Circadian and Wake-Dependent Effects on the Pupil Light Reflex in Response to Narrow-Bandwidth Light Pulses

Abstract

Purpose.: Nonvisual light-dependent functions in humans are conveyed mainly by intrinsically photosensitive retinal ganglion cells, which express melanopsin as photopigment. We aimed to identify the effects of circadian phase and sleepiness across 24 hours on various aspects of the pupil response to light stimulation.

Methods.: We tested 10 healthy adults hourly in two 12-hour sessions covering a 24-hour period. Pupil responses to narrow bandwidth red (635 ± 18 nm) and blue (463 ± 24 nm) light (duration of 1 and 30 seconds) at equal photon fluxes were recorded, and correlated with salivary melatonin concentrations at the same circadian phases and to subjective sleepiness ratings. The magnitude of pupil constriction was determined from minimal pupil size. The post-stimulus pupil response was assessed from the pupil size at 6 seconds following light offset, the area within the redilation curve, and the exponential rate of redilation.

Results.: Among the measured parameters, the pupil size 6 seconds after light offset correlated with melatonin concentrations (P < 0.05) and showed a significant modulation over 24 hours with maximal values after the nocturnal peak of melatonin secretion. In contrast, the post-stimulus pupil response following red light stimulation correlated with subjective sleepiness (P < 0.05) without significant changes over 24 hours.

Conclusions.: The post-stimulus pupil response to blue light as a marker of intrinsic melanopsin activity demonstrated a circadian modulation. In contrast, the effect of sleepiness was more apparent in the cone contribution to the pupil response. Thus, pupillary responsiveness to light is under influence of the endogenous circadian clock and subjective sleepiness.

Introduction

Image formation is a key function of light detection by the outer retinal photoreceptors, the rods and cones. In addition, the eye has several equally important non-image-forming functions that also are dependent on light. ^1–3 These include entrainment of the mammalian circadian clock in the suprachiasmatic nucleus (SCN); ⁵ acute photic effects, such as suppression of nocturnal pineal melatonin synthesis ⁶ ; changes in alertness ⁷ ; and mediation of the pupil light reflex (PLR). ⁸ Nonvisual light detection and photo transduction occur primarily in non-rod and non-cone photoreceptors, which are, in fact, a small group of intrinsically photosensitive retinal ganglion cells (ipRGCs) that express the photo pigment melanopsin. ^9–12 So far, at least three (M1–M3) out of several subtypes of ipRGC demonstrate distinct anatomic projections with clear functional differences. ^13–15 One major site of ipRGC projection is the SCN in the hypothalamus. ^16–18  

Another site of ipRGC projection is the olivary pretectal nucleus (OPN), the central integrator of the PLR. ^8,10 The ipRGCs receive or project modulatory signals from and to neighboring cells, including rods and cones, ^19–22 which themselves act as local circadian clock cells in the retina. ^23,24 The relative contribution of rods, cones, and intrinsic melanopsin to the ipRGC activity at any given moment varies with light characteristics, such as wavelength, intensity, and exposure duration. ^9,11,19,22  

Understanding the trivariate contributions to the afferent pupillomotor signal of ipRGCs has led to renewed interest in using the pupil to detect and monitor outer retinal photoreceptor function as well as inner retinal integrity. ^25–28 Three main parts of the PLR are distinguishable on a pupillogram: (1) the immediate pupillary constriction of the dark or dim light adapted pupil to the abrupt onset of a light stimulus (transient constriction), which largely is due to signaling from activated rod-cones; (2) a more sustained state of the pupillary constriction during continuous light stimulation to which all photoreceptive elements may contribute; and (3) under certain conditions an extended phase of pupillary constriction that persists for some time after stimulus light termination. This post-stimulus pupil constriction derives primarily from intrinsic melanopsin activation of the ipRGCs. ^20,25,29 Any or all of these aspects of the PLR may be affected by non-light influences, such as the endogenous biologic clock. 

