FormalPara Key Summary Points

Why carry out this study?

Chromatic pupillometry presents high efficacy in the detection of pupillary light response (PLR) dysfunction in glaucoma.

The current study investigated the changes in PLR induced by full-field and quadrant-field stimuli in primary open-angle glaucoma (POAG), as well as examining the correlation between PLR features and macular measurements.

As far as we are aware, this is the first study to examine the correlation between PLR features and macular measurements.

What was learned from this study?

Our findings highlight the potential effectiveness of utilizing a head-mounted chromatic pupillometer as an objective method for detecting glaucomatous damage.

The observed impairment of PLR may signify concomitant structural macular damage, underscoring the diagnostic value of this approach.

Introduction

Glaucoma is one of the leading causes of irreversible blindness worldwide [1, 2]. The glaucomatous neuropathy is characterized by loss of retinal ganglion cells (RGCs), leading to the presence of structural hallmarks such as thinning of RGC-related retinal layers, and enlargement of the cup-to-disc ratio [2,3,4]. The most common form of glaucoma is primary open-angle glaucoma (POAG) with a prevalence of approximately 3.1% globally [2]. The current diagnostic tools for detecting glaucoma, such as optical coherence tomography (OCT) and standard automated perimetry (SAP), require expensive and highly calibrated equipment. However, SAP is subjective and time-consuming, and may not always reflect the glaucomatous structural changes [4]. The macular region contains more than half of all RGCs and it was previously believed that POAG spared this region until the late stage of the disease [5]. However, recent studies have shown that OCT measurements of macular thickness can provide a highly accurate diagnostic performance for detecting glaucoma, even in the early stage [6, 7].

Chromatic pupillometry is a quantitative and objective assessment that measures the pupil light response (PLR) to different wavelengths and irradiance of light stimulus, providing a means to evaluate the integrity of retinal and optic nerve health [8,9,10,11]. With advancements in instrumentation and analysis techniques, the precise quantification of PLR features is now possible [12]. By employing various light wavelengths and irradiances, one can evaluate the overall integrity of PLR circuits driven by different retinal photoreceptor types, including rods, cones, and intrinsically photosensitive retinal ganglion cells (ipRGCs) [8, 11, 13, 14]. Rod and cone photoreceptors as well as ipRGCs differ in the properties of their response, thus providing an indication of the function of each cell type [11]. ipRGCs are a group of RGCs that transmit information to the suprachiasmatic nucleus (SCN) and the olivary pretectal nucleus (OPN), mediating non-image/image-forming vision, PLRs, and circadian photoentrainment [15,16,17,18,19]. Although ipRGCs constitute a small portion of total RGCs, several studies showed that ipRGC loss or dysfunction could be detected in patients with glaucoma [12, 20,21,22,23,24,25,26,27]. The shortened PLR constriction amplitude and sluggish post-illumination pupil response (PIPR) representing melanopsin dysfunction have been observed in patients with glaucoma [21,22,23,24, 26].

Although chromatic pupillometry presents promising results in the detection of PLR dysfunction in glaucoma, the situational constraints, interpretation of tests, and complexity of the devices make it challenging to implement in clinical or community-based settings. Little is known about the impact of glaucomatous macular damage, specifically macular RGC loss, on PLRs measured by chromatic pupillometry.

In this study, we aimed to explore whether the alterations in PLR features are related to glaucomatous RGC damage in the macular region, and to evaluate the feasibility of binocular chromatic pupillometer in opportunistic POAG diagnosis. The current study employed a head-mounted binocular pupillometer to investigate the changes in PLR features induced by full-field and quadrant-field stimuli in POAG. To the best of our knowledge, this is the first study that addresses the association between PLR characteristics and macular measurements in POAG.

