Post-Illumination Pupil Response and Sleep Quality in Patients With Glaucoma: The LIGHT Study

Abstract

Purpose: This study aimed to investigate whether intrinsically photosensitive retinal ganglion cell function evaluated using post-illumination pupil response (PIPR) in patients with glaucoma is associated with sleep quality.

Methods: This cross-sectional study measured the PIPR in 138 patients with glaucoma (mean age, 70.3 years) using pupil diameter after red and blue light exposure. The net PIPR change was classified into three groups according to tertiles (i.e., low, intermediate, and high groups), with lower net PIPR change indicating lower intrinsically photosensitive retinal ganglion cell (ipRGC) function. Subjective and objective sleep qualities were assessed using the Pittsburgh Sleep Quality Index (PSQI) questionnaire and actigraphy, respectively, with a total PSQI score of ≥6 indicating sleep disturbance.

Results: The prevalence of subjective sleep disturbance significantly increased with decreasing tertile groups of net PIPR change (P = 0.036). Subgroup analysis obtained the same results in the severe glaucoma group (P = 0.004) but not in the non-severe glaucoma group. In the severe glaucoma group, multivariable logistic regression analysis adjusted for potential confounders showed a higher odds ratio for subjective sleep disturbance in the low-tertile group of net PIPR compared with the high-tertile group (odds ratio = 6.22; 95% confidence interval, 1.76–21.90; P = 0.004). Significant associations between PIPR and objective sleep quality (total sleep time, sleep efficiency, and wake after sleep onset) were found in the severe glaucoma group (P = 0.015, P = 0.013, and P = 0.015, respectively).

Conclusions: The PIPR in patients with glaucoma was significantly associated with decreased sleep quality, independent of potential confounders.

Glaucoma, a progressive neurodegenerative disease of the eye, is characterized by retinal ganglion cell death and subsequent visual field loss.¹ In addition, experimental and clinical studies have reported that glaucoma is involved in the structural and functional impairment of intrinsically photosensitive retinal ganglion cells (ipRGCs).^2,3 These cells express a photopigment, melanopsin, and are important for the regulation of circadian rhythm through the transmission of non-image–forming light to the suprachiasmatic nuclei of the hypothalamus, which is known as the circadian master clock.^4,5 In fact, lower circadian rhythm parameters, such as melatonin levels, have been reported in patients with glaucoma.⁶ Thus, impaired ipRGC function in patients with glaucoma may cause circadian disruptions. 

Blindness caused by various neurodegenerative and retinal diseases, such as glaucoma, age-related macular degeneration, and diabetic retinopathy, has been associated with sleep disturbance.⁷ Light stimulus to the ipRGCs regulates sleep rhythm by entraining the circadian rhythm in the suprachiasmatic nuclei.⁸ Moreover, experimental studies in mice have shown that ipRGCs contribute to sleep rhythm through various pathways, such as the preoptic circuit in the brain and endogenous opioid signaling in the retina.^9,10 Several clinical studies have reported that glaucoma was associated with lower subjective sleep quality^11,12; however, clinical evidence for the involvement of ipRGCs in the sleep quality of glaucoma patients has been controversial. Only one observational study of 32 patients with glaucoma and 13 healthy participants reported that impaired ipRGC function was associated with objective sleep quality evaluated using polysomnography.¹³ In contrast, another study of 15 patients with glaucoma and 17 healthy participants showed no significant association between ipRGC function and subjective sleep quality.¹⁴ 

ipRGCs also project to the olivary pretectal nucleus in the midbrain, which is involved in regulation of the pupillary light reflex.⁵ Notably, the ipRGC-mediated pupillary light reflex is characterized by persistent constriction and delayed recovery after short-wavelength, blue-light stimulus offset.¹⁵ As such, the post-illumination pupil response (PIPR) after blue and red light stimuli has been used to evaluate ipRGC function,¹⁵ with several studies showing a relationship between dysfunction of ipRGC evaluated using PIPR and several ophthalmic and neurodegenerative diseases, including glaucoma.^3,16,17 

The present cross-sectional study evaluated the association between ipRGC function and sleep quality in 138 patients with glaucoma. ipRGC function, subjective sleep disturbance, and objective sleep quality were assessed using PIPR, the Pittsburgh Sleep Quality Index (PSQI), and actigraphy, respectively. 

