eFigure 1. Classification of diffuse macular damage.
eFigure 2. Classification of focal macular damage.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Hirji SH, Hood DC, Liebmann JM, Blumberg DM. Association of Patterns of Glaucomatous Macular Damage With Contrast Sensitivity and Facial Recognition in Patients With Glaucoma. JAMA Ophthalmol. 2021;139(1):27–32. doi:10.1001/jamaophthalmol.2020.4749
What is the association of focal and diffuse glaucomatous macular damage with the clinical outcomes of facial recognition and contrast sensitivity?
In this cohort study including 144 eyes from 72 patients with glaucoma, across disease severity, diffuse but not focal glaucomatous macular damage was associated with facial recognition and contrast sensitivity impairments.
These results suggest that diffuse glaucomatous macular damage is disproportionately associated with the ability of patients with glaucoma to recognize faces, which may directly impact patient quality of life.
Facial recognition is a critical activity of daily living that relies on macular function. Glaucomatous macular damage may result in impaired facial recognition that may negatively affect patient quality of life.
To evaluate the association of patterns of glaucomatous macular damage with contrast sensitivity and facial recognition among patients with glaucoma.
Design, Setting, and Participants
In this prospective cohort study at a single tertiary care center, 144 eyes of 72 consecutive patients with glaucoma with good visual acuity (20/40 or better in each eye) were studied. Data were collected from March to April 2019.
Eyes with macular damage were categorized as having focal, diffuse, or mixed (focal and diffuse) damage based on optic disc and macular spectral-domain optical coherence tomography and 10-2 visual field (VF) damage. Only eyes with focal or diffuse damage were included. Higher-acuity and lower-acuity eyes were determined by 10-2 VF mean deviation (MD). Facial disability was defined as facial recognition scores at the 2% level of normal participants.
Main Outcomes and Measures
(1) Monocular contrast threshold as measured by the Freiburg Visual Acuity and Contrast Test and (2) binocular facial recognition as measured by the Cambridge Face Memory Test.
Of the 72 included patients, 49 (68%) were White and 41 (57%) were female, and the mean (SD) age was 67.0 (11.6) years. Eyes with diffuse damage had greater contrast impairment compared with eyes with focal damage (β = −0.5; 95% CI, −0.6 to −0.4; P < .001) after adjusting for 10-2 VF MD, 24-2 VF MD, age, presence of an early cataract, and number of drops. Similarly, Cambridge Face Memory Test scores were significantly lower in patients with diffuse rather than focal macular damage, regardless of eye (better-seeing eye: β = 10.0; 95% CI, 2.0 to 18.2; P = .001; worse-seeing eye: β = 5.5; 95% CI, 0.8 to 10.0; P = .23). Relative risk of facial disability was greater for patients with diffuse but not focal macular damage in the better-seeing eye (relative risk, 86.2; 95% CI, 2.7 to 2783.3; P = .01).
Conclusions and Relevance
In this cohort study, diffuse rather than focal glaucomatous macular damage was associated with diminished facial recognition and contrast sensitivity. Evaluation of macular optical coherence tomography and 10-2 VF and resultant detection of diffuse macular damage may help minimize glaucoma-related visual disability.
Facial recognition is an activity of daily living that is vital for social interaction and forming and maintaining relationships. Impairment in facial recognition negatively impacts quality of life by affecting patients’ ability to interact socially.1-3 This is a complex visual task that relies on central visual function and contrast sensitivity (CS).4,5 For this reason, facial recognition impairment has been explored more thoroughly in pathologies that classically affect the macula, such as age-related macular degeneration.6,7
Although previous reports have suggested that facial recognition is impaired in patients with glaucoma,8,9 the mechanism has been poorly understood. Our recent work suggested a correlation between severity of glaucomatous macular damage and degree of facial recognition impairment.10 Further, we hypothesized that because CS is a clinical measure of macular function and is critical to successful facial recognition,11-13 it may be an important component to the association of macular damage with impairment of facial recognition.