In this regard, one indirect influence on the PLR may be the time of day. A previous study by Figueiro et al. suggested a temporal change in the sensitivity of circadian photoreceptors when tested at two different times of the night. ³⁰ Zele et al. showed that there is a differential phase relationship between the predominantly melanopsin-mediated and the mainly cone-driven post-illumination pupil responses relative to the onset of melatonin secretion. ³¹ It also is possible that circadian modulation via external, circulating, and/or central stimuli have a differential effect on rod, cone, and ipRGC sensitivity, which may be detectable through the PLR. Most recently, Owen et al. reported that mutations of circadian clock genes do not alter ipRGC number or melanopsin expression, but do reduce the sensitivity of the PLR at different wavelengths of light. ³² These and other studies lend support to the notion that there is a circadian modulation of the sensitivity of the nonvisual light photoreception system to light as already has been shown for the visual photoreceptive system. ^33–35  

Given that melanopsin is the primary circadian photoreceptor, we hypothesize that the intrinsic melanopsin-mediated pupil response may show a 24-hour variation that is influenced by the circadian clock and, therefore, will persist under constant dim lighting conditions. In our study, we addressed the question of whether the PLR varies with time of day. It might be that prolongation of wakefulness also affects rod, cone, and melanopsin activity measured in the PLR. More specifically, we aimed to identify which measurable parameters best reflect the mainly ipRGC driven part of the pupil response in humans to narrow bandwidth light stimulations. We evaluated further subjective sleepiness during the same 24-hour periods and correlated it with the pupil responses. 

Methods

Subjects

Subjects were recruited from flyers posted at the Swiss Federal Institute of Technology in Lausanne, Switzerland. Ten healthy nonsmoking subjects (age 24.5 ± 7.2 years; 3 women, 7 men) without medical or psychiatric disorders were included in the study. Subjects underwent a screening ophthalmologic exam, which included Snellen visual acuity, color vision testing with Ishihara plates, clinical pupil testing for a relative afferent pupillary defect, and funduscopy, and all subjects had normal examination results. All subjects were healthy and had no sleep problems, as assessed by an entrance questionnaire, an interview and the Pittsburgh Sleep Quality Index (PSQI score, mean ± SD = 2.7 ± 0.8). ³⁶ None of the subjects was an extreme chronotype as assessed by the Horne-Östberg morning-evening type questionnaire (mean ± SD = 55.5 ± 6). ³⁷ Seven days before the study, subjects were asked to keep a regular sleep-wake cycle with bedtime and wake time at self-selected target times that varied by no more than ±30 minutes, and having a total sleep time of approximately 8 hours. Compliance was verified by activity watches (Daqtix, Oetzen-Süttorf, Germany), worn on the non-dominant wrist as well as sleep logs. Also in the week before testing, consumption of beverages containing alcohol and caffeine was restricted to moderation, whereas on study days, subjects were asked to abstain from these substances completely. All subjects gave written informed consent for study participation, and all study procedures were approved by the local ethical board for human research and conformed to the tenets of the Declaration of Helsinki. 

Study Design

The study was designed to cover an entire 24-hour period under controlled laboratory conditions. Subjects were tested individually in two sessions of 12 hours each. The first 12-hour session started within the first hour after habitual wake time (wake time 7:54 ± 43 minutes, study start time 8:39 ± 49 minutes, mean ± SD). The laboratory illumination was maintained steadily at less than 6 lux, except during the pupil testing (see below). In the laboratory, subjects remained seated, and were permitted to read, listen to music, or engage in conversation. A trained assistant was present to ensure that subjects did not fall asleep or use external light sources, such as a cellular phone, and to provide snacks and beverages regularly. Every 30 minutes, subjects were asked to assess their level of sleepiness by filling out a visual analogue scale (see below). Once every hour, subjects were asked to produce a saliva sample for melatonin analyses, and undergo pupil testing (see below). After the first 12-hour session subjects were requested to go to bed and to arise at their habitual wake time. 

Subjects returned for the second 12-hour session on the next evening, that is 13 hours after their habitual wake time. In the interval between the two study sessions (one night and one day at home), subjects were exposed to their usual environmental light conditions. All in-laboratory study procedures were identical for the two 12-hour sessions for all subjects. 

Saliva Samples

Saliva samples were collected hourly throughout the study duration, and then centrifuged and stored at −20°C. Upon study completion, samples were sent to the Centre Hospitalier Universitaire Vaudois (Lausanne, Switzerland) for melatonin concentration assessment by using the radioimmunoassay method. Detection threshold for melatonin was 0.2 pg/mL (intra assay coefficient <5.8%; interassay coefficients <11.4%; Bühlmann Laboratories, Schönenbuch, Switzerland). 

Subjective Sleepiness

Subjective sleepiness was assessed every 30 minutes by means of a paper-based 100 mm visual analogue scale (VAS), where 0 mm indicates extremely alert and 100 mm indicates extremely sleepy. 

