Although the clinical benefit of OCTA has yet to be fully elucidated, recent advancements in OCTA technology have enabled more precise descriptions and analyses of the retinal microvasculature. Conventionally, fluorescein angiography has been the gold standard imaging modality for identifying and evaluating the retinal capillary system. However, this evaluation has been predominantly limited to the SCP as the deep capillary system is poorly visualized with conventional dye-based angiography systems.
5,24 By contrast, OCTA provides depth-resolved imaging and segmentation of the microvascular layers of the macula, allowing separate evaluation and quantification of the SCP versus the DCP. This has been a critical breakthrough as it has become evident that there can be preferential ischemia at the level of the DCP in various retinal vascular diseases such as diabetic retinopathy, retinal vascular occlusion, and sickle cell retinopathy.
3,4,25–31 These studies underscore the importance of reporting data from the SCP and DCP in normal eyes in order to validate the abnormal OCTA findings in retinal vascular diseases.
This is the first study of its size to report the retinal microvascular density and the FAZ dimensions at the level of the SCP and DCP by decade of life in a heterogeneous normal population. Previous studies have shown a decrease in total retinal blood flow associated with an increase in FAZ area with increasing age.
14,32,33 In a study of Chinese patients with healthy eyes, using OCTA, Yu et al.
14 found decreases in the SCP parafoveal flow index and vessel area density with increasing age at a rate of 0.6% and 0.4% per year, respectively. Shahlaee et al.
15 reported mean parafoveal vascular densities of 46% and 52% in the superficial and deep vascular networks, respectively, and also noted negative correlations with age. In a separate study, Shahlaee et al.
23 measured a mean superficial FAZ area of 0.27 mm
2 and deep FAZ area of 0.34 mm
2. That study also noted a larger mean horizontal FAZ length than vertical length in the SCP, as well as a 17% and 13% increase in the horizontal and vertical length, respectively, when measured in the DCP. Comparing histological and speckle variance OCT images, Mammo et al.
34 demonstrated that capillaries in the DCP terminated farther from the fovea than capillaries in the SCP, leading to an apparently larger FAZ in the deep layer.
The present study corroborates and expands upon previously published results of vessel density and FAZ measurements by OCTA analysis in several important ways. First, our data provide vessel density analysis of the entire en face area of both the 3 × 3-mm and the 6 × 6-mm OCTA scans, which is more clinically feasible and applicable than separate foveal and parafoveal calculations. Second, we report vessel density in units of mm−1, which more accurately reflects perfusion of both retinal capillary plexuses. Third, our study sought to quantify and validate annual rates of change in vessel density and FAZ measurements. In addition, we performed a reproducibility analysis and validated the accuracy of repeatable measurements, particularly with the 3 × 3-mm scans. Reproducibility of 6 × 6-mm scans of the DCP was not validated by this study.
Our study analyzed 113 healthy eyes from 70 subjects of various ethnic backgrounds whose ages ranged from 9 to 88 years old (mean: 48 ± 20 years). Compared to the study of parafoveal vessel area density by Yu et al.,
14 our results show a smaller rate of change in vessel density with increasing age occurring in the SCP with an annual rate of reduction of 0.04 mm
−1 (0.26%) per year in 3 × 3-mm scans. The annual change in density at the level of the DCP was similar at 0.06 mm
−1 (0.27%) per year. It should be emphasized that previous studies calculated vessel area density by comparing the area occupied by retinal vessels to defined areas of interest (i.e., foveal or parafoveal regions), reporting vessel area density as a percent of the total.
We report vessel density in units of mm
−1 since we believe this value is more important for comparison in pathologic studies. Previous reports on neovascular complexes in AMD
9,35,36 and paracentral acute middle maculopathy lesions in retinal vascular diseases
28 have quantified changes in vessel density based on the presence or absence of finer vessels and capillaries which are more accurately quantified in units of mm
−1. Since current OCTA technology produces two-dimensional en face representations of volumetric scan data, quantitative analysis of capillary vessel density is limited to the measurement of vessel length per surface area (i.e., mm/mm
2 = mm
−1) as opposed to vessel length per volume (mm/mm
3 = mm
−2). Quantifying the presence of vessels, that is, vessel length, in the macula holds more significance with regard to normal retinal tissue perfusion than the caliber of those vessels, especially when interrogating a capillary plexus. OCTA images segmented at the level of the SCP clearly include large arteries and veins, which do not participate in oxygen and nutrient exchange in the retina. Therefore, calculations that take into account the caliber of retinal vessels, such as total vessel area per area of interest (percentage) via thresholded images,
14–16 or local fractal dimensions,
37 may overestimate retinal tissue perfusion.
In the analysis of 3 × 3-mm scans, our results show a decrease in mean vessel density between subjects 40 to 49 years of age and those 50 to 59 years of age that is trending toward significance in the SCP (
P = 0.097) and is statistically significant (
P = 0.033) in the DCP (
Fig. 3). Furthermore, a statistically significant decrease in mean vessel density exists in both the SCP (
P = 0.029) and DCP (
P = 0.003) in a comparison of subjects 60 to 69 to subjects 70 to 79 years of age. We also analyzed the 3 × 3-mm scan data by comparing larger near-tertile age groups, that is, subjects 35 years of age or less, 36 to 64 years of age, and 65 years of age or older. We found statistically significant decreases in SCP and DCP vessel density between the lowest and middle tertiles (
P = 0.007 in SCP;
P = 0.002 in DCP) and between the middle and highest tertiles (
P = 0.035 in SCP;
P = 0.001 in DCP). This is the first study to validate an age-dependent change in the retinal capillary plexus vessel density. The FAZ area was not found to be statistically different between any of these tertiles at either the SCP or DCP. Larger data sets analyzing healthy eyes from all age groups and ethnicities will be necessary in order to develop screening guidelines to detect variations in vessel density which could serve as an early indicator of change secondary to AMD or retinal vascular diseases that commonly affect older patients.
