Macular Microstructures and Prognostic Factors in Myopic Subretinal Hemorrhages

Abstract

Purpose.: To investigate microstructural changes and visual prognosis in myopic subretinal hemorrhages (mSH) without choroidal neovascularization (CNV).

Methods.: In this retrospective, observational case series, 13 consecutive eyes with mSH were followed for 6 months. The medical records, fluorescein angiography (FA), and spectral-domain optical coherence tomography (OCT) were reviewed. Fluorescein angiography confirmed the absence of CNV. The baseline and 6-month findings/parameters were investigated, including the maximal hemorrhagic height, intraretinal hyperscattering signal across the retina at 6 months (i.e., intraretinal hyperreflective sign), and integrity of the photoreceptor inner and outer segment (IS/OS) and external limiting membrane (ELM) lines.

Results.: The final visual acuity (VA) improved significantly (P = 0.001), and the hemorrhages resolved in 12 (92.3%) eyes by 6 months. The tops of the hemorrhages reached the outer nuclear layer (ONL) in three eyes (23.1%), internal limiting membrane (ILM) in five (38.5%), and between the two layers in five (38.5%). The intraretinal hyperreflective sign in all eyes extended into the ONL in five eyes (38.5%), to the ILM in four (30.8%), and between the two layers in four (30.8%). The location of the hyperreflective signs at 6 months coincided with the ruptured retinal layers at baseline in all eyes. The IS/OS line and the ELM were each intact in six (46.2%) eyes. The final VA was associated significantly with the IS/OS (P < 0.05) and ELM (P < 0.01) integrity.

Conclusions.: The intraretinal hyperreflective sign, presumed to be scarring that enters through the disrupted outer retina, is correlated closely with photoreceptor function.

Introduction

Myopia is attributable to maculopathy, such as a retinal detachment from a macular hole, ¹ myopic foveoschisis, ^2,3 chorioretinal atrophy, axial elongation, and choroidal neovascularization (CNV), ^4,5 and is a leading cause of blindness especially in east Asian populations. ^6,7 A submacular hemorrhage, sometimes referred to as a myopic subretinal hemorrhage (mSH), is a specific complication, the prevalence of which is approximately 3.1% in highly myopic eyes. ^8–10 An mSH is thought to be bleeding from the choriocapillaris because of axial elongation and consequent rupture of Bruch's membrane. ¹¹ Lacquer cracks, which have an incidence of approximately 4%, are believed to be a scarring reaction after rupture. ^4,12 The typical mSH, which occurs without development of CNV, has a relatively favorable outcome based on observation of 24 eyes. ¹³ However, the natural course is poorly understood because of the small number of studies. ^13–15  

Subretinal hemorrhages are also seen and are important in the disruption of the central visual function in pathologies such as age-related macular degeneration. Hemorrhages beneath the retina are thought to damage the photoreceptors. Toth et al. ¹⁶ reported a sequence of clot organization with tearing of the sheets of the photoreceptor inner and outer segment (IS/OS) in a cat model. It is generally accepted that histologic findings from postmortem eyes or animal models are informative; however, to our knowledge no reports have documented the in vivo retinal microarchitectural changes after subretinal hemorrhages. 

Recent technologic advances especially in optical coherence tomography (OCT) have enabled visualization of the retinal microarchitectural changes in vivo. Spectral-domain OCT (SD-OCT) obtains optimized images from the vitreous to the deep choroid. Because of the rapid scanning speed, SD-OCT minimizes the effect of ocular movement, allowing a more precise view of the retinal/choroidal architecture. ¹⁷ In the current study, we used SD-OCT to document the morphologic retinal changes after mSH and their relation to visual prognosis. 

Methods

Patients

We retrospectively studied 13 consecutive eyes of 13 patients with an mSH who presented to Osaka University Hospital, Osaka, Japan, from February 2008 to March 2012. An mSH was diagnosed as a subretinal hemorrhage without CNV in highly myopic eyes associated with lacquer cracks. High myopia was defined as a spherical equivalent refractive error (SERE) exceeding −6 diopters (D) or axial length (AL) longer than 26.0 mm. The inclusion criteria were a symptomatic mSH that included the fovea diagnosed by fundus observation and fluorescein angiography (FA); a duration of symptoms less than 2 weeks; follow-up visits at least after 1, 3, and 6 months; the availability of color fundus photographs and SD-OCT images from baseline and every follow-up visit; and the presence of new lacquer cracks beneath the hemorrhages from which the hemorrhages were suspected of originating, confirmed by indocyanine green angiography (ICGA). 

