Correlation Between Reflectivity of Subretinal Fluid in OCT Images and Concentration of Intravitreal VEGF in Eyes With Diabetic Macular Edema

Abstract

Purpose.: The reflectivity of optical coherence tomographic (OCT) images has been used to evaluate retinal diseases. The purpose of our study was to determine whether a significant correlation exists between the reflectivity of the subretinal fluid (SRF) and the concentration of intravitreal cytokines in eyes with diabetic macular edema (DME).

Methods.: A retrospective comparative study was done of eyes with DME with SRF before vitrectomy. The reflectivity of the SRF was determined from the OCT images. Vitreous samples were collected during vitrectomy, and analyzed for the concentrations of VEGF, IL-6, and IL-8. To determine the factors in the SRF that could affect the reflectivity, the aqueous humor of isolated swine eyes was replaced by saline with plasma, albumin, or fibrinogen, and the reflectivity of the anterior chamber was determined by anterior segment OCT.

Results.: The average OCT reflectivity of the SRF was 3.52 arbitrary units (AU; 15 eyes; range, 0.01–20.7 AU). The average concentration of VEGF was 870.1 pg/mL, that of IL-6 was 131.7 pg/mL, and that of IL-8 was 224.1 pg/mL. The degree of OCT reflectivity was correlated significantly with the intravitreal VEGF concentration (r = 0.516, P = 0.049, Spearman's rank correlation coefficient), but not with IL-6 or IL-8. In the swine eyes, the presence of plasma, bilirubin, and fibrinogen in the anterior chamber led to significant increases in the reflectivity.

Conclusions.: The significant correlation between the reflectivity of SRF and intravitreal VEGF indicates that OCT can be used to monitor the level of VEGF in eyes with DME.

Introduction

Optical coherence tomography (OCT) has been used to detect morphologic changes of the macro- and microstructures of the retina of eyes affected by various retinal diseases. For example, in situ measurements of the retinal thickness have enabled clinicians evaluate the effect of retinal diseases on the thickness. The information obtained has led to accelerated development of new drugs to treat retinal diseases, such as the anti-VEGF antibodies. ^1,2 The diagnosis of a macular hole has become easier and more accurate with OCT, and the visual acuity after a closure of the macular hole can be predicted from the integrity of the microstructures of the outer retina seen in the OCT images. ^3,4  

Other information can be obtained from the OCT images that potentially can be valuable in evaluating the condition of the retina. Thus, Barthelmes et al. used the optical density or reflectivity of retinal spaces to determine the pathogenesis of various retinal diseases, ⁵ and Ahlers et al. used the optical density or reflectivity of the OCT images for diagnosis and prognosis in eyes with exudative macular diseases. ⁶  

Recently, Neudorfer et al. evaluated the reflectivity of the subretinal fluid (SRF) in the OCT images of eyes with different retinal diseases. ⁷ They found that the OCT reflectivity of the SRF was higher in eyes with diabetic retinopathy (DR) and age-related macular degeneration, but lower in eyes with rhegmatogenous retinal detachment. Horii et al. reported that the OCT-determined fluorescein intensity and heterogeneity in the retinal cystoid spaces was related to the degree of diabetic macular edema (DME). ⁸ These findings suggested that the reflectivity of the OCT images can be used as an indicator of the disease process and activity. 

DME has been studied extensively by OCT, and the findings have been critical for diagnosing the disease. ^9–12 It is accepted widely that the pathogenesis of DME is due to different bioactive molecules, for example, VEGF. ^10,12–14 These findings led to the successful use of anti-VEGF to treat DME. ¹⁵ At present, anti-VEGF therapy is a standard therapy for DME; however, there still are unanswered questions related to the use of anti-VEGF agents, for example, frequency, time, and duration of treatment. Thus, it would be valuable if the intraocular VEGF concentration could be determined rapidly and noninvasively. This would allow clinicians to monitor the level of the VEGF and to determine whether the therapy being used was effective. 

Thus, examining OCT images was considered to be a possible way to assess the VEGF levels because OCT can be performed easily with high repeatability and reproducibility. It also can be done noninvasively in a standard clinical setting. We hypothesized that the degree of reflectivity of the SRF in eyes with DME would be correlated significantly with the level of VEGF in the SRF. To test this hypothesis, we determined the degree of reflectivity of the SRF in the OCT images of eyes with DME just before vitrectomy. The level of VEGF was measured in the vitreous samples collected during the vitrectomy. In addition, we studied swine eyes to try to determine how the presence of different blood proteins affects the OCT reflectivity. 

Methods

Study Population

Patients with type 2 diabetes who had DR and were scheduled to undergo pars plana vitrectomy (PPV) were studied. When both eyes met the inclusion criteria, only the right eyes were analyzed. All patients had clinically significant DME, that is a thickening of the macular area or an area within 500 μm of the foveola, presence of hard exudates at or within 500 μm of the fovea with thickening of the adjacent retina, or the presence of retinal thickening one disk area or larger, any part of which was within one disk diameter from the fovea with serous retinal detachment. ¹⁶ All diagnoses were confirmed by at least 2 of the authors independently at the time of admission. 

For controls, eyes with rhegmatogenous retinal detachment that underwent vitrectomy during the same period were used. Eyes that had intraocular diseases, such as uveitis, diabetic retinopathy, glaucoma, age-related macular degeneration, and retinal vein diseases, were excluded. The measurement of OCT reflectivity, collection of vitreous samples, and analysis were performed as on the DME eyes. 

A written informed consent was obtained from all patients before treatment, and all surgeries were performed at the Kagoshima University Hospital. This study protocol was approved by the Ethics Committee of the Kagoshima University Hospital, Kagoshima, Japan, and the procedures conformed to the tenets of the Declaration of Helsinki. All of the animal researches were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

The clinical histories of all patients were obtained from their medical records. Slit-lamp biomicroscopy, funduscopy, fluorescein angiography, and gonioscopy were used to classify eyes as having active PDR in which there was perfused retinal and/or iris neovascularization. Eyes with massive vitreous hemorrhages or media opacities were excluded because clear OCT images could not be obtained. To minimize individual variations, patients with a tractional retinal detachment, or those who had undergone intraocular surgery or retinal photocoagulation within 2 months before the PPV were excluded. Eyes that had received any intraocular injection ≤3 months to the surgery also were excluded. 

OCT Scanning Protocols

OCT examinations were performed within one week of the PPV with a Topcon 3D OCT-1000 MARK II (Topcon, Tokyo, Japan). Sectional images of the retina were obtained by infrared fundus photography. For qualitative and quantitative analysis of the OCT images, 20° radial scans centered on the fovea were obtained in a clockwise manner. We used 49 raster scans to evaluate the mean retinal thickness from the internal limiting membrane (ILM) to the RPE at the fovea (radius, 500 μm) and 4 subfields (superior, nasal, inferior, and temporal) in the parafoveal areas (radius, 500–1500 μm) according to the manufacturer's protocol. Measurements of the central macular thickness (CMT), height of the SRF, and largest diameter of the SRF were done by a single examiner (HT) who was masked to the information of the eyes as described in detail in our earlier publications. ¹⁷  

Reliability of OCT Reflectivity

All OCT examinations were performed according to the analysis protocol recommended by the manufacturer of the OCT instrument. Following each examination, the best image was displayed on a computer screen and evaluated by 3 independent, masked graders (MS, SS, HO). When 2 or more graders determined that the retinal image was sufficiently clear, the image was used for the following analyses. 

Measurement of the OCT reflectivity of the SRF was done in a masked fashion with no information of the characteristics of the eyes known by the examiner. For intrarater comparisons, the measurement was done three separate times by Examiner 1 (MS), and the results were used for the analysis. For inter-rater comparisons, each image was measured by three independent masked raters (MS, SS, HO) with no information of eyes or the results of the other rater. These data were used for the statistical analyses. 

Quantification of OCT Reflectivity

The reflectivity of each area of the OCT image was measured according to a described method with some modifications. ⁸ The ImageJ (National Institutes of Health, Bethesda, MD; available in the public domain at http://rsb.info.nih.gov/ij/index.html), an open code Java-based image processing software, was used. ¹⁸ To quantify the reflectivity, the margin of the SRF was traced manually on an inverted grayscale image. The average reflectivity in the area encircled was measured using an image processing software (Photoshop; Adobe Systems, San Jose, CA). We used the reflectivity levels of the vitreous cavity and nerve fiber layer (NFL) as the standard in each image. After measuring the averaged reflectivity in each area, we defined the level in the vitreous as 0 and in NFL as 100. The reflectivity values of the SRF were calculated as an arbitrary unit (AU) by the formula:  

 

 

The measurement was performed three times and the average was used. In measuring the SRF, care was taken not to include the edge of the retina or RPE because the edge is hyperreflective (Fig. 1). 

Figure 1

 

Representative OCT images used to measure the reflectivity of the SRF. The measurement area is outlined by the yellow line. (A) A case with SRF of low OCT reflectivity. (B) A case with SRF with high OCT reflectivity.

Disease Activity by Morphologic Analysis

The activity of the DME was evaluated by the fluorescein angiograms within 1 week of the OCT examination. Fluorescein angiograms at the mid and late phases were projected onto a computer screen, and the following three points were evaluated; the presence of diffuse dye leakage, presence of neovascularization, and presence of a fibrous membrane. These were determined by three masked observers. If the classification was split, the majority was taken as the final result. The presence of each point was counted as +1, and the activity was expressed by the sum of each point ranging from +3 to 0 (total FA score). The CMT, height of SRF, and largest diameter of SRF (width of SRF) also were used as morphologic indicators of disease activity. 

Vitreous Collection

Undiluted vitreous samples (0.5–0.7 mL) were collected during PPV, immediately placed in sterile tubes, centrifuged to remove cells and debris, and stored at −80°C as described. ^19,20  

ELISAs and a multiplex cytometric bead array (CBA) assay were used to determine the vitreous levels of the bioactive molecules in the samples as described. ^12,19 The level of VEGF was quantified for each sample by a commercial human VEGF ELISA kit (R&D Systems, Minneapolis, MN). Each assay was performed in duplicate in accordance with the manufacturer's instructions, with a 50 μL vitreous sample (prediluted to 100 μL) per well for the VEGF ELISA and 100 μL (without predilution). Under these conditions, the detection limit of VEGF was 18 pg/mL. 

In our earlier studies, we found that the levels of intravitreal IL-6 and IL-8 were correlated significantly with the degree of DR. ¹² So, the concentrations of IL-6 and IL-8 were quantified using a commercial multiplex CBA human inflammation Kit (BD Biosciences Pharmingen, San Diego, CA), according to the manufacturer's protocols. The sample requirement was 50 μL, so 50 μL of undiluted vitreous samples were used for the CBA kit. The detection limits for IL-6 and IL-8 were 3.6 and 3.3 pg/mL, respectively. All concentrations less than the detection level were assigned a value of 0 for the subsequent analyses. 

Plasma Components Associated With OCT Reflectivity in Swine Eyes

The SRF is composed of various components of the blood, and the effect of different blood components on the OCT reflectivity was evaluated in swine eyes. Swine eyes were obtained from a local abattoir and were used within 6 hours after enucleation. The aqueous humor was replaced by balanced salt solution (BSS plus; Alcon Japan, Tokyo, Japan) that contained the following concentrations of proteins: 98, 49, 24.5, 12.25, and 6.13 mg/mL of albumin; 2.6, 1.3, 0.65, 0.33, and 0.16 mg/dL of bilirubin; or 6.7, 3.35, 1.68, 0.84, and 0.42 mg/mL of fibrinogen. The highest concentrations were twice the concentration of each protein in the normal blood of humans. In addition, to determine whether the OCT reflectivity is dependent upon the quantity or concentration of the protein, we used different concentrations of albumin up to the maximum solubility level 588.0 mg/mL and then the OCT reflectivity was determined. Plasma from a healthy volunteer (SS) also was used and diluted with BSS. Care was taken not to let the corneal surface dry, and all the procedures were performed at room temperature of 20 ± 2°C. 

Anterior segment OCT (AS-OCT) was used to obtain images of the cornea and anterior chamber of the swine eyes (Heidelberg Spectralis-OCT; Heidelberg Engineering, Heidelberg, Germany). ²⁰ AS-OCT is able to obtain images of the fine structures of the anterior segment, including the anterior chamber of humans and animals, such as rabbits. ^21,22 For the analysis of the OCT images, a 7-line (15° × 3°) raster scan centered on the cornea was taken with 100 frames averaged. 

We calculated the mean reflectivity of square areas enclosed within the anterior chamber, or the cornea, or an area outside of the eye. The squares were centered on a line connecting the anterior pole of the lens and the corneal vertex, and the mean reflectivity was measured with the “measure” analyses program of ImageJ. Randomly selected points in the anterior chamber also were measured. The reflectivity outside the eye was used as a reference. From these, the relative reflectivity levels were determined according to the formula below. The average of three randomly selected points in the anterior chamber was used as the OCT reflectivity for each protein. The reflectivity of the anterior chamber was calculated by the following formula:  

 

 

Statistical Analyses

The significance of the correlation between one of the indicated variables and the OCT reflectivity or VEGF concentration was determined by Spearman's rank correlation coefficient. The intraclass and inter-rater correlation coefficients were determined by a two-way mixed effects model for measurements of absolute agreement. The comparison between DME group and control group was done by the Mann-Whitney U test or Fisher's exact test. One-way ANOVA with Tukey's posttest was used to compare the OCT reflectivity at the different concentrations of plasma, albumin, bilirubin, and fibrinogen. All statistical analyses were performed with SPSS statistics 19 for Windows (SPSS Inc., IBM, Somers, NY). A P value <0.05 was considered to be statistically significant. 

Results

A total of 144 eyes underwent PPV for diabetic retinopathy and 15 eyes met the inclusion criteria. The clinical demographic findings of the cases are shown in Table 1. 

Table 1

 

Background Characteristics of the Studied Eyes

	Average ± SD	Range
Sex, M/F	11/4
Age, y	62.6 ± 8.4	48–78
Visual acuity, logMAR	0.64 ± 0.23	0.22–1.00
HbA1c, %	7.71 ± 1.66	6.5–12.1
CMT, μm	599.3 ± 154.8	417–964
SRF reflectivity	3.52 ± 4.75	0.01–20.7

The concentrations of VEGF, IL-6, and IL-8 were determined in 15 eyes from the vitreous samples collected during vitrectomy. The vitrectomy was performed within one week of the SD-OCT recordings. 

Reliability of OCT Reflectivity Results

Because the reflectivity of the SRF in the OCT images is a relatively new type of retinal examination, its repeatability was evaluated by intra- and interexaminer agreements. The results showed that our method of measuring the reflectivity was repeatable with high intraexaminer and interexaminer agreement ratios (Table 2). 

Table 2

 

Reliability of OCT Reflectivity Results

Intra-Examiner Reliability	Session 1 vs. 2	Session 2 vs. 3	Session 3 vs. 1	P*	Examiner 1 vs. 2	Examiner 2 vs. 3	Examiner 3 vs. 1	P*
ICC (95% CI)	0.994 (0.988–0.997)	0.991 (0.982–0.996)	0.990 (0.980–0.995)	<0.0001	0.988 (0.976–0.994)	0.977 (0.952–0.989)	0.974 (0.947–0.988)	<0.0001

 ICC, intraclass correlation coefficient; CI, confidence interval.

*  Two-way mixed effects model for measurements of absolute agreement.

OCT-Determined Reflectivity of SRF in Eyes With DME

All eyes had detectable OCT signals in the SRF that were higher than that of the vitreous. We calculated the OCT reflectivity from the SRF, vitreous, and RPE of all eyes. One of the 15 eyes had a mild nonuniform reflectivity of the SRF, while the other 14 eyes had a uniform reflectivity of the SRF in the OCT images. A vitreomacular traction was noted in 3 eyes. 

The correlations between OCT reflectivity of SRF and each clinical variable were calculated (Table 3). The results showed that no clinical variable was correlated significantly with the degree of OCT reflectivity of the SRF. 

Table 3

 

Correlation Between OCT Reflectivity of SRF and Clinical Variables

	Average ± SD	Correlation With OCT Reflectivity of SRF
		P	R
Sex, M/F	11/4
Age, y	62.6 ± 8.4	0.864	−0.048
Visual acuity, logMAR units	0.64 ± 0.23	0.792	−0.074
HbA1c, %	7.71 ± 1.66	0.338	−0.266
SRF height, μm	149.4	0.193	0.356
SRF width, μm	1794.3	0.24	0.324
CMT, μm	599.3 ± 154.8	0.337	0.266

 Spearman's rank correlation coefficient.

The correlations between the degree of OCT reflectivity and the intravitreal concentrations of the three cytokines were calculated (Fig. 2). The OCT reflectivity was 3.52 ± 4.75 AU (average ± SD) and ranged from 0.01 to 20.7 AU. The concentration of VEGF was 870.1 ± 909.6 pg/mL, that of IL-6 was 131.7 ± 71.6 pg/mL, and that of IL-8 was 224.1 ± 124.9 pg/mL. The intravitreal concentration of VEGF was correlated significantly with the degree of OCT reflectivity of the SRF (r = 0.516, P = 0.049, Spearman's rank correlation coefficient). The intravitreal concentrations of IL-6 and IL8 were not correlated significantly with the degree of OCT. 

Figure 2

 

Correlation of OCT reflectivity, and intravitreal concentrations of VEGF, IL-6 and IL-8. The intravitreal concentration of VEGF was correlated significantly with the degree of OCT reflectivity of the SRF (r = 0.516, P = 0.049). Neither the intravitreal concentration of IL-6 nor that of IL-8 was correlated significantly with the degree of OCT reflectivity. Spearman's rank correlation coefficient.

Comparisons With Controls

During the study period, vitreous samples were collected from 12 eyes with rhegmatogenous retinal detachment during vitrectomy. The vitreous fluid was collected and examined as controls (Table 4). The sex distribution, age, and preoperative visual acuity were not significantly different between the DME and controls. On the other hand, the VEGF level was significantly higher in eyes with DME than with a retinal detachment (P < 0.001). The IL-6 concentration in the eyes with DME was not significantly different from that of the controls (P = 0.75). However, the concentration of IL-8 was significantly higher in DME than in rhegmatogenous retinal detachment (P = 0.04). The OCT reflectivity of SRF was significantly higher in the eyes with DEM than in eyes with a retinal detachment (P < 0.0001). 

Table 4

 

Demographics of Eyes With DME and Retinal Detachment

Variables	DME	Retinal Detachment, Controls	P Value
Sex, M/F	11/4	6/6	0.20*
Age, y	62.6 ± 8.4	63.5 ± 7.3	0.76
Visual acuity, logMAR	0.64 ± 0.23	0.98 ± 0.63	0.20
VEGF, pg/mL	870.1 ± 909.6	0 ± 0	<0.001
IL-6, pg/mL	131.7 ± 71.6	84.9 ± 88.5	0.071
IL-8, pg/mL	224.1 ± 124.9	99.2 ± 99.6	0.004
SRF reflectivity	3.52 ± 4.75	−0.7 ± 1.9	0.011

*  Fishers's exact t–test; other analysis, Mann-Whitney U test.

The intravitreal concentration of VEGF was lower than the detection level in all of the control eyes, (Table 5), thus the statistical analysis could not be done. However, it is highly likely that the correlation between intravitreal concentration of VEGF and OCT reflectivity of SRF would not be significant. In the controls, there was no significant correlation between the intravitreal concentration of IL-6 and the OCT reflectivity of the SRF (r = −0.007, P = 0.983), or the concentration of IL-8 and the OCT reflectivity of the SRF (r = 0.126, P = 0.697) by Spearman's rank correlation coefficient. However, there was a significant correlation between the intravitreal concentration of IL-6 and IL-8 (r = 0.650, P = 0.022, Spearman's rank correlation coefficient). 

Table 5

 

Correlation Between Intravitreal VEGF and Clinical Variables

	Average ± SD	Correlation With Intravitreal VEGF
		P	R
Total FA score	1.47 ± 0.92	0.34	0.265
SRF height, μm	149.4 ± 61.0	0.426	0.222
SRF width, μm	1794.3 ± 612.9	0.106	0.434
CMT, μm	599.3 ± 154.8	0.337	0.204

 Spearman's rank correlation coefficient.

Activity of Diabetic Retinopathy and Intravitreal VEGF

The results showed that the reflectivity of the SRF was correlated significantly with the intravitreal level of VEGF. We then examined whether there was a significant correlation between activity of the DR, evaluated by fluorescein angiography and OCT, and the intravitreal level of VEGF. We found that none of the clinical parameters, such as the total FA score, SRF height, or SRF width was correlated significantly with the intravitreal VEGF (Table 4). 

Plasma Components Involved on OCT Reflectivity in Swine Eyes

We used swine eyes to determine the effect of different blood proteins in the aqueous on the degree of OCT reflectivity of the anterior chamber (Fig. 3). Our results showed that the OCT reflectivity increased as the concentration of the plasma increased (Fig. 4). 

Figure 3

 

OCT image of swine anterior chamber filled with plasma. OCT reflectivity increases as the concentration of plasma increases.

Figure 4

 

OCT reflectivity of anterior chamber of swine eyes filled with different concentrations of plasma. OCT reflectivity decreases with increasing dilutions of the plasma. OCT reflectivity is expressed in AU. N = 4 (see Methods). **P < 0.001, 1-way ANOVA with Tukey's post-test. Average ± SD.

To determine the effect of proteins on the degree of OCT reflectivity, the three major plasma proteins, that is albumin, bilirubin (in 0.1% dimethyl sulfoxide [DMSO]), and fibrinogen, were diluted in balanced salt solution (BSS) and injected into the anterior chamber of swine eyes. The presence of albumin did not affect the OCT reflectivity even with twice the normal concentration in blood. However, bilirubin and fibrinogen increased the OCT reflectivity, with the increase dose-dependent (Fig. 5). 

Figure 5

 

OCT reflectivity of anterior chamber of swine eyes filled with the indicated protein. Fibrinogen and bilirubin increased the OCT reflectivity as the concentration of the protein increased. The presence of albumin did not increase the OCT reflectivity significantly within the examined range from 1/8× to 2× of the concentration in normal plasma. *P < 0.05, **P < 0.01, 1-way ANOVA with Tukey's posttest. Average ± SD.

To determine whether the low OCT reflectivity of albumin was due to its low concentration, the OCT reflectivity of BSS with higher concentrations of albumin was examined. The results showed that the OCT reflectivity remained low even at the highest soluble concentration of 588.0 mg/mL (Supplementary Fig. S1). 

Discussion

We hypothesized that the OCT reflectivity of the SRF is an indicator of the intraocular concentration of VEGF, because VEGF induces vascular hyperpermeability and extravasation of blood proteins into the tissues. To test this, we measured the degree of OCT reflectivity of the SRF in eyes with DME, because SRF is optically homogeneous in comparison with intraretinal cysts. ⁸ The results showed that the degree of OCT reflectivity of the SRF was correlated significantly with the intravitreal concentration of VEGF in eyes with DME. However, we also found that the activity of DME, determined by conventional OCT morphologic parameters or by fluorescein angiography, was not correlated significantly with the intraocular level of VEGF. 

Barthelmes et al. used the optical density of retinal cysts to determine the pathogenesis of several retinal diseases. ⁵ Using a similar method, Ahlers et al. found that the optical density of the SRF was correlated significantly with the visual acuity of eyes with neovascular age-related macular degeneration. ⁶ They found that the optical density of the SRF was the only parameter that was correlated significantly with the presence of neovascular age-related macular degeneration, but not for central serous chorioretinopathy. Neudorfer et al. reported that the OCT reflectivity in eyes with DR was higher than that in eyes with a retinal detachment, ⁷ which is consistent with our results. Furthermore, conventional morphometric analysis of the OCT images has limited predictive value of the visual function. ⁶ However, to the best of our knowledge, there is no report comparing the OCT reflectivity and intraocular cytokines. 

VEGF is a key player in the vascular hyperpermeability present in eyes with DR ^10,13,23 There are no anatomic barriers blocking the movement of fluids within the retina, although the external limiting membrane has narrow channels. ²⁴ A breakdown of the blood–retinal barrier causes extravasation of lipids and proteins into the retina, but they cannot pass through the external limiting membrane, thus they accumulate anterior to the external limiting membrane. ²⁴ When the external limiting membrane is damaged, the fluid that has accumulated in the outer retina moves into the subretinal space, leading to serous retinal detachment. ²³ Of importance is that the disruption of the external limiting membrane alone is not sufficient to cause a serous retinal detachment, but the leakage of fluid and proteins from the impaired blood–retinal barrier would be necessary. ²³ The vascular hyperpermeability caused by VEGF leads to an increase in the concentrations of proteins/lipids in the retina, which may result in an increase of OCT reflectivity of the SRF. In addition, VEGF can induce coagulation by upregulating tissue factors that cause the formation of fibrinogen and fibrin. ²⁵ This also may cause an increase of the OCT reflectivity. 

