Abstract
Purpose:
To study the association between the distributions of microaneurysms detected by en face optical coherence tomography angiography (OCTA) and diabetic macular edema (DME).
Methods:
The study design was a retrospective chart review of 27 patients (33 eyes) with DME. The eyes were scanned using OCTA (6 × 6 mm) and spectral-domain (SD) OCT macular cube. Each of the images of the capillary plexus was overlaid onto the image of the topographic map, and the densities of the microaneurysms were measured by ImageJ software. The association between the distribution of microaneurysms and macular edema was evaluated.
Results:
For microaneurysms in areas with and without edema, 77.3 ± 8.1% of these microaneurysms were located in the deep capillary plexuses (DCP). However, in areas of edema where the retinal thickness was more than 400 μm, 91.3 ± 9.1% of the microaneurysms were found in the DCP. This difference was statistically significant (P < 0.001). In the macular edema area, there was a significantly higher density of microaneurysms in the DCP compared to the superficial capillary plexuses (1.71/mm2 vs. 0.17/mm2, P < 0.001). There was also a significant correlation between the macular volume and the density of microaneurysms in the DCP in edema (r = 0.63, P < 0.001).
Conclusions:
Our study demonstrated a high proportion of microaneurysms in the DCP, as well as a novel association between the distributions of microaneurysms detected by OCTA and DME. Results also indicated that microaneurysms located in the DCP contribute to the pathogenesis of DME.
Diabetic macular edema (DME) is a leading cause of blindness in most industrialized nations.
1,2 Although the pathogenesis of DME remains unclear, it is thought to be related to the breakdown of the blood–retina barrier, leaking microaneurysms (MAs), and upregulated cytokine, among other factors. Enhanced expression of the proangiogenic cytokine vascular endothelial growth factor (VEGF)-A has been validated in patients with DME,
3,4 and anti-VEGF-A therapy is the current standard of care for DME.
5–7 However, this treatment requires repeated intravitreous injections for an indefinite period, and safety concerns regarding continual blockade of VEGF-A, which is constitutively expressed in the normal adult human retina, are emerging.
8–12 Although previous clinical studies for DME have proven that focal laser treatment for MAs is effective in focal edema that originated from leaking MAs,
13,14 other clinical studies that compared anti-VEGF therapy and focal laser treatment have shown the superiority of the anti-VEGF therapy in the treatment of DME.
5–7 Recently, precise focal photocoagulation using navigated laser photocoagulation combined with anti-VEGF antibody therapy showed promising efficacy for reducing the burden of the anti-VEGF agent.
15 To conduct precise focal photocoagulation, an accurate determination of the leaking MAs that are responsible for the edema is required. Currently, only fluorescein angiography can be used to clarify whether the type of edema is focal or diffuse and determine the leaking MAs. However, issues associated with the dye injection, such as renal dysfunction and anaphylaxis, are of concern for both patients and doctors.
16
The use of optical coherence tomography (OCT) for in vivo vascular imaging was first reported for Doppler OCT, with this technique subsequently referred to as optical coherence angiography.
17 Recent developments have led to several new variations of the technique,
18–22 which are now collectively referred to as OCT angiography (OCTA). Optical coherence tomography angiography can be used to visualize the chorioretinal microcirculation without dye injection. The AngioVue system (Avanti OCT; Optovue, Fremont, CA, USA) using the split-spectrum amplitude decorrelation angiography (SSADA) algorithm can divide the retina–choroid layer into four layers: the superficial capillary plexuses (SCP), deep capillary plexuses (DCP), outer retina, and choroidal vessels.
23–28 Recent reports have demonstrated that OCTA can detect the characteristics of the vasculature in eyes with diabetic retinopathy.
29,30
In the current study, we sought to further explore the association between the distribution and density of MAs detected by OCTA and macular edema detected by spectral-domain (SD)-OCT (Cirrus HD-OCT; Carl Zeiss Meditec, Dublin, CA, USA) in patients with DME.
In each eye, we overlaid the retinal thickness topographic map obtained by the Cirrus HD-OCT onto the images of SCP and DCP obtained by OCTA using Adobe Photoshop (cs2; Adobe Systems, Inc., San Jose, CA, USA). Due to issues related to the color contrast of the topographic map, it was hard to identify the area where the retinal thickness was more than 300 or 350 μm (
Fig. 2). Therefore, we defined the area where the retinal thickness was more than 400 μm (orange-red to white in the Cirrus HD-OCT color chart) as the area of edema. The measured area of the retinal edema (mm
2) and the number of MAs were determined by using the ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD, USA; available in the public domain at
http://rsb.info.nih.gov/ij/index.html).
