Association of Structural and Functional Measures with Contrast Sensitivity in Glaucoma

Abstract

PURPOSE

To test the hypothesis that structural and functional measures predict contrast sensitivity (CS) outcomes in glaucomatous eyes.

DESIGN

Cross-sectional prospective study.

METHODS

105 eyes of 65 patients who underwent macular SD-OCT imaging, 24-2 standard achromatic visual fields (VF) and CS measurement on the same day were enrolled. Association of CS at 4 spatial frequencies (3, 6, 12 and 18 cycles per degree, cpd), with structural and functional outcomes was explored with correlation and regression analyses.

RESULTS

The median (IQR) 24-2 visual field mean deviation was −7.6 (−11.1 to −3.0). Significant correlations were found between CS at 6 cpd and ganglion cell/inner plexiform layer thickness at inferotemporal and inferonasal macular sectors (ρ =0.222, p=0.023 and ρ =0.209, p=0.032, respectively). CS at 6 cpd demonstrated higher correlations with full macular thickness measurements, the strongest of which was with the central macular thickness in the superior 6×3 degree region (ρ =0.311, p=0.001). Contrast sensitivity at 6 cpd also had the strongest correlation with mean deviation of the 4 central VF points (ρ=−0.420; p<0.001). There was a significant correlation between logMAR visual acuity and contrast sensitivity at 6, 12, and 18 cpd (ρ =−0.306, −0.348 and −0.241, p<0.013, respectively).

CONCLUSIONS

Structural and functional measures showed a fair relationship with contrast sensitivity. This association was most prominent between full thickness macular measures or central VF parameters and CS at 6 cpd. Contrast sensitivity was not a reliable surrogate for glaucoma severity in this cross-sectional study.

Introduction

Contrast sensitivity (CS) is an important aspect of human vision and is frequently affected in patients with glaucoma with various degrees of damage.^1–3 Conventional psychophysical measures of visual function such as visual fields reveal the location and extent of the visual defect and allow the clinician to determine the severity or progression of the disease. However, these tests have limited ability to indicate the level of disability experienced by the patient.^3,4 It has been proposed that assessing monocular CS might be more useful in monitoring the progression of functional visual loss than testing visual fields, which is more costly and time-consuming.⁵ Subjects with glaucoma can have impaired CS despite good visual acuity.^6–8 Changes in CS of glaucomatous eyes have been detected prior to visible damage to the retinal nerve fiber layer, manifest defects on standard automated perimetry, or a decrease in visual acuity (VA).^3,9,10

Glaucoma preferentially affects the retinal ganglion cells (RGC) and their neural processes located in the retinal nerve fiber layer (RNFL) and the inner plexiform layer (IPL), which contains the RGC dendrites.^11–15 The inner retinal layers can now be measured with reasonable accuracy with spectral domain optical coherence tomography (SD-OCT). Recent advances in OCT segmentation algorithms have facilitated visualization and measurement of individual retinal layers in the macular region with SD-OCT.^16–23 There is no study, to date, that has addressed the relationship between CS and inner macular measurements in glaucomatous eyes.

The present study was carried out to test the hypothesis that structural and functional measures predict CS outcomes in eyes with glaucoma. In case such associations are strong, measurement of CS could be advocated for monitoring patients with glaucoma, especially those with advanced disease in whom structural and functional tests have limited utility.

Methods

Study sample

Patients from the Advanced Glaucoma Progression Study cohort who met specific inclusion criteria were enrolled in this study. The Institutional Review Board’s approval was obtained and all patients gave their written informed consent. Our study protocol was carried out in accordance with the principles of the Declaration of Helsinki and The Health Insurance Portability and Accountability Act.

All patients had a comprehensive eye exam including visual acuity, automated refraction, measurement of intraocular pressure (IOP), gonioscopy, slit-lamp exam, dilated fundus exam, 24-2 standard achromatic perimetry (SAP), macular SD-OCT imaging with Cirrus and Spectralis devices and CS measurement with the CSV-1000 device.

