nogram (PSG) as well as multiple sleep latency testing. (MSLT) prior to diagnosis. ..... Latency to stage 1 as a function of group and MSL T session. (mean, SEM).
Sleep, 14(2): 121-129 © 1991 Association of Professional Sleep Societies
Pupillometric Assessment of Excessive Daytime Sleepiness in Narcolepsy-Cataplexy Jan Newman and Roger Broughton Schools of Psychology and Medicine, University of Ottawa, Ottawa, Canada
Summary: Ten untreated patients with narcolepsy--cataplexy and age- and sex-matched normals between the ages of 20 and 71 years underwent pupillometric analyses immediately prior to each of five multiple sleep latency test sessions. Although narcoleptics were sleepier as indicated both by their Stanford Sleepiness Scale ratings and by their latencies to sleep onset, the baseline pupil diameter, pupillary light reflex, and pupillary orienting response did not differentiate between groups. Narcoleptics did, however, exhibit a significantly greater frequency of spontaneous oscillations in the dark-adapted state than did controls. These findings indicate that pupillary stability may serve as a supplementary diagnostic tool for narcolepsy--cataplexy. The results are discussed with the view that psychosensory restoration of alertness, among other extraneous variables, must be controlled when utilizing pupillometric techniques. A review of the literature indicates a variety of methodological and statistical shortcomings that must be amended. Suggestions are made for improving the reliability and validity of the pupillometric approach. Key Words: Pupillometry-Narcolepsy-Cataplexy-Excessive daytime sleepiness-Orienting response-Psychosensory restitution.
For the majority of individuals, the preservation of arousal and the initiation and maintenance of sleep are easily regulated processes. For narcoleptics however, the ability to ward off sleepiness and to dictate the timing of sleep is generally beyond voluntary control. Accordingly, the patient with narcolepsy--cataplexy presents with chronic excessive daytime sleepiness (EDS) often also expressed as irresistible sleep attacks, apparent memory deterioration, impaired job performance, and, at least when untreated, increased vehicular accidents (1-4). Unfortunately, this unrelenting pressure for sleep remains relatively refractory to treatment (5) and is in critical need of further investigation and understanding. A relatively unexplored approach for the detection and quantification ofEDS is that of pupillometry. Electronic pupillometry is a precise technique of infrared scanning and recording of pupillary diameter at rest or in response to stimuli (6). Its current employment has been based upon the belief that a large and stable pupil diameter is associated with alertness, whereas a constricted and unstable pupil is associated with sleepiness Accepted for publication November 1990. Address correspondence to J. Newman, Human Neurosciences Research Unit, Health Sciences Building, 451 Smyth Rd., Ottawa, Canada KIH 8M5. Address reprint requests to Roger Broughton M.D., Ottawa General Hospital, 501 Smyth Rd., Ottawa, Canada KIH 8L6.
(7,8). These findings, in addition to the notion that pupillary diameter is largely beyond voluntary control, have prompted some investigators to adopt pupillary change as an apparently sensitive and reliable index of arousal level. In keeping with this view, some research has shown that narcoleptics exhibit a smaller resting pupil diameter than controls (8-11). Contrary to these findings, others have been unable to distinguish between groups on the basis of pupil size alone (12). Moreover, this measure has not been able to discriminate sleep-deprived normals from alert controls (13) nor detect ultradian fluctuations in alertness (14). Pupillary stability as an index of pathological sleepiness has also yielded conflicting results. Although some research indicates that narcoleptics experience greater and/or more frequent spontaneous pupillary oscillations than normal (8,15), other studies suggest that the inverse is true (9) or that no differentiation of groups is possible using this measure (10). Such incongruity is not present in normative data, as drowsiness due to sleep deprivation (13) and to ultradian periodicity (14) has been shown to increase pupillary variability. However, as these latter two studies remain unconfirmed, further research is necessary to determine their validity. It has also been suggested that changes in the pupillary light reflex (PLR) indicate the presence of EDS
121
122
J. NEWMAN AND R. BROUGHTON
