- Email: [email protected]
Association between Perifoveal Hyperfluorescence and Serous Retinal Detachment in Diabetic Macular Edema
Association between Perifoveal Hyperfluorescence and Serous Retinal Detachment in Diabetic Macular Edema Tomoaki Murakami, MD, Akihito Uji, MD, Ken Ogino, MD, Noriyuki Unoki, MD, Takahiro Horii, MD, Shin Yoshitake, MD, Kazuaki Nishijima, MD, Nagahisa Yoshimura, MD Objective: To study the association between the fluorescence levels on fluorescein angiography images and the characteristics on spectral-domain optical coherence tomography (SD OCT) images in diabetic macular edema (DME). Design: Retrospective, observational, cross-sectional study. Participants: One hundred sixty-seven consecutive eyes of 116 patients with diabetic retinopathy for whom FA and SD OCT were performed on the same day. Methods: Fluorescein angiography using the Heidelberg Retina Angiograph 2 and OCT images using Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany) were obtained. The leakage of fluorescein dye in each subfield of the Early Treatment Diabetic Retinopathy Study (ETDRS) grid was quantified and defined as fluorescence levels, which were compared with the retinal thickness and foveal pathomorphologic features evaluated by SD OCT. Main Outcome Measures: The relationship between fluorescence levels and the foveal pathomorphologic features on SD OCT images. Results: One hundred twelve (67%) eyes with center-involved DME had significantly higher fluorescence levels in all subfields of the ETDRS grid than 55 (33%) eyes without DME. Fluorescence levels were correlated modestly with the retinal thickness in individual subfields in eyes with center-involved DME. Thirty-seven eyes with foveal serous retinal detachment (SRD) had greater retinal thickness in all subfields and higher levels of fluorescence in most subfields, except the superior subfield of the inner ring. After adjusting for the central retinal thickness using multivariate analyses, eyes with SRD had significantly (P ¼ 0.0085) higher fluorescence levels in the nasal subfield of the inner ring and the superior, nasal, and inferior subfields of the outer ring (P ¼ 0.0117, P ¼ 0.0020, and P ¼ 0.0017, respectively). However, the fluorescence levels in any subfields of the inner or outer ring did not differ significantly between eyes with and without foveal cystoid spaces. Conclusions: The correlation between the fluorescence levels and retinal thickness suggests that the vascular hyperpermeability in the perifovea contributes to the pathogenesis of foveal SRD in DME. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Ophthalmology 2013;120:2596-2603 ª 2013 by the American Academy of Ophthalmology.
Diabetic macular edema (DME) is a leading cause of visual dysfunction in patients with diabetes.1 Diabetes induces breakdown of the blooderetinal barrier (BRB) in the retinal vasculature and concomitantly exacerbates the morphologic and functional changes in the retinal neuroglial components.2,3 Although several therapeutic strategies, including photocoagulation, antievascular endothelial growth factor (VEGF) drugs, and steroids, are used clinically to prevent vascular hyperpermeability in DME,4e8 many patients have a poor visual prognosis, and the pathogenesis of DME remains to be investigated. Diabetic retinopathy is a diabetic microangiopathy, and fluorescein angiography (FA) was introduced clinically to delineate better the retinal microvascular changes, including neovascularization, microaneurysms, intraretinal
2596
2013 by the American Academy of Ophthalmology Published by Elsevier Inc.
microvascular abnormalities, and venous beading. In addition to these lesions, hyperfluorescence often indicates the breakdown of the BRB induced by diabetes. Although fluorescein sodium is a smaller molecule, the physiologic retinal vasculature has highly integrated intercellular junctions containing tight junctions and adherens junctions and less fenestration and prevents the dye extravasation. Diabetes increases the paracellular and transcellular permeability mediated via biochemical pathways and growth factors, including vascular endothelial growth factor.9e12 These changes are reflected in the clinical findings, focal or diffuse fluorescein leakage, and fluorescein pooling.5,13,14 However, only a few studies have reported how the hyperfluorescence corresponds to retinal thickening or morphologic changes.15,16 ISSN 0161-6420/13/$ - see front matter http://dx.doi.org/10.1016/j.ophtha.2013.06.014
Murakami et al
Perifoveal Hyperfluorescence and SRD in DME
Optical coherence tomography (OCT) objectively quantifies the retinal thickness and delineates the morphologic changes in the macular area. The macular thickness is correlated modestly with visual dysfunction in DME,17 and macular pathomorphologic features, for example, serous retinal detachment (SRD), cystoid macular edema, and sponge-like retinal swelling in DME, also have been reported.18 Intriguingly, the cystoid spaces in the inner nuclear layer and outer plexiform layer on OCT images correspond to a honeycomb-like or petaloid pattern of fluorescein pooling.15,16 In addition, foveal cystoid spaces often are accompanied by an enlarged foveal avascular zone and surrounding microaneurysms.19 Compared with such changes, few reports have described the fluorescein findings in eyes with SRD in DME. Serous retinal detachment in central serous chorioretinopathy often is associated with fluorescein pooling in the subretinal spaces, whereas SRD in DME is not accompanied by fluorescein pooling in the corresponding areas.16,20,21 In addition, the optical density of the subretinal spaces on OCT images represents the differences between these diseases.22 Therefore, we questioned how vascular hyperpermeability induces development or maintenance of SRDs in DME. In the current study, we investigated the relationship between foveal pathomorphologic features and fluorescein leakage in the individual subfields of the Early Treatment Diabetic Retinopathy Study (ETDRS) grid and showed an association between foveal SRD and perifoveal hyperfluorescence in DME.
Methods Patients We retrospectively reviewed 167 eyes of 116 patients (age range, 49e85 years; mean age, 65.29.6 years; 4 eyes with mild nonproliferative diabetic retinopathy, 84 with moderate nonproliferative diabetic retinopathy, and 79 with severe nonproliferative diabetic retinopathy) who visited the Department of Ophthalmology of Kyoto University Hospital from January 2008 through March 2011. The inclusion criterion was the availability of FA and OCT images of sufficient quality that were acquired on the same day. The major exclusion criterion
Fluorescein Angiography After measuring the best-corrected visual acuity and performing fundus biomicroscopy, FA images were obtained using the Heidelberg Retina Angiograph 2 (Heidelberg Engineering, Heidelberg, Germany) in the early phase (30e90 seconds after intravenous administration of fluorescein dye) and late phase (300e600 seconds after dye administration). At image acquisition, the gain in fluorescein signals was optimized to obtain clear images at each time point, and concomitantly, the gain was not constant. Fluorescein angiography images were exported in TIFF format with 256 different levels of grayscale. Previous reports had evaluated the fluorescein findings in the late phase containing focal or diffuse fluorescein leakage and fluorescein pooling in individual subfields of the ETDRS grid,14,23,24 which encouraged us to quantify the fluorescence levels in those subfields (Fig 1). We first checked the quality of the FA images, and after excluding decentered images, saturated signals in any pixels, or shadows in the peripheral areas by aperture, further processes were applied. The signal intensities of the early-phase images were adjusted to that in the late phase according to the fluorescein intensity of major vessels in each quadrant of the image. Briefly, we quantified the signal intensity at the bifurcation of the major vessels in each quadrant of the FA images using ImageJ (National Institutes of Health, Bethesda, MD) and then calculated the averaged intensity in the early and late phases. The mean fluorescein intensity of major vessels in the early phase was almost the same as that in the late phase (ratio, 0.9670.104; P ¼ 0.1179) in the 26 eyes from which FA images were obtained without changing the excitation or gain levels, suggesting that the intensity of major vessels may be a good internal control. Considering the differences in signal intensity in the major vessels, the signal levels in all pixels of the early-phase images were adjusted to those in the late phase using the adjust brightness function in that software. The late-phase FA images and the early-phase adjusted images were exported to the image processing software (Photoshop; Adobe, San Jose, CA), and the mean signal intensity was calculated in the individual subfields of the ETDRS grid. The early-phase intensity was subtracted from the late-phase intensity to eliminate the background signals derived from the retinal vessels, choroidal tissues, autofluorescence, and so on. Considering the linear increases in fluorescein intensity (Fig 2, available at http://aaojournal.org), we then normalized the intensity by dividing the duration between the early and late phases (seconds) and the averaged signal intensity at 3 points in the major venules within 2 disc diameters from the optic disc margin. The formula used to calculate the fluorescence levels was:
Fluorescence level ðA:U:Þ in each subfield ¼
ðfluorescein intensity ½lateÞ ðadjusted fluorescein intensity ½earlyÞ ðduration of fluorescein perfusion ½secondsÞ ðfluorescein intensity in major venulesÞ
was the presence of proliferative diabetic retinopathy, ischemic maculopathy, or other chorioretinal diseases, because intravitreal hyperfluorescence from neovascular tissue or lesions in other diseases makes quantification of the exact levels of intraretinal fluorescence difficult. Eyes that underwent any intervention for macular lesions also were excluded. All research and measurements adhered to the tenets of the Declaration of Helsinki. The ethics committee of our institution approved the study protocol.