The exclusion criteria were a history of other ocular disorders and/or intraocular surgery or scleral buckling that might affect the visual acuity (VA) or AL, an mSH outside the fovea, the lack of clear FA and ICGA images from the initial visit or SD-OCT images from the initial and follow-up visits, and the presence of severe myopic chorioretinal atrophy seen as whitish round or oval lesions at the fovea, which may affect the VA. Ten patients were excluded because of unavailable SD-OCT examinations (n = 2), a follow-up period less than 6 months (n = 4), location of the mSH outside of the fovea (n = 1), a past history of mSH (n = 2), and severe retinal pigment epithelial (RPE) damage at the macula (n = 1). 

We reviewed the patient records and recorded the age, sex, mSH duration, best-corrected VA (BCVA), intraocular pressure, SERE, and AL; we also reviewed the color fundus photographs and SD-OCT images at baseline and 1, 3, and 6 months later. All patients provided informed consent for all examinations. The review board of Osaka University Hospital approved this retrospective study. The research adhered to the tenets of the Declaration of Helsinki. 

SD-OCT Examination and Interpretation

The SD-OCT scan protocol was a 6- × 6-mm cube scan containing 512 × 128 A-scans with 128 horizontal lines obtained using Cirrus HD-OCT (Carl Zeiss Meditec, La Jolla, CA) in 12 eyes and 73 raster lines in a 5.8- × 4.35-mm area using Spectralis HRA (Heidelberg Retinal Angiography)+OCT (Heidelberg Engineering, Heidelberg, Germany) in two eyes, or a 9- × 9-mm macular map scan containing 512 × 128 A-scans in RS-3000 (Nidek, Gamagori, Japan) in three eyes. 

One author (TA) reviewed all B-scan images. The parameters, including the maximal thicknesses of the hemorrhages and the distance from the central fovea, were measured using the OCT software. We also explored the presence/absence of intraretinal invasion of the hemorrhages, usually appearing as a hyperscattering signal on top of the hemorrhages at baseline (Fig. 1); the presence/absence of a thin hyperscattering signal across the neural retina at 6 months (intraretinal hyperreflective sign) (Fig. 2); and the integrity of the photoreceptor IS/OS line and the external limiting membrane (ELM) at baseline and 6 months. The maximal height of the intraretinal hyperreflective sign and the distance between the foveal center and the foot of the sign also were measured (Fig. 2). We evaluated the level of invasiveness or intraretinal hyperreflective sign by referring to the retinal layers as follows: outer nuclear layer (ONL) (level 1), between the outer plexiform layer (OPL) and the nerve fiber layer (level 2), and to the internal limiting membrane (ILM) (level 3). If multiple intraretinal hyperreflective signs were seen in one eye, the highest was measured. 

Figure 1

 

An OCT image shows the typical appearance of an mSH (asterisk) and invasion site (arrows) at baseline. The invasion site appears as a mound on top of the hemorrhage.

Figure 2

 

The height of the intraretinal hyperreflective sign is defined as the vertical distance (Height) from the RPE to the top of the intraretinal hyperreflective sign (arrow). The distance (D) from the foveal center (vertical dotted line) to the bottom of the intraretinal hyperreflective sign is measured (arrow).

Other Examinations

The size of the mSH lesion was measured on the baseline FA images using PDT software (Image Net; Topcon, Tokyo, Japan). The refractive error was measured by noncontact refractometry (Nidek), the IOP by noncontact tonometry (Topcon), and the AL by partial inferometry (IOLMaster; Carl Zeiss Meditec). Late-phase ICGA with confocal system (HRA2; Heidelberg Engineering) was used to locate lacquer cracks. 

Statistical Analysis

The paired t-test and correlation test were used to determine statistical significance between the various parametric comparisons (JMP version 8.0 software; SAS System, Inc., Cary, NC). The paired t-test was used to compare the initial and final VA levels. Linear regression analysis was used to explore the association of parameters. Analysis of variance was used to compare parameters among multiple categories. P < 0.05 was always considered significant. 

