The range of subretinal hyperreflective material (SHRM) seen in macular disease includes type 2 macular neovascularization, fibrosis, exudation, vitelliform material, and hemorrhage. The prognostic significance of SHRM has been evaluated retrospectively in clinical trials, but discriminating SHRM subtypes traditionally requires multiple imaging modalities. The purpose of this study is to describe optical coherence tomography angiography (OCTA) flow characteristics and artifacts that might help to distinguish SHRM subtypes.
Patients with age-related macular degeneration (AMD), myopia, pachychoroid disease, and macular dystrophy, manifesting SHRM on optical coherence tomography (OCT), were recruited. Clinical chart review and multimodal imaging established the SHRM subtype. All patients underwent OCTA. OCT and OCTA images were examined together for (1) intrinsic flow, (2) retinal projection onto the anterior SHRM surface (strong, weak, absent), (3) retinal projection through SHRM onto retinal pigment epithelium (RPE), and (4) masking of choriocapillaris flow.
Thirty-three eyes of 25 patients were included (type 2 neovascularization ×3; fibrosis ×4; exudation ×10; hemorrhage ×5; vitelliform ×17). Mean age per eye was 76 years (standard deviation: 12). Intrinsic flow was strongest in type 2 neovascularization. Subretinal fibrosis showed limited flow in residual large-caliber vessels and branches. Flow was not detected within foci of exudation, hemorrhage, or vitelliform lesions. Retina-SHRM surface projection was strongest onto smooth-surfaced SHRM and weaker onto exudation. Retinal projection was weakest on the surface of vitelliform lesions. Retina-RPE projection was masked by dense hemorrhage and vitelliform material. In compound SHRM, OCTA distinguished between vascular and avascular components.
Optical coherence tomography angiography can distinguish vascular from avascular SHRM components. OCTA artifacts may distinguish certain avascular SHRM components.
Separation of the neurosensory retina from the retinal pigment epithelium (RPE) occurs pathologically and the resulting subretinal space may contain materials, deposits, and tissue components that vary considerably in their composition, natural history, and response to intervention.
Materials other than serous fluid share the characteristic that they are, to some degree, hyperreflective on optical coherence tomography (OCT) imaging. The range of subretinal hyperreflective material (SHRM) seen in macular disease includes neovascular tissue, fibrosis, exudate, vitelliform material, hemorrhage, and reticular pseudodrusen (subretinal drusenoid deposits). Subretinal fibrin may be seen in central serous chorioretinopathy and posterior uveitis. Subretinal hyperreflective exudation (SHE) refers to a form of avascular SHRM which is associated with active type 1, 2, or 3 neovascular tissue and which often regresses as the neovascularization becomes quiescent in response to antiangiogenic therapy.
Advances in in vivo imaging, particularly spectral-domain OCT, have improved the detection and characterization of SHRM. It has also been shown that the presence and composition of SHRM in eyes with age-related macular degeneration (AMD) has some relevance to visual prognosis. However, the appearances of different forms of SHRM are not sufficiently specific for them to be distinguished unambiguously by OCT alone, and other imaging modalities must be employed to resolve the differential diagnosis. Vitelliform material, for example, may be distinguished from neovascular tissue and hemorrhage by its hyperautofluorescence, and dye-based angiography is traditionally employed to diagnose neovascularization.
Fluorescein angiography and indocyanine green angiography reveal time-resolved tissue perfusion characteristics and contribute to the diagnosis of vascular lesions. However, both modalities are limited in depth resolution and confounded by leakage and staining effects, which can compromise lateral resolution within microvascular networks. Optical coherence tomography angiography (OCTA) employs decorrelation algorithms to render depth-resolved microvascular flow maps derived from dense raster structural reflectivity data and, under favorable conditions, can overcome some of the limitations of dye angiography. An additional benefit of this modality is that it captures structural and dynamic data in a single acquisition.
The purpose of this study is to examine vascular and avascular forms of SHRM occurring in AMD with OCTA, to describe flow characteristics that are either intrinsic to the SHRM or related to its adjacent tissues and to identify features that may resolve differential diagnosis.
This validity analysis was approved by the Western Institutional Review Board (Olympia, Washington, USA), the Tufts Medical Center Institutional Review Board (Boston, Massachusetts, USA), the University of California Los Angeles Institutional Review Board (Los Angeles, California, USA), and the San Raffaele Hospital Institutional Review Board (Milan, Italy). It complied with the Health Insurance Portability and Accountability Act of 1996 and followed the tenets of the Declaration of Helsinki.
Participants and Their Diagnoses
Patients were recruited from the practices of 6 retina specialists (K.B.F., L.A.Y., J.S.D., N.W., D.S., G.Q.). Patients were included if they exhibited a neurosensory detachment on cross-sectional OCT with hyperreflective material in the subretinal space, in the context of AMD, myopic macular degeneration, pachychoroid spectrum disease, or retinal dystrophy. Eyes with non-AMD diagnoses were retained for study if they harbored types of SHRM that might be seen in AMD (eg, vitelliform lesion in Best disease) or that can occur with AMD in “overlap” conditions.