Optical Coherence Tomography Retinal sectional images of the macula were obtained using spectral domain (SD) OCT (Spectralis OCT; Heidelberg Engineering) with infrared images. Raster scans (30 25 ) were acquired, and the OCT maps with the ETDRS grid (central 1 mm, each quadrant of the inner ring [1e3 mm] or outer ring [3e6 mm]) were applied to further investigations. We also determined the macular pathomorphologic features at
2597
Ophthalmology Volume 120, Number 12, December 2013
Figure 1. Quantification of fluorescence levels in individual subfields of the Early Treatment Diabetic Retinopathy Study grid. Shown are results from a representative case with diabetic retinopathy but not center-involved diabetic macular edema (DME). Fluorescein angiography (FA) images in the (A) early and (B) late phases. C, The grayscale in the early-phase FA images is adjusted to the levels of the late phase according to the signal intensity of the major vessels, followed by quantification of the averaged signal intensity in the individual subfields. D, Quantification of the averaged signal intensity in the late phase. E, Fluorescence levels are normalized by the signal intensity in the major venules and the duration of fluorescein perfusion. F, G, Optical coherence tomography map showing no center-involved DME.
the presumed foveal center in the cross-hair sectional images with 30 .19 Center-involved DME on SD OCT was identified according to a recent publication.25 Briefly, 60 eyes of men and 52 eyes of women were diagnosed with center-involved DME and another 55 eyes were diagnosed with no DME in the current study. We first compared the fluorescence levels in eyes with center-involved DME with those in eyes with no DME and also evaluated the fluorescence levels in eyes with SRD at the fovea in the 112 eyes with center-involved DME (Table 1, available at http:// aaojournal.org).
2598
Statistical Analysis The results were expressed as the mean standard deviation. The Student’s t test was used to compare the quantitative data populations with normal distributions and equal variance. The data were analyzed using the ManneWhitney U test for populations with nonnormal distributions or unequal variance. Linear regression analysis (with Bonferroni correction, if necessary) was performed to test the statistical correlation. Multivariate analyses were used to adjust for confounding factors. P<0.05 was considered statistically significant.
Murakami et al
Perifoveal Hyperfluorescence and SRD in DME
Table 3. Association between Fluorescence Levels and Retinal Thickness in Individual Subfields in Diabetic Macular Edema Subfield
Correlation
Central Inner ring Superior Nasal Inferior Temporal Outer ring Superior Nasal Inferior Temporal
R ¼ 0.316, P ¼ 0.0006 R R R R
¼ ¼ ¼ ¼
0.213, 0.479, 0.421, 0.374,
P ¼ 0.0238 P<0.0001 P<0.0001 P<0.0001
R R R R
¼ ¼ ¼ ¼
0.342, 0.376, 0.330, 0.216,
P ¼ 0.0002 P<0.0001 P ¼ 0.0003 P ¼ 0.0225
Results Hyperfluorescence in Eyes with Center-Involved Diabetic Macular Edema When we investigated the levels of fluorescein leakage in eyes with center-involved DME and compared the results with those in eyes without center-involved DME, we found that 112 eyes with centerinvolved DME had higher fluorescence levels in all ETDRS subfields than 55 eyes without center-involved DME (Table 2, available at http://aaojournal.org). We then analyzed the association between fluorescein leakage and retinal thickening measured by OCT in the 112 eyes with center-involved DME and showed that the fluorescence level in each subfield was correlated positively with the mean retinal thickness in the corresponding subfield (Table 3). The associations were stronger in the nasal and inferior subfields in the inner ring and the nasal subfield in the outer ring of the ETDRS grid. Intriguingly, the retinal thickness in the central subfield had stronger correlation with the fluorescence levels in the nasal and inferior subfields in the inner ring (Table 4, available at http://aaojournal.org).
Higher Fluorescence Levels in Eyes with Serous Retinal Detachment Thirty-seven eyes with SRD had higher levels of fluorescence in most subfields than the 75 eyes without foveal SRD (Table 5, Fig 3). Interestingly, the differences in the subfields of the outer Table 5. Perifoveal Hyperfluorescence in Eyes with Foveal Serous Retinal Detachment in Diabetic Macular Edema
Subfield Central Inner ring Superior Nasal Inferior Temporal Outer ring Superior Nasal Inferior Temporal
No Serous Retinal Detachment (75 Eyes; 310L4)
Serous Retinal Detachment (37 Eyes; 310L4)
8.835.53
11.708.40
11.6215.50 9.405.67 9.795.71 11.116.58
16.1410.54 16.4411.70 15.1014.82 15.3911.46
0.1132 <0.0001 0.0074 0.0137
8.785.59 7.905.58 8.084.89 9.685.92
15.1510.75 14.208.96 14.689.55 14.349.69
<0.0001 <0.0001 <0.0001 0.0021
P Value 0.0332
ring were greater than those in the subfields of the inner ring (Table 5). The differences in the nasal subfield were greatest in the subfields of the inner ring, whereas we did not find differences in the superior subfield (Table 5). In addition, the retinal thickness in eyes with SRD was greater than in those without SRD in all subfields (Table 6, available at http:// aaojournal.org). After adjusting for the central retinal thickness by multivariate analysis, eyes with SRD had higher levels of fluorescence than those without SRD in the nasal subfield of the inner ring and in the superior, nasal, and inferior subfields of the outer ring (Table 7, available at http://aaojournal.org). The retinal thickness in eyes with foveal SRD was greater than that in eyes without SRD in all subfields except the temporal subfield of the inner ring (Table 8, available at http://aaojournal.org). Fluorescein pooling is a classic finding that often is delineated in the foveal cystoid spaces in DME. We compared the fluorescence levels in 43 eyes with foveal cystoid spaces alone with those in 32 eyes with neither cystoid spaces nor SRD and found the significant differences in fluorescence levels (Table 9, Fig 4) and retinal thickness of the central subfield (Table 10, available at http://aaojournal.org). However, after the adjustment for central thickness, there were no differences in fluorescence levels in the central subfield (Table 11, available at http://aaojournal.org). The retinal thickness or fluorescence levels of the inner and outer rings did not differ (Table 9; Table 10, available at http:// aaojournal.org).