The MAs in the SCP and DCP were measured separately. Based on the findings by Ishibazawa et al.,
29 we defined the MAs as round, saccular, or fusiform capillary dilation. We also separated the number of MAs inside of the retinal edema from the number of MAs outside of the retinal edema.
The number of MAs/mm2 in the SCP and DCP was calculated as the density of the MAs. In addition, we also separately calculated the densities of the MAs inside and outside of the edema.
In order to emphasize the density of the MAs within the area of the edema, we calculated the ratio of the density of MAs (/mm2) inside the edema to the density of the MAs (/mm2) outside the edema.
We observed that some of the MAs that were located near the border were found in both the SCP and the DCP. Thus, we decided to select eight eyes at random and then evaluate the number of MAs that could be found at the same point in both the SCP and the DCP.
All measurements were performed twice by one inspector (NH) masked to the area of edema that was shown in the topographic map. Mean values of the two measurements were analyzed and used as the result. We also calculated the correlations for each of the values that were measured twice.
All results are expressed as the mean ± SD. Data were collected and analyzed by a Student's t-test and the Tukey-Kramer test. For all statistical tests, P < 0.05 was considered as the significance level. Statistical analysis was performed using Ekuseru-Toukei 2012 (Social Survey Research Information Co., Ltd., Tokyo, Japan) and IBM SPSS Statistics for Windows, Version 22.0 (IBM Corp., Armonk, NY, USA). Pearson correlation coefficient was calculated to assess relationships between the number of MAs and the MV (mm3), with r2 > 0.4 considered to be strongly correlated.
Mean correlation coefficient between the two measurements was 0.93 (0.89–0.99) in counting MAs, and mean correlation coefficient was more than 0.99 in the edema measurement area (>400 μm).
For retinal edema, the mean area (>400 μm) for the 6 × 6–mm topographic map was 5.1 ± 4.5 mm2 (0.5–22.1). The mean number of MAs detected by 6 × 6–mm en face OCT angiography was 17.2 ± 6.8 (6–29).
For the total number of MAs, 77.3 ± 7.6% of the MAs were located in the DCP, whereas in the retinal edema area, 91.3 ± 8.7% of the MAs were located in the DCP. There was a statistically significant difference between the two values (
P < 0.001) (
Fig. 3).
As seen in
Figure 4, the analysis showed that there was no correlation between the ratio of the density of MAs in the DCP to the SCP and the total number of MAs (
r = 0.07,
P = 0.69).
The densities of the MAs (/mm
2) inside and outside of the edema in the SCP were 0.17 ± 0.25 and 0.10 ± 0.06, respectively. Likewise, the densities of the MAs (/mm
2) inside and outside of the edema in the DCP were 1.71 ± 1.21 and 0.24 ± 0.10, respectively (
Fig. 5). Higher densities of MAs were found inside and outside of the edema in the DCP versus the SCP, with the highest density observed inside the edema in the DCP (
P < 0.001).
In the eight eyes that were randomly selected from the subject group, 3.53 ± 1.55% (range, 0–7.40) of the total MAs were detected at the same point in both the SCP and the DCP.
Our current study examined the association between the distribution of MAs detected by OCTA and DME. Although the density of MAs in the SCP exhibited no significant difference between inside and outside of the edema, there was a significant increase in the density inside the edema in the DCP. These results demonstrated that the density of the MAs in the DCP was significantly associated with the MV. Moreover, our results also indicated that the MAs in the DCP contributed to the pathogenesis of DME, especially for the CME type of DME. To the best of our knowledge, this is the first study to evaluate the association between the distribution of MAs in each capillary plexus layer and DME using three-dimensional OCTA imaging.
Moore et al.
33 used immunohistochemistry to investigate human diabetic postmortem eyes, and showed that the majority of the MAs (approximately 80%) originated from the INL and therefore were in the deeper part of the inner retinal capillary plexus. Horii et al.
34 also reported that 80.3% of the MAs were located in the INL, while Ishibazawa et al.
29 used OCTA to examine diabetic eyes and showed that the MAs were mainly located in the deep plexus. Couturier et al.
30 used an OCTA 3 × 3–mm scan area and reported that the mean number of MAs located in the DCP was 4.4 while it was 2.9 in the SCP in the eyes with diabetic retinopathy. Although these previous studies did include diabetic eyes without DME, their findings were compatible with our current results, which showed that 77.3% of the MAs were located in the DCP.
