Relationships between mesopic visual sensitivity and macular inner and outer retinal layer thickness in healthy younger, middle‐aged and older adults

To examine relationships between mesopic visual sensitivity measurements on microperimetry and macular inner and outer retinal layer (IRL and ORL) thicknesses in healthy younger, middle‐aged and older subjects.


| I N T RODUC T ION
As a consequence of population ageing, the number of visually impaired people continues to increase (GBD 2019 Blindness andVision Impairment Collaborators, 2021).During normal ageing, the ability to see in low-light conditions, such as driving at night, and to identify lowcontrast objects substantially worsens (Owsley, 2011).While optical factors such as pupillary miosis and changes in the lens are able to explain some age-related decline in visual function, the losses produced over time in factors such as contrast sensitivity at low luminance or mesopic light levels are better explained by neural changes (Andersen, 2012;Gillespie-Gallery et al., 2013;Keuken et al., 2022;Owsley, 2011;Spear, 1993).These neural changes produced with normal ageing include changes in rod density (Curcio et al., 1993), and a reduced capacity of cones to absorb photons, which impairs contrast sensitivity at low luminance intensities in healthy older adults (Silvestre et al., 2019).Postreceptor changes have also been identified such as reduced ganglion cell density which contributes to the worsening of spatial vision that occurs in normal human aging (Calkins, 2013;Gao & Hollyfield, 1992;Harman et al., 2000).These neural changes are characterized by high-inter-individual variability and may increase an individual's susceptibility to age-related neurodegenerative diseases such as glaucoma or age-related macular degeneration (AMD).These findings determine a need for a better understanding of the role played by underlying neural mechanisms in the decline in visual function produced during ageing.
While so far several studies have examined how modifications in retinal morphology determined by spectral domain optical coherence tomography (SD-OCT) correlate with visual function in AMD (Schmidt-Erfurth & Waldstein, 2016) and glaucoma (Yohannan & Boland, 2017), this issue has hardly been addressed in healthy subjects.In young and older healthy eyes, structure-function relationships have been detected between retinal layer thicknesses and contrast sensitivity (Puell et al., 2018), mesopic visual acuity and low-luminance deficit (Puell et al., 2019) and flicker modulation sensitivity (Pérez-Carrasco et al., 2021).Hence, information about retinal structure-function relationships at different ages in healthy eyes may be useful to elucidate the mechanisms behind the loss of visual function caused by aging.In addition, this knowledge could help to early identify subjects at high risk of developing retinal diseases so that appropriate control and prevention measures can be taken.
Ocular structure-function relationships have been extensively studied through standard automated perimetry (SAP).However, with this technique, the projection of stimuli onto the retina can vary anatomically due to fixation losses and head misalignments.In contrast, microperimetry or fundus-controlled perimetry, is able to superimpose functional measurements on corresponding structural images.This technique can determine visual sensitivity, specifically differential luminance sensitivity (DLS), at specific retinal locations due to real-time eye movement tracking (Talib et al., 2018) with applications as an outcome measure in clinical trials (Pfau et al., 2020).Microperimetry has been used to assess central retinal function mostly in persons with AMD (Cassels et al., 2018;Takahashi et al., 2016).Studies have also shown that mean retinal sensitivity declines with age in healthy eyes (Molina-Martín et al., 2017;Sabates et al., 2011).While microperimetry and SD-OCT combined has been used to investigate macular structure-function relationships in AMD (Pilotto et al., 2016;Roh et al., 2019;Saßmannshausen et al., 2018;Sayegh et al., 2014;Steinberg et al., 2016;Tepelus et al., 2017;Wu et al., 2014), glaucoma (Garcia-Medina et al., 2021) and diabetes without retinopathy (Montesano et al., 2021;Orduna-Hospital et al., 2021), to the best of our knowledge, no prior work has tried to relate macular retinal layer thickness to mesopic visual sensitivity in younger, middle-aged and older healthy adults.