Results

Baseline Demographic Data

The mean patient age of the nine women and four men at baseline was 33.5 ± 11.7 years (range, 18–52 years). The baseline variables are shown in Table 1. The mean SERE was −13.11 ± 2.62 D (range, −19 to −7.875 D); the mean AL was 29.63 ± 1.38 mm (range, 26.99–31.46 mm); the mean logarithm of the minimum angle of resolution (logMAR) VA was 0.50 ± 0.25 units (range, 0–0.82); and the mean IOP was 15.8 ± 2.9 mm Hg (range, 10–20 mm Hg) (Table 1). 

Table 1

 

Baseline Demographic Data From 13 Patients With mSH

Parameter	Mean ± SD	Range
Sex	9 men, 69%	4 women, 31%
Age, y	33.5 ± 11.7	18–52
Spherical equivalent refractive error, D	−13.11 ± 2.62	−19 to −7.875
Intraocular pressure, mm Hg	15.8 ± 2.9	10.0–20.0
Axial length, mm	29.63 ± 1.38	26.99–31.46
Baseline logMAR units	0.50 ± 0.25	0–0.82
Lesion size at baseline, mm2	1.3 ± 0.6	0.6–2.3

Representative Case Reports

Case 1 was that of a 30-year-old man with high myopia referred for decreased vision and a central scotoma in the right eye. The BCVA was 20/63, IOP 15 mm Hg, and SERE −15.0 D in the right eye. A fundus examination showed a tigroid appearance with a submacular hemorrhage (Fig. 3A) in the right eye. Fluorescein angiography (Fig. 3B) and ICGA (Fig. 3C) confirmed the presence of an mSH without CNV. Baseline SD-OCT images showed a foveal hemorrhage (Figs. 3D, 3E). A magnified view showed that the hemorrhage was not a lesion that resembled a mound but was irregularly shaped and invaded the retinal layers (Fig. 3F). The hemorrhage reached to between the INL and OPL (Fig. 3F). 

Figure 3

 

Fundus photography, FA, ICGA, and SD-OCT at baseline and 6 months show the left eye of a 30-year-old man with an mSH. (A) A fundus photograph shows a dark reddish circular lesion at the macula from the subretinal hemorrhage. (B) Fluorescein angiography and (C) ICGA show hypofluorescent blocking from the hemorrhage. There are no hyperfluorescence signals in either image, indicating the absence of CNV. (D) A scanning laser ophthalmoscopy (SLO) OCT image shows the location of the OCT scans at baseline. (E) An OCT image shows a hyperscattering signal from the hemorrhage with unique morphology with invasion of the hemorrhage into the retinal layers. (F) The magnified image shows two processes from the hemorrhage (invasion sites). The dashed line indicates the border of the hemorrhage. (G) An SLO-OCT image at 6 months shows that the location is similar to the baseline location in (D). (H) An OCT image shows no subretinal hemorrhage but intraretinal hyperreflective signs across the retina (intraretinal hyperreflective sign). (I) A magnified image shows two intraretinal hyperreflective signs (arrows). The left one is perpendicular to the RPE; however, the right one is outside of the foveal center.

At 6 months, the mSH had resolved, and the VA improved to 20/30 (Figs. 3H, 3I). Spectral-domain OCT showed a hyperscattering signal vertically and across the retinal layers (intraretinal hyperreflective sign) (Fig. 3I). These two signs were exactly coincident with two lateral expansions of the hemorrhage initially seen as invasion sites at baseline (Fig. 3F), both of which seemed to reach approximately to the inner plexiform layer. We confirmed that the locations of the invasion sites of the hemorrhage and intraretinal hyperreflective sign coincided on the scanning laser ophthalmoscope OCT images (Figs. 3D, 3E, 3G, 3H). 

VA Time Course

The mean logMAR VA improved to 0.58 ± 0.31 (range, 0.22–1.00, P = 0.62 versus baseline) at 1 month, 0.35 ± 0.29 (range, 0–1.00, P = 0.08) at 3 months, and 0.21 ± 0.21 (range, −0.08 to 0.52, P < 0.01) after 6 months. The mean gains in the logMAR VA compared with baseline were −0.08 ± 0.30 (range, −0.48 to 0.60) at 1 month, 0.15 ± 0.24 (range, −0.30 to 0.48) at 3 months, and 0.30 ± 0.25 (range, −0.12 to 0.70) at 6 months. The mean duration before absorption of the hemorrhage based on color fundus photography was 3.5 ± 1.6 months (range, 2–7 months). No patient had a recurrence. 

