were therefore treated as ordinal data by converting them to percentiles based on the alertness ratings obtained from that subject. Mean levels of subjective alertness before and after meals were then calculated by the following method for subjects in each group (narcolepsy and control) and condition (scheduled and free-running): the percentiles were replaced by estimates calculated from them by linear interpolation at 5 min intervals, from 120 min before meal onset to 120 min after meal onset. (The analysis period was limited in order to minimize mixing the effects of adjacent meals.) Estimates at corresponding times were then averaged over all meals eaten by a subject under either the scheduled or free-running condition. After inspection, the resulting mean curves of the subjects were aligned and averaged to yield curves of subjective alertness for each group of subjects. The statistical significance of changes in mean alertness was determined by analysis of variance for repeated measures (7). RESULTS Temporal organization of meals During the initial, scheduled portion of the experiment, narcoleptic subjects ate more often than controls and nearly as often as meals were offered to them (3.9 times/day or 0.24 times/h spent out of bed). Later, while free-running, they ate at a frequency similar to that of the controls (0.19 meals/h out of bed). The number of meals per biological day remained high, but the meals were spread over a longer period spent out of bed (Table 2). The most common free-running meal sequence of both subject groups was breakfastlunch-dinner, in that order (Table 3). A breakfast-lunch-dinner-snack sequence was more frequent in the narcoleptic subjects, suggesting that their extra meals were snacks eaten as the final meal of the day. While free-running, all narcoleptic subjects and controls maintained a circadian pattern of periods spent in bed in a dark room (rectangles in Fig. 1). In two of three narcoleptic subjects whose biological days spontaneously lengthened and lost their synchrony with other circadian rhythms, intermeal intervals increased in proportion to the increase in the biological day. These relationships are shown by scattergrams (Fig. 2) and can also be seen in raster plots of meals and biological days (Fig. 1). Thus, the meal and sleep patterns of at least some narcoleptic subjects appeared to be governed by a common timing mechanism, as previously found for normal subjects (1). Nutrients consumed The nutritional intake of the narcoleptic subjects and controls was calculated in three ways: nutrients per 24 h, nutrients per biological day, and nutrients per meal (Table 2). Four nutritional measures were calculated: total kilocalories and grams of protein, Sleep. Vol. 13, No.6, 1990

5 EATING IN NARCOLEPSY 471 TABLE 2. Meal occurrences and nutrients consumed Scheduled Free-running Narcoleptic Narcoleptic subjects Controls subjects Controls Meal occurrences Duration of out-of bed periods (min) (22.6) 1,004.3 (40.9) 1,147.4 (82.1)* 1,029.4 (80.6) Meals/24 h 3.9 (0.08)* 3.4 (0.63) 3.1 (0.52) 2.8 (0.43) Mealslbiological day 3.9 (0.09)* 3.4 (0.63) 3.5 (O.4)t 3.1 (0.4) Mealslh out of bed 0.24 (0.01)* 0.20 (0.04) 0.19 (0.03) 0.18 (0.03) Nutrients/24 h Total (kcal) 1,719.0 (259.7) 2,024.6 (753.0) 1,393 (227.4) 1,733.1 (630.0) Protein (g) 93.2 (12.6) 96.4 (27.4) 72.9 (21.6) 88.6 (26.1) Carbohydrate (g) (70.4) (105.0) (44.7) (88.6) Fat (g) 62.6 (17.7) 70.5 (33.3) 44.7 (17.1) 64.8 (26.8) N utrientslbiological day Total (kcal) 1,717.8 (259.2) 2,024.6 (752.8) 1,542.7 (266.8) 1,884.5 (674.3) Protein (g) 93.1 (12.6) 96.4 (27.4) 79.5 (19.1) 96.7 (29.3) Carbohydrate (g) (70A) (104.9) (87.1) (93A) Fat (g) 62.5 (17.6) 70.5 (33.3) 49.7 (20.0) 70.8 (29.2) Nutrients/meal Total (kcal) (72.5) (211.7) (112.4) (148.2) Protein (g) 23.9 (3.2) 28.7 (7.5) 23.9 (5.3)* 31.0 (6.0) Carbohydrate (g) 54.3 (18.7) 78.3 (31.0) 64.5 (28.9) 69.3 (23.5) Fat (g) 16.2 (4.8) 21.1 (10.1) 15.3 (6.6) 23.1 (8.2) Values are mean (SD). Differs from controls by t test, *p < 0.05; tp < carbohydrate, and fat. During the scheduled portion of the experiment, the mean of each of the resulting 12 measures of intake was smaller in the narcoleptic subjects, but none differed significantly. The mean of each measure was also smaller when the narcoleptic subjects were free-running, but the only significant difference was in protein per meal. Much of the difference between groups was accounted for by a male control subject of average weight and body mass index whose intake was unusually large (2,967 kcall24 h, 3,180 kcal/cycle, 880 kcal/meal). Both the narcoleptic subjects and the controls lost weight during temporal isolation. The effect was more consistent among the narcoleptic subjects, but their mean weight change did not significantly exceed that of the controls (Table I). Alertness and naps around mealtimes Meal-associated changes in alertness were examined in the scheduled and freerunning conditions for breakfasts, lunches, dinners, and snacks. When all scheduled meals of the narcoleptic subjects were considered jointly, they were found to be preceded by increasing levels of SUbjective alertness. Alertness increased from 90 to 120 min before meal onset to a peak at the time of meal onset, and then rapidly decreased over a period of min to below the premeal baseline (Fig. 3, left panel). The F ratio for 30 and 150 dfwas 5.1 (p < 0.001). In the free-running condition, subjective alertness was stable until 5-10 min before meal onset, when it increased to a peak before decreasing over min to below the premeal baseline (Fig. 3, middle panel). F for 30 and 150 dfwas 2.77 (p < 0.001). The controls also rated themselves as being more alert at meal onset. Instead of decreasing after meals, however, the increased alertness persisted for> 1 h after meal onset before returning to the premeal baseline (Fig. 3, right panel). F for 30 and 180 dfwas 1.88 (p < 0.01). Alertness did not change significantly in the controls when meals were scheduled. Sleep, Vol. 13, No.6, 1990

6 472 C. P. POLLAK AND J. GREEN \<IN I I I I I en o til c=:::j B L o ( c=j c::::==j B L 0 c::==j c::::==j B L 0 c=j c=jb L o ( c=j c::==j B L o c c::==j c:=::j B L o c r==:::j c:::==::::j B L 0::, I I I IB L P I I I IB L 0 I I I IB L 0 c::==j I c=jb L 0 B L o c::::==j c=jb L 0 c=:::j c==jb L o c:==j I IB L o c:::j CJ Tire of Day FIG. 1. Raster of bed--dark periods and meals of a narcoleptic subject (code name "ZZ02"). Rectangles represent periods spent in bed and darkness. Sequential pairs of such periods are shown on the same line to delimit the periods spent out of bed in the light when eating took place. B: breakfast. L: lunch. D: dinner. S: snack. MN: solar midnight. For the first six calendar days, as indicated by the numbers in the left-hand margin, the bed--dark periods and the meals were scheduled by the experimenters on a 24-h cycle. For days 7-16, the bed--dark periods were selected by the subject ("free-running"). All narcoleptic subjects napped while they were out of bed in the light. The naps could be considered pathological, since instructions had been given to avoid naps and all controls were able to comply with the instructions. The occurrence and duration of such naps provided a window on sleep tendency near meal times that is not available in other subjects. The mean temporal pattern of naps is shown with that of dietary intake in Fig. 4. Intake of the controls is also shown. When meals were scheduled, eating was concentrated in three narrow bands plus an additional period of intake (snacks) late in the day in both groups of subjects (Fig. 4, left panel). The narcoleptic subjects spent less time awake after each of the major meals. The sleep tendency was most prominent after lunch. During the free-running portion of the experiment, when the day length sometimes varied widely (Fig. 2), eating by both groups can again be seen to occur in three major episodes, although the peaks are flatter (Fig. 4, right panel). The caloric intake of the narcoleptic subjects lagged behind that of the controls, and there was a greater tendency to snack late in the day. The only distinct increase in nap tendency occurred after the second eating peak (lunch). The lack of distinct, meal-related fluctuations in sleep tendency in the free-running condition could result from misalignments of meals and naps, which occurred at different times on different days. The intervals from naps to meal onsets were therefore I MN Sleep. Vol. 13, No.6, 1990

7 EATING IN NARCOLEPSY 473 SLbject ZZOl Subject ZZ02 SL..bject ZZ()5 ro > 0' 8 '- Q) 0...,...,. ~ x * (f) Q) ro 4 Q)... * * * li"*,. E :::J * * '- C,. * * Q).~... C H r=0.68 r=d.73 r= Al Al Biological Day Length (minutes x 100) FIG. 2. Scattergrams relating biological day length to intermeal interval for three free-running narcoleptic subjects for whom the biological day spontaneously lengthened and de synchronized from other circadian rhythms. Regression lines and correlation coefficients are also shown. In two of these cases (ZZOI and ZZ02), the regression lines pass close to the origin, indicating that changes in intermeal interval changed in proportion to changes in the length of the biological day. When the uppermost, outlying point of ZZ06 was removed, r = 0.27 and the regression line passed well above the origin. measured directly. The frequency of naps was found to differ markedly before and after