of the retinal microvasculature are now available, including measurement of the foveal avascular zone (FAZ), capillary vascular density (CVD), and fractal analysis of vessel branching complexity (fractal dimensions; FD) at the superficial and deep capillary plexus, as well as RPC [34–40].
Although visualization of thin (<2 mm) structures within the scan depth of any OCT machine provides excellent images, lesions or elevated tumors with a thickness approaching or greater than the OCT scan depth can lead to segmentation errors, mirror artifacts, and generally a loss of image quality [41, 42]. Similarly, OCTA is also susceptible to the limitations above, and with a greater length of time required for imaging, it is not surprising that OCTA are more susceptible to artifacts and image degrading [43–45]. In the case of OCTA scans taken over elevated tumors, loss of vascular detail leads to diminished eye tracking capabilities and an overall loss of image quality [45]. Cennamo et al. [46] evaluated 116 patients with various choroidal tumors using EDI-OCT and OCTA imaged with the Optovue Avanti RTVue XR (Optovue Inc., Fremont, CA, USA). The authors took images centered on each tumor, and reported its features at the level of the outer retina (outer plexiform layer to Bruch’s membrane) and the choriocapillaries (30 μm below Bruch’s membrane to 30–60 μm below the RPE) using manual segmentation [46]. They reported a normal avascular outer retina and normal choriocapillaries in eyes with choroidal nevus and optic nerve melanocytoma, avascular outer retina and a choriocapillaries flow void in choroidal metastasis, and greatly vascular outer retina and choriocapillaries with choroidal melanoma, choroidal hemangioma, and choroidal osteoma [46]. However, review of their methods and images reveal several inconsistencies. First, the mean tumor thickness for each tumor ranged from 1.23 mm for choroidal nevi to 3.49 mm for choroidal metastasis; hence, it is not surprising that there is an absence of blood flow in their cases of choroidal metastasis, as 3.49 mm is beyond the maximum scan depth of the Optovue Avanti RTVue XR machine (approx. 3.0 mm) and any segmentation will generate errors from loss of signal at the tumor apex or base, depending on the point of focus [46]. Second, several of their figures also show projection artifacts and areas of RPE thinning/atrophy on OCT, which are well-described masqueraders in OCTA [46, 47]. Lastly, the authors themselves stated that their study had several limitations, and “many OCT-angiography scans were of insufficient quality” but did not report on the mean signal strength, frequency of artifacts, or percentage of useable images in their series [46]. Therefore, current commercially available OCTA machines are unable to consistently generate OCTA images of sufficient quality to allow analysis of CVD, FD, or even intratumoral vessels over elevated tumors.
Fig. 1. Comparative analysis of CVD of affected and fellow eyes between patients with unilateral choroidal nevus and melanoma. Fundus photo of 2 patients, one (a) with a superotemporal choroidal nevus with fibrous metaplasia and no subretinal fluid or orange pigment, while the other (b) had a nasal choroidal melanoma (image not focused due to the tumor height), with sparing of the macula. Both patients had affected right eyes. Comparison of normal fellow (c, d, left) and affected (c, d, right) superficial CVD maps show no significant change in the choroidal nevus (c), but significant reduction (cooler blue color) in choroidal melanoma (d). e, f Comparison of normal fellow (e, f, left) and affected (e, f, right) deep CVD maps show no significant change in the choroidal nevus (d), but significant reduction (cooler blue color) in the choroidal melanoma (f).
Despite the limitations in scan depth, OCTA is still useful in the evaluation of choroidal melanomas by studying its effects on the parafoveal microvasculature. Li et al. [36] studied 30 consecutive patients with treatment-naïve unilateral choroidal melanoma and found no significant difference in size of the superficial plexus FAZ in melanoma-affected compared to fellow eyes (0.242 vs. 0.251 mm2, p = 0.316), but significantly enlarged deep plexus FAZ (0.546 vs. 0.354 mm2, p < 0.001) in eyes with melanoma. Furthermore, they found a significant reduction in superficial (49 vs. 52%, p = 0.004) and deep (54 vs. 57%, p < 0.001) plexus CVD in melanoma-affected compared to paired fellow eyes, in parallel to the significantly greater CMT (312 vs. 266 μm, p < 0.001) [36]. The parafoveal microvascular changes on OCTA, similar to subclinical macular edema, are likely explained by the presence of pro-inflammatory cytokines in eyes harboring uveal melanoma [36]. Missotten et al. [48] and others have found increased VEGF in eyes with uveal melanoma, which increases in parallel to tumor basal diameter and thickness [48–50]. These parafoveal changes, however, are not isolated to choroidal melanoma, as FAZ enlargement and reductions in CVD are also found in eyes with diabetic retinopathy and retinal vein occlusion, both of which are also associated with high VEGF [51–58]. Hence, past medical history and prior ocular pathology should be taken into account when analyzing changes in parafoveal microvasculature using OCTA. In terms of the RPC, Skalet et al. [59] imaged 10 eyes with uveal melanoma before plaque radiotherapy, but did not find any significant differences compared to the fellow eyes.
Table 1. OCTA in the evaluation of choroidal melanoma and management of radiation retinopathy
FAZ enlargement and reduction in parafoveal CVD could be useful in distinguishing between melanomas and pseudomelanomas (Fig. 1; Table 1). Valverde-Megias et al. [37] compared parafoveal OCTA features between affected and fellow eyes in patients with unilateral unifocal choroidal nevus and choroidal melanoma, reporting no significant difference in CMT, FAZ, and CVD in patients with choroidal nevi, but increased CMT and FAZ, with decreased CVD in melanoma-affected eyes. Sioufi et al. [60] described parafoveal OCTA features of 14 eyes with circumscribed choroidal hemangioma and reported significantly increased CMT (312 vs. 264 μm, p = 0.042), no significant change in superficial (0.254 vs. 0.298 mm2, p = 0.327) and deep FAZ (0.414 vs. 0.381, p = 0.563), no significant change in superficial CVD (48 vs. 51%, p = 0.159), but decreased deep CVD (50 vs. 54%, p = 0.010) in hemangioma-affected compared to fellow eyes. However, subgroup analysis showed that only eyes with choroidal hemangioma that had previous/current CME and/or subretinal fluid had a reduction in deep plexus CVD, while eyes that never had documented