There was a significant (P < 0.05) negative correlation between the VA at 6 months and the period of hemorrhagic absorption. However, there was no correlation between the final VA and sex, age, SERE, IOP, AL, baseline VA, or baseline lesion size (Table 2). 

Table 2

 

Correlation of Various Factors With Final VA at 6 Months in 13 Eyes With an mSH

Factors	P Value	R 2
Sex	0.45	−
Age, y	0.25	0.12
Spherical equivalent refractive error, D	0.66	0.02
Intraocular pressure, mm Hg	0.57	0.03
Axial length, mm	0.62	0.02
Baseline logMAR units	0.18	0.16
Absorption period, mo	<0.01*	0.53
Lesion size at baseline, mm2	0.95	0

*  Significant.

OCT Appearance at the Initial Visit

Cleavage of the outer retinal layers was detected over the hemorrhage in all 13 (100%) eyes. The cleavage had a variety of shapes and sometimes had multiple sites of invasiveness into the retinal layers (Fig. 3E). The innermost retinal layer that the hemorrhage reached was within the ONL in three (23%) eyes, between the ONL and ILM in five (38.5%) eyes, and in the ILM in five (38.5%) eyes. The greatest mean thickness of the hemorrhage was 228.0 ± 51.7 μm (range, 122–292 μm). The mean foveal thickness was 281.9 ± 56.5 μm (range, 214–412 μm). No factors were associated significantly with the final VAs (Table 3). 

Table 3

 

Relationship Between Baseline OCT Findings and Final VA in 13 Eyes With an mSH

OCT Finding	Mean	P Value	R 2
Foveal retinal thickness, μm	281.9 ± 56.5	0.24	0.13
Maximal hemorrhagic height, μm	228 ± 51.7	0.45	0.05
Grading of the highest layer of the hemorrhage	Level 1, 5 eyes, 38.5%	0.63	−
	Level 2, 5 eyes, 38.5%
	Level 3, 3 eyes, 23.0%

 Level 1, ONL; level 2, between OPL and nerve fiber layer; level 3, ILM.

OCT Appearance at 6 Months

The intraretinal hyperreflective signs seen in all 13 (100%) eyes involved the foveal center in 10 (77%) eyes. The location of the intraretinal hyperreflective sign corresponded exactly to that of the invasion site of the hemorrhage in all 13 eyes. The mean distance from the fovea to the foot of the sign was 23.9 ± 56.3 μm (range, 0–160 μm). The intraretinal hyperreflective sign extended to the ONL in five (38.5%) eyes, between the ONL and ILM in four (30.8%) eyes, and to the ILM in four (30.8%) eyes. The mean height of the intraretinal hyperreflective sign was 160.3 ± 37.7 μm (range, 91–221 μm). The IS/OS line was intact in six (46.2%) eyes and disrupted in seven (53.8%) eyes. The ELM was intact in six (46.2%) eyes and disrupted in seven (53.8%) eyes. Nineteen intraretinal hyperreflective signs were found, and six (31.6%) were perpendicular to the RPE. Thirteen (68.4%) signs were oblique to the RPE, seven of which (53.8%) were not near the foveal center but seemingly along Henle's nerve fiber, and six (46.2%) seemed to start from outside and extend toward the foveal center. All signs coincided with lacquer cracks. 

Relationship Between OCT Findings and Visual Outcomes

The final BCVA was correlated significantly with disruption of the IS/OS line (P < 0.05) and disruption of the ELM line (P < 0.01) at 6 months; however, the final BCVA was not correlated with other OCT findings (Table 4). The disruption of the IS/OS junction at 6 months was correlated significantly with the height of the intraretinal hyperreflective sign (P < 0.05) and disruption of the ELM signal (P < 0.001) at 6 months and tended to be correlated (P = 0.06) with the period of hemorrhagic absorption. However, there were no correlations between the disruption of the IS/OS and other OCT findings or the initial size of the mSH lesion (Table 5). 

