described similar changes in 37 eyes with small choroidal melanoma treated with proton beam radiation. In contrast to other studies, the authors only included those that received the full radiation dose (60 Gy) to the macula and found OCTA abnormalities (FAZ disruption, non-perfusion, vascular dilation, microaneurysms, intraretinal flow voids, or cysts) in all eyes [77]. In contrast, only 47% had OCT-evident radiation retinopathy and 77% had clinically evident radiation retinopathy [77]. The authors also described “signal voids” at the level of the choriocapillaries in 88% of eyes, as well as a significant reduction in retinal global CVD (combined superficial and deep plexus CVD, 23 vs. 26%, p < 0.001) and choriocapillaries CVD (79 vs. 99%, p < 0.001) compared to matched healthy controls [77]. Matet et al. [40] also described OCTA features in 35 eyes with uveal melanoma treated with proton beam radiation with at least 12 months post-radiation follow-up using paired fellow eyes as controls. They found similar enlargement of FAZ and CVD at both superficial and deep plexuses, along with reduction in FD (1.83 vs. 1.91, p < 0.001 and 1.82 vs. 1.90, p < 0.001) at the level of the superficial and deep plexus in irradiated eyes relative to fellow control eyes [40]. Retinal FD is a measurement of vessel branching complexity, and reflects the overall health of retinal microvasculature [38, 39]. Changes in FD have been described in diabetic retinopathy and reduction in FD after radiation refers simply to a diminished vascular branching pattern [40, 56, 57]. It remains to be determined whether changes in FD occurs earlier, later, or in parallel to changes in CVD and FAZ. Lastly, Skalet et al. [59] analyzed the RPC in eyes with choroidal melanoma before and after plaque radiotherapy and found a significant reduction in mean RPC CVD between treated compared to fellow eyes (53 vs. 73%, p < 0.004). Furthermore, they also found a significant inverse linear correlation between optic disc radiation dose and RPC CVD (r = –0.528, p = 0.043), as well as visual acuity and RPC CVD (r = –0.564, p = 0.028) [59]. This reduction in RPC CVD was seen only after radiation, in contrast to parafoveal CVD, wherein reductions can be appreciated even prior to treatment [36, 37].
Fig. 2. Comparison of NVD by OCTA and IVFA. a Fundus photo of NVD associated with choroidal melanoma. Optic nerve head OCTA (b) with segmentation through the full thickness of the optic nerve indistinctly shows the NVD, while adjustment of segmentation at the level of the vitreous (c) clearly identifies the full extent of NVD. IVFA of the same eye shows the choroidal melanoma incidentally at the superotemporal quadrant, and the NVD with early hyperfluorescence (d) and late leakage (e). Leakage can only be seen by IVFA, but its early hyperfluorescence and profuse leakage, as well as poorer resolution, does not allow clear visualization of NVD.
OCTA can also be useful as a guide in treatment planning and monitoring treatment response. When segmentation is set at the pre-internal limiting membrane or vitreoretinal interface, visualization of preretinal neovascularization is enhanced [78]. Given its improved contrast compared to color photos, it can enable the physician to monitor regression after treatment [78]. However, it is important to realize that OCTA cannot measure vascular leakage like IVFA (Fig. 2), hence, regression can only be assessed according to change in area or size of neovascularization or a change in appearance [28, 29, 78]. Ishibazawa et al. [78] used OCTA to describe two distinct OCTA patterns of neovascularization of the disc (NVD) and neovascularization elsewhere (NVE) in eyes with proliferative diabetic retinopathy that correlated with vascular activity (leakage) on IVFA. The authors described eyes with exuberant vascular proliferation (EVP), defined by the authors as “irregular proliferation of fine vessels,” correlated with eyes that were treatment naïve exhibiting extensive early leakage on IVFA [78]. In contrast, NVD/NVE that did not exhibit features of EVP, but instead had “pruned vascular loops of filamentous new vessels,” correlated with eyes that had prior panretinal photocoagulation with only faint leakage on IVFA [78]. Aside from visualization of NVD or NVE, use of longer scan lengths (up to 8 mm) can also be used to detect areas of capillary dropout for planning of additional fill-in laser photocoagulation in cases of poorly responsive proliferative diabetic retinopathy (Fig. 3). Keep in mind that in contrast to IVFA, the field of view is limited with OCTA, but it can be beneficial in eyes with active NVD or NVE wherein extensive neovascular leakage can sometimes mask the extent of neovascular tissue [79]. Lastly, because OCTA only allows visualization of microaneurysms, NVD/NVE or non-perfusion, without clearly defining its activity (leakage), OCTA cannot be used in treatment planning for focal laser [79]. Considering their differences, OCTA is not meant as a replacement for IVFA, and should be considered an adjunct. A summary of the advantages and disadvantages of IVFA and OCTA is presented in Table 2.
Fig. 3. Comparison of non-perfusion and leakage in proliferative radiation retinopathy by OCTA and IVFA. a Fundus photo of treated choroidal melanoma with active proliferative radiation retinopathy despite laser. b Montage of multiple OCTA images does not allow visualization of the full extent of neovascularization. c A single wide-field IVFA image showing the entire region of non-perfusion, and a small hyperfluorescent patch of neovascularization at the disc not seen on OCTA.
Table 2. OCTA in the evaluation of choroidal melanoma and management of radiation retinopathy: OCTA versus IVFA for evaluation and management of choroidal melanoma and radiation retinopathy
Current Limitations and Future Direction of OCTA for Choroidal Melanoma and Radiation Retinopathy
Despite advances provided by OCTA in the differentiation of choroidal melanoma and pseudomelanomas, and particularly in our understanding of radiation retinopathy, certain limitations still exist. First, OCTA is very dependent on image quality and the limitation of image artifacts [43–45].