(16). The PLR, as the name implies, refers to the reflex constriction and redilation of the pupil in response to light stimuli. Its shape is dependent upon the interplay of the parasympathetic and sympathetic nervous systems and is presumably altered in a predictable fashion with drowsiness or fatigue (17). The results, however, have proven confusing. Although the majority of fi~ search suggests that sleepiness, whether pathological (8,11,18) or normal (13,14), is not manifested in the PLR, other studies (9) have found that narcoleptics exhibit a reduced contraction to light. In opposition to the latter, an increase in the extent of pupillary constriction has been described in sleepy normals (17) and in tired insomniacs (6). Such marked discrepancies found throughout the pupillometry literature suggest that the utility of this approach, as currently applied, is dubious. Baseline pupil diameter, pupillary stability, and the PLR do not appear to systematically vary with changes in arousal. Inadequacies in methodology and in statistical applications probably can account for a proportion of the discrepancies and deserve mention. First, there has been a tendency to utilize very small samples (10), which limits the generalizability ofthe results. Second, as already noted by others (12), in many research designs (8,9) no normal controls have been included. . Although there may be some interest in comparing various patient populations without controls, it is critical that any diagnostically useful technique be able to discriminate between pathological sleepiness and normal alertness. Finally, underreporting of statistical details including approach used, significance levels, and summary tables has made interpretation and replication of some existing findings (8,9) difficult. The main objective of this study was therefore to clarify the efficacy, or lack of efficacy, of pupillometric measures in the detection and quantification of EDS using a methodologically correct experimental design. In addition to the standard measures of pupillary diameter and reactivity, the pupillary orienting response (OR) was included as an alternate index of arousal. This supplemental reflex pupillary response was chosen as it is purportedly a relatively uncontaminatl~d measure of arousal level (19).
METHODS Subjects Six male and 4 female patients diagnosed as having had narcolepsy-cataplexy for at least 5 years (all presenting with sleep attacks, cataplexy, and EDS) (ASDC criteria) (20) and 10 age- and sex-matched nonnapping individuals with normal nocturnal sleep duration (6.5Sleep, Vol. 14, No.2, 1991
8.0 h of sleep/night) were studied. Narcoleptics ranged in age from 21 to 70 years [mean 48.6 ± 13.6 (SD)] and controls were 20 to 71 years (49.0 ± 13.8). All subjects underwent an overnight screening polysomnogram (PSG) as well as multiple sleep latency testing (MSLT) prior to diagnosis. Subjects were excluded from the study if their sleep apnea index exceeded five per hour or if more than 20% of their nocturnal sleep period was characterized by periodic movements in sleep (PMS). All subjects were further screened for suitability through the use of sleep histories, personal sleep logs, and audiometric testing (Amplivox Audiometer Model 2) for normal hearing. In narcoleptics, tricyclics were withdrawn for a 3-week period preceding the study (four patients had been taking anticataplectic medications) and central nervous system (CNS) stimulants (seven patients had been taking methylphenidate; no subjects were receiving amphetamines) were withdrawn for 1 week. Control subjects were remunerated for their participation. The protocol was accepted by the local human experimental procedures committee and consent forms were signed.
Apparatus The pupillometer is an instrument designed to record and transmit changes in pupil diameter. The polymetric pupillometer (Model V-1165-lR) used in this study consisted of a metal frame to which a chin rest, an RCA video camera, and a Bausch and Lomb infrared light apparatus were mounted. Monocular video signals from the system were monitored on a television screen and charted on a 10-channel Grass Model 6 electroencephalographic (EEG) machine. The hard copy records were then manually scored with reference to calibration signals. The system had a sensitivity of 0.1 mm for pupillary diameter. Light stimuli were administered to both eyes through a multiple stimulus display box manufactured by the Lafayette Instrument Company. The magnitude of the stimulus and the area of illumination varied as a function of the desired response. Auditory stimuli for orienting response (OR) elicitation were delivered to both ears by headphones. They consisted of 90-dB tones with a 50-ms rise-time and 2-s duration. The positioning of a toggle switch controlled tone frequency (1,000 or 2,000 Hz). On day 1, 24-h ambulatory monitoring of sleepwake variables was accomplished through the use of a four-channel Medilog recorder (Oxford Medical Systems). Standard polysomnographic variables were monitored continuously during the following lO-h test day by the EEG machine along with pupillary diameter and synch pulse during the pupillometric sessions.
PUPILLOMETRIC ASSESSMENT OF EDS
Procedure
123
then recorded. During the final phase of the test session, the PLR was elicited by the administration of 10 O.I-s low intensity (30 lux) light flashes (3.6 x 4.9 cm display) at 10-20-s intervals. Half of the participants received the auditory stimuli first and the other half received the visual sequence first. All stimuli were presented manually to prevent coincidence of stimuli and recording artifacts, including blinks and ptosis, which were monitored in the PSG. If at any time lid closure occurred (evident by direct observation on the videomonitor or on the polygraphic recording), the subject was verbally alerted through an intercom system. (Dozing-off rarely occurred during test sessions in either group.) As dilatation naturally results from the alerting procedure itself, stimulus presentation was resumed only upon pupil restabilization to baseline diameter and was terminated when a total of 10 artifact-free PLR responses was recorded. (The use of an eyelid crutch was found to be both awkward and disruptive.) The SSS was completed every 30 min as well as promptly following each MSLT session.
Day 1: ambulatory monitoring ofprior sleep. At 0800 h, subjects reported to the lab for audiometric testing and electrode application. C3-A2 EEG, right and left outer canthi electrooculograms (EOG), and submental electromyograms (EMG) were recorded using the portable Medilog unit. Participants wore the recording device for a 24-h period beginning at 0830 h. This permitted documentation of prior night sleep, which might influence subsequent pupillometric measures. During this time, they were to adhere closely to their daily routines and to complete the Stanford Sleepiness Scale (SSS) (21) every 30 min during wakefulness. Day 2: pupillometry and MSLT. Upon returning to the lab at 0830 h, the Medilog system was removed and electrodes for the MSLT were applied. These included C3-A2, C4-A 1, 02-A 1, right and left EOG, and submental EMG. Subjects were then briefed on the protocol and familiarized with the equipment. Pupillary and polygraphic calibrations were made. The MSLT was administered at 1000, 1200, 1400, 1600, and 1800 h. Each scheduled nap was terminated following either 20 min of wakefulness or 10 min of Pupillary variables sleep (with sleep onset defined as two consecutive 40-s epochs of stage 1 or REM). Sleep stage criteria were 1) Dark-adapted pupil diameter: The mean pupil dithose of Rechtschaffen and Kales (22). ameter derived from 30 samples taken from the Just prior to each MSLT nap, an I8-min pupillomefirst recorded minute of each test session (mm). try protocol was followed. 2) Latency to maximum constriction: The latency The subject was fitted with welding glasses for a from stimulus onset to the point of minimum 7-min dark-adaptation period. Not only did this appupil diameter (s). proach serve to increase pupil diameter to accepted 3) Extent of constriction: The difference between pudark-adapted levels, it also permitted subjects to move pil diameter at the onset of the stimulus and that freely about the laboratory and to thereby maintain at the point of maximum constriction (mm). alertness. The individual was then instructed to sit in 4) Rate of constriction: The ratio of extent of cona darkened room at the pupillometer with his/her chin striction to the latency to maximum constriction on the chin-rest and to fixate on a dimly illuminated (mm/s). (0.02 lux) square display (1.2 x 1.2 cm) at eye level 5) Constriction velocity: The slope of the primary some 80 cm in front. This target restricted movement constriction phase. of the eyes away from the camera and permitted rel6) Maximum redilated diameter: The largest pupil atively artifact-free recording of the dark-adapted pupil diameter achieved during the 10-s period followfor a I-min period. ing stimulus offset (mm). Immediately thereafter, 11 tone bursts at 10-30-sec 7) Rate of redilation: The ratio of pupil diameter to intervals were presented. The frequency of the tones the latency to maximum redilation (taken at I-mm alternated between 1,000 Hz and 2,000 Hz. Half of intervals and then averaged) (mm/s). the subjects heard 1,000-Hz tones prior to their first, 8) Redilation velocity: The slope of the secondary third, and fifth naps and 2,000-Hz tones prior to their redilation phase. second and fourth naps. The remaining subjects heard 9) Pupillary OR: The difference between prestimulus the inverse pattern beginning with the 2,000-Hz series. pupil diameter (mean of five samples taken at I-s To dishabituate within each session, the 11 th tone of intervals immediately preceding the tone) and pueach series differed from the first 10 presentations. pil diameter achieved following the stimulus (mean of three samples taken from the I-s period followFollowing the final auditory stimulus, subjects were ing the tone) (mm). given a I-min break during which fixation was still required but no stimuli were presented. An additional 10) Spontaneous oscillations: Number of pupillary fluctuations ::::0.2 mm (no.lmin). I-min period of spontaneous pupillary fluctuations was Sleep, Vol. 14, No.2, 1991
J. NEWMAN AND R. BROUGHTON
124 Data analysis
SUbjective sleepiness measures for narcoleptics and controls were analyzed across test sessions, as well as immediately preceding versus promptly following each MSLT session, using a 2 x 2 x 5 related samples analysis of rank means (23). Comparisons ofSSS scores every half hour across days 1 and 2 for both groups were made using a 2 x 2 x 21 related samples analysis of rank means (23). Post hoc simple effects were extracted using a modified nonparametric Tukey honestly significant difference (HSD) statistic suggested by Marascuilo and McSweeney (24). MSLT variables, including the latencies to stages 1, 2, 3, 4, and rapid eye movement (REM) sleep we:re analyzed between groups and across sessions with 2 x 5 analyses of variance (ANOVA). All statistical comparisons of pupillary variables were made between groups across the five test sessions and for each individual stimulus where applicable. Intergroup variations in dark-adapted pupil diameter and in the frequency of spontaneous oscillations were investigated using a 2 x 5 ANOVA and a 2 x 5 related samples analysis of rank means (23), respectively. The PLR indices, including latency to maximum constriction, extent of constriction, rate of constriction, maximum redilated diameter, and rate of redilation, were compared using 2 x 5 x 10 ANOVAs. Constriction velocity and redilation velocity for the first stimulus of each session were analyzed using 2 x 5 ANOVAs. Due to the additional dishabituation stimulus presented in the auditory series, the pupillary orienting responses were tested using 2 x 5 x 11 ANOVAs. Repeated measures ANOVAs were used in order to determine the mean effects of visual and auditory stimuli as well as the impact of each individual stimulus within and across sessions for both groups. Significant stimulus effects were further scrutinized using the Newman-Keuls test (25). Prior night sleep effects were removed from the pupillary data by repeated measures analyses of covariance (ANCOVA) (25). Significance achieved in ANOVA and ANCOVA interactions was determined by the conservative Geisser-Greenhouse (GG) approach (25). Within group MSLT and spontaneous oscillation data were collapsed across sessions and correlated using the Pearson product-moment correlation (26). These data were then correlated with prenap SSS scores using the Spearman rank correlation (27). RESULTS SSS findings When SSS scores for both groups were analyzed across days I and 2 at 30-min intervals, the data inSleep, Vol. 14, No.2, 1991
---- Narcoleptics ---- Controls
4
3
2
0830
1030
1230
1430
1630
1830
Time (hrs) FIG. 1. Mean SSS ratings for narcoleptics and controls recorded at 30-min intervals. Data are collapsed across day 1 (with home 24-h ambulatory sleep-wake monitoring) and day 2 (laboratory day with pupillometry and MSLT testing).
dicated that narcoleptics felt subjectively sleepier (mean 2.95 ± SEM 0.56) than controls (1.74 ± 0.32) overall, H(1) = 106.32, p < 0.001 (Fig. 1). In addition to this main group effect, day main effects (ambulatory day versus testing day) and time of day main effects (based on 21 time points corresponding to 0830 through 1830 h) were also present. Specifically, both narcoleptics and controls rated themselves as feeling sleepier on day 1 (2.71 ± 0.56) than on day 2 (1.97 ± 0.34), H(1) = 29.50, p < 0.001 (Fig. 2). Both groups also displayed pronounced sleepiness at particular times of the day, H(20) = 35.15, P < 0.02 with increases in sleepiness occurring at 1000 h, from 1300 to 1400 h, from 1500 to 1600 h, and at 1800 h. Accordingly, examination of the day x time interaction indicated that sleepy periods on day 1 significantly differed from those on day 2, H(20) = 33.04, p < 0.05. Post-hoc analyses indicated that SSS scores at 1100, 1300, 1330, and 1500 h were significantly greater in the home relative to the lab. Analysis of SSS scores for the MSLT day alone revealed that narcoleptics reported themselves as feeling sleepier (2.62 ± 0.41) than controls (2.12 ± 0.35) in general, H(1) = 5.35, p < 0.03. In addition to this main group effect, all subjects proved sleepier in the prenap condition (2.53 ± 0.31) relative to the postnap condition (2.21 ± 0.23), H(1) = 4.27, p < 0.05. SSS scores did not significantly correlate with latency to sleep onset in the narcoleptic (r = -0.26, p > 0.01) or in the control (r = -0.28, P > 0.01) group. MSLT findings Statistical analyses of the MSLT data revealed that narcoleptics experienced shorter latencies to stage 1 (3.35 ± 1.48) compared to controls (10.05 ± 2.10),
125
PUPILLOMETRIC ASSESSMENT OF EDS Day 1 Day 2
4
N arcoleptics Controls
14
--->:: 's
12 10
:>.,
u
>::
....,ill
08
ro
06
Po
04
!:
02
....:I ill
lfJ
0 0830
1230
1030
1430
1630
1000
1830
1200
1600
1400
1800
Time (hrs)
Time (hrs) FIG. 2. Mean SSS ratings at 30-min intervals on day I (with home 24-h ambulatory sleep-wake monitoring) and day 2 (laboratory day with pupillometry and MSLT testing). Data are collapsed across groups. Increased subjective sleepiness in the home environment is evident.
F(1,18) = 13.10, p < 0.002 (Fig. 3). As also expected,
the data indicated that narcoleptics entered REM sleep earlier (9.67 ± 2.40) than their counterpart controls who showed no REM patterns (20.00 ± 0.00), F(1, 18) = 29.57, p < 0.001. In essence, the MSLT findings reaffirmed the diagnosis of narcolepsy~ataplexy. In contrast, latencies to stages 2 and 3 did not differentiate groups. Stage 4 was not achieved by either group. Interrater reliability for sleep staging was 0.90. Pupillary findings
FIG. 3. Latency to stage 1 as a function of group and MSLT session (mean, SEM).
8: 5.55 ± 0.16; stirn 9: 5.51 ± 0.16; stirn 10: 5.48 ± 0.16) F(9,162) = 3.16, p < 0.002 and an increase in the magnitude of the pupillary OR following the 11 th tone (stirn 1: 0.32 ± 0.05; stirn 2: 0.12 ± 0.05; stirn 3: 0.17 ± 0.04; stirn 4: 0.19 ± 0.03; stirn 5: 0.09 ± 0.03, stirn 6: 0.12 ± 0.06; stirn 7: 0.12 ± 0.05; stirn 8: 0.05 ± 0.05; stirn 9: 0.05 ± 0.04; stirn 10: 0.20 ± 0.05; stirn 11: 0.29 ± 0.05) F(10,180) = 3.99, p < 0.001. The former may reflect habituation of one component of the PLR, whereas the latter indicates the successful employment of a dishabituation stimulus in the tone series. No additional significance was achieved by any pupillary variable upon the statistical removal of prior amount of sleep. The frequency of spontaneous oscillations significantly correlated with MSLT latencies in the controls (r = -0.52, p < 0.001) but not in the narcoleptics (r = -0.16, p > 0.01). No significant correlations were found between spontaneous oscillations and SSS scores in either group.
There were no significant group differences whatsoever in baseline pupil diameter, in any component of the PLR, or in the amplitude of the pupillary OR (see Tables 1-4 and Figs. 4 and 5). Number of spontaneous oscillations per min in narcoleptics (25.76 ± 4.50) was, however, significantly increased over controls (11.36 ± 2.78), H(I) = 21.27, p < 0.001 (Fig. 6). Other significant findings included a reduction in the magnitude ofthe redilated diameter across the 10 light DISCUSSION stimuli (stirn 1: 5.81 ± 0.15; stirn 2: 5.67 ± 0.16; stirn 3: 5.51 ± 0.16; stirn 4: 5.51 ± 0.16; stirn 5: 5.52 ± In addition to the overall group difference in sub0.15; stirn 6: 5.56 ± 0.16; stirn 7: 5.61 ± 0.16; stirn jective sleepiness, the SSS results indicated that all TABLE 1. Means and standard errors for dark-adapted pupil diameter (mm) Time (h) 1000 1200 1400 1600 1800 Overall group mean ns, Not significant.
Narcoleptics 5.47 4.91 5.81 5.67 5.37 5.45
± ± ± ± ± ±
0.62 0.34 0.51 0.41 0.45 0.47
TABLE 2. Means and standard errors for pupillary constriction (s)
Controls
p value
Time (h)
± ± ± ± ± ±
ns ns ns ns ns ns
1000 1200 1400 1600 1800 Overall group mean
6.04 5.76 5.93 5.86 6.17 5.95
0.41 0.54 0.46 0.43 0.52 0.47
Narcoleptics 1.03 .94 1.05 1.06 1.04 1.02
± ± ± ± ± ±
0.17 0.15 0.16 0.13 0.17 0.16
Controls
p value
± ± ± ± ± ±
ns ns ns ns ns ns
1.11 1.15 1.13 1.15 1.16 1.14
0.10 0.08 0.08 0.09 0.09 0.09
ns, Not significant. Sleep, Vol. 14, No.2, 1991
126
J. NEWMAN AND R. BROUGHTON 7
8 8 ...
(l) .....,
1.2
Narcoleptics Controls
S S bD
II
(l)
8ell
I
..... Q
...... ..... p.,
1.0
(l)
6
5
;:l
0....
>::
0.9
..c:
O.B
ell
u
;>.,
...
ell ...... ....... ....
1000
1200
1400
1600
0.7 0.6
p., ;:l
0....
Narcoleptics Controls
1.1
0.5 1000
1800
1200
1400
1600
1800
Time (hrs)
Time (hrs) FIG. 4. Maximum redilated pupillary diameter achieved during the pupillary light reflex as a function of group and testing session (mean, SEM).
FIG. S. Extent of pupillary constriction achieved during the pupillary light reflex as a function of group and testing session (mean, SEM).
subjects experienced greater sleepiness on day 1 relative to day 2. No doubt this was due to the unstructured nature of the time spent at home as well as to the enforcement of alertness between MSLT naps on the laboratory day (28). The decreases in sleepiness immediately following MSLT naps confirmed the subjective recuperative effect of naps in general (29,30). The greatly increased sleepiness experienced by the narcoleptics was objectified by their significantly shorter MSLT latencies. Unexpectedly, this greater slee:piness in narcoleptics was not reflected in the majority of the pupillographic measures. However, this failure to discriminate sleepy narcoleptics from alert controls using most pupillary measures does not necessarily negate the value of the pupil as an index of sleepiness. In fact, some aspects of pupillary physiology not only indicate the level of alertness, they can also reflect a variety of psychophysiological functions including mental effort, attention, anxiety, excitement, motivation, thought, counting, and muscle tension (31-35). Bumke (36) aptly summarized the situation when he stated: "In general every active intellectual prOCI~SS, every psychical effort, every exertion of attention, every active mental image, regardless of content, particularly every affect just as truly produces pupil enlarge-
ment as does every sensory stimulus (p. 60)." In neurophysiological terms, psychosensory and physical stimuli alter the existing sympathetic-parasympathetic balance and thus modify the course of arousal (37). These often uncontrolled factors may explain why both intersubject and intrasubject pupillary variability are quite high (38). To control for the many extraneous sources of error and thereby tease out the effects of level of alertness alone, it has been suggested that cognitive relaxation be emphasized in pupillary studies (39). This condition was met in the current study only during the 2-min reprieves from sensory stimuli. As subjects were then permitted to temporarily blink freely and were well informed of the existence and length of the break, it is conceivable that this segment of the experiment was experienced as less demanding or stressful and thus was more likely to promote sleepiness. It is notable that the only pupillary variable that significantly distinguished narcoleptics from controls was the greater number of spontaneous oscillations experienced during this 2-min interval. The failure to detect similar group differences in pupillary fluctuations during the baseline recording periods may have been due to feelings of anticipation or anxiety at the onset of each test.