Our working hypothesis was that regional retinal sensitivities will correlate with their corresponding retinal layer thicknesses in healthy subjects.The present study was therefore designed to examine whether regional mesopic visual sensitivity (differential luminance sensitivity, DLS) measured by microperimetry correlates directly with differences in SD-OCT macular inner retinal layer (IRL, a ganglion cell (GC)-related layer) and outer retinal layer (ORL) thicknesses in healthy eyes, and to establish how these relationships vary between age groups.The findings of this study may help to understand how neural performance, as reflected by retinal layer thickness, affects visual function in healthy eyes across all age groups.

| M AT ER I A L S A N D M ET HOD S
This was a cross-sectional study carried out at the Faculty of Optics and Optometry, Universidad Complutense de Madrid, Madrid, Spain.Participants were 154 healthy subjects aged 20-80 years (mean 43.6 ± 17.9 years of age) recruited among university undergraduates, students of an older adult studies program (Universidad para los Mayores, UCM) and staff, and among those visiting the outpatients University Optometry Clinic for a routine ocular examination.The study protocol complied with the principles of the Declaration of Helsinki and received institutional review board approval.Written informed consent was obtained from each participant.
Inclusion criteria were a best-corrected visual acuity (BCVA) of 0.1 logMAR (20/25 Snellen) or better, a refractive error no greater than 3.50 diopters (D) of sphere or 1.50 D of cylinder and normal findings in the eye examination.Exclusion criteria were systemic diseases, such as diabetes, previous ocular surgery, lens opacities LOCS III classification grade 2 or greater, medications, or an ocular disease such as glaucoma, amblyopia and retinal abnormality.Measurements were made in only one eye of each subject.If both eyes fulfilled the inclusion criteria the right eye was selected.
All the participants were subjected to a full ocular examination at the University Optometry Clinic to detect loss of visual function or the presence of clinically recognized disease.Ophthalmic assessments included BCVA, subjective refraction, optical biometry, anterior pole slit-lamp biomicroscopy and fundus examination.Normal retinal health was confirmed by colour fundus photography and SD-OCT.
BCVA was measured monocularly using an ETDRS chart placed in an illuminator cabinet 4 m from the patient.Participants were encouraged to guess letters, even if they were unsure.Each letter read correctly on each line was given a score of 0.02 log units.Values were expressed as the logarithm of the minimum angle of resolution (logMAR).Axial length (AL) and pupil size were measured by optical biometry (Lenstar LS 900, Haag-Streit Diagnostics).Means of five repeated and reliable measurements were recorded.As retinal illuminance may affect retinal sensitivity, we used pupil size to calculate retinal illuminance expressed in trolands.
Retinal layer thicknesses were measured at the macula in the selected eye in all participants with the 3.3 iVue OCT system (Optovue Inc.).The Retina Map Scan protocol was used for macular retinal layer thicknesses using the grid employed in the Early Treatment Diabetic Retinopathy Study (ETDRS).Scanning was performed through an undilated pupil in conditions of dark room lighting, and only high-quality images showing a Scan Quality Index >65 were accepted.
In this study, retinal layer thickness was determined only in the 5 subfields of the ETDRS grid (3 × 3 mm): central fovea (1 mm diameter) and the quadrants temporal (T), nasal (N), superior (S) and inferior (I) of a ring of outer diameter 3 mm and inner diameter of 1 mm centered at the fovea.Mean retinal layer thicknesses were determined for each of three retinal segmentations automatically measured in the central foveal subfield and parafoveal ring: (1) total retinal layer, measured from the internal limiting membrane to the retinal pigment epithelium; (2) inner retinal layer (IRL) measured from the internal limiting membrane to the outer limit of the inner plexiform layer (IPL) including the macular RNFL, GC layer and IPL; and (3) ORL measured from the outer limit of the IPL to the retinal pigment epithelium (Figure 1).Total, IRL and ORL thicknesses were measured in the five ETDRS subfields.Furthermore, overall parafoveal thickness was calculated as the average of the values measured at each parafoveal ring quadrant.