Early glaucomatous damage to the macula region can be focal and/or diffuse.14 These 2 patterns of macular damage are best differentiated by comparing abnormal regions on the 10-2 visual field (VF) with the corresponding regions on retinal ganglion cell (RGC) and retinal nerve fiber layer (RNFL) probability maps obtained from spectral-domain optical coherence tomography (SD-OCT) macular cube and optic disc scans.14-16 On central VF testing, eyes with diffuse macular damage have a mild generalized depression, whereas eyes with focal macular damage have a dense paracentral defect.14 To date, it is unclear how each of these types of macular damage contributes to visual disability. The primary purpose of this study is to evaluate the association of the pattern of glaucomatous macular damage with 2 dimensions of visual function: CS and facial recognition. Although the characterizations of focal and diffuse macular damage are not mutually exclusive and frequently co-occur in advanced glaucoma, our intent was to gain a better understanding of the relative contributions of focal and diffuse macular damage in visual disability. As such, we have limited our analyses to eyes with focal or diffuse damage.
Patients with glaucoma were enrolled in a prospective cohort study conducted at the Department of Ophthalmology at Columbia University Irving Medical Center, New York, New York, from March 25 to April 30, 2019. The Institutional Review Board of Columbia University Medical Center approved the study and its methods. The study adhered to the regulations of the Health Insurance Portability and Accountability Act. Due to institutional review board determination that the study was of no risk to patients, verbal informed consent was sufficient and obtained from all participants. Patients were not offered incentives or compensation to participate in the study. All study methods adhered to the tenets of the Declaration of Helsinki.
A total of 72 eligible consecutive patients were recruited for this study. All participants underwent a comprehensive ophthalmologic examination, including medical history review, best-corrected visual acuity, slitlamp biomicroscopy, intraocular pressure measurement, gonioscopy, dilated ophthalmoscopy, and standard automated perimetry (24-2 and 10-2 programs using the Swedish interactive threshold algorithm; Carl Zeiss Meditec). Individuals were included if they had a diagnosis of primary open-angle glaucoma in 1 or both eyes as determined by a glaucoma specialist. Glaucoma was determined based on characteristic optic nerve damage on stereoscopic examination with localized or diffuse RNFL thinning on SD-OCT and open anterior chamber angles on gonioscopy. Glaucomatous eyes, both with and without macular damage as determined by 10-2 VF and macular OCT, were included. Additional inclusion criteria were a visual acuity of 20/40 or better in each eye and a score of 30 or more on the Short Test of Mental Status (STMS). Those with ocular or systemic disease that could affect the optic nerve, macula, or VF were excluded. Specifically, patients with epiretinal membranes, drusen, or age-related macular degeneration were eliminated from study entry. Lenses were graded according to the LOCS III criteria.17 Patients with 2 or more nuclear sclerotic changes or any posterior subcapsular or cortical lenticular changes were excluded. Patients with posterior capsule opacification, as assessed by slitlamp examination and best-corrected visual acuity, were excluded. Patients receiving miotic therapy or those whose pupil diameter was less than 3 millimeters as measured by the Humphrey VF auto pupil feature were excluded. Patients with severe dry eye or ophthalmic surface disease as determined by the glaucoma specialist or by diminished signal strength (less than 7) on OCT were also excluded. Eyes were required to be orthophoric for inclusion. Patients were required to have a reliable VF, defined as less than 33% fixation losses or false-negative errors and less than 15% false-positive errors. To increase specificity, the abnormal regions of the RNFL, RGC plus inner plexiform layer (RGC+), SD-OCT, and VF were required to show topographical agreement for study inclusion. Therefore, all 72 included patients had strict structure-function correlation. For the purposes of this study, structure-function agreement was required for classification of macular damage. Therefore, eyes without corresponding damage on both SD-OCT and VF were classified as having no damage. Specifically, eyes with RGC+ thinning without corresponding VF loss were classified as having no damage.
SD-OCT images (Cirrus SD-OCT Macular Cube 512 × 128 scan; Carl Zeiss Meditec) were acquired by a trained ophthalmic photographer. Images were excluded if they contained motion or blinking artifacts, incorrect placement of the measurement circle, segmentation error, or poor image quality (signal strength less than 7). VF testing was performed with the Humphrey Field Analyzer II (Carl Zeiss Meditec) with appropriate refractive correction. The 10-2 and 24-2 Swedish interactive threshold algorithm standard automated perimetry programs were used for monocular VF testing. All VF testing was performed by a trained ophthalmic technician. For this study, better-seeing and worse-seeing eyes were defined based on the 10-2 VF mean deviation (MD).
