Thirty-nine myopic (n = 20, male = 8; female = 12), hyperopic (n = 5, male = 3; female = 2) and emmetropic (n = 14, male = 5; female = 9) participants between the ages of 20 and 35 years (mean age ± SD, 23.36 ± 3.06 years) were recruited to examine the association between the ipRGC-driven pupil response and refractive error. Prior to participation, all subjects underwent a comprehensive eye examination to assess their refractive status and ocular health. The mean spherical equivalent refraction (SER) was − 3.50 ± 1.82, +0.58 ± 0.23 and −0.02 ± 0.12 DS for participants in the myopic, hyperopic and emmetropic groups, respectively. None of the pupil metrics data were significantly different between the low hyperopic and emmetropic participants (p > 0.05, data not shown), and hence they were grouped together as one “non-myopic” group (n = 19; mean SER, +0.28 ± 0.23 DS). All subjects had normal visual acuity of 0.00 logMAR or better, and astigmatic refractive error of ≤1.00 DC. No participants had ocular pathology or a history of any major eye or refractive surgery.

No participants were taking any prescription medication known to affect the pupil size or sleep patterns (such as melatonin). In addition, participants were asked to refrain from alcohol, caffeine, and nicotine for 12 h prior to the pupil measurements. All participants were tested between 9:00 am and 12:00 pm to minimize the effects of circadian variation on ipRGC function and the PIPR.39 The study was approved by the Southern Adelaide Local Health Network (SALHN, ID: 156.17) ethics committee, and all participants provided written informed consent prior to their participation. All subjects were treated in accordance with the Declaration of Helsinki.

The PIPR was measured using a custom-built optical system, similar to the one described by Kankipati et al.,31 as shown in Fig. 1A and B. The illumination system consisted of a set of red and blue light-emitting diodes (LEDs). The light from the red and blue LEDs was transmitted to the right eye via two Fresnel lenses; F1 and F2, each 10.16 cm in diameter and with a 10.16 cm focal length (Edmund Optics, Barrington, NJ). The blue (470 nm, 3 mm diameter, full width at half maximum [FWHM] 22 nm) and red LEDs (625 nm, 3 mm diameter, FWHM 20 nm) (Jaycar Electronics, Rydalmere, Australia) were positioned at the focal length of the first Fresnel lens, F1. The two Fresnel lenses were kept 20.32 cm apart (i.e. separated by twice their focal length). A 20.32 × 20.32 cm holographic diffuser of 5-degree diffusing angle (Edmund Optics, Barrington, NJ) was placed in front of the second Fresnel lens (F2), and the participant’s right eye was positioned at the focal point of F2. During the PIPR measurement, the right eye was presented with the light stimulus, and the effect of light stimulation was measured in the contralateral left eye. A modified Logitech C920 HD Pro webcam (Logitech, Newark, CA) with illuminating infrared LEDs (940 nm, 5 mm diameter, Core Electronics, NSW, Australia) was used to record the pupil responses from the left eye at a rate of 15 frames/s. The presentation of light stimulus and the duration of PIPR recording were controlled via a small single-board computer, Raspberry Pi 3 Model B (Core Electronics, NSW, Australia). During the experiment, the participants were positioned in a chinrest and instructed to look straight ahead at a small red laser spot on the wall at a distance of 4 m.

Figure 1.

Overview of the optical system used in the study; sketch (A) and real image (B). Two Fresnel lenses, F1 and F2 (both 10.16 cm focal length and diameter), were placed at twice their focal length apart. Red and blue LEDs were placed at one end of the optical system. The subject’s dilated right eye was aligned at the other end, while the undilated left eye was recorded by the infrared camera (IR camera) attached to the computer. The diffuser had a 5 deg diffusing angle.

(0.25MB).

The red and blue LEDs illuminating the eye flickered at 10 Hz with a duty cycle of 80%. The two stimulus durations of 1 s and 5 s used in this study were within the range of previously published studies.29,37 The corneal irradiance levels, measured using an optical power meter (Newport Corporation, Irvine, CA), were 3.24 × 1014 photons/cm2/s for the blue stimulus (470 nm) and 3.68 × 1014 photons/cm2/s for the red stimulus (625 nm). These corneal irradiances were close to previously used irradiance levels in young, healthy participants.37,38,40Table 1 shows excitation for each photoreceptor class estimated using the toolbox provided by Lucas et al.41 The L cones have higher sensitivity to the 625-nm light, whereas melanopsin, rods, and S cones have higher excitation to the 470-nm light compared to L cones.

