Secondary Logo

Journal Logo

CORRELATION BETWEEN CYSTOID SPACES IN CHRONIC DIABETIC MACULAR EDEMA AND CAPILLARY NONPERFUSION DETECTED BY OPTICAL COHERENCE TOMOGRAPHY ANGIOGRAPHY

Mané, Valérie MD; Dupas, Bénédicte MD; Gaudric, Alain MD; Bonnin, Sophie MD; Pedinielli, Alexandre MD; Bousquet, Elodie MD, PhD; Erginay, Ali MD; Tadayoni, Ramin MD, PhD; Couturier, Aude MD

doi: 10.1097/IAE.0000000000001289
Original Study
Open

Purpose: To study the relationship between the location of cystoid spaces and retinal capillary nonperfusion areas in diabetic cystoid macular edema (DCME).

Methods: In this retrospective study, 24 eyes of 21 patients with chronic DCME were followed using optical coherence tomography angiography. The capillary density of the superficial capillary plexus and deep capillary plexus was measured using AngioAnalytics software in all DCME eyes and in 20 healthy controls. Diabetic cystoid macular edema improved spontaneously or after treatment in 11 eyes.

Results: The intraretinal cystoid spaces were surrounded by capillary-flow void areas in the superficial capillary plexus in 71% of cases and in the deep capillary plexus in 96% of cases. The deep capillary plexus had lost its regular pattern in all cases. The capillary density was decreased in both plexus (mean decrease of −23.0% in the superficial capillary plexus and −12.4% in the deep capillary plexus vs. normal). In the 11 cases with DCME resolution, the capillary did not reperfuse in areas of resolved cystoid spaces, and the capillary density did not change significantly.

Conclusion: In chronic DCME, cystoid spaces were located within capillary dropout areas. No reperfusion occurred after DCME resolution. The impact of the severity of this nonperfusion on the risk of recurrence of DCME remains to be clarified.

In 24 eyes with chronic diabetic cystoid macular edema, optical coherence tomography angiography showed that the cystoid spaces were located within the larger areas of capillary-flow voids. Capillary density was decreased in the superficial and deep capillary plexus. No reperfusion occurred in the areas of resolved cystoid spaces.

Department of Ophthalmology, Hôpital Lariboisière, AP-HP, Université Paris 7—Sorbonne Paris Cité, Paris, France.

Reprint requests: Aude Couturier, MD, Hôpital Lariboisière, Service d'Ophtalmologie, 2 Rue Ambroise Paré, 75475 Paris, CEDEX 10, France; e-mail: aude.couturier@aphp.fr

The Department of Ophthalmology of Lariboisière hospital received an independent research grant from Novartis Pharma SAS. The funding organization had no role in the design or conduct of this research.

None of the authors have any financial/conflicting interests to disclose.

This is an open access article distributed under the terms of the Creative Commons Attribution-Noncommercial-Noderivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially.

In contrast with the significant advances in the treatment of diabetic cystoid macular edema (DCME), the factors that might predict the trend to edema recurrence or the magnitude of visual improvement remain poorly identified. It is well known that diabetic macular edema may be caused by several factors, including glial and neuronal cell dysfunction, inflammatory reaction, blood–retinal barrier breakdown, and capillary dropout.1–5 However, the role of capillary nonperfusion in the severity of DCME remains largely unknown and may have been underestimated by fluorescein angiography because of its inability to show the deep capillary plexus (DCP). The emergence of optical coherence tomography angiography (OCTA) has allowed for the first time analyzing both the superficial capillary plexus (SCP) and the DCP. The ability of OCTA to detect vascular changes in diabetic retinopathy (DR) has already been evaluated6–9 and opens up new possibilities for assessing the capillary perfusion in DCME.

The aim of this study was to describe the retinal structural damages observed in long-standing DCME in both the SCP and DCP, and especially the relationship between cystoid spaces and the capillary nonperfusion during the course and after resolution of the edema.