Discussion Diabetic macular edema has been characterized as retinal thickening seen on ophthalmoscopy and leakage of blood components on FA images. However, clinicians often believe that there is only modest or even lower correlation between these findings, despite the clinical significance of fluorescein leakage as described previously.5,26 In the current study, we showed for the first time that eyes with foveal SRD had higher fluorescence levels in the macular areas than those without SRD. The differences in fluorescence levels also were the most significant in the perifoveal subfields, but not in the central subfield. These data suggested that vascular leakage in these subfields contributes at least partly to the pathogenesis of foveal SRD. Further studies should be planned to determine whether photocoagulation to the parafoveal or perifoveal areas with hyperfluorescence is effective for diabetic patients with foveal SRD. Clinically, center-involved DME is diagnosed mainly based on retinal thickening at the fovea where there is no retinal vasculature. The current results support the hypothesis that BRB breakdown in the parafoveal or perifoveal vasculature induces increased retinal thickness at the fovea. We first investigated the characteristics of eyes with center-involved DME and compared them with eyes without DME and found higher levels of fluorescence in all subfields of the ETDRS grid, which suggested the relevance of hyperfluorescence in DME and encouraged us to evaluate the correlation between fluorescence levels and retinal thickness measured by OCT. Interestingly, the correlation was stronger in the subfields of the inner ring than in the outer ring, except in the superior subfields. The displacement of the inner retinal layers is greater in the parafovea, and neuroglial components in the perifovea have minimal
2599
Ophthalmology Volume 120, Number 12, December 2013
Figure 3. Higher levels of fluorescence in a representative case with foveal serous retinal detachment (SRD) in diabetic macular edema. The fluorescein intensity levels in the individual subfields were quantified in a fluorescein angiography image in (A) the early phase with adjusted grayscale or (B) the late phase. C, Normalized fluorescence levels had higher levels in the nasal and inferior subfield of the outer ring. D, E, Optical coherence tomography map showing the increased retinal thickness in all subfields, especially in the nasal and inferior subfields of the perifovea. F, Serous retinal detachment at the fovea in the sectional image.
horizontal shifts.27 That is, the obliquely arrayed secondary neurons or Müller cells may have a higher capability to be reservoirs for the extravasated blood components in the Table 9. Fluorescence Levels in Eyes with Foveal Cystoid Spaces in Diabetic Macular Edema
Subfield Central Inner ring Superior Nasal Inferior Temporal Outer ring Superior Nasal Inferior Temporal
2600
Neither Foveal Cystoid Spaces Nor Serous Retinal Detachment (32 Eyes; 310L4)
Foveal Cystoid Spaces Alone (43 Eyes; 310L4)
P Value
6.914.40
10.185.88
0.0105
13.3523.22 8.335.38 8.565.53 10.256.26
10.405.82 10.155.80 10.665.73 11.726.80
0.4214 0.1729 0.1161 0.3455
8.265.10 6.835.23 7.064.87 9.715.53
9.145.94 8.645.75 8.804.83 9.666.24
0.5053 0.1678 0.1301 0.9730
outer plexiform layer in the parafoveal areas and may facilitate retinal thickening, which results in a better correlation between the fluorescence levels and retinal thickness. In contrast, because of the lower capacity in the perifoveal areas, extravasated fluids may migrate toward the parafoveal or central area. Another factor was the modest correlation between the fluorescence levels and retinal thickness in the central subfield. Optical coherence tomography identified several patterns of pathomorphologic features, specifically, SRD, cystoid macular edema, and sponge-like retinal swelling, suggesting a variety of pathogeneses of retinal thickening in the central subfield.18 Further, the central thickness was associated more significantly with fluorescence levels in the nasal and inferior subfields in the inner ring and the superior, nasal, and inferior subfields in the outer ring. These data encouraged us to investigate how vascular leakage in the individual subfields contributes to the pathomorphologic features at the fovea. Therefore, we studied the correlation between the foveal SRD and fluorescence levels and found that eyes with SRD had higher levels of fluorescence in almost all subfields than
Murakami et al
Perifoveal Hyperfluorescence and SRD in DME
Figure 4. Fluorescence levels in a representative case with foveal cystoid spaces in diabetic macular edema. The fluorescein intensity was quantified in the individual subfields of the Early Treatment Diabetic Retinopathy Study grid (A) in the early phase with adjusted grayscale and (B) in the late phase. C, Normalized fluorescence levels were higher in most subfields. D, E, Optical coherence tomography map showing that the prominent increase in the central retinal thickness. F, Retinal sectional images delineating the cystoid spaces at the fovea.
those without SRD. However, we did not observe fluorescein pooling in the subretinal space in those eyes.20 These data suggested that higher levels of vascular leakage from the retinal vasculature in the macular area contribute to development or maintenance of SRD at the fovea. Compared with smaller molecules such as water and solutes, extravasated macromolecules containing lipoproteins and albumins cannot be absorbed from the retinal pigment epithelium or retinal vasculature with intact BRB. Concomitantly, increased oncotic or osmotic pressure in the intraocular fluids may lead to development of SRD, as Marmor28 described in his clinical and experimental studies. Although the pathogenesis of SRD in DME remains ill-defined, the higher levels of vascular leakage may be a major contributor, in addition to vitreomacular traction.29,30 There were the greatest differences in the fluorescence levels in the perifovea, especially in the superior, nasal, and inferior subfields, between eyes with and without foveal SRD. The fluorescence levels in the nasal and inferior subfields of the inner ring also had greater differences than those in the central subfield (Table 5). This may indicate that the vascular hyperpermeability in those perifoveal subfields
contributes more to the pathogenesis of foveal SRD, which is to some extent consistent with the correlation between the central retinal thickness and the perifoveal fluorescence levels in the current study (Table 4, available at http:// aaojournal.org). In addition, these data may be explained by distant effects that hyperpermeability in the retinal vasculature outside of the macula often is accompanied by SRD, but not cystoid macular edema, at the fovea in several retinal vascular diseases.31,32 A recent study reported that eyes with SRD had more intact retinal vasculature around the fovea than those with foveal cystoid spaces in DME.19 Based on that, we speculated that the extravasated macromolecules from the perifovea become concentrated in the central areas, with concomitant foveal SRD development. Several publications have reported a relationship between the cystoid spaces seen on OCT and the impressive FA findings of honeycomb-like or petaloid-pattern fluorescein pooling in DME.15,16,30 However, we did not find significant differences in the fluorescence levels in any subfields in the parafovea or perifovea between eyes with and without foveal cystoid spaces compared with the significant differences in the central retinal thickness. The data suggested that
2601
Ophthalmology Volume 120, Number 12, December 2013 the presence of foveal cystoid spaces does not depend on the levels of vascular hyperpermeability in the parafovea or perifovea, which seems to contradict the clinical impression that fluorescein pooling represents pathogenesis in the cystoid spaces. This prompted us to consider a few possible explanations. In the current study, we often found the diffuse fluorescein leakage with higher intensity or fluorescein pooling in eyes without foveal cystoid spaces on OCT images. It suggests that higher levels of fluorescence did not necessarily correspond to the cystoid spaces on OCT images. In addition, it was reported recently that there were various intensities of fluorescein pooling in foveal cystoid spaces.21 Another possible explanation may be ischemia. Experimental studies have reported that microinfarctions in talc retinopathy induce cystoid spaces in the outer plexiform layer,33 and clinically, eyes with DME with foveal cystoid spaces have larger foveal avascular zones.19 Therefore, we hypothesized that the pathogenesis of the foveal cystoid spaces depends on vascular leakage as well as other mechanisms, that is, retinal ischemia. This retrospective study had several limitations. The fluorescence levels in the FA images depend partly on the fluorescence dye in the retina and intraocular humor at least in part. Although we excluded an eye with proliferative diabetic retinopathy in which higher fluorescence levels in the vitreous would be expected, we could not confirm that the measured fluorescence levels were correlated completely with those in the fluorescein dye that extravasated into the retinal tissue alone. In addition, because theoretically the zero point of fluorescence intensity in the FA images, the degree of active transport from the retinal pigment epithelium, and excretion from the kidneys cannot be determined accurately,34,35 the fluorescence levels calculated in this study were approximate but not absolute values. We did not confirm the reproducibility of the image acquisition. Finally, regarding the method of segmentation, we used the ETDRS grid to compare the fluorescence levels with the retinal thickness in each subfield, whereas previous studies used automated segmentation of the hyperfluorescent areas, which is better when considering the source or diffusion of the fluorescein dye.36 In conclusion, the current study showed a correlation between the foveal pathomorphologic features and fluorescence levels in DME, suggesting that severe vascular leakage contributes partly to the pathogenesis of foveal SRD.
References 1. 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;102:7–16. 2. Gardner TW, Antonetti DA, Barber AJ, et al. Diabetic retinopathy: more than meets the eye. Surv Ophthalmol 2002;47(suppl):S253–62. 3. Antonetti DA, Klein R, Gardner TW. Diabetic retinopathy. N Engl J Med 2012;366:1227–39. 4. Mohamed Q, Gillies MC, Wong TY. Management of diabetic retinopathy: a systematic review. JAMA 2007;298:902–16.
2602
5. Early Treatment Diabetic Retinopathy Study Research Group. Photocoagulation for diabetic macular edema. Early Treatment Diabetic Retinopathy Study report number 1. Arch Ophthalmol 1985;103:1796–806. 6. Jonas JB, Sofker A. Intraocular injection of crystalline cortisone as adjunctive treatment of diabetic macular edema. Am J Ophthalmol 2001;132:425–7. 7. Macugen Diabetic Retinopathy Study Group. A phase II randomized double-masked trial of pegaptanib, an antivascular endothelial growth factor aptamer, for diabetic macular edema. Ophthalmology 2005;112:1747–57. 8. Diabetic Retinopathy Clinical Research Network Writing Committee; Elman MJ, Bressler NM, Qin H, et al. Expanded 2-year follow-up of ranibizumab plus prompt or deferred laser or triamcinolone plus prompt laser for diabetic macular edema. Ophthalmology 2011;118: 609–14. 9. Antonetti DA, Barber AJ, Bronson SK, et al; JDRF Diabetic Retinopathy Center Group. Diabetic retinopathy: seeing beyond glucose-induced microvascular disease. Diabetes 2006;55:2401–11. 10. Murakami T, Frey T, Lin C, Antonetti DA. Protein kinase cbeta phosphorylates occludin regulating tight junction trafficking in vascular endothelial growth factor-induced permeability in vivo. Diabetes 2012;61:1573–83. 11. Dvorak AM, Feng D. The vesiculo-vacuolar organelle (VVO). A new endothelial cell permeability organelle. J Histochem Cytochem 2001;49:419–32. 12. Joussen AM, Smyth N, Niessen C. Pathophysiology of diabetic macular edema. Dev Ophthalmol 2007;39:1–12. 13. Browning DJ, Altaweel MM, Bressler NM, et al. Diabetic macular edema: what is focal and what is diffuse? Am J Ophthalmol 2008;146:649–55. 14. Early Treatment Diabetic Retinopathy Study Research Group. Classification of diabetic retinopathy from fluorescein angiograms. ETDRS report number 11. Ophthalmology 1991;98: 807–22. 15. Otani T, Kishi S. Correlation between optical coherence tomography and fluorescein angiography findings in diabetic macular edema. Ophthalmology 2007;114:104–7. 16. Bolz M, Ritter M, Schneider M, et al. A systematic correlation of angiography and high-resolution optical coherence tomography in diabetic macular edema. Ophthalmology 2009;116: 66–72. 17. Diabetic Retinopathy Clinical Research Network. Relationship between optical coherence tomography-measured central retinal thickness and visual acuity in diabetic macular edema. Ophthalmology 2007;114:525–36. 18. Otani T, Kishi S, Maruyama Y. Patterns of diabetic macular edema with optical coherence tomography. Am J Ophthalmol 1999;127:688–93. 19. Murakami T, Nishijima K, Sakamoto A, et al. Foveal cystoid spaces are associated with enlarged foveal avascular zone and microaneurysms in diabetic macular edema. Ophthalmology 2011;118:359–67. 20. Ozdemir H, Karacorlu M, Karacorlu S. Serous macular detachment in diabetic cystoid macular oedema. Acta Ophthalmol Scand 2005;83:63–6. 21. Horii T, Murakami T, Nishijima K, et al. Relationship between fluorescein pooling and optical coherence tomographic reflectivity of cystoid spaces in diabetic macular edema. Ophthalmology 2012;119:1047–55. 22. Neudorfer M, Weinberg A, Loewenstein A, Barak A. Differential optical density of subretinal spaces. Invest Ophthalmol Vis Sci 2012;53:3104–10.
Murakami et al
Perifoveal Hyperfluorescence and SRD in DME
23. Neubauer AS, Chryssafis C, Priglinger SG, et al. Topography of diabetic macular oedema compared with fluorescein angiography. Acta Ophthalmol Scand 2007;85:32–9. 24. Soliman W, Sander B, Hasler PW, Larsen M. Correlation between intraretinal changes in diabetic macular oedema seen in fluorescein angiography and optical coherence tomography. Acta Ophthalmol 2008;86:34–9. 25. Chalam KV, Bressler SB, Edwards AR, et al; Diabetic Retinopathy Clinical Research Network. Retinal thickness in people with diabetes and minimal or no diabetic retinopathy: Heidelberg Spectralis optical coherence tomography. Invest Ophthalmol Vis Sci 2012;53:8154–61. 26. Asrani S, Zeimer R, Goldberg MF, Zou S. Application of rapid scanning retinal thickness analysis in retinal diseases. Ophthalmology 1997;104:1145–51. 27. Sjostrand J, Popovic Z, Conradi N, Marshall J. Morphometric study of the displacement of retinal ganglion cells subserving cones within the human fovea. Graefes Arch Clin Exp Ophthalmol 1999;237:1014–23. 28. Marmor MF. Mechanisms of fluid accumulation in retinal edema. Doc Ophthalmol 1999;97:239–49. 29. Kaiser PK, Riemann CD, Sears JE, Lewis H. Macular traction detachment and diabetic macular edema associated with posterior hyaloidal traction. Am J Ophthalmol 2001;131:44–9.
30. Kang SW, Park CY, Ham DI. The correlation between fluorescein angiographic and optical coherence tomographic features in clinically significant diabetic macular edema. Am J Ophthalmol 2004;137:313–22. 31. Otani T, Yamaguchi Y, Kishi S. Serous macular detachment secondary to distant retinal vascular disorders. Retina 2004;24: 758–62. 32. Tsujikawa A, Sakamoto A, Ota M, et al. Retinal structural changes associated with retinal arterial macroaneurysm examined with optical coherence tomography. Retina 2009;29: 782–92. 33. Tso MO. Pathology of cystoid macular edema. Ophthalmology 1982;89:902–15. 34. Sander B, Larsen M, Moldow B, Lund-Andersen H. Diabetic macular edema: passive and active transport of fluorescein through the blood-retina barrier. Invest Ophthalmol Vis Sci 2001;42:433–8. 35. Knudsen ST, Bek T, Poulsen PL, et al. Macular edema reflects generalized vascular hyperpermeability in type 2 diabetic patients with retinopathy. Diabetes Care 2002;25:2328–34. 36. Phillips RP, Ross PG, Tyska M, et al. Detection and quantification of hyperfluorescent leakage by computer analysis of fundus fluorescein angiograms. Graefes Arch Clin Exp Ophthalmol 1991;229:329–35.
