Suspended scattering particles in motion using OCT angiography in branch retinal vein occlusion disease cases with cystoid macular edema

We aimed to investigate the clinical implication of suspended scattering particles in motion (SSPiM) using optical coherence tomography angiography (OCTA) among branch retinal vein occlusion disease (BRVO) cases with macular edema (ME). Medical records of BRVO patients were reviewed. Central retinal thickness (CRT), ME type, and cyst size on optical coherence tomography images were evaluated before and after intravitreal bevacizumab injection. Nonperfusion area, SSPiM, and microvascular abnormalities in OCTA images were evaluated using a Heidelberg machine. SSPiM was identified in 24 of 56 cases. There were no differences in baseline characteristics between groups with and without SSPiM. Disease duration, disease-free duration, previous injection number, microaneurysms in the superficial vascular complex, and microaneurysms in the deep vascular complex (DVC) (p = 0.003, 0.013, 0.028, 0.003, < 0.001, respectively) differed significantly between the two groups. After multivariate logistic analysis, microaneurysms in the DVC were the only different factor between the two groups (odds ratio [OR]: 0.091; p = 0.001). Furthermore, SSPiM in the DVC (OR 10.908; p = 0.002) and nonperfusion grade (OR 0.039; p < 0.001) were significantly associated with cyst response after intravitreal injection. SSPiM may be correlated with microaneurysms in the DVC and a poor anatomical response after intravitreal injection.


Results
patient characteristics. Among the total 135 patients identified during the chart review period, 43 cases were excluded due to no history of intravitreal injections after symptomatic macular edema, and another 36 cases were excluded due to inadequate image quality and follow-up loss. Finally, 56 cystoid ME cases treated with intravitreal Avastin injection were included. Thirty-two cases did not show SSPiM, and 24 cases showed SSPiM. Mean age was 62.35 ± 9.92 years, and 34 patients were females. Mean disease duration was 10.74 ± 15.24 months, and mean disease-free duration was 6.43 ± 10.24 months. All cases received at least one injection, and the mean previous injection number was 2.43 ± 2.53. Mean follow-up duration was 8.75 ± 3.48 months.
OCT and OCTA findings. Patients with SSPiM (group 1) showed no significant differences in initial vision, age, sex, intraocular pressure (IOP), ocular perfusion pressure (OPP), or central retinal thickness (CRT) compared with patients without SSPIM (group 2). Group 1 had a longer disease duration (16.13 ± 17.52 months vs 6.69 ± 12.04 months, p = 0.003 by Mann-Whitney U test), longer disease-free duration (7.29 ± 8.93 months vs 5.78 ± 11.22 months, p = 0.013 by Mann-Whitney U test), and a greater number of previous injections (3.08 ± 2.26 vs 1.91 ± 2.64, p = 0.028 by Mann-Whitney U test) than group 2. Group 1 showed a greater tendency for microaneurysms in the superficial vascular complex (SVC, yes/no; 19/5 vs 12/20; p = 0.002) and microaneurysms in the deep vascular complex (DVC, yes/no; 20/4 vs 10/22; p = 0.001) than group 2. All patients had telangiectasia in the SVC and most patients had telangiectasia in the DVC. None of the cases had definite collateral vessels in either the SVC or DVC. After one intravitreal bevacizumab injection, mean CRT showed no significant difference between the two groups. Cystoid changes after treatment in group 1 were less extensive than those in group 2 (Table 1).

Sub-analysis of SSpiM after intravitreal injection.
Among 24 cases with SSPiM, only three cases resolved completely after one intravitreal Avastin injection while the other 21 cases received 1.58 ± 0.93 subsequent injections. Fourteen cases with persistent SSPiM after one intravitreal injection did not have larger CRT values than the 10 cases with transient SSPiM, but there were significant differences in the number of previous intravitreal injections (5.20 ± 1.32 vs 1.57 ± 1.39, p < 0.001), disease duration (26.00 ± 19.86 months vs 9.07 ± 11.88 months, p = 0.003), and disease-free duration (12.80 ± 11.02 months vs 5.50 ± 6.16 months, p = 0.036) between the sub-group with persistent SSPiM and the sub-group with transient SSPiM. During follow-up, cyst size decreased on OCT images in most patients, and this corresponded with SSPiM on OCTA images. Most SSPiM were located at the vascular-nonvascular junction (junction between the perfused retina and the nonperfused retina), even when multiple SSPiMs were simultaneously located in a single cyst (Fig. 1). As mentioned above, BCVA was not significantly different between groups 1 and 2. In one case, there was a consistent central cystoid lesion corresponding to SSPiM during 12 months of follow-up, but this patient's visual acuity remained 20/20 (Fig. 2).