Back to Top | Article Outline

Patients and Methods

The records of all consecutive patients with diabetes with DCME imaged by OCTA between January and December 2015 in Lariboisière Hospital, Paris, were retrospectively reviewed. Inclusion criteria were patients with DR associated with DCME and for whom high-quality OCTA images had been recorded on at least two different visits. High-quality OCTA images were defined by a signal strength of at least 60. Cases with motion artifacts preventing the accurate analysis of the microvascularization were excluded. Diabetic cystoid macular edema was diagnosed on standard optical coherence tomography (OCT) B-scan. Patients with any other retinal disorder, including a history of vitreous surgery or the presence of media opacities such as vitreous hemorrhage or cataract, were not included. Data collected included baseline demographics (sex, mean diabetes duration, mean DCME duration, hemoglobin A1c, and previous DCME treatment) and current ophthalmologic examination findings (best-corrected visual acuity, slit-lamp biomicroscopy, fundus examination, spectral domain OCT, and OCTA).

All patients with DCME and healthy eyes of patients without diabetes were imaged using the AngioVue OCTA device (Optovue, Inc, Freemont, CA). Optical coherence tomography angiography volumes were acquired on the 3-mm × 3-mm macular region. The device operates 70,000 A-scans per second to create volumes of 304-μm × 304-μm A-scans. Angiograms of the SCP and DCP were automatically segmented using preset parameters. The automated segmentation of the SCP and DCP was used for measuring the capillary density. Customized settings were also used to analyze thinner slabs providing a better visualization of the SCP and DCP. The customization of the slabs was decided, in agreement with the 2 readers, as thinner slabs of 20 μm that were moved progressively from the outer retina to the inner plexiform layer. Customized slabs were mainly used to confirm the absence of capillaries in the areas of cystoid spaces or capillary dropout and to identify the presence of projection artifacts in some cases. Full central macular thickness (CMT), internal retinal thickness, and angioflow density were calculated for each plexus using AngioAnalytics software (version 2015.1.0.71; Optovue, Inc). Details of the density calculation are undisclosed but AngioAnalytics calculates the percentage of the angioflow surface to the total surface of the angiogram. We used the mean value of the capillary density measured in a series of 20 healthy eyes in both the SCP and DCP, for comparison with the capillary density in DCME cases.

The location of cystoid spaces was determined on the en-face image using a semiautomatic delineation of cystoid spaces with GIMP software (v2.8; GNU GPL v3) using a grayscale. This delineation obtained on the en-face image was then overlaid on the angiograms of the SCP and DCP (Figure 1). The nonperfusion areas were easily differentiated from the cystoid spaces,10 and their location and extent were analyzed qualitatively on the angiograms of both plexus. The location of the cystoid spaces and their relationship with nonperfusion areas were studied on the initial and follow-up OCTA images. The OCTA images were analyzed together with the structural en-face OCT scans by each reader. The analyses of the structural en-face OCT scans were helpful for assessing the location of the cystoid spaces. The relationship between cystoid spaces and the capillary nonperfusion was assessed qualitatively by two masked examiners. The areas located on the edges of the cystoid spaces were first analyzed. A colocalization was noted as “present” if the cystoid spaces were surrounded by areas devoid of capillaries or as “absent” if the capillaries could be seen at the periphery of the cystoid spaces in each capillary plexus. In case of discrepancy, the images were reviewed and discussed by the two examiners. Then, the other nonperfusion areas in the entire 3-mm × 3-mm scan were analyzed separately (Figure 2). The Fleiss kappa reliability coefficient was calculated to determine the interexaminer agreement.

Fig. 1

Fig. 1

Fig. 2

Fig. 2

For statistical analysis, continuous variables are presented as mean ± SD. Continuous variables were analyzed using an independent t-test or Mann–Whitney test when appropriate. Correlations were determined using the Spearman correlation coefficient. A P value less than 0.05 was considered significant. Statistical analyzes were performed using XLstat software (Addinsoft, Paris, France) version 2014.6.01.

This study was conducted in compliance with the tenets of the Declaration of Helsinki. All study-related data acquisitions were approved by our institutional review board (CEERB d'Ile de France, Paris, France). Informed consent was obtained routinely from all examined patients to participate in this research.