Footnotes and Financial Disclosures Originally received: February 2, 2013. Final revision: June 5, 2013. Accepted: June 5, 2013. Available online: August 13, 2013.
Manuscript no. 2013-189.
Department of Ophthalmology and Visual Sciences, Kyoto University Graduate School of Medicine, Kyoto, Japan.
Correspondence: Tomoaki Murakami, MD, Department of Ophthalmology and Visual Sciences, Kyoto University Graduate School of Medicine, 54 ShogoinKawaracho, Sakyo, Kyoto 606-8507, Japan. E-mail: mutomo@kuhp. kyoto-u.ac.jp.
Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article.
2603
Diabetic macular edema (DME) is a leading cause of visual dysfunction in patients with diabetes.1 Diabetes induces breakdown of the blooderetinal barrier (BRB) in the retinal vasculature and concomitantly exacerbates the morphologic and functional changes in the retinal neuroglial components.2,3 Although several therapeutic strategies, including photocoagulation, antievascular endothelial growth factor (VEGF) drugs, and steroids, are used clinically to prevent vascular hyperpermeability in DME,4e8 many patients have a poor visual prognosis, and the pathogenesis of DME remains to be investigated. Diabetic retinopathy is a diabetic microangiopathy, and fluorescein angiography (FA) was introduced clinically to delineate better the retinal microvascular changes, including neovascularization, microaneurysms, intraretinal
2596
2013 by the American Academy of Ophthalmology Published by Elsevier Inc.
microvascular abnormalities, and venous beading. In addition to these lesions, hyperfluorescence often indicates the breakdown of the BRB induced by diabetes. Although fluorescein sodium is a smaller molecule, the physiologic retinal vasculature has highly integrated intercellular junctions containing tight junctions and adherens junctions and less fenestration and prevents the dye extravasation. Diabetes increases the paracellular and transcellular permeability mediated via biochemical pathways and growth factors, including vascular endothelial growth factor.9e12 These changes are reflected in the clinical findings, focal or diffuse fluorescein leakage, and fluorescein pooling.5,13,14 However, only a few studies have reported how the hyperfluorescence corresponds to retinal thickening or morphologic changes.15,16 ISSN 0161-6420/13/$ - see front matter http://dx.doi.org/10.1016/j.ophtha.2013.06.014
Murakami et al
Perifoveal Hyperfluorescence and SRD in DME
Optical coherence tomography (OCT) objectively quantifies the retinal thickness and delineates the morphologic changes in the macular area. The macular thickness is correlated modestly with visual dysfunction in DME,17 and macular pathomorphologic features, for example, serous retinal detachment (SRD), cystoid macular edema, and sponge-like retinal swelling in DME, also have been reported.18 Intriguingly, the cystoid spaces in the inner nuclear layer and outer plexiform layer on OCT images correspond to a honeycomb-like or petaloid pattern of fluorescein pooling.15,16 In addition, foveal cystoid spaces often are accompanied by an enlarged foveal avascular zone and surrounding microaneurysms.19 Compared with such changes, few reports have described the fluorescein findings in eyes with SRD in DME. Serous retinal detachment in central serous chorioretinopathy often is associated with fluorescein pooling in the subretinal spaces, whereas SRD in DME is not accompanied by fluorescein pooling in the corresponding areas.16,20,21 In addition, the optical density of the subretinal spaces on OCT images represents the differences between these diseases.22 Therefore, we questioned how vascular hyperpermeability induces development or maintenance of SRDs in DME. In the current study, we investigated the relationship between foveal pathomorphologic features and fluorescein leakage in the individual subfields of the Early Treatment Diabetic Retinopathy Study (ETDRS) grid and showed an association between foveal SRD and perifoveal hyperfluorescence in DME.
Methods Patients We retrospectively reviewed 167 eyes of 116 patients (age range, 49e85 years; mean age, 65.29.6 years; 4 eyes with mild nonproliferative diabetic retinopathy, 84 with moderate nonproliferative diabetic retinopathy, and 79 with severe nonproliferative diabetic retinopathy) who visited the Department of Ophthalmology of Kyoto University Hospital from January 2008 through March 2011. The inclusion criterion was the availability of FA and OCT images of sufficient quality that were acquired on the same day. The major exclusion criterion
Fluorescein Angiography After measuring the best-corrected visual acuity and performing fundus biomicroscopy, FA images were obtained using the Heidelberg Retina Angiograph 2 (Heidelberg Engineering, Heidelberg, Germany) in the early phase (30e90 seconds after intravenous administration of fluorescein dye) and late phase (300e600 seconds after dye administration). At image acquisition, the gain in fluorescein signals was optimized to obtain clear images at each time point, and concomitantly, the gain was not constant. Fluorescein angiography images were exported in TIFF format with 256 different levels of grayscale. Previous reports had evaluated the fluorescein findings in the late phase containing focal or diffuse fluorescein leakage and fluorescein pooling in individual subfields of the ETDRS grid,14,23,24 which encouraged us to quantify the fluorescence levels in those subfields (Fig 1). We first checked the quality of the FA images, and after excluding decentered images, saturated signals in any pixels, or shadows in the peripheral areas by aperture, further processes were applied. The signal intensities of the early-phase images were adjusted to that in the late phase according to the fluorescein intensity of major vessels in each quadrant of the image. Briefly, we quantified the signal intensity at the bifurcation of the major vessels in each quadrant of the FA images using ImageJ (National Institutes of Health, Bethesda, MD) and then calculated the averaged intensity in the early and late phases. The mean fluorescein intensity of major vessels in the early phase was almost the same as that in the late phase (ratio, 0.9670.104; P ¼ 0.1179) in the 26 eyes from which FA images were obtained without changing the excitation or gain levels, suggesting that the intensity of major vessels may be a good internal control. Considering the differences in signal intensity in the major vessels, the signal levels in all pixels of the early-phase images were adjusted to those in the late phase using the adjust brightness function in that software. The late-phase FA images and the early-phase adjusted images were exported to the image processing software (Photoshop; Adobe, San Jose, CA), and the mean signal intensity was calculated in the individual subfields of the ETDRS grid. The early-phase intensity was subtracted from the late-phase intensity to eliminate the background signals derived from the retinal vessels, choroidal tissues, autofluorescence, and so on. Considering the linear increases in fluorescein intensity (Fig 2, available at http://aaojournal.org), we then normalized the intensity by dividing the duration between the early and late phases (seconds) and the averaged signal intensity at 3 points in the major venules within 2 disc diameters from the optic disc margin. The formula used to calculate the fluorescence levels was:
Fluorescence level ðA:U:Þ in each subfield ¼
ðfluorescein intensity ½lateÞ ðadjusted fluorescein intensity ½earlyÞ ðduration of fluorescein perfusion ½secondsÞ ðfluorescein intensity in major venulesÞ
was the presence of proliferative diabetic retinopathy, ischemic maculopathy, or other chorioretinal diseases, because intravitreal hyperfluorescence from neovascular tissue or lesions in other diseases makes quantification of the exact levels of intraretinal fluorescence difficult. Eyes that underwent any intervention for macular lesions also were excluded. All research and measurements adhered to the tenets of the Declaration of Helsinki. The ethics committee of our institution approved the study protocol.