Clinical charts and multimodal imaging data were reviewed including fundus photography, fluorescein angiography, fundus autofluorescence, and OCT, all of which had been performed as clinically indicated.
Fluorescein angiographic appearances of “choroidal” neovascular lesions were recorded in terms of the original descriptions of “occult” and “classic” patterns, but type 1, type 2, and type 3 lesions themselves were defined and classified anatomically using both fluorescein angiography and cross-sectional OCT.
Subretinal fibrosis was identified by the fundus biomicroscopy finding of an elevated mound of yellow-white tissue with late staining and minimal leakage on fluorescein angiography. Subretinal hyperreflective exudation was recognized on cross-sectional OCT in association with neovascularization as previously described. Acquired vitelliform lesions were defined as subfoveal accumulations of yellow-white material exhibiting moderate to intense hyperautofluorescence and were described in terms of their amorphous and granular components.
All patients had undergone OCTA of the study eye. Patients for whom gradable OCTA images were unavailable owing to media opacity or poor fixation were excluded. Multimodal images and OCT and OCTA segments of SHRM foci in included patients were sent to one of the authors (K.K.D.) for grading and classification.
Imaging devices used were as follows: TRC-50IX fundus camera (Topcon Corporation, Tokyo, Japan), Spectralis HRA+OCT scanning laser ophthalmoscope (Heidelberg Engineering, Heidelberg, Germany), and RTVue XR “Avanti” (Optovue, Fremont, California, USA).
The RTVue XR is a spectral-domain OCT device that employs a split-spectrum amplitude-decorrelation algorithm for angiographic processing. Horizontal and vertical raster patterns are combined to correct eye movement artifacts in post-processing. The viewing software performs automated tissue boundary detection to assist in segmenting retinal and choroidal layers for en face projection. Four preset curve pairs are offered (superficial, deep, outer retina, choriocapillaris), as well as a “flat” line pair with slant adjustment. In cross-sectional mode, the viewing software offers the option of overlaying thresholded (“all-or-nothing” encoded) flow signals on structural reflectance scans, enabling point-to-point correlation between cross-sectional structural, cross-sectional angiographic, en face structural, and en face angiographic views.
Since a SHRM focus represents, in geometric terms, a “space-occupying lesion” its presence violates the assumptions regarding lamellar tissue arrangement on which automated segmentation algorithms rely. To interpret scans with segmentation errors, each SHRM focus was studied using multiple preset boundary curves, with manual adjustment of the anteroposterior positions of those curves. Lesions were also “scrolled through” to build up a mental 3-dimensional image of each lesion. OCTA imaging artifacts were described using recently published nomenclature.
Optical Coherence Tomography Angiography Outcome Measures
Each SHRM focus was examined for intrinsic flow that could be demonstrated in both cross-sectional scans with angiographic overlays and in en face views. In cross-section, the presence of a flow signal at the level of the RPE could be interpreted as intrinsic rather than projected if there was no corresponding flow signal in the inner retina along the same A-scan axis. Furthermore, to qualify SHRM definitively as vascularized, the en face morphology of its (intrinsic) flow signature had to differ from that of the retinal circulation, distinguishing it from projection artifact.
Additionally, each SHRM focus was examined systematically from anterior to posterior aspects using en face planes to study projection characteristics. The anterior surface was studied to determine whether it produced a well-resolved projection of the flow signature of the overlying retina. The projection onto the anterior surface of the SHRM was compared qualitatively with the projection of the retinal circulation onto the surrounding unmasked RPE, and graded “strong” (comparable, eg, Figure 4 ), “weak” (attenuated or reduced detection), absent, or ungradable. Projection of the retinal microvasculature through SHRM onto the RPE was similarly assessed, and weak retina-RPE projection was attributed to SHRM opacity.
The inner choroid was examined en face, directly beneath the SHRM lesion. After accounting for segmentation errors, the sub-SHRM choriocapillaris and inner Sattler layers were compared with the homogenous, physiological flow signature of the more peripheral choroid. Flow signal attenuation could be attributed to masking (shadow) or inner choroidal atrophy. The choriocapillaris was considered to be masked by the SHRM if it appeared hyporeflective on the en face structural OCT and if the outline of the shadow matched the en face silhouette of the SHRM.
All images were examined independently by 2 retinal specialists (K.K.D. and F.G.) and any discrepancies were resolved by the senior author (K.B.F.). The outcome measures described above are illustrated in the accompanying figures, which describe example cases of each.
A total of 33 eyes of 25 patients were included. Mean age per eye was 76 years (standard deviation: 12). A list of study eyes with individual diagnoses and features of interest is presented as Table 1 . The diagnostic classification of each eye is summarized in Table 2 together with summarized OCTA findings for each diagnosis.
|Patient||Age||Sex||Eye||Eye #||Diagnosis||SHRM Types||Figure|
|1||73||M||L||1||Myopia||Type 2 NV||1|
|2||83||M||L||2||AMD||Type 2 NV; exudation; hemorrhage||2|
|3||85||F||R||3||AMD||Type 2 NV|
Type 1 NV
|10||81||F||R||11||Type 1 NV||Exudation|
|L||12||Type 1 NV||Exudation|
Type 1 NV
Type 1 NV
|Exudation (thin layer)|
Treatment-naïve type 1 NV
|17||61||20||Adult vitelliform dystrophy||Vitelliform|
|18||81||21||Adult vitelliform dystrophy||Vitelliform|
|20||56||24||Adult vitelliform dystrophy||Vitelliform|
|21||66||R||25||Adult vitelliform dystrophy||Vitelliform|
|22||51||R||27||Adult vitelliform dystrophy||Vitelliform|
|24||77||R||31||Adult vitelliform dystrophy||Vitelliform|
|25||85||33||Adult vitelliform dystrophy||Vitelliform|
|Subretinal Hyperreflective Material||Eyes||Optical Coherence Tomography Angiography|
|Intrinsic Flow||Surface Projection||RPE Projection||Choriocapillaris Masking|
|Type 2 neovascular tissue||3||Yes (primary)||Unresolved||Ungradable|
|Fibrosis/disciform scar||4||Yes (residual)||Strong||Absent||Ungradable|
|Subretinal hyperreflective exudation||10||No||Strong or weak||Strong or weak||No|
|Vitelliform||17||No||Weak or absent||Absent||Amorphous: no |
Type 2 Neovascularization
Three eyes of 3 patients were identified with type 2 neovascularization. Each of these exhibited well-defined early hyperfluorescence on fluorescein angiography, and flow within the type 2 lesions was detected readily by OCTA in both cross-sectional overlay and en face views ( Figures 1 and 2 ).
Eye #1 featured a hypopigmented myopic fundus with a thin choroid (<115 μm) and subfoveal type 2 neovascularization. Enhanced infrared transmission was seen through the choroid, behind the neovascular lesion, on cross-sectional OCT. Deep en face OCTA segments showed marked projection of the flow signature of the type 2 lesion onto the sclera , but in spite of this depth of tissue penetration by the scanning infrared laser, flow within the lumina of dilated outer choroidal vessels could not be detected.
In Eye #2, with type 2 neovascularization anterior to a pigment epithelial detachment (PED) ( Figure 2 ), the proximity of the flow signal to the elevated RPE made it difficult, in both en face and cross-sectional viewing modes, to determine definitively whether the neovascular tissue was predominantly subretinal or sub-RPE. However, the en face morphology of the flow signature showed greater resemblance to type 2 than type 1 neovascularization when compared with previously published images. Fluorescein angiography was helpful, showing a “predominantly classic” configuration.
In each of the above cases, retinal projection, onto either the type 2 lesion or the RPE, could not be graded owing to the intense flow signature of the type 2 lesion itself, which appeared qualitatively to have saturated the detection threshold of the device.
Four eyes were identified with subretinal fibrosis based on multimodal imaging. In Eye #4 subfoveal fibrosis had occurred adjacent to a PED. OCTA showed a tangled neovascular network within the PED with a trunk vessel that branched and coursed anteriorly into the subfoveal scar ( Figure 3 ). Pathologic disruption of the anteroposterior OCT reflectivity profiles in this eye resulted in failure of automated segmentation, necessitating the use of an arbitrary flat tissue boundary, with slant adjustment for best visualization.
OCTA did not show any flow intrinsic to subretinal fibrosis in the other 3 cases, but in Eye #6 ( Figure 4 ) and Eye #7 the smooth inner surface of the hyperreflective fibrotic lesion produced strong, well-resolved projection artifacts of the overlying retinal vessels with large decorrelation amplitudes represented by bright en face pixels. Although dense foci of subretinal fibrosis appeared to mask both the retina-RPE projection and the visualization of choriocapillaris in en face views, these observations were difficult to validate with reference to cross-sectional scans, which showed indistinct tissue topology at the level of the RPE owing to choriocapillaris atrophy.
Subretinal Hyperreflective Exudation and Subretinal Hemorrhage
Subretinal hyperreflective exudate was identified in 10 eyes of 9 patients and was associated with type 1 neovascularization (4 eyes) and with type 2 neovascularization (Eye #2, described above, Figure 2 ). Combined evaluation of structural and OCT angiographic data enabled exudation to be distinguished from type 2 neovascular tissue within a compound SHRM focus containing both lesion types.
In Eye #2 subretinal hyperreflective exudation was located within the foveal avascular zone and retinal projections were therefore ungradable. Where gradable, the strength of projection artifacts related to exudation was variable ( Table 2 ). Figure 5 summarizes the findings in the left eye of an 86-year-old patient with type 1 neovascular AMD and subretinal hyperreflective exudation (Eye #9). In this eye OCTA showed strong retina-SHRM projection and weak retina-RPE projection.