Methods

Patients With Glaucoma

A total of 172 patients with glaucoma were enrolled in Longitudinal study of biological circadian rhythms In Glaucoma patients: Home Testing of circadian intraocular pressure and biological parameters (LIGHT) study between May 2017 and September 2020. We recruited the patients with at least one eye affected by glaucoma from Nara Medical University Hospital and included those who consented to participate in the LIGHT study protocol. The LIGHT study was approved by the Ethics Committee of Nara Medical University (number 1314) and registered with the University Hospital Medical Information Network Clinical Trials Registry (no. UMIN000027299). The protocol of the LIGHT study adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from all enrolled patients with glaucoma. All patients with glaucoma had completed the following examinations for the diagnosis of glaucoma: slit-lamp biomicroscopy, ophthalmoscopy, gonioscopy, best-corrected visual acuity testing, Goldmann applanation tonometry to determine intraocular pressure, standard automated perimetry to assess the visual field, and spectral-domain optical coherence tomography. Patients with glaucoma were diagnosed by one glaucoma specialist (T.Y.) based on the glaucomatous optic disc and consistent glaucomatous visual field defects.¹⁸ Bilateral glaucoma and unilateral glaucoma were defined as the presence of glaucoma in both eyes or one eye, respectively. For patients with bilateral glaucoma, the eye with worse glaucoma was analyzed in this study. Among the 172 eyes of 172 patients with glaucoma, 34 were excluded due to missing and unreliable data on net PIPR change, leaving a total of 138 eyes. Briefly, the exclusion criteria with regard to PIPR were as follows: missing data (n = 2), interrupted measurement data of pupillary light response (n = 3), unstable pupil during measurement of pupillary light response (n = 17), and unreliable pupillary light response after red and blue light stimulus (n = 12), which have been described in detail previously.³ Out of 138 eyes, 110 eyes (79.7%) exhibited primary open-angle glaucoma, 11 eyes (8.0%) exhibited primary angle-closure glaucoma, and 17 eyes (12.3%) exhibited secondary glaucoma, including exfoliation glaucoma. 

Assessment of Glaucoma Severity

Visual field examinations for the diagnosis of glaucoma and determination of its severity were conducted using the Humphrey Field Analyzer II (Humphrey, Carl Zeiss Meditec, Dublin, CA, USA) using the 30-2 Swedish interactive threshold algorithm (SITA) standard program. Eyes with false-positive rates > 15% were excluded from analysis based on a previous study.¹⁹ Among the eyes with a false-positive rate > 15% in the SITA standard program, 10 were substituted with data (false-positive rate <15%) from the SITA fast program. Based on previous criteria for glaucoma severity, two eyes with unmeasurable 30-2 SITA standard program data were characterized as having severe glaucoma due to central scotoma.²⁰ Patients with glaucoma were then divided into the following two groups based on the visual field mean deviation (MD) in the worse eye: non-severe glaucoma group, MD > −12 dB (n = 57); severe glaucoma group, MD ≤ −12 dB (n = 81). Of the 138 patients, 101 (73.2%) had bilateral glaucoma. Of the 57 eyes with non-severe and 81 eyes with severe glaucoma, 37 (64.9%) and 64 (79.0%) had bilateral glaucoma, respectively. 

