Association Between Contrast Sensitivity and Central Subfield Thickness in Center-Involving Diabetic Macular Edema

Abstract

Purpose: To evaluate the association between contrast sensitivity (CS) and central subfield thickness (CST) in diabetic macular edema (DME). Methods: This prospectively recruited, cross-sectional study included eyes with DME evaluated from November 2018 to March 2021. CST was measured using spectral-domain optical coherence tomography on the same day as CS testing. Only eyes with center-involving DME (CST >305 µm for women; >320 µm for men) were included. CS was evaluated using the quantitative CS function (qCSF) test. Outcomes included visual acuity (VA) and the following qCSF metrics: area under the log CS function, contrast acuity (CA), and CS thresholds at 1 to 18 cycles per degree (cpd). Pearson correlation and mixed-effects regression analyses were performed. Results: The cohort included 52 eyes of 43 patients. Pearson correlation analysis showed a stronger association between CST and CS thresholds at 6 cpd (r = −0.422, P = 0.002) than CST and VA (r = 0.293, P = 0.035). Mixed-effects univariate and multivariate regression analyses showed significant associations between CST and CA (β = −0.001, P = .030), CS at 6 cpd (β = −0.002, P = .008), and CS at 12 cpd (β = −0.001, P = .049) but no significant associations between CST and VA. Among the visual function metrics, the effect size of CST was largest on CS at 6 cpd (β_Standardized = −0.37, P = .008). Conclusions: In patients with DME, CS may be more strongly associated with CST than VA. Including CS as an adjunct visual function outcome measure in eyes with DME may prove clinically valuable.

Introduction

Diabetic retinopathy (DR) is a leading cause of blindness worldwide.^1,2 Diabetic macular edema (DME), defined as accumulation of fluid in the outer plexiform and inner nuclear layers around the fovea, ³ can occur during any DR stage and is the most common cause of vision loss in type 2 diabetes. ⁴ Assessment of DME includes structural evaluation of the retina based on imaging and clinical examination coupled with visual function evaluation to objectively assess how well the patient can see. Optical coherence tomography (OCT) is the gold standard for structural evaluation of DME, and OCT-derived central subfield thickness (CST), defined as retinal thickening in the central 1.0 mm of the macula, has historically been the key biomarker for the diagnosis and clinical management of center-involving DME. CST is used as a cutoff for inclusion criteria and is the main structural endpoint in DME clinical trials.^5,6

Assessment of visual function traditionally relies on measurement of visual acuity (VA) by the Snellen or the Early Treatment Diabetic Retinopathy Study (ETDRS) eye chart. Despite the widespread use of these measurements, studies have shown a fairly weak correlation of CST with the baseline VA in center-involving DME^7–10 and a poor association between changes in CST and changes in VA after intravitreal injection.^5,11 Therefore, there is value in assessing whether alternative visual function metrics, provided they are still clinically feasible, could deliver stronger structure-function associations with CST in center-involving DME.

Contrast sensitivity (CS) is one visual function metric that, compared with VA, may detect more subtle changes in vision, be affected at an earlier stage of neurodegenerative disorders, and better reflect subjective visual impairment and everyday vision-related tasks.^12–16 Studies suggest that in DME, CS plays a more important role than VA in reading speed, ¹⁷ may improve with intravitreal injection or laser treatment,^18–20 and may reflect residual visual impairment despite normal VA.^21,22 However, significant limitations of currently available CS testing methods have prevented incorporating CS evaluation into routine clinical practice. ²³ Prior tests only evaluated CS at a single spatial frequency, ²⁴ had poor test–retest reliability,^23,25 or were time intensive. ²³

A promising tool to assess CS is the quantitative CS function (qCSF) test, which evaluates a wide range of contrast thresholds at multiple spatial frequencies, has good test–retest reliability, only requires 2 to 5 minutes per eye,^26–28 and has been used for various retinal conditions.^23,29–34 Although qCSF has been used in studies of DR,^31,32 to our knowledge no study has evaluated its utility in the presence of DME.

In this study, we measured CST in a cohort of eyes with center-involving DME and evaluated for structure-function associations between CST and both VA and CS measured with the qCSF test.

Methods

Study Design

This prospectively recruited, cross-sectional observational study comprised patients with a known diagnosis of DME at Massachusetts Eye and Ear (MEE) from November 2018 to March 2021. The study was performed in accordance with the tenets set forth in the Declaration of Helsinki, was compliant with the US Health Insurance Portability and Accountability Act of 1996, and received approval from the Institutional Review Board at MEE before its initiation.

Patient Enrollment

Inclusion criteria was age greater than 18 years, a clinical diagnosis of DME with intraretinal fluid on the day of the visit established by a retina specialist based on OCT imaging and clinical examination, a CST greater than 305 µm for women and greater than 320 µm for men on spectral-domain OCT (SD-OCT) imaging,^35,36 and CS testing on the same day as SD-OCT imaging. The cutoff values for CST were chosen based on previous studies evaluating the normative values for CST on SD-OCT imaging from the system used in the current study (Spectralis, Heidelberg Engineering Inc).^35,36 Exclusion criteria included VA improvement with pinhole greater than 5 letters on the Snellen chart, previous retinal detachment, active vitreous hemorrhage, a diagnosis of glaucoma, other coexisting vitreoretinal or corneal I pathology, and an inability to complete testing.

All patients had a comprehensive ophthalmic examination, including history taking, VA measurement with Snellen charts, intraocular pressure measurement, color fundus photography, SD-OCT imaging, slitlamp examination, and dilated fundus examination. Demographic characteristics and clinical characteristics, such as the stage of DR and lens status, were recorded for all eyes.

Contrast Sensitivity Testing

CS was measured using the qCSF method on the AST platform (Adaptive Sensory Technology) as previously described. ²³ In brief, the test begins by presenting 3 letters at a single spatial frequency and decreasing levels of contrast. A technician records the patient’s responses for each letter, and then an active learning algorithm selects the next combination of letters at a certain contrast and spatial frequency, which maximizes information gain and decreases testing time. After 25 personalized trials, requiring 2 to 5 minutes per eye, the software develops a 2-dimensional curve that shows CS thresholds as a function of spatial frequency, from which various CS metrics can be calculated. The area under the logarithm of CS function (AULCSF) represents CS estimates integrated across spatial frequencies ranging from 1.5 to 18 cycles per degree (cpd). ²⁶ Contrast acuity (CA) represents the smallest optotype seen at full 100% contrast, the intersection of the CS function curve with the x-axis and is reported as logarithm10(CA). CS thresholds at 1, 1.5, 3, 6, 12, and 18 cpd are measured, which represent the least amount of contrast seen at each spatial frequency.

Spectral-Domain Optical Coherence Tomography Imaging and Analysis

SD-OCT imaging was performed by trained imaging technicians. The retinal layers on each SD-OCT scan were segmented semiautomatically using the built-in software of the SD-OCT system. Retinal thickness was analyzed using the ETDRS grid, and CST was recorded by evaluating the 1.0 mm central fovea.

Statistical Analysis

The population demographics and clinical characteristics were described using traditional descriptive methods. VA was converted to logMAR notation. The Pearson correlation coefficient and mixed-effects univariate and multivariate linear regression models were used to assess for associations between CST and the visual functional metrics, which included VA and the CS metrics. Variables with a P value less than 0.05 on univariate analysis were included in the multivariate models. Standardized β coefficients were calculated by refitting the standardized data. Statistical analysis was performed using RStudio software (version 2021.09.1, build 372) and Prism software (version 8.0.0, GraphPad). All tests were 2-tailed, and statistical significance was considered when the P value was less than 0.05.

Results

Demographics and Clinical Characteristics

The cohort included 52 eyes of 43 patients with center-involving DME. Table 1 shows the demographics and clinical characteristics. Seven eyes had mild nonproliferative DR (NPDR), 16 had moderate NPDR, 7 had severe NPDR, and 22 had PDR. The median logMAR best-corrected VA was 0.28 (interquartile range [IQR], 0.1-0.42) (Snellen equivalent 20/40 [IQR, 20/25-20/50]). Table 2 shows the CS outcome measures. The median CST was 378.5 µm (IQR, 334.3-425.5).

Table 1.

Demographics and Clinical Characteristics.