TABLE 3. Means and standard errors/or pupillary OR to tones (mm)
TABLE 4. Means and standard errors/or redilation velocity (slope)
Time (h) 1000 1200 . 1400 1600 1800 Overall group mean ns, Not significant. Sleep. Vol. 14, No.2, 1991
Narcoleptics 0.23 0.19 0.20 0.15 0.16 0.19
± ± ± ± ± ±
0.20 0.26 0.16 0.16 0.13 0.18
Controls
p value
Time (h).
± ± ± ± ± ±
ns ns ns ns ns ns
1000 1200 1400 1600 1800 Overall group mean ns, Not significant.
0.14 0.12 0.14 0.12 0.09 0.12
0.09 0.12 0.10 0.08 0.09 0.10
Narcoleptics 0.35 0.23 0.29 0.21 0.26 0.27
±0.09 ± 0.05 ± 0.07 ± 0.05 ± 0.09 ± 0.07
Controls 0.26 0.24 0.19 0.15 0.22 0.21
± ± ± ± ± ±
0.06 0.06 0.04 0.03 0.04 0.05
p value ns ns ns ns ns ns
PUPILLOMETRIC ASSESSMENT OF EDS
""
Narcoleptics Controls
40
(j)
.........
35
lfJ
>:: 0
30
~
25
:;; '(j
1
lfJ
0
20
lfJ
;:l 0
.::'"
...,c:l >:: 0
15 10
r
Po. [JJ
1000
1200
1400
1600
1800
Time (hrs) FIG. 6. Frequency of spontaneous oscillations as a function of group and testing session (mean, SEM).
Narcoleptics are in fact capable of producing normal pupillographic tracings during periods of heightened arousal (40). These findings, in conjunction with the observed significant correlation between spontaneous oscillations and MSLT latencies in the controls, suggest that spontaneous oscillations may be a robust measure of level of alertness, which in turn may be more clearly isolated when the individual is at his or her ease. Under the latter condition, boredom is promoted and the effort needed to remain awake is maximal, particularly for the chronically sleepy narcoleptic group. It is this very struggle to maintain alertness that is believed to result in autonomic imbalance and consequent pupillary fluctuations (17). More specifically, as the individual becomes sleepy, sympathetic influence begins to deteriorate and supranuclear inhibition gradually decreases, resulting in pupillary constriction (37). As sensations of sleepiness become even greater, the individual makes conscious efforts to remain awake, which result in temporary inhibition of the oculomotor nucleus, an increase in sympathetic tone, and pupillary dilation. As the individual has periodic success at alerting himself or herself, the pupil rhythmically constricts and redilates (17). On the basis of this knowledge, it makes sense to utilize a protocol that promotes sleepiness yet requires alertness when measuring spontaneous oscillations as a function of arousal. As narcoleptics appear to be particularly vulnerable to such conditions (41), it seems likely that, given the same circumstances, other EDS patients will also display increased pupillary activity. Perhaps the conflicting reports in the literature regarding the relationship between spontaneous pupil activity and arousal level (15,17 versus 9,10,12) are largely due to the differing conditions under which these oscillations have been measured. It is expected
127
that procedural standardization will aid in clearly demonstrating that pupil variability during conditions of cognitive relaxation is the most reliable pupillometric measure of EDS. In contrast to pupil stability, the PLR and pupillary OR did not differentiate between groups. The relative ineffectiveness of these measures may have been due to the alerting effects of the stimuli themselves (39) and/or to the probable intrusion of various psychosensory factors including counting (37). As sleepiness is discouraged under such conditions, it is not surprising that pupil changes as a function of EDS were undetected. Perhaps the methodology can be altered in the future such that the alerting effects of the stimuli are curtailed (i.e., lower intensity) and the struggle to remain awake is enhanced (i.e., longer intervals between stimuli). Although there has been some suggestion in previous work that the PLR may be used to distinguish sleepy from alert individuals (9,16), the majority of recent findings indicate otherwise (11,13,18). Perhaps such inconsistencies in the literature are due to poor or varying design, a lack of standardization of experimental conditions, interfering sensory or emotional stimuli (42), andlorinterindividual differences. Indeed the PLR may well vary a great deal among individuals, as well as quite erratically within the same individual from moment to moment (43). In addition to the alerting effects of the stimuli, it is likely that such variability (44) also accounts for the inadequacy of the pupillary orienting response (OR) as a measure of EDS. Moreover, one must question the validity of this measure, as the observed minute changes in diameter following the tones may have been due to spontaneous activity rather than to the orienting reflexes themselves (45). In keeping with the PLR and OR data, similar null results were also achieved for the baseline diameter variable. These findings may be explained in terms of the alerting effects of motivational factors andlor feelings of anxiety experienced upon the initiation of each test. Yet such factors cannot account for the negative results obtained during the supposedly relaxing 2-min breaks. Why is it that pupil diameter does not serve to differentiate groups at a time when the narcoleptics, un~ like the controls, are clearly struggling to remain awake and have exceedingly short sleep latencies? If one expects the spontaneous oscillations to cancel one another out, the underriding baseline diameter should remain uncontaminated and thereby reflect these differing levels of alertness. Perhaps in the final analysis, resting pupil diameter may not prove to be a sensitive measure of arousal level in the usual laboratory setting, as some have proposed (7,8). As these authors have Sleep. Vol. 14. No.2. 1991
128
1. NEWMAN AND R. BROUGHTON
reported decreases in pupil diameter with increases in sleepiness, it is difficult to understand why our untreated and very sleepy narcoleptic patients did not show this change. Perhaps the methodological shortcomings outlined in the introduction are accountable to some degree. In order to fully clarify the potential usefulness of pupillometry for the detection and assessment ofEDS, corrective action in a variety of areas must be taken. First, there must be standardization of dependent measures. In this study, any unelicited fluctuation in pupil diameter of 0.2 mm or more was considered to be a spontaneous oscillation. A specific amplitude criterion, however, has not been universally adopted. The quantification of pupillary instability has varied from strictly statistical measures such as the standard deviation (13) or range (14) of movement to classification on the basis of amplitude and duration (46). If any degree of reliability across studies is to be achieved,. it is critical that uniform measures be agreed upon. As a corollary, it is essential that the conditions under which each variable is measured be carefully controlled and be clearly detailed. Second, the experimental design itself must minimize unwanted error variance due to such factors as age, time of day effects, prior amount of sleep, the nature of the required response, and psychosensory restoration of alertness, to name a few. Finally, and foremost, the validity of pupillary response as a measure of sleepiness, both in nomlal sleepers and in those who present with EDS as symptomatic of differing disorders, must be further assessed. As the number of spontaneous oscillations did not significantly correlate with SSS scores in either group or with MSLT latencies in the narcoleptics, it is difficult to state with any certainty what precisely is being measured. It in fact still remains unclear whether pupillometry primarily reflects variations in the alertnesssleepiness dimension or, on the other hand, variations in autonomic arousal (33). Acknowledgments: This study was supported by the Medical Research Council of Canada (grant MT329 1). We express special thanks to D. Wayne Dunham for his statistical expertise. This work was presented in part at the Association of Professional Sleep Societies Annual Meeting, Washington, DC, U.S.A., June 1989.
REFERENCES 1. BrOUghton R, Ghanem Q. The impact of compound narcolepsy on the life of the patient. In: Guilleminault C, Dement W, Passouant P, eds. Narcolepsy. New York: Spectrum Publications, 1976, pp 201-20. 2. Pressman M, Spielman A, Pollak C, Weitzman E. Long-latency auditory evoked responses during sleep deprivation and in narcolepsy. Sleep 1982;5:S147-56. Sleep. Vol. 14. No.2. 1991
3. Aguirre M, Broughton R, Stuss D. Does memory impairment exist in narcolepsy-