Optical Coherence Tomography Retinal sectional images of the macula were obtained using spectral domain (SD) OCT (Spectralis OCT; Heidelberg Engineering) with infrared images. Raster scans (30 25 ) were acquired, and the OCT maps with the ETDRS grid (central 1 mm, each quadrant of the inner ring [1e3 mm] or outer ring [3e6 mm]) were applied to further investigations. We also determined the macular pathomorphologic features at
2597
Ophthalmology Volume 120, Number 12, December 2013
Figure 1. Quantification of fluorescence levels in individual subfields of the Early Treatment Diabetic Retinopathy Study grid. Shown are results from a representative case with diabetic retinopathy but not center-involved diabetic macular edema (DME). Fluorescein angiography (FA) images in the (A) early and (B) late phases. C, The grayscale in the early-phase FA images is adjusted to the levels of the late phase according to the signal intensity of the major vessels, followed by quantification of the averaged signal intensity in the individual subfields. D, Quantification of the averaged signal intensity in the late phase. E, Fluorescence levels are normalized by the signal intensity in the major venules and the duration of fluorescein perfusion. F, G, Optical coherence tomography map showing no center-involved DME.
the presumed foveal center in the cross-hair sectional images with 30 .19 Center-involved DME on SD OCT was identified according to a recent publication.25 Briefly, 60 eyes of men and 52 eyes of women were diagnosed with center-involved DME and another 55 eyes were diagnosed with no DME in the current study. We first compared the fluorescence levels in eyes with center-involved DME with those in eyes with no DME and also evaluated the fluorescence levels in eyes with SRD at the fovea in the 112 eyes with center-involved DME (Table 1, available at http:// aaojournal.org).
2598
Statistical Analysis The results were expressed as the mean standard deviation. The Student’s t test was used to compare the quantitative data populations with normal distributions and equal variance. The data were analyzed using the ManneWhitney U test for populations with nonnormal distributions or unequal variance. Linear regression analysis (with Bonferroni correction, if necessary) was performed to test the statistical correlation. Multivariate analyses were used to adjust for confounding factors. P<0.05 was considered statistically significant.
Murakami et al
Perifoveal Hyperfluorescence and SRD in DME
Table 3. Association between Fluorescence Levels and Retinal Thickness in Individual Subfields in Diabetic Macular Edema Subfield
Correlation
Central Inner ring Superior Nasal Inferior Temporal Outer ring Superior Nasal Inferior Temporal
R ¼ 0.316, P ¼ 0.0006 R R R R
¼ ¼ ¼ ¼
0.213, 0.479, 0.421, 0.374,
P ¼ 0.0238 P<0.0001 P<0.0001 P<0.0001
R R R R
¼ ¼ ¼ ¼
0.342, 0.376, 0.330, 0.216,
P ¼ 0.0002 P<0.0001 P ¼ 0.0003 P ¼ 0.0225
Results Hyperfluorescence in Eyes with Center-Involved Diabetic Macular Edema When we investigated the levels of fluorescein leakage in eyes with center-involved DME and compared the results with those in eyes without center-involved DME, we found that 112 eyes with centerinvolved DME had higher fluorescence levels in all ETDRS subfields than 55 eyes without center-involved DME (Table 2, available at http://aaojournal.org). We then analyzed the association between fluorescein leakage and retinal thickening measured by OCT in the 112 eyes with center-involved DME and showed that the fluorescence level in each subfield was correlated positively with the mean retinal thickness in the corresponding subfield (Table 3). The associations were stronger in the nasal and inferior subfields in the inner ring and the nasal subfield in the outer ring of the ETDRS grid. Intriguingly, the retinal thickness in the central subfield had stronger correlation with the fluorescence levels in the nasal and inferior subfields in the inner ring (Table 4, available at http://aaojournal.org).
Higher Fluorescence Levels in Eyes with Serous Retinal Detachment Thirty-seven eyes with SRD had higher levels of fluorescence in most subfields than the 75 eyes without foveal SRD (Table 5, Fig 3). Interestingly, the differences in the subfields of the outer Table 5. Perifoveal Hyperfluorescence in Eyes with Foveal Serous Retinal Detachment in Diabetic Macular Edema
Subfield Central Inner ring Superior Nasal Inferior Temporal Outer ring Superior Nasal Inferior Temporal
No Serous Retinal Detachment (75 Eyes; 310L4)
Serous Retinal Detachment (37 Eyes; 310L4)
8.835.53
11.708.40
11.6215.50 9.405.67 9.795.71 11.116.58
16.1410.54 16.4411.70 15.1014.82 15.3911.46
0.1132 <0.0001 0.0074 0.0137
8.785.59 7.905.58 8.084.89 9.685.92
15.1510.75 14.208.96 14.689.55 14.349.69
<0.0001 <0.0001 <0.0001 0.0021
P Value 0.0332
ring were greater than those in the subfields of the inner ring (Table 5). The differences in the nasal subfield were greatest in the subfields of the inner ring, whereas we did not find differences in the superior subfield (Table 5). In addition, the retinal thickness in eyes with SRD was greater than in those without SRD in all subfields (Table 6, available at http:// aaojournal.org). After adjusting for the central retinal thickness by multivariate analysis, eyes with SRD had higher levels of fluorescence than those without SRD in the nasal subfield of the inner ring and in the superior, nasal, and inferior subfields of the outer ring (Table 7, available at http://aaojournal.org). The retinal thickness in eyes with foveal SRD was greater than that in eyes without SRD in all subfields except the temporal subfield of the inner ring (Table 8, available at http://aaojournal.org). Fluorescein pooling is a classic finding that often is delineated in the foveal cystoid spaces in DME. We compared the fluorescence levels in 43 eyes with foveal cystoid spaces alone with those in 32 eyes with neither cystoid spaces nor SRD and found the significant differences in fluorescence levels (Table 9, Fig 4) and retinal thickness of the central subfield (Table 10, available at http://aaojournal.org). However, after the adjustment for central thickness, there were no differences in fluorescence levels in the central subfield (Table 11, available at http://aaojournal.org). The retinal thickness or fluorescence levels of the inner and outer rings did not differ (Table 9; Table 10, available at http:// aaojournal.org).
Discussion Diabetic macular edema has been characterized as retinal thickening seen on ophthalmoscopy and leakage of blood components on FA images. However, clinicians often believe that there is only modest or even lower correlation between these findings, despite the clinical significance of fluorescein leakage as described previously.5,26 In the current study, we showed for the first time that eyes with foveal SRD had higher fluorescence levels in the macular areas than those without SRD. The differences in fluorescence levels also were the most significant in the perifoveal subfields, but not in the central subfield. These data suggested that vascular leakage in these subfields contributes at least partly to the pathogenesis of foveal SRD. Further studies should be planned to determine whether photocoagulation to the parafoveal or perifoveal areas with hyperfluorescence is effective for diabetic patients with foveal SRD. Clinically, center-involved DME is diagnosed mainly based on retinal thickening at the fovea where there is no retinal vasculature. The current results support the hypothesis that BRB breakdown in the parafoveal or perifoveal vasculature induces increased retinal thickness at the fovea. We first investigated the characteristics of eyes with center-involved DME and compared them with eyes without DME and found higher levels of fluorescence in all subfields of the ETDRS grid, which suggested the relevance of hyperfluorescence in DME and encouraged us to evaluate the correlation between fluorescence levels and retinal thickness measured by OCT. Interestingly, the correlation was stronger in the subfields of the inner ring than in the outer ring, except in the superior subfields. The displacement of the inner retinal layers is greater in the parafovea, and neuroglial components in the perifovea have minimal
2599
Ophthalmology Volume 120, Number 12, December 2013
Figure 3. Higher levels of fluorescence in a representative case with foveal serous retinal detachment (SRD) in diabetic macular edema. The fluorescein intensity levels in the individual subfields were quantified in a fluorescein angiography image in (A) the early phase with adjusted grayscale or (B) the late phase. C, Normalized fluorescence levels had higher levels in the nasal and inferior subfield of the outer ring. D, E, Optical coherence tomography map showing the increased retinal thickness in all subfields, especially in the nasal and inferior subfields of the perifovea. F, Serous retinal detachment at the fovea in the sectional image.