In eyes with a retinal detachment, comparatively large volumes of SRF samples can be collected, and the most abundant protein in the SRF was found to be albumin. ²⁶ However, it is difficult to analyze the composition of SRF of DR because of its small volume and lack of a sampling method. ⁷ To simulate the SRF, we developed a swine anterior chamber model. We injected various blood components, such as plasma, bilirubin, albumin, or fibrinogen, into the anterior chamber because the OCT hyperreflectivity in the retina is attributed, at least in part, to the blood components. ⁸ As expected, the OCT reflectivity was higher in aqueous solutions containing high concentrations of plasma. However, this was not the case for every protein. For example, the concentration of albumin did not affect the OCT reflectivity in the range of 1/8× to 2× the normal blood concentration, but the concentrations of fibrinogen and bilirubin did affect the OCT reflectivity in the same range. The OCT reflectivity of the SRF might be affected by the concentration of the protein in the fluid being examined. However, we found that even with the highest soluble concentration of albumin, there was no increase of the OCT reflectivity. Thus, it is possible that the SRF reflectivity might not be affected by the concentration of the protein. However, the SRF reflectivity might be affected by the type of protein in the SRF. 

The increase of OCT reflectivity in the eye might suggest a change in the activity of the DR, because intraocular VEGF is related closely to the severity or activity of the DR. ^27,28 However, we found that the concentration of intravitreal VEGF was not correlated significantly with the disease activity evaluated by fluorescein angiography and conventional OCT. Although we cannot explain the discrepancy between our results and those of Funatsu et al., we suggest the following. First, we used different methods to classify the activity of the DME; they used 2 stages, active or nonactive DME, while we used 4 stages of severity from 0 to +3. Each method was subjective and, thus, we cannot eliminate a potential bias. Second, our cases all were categorized as moderate-to-severe DME, so that the clinical activity was most likely different, at least by the present fluorescein angiography and conventional OCT. On the other hand, it is likely that the intraocular VEGF may vary greatly even between the eyes with similar clinical activity. Additionally, the number of eyes may have been too few to reach significance. These factors might have affected our results. 

Our findings suggested that the OCT reflectivity may become a new indicator of intraocular VEGF or an assessor of the effectiveness of the treatment. For example, the DME eyes with high OCT reflectivity would require immediate anti-VEGF treatment rather than retinal photocoagulation. If the OCT reflectivity is high in the SRF, anti-VEGF therapy might be suggested even between the sessions of retinal photocoagulation. Additionally, it might be possible that the eyes with high OCT reflectivity tend to progress to neovascular glaucoma. These issues should be investigated to take advantage of the present findings. 

There are several limitations in our study. Because this was a retrospective study, sampling bias cannot be eliminated. In addition, the number of eyes was small, and it may have affected the strength of the statistical analyses. For example, the P value for the correlation between IL-8 and OCT reflectivity was 0.072. This might have reached a significant level if a larger number of patients had been studied. Because this was a pilot study, this was not done. Power analysis was not possible from the present data. However, the present data would be useful for power analysis in future studies with more subjects. The number of eyes was not large because the use of vitrectomy for DME is decreasing ²⁹ and, thus, the collection of vitreal samples is more difficult. Our method to calculate the OCT reflectivity of the SRF had good intraexaminer and interexaminer agreement, but the intermachine agreement is not known. Because our Topcon 3D-OCT 1000 was not equipped with an anterior segment analyzing system, we used Spectralis-OCT for the animal studies. Even though, the imaging characteristics of retinal microstructure were proven to be equivalent between the present two machines, ¹⁷ this limitation should be remembered. We studied three major molecules, VEGF, IL-6, and IL-8, in eyes with DME. However, many other molecules may have a more important role in the SRF reflectivity. This possibility cannot be eliminated. Above all, this evaluation could not be done in eyes with moderate-to-severe cataract, vitreous opacity, or hemorrhage. The effect of intraretinal fluid and cystoid changes on the SRF reflectivity is not known. These limitations should be remembered in interpreting or generalizing our results. 

In conclusion, the OCT reflectivity of SRF is correlated significantly with the intravitreal VEGF in eyes with DME. Because the OCT reflectivity is related to specific proteins, OCT studies will allow us to obtain not only quantitative morphologic information, but also qualitative information. Because it is less invasive, has high reproducibility, and is easy to use, OCT will be more useful for diagnosis and determining the effectiveness of treatment of retinal diseases. 

Supplementary Materials

pdf S1 

Acknowledgments

Supported by a grant from the Research Committee on Chorioretinal Degeneration and Optic Atrophy, Ministry of Health, Labor, and Welfare, Tokyo, Japan; and by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, and Culture of the Japanese Government. The authors alone are responsible for the content and writing of this paper. 

Disclosure: S. Sonoda, None; T. Sakamoto, None; M. Shirasawa, None; T. Yamashita, None; H. Otsuka, None; H. Terasaki, None