Pupil Recordings

During the 24-hour study protocol subjects were seated every hour in front of a computerized infrared pupillometer. This device comprised an integrating sphere to provide light stimulation (ColorDome Ganzfeld ERG sphere; Diagnosys LLC, Lowell, MA), a chinrest, dual-channel binocular eye tracking, and pupil recording system mounted on an eye frame, worn by the subject (Arrington Research Inc., Scottsdale, AZ). Narrow-bandwidth red (635 ± 18 nm) or blue (463 ± 24 nm) light stimuli of equal photon density (1.0 × 10¹⁴ photons/cm²/s) were selected based on prior investigations in the literature and preliminary testing in our laboratory, which suggested this irradiance level would activate melanopsin. ^29,38,39 Two stimulus durations of 1 and 30 seconds were used. Recent work by Park et al. on isolating the predominantly melanopsin-driven pupil response, found that the maximum difference in the post stimulus pupil response between blue versus red light stimulation occurred at 6 seconds after termination of a 1-second light stimulus. ²⁵ Based on their findings, we chose to use a 1-second stimulus duration in addition to the more commonly used continuous light stimulation (30 seconds in our study) as a means to activate preferentially melanopsin-mediated phototransduction in ipRGCs. 

All recordings were performed monocularly on nondilated pupils by covering one eye with a light-occluding eye patch. The pupil responses were recorded at a frequency of 60 Hz, which resulted in >16,000 data points per tracing. The red stimulus always was presented to the right eye, which always was tested first. After 60 seconds of dark adaptation, a narrow bandwidth red light stimulus 1 second in duration was presented and then followed by 60 seconds of darkness (Fig. 1 for more details). Then, a 30-second red light stimulus was presented, which again was followed by 60 seconds of darkness. Following an interval of 3 to 5 minutes, the same stimulus protocol was repeated with the left eye using a narrow-bandwidth blue light stimulus. 

Figure 1. 

 

(A) Overview on the pupil protocol. Averaged pupil response curves are shown for 10 subjects to 1 and 30 seconds of red and blue light stimuli (averaged across 24 recordings per subject, expressed relative to BL). Outcome variables are indicated schematically: 0, BL (baseline 1); 1, MPS-1s (minimum pupil size); 2, PSPS-1s and PSPS-30s (post-stimulus pupil size 6 seconds after light termination, dashed lines); 3, ERR-1s and ERR-30s (exponential redilation rate, fitted curves); 4, ARS-1s and ARS-30s (asymptotic redilation size, fitted curves); 5, AUC-1s (area under the curve, enlarged small inlay pupil tracing to 1-second light stimulus. Grey area: AUC); 6, MPS-30s (sustained pupil size), and 7 = LS-30s (linear slope for linear regression). Grey hatched boxes: timing and duration of the two light stimuli. (B, C) Overview of the averaged pupil tracings (relative to BL, mean ± SEM, separate for red and blue light stimuli) for the 1-second light stimulus (B) and the 30-second light stimulus (C) separately. Bold lines: mean values (N = 10, n = 24 recordings). Dashed lines: ± SEM.

Data Analysis

The time of dim light melatonin onset (DLMO) from salivary melatonin concentrations in the evening was calculated for each subject separately (one missing value in one subject was interpolated linearly). The DLMO was defined as the time when salivary melatonin concentrations exceeded 2 SDs, compared to averaged (low) daytime values (summarized by Benloucif et al.⁴⁰). Individual DLMOs were used to align hourly pupil recordings to circadian phase bins. This resulted in 24 circadian phase bins of 15 degrees, from 0°–360°, whereas the 0/360-degree bins equal the circadian phase of the DLMO. Subjective sleepiness data were averaged per hour and plotted as two bins across subjects (missing data for two bins in two subjects were interpolated linearly). Blinking and eye movement artifacts were removed from the raw pupil tracings by applying a customized semiautomated filter function (Microsoft Visual Basic 6.5). Then, the pupil tracings were inspected visually by a technician naive to the study aims, who removed any remaining large amplitude artifacts and noise due to pupil movements or pupil tracking problems related to eye strain. A total of 474 recordings was analyzed in this manner, and 6 pupil recordings were excluded due to technical problems or very poor recording quality. All analyzed pupil tracings were smoothed by a polynomial smoothing function (Savitzky-Golay, Origin Pro v.8.50 SRO; OriginLab, Northampton, MA). Thereafter, an exponential fitting was applied on smoothed tracings to obtain post-light stimulus recovery curves (after 1 and 30 seconds of light exposure) by using an asymptotic exponential function according to the formula: $y=a-b{c}^x$ (where a is the asymptotic maximum, b is the range, and c is the rate). Mean pupil size during the first 60 seconds of recording in darkness was defined as baseline (BL; in mm). Actual pupil sizes were converted to relative pupil sizes (RPS) and expressed as ratio of BL sizes, defined as actual pupil size divided by baseline pupil size. 

The main pupil outcome parameters for our study were the following: 

1. Minimum pupil size to a light flash (MPS-1s): This was measured as RPS during 1 second of light stimulation. MPS-1s is representative of the initial maximal pupillary constriction. 

2. Post-stimulus pupil size (PSPS-1s and PSPS-30s): This was the RPS at 6 seconds after light termination, and calculated as mean RPS between 5.5 and 6.5 seconds after termination of both light stimuli. The PSPS is representative of the persistent pupil constriction after light offset. 

3. Exponential redilation rates of RPS after 1 and 30 seconds of light stimulation (ERR-1s and ERR-30s): These are the redilation kinetics taken from the nonlinear fitting function and indicated in mms⁻¹. 

4. Asymptotic redilation size after both light stimuli (ARS-1s and ARS-30s): This parameter was defined as RPS at the asymptotic maximum after light stimulation (derived from fitted curves). 

5. Area under the curve between RPS and the extrapolated maximal baseline after 1 second of light stimulation until 60 seconds after lights off (AUC-1s). AUC-1s is indicated in arbitrary units derived from RPS or in mm². 

6. Minimum pupil size during continuous light stimulation (MPS-30s): This was measured as RPS during 30 seconds of light stimulation. MPS-30s is representative of the sustained pupil constriction (between 5 and 30 seconds after light onset). 

7. Linear slopes (from linear regression) of RPS during the 30 seconds light exposure (LS-30s). Table 1 summarizes all abbreviations for the aforementioned pupil variables, and Figure 1A graphically depicts the outcome variables on a continuous pupil tracing. 

Table 1. 

 

Summary of the Measured Parameters (with Abbreviations) of the Pupillary Response Evoked by (Narrow-Band Width) Red and Blue Light Stimuli

Abbreviation	Variable
BL	Baseline pupil size = mean (mm) during first 60s of darkness (relative pupil size BL = 1)
RPS	Relative pupil size = absolute pupil size/BL pupil size
MPS-1s	Minimum pupil size during 1s light stimulus (relative RPS during 1s light stimulus)
PSPS-1s	Post-stimulus pupil size = RPS at 6s after 1s light stimulus offset (relative to BL pupil size)
ERR-1s	Exponential redilation rate after 1s light stimulus offset (mm/s)
AUC-1s	Area under the curve (arbitrary units) after 1s stimulus until 60s after lights off
ARS-1s	Asymptotic redilation size after 1s light from fitted curves (relative to BL pupil size)
MPS-30s	Sustained pupil size during 30s light stimulus (relative to BL pupil size)
PSPS-30s	Post-stimulus pupil size = RPS at 6s after 30s light stimulus offset (relative to BL pupil size)
ERR-30s	Exponential redilation rate after 1s light stimulus offset (mm/s)
ARS-30s	Asymptotic redilation size after 30s light stimulus from fitted curves (relative to BL)
LS-30s	Linear slope during 30s light stimulus (from linear regression curve)

Pupil outcome parameters either were aligned to elapsed time (since wake time) or to circadian phase, where 0°/360° equals individual DLMOs (see above). We also assigned circadian bins to either “day” or “night,” in which “day” refers to those averaged circadian phase bins when melatonin concentrations were below DLMO threshold, and “night” refers to circadian phase bins with melatonin concentrations above DLMO threshold. Statistics were performed on log-transformed data (for pupil responses) by using a mixed linear regression model (PROC MIXED) with the factors “color” and ”elapsed time,” or “circadian phase” or “day-night” (SAS, v. 9.2; SAS Institute Inc., Cary, NC), with P values based on Kenward-Rogers corrected degrees of freedom. If one of the parameters within a single pupil recording could not be assessed due to eye movement artifacts, blinks, or insufficient fit, this single parameter was excluded. This resulted in slightly varying degrees of freedom. For post-hoc analyses t-tests, F-tests, and differences of least square means were used, and P values were adjusted for multiple comparisons according to Tukey-Kramer. 

Results

Salivary Melatonin Concentrations and Subjective Sleepiness

Salivary melatonin concentrations followed a circadian rhythm with an average DLMO threshold of 2.2 ± 0.4 pg/mL, which occurred at 21:34 hours ± 85 minutes (mean ± SD, n = 10, Fig. 2A), which was on average 13:40 hours ± 43 minutes after habitual wake time. Subjective sleepiness increased with elapsed time of wakefulness (main effect of “elapsed time,” F[23,446] = 33.8, P < 0.0001). Subjects became significantly sleepier 17 hours after their habitual wake time (P < 0.0001, Tukey-Kramer, Fig. 2B). There was no significant difference in subjective sleepiness at the transition between the day and night session (P > 0.4). 

Figure 2. 

 

(A) Salivary melatonin concentrations averaged per circadian phase (15° = 1 hour) and double plotted (mean ± SEM, n = 10). Dashed line: averaged DLMO threshold for melatonin concentration (2.2 pg/mL). (B) Averaged subjective sleepiness ratings collapsed into 2-hour bins with respect to elapsed time since wakefulness (mean ± SEM). Grey area: the 12-hour nighttime session.

Pupil Light Responses

BL pupil size was similar for both eyes: 7.3 ± 1.3 mm for the left and 7.4 ± 1.2 mm for the right (mean ± SD, F[1,416] = 2.34, P = 0.13), without significant changes across 24 hours (main effect of “elapsed time” F[23,416] = 0.76, P = 0.8). With respect to differences in the pupil response to the two colors of light at equal photon fluxes, there was a greater pupil constriction following the onset of blue light when compared to red light. This was evident in the smaller MPS-1s and smaller MPS-30s evoked by blue light compared to red light (F[1,416] > 838, P < 0.0001, Figs. 1B, 1C, mean values ± SD are reported in Table 2). There also was a significant flattening of the slope (LS-30s) during the 30-second blue light stimulus compared to that for red light (F[1,416] = 210.24, P < 0.0001). Following termination of the blue light (1 and 30 seconds), the pupil stayed more constricted, compared to red light. This was detectable in several parameters: smaller PSPS-1s and PSPS-30s, slower ERR-1s and ERR-30s, and larger AUC-1s for blue light (F[1,415] > 7.3, main effect of “color,” P < 0.01). The asymptotic redilation size following 1 and 30 seconds of light stimulation (ARS-1 and ARS-30s) did not differ between colors (F[1,415] < 3.4, P > 0.06). 

Table 2. 

 

Overview on the Relative Pupil Light Reflex Variables (Relative to BL) Tested in Response to Two Different Light Stimuli (N = 10)

	Relative Pupil Size (Relative To BL)		Absolute Pupil Size (Mm)
Pupil Variable	Red (Mean ± SD)	Blue (Mean ± SD)	Red (Mean ± SD)	Blue (Mean ± SD)	P Value
BL	1	1	7.4 ± 1.2	7.3 ± 1.3	ns*
MPS-1s	0.64 ± 0.05	0.54 ± 0.05	4.7 ± 0.4	3.9 ± 0.4	†
PSPS-1s	0.91 ± 0.05	0.72 ± 0.10	6.8 ± 0.4	5.3 ± 0.7	†
ERR-1s ‡	−4.92 ± 0.17	−4.73 ± 0.08	−0.364 ± 0.012	−0.345 ± 0.006	†
AUC-1s §	4.82 ± 2.18	8.32 ± 2.8	35.7 ± 16.2	60.7 ± 20.6	†
ARS-1s	0.99 ± 0.05	0.98 ± 0.06	7.3 ± 0.3	7.2 ± 0.4	ns*
MPS-30s	0.55 ± 0.07	0.45 ± 0.04	4.1 ± 0.5	3.3 ± 0.3	†
LS-30s	0.30 ± 0.24	0.07 ± 0.11	—	—	†
PSPS-30s	0.84 ± 0.07	0.77 ± 0.08	6.2 ± 0.5	5.6 ± 0.6	†
ERR-30 ‡	−4.77 ± 0.05	−4.75 ± 0.05	−0.353 ± 0.004	−0.347 ± 0.003	||
ARS-30s	0.98 ± 0.04	0.97 ± 0.05	7.2 ± 0.3	7.1 ± 0.4	ns*

 Mean values ± SD are shown for colors and durations of light stimuli. Values are indicated relative to BL (in arbitrary units for AUC), and in absolute values (in mm for BL, MPS-1s, PSPS-1s, PSPS-30s, and MPS-30s; in mm/s- ¹ for ERR, and in mm² for AUC). PSPS, post-stimulus pupil size; LS, linear slope; ns, not significant.

*   P > 0.06.

†   P < 0.0001.

‡  Values expressed as %/s and mm/s.

§  Arbitrary units and mm. ²

|  | P < 0.05.

With respect to elapsed time awake, minimal pupil size (MPS-1s) became significantly smaller over a 24-hour period for red but not for blue light (slope for red −0.15 ± 0.20, P = 0.04; slope for blue light −0.09 ± 0.24, P = 0.3; mean ± SD; Fig. 3). The AUC-1s also exhibited a significant change over time (main of “elapsed time” F[23,416] = 3.49, P < 0.0001), showing higher nighttime than daytime values for both colors of light (F[1,27] = 4.32, P = 0.047; Fig. 4; main effect of “day-night”). 

Figure 3. 

 

Minimum pupil size (MPS) following onset of 1-second red and blue light stimulus (averaged per hour) over a 24-hour period. Black lines: trend lines for both curves (n = 10, ± SEM). Mean values per hour are plotted with respect to elapsed time awake (hours since wake time).

Figure 4. 

 

Area under the curve (AUC) in response to the1-second light stimulus. Red line: AUC after red light stimulus. Blue line: AUC after blue light stimulus (mean + SEM, n = 10, main effects of color and time, P < 0.05). The data are double plotted and indicated in arbitrary units. Filled grey area: time course of subjective sleepiness (VAS), averaged across 24 hours (n = 10), aligned to elapsed time awake.

When the post-stimulus pupil size to 1-second light stimuli (PSPS-1s) was related to circadian bins, there was a significant interaction with the factors “color” and “circadian phase” (F[23,417] = 2.7, P < 0.0001). Post-hoc analyses revealed no significant variation with circadian phase of the PSPS-1s after the 1-second red light stimulus (test of effect slices P > 0.9). In contrast, the PSPS-1s following blue light showed a significant modulation over a 24-hour period with a maximum (when pupil size becomes larger) at around 150° (Fig. 5, P < 0.001). In other words, the amount of pupillary constriction after blue light termination was least at approximately 10 hours after the DLMO (Table 2, Fig. 5). The ERR-1s after the 1-second light stimulus showed a significant day-night difference only after red light, with a slower ERR-1s during night than daytime (F[1,27] = 7.0, “color” × “day-night,” P = 0.01, Table 1, Fig. 6A). For both colors of light the post-stimulus pupil size to continuous light (PSPS-30s) was significantly smaller during night than daytime (F[1,27] = 4.4; main effect of “day-night,” P = 0.045, Fig. 6B). 

Figure 5. 

 

Relative post-stimulus pupil size after 1-second light stimulus (PSPS-1s), collapsed into 2-hour bins across circadian phases (0°–360° degrees, 0°/360° indicates DLMO). All values are aligned to circadian phase 0° (DLMO) and are double plotted (n = 10, mean + SEM). Red line: PSPS-1s after red light stimulus. Blue line: PSPS-1s after blue light stimulus. Grey filled area: averaged salivary melatonin concentrations across 24 hours (double plotted, n = 10), aligned to circadian phase (0°/360° = DLMO). Dashed line: mean salivary DLMO threshold.

Figure 6. 

 

(A) Exponential redilation rate (ERR; mm/s) after 1-second red (red bars) and blue (blue bars) light stimulus, averaged for day and nighttime tests (n = 10, mean ± SEM, *P < 0.05, not significant [ns] p > 0.3). (B) Post-stimulus pupil size after the 30-second light stimulus (PSPS-30s) averaged for day and nighttime tests. Red and blue bars: data for the response to the red and blue light stimulus, respectively (n = 10, mean ± SEM, *P < 0.05).

To analyze whether the differences between the two colors and the dynamics of the pupil light responses reported above could be explained by a relationship with either circadian or sleep-dependent variables, we performed correlation analyses. Regarding the relationship of salivary melatonin and pupil responses, greater salivary melatonin concentrations were correlated significantly with smaller MPS-1s and greater AUC-1s for the 1-second red light stimulus (P < 0.05, for R values see Table 3). For the 1-second blue light stimulus, greater melatonin concentrations were correlated significantly with greater PSPS-1s (i.e., less pupillary constriction) and faster ERR-1s, (P < 0.05, Table 3). During the 30-second light stimulus, larger MPS-30s and steeper LS-30s were associated with greater melatonin concentrations for both stimuli colors (P < 0.05, Table 3). 

Table 3. 

 

Summary of Correlations between Relative Pupil Light Reflex Variables (Relative to BL) and Melatonin or Subjective Sleepiness Assessments (Spearman Correlation)

		Correlations
		Melatonin		VAS
Light Stimulus	Pupil Variables:	Red Light	Blue Light	Red Light	Blue Light
1 s	MPS-1s	−0.18*	ns	−0.17*	−0.25*
	PSPS-1s	ns	0.28*	−0.22*	ns
	ERR-1s	ns	−0.33*	ns	ns
	AUC-1s	0.20*	ns	0.37*	0.30*
30 s	MPS-30s	0.18*	0.19*	ns	ns
	LS-30s	0.15*	0.15*	ns	ns
	PSPS-30s	ns	ns	−0.24*	−0.14*
	ERR-30s	ns	ns	0.29*	ns

 ns, non-significant (P > 0.05).