Measurement of PIPR

ipRGC function was evaluated using the PIPR measured on a pupillometer (RAPDx; Konan Medical USA, Inc., Irvine, CA, USA). ipRGCs were excited by short-wavelength blue light with a peak at 470 to 480 nm.²¹ Additionally, the characteristics of pupillary light reflex caused by ipRGCs are slow recovery and sustained constriction after blue light stimulus offset.²² Therefore, blue light with a 448-nm peak wavelength and irradiance of 2.70 × 10¹² photons/s/cm² was used to evaluate the ipRGC function. Red light with a 608-nm peak wavelength and irradiance of 2.58 × 10¹² photons/s/cm² was measured to evaluate the outer retina as a control. The protocol for measuring PIPR is described in detail elsewhere.³ Briefly, the eye with worse glaucoma was dilated using a mydriatic eye drop, except for primary angle-closure glaucoma, and was stimulated with blue light after red light. The pupil diameter of the fellow eye without mydriasis was measured to determine the PIPR parameters. The procedure for measuring PIPR was as follows: (1) dark adaptation, 5 minutes; (2) measurement of baseline pupil diameter before light stimulus, 7 seconds; (3) measurement of initial pupil diameter during light stimulus, 10 seconds; and (4) measurement of the PIPR after light stimulus offset, 40 seconds. Net PIPR change was used to evaluate ipRGC function and was defined based on a previous study¹⁵ as follows: Net PIPR change (%) = Blue sustained PIPR change (%) – Red sustained PIPR change (%) (Fig. 1). Sustained PIPR change (%) was calculated using the following formula: (Sustained PIPR/Baseline pupil diameter) × 100, where sustained PIPR (mm) is the baseline pupil diameter minus the mean sustained pupil diameter for a duration of 30 seconds starting from 10 seconds after light stimulus offset to 40 seconds. Lower net PIPR change indicates lower ipRGC function due to pupillary response characterized by sustained constriction and slow recovery after excitation of blue light. 

Figure 1.

 

Time trace plots of the mean pupil diameter of pupillary light reflex in the high-tertile group according to net PIPR change. The x-axis and y-axis indicate the time (seconds) and pupil diameter (percent baseline), respectively.

Evaluation of Subjective and Objective Sleep Quality

Subjective sleep quality was assessed using the PSQI questionnaire.²³ This questionnaire evaluates sleep quality and sleep disturbance over a 1-month period and consists of seven components, including sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbance, use of sleep medication, and daytime dysfunction. The total PSQI score, which can range from 0 to 21, is calculated by adding the scores for all seven components, each of which ranges from 0 to 3, with higher total scores indicating lower sleep quality. Subjective sleep disturbance was defined as a total PSQI score of ≥6. 

Objective sleep quality was evaluated using an accelerometer (ActiGraph GT3X-BT; ActiGraph LLC, Pensacola, FL, USA) with a 30-Hz sampling frequency in 1-minute epochs. Patients with glaucoma were instructed to wear an ActiGraph on the non-dominant wrist for six consecutive nights, including weekdays and weekends, and to keep a sleep diary for the evaluation of bedtime and rising time. 

Evidence has confirmed that sleep parameters measured using actigraphy are in moderate to high agreement with those measured using polysomnography.²⁴ Sleep/wake states were distinguished using the Cole–Kripke algorithm.²⁵ Sleep onset and offset were determined to be the beginning of the first 10 consecutive minutes recorded as sleep after the bedtime and the end of the last 10 consecutive minutes recorded as wake before the rising time, respectively. Bedtime and rising time were obtained from a sleep diary forming part of the self-reported questionnaire. Objective sleep parameters included total sleep time (i.e., the sum of sleep time between sleep onset and sleep offset), sleep efficiency (i.e., the percentage calculated by dividing the total sleep time by the time between bedtime and rising time), wake after sleep onset (i.e., awake time after sleep onset), and sleep-onset latency (i.e., time between bedtime and sleep onset). 

