Diabetic macular edema (DME) is a vision-threatening complication of diabetic retinopathy.1 DME is characterized by particular morphologic patterns of focal or diffuse expansion of the macular area with fluid accumulation. The mechanism of DME is not completely elucidated, but the proposed pathophysiology varies according to the optical coherence tomography (OCT) pattern.2 The treatment responses and lesion characteristics also differ according to the morphologic pattern.3–7
DME is associated with leakage from the retinal vessels and induces the formation of hard exudates (HEs) in the intraretinal space, especially within the outer plexiform layer (OPL) and outer nuclear layer (ONL) adjacent to the deep capillary plexus of the retina vascular network. Both persistent macular edema and HEs are related to a poor long-term prognosis in diabetic retinopathy.8 Therefore, quantitative evaluation of these lesions following prompt treatment is crucial for managing DME.
Several studies have evaluated the importance of detecting HEs in diabetic retinopathy during the past few years.9,10 The methodology has mainly included computerized algorithm-based detection of HEs using color fundus images. The usefulness of spectral-domain OCT (SD-OCT) for detecting HEs, however, has not been fully evaluated. Some recent reports examined the applicability of SD-OCT for detecting HEs, but the analyses were only conducted with cross-sectional images.11,12
SD-OCT is useful for precise stratification of the retinal microstructure.13 In particular, en face visualization is a novel application of SD-OCT technology that facilitates the noninvasive assessment of specific layers of the retina and choroid and allows for assessment of the topographic characteristics of lesions and quantification of their extent. However, to the best of our knowledge, only a few studies have systematically investigated the anti-vascular endothelial growth factor (VEGF) effects on HEs in DME based on en face images of SD-OCT.
In this study, we aim to evaluate en face SD-OCT findings of the outer retinal HEs in patients with DME following intravitreal ranibizumab (IVR) (Lucentis; Genentech, South San Francisco, CA) therapy according to particular type of OCT pattern.
Patients and Methods
This exploratory, post-hoc study was performed with a previous study designed as a prospective, open-label, consecutive, comparative case series of IVR in center-involved DME. Treatment protocols and clinical findings, including visual and anatomical outcomes, have been described previously.6 Informed consent was obtained from all patients prior to enrollment. Approval for the prospective clinical trial was obtained from the institutional review board of KyungHee University Hospital, and the study adhered to the tenets of the Declaration of Helsinki.
Patients with either type 1 or type 2 diabetes mellitus were included in this study. All patients had center-involved DME with a central subfield thickness (CST) greater than 300 µm and best-corrected visual acuity (BCVA) of 20/160 to 20/32 Snellen equivalents based on the Early Treatment Diabetic Retinopathy Study (ETDRS) charts. Complete inclusion and exclusion criteria are described in previous reports.6
All patients were assigned to intravitreal administration of 0.5 mg ranibizumab. A loading dose of three injections was given at monthly intervals. Additional injections were given after the loading phase if any of the retreatment criteria were met, which has also been detailed in previous reports.6
All study participants were followed-up monthly for 12 months and underwent a complete ophthalmologic examination, including measurements of BCVA using the standard 4-meter ETDRS charts, intraocular pressure, slit lamp examination, fundus photography, and SD-OCT at every visit. DME was classified according to the OCT pattern into three types: diffuse retinal thickening (DRT), cystoid macular edema (CME), and serous retinal detachment (SRD), as described previously.6,7
En face SD-OCT scans were obtained using the Macular Cube 512 × 128 scan of the Cirrus HD-OCT (Carl Zeiss Meditec, Dublin, CA) at each patient's first visit and months 1, 2, 3, 6, and 12. The scanned data were processed using embedded tools of advanced visualization. The partial thickness SD-OCT fundus images of the ONL was generated by setting the regions of interest to a 20-µm thick layer that was parallel and adjacent to the retinal pigment epithelium. The upper border of the ONL slab was manually generated at the lower border of the hyperreflective OPL (Figures 1A and 1B).