Methods

Participants

Forty-six patients with POAG (16 female, 30 male) and 23 visually healthy control subjects (12 female, 11 male) participated. Glaucoma participants were recruited from the Eye and ENT Hospital of Fudan University (Shanghai, China). The research followed the tenets of the Declaration of Helsinki, and the study was approved by the human subjects review committee of the Eye and ENT Hospital of Fudan University. Written informed consent was obtained from each participant. All subjects received PLR tests on a headset (Fig. 1) and comprehensive ophthalmologic examinations including intraocular pressure measurement, slit-lamp stereo biomicroscopy, direct ophthalmoscopy, OCT, and SAP. Subjects with retinopathies, other causes of optic neuropathy, or ocular motor disorders were excluded from the study. Subjects with diabetes, psychiatric or neurologic disorders, including cognitive impairment or dementia, as well as those who had previously undergone intraocular surgery or were taking psychotropic or other medications that could potentially affect the PLR or alertness were also excluded.

Fig. 1
figure 1

a Subject underwent the PLR test on a headset, and the PLR curve was recorded by the surrounding IR cameras. b Scheme of the binocular chromatic pupillometer

Full size image

SAP Tests

SAP was evaluated with the Octopus 900 perimeter to test the visual field (VF) (Haag-Streit, Koeniz, Switzerland), G standard white/white TOP program. All SAP tests were required to meet reliability criteria. VF tests were defined as abnormal when they met one of the following conditions: (1) the presence of three or more adjacent points in the superior or inferior field with P < 5% probability of normal range and one or more points with P < 1% probability of normal range; (2) presence of two or more adjacent points with P < 1% probability of normal range and one or more points with P < 2% probability. Patients with POAG were further divided into mild–moderate stage (mean deviation (MD) scores < 12 dB) and severe-stage (MD scores ≥ 12 dB) groups according to Hodapp–Parrish–Anderson criteria [28].

OCT Examinations

Optical coherence tomography was performed with spectral domain optical coherence tomography OCT (Optovue, Inc., Fremont, CA, USA). All OCT scans were obtained using the MM5 protocol consisting of a 5 mm by 5 mm grid of horizontal and vertical scans through the macula, and a standard ganglion cell complex (GCC) scan. The inner retina was defined as the retinal nerve fiber layer, ganglion cell layer, and inner plexiform layer. Thickness and volume data were generated by the Optovue software and were recorded from thickness maps (Fig. 2c–f). The measurement regions and inner retinal segmentation boundaries are illustrated in Fig. 2c, d. For thickness and volume measurements, evaluated regions include the superior hemisphere, and inferior hemisphere areas of the parafoveal and perifoveal regions, as well as the central foveal zone [29].

Fig. 2
figure 2

a Chromatic pupillometry protocol of sequenced PLR tests. b PLR curve of one healthy eye (red line) and a glaucomatous eye (blue line). Representative optical coherence tomography macular scan used for thickness and volume measurements of a healthy control (c) and in a patient with POAG (d). The green line represents the boundaries of the inner retina. The fovea (yellow), parafovea (orange), and perifovea (red) regions are denoted. Representative thickness map of macular regions used for thickness and volume measurements in a healthy control (e) and in a patient with POAG (f). More green color represents a thinner inner retina

Full size image

Apparatus and Calibration of Binocular Pupillometer

The pupillometer setup consists of an HTC Vive Pro, with an integrated eye tracker (Tobii pro, Tobii Technology, Stockholm, Sweden). The visual stimuli are generated in Unity 2018, utilizes SteamVR 2.0, and runs on a laptop computer (MEHREV Z2, NVIDIA GeForce GTX 2060 i7-10870H@2.20 GHz, 16 GB RAM). Subjects were fitted with helmet-mounted displays (HMDs), using near-infrared (940 nm) illuminating diodes and infrared cameras to track pupillary data (sample rate 90 Hz). Both eyes were tested in turn. Stimulus wavelength and luminance were verified with a spectroradiometer (Konica CS-100A; Konica Minolta, Inc., Tokyo, Japan).

Stimuli and Procedures

After calibration of HMDs, dark adaption was performed by displaying a dark screen for 120 s. Full-field (110° in diameter) of short-wavelength (“blue” dominant wavelength of 465 nm, luminance of 100 cd/m2) pulses of light were shone for 1 s after dark adaption. After 60 s rest, the superior and inferior quadrant flash stimuli were presented in turn (Fig. 2a). The pupillary response curve was recorded (Fig. 2b).

Data Acquisition and Feature Extraction

During the experiment, real-time pupil diameter was recorded (90 Hz). Data were analyzed offline using custom scripts programmed in MATLAB (MathWorks, Inc., Natick, MA, USA), which automated interpolated eye movement and filtered short-time (< 500 ms) eye blinks.

The following components were extracted in three experiments:

Baseline pupil diameter (BPD) (mm): mean pupillary diameter during the initial 10 s before light onset.

Maximum constriction amplitude (%) = BPD-maximum constriction pupil diameter/BPD × 100%.

Time to maximum constriction (s): the interval between the beginning stimulus and peak constriction.

Time to pupil dilation (s): the interval between the beginning of light stimulus and the pupil diameter recovered to 90% BPD.

Constriction velocity (mm/s): BPD-maximum constriction pupil diameter (mm)/time to maximal constriction (s).

Dilation velocity (mm/s): BPD-maximum constriction pupil diameter (mm)/time to dilation(s).

Sustained response (mm): the pupil diameter at 6 s after the flashlight offset.

PIPR: BPD-PD at 6 s after flash offset/BPD × 100%.

Statistical Analysis

Ophthalmologic examination outcomes and pupillometric features are represented as means ± standard deviation or median (interquartile range). Comparisons between healthy controls and the different glaucoma severity groups were evaluated by using one-way analysis of variance (ANOVA) or Kruskal–Wallis one-way analysis of variance on ranks test. Post hoc analysis (LSD) or Dunn’s method was further performed to identify significant group differences. One-way ANOVA or the Wilcoxon rank-sum test was used to compare pairs of groups (Table 1). The correlation between pupillometric features and macular measures or MD score was assessed using Spearman’s rank correlation. A receiver operating characteristic curve (ROC) and the area under the ROC curve (AUC) analysis was calculated to compare the diagnostic value using the pROC package and reportROC package. The sensitivity, specificity, and 95% CI were calculated at the best statistical cutoff. Comparison between the AUCs of pupillometric features and macular structures was performed using DeLong’s method with Bonferroni correction. The threshold for significance for all statistical tests was set at 0.05. Statistical procedures were performed using R version 4.2.2 (R: a language and environment for statistical computing, R Core Team, Vienna, Austria).

Table 1 Ocular characteristics of the study eyes in the control and glaucoma groups
Full size table

Results

Of the 138 eligible eyes included in this study, 46 were assigned to the healthy control group and 92 were assigned to the glaucoma group: 72 eyes to the mild–moderate glaucoma group, and 20 to the severe glaucoma group. No significant differences in age and gender were found between the control group and the glaucoma group or among different staged glaucoma groups. The mean age of control subjects was 42.00 ± 11.08 years old, and the mean age of the glaucoma group was 41.00 ± 13.03 years old. There was no statistical difference in age distribution between the patients with glaucoma and healthy controls.

Macular Structural Features

Figure 2c, d illustrates the macular measurements for cases and controls. Figure 2e, f depicts the thickness maps of the inner retina. Table 1 and Fig. 3 summarize the macular thickness and volume measurements. The thickness (P = 0.445) and volume (P = 0.276) measurements of the fovea did not differ between the control and the glaucoma group. Most inner retinal measurements, namely parafoveal thickness, parafoveal volume, perifoveal thickness, and perifoveal volume, were reduced in glaucoma groups (Fig. 3A). The macular measurements in both hemispheres in the perifoveal and parafoveal areas were decreased in the glaucoma groups (Fig. 3B, C). The most discriminating parameter was the thickness of the inner retina perifoveal inferior hemisphere (13.5% decline in glaucoma), followed by the volume of the inner retina perifoveal inferior hemisphere (13.2% decline in glaucoma).

Fig. 3
figure 3

Macular measurements comparisons between control, mild–moderate glaucoma, and severe glaucoma. A (a–f) Central macular measurements. B (a–e) Inferior hemisphere of the macular region. C (a–e) Superior hemisphere of the macular region. Y-axis shows the log value of measurements

Full size image

Pupillometric Features

Figure 2b presents the PLR curves of a control subject (red) and a patient with glaucoma (blue) while being exposed in sequence to a full-field stimulus, a superior quadrant-field stimulus, and an inferior quadrant-field stimulus. The PLR curve presents as a transient constriction response followed by a dilation to baseline in response to three stimuli. The subject age or gender does not have a significant impact on pupillometric features.

Table 2 summarizes the pupillometric features of healthy controls and patients with different stages of glaucoma. In response to the full-field stimulus, the constriction amplitude was greater in controls compared to patients with glaucoma (P = 0.001), but there was no significance between groups with different stages of glaucoma (P = 0.506). The PIPR was decreased in the glaucoma group compared to the control group (P = 0.015), but no significant difference between groups with different stages of glaucoma was observed (P = 0.116). The time to pupil dilation was significantly prolonged in glaucoma groups (P < 0.001), and significantly different between the two glaucoma groups (P = 0.002). The constriction-related variables did not differ between controls and patients with glaucoma.

Table 2 Pupillometric features of the study eyes in the control and glaucoma groups
Full size table

In response to the superior quadrant-field stimulus, the most distinguishable PLR feature was the time to pupil dilation (P < 0.001). The constriction amplitude was greater in controls compared to patients with glaucoma (P = 0.007), and the time to maximum constriction was faster in controls compared to patients with glaucoma (P = 0.035). The PIPR to superior quadrant-field stimulus was decreased in the severe glaucoma group compared to the control group (P = 0.018).

In response to the inferior quadrant-field stimulus, the most distinguishable PLR feature was the time to pupil dilation (P < 0.001). The PIPR to inferior quadrant-field stimulus was significantly decreased in the severe glaucoma group compared to the control group (P = 0.035). Additionally, the velocity of constriction was also reduced in the glaucoma group (P = 0.011).

Correlation of Pupillometric Features and Ocular Characteristics

Spearman’s correlation coefficients were calculated to assess the relationship between pupillometric features and macular measurements (Fig. 4). In the full-field stimulus experiment, the time to pupil dilation was inversely correlated with the parafoveal thickness (r = − 0.284, P < 0.001), the parafoveal volume (r = − 0.279, P < 0.001), the perifoveal thickness (r = − 0.429, P < 0.001), and the perifoveal volume (r = − 0.364, P < 0.001). The PIPR to full-field stimulus was negatively correlated with the perifoveal thickness (r = − 0.290, P < 0.001) and the perifoveal volume (r = − 0.213, P = 0.011) (Fig. 4a).

Fig. 4
figure 4

Correlations of PLR features and macular structural measurements in full-field stimulus (a), superior quadrant-field stimulus (b), and inferior quadrant-field stimulus (c). X-axis shows the Spearman’s correlation coefficient. Symbols denote regions that are statistically significant between the cases and controls (*P < 0.05, **P < 0.01, ***P < 0.001). T time, Vel velocity, PIPR post-illumination pupil response

Full size image

In response to the superior quadrant stimulus, the time to pupil dilation was negatively correlated to the inferior hemisphere GCC thickness (r = − 0.531, P < 0.001), the inferior parafoveal thickness (r = − 0.232, P = 0.006), the inferior parafoveal volume (r = − 0.284, P < 0.001), the inferior perifoveal thickness (r = − 0.451, P < 0.001), and the inferior perifoveal volume (r = − 0.417, P < 0.001). The PIPR to the superior quadrant stimulus was negatively correlated to the inferior parafoveal thickness (r = − 0.210, P = 0.014), the inferior parafoveal volume (r = − 0.185, P = 0.030), and the inferior perifoveal thickness (r = − 0.205, P = 0.016). The constriction amplitude was inversely correlated to the inferior hemisphere GCC thickness (r = − 0.218, P = 0.010), the inferior perifoveal thickness (r = − 0.207, P = 0.014), and the inferior perifoveal volume (r = − 0.208, P = 0.014).

In response to the inferior quadrant stimulus, the time to pupil dilation was negatively correlated with the superior hemisphere GCC thickness (r = − 0.383, P < 0.001), the superior parafoveal thickness (r = − 0.218, P = 0.010), the superior parafoveal volume (r = − 0.189, P = 0.026), the superior perifoveal thickness (r = − 0.299, P < 0.001), and the superior perifoveal volume (r = − 0.304, P < 0.001). The PIPR to the inferior quadrant stimulus was negatively correlated to the superior parafoveal thickness (r = − 0.273, P = 0.001), the superior parafoveal volume (r = − 0.251, P = 0.003), the superior perifoveal thickness (r = − 0.310, P < 0.001), and the superior perifoveal volume (r = − 0.290, P < 0.001). The dilation velocity in response to the inferior quadrant stimulus was correlated with the superior perifoveal thickness (r = − 0.211, P = 0.013) and volume (r = 0.203, P = 0.207).

Table 3 summarizes the correlations between pupillometric features and vision field median deviation. In response to the full-field stimulus, the constriction amplitude (r = − 0.372, P < 0.001), constriction velocity (r = − 0.293, P < 0.001), and the dilation velocity (r = − 0.225, P = 0.008) were inversely correlated to MD score. The post-illumination response (r = 0.280, P < 0.001) and dilation time (r = 0.479, P < 0.001) were positively correlated with MD score. In response to the superior quadrant-field stimulus, the post-illumination response (r = 0.184, P = 0.030) and the dilation time (r = − 0.502, P < 0.001) were positively correlated with the MD score. The dilation velocity (r = − 0.205, P = 0.016) was negatively correlated with MD score. In response to the inferior quadrant-field stimulus, the dilation time (r = 0.515, P < 0.001) was positively correlated with the MD score, and the velocity in constriction (r = − 0.177, P = 0.037) and velocity in dilation (r = − 0.238, P = 0.005) were negatively correlated with MD score. Overall, the strongest correlations with MD score were the dilation time in response to the inferior quadrant stimulus, followed by the dilation time in response to the superior quadrant and full-field stimulus. The maximum constriction amplitude and PIPR in full-field stimulus also showed moderate correlations.

Table 3 Spearman’s rank correlations for pupillometric features and vision field median deviation
Full size table

Diagnostic Performance of Pupillometric Features

Twenty-one pupillometric features were evaluated for the diagnostic performance of glaucoma (Table 4). The ROC curves are presented in Fig. 5. The variable with the greatest AUC was the dilation time in response to the superior quadrant-field stimulus (AUC 0.909, sensitivity 0.837, specificity 0.935), followed by the dilation time in response to the full-field stimulus (AUC 0.833, sensitivity 0.739, specificity 0.761), and the dilation time in response to the inferior quadrant-field stimulus (AUC 0.821, sensitivity 0.696, specificity 0.804). The constriction amplitude (AUCfull-field 0.681, sensitivityfull-field 0.598, specificityfull-field 0.761; AUCsuperior quadrant 0.641, sensitivitysuperior quadrant 0.815, specificitysuperior quadrant 0.435; AUCinferior quadrant 0.610, sensitivitysuperior quadrant 0.348, specificitysuperior quadrant 0.870) and PIPR (AUCfull-field 0.627, sensitivityfull-field 0.554, specificityfull-field 0.739; AUCsuperior quadrant 0.586, sensitivitysuperior quadrant 0.891, specificitysuperior quadrant 0.304; AUCinferior quadrant 0.597, sensitivitysuperior quadrant 0.859 specificitysuperior quadrant 0.391) also showed good diagnostic value in three experimental tasks.

Table 4 Diagnostic value of pupillometric parameters
Full size table
Fig. 5
figure 5

Diagnostic values of pupillometric features. a–g ROC curves of PLR features in full-field stimuli; h–n ROC curves of PLR features in superior quadrant-field stimuli; o–u ROC curves of PLR features in inferior quadrant-field stimuli. Black line = AUC in discriminating control vs glaucoma, red line = AUC in discriminating control vs mild–moderate glaucoma; blue line = AUC in discriminating control vs severe glaucoma; orange line = AUC in discriminating mild–moderate glaucoma vs severe glaucoma. Data are represented as average (95% CI). ROC receiver operating characteristic curve, AUC area under the ROC curve, T time, Vel velocity, PIPR post-illumination pupil response

Full size image

Discussion

Using a head-mounted chromatic pupillometer, we conducted measurements of the pupillary response to full-field and sectorial light stimuli and discovered a correlation between PLR features and macular structures in glaucoma. These findings suggest that PLR tests could serve as a rapid and quantitative diagnostic evaluation tool for glaucoma, reflecting the structural macular impairment.

It has long been believed that glaucomatous arcuate damage spares the macular region until the end stage [30]. However, macular measures are of special interest now because of the density of RGCs located in this region and the realization that, contrary to conventional opinion, the macula often is involved early in the glaucomatous process [6, 31, 32]. Structural biomarkers can now be measured in patients with glaucoma in the macula with an increasing number of OCT parameters available [33, 34]. The macular measurements have been shown to possibly be of value for monitoring eyes with advanced glaucoma, beyond the floor observed in circumpapillary retinal nerve fiber layer (cpRNFL) measurements [35]. Wan et al. found that the inner macular thickness could discriminate glaucomatous eyes from healthy eyes even in the early stage [7]. The current study also expands on previous research by analyzing measurements of the inner macula. The inner retinal thickness and volume at the fovea do not alter significantly in patients with glaucoma in our study. The most discriminating parameters are the thickness and volume of the inner retina perifoveal inferior hemisphere. These results are in line with a previous study reporting that the average RGC-related layer shows greater glaucomatous thinning in the inferior retina in patients with or suspected of having glaucoma, since most of the inferior macula projects to the inferior quadrant of the disc, a region known to be particularly vulnerable to glaucomatous damage [31].

Functional glaucomatous progression has both personal and societal costs and therefore has a serious impact on quality of life [36, 37]. The VF examinations, however, are subjective, time-consuming, and lack reproducibility in some circumstances [37,38,39]. The chromatic pupillometry has been proven useful for objectively evaluating the ipRGC dysfunction in patients with glaucoma [11]. Measurement of the PLR features induced by bright blue light stimuli showed impaired pupillary responses in patients with glaucoma and a reduced post-illumination pupillary response after light offset, representing the melanopsin dysfunction [11, 22,23,24, 26, 27]. Some studies indicated that the magnitude of the PIPR after exposure to a bright blue light stimulus correlated with the magnitude of visual field loss, and retinal nerve fiber layer (RNFL) thinning [22, 40, 41]. Rukmini et al. showed that the PLR was reduced in patients with POAG at high irradiance levels, associating strongly with disease severity with a significant linear correlation between pupil diameter and mean deviation, as well as cup-to-disc ratio [26]. Gracitelli et al. observed that the sustained pupillary response to the blue flash was associated with the average RNFL thickness as well as VF loss [40]. Although some studies indicated that ipRGCs are relatively resistant to glaucomatous damage and did not decrease until the late stage [25, 40, 42], Najjar et al. showed that patients with early-stage POAG exhibit reduced PLR to moderate and high irradiances of chromatic lights, and the PLR alteration correlates with RNFL thinning but not the VF scores [24]. Using quadrant field stimuli, Adhikari et al. detected attenuated PIPR in individuals with early-stage glaucoma, preperimetric glaucoma, and suspected glaucoma [20]. In this study, we utilized HMDs with binocular screens to conduct the examination (Fig. 1), without the need for an eyepatch or a dark room. Our findings demonstrated that a cost-effective and time-efficient test, which can be performed in simple circumstances, has the potential to detect glaucomatous damage.

Findings from previous studies have primarily focused on the classification performance of PLR for the detection of overall functional loss in glaucoma [23, 24, 27, 40, 41]. In our full-field stimulus experiment, we found that dilation time back to baseline (AUC 0.833, 95% CI 0.764–0.902) demonstrated better diagnostic performance than constriction amplitude (AUC 0.681, 95% CI 0.583–0.778) and PIPR (AUC 0.627, 95% CI 0.535–0.719). The dilation time in the full-field experiment also showed good performance in discriminating mild–moderate glaucoma from healthy eyes (AUC 0.800, 95% CI 0.720–0.880), and in discriminating mild–moderate glaucoma from severe glaucoma (AUC 0.765, 95% CI 0.653–0.877) (Fig. 5, Table 4). In our study, the glaucoma eyes exhibit sluggish dilation due to afferent defects, a response which could be used to detect or monitor the progression of glaucoma. The altered PLR features in the full-field stimulus are correlated with MD scores (amplitude full-field r = − 0.372, P < 0.001; PIPR full-field r = 0.280, P < 0.001; T dilation full-field r = 0.479, P < 0.001). These PLR features also presented a stronger association with parafoveal and perifoveal measurements (Fig. 4a). The patients with glaucoma included in our study are at a relatively early disease stage (MD = 5.65 dB), which suggests that the altered PLR features (amplitude full-field, PIPR full-field, T dilation full-field) in full-field experiments may indicate ipRGC dysfunction or loss, reflecting macular structural damage at an early stage.