Other Measurements

We evaluated potential confounders by measuring basic, clinical, and circadian rhythm parameters (Table 1). Body mass index was calculated by dividing the patient’s weight in kilograms by their height in meters squared. Alcohol consumption (>30 g/d) and benzodiazepine use were assessed using a self-reported questionnaire. Hypertension was determined using the medical history or current use of antihypertensive drugs. Diabetes was determined based on the medical history, current use of antidiabetic drugs, or fasting plasma glucose levels (≥126 mg/dL) and glycated hemoglobin levels (≥6.5%). Chronic kidney disease was determined based on an estimated glomerular filtration rate (eGFR) of <60 mL/min per 1.73 m². Depressive symptoms were determined based on a Geriatric Depression Scale score of ≥6. The photoperiod from sunrise to sunset time in Nara was calculated using the National Astronomical Observatory of Japan website. Duration in bed for the scotoperiod was determined from the sleep diary. Physical activity was assessed using the International Physical Activity Questionnaire. History of cataract surgery was evaluated using ophthalmic examination. 

Table 1.

 

Basic, Clinical, and Circadian Rhythm Parameters by Tertile Group According to Net PIPR Change and Sleep Quality

 

Statistical Analyses

Normally and asymmetrically distributed data were expressed as mean ± standard deviation and median (interquartile range), respectively. Sleep-onset latency values were naturally log-transformed due to skewed distribution. The net PIPR change was classified into three groups according to the following tertiles: low (−12.52% to −0.11%), intermediate (−0.07% to 3.51%), and high (3.60% to 22.45%) (Supplementary Fig. S1). Trends analyses were conducted to evaluate the association between the tertile groups of net PIPR change and normally distributed or categorical variables using linear or logistic regression models, respectively. The Jonckheere–Terpstra test was used for the analysis of variables with asymmetrical distribution. Categorical data were analyzed using the χ² test, whereas continuous data with a normal and asymmetrical distribution were analyzed using an unpaired t-test and the Mann–Whitney U test, respectively. Adjusted mean objective sleep parameters (total sleep time, sleep efficiency, wake after sleep onset, and sleep-onset latency) were assessed using analysis of covariance. Odds ratios (ORs) for subjective sleep disturbance with 95% confidence intervals (CIs) were calculated using a multivariable logistic regression model. Age and other variables that were marginally associated with net PIPR change and sleep disturbance (P < 0.2) in the univariate model were selected as potential confounders, including age (per year), sex (male or female), alcohol consumption (>30 g/d), diabetes (yes or no), benzodiazepine use (yes or no), and day length (per tertile). The interaction effects of net PIPR change (χ¹: high-, intermediate-, and low-tertile groups according to net PIPR changes of 0, 1, and 2, respectively) and the severity of glaucoma (χ²: non-severe glaucoma, 0; severe glaucoma, 1) on subjective sleep disturbance and objective sleep quality were assessed by adding an interaction term (χ¹ × χ²) to the logistic and linear regression formulas, respectively. All statistical analyses were performed using SPSS Statistics 28 (IBM, Chicago, IL, USA), with a two-side P < 0.05 indicating statistical significance. 

Results

The mean age of the 138 patients with glaucoma was 70.3 ± 11.5 years, among whom 67 (48.6%) were men, 58 (42.0%) had subjective sleep disturbance, and 81 (58.7%) had severe glaucoma. The mean net PIPR change was 1.69% ± 5.32%. The mean baseline pupil sizes before the red and blue light stimuli were 5.11 ± 1.08 mm and 4.86 ± 1.03 mm, respectively. No significant associations were observed between the tertile groups for net PIPR change and basic, clinical, and circadian rhythm parameters (Table 1). Patients with sleep disturbance were significantly less likely to be male (P = 0.005), had greater benzodiazepine use (P < 0.001), and had significantly longer day length than did those without sleep disturbance (P = 0.035) (Table 1). 

The prevalence of the patients with subjective sleep disturbance significantly increased with decreasing tertile groups according to net PIPR change (P for trend = 0.036) (Fig. 2). Subgroup analysis according to glaucoma severity found similar results in the severe glaucoma group (P = 0.004) but not in the non-severe glaucoma group (P = 0.81) (Fig. 2). Moreover, glaucoma severity and the tertile groups according to net PIPR change showed a significant interaction effect with subjective sleep disturbance (P = 0.037 for interaction). In the severe glaucoma group, multivariable logistic regression analysis adjusted for potential confounders, including age, sex, alcohol consumption, diabetes, benzodiazepine use, and day length, showed that the low-tertile group according to net PIPR change had a significantly higher adjusted OR for subjective sleep disturbance than did the high-tertile group (age-adjusted model: OR = 6.13; 95% CI, 1.78–21.10; P = 0.004; model 3: OR = 6.22, 95% CI, 1.76–21.90; P = 0.004) (Table 2). 