In the current study we found no statistical difference between males and females in the mean vessel density of the SCP or DCP in both 3 × 3-mm and 6 × 6-mm OCTA scans. The annual decrease in vessel density at the level of the SCP was greater for males at 0.0528 mm
−1 (0.34%) per year than for females at 0.0344 mm
−1 (0.23%) per year with 3 × 3-mm scans. In the DCP the annual decreases in vessel density were similar for both sexes at 0.0557 mm
−1 (0.26%) per year for males and 0.0578 mm
−1 (0.27%) per year for females. These results support those in the study by Yu et al.,
14 showing that annual decreases in the parafoveal vessel area density of the full retinal thickness were greater in male than in female subjects.
15
The mean FAZ area was found to be 0.289 ± 0.108 mm
2 and 0.614 ± 0.200 mm
2 in the SCP and DCP, respectively, demonstrating a twofold increase in FAZ area at the level of the deep plexus. In both males and females, the horizontal GLD was greater than the vertical GLD at the level of both capillary plexuses, which is consistent with findings by Shahlaee et al.
23 As previously reported, the FAZ borders were more clearly delineated at the level of the SCP than at the level of the the DCP.
11,38 Because split-spectrum amplitude-decorrelation angiography measurements are derived from changes in reflections and backscattering of light, the larger vessels of the SCP often resulted in projection artifacts that were visualized at the level of the DCP, obscuring the FAZ borders at that level.
15,23 Careful comparison between the retinal layers was needed at times to resolve the FAZ borders of the deep plexus. Our results showed FAZ areas were not significantly different between males and females in the SCP (
P = 0.174) or the DCP (
P = 0.383). These results support the findings of Samara et al.,
11 who demonstrated no differences in FAZ area between males and females. Across all subjects, our results showed a positive correlation of the FAZ area with increasing age in the SCP. This is in contrast to Samara et al.
11 but consistent with Shahlaee et al.
23
A reproducibility analysis was performed in a subset of 27 eyes (24%) in order to confirm the reliability of vessel density and FAZ dimension calculations on OCTA images. The results from this analysis validated the reproducibility of capillary density and FAZ measurements made on 3 × 3-mm OCTA scans. For 6 × 6-mm scans, however, the results demonstrated a statistically significant change in DCP vessel density at follow-up. This indicates further investigation is needed to validate the ability of the RTVue-XR Avanti device to accurately resolve the deep retinal capillary plexus in 6 × 6-mm OCTA scans.
Of note, three healthy subjects without history of albinism or nystagmus showed minimal or potentially absent FAZs despite the presence of a normal foveal pit and preserved visual function. Previously, Marmor et al.
39 studied four patients who showed absence or near absence of the FAZ in the setting of fovea plana. In these patients, vessels were seen crossing the foveal center on fundus photos and fluorescein angiography, and multifocal electroretinography responses were within normal limits and with normal waveforms across the fovea and posterior pole. The authors concluded that neither the FAZ nor the foveal pit is critical to development of cone lengthening and spatial packing that presumably results in higher visual resolution. Dolz-Marco et al.
40 reported OCTA findings from three patients with fovea plana, demonstrating fusion of the SCP and DCP, absence of an FAZ, and flattening of the foveal pit. In our study, by coregistering the OCTA scans with OCT B-scans from the same location, we observed normal foveal pits with widened outer nuclear layers and cone outer segment lengthening despite minimal or absent FAZ (
Fig. 4). Because each subject maintained normal visual acuity, it was determined that they were normal variants and were included in the final analysis. These results support the conclusion of Marmor et al.
39 that the FAZ is not critical for the normal structural development of the high visual acuity fovea or foveal pit.
Despite continued advancement in OCTA technology, we acknowledge some limitations in our study. As SSADA technology relies on the detection of blood flow, any movement of the patient's head or eyes during image acquisition resulted in varying degrees of motion artifact and decreased image quality. These complications were noted more often in the extremes of age groups. In addition, projection artifact of the large vessels of the SCP onto deeper retinal layers may have lead to artificially increased values of vessel density in the DCP.
41 More advanced algorithms that remove or correct for these types of artifacts will greatly improve the accuracy with which OCTA can identify vessels in their native location. This imminent advancement will permit a more real and distinct separation of superficial, intermediate, and deep retinal capillary plexuses and will vastly improve the reliability of measurements in both normal and pathological eyes. Finally, this study only used one trained reader to measure the dimensions of the FAZ. Previous studies have shown that manual identification of the FAZ using OCTA is reliable and accurate in the SCP but may vary in the DCP due to projection artifact obscuring the FAZ borders.
23 Even with these limitations, the results from this study are noteworthy as they include data from subjects in nearly every decade of life.
Age-matched measurements of the SCP and DCP in normal eyes provide important standardized values and may allow more accurate diagnosis and management of retinal vascular disorders. By continuing to perform OCTA analysis of healthy eyes across the lifespan, we hope to eventually establish normative data for vessel density and FAZ dimensions in both the superficial and the deep retinal capillary plexuses to aid in the early diagnosis of macular pathology and the preservation of vision into old age.