Purpose
To investigate relationship between macular sensitivity and retinal thickness in diabetic macular edema (DME).
Design
Prospective observational study.
Methods
settings: University-based retina practice. patients: Twenty-two eyes of 11 patients with DME. procedure: Fundus microperimetry and retinal thickness tomography were performed simultaneously using an automatic fundus perimetry/tomography system. main outcome measures: Quantification of macular sensitivity, fixation pattern, and relationship between macular sensitivity and retinal thickness.
Results
Fixation stability revealed that 21 eyes (95.4%) had stable fixation (>75% within central 2 degrees of point of fixation) and 1 eye (4.5%) had relatively unstable fixation (<75% of fixation points located within 2 degrees, >75% located within 4 degrees). Evaluation of fixation location revealed that 15 eyes (68.2%) had central (>50% of fixation points within 0.5 mm of fovea), 3 eyes (13.6%) pericentral (25% to 50% within 0.5 mm of fovea), and 4 eyes (18.2%) eccentric (<25% of fixation points within 0.5 mm of fovea) fixation location. Macular sensitivity increased by an average of 0.03 decibel (dB) (95% confidence interval [CI]: 0.00, 0.06) per 1-micron (μm) increase in retinal thickness for thickness values ≤280 μm measured with the OPKO/OTI spectral-domain OCT. The macular sensitivity decreased by an average 0.05 dB (95% CI: −0.08, −0.02) per 1-μm increase in thickness for thickness values >280 μm.
Conclusions
In this pilot study, the majority of eyes with DME had stable, central fixation. Macular sensitivity varied depending on the thickness of the retina. Additional studies are needed to determine the role of microperimetry in eyes with DME.
Diabetic macular edema (DME) is a major health problem and is the leading cause of visual impairment in developed countries among the working-age population.
Despite significant progress in the techniques to image the retina and connect the in vivo histologic diagnosis with the clinical diagnosis, physicians as well as patients still experience discrepancy between the microanatomy and visual function. Visual acuity is usually considered the gold standard for determination of damage attributable to DME; however, visual acuity may be inadequate to describe the functional impairment caused by DME.
Microperimetry has made it possible to evaluate macular sensitivity in areas of macular edema. The deterioration of macular sensitivity in patients with DME has been reported in previous studies. In the present study, we investigate the relationship between macular sensitivity, retinal thickness, and associated factors in DME patients, quantified with an automatic fundus perimetry/tomography system (scanning laser ophthalmoscope/spectral-domain optical coherence tomography [SLO/OCT]; OPKO/OTI, Toronto, Ontario, Canada).
Methods
Adult patients with DME who were evaluated at the Wilmer Eye Institute, Johns Hopkins University (Baltimore, Maryland, USA) were eligible for participation. The diagnosis of DME was made by 2 retina specialists (D.V.D. and Q.D.N.) using contact lens biomicroscopy and confirmed with optical coherence tomography (OCT). Additional study assessments included Early Treatment Diabetic Retinopathy Study (ETDRS) best-corrected visual acuity (BCVA) and fluorescein angiography. The level of Hb A1C was obtained from subjects’ medical records.
OCT imaging was performed with the spectral-domain OCT module integrated in the OPKO/OTI device. The Spectral OCT/SLO module of the device generated 3-dimensional (3D) retinal maps. It captured 28 000 A-scans per second, which enabled the acquisition of up to 128 longitudinal OCT scans in 2 seconds over a 5-mm area in the macula. Retinal thickness was defined as the distance between the retinal nerve fiber layer and the hyporeflective line above the retinal pigment epithelium, and this distance was measured automatically by the OPKO software algorithm.
A specific microperimetry circular test pattern, the POLAR 3 (28 dots: 4 central, 12 mid, and 12 outer rings), was applied to all patients. The following features were incorporated in the POLAR 3 pattern: Goldmann III stimulus size, 200-ms stimulus duration, and a 1000-ms interval between stimuli presentation. The POLAR 3 pattern covers an area of 12 degrees around the center of the fovea, and consists of 4 central dots within 4 degrees, 12 mid-periphery dots within 8 degrees, and 12 outer dots within 12 degrees from the center of the fovea.
Fundus localization based on retinal vessel alignment was automatically tracked by the Spectral OCT/SLO system. The images from retinal topography and microperimetry were aligned and 3D overlay images were created. For each of the 28 loci of the POLAR 3 test pattern, retinal thickness values were obtained. The paired data of microperimetry threshold and corresponding retinal thickness measurements were used to evaluate the relationship between retinal thickness and sensitivity.
The fixation pattern was evaluated as fixation stability and fixation location. Fixation stability was classified into 3 categories: stable, relatively unstable, or unstable. If >75% of fixation points were located within a 2-degree-diameter circle, regardless of their position in relation to the foveal center, the fixation was classified as stable . If <75% of fixation points were located within a 2-degree circle, but >75% of fixation points were located within a 4-degree circle, fixation was classified as relatively unstable . If <75% of fixation points were located within a 4-degree circle, fixation was classified as unstable . Fixation location was documented in 3 categories: central, pericentral, and eccentric. If >50% of fixation points were within 0.5 mm of the foveal center, fixation was classified as central . If 25% to 50% of the fixation points were within 0.5 mm of the foveal center, fixation was classified as pericentral . If <25% of fixation points were within 0.5 mm of the foveal center, fixation was classified as eccentric .
Statistical Analysis
Exploratory data analyses included graphical (histograms, tables, side-by-side box plots, and scatterplots) and statistical summaries (means, standard deviations, quantiles, and correlation coefficients) for 1) the distribution of each study variable, 2) the relationship between macular sensitivity and retinal thickness and potential confounding variables, and 3) the relationship between retinal thickness and potential confounding variables.
The potential confounding variables included level of Hb A1C, BCVA, fixation stability, fixation location, age, and gender. Age was treated as a categorical variable with levels: age <50, ≥50 to <60, ≥60 to <70, and ≥70 years.
Linear mixed models were used to estimate the mean macular sensitivity as a function of retinal thickness adjusting for the potential confounding variables. The relationship between the mean macular sensitivity and retinal thickness was estimated separately for retinal thickness ≤280 μm and >280 μm; the cut point of 280 μm was selected based on the exploratory data analysis. Presumably, the very low retinal thicknesses represent atrophy or loss of neural retina while increase in retinal thickness often is a sign of retinal edema, which might lead to retinal dysfunction, especially if chronicity occurs.
There were several potential sources of correlation within the data: measurements correlated within subjects, visits, eyes, and circular test pattern (central, mid, and outer rings around the foveal center). Exploratory analysis of the potential sources of correlation was performed by fitting the mean model described above, assuming the data were independent, and examining the residuals of that model for the various sources of correlation. The linear mixed model included random intercepts for subject, eye, and circular test pattern.
Several sensitivity analyses were performed. The linear mixed model was fit allowing the cut point to vary from 270 to 300 μm. In addition, an analysis was performed to evaluate the effect of potential influential observations of retinal thickness by removing the upper and lower 1% of retinal thickness values from the data.
The statistical analysis was performed using STATA version 10.1 (STATACORP, College Station, Texas, USA). Statistical significance was reported if P < .05.
Results
Twenty-two measurements from eyes of 11 eligible patients were included in this study. Among the 11 patients, 5 patients contributed both eyes (10 eyes/10 measurements) and 6 patients contributed 1 eye (6 eyes/6 measurements) at the initial visit. Four of the 11 patients had follow-up visits and repeated measurements, which added 6 more measurements to the total sample size.
All the included eyes in the study had DME at the time of microperimetry measurement. Table 1 describes the characteristics of the subjects included in the study. The age of the patients ranged from 35.1 to 88.2 years (median: 66.3 years) and 36.4% (4/11) were women. Hb A1C ranged from 6.3% to 11.1% (median: 7.2%). The BCVA (Snellen equivalent) at the time of microperimetry measurement ranged from 20/25 to 20/250 (median: 20/50). Of the total 22 examined eyes, 21 eyes (95.4%) had stable fixation and 1 eye (4.5%) had relatively unstable fixation. The evaluation of fixation location revealed that 15 eyes (68.2%) had central, 3 eyes (13.6%) peri-central, and 4 eyes (18.2%) eccentric fixation.
| Patient ID | Age at Time of Visit | Gender | Visit Number | Laterality a | Hb A1C | BCVA | Fixation Stability b | Fixation Location c |
|---|---|---|---|---|---|---|---|---|
| 1 | 67.5 | M | 1 | L | 8.7 | 20/40 | S | P |
| 2 | L | 8.6 | 20/63 | S | C | |||
| 2 | 66.3 | F | 1 | R | 7.2 | 20/50 | S | C |
| 1 | L | 7.2 | 20/63 | S | C | |||
| 3 | 64.8 | F | 1 | R | 11.1 | 20/80 | S | C |
| 1 | L | 11.1 | 20/25 | S | C | |||
| 2 | R | 11.1 | 20/80 | S | C | |||
| 2 | L | 11.1 | 20/40 | S | P | |||
| 4 | 88.2 | M | 1 | R | 6.5 | 20/32 | S | C |
| 5 | 35.1 | M | 1 | R | 8.2 | 20/32 | S | C |
| 6 | 72.5 | M | 1 | R | 6.6 | 20/100 | S | E |
| 7 | 64.7 | F | 1 | R | 9.2 | 20/125 | R | E |
| 2 | R | 9.2 | 20/80 | S | E | |||
| 8 | 87.3 | M | 1 | R | 6.3 | 20/40 | S | C |
| 1 | L | 6.3 | 20/200 | S | E | |||
| 2 | R | 6.3 | 20/50 | S | C | |||
| 2 | L | 6.3 | 20/250 | S | C | |||
| 9 | 71.8 | F | 1 | R | 7.2 | 20/60 | S | C |
| 1 | L | 7.2 | 20/40 | S | C | |||
| 10 | 61.3 | M | 1 | R | 6.7 | 20/40 | S | C |
| 11 | 59.5 | F | 1 | R | 7.2 | 20/63 | S | P |
| 1 | L | 7.2 | 20/25 | S | C |
a Laterality: R = right; L = left.
b Fixation stability: S = stable; R = relatively unstable; U = unstable.
c Fixation location: C = central; P = pericentral; E = eccentric.
Macular sensitivity and corresponding retinal thickness were measured for 28 loci located on 1 of the 3 circular patterns of each of the 22 eyes, with a total number of 616 pair measurements of macular sensitivity and retinal thickness. The mean retinal thickness was 369.1 microns (μm) (standard deviation [SD]: 78.9 μm) and the mean macular sensitivity was 7.5 decibels (dB) (SD: 4.9 dB). We reported the mean macular sensitivity among different groups of retinal thickness measurements. The ranges are defined based on the available normal data as well as SD of the mean. The mean macular sensitivity was 5.0 dB in areas with retinal thickness of less than 200 μm, 4.7 dB in areas of equal to or more than 200 μm but less than 240 μm, 7.2 dB in areas equal to or more than 240 μm and less than 280 μm, 9.4 dB in areas equal to or more than 280 μm and less than 320 μm, 8.3 dB in areas equal to or more than 320 μm and less than 360 μm, and 6.4dB in areas of 360 μm and more.
The Figure displays the mean macular sensitivity as a function of retinal thickness estimated by a lowess smooth function (thick gray line). At a thickness of 200 μm the mean sensitivity is roughly 3.5 dB, then raises linearly to 8 dB by a thickness of 300 μm, then decreases linearly to 4 dB by a thickness of 600 μm. This figure also presents the fitted model for a possible relationship between macular sensitivity and retinal thickness (represented by the thick black line) as well as 95% confidence interval (CI) (dashed lines). Table 2 displays the estimated relationships between the macular sensitivity and retinal thickness as well as the potential confounders based on the linear mixed-effects model. After removing the effects of retinal thickness and the potential confounders, approximately 19% of the variation in macular sensitivity could be attributed to differences across patients and 16% of the variation could be attributed to differences between 2 eyes of each patient. An additional 18% of the variation could be explained by variation across macular areas on 3 rings around the foveal center in each eye. After accounting for the potential confounding variables, the macular sensitivity increased by an average of 0.03 dB (95% CI: 0.00, 0.06) per 1-μm increase in the retinal thickness for the thickness values of 280 μm or less measured with the OPKO/OTI spectral-domain OCT. The macular sensitivity decreased by an average of 0.01dB (95% CI: −0.02, −0.01) per 1-μm increase in the thickness for thickness values of more than 280 μm. The estimated change in mean macular sensitivity for retinal thickness of 280 μm or less was different from the estimated change for values greater than 280 μm ( P < .0001).