Figure 2.

 

Bar graph showing the prevalence of the patients with sleep disturbance in the tertile groups according to net PIPR change. The P value indicates the results of the trend analysis.

Table 2.

 

Odds Ratios for Subjective Sleep Disturbance

 

Table 3 shows the association between the tertile groups according to net PIPR change and objective sleep parameters. In the severe glaucoma group, the age-adjusted model showed that the low-tertile group exhibited a significantly shorter total sleep time, lower sleep efficiency, and longer wake after sleep onset than did the high-tertile group according to net PIPR change (P = 0.009, P = 0.027, and P = 0.035, respectively). Similar associations were observed in the multivariable models adjusted for age, sex, alcohol consumption, diabetes, benzodiazepine use, and day length. However, in the non-severe glaucoma group, multivariable analysis showed no significant association between the tertile groups according to net PIPR change and objective sleep parameters (Table 3). In addition, the interaction terms of glaucoma severity and tertile groups according to net PIPR change were significantly associated with subjective sleep disturbance (total sleep time, P = 0.014; sleep efficiency, P = 0.001; sleep-onset latency, P = 0.002). 

Table 3.

 

Association Between PIPR and Objective Sleep Parameters

 

Discussion

Our cross-sectional study of 138 patients with glaucoma investigated the association between sleep quality and ipRGC function evaluated using PIPR. The novel results of the present study showed that impaired ipRGC function in patients with glaucoma was significantly associated with decreased subjective and objective sleep quality. In addition, our findings revealed that glaucoma severity and PIPR had an interaction effect on sleep disturbance. Previous studies on the matter performed univariable analysis, possibly leading to inaccurate results due to the presence of confounding factors. Our multivariable analysis, which adjusts for potential confounders, provides robust evidence regarding the association between ipRGC function and sleep quality in patients with glaucoma. The strengths of this study include our large sample size, multivariable analysis adjusting for potential confounders, and objective measurement of sleep quality. 

Previous studies on the association between PIPR and sleep quality in patients with glaucoma have shown mixed results. One cross-sectional study using polysomnography to evaluate 32 patients with glaucoma and 13 participants without glaucoma showed that ipRGC function was partly associated with objective sleep quality, such as the number of arousals during the night. However, another clinical study using the PSQI questionnaire to evaluate 15 patients with glaucoma and 17 participants without glaucoma reported no significant association between ipRGC function and subjective sleep quality. This inconsistency in results may be attributed to small sample size and glaucoma severity. To address this issue, the present study supported this association by evaluating 138 patients with glaucoma, subsequently showing an association between ipRGC function and sleep quality in patients with severe glaucoma but not non-severe glaucoma. This indicates that discrepancies in results could have been attributed to the low the number patients with severe glaucoma in the earlier studies. 

Our results suggest that impaired ipRGC function is a potential risk factor for sleep disturbance in patients with glaucoma. Although a few clinical studies have reported an association between glaucoma and sleep disturbance,^11,12 the cause of the sleep disturbance in such patients remains unclear. The sleep cycle is regulated via two processes—namely, sleep–wake-dependent homeostasis and circadian clock controlled by suprachiasmatic nuclei.²⁶ One cross-sectional study reported that ipRGC function evaluated using the PIPR was significantly worse in participants with delayed sleep–wake phase disorder than in controls.²⁷ Histological and clinical evidence has shown that patients with glaucoma exhibit loss of ipRGCs.^2,3 Thus, impaired ipRGC function caused by glaucoma may lead to sleep disturbance through reduced light transmission. 

