Introduction

Top Abstract Introduction Methods Results Discussion References

Impaired alertness chronically affects tens of millions of people (because of narcolepsy, aging, jet-lag, shift-work and other disorders of sleep or circadian timekeeping) and is a major cause of accidents and death in modern society (National Commission on Sleep Disorders Research, 1993). It is remarkable, therefore, that there are few pharmacological therapies available that can selectively enhance waking. Drugs that facilitate catecholamine release and/or block their reuptake, such as amphetamines and the nonamphetamine stimulant methylphenidate, are commonly used to treat impaired alertness in severe disorders such as narcolepsy, but have intractable side effects that limit their general use (Leith and Barrett, 1976; Koob and Bloom, 1988). An alternative wake-promoting compound, modafinil [(diphenyl-methyl)-sulfinyl-2-acetamide], has gained attention for its apparent selectivity in promoting wakefulness and potential for treating narcolepsy, hypersomnia and related performance deficits (Bastuji and Jouvet, 1988; Boivin et al., 1993; Lagarde et al., 1995; Pigeau et al., 1995). Although its mechanism of action is not yet known, modafinil potently increases waking in rodents (Touret et al., 1995), cats (Lin et al., 1992), dogs (Shelton et al., 1995), monkeys (Hermant et al., 1991; Lagarde and Milhaud, 1990) and humans (Lyons and French, 1991) with less peripheral or central side effects (Simon et al., 1994; Buguet et al., 1995) and abuse liability than amphetamines (Gold and Balster, 1996). The relative increase in locomotor stimulation versus wakefulness (an index of the drug's wake-promoting specificity) and the time course of compensatory sleep responses after modafinil waking (a measure of drug interaction with sleep-wake regulatory mechanisms), however, have not been fully characterized within a single species.

Modafinil is pharmacologically distinct from classical psychomotor stimulants (Simon et al., 1995). Unlike amphetamines and methylphenidate, modafinil exhibits only weak affinity for the dopamine uptake carrier site (Mignot et al., 1994), and does not stimulate striatal dopamine release in rodents (de Sereville et al., 1994). Dopamine receptor blockade with haloperidol, or tyrosine hydroxylase inhibition with -methylparatyrosine, block the stimulating effects of amphetamine, but not modafinil, in cats and mice (Lin et al., 1992; Duteil et al., 1990; Simon et al., 1995). Behaviorally, amphetamines elicit anxiety and repetitive stereotyped movements, but modafinil does not (Duteil et al., 1990; Hermant et al., 1991). Although both amphetamine and modafinil increase locomotor motor activity in mice and monkeys (Duteil et al., 1990), it is not known if these effects are proportional to the wakefulness-promoting effects for both compounds. Modafinil does not inhibit spontaneous activity of noradrenergic and dopaminergic brainstem neurons (Akaoka et al., 1991), and unlike psychomotor stimulants, induces cFOS expression in the anterior hypothalamus adjacent to the suprachiasmatic nucleus (Lin et al., 1996); a structure responsible for the generation of sleep-wake circadian rhythms by promoting cortical and behavioral activation at particular times of day (Edgar et al., 1993). Modafinil does not bind to any known adrenergic receptors; nonetheless, the stimulant-like effects of modafinil in cats, mice and monkeys are prevented by pretreatment with the alpha-1 antagonist prazosin (Lin et al., 1992). Furthermore, the beta antagonist propranolol and the alpha antagonist phentolamine attenuate the EEG-wake-promoting effect of modafinil in cats (Lin et al., 1992), but not the locomotor-activity-promoting effect in mice (Duteil et al., 1990). These compounds do not block amphetamine stimulation in cats (Lin et al., 1992). Although modafinil promotes waking in narcoleptic dogs (Shelton et al., 1995), it is ineffective in blocking cateplexy, a symptom that is usually responsive to modulators of adrenergic transmission (Mignot et al., 1994). It therefore seems plausible that modafinil may facilitate waking through indirect mechanisms, perhaps through decreased GABAergic transmission (Tanganelli et al., 1995; Ferraro et al., 1996; Lin et al., 1996).