Back to Top | Article Outline

Results

Twenty-four eyes of 21 patients with DR with DCME (mean age: 56.6 ± 12.2 years), for whom high-quality OCTA images were obtained on at least 2 different visits, were included. Two patients had Type 1 diabetes and 19 had Type 2 diabetes. The mean diabetes duration was 16.1 ± 8.2 years (range: 2–29) and the mean DCME duration was 33.2 ± 25.6 months (range: 8–97). Patient demographics are summarized in Table 1. Each eye was imaged using OCTA on 2.3 ± 0.40 different visits (range: 2–5 visits) over a mean time of 90.21 ± 47.83 days (range: 27–189). Four eyes were imaged before treatment initiation and the 20 remaining eyes were imaged under intravitreal therapy, at the time of DCME recurrence. The mean number of anti-vascular endothelial growth factor (anti-VEGF) injections in these 20 eyes before OCTA examination was 6.1 ± 5.8 (range: 1–20). Diabetic cystoid macular edema regressed in 11 of the 24 eyes, with complete resolution in 4 cases and only a few residual cysts in 7 cases (Table 2). Diabetic cystoid macular edema resolved spontaneously in 6 eyes and after anti-VEGF injections in 5 eyes. The initial OCT angiograms of all the 24 eyes and the follow-up angiograms of the 11 eyes with DCME resolution (defined as a CMT < 250 μm and/or CMT decreased by at least 20%) were analyzed. The follow-up angiograms of the 13 remaining eyes with stable DCME (CMT > 250 μm and/or CMT decreased by less than 20%) are not shown in this report as no significant change in capillary perfusion and edema profile was observed (mean change in CMT: −5.82 ± 41.72 μm).

Table 1

Table 1

Table 2

Table 2

On all the initial OCT angiograms, DCME was visualized as roundish black areas completely devoid of flow signal and perfectly matched with cystoid areas delineated on the companion en-face image. The cysts were more visible and more numerous in the DCP than in the SCP which is consistent with previous OCTA studies and histopathologic studies of DCME.10,11 No capillary perfusion was present within the cystoid spaces in any case. Some vessels that seemed to cross the capillary voids were in fact those located at the roof of the cysts (Figure 1).

In all the 24 eyes, capillary nonperfusion areas were seen as gray irregular areas surrounded by adjacent capillaries,10 and their relationship with intraretinal fluid was analyzed in both plexus on OCT angiograms. The intraretinal cystoid spaces were surrounded by capillary dropout areas in the SCP in 71% (17 of 24 eyes) of cases and in the DCP in 96% (23 of 24 eyes) of cases, indicating a colocalization between the cysts and nonperfusion areas in DCME (Figures 2 and 3). The Fleiss kappa reliability coefficient, reflecting the interexaminer agreement for the analysis of the capillary dropout, was 0.57 for the SCP and 0.87 for the DCP. The DCP had lost its regular pattern in the whole field of the angiogram, even outside nonperfusion areas (Figure 4).

Fig. 3

Fig. 3

Fig. 4

Fig. 4

The capillary density on 3-mm × 3-mm OCT angiograms was measured in both plexus and compared with the mean density in the 20 healthy nondiabetic eyes. The mean capillary density was reduced in both plexus in DCME eyes compared with healthy nondiabetic eyes (44.98 ± 3.32% vs. 58.43 ± 2.28%, P < 0.0001, and 51.55 ± 3.41% vs. 58.84 ± 2.43%, P < 0.0001, for the SCP and DCP, respectively). No correlation was found between the full CMT and the capillary density of the SCP or that of the DCP (P = 0.84 and 0.79, respectively). The comparison between treated and untreated eyes showed no significant difference in capillary density, CMT, and qualitative grading parameters (Figure 5).

Fig. 5

Fig. 5

Among the 11 eyes that experienced DCME resolution, the mean capillary density did not change significantly after edema resolution: 43.40 ± 5.1% after DCME resolution in the SCP compared with 43.89 ± 3.6% on the initial OCTA (P = 0.80), and 49.02 ± 5.8% after DCME resolution in the DCP compared with 51.14 ± 4.1% on the initial OCTA (P = 0.33). This stability of the capillary density after edema resolution was observed both in cases of spontaneous resolution and after anti-VEGF treatment. Similarly, after DCME resolution, the areas of resolved cystoid spaces and the nonperfusion areas outside the cystic spaces in all the 11 eyes remained devoid of capillaries. The few capillaries visible in the areas of resolved edema were either the projection artifact of the SCP (Figures 4 and 6) or capillaries already present at the roof of the cystoid spaces (Figure 1). This phenomenon was even more obvious in cases in which the macula became thinner than normal after treatment.