Cross-sectional and en face optical coherence tomography images of diabetic macular edema with hard exudates (HEs). (A) Selection of outer nuclear layer (ONL) slab is indicated (green dot lines) in cross-sectional image. (B) En face ONL slab image corresponding to cross-sectional image. (C) Selection of area corresponding to hyperreflective HEs (red) in en face ONL slab image.
The percentile of the en face ONL slab HEs area was calculated by ImageJ software (version 1.46, National Institutes of Health, Bethesda, MD) using the threshold adjustment and area fraction measurement (Figure 1C). The boundaries of ONL HEs are automatically detected and highlighted in red after threshold adjustment. Detected ONL HEs was also automatically calculated in area fraction by embedded tools of measurement. The maximum and minimum gray value of the ONL slab was standardized with facets outside the edge of the slab, consisted with true black (value = 0) and true white (value = 255). The size of ONL slab was fixed with 902 pixels in width and 924 pixels in heights including gray value standardizing facets. The presence of HEs within the central 1-mm ETDRS subfield was also assessed using ONL slab image.
Two different physicians performed en face slab generation, followed by HEs area calculation. Inter-examiner reproducibility of area calculations after imaging processes was assessed in a blinded manner. In addition, the mean gray value of the vitreous cavity was determined to assess differences in signal strength as previously described by Oh et al.14 The interexaminer reproducibility and signal strength variation are detailed in the Results section of this study.
Statistical analyses were performed with the Statistical Package for the Social Sciences (SPSS) for Windows (Version 18.0; SPSS, Chicago, IL). A P value of less than .05 was considered to indicate a significant difference. A paired t test was used to analyze the pretreatment and posttreatment changes. The Kruskal-Wallis test was used to compare the mean parameter values between DME patterns. Spearman correlation analyses were used to verify the correlation between visual acuity and area fraction of HEs at each interval. Reproducibility was evaluated with intraclass correlation coefficient (ICC), kappa statistics, and Bland-Altman plots for area fraction measurement.
SD-OCT ONL slab images were available for all 55 enrolled patients at baseline and 49 of 55 (89.1%) patients at month 12. All 55 study eyes had detectable HEs in ONL slab images at baseline. Data for six patients were excluded from the analysis; four were lost to follow-up and two withdrew their consent. The results at the most recent visit were used to analyze the final outcome. DME type included 23 eyes with DRT (41.8%), 16 eyes with CME (29.1%), and 16 eyes with SRD (29.1%). The diabetic characteristics did not differ significantly among DME types.
Reproducibility of Measurements
The interexaminer reproducibility was assessed between two examiners in blinded manner. All examiners generated ONL slab and calculated the area fraction of HEs with baseline SD-OCT scans for the 55 patients. The ICC of area fraction measurement was 0.906 (95% confidence interval [CI], 0.869 to 0.937) and kappa statistics was 0.77 (95% CI, 0.73 to 0.80).
The signal strength variation of OCT image was assessed with mean gray value of vitreous at each visit. The mean baseline gray value was 34.07 ± 1.86 (range: 29.18 to 37.03). No significant difference was found at months 1, 2, 3, 6, or 12 (35.11 ± 1.54, P = .462; 34.56 ± 1.66, P = .567; 34.98 ± 1.94, P = .498; 35.09 ± 1.39, P = .487; 34.29 ± 1.72, P = .512, respectively).
Area Fraction of ONL HEs
The area percentile of HEs in the en face ONL slab images significantly increased following IVR treatments. The overall mean baseline area fraction was 1.84% ± 1.28%, and it was significantly increased to 2.40% ± 2.48% at month 12 (P = .018). The overall area fraction gain was greatest after the second injection, at month 2, and tended to remain unchanged after the third injection (Figure 2). Eyes with CME exhibited the greatest increase, from 2.24% ± 1.85% at baseline to 3.25% ± 1.52% at 12 months (P = .013). Eyes with DRT also exhibited a significant increase in area fraction, from 1.45% ± 1.22% at baseline to 2.24% ± 1.31% at 12 months (P = .021). The area fraction of eyes with SRD did not significantly change following the injections (P = .462). The area fraction gains during the initial 3 months were 0.83% for DRT and 1.25% for CME. The HEs in CME was greater than that in the DRT and SRD at baseline (P < .001 and P = .016, respectively) and months 1, 2, 3, 6, and 12 (P < .001, all). A representative case of DRT and CME with serial changes in the en face ONL slab HEs are shown in Figures 3 and 4, respectively.