Several reports have also suggested that, unlike amphetamines, modafinil-induced waking may not elicit strong compensatory sleep responses in various animal species (Lagarde and Milhaud, 1990; Hermant et al., 1991; Lin et al., 1992; Touret et al., 1995); however, controlled, statistical comparisons of recovery sleep (both immediate and delayed) after drug-induced waking are lacking. In a study of humans in which modafinil or amphetamine was administered during sustained sleep deprivation, modafinil-treated subjects showed reduced need for long recovery sleep (Buguet et al., 1995), consistent with animal studies. Unfortunately, possible differences in prior sleep history (e.g., accumulated sleep loss) between the parallel treatment groups prevented a definitive conclusion.

To further determine whether modafinil has selective influences on waking distinct from psychomotor stimulants, we examined the interrelationship between drug-induced wakefulness, motor activity as a function of waking and subsequent compensatory sleep responses in the rat. With use of continuous computerized sleep scoring, we compared the immediate and latent time course of recovery sleep after statistically equal amounts of wakefulness induced by modafinil and the prototypical stimulant METH in rats with matched prior sleep histories.

	Methods

Top Abstract Introduction Methods Results Discussion References

Animal surgery. Adult, male Wistar rats (275-320 g at time of surgery, Charles River Laboratories, Wilmington, MA) were anesthetized (Nembutal, 60 mg/kg) and surgically prepared with a cranial implant that permitted chronic EEG and EMG recording. Body temperature and LMA were monitored via a miniature transmitter (Barrows, Palo Alto, CA) surgically placed in the abdomen. The cranial implant consisted of stainless steel screws (2 frontal [+3.9 AP from bregma, ±2.0 ML] and 2 occipital [6.4 AP, ±5.5 ML]) for EEG recording. Two Teflon-coated stainless steel wires were positioned under the nuchal trapezoid muscles for EMG recording. All leads were soldered to a miniature connector (Microtech, Boothwyn PA) and gas sterilized with ethylene oxide before surgery. The implant assembly was affixed to the skull by the combined adhesion of the EEG recording screws, cyanoacrylate applied between the hermetically sealed implant connector and skull and dental acrylic. An antibiotic (Gentamycin) was administered for 3 to 5 days postsurgery. At least 3 weeks were allowed for postsurgical recovery.

Recording environment. Rats were housed individually within specially modified Nalgene microisolator cages equipped with a low-torque slip-ring commutator (Biella Engineering, Irvine CA) and a custom polycarbonate filter-top riser. These cages were isolated in separate, ventilated compartments of a stainless steel sleep-wake recording chamber. Food and water were available ad libitum and ambient temperature was 24 ± 1°C. A 24-h light-dark cycle (LD 12:12) was maintained throughout the study by 4-watt fluorescent bulbs located approximately 5 cm from the top of each cage. Light intensity was 30 to 35 lux at midlevel inside the cage. Animals were undisturbed for 3 days both before and after treatments.

Automated data collection. Sleep and wake stages were determined with SCORE, a microcomputer-based sleep-wake and physiological monitoring system. SCORE^TM design features, validation in rodents and utility in preclinical drug evaluation have been reported elsewhere (Van Gelder et al., 1991; Edgar et al., 1991,1997; Seidel et al., 1995). In the present study, the system monitored amplified (×10,000) EEG (bandpass, 1-30 Hz; digitization rate, 100 Hz), integrated EMG (bandpass, 10-100 Hz, root mean square integration) and telemetered body temperature and non-specific LMA from up to 64 rodents simultaneously. Arousal states were classified on-line as NREM sleep, REM sleep, wake or theta-dominated wake every 10 s by use of EEG period and amplitude feature extraction and ranked membership algorithms. Individually taught EEG-arousal-state templates and EMG criteria differentiated REM sleep from theta-dominated wakefulness (Welsh et al., 1985). LMA was automatically recorded as discrete events every 10 s. Body temperature was recorded each minute. LMA was detected in both horizontal and vertical planes by a customized telemetry receiver (Data Sciences Inc., St. Paul, MN) located beneath the cage. Telemetry measures (LMA and body temperature) were not part of the SCORE arousal state determination algorithm. Thus, sleep-scoring and telemetry data were parallel but independent measures. Data quality was assured by frequent on-line inspection of the EEG and EMG signals. Raw data quality and sleep-wake scoring was scrutinized further by a combination of graphical and statistical assessments of the data as well as visual examination of the raw EEG wave forms and distribution of integrated EMG values.

