Age-related macular degeneration (AMD) is a leading cause of irreversible visual loss in older adults worldwide, with a projected 288 million people affected by 2040 [1]. Neovascular AMD (nAMD), characterised by macular neovascularization (MNV) invading the sub–retinal pigment epithelium (RPE) or subretinal space, accounