Imaging Histology Correlations of Intraretinal Fluid in Neovascular Age-Related Macular Degeneration

Abstract

Purpose: Fluid presence and dynamism is central to the diagnosis and management of neovascular age-related macular degeneration. On optical coherence tomography (OCT), some hyporeflective spaces arise through vascular permeability (exudation) and others arise through degeneration (transudation). Herein we determined whether the histological appearance of fluid manifested this heterogeneity.

Methods: Two eyes of a White woman in her 90s with anti-vascular endothelial growth factor treated bilateral type 3 neovascularization secondary to age-related macular degeneration were osmicated, prepared for submicrometer epoxy resin sections, and correlated to eye-tracked spectral domain OCT. Examples of intraretinal tissue fluid were sought among similarly prepared donor eyes with fibrovascular scars, in a web-based age-related macular degeneration histopathology resource. Fluid stain intensity was quantified in reference to Bruch's membrane and the empty glass slide.

Results: Exudative fluid by OCT was slightly reflective and dynamically responded to anti-vascular endothelial growth factor. On histology, this fluid stained moderately, possessed a smooth and homogenous texture, and contained blood cells and fibrin. Nonexudative fluid in degenerative cysts and in outer retinal tubulation was minimally reflective on OCT and did not respond to anti-vascular endothelial growth factor. By histology, this fluid stained lightly, possessed a finely granular texture, and contained mainly tissue debris. Quantification supported the qualitative impressions of fluid stain density. Cells containing retinal pigment epithelium organelles localized to both fluid types.

Conclusions: High-resolution histology of osmicated tissue can distinguish between exudative and nonexudative fluid, some of which is transudative.

Translational Relevance: OCT and histological features of different fluid types can inform clinical decision-making and assist in the interpretation of newly available automated fluid detection algorithms.

Introduction

Age-related macular degeneration (AMD) causes legal blindness in older adults worldwide.¹ Exudative macular neovascularization (MNV) resulting in fluid accumulation affects 15% of AMD patients.² MNV is managed by intravitreal injections of anti-vascular endothelial growth factor (VEGF).³ Injections are monitored by optical coherence tomography (OCT) and automated measures of tissue thickness, a surrogate for edema, to inform treatment decisions. An emerging approach of disease management is automated, direct measures of MNV fluid volume in different tissue compartments.⁴ Clinicopathological correlation of eyes with MNV fluid imaged during life, the subject of this report, can help to validate this new approach to managing exudative AMD. 

Fluid in intraretinal, subretinal, and subretinal pigment epithelium (RPE) basal lamina⁵ compartments is an important clinical biomarker in MNV.⁶ The presence and dynamics of fluid are central to guiding management and anti-VEGF treatment frequency. In practice, hyporeflective spaces on OCT are often thought to represent VEGF-mediated leaks from vessels (exudation) that will respond to anti-VEGF therapy. On OCT, optical density relative to internal references like vitreous is higher in hyporeflective subretinal spaces of eyes with exudative neovascular AMD than in similar spaces of central serous chorioretinopathy.⁷ However, some hyporeflective areas are considered transudation fluid owing to RPE pump failure.^8–10 Moreover, structural changes in retinal tissue, such as degenerative cavitation or outer retinal tubulation (ORT), can produce hyporeflective areas mimicking fluid, yet are not responsive to anti-VEGF.¹¹ Thus, there may be a spectrum of (intra)retinal fluid ranging between exudative and transudative. It is unclear precisely what component of intraretinal fluid is detected by OCT in vivo. Thus far, both proteins and lipids have been proposed,^12,13 and experimental determination of blood component reflectivity suggest that triglycerides (as in plasma lipoproteins) are important determinants.¹⁴ Finally, there are no established animal models for exudation in which these questions could be addressed experimentally.¹⁵ Thus, clinicopathological correlation of human tissue remains a potential source of information about intraretinal fluid types. 

Our recent histological analyses of neovascular AMD eyes revealed intra- and subretinal collections of fluid, as well as serous retinal pigment epithelium detachment.^16–18 Because these tissues had been osmicated and stained with toluidine blue, the fluid appeared blue gray, a color possibly attributable to the staining of lipoproteins from plasma and debris containing cellular membranes. Conversely, fluid in ORT and in degenerative cysts appeared light blue or clear.¹⁹ These findings prompted a systematic investigation to seek a histological basis of different intraretinal fluid types. We herein report histological analyses of clinically documented intraretinal fluid in two index eyes treated with anti-VEGF injection owing to type 3 MNV^20,21 and 23 donor eyes with fibrovascular scar secondary to neovascular AMD. 

Methods

Compliance

This study was approved by Institutional review at the University of Alabama at Birmingham (protocol #300004907) and conducted in accordance with the tenets of the Declaration of Helsinki and the Health Insurance Portability and Accountability Act of 1996. 

Clinical Course

A White, pseudophakic woman in her 90s received a comprehensive ophthalmologic examination and multimodal imaging during a 5-year follow-up owing to bilateral type 3 MNV secondary to AMD. The patient presented in 2014 with exudative type 3 MNV in the right eye. Over 5 years, the right eye received a total of 37 intravitreal anti-VEGF injections over approximately 6 fluid resorption cycles (12 × 0.5 mg/0.05 mL ranibizumab then 25 × 2 mg/0.05 mL aflibercept). One fluid resorption cycle is defined as the time in weeks and the number of injections needed from initial detection of intraretinal/subretinal edema on OCT until complete absence of edema on OCT. The left eye was diagnosed with exudative type 3 MNV 4 years after the right eye. Over 9 months, the left eye received a total of six intravitreal anti-VEGF injections over approximately two fluid resorption cycles in the left eye (12 × 0.5 mg/0.05 mL ranibizumab). Her general medical history included chronic dyslipidemia and paroxysmal atrial fibrillation. In late 2018, the patient was diagnosed with terminal gallbladder adenocarcinoma. Her last ophthalmic evaluation and anti-VEGF treatment of the right eye was in January 2019, 2 months before her death owing to adenocarcinoma.²⁰ Subsequent analysis revealed a total of seven vascular lesions in the two eyes, six of which were found by histology.^20,21 Three lesions were classified as type 3 MNV, and three were classified as deep retinal age-related microvascular anomalies.²² 

Clinical Image Capture and Analysis

All images were acquired using Spectralis HRA + OCT (Heidelberg Engineering, Heidelberg, Germany). Available for review were 11 (right eye) and 7 (left eye) eye-tracked spectral domain OCT volumes (6 mm × 6 mm horizontal and radial scans; 20° × 20° field, 127 µm distance between B-scans) and fluorescein angiography at first presentation and 4 years later. One OCT angiography volume of the right eye was qualitatively evaluated for vascular dropout in the deep capillary plexus. 

Histology Preparation and Image Analysis of the Index Case

As described,²³ globes were recovered 2:05 hours after death, opened anteriorly,²⁴ and immersion fixed in buffered 1% paraformaldehyde–2.5% glutaraldehyde. Post-mortem OCT volumes of opened and preserved globes were captured and registered via eye-tracking to pre-mortem OCT volumes of the same globes, as described.^24,25 A 8 mm × 12 mm rectangular tissue block containing fovea and optic nerve was post-fixed in 1% osmium–tannic acid–paraphenylenediamine and embedded in epoxy resin. Submicrometer sections stepped at 30- to 60-µm intervals were stained with toluidine blue and scanned using 60× or 100× oil immersion objectives.²⁶ Tissue sections spanning a distance of 5453 µm (right eye, 112 glass slides) and 4243 µm (left eye, 87 glass slides) were matched to clinical OCT scans by comparing overall tissue contours.²⁰ 

Histology Preparation and Image Analysis of Donor Eyes (Project MACULA)

Eyes were accessioned from White, nondiabetic donors by the Advancing Sight Network between 1995 and 2012, opened to allow preservative entry, and retained in preservative long term.²⁷ Ophthalmic histories were not available. Eyes with neovascular AMD were accessioned largely before anti-VEGF therapy was introduced in 2006 and may have received other treatments available during that period. Eyes with neovascular AMD were recognized by fibrovascular scars in the subretinal or subRPE–basal lamina compartments, in the setting of soft drusen material and basal laminar deposits, as described.²⁸ 

All eyes were preserved in 1% paraformaldehyde and 2.5% glutaraldehyde in a 0.1 M phosphate buffer with an average death-to-preservation time of 3:49 (range, 0:40–11:40). Macula-wide, full-thickness tissue punches were obtained with an 8.25-mm diameter trephine (#68825-L; Howard Instruments, Tuscaloosa, AL) and processed as described above in the previous section. They were sectioned at 0.8 µm starting at the superior edge of the punch and stained with toluidine blue. Sections at each of two locations (foveal rod-free zone and perifovea 2 mm superior) were scanned along their entire length and digitized using image stitching software (CellSens; Olympus, Center Valley, PA). Sections were systematically reviewed and features of interest were photo-documented with a 60× oil immersion objective (numerical aperture, 1.4).^29–31 

For figures, images were cropped, adjusted to maximize the intensity histogram for contrast and white balance, and composited (Photoshop CS6; Adobe Systems, San Jose, CA). To quantify our impressions of staining differences between intraretinal fluid types, we measured pixel color intensities of fluid (red, blue, green) (Photoshop CS6; Adobe Systems). Pixel color intensities in three channels of Bruch's membrane and a blank glass slide within the same section served as positive and negative control for stain intensity, respectively. Bruch's membrane stained deeply so that the red, green, and blue intensities were low, whereas intensities of a blank area on the slide were high. 

Results

Clinically Documented Case

In the two index eyes, significant macular exudation responded well to anti-VEGF intravitreal injection, but required ongoing treatment.^20,21 In the right eye, 6 fluid absorption cycles over 37 anti-VEGF injections resulted in structural damage and reorganization in the inner and outer retina. 

Figure 1 illustrates a hyporeflective cyst with a discontinuous and faint hyper-reflective edge on OCT 11 months before death. Located in the inferior parafovea (Fig. 1A1), this cyst sat above confluent soft drusen (Fig. 1A2). Several hyper-reflective specks formed a column within the Henle fiber layer (HFL) and pointed toward the cyst (Fig. 1A2). The cyst persisted under treatment, unlike a dynamic pocket of exudation next to it (Fig. 1B2). On histology (Fig. 1C), large and small fluid filled cysts are present within the HFL. A large cyst contained a cell with loosely packed RPE granules (right side of Fig. 1D) and corresponded with the clinically stable feature seen on OCT (Figs. 1A2 and 1B2). The fluid within the large cyst (Fig. 2B) appeared pale-stained with granular texture. In contrast, the small cyst (Fig. 2A) corresponded with the clinically dynamic pocket of exudation and contained darker stained fluid with a homogeneous texture. On OCT angiography (Supplementary Fig. S1) vascular dropouts of the deep capillary plexus was present in the region of intraretinal fluid around one type 3 MNV lesion in the right eye. 

Figure 1.

 

Intraretinal cyst with transudative fluid in tissue damaged by previous exudation is stable compared with exudative fluid. (A1, A2) Near infrared reflectance (NIR) (A1) acquired 11 months before death shows two locations of histologically confirmed, exudative type 3 MNV (white arrowheads; 8 weeks after the prior anti-VEGF injection, 11 months before death). The green line (A1) represents the plane of OCT B-scan (A2). (A2) On horizontal OCT scan, a hyporeflective degenerative cyst (light blue arrowhead) in the inner nuclear (INL) and HFL is located above hyper-reflective material, which is in turn located above a retinal pigment epithelium (RPE) elevation with a hyper-reflective core within soft drusen material (orange arrowhead). (B1, B2) On a radial OCT B scan acquired 2 months before death, fluid seen in A (light blue arrowhead) is overall stable and is considered transudative. In contrast, a recurrent hyporeflective exudative fluid pocket (dark blue arrowhead) exhibits dynamic changes. (C) Panoramic histology shows atrophy of RPE and outer retina, with intraretinal fluid (light blue arrowhead) above several calcified drusen. The retina largely detached during histological tissue processing. It remained attached at sites of drusen-related RPE atrophy owing to glial adhesion to basal laminar deposit (BLamD). (D) Magnified histology shows a large and small intraretinal cyst (light and dark blue arrowheads, respectively). The large cyst contains pale-stained fluid with granular texture, representing transudative fluid, and a nucleated cell containing some RPE organelles (magnified inset, light blue dotted rectangle) (Fig. 2). The small cyst contains darker stained fluid with a smooth and homogenous appearance, representing residual exudative fluid (magnified inset, dark blue dotted rectangle) (Fig. 2). To the left of the small cyst, structural damage to the HFL (dark blue asterisk) suggests an area of prior intraretinal fluid. The outer plexiform layer (OPL) and the outer nuclear layer are thinned without dyslamination. Relative to the external limiting membrane descent (yellow arrowhead), the RPE layer is disintegrated to the left (temporal) and continuous but dysmorphic to the right (nasal). BLamD separates the RPE from the underlying druse, which contains avascular fibrosis and evidence of dislodged calcific nodules (black asterisk).

Figure 2.

 

Intraretinal exudative and transudative fluid in donor eyes with neovascular AMD possess different stain intensities. Magnified histology of Figure 1D compares red, green, blue (RGB) pixel color intensity of Bruch's membrane (a positive control, fuchsia rectangle) and a blank area of the same glass slide (negative control, orange rectangle) and presumed exudative (A) and transudative (B) fluid in donor eyes with neovascular AMD (scale bar, 100 µm). Three-channel stain intensities are shown in Supplementary Table S1. (A) A small cyst contains darker stained fluid with a smooth and homogenous appearance, representing residual exudative fluid, compared with fluid shown in B. (B) The edge area of a large cyst contains pale-stained fluid with granular texture, representing transudative fluid, and a nucleated cell with some RPE organelles (scale bar, 25 µm).

Figure 3 illustrates exudative fluid, recognized by its dynamic response to anti-VEGF. Enlarged intraretinal cysts located at the level of the HFL responded to intravitreal anti VEGF injections (Figs. 3A2 and 3B2). The source of the fluid was a pyramidal type 3 MNV lesion (Figs. 3C, D), visible as a hyper-reflective lesion atop a druse with calcific nodules³² (Fig. 3A2). The type 3 lesion was bounded by RPE cells and extended from the deep capillary plexus to the basal laminar deposit draping a calcified druse. On histology, the intraretinal fluid appeared homogeneously blue grey (Fig. 3C) and was comparable with intravascular plasma. The fluid was located next to an area of swollen processes (Fig. 3D). 

Figure 3.

 

Residual intraretinal exudative fluid. Green lines on near-infrared reflectance (NIR) (A1, B1) represent OCT B-scans (A2, B2). The blue line indicates histology sections in C and D. Different red (R), green (G), and blue (B) pixel color intensities are measured within dotted rectangles in C and D as reported in Supplementary Table S1. Bruch's membrane and areas without tissue were positive and negative controls (fuchsia and orange rectangles, respectively. (C) Adapted from Berlin et al.²¹ (A1-2) OCT B-scan (A2) displays enlarged intraretinal cysts (dark blue arrowhead) next to a hyper-reflective lesion (green arrowhead). This lesion sits atop a calcified druse, creating a double layer sign appearance (5 total injections, 0.5 mg/ 0.05 mL ranibizumab, 8 weeks after the prior injection, 4 months before death). (B1-2) Intraretinal fluid is significantly reduced on OCT B-scan (B2) after six injections. This illustrates the dynamic of exudative fluid during anti-VEGF treatment (6 weeks after the prior injection, 2 months before death). Numerous hyper-reflective foci are visible on the left side of the B-scan. (C and D). On magnified histology of two sections 40 µm apart, intraretinal exudative fluid in the HFL appears homogeneously blue grey (dark blue asterisks, dark blue dotted rectangle). This fluid represents recurrent exudation and is comparable to intravascular plasma seen in a capillary (red asterisk). A pyramidal type 3 lesion (green arrowhead) (C) is the main source of this fluid. It is located next to an area of swollen cell bodies in the HFL (D, dark blue asterisk). The type 3 lesion is bounded by retinal pigment epithelial (RPE) cells and extends from the inner nuclear layer (ONL)/ outer plexiform layer (OPL) border to basal laminar deposit draping a calcified druse. Intraretinal RPE organelles, including some in the fluid pocket (dark blue arrowhead), are indicated (light brown arrowheads).

Donor Eyes With Neovascular AMD

In 23 nondiabetic donor eyes with neovascular AMD and fibrotic scars, ORT, intraretinal fluid, and intravascular plasma could be identified on histology. These features were correlated with ex vivo OCT in one case. Lumens of ORT form a continuous compartment with the subretinal space, as Müller glia scroll the photoreceptors into a tube.³³ Fluid in ORT may thus derive from several sources (physiological, exudative, and transudative). In the description provided below, we use the term exudative for fluid clearly derived from compromised vessels to distinguish these from other sources. 

Figure 4 illustrates examples of intraretinal fluid and intravascular plasma. Degenerative cysts appeared lightly stained and transparent with a fine granular texture (Fig. 4A). Cellular blood components were not present. Nonexudative fluid within ORT looked similar (Fig. 4B). Although debris of photoreceptor outer segments and transdifferentiated RPE could be found within ORT, blood cells were not found (Fig. 4B). Conversely, Figure 4C demonstrated an instance of blue–greyish homogenously appearing intraretinal fluid containing blood cells. This intraretinal fluid pocket (Fig. 4C) appeared similar to intravascular plasma (Fig. 4D). 

Figure 4.

 

Intraretinal fluid versus intravascular plasma in eyes with neovascular AMD. Different red (R), green (G), and blue (B) pixel color intensities are measured within dotted rectangles in (A–D), as reported in Supplementary Table S1. Red, green, blue (RGB) pixel color intensity of Bruch's membrane (fuchsia dotted rectangle) and blank glass slide (not shown) serves as positive and negative control, respectively. (A) Intraretinal transudative fluid (light blue arrowhead) in two degenerative cysts appears lightly stained and transparent with a fine granular texture (light blue dotted rectangle; right light blue dotted rectangle). Cellular blood components are not detected. An 87-year-old female donor. (B) Intraretinal transudative fluid within an ORT, a degenerative tissue feature, contains non exudative fluid similar appearing as in degenerative cysts in (A). Fluid appears very lightly stained, and the texture is finely granular (light blue dotted rectangle). Within this ORT, photoreceptor outer segment debris and transdifferentiated retinal pigment epithelium (RPE) (orange arrowhead) are shown. However, cellular blood components are not seen. An 87-year-old female donor. (C) Intraretinal exudative fluid (dark blue arrowhead, dark blue dotted rectangle) appears blue greyish with homogenous texture and contains red (fuchsia arrowhead) and white blood cells. A 94-year-old female donor. (D) Intravascular plasma (dark blue arrowhead, dark blue dotted rectangle) in a retinal artery appears blue grey and darker than tissue-free areas on the same glass, at the top of the panel (orange dotted rectangle, negative control). The texture is homogenous and red blood cells are seen (fuchsia arrowhead). A 90-year-old female donor.

In Figure 5, intraretinal fluid containing presumed fibrin, RPE-derived cellular components, and cellular blood components were correlated to ex vivo OCT scans. Presumed fibrin appeared as a hyper-reflective lesion within a hyporeflective cyst on OCT. On histology, the hyper-reflective lesion corresponded with a darkly stained and solid ovoid droplet. Another hyper-reflective lesion within a hyporeflective cyst on ex vivo OCT (Fig. 5B) was correlated with a cell with organelles typical of RPE. This cyst also contained several erythrocytes. 

Figure 5.

 

Intraretinal exudative and transudative fluid in donor eyes with neovascular AMD. Different red (R), green (G), and blue (B) pixel color intensities are measured within dotted rectangles in (A4, B3, and B5), as reported in Supplementary Table S1. Red, green, blue (RGB) pixel color intensity of Bruch's membrane (not shown) and blank glass slide (orange dotted rectangle) serve a positive and negative control respectively. (A) (A1-A5 and B1–B5) Ex vivo OCT and histology at two horizontally oriented planes, 120 µm apart, through one peripapillary fluid pocket, in a 79-year-old male eye donor. (A1) Ex vivo B-scan shows macular atrophy and an intraretinal lesion (dark blue dotted rectangle). Retina is artifactually detached. (A2) Magnified B-scan shows a hyper-reflective spot (yellow asterisk) within a hyporeflective space. (A3) Panoramic histology matches (A1) (dark blue arrowhead, intraretinal fluid). (A4) Magnified histology shows intraretinal fluid (dark blue arrowhead, dark blue dotted rectangle) with a presumed fibrin droplet⁵⁶ in the center (yellow asterisk, yellow dotted rectangle) within the HFL and compressing the outer nuclear layer. (A5) The fluid is blue–greyish and contains lighter stained round areas with darker central dots. The texture of the fluid is otherwise homogenous and is considered exudation. (B) Ex vivo OCT B-scan (B1) shows a second hyporeflective intraretinal lesion with a hyper-reflective focus (HRF) (light blue and light brown arrowheads, respectively). (B2) The hyporeflective space next to the HRF is more hyporeflective and homogenous than the space in (A2). (B3) Panoramic histology matching the peripapillary region in (B1). (B4) RGB pixel color intensity of blank glass slide (B3, orange dotted rectangle) serve a negative control. Magnified histology shows three intraretinal fluid pockets (dark and light blue arrowheads for gray-stained and clear, respectively). The stained fluid contains a cell (light brown arrowhead). The middle and right pockets are shown in (B5). (B5) One fluid pocket (dark blue arrowhead, dark blue dotted rectangle), presumed exudative, stains gray with a homogenous and smooth texture. It contains one cell with organelles typical of retinal pigment epithelium plus several erythrocytes. The other fluid pocket (light blue arrowhead, light blue dotted rectangle), presumed transudative, stains pale gray with a fine granular texture, lacking cells.

Pixel Color Intensities of Exudative and Transudative Intraretinal Fluid

Table 1 summarizes the mean pixel intensities of exudative and nonexudative fluids in the figures. Raw pixel intensities of all instances are listed in Supplementary Table S1. Bruch's membrane and blank glass slide served as positive (low intensities) and negative (high intensities) controls, respectively. This analysis showed that exudative fluid had consistently lower pixel intensities than nonexudative fluid, supporting the qualitative assessments. Although swollen cell bodies in Figure 3D possess similar pixel intensities as transudative fluid, presumed fibrin within exudative fluid in Figure 5A had pixel intensities comparable to stained Bruch's membrane (Supplementary Table S1). 

Table 1.

 

Different Mean Red (R), Green (G), and Blue (B) Pixel Color Intensities of Intraretinal Exudative and Nonexudative Fluid in Donor Eyes With Neovascular AMD

 

Summary of Findings

Table 2 summarizes the fluid appearance in the 2 index eyes and in the 23 donor eyes. When compared with intravascular blood as an internal reference, exudative and nonexudative fluids (including transudative fluid) were distinguishable regarding OCT reflectivity, response to anti-VEGF, appearance on histology, and cellular components. Compared with nonexudative fluid, exudative fluid exhibited more reflectivity on OCT, dynamically responded to intravitreal anti-VEGF injections, stained more darkly in 1% osmium–tannic acid–paraphenylenediamine postfixed tissue, possessed a homogenous texture, and contained blood cells and fibrin. In contrast, transudative fluid exhibited less reflectivity on OCT, remained stable in response to anti-VEGF injection, stained lighter, possessed a granular texture, and contained mainly tissue debris. Cells containing RPE organelles could be found in both intraretinal fluid types. 

Table 2.

 

Qualitative Features of Intraretinal Exudative and Nonexudative Seen by Histology in Donor Eyes

 

Discussion

This clinicopathological correlation uses histological appearances to provide initial proof of concept for different intraretinal fluid types in neovascular AMD—exudative and nonexudative, which includes transudative. This distinction is important because repeated exudation and fluid absorption under anti-VEGF treatment leads to tissue damage and tissue remodeling, resulting in the accumulation of transudative fluid. Importantly, our data offer clinically relevant insights to support diagnostic and treatment decisions. 

On OCT, fluid is routinely described as a hyporeflective space within the retinal layers.³⁴ In neovascular AMD, fluid accumulates as exudation exceeds the local capacity to remove it.³⁵ To understand location, size, shape, and components of fluid, it is important to understand its underlying pathophysiology. Fluids in the retinal layers result from an imbalance of inflow and outflow mechanisms. Regarding inflow, leakage of fluid can develop as a result of two basic pathways: transudation or exudation.⁸ Regarding outflow, the Na⁺–K⁺ ATPase pump of the RPE and the intraretinal Muller cells are critical facilitators. These actively maintain a water-tight state of deturgescence.^10,36,37 

Hyporeflective spaces on OCT can also be encountered in eyes with non-neovascular AMD. The interior of degenerative cavitation or ORT (Fig. 4A and B, respectively), can mimic fluid with a hyporeflective appearance on OCT.^11,19 ORT are considered a sign of neurodegeneration and gliosis in the outer nuclear layer, consisting almost exclusively of cones lacking outer and/or inner segments.¹⁹ The identity as cones is based on the similar ultrastructure of remaining inner segments to those in retinitis pigmentosa,³⁸ cone-type–specific carbonic anhydrase histochemistry,³⁹ and the persistence of cone-specific inclusion bodies.¹⁹ Similar in appearance to ORT on OCT, intraretinal microcysts, originally referred to as pseudocysts, can occur over areas of RPE and outer retinal atrophy in the absence of neovascularization.⁴⁰ They appear as hyporeflective, circular cystic spaces that do not leak on fluorescein angiography and do not require anti-VEGF treatment. Pseudocysts possibly represent loss of cells, for example, Müller glia. 

Our histological description of fluid expands a sparse literature. The few animal models for retinal exudation have focused on demonstrating fluorescein leakage in neovascularization and not fluid appearance in histology or OCT.^41–44 Thus, it is notable that high-resolution histology of osmicated tissues in our studies allows visualization of tissue fluid in cases of MNV^16,18 and in retinal pigment epithelium detachment associated with AMD.¹⁷ In our published cases, intraretinal fluid assumed a grayish appearance within spaces with borders that were either smooth cyst like^18,45 or ragged,¹⁶ and sometimes containing cells. One retinal pigment epithelium detachment was seen to have fluid phases differing in osmophilia, signifying a variation in lipid content, along with presumed cholesterol clefts.⁴⁵ In comparison with these published fluid instances, our examples of intraretinal fluid in the index case shown in Figures 1 and 3 are less symmetrical in shape and contain fluid as well as tissue components within them. Of note, all intraretinal fluid instances reported via clinicopathological correlation thus far occurred in the HFL. 

The presence of RPE cells within fluid pockets represents breaches of the external limiting membrane. Previously, we speculated that RPE cells could appear in ORT if they separated from the orthotopic layer (as the “sloughed” phenotype) and floated into the subretinal space. This space is scrolled by Müller glia into a tube that is lined by photoreceptors. If RPE cells appear within exudative fluid, they had to cross the external limiting membrane and travel through the retina.^19,46 Thus, RPE cells within these pockets did not result from a breach of the blood–retina barrier at the vessel wall. Moreover, the presence of RPE cells within fluid pockets suggests an active involvement of these cells in the pathogenesis and persistence of intraretinal fluid. 

Cellular swelling secondary to vascular hyperpermeability at an altered blood–retina barrier (vasogenic edema)^47–50 is one possible mechanism in neovascular AMD. In our index case, we identified only a few moderately swollen cell bodies (Fig. 3D) near one type 3 MNV neovessel stalk. However, owing to the ongoing long-term anti VEGF therapy and lack of electron microscopy, we cannot directly address hyperpermeability in the current case. Comparing retinal vascular diseases⁵⁰ with neovascular AMD, macular edema demonstrates different primary origins.⁵¹ In pars planitis,⁵² Irvine Gass syndrome,^52–54 retinitis pigmentosa,^47–49 diabetic retinopathy,^53,54 or branch retinal vein occlusion,^53,54 macular edema is associated with intracellular swelling (cytotoxic edema), mainly Müller glia, as well as vasogenic edema.^47–50 

Our data complement literature on fluid reflectivity in OCT. Barthelmes et al.¹² were the first to investigate optical densities of intraretinal spaces on OCT. Ahlers et al.⁵⁵ measured densities of exudative fluid relative to vitreous in neovascular AMD. Liang et al.⁵⁶ reported hyper-reflective solid material within retinal cysts on OCT in diabetic retinopathy, some comparable with our Figure 5A.These authors hypothesized that fibrin or other inflammatory byproducts account for this appearance. Horii et al.⁵⁷ correlated fluorescein pooling with reflective cystoid spaces and the presence of hard exudates. In diabetic macula edema, lipoprotein particles enter from plasma as well as proteins.⁴⁶ Kashani et al.,¹³ using OCT angiography, showed a strong decorrelation signal in cystoid spaces that could be modeled experimentally by an emulsion of 1% intralipid in 2% gelatin. The OCT angiography signal was attributed to Brownian motion of the emulsion. 

Our differences in intraretinal fluid reflectivity and stain intensity can be related to optical densities measured in subretinal fluid (SRF).^7,58 In SRF, optical density relative to internal reflectivity standards varies, implying different SRF composition. Thus, SRF was more reflective in exudative neovascular AMD than in central serous chorioretinopathy.⁷ Moreover, in eyes with rhegmatogenous retinal detachments, relative optical density was observed to increase over time.^58–60 SRF can be collected more easily than other fluids and is, thus, amenable to direct assay.⁶¹ Lipids detected in the SRF include membranes of damaged photoreceptor cells.⁶² 

Our stain intensity comparison pertains to proposed differences in reflectivity¹² and clinical lifecycle⁸ among fluids of exudative and degenerative origins. Our use of osmium postfixed material allowed visualization and quantification of stain density of lipid-containing fluid, including plasma in intact tissues. We found that all examples of exudative fluid stained darker than degenerative fluid in the same tissue section. We speculate that exudative fluid is stained darker than nonexudative fluid owing to higher vessel permeability to bigger particles, such as lipoproteins, in exudation compared with transudation. Tissue composition can impact OCT reflectivity via Mie and Raleigh scattering of light, depending on the size of the scattering particles.^63,64 Solid intravascular particles like blood cells and lipoprotein particles are, thus, key components of hyper-reflectivity in intraretinal fluid. Blood contains 45% volume of cells and 55% volume of plasma. Plasma contains lipoprotein particles, which are multimolecular assemblies specialized for transport of core lipids (triglycerides and esterified cholesterol) and surface lipids (phospholipids and unesterified cholesterol) and apolipoproteins.^65–67 Although 17% of plasma volume consists of lipids, only 7%⁶⁸ consists of proteins (Supplementary Material). Hence, with 2.4-fold more lipids than proteins in blood to provide refractive index boundaries, lipids and lipoproteins are important potential sources of OCT reflectivity in exudative fluid. 

The location of intraretinal fluid accumulation might be associated with capillary loss in the deep capillary plexus on OCT angiography, as suggested by the limited OCT angiography data from our index case. The relation between capillary dropout and fluid accumulation could be qualitatively confirmed (Supplementary Fig. S1). Previous OCT angiography studies suggested that fluid accumulates at sites of capillary loss in diabetic macular edema and branch retinal vein occlusion.^69,70 In chronic diabetic cystoid macular edema, cystoid spaces were located within capillary dropout areas, and tissue did not reperfuse after the edema resolved.⁶⁹ In branch retinal vein occlusion, nonperfusion in parafoveal capillaries colocalized with leakage on fluorescein angiography, with capillary tortuosity and dilated vessels surrounding areas of dropout involving both superficial and deep plexuses. Thus, future investigation of larger samples is needed to confirm this finding. 

Our findings support artificial intelligence (AI)-based image analysis for functional outcomes in exudative disease management.⁷¹ AI-based algorithms can precisely and reliably segment fluid for objective quantification of fluid, volumes, and resolution over time.^72,73 With automated AI algorithms becoming commercially available, OCT evaluation of fluid dynamics under therapy will allow a distinction between harmful and stable macular fluids, and hence predict a patient's need for treatment.⁷⁴ Moreover, continuous anti-VEGF levels via intraocular delivery systems^75,76 and, potentially, tailored multitarget therapies^77,78 might affect fluid dynamics as well in the future. 

Strengths of this study include the availability of eye-tracked OCT imaging for one case, rapid preservation that largely maintained retinal attachment, comprehensive histological and photomicroscopy techniques to reveal tissue elements including lipid, and access to many similarly prepared donor eyes. Limitations include the facts that the clinicopathological correlation represents only one case, thus limiting generalizability, and that donor eyes lacked clinical documentation. In conclusion, we demonstrated retinal tissue damage in areas of prior exudation, interstitial and intracellular fluid accumulation, plus we systematically categorized exudative and nonexudative fluids, and contextualized different fluid appearances on OCT. Moving forward, further studies are warranted to investigate fluid composition using techniques for molecular characterization. We expect OCT technologies with greater axial resolution to provide more insight into in vivo fluid appearances.^79,80 Finally, distinguishing exudative from degenerative fluid, particularly with regard to dynamism, can be approached today with clinical or AI-assisted methods of large clinical trial imaging datasets. 

Acknowledgments

Supported by Genentech/Hoffman LaRoche, The Macula Foundation, Inc., New York, New York; unrestricted funds to the Department of Ophthalmology and Visual Sciences (UAB) from Research to Prevent Blindness, Inc., and EyeSight Foundation of Alabama. AB reports grants from the Werner Jackstädt-foundation. Purchase of the slide scanner was made possible by the Carl G. and Pauline Buck Trust. 

The Project MACULA website and recovery of human donor eyes for research was supported by NIH (grants R01EY06109 and P30 EY003039), EyeSight Foundation of Alabama, International Retinal Research Foundation, Edward N. and Della L. Thome Foundation, the Arnold and Mabel Beckman Initiative for Macular Research, and Research to Prevent Blindness, Inc. 

Author Contributions: All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Curcio and Berlin had full access to all the data in the clinical picture and take responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design: AB, LC, CB, RM, DF, KBF, and CC. Acquisition of data: RM, CB, JM, LC, AB, DC, KBF, and CC. Analysis and interpretation of data: AB, PR, LC, DF, KBF, and CC. Writing of manuscript: AB, PR, CB, LC, RM, DF, KBF, and CC. 

The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication. 

Disclosure: A. Berlin, None; J.D. Messinger, None; C. Balaratnasingam, None; R. Mendis, None; D. Ferrara, Genentech/Roche (E); K.B. Freund, Genentech (C), Zeiss (C), Heidelberg Engineering (C), Allergan (C), Bayer (C), Novartis (C); C.A. Curcio, Genentech/Roche (F), Apellis (C), Astellas (C), Boehringer Ingelheim (C), Character BioSciences (C), Osanni (C)