Area fraction of outer nuclear layer hard exudates according to diabetic macular edema patterns during a period of 12 months after treatment with intravitreal ranibizumab. Bars presenting 95% confidence intervals. DRT = diffuse retinal thickening; CME = cystoid macular edema; SRD = serous retinal detachment.
Serial changes in en face outer nuclear layer (ONL) slab hard exudates (HEs) after intravitreal ranibizumab treatment: a 69-year-old male patient with diffuse retinal thickening. Color fundus images (top), en face ONL slab images indicated with HEs area in red (middle), and cross-sectional images of central subfield corresponding to the en face image (bottom). The area fraction ONL HEs in en face ONL slab images are 1.68% at month 1, 2.80% at month 2, 3.18% at month 3, 3.42% at month 6, and 3.43% at month 12.
Serial changes in en face outer nuclear layer (ONL) slab hard exudates (HEs) after intravitreal ranibizumab treatment: a 60-year-old male patient with cystoid macular edema. Color fundus images (top), en face ONL slab images indicated with HEs area in red (middle), and cross-sectional images of central subfield corresponding to the en face image (bottom). The area fraction ONL HEs in en face ONL slab images are 1.36% at month 1, 2.95% at month 2, 3.49% at month 3, 3.61% at month 6, and 3.82% at month 12.
Although the increase in area fraction occurred mainly among eyes with DRT and CME, two eyes (8.7%) with DRT and one eye (6.3%) with CME exhibited a decrease in HEs at month 12. Seven eyes (43.8%) with SRD also exhibited a decrease in HEs at month 12 compared to baseline. A representative case of SRD is shown in Figure 5.
Serial changes in en face outer nuclear layer (ONL) slab hard exudates (HEs) after intravitreal ranibizumab treatment: a 59-year-old male patient with serous retinal detachment. Color fundus images (top), en face ONL slab images indicated with HEs area in red (middle), and cross-sectional images of central subfield corresponding to the en face image (bottom). The area fraction ONL HEs in en face ONL slab images are 1.50% at month 1, 1.49% at month 2, 1.35% at month 3, 1.66% at month 6, and 1.74% at month 12.
The overall correlation between baseline visual acuity and area fraction of ONL HEs was not significant (R2 = −0.097; P = .375). There was also no significant correlation across all DME types (DRT: R2 = −0.047, P = .475; CME: R2 = −0.084, P = .312; SRD: R2 = −0.114, P = .384).
Presence of Central HEs and Visual Outcome
The presence of central 1-mm ETDRS subfield HEs was found in 23 eyes (41.8%) at baseline. Visual acuity (logMAR) of both eyes with and without central HEs at baseline improved from 0.48 ± 0.22 to 0.33 ± 0.18 and from 0.52 ± 0.28 to 0.31 ± 0.21 at month 12, respectively (Figure 6). No significant difference was obtained between two subgroups throughout the visit. The number of eyes presenting with central HEs at each time of analyses are described in Figure 7. According to DME pattern, central HEs was frequently found in DRT with 15 eyes (65.2%) at baseline, followed by SRD with five eyes (31.3%) and CME with 3 eyes (18.9%). Across all types, the number of eyes with central HEs is decreased after IVR treatment. However, the overall correlation between visual acuity and presence of central HEs was not significant (R2 = 0.106; P = .315). Across all types, no significant correlation was obtained between visual acuity and presence of central HEs (DRT: R2 = 0.117, P = .248; CME: R2 = 0.094, P = .452; SRD: R2 = 0.102, P = .397).