Discussion
Nonvascular decorrelation signals can lead to challenges or difficulties in finding the real retinal microvasculature during OCTA image analysis. However, nonvascular decorrelation signals are not always artifacts and could have clinical and anatomical meaning [15][16][17]20 . In our study, we investigated if SSPiM in OCTA images had clinical relevance.
Cases with SSPiM (group 1) had a longer disease duration than cases without SSPiM (group 2). There were also significant differences in disease-free duration and previous injection number between the two groups. One hypothesis to explain SSPiM are that these are particles remaining after resolution of a retinal hemorrhage plus fluid leakage from the broken BRB following venous obstruction. Another hypothesis is that SSPiM are related to microaneurysms because microaneurysms in BRVO are related to chronic disease and are known to be factors for ME relapse 2,21,22 . Microaneurysms in BRVO can occur at the junction of the nonperfused and perfused retina 23 , and one study suggested that microaneurysms are a secondary change caused by local hypoxia and increased  24 . Several studies in humans and animals have also demonstrated that retinal vasculatures, especially venules, may connect to the DVC without direct connection with the SVC and that the DVC may function as a venous route 6,7,[25][26][27] . Furthermore, the outer plexiform layer comprises highly active metabolic tissue 28,29 . The middle retina may be more vulnerable to ischemia than the inner retina because the middle retina is far from both the choroidal and retinal circulations. In contrast, the inner retina has low oxygen demand because oxygen diffusion from the vitreous is also available. These results are consistent with the significant association between SSPiM of the DVC and treatment response and our finding that there were more microaneurysms in the DVC than in the SVC (although this difference was not statistically significant).
However, the results of the current study contradict some of our those of our previous study of SSPiM in DME. The factor most associated with treatment response in our previous study of DME patients was microaneurysms in the SVC. In DME, both the SVC and DVC have vascular diseases, and it appears that in these patients, the change in the SVC is larger than that in the DVC. Most SSPiM in BRVO cases in this study were observed in both the SVC and DVC. Abnormalities of the DVC may be more common than those of the SVC in vascular diseases such as BRVO. It is also possible that the SSPiM may sink toward the DVC while lying in bed at night because cysts corresponding to SSPiM on OCTA were composed of denser mass-like substances than cysts in DME. Furthermore, most BRVO cases with SSPiM showed more incomplete anatomical resolution than DME cases with SSPiM. Mean change in cyst size in the group with SSPiM before and after treatment was 43% in the current study. Further studies are needed to determine whether the positioning of the SSPiM changes as posture changes in both BRVO and DME patients.
Anatomical response of cysts to anti-VEGF injection was significantly related to nonperfusion grade and SSPiM in the DVC in multivariate analysis, while there were no significant factors related to the anatomical response of the CRT after anti-VEGF injection. Previous fundus autofluorescence studies showed that retinal nonperfusion was correlated with ME in BRVO cases 30,31 . In addition, foveal capillary nonperfusion on FAF images and macular thickness on OCT images were related to the recurrence of ME after intravitreal anti-VEGF www.nature.com/scientificreports/ treatments in BRVO 5 . Although our previous study focused on acute BRVO cases with ME, nonperfusion areas on OCTA images were consistently associated with ME recurrence 32 . This study had several limitations. First, it was a retrospective, cross-sectional study with a small number of cases and therefore low statistical power. Due to the retrospective nature of the study, we were not able to assess the overall frequency of SSPiM. Furthermore, we did not find factors significantly related to CRT response. Additionally, we evaluated OCTA images using the Spectralis imaging system, while other OCTA analyses of BRVO cases used other OCA devices [8][9][10]12,16 . Differences in the settings for the SVC and DVC in different OCTA software programs may affect the en face OCTA images. Additionally, we only used one anti-VEGF drug (bevacizumab) for treatment. Further studies of other anti-VEGF drugs such as ranibizumab, aflibercept, triamcinolone, or dexamethasone are required to further evaluate the treatment response of BRVO patients with SSPiM. Last, www.nature.com/scientificreports/ we could not accurately evaluate vessel density and FAZ area because retinal hemorrhage and exudate were frequently observed in our BRVO patients.
In conclusion, we demonstrated that SSPiM on OCTA images are present in chronic BRVO cases and may be correlated with a poor response to intravitreal bevacizumab treatment. Nonperfusion grade in addition to SSPiM in the DVC were significantly associated with a poor anatomical response of cysts to treatment. Eyes with SSPiM had more microaneurysms in the DVC than eyes without SSPiM. Future prospective studies using one OCTA machine in acute BRVO patients with a long-term follow-up duration will be helpful for further elucidating the clinical relevance of SSPiM.