The interaction effect of glaucoma severity and PIPR on sleep quality may be explained by the influence of the fellow eye on photoentrainment and melatonin levels. Notably, one study found an abnormal circadian shift in participants with enucleation of both eyes,²⁸ whereas another clinical study on 18 totally visually blind participants showed a deficiency in light-induced regulation of melatonin secretion in bilaterally blind participants.²⁹ These results indicate that bilateral severe ocular diseases likely cause circadian misalignment. In the present study, 37 (64.9%) of the 57 eyes with non-severe glaucoma had bilateral glaucoma, suggesting that the fellow eye may have compensated for photoentrainment in the same group. Another possible mechanism is the difference in the degree of melatonin levels between patients with severe and non-severe glaucoma. One of the biologic functions of melatonin involves regulation of sleep.³⁰ One clinical study showed that urinary melatonin secretion was 0.30 log ng/mg creatinine lower in patients with severe glaucoma than in those with non-severe glaucoma.⁶ Thus, impaired ipRGC function in eyes with severe glaucoma in addition to lower melatonin levels may promote sleep disturbance. 

The present study has several limitations. First, PIPR was evaluated in the unilateral eye with worse glaucoma following the PIPR measurement protocol of providing light stimulus to the mydriatic eye and evaluating the fellow eye without mydriasis. Non-image–forming light information mediated by ipRGCs of the fellow eye might have influenced circadian photoentrainment. However, despite using unilateral PIPR data from the eye with worse glaucoma, the analysis revealed a significant association between PIPR and decreased sleep quality in patients with glaucoma. Second, after dividing our patients into three groups according to net PIPR change for subgroup analyses, the number of patients in each group was relatively small. Consequently, adjustment for potential confounders during multivariable analysis was restricted, and wide 95% CI ranges were obtained during statistical analyses. Third, the influence of melatonin levels on sleep quality in patients with glaucoma is unclear. Impaired ipRGC function in patients with glaucoma leads to decreased melatonin levels through circadian misalignment, possibly resulting in sleep disturbance. Further longitudinal studies are needed to determine whether ipRGC function affects sleep quality through decreased melatonin levels. Fourth, given that our study used a cross-sectional research design, we could not determine whether impaired ipRGC function causes poor sleep quality. Fifth, out of 172 eyes, 29 eyes (16.9%) were excluded due to unstable and unreliable pupillary light reflex. One study evaluating ipRGC function using pupillary light reflex among 214 participants reported that 43 eyes (20.1%) with an unreliable pupillary light reflex were excluded.³¹ Another study of 45 participants without ocular disease reported the exclusion of three participants (6.6%) due to unstable data.¹⁵ Thus, the exclusion number in our study was consistent with the results of these previous studies. Sixth, lens status (i.e., the presence of cataracts) was not evaluated. Age-related lens opacities decrease in light transmission to the retina. An increase in lens opacity blocks the blue light that excites ipRGCs, possibly affecting the net PIPR change. Therefore, the negative value of net PIPR change in the low-tertile group may have been caused by cataracts. Finally, no age-matched control group without glaucoma was included. Therefore, whether PIPR in the elderly participants without glaucoma is associated with sleep quality remains unclear. 

In conclusion, ipRGC dysfunction evaluated using PIPR was significantly associated with subjective and objective sleep disturbance in patients with severe glaucoma but not non-severe glaucoma. This association was independent of potential confounders, including age and sex. 

Acknowledgments

The authors thank Michiru Higuchi and Yuki Ouchi for their help with data collection. 

Supported by grants from JSPS KAKENHI (22K09797, 19K09956); by the Mitsui Sumitomo Insurance Welfare Foundation; by the Osaka Gas Group Welfare Foundation; by research grants from Novartis Pharma and Alcon; by a Nara Medical University Grant-in-Aid for Young Scientists; by Setsuro Fujii Memorial–The Osaka Foundation for Promotion of Fundamental Medical Research; by the Osaka Community Foundation; and by a Bayer Retina Award. 

Disclosure: H. Jimura, None; T. Yoshikawa, None; K. Obayashi, None; K. Miyata, None; K. Saeki, None; N. Ogata, None