Clinical applications of optical coherence tomography angiography: What we have learnt in the first 3 years

Age-related macular degeneration

In dry age-related macular degeneration (AMD), OCTA has revealed an impaired choriocapillaris flow in areas of nascent geographic atrophy (GA) or drusen-associated GA in all eyes studied in a small series. ⁴ A patchy thinning of the choriocapillaris has also been observed. ⁵ In early and intermediate AMD, the superficial and deep retinal vessel density has also been reported to be less than that of controls,^6,7 but other authors have not found any difference in retinal vascular layers in the case of GA compared to controls. ⁸ In more advanced GA, OCTA has systematically shown a marked reduction in choriocapillaris blood flow in the affected areas, while the margins of atrophic lesions may also present alterations in the choriocapillaris.^5,9

Most studies have, however, focused on exudative AMD. Since the first reports, OCTA has been shown to noninvasively allow visualizing the abnormal blood flow corresponding to neovascularization.^10–12 The sensitivity was initially limited, but improved very rapidly. In type 2 neovascularization, a 100% sensitivity has been reported in some series.^13,14 The authors have described different neovascularization patterns, namely, a “medusa-shaped” or “glomerulus-shaped” lesion, and shown an excellent correlation with images obtained by fluorescein angiography (FA). In type 3 neovascularization, the lesions were inconsistently recognized with the first versions of the devices, ¹⁵ but their recognition improved thereafter to reach 100% sensitivity in recent studies.^16–18 Retinal angiomatous proliferation lesion has been described as a tuft located in the outer retina, and OCTA has frequently allowed showing retinal–retinal anastomosis in the early stages of the disease.^15,17 In type 1 neovascularization, different interesting findings have been reported. Choroidal neovascularization (CNV) usually had a tangled morphology. Vessels were considered more or less mature depending on the number of small capillaries at the lesion periphery. ¹⁹ The neovascularization size was smaller when analyzed by OCTA than by indocyanine green angiography (ICGA). ²⁰ More interestingly, OCTA has helped to diagnose nonexudative or “quiescent” type 1 neovascularization in asymptomatic eyes.^21–24 The management of these lesions is still debated. OCTA sensitivity has been considered good for the diagnosis of type 1 CNV. Compared to FA and structural OCT, 87.5% sensitivity was measured in a large prospective multicentric study. ²⁵ Regardless of the type of neovascularization, swept-source OCTA (SS-OCTA) seems to show a larger lesion than spectral-domain OCTA (standard deviation (SD)-OCTA). ²⁶ End-stage neovascularization, known as subretinal fibrosis, has also been analyzed by OCTA, and a vascular blood flow was visible within the fibrotic scar in more than 90% of cases, while exudative signs had disappeared. The vessels had a typical pattern with a pruned vascular tree that was more commonly found than a tangled network pattern. Vascular loops, large flow voids, and dark halos have also been reported. ²⁷

The analysis of the different patterns observed by OCTA has led to conduct studies aimed to correlate the morphology and the neovascular activity, frequently assessed based on the presence of exudative manifestations observed on SD-OCT B-scans. In a prospective series of 80 eyes, the authors have shown that some morphological signs were usually associated with a neovascular activity such as branching, numerous tiny capillaries, vascular peripheral arcade, or a dark perilesional halo. Conversely, the absence of previous signs or a dead tree pattern generally corresponded to inactive lesions. ¹² Others have proposed to classify the OCTA patterns into “well circumscribed” or “poorly circumscribed” and, on the contrary, have observed no significant correlation between these patterns and the clinical activity. ²⁸

After treatment with anti-vascular endothelial growth factor (VEGF) agents, all studies have shown a reduction in neovascular lesion size and vascular density.^12,29,30 The morphology of the neovascular lesion significantly changed after treatment, usually with capillary rarefaction at the vascular network periphery. A disappearance or reduction of the dark halo has also been reported.^31,32 An arteriolization of the neovascular lesion has been described as a vascular abnormalization, meaning that the small capillaries were reduced, while the main trunks of the vascular network became larger and developed many direct vascular anastomotic connections. These findings could explain the chronicity of the neovascular process observed in wet AMD treated with anti-VEGF agents. ³³

Many questions remain regarding the role of OCTA in wet AMD. Is OCT, both structural OCT and OCTA, sufficient for the diagnosis of wet AMD, allowing to prevent the use of invasive tools? Are the changes observed by OCTA sufficient to re-treat an eye with anti-VEGF agents in the absence of hemorrhages or recurrence of exudative manifestations? Could OCTA play a role in future clinical trials as recently discussed? ³⁴ Further studies are needed to answer all these questions.