Time of treatment. All treatments were administered 5 h after lights-on. This time-point is designated "CT-5" (CT, circadian time; CT-0, lights-on). In both rats and humans, drug efficacy can interact with prior sleep loss (Roehrs et al., 1989; Edgar et al., 1991; Trachsel et al., 1992). Thus, time of treatment can be an important factor in the sensitivity of a preclinical sleep assay. We have empirically determined that, at CT-5, rats normally spend about two thirds of their time asleep, and the effects of stimulants are readily apparent relative to the waking effects of vehicle treatment alone. Testing earlier than CT-5 can infringe on the time of day when the statistical variance in the amount of sleep is high and extremely variable from hour to hour. Testing later than CT-5 can risk allowing the normal circadian surge of wakefulness at lights-off to mask the stimulant effect being measured.

Drug administration and study design. The test compounds, modafinil (Cephalon, Inc., West Chester, PA), and methamphetamine (Sigma, St. Louis, MO), were suspended in sterile 0.25% methylcellulose (pH = 6.2; Upjohn Co., Kalamazoo, MI) and injected intraperitoneally in a volume of 1 ml/kg. Rats were randomly divided into parallel treatment groups. Samples sizes (n) for all monitored variables were between 9 and 12 for active treatment groups. For each of the five active treatments, a baseline matched vehicle-control group, n = 20, was objectively selected from a pool of 90 vehicle-treated control rats. A computerized algorithm identified the 20 vehicle-treated animals whose pretreatment 24-h baseline sleep-wake circadian waveforms best fit (computed by least squares) the mean 24-h pretreatment baseline circadian waveform for the treatment group. This procedure assures that the pretreatment sleep-wake base-line (prior sleep history) in the respective vehicle and active treatment groups are comparable (Edgar et al., 1991, 1997; Seidel et al., 1995).

EEG spectral analysis. Each 10-s epoch of raw EEG signal was digitized (100 Hz) for 24 h after the two higher doses of modafinil and METH, and a randomly selected subset (n = 13) of vehicle controls. All raw EEG data files were carefully scrutinized and marked for artifacts, and analyzed using Hartley's modification of the Fast Fourier Transform (Bracewell, 1986). For each hour post-treatment, EEG power during NREM was calculated in each band (delta = 0.5-4 Hz, theta = 4.1-8 Hz, alpha = 8.1-12 Hz, beta = 12.1-20 Hz) and then expressed as a percentage of total EEG power. All post-treatment epochs scored as NREM and not tagged as artifact were used in the analysis. For the purposes of this study, emphasis was placed on EEG delta power during NREM, which was plotted as a percent of total power. This analysis and applied statistics (noted in "Results") are consistent with methods described previously (Edgar et al., 1991; Seidel et al., 1995).

Data analysis and statistics. The principal variables recorded were minutes per hour of wake, NREM sleep, REM sleep, counts per hour of LMA, and average hourly deviation from the 24-hr mean body temperature. LMA "intensity," calculated as LMA counts per minute awake, was also computed. For each animal, the mean hourly value for each variable was computed for 30 h before and after treatment. Base-line values averaged across the 24-h period preceding treatment were compared using one-way analysis of variance (ANOVA) across groups for all variates, and differences were nonsignificant. Treatment groups were then compared post-treatment by repeated-measures ANOVA. In the presence of a significant main effect, Dunnett's contrasts ( = 0.05) assessed differences between active treatment groups and vehicle controls, unless otherwise specified.

View larger version (61K): [in this window] [in a new window]   Fig. 1.   Waking pharmacodynamic profiles in rats treated with modafinil or METH at CT-5 (5 h after lights-off). Both drugs produced a significant dose-dependent increase in waking relative to vehicle controls (vehicle mean ± S.E.M. = shaded line, active treatments plotted as mean ± S.E.M.).

	Results

Top Abstract Introduction Methods Results Discussion References

Wake-promoting effects. Both modafinil and METH produced dose-dependent increases in EEG-defined waking and rapid onset of action, with differences from vehicle controls evident within 15 min after treatment (see fig. 1). The total, cumulative duration of wake-promoting activities of modafinil (300 mg/kg) and METH (1 mg/kg) was similar and appeared to subside completely within 4 h. During the first 4 h post-treatment, rats given modafinil (300 mg/kg) were awake for 108.7 min more than they had been during the corresponding 4-h baseline period 24 h earlier. For METH (1 mg/kg), awake time increased by 116.0 min (see fig. 2). The difference between these two groups did not approach statistical significance (see table 1). Thus, with respect to both the duration of wake-promoting activity and the total increase in the minutes of wake over base-line, these two doses were essentially equipotent. Modafinil (100 mg/kg) and METH (0.5 mg/kg) were nearly equipotent (see fig. 2 and table 1) although there was a trend (P = .09) for METH (0.5 mg/kg) to be slightly more effective at promoting wake than modafinil (100 mg/kg). Modafinil (30 mg/kg) also evidenced a small but statistically reliable wake-promoting effect, verified by simultaneous contrasts of waking levels 2 h post-treatment with use of the Ryan-Einot-Gabriel-Welsh Multiple F Test (alpha = 0.05, dF = 140).

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Total cumulated minutes of wakefulness (relative to base line 24 h earlier) after modafinil (MODAF) or METH treatment in the rat. * P < .01, **P < .0001 compared with vehicle controls.

Consolidation of waking. Although modafinil (300 mg/kg) and METH (1.0 mg/kg) promoted wake equipotently, the average duration of continuous bouts of wakefulness was longer after modafinil than after METH (see table 2). To simplify the formal statistical comparison between groups, given that the duration of effect varied with dose, only wake bouts initiated within the first 2 h were examined. For this period, the highest dose of modafinil produced significantly longer bouts of continuous wakefulness than METH (1 mg/kg; P < .0001, table 2). For modafinil (100 mg/kg) versus METH (0.5 mg/kg), however, the difference was not statistically significant. Thus, at the highest doses tested, modafinil but not METH suppressed even very brief (20 seconds or more) intrusions of sleep.

LMA. Both doses of METH and the higher 2 doses of modafinil significantly stimulated LMA in a dose-related manner (fig. 3). The duration of increased LMA after METH closely paralleled the duration of its effect on EEG wakefulness, whereas for each dose of modafinil, the LMA stimulation subsided to control levels about one hour before its wake-promoting effect ended. During the first 2 h after treatment, the amount of LMA after METH significantly exceeded modafinil, regardless of dose (fig. 4, table 2). Paradoxically, the initial reaction to modafinil was a slight reduction in LMA; the small decrease from vehicle-control values was observed within 5 min of injection and lasted about 20 min, but did not appear to be strongly dose-related.

View larger version (49K): [in this window] [in a new window]   Fig. 3.   LMA pharmacodynamic profiles in rats treated with modafinil and METH. Vehicle plotted as mean ± S.E.M. (shaded line).

View larger version (19K): [in this window] [in a new window]   Fig. 4.   LMA counts during the initial 2 h after treatment with modafinil (MODAF) or METH in the rat. *P < .05, ** P < .005.

Intensity of LMA. Because wakefulness and LMA both increased after treatment, another variate was required to assess increases in LMA independently of increased wake. Therefore, "LMA intensity" was defined as the number of LMA counts per minute of wakefulness. Figure 5 shows that LMA intensity was not increased above vehicle control levels by modafinil and was not dose-related. With respect to both increased LMA counts and LMA intensity, however, METH greatly exceeded modafinil (P < .0001, table 2). Modafinil dose-dependently increased LMA and wake, but did not disproportionately stimulate motor activity; that is, LMA intensity after modafinil did not differ from vehicle controls. In contrast, METH promoted LMA more potently than it promoted wake, so that LMA intensity increased significantly over controls (P < .0001), and this increase was dose-related.

View larger version (20K): [in this window] [in a new window]   Fig. 5.   LMA "intensity" (activity counts per minute awake) during the initial 2 h after treatment with modafinil (MODAF) or METH in the rat. **P < .0001. Note that LMA was proportional to wake time at all MODAF doses. METH produced a disproportionate and dose-dependent increase in LMA intensity.

Body temperature. As with LMA, modafinil initially resulted in a small drop in body temperature relative to vehicle controls (0.6°C maximum difference), lasting about 30 min (fig. 6). This drop may be related to the small concomitant decrease in LMA (compare with fig. 4). Subsequently, body temperature was slightly elevated for about 4 h (+0.5°C at peak). In contrast, METH dose-dependently increased body temperature. Both doses rapidly increased body temperature above vehicle control levels (+1.1°C at peak for 1 mg/kg METH) and with a time course that closely paralleled its effects on LMA and EEG wake.

View larger version (39K): [in this window] [in a new window]   Fig. 6.   Body temperature profiles during the initial 5 h post-treatment with modafinil or METH. Body temperature was plotted as the °C change from base line, (e.g., normalized with respect to the 24-h (circadian) mean).