horizontal shifts.27 That is, the obliquely arrayed secondary neurons or Müller cells may have a higher capability to be reservoirs for the extravasated blood components in the Table 9. Fluorescence Levels in Eyes with Foveal Cystoid Spaces in Diabetic Macular Edema
Subfield Central Inner ring Superior Nasal Inferior Temporal Outer ring Superior Nasal Inferior Temporal
2600
Neither Foveal Cystoid Spaces Nor Serous Retinal Detachment (32 Eyes; 310L4)
Foveal Cystoid Spaces Alone (43 Eyes; 310L4)
P Value
6.914.40
10.185.88
0.0105
13.3523.22 8.335.38 8.565.53 10.256.26
10.405.82 10.155.80 10.665.73 11.726.80
0.4214 0.1729 0.1161 0.3455
8.265.10 6.835.23 7.064.87 9.715.53
9.145.94 8.645.75 8.804.83 9.666.24
0.5053 0.1678 0.1301 0.9730
outer plexiform layer in the parafoveal areas and may facilitate retinal thickening, which results in a better correlation between the fluorescence levels and retinal thickness. In contrast, because of the lower capacity in the perifoveal areas, extravasated fluids may migrate toward the parafoveal or central area. Another factor was the modest correlation between the fluorescence levels and retinal thickness in the central subfield. Optical coherence tomography identified several patterns of pathomorphologic features, specifically, SRD, cystoid macular edema, and sponge-like retinal swelling, suggesting a variety of pathogeneses of retinal thickening in the central subfield.18 Further, the central thickness was associated more significantly with fluorescence levels in the nasal and inferior subfields in the inner ring and the superior, nasal, and inferior subfields in the outer ring. These data encouraged us to investigate how vascular leakage in the individual subfields contributes to the pathomorphologic features at the fovea. Therefore, we studied the correlation between the foveal SRD and fluorescence levels and found that eyes with SRD had higher levels of fluorescence in almost all subfields than
Murakami et al
Perifoveal Hyperfluorescence and SRD in DME
Figure 4. Fluorescence levels in a representative case with foveal cystoid spaces in diabetic macular edema. The fluorescein intensity was quantified in the individual subfields of the Early Treatment Diabetic Retinopathy Study grid (A) in the early phase with adjusted grayscale and (B) in the late phase. C, Normalized fluorescence levels were higher in most subfields. D, E, Optical coherence tomography map showing that the prominent increase in the central retinal thickness. F, Retinal sectional images delineating the cystoid spaces at the fovea.
those without SRD. However, we did not observe fluorescein pooling in the subretinal space in those eyes.20 These data suggested that higher levels of vascular leakage from the retinal vasculature in the macular area contribute to development or maintenance of SRD at the fovea. Compared with smaller molecules such as water and solutes, extravasated macromolecules containing lipoproteins and albumins cannot be absorbed from the retinal pigment epithelium or retinal vasculature with intact BRB. Concomitantly, increased oncotic or osmotic pressure in the intraocular fluids may lead to development of SRD, as Marmor28 described in his clinical and experimental studies. Although the pathogenesis of SRD in DME remains ill-defined, the higher levels of vascular leakage may be a major contributor, in addition to vitreomacular traction.29,30 There were the greatest differences in the fluorescence levels in the perifovea, especially in the superior, nasal, and inferior subfields, between eyes with and without foveal SRD. The fluorescence levels in the nasal and inferior subfields of the inner ring also had greater differences than those in the central subfield (Table 5). This may indicate that the vascular hyperpermeability in those perifoveal subfields
contributes more to the pathogenesis of foveal SRD, which is to some extent consistent with the correlation between the central retinal thickness and the perifoveal fluorescence levels in the current study (Table 4, available at http:// aaojournal.org). In addition, these data may be explained by distant effects that hyperpermeability in the retinal vasculature outside of the macula often is accompanied by SRD, but not cystoid macular edema, at the fovea in several retinal vascular diseases.31,32 A recent study reported that eyes with SRD had more intact retinal vasculature around the fovea than those with foveal cystoid spaces in DME.19 Based on that, we speculated that the extravasated macromolecules from the perifovea become concentrated in the central areas, with concomitant foveal SRD development. Several publications have reported a relationship between the cystoid spaces seen on OCT and the impressive FA findings of honeycomb-like or petaloid-pattern fluorescein pooling in DME.15,16,30 However, we did not find significant differences in the fluorescence levels in any subfields in the parafovea or perifovea between eyes with and without foveal cystoid spaces compared with the significant differences in the central retinal thickness. The data suggested that
2601
Ophthalmology Volume 120, Number 12, December 2013 the presence of foveal cystoid spaces does not depend on the levels of vascular hyperpermeability in the parafovea or perifovea, which seems to contradict the clinical impression that fluorescein pooling represents pathogenesis in the cystoid spaces. This prompted us to consider a few possible explanations. In the current study, we often found the diffuse fluorescein leakage with higher intensity or fluorescein pooling in eyes without foveal cystoid spaces on OCT images. It suggests that higher levels of fluorescence did not necessarily correspond to the cystoid spaces on OCT images. In addition, it was reported recently that there were various intensities of fluorescein pooling in foveal cystoid spaces.21 Another possible explanation may be ischemia. Experimental studies have reported that microinfarctions in talc retinopathy induce cystoid spaces in the outer plexiform layer,33 and clinically, eyes with DME with foveal cystoid spaces have larger foveal avascular zones.19 Therefore, we hypothesized that the pathogenesis of the foveal cystoid spaces depends on vascular leakage as well as other mechanisms, that is, retinal ischemia. This retrospective study had several limitations. The fluorescence levels in the FA images depend partly on the fluorescence dye in the retina and intraocular humor at least in part. Although we excluded an eye with proliferative diabetic retinopathy in which higher fluorescence levels in the vitreous would be expected, we could not confirm that the measured fluorescence levels were correlated completely with those in the fluorescein dye that extravasated into the retinal tissue alone. In addition, because theoretically the zero point of fluorescence intensity in the FA images, the degree of active transport from the retinal pigment epithelium, and excretion from the kidneys cannot be determined accurately,34,35 the fluorescence levels calculated in this study were approximate but not absolute values. We did not confirm the reproducibility of the image acquisition. Finally, regarding the method of segmentation, we used the ETDRS grid to compare the fluorescence levels with the retinal thickness in each subfield, whereas previous studies used automated segmentation of the hyperfluorescent areas, which is better when considering the source or diffusion of the fluorescein dye.36 In conclusion, the current study showed a correlation between the foveal pathomorphologic features and fluorescence levels in DME, suggesting that severe vascular leakage contributes partly to the pathogenesis of foveal SRD.