Polypoidal choroidal vasculopathy

Studies assessing polypoidal choroidal vasculopathy (PCV) with OCTA have shown that polyps were inconsistently detected and that OCTA appeared to be less effective than ICGA for their visualization. Indeed, the detection rate of polyps with OCTA ranged from 40% to 75% of cases.^35–38 Polyps located just beneath the retinal pigment epithelium (RPE) were the more difficult to detect. ³⁹ Manual segmentation could more effectively allow visualizing them than automatic segmentation. ⁴⁰ Others have suggested analyzing the flow on OCTA B-scans rather than on en face images. ⁴¹ When visible, polyps have been described as a hyperflow round structure surrounded by a hypointense halo or as a hypoflow round structure. ³⁵ Unlike polyps, the branching vascular network observed in PCV was more easily recognized with OCTA than with ICGA in all studies.^35–42

Based on the current literature, OCTA does not seem to replace ICGA for the diagnosis of PCV. Further studies are thus needed to assess its possible role in the monitoring of the disease.

Pathologic myopia

In pathologic myopia, OCTA has shown a reduced retinal vascular density in the peripapillary atrophic area of eyes with tessellated fundus, compared to controls. ⁴³ Posterior scleral reinforcement did not change the macular flow density. ⁴⁴ Another study has reported numerous segmentation artifacts in myopic eyes, occurring in 73.9% of cases in a series of 92 eyes. The authors have suggested to position the segmentation layer behind the choroid, within the sclera, to better visualize the choroidal blood flow. ⁴⁵

However, most studies have focused on the diagnosis of CNV. The detection rate of CNV was 94% with OCTA versus 100% with FA. ⁴⁶ OCTA has been considered by authors as sufficiently effective to prevent the need for using FA in most cases of myopic CNV. However, one CNV with a very small surface (0.01 mm²) could not be detected by OCTA. ⁴⁶ SS-OCT has also been considered to be superior to SD-OCT for the detection of CNV in a case report. ⁴⁷ Two studies have described the pattern of myopic CNV examined with OCTA. The authors have found that CNV had one of the two following patterns: “interlacing” or “tangled.” ⁴⁸ Most active CNV (81.8%) had an “interlacing” pattern, while two-thirds of inactive CNV had a “tangled” pattern. Thus, recognizing these patterns could help to assess the neovascular activity. ⁴⁹

In summary, OCTA appears to be a useful method for the detection of CNV in pathologic myopia. However, further studies are needed to determine its sensitivity and specificity.

Heredodegenerative disorders

Heredodegenerative disorders are a very heterogeneous group of hereditary chorioretinal diseases. A few studies have used OCTA in these disorders.^3,50–56 In Stargardt’s disease, OCTA has revealed a very typical pattern known as “dark atrophy,” characterized by the complete disappearance of the choriocapillaris. This pattern contrasted with that of eyes with GA in which the choriocapillaris was rarefied but present. ⁵⁰ A second study has also shown a rarefaction of the superficial and deep vascular retinal plexuses. ⁵¹ In choroideremia, OCTA has allowed analyzing separately the retinal and choroidal vascular dysfunctions and could, in the future, help to predict progression and therapeutic response. ³ Detection of CNV was allowed by OCTA in one patient with choroideremia, as in three other dystrophies, that is, EFEMP1-related retinopathy, Best vitelliform dystrophy, and adult-onset foveomacular vitelliform dystrophy (AOFVD). ⁵² Two additional studies focused on AOFVD. They reported vascular rarefaction in the superficial and deep capillary plexuses, but also in the choriocapillaris layer, and the possibility of visualization of CNV complicating the condition in most cases^53,54 There were no retinal vascular-specific changes in X-linked retinoschisis. ⁵⁵ In retinitis pigmentosa, a large study of 32 eyes compared the vascular plexuses with normal subjects. The study showed a reduced vascular density in the superficial and deep capillary plexuses and in the choriocapillaris. Most important defects were found at the level of the deep capillary plexus. ⁵⁶

Central serous chorioretinopathy

Central serous chorioretinopathy (CSC) is characterized by an acute or chronic serous retinal detachment, usually associated with a pachychoroid, and complicated by areas of RPE changes and atrophy. In acute CSC, one or more leaking points are seen on FA. Chronic CSC may be mistaken for type 1 neovascularization of AMD, and CNV may complicate chronic CSC. The most difficult cases to be diagnosed are characterized by a shallow or flat irregular RPE detachment in which the diagnosis of CNV is frequently doubtful.^57,58

Studies assessing CSC with OCTA have shown that no specific change was associated with the leaking point. ⁵⁹ The choroidal flow has been reported to be normal in some patients, ⁵⁵ but was often impaired.^59–62 The affected areas corresponded to staining or hypofluorescence areas observed on FA and ICGA. ⁶¹ More precisely, studies have described dark areas and dark spots, and slightly enlarged choroidal capillaries, well visible within the dark areas.^60,62 The findings may be difficult to differentiate from occult or type 1 CNV and should be interpreted with caution. ⁶² However, despite these limitations, OCTA appeared to be very useful to diagnose CNV complicating CSC.^60,62,63 OCTA sensitivity was considered to be as good as that of OCT combined with FA, ⁶⁴ and the frequency of CNV was considered higher than previously reported. For example, OCTA allowed diagnosing CNV in 13 (41.9%) out of the 31 eyes with irregular RPE detachments. ⁶⁵

In summary, the diagnosis of acute and chronic CSC is not based on OCTA. It is still based on the clinical examination and OCT combined with angiography. However, in chronic CSC, especially when a flat or shallow irregular RPE detachment is present, OCTA is very helpful to show an abnormal blood flow corresponding to CNV complicating the disease.

Chorioretinal tumors

Retinal astrocytic hamartomas are benign intraocular tumors described as a yellow–white mass, often vascularized, with apparent continuity between the mass and the inner retinal layers. It may be isolated or part of a tuberous sclerosis complex. In one study and one case report, OCTA has shown the presence of a dense vascular network within the tumor.^66,67 Thus, when the clinical examination and SD-OCT findings are consistent with the diagnosis of retinal astrocytic hamartoma, the presence of a vascular network observed on OCTA may support the diagnosis without using FA.

Choroidal osteomas are ossifying tumors mainly observed in the peripapillary area in young women. CNV is the main complication of the disease. Two case reports have shown that OCTA allowed demonstrating the presence of CNV as a blood flow corresponding to dilated vessels with a medusa appearance.^68,69 In one of these cases, OCTA has been identified as superior to FA for the detection of CNV. ⁶⁹

However, most studies have focused on treatment-naive or treated choroidal melanoma. After plaque radiotherapy, eyes usually develop radiation maculopathy that may be diagnosed with FA and OCT. But OCTA has allowed visualizing a decrease in capillary density in the parafoveal superficial and deep plexuses in irradiated eyes, while it was not visible with the other tools, ⁷⁰ despite the presence of numerous artifacts. ⁷¹ Additional studies have confirmed that radiation retinopathy may be diagnosed before any change is observed on FA or SD-OCT.^72,73 Based on these findings, a new classification of radiation retinopathy, including OCTA changes, has been proposed. ⁷³ However, another study revealed that the decreased capillary density was also observed in eyes with melanoma, before any radiation therapy. ⁷⁴ Thus, OCTA has also been used to differentiate eyes with melanoma from eyes with benign choroidal nevus. Compared to the fellow eyes, eyes with nevus showed no difference in the foveal avascular zone (FAZ) or capillary vascular density. Conversely, eyes with choroidal melanoma showed a significantly decreased capillary vascular density, in both the superficial and deep plexuses, and a significantly enlarged FAZ compared to fellow eyes. ⁷⁵

In summary, studies assessing choroidal melanoma with OCTA have revealed surprising findings. Indeed, while a precise analysis of capillaries in radiation retinopathy/maculopathy with OCTA was expected, the presence of perifoveal capillary changes in eyes with melanoma, before any radiation, is quite surprising. If these findings are confirmed, they could have many clinical implications for the monitoring of eyes with choroidal nevi and perhaps the diagnosis of challenging cases of malignant transformation.

Chorioretinal infections and inflammations

To date, most studies assessing infectious/inflammatory chorioretinal disorders with OCTA are case reports, but they have shown interesting findings for their diagnosis or management. In Birshot chorioretinopathy, a study in four patients has shown a decrease in choroidal flow in depigmented areas. The analysis of retinal capillaries has also shown capillary dilations and loops, abnormal telagiectatic vessels, and increased intercapillary spaces. ⁷⁶ A case report has also described a decrease in plexus capillary density, primarily in the deep plexus. ⁷⁷

In Vogt–Koyanagi–Harada disease, OCTA results have been compared to ICGA findings in the acute phase of the disease. OCTA has shown multiple dark areas, with flow void that correlated with the ICGA findings. Interestingly, some areas disappeared after treatment initiation and were interpreted as inflammatory foci suggestive of choriocapillaris hypoperfusion. ⁷⁸

In acute posterior multifocal placoid pigment epitheliopathy (APMPPE), four case reports have demonstrated that in the acute phase, multiple areas of choriocapillaris and inner choroid non-reduced or reduced perfusion were present.^79–82 Over time, these areas reperfused, but other areas were involved in the recurrence of the inflammatory process. The authors have described relatively rapid changes in affected areas. ⁸²

In serpiginous choroiditis, OCTA has shown dark areas correlating with the intermediate- and late-phase hypoperfusion observed on ICGA. ⁸³

In multiple evanescent white dot syndrome (MEWDS), while en face OCT appeared very useful, the OCTA analysis of the retinal and choroidal vasculature was unremarkable. The authors have concluded that the disease could primarily affect the photoreceptor outer segments without altering the retinal and choroidal vascular networks. ⁸⁴ However, OCTA was useful for the diagnosis of CNV complicating a case of MEWDS. ⁸⁵

In multifocal choroiditis (MFC) and punctate inner chorioretinopathy (PIC), the diagnosis of CNV that frequently complicates these disorders may be very challenging. Indeed, the findings of the clinical examination and multimodal imaging (MMI) are often not conclusive, because both CNV and inflammatory lesions appear as white to yellow deep chorioretinal macular lesions with early iso- or hyperfluorescence, and late staining and leakage, and appear as a hyper-reflective material infiltrating the subretinal space on OCT. OCTA has been shown to be very effective in differentiating inflammatory lesions from CNV, showing in the latter case an abnormal blood flow in active and scarred lesions.^86,87

In Behçet’s uveitis, a large series including 25 patients has shown that perifoveal microvascular changes were more commonly observed on OCTA than on FA. These anomalies included a disruption of the perifoveal capillary arcade, areas of retinal capillary nonperfusion or hypoperfusion, and rarefied or dilated vessels. The FAZ was also enlarged compared to controls. ⁸⁸

In infectious chorioretinal lesions, OCTA findings were reported in tuberculosis, more precisely in tubercular serpiginous-like choroiditis. Studies have shown discrete areas of flow void beneath the RPE, suggestive of choriocapillaris hypoperfusion. These areas were better defined on OCTA than the nonperfused areas visualized on ICGA. ⁸⁹ The same authors have also shown that flow void areas were increased after antituberculosis treatment initiation. This phenomenon, known as paradoxical worsening, could be precisely analyzed with OCTA. ⁹⁰

In other infectious conditions, OCTA has been used for the diagnosis of CNV in cat-scratch disease due to Bartonella henselae ⁸⁴ and in toxoplasmosis. ⁹¹

Retinal vascular diseases

Other retinal vascular disorders

In Coats’ disease, OCTA has shown that inner retinal vessels passing through the FAZ were commonly observed in affected eyes, but also in nonaffected fellow eyes. ¹³¹ OCTA has also been used to monitor recurrence and assess laser treatment efficacy in hemangioblastomas of von Hippel–Lindau disease.^132,133 A case of type 2 retinal arteriovenous malformation has also been imaged with OCTA. ¹³⁴

In summary, retinal vascular changes are easy to diagnose with OCTA. The technique is considered superior to FA in all conditions affecting the deep capillary plexus. It has improved the understanding of the condition known as PAMM. The need for using FA is even questionable in the diagnosis of Mactel 2 because OCTA clearly and noninvasively shows changes in the retinal vasculature in the temporal area of the fovea. In incomplete RAO and RVO, OCTA may help to assess the visual prognosis, because some anatomical features have been correlated with the loss of photoreceptors and/or visual acuity.

Diabetic retinopathy and maculopathy

Diabetes may affect the retinal capillaries in various ways, leading to their increased permeability, capillary defects, and the occurrence of vascular sprouting and retinal neovascularization. Thus, the noninvasive visualization of the different layers of retinal capillaries is essential to the diagnosis and monitoring of diabetic patients.

OCTA has allowed easily diagnosing neovascularization in the disk or other locations.^135,136 However, most studies have focused on the analysis of diabetic maculopathy. Initial studies have shown that OCTA could clearly visualize microaneurysms and retinal nonperfused areas and showed FAZ enlargement and distortion. The technique did not show the capillary leakage, but clearly delineated areas of capillary changes.^137,138 The cystoid spaces in diabetic macular edema (DME) appeared devoid of flow and were easily differentiated from areas of capillary dropout, ¹³⁹ despite the fact that cystoid spaces were located within the capillary dropout areas. ¹⁴⁰

The analysis of patients without apparent retinopathy has shown that the presence of an unsuspected FAZ enlargement was common,^141–143 and OCTA sometimes revealed retinal microvascular abnormalities and an impaired choriocapillaris flow. ¹⁴⁴ Studies comparing OCTA and FA have considered that FA was more effective to diagnose microaneurysms, but less effective to detect capillary nonperfused areas. ¹⁴⁵ However, others have also shown that microaneurysms located in the deep capillary plexus could be better imaged with OCTA ¹⁴⁶ and that their presence was associated with DME. ¹⁴⁷ The analysis of the FAZ was also more accurate with OCTA than with FA. ¹⁴⁸ OCTA has then been considered as a valid and reliable tool for the analysis of ischemic diabetic maculopathy.^149,150 For all these reasons, OCTA is now considered a new step forward for the detection of diabetic macular ischemia. ¹⁵¹ Its reproducibility seems to be good, allowing obtaining gradable images^152–154 and quantification of capillary vascular density. ¹⁵⁵

More interestingly, a correlation has been found between capillary changes observed with OCTA and macular photoreceptor disruption ¹⁵⁶ and diabetic retinopathy severity.^157,158 The FAZ enlargement also correlated with visual acuity loss.^159,160 The presence of damage to the deep capillary plexus has also been associated with a poor response to intravitreal anti-VEGF agents. ¹⁶¹

The next step might be the automatic detection and quantification of diabetic macular changes. Various teams have published interesting results on this topic and shown a high detection rate using automatic algorithms.^162–164

For all these reasons, OCTA is undoubtedly a very useful tool for the early detection of lesions and monitoring of disease progression in diabetic patients. ¹⁶⁵

Conclusion

OCTA is obviously a promising technique. OCTA may allow the diagnosis of CNV in cases where FA and/or ICGA are not conclusive, including in CNV complicating CSC, MFC, or in pathologic myopia. OCTA could also replace FA for the initial diagnosis of typical exudative AMD and prevent the use of invasive tools in elderly patients. FA utility may also be challenged to diagnose and monitor Mactel 2. OCTA also allows the early diagnosis of diabetic maculopathy, and a noninvasive monitoring of diabetic maculopathy and RVO seems possible and easy. OCTA has also improved the identification of incomplete RAO, a condition initially described as PAMM. If confirmed, the observation of a reduced retinal vascular flow in eyes with choroidal melanoma could also have great implications in the monitoring of eyes with choroidal nevus.

The first 3 years of OCTA use have been full of discoveries and suggest its relevant applications, not only in research but also in the daily clinical practice. However, further studies are needed to improve and standardize OCTA image acquisitions, and to better assess its role in the multimodal diagnosis of chorioretinal diseases.