As we and others previously demonstrated,
19–24 en face OCT angiography can directly visualize well-delineated microvascular structures of new vessels and differentiate them from original retinal vascular structures. However, the detailed morphologic features of these newly formed vessels have not been extensively characterized. The current study revealed that new vessels in patients with PDR could be morphologically divided into two main types: new vessels with EVP and those without EVP (
Fig. 1). EVP, which is the intense growth of irregular small-caliber vessels located at the margin of new vessels, likely represents active proliferation, as almost all treatment-naïve eyes with PDR (95%) had lesions with EVP (
Fig. 2). In contrast, the rate of EVP in new vessels in eyes treated with PRP was 65%, and the remaining new vessels without EVP (35%) in these previously treated eyes had only filamentous vascular loops (
Table;
Figs. 3,
4). These looped structures without EVP detected by OCT angiography were observed as whitish fibrovascular membranes without obvious red blood columns by funduscopy. The degree of leakage in FA also strongly supported the conclusion that these were indeed active new vessels. A previous study clearly demonstrated that immature (i.e., young) neovascularization had a much faster and greater leakage of fluorescein than mature (old) neovascularization.
27 The current results demonstrated that all new vessels with EVP had excessive leakage in early-phase FA, whereas 83.3% of the new vessels without EVP had faint leakage. This concordance between the presence of EVP in OCT angiography and remarkable leakage in early-phase FA indicated that the presence of EVP on OCT angiograms should be interpreted as an active sign of new vessels in clinical practice.
Corresponding to the reduction in fluorescein leakage after PRP, new vessels with EVP before PRP were predictably pruned in the vascular area and signs of EVP were decreased (
Figs. 5,
6). Moreover, the mean duration after PRP in eyes with EVP was significantly shorter than that in eyes without EVP. Previous studies clearly demonstrated that VEGF concentration in the ocular fluid of patients with active PDR was higher than that in patients with quiescent PDR,
28,29 and that VEGF levels decreased following successful laser photocoagulation.
28 Inhibition of VEGF by intravitreous injection of bevacizumab led to complete, or at least partial, resolution of leakage from new vessels in FA, with a concomitant reduction in the caliber or presence of perfused blood vessels.
30,31 Although we did not directly measure the VEGF concentrations in the ocular fluid of patients with PDR either before or after PRP in this study, we predicted that the VEGF concentrations after PRP treatment were decreased in parallel to decreased activity of new vessels and decreased EVP, as detected by OCT angiography.
Previously reported histologic findings have provided evidence on the role of VEGF in the development of EVP; for example, an angio-fibrotic switch in PDR has previously been proposed by several studies.
32,33 Injection of anti-VEGF drugs could lead to a profibrotic switch with significant reduction in the neovascular component, as determined by decreased expression of the panendothelial marker CD34 and the marked increase in the contractile elements, smooth muscle actin and collagen, of the proliferative membrane over time.
33 These histologic findings provided evidence that EVP observed by OCT angiography might represent newly growing sections of active new vessels as they are significantly affected by VEGF inhibition. Conversely, pruned new vessels with no sign of EVP likely reflect the inactive fibrotic stage of neovascular membranes.
Furthermore, Spaide
25 recently evaluated the morphologic vascular abnormalities associated with periodic antiangiogenic therapy for CNV on OCT angiograms. He proposed that high levels of cytokines, such as VEGF, led to exuberant proliferation of new vessels (i.e., angiogenesis); however, after anti-VEGF injection, the levels of free VEGF dropped precipitously with subsequent regression of the newly growing vessels, particularly those with poor pericyte coverage. Some vessels with pericyte coverage remained, and arteriogenesis led to an increase in vascular diameter with increased blood flow through the remaining vessels. The consequent increase in VEGF and repeated inhibition of VEGF enhanced pruning of new vessels, leading to the formation of large-sized, abnormal anastomotic connections similar to those observed in shunt vessels.
25 These processes occurred in many blood vessels throughout the body and were likely to occur in new vessels in the eyes of patients with PDR. Our current observations showing that new vessels with EVP were converted to pruned new vessels with residual, larger vascular loops after PRP suggested that new vessels in the eyes of patients with PDR might be at least partially undergoing these processes.
In the current study, for the first time, we circumstantially reported the growth process of NVD in PDR (
Figs. 7,
8): OCT angiography clearly demonstrated a vascular bud merging with another bud to distinctively undergo neovascularization with concurrent signs of EVP. Vascular spouting observed in this case was very similar to that of NVE from the superficial capillaries in hemicentral retinal vein occlusion, as we previously described.
34 Moreover, the exuberant NVD was pruned after the start of PRP and became a large-trunk vessel, whereas additional new vessels growing toward the macula had EVP. This patient did not have posterior vitreous detachment; thus, we predicted that the growing NVD trailed the posterior hyaloid toward the macula where laser treatment was not performed.
This study has several limitations. First, in some instances, OCT angiogram images vaguely showed the vessels of the superficial vascular plexus and/or radial peripapillary capillaries. En face segmentation of the outer border on OCT angiograms was set just below the ILM as the default setting of vitreous slab on AngioVue software, because thin fibrovascular membranes including new vessels could occasionally be misidentified as the ILM. Differentiation between neovascularization and IRMA should be considered,
9 as IRMA usually has minimal or no leakage in FA.
35 In the current study, we confirmed the presence of overlaying flow signals on OCT B-scan images in structures breaching the ILM and/or the posterior hyaloid; therefore, the potential interference of IRMA in evaluation of the extent of NVD/NVE was minimized. Second, we did not perform FA in 37.5% (3/8) of the eyes that had looped new vessels without EVP on OCT angiograms. However, in those three eyes, pruned new vessels on OCT angiograms were observed as whitish fibrovascular membranes with no obvious red blood columns on fundoscopy and were clinically diagnosed as inactive new vessels.
1,10 A third limitation was the classification of buds (sprouts) of new vessels that could sometimes be detected using FA and OCT angiography. The morphology of the vascular buds observed with OCT angiography did not resemble an exuberant form, but instead was shaped like a tuft or a small loop; thus, buds could potentially have been classified as small new vessels without EVP in this study. However, vascular buds can sprout and grow into new vessels, as that was shown in our case (
Fig. 7); thus, growing buds should be considered as active. The morphology of small new vessels by OCT angiography should be carefully evaluated, and might need to be supported with FA to determine the activity of particularly small new vessels especially. A fourth limitation was the very narrow field of view on 3 × 3 mm OCT angiography scans during screening for peripheral NVE. Although OCT angiography scans in the larger size of 6 × 6 mm can better visualize peripheral NVE,
23 the detailed features, such as the presence of EVP, are not clearly depicted due to low resolution. However, the development of wider and higher-resolution scans for OCT angiography should address this concern in the future.
In conclusion, en face OCT angiography allowed us to observe structures of retinal neovascularization at the microcirculation level and to evaluate the morphologic features of new vessels. Exuberant vascular proliferation observed in new vessels on OCT angiograms can be an active sign of neovascularization. Although a wider scan area is desired for future screening, morphologic evaluation of neovascularization using OCT angiography may be useful for clinicians to estimate the activity of each neovascularization in the eyes of patients with PDR.