Recovery sleep after drug-induced waking. The analysis of sleep loss and sleep recovery was somewhat complicated because the duration of wake-promoting activity and subsequent recovery sleep differed as a function of treatment. Thus, to meaningfully assess the total treatment effect between groups, the data were necessarily blocked according to the timing of the initial wake-promoting effect and the subsequent recovery-sleep (see "Methods"). Table 3 presents the data for NREM sleep reduction during the period of drug stimulation and the subsequent recovery of sleep. In the 4 h after treatment, rats administered modafinil (300 mg/kg) obtained 83.9 min less NREM sleep than they had during the corresponding 4-h baseline interval 24 h earlier. For METH (1 mg/kg), this "NREM deficit" was 91.9 min [not significantly different from modafinil (300 mg/kg)]. By 24 h post-treatment, rats administered modafinil (300 mg/kg) had recovered 65% (54.3 of 83.9 min) of their NREM deficit, and those given METH (1 mg/kg) had recovered 63% of their NREM deficit. (These values were not significantly different.) Thus, both treatments resulted in similar NREM deficits and were followed by a similar amount of NREM recovery sleep. The timing of the NREM recovery sleep, however, was distinctly different between the two drugs.

View larger version (37K): [in this window] [in a new window]   Fig. 7.   Circadian rhythms of nonREM sleep (NREM) before and after treatment with METH or modafinil. After drug-induced waking (decrease in NREM), METH-treated rats showed an immediate rebound hypersomnolence that was not evident after modafinil treatment. Arrow indicates hour of treatment (CT-5). Vehicle mean ± S.E.M. = shaded line. Black bars along the abscissa indicate times when the lights were off.

View larger version (38K): [in this window] [in a new window]   Fig. 8.   NREM deficit profiles in rats treated with higher (upper panel) or lower (lower panel) doses of modafinil or METH. Data are plotted as the cumulative minutes of NREM sleep post-treatment relative to corresponding hours during base line (24 h before treatment). A negative slope in the plot indicates a wake-promoting effect, and a positive slope indicates recovery sleep (relative to base line). Hour 0 corresponds to CT 05:00 Vehicle mean ± S.E.M. = shaded line. Black bar along abscissa indicates when the lights were off.

REM sleep. Modafinil (300 mg/kg) potently interfered with REM sleep for about 8 h post-treatment (see fig. 9). During that time, rats obtained a total of 30.2 min less REM sleep than they had during the corresponding 8-h baseline interval 24 h earlier. By 24 h post-treatment, 57% of this "REM deficit" had been recovered. Figure 9 reveals that the greatest increase in REM sleep relative to vehicle occurred in the first 3 h of the subsequent light phase (e.g., 19 h after treatment). METH (1 mg/kg) produced similar effects (see table 4). A closely similar, dose-related pattern was observed after modafinil (100 mg/kg) versus METH (0.5 mg/kg). Thus modafinil and METH had similar effects regarding both interference with REM sleep and subsequent recovery. The duration of reduced REM sleep was consistently longer after modafinil, however, when compared with doses of METH "equipotent" for waking effects. For example, REM inhibition persisted for 8 h after modafinil (300 mg/kg) but only 5 h after METH (1 mg/kg) (table 4).

View larger version (45K): [in this window] [in a new window]   Fig. 9.   Circadian rhythms of REM sleep before and after treatment with METH or modafinil. Note the REM rebound beginning at CT-0 approximately 19 h after treatment. Arrow indicates hour of treatment (CT-5). Vehicle mean ± S.E.M. = shaded line. Black bars along the abscissa indicate times when the lights were off.

EEG spectra in NREM sleep. As the wake-promoting effect of modafinil (300 mg/kg) and METH (1 mg/kg) subsided, EEG power during NREM sleep shifted to lower frequencies (delta; 0.5-4.0 Hz), and total EEG power during NREM sleep increased (fig. 10). A post hoc analysis of the increased delta EEG during NREM sleep during the period 4 to 9 h post-treatment (depicted in fig. 10) showed a significant treatment effect (F[2,30] = 3.77, P < .05). Scheffé contrasts ( = 0.05) verified that modafinil (300 mg/kg), but not METH (1 mg/kg), increased the relative delta power in NREM sleep during this period compared with vehicle controls; the difference between modafinil (300 mg/kg) and METH (1 mg/kg) was not statistically significant.