The SD-OCT RNFL and RGC+ and 10-2 VF data were examined for the presence of corresponding regions of macular damage on VF and OCT probability maps to classify eyes as having focal, diffuse, or mixed macular damage. The method of classifying macular damage as focal or diffuse has been described elsewhere.14 In brief, diffuse macular loss on SD-OCT was defined as RGC+ macular thinning that was widespread and included both hemiretinas, with corresponding thinning of the circumpapillary RNFL and corresponding shallow, widespread reduction of sensitivity on 10-2 VFs.14 Diffuse loss is demonstrated in eFigure 1 in the Supplement. Focal damage was defined as a local region of circumpapillary RNFL thinning, usually found at the border of temporal and inferior quadrants, with preservation of RNFL thickness in the temporal disc14 with corresponding RGC loss and paracentral VF loss, as seen in eFigure 2 in the Supplement. Specifically, diffuse macular damage was defined as 3 or more abnormal (5% or less) points on both hemifields of the 10-2 VF total deviation probability maps and abnormal (5% or less) superior and inferior RGC+ thinning with corresponding temporal RNFL thinning.14 Focal macular thinning was defined as 3 or more abnormal (2% or more) contiguous points on the 1 hemifield of the 10-2 pattern deviation probability maps with abnormal (2% or less) contralateral RGC+ and RNFL thinning.14 For the purposes of this study, structure-function agreement was required for a classification of macular damage. Any eye with mixed macular damage that could not be classified as having primarily focal or diffuse damage was not included in study analyses, although the fellow eye was eligible for study inclusion.
Monocular visual acuity was measured using a rear-illuminated Early Treatment Diabetic Retinopathy Study screen. Visual acuity was scored as the total number of letters identified correctly and was converted to logMAR. Monocular contrast threshold was measured using the Freiburg Visual Acuity and Contrast Test (FrACT) according to FrACT protocol.18 If a patient’s contrast threshold is high, a greater difference in contrast is needed to distinguish 2 objects from one another. Therefore, larger contrast thresholds determined by FrACT correlate to greater impairment in CS. Specifically, contrast threshold is inversely proportional to CS. FrACT was administered on a 15.4-inch 2015 Macbook Pro laptop (Apple) at a resolution of 1680 × 1050 at 60 Hz. All participants completed the FrACT contrast test at a viewing distance of 40 cm and had their head mounted in a chin rest.
Facial recognition was assessed using the Cambridge Face Memory Test (CFMT). The CFMT is a freely available, validated test.19 It has been used to quantify facial recognition defects in a variety of clinical conditions, including macular degeneration and glaucoma.8,20 A full description of the methods and test validation is described by Duchaine and Nakayama.19 The CFMT has been found to produce both accurate and reliable results.21,22 Briefly, participants binocularly view 6 target faces at 3 different viewing angles. Their recognition of these faces is subsequently tested in a series of forced-choice recognition trials in which they must identify the previously seen target face from an additional 2 unfamiliar faces. The outcome measure for the test is the total number of correctly identified faces, with a maximum score of 72 and a mean (SD) score in healthy participants of 57.9 (7.9).19 The testing protocol was identical to that used for FrACT, although patients performed the testing binocularly. The image size and viewing angle were calculated to mirror viewing faces at a distance of roughly 1 meter in the real world.
Comparisons between eyes with focal and diffuse macular damage were performed using an independent, 2-tailed, unpaired t test for means and χ2 statistics for proportions. Since it is believed that the visual system responds logarithmically to changes in stimulations,23 contrast measurements were converted to logarithm of contrast for all analysis. Since contrast testing was performed monocularly, the association of pattern of macular damage (focal or diffuse) with contrast was assessed using a linear mixed-effects model to adjust for intereye correlation.
For facial recognition, better-seeing and worse-seeing eye univariable and multivariable linear regression analyses using ordinary least squares was conducted using pattern of macular damage (focal or diffuse) as the predictor and CFMT score as the outcome measure. Facial disability was defined as facial recognition values that were at the 2% level of normal participants or 2 SDs below the mean. Better-seeing and worse-seeing eye univariable and multivariable logistic regression analyses were conducted using pattern of macular damage as the predictor and facial disability as the outcome measure. For both contrast threshold and CFMT, variables that were related to the outcome scores (P ≤ .20) in the univariable analyses were then entered into the multivariable regression model.