Table 4

 

Relationship Between Final OCT Findings at the Macula and VA at 6 Months in 13 Eyes With mSH

	Mean ± SD	P Value	R 2
Categorical findings
Disruption of IS/OS line, present/absent		0.02*
Disruption of ELM line, present/absent		<0.01*
Numerical parameters
Foveal retinal thickness, μm	207.3 ± 17.4	0.20	0.16
Grading of the highest layer  of the intraretinal hyperreflective sign	Level 1, 5 eyes, 38.5%; level 2, 4 eyes, 30.8%; level 3, 4 eyes, 30.8%	0.18	0.29
Maximal height of intraretinal  hyperreflective sign, μm	160.3 ± 37.7	0.06	0.30

 Levels defined as for Table 3.

*  Significant.

Table 5

 

Correlation Between Disruption of IS/OS Line and Other OCT Parameters/Findings

	P Value	R 2
Baseline
Absorption period, mo	0.06	0.29
Foveal retinal thickness, μm	0.27	0.11
Hemorrhage invasion level	0.72	0.02
Hemorrhage height, μm	0.12	0.21
Hemorrhage lesion size, mm2	0.51	0.04
Final visit, 6 mo
Level of intraretinal hyperreflective sign	0.33	0.08
Height of intraretinal  hyperreflective sign, μm	0.03*	0.35
Distance from intraretinal hyperreflective  sign to foveal center, μm	0.87	0.00
Disruption of ELM line	<0.001*	0.68
Foveal retinal thickness, μm	0.24	0.14

*  Significant.

The disruption of the ELM at 6 months was correlated significantly with the disruption of the IS/OS at 6 months (P < 0.001) and the period of hemorrhagic absorption (P < 0.05), and it tended to be correlated with the height of the intraretinal hyperreflective sign at 6 months (P = 0.07) However, there was no correlation with other OCT findings (Table 6). 

Table 6

 

Correlation Between Disruption of the ELM Line and Various Factors

	P Value	R 2
Baseline
Absorption period, mo	0.03*	0.35
Foveal retinal thickness, μm	0.15	0.18
Hemorrhage invasion level	0.72	0.02
Hemorrhage height, μm	0.16	0.17
Hemorrhage lesion size, mm2	0.61	0.02
Final visit, 6 mo
Level of intraretinal hyperreflective sign	0.56	0.04
Height of intraretinal hyperreflective sign	0.07	0.27
Distance from intraretinal hyperreflective  sign to foveal center	0.23	0.14
Disruption of IS/OS line	<0.001*	0.68
Foveal retinal thickness	0.19	0.16

*  Significant.

Discussion

Other investigators have documented the OCT appearance at the onset of mSH and found significantly higher rates of IS/OS irregularity or defects in eyes with a final BCVA below 0.7. ¹⁵ However, according to a PubMed search, the current study was the first to follow the changes in the foveal microstructure of the mSH without CNV using SD-OCT, and explored the association with the outcomes. We also identified unique and specific findings, such as the invasion site and the intraretinal hyperreflective sign after subretinal hemorrhages, that have not been reported previously. These signs were found in highly myopic eyes; however, subretinal bleeding is very common in other macular diseases, and the current findings are highly likely to occur frequently in other macular conditions. For instance, age-related macular degeneration, polypoidal vascular choroidopathy, and macroaneurysms often present with subretinal bleeding. ¹⁸ However, unfortunately, the hemorrhages are usually much larger. The OCT appearance also is much more complicated to interpret due to a variety of pathologies beneath the retina, such as CNV formation, a fibrin reaction, and pigment epithelial detachments. 

The current study showed that mSH invaded the retinal tissue and reached beyond the ONL in eight (61.5%) eyes. This finding differed from those in previous reports, which described the microstructural changes after experimental submacular hemorrhages. Toth et al. ¹⁶ reported a model of subretinal hemorrhage in the domestic cat and found that fibrin was associated with tearing of sheets of the photoreceptor IS/OS. Later degeneration progressed to involve all retinal layers overlying the densest areas of fibrin in the clots. However, the investigators did not mention hemorrhages that invaded the photoreceptor segments. The histologic differences in the current study from those reported previously may have resulted from a few factors. One possibility is that the current study was conducted in humans and that of Toth et al. ¹⁶ was in cats. Another is that in the previous study, hemorrhages were induced by transsclerally creating a focal neurosensory retinal detachment using a micropipette technique with the animals under general anesthesia, while the hemorrhages in the current study were not. 