Snodderly et al.
35 reported that the capillary network bracketing the INL, which contains the DCP, is more voluminous than the capillary network associated with the ganglion cell and nerve fiber layers, which contains the SCP. Anatomically, the majority of arteries and veins lie in the superficial vascular plexus, while the majority of capillaries lie in the deep plexus.
36 While the precapillary arterioles are enveloped by a single continuous layer of smooth muscle, the capillaries are surrounded only by pericytes.
37 Considering the distribution of the capillaries, we think it is reasonable that a majority of the MAs should be found in the DCP. In our study, we found no correlation between the ratio of the density of MAs in the DCP to the SCP and the total number of MAs. This indicates that the ratio of the number of MAs in the DCP to the SCP did not differ among the eyes examined, regardless of the severity of the diabetic retinopathy. These findings also support the hypothesis that the formation of MAs is primarily dependent on the vessel type, that is, capillaries or other vessels.
Previous OCT (B-scan) analysis has shown that MAs are the major source of focal fluid accumulation and are located deep inside the INL in focal edema.
38 Our current results also demonstrated that the MAs located in the DCP were significantly associated with the area of the edema, and that the density of the MAs in the DCP was significantly associated with the CME type of edema. Murakami et al.
32 reported that the CME type had significantly larger numbers of MAs compared with the SRD type or the retinal swelling type, and that the cystoid changes were usually located in the INL and outer plexiform layer.
33,35 These previous reports also support our speculation that the MAs in the DCP are responsible for pathogenesis of the CME type of edema.
Liegl et al.
15 reported that the use of precise focal laser photocoagulation for leaking MAs in conjunction with anti-VEGF therapy was a promising treatment option, as it reduced the number of additional anti-VEGF injections that were required. In order to detect leaking MAs, eyes need to be examined by fluorescein angiography. Unfortunately, some patients cannot tolerate the dye injection and thus fluorescein angiography cannot be used. In our present study, we used OCTA and demonstrated that there was a significant correlation between the MAs in the DCP and DME. Although OCTA cannot be used to visualize leakage, our results do indicate that OCTA might have use as a possible therapeutic approach for detecting MAs with adequate flow.
The limitations of our present study include the fact that it was a retrospective study and examined only a limited number of patients. In addition, in order to be able to quantify the distribution of the MAs, we excluded eyes with poor OCTA image quality. Although we evaluated 94 eyes, a total of 61 had to be excluded due to poor image quality caused by cataract and/or poor fixation and/or segmentation error. In addition, our exclusion criteria might have eliminated severe DME from the cases examined in this study. Furthermore, segmentation errors and projection artifacts have the potential to complicate OCTA analyses. In the current study, we found that there were several MAs that were located at the same point in both the SCP and the DCP. Thus, since these MAs were considered to be projection artifacts or to exist near the border of the segmentation, this could have led to errors during the counting of the MAs. To examine this potential problem, we subsequently tried to quantify the number of MAs that were found at the same point in the SCP and the DCP. Our results showed that less than 3.53% of all MAs were detected at the same point in the SCP and the DCP. This suggests that our results showing that the MAs in the DCP were associated with DME were not influenced by these errors. Also, our study did not compare MAs detected by OCTA with those detected by fluorescein angiography. Couturier et al.
30 reported that only 62% of MAs detected by fluorescein angiography could be visualized when using OCTA. Since OCTA can detect only blood flow, slow flow, or flow turbulence
29 in MAs, this might be the reason why visualization could not be achieved with use of OCTA. Since our findings did show that the MAs in the DCP were associated with DME, a further study that compares MAs detected by OCTA and by fluorescein angiography will need to be undertaken in the future.
In summary, the findings of our examination of the association between the distribution of MAs detected by OCTA and DME suggest that MAs in the DCP might contribute to the pathogenesis of DME. Optical coherence tomography angiography is a new technology that can divide the SCP and DCP, which makes it possible to perform a three-dimensional evaluation of the structure of edema. Our results additionally showed the potential benefit of being able to treat DME by targeting MAs detected by OCTA in the DCP.
The authors thank Fumie Shibuya, Yasuyo Matsuda, Ayano Maruyama, and Sayaka Oshio for their technical assistance.
Disclosure: N. Hasegawa, None; M. Nozaki, None; N. Takase, None; M. Yoshida, None; Y. Ogura, None