Highlights
-
•
One-third of eyes with acute SMH retained driving-level vision at 12 months.
-
•
Subfoveal choroidal thickness independently predicted visual recovery.
-
•
Larger hemorrhage size and diabetes mellitus predicted poor outcome.
-
•
Structural OCT recovery parallels functional visual improvement.
-
•
Anti-VEGF monotherapy achieved meaningful visual improvement in most eyes.
Purpose
To identify clinical and imaging predictors of 12-month visual outcome in patients with fovea-involving submacular hemorrhage (SMH) due to neovascular age-related macular degeneration (nAMD) and polypoidal choroidal vasculopathy (PCV).
Design
Prospective observational clinical cohort study.
Subjects
Asian patients with nAMD or PCV presenting with fovea-involving SMH.
Methods
Between January 2016 and July 2024, eyes presenting with fovea-involving SMH ≥ 1 disc diameter with no evidence of blood organization were included. All patients received standard-of-care treatment. Clinical and multimodal imaging parameters were analyzed to identify predictors of visual outcome at 12 months.
Main Outcome Measures
Good (≤0.3 logMAR) and poor (≥1.0 logMAR) visual outcomes at 12 months.
Results
Among 487 treatment-naïve eyes, 75 met inclusion criteria for fovea-involving SMH. Mean (SD) age was 71.40 (9.59) years, 74.4% were male, and 76.0% had PCV. Mean hemorrhage size was 19.45 (22.65) mm². All eyes received anti-VEGF therapy, 2.6% underwent additional pneumatic displacement, and 5.3% received photodynamic therapy. Best-corrected visual acuity (BCVA) improved from 1.04 (0.67) logMAR to 0.69 (0.62) at 12 months. 25 eyes (33.3%) achieved good visual outcome, while 22 eyes (29.3%) had poor visual outcome. On multivariable analysis, younger age (OR = 0.47 per decade, 95% CI: 0.26-0.86; P =.02), better baseline BCVA (OR = 0.87 per 0.1 logMAR, 95% CI: 0.79-0.97; P <.01), and thicker subfoveal choroid (OR = 1.50 per 50 µm, 95% CI: 1.07-2.10; P =.02) independently predicted good visual outcome. Conversely, larger hemorrhage size predicted poor visual outcome (OR = 1.04 per mm 2, 95% CI: 1.01-1.07; P =.02).
Conclusion
One-third of eyes with SMH achieved 0.3 logMAR (Snellen equivalent 20/40) or better vision. In addition to age, baseline BCVA, and hemorrhage size, we identified choroidal thickness as a biomarker associated with visual outcome.
INTRODUCTION
S ubmacular hemorrhage (SMH) is an uncommon, vision-threatening complication of neovascular age-related macular degeneration (nAMD). In particular, polypoidal choroidal vasculopathy (PCV), a subtype of nAMD, has been associated with hemorrhagic presentation. The natural history of untreated large, central SMH is poor, with many eyes progressing to a final visual acuity (VA) worse than 20/200. Even with treatment, the prognosis is considered poor generally. , The presence of submacular blood is toxic to retinal tissue, resulting in iron-related oxidative damage, impaired metabolic exchange at the photoreceptor-retinal pigment epithelium (RPE) interface, and subsequent organization and fibrosis, which may rapidly lead to irreversible photoreceptor loss and macular scarring.
Furthermore, the optimal management strategy for SMH remains unclear. Management strategies include intravitreal anti-vascular endothelial growth factor (VEGF) therapy, pneumatic displacement with or without recombinant tissue plasminogen activator (tPA), pars plana vitrectomy, and combination approaches. The ongoing TIGER randomized controlled trial, comparing aflibercept monotherapy with aflibercept combined with early surgical displacement, is expected to provide important evidence to guide optimal treatment selection.
In view of the heterogeneity of clinical features and the variable outcomes, previous researchers have evaluated anatomical features in SMH. Several clinical and imaging factors have been suggested to influence prognosis in SMH, including baseline VA, duration of SMH, height and area of hemorrhage, and integrity of retinal microstructure. ,,, These studies suggest that imaging-derived biomarkers may improve prognostication in treatment-naive cases and support individualized management decisions. However, existing evidence is limited by relatively small sample sizes, retrospective design, or heterogeneous treatment regimens.
Therefore, the current prospective study aimed to describe the visual outcome of a cohort of patients with SMH secondary to nAMD, and to evaluate clinical and multimodal imaging predictors of visual outcome at month 12. Specifically, we sought to identify early structural features associated with achieving good vs poor best corrected visual acuity (BCVA ≤0.3 and BCVA ≥ 1.0 logMAR, respectively), to improve prognostication and inform patient counselling in routine clinical practice.