A sample size calculation was performed a priori for an expected difference of 5 faces between groups. All analyses were run for better-seeing and worse-seeing eyes. Significance was set at a P value less than .05. Statistical analyses were performed using Stata version 15 (StataCorp).
A total of 73 participants were recruited for this study. One participant failed the STMS and was not included in the analyses, leaving 72 participants. Of the 72 included patients, 49 (68%) were White and 41 (57%) were female, and the mean (SD) age was 67.0 (11.6) years. Age, race/ethnicity, and sex were not found to be significant univariable predictors of facial recognition.
Of the 72 better-seeing eyes, 44 eyes (61%) had macular damage and 28 (39%) did not. Of 44 eyes with macular damage, 22 (50%) had focal damage, 15 (34%) had diffuse damage, and 7 (16%) had mixed focal and diffuse damage. Of the 72 worse-seeing eyes, 63 eyes (88%) had macular damage and 9 (12.5%) did not. Of 63 eyes with macular damage, 26 (41%) had focal damage, 19 (30%) had diffuse damage, and 18 (29%) had mixed focal and diffuse damage.
For both better-seeing and worse-seeing eyes, the only significant difference in baseline ocular characteristics between focal and diffuse groups was age. Despite this, there was no significant difference in cognitive function between the focal and diffuse groups as reflected by STMS performance.
The presence of diffuse macular damage was associated with significant contrast impairment compared with eyes with focal macular damage (β = −0.5; 95% CI, −0.7 to −0.3; P < .001). Diffuse macular damage remained a significant predictor of contrast impairment (β = −0.5; 95% CI, −0.6 to −0.4; P < .001) even after adjusting for age, visual acuity, 10-2 VF MD, and 24-2 VF MD.
As the relative contribution of each eye in facial recognition is unknown, all regression analyses were performed separately for better-seeing and worse-seeing eyes. The mean (SD) number of faces recognized was 50.2 (8.4).
Among better-seeing eyes as determined by 10-2 VF MD, of 72 faces on the CFMT, eyes with focal macular damage recognized 10.0 (95% CI, 2.0-18.2; P = .001) more faces than eyes with diffuse macular damage, even after adjusting for disease severity (10-2 VF and 24-2 VF MDs) as well as number of drops, age, visual acuity, and CS. Results of the multivariable linear regression model used to calculate difference in CFMT score (β) are listed in Table 1. We next performed univariable and multivariable logistic regression analyses to identify risk factors associated with facial disability. As noted above, facial disability was defined as facial recognition in the lowest 5% of normal.19 In the univariable model, the relative risk of facial disability was significantly higher in patients with diffuse rather than focal damage (relative risk, 11.4; 95% CI, 1.9-67.3; P = .007). In the multivariable model, the relative risk of facial disability was 86.2 (95% CI, 2.7-2783.3; P = .01) for patients with diffuse rather than focal damage, even after adjusting for number of intraocular pressure–lowering medications, 10-2 VF MD, visual acuity, 24-2 VF MD, and CS. Results of the multivariable model are listed in Table 2.
Among worse-seeing eyes as determined by 10-2 VF MD, of 72 faces on the CFMT, eyes with focal macular damage recognized 5.5 (95% CI, 0.8-10.0; P = .02) more faces than eyes with diffuse macular damage, even after adjusting for disease severity (10-2 VF and 24-2 VF MDs) as well as number of drops, age, visual acuity, and CS. Results of the multivariable linear regression model used to calculate difference in CFMT score (β) are listed in Table 3. However, in contrast to the better-seeing eye, we found no significant difference in relative risk of facial disability between worse-seeing eyes with focal and diffuse macular damage.
Glaucomatous damage has been previously reported to affect facial recognition.8,9,24,25 Our recent work found an association specifically of glaucomatous macular damage with deficits in facial recognition.10,26 This study extends our research by classifying macular damage into focal and diffuse patterns.14 Our results suggest that diffuse macular damage is disproportionately associated with both CS and facial recognition impairments. Although contrast may partially mediate the association of macular damage with facial recognition, diffuse macular damage remains an independent predictor of facial recognition in the multivariable model, suggesting that contrast impairment may be only one dimension of visual disability from diffuse macular damage.