Materials and methods
patients. The Institutional Review Board of Korea University Guro Hospital approved this study and waived the requirement to obtain signed consent for study participation. All research and data collection were conducted following the tenets of the Declaration of Helsinki.
As a retrospective cross-sectional study, chart review was done for consecutive patients diagnosed with unilateral BRVO with ME at Korea University Guro Hospital between March 2018 and May 2019. All patients received complete ophthalmic examinations at their visit, including best-corrected visual acuity (BCVA) measurement (expressed as logMAR units), slit-lamp examination, dilated fundus examination, and OCT and OCTA imaging. During follow-up, all examinations were performed again and analyzed. In addition, total disease duration, disease-free duration, blood pressure, and intraocular pressure (IOP) before intravitreal bevacizumab injection were reviewed. Disease-free duration was defined as the time interval between the previous injection period and the recurrence period. IOP was measured by Goldmann applanation tonometry. Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured with a digital automatic blood pressure machine 33 . Mean ocular perfusion pressure (MOPP) was defined as 1/3 (mean arterial pressure-IOP) and mean arterial pressure (MAP) was defined as DBP + 1/3 (SBP−DBP) 33 .
Study inclusion criteria were as follows: (1) unilateral BRVO with cystoid ME; (2) treatment with one or more intravitreal bevacizumab injections; (3) OCTA images taken before and after intravitreal bevacizumab injection; (4) central retinal thickness (CRT) of ≥ 300 μm as measured by OCT before intervention; (5) OCT with a signal-to-noise ratio of 0.6 or greater; and (6)  image analysis. The Heidelberg image viewer software defined the SVC as the region from the internal limiting membrane to the outer border of the inner plexiform layer (IPL). The DVC was considered the area from the outer border of the IPL to the outer border of the outer plexiform layer. SSPiM in en face OCTA images was defined as a cystic lesion with a nonvascular decorrelation signal (Fig. 3). Cases with SSPiM were assigned to group 1, and cases without SSPiM were assigned to group 2. We could not evaluate the area of the foveal avascular zone and vessel density accurately due to cystic lesions on OCTA images; nevertheless, vessel density reduction can be used to predict whether recurrences of macular edema will develop after the initial anti-VEGF injection in BRVO cases 34,35 . OCTA images in our macular edema cases were not appropriate for accurate evaluate of vessel density and FAZ. Instead, we assessed the simplified nonperfusion grade. In our previous study of the nonperfusion area in BRVO using OCTA images, receiver operating characteristic curve analysis was used to identify the cutoff value; this was 3/8 of the area in the 3-mm zone, which had a sensitivity of 48% and specificity of 93.33% 32 . The degree of nonperfusion in the fovea 3-mm zone was divided into two grades according to the angle of the nonperfusion area (grade 1: < 135 axis degrees vs grade 2: ≥ 135 axis degrees) (Figs. 1 and 3). In addition, CRT and ME type were evaluated. CRT was defined as the mean distance between the upper border of the retinal pigment epithelium and the internal limiting membrane in a 1-mm diameter foveal zone. ME type was evaluated in the initial OCT images and classified as follows 5 . Serous retinal detachment was defined if subretinal fluid was present, while cystoid ME was defined as the presence of an intraretinal cystoid. The third classification type was mixed serous retinal detachment and cystoid ME. Microvascular abnormalities such as capillary telangiectasias, collateral vessels (venovenous drainage), and microaneurysms were also evaluated 9 . In particular, we investigated if microaneurysms were located near the cystoid ME or not.
Changes in intraretinal cyst size and CRT were also evaluated. Cyst responders were defined as cases where there was a reduction of ≥ 20% in preinjection cyst size after intravitreal injection. Cyst nonresponders were defined as those cases where there was cyst enlargement or a reduction in cyst size of < 20% after intravitreal injection. CRT responders were defined as patients with a reduction in CRT of ≥ 10% of preinjection CRT after www.nature.com/scientificreports/ intravitreal injection. CRT nonresponders were defined as those patients with enlargement of CRT or with a reduction in CRT of < 10% the preinjection CRT after intravitreal injection. Images were graded independently by two retinal experts (K. E. C. and S. W. K.). Final grade was determined by consensus and was used in subsequent analyses. Statistical analysis. Baseline factors were compared between cases with SSPiM and those without SSPiM using the Mann-Whitney U test, chi-squared analysis (or Fisher's exact test), Wilcoxon signed-rank test, or linear-by-linear test. Multivariate logistic analysis of significant parameters identified in the univariate analyses www.nature.com/scientificreports/ as independent variables was used to find significant factors associated with SSPiM. Statistical analyses were performed using SPSS version 21.0.0.0 (IBM, Armonk, NY, USA) or MedCalc V.12.1.3.0 (MedCalc Software bvba, Ostend, Belgium) software. All statistics were two-tailed and p values less than 0.05 were considered to be statistically significant.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.