Glaucoma was defined as presence of glaucomatous optic nerve damage (i.e., vertical cup-to-disc ratio of >0.6, or cup to disc asymmetry >0.2, or presence of focal thinning or notching) and an associated visual field defect on standard achromatic perimetry. A visual field defect was considered to be present if both of the following criteria were met: (1) Glaucoma Hemifield Test outside normal limits; and (2) four abnormal points with p <5% on the pattern deviation plot, both confirmed at least once. These criteria have been shown to be highly specific and reasonably sensitive for detection of early glaucomatous visual field loss.²⁴ Patients were required to have a visual field mean deviation (MD) worse than −6 dB or evidence of involvement of two or more points within the central 10 degrees of the 24-2 field (p values <5% on the pattern deviation plot) confirmed at least once.

Other inclusion criteria for the study subjects were age 40–85 years, best corrected visual acuity ≥20/50 in the eligible eye, spherical refractive error ≤5 D and ≤3 D of astigmatism, and no significant retinal or neurological disease including diabetic retinopathy or age-related macular degeneration. Eligible patients could have had prior incisional or laser glaucoma surgery or cataract extraction as long as the above criteria were met.

The CSV-1000 (VectorVision, Grenville, OH) measures contrast sensitivity at 3, 6, 12, and 18 cycles/degree frequencies. The device automatically calibrates the light level to 85 cd/m². The patients were examined at a distance of 8 feet from the device’s screen. The device projects 4 double rows (rows A, B, C and D) displaying circles of decreasing contrast sensitivity at 3, 6, 12 and 18 cycles/degree, respectively. Each row consists of 17 circles, with the first circle of each row displaying the highest contrast. The remaining 16 circles are presented in 2 rows consisting of 8 pairs of circles. The patient is instructed to choose the one circle out of each pair showing the grid pattern. The last correct response for each level of contrast is defined as the contrast threshold for that spatial frequency.

The Posterior Pole Algorithm of Spectralis^® SD-OCT (Heidelberg Engineering, Germany) and the Macular Cube 200 × 200 protocol of Cirrus high-definition OCT (HD-OCT, Model 4000; software version 6.0, Carl Zeiss Meditec Inc., Dublin, CA, USA) were used for macular imaging. The SD-OCT images were performed after dilation. The xml data from both devices were exported to a personal computer and processed. All the images were reviewed by one of the investigators (SN), and images with signal strength <7 (Cirrus HD-OCT) or quality factor <15 (Spectralis SD-OCT), missing data, obvious motion artifact, or incorrect segmentation were excluded. The macular imaging algorithm of the Spectralis SD-OCT (Posterior Pole Algorithm) consists of 61 horizontal B-scans each consisting of 768 A-scans spanning a 30×25° wide area. At each position, acquisition of B-scans is repeated 9–11 times to decrease speckle noise (Automatic Real Time or ART =9–11). The data are then averaged and an 8×8 grid of thickness measurements (64 superpixels within the central 24×24°, each 3° wide) for the full macular thickness is created (see Figure 1). The Macular Cube 200 × 200 algorithm measures 40,000 axial scans (in a 6×6×2 mm cube) centered on the fovea. The ganglion cell analysis available on the Cirrus software version 6.0 (or higher) measures the combined thickness of the ganglion cell and inner plexiform layers (GC/IPL) in a 4.8 × 4.0 mm oval area with a longer horizontal axis (Figure 2). It provides GC/IPL measurements in 6 wedge-shaped sectors after excluding the central foveal region (1.2×1 mm in diameter) along with a pseudocolor scheme for the GC/IPL thickness and a deviation map.¹⁴

Figure 1.

Definition of macular sectors on the 8×8 array of superpixels derived from the Posterior Pole algorithm of Spectralis SD-OCT. The measurement grid is centered on the foveal center and covers the central 24×24 degrees of the posterior pole. Each superpixel is 3 degrees wide.

Figure 2.

The Cirrus high-definition optical coherence tomography provides ganglion cell /inner plexiform layer (GC/IPL) thickness measures in a 4.8 × 4.0 mm oval area with longer horizontal axis. The oval area is divided into 6 wedge-shaped regions and thickness measurements are provided for the superotemporal, superior, superonasal, inferonasal, inferior and inferotemporal sectors. A central oval area (1.2×1.0 mm) is excluded as the GC/IPL thickness is close to zero in this region.