Fig. 6

Fig. 6

Back to Top | Article Outline

Discussion

In chronic DCME, capillary changes variably combine some degree of capillary closure and leakage from perfused capillaries and microaneurysms.5 It is likely that vasoregression is the first event in the sequence of DR progression.3,4 Although the exact combination of events has not been identified in humans, animal studies have shown that pericytes first disappear followed by endothelial cells, resulting in acellular capillaries and capillary closure.3,4 As a consequence, glial cell activation and increased local VEGF release result in blood–retinal barrier breakdown of the neighboring capillaries.1 In DCME, capillary closure and leakage from perfused capillaries are thus closely linked. The OCTA technology enables to compare changes in capillary perfusion and structural retinal changes in a dataset acquisition. It was then possible to study the relationship between the capillary density, capillary dropout, and intraretinal cystic spaces induced by capillary leakage. On OCTA, both nonperfusion areas and cystoid spaces appear black without any decorrelation signal. However, cystoid spaces may easily be differentiated from nonperfusion areas by their shape,10 and because they appear identical on structural OCT and on the companion en-face image (Figures 2 and 3).

We found that cystoid spaces were colocalized within nonperfusion areas, especially in the DCP where they were surrounded by a wider area of poor capillary perfusion, suggesting a possible relationship between the capillary dropout and the occurrence of the edema. Furthermore, in this series, no reperfusion occurred in areas of cystoid spaces after DCME resolution. A decrease in capillary density was found in the SCP and DCP in all DCME eyes, without significant change after treatment. Considering these results, it seems unlikely that the cystoid spaces had only displaced laterally the capillaries or reduced their visualization in chronic DCME. It could be assumed that fluid preferentially accumulated in areas of capillary dropout. This is in agreement with findings in retinal vein occlusion showing that edema occurs in areas of poor perfusion in eyes with branch retinal vein occlusion.12,13 However, it could also be assumed that some capillaries were not detected by OCTA because their blood flow was too slow and not detected by split-spectrum amplitude–decorrelation angiography. Thus, nonperfusion areas might either be void of capillaries or show a severely decreased flow signal. In DCME, de Carlo et al10 have reported that the microvasculature reappeared in some cases in the areas of intraretinal fluid after treatment and hypothesized either a reperfusion after anti-VEGF treatment or a displacement of the vessels by the cystoid spaces. We did not find such an evolution. In our cases, we only observed some capillaries located at the roof of the cysts projecting partially into the cystoid spaces and still visible after edema resolution, or some capillaries reappearing with changes in the segmentation plane. In all cases, they were already present at the baseline pretreatment examination. However, it should be noted that our case selection mainly included chronic or recurrent DCME with already well-established areas of capillary nonperfusion. At an earlier stage of the disease, DCME could eventually develop without significant capillary dropout. This relationship between the capillary closure and the occurrence of DCME has previously been stressed in reports on the microaneurysm turnover.14,15 It is well known that microaneurysms develop in areas of acellular capillaries and are the clinical marker for capillary nonperfusion in early DR.16,17 Yet, a high microaneurysm turnover has been associated with a higher risk of DCME development in nonproliferative DR eyes, suggesting also a possible relationship between the capillary nonperfusion and the occurrence of DCME.14,15 Further long-term studies of diabetic eyes with early-onset DCME are needed to clarify the relationships between the progression of capillary dropout and the occurrence of DCME.

We also found that in patients with DCME, the capillary density was less than that found in healthy controls, confirming the findings of the report by Agemy et al18 for DR in general. However, the loss of density capillary was greater in the SCP than in the DCP, which contrasts with the results of Agemy et al. In our series, the capillary density in the DCP was obviously overestimated by AngioAnalytics software for several reasons. First, some reflective areas were falsely interpreted as a correlation signal such as the walls between cystoid spaces or the very small lipid deposits in the cystoid spaces themselves, and overestimated the flow density (Figure 6). Second, projection artifacts of the SCP on the DCP also contributed to a false increase in capillary density in the DCP (Figures 4 and 6).19 However, although until now it was not possible to correlate capillary loss and retinal tissue thinning or disorganization, the retinal capillary density may nowadays be measured, although imperfectly, by OCTA. Future improvements in density measurement software could enable to clarify the relationship between the capillary perfusion, tissue loss, and visual function. It remains to be demonstrated whether the amount of capillary dropout before treatment may have a prognostic value for the functional recovery.

This study had some limitations, including its retrospective design and small sample size. Cases were mainly chronic recurrent DCME, which might not be representative of all types of DCME. The presence of cystoid spaces could have limited the segmentation accuracy and the assessment of capillary changes. False decorrelation signal and artifactual projections of the SCP on the DCP impaired the interpretation of DCP angiograms and the measurement of capillary density. The measurement tool for the vessel density has not been validated in a large series and has no corrections for artifacts. The fact that most patients had previously received anti-VEGF therapy could also have influenced the capillary changes seen; however, the four treatment-naive eyes showed similar capillary changes on OCTA.