METHODS
This analysis is derived from the Asian Age-Related Macular Degeneration Phenotyping Study, a prospective, observational cohort of treatment-naïve, central-involving nAMD patients. For the current study, we identified a subset of consecutive eyes presenting with SMH at first clinical presentation and recruited between January 1, 2016 and July 1, 2024. The study adhered to the tenets of the Declaration of Helsinki, received approval from the centralized Institutional Review Board (Singapore) (Protocols R697/47/2009 and R1338/24/2016). Written informed consent was obtained from all participants.
ELIGIBILITY CRITERIA
Eyes were eligible if they presented with SMH involving the foveal center, measuring at least 1 disc diameter (∼2 mm) in greatest linear dimension and arising from newly diagnosed macular neovascularization (MNV). Hemorrhages were required to appear red or dark red, without signs of fibrosis or blood organization at presentation. Eyes with established organized hemorrhage, pre-existing macular scars, or subretinal fibrosis were excluded. Additional exclusion criteria included co-existing retinal disease that could confound visual outcomes (such as diabetic retinopathy, retinal vein occlusion, or myopic macular neovascularization), prior AMD treatment in the study eye, and poor-quality imaging precluding adequate baseline assessment.
IMAGING ACQUISITION AND EVALUATION
All patients underwent comprehensive examination and standardized multimodal imaging at baseline, month 3, and month 12. Imaging included color fundus photography (CFP) (TRC-50DX; Topcon, Tokyo, Japan), spectral-domain optical coherence tomography (OCT) with enhanced-depth with 25 B-scans per volume scan, near-infrared reflectance (NIR) imaging (Spectralis; Heidelberg Engineering, Heidelberg, Germany), and fluorescein and indocyanine green angiography (FA, ICGA) (Spectralis Heidelberg Retina Angiograph (HRA); Heidelberg engineering, Heidelberg, Germany or TRC-50DX; Topcon, Tokyo, Japan).
SMH was characterized as a red lesion on CFP corresponding to a hyporeflective area on NIR and a hyperreflective material under the neurosensory retina or the RPE on OCT. It was identified primarily on CFP, and its size was measured using calibrated digital calipers on Topcon CFP images. When borders were indistinct, lesion extent was verified on OCT B-scans, outlined on the corresponding NIR image, and measured by the in-built caliper tool ( Figure 1 ).
Representative examples of fovea-involving submacular hemorrhage (SMH). Color fundus photographs (CFP) and corresponding optical coherence tomography (OCT) B-scans from 3 representative eyes with central SMH secondary to neovascular age-related macular degeneration or polypoidal choroidal vasculopathy. (A1-C1) Baseline CFP demonstrating fovea-involving submacular hemorrhage of varying extent. (A2-C2) Corresponding central foveal OCT B-scans showing hyperreflective subretinal material consistent with hemorrhage, associated subretinal fluid, pigment epithelial detachment, and disruption of outer retinal layers. (A3-C3) CFP with overlay illustrating the measured hemorrhage area used for quantitative analysis.
Fibrosis and atrophy were characterized on multimodal imaging as well. Fibrosis was defined as well-demarcated, hyperreflective subretinal material with associated shadowing and loss of outer retinal layers on OCT, often corresponding to a yellowish, glistening lesion on CFP and hyperreflective area on NIR. Areas of chorioretinal atrophy were identified as well-demarcated pale lesions with visible underlying choroidal vessels on CFP, confirmed by corresponding hypo or absent autofluorescence on fundus autofluorescence (FAF) imaging when available, and by OCT evidence of outer retinal and RPE loss with choroidal hypertransmission.
MNV subtype was determined from baseline FA, ICGA, and OCT according to the Consensus on Neovascular AMD Nomenclature study group definitions. Lesions were classified as type 1, type 2, type 3, or PCV. Mixed type MNV was interpreted as type 1. MNV location was categorized as subfoveal or nonsubfoveal based on involvement of the foveal center on multimodal imaging. When baseline angiography was obscured by hemorrhage, month 3 imaging was used for lesion subtype classification.
OCT-derived parameters were evaluated at each time point and included:
-
•
Central subfield thickness (CST) automatically derived from the ETDRS thickness map.
-
•
Sub-foveal choroidal thickness (SFCT) measured manually on the central foveal B-scan from the outer border of the RPE to the choroid-scleral interface; SFCT measurements were not obtainable in some eyes due to signal attenuation from dense submacular hemorrhage.