Characteristic	Value
Patients, n	43
Sex, n (%)
Female	14 (32.6)
Male	29 (67.4)
Mean age (y) ± SD	59.6 ± 11.1
Diabetes type, n (%)
Type 1	8 (18.6)
Type 2	35 (81.4)
Eyes, n	52
Laterality, n (%)
Right eye	28 (53.9)
Left eye	24 (46.2)
Diabetic retinopathy stage, n (%)
Mild NPDR	7 (13.5)
Moderate NPDR	16 (30.8)
Severe NPDR	7 (13.5)
PDR	22 (42.3)
Lens status, n (%)
Pseudophakic	18 (34.62)
Normal lens	8 (15.4)
1+ lens opacification	9 (17.3)
2+ lens opacification	12 (23.1)
3+ lens opacification	5 (9.62)
4+ lens opacification	0
LogMAR BCVA
Median	0.28
IQR	0.11-0.42

Abbreviations: BCVA, best-corrected visual acuity; IQR, interquartile range; NPDR, nonproliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy.

Table 2.

Contrast Sensitivity Results Measured With the qCSF Test.

Metric	Median (IQR)
AULCSF	0.59 (0.44-0.84)
Contrast acuity	0.95 (0.79-1.17)
Contrast sensitivity
At 1 cpd	1.07 (0.96-1.23)
At 1.5 cpd	1.11 (0.93-1.25)
At 3 cpd	0.97 (0.76-1.17)
At 6 cpd	0.47 (0.042-0.89)
At 12 cpd	0.00 (0.00-0.29)
At 18 cpd	0.00 (0.00-0.00)

Abbreviations: AULCSF, area under the logarithm of contrast sensitivity function; BCVA, best-corrected visual acuity; cpd, cycles per degree; IQR, interquartile range; qCSF, quantitative contrast sensitivity function.

Pearson Correlation

Table 3 shows the Pearson Correlation results. Increased CST was significantly correlated with worse logMAR VA (r = 0.293, P = .035). Of the CS outcome measures, increased CST was significantly correlated with a decreased AULCSF (r = −0.296, P = .033), CA (r = −0.337, P = .015), and CS thresholds at 6 cpd (r = −0.422, P = .002) and 12 cpd (r = −0.290, P = .037). CS thresholds at 18 cpd and CS thresholds of 3 cpd or less were not significantly associated with CST.

Table 3.

Pearson Correlation Results Between Central Subfield Thickness and Visual Acuity and Between Central Subfield Thickness and Contrast Sensitivity Metrics.

	Pearson Correlation
Metric	r Value	P Value
LogMAR BCVA	0.293	.035 a
AULCSF	−0.296	.033 a
Contrast acuity	−0.337	.015 a
Contrast sensitivity
At 1 cpd	0.04	.76
At 1.5 cpd	0.001	.99
At 3 cpd	−0.14	.33
At 6 cpd	−0.422	.002 a
At 12 cpd	−0.290	.037 a
At 18 cpd	−0.19	.19

Abbreviations: AULCSF, area under the logarithm of contrast sensitivity function; BCVA, best-corrected visual acuity; cpd, cycles per degree.

^aStatistically significant (P < .05).

Univariate Regression Analyses

Table 4 shows the mixed-effects univariate regression analysis results. Analysis between CST and VA did not reach statistical significance (β = 0.001, P = .11). Of the CS outcome metrics, increased CST was significantly associated with decreased CA (β = −0.001, P = .020) and decreased CS thresholds at 6 cpd (β = −0.002, P = .005) and 12 cpd (β = −0.001, P = .048).

Table 4.

Mixed-Effects Univariate Regression Analysis Results Evaluating for Significant Associations Between Clinical Characteristics and Visual Function Metrics, Including VA and Multiple CS Metrics.

	Vision Function Metric			CS Across Spatial Frequencies
Clinical Feature	VA	AULCSF	CA	1 cpd	1.5 cpd	3 cpd	6 cpd	12 cpd	18 cpd
Sex
β	0.009	−0.058	−0.07	−0.03	−0.04	−0.035	−0.119	−0.034	−0.006
P	.91	.57	.37	.68	.63	.77	.40	.7	.9
Age
β	0.003	−0.007	−0.004	−0.003	−0.003	−0.008	−0.008	−0.004	−0.002
P	.33	.13	.18	.39	.37	.11	.20	.35	.38
Laterality
β	0.010	−0.082	−0.054	−0.106	−0.108	−0.107	−0.031	−0.038	−0.04
P	.84	.22	.35	.07	.09	.17	.71	.59	.29
DM type
β	0.123	−0.216	−0.106	−0.15	−0.173	−0.274	−0.21	−0.109	−0.08
P	.20	.07	.25	.08	.08	.044 a	.21	.30	.14
DR stage
β	0.021	−0.035	−0.018	0.004	0.005 × 10−1	0.004 × 10−1	−0.05	−0.075	−0.038
P	.54	.43	.59	.90	.99	.99	.40	.043 a	.046 a
Lens status
β	0.038	−0.014	−0.016	0.001	−0.003	−0.004	−0.017	−0.024	−0.002
P	.13	.66	.53	.98	.92	.91	.69	.42	.88
CST
β	0.001	−0.001	−0.001	0.001 × 10−1	−0.002 × 10−2	−0.004 × 10−1	−0.002	−0.001	−0.003 × 10−1
P	.11	.051	.020 a	.86	.95	.45	.005 a	.048	.24

Abbreviations: AULCSF, area under the logarithm of contrast sensitivity function; CA, contrast acuity; cpd, cycles per degree; CS, contrast sensitivity; CST, central subfield thickness; DM, diabetes mellitus; DR, diabetic retinopathy; VA, visual acuity.

^aStatistically significant (P < .05).

Multivariate Regression Analyses

Table 5 shows the mixed-effects multivariate regression analysis results evaluating for associations between visual function metrics and CST. When controlling for diabetes mellitus (DM) type and DR stage, the association between VA and CST did not reach statistical significance (β = 0.001, P = .15). When controlling for DM type and DR stage, CST continued to be significantly associated with CA (β = −0.001, P = .030), CS at 6 cpd (β = −0.002, P = .008), and CS at 12 cpd (β = −0.001, P = .049). Of the statistically significant associations, the standardized β coefficients were largest between CST and CS at 6 cpd (β_Standardized = −0.37, P = .008) compared with CA (β_Standardized = −0.32, P = .030) or CS at 12 cpd (β_Standardized = −0.27, P = .049).

Table 5.

Mixed-Effects Multivariate Regression Analysis Results Evaluating for Significant Associations Between CST and Vision Function Metrics, Controlling for DM Type and DR Stage.

Clinical Feature	Vision Function Metric			CS Across Spatial Frequencies
	VA	AULCSF	CA	1 cpd	1.5 cpd	3 cpd	6 cpd	12 cpd	18 cpd
DM type
β	0.105	−0.187	−0.071	−0.160	−0.178	−0.265	−0.148	−0.080	−0.073
β a	0.43	−0.58	−0.29	−0.69	−0.68	−0.75	−0.34	−0.28	−0.49
P	.27	.12	.44	.07	.08	.06	.35	.42	.16
DR stage
β	0.026	−0.042	−0.023	0.004	−0.002	−0.007	−0.059	−0.080	−0.040
β a	0.12	−0.15	−0.10	0.02	−0.01	−0.02	−0.15	−0.31	−0.30
P	.43	.31	.49	.89	.96	.89	.29	.027 b	.032 b
CST
β	0.001	−0.001	−0.001	0.002 × 10−1	0.001 × 10−1	−0.002 × 10−1	−0.002	−0.001	−0.002 × 10−1
β a	0.22	−0.24	−0.32	0.08	0.04	−0.06	−0.37	−0.27	−0.15
P	.15	.09	.030 b	.57	.78	.67	.008 b	.049 b	.28

Abbreviations: AULCSF, area under the logarithm of contrast sensitivity function; CA, contrast acuity; cpd, cycles per degree; CS, contrast sensitivity; CST, central subfield thickness; DM, diabetes mellitus; DR, diabetic retinopathy.

^aStandardized β coefficient.

^bStatistically significant (P < .05).

Conclusions

In this study we used the qCSF test to characterize CS in eyes with center-involving DME and found significant associations with CST that were stronger than the associations between CST and VA. Our findings build on previous literature of the qCSF test in various retinal disorders,^{23,29,31–34} validating the value of CS as a functional outcome metric in DME.

Pearson correlation coefficient results between VA and CST were rather weak, similar to findings in previous studies that reported correlation coefficients ranging from 0.2 to 0.3.^8–10 We found significant associations and larger correlation coefficients between CST and multiple CS outcome metrics, with the largest correlation coefficient between CST and CS thresholds at 6 cpd. In addition, multivariate regression results showed that of the visual function metrics, the effect size of CST was largest on CS at 6 cpd, as evidenced by the standardized β coefficient results. These results suggest that DME assessed by CST on SD-OCT may disproportionately affect the neural channels contributing to contrast vision at intermediate spatial frequencies. ³⁷ This is of particular interest because CS thresholds at the spatial frequency of 6 cpd are strongly linked with vision-related everyday life activities,^38,39 such as finding a door on a wall or discerning a bus from a car. ³⁸ CS at 6 cpd was also found to be the best predictor of road sign and object identification. ³⁹

On mixed-effects regression analyses, we found no significant associations between CST and VA on univariate and multivariate analyses. However, CA and CS thresholds at 6 cpd and 12 cpd were significantly associated with CST on both univariate and multivariate analyses controlling for DM type and DR stage. Of note, such changes in CS and their respective associations with CST would have been missed if CS had been measured using the traditional Pelli-Robson chart, which only measures CS thresholds at 0.5 cpd to 1 cpd. ²⁴

Few studies have evaluated for associations between CS and CST in patients with DME.^21,40 One study evaluated CS with the Pelli-Robson chart and, in line with our results, found a stronger relationship between changes in CST and CS than in CST and VA in 40 eyes with DME. ²¹ A similar study by the same group evaluated CS with 2 tests (Pelli-Robson and CamBlobs) and found that after injection with aflibercept, a reduction in CST was associated with an improvement in CS. ⁴¹

Previous studies have evaluated CS in patients with DME after intravitreal injection but did not evaluate for associations with CST.^18,42,43 One study evaluated patients with DME 24 weeks after intravitreal injection and found that although there were significant improvements in CST and VA, there were no significant improvements in CS measured with the Pelli-Robson chart. ⁴² It is possible that an improvement in CS was missed because of the limited spatial frequencies measured by the Pelli-Robson and improvement may have been measured at a higher spatial frequency. A different study found that intravitreal injections were associated with significant changes in both CS measured with the Pelli-Robson chart and CST. ⁴³ Another study used the CSV-1000 chart, which uses letter optotypes, each of which is the same size and of a low spatial frequency (2.4 cpd). ¹⁸ In a cohort of 20 eyes, the authors found that intravitreal injection temporarily improved CS and CST but did not improve VA. ¹⁸ However, unlike the current study, none of the aforementioned studies evaluated for associations between CST and CS.^18,42,43

To our knowledge, this is the first study to evaluate CS with the comprehensive, reliable, and clinically feasible qCSF test in patients with DME. We found significant associations between CS outcome measures and CST. CS is gaining ground as a valuable visual function metric in various retinal diseases; thus, it seems it has particularly useful clinical applications in DR. In patients with DME, the timing and threshold for administering intravitreal injections and laser photocoagulation may be improved by monitoring CS with a test highly sensitive to subtle changes in visual function, such as the qCSF method, as opposed to VA alone.^19,40,43,44 A subset of patients retain good VA despite the presence of DME on clinical examination; our study suggests CS could offer additional insights that would be valuable in the clinical management of these patients. ⁴⁵ This is in line with previous work by our team showing that eyes with various maculopathies can exhibit CS deficits even when they maintain VA as good as 20/30 or even 20/20⁻¹. ⁴⁶ It is also similar to previous studies that found CS deficits in patients with DM and no signs of DR, suggesting CS measurement might aid in earlier identification of visual function impairment.^31,47,48

Our results suggest that integrating CS testing into the routine retina evaluation may be valuable. In addition to assessing VA, recording CS with the qCSF test at each retina evaluation may be a better quantitative measure of the visual changes patients experience and could guide a more personalized approach to the timing of intravitreal injections.

Limitations of the current study include its cross-sectional nature, which means causality between CST and the visual function metrics cannot be inferred. Also, no intervention was studied and the sample size was moderate. Future work includes evaluation of CS with qCSF before and after intravitreal injections in patients with DME and evaluating the association between qCSF and other structural OCT parameters. These studies could help elucidate how much CS can be regained through intravitreal injections and the level of CS impairment at which initiation of intravitreal injections should occur. A direct comparison of various CS tests in patients with DME could also help determine what CS testing modality is best for clinical and research purposes.

In summary, assessment of CS in center-involving DME could be clinically valuable because it appears to be more strongly associated with CST than VA. Including CS as an adjunct visual function outcome measure in eyes with DME may lead to a more comprehensive understanding of treatment outcomes and allow for more timely or better spaced therapeutic interventions.