View larger version (34K): [in this window] [in a new window]   Fig. 10.   EEG delta power during nonREM sleep after modafinil or METH treatment in rats. In contrast to METH, Modafinil produced a long-lasting elevation in EEG delta power. EEG delta power (0.5-4.0 Hz) is plotted as a percentage of total power (0.5-20 Hz) during corresponding nonREM epochs. Vehicle mean ± S.E.M. = shaded line. Black bar along the abscissa indicates the time when the lights were off.

	Discussion

Top Abstract Introduction Methods Results Discussion References

The present study shows that modafinil has potent wake-promoting effects that are more specific than those of the classical psychomotor stimulant METH. Several distinct features of modafinil differentiate it from prototypical stimulants. We found that modafinil increased LMA in rodents consistent with previous reports (Duteil et al., 1990; Simon et al., 1996), but only in proportion to the amount that was expected because of increased time awake. This was in sharp contrast to METH, in which LMA disproportionately and dose-dependently increased relative to waking time. The difference in LMA intensity measures between modafinil and METH likely reflects important differences in their central effects on striatal dopamine release, which is potently induced by amphetamines, but not by modafinil (Akaoka et al., 1991; de Serevile et al., 1994). By definition, if LMA per unit time awake is increased above base-line, an animal is "hyperactive." It is our contention that, in a preclinical drug evaluation, an ideal (e.g., selective) wake-promoting therapeutic will enhance alertness without inducing hyperkinesis. It is perhaps noteworthy that, in addition to METH, methylphenidate and caffeine increase LMA intensity (D.M. Edgar and W.F. Seidel, unpublished data). Caffeine-induced anxiety and motor side effects are well documented in humans (cf., Nickell and Uhde, 1994) and, despite public toleration of these side effects, renders it less than ideal as a wake-promoting therapeutic. It is not yet known, however, if our preclinical LMA intensity measures in rats will reliably predict anxiety or anxiety-related motor side effects in humans. Because LMA measures have not been routinely described as a function of a drug's wake-promoting efficacy, additional studies across a wide variety of wake-promoting compounds will be needed to confirm the predictive utility of this measure.

Another distinguishing feature of modafinil was its clear lack of "rebound hypersomnolence," an intense interval of compensatory sleep after drug-induced waking. Sleep homeostasis, a process with unknown mechanisms, usually invokes compensatory sleep responses (e.g., increase in sleepiness, sleep time and depth of sleep) that are proportional to prior wake duration (cf., Borbely and Neuhaus, 1979; Carskadon and Dement, 1979; Dijk et al., 1990; Daan et al., 1984). The present study suggests that some wake-promoting compounds (e.g., modafinil) can influence the time course of recovery sleep so that it is more gradual. Modafinil's effects on recovery sleep were thus quite distinct from METH, which exhibited strong rebound hypersomnolence, a response similar to that after total sleep deprivation.

Previous studies have suggested that modafinil may lack strong compensatory sleep responses (Lagarde and Milhaud, 1990; Lin et al., 1992; Touret et al., 1995), but the present work constitutes the first systematic approach and statistical evaluation of recovery sleep after drug-induced waking in a laboratory animal. Our results are consistent with a cursory report of waking and compensatory sleep after modafinil and d-amphetamine treatment in rats (Touret et al., 1995); unfortunately, that study presented no statistics to support any claim regarding NREM or REM compensatory sleep in the rat. Our results showed that approximately 1 h of NREM sleep deficit imposed by the waking effects of modafinil (100 mg/kg) or METH (0.5 mg/kg) was gradually but ultimately completely recovered about 15 h after initial treatment. In contrast, only about 64% of the approximately 90 min of NREM sleep deficit imposed by higher doses of these drugs was ultimately recovered. The latter finding is generally consistent with the notion that larger amounts of sleep deprivation are only partially recovered (Mistlberger et al., 1983). One factor that may have influenced the rate of recovery sleep at both doses was time of treatment. It is plausible that there may be one or more interactions between circadian factors modulating the expression of sleep-wakefulness and the recovery sleep process (Borbely and Neuhaus, 1979; Edgar et al., 1993), or possibly interactions caused by ceiling effects (e.g., rats rarely sleep more than 75-80% per hour during the rest phase; see fig. 7 and Edgar et al., 1991). The opportunity to increase sleep time during the usual activity phase (lights-off) in the rat could possibly be greater than during the rest phase if, during the latter, the animals are already sleeping maximally. Controlled studies of equipotent drug effects at multiple times of day in intact and suprachiasmatic nuclei-lesioned animals would be necessary to establish the presence of temporal interactions.

