SPECKLED HYPOAUTOFLUORESCENCE AS A SIGN OF RESOLVED SUBRETINAL HEMORRHAGE IN NEOVASCULAR AGE-RELATED MACULAR DEGENERATION

Abstract

Purpose:

To describe patterns of hypoautofluorescence in eyes with neovascular age-related macular degeneration (AMD) occurring after subretinal hemorrhage.

Methods:

This was a retrospective descriptive analysis of neovascular AMD eyes presenting with subretinal hemorrhage over the last 5 years that underwent serial multimodal imaging. A review of color fundus photographs (CFP), fundus autofluorescence (FAF), near-infrared reflectance, and optical coherence tomography was performed at baseline and all available follow-up visits to document the course and evolution of subretinal hemorrhage in these eyes.

Results:

Eleven eyes of 10 patients (9 female, 1 male; mean age: 84.1 years, range: 72–99 years) with a mean follow-up of 19.8 months (range: 3–68 months) were included. Color fundus photographs showed subretinal hemorrhage that resolved over a mean of 5.5 months. During and after hemorrhage resolution, all eyes showed hypoautofluorescence which appeared distinct from that due to retinal pigment epithelium (RPE) loss. Discrete multifocal punctate hyperpigmented lesions were observed in 90% of eyes, being markedly hypoautofluorescent and producing a speckled pattern on FAF.

Conclusion:

Areas of hypoautofluorescence in the absence of RPE atrophy, often with a speckled pattern, delineate areas of prior subretinal hemorrhage long after its resolution in patients with neovascular AMD. Potential mechanisms for development of this pattern are proposed.

Summary Statement:

This study describes a pattern of speckled hypoautofluorescence without RPE atrophy occurring in eyes with neovascular AMD following resolution of subretinal hemorrhage. We document the course and evolution of subretinal hemorrhage in treated AMD with multimodal imaging using color fundus photographs, fundus autofluorescence, near-infrared reflectance, and optical coherence tomography.

Introduction:

Subretinal hemorrhage is a serious complication of neovascular age-related macular degeneration (AMD), often resulting in severe and permanent loss of central vision.¹ Risk factors for subretinal hemorrhage in neovascular AMD include RPE tears, aneurysmal type 1 neovascularization (polypoidal choroidal vasculopathy), and the use of anticoagulants.^2–4 In addition, one study showed that eyes with subretinal hemorrhage due to neovascular AMD have a greater than 50% risk of recurrent hemorrhage within 2 years.⁵ Subretinal hemorrhage can occur shortly following an examination in which both funduscopy and optical coherence tomography (OCT) have shown no evidence of intraretinal or subretinal fluid.⁶

Fundus autofluorescence (FAF) is a non-invasive imaging modality that is often used to assess the status of the retinal pigment epithelium (RPE). FAF signal originates predominantly from lipofuscin and melanolipofuscin inclusion bodies within the RPE and is modulated by presence of other fluorophores found in adjacent tissues.⁷ Characteristic FAF patterns have been described for each of many genetic, degenerative, and inflammatory retinal diseases.⁸ In addition, the status of the neurosensory retina can be indirectly evaluated with FAF, as reduced photopigment density due to transient or permanent outer retinal disruption can unmask the FAF of healthy RPE producing focal or zonal areas of hyperautofluorescence.⁹ In AMD, the presence of diffuse hyperautofluorescence surrounding areas of neovascularization correlates with areas of prior subretinal fluid (“exudative flood plain”).¹⁰ The FAF findings following resolution of subretinal hemorrhage in AMD have not been described in detail.

Herein we describe areas of hypoautofluorescence, often with a speckled pattern, observed in patients with neovascular AMD during and after resolution of subretinal hemorrhage. This pattern of hypoautofluorescence due to prior subretinal hemorrhage differs distinctly from that related to prior subretinal fluid, as it overlies an area of RPE that does not appear to be attenuated on OCT.

Methods:

The study complied with the Health Insurance Portability and Accountability Act of 1996, followed the tenets of the Declaration of Helsinki, and received approval from the Western Institutional Review Board (Olympia, WA). Patients had signed informed consent forms for their charts and images to be utilized in an anonymous manner for educational and research purposes.

This was a retrospective review of consecutive patients with neovascular AMD who presented with subretinal hemorrhage and were followed with multimodal imaging during the course of anti-vascular endothelial growth factor therapy (Table). All study patients underwent a dilated fundus examination by a single retina specialist (KBF). Color fundus photographs (CFPs), pseudo-color ultra-wide field fundus photographs, and FAF images were acquired using the Topcon TRC-50XF fundus camera (Topcon Medical Systems, Paramus, New Jersey, USA) and Optos 200Tx (Optos, Dunfermline, United Kingdom). FAF images were acquired using 535–585 nm excitation wavelengths with the flood illuminated Topcon TRC-50XF, whereas the Optos non-confocal scanning laser ophthalmoscope (SLO) system uses a 532 nm excitation wavelength. Spectral-domain OCT with corresponding near-infrared reflectance (NIR) was performed using the Spectralis HRA + OCT (Heidelberg Engineering, Heidelberg, Germany). Eye tracking and image registration functions were enabled during image acquisition to compare the same OCT B-scans for successive visits. Quantitative fundus autofluorescence, when available, was performed on the Spectralis HRA + OCT (Heidelberg Engineering), a confocal SLO with a 488 nm excitation wavelength.

Table.

Summary of AMD patients with Subretinal Hemorrhage included in the Study

Case	Age (Yrs.)	Gender	Eye	Follow-up (Mos.)^	VA Prior Visit*	Treatment	Most Recent VA	Time to SRH Resolution (Mos.)
1	80	F	OD	42	20/70	Pneumatic/tPA/Ranibizumab	20/400	10
2	91	F	OD	11	20/300	Aflibercept	20/300	4
3	80	F	OS	14	CF	Aflibercept	CF	4
4	84	F	OD	11	20/40	Aflibercept	20/40	7
4	84	F	OS	15	20/40	Aflibercept	20/30	6
5	72	F	OD	9	20/400	Ranibizumab	20/200	4
6	90	F	OD	68	N/A	N/A	20/150	N/A
7	88	F	OD	14	N/A	Ranibizumab	20/30	5
8	81	F	OS	12	20/200	Aflibercept	20/250	5
9	74	M	OD	21	NA	Aflibercept	20/50	5
10	99	F	OS	3	CF	Aflibercept	CF	N/A

F=female, M=male, OD=right, OS=left, VA=visual acuity, Yrs.= years, Mos.=months, tPA=tissue plasminogen activator, SRH=subretinal hemorrhage, N/A=not available

^{^}Indicates follow up since the onset of hemorrhage

^*VA prior visit indicates visual acuity at the most recent visit prior to onset of subretinal hemorrhage

A detailed analysis of the multimodal imaging of these cases of subretinal hemorrhage was performed and compared with a typical case of serous PED associated with subretinal fluid in AMD by three investigators (KBF, SAH, RDM) independently (Figure 1). The presence of subretinal hemorrhage, de-hemoglobinized blood, subretinal fibrosis, hypopigmentation indicating macular atrophy, and multifocal punctate hyperpigmented lesions was evaluated on CFPs. The presence of hypo- or hyperautofluorescence in areas of prior subretinal hemorrhage was analyzed on FAF. De-hemoglobinized blood was characterized by a white-yellowish appearance on CFPs, showing an intense hyperautofluorescence that allowed us to differentiate it from subretinal fibrosis, which appeared hypoautofluorescent. OCT B-scans were examined in detail in order to correlate tomographic features with the different findings on CFPs, ultra-wide field pseudo-color images, NIR, and FAF, in particular the multifocal punctate hyperpigmented lesions and areas of hyper/hypoautofluorescence. OCT terminology recommended by the Classification of Atrophy Meetings (CAM) program was used to describe the observed outer retinal changes.¹¹ Attenuation of outer retinal layers including the interdigitation zone (IZ), ellipsoid zone (EZ) and outer nuclear layer (ONL) were described as outer retinal atrophy (ORA). When ORA occurred in areas of choroidal hypertransmission due to disruption of the RPE band, the term “complete RPE and ORA (cRORA)” was used.¹¹ The terms incomplete ORA (iORA) and incomplete RORA (iRORA) were used to designate discontinuous changes (Figure 1).

Figure 1.

Persistent hyperautofluorescence caused by prior subretinal fluid in the right eye of a 74-year-old man with neovascular age-related macular degeneration. Color fundus photograph (CFP, first column); fundus autofluorescence (FAF, second column); near-infrared reflectance (NIR, third column), and optical coherence tomography (OCT) B-scan (fourth column). Green arrows on NIR indicate location and direction of the corresponding OCT B-scans. The blue-dashed area corresponds with the magnification of the OCT shown at the bottom row. At baseline (top row), CFP shows a central pigment epithelial detachment (PED) with overlying subretinal fluid (orange-arrowheads). The PED appears hypoautofluorescent on FAF, and the area of subretinal fluid shows hyperautofluorescence (orange-arrowheads). NIR demonstrates mild hyporeflectivity within the area of the PED and subretinal fluid (orange-arrowheads). The OCT shows subretinal fluid (orange-arrowheads) in association with subretinal hyperreflective material at the uppermost part of the serous PED as well as hyperreflective material adherent to the posterior surface of the elevated retinal pigment epithelium (RPE). At 4-year follow-up (second row), CFP reveals a central area of atrophy appearing hypoautofluorescent on FAF. The atrophy appears hyperreflective on NIR and as complete retinal pigment epithelium (RPE) and outer retinal atrophy (cRORA) on OCT, as indicated by areas of choroidal hypertransmission. The PED and subretinal fluid have resolved leaving persistent hyperautofluorescence where previously there had been subretinal fluid (orange-arrowheads), corresponding to an area of disrupted ellipsoid zone and thinned outer nuclear layer on OCT. The RPE band is preserved in this area indicating complete outer retinal atrophy (cORA; orange-arrowheads). At 7-year follow-up (third row), CFP demonstrates growth of macular atrophy showing a larger area of hypoautofluorescence, increased NIR, and OCT hypertransmission representing cRORA. The hyperautofluorescence related to prior subretinal fluid persists in the area of cORA (orange-arrowheads).On a magnified view of 7-year follow-up (bottom row) a clearly visualized area with attenuation of RPE and outer retinal layers (cRORA, green-arrowheads) contrasts with an area of previous subretinal fluid, in the absence of choroidal hypertransmission (cORA, blue-arrowheads).

Results:

Eleven eyes of 10 AMD patients (9 female, 1 male; mean age: 84.1 years, range 72–99) who presented with subretinal hemorrhage were included in the present study. Mean follow-up period from the diagnosis of hemorrhage was 19.8 months (range: 3–68). Demographic and clinical data are summarized in the Table All patients had color photographs, FAF, and OCT B-scans over the follow-up period. Color photographs revealed initial subretinal hemorrhage in all cases with gradual reabsorption over time (Figures 2, 3, and 4). Areas of de-hemoglobinized blood were observed as white-yellowish lesions within the area of hemorrhage (Figures 2 and 3). FAF images revealed marked hypoautofluorescence in the presence of subretinal hemorrhage (Figures 2, 3). With the resolution of the hemorrhage, the hypoautofluorescence became less intense, but persisted in all cases (Figures 2, 3). De-hemoglobinized blood was observed as bright hyperautofluorescent area (Figures 2 and 3, green arrowhead). The OCT B-scans demonstrated acute subretinal hemorrhages as heterogeneous hyperreflective subretinal material. As hemorrhages reabsorbed, varying degrees of attenuation of the outer retinal bands were observed overlying a preserved hyperreflective RPE band (cORA); notably, there areas were not associated with choroidal hypertransmission (Figures 2 and 3).

Figure 2.

Persistent hypoautofluorescence associated with prior subretinal hemorrhage in the right eye of an 84-year-old woman (Case 4) with neovascular age-related macular degeneration. Color fundus photograph (CFP, first column); fundus autofluorescence (FAF, second column); near-infrared reflectance (NIR, third column) and optical coherence tomography (OCT) B-scans (fourth column). The green arrows on NIR indicate the location and direction of the corresponding OCT B-scans. In all FAF panels, the orange arrowheads indicate the maximal extent of the hemorrhage, seen at 3-month follow-up.At baseline (first row) CFP reveals an area of subretinal hemorrhage adjacent to a vascularized pigment epithelial detachment. FAF demonstrates marked hypoautofluorescence in the area of hemorrhage, which also shows reduced signal on NIR. The OCT shows hyperreflective subretinal hemorrhage (blue-arrowheads) with some adjacent hyporeflective subretinal fluid. At 3-month follow-up (second row) CFP demonstrates partial reabsorption of the hemorrhage and an area of de-hemoglobinized blood (green arrowhead), with slight expansion of the area of hemorrhage seen on NIR (orange-arrowheads). OCT shows marked reduction in subretinal hemorrhage (blue-arrowheads). At 4-month follow-up (third row), reabsorption of the blood continues with minimal residual hemorrhage seen on CFP, and a small area of de-hemoglobinized hemorrhage (green-arrowhead). Persistent mottled hypoautofluorescence is evidenced on FAF, whereas NIR reveals isoreflectivity in this area of prior hemorrhage. The OCT shows complete reabsorption of the blood with evidence of incomplete retinal pigment epithelium (RPE) and outer retinal atrophy (iRORA) indicated by disruption of the ellipsoid zone and outer nuclear layer thinning (blue-arrowheads). The absence of hypertransmission indicates an absence of complete RPE and outer retinal atrophy. At 10-month follow-up (bottom row), CFP shows complete resolution of the hemorrhage with persistent hypoautofluorescence indicating its prior location (orange-arrowheads). This area is isoreflective on NIR, and corresponds with an area iRORA on OCT (blue arrowheads).

Figure 3.

Left eye of the 84-year-old woman shown in Figure 2 (Case 4) that developed subretinal hemorrhage 3 months after its onset in the fellow right eye. Color fundus photograph (CFP, first column) and fundus autofluorescence (FAF, second column) at baseline, 2, 3 and 4-month follow-up show the progressive resorption of the subretinal hemorrhage leaving persistent speckled hypoautofluorescence in the corresponding area (orange-arrowheads) that is distinct from an area of macular atrophy (red asterisk). An area of intensely hyperautofluorescent de-hemoglobinized blood that later evolves into subretinal fibrosis is first seen at 2-month follow-up (green arrowhead). At 7-month follow-up (third and fourth columns) CFP reveals complete resolution of the blood, with persistent speckled hypoautofluorescence in the corresponding area on FAF (orange-arrowheads). The optical coherence tomography (OCT) scan corresponding to the indicated area on FAF (green arrow) demonstrates incomplete retinal pigment epithelium and outer retinal atrophy (iRORA) seen as disruption of the ellipsoid zone and thinning of the outer nuclear layer. The absence of hypertransmission on OCT indicates preservation of the RPE. Magnified CFP (contrast adjusted), FAF and OCT images at 7-month follow-up corresponding to the area indicated by the white box in the 3^rd column are shown in the 4^th column. Multifocal punctate hyperpigmented lesions are seen on CFP which appear intensely hypoautofluorescent. These lesions correspond to focal areas of hyperreflective material at the anterior border of the RPE-Bruch’s membrane band on OCT.

Figure 4.

Long-term follow-up of persistent hypoautofluorescence in an area of prior subretinal hemorrhage in the right eye of an 88-year-old woman (Case 6) with neovascular age-related macular degeneration. Color fundus photograph (CFP, left column) and fundus autofluorescence (FAF, right column). At baseline (top row), an area of subretinal hemorrhage seen on CFP appears hypoautofluorescent on FAF (orange-arrowheads). At 2-month follow-up (middle row) some hemorrhage persists, but hypoautofluorescence is still evident in areas where there is no hemorrhage observed. At 5-year follow-up (bottom row) small areas of hypopigmentation representing macular atrophy and hyperpigmentation are present on CFP, with no evidence of subretinal hemorrhage. On FAF, the hypoautofluorescence corresponding to the prior subretinal hemorrhage is still evident (orange-arrowheads), as well as other areas of hypoautofluorescence corresponding with 2 areas of atrophy of the retinal pigment epithelium (red asterisk) and an area of hyperpigmentation (green asterisk).

Following resolution of hemorrhage, all eyes showed persistent diffuse hypoautofluorescence in areas of cORA; these areas did not change in size over the duration of follow-up up to 5 years (Figure 4). Notably, unlike geographic atrophy, the RPE band was not attenuated on OCT in these areas of hypoautofluorescence. Multifocal punctate hyperpigmented lesions were observed within the areas of persistent hypoautofluorescence in 9 eyes (90%) after resolution of hemorrhage (Figure 5, yellow arrowheads). These lesions were markedly hypoautofluorescent giving the larger zones of reduced hypoautofluorescence a characteristic speckled appearance (Figures 3 and 5). The multifocal punctate hyperpigmented lesions corresponded to focal areas of subretinal hyperreflective material at the anterior border of the RPE-Bruch’s membrane band on OCT (Figure 5). Using the false-coloring feature of quantitative FAF software, visualization of reduced FAF in the areas of prior hemorrhage was improved (Figure 6). Ultra-wide field pseudo-color and FAF imaging showed that the speckled hypoautofluorescent pattern following resolution of subretinal hemorrhage could occur outside of macula (Figure 7).

Figure 5.

Speckled hypoautofluorescence in areas of prior subretinal hemorrhage showing multifocal punctate hyperpigmented lesions (Case 9, top row and Case 2, bottom row).Color fundus photograph (CFP, left), fundus autofluorescence (FAF, right), and optical coherence tomography (OCT, bottom). Multifocal punctate hyperpigmented lesions seen on CFP appear markedly hypoautofluorescent on FAF, and correspond with subretinal hyperreflective material at the anterior border of the RPE-Bruch’s membrane band on OCT (orange-arrowheads).

Figure 6.

Quantitative fundus autofluorescence (qFAF) demonstrating marked hypoautofluorescence in an area of prior subretinal hemorrhage (Case 9). At baseline (top row) color fundus photograph (CFP) shows an area of subretinal hemorrhage. Fluorescein angiography demonstrates this area as intensely hypofluorescent due to blocking by the hemorrhage. At 1-year follow-up (middle-row), CFP shows complete resolution of the hemorrhage. Fundus autofluorescence (FAF) delineates an area of hypoautofluorescence in the area of previous hemorrhage. At 1-year follow-up (bottom row) there is intense hypoautofluorescence in both the color-coded (left) and mean (right) qFAF images.

Figure 7:

Wide field imaging showing peripheral subretinal hemorrhage due to peripheral exudative hemorrhagic chorioretinopathy in a 99-year-old woman with neovascular age-related macular degeneration (Case 10).First row (baseline), middle row (6-month follow-up) and bottom row (9-month follow-up). Pseudocolor fundus photograph (left column) and fundus autofluorescence (right column) demonstrate extensive areas of atrophy with a central area of subretinal fibrosis (red dashed area) showing different degrees of hypoautofluorescence. At 6-month follow-up, a large area of subretinal hemorrhage is observed in the superior periphery demonstrating intense hypoautofluorescence (orange-arrowheads). At 9-month follow-up, much of the hemorrhage has been reabsorbed leaving pigmentary changes (blue-asterisk). There is an area of temporal de-hemoglobinized hemorrhage (green-asterisk). Persistent hypoautofluorescence is observed in the area of prior subretinal hemorrhage (orange-arrowheads), with marked hyperautofluorescence in the area of de-hemoglobinized blood (green-asterisk)

Discussion:

Lipofuscin and melanolipofuscin are lysosome-related inclusion bodies that contain metabolic byproducts of the visual cycle and are the primary source of fundus autofluorescence (FAF) signal. Increased production or decreased clearance of lipofuscin/melanolipofuscin can cause hyperautofluorescence. Other causes of hyperautofluorescence include dysmorphia or multi-layering of RPE cells that increase path length of exciting light through fluorophores, autofluorescent subretinal debris originating from shed photoreceptor outer segments, and an unmasking of normal RPE FAF due to loss of overlying macular luteal pigment or photopigment.^12,13 In contrast, hypoautofluorescence can result from blockage of excitation and emitted light by overlying features such as media opacities and luteal pigment, and reduced fluorophore concentration in individual RPE cells due to loss or re-rearrangement of organelles.^8,14 The key finding we report in this long-term study is a speckled pattern of persistent hypoautofluorescence without RPE atrophy corresponding to areas of resolved subretinal hemorrhage in eyes with neovascular AMD.

Eyes with resolved subretinal hemorrhage and hypoautofluorescence showed findings of complete outer retinal atrophy (cORA) which on OCT included attenuation of the ellipsoid zone, outer nuclear layer thinning, a preserved RPE band, and absence of choroidal hypertransmission, which in one case persisted for up to 7 years of follow-up (Figure 4). This FAF pattern is highly specific and differs from the well-known pattern of hyperautofluorescence produced by persistent subretinal fluid in central serous chorioretinopathy (CSR) and neovascular AMD (Figure 1). Whereas hyperautofluorescence is seen in cases of subretinal fluid from a combination of unmasking of RPE autofluorescence due to loss of overlying photopigment and accumulation of photoreceptor debris in the subretinal space due to loss of apposition between the RPE and outer retina, hypoautofluorescence is observed in cases of resolved subretinal hemorrhage.^9,15 Also, the RPE band on OCT does not show attenuation or choroidal hypertransmission indicative of RPE atrophy that would explain the observed hypoautofluorescence. Lastly, a speckled hypoautofluorescent pattern was observed which is most evident in Figures 2, 3, and, 5. This pattern though subtle was also seen outside of the macula following resolution of subretinal hemorrhage in a patient with neovascular AMD and subretinal hemorrhage due to peripheral exudative hemorrhagic chorioretinopathy (Figure 7).

We build on previous reports by providing long-term follow-up of FAF patterns and high-resolution structural OCT findings documenting the natural course of subretinal hemorrhage in AMD. Previous studies have described a similar hypoautofluorescence to occur after bleeding in highly myopic eyes both in the presence and absence of underlying CNV, but not in AMD.^16,17 Though Moriyama et al. speculated that this pattern might be specific to highly myopic eyes and does not occur in AMD, we observed it universally in our series of eyes with subretinal hemorrhage related to AMD.¹⁶ In addition, Sawa et al. described intense hyperautofluorescence from “yellow, devitalized blood” after subretinal hemorrhage in a patient with a ruptured macroaneurysm that resolved over a year without treatment.¹⁸ They proposed that lipids in photoreceptors undergo iron-mediated transformation into toxic autofluorescent compounds that are phagocytosed by the RPE, thus causing damage to both photoreceptors as well as the RPE. Hyperautofluorescence from de-hemoglobinized blood has also been described in cases of subretinal hemorrhage resulting from blunt trauma.^19,20 We also observed focal intense hyperautofluorescence in some of our cases in the acute phase following the onset of subretinal hemorrhage (Figures 2 and 3).

The diffuse hypoautofluorescence occurring in these cases may be due to an effect of blood on RPE that is not detectable on OCT. False-coloring of gray-scale FAF (Figure 6) makes reduced FAF in these areas of previous hemorrhage more apparent. Animal studies have shown that iron is taken up from sub-RPE hemorrhage by RPE cells,^21,22 and this iron leads to oxidative degradation of the bisretinoid fluorophore A2E and photoreceptor damage. Recent studies have shown that A2E levels in human and monkey macula is low, ^23–25 and yet other to-be-discovered bisretinoids in these regions may contribute to these variations in FAF. Gelfand et al. demonstrated that in mice subretinal iron injection causes toxicity to the RPE that is mediated by the NLRP3 inflammosome.²⁶ Also in mice, work by Dunaief and co-workers has shown higher iron accumulation and RPE vacuolization in animals receiving IV iron compared to controls²⁷ as well as iron-induced RPE degeneration in knockout mouse models.^{28, 29} Iron toxicity is implicated in direct injury to the RPE that can impair its ability to process bisretinoids and lead to hypoautofluorescence in these experiments. In experimental subretinal hemorrhage in a rabbit model, the RPE was Prussian-blue positive for iron deposition at 2 weeks’ post-injury; longer time points were not reported. Recent work by Sparrow et al. has shown that in mice treated with an iron chelator, intracellular iron decreased with subsequent increase in bisretinoid levels and quantitative FAF.³⁰ Based on these reports, it is conceivable that iron toxicity following subretinal hemorrhage impairs RPE function, resulting in clinical hypoautofluorescence, with the exact mechanism to be determined.