Standard automated perimetry (SAP) 24-2 with Swedish Interactive Thresholding Algorithm was performed for all patients. Visual fields with false negative response rate >33%, false positive response rate more than 15% and fixation loss rate >20% were excluded. As we expected that CS would be more closely related to central visual function, we defined and calculated the sectoral mean deviation (MD) for the following sectors or regions in addition to the global MD provided by the device (Figure 3): central 4 points, superior paracentral 4 points, inferior paracentral 4 points, central 10-degree visual field (includes 12 test locations in a cross-shaped pattern), superior central 10-degree hemifield, inferior central 10-degree hemifield, inferior 24-2 hemifield and superior 24-2 hemifield. The total deviation values were converted to 1/Lambert units before averaging and then were reconverted back to dB values. Similarly, we defined various regions on the macular 8×8 grids from Spectralis as follows (Figure 1): central 2×2, superior central 2×1, inferior central 2×1, central 4×4, superior central 4×2, inferior central 4×2, central 6×6, superior central 6×3 and inferior central 6×3 sectors.

Figure 3.

The 24-2 visual field sectors defined to estimate correlations with contrast sensitivity.

Statistical methods

The correlation between various global and regional structural (full macular thickness and GC/IPL) or functional outcomes (regional and global MD values) with CS was estimated with Spearman’s correlation. We also repeated the same analyses with regression analyses with adjustment for inclusion of both eyes of some patients.

All left eye data were flipped to right eye format. As the current study was considered exploratory and since many of the outcomes of interest were correlated, no correction for p values was deemed appropriate and a p value of <0.05 was considered statistically significant. The Stata software version 12.0 (Stata Corps, College Station TX) was used for all analyses.

Results

One hundred-five eyes of 65 subjects were included in the current study. Table 1 shows the demographic and clinical characteristics of the enrolled subjects. The mean (± SD) age of the patients was 67.6 (± 10.4) years. Forty-four (41.9%) eyes were pseudophakic. The median (interquartile range or IQR) logMAR best corrected visual acuity was 0 (0–0.1). The mean (± SD) contrast sensitivity for 3, 6, 12 and 18 cpd was 1.58 (±0.23), 1.77 (±0.21), 1.41 (±0.28) and 0.97 (±0.26) logarithmic units, respectively. The mean (± SD) central macular thickness for the 6×6 superpixel grid (Spectralis SD-OCT) was 286.9 (± 16.81) µm. The median (IQR) average GC/IPL thickness was 64 (58–70) µm. The median (IQR) 24-2 visual field MD was −7.6 (−11.1 to −3.0) dB. The mean (±SD) intraocular pressure (IOP) at the session where CS was measured was 12.5 (±3.4) mmHg.

Table 1.

Demographic and clinical characteristics of the study sample.

Number of eyes (patients)	105 (65)
Gender (Female/male)	36 (55.4%) / 29 (44.6%)
Age (mean ± SDa)	67.6 (±10.4)
Lens status (phakic/pseudophakic)	61 (58.1%) / 44 (41.9%)
LogMAR visual acuity (median and IQRb)	0 (0–0.1)
Contrast Sensitivity at 3 cpdc (mean ±SD)	4.65 (±1.52)
Contrast Sensitivity at 6 cpd (mean ±SD)	4.54 (±1.33)
Contrast Sensitivity at 12 cpd (mean ±SD)	4.14 (±1.73)
Contrast Sensitivity at 18 cpd (mean ±SD)	4.17 (±1.62)
24-2 visual field mean deviation (dB, median, IQR)	−7.46 (−11.08—2.97)
Average GC/IPLd thickness (µm, median, IQR)	64 (58–70)
Axial length (mm, mean ± SD)	24.4 (± 1.43)
IOPe (mmHg, mean ± SD)	12.5 (±3.38)

^aSD = standard deviation

^bIQR = interquartile range

^ccpd = cycle per degree

^dganglion cell layer and inner plexiform layer

^eIntra ocular pressure

The axial length, refractive error and IOP did not demonstrate a significant correlation with CS (p >0.713, >0.133, and >0.194 for all correlations, respectively). There was a significant correlation between logMAR visual acuity and contrast sensitivity at 6, 12, and 18 cpd (ρ =−0.306, p =0.001; ρ =−0.348, p <0.001; and ρ =−0.241, p =0.013, respectively). The correlation between CS at 3 cpd and logMAR visual acuity was not statistically significant (ρ =−0.122; p =0.215).