*  Significant correlations, R values are indicated (P < 0.05).

Correlation with higher subjective sleepiness (VAS) showed smaller MPS-1s as well as a greater AUC-1s in response to both colors, 1-second red and blue light stimuli (P < 0.05, for R values see Table 3). Greater sleepiness was correlated with smaller PSPS-1s after termination of the 1-second red light (P < 0.05), but not the blue light (P > 0.7). In response to the 30-second light stimulus, smaller PSPS-30s was correlated significantly with higher subjective sleepiness for both stimulus colors (P < 0.05, Table 3). Lastly, greater subjective sleepiness correlated significantly with a slower ERR-30s after the red 30-second light stimulus (P < 0.05, Table 3). 

Taken together, we found that the pupil constricted more (smaller MPS-1s and MPS-30s) to blue light compared to red light presented for either 1 or 30 seconds. Persistence of pupillary constriction after light termination was noted only after (blue) light termination for 1 and 30 seconds (smaller PSPS-1s and PSPS-30s). The time course of PSPS-1s in response to blue light resulted in a significant modulation over 24 hours, with maximal post-stimulus pupil size at around 150 degrees. Greater melatonin concentrations were associated significantly with greater PSPS-1s after termination of a 1-second blue light stimulus. On the other hand, greater subjective sleepiness could be related to smaller PSPS-1s following a 1-second red light flash. 

Discussion

Our study was designed to test the pupillary light response to two different colors of light across an entire 24-hour cycle under controlled dim light conditions. Recorded at equal photon fluxes, pupil responses to blue light were significantly more pronounced when compared to red light at most times of the 24-hour period, with the exception of asymptotic redilation sizes (ARS). To assess which aspect of the pupil response related best to either circadian or wakefulness-dependent dynamics, hourly pupil responses were related to salivary melatonin secretion and subjective sleepiness. The post-stimulus pupil size after termination of the 1-second light stimulus (PSPS-1s) turned out to be the most promising marker, as it resulted in a significantly different response to circadian and homeostatic, that is wakefulness-dependent, functions after blue and red light, respectively. 

The endogenous biological rhythm of our study participants was well synchronized with the environmental 24-hour day, and melatonin secretion onset (DLMO) occurred 2–3 hours before their habitual bedtime. This is consistent with the published literature on DMLO and circadian timing. ⁴⁰ The progressive increase of subjective sleepiness in our study subjects after a 16-hour waking-day also follows the well-known time course under prolonged wakefulness conditions in dim light. ⁴¹  

Our selected pupillary outcome measures were based on existing literature describing the rod, cone, and melanopsin contributions to different components of the PLR. Several investigators have reported that the maximal amplitude of the initial pupil constriction as well as the sustained pupil constriction during light stimulations mainly are conveyed by rods and cones, with a smaller contribution by intrinsic, melanopsin activation of ipRGC. ^{20,25,29,42,43}  

In contrast, the pupillary response after termination of a bright light stimulus appears to be the better parameter for assessing melanopsin activation. ^29,43–45 Evidence that the post-stimulus pupil response in primates and humans is dominated by the intrinsic input from ipRGC has been demonstrated convincingly by Gamlin et al. ²⁹ After pharmacologic blockade of rods and cones in behaving macaques, the authors still could record pupillary constriction that persisted for some time following termination of a continuous 10-second bright blue light stimulus. To our knowledge, the specific aspect of the post-stimulus pupil response that best defines melanopsin activity has not been established yet. Thus, in our study, we evaluated several outcome measures of the post-stimulus response (pupil size, area under curve, redilation rate, and asymptotic amplitude). 

The stimulus luminance in our study (i.e., 15 cd/m² for blue light and 34 cd/m² for red light) was intermediate between luminance used previously for exploring rod and cone-weighted responses, ^25,26 and our subjects were dim-light adapted. Thus, the pupil constrictions to the 1-second stimulus most likely derived from mixed rod and cone inputs. The far greater proportion of rod photoreceptors, which have a greater sensitivity to shorter wavelengths of light, is likely why the pupil constrictions consistently were larger to blue light compared to red light. However, sensitivity to long wavelength light is a characteristic of M/L cones, so it generally is accepted that pupil responses to high intensity red light are dominated by cone inputs. In rodents, melanopsin-dependent responses recorded in the olivary pretectal nucleus dominate when using short wavelength light, but recent evidence points to some contribution of S-cones as well. ^46,47 Therefore, while we selected stimulus conditions that permit one photoreceptive element to dominate the pupillomotor input, we acknowledge that none of the pupil parameters of our study can be said to represent the activity of a single photoreceptor system. 

The most significant finding in our study was the 24-hour modulation of the post-stimulus pupil response (PSPS-1s) after termination of 1 second of blue light. The blue light PSPS-1s was highest after the peak of salivary melatonin concentration, close to habitual wake time. In addition, only the PSPS-1s to blue light was correlated negatively with melatonin secretion, meaning that pupils were less contracted at times with higher melatonin secretion. This suggests that the intrinsic melanopsin system becomes less sensitive to light in the second half of the night, after the peak of melatonin secretion and closer to wake time. One possible reason for this is the phenomenon of light-evoked rod shedding close to wake time. ^34,48 However, there still must be a central clock influence for the following reasons: a peak in the PSPS-1 occurs only once in constant dim light across 24 hours, and the timing of the peak is in phase with melatonin secretion. These and the absence of red light PSPS-1 modulation suggest a circadian regulation of the intrinsic, melanopsin-driven pupil response. 

The finding of lesser pupil responsiveness, that is larger pupil size, at first glance might appear counterintuitive to clinical observations of smaller pupils associated with sleepiness. It is important to remember that the PSPS-1s and PSPS-30s represent the pupil sizes 6 seconds after light termination, and are not a baseline size. A circadian modulation of the PLR to blue light (470 nm) was shown recently in mice lacking outer photoreceptors (rd/rd mice). The study revealed increased sensitivity during daytime hours in these nocturnal animals, with lowest sensitivity shortly before their activity onset. This gave further evidence that ipRGCs are under circadian clock control. ³²  

One could argue that if BL pupil sizes already were smaller before the blue light stimulus, the PSPS-1s might be more attenuated as the pupil could be constrained mechanically from further constriction. However, differences in BL before red light versus blue light exposure were not present in our subjects. We found only a decreasing linear trend over time in MPS-1s for the pupil response to the red color of light. In a previous study by Yasukouchi et al., there was no linear relationship between the PLR and melatonin suppression levels (in response to different light intensities, ranging from 1–30 lux for the PLR, and from 30–600 lux for melatonin suppression). ³⁵ We also did not find any significant association between baseline dim light pupil size and salivary melatonin concentrations. 

In our study, we also found higher order cortical effects that very clearly reflected subjective sleepiness in our pupil recordings. With elapsed time awake, pupil responses became progressively affected with greater blink rates and fatigue waves. This was reflected in greater AUC-1s during nighttime where some of the fatigue waves contributed to the AUC-1s calculation, and a greater overall variability in the pupil recordings. 

In conclusion, we find that the post-stimulus pupil response to 1-second blue light (PSPS-1s) shows a 24-hour variation and appears to be the superior pupillary parameter of the intrinsic melanopsin activity. This may be related to a greater resistance of the early post-stimulus phase of the pupil response to the central influences, such as fatigue and sleepiness. ^24,25,32 The PSPS-1s presumably reflects circadian modulation. Further studies are needed to determine if this pupil parameter can be used to estimate circadian variables reliably. On the other hand, the cone-mediated pupil response to red light varies more with sleepiness. However, a complete separation of circadian and homeostatic (e.g., accumulation of sleepiness) influences on the pupil response would require a different study design, for example a forced desynchrony protocol. ⁴⁹  

Thus, this study provides further evidence that pupillary responsiveness to light response not only is mediated by rods, cones, and melanopsin through ipRGC activity in a wavelength-dependent manner, but also is under central modulation of the endogenous circadian clock and subjective sleepiness. These and other biologic influences on the pupil light reflex may be important issues to consider in any future studies using quantitative pupillometry in the clinical setting. 

Acknowledgments

We recognize the subjects for their study participation. Francois Bourqui and Laurent Deschamps, from the LESO-IT group, programmed the semi-automated artifact filter. J-L Scartezzini (Director LESO-PB) provided generous study support and Pierre Loesch provided excellent technical help with the study setup. Daniel Hulliger (St. Sulpice) provided valuable help with data pre-processing. 

Footnotes

 Supported by the Velux Foundation (Switzerland, MM), the Loterie Romand/Open Eyes Foundation (Switzerland, AK) and by a study grant from the Grieshaber Ophthalmologic Research Foundation (Switzerland, MM and AK).

Footnotes

 Disclosure: M. Münch, None; L. Léon, None; S.V. Crippa, None; A. Kawasaki, Bayer SpA (C)