In summary, this OCTA study of areas of nonperfusion in chronic DCME shows that cystoid spaces were colocalized within areas of capillary nonperfusion. No capillary reperfusion occurred after DCME resolution. Further studies in early stages of untreated DCME are needed to confirm the relationship between the occurrence of edema and the capillary dropout.

Back to Top | Article Outline

References

1. Antonetti DA, Klein R, Gardner TW. Diabetic retinopathy. N Engl J Med 2012;366:1227–1239.
2. Mizutani M, Gerhardinger C, Lorenzi M. Müller cell changes in human diabetic retinopathy. Diabetes 1998;47:445–449.
3. Hammes H-P, Feng Y, Pfister F, Brownlee M. Diabetic retinopathy: targeting vasoregression. Diabetes 2011;60:9–16.
4. Frank RN. Diabetic retinopathy. N Engl J Med 2004;350:48–58.
5. Das A, McGuire PG, Rangasamy S. Diabetic macular edema: pathophysiology and novel therapeutic targets. Ophthalmology 2015;122:1375–1394.
6. Ishibazawa A, Nagaoka T, Takahashi A, et al. Optical coherence tomography angiography in diabetic retinopathy: a prospective pilot study. Am J Ophthalmol 2015;160:35–44.
7. Jia Y, Bailey ST, Hwang TS, et al. Quantitative optical coherence tomography angiography of vascular abnormalities in the living human eye. Proc Natl Acad Sci U S A 2015;112:E2395–E2402.
8. Couturier A, Mané V, Bonnin S, et al. Capillary plexus anomalies in diabetic retinopathy on optical coherence tomography angiography. Retina 2015;35:2384–2391.
9. Stanga PE, Papayannis A, Tsamis E, et al. New findings in diabetic maculopathy and proliferative disease by swept-source optical coherence tomography angiography. Dev Ophthalmol 2016;56:113–121.
10. De Carlo TE, Chin AT, Joseph T, et al. Distinguishing diabetic macular edema from capillary nonperfusion using optical coherence tomography angiography. Ophthalmic Surg Lasers Imaging Retina 2016;47:108–114.
11. Byeon SH, Chu YK, Hong YT, et al. New insights into the pathoanatomy of diabetic macular edema: angiographic patterns and optical coherence tomography. Retina 2012;32:1087–1099.
12. Finkelstein D. Ischemic macular edema. Recognition and favorable natural history in branch vein occlusion. Arch Ophthalmol 1992;110:1427–1434.
13. Spaide RF. Volume rendered optical coherence tomography of retinal vein occlusion pilot study. Am J Ophthalmol 2016;165:133–144.
14. Nunes S, Ribeiro L, Lobo C, Cunha-Vaz J. Three different phenotypes of mild nonproliferative diabetic retinopathy with different risks for development of clinically significant macular edema. Invest Ophthalmol Vis Sci 2013;54:4595–4604.
15. Ribeiro ML, Nunes SG, Cunha-Vaz JG. Microaneurysm turnover at the macula predicts risk of development of clinically significant macular edema in persons with mild nonproliferative diabetic retinopathy. Diabetes Care 2013;36:1254–1259.
16. Bresnick GH, Davis MD, Myers FL, de Venecia G. Clinicopathologic correlations in diabetic retinopathy. II. Clinical and histologic appearances of retinal capillary microaneurysms. Arch Ophthalmol 1977;95:1215–1220.
17. Kohner EM, Henkind P. Correlation of fluorescein angiogram and retinal digest in diabetic retinopathy. Am J Ophthalmol 1970;69:403–414.
18. Agemy SA, Scripsema NK, Shah CM, et al. Retinal vascular perfusion density mapping using optical coherence tomography angiography in normals and diabetic retinopathy patients. Retina 2015;35:2353–2363.
19. Spaide RF, Fujimoto JG, Waheed NK. Image artifacts in optical coherence tomography angiography. Retina 2015;35:2163–2180.
Keywords:

anti-VEGF; capillary density; capillary dropout; capillary nonperfusion; capillary plexus; cystoid macular edema; diabetic macular edema; diabetic retinopathy; optical coherence tomography angiography

© 2016 by Ophthalmic Communications Society, Inc.