The speckled hypoautofluorescence occurs in areas where the reflective RPE band does not appear to be attenuated on OCT, as indicated by the absence of choroidal hypertransmission. This finding readily distinguishes it from geographic atrophy and complete RPE and outer retinal atrophy (cRORA), a recently defined term.¹¹ The speckled appearance was associated with the presence of multifocal punctate hyperpigmented lesions. Similar hyperpigmented lesions have been recently described after resolution of submacular hemorrhage in the setting of aneurysmal type 1 neovascularization (polypoidal choroidal vasculopathy).³¹ The authors of that study hypothesized that the intense outer retinal disruption seen in these cases may induce hyperplasia, clumping and migration of RPE cells. Another possible mechanism is the deposit of hemosiderin.³¹ Speckled pigmentary changes have been also described in eyes with age-related choroidal atrophy.³² These cases may show round focal areas of hyperpigmentation on clinical examination and color photographs corresponding to outer retinal hyperreflective deposits on structural OCT. In our cases, we observed multifocal pigmentary deposits in 90% of the eyes, corresponding to focal thickening of the RPE or discrete hyperreflective blobs above the RPE band. The mechanism proposed for punctate hyperpigmented lesions in age-related choroidal atrophy is outer retinal ischemia.³² The nature of the focal subretinal material accounting for speckles after hemorrhage (Figure 5) remains to be determined but could be anteriorly migrated RPE cells, phagocytes of retinal or systemic origin that assemble in response to hemorrhage, or extracellular deposits of cellular debris. One possible explanation for multifocal punctate hyperpigmented and reflective lesions is direct toxic damage to the RPE. Previous studies showing iron accumulation in Bruch’s membrane and RPE cells demonstrated patchy hyperpigmentation in some of the evaluated eyes,³³ but did not describe or illustrate a speckled or punctate pigmentary pattern. However, in cultured human fetal RPE and ARPE-19 cell line, iron can upregulate melanogenesis,³⁴ possibly contributing to a speckled pigmented appearance in our cases, since subretinal iron levels are likely high due to the hemorrhage. In experimental subretinal hemorrhage in a rabbit model, the RPE was Prussian-blue positive for iron deposition at 2 weeks.²²

In conclusion, we describe a speckled pattern of persistent hypoautofluorescence without RPE atrophy corresponding to areas of resolved subretinal hemorrhage in eyes with neovascular AMD. Although the flood-illuminated Topcon fundus camera uses a band of longer excitation wavelengths (535–585 nm) compared to the Optos scanning laser ophthalmoscope (SLO) using a single 532 nm wavelength, the FAF pattern we describe herein was readily visible with both systems. Similarly, the FAF pattern was seen clearly with FAF acquired on the Heidelberg Spectralis HRA+OCT (Heidelberg Engineering, Germany), a confocal SLO system, using a shorter 488 nm excitation wavelength (images not shown). It is important to note that subtle differences in imaging results may exist between different FAF platforms, but these differences could relate not only to different excitation wavelengths, but other technical differences between imaging systems. The strengths of the present study include multimodal imaging analysis and long-term follow-up. Limitations of our observational study include the small number of cases, its descriptive and retrospective nature, and lack of quantitative analysis. A direct clinicopathologic correlation will be helpful in elucidating the pathophysiology underlying this speckled hypoautofluorescent pattern without RPE atrophy.