References 1. 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;102:7–16. 2. Gardner TW, Antonetti DA, Barber AJ, et al. Diabetic retinopathy: more than meets the eye. Surv Ophthalmol 2002;47(suppl):S253–62. 3. Antonetti DA, Klein R, Gardner TW. Diabetic retinopathy. N Engl J Med 2012;366:1227–39. 4. Mohamed Q, Gillies MC, Wong TY. Management of diabetic retinopathy: a systematic review. JAMA 2007;298:902–16.
2602
5. Early Treatment Diabetic Retinopathy Study Research Group. Photocoagulation for diabetic macular edema. Early Treatment Diabetic Retinopathy Study report number 1. Arch Ophthalmol 1985;103:1796–806. 6. Jonas JB, Sofker A. Intraocular injection of crystalline cortisone as adjunctive treatment of diabetic macular edema. Am J Ophthalmol 2001;132:425–7. 7. Macugen Diabetic Retinopathy Study Group. A phase II randomized double-masked trial of pegaptanib, an antivascular endothelial growth factor aptamer, for diabetic macular edema. Ophthalmology 2005;112:1747–57. 8. Diabetic Retinopathy Clinical Research Network Writing Committee; Elman MJ, Bressler NM, Qin H, et al. Expanded 2-year follow-up of ranibizumab plus prompt or deferred laser or triamcinolone plus prompt laser for diabetic macular edema. Ophthalmology 2011;118: 609–14. 9. Antonetti DA, Barber AJ, Bronson SK, et al; JDRF Diabetic Retinopathy Center Group. Diabetic retinopathy: seeing beyond glucose-induced microvascular disease. Diabetes 2006;55:2401–11. 10. Murakami T, Frey T, Lin C, Antonetti DA. Protein kinase cbeta phosphorylates occludin regulating tight junction trafficking in vascular endothelial growth factor-induced permeability in vivo. Diabetes 2012;61:1573–83. 11. Dvorak AM, Feng D. The vesiculo-vacuolar organelle (VVO). A new endothelial cell permeability organelle. J Histochem Cytochem 2001;49:419–32. 12. Joussen AM, Smyth N, Niessen C. Pathophysiology of diabetic macular edema. Dev Ophthalmol 2007;39:1–12. 13. Browning DJ, Altaweel MM, Bressler NM, et al. Diabetic macular edema: what is focal and what is diffuse? Am J Ophthalmol 2008;146:649–55. 14. Early Treatment Diabetic Retinopathy Study Research Group. Classification of diabetic retinopathy from fluorescein angiograms. ETDRS report number 11. Ophthalmology 1991;98: 807–22. 15. Otani T, Kishi S. Correlation between optical coherence tomography and fluorescein angiography findings in diabetic macular edema. Ophthalmology 2007;114:104–7. 16. Bolz M, Ritter M, Schneider M, et al. A systematic correlation of angiography and high-resolution optical coherence tomography in diabetic macular edema. Ophthalmology 2009;116: 66–72. 17. Diabetic Retinopathy Clinical Research Network. Relationship between optical coherence tomography-measured central retinal thickness and visual acuity in diabetic macular edema. Ophthalmology 2007;114:525–36. 18. Otani T, Kishi S, Maruyama Y. Patterns of diabetic macular edema with optical coherence tomography. Am J Ophthalmol 1999;127:688–93. 19. Murakami T, Nishijima K, Sakamoto A, et al. Foveal cystoid spaces are associated with enlarged foveal avascular zone and microaneurysms in diabetic macular edema. Ophthalmology 2011;118:359–67. 20. Ozdemir H, Karacorlu M, Karacorlu S. Serous macular detachment in diabetic cystoid macular oedema. Acta Ophthalmol Scand 2005;83:63–6. 21. Horii T, Murakami T, Nishijima K, et al. Relationship between fluorescein pooling and optical coherence tomographic reflectivity of cystoid spaces in diabetic macular edema. Ophthalmology 2012;119:1047–55. 22. Neudorfer M, Weinberg A, Loewenstein A, Barak A. Differential optical density of subretinal spaces. Invest Ophthalmol Vis Sci 2012;53:3104–10.
Murakami et al
Perifoveal Hyperfluorescence and SRD in DME
23. Neubauer AS, Chryssafis C, Priglinger SG, et al. Topography of diabetic macular oedema compared with fluorescein angiography. Acta Ophthalmol Scand 2007;85:32–9. 24. Soliman W, Sander B, Hasler PW, Larsen M. Correlation between intraretinal changes in diabetic macular oedema seen in fluorescein angiography and optical coherence tomography. Acta Ophthalmol 2008;86:34–9. 25. Chalam KV, Bressler SB, Edwards AR, et al; Diabetic Retinopathy Clinical Research Network. Retinal thickness in people with diabetes and minimal or no diabetic retinopathy: Heidelberg Spectralis optical coherence tomography. Invest Ophthalmol Vis Sci 2012;53:8154–61. 26. Asrani S, Zeimer R, Goldberg MF, Zou S. Application of rapid scanning retinal thickness analysis in retinal diseases. Ophthalmology 1997;104:1145–51. 27. Sjostrand J, Popovic Z, Conradi N, Marshall J. Morphometric study of the displacement of retinal ganglion cells subserving cones within the human fovea. Graefes Arch Clin Exp Ophthalmol 1999;237:1014–23. 28. Marmor MF. Mechanisms of fluid accumulation in retinal edema. Doc Ophthalmol 1999;97:239–49. 29. Kaiser PK, Riemann CD, Sears JE, Lewis H. Macular traction detachment and diabetic macular edema associated with posterior hyaloidal traction. Am J Ophthalmol 2001;131:44–9.
30. Kang SW, Park CY, Ham DI. The correlation between fluorescein angiographic and optical coherence tomographic features in clinically significant diabetic macular edema. Am J Ophthalmol 2004;137:313–22. 31. Otani T, Yamaguchi Y, Kishi S. Serous macular detachment secondary to distant retinal vascular disorders. Retina 2004;24: 758–62. 32. Tsujikawa A, Sakamoto A, Ota M, et al. Retinal structural changes associated with retinal arterial macroaneurysm examined with optical coherence tomography. Retina 2009;29: 782–92. 33. Tso MO. Pathology of cystoid macular edema. Ophthalmology 1982;89:902–15. 34. Sander B, Larsen M, Moldow B, Lund-Andersen H. Diabetic macular edema: passive and active transport of fluorescein through the blood-retina barrier. Invest Ophthalmol Vis Sci 2001;42:433–8. 35. Knudsen ST, Bek T, Poulsen PL, et al. Macular edema reflects generalized vascular hyperpermeability in type 2 diabetic patients with retinopathy. Diabetes Care 2002;25:2328–34. 36. Phillips RP, Ross PG, Tyska M, et al. Detection and quantification of hyperfluorescent leakage by computer analysis of fundus fluorescein angiograms. Graefes Arch Clin Exp Ophthalmol 1991;229:329–35.
Footnotes and Financial Disclosures Originally received: February 2, 2013. Final revision: June 5, 2013. Accepted: June 5, 2013. Available online: August 13, 2013.
Manuscript no. 2013-189.
Department of Ophthalmology and Visual Sciences, Kyoto University Graduate School of Medicine, Kyoto, Japan.
Correspondence: Tomoaki Murakami, MD, Department of Ophthalmology and Visual Sciences, Kyoto University Graduate School of Medicine, 54 ShogoinKawaracho, Sakyo, Kyoto 606-8507, Japan. E-mail: mutomo@kuhp. kyoto-u.ac.jp.
Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article.
2603