-
•
Pigment epithelial detachment (PED) recorded as present or absent at the foveal center; when present, PED height was measured at the fovea.
-
•
Hyperreflective material (HRM) was defined as hyperreflective material located between the neurosensory retina and the RPE, beneath the RPE, or spanning both compartments. HRM was further classified as well-defined or ill-defined based on its margins and internal reflectivity. In eyes with submacular hemorrhage, subretinal HRM at baseline was taken to represent the hemorrhagic material, and its height was measured at the fovea.
-
•
Intra- and subretinal fluid (IRF/SRF) were documented as present or absent and were assessed across the entire scan volume.
OCT images were graded independently by 2 retina fellows for each eye. N.G. graded all images, and the second grading was performed by either M.S. or W.K. In cases of disagreement, grading was resolved by open discussion and consensus.
TREATMENT APPROACH
All patients received anti-VEGF therapy as first-line treatment following standard-of-care protocols. Pneumatic displacement was performed in selected cases (n = 2) at the discretion of the treating clinician, and photodynamic therapy (PDT) was applied in PCV eyes where clinically indicated (n = 4). No predefined interventional protocol was mandated.
OUTCOME MEASURES
Visual acuity was measured as best corrected visual acuity, with refraction performed as part of the study assessment. The primary outcome was 12-month visual outcome, categorized as good (BCVA ≤0.3 logMAR; Snellen equivalent ≥20/40) or poor (BCVA ≥ 1.0 logMAR; Snellen equivalent ≤20/200). Visual gain at 12 months was additionally categorized by change in BCVA as large gain (≥0.3 logMAR improvement), small gain (0-0.3 logMAR improvement), or no gain (≤0 logMAR change), and analyzed as an exploratory outcome. Secondary analyses assessed baseline predictors of visual outcome and described longitudinal changes in SMH and OCT-derived structural parameters over a 12-month period.
STATISTICAL ANALYSIS
Continuous variables were summarized as mean ± SD or median (IQR), depending on distribution; categorical variables were reported as counts and percentages. Comparisons between eyes achieving good vs poor visual outcomes were performed using appropriate parametric (t-test) or nonparametric (Mann-Whitney U) tests for continuous variables, and Fisher’s exact or X² tests for categorical variables. For exploratory analyses comparing visual gain categories (large gain, small gain, and no gain), continuous variables were compared using Kruskal–Wallis tests and categorical variables using Fisher’s exact tests. For longitudinal structural parameters, within-eye changes across baseline, month 3, and month 12 were evaluated using Friedman tests for continuous variables and Cochran’s Q tests for binary variables. Post hoc Wilcoxon signed-rank or McNemar tests were performed where appropriate. To identify predictors of visual outcome, univariate logistic regression analyses were first performed. Variables with clinical relevance or univariate association were then evaluated in multivariable logistic regression, with key analyses adjusted a priori for age and baseline BCVA. Odds ratios (OR) with 95% CIs (CI) were reported. Exploratory regression analyses were conducted for visual gain outcomes. A two-sided P -value <.05 was considered statistically significant. All analyses were performed using R statistical software (version 4.3.0).
RESULTS
Between January 2016 and July 2024, a total of 487 treatment-naïve eyes with newly diagnosed neovascular AMD or PCV were enrolled in the phenotyping cohort. Among hemorrhagic cases screened for eligibility, 58 eyes were excluded because of blood organization, fibrosis, or scarring at baseline. Ultimately, 78 eyes (16% of the full phenotyping cohort) met criteria for fovea-involving SMH ≥ 1 disc diameter, of which 75 (96.2%) completed 12-month follow-up and comprised the analysis cohort. Among these, 64 eyes (82.1%) completed multimodal imaging at month 12 and were included in longitudinal analyses.
The mean age of included SMH patients was 71.26 ± 9.57 years, compared with 71.15 ± 8.55 years in the remainder of the cohort ( P =.92). Males comprised 76.0% and 64.3% of these 2 groups, respectively ( P =.09). Systemic comorbidities were frequent and broadly comparable between groups. Hypertension was present in 63.5% vs 64.3%, diabetes mellitus in 28.0% vs 26.6%, history of smoking in 42.7% vs 41.9%, previous myocardial infarction in 13.5% vs 8.0%, and stroke in 6.7% vs 6.2% of patients, respectively (all P >.05). Aspirin therapy was used in 20.3% of SMH patients compared to 16.8% in the rest of the cohort ( P =.57) ( Table 1 ).