Other evidence of invasion of blood into the retinal tissue was the intraretinal hyperreflective sign, the hyperscattering thin and vertical signal extending from the RPE to the intraretinal layers that was seen in all eyes on SD-OCT at the final visit. We confirmed from the three-dimensional OCT scan that the location of the intraretinal hyperreflective sign coincided with the lesion of the hemorrhagic site of invasion at baseline. These findings led us to conclude that intraretinal hyperreflective sign is a scarring reaction of the glial cells to hemorrhages and sheared synapses after invasion. However, further studies are needed to confirm the origin of this signal. 

There were significant negative correlations between the final VA and the disruption of the foveal IS/OS and ELM signals at the final examination, which agreed with previous reports. ¹⁵ The lack of integrity of the IS/OS and ELM line, which indicated damage to the photoreceptors resulting from a hemorrhage, is a critical indicator of visual prognosis in various macular pathologies. ^19–23 The disruption of the ELM signal was correlated significantly with the duration of the hemorrhages in the current study, which showed that the presence of a persistent hemorrhage prevents restoration of the photoreceptor IS/OS and nuclear layers, resulting in poor visual recovery. This agrees with previous studies of subretinal hemorrhages. ^18,24–27 However, the height of the hemorrhage at the first examination was not correlated with the final disruption of the IS/OS or ELM signal, which led us to hypothesize that damage to the photoreceptors may be affected not only by the hemorrhagic volume but also by multiple complex factors. 

The height of the intraretinal hyperreflective sign was correlated significantly with disruption of the IS/OS and ELM signal, indicating that with more excessive hemorrhagic invasion into the retina, greater damage to the photoreceptors and subsequent poor visual recovery can occur. The organized hemorrhagic material releases iron, which is toxic to the retinal circulation and choriocapillaris ²⁸ ; contraction of the subretinal clot causes irreversible damage to the photoreceptors; and thick hemorrhages prevent the exchange of nutrients, oxygen, and metabolites between the retina and choriocapillaris, ^18,27–29 all leading to photoreceptor damage. The height of the intraretinal hyperreflective sign was not correlated directly with the VA. The visual recovery might be more sensitive to the damage to the central foveal photoreceptors. For instance, if the intraretinal hyperreflective sign is slightly outside the central area, the VA should not be affected. 

According to the pathology literature, ³⁰ “Because fibers in Henle's fiber layer are quite delicate and loosely arranged, this layer is very susceptible to deposition of transudates, exudates, hemorrhage, and other products.” Therefore, subretinal hemorrhages easily enter between the synapses. Histologic confirmation is awaited; however, we believe the hemorrhages coursed along with the retinal neurons, since many intraretinal hyperreflective signs run obliquely and away from the foveal center. The Müller cells are sandwiched between the synapses, and the hemorrhages are likely to damage the glial cells, which support the neurons in the fovea. Therefore, the photoreceptors in the fovea were damaged not only by the hemorrhages and the aforementioned mechanism but also by intraretinal shearing of the glial cell processes and synapses. This may partly explain why the hemorrhagic size/thickness and the invasion site and intraretinal hyperreflective signs were not as strongly associated with the VA as the IS/OS and/or ELM lines. These two signs more directly represent the photoreceptor function. 

The limitations of the current study included the retrospective design, the small number of eyes, and the relatively short-term follow-up period. In addition, the volume scan of the SD-OCT did not provide the best-quality images, since the images were not averaged. The next-generation SD-OCT instrument will have an averaging function in the volume scan, which should facilitate more accurate evaluation of the hemorrhages, intraretinal hyperreflective signs, and integrity of IS/OS and ELM line. The disruption of the IS/OS and ELM line also does not directly represent the photoreceptor cell damage, and there is no direct evidence regarding the origin of the hyperscattering signals seen at the baseline and final visits. It is generally agreed that the vertical distance is accurate in OCT images, but we used three kinds of OCT machines. Although the difference is reportedly as small as 3 to 4 μm, ³¹ we could not rule out the difference between different OCT machines. Further investigation is required. 

Acknowledgments

Disclosure: T. Asai, None; Y. Ikuno, None; K. Nishida, None