The mean visual acuity change from baseline by the presence or absence of outer nuclear layer hard exudates (HEs) in the central 1-mm ETDRS subfield after intravitreal ranibizumab treatment. Lines corresponds to best-corrected visual acuity in eyes without HEs (black) and in eyes with HEs (gray) in central subfield at baseline. Bars represent 95% confidence intervals.
Changes in the number of eyes with hard exudates in the central 1-mm ETDRS subfield after intravitreal ranibizumab treatment. DRT = diffuse retinal thickening; CME = cystoid macular edema; SRD = serous retinal detachment.
Quantification and interpretation of HEs in center-involved DME is essential for establishing a therapeutic strategy since the prolonged HEs in macula can evolve into subretinal fibrosis.15 Several recent attempts to detect HEs automatically in diabetic retinopathy resulted in greater improvement in the sensitivity and specificity.9,16,17 The possibilities of robust algorithms for comprehensive evaluation of HEs in DME have also been demonstrated.9,10 Moreover, the number of serial changes in the HEs in DME after intravitreal anti-VEGF therapy is increased and lasts for various durations.18,19
Jeon et al.18 reported that the number of HEs, as determined by color fundus images, significantly increased following intravitreal injections of bevacizumab (Avastin; Genentech, South San Francisco, CA) during the course of 6 months without improvement. The result of the current study also revealed the increase of HEs following IVR for 12 months. In addition, Jeon et al.18 reported that the greatest increase in HEs was detected 3 months after initiating treatment in our subjects, and during the same period, the reduction in CST was greatest. The reorganized HEs were associated with a rapid decrease in edematous space. This result also showed good concordance with our analyses.
A recent report from the Pemp et al.19 indicated that the distribution of HEs changes after anti-VEGF treatment in DME. They used a cross-sectional image of SD-OCT, which revealed that the amount of exudate increased significantly after three consecutive treatments and was related to pronounced changes in retinal thickness and macular volume. The present findings are consistent with these findings, especially in terms of the downward distribution of HEs in the retina. The area of HEs in the outer retina might increase following the IVR. Breakdown of the blood-retina barrier and changes in the osmotic gradient related with the extravasation of macromolecules play an important role in this phenomenon. The changes in hyperreflective HEs in OCT images can be more precisely detected with en face images due to its superiority in revealing the topography over the single cut of a cross-sectional image.
Recent reports from the RIDE and RISE phase 3 clinical trials revealed that monthly IVR injections resulted in a significantly greater reduction of HEs compared with sham during a 24-month period.20 However, the current study revealed an increase in HEs following IVR for 12 months, with the most rapid gain during the first 3 months. The difference might be derived from the different follow-up periods and the total number of study eyes. According to the RIDE and RISE studies, the HEs reduction was obtained gradually despite the rapid effect of IVR on macular edema. The proportion of eyes with HEs did not change until month 12. Therefore, it is possible that our current results could be different after extending the follow-up periods beyond 12 months.
Based on our study, the greatest increase in HEs was observed in eyes with CME, but the reason for this is unclear. This may be due to liquefaction necrosis of Müller cells and the breakdown of the two plexiform layers in the pathogenesis of the CME, leading to the accumulation of more inflammatory mediators and related macromolecules in the edematous cavity and subsequently, in the outer retina after reduction of their volume.2,21 In contrast, our findings indicated that the HEs in eyes with SRD did not change from the baseline despite improvement in both central subfield thickness and visual acuity. In the pathophysiology of SRD, the external limiting membrane and ellipsoid zone are damaged by a subsequent increase in oncotic pressure, which can result in fluid accumulation in the subretinal space.21,22 Therefore, the accumulation of macromolecules in the outer retina is less likely to occur due to the directly damaged photoreceptor layer, which is not frequently found in eyes with DRT and CME.
In an assessment of the correlation between visual outcomes and central HEs, the present result failed to derive significant correlations. Previous reports, including the RIDE and RISE studies, also showed no meaningful correlation between baseline central HEs and visual acuity.18–20 According to the present study, especially in eyes with DRT, the number of eyes with central HEs decreased after IVR and the area fraction of HEs increased during the same period. This may indicate that even the redistribution of HEs from the center to beyond the parafoveal area cannot influence the visual outcome in DME.