TABLE 1
Baseline Systemic and Ocular Characteristics of Eyes With Submacular Hemorrhage (SMH) Compared With the Remainder of the Phenotyping Cohort.
| Variable | SMH (n = 75) | Rest of Phenotyping Cohort (n = 409) | P -Value |
|---|---|---|---|
| Systemic features | |||
| Age, mean (SD) | 71.26 (9.57) | 71.15 (8.55) | .92 |
| Gender (male), n (%) | 57 (76.0%) | 263 (64.3%) | .09 |
| Race- Chinese, n (%) | 64 (85.3%) | 375 (91.7%) | .09 |
| Hypertension, n (%) | 47 (63.5%) | 258 (64.3%) | .90 |
| BMI, mean (SD) | 23.87 (4.36) | 24.65 (4.19) | .15 |
| Smoking, n (%) | 32 (42.7%) | 169 (41.9%) | .90 |
| Past myocardial infarction, n (%) | 10 (13.5%) | 32 (8.0%) | .12 |
| Past stroke, n (%) | 5 (6.7%) | 25 (6.2%) | .80 |
| Diabetes mellitus, n (%) | 21 (28.0%) | 107 (26.6%) | .78 |
| HbA1C, mean (SD) | 5.98 (1.14) | 6.06 (1.16) | .64 |
| Aspirin use, n (%) | 12 (20.3%) | 51 (16.8%) | .57 |
| Ocular features | |||
| LogMAR baseline BCVA, mean (SD) | 1.04 (0.67) | 0.71 (0.49) | <.01 |
| Axial length, mean (SD) | 23.54 (1.19) | 23.76 (1.39) | .17 |
| Lens status, Pseudophakic, n (%) | 26 (34.7%) | 142 (35.2%) | >.99 |
| Baseline CST, µm, mean (SD) | 658.42 (372.93) | 433.86 (161.34) | <.01 |
|
MNV type, n (%)
Type 1 type 2 Type 3 PCV |
(n = 75)
11 (14.7%) 3 (4.0%) 4 (5.3%) 57 (76.0%) |
(n = 402)
134 (33.3%) 63 (15.7%) 23 (5.7%) 182 (45.3%) |
<.01 |
BCVA = best-corrected visual acuity; CST = central subfield thickness; MNV = macular neovascularization.
Percentages calculated among eyes with available data.
MNV subtype available in 402 eyes in the non-SMH cohort.
Ocular characteristics were also generally comparable between groups. The mean axial length in the SMH group was 23.54 ± 1.19 mm vs 23.76 ± 1.39 mm in the remainder of the cohort ( P =.17). Pseudophakic eyes accounted for 34.7% vs 35.2% of eyes, respectively ( P >.9). As expected, significant differences were observed in visual and structural parameters. Baseline BCVA was markedly worse in SMH eyes (1.04 ± 0.67 vs 0.71 ± 0.49; P <.01), and CST was significantly greater (658.42 ± 372.93 µm vs 433.86 ± 161.34 µm; P <.01). Notably, the distribution of MNV subtypes differed between groups, with PCV accounting for 76.0% of SMH cases vs 45.3% in the rest of the cohort ( P <.01), while type 1 MNV was less frequent (14.7% vs 33.3%). Type 2 and type 3 lesions occurred in 4.0% vs 15.7% and 5.3% vs 5.7%, respectively. At baseline, mean hemorrhage size was 19.45 ± 22.65 mm², with 37.3% showing combined subretinal and sub-RPE involvement, 61.3% confined to the subretinal space, and 1.3% involving only the sub-RPE compartment.
FUNCTIONAL OUTCOMES
BCVA improved over time, from 1.04 ± 0.67 logMAR at baseline to 0.84 ± 0.66 logMAR at month 3 and 0.69 ± 0.62 logMAR at month 12, whereas in the remainder of the phenotyping cohort with available follow-up data (338/409 eyes), mean BCVA improved from 0.71 ± 0.49 logMAR at baseline to 0.43 ± 0.43 logMAR at month 12. At month 12, 25 SMH eyes (33.3%) achieved good visual outcome (BCVA ≤0.3 logMAR), while 22 eyes (29.3%) had poor visual outcome (BCVA ≥ 1.0 logMAR). The duration of reported visual symptoms before presentation was available for 61 eyes and averaged 34.96 ± 69.02 days. Among these, 40 eyes reported symptom duration of 14 days or less, while 21 eyes reported more than 14 days. The proportion presenting within 14 days was 76.2% in the good visual outcome group, 56.5% in the intermediate group, and 64.7% in the poor visual outcome group, and mean symptom duration did not differ significantly between groups. MNV subtype distribution, and subfoveal lesion location did not differ between outcome groups ( Table 2 ).
TABLE 2
Baseline Characteristics Stratified By 12-Month Visual Outcome.
| Characteristic | N |
Good Visual Outcome
(BCVA ≤0.3 logMAR) N = 25 |
Intermediate Visual Outcome
(0.3 < BCVA < 1.0 logMAR) N = 28 |
Bad Visual Outcome
(BCVA ≥ 1.0 logMAR) N = 22 |
P
-Value
Overall |
P
-Value
Good vs Poor Outcome |
|---|---|---|---|---|---|---|
| Systemic | ||||||
| Age, years, mean (SD) | 75 | 67.71 (9.24) | 71.42 (9.39) | 75.10 (9.02) | .04 | <.01 |
| Male, n (%) | 75 | 18 (72.0%) | 21 (75.0%) | 18 (81.8%) | .73 | .51 |
| BMI, mean (SD) | 75 | 24.50 (3.87) | 22.85 (4.34) | 24.44 (4.84) | .50 | .70 |
| Smoking, n (%) | 75 | 10 (40%) | 12 (42.9%) | 10 (45.5%) | .93 | .77 |
| Hypertension, n (%) | 75 | 12 (48.0%) | 21 (75%) | 13 (61.9%) | .13 | .35 |
| Hypercholesterolemia, n (%) | 75 | 15 (60.0%) | 18 (64.3%) | 14 (66.7%) | .89 | .64 |
| Myocardial infarction, n (%) | 75 | 3 (12.0%) | 2 (7.1%) | 5 (22.7%) | .27 | .40 |
| Stroke, n (%) | 75 | 1 (4.0%) | 1 (3.6%) | 3 (13.6%) | .43 | .30 |
| Diabetes mellitus, n (%) | 75 | 4 (16.0%) | 7 (25.0%) | 10 (45.5%) | .07 | .05 |
| Aspirin use, n (%) | 60 | 1 (5.0%) | 4 (20.0%) | 7 (35.0%) | .07 | .04 |
| Ocular | ||||||
| LogMAR baseline BCVA, mean (SD) | 75 | 0.75 (0.66) | 0.96 (0.52) | 1.47 (0.65) | <.001 | <.01 |
| Days since subjective vision drop, mean (SD) | 61 | 17.86 (23.26) | 22.42 (27.12) | 17.76 (18.59) | .93 | .99 |
| Phakic, n (%) | 75 | 16 (64%) | 20 (71.4%) | 13 (59%) | .65 | .80 |
| Axial length, mm, mean (SD) | 71 | 23.73 (1.21) | 23.53 (1.18) | 23.35 (1.20) | .56 | .29 |
| Hemorrhage size, mm², mean (SD) | 75 | 15.36 (18.89) | 14.22 (18.64) | 30.76 (28.06) | <.01 | .04 |
| PCV subtype, n (%) | 75 | 20 (80.0%) | 21 (75.0%) | 17 (77.3%) | .91 | >.99 |
| Subfoveal MNV, n (%) | 74 | 10 (40.0%) | 13 (46.4%) | 10 (47.6%) | .84 | .56 |
| CST, µm, mean (SD) | 75 | 599.68 (207.95) | 630.14 (204.95) | 794.95 (279.72) | .01 | .01 |
| SFCT, µm, mean (SD) | 66 | 265.96 (109.07) | 201.16 (100.23) | 185.28 (69.12) | .02 | <.01 |
| SFCT at month 3, µm, mean (SD) | 71 | 263.72 (111.54) | 198.96 (111.20) | 188.26 (82.48) | .02 | <.01 |
| PED, n (%) | 75 | 25 (100.0%) | 26 (92.9%) | 21 (95.5%) | .51 | .70 |
|
HRM location, n (%)
Subretinal Sub-RPE Both |
75
75 75 |
14 (56.0%)
1 (4.0%) 10 (40.0%) |
19 (67.9%)
0 9 (32.1%) |
13 (59.1%)
0 9 (40.9%) |
.72 | .637 |
| HRM height, µm, mean (SD) | 75 | 390.08 (214.79) | 387.43 (161.73) | 481.27 (214.34) | .13 | .15 |
| Treatment | ||||||
| IVT number, mean (SD) | 47 | 6.00 (2.48) | 6.64 (2.42) | 4.95 (2.87) | .07 | .19 |
| Photodynamic therapy, n (%) | 47 | 3 (12.0%) | 2 (7.1%) | 1 (4.5%) | .76 | .61 |
| Pneumatic displacement, n (%) | 47 | 1 (4.0%) | 0 | 1 (4.5%) | .53 | >.99 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