We estimated the correlation between various global or regional visual field summary indices and the four CS levels (Table 2). It can be observed that CS at all spatial frequencies, especially at 6 and 12 cpd, demonstrated an association with the VF indices particularly with the central indices. The highest correlation was detected between CS at 6 cpd and the central 4-point MD (ρ =0.420; p <0.001) (Figure 4). Overall, the central summary indices showed the highest correlation with CS regardless of the frequency and there was no obvious preference for either superior or inferior central field indices to demonstrate a higher correlation with any CS frequency (Figure 5). Table 3 lists the correlation of various visual field indices with logMAR visual acuity. The central 10-degree MD demonstrated the highest correlation with logMAR visual acuity (ρ =−0.351, p <0.001), followed by central 4 points MD (ρ =−0.329, p =0.001) (Figure 6). We also did the same analysis with the actual sensitivity at each sector (sensitivity in apostilb unit) the results were less significant as expected and the magnitude of the correlation has decreased. Data are shown in table 3. The p values have become larger and the magnitude of the correlation has decreased.

Table 2.

Median and interquartile ranges for global and sectoral visual field measurements in the study sample (leftmost column) and correlation of various visual field indices with contrast sensitivity at four different spatial frequencies. For correlation coefficients, the top number represents the ρ and the bottom number in parentheses is the p-value.

	MDa (median and IQRb), dB	CSc at 3 cpdd	CS at 6 cpd	CS at 12 cpd	CS at 18 cpd
Global MD	−7.6 (−11.1 – −3.0)	0.227 (0.020)	0.287 (0.003)	0.285 (0.003)	0.236 (0.016
Superior 24-2 hemifield	−2.80 (−6.63 – −1.18)	0.196 (0.045)	0.266 (0.006)	0.299 (0.002)	0.197 (0.044)
Inferior 24-2 hemifield	−3.43 (−5.97 – −1.55)	0.205 (0.035)	0.273 (0.005)	0.238 (0.014)	0.197 (0.044)
Central 10-degree visual field	−3.23 (−4.94 – −1.60)	0.194 (0.047)	0.399 (<0.001)	0.371 (<0.001)	0.258 (0.008)
Superior central 10-degree hemifield	−3.60 (−7.11 – −1.40)	0.149 (0.128)	0.291 (0.003)	0.316 (0.001)	0.147 (0.134)
Inferior central 10-degree hemifield	−3.07 (−4.76 – −1.59)	0.190 (0.053)	0.350 (<0.001)	0.260 (0.007)	0.240 (0.014)
Central 4 points	−2.12 (−4.35 – −0.73)	0.211 (0.031)	0.420 (<0.001)	0.383 (<0.001)	0.228 (0.019)
Superior central 4 points	−2.89 (−6.56 – −0.89)	0.181 (0.065)	0.367 (<0.001)	0.369 (<0.001)	0.190 (0.052)
Inferior central 4 points	−1.89 (−3.56 – −0.89)	0.163 (0.096)	0.315 (0.001)	0.208 (0.033)	0.164 (0.095)

^aMD = mean deviation

^bIQR = interquartile range

^cCS = contrast sensitivity

^dcpd = cycles per degree

Figure 4.

Box plot demonstrating the relationship between the central 4-point mean deviation and contrast sensitivity at 6 cycles per degrees (ρ =0.420; p <0.001).

Figure 5.

Right, the scatter plot demonstrates the correlation between contrast sensitivity at 6 cycles per degree and the central 10-degree visual field mean deviation (ρ =0.399, p <0.001). Left, the scatter plot describes the correlation of contrast sensitivity at 6 cycles per degree and the average central 2×2 macular thickness (ρ =−0.311, p =0.001) (also see Figure 1).

Table 3.

Correlation of actual visual field sensitivity indices in apostilbs with contrast sensitivity at four different spatial frequencies. For correlation coefficients, the top number represents the ρ and the bottom number in parentheses is the p-value.