Mesopic visual sensitivity (DLS) at specific retinal locations was measured using the commercially available macular integrity assessment (MAIA) microperimeter (Centervue) in a dark room.Briefly, in this test white LED stimuli of Goldmann III size (0.43 degrees diameter) are presented to the subject over different locations and a 4-2 decibels (dB) staircase strategy is used to measure the minimum luminance (threshold luminance whose inverse is the luminance sensitivity) necessary to perceive spot stimuli against a mesopic (1.27 cd/ m 2 ) background (differential luminance sensitivity; Talib et al., 2018).The dynamic range of each stimulus is 36 dB.The stimuli grid is composed of 37 testing locations covering the central 10-degree diameter circle of the macular area.Twelve light stimuli are presented at each of the 1-degree, 3-degree and 5-degree radii, and one stimulus at the central fovea (Figure 1).A real-time fundus image is generated with a confocal scanning laser ophthalmoscope and an eye-tracking system ensures that there is correspondence between the light stimulus and retinal location on the fundus image throughout the examination.Tests were preceded by a practice run with the "Fast" protocol to minimize the learning effect.To examine correlations between the measured structure-function variables, we related the location of the MAIA stimuli to the corresponding macular SD-OCT regions by establishing five corresponding subfields (one central and 4 parafoveal subfields) according to the 3 × 3 mm ETDRS grid used in SD-OCT and MAIA (Figure 1).In the central region (C), the central and the 12 sensitivity points located in the 1° radius (diameter of 0.6 mm) were averaged for each subject to produce a measure of central DLS that corresponds to the central foveal ETDRS subfield (1-mm diameter).The parafoveal region of the MAIA (3°-5° radius corresponding to diameters of 1.8 and 3 mm, respectively) corresponds to the 3-mm diameter parafoveal ETDRS ring.In this parafoveal ring, average DLS values of the sensitivity points located in the S, T, I and N quadrants were calculated for each participant.These four regions correspond to the S, T, I and N quadrants of the parafoveal ETDRS ring used in SD-OCT.Furthermore, overall parafoveal DLS was calculated as the average of the values measured in each quadrant of the parafoveal ring.The mean sensitivity obtained for all the stimulus locations was also considered.

| Statistical analysis
The normality of the data was checked through graphical inspections and the Shapiro-Wilk test.One-way analysis of variance (ANOVA) was used to compare normally distributed variables, while the Kruskal-Wallis test was used for non normally distributed continuous variables.For multiple comparisons, we used the Holm-Bonferroni post-hoc correction.The post-hoc Fisher's least significant difference procedure was used to determine which means were significantly different.In an initial analysis, mean sensitivities measured across all the stimulus locations were compared for six decades of age to determine which age groups were homogeneous for this factor given the known decline in mean sensitivity that occurs with age.As a result, participants were divided into three age groups: younger (20-39 years; n = 74), middle-aged (40-59 years; n = 42) and older (60-79 years; n = 38) based on their homogenous mean sensitivity values.The minimum sample size estimated to detect a significant anticipated correlation coefficient of 0.50, an alpha risk of 5.0% and a power of 90% was 38.
Associations between each of the dependent DLS variables and corresponding retinal layer thicknesses (independent variables) in each of the five subfields were examined in each age group by estimating Spearman correlation coefficients.Central DLS was correlated with central foveal thickness.Average parafoveal DLS values for each of the N, S, T, I subfields were correlated with averages for the total, ORL and IRL thicknesses of each corresponding quadrant of the parafoveal ring.Furthermore, overall parafoveal DLS was correlated with average macular thicknesses (total, ORL and IRL) at the parafoveal ring.To test the same hypothesis in the three sections (total, ORL, IRL) of each retinal thickness variable, the Holm-Bonferroni corrected p-values were calculated.