Compensatory REM sleep was also observed in the 24 h after modafinil-induce waking, with largest amounts occurring at the beginning of the usual rest phase (lights-on). This observation raises questions about the generality of a previous report that modafinil displaces and/or inhibits REM sleep without subsequent compensatory REM sleep (Touret et al., 1995). One possible explanation, again, is that aspects of recovery sleep may be gated by or otherwise interact with the circadian timing system. In humans, REM sleep is under strong circadian control (Czeisler et al., 1980). Rats, too, exhibit a robust REM sleep circadian rhythm (see fig. 9). In the study by Touret and colleagues, rats were treated at lights-on; 5 h earlier in the circadian cycle than in the present study. Thus, they may have missed REM recovery in the subsequent light-phase (beyond their 24 h post-treatment measures). The timing of REM sleep recovery in our present study was also interesting in that it was not paralleled by a concomitant elevation in NREM sleep (relative to base-line). One model of sleep regulation posits that REM sleep propensity is a function of prior NREM (Benington and Heller, 1994). On a hour-to-hour basis, we observed that NREM sleep was elevated only slightly relative to base-line and vehicle controls after the primary waking effect subsided. No additional compensatory NREM occurred in the light phase subsequent to treatment. Thus, consistent with other studies (Seidel et al., 1995), compensatory REM sleep can occur with a time course independent from that of NREM sleep.

Although modafinil did not show an intense compensatory sleep response after its primary waking effect, EEG spectral analyses supported the NREM recovery data presented in this study, which suggests that there may have been an underlying sleep debt that was gradually repaid. Elevated EEG power in the delta band (0.5-4.0 Hz) during NREM sleep is related to the duration of prior sleep loss (Borbely and Neuhaus, 1979; Dijk et al., 1990), and is a correlate of depth of sleep. During the gradual course of modafinil recovery sleep, EEG delta power was elevated relative to METH- and vehicle-treated animals. This could be interpreted as reflecting an augmented underlying sleep need subsequent to drug-induced waking. Alternatively, the EEG patterns may reflect a greater depth of sleep that need not be related to the status of the underlying sleep drive. Finally, the increased EEG delta power may reflect a dissociation of the dominant EEG frequency in NREM sleep, independent of the homeostatic state of sleep regulation in the animal. Studies of modafinil interaction with preexisting sleepiness from sleep deprivation (recently completed in our laboratory) should help differentiate these possibilities.

Both modafinil and METH significantly increased total wake time, but modafinil produced more consolidated bouts of wakefulness. This was an unexpected and interesting feature that further differentiated these classes of compounds. In humans, the quality of wakefulness per se (e.g., independent of performance or cognition measures) can be difficult to assay because the daily episode of waking is normally highly consolidated. In contrast, the rat normally has polyphasic sleep-wake patterns, and the individual bout durations of sleep or wakefulness can offer potentially useful insights into drug interaction with arousal state maintenance. By treating rats at CT-5 in the present study, the pharmacodynamic properties of modafinil and METH were tested at a time in which natural sleep propensity was high (akin to elevated physiological sleepiness associated with disorders of excessive sleepiness). Under these conditions, modafinil appeared more effective in staving off intervening sleep (that is, opposing preexisting drive toward sleep or sleep "pressure") than METH.

In summary, modafinil potently promoted wake time and wake consolidation without intensifying LMA or producing rebound hypersomnolence. The specificity of modafinil's wake-promoting effects further differentiate it from classical psychomotor stimulants and raise important practical issues. When assessing the utility of a wake-promoting compound for the general population, rebound hypersomnolence after the primary waking effect presents a serious safety concern, especially for individuals operating motor vehicles or performing other hazardous functions. Modafinil's lack of rebound hypersomnolence could be a favorable and important requisite for novel wake-promoting drugs (Edgar, 1996). Finally, the selective wake-promoting effects of modafinil bring to the forefront the importance of direct EEG sleep-wake assessments in the sleep-wake therapeutic discovery process. Drugs with seemingly little potential based on locomotion-dependent behavioral pharmacology techniques could go unappreciated without direct measures of arousal state.