Absolute mean sensitivity	CSa at 3 cpdb	CS at 6 cpd	CS at 12 cpd	CS at 18 cpd
24-2 mean sensitivity	−0.125 (0.205)	−0.160 (0.103)	−0.141 (0.151)	−0.147 (0.134)
Superior 24-2 hemifield	−0.099 (0.317)	−0.200 (0.041)	−0.218 (0.026)	−0.141 (0.154)
Inferior 24-2 hemifield	−0.114 (0.249)	−0.135 (0.170)	−0.118 (0.0231)	−0.102 (0.299)
Central 10-degree visual field	−0.125 (0.205)	−0.200 (0.041)	−0.141 (0.151)	−0.186 (0.058)
Superior central 10-degree hemifield	−0.113 (0.249)	−0.237 (0.015)	−0.242 (0.013)	−0.124 (0.208)
Inferior central 10-degree hemifield	−0.169 (0.085)	−0.264 (0.006)	−0.213 (0.029)	−0.255 (0.009)
Central 4 points	−0.208 (0.034)	−0.349 (<0.001)	−0.275 (0.005)	−0.155 (0.116)
Superior central 4 points	−0.116 (0.238)	−0.250 (0.010)	−0.257 (0.008)	−0.119 (0.226)
Inferior central 4 points	−0.118 (0.062)	−0.254 (0.009)	−0.203 (0.038)	−0.229 (0.0119)

^aCS = contrast sensitivity

^bcpd = cycle per degree

Figure 6.

Box plots display a dose-response relationship between the central 4-point mean deviation and logMAR visual acuity (ρ =−0.329, p =0.001).

Table 5 describes the GC/IPL thickness measurements and their correlations with CS at 4 measured frequencies. There were only few statistically significant correlations between the CS and GC/IPL sectors. For the CS at 6 cpd, a significant correlation was found with the inferior temporal and inferior nasal macular sectors (ρ =0.222, p =0.023 and ρ =0.209, p =0.032, respectively). The CS at 12 cpd had a statistically significant correlation only with the inferior temporal macular GC/IPL sector (ρ =0.248; p =0.023). None of the other spatial frequencies showed a significant correlation with the GC/IPL thickness measures.

Table 5.

Ganglion cell/inner plexiform layer (GC/IPL) sector thickness measurements and their correlation with contrast sensitivity at various measured frequencies. For correlations between contrast sensitivity and GC/IPL thickness the top number represents the ρ and the bottom number in parentheses is the p value.

	Thickness (µm) (median and IQRa)	CSb at 3 cpdc	CS at 6 cpd	CS at 12 cpd	CS 18 cpd
Average GC/IPL	64 (58–70)	0.149 (0.129)	0.187 (0.056)	0.213 (0.829)	−0.012 (0.901)
Inferotemporal sector	60 (54–68)	0.145 (0.140)	0.222 (0.023)	0.248 (0.011)	0.059 (0.552)
Superotemporal sector	64 (55–71)	0.117 (0.233)	0.1756 (0.073)	−0.008 (0.937)	0.042 (0.674)
Superior sector	65 (57–72)	0.004 (0.969)	0.068 (0.490)	−0.043 (0.665)	−0.016 (0.872)
Superonasal sector	68 (59–77)	0045 (0.650)	0.123 (0.212)	−0.016 (0.875)	−0.030 (0.766)
Inferonasal sector	66 (58–72)	0.109 (0.268)	0.209 (0.032)	0.068 (0.490)	0.046 (0.643)
Inferior sector	60 (54–68)	0.0115 (0.242)	0.162 (0.099)	0.1080 (0.273)	0.047 (0.636)

^aIQR = interquartile range

^bCS = Contrast sensitivity

^ccpd = cycle per degree

Table 6 shows the correlations between full macular thickness parameters derived from the Spectralis SD-OCT and CS at various frequencies. The CS at 18 cpd did not have a significant correlation with any of the full macular thickness parameters. Other CS parameters demonstrated significant correlations with full macular thickness parameters especially with the central and inferior regions of the macula. The strongest correlation was between the CS at 6 cpd and central 2×2 full macular thickness (ρ =0.311, p =0.001) (Figure 5). The correlations still remained statistically significant after adjusting for inclusion of both eyes of some patients and lens status or IOP with regression analyses regardless of the VF units used.

Table 6.

Full thickness measurements in different macular sectors and their correlation with contrast sensitivity. For correlations between contrast sensitivity and full macular thickness the top number represents ρ and the bottom number in parentheses is the p value.

	Average (µm) (mean ± SDa)	CSb at 3 cpdc	CS at 6 cpd	CS at 12 cpd	CS 18 cpd
Inferior central macular thickness (6×3 superpixels)	283.1 (±18.9	0.208 (0.033)	0.228 (0.019)	0.214 (0.028)	0.079 (0.424)
Central macular thickness (4×4 superpixels)	304.4 (±18.0)	0.211 (0.031)	0.228 (0.020)	0.144 (0.143)	0.117 (0.235)
Superior central macular thickness (4×2 superpixels)	306.7 (±19.6)	0.180 (0.067)	0.159 (0.105)	0.065 (0.509)	0.116 (0.239)
Inferior central macular thickness (4×2 superpixels)	302.1 (±18.9)	0.194 (0.048)	0.242 (0.013)	0.187 (0.056)	0.085 (0.387)
Central macular thickness (2×2 superpixels)	319.5 (±19.3)	0.243 (0.012)	0.311 (0.001)	0.198 (0.043)	0.172 (0.079)
Superior central macular thickness (2×1 superpixels)	319.7 (±20.2)	0.216 (0.027)	0.273 (0.005)	0.178 (0.070)	0.1665 (0.090)
Inferior central macular thickness (2×1 superpixels)	319.2 (±19.9)	0.236 (0.016)	0.303 (0.002)	0.182 (0.063)	0.150 (0.127)
Central macular thickness (6×6 superpixels)	286.9 (±16.8)	0.232 (0.017)	0.244 (0.012)	0.170 (0.084)	0.117 (0.233)
Superior central macular thickness (6×3 superpixels)	290.8 (±18.8)	0.172 (0.080)	0.140 (0.154)	0.028 (0.776)	0.077 (0.435)

^aSD = standard deviation

^bCS = contrast sensitivity

^ccpd = cycle per degree

We also assessed the correlation between logMAR VA and structural measures. We did not find any statistically significant correlation between logMAR VA and any full macular thickness parameters from the Spectralis SD-OCT. LogMAR VA correlated significantly with the inferotemporal GCL/IPL sector thickness (ρ =−0.262, p =0.007).

Discussion

The aim of this study was to test the hypothesis that functional and structural measures predict contrast sensitivity in glaucoma patients. If this hypothesis were proven and the magnitude of such correlations were large enough, CS measurement could be used as an outcome measure for confirming worsening of glaucoma over time. This would be especially useful in eyes with moderate to advanced glaucoma as detection of progression in such eyes is challenging.²⁵ Some prior studies reported a decrease in CS at 3 and 6 cpd in glaucoma,^26–28 whereas a few others found a more prominent loss at 6 and 12 cpd.^29,30 Our study confirms the prominent involvement of CS at 6 cpd frequency in glaucoma patients. Contrast sensitivity at 6 cpd was most frequently involved and demonstrated the strongest associations with structural or functional measures.

Our analyses showed fair correlations between contrast sensitivity and VF loss especially in the central region. When VF measures were converted to 1/L units, the correlations became weaker and the p values became larger (i.e., less significant) although the overall trend of correlations was similar. Also, VF correlations with CS were generally larger than those for macular structural measures (highest correlation coefficient of 0.421 compared with 0.311). This was expected since both VF and CS reflect different functional aspects of the human visual system. Wilensky et al. reported an association between decreased contrast sensitivity and worse visual field MD on 24-2 tests.⁹ The correlation was stronger among patients with primary open angle glaucoma than patients with ocular hypertension. Tochel et al. explored whether the site of VF loss was associated with decreased CS when the latter was measured in 4 quadrants between 10–20 degrees from fixation.³¹ A reduction in CS was in agreement with the location of VF loss only in half of glaucoma patients. They speculated that a decrease in contrast sensitivity represents a different kind of glaucomatous damage, as it was not always related to visual field loss. As mentioned above, in our study, the central visual field sensitivity seemed to be the best predictor of CS loss, especially for spatial frequencies at 6 and 12 cpd. We have to emphasize that the association between contrast sensitivity and various global or regional VF summary indices was modest to fair at best in our study. The average central 4 points had the highest correlation with CS parameters (ρ =0.421).

There are few prior reports in the literature on the potential association of structural measures with CS in glaucoma patients. Agrawal and colleagues found that CS and BCVA were strongly related to macular thickness with Cirrus HD-OCT in patients with primary open angle glaucoma.³² Because of the redundancy of retinal ganglion cells in the human visual system, visual field damage tends to emerge when at least 25–30% retinal ganglion cells are lost.^33–36 A puzzling finding of our study was that the magnitude of the correlations between GC/IPL thickness measures and CS was not as high as the correlations of the latter with FT thickness measurements. For both macular outcomes, the strongest correlations were observed with CS at 6 cpd. The reason for the stronger correlations observed between CS and FT macular measurements is not immediately apparent but could possibly be attributed, at least partially, to the lower precision in GCIPL measurements as the segmentation of the inner GCL and outer IPL borders is more challenging compared to that of the internal limiting membrane and retinal pigment epithelium for FT parameters. Prior studies exploring utility of macular measurements actually found the GC/IPL layer thickness to perform better for discriminating glaucoma from normal eyes.^37–39 One alternative explanation might be that the outer retinal circuitry may somehow influence CS in glaucoma patients. However, existence of outer retinal damage in glaucoma patients is controversial.^40–42 Although these structural parameters are highly associated with the presence of VF defects,^43,44 based on our findings, they don’t seem to be strong predictors of CS.

Another interesting finding of our study was that the only significant observed correlations with macular GC/IPL measurements were seen in the inferior macula (inferotemporal and inferonasal sectors with CS at 6 and 12 cpd). This raises the question whether specific subsets or regions of retinal ganglion cells determine CS. Alternatively, this could be due to the fact that the inferotemporal GCIPL is the macular sector most commonly affected in early glaucoma^14,45 and therefore, it had a slightly higher range of variation. Whether this inferior localization of macular damage has any pathophysiological significance is not quite clear at this point especially since it was not observed with FT macular measurements.

There was a statistically significant correlation between logMAR VA and CS at 6, 12, and 18 cpd. It has been reported that glaucoma patients with good VA have worse contrast sensitivity compared to normal people.⁶ Kim et al. demonstrated a curvilinear correlation between BCVA and SD-OCT parameters and found that BCVA had the highest correlation with the average RNFL thickness (ρ =−0.447) compared with regional RNFL and GCC parameters (global, superior and inferior GCC thickness).⁴⁶ The VA was related mostly to central VF parameters as expected. Wilensky et al., reported that VA was correlated with the visual field MD deviation (ρ =−0.193), but they did not explore other summary indices.⁹ Visual acuity was not related to any of the full thickness macular measures; interestingly, the only significant correlation between VA and OCT measures was found with the inferotemporal sectoral thickness.

The results of our study should be interpreted in the light of its potential shortcomings. Our findings likely represent a best-case scenario for the association between CS and structural and functional measures as our study sample was specifically chosen to have at least very early evidence of central field loss and macular damage. However, our patient sample consisted of a group of eyes where severity had a wide range; while the median MD was −7.5, the IQR ranged from −11.1 to −3.0 dB meaning that 25% of our patients had an MD better than −3 dB. Although we could not control for cataract in our study sample, the median VA was 20/20 in this study and 42% of the eyes were already pseudophakic. Our study consisted of a cross-sectional sample of eyes and therefore, inter-individual variability among study eyes may have diluted correlations between structural and functional measures with CS in glaucoma. All the included eyes were under treatment. Given reports on partial reversibility of CS loss in patients after treatment,^47–49 it is not clear how this might have affected the findings.

In summary, we found fair correlations between central VF summary indices and central full macular thickness measurements with CS, most markedly at 6 cpd. Weaker associations were observed between inferior GC/IPL thickness parameters and CS at 6 and 12 cpd. Given our findings, CS outcomes do not seem to be adequate surrogates to be used for detection of disease worsening in glaucoma eyes beyond the very early stages. However, longitudinal studies are needed to better investigate the association of changes in macular structural and functional measures with changes in CS over time in glaucoma. We expect that long-term follow-up of patients enrolled in our study will provide more definitive answers in this regard.

Table 4.

Correlation of various visual field indices with logMAR visual acuity. For correlation coefficients, the first value represents the ρ and the second value in parentheses is the p-value.

Visual field index	logMAR VA
24-2 Visual field	−0.307 (0.001)
Superior 24-2 hemifield	−0.251 (0.010)
Inferior 24-2 hemifield	−0.268 (0.006)
Central 10-degree visual field	−0.351 (<0.001)
Superior central 10-degree hemifield	−0.295 (0.002)
Inferior central 10-degree hemifield	−0.250 (0.010)
Central 4 points	−0.329 (0.001)
Superior central 4 points	−0.265 (0.006)
Inferior central 4 points	−0.284 (0.003)