On the day of experiment, the participant’s right eye was dilated with 1% Mydriacyl (tropicamide, Alcon, Fort Worth, TX) to ensure consistent retinal illumination within and between subjects during the PIPR measurement. After 20-minutes, all subjects were dark adapted for 5 min (∼2–5 lux) before commencing the pupil measurements. After dark adaptation, the right eye was presented with 1 and 5 s long-wavelength (red) and short-wavelength (blue) narrowband light stimuli while the consensual pupil response was measured in the undilated left eye as a measure of the ipRGC-induced PIPR, as shown in Fig. 2. The order of stimulus presentation was: 1 s red, 1 s blue, 5 s red, and 5 s blue. Testing with 1 s light pulses always preceded the 5 s pulses. Red and blue stimuli were alternated in all sessions, similar to previously published experiments,30 to control for the effect of melanopsin bistability.42 As light-induced melanopsin response could persist for up to 3–5 min after light offset,29,43 the 5-minute dark adaptation period between 1 s and 5 s trials was necessary to avoid potentiation of the response from previous light stimulation.44 Two repeats for each stimulus (470 nm and 625 nm) were recorded for each stimulus duration and were averaged for further analysis.

Figure 2.

Pupil stimulation protocols for the experiment. Dark adaptation (5 min) was followed by a 10 s baseline and 1 s red stimulus with 60 s pupil measurements after stimulus offset. After another 10 s baseline, the same protocol was repeated for 1 s blue stimulus. Following 1 s measurements, a dark adaptation period of 5 min was observed before repeating the same protocol for 5 s red and blue stimuli. Two repeats for each stimulus (470 nm and 625 nm) were recorded for each stimulus duration and were averaged.

(0.07MB).

The change in pupil diameter in response to red and blue stimuli was measured from the pupil camera recordings using a custom Matlab program (Matlab 2017b, version 9.3, MathWorks, Natick, MA). For both 1 and 5 s trials, the Matlab program analysed the change in pupil area relative to the average baseline pupil area for each wavelength (i.e. the average of 10 s pre-stimulus period before red and blue stimulation). To calculate the pupil area, the Matlab algorithm cycled through each frame of the recording, and the ‘starburst’ algorithm was used to detect the pupil outline and fit an elliptical shape to the boundary. For each frame, the area of the ellipse was determined using the formula, A = a*b*pi (where a and b are the two semi-axes of the ellipse). Frames where an ellipse could not be detected due to blinks or poor fixation were automatically removed from the analysis. Data was smoothed using the moving average filter with a window of length 1 s. Finally, a time-stamped series of relative pupil responses was generated for further analysis. As outlined in Table 2, the PIPR was described by 5 metrics, the peak constriction, the 6 and 30 s PIPR, and the early and late area under the curve (AUC). These are well defined, robust and reliable metrics that have been frequently used by previous studies.29,37,38,45–47 All pupil metrics are shown as “normalized change” to the average baseline pupil diameter (expressed as a percentage). Whilst peak pupil constriction represents both rod/cone and inner retinal activity, the other four metrics are commonly used to describe the ipRGC activity.29,30

Statistical analyses were performed using commercial software (SigmaStat 3.5, Aspire Software International, Ashburn, VA). For both 1 and 5 s stimulus durations, the difference in pupil metrics between myopes and non-myopes with red and blue stimuli were analysed with two-way analysis of variance (ANOVA) and Holm-Sidak post-hoc tests for statistical significance, using “refractive error” as the between-subjects factor and “wavelength” as the within-subjects factor. To determine the within-subject variability of the PIPR metrics, the intrasession coefficient of variation (CV or SD/mean) was calculated. The CV provides a reliable measurement of variability because it is dimensionless and is not affected by the changes in measurement units.48 A p-value of less than 0.05 was considered to be statistically significant. All data are expressed as mean ± standard error of mean (SEM).