meals (Fig. 5). Considering scheduled meals first, nap frequency decreased ~90 min before meal onset and increased sharply ~ 30 min after meal onset. Postmeal nap frequency remained high until min after meal onset, when it sharply decreased to the premeallevel. The F ratio for 11 and 55 dfwas 8.26 (p < 0.001). Free-running meals were preceded by only a slight reduction of nap frequency but were also followed by a marked increase of napping. F for 11 and 55 df was 4.17 (p < 0.001). Similar TABLE 3. Meal sequences during biological days Narcoleptic subjects Controls Meal No. of % of No. of % of sequence days days days days B-L-D B-L-D-S 23 40*** 3 5 B-S-D 0 0** 8 13 B-L 0 0* 6 10 S-B-L-S B-L-S-D B-S-L-D S-B-S-L B-L-S-D-S I 2 I 2 B-S-S-D 0 0 I 2 S-8-L-S-S I B-D-S I B-D I Total >3 meals 30 52*** Any snack 31 53* breakfast; L-Iunch; D-dinner; S-snack. Differs from controls by X 2, df = I, ***p < **p < *p < Sleep, Vol. 13, No.6, 1990

8 474 C. P. POLLAK AND J. GREEN Q) Narc - Sched Narc - Free-run Control - Free-run..., c 100 ~ <- 'B. 80 ~ (f) (f) 60..., ~ <- Q) ro 40 ~111"If+i Q) 20 >..., &l..., 0.0 ell Time from Meal Onset (hours) FIG. 3. Percentiles of subjective alertness by time before and after meal onset (-120 to 180 min) for six narcoleptic subjects under the scheduled condition (left panel) and free-running condition (middle panel), and seven free-running controls (right panel). Mean :t SEM. effects were present separately for breakfast, lunch, and dinner. No conclusions could be drawn regarding snacks, most of which were eaten within a few hours of bedtime. The effect of meal content on nap frequency was analyzed in the following way: Meals eaten by the narcoleptic subjects under the scheduled and free-running conditions were divided into two categories, i.e., those whose onset was followed by the onset of sleep min later and those not followed by sleep. This was the interval during which the frequency of postmeal sleep onsets was most markedly elevated (Fig. 5). It was found that mean meal size as well as mean protein, carbohydrate, and fat content did not differ significantly between meals of the two categories. A similar SchedUled Free-running (f) QJ 75 co <- 300 ~ 0 ~ ~ , , :;,!Ii u < Q) a... o o o Time after arising (circadian degrees) FIG. 4. Wakefulness and caloric intake by time of the biological day during the scheduled portion of the experiment (left panel) and the free-running portion (right panel). Upper traces: mean percent of time that narcoleptic subjects were awake (scale at far right). Each data point represents 15 or 'I24th of the biological day, measured from waketime to waketime. Heavy lower traces: calories consumed by narcoleptic subjects (scale at left). Light lower traces: calories consumed by controls. 100 o QJ ~ Sleep, Vol. 13, No.6, 1990

9 EATING IN NARCOLEPSY 475 Scheduled Free-rLllning If)..., w If) S 20 \ilw Ui 15 <to..., c w 10 () '- w a Time from Meal Onset (hours) FIG. S. Distributions of intervals from sleep onset (vertical line) to meal onset of narcoleptic subjects under scheduled and free-running conditions. All meals (breakfasts, lunches, dinners, and snacks) were included. The intervals from 2.5 h before to 3.5 h after meal onset were sorted into 30 min bins. No additional meal onsets occurred during this interval. The frequencies were normalized as percentages of the total number of sleep onsets within the interval and averaged across subjects. Mean ± SEM. comparison was made between the two categories of meals for each subject and each nutrient measure; no significant differences were found. We conclude that while naps were more likely to follow meals than to precede them, they were not more likely to follow larger meals or meals containing more of any macronutrient. DISCUSSION There were no significant differences in the number, frequency, size, or macronutrient composition of meals consumed by the narcoleptic subjects and controls when they were free to eat ad libitum in the free-running condition. This was true whether meal occurrences were calculated on the basis of the 24-h day or the biological day. The narcoleptic subjects did eat more snacks than the controls, but this could be explained by proportionately longer periods spent out of bed and therefore longer opportunities to eat. The findings do not support the existence of an eating disorder intrinsic to narcolepsy. It must be noted, however, that the narcolepsy subjects accepted a larger number of meals during the initial portion of the experiment when meals were offered to them four times a day. Since these meals were smaller than the meals of the controls, overall nutrient intake of the groups did not differ even then. The biological days of three of the six free-running narcoleptic subjects spontaneously lengthened and lost synchrony with other circadian rhythms for at least a few cycles. This event has been observed in 46% of the normal subjects who free-ran in our laboratory and about one-half of the subjects studied by Aschoff and Wever (8). As previously observed in normal subjects (1,2,9), the intermeal interval remained proportionate to the biological day in two of the three narcoleptic subjects whose rhythms Sleep, Vol. 13, No.6, 1990