Previous work has shown that the distinction between focal and diffuse macular damage has a significant effect on patient quality of life. Patients with diffuse but not focal thinning on SD-OCT RGC+ macular scans had decreased National Eye Institute Visual Function Questionnaire scores.27 Furthermore, these 2 patterns of damage vary in their symptomology under low luminance conditions. Patients with diffuse but not focal macular damage were shown to have diminished scores on the Low Luminance Questionnaire relative to patients without macular damage.28 These findings suggest that it is important to both detect macular damage and determine the pattern of said damage to better understand the visual symptoms experienced by patients with glaucoma.
Aging, nuclear sclerosis, refractive error, pupillary size, dry eyes, and macular disease may all affect CS and potentially facial recognition.11,29-32 Therefore, we had strict inclusion criteria and ran multivariable regressions to adjust for potential confounding. As our study criteria excluded patients with diminished visual acuity, visually significant cataract, posterior capsule opacification, severe dry eye, macular disease, and miotic pupils, we did not find these potential confounders to be associated with facial recognition in our sample. In a population-based sample, it is likely that these factors would contribute to diminished facial recognition. In this study, we followed the guidelines for ambient lighting as recommended by the FrACT protocol,18 measured by the mobile Light Meter app (Nuwaste Studios). Future studies should evaluate the role of binocular CS in facial recognition and assess this association in low-luminance conditions.
Our study had limitations. Translating an eye-based test into a person’s visual ability poses inherent difficulty. It is unclear how better-seeing and worse-seeing eyes interact in facial recognition, so we performed analyses for each eye independently. As facial recognition is a binocular task, patients with the same pattern of macular damage in both eyes may experience variable disability compared with patients with different patterns of macular damage. It is also possible that ocular dominance affected our results. Future studies should investigate the interaction between patients’ 2 eyes in the process of facial recognition.
As glaucoma progresses, the distinction between focal and diffuse damage may be lost. We therefore had a group of eyes with mixed macular damage that could not be classified as predominantly focal or diffuse. We are not suggesting that the characterizations of focal and diffuse are mutually exclusive. Instead, we assigned patents to focal or diffuse categories so that we could better understand the pathophysiology of how macular damage impairs visual ability.
In summary, this work increases our understanding of the visual disability experienced by patients with glaucoma. In particular, a diffuse pattern of macular damage was associated with greater impairment in both CS and facial recognition compared with a focal pattern of damage. Although diffuse damage may be more difficult to identify than focal damage, it is an important determinant of visual function and should prompt early and aggressive intervention. Therefore, careful evaluation of the macular OCT and 10-2 VF and resultant early detection of diffuse macular damage is critical to minimize glaucoma-related visual disability.
Accepted for Publication: September 21, 2020.
Published Online: November 5, 2020. doi:10.1001/jamaophthalmol.2020.4749
Corresponding Author: Dana M. Blumberg, MD, MPH, Bernard and Shirlee Brown Glaucoma Research Laboratory, Department of Ophthalmology, Edward S. Harkness Eye Institute, Columbia University Irving Medical Center, 635 W 165th St, PO Box 77, New York, NY 10032 (email@example.com).
Author Contributions: Dr Hirji had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Hirji, Hood, Blumberg.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Hirji, Blumberg.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Blumberg.
Administrative, technical, or material support: All authors.
Study supervision: Liebmann, Blumberg.
Conflict of Interest Disclosures: Dr Hood has received grants, consulting fees, and equipment from Heidelberg Engineering and Topcon and consulting fees from Novartis. Dr Liebmann has received grants from Heidelberg Engineering and the National Eye Institute; consulting fees from Aerie Pharmaceuticals, Allergan, Carl Zeiss Meditech, Galimedix Therapeutics, Heidelberg Engineering, Infocus Capital Partners, Merck & Co, Quark Pharmaceuticals, Qura, Sensimed, and Sustained Nano Systems; holds patents and receives royalties from Théa Pharmaceuticals; and owns equity in Diopsys Corporation, Galimedix Therapeutics, Qura, SOLX, and Sustained Nano Systems. No other disclosures were reported.