The main weakness of the present study is the post-hoc nature of the analysis, since the previous prospective study was designed to evaluate the different efficacies of IVR on visual and anatomical outcome according to DME patterns. Moreover, with its limited detection accuracy, HEs in OPL and additional HEs in ONL, which were not included in the 20-µm thick en face slab, were not accounted for in the analysis. There is also the possibility of missing HEs in subretinal space where some HEs could be accumulated. However, the strength of the present study is that current results revealed the different efficacy of IVR on HEs in each DME pattern by en face SD-OCT imaging modality. Although our study demonstrates the feasibility of en face OCT visualization on ONL HEs, evidence suggesting the increase in outer retinal HEs after IVR might be strengthened by additional data. The signal strength was determined and compensated for the reflectivity of the vitreous; however, differences in the signal strength of each image and its effect on the reflectivity could not be fully eliminated. The relatively limited number of subjects is another limitation, especially the small number of each subgroup. Other limitations include the use of manual segmentation of regions of interest and imaging processes, lack of comparison with color fundus images of HEs, and the referral bias inherent in a tertiary center study.
In conclusion, our findings revealed the ONL HEs area significantly increased in eyes with DRT and CME, and the greatest increase occurred during the initial IVR loading phase. The HEs gain remained in both DRT and CME despite repeated rescue injections, but there was no correlation with visual outcome. The presence of central HEs in the outer retina was not correlated with visual acuity after IVR. Although larger studies should be conducted for further validation, these findings suggest that en face SD-OCT images are a useful imaging modality for evaluating outer retina HEs in patients with DME.
- Klein R, Klein BE, Moss SE, Davis MD, DeMets DL. The Wisconsin epidemiologic study of diabetic retinopathy. III. Prevalence and risk of diabetic retinopathy when age at diagnosis is 30 or more years. Arch Ophthalmol. 1984;102(4):527–532. doi:10.1001/archopht.1984.01040030405011 [CrossRef]
- Bringmann A, Reichenbach A, Wiedemann P. Pathomechanisms of cystoid macular edema. Ophthalmic Res. 2004;36(5):241–249. doi:10.1159/000081203 [CrossRef]
- Wu PC, Lai CH, Chen CL, Kuo CN. Optical coherence tomographic patterns in diabetic macula edema can predict the effects of intravitreal bevacizumab injection as primary treatment. J Ocul Pharmacol Ther. 2012;28(1):59–64. doi:10.1089/jop.2011.0070 [CrossRef]
- Shimura M, Yasuda K, Yasuda M, Nakazawa T. Visual outcome after intravitreal bevacizumab depends on the optical coherence tomographic patterns of patients with diffuse diabetic macular edema. Retina. 2013;33(4):740–747. doi:10.1097/IAE.0b013e31826b6763 [CrossRef]
- Elman MJ, Ayala A, Bressler NM, et al. Intravitreal Ranibizumab for diabetic macular edema with prompt versus deferred laser treatment: 5-year randomized trial results. Ophthalmology. 2015;122(2):375–381. doi:10.1016/j.ophtha.2014.08.047 [CrossRef]
- Seo KH, Yu SY, Kim M, Kwak HW. Visual and morphologocal outcomes of intravitreal ranibizumab for diabetic macular edema based on optical coherence tomography patterns. Retina. 2016;36(3):588–595. doi:10.1097/IAE.0000000000000770 [CrossRef]