10 476 c. P. POLLAK AND 1. GREEN became de synchronized. This suggests that their meals and sleep periods were timed by a common mechanism. The timing mechanism for sieep has been described either as a second circadian oscillator that is distinct from the oscillator that governs the circadian body temperature and plasma cortisol rhythms (10,11) or as a homeostatic process governing primarily slow-wave sleep (12,13). In either case, the present data provide no reason to think that the mechanism functions abnormally in people with narcolepsy. When the meals of the narcoleptic subjects were scheduled every 24 h (as was a daily period of darkness and bed rest), they were preceded by an ~90-min period of decreased napping and increased subjective alertness. Alertness peaked at the time eating began. Changes in sleep tendency and subjective alertness were less marked before free-running meals, and controls showed no premeal activation under either condition. To our knowledge, increased activation preceding and during the act of eating has not previously been described in humans, while it has long been known that animals become motorically active as the time for a meal approaches (14). Such food anticipatory activity rhythms have been shown to depend on a circadian timing system (15,16), but the small degree of activation preceding subject-initiated meals suggests that an additional mechanism (such as activation conditioned to circadian time) operated in our narcoleptic subjects. Meal onset in the narcoleptic subjects was followed within 10 min by a decrease in subjective alertness. This could represent a loss of premeal activation, but the decrease continued to below the premeal baseline; an additional, deactivating effect was therefore present. This was confirmed by an increase in the frequency of naps far above the premeallevel. These changes in alertness and napping lasted several hours. Despite the degree of deactivation, however, we have found that cognitive and motor performance were maintained close to premeal levels (17). Somnolence and sleep have been described as components of a characteristic sequence of postmeal behaviors in animals (18). If this postmeal effect were mediated by absorbed nutrients, we would expect the loss of alertness and the frequency of sleep onsets to be correlated with meal size and composition. We found instead that meals that were followed by naps did not differ in size or macronutrient composition from meals that were not followed by naps. This does not necessarily mean that diets are of no use in the treatment of narcolepsy. The narcoleptic subjects studied by us varied the size and content of their meals without apparent regard for the effects this might have on their symptoms. Larger or more consistent changes in diet might have measurable effects on postprandial alertness and sleep tendency. Alternatively, preabsorptive effects of the meal may have been responsible for the tendency of the narcoleptic subjects to fall asleep after meals. Gastric distention, for example, stimulates vagal afferents and activation of certain vagal fibers can induce cortical synchronization similar to that of slow-wave sleep (19). The absence of postmeal deactivation or napping in the normal controls could then be explained by supposing that the vagal effects of eating were too weak to affect subjects who were not sensitized by the underlying pathophysiology of narcolepsy. Another possibility is that reduced nap frequency around mealtime might induce a compensatory increase in postmeal sleep tendency, since narcoleptic naps fill part of a required sleep quota (3). This would provide a different explanation of why only the narcoleptic subjects showed postmeal deactivation. A final possibility is that neither the increase in postmeal napping nor the decrease in subjective alertness was contingent on the meal itself. Evidence against a sleep- Sleep. Vol. /3, No.6, 1990