To correct for potential confounding factors, a forward stepwise multiple linear regression analysis was performed.In this analysis, we identified predictors of regional DLS in each age group separately and corrected for the covariates age, gender, BCVA, spherical equivalent, axial length, retinal illuminance and each corresponding retinal thickness.These retinal thickness predictors were log 10 transformed prior to analysis to match the scale of DLS loss.p-Values were corrected through the Holm-Bonferroni adjustment.Furthermore, an analysis of covariance was performed to compare regression lines relating DLS and retinal layer thicknesses for the three age groups.All the statistical tests were performed using the package Statgraphics 19-X64 (Statpoint Technologies, Inc.).

| R E SU LT S
Demographic and ocular characteristics for the young, middle-aged and older age groups are provided in Table 1.
In Figure 2, we display the distribution of mean mesopic sensitivity measured by microperimetry over the total 10°-diameter macular visual field for six age intervals or decades.Through ANOVA, a significant effect of decade of age on mean DLS was observed (p < 0.0001).Mean sensitivity obtained across all stimulus locations showed a 2.56 dB difference between the older and younger age groups.According to our post-hoc analysis, mean DLS failed to vary significantly from one decade to the next yet differed to the remaining age groups.This meant that the age groups 20-29 and 30-39 years (young age groups) were similar and did not differ significantly from each other in terms of mean sensitivity.Similarly, mean DLS did not differ significantly between the age groups 40-49 and 50-59 years (middle-aged groups) or age groups 60-69 and 70-79 years (older age groups).However, DLS means did differ when the three age groups with homogeneous DLS values (young, middle-aged and older) were compared to the other age groups, with the older age group showing higher standard deviations for DLS values obtained for each macular region.Table 2 shows regional mean mesopic DLS determined in the macular central subfield and in each of the parafoveal ring quadrants (S, T, N, I and overall) for the three age groups homogenous in terms of mean DLS.Regional means for DLS were significantly different in the three age groups (all p < 0.001).No significant differences were detected in total mean DLS or in mean DLS values for the central subfield or parafoveal quadrants between men and women in each of the three age groups.
Mean total and ORL thicknesses measured by SD-OCT were not significantly different in the central subfield and each of the four parafoveal subfields of the ETDRS grid for the three age groups.However, age had a significant effect on IRL thickness in each quadrant of the parafoveal ring.Retinal layer (ORL and IRL) thicknesses recorded in the four quadrants of the parafoveal ring in the younger, middle-aged and older age groups are shown in Figure 3.In the S, N and I subfields, mean IRL thicknesses were significantly lower in the older age group compared to the other age groups (all p < 0.0001).In the T quadrant, mean IRL thickness decreased with age and was significantly different for each of the three age groups (p < 0.0001).
Spearman correlations for mean DLS versus mean retinal thickness in each corresponding subfields were examined in each age group.No significant correlations were recorded between DLS and retinal layer thicknesses in the younger age group.Mean DLS measured in the inferior parafoveal ring quadrant was significantly  correlated with mean IRL thickness in the middle-aged group (rho = −0.50;p = 0.0045) and with ORL thickness in the older age group (rho = −0.52;p = 0.0045).These correlations were negative, meaning that a thicker retinal layer was related to a worse DLS.For the other subfields (central, superior, temporal, nasal), no significant correlations between mean DLS and corresponding total, outer or inner retinal thicknesses emerged in the middleaged and older age groups.Forward stepwise multiple linear regression analyses were conducted to identify the independent predictors of DLS in each age group.The independent variables, or predictors, were retinal thicknesses, age, sex, SE, axial length, BCVA and retinal illuminance.The results of the models with significant outputs are shown in Table 3. IRL thickness in the inferior region of the parafoveal ring and SE emerged as significant predictors of DLS in the inferior quadrant (R 2 model = 33%; p = 0.0004) in the middle-aged group.In this age group, a thicker IRL was correlated with a worse DLS (p = 0.0207).In the older age group, ORL and IRL thickness in the inferior parafoveal ring were identified as significant contributing factors to inferior parafoveal DLS (ORL p < 0.0001; IRL p = 0.0003).In this age