- Kim M, Lee P, Kim Y, Yu SY, Kwak HW. Effect of intravitreal bevacizumab based on optical coherence tomography patterns of diabetic macular edema. Ophthalmologica. 2011;226(3):138–144. doi:10.1159/000330045 [CrossRef]
- Klein R, Klein BE, Moss SE, Cruickshanks KJ. The Wisconsin Epidemiologic Study of Diabetic Retinopathy. XV. The long-term incidence of macular edema. Ophthalmology. 1995;102(1):7–16. doi:10.1016/S0161-6420(95)31052-4 [CrossRef]
- Zhang X, Thibault G, Decencière E, et al. Exudate detection in color retinal images for mass screening of diabetic retinopathy. Med Image Anal. 2014;18(7):1026–1043. doi:10.1016/j.media.2014.05.004 [CrossRef]
- Youssef D, Solouma NH. Accurate detection of blood vessels improves the detection of exudates in color fundus images. Comput Methods Programs Biomed. 2012;108(3):1052–1061. doi:10.1016/j.cmpb.2012.06.006 [CrossRef]
- Lammer J, Bolz M, Baumann B, et al. Detection and analysis of hard exudates by polarization-sensitive optical coherence tomography in patients with diabetic maculopathy. Invest Ophthalmol Vis Sci. 2014;55(3):1564–1571. doi:10.1167/iovs.13-13539 [CrossRef]
- Ota M, Nishijima K, Sakamoto A, et al. Optical coherence tomographic evaluation of foveal hard exudates in patients with diabetic maculopathy accompanying macular detachment. Ophthalmology. 2010;117(10):1996–2002. doi:10.1016/j.ophtha.2010.06.019 [CrossRef]
- Hee MR, Puliafito CA, Wong C, et al. Quantitative assessment of macular edema with optical coherence tomography. Arch Ophthalmol. 1995;113(8):1019–1029. doi:10.1001/archopht.1995.01100080071031 [CrossRef]
- Oh JH, Oh J, Togloom A, Kim SW, Huh K. Characteristics of cystoid spaces in type 2 idiopathic macular telangiectasia on spectral domain optical coherence tomography images. Retina. 2014;34(6):1123–1131. doi:10.1097/IAE.0000000000000038 [CrossRef]
- Fong DS, Segal PP, Myers F, Ferris FL, Hubbard LD, Davis MD. Subretinal fibrosis in diabetic macular edema. ETDRS report 23. Early Treatment Diabetic Retinopathy Study Research Group. Arch Ophthalmol. 1997;115(7):873–877. doi:10.1001/archopht.1997.01100160043006 [CrossRef]
- Wu H, Zhang X, Geng X, Dong J, Zhou G. Computer aided quantification for retinal lesions in patients with moderate and severe non-proliferative diabetic retinopathy: a retrospective cohort study. BMC Ophthalmol. 2014;14:126. doi:10.1186/1471-2415-14-126 [CrossRef]
- Jaafar HF, Nandi AK, Al-Nuaimy W. Decision support system for the detection and grading of hard exudates from color fundus photographs. J Biomed Opt. 2011;16(11):116001. doi:10.1117/1.3643719 [CrossRef]
- Jeon S, Lee WK. Effect of intravitreal bevacizumab on diabetic macular edema with hard exudates. Clin Ophthalmol. 2014;8:1479–1486.
- Pemp B, Deák G, Prager S, et al. Distribution of intraretinal exudates in diabetic macular edema during anti-vascular endothelial growth factor therapy observed by spectral domain optical coherence tomography and fundus photography. Retina. 2014;34(12):2407–2415. doi:10.1097/IAE.0000000000000250 [CrossRef]
- Domalpally A, Ip MS, Ehrlich JS. Effects of intravitreal ranibizumab on retinal hard exudate in diabetic macular edema: findings from the RIDE and RISE phase III clinical trials. Ophthalmology. 2015;122(4):779–786. doi:10.1016/j.ophtha.2014.10.028 [CrossRef]
- Zhang X, Wang N, Barile GR, Bao S, Gillies M. Diabetic retinopathy: neuron protection as a therapeutic target. Int J Biochem Cell Biol. 2013;45(7):1525–1529. doi:10.1016/j.biocel.2013.03.002 [CrossRef]
- Scholl S, Augustin A, Loewenstein A, Rizzo S, Kupperman B. General pathophysiology of macular edema. Eur J Ophthalmol. 2011;21Suppl 6:S10–19. doi:10.5301/EJO.2010.6050 [CrossRef]