11 EATING IN NARCOLEPSY 477 inducing effect of meals can be found in the literature: administering an -1,000 kcal meal to normal subjects at 10 a.m. did not hasten the onset of sleep or increase subjective sleepiness (20), and replacing lunch with small, frequent meals did not abolish the normally enhanced ability of normal subjects to fall asleep in the afternoon (21). Rather than the meals themselves, the trigger for postmeal deactivation may be the timing mechanism for meals that was alluded to earlier. Since the postulated timing mechanism governs the timing of both sleep periods and meals, it might also influence the timing of naps or, alternatively, periods of wakefulness. This would explain the strong tendency of narcoleptic naps to follow meals rather than precede them, since wakefulness would be maintained while the propensity to eat is increasing. After the onset of a meal, both the propensity to eat and to remain awake would decline and, if the subject is narcoleptic, naps would intrude. Our data do not provide crucial evidence for either a meal-related or chronobiological mechanism of postmeal deactivation. The question might be resolved by experiments that manipulate the sizes and macronutrient compositions of meals administered to free-running narcoleptic subjects. Whatever the physiological relationships between eating and sleeping in narcolepsy turn out to be, the finding that naps are much more likely to follow meals than to precede them suggests that mealtimes must be taken into account in any analysis of the occurrence and timing of narcoleptic naps. Acknowledgment: We are grateful to M. Moline, A. Stroud, and other members of the Institute of Chronobiology who were instrumental in collecting the data and to G. P. Smith for reviewing the manuscript. This research was supported by National Institute of Mental Health grant POI MH37814 and National Institute on Aging grant POl AG04135 to C. P. Pollak and a grant from the Center for Applied Science of William Paterson College to J. Green. REFERENCES I. Green J, Pollak CP, Smith GP. The effect of desynchronization on meal patterns of humans living in time isolation. Physiol Behav 1987;39: Green J, Pollak CP, Smith GP. Meal size and intermeal interval in human subjects in time isolation. Physiol Behav 1987;41: Pollak CP, Wagner DR, Moline ML. The duration of sleep and short-term need for sleep are normal in narcoleptic subjects living without temporal cues (submitted for publication). 4. BellI. Diet histories in narcolepsy. In: Guilleminault C, Dement WC, Passouant P, eds. Narcolepsy. New York: Spectrum Publications, Inc., 1976: Carskadon MA, Dement WC, Mitler MM, Roth T, Westbrook PR, Keenan S. Guidelines forthe multiple sleep latency test (MSLT): a standard measure of sleepiness. Sleep 1986;9: Mitler MM, Van den Hoed J, Carskadon MA, et al. REM sleep episodes during the multiple sleep latency test in narcoleptic patients. Electroencephalogr Clin NeurophysioI1979;46: Winer BJ. Statistical principles in experimental design, second edition. New York: McGraw-Hili Book Company, Wever R. The circadian system of man. New York: Springer-Verlag, Inc., Aschoff J, von Goetz C, Wildgruber C, Wever RA. Meal timing in humans during isolation without time cues. J Bioi Rhythms 1986;1: Kronauer RE, Czeisler CA, Pilato SF, Moore-Ede MC, Weitzman ED. Mathematical model of the human circadian system with two interacting oscillators. Am J PhysioI1982;242:R Kronauer RE. Modeling principles for human circadian rhythms. In: Czeisler CA, ed. Mathematical models of the circadian sleep-wake cycle. New York: Raven Press, 1984; Borbely AA. A two-process model of sleep regulation. Hum Neurobiol 1982;1: Daan S, Beersma DGM, Borbely AA. Timing of human sleep: recovery process gated by a circadian pacemaker. Am J Physiol 1984;246:RI Richter CPo A behavioristic study of the activity of the rat. Comp Psychol Monogr 1922;1: Bolles RC, DeLorge J. The rat's adjustment to a-diurnal feeding cycles. J Comp Physiol Psychol 1962;55: Sleep, Vol. 13, No.6, 1990

12 478 c. P. POLLAK AND J. GREEN 16. Bolles RC, Stokes LW. Rat's anticipation of two meals a day. J Comp Physiol PsychoI1965;69: Pollak CP; Moline ML, Wagner DR. Cognitive and motor performance in narcoleptic subjects living in temporal isolation. Presented at the Annual Meeting, Northeastern Sleep Society, Boston, March 31- April I, Antin J, Gibbs J, Holt J, Young RC, Smith GP. Cholecystokinin elicits the complete behavioral sequence of satiety in rats. J Comp Physiol Psychol 1975;89: Chase MH, Nakamura Y, Clemente CD, Sterman MB. Afferent vagal stimulation: neurographic correlates of induced EEG synchronization and desynchronization. Brain Res 1967;5: Stahl ML, Orr WC, Bollinger C. Postprandial sleepiness: objective documentation via polysomnography. Sleep 1983;6: Carskadon MA, Dement WC. Daytime sleepiness: quantification of a behavioral state. Neurosci Biobehav Rev 1987;11: Sleep, Vol. /3, No.6, 1990