group, a thicker ORL and thinner IRL showed correlation with a worse DLS.The estimated effects of a 0.1 log micron (1.26 micron) increase in ORL thickness resulted in a 5.5 dB decrease in DLS.On the contrary, 0.1 log micron increase in IRL thickness resulted in a 2.2 dB increase in DLS.The model explained 51% of the variance of inferior parafoveal DLS (p = 0.0257) in the older age group.Although Spearman correlation between mean DLS and inferior quadrant IRL thickness was not significant, both variables were correlated (p = 0.0003) in the forward stepwise multiple regression analysis.One explanation could be that correlation between DLS and IRL depends on the range of ORL values.When only eyes with ORL values equal to or greater than the median value were considered, the univariate correlation observed between DLS and IRL was high (r = 0.69, p = 0.0009).In contrast, no correlation was detected for eyes with ORL values lower than the median.Total retinal thickness, age, sex, SE, axial length, BCVA and retinal illuminance were not found to be significant predictors of inferior mean DLS. Figure 4 shows the relationships detected between mesopic DLS and ORL and IRL thicknesses in the inferior parafoveal ring quadrant for the younger, middle-aged and older age groups.We also compared regression lines relating inferior DLS to retinal layer thicknesses in the three age groups.Comparisons were significant for both IRL thickness (p = 0.0001) and ORL thickness (p = 0.0001).Significant differences emerged between the slopes of regression lines relating DLS to IRL thickness (p = 0.0027) or ORL thickness (p = 0.0020) in the three age groups, and between the intercepts of regression lines (IRL p = 0.0001; ORL p = 0.0001).When considering IRL thickness, the model was able to explain 44% of the variability in DLS, and when considering ORL thickness, the model explained 46% of this variability.

| DI SC US SION
This study examines relationships between regional macular IRL (GC-related layer) and ORL thicknesses imaged by SD-OCT and their corresponding mesopic retinal sensitivities (DLS) measured through MAIA microperimetry in healthy adults of different age.Our main finding was that macular IRL and ORL thickness were significant independent predictors of DLS which were differentially and significantly affected by age.In the inferior quadrant of the parafoveal ring, a worse DLS was associated with thicker IRL in our middleaged participants and with a thicker ORL and thinner IRL thickness in our older age group.No significant correlations were detected between DLS and macular thicknesses in the younger age group.
In agreement with the finding that mean retinal sensitivity as measured with the MAIA microperimeter decreases with age in healthy eyes (Molina-Martín et al., 2017), we found that the means for mesopic DLS obtained across all the stimulus locations decreased with age but there were three age groups that were homogenous in terms of DLS (younger, middle-aged and older).As the DLS means recorded in each of the parafoveal ring quadrants (S, T, N, I) were significantly lower in the older age group than middle-aged and younger groups, which in turn also differed significantly from each other, relationships between retinal layer thicknesses and mesopic DLS were analysed separately in each of these age groups showing similar DLS values.To our knowledge, this structure-function relationship has not been examined previously in healthy eyes of different ages.
To control for potential confounding factors such as age, sex, BCVA, spherical equivalent, axial length, retinal illuminance and retinal layer thicknesses, a stepwise forward multiple regression analysis was conducted.Significant structure-function relationships were found in the inferior quadrant of the parafoveal ring and these relationships differed significantly across age groups.While no association was detected for the younger age group, the negative correlation noted between macular IRL thickness and mesopic DLS in the middle-aged group changed to positive correlation in the older age group.In this last group, a thinner IRL was a predictor of worse DLS, in agreement with the findings of a study in which macular IRL thinning was associated with reduced contrast sensitivity measured in photopic and mesopic light conditions in older healthy eyes (Puell et al., 2018).In AMD, mean retinal sensitivity as determined by MAIA microperimetry decreases with increasing age, presence of advanced AMD and retina thinning (Roh et al., 2019).In the present study, age had a significant effect on GC-related layer (IRL) thickness in each T A B L E 3 Forward stepwise multiple regression models used to identify independent predictors of differential luminance sensitivity as measured by mesopic microperimetry in the inferior parafoveal quadrant.

Dependent variable
Predictor subfield of the parafoveal ring.Thus, mean IRL thickness was significantly lower in the older age group compared to the other groups.Several studies have shown intra-retinal layer thickness changes with aging (Altay et al., 2017;Demirkaya et al., 2013;Kim et al., 2011;Ooto et al., 2011;Won et al., 2016).While findings have not always been consistent, most studies have found that the thickness of the inner macular layers (including ganglion cell-inner plexiform layer or the ganglion cell complex) decreases with age.The IRL thinning associated with a worse mesopic DLS in some of our older eyes could therefore reflect a loss of ganglion cells (Altay et al., 2017).
In contrast with the positive association found for the older age group, correlation between IRL thickness and DLS was negative for the middle-aged group, meaning that an increased thickness gave rise to a worse mean DLS in the inferior region of the parafoveal ring.The IRL thickening observed in some, supposedly healthy, middle-aged eyes could be the consequence of increased glial tissue (Vecino et al., 2016) leading to transient macular IRL thickening which might be responsible for the lower DLS.
In our study, inferior IRL and ORL thickness explained 51% of the variance of inferior DLS (p = 0.0257) in the older age group.However, a 0.1 log micron (1.26 micron) change in ORL thickness had a greater effect on DLS than the same change in IRL thickness (5.5 dB and 2.2 dB, respectively).Accordingly, we found that a greater ORL thickness was associated with a worse DLS in the older age group, despite the fact that mean ORL thicknesses were not significantly different for the three age groups.As commented previously, the changes in intra-retinal layers produced with aging remain unclear but in a recent study no association was observed between age and ORL thickness in healthy subjects over 60 years of age (Altay et al., 2017).The increase produced in ORL thickness in some of our older eyes could perhaps be attributed to low-grade inflammation in the retina likely because of mild neural tissue swelling, causing a discrete increase in retinal thickness affecting DLS.Some authors have reported low-grade inflammation in the healthy aging retina in physiological conditions (Xu et al., 2009).In agreement with our observations, in older healthy eyes with good BCVA, increased parafoveal ORL thickness and/ or central foveal thickness (1 mm-central-subfield) have been related to a worse mesopic VA and greater low-luminance deficit (difference between photopic and mesopic VA), respectively (Puell et al., 2019).Furthermore, in primary open-angle glaucoma, negative correlations were found between retinal sensitivity, as assessed by microperimetry, and the thicknesses of the macular ORLs, but positive correlations were also found for the IRL.These structure-function relationships showed different behaviour depending on the retinal layer considered in agreement with our finding in the older age group (Garcia-Medina et al., 2021).
In our younger age group, no significant relationships were detected between macular layer thicknesses and mesopic DLS.Consistently, no association was found by others between central foveal and parafoveal thicknesses and mesopic visual acuity in young eyes (Puell et al., 2019).On the contrary, increased macular IRL thickness has been correlated with a worse contrast sensitivity in young eyes (Puell et al., 2018).It is thus likely that the sensitivity of the laboratory technique used to measure contrast sensitivity was able to capture subtle changes that microperimetry or visual acuity could not.
Significant structure-function relationships were found here only in the inferior quadrant of the parafoveal ring.glaucoma, RGC axons in the inferior hemisphere have been found to be most susceptible to glaucomatous damage (Na et al., 2011;Nakatani et al., 2011), and the inferior outer sector in the macular area was shown to have the greater diagnostic capacity (Na et al., 2011), and inferior inner macular volume to correlate most with visual field mean deviation (Nakatani et al., 2011).Furthermore, in a study in persons with primary openangle glaucoma, correlations detected between retinal sensitivity measured by microperimetry and macular GC layer thickness were stronger in the inferior hemisphere (Garcia-Medina et al., 2021).
To our knowledge, this is the first study to show that relationships between mesopic retinal sensitivity and both IRL and/or ORL thickness in the inferior macular region differ significantly among age groups.The slopes of the regression lines used to predict DLS from retinal layer thickness were significantly different for the comparisons involving IRL thickness and age group (p = 0.0001) and ORL thickness and age group (p = 0.0001).Our regression models were able to explain 44% (IRL) and 46% (ORL) of the variability noted in DLS.Hence, it could be that the IRL thickening associated with a worse retinal sensitivity in some middle-aged eyes progressed to IRL thinning and/or ORL thickening with increasing age.Longitudinal studies are needed to assess changes in structure-function relationships associated with ageing to understand how early neural retinal damage can advance to affect visual function.
A limitation of our study is its cross-sectional design, so no conclusions can be drawn about how aging causes changes in neural performance (reflected by retinal layer thickness) that affect visual performance in healthy eyes.Another limitation is the method used to measure retinal layer thicknesses.Retinal segmentation methods capable of measuring individual retinal layers might help better identify the underlying neural changes responsible for the retinal sensitivity changes observed here.
In conclusion, microperimetry-derived mesopic visual sensitivity was predicted by ORL and IRL (GC-related layer) thicknesses measured in the inferior parafoveal ring quadrant, but the sign of this relationship changed across age groups.Accordingly, a worse DLS was related to a thicker macular IRL in middle-aged eyes and to a thicker ORL and thinner IRL in older eyes.Considering the great individual variability in the effects of aging, measuring retinal sensitivity in combination with macular retinal layer thicknesses could be a useful tool to detect early changes possibly leading to neurodegenerative diseases of the eye.

AC K NO W L E DGE M E N T S
The authors thank Marcos Sanz-Velasco MsC and Marina Casado-Velasco MsC for their help with participant recruitment and clinical measurements, and thank Ricardo García-Mata (Computer Services and Research Support at UCM) for his support in the statistical analysis of our data.

F
I G U R E 1 Example of the right eye of a healthy study subject showing the corresponding regions of the macular area subjected to SD-OCT and MAIA.(a) Inner (GC-related) and outer retinal layer segmented automatically by the SD-OCT system.The yellow and green dots indicate the location of the light stimuli in the horizontal meridian.(b) Retinal fundus image showing on the 3 × 3 mm ETDRS grid the 37 test locations covering the central 1 mm subfield and four quadrants (superior (S), temporal (T), inferior (I), nasal (N)) of the parafoveal ring (1-3 mm).
Demographic and ocular characteristics of the younger, middle-aged and older age group participants of this study.Mean ± SD (min, max)., best corrected visual acuity measured using ETDRS letter charts.F I G U R E 2 Box plots of mean retinal sensitivity (dB) measured by mesopic microperimetry across all the stimulus locations in the 10° (diameter) macular field of the study eyes in subjects stratified into decades of age.The boxes enclose the interquartile range, the whiskers indicate the 90th and 10th percentiles, and outliers show the 95th and 5th percentiles.T A B L E 2 Mean mesopic visual sensitivity determined in the macula region (10° diameter): central subfield and temporal, nasal, superior and inferior quadrants of a parafoveal ring in the younger, middle-aged and older age groups.

F
Box plots of retinal thickness measured by spectral-domain optical coherence tomography: outer (a) and inner (b) retinal layer thicknesses recorded in the four quadrants (superior, temporal, nasal and inferior) of the parafoveal ring of the ETDRS grid in the younger, middle-aged and older age groups.The boxes enclose the interquartile range, the whiskers indicate the 90th and 10th percentiles, and outliers show the 95th and 5th percentiles.

F
Relationships between mesopic visual sensitivity and inner (○) and outer (•) retinal layer thickness in the inferior parafoveal quadrant.Panel A: younger age group, Panel B: middleaged group, Panel C: older age group.
The covariates age, gender, BCVA, spherical equivalent, axial length, retinal illuminance and each corresponding retinal thicknesses were entered into the model.Partial p value in brackets. Note: