Introduction

Age-related macular degeneration (AMD) is the main cause of visual impairment in developed countries in individuals with more than 50 years of age [1, 2]. Multimodal imaging has led to a remarkable improvement in our understanding of the physiopathology of macular diseases. Combining the different imaging modalities has allowed the retinal specialists not only to understand disease progression but also, most importantly, to monitor treatment response. Fluorescein angiography (FA) is the method of choice for classifying choroidal neovascularization (CNV) as classic, occult, or combination subtypes, essential in the era of photodynamic therapy [3], but continue to be useful for detection of “active” neovascularization. It is an irreplaceable method to analyze and monitor leaks and compare them with staining and window defects. In contrast to FA, indocyanine green angiography (ICGA) uses a dye that is 98% protein bound, providing more detailed images of the choroid and thus identifying the entire extension of the CNV. However, both modalities are invasive, requiring intravenous dye injection, which can cause some side effects such as nausea and anaphylaxis [4, 5]. Moreover, FA and ICG are time consuming in a routine use and can only capture early transit in one eye. Optical coherence tomography (OCT) has revolutionized the diagnostic approach and treatment of macular diseases. Particularly in AMD, OCT is able to detect the presence of drusen, retinal pigmentary epithelium (RPE) atrophy, fibrovascular complex, and sub- and intra-retinal fluid, among other features [6].

Optical coherence tomography angiography (OCTA) has been studied for more than 10 years but only recently became available to retinal specialists. Unlike traditional angiography, OCTA is safer and non-invasive, because it does not require the use of exogenous dyes. It is a novel imaging modality that allows direct visualization of the retinal and choroidal vasculature in vivo without dye leakage and staining that could possibly obscure the limits and anatomy of pathologies [7]. OCTA obtains multiple B-scans at the same position and detects erythrocyte movement by analyzing the signal decorrelation between scans [8]. It is fast, which is an enormous advantage to other methods when evaluating patients that require frequent visits. It is three-dimensional, unlike the two-dimensional image of FA and ICGA, allowing visualization of the exact localization and the dimensions of a lesion [7]. Structural en face OCT combines OCT and monochromatic or angiographic confocal scanning laser ophthalmoscope analysis. The en face acquisition areas range from 2 × 2 mm to 12 × 12 mm. The scan quality decreases with a widened field of view because of the limited number of OCT B-scans for all scanning areas (Fig. 1). This range has been quite sufficient for analysis of AMD since the 3 × 3mm central scan encompasses most of the AMD disease. In other hand, this new technology is prone to several image artifacts, like shadow, projection, and motion artifacts [9•, 10•, 11,12,13]. OCTA can also miss areas of slow or turbulent blood flow below its minimum threshold of detectable flow [9•, 10•]. Furthermore, there is a difference between the two main technologies of OCTA, the spectral domain (SD-OCTA) and the swept source OCTA (SS-OCTA). Most of the commercial devices are SD-OCTA and uses a shorter wavelength, approximately 840 nm, which is substantially attenuated by the RPE. The sub-RPE tissue attenuation may be more important in drusen or RPE thickening, since the RPE is an important hyperreflective barrier, making it difficult for the light to pass through it [14,15,16]. The SS-OCTA uses a longer wavelength, approximately 1050 nm, thereby being able to penetrate deeper into the choroid. The different light source also contributes for the improved visualization of the sub-RPE structures [15].

Fig. 1
figure 1

Spectral-domain (SD) optical coherence tomography angiography of a type 1 choroidal neovascularization from one eye of patient with exudative age-related macular degeneration. a 3 × 3 mm en face flow angiogram extracted from the slab shown on c revealing the neovascular network of a type 1 choroidal neovascularization (CNV). b 6 × 6 mm en face flow angiogram extracted from the slab shown on d. Differently from the 3 × 3 mm angiogram shown on A, the CNV is almost indistinguishable from the surrounding choriocapillaris due to lower resolution of the 6 × 6 angiogram. c, d B-scan from which the 3 × 3 and 6 × 6 mm angiogram on a and c, respectively, were extracted demonstrating the boundaries of the slab; top boundary under the retinal pigment epithelium and bottom boundary under the Bruch’s membrane

Full size image

OCTA of Dry AMD

Dry AMD accounts for 85–90% of all cases of AMD [17]. It is characterized by drusen, pigmentary changes, and photoreceptor and RPE loss. OCTA images in early dry AMD eyes suggest that there is a general reduction in choriocapillaris (CC) density in comparison to age-matched normal controls, interpreted as patches of CC loss or flow impairment [18]. It is important to remind that OCTA may produce false-positive images of flow impairment mostly in early AMD and under drusen when the RPE is healthy or thickened enough to block the signal from the CC. This is called shadow or masking artifact and is more prone to happen in SD-OCTA devices. Therefore, one should be really careful when stating there is a reduction in the flow of sub-RPE layers, especially in SD-OCTA images [9•, 10•, 12, 15].

With the progression of the dry AMD, the development of geographic atrophy (GA) occurs with increasing photoreceptor and RPE loss. In patients with GA, OCTA shows worsening of CC flow impairment under the regions of GA and this alteration may not be limited to the area of GA with evidence of flow impairment beyond that area [19]. The changes underlying the area of GA are usually well visualized on both spectral domain and swept source OCTA since the hyperreflective barrier of the RPE is weakened in patients with GA and does not attenuate the OCTA signal.

Another important finding in dry AMD is the detection of quiescent, non-exudative, treatment-naïve CNVs [7, 20,21,22]. The existence of these lesions was described before in histological specimens and with ICGA [23, 24], and with OCTA, they are quickly detected under unsuspicious RPE elevations and drusen. We emphasize that a proper analysis of OCTA images is mandatory to avoid false-positive images. The projection artifact over drusen and RPE detachments can be easily misinterpreted as a CNV [16, 25]. Until natural history data of quiescent, non-exudative, naïve CNVs is known, the recommendation is not to treat them. There is a possibility that these asymptomatic lesions may be supporting nutritionally the RPE and photoreceptors, thus anti-VEGF therapy could induce macular atrophy [26]. Also, there is a lack of a goal for the therapy or how frequent the treatment should be performed.

OCTA of Exudative AMD

In 2014, Jia et al. first described the ability of a prototype SS-OCTA system to visualize and quantify CNV [27]. Several studies followed Jia’s confirming OCTA’s capacity of detection of all three types of CNV [28,29,30,31,32]. The sensitivity of the OCTA in detecting CNV lesions varies from 50 to 100% [27,28,29]. Morphologically, a variety of CNV formats have been described. The medusa shape corresponding with vessels radiating from a central trunk vessel to all directions (Fig. 2a); the seafan shape, with vessels branching to one side of the main trunk (Fig. 2b); the glomerular shape, defined as globular structures with intertwined vessels (Fig. 2c); and the “dead tree” shape, composed of a main pruned vascular trunk (Fig. 2d) [31, 33]. These patterns still have questionable clinical relevance.

Fig. 2
figure 2

En face flow angiograms from different morphologic patterns of choroidal neovascularization (CNV). a Medusa shape, showing vessels radiating from the center to all directions. b Seafan shape, presenting with vessels branching to one side of the main trunk. c Glomerular shape, defined as globular structures with intertwined vessels. d “dead tree” shape, composed of a main pruned vascular trunk

Full size image

In comparison to FA, OCTA proved to be a valuable tool for evaluation of CNV activity, even not recognizing leakage [34]. Many studies reported a decrease on the area of the CNV after treatment with anti-VEGF [35,36,37,38]. Others found persistence of the neovascular network after therapy, even long time after injection in quiescent CNVs and macular fibrosis [38,39,40,41]. Several artefactual pitfalls on OCTA can result in misinterpretation of CNV size change. First, the quality for detection of neovascular network between SD-OCTA and SS-OCTA instruments are significantly different [14, 15]. The SS-OCTA technology is able to provide images deeper into the sub-RPE tissue, therefore is more reliable in the detection of the entire neovascular complex, including secondary smaller capillaries [14, 42]. Measurements based on SD-OCTA images may be underestimating the size of the CNV. Second, when comparing images from the same CNV complex pre- and post-treatment, it is imperative to utilize the exact same segmentation in order to ensure that the entire neovascular tissue was included in both images. However, the automatic segmentation is not reliably reproducible in diseases where the outer retina and RPE layers could be disorganized as in AMD [9•, 10•]. Third, areas of atrophy, mainly when next to CNV, can acquire the appearance of a neovascular complex, allowing for inaccurate measurement, making precise evaluation challenging. This is called unmasking artifact [9•, 10•]. Last is the questionable ability to differentiate between real absence of flow or merely reduced flow in treated lesions. If the flow in the vessels is too slow, beyond the threshold of detection, it will appear as absence of flow. So it is possible that the CNV area is kept stable after treatment, but the flow within the neovascular network become slower, thus undetectable. New softwares, such as the variable interscan time analysis, or VISTA, may help to detect sub-threshold flow signals and distinguish slow flow from flow absence [43•].

Conclusion

Fluorescein angiography, the gold standard for diagnosing CNV, is a dynamic exam able to show different patterns of dye transit and leakage [44, 45], however is invasive and time consuming. Moreover, it generates a two-dimensional image. In the last decade, OCT became essential in the diagnosis and management of AMD. Recently, OCTA with its cutting edge new technology allows for simultaneous evaluation of functional (OCT angiograms) and anatomical (OCT B-scans) information. Furthermore, it is safe, fast and non-invasive. It is increasingly being tested in retina and research centers, not only remodeling our level of understanding and interpretation of the images but allowing for the identification of the limitations of the modality, such as artifacts and differences between commercially available devices. Careful designed prospective clinical trials are critically necessary to best determine the optimal role and use of OCTA. Developing companies are facing a technological run to improve the eye-tracking, the speed and number of scans, the segmentation, and the quantification, among other software and hardware advances to provide reliable and reproducible quantitative analysis. OCTA may grant us the opportunity to decrease morbidity through early detection and intervention in macular diseases.

OCTA is still an unfolding technology with several limitations, but it will not be surprising if OCTA becomes the gold standard for diagnosis and follow-up of macular diseases, particularly AMD.