Several previous studies used the Ganzfeld dome [24, 26, 40] or the Maxwellian system [20, 43] as the light source in PLR tests, generating a relatively narrow area of the light stimulus. In our study, we used the head-mounted display (110° in diameter), providing stimuli of a larger area. Further, we used the sectorial stimuli to evaluate if the regional structural defect could be detected by chromatic pupillometry. Similar to the finding in the full-field stimulus experiment, the most distinct PLR feature was the time to pupil dilation in response to both quadrant stimuli (P < 0.001 in both comparisons), the PIPR to quadrant-field stimulus was decreased in the severe glaucoma group relative to the healthy control group (superior quadrant, P = 0.018; inferior quadrant, P = 0.035). The alteration in PLRs correlated well with regional macular measurements (Fig. 4B, C), especially the dilation time to superior and inferior quadrant-field stimulus. These results suggest that the structure–function association between inner macular measurements (parafoveal thickness and volume, perifoveal thickness and volume) and PLR features translates to ipRGC loss or dysfunction at the corresponding region. Of note, we found that the dilation time to superior quadrant stimulus (AUC 0.909, 95% CI 0.855–0.963) and inferior quadrant stimulus (AUC 0.821, 95% CI 0.750–0.892) demonstrated excellent diagnostic accuracy in discriminating glaucoma eyes from healthy eyes. The PLR testing paradigm of our chromatic binocular pupillometry could be used in community settings, general practitioner’s clinics, and in telemedicine, reducing the cost of accurate glaucoma detection. Longitudinal studies could be carried out to further evaluate the potential of chromatic binocular pupillometry in tracking glaucoma progression.

We acknowledge some limitations with our study. First, the PLR to stimuli could not be entirely isolated to the ipRGC circuit. However, the PLR to the melanopsin-weighted stimuli is considered to be weighted by ipRGCs because they participate to a major extent in the response [44]. Second, abnormal PLRs may occur in multiple ocular diseases [8, 10, 11, 44,45,46,47] and other systemic diseases [48,49,50,51]. For example, previous studies showed that patients with outer retinal disease (retinitis pigmentosa or Leber congenital amaurosis) presented attenuated PLRs to the rod-weighted dim blue light stimulus, as well as weak responses to the cone-weighted red light stimulus. In contrast, the PIPR for the melanopsin-weighted stimulus appeared normal in patients with retinitis pigmentosa or Leber congenital amaurosis [10, 45]. Patients with non-proliferative diabetic retinopathy (NPDR) presented normal PLRs to the rod-weighted stimulus and reduced PLRs to the cone-weighted stimulus. Patients with mild or moderate-to-severe NPDR also presented an attenuated PIPR in the melanopsin-weighted stimulus [47]. In patients with early age-related macular degeneration, PIPR amplitude to melanopsin-weighted stimulus was attenuated, but not the latency or transient amplitude [46]. By comparing different PLR features responding to different stimuli one could distinguish glaucoma from other diseases. Although this alteration may lower the specificity of pupillometry for glaucoma detection, the ability to detect other diseases is beneficial for screening strategies in opportunistic examinations.

Conclusion

The PLRs of glaucomatous eyes are impaired in response to full-field and sectoral light stimulation. The pupil redilation after constriction, the constriction amplitude, and the PIPR are sensitive parameters for the detection of glaucomatous damage. The altered PLR features correlates well with the macular structural change in patients with glaucoma. A major advantage of our approach is that it can be adapted for population screening in compact devices, providing objective data on the integrity of the PLR circuits. A long-term study with both PLR tests and macular structural measurements is required to ascertain the temporal sequence of change in macular structure and PLR features during the development and progression of glaucoma.