The change in pupil metrics with 1 and 5 s long-wavelength (red) and short-wavelength (blue) stimuli for myopes and non-myopes is shown in Table 3. Following 5 min of dark adaptation, baseline pupil area of the undilated left eyes was not significantly different between 1 s red (mean ± SEM for the two refractive groups, 100.96 ± 0.94%) and blue stimuli (100.85 ± 0.53%, two-way ANOVA main effect of wavelength F(1,75) = 0.017, p = 0.895). For 36 participants (92%), pupils re-dilated rapidly after light offset following red stimulation; whereas, with blue stimulation the rate of re-dilation to the baseline pupil diameter was considerably slower (Fig. 3). Exposure to the blue stimulus caused a greater constriction of the pupil than the red stimulus (red stimulus, 36.65± 1.60%;blue stimulus, 29.42± 1.82%,two-way ANOVA main effect of wavelength, p < 0.001, Table 3). Compared to the 1 s red stimulus, the 6 s (red stimulus, 81.02± 1.51%; blue stimulus, 56.93± 4.20%) and 30 s PIPR (red stimulus, 98.18± 0.77%; blue stimulus 86.11± 2.83%) were significantly smaller for the blue stimulus across both refractive groups (two-way ANOVA main effect of wavelength, p < 0.001, Fig. 3A and B). In addition, the early (red stimulus, 1.06 ± 0.03, blue stimulus 1.28 ± 0.04) and late AUC (red stimulus, 0.60 ± 0.07, blue stimulus 1.15 ± 0.09) were significantly longer for the blue light than the red light (two-way ANOVA main effect of wavelength, p < 0.001, Fig. 3C and D). These results indicate a strong ipRGC-induced PIPR following short-wavelength stimulation.

Figure 3.

Change in pupil metrics with 1 s red and blue stimulation for myopes (n = 20) and non-myopes (n = 19). The 6 s (A) and 30 s (B) post-illumination pupil responses (PIPR) were significantly lower with the blue light compared to the red light (two-way ANOVA main effect of wavelength, p < 0.001). The early (C) and late (D) area under the curve (AUC) were significantly greater following blue light stimulation compared to red light stimulation (two-way ANOVA main effect of wavelength, p < 0.001). The PIPR values are shown as normalized change relative to the baseline pupil diameter; whereas the AUC values are shown in log units. None of the pupil metrics were significantly different between myopic and non-myopic participants (two-way ANOVA main effect of refractive error, p > 0.05). Error bars represent standard error of the mean. (E) Normalized change in pupil size for 1 s red and blue pulses across the two refractive groups. Pupil metrics include baseline, peak constriction, 6 s PIPR, 30 s PIPR, early AUC, late AUC. Shaded regions represent 95% confidence intervals. Stimulus is shown in yellow.

(0.57MB).

As illustrated in Fig. 4, 5 s blue stimulation induced a slightly stronger melanopsin response compared to 1 s stimulation with the same wavelength. All pupil metrics, including the peak constriction, the 6 s and 30 s PIPR, and the early and late AUC indicated a strong PIPR in response to blue stimulation for both groups (two-way ANOVA main effect of wavelength, p < 0.001 for all, Table 3).

Figure 4.

Change in pupil metrics with 5 s red and blue stimulation for myopes (n = 20) and non-myopes (n = 19). The 6 s (A) and 30 s (B) post-illumination pupil responses (PIPR) were significantly lower with the blue light compared to the red light (two-way ANOVA main effect of wavelength, p < 0.001). The early (C) and late (D) area under the curve (AUC) were significantly greater following blue light stimulation compared to red light stimulation (two-way ANOVA main effect of wavelength, p < 0.001). The PIPR values are shown as normalized change relative to the baseline pupil diameter; whereas the AUC values are shown in log units. None of the pupil metrics were significantly different between myopic and non-myopic participants (two-way ANOVA main effect of refractive error, p > 0.05). Error bars represent standard error of the mean. (E) Normalized change in pupil size for 5 s red and blue pulses across the two refractive groups. Pupil metrics include baseline, peak constriction, 6 s PIPR, 30 s PIPR, early AUC, late AUC. Shaded regions represent 95% confidence intervals. Stimulus is shown in yellow.

(0.58MB).

As shown in Figs. 3 and 4, none of the pupil metrics were significantly different between myopic and non-myopic participants for either 1 s stimulus or 5 s stimulus (two-way ANOVA main effect of refractive error, p > 0.05 for all).

To quantify the within-subject variability in the PIPR metrics, we calculated the intrasession CV for each of the pupil metrics for both 1 and 5 s stimuli (Table 4 and Supplementary Figure A.1). For both the 1 and 5 s stimuli, the intrasession CV for the peak constriction and the 6 and 30 s PIPR were generally greater for the blue stimulus compared to the red stimulus, but they were all <20%, which is considered low and acceptable for PIPR measurements.29 The intrasession CV for both 1 and 5 s stimuli were significantly greater for AUC parameters, particularly for the late AUC with CV > 20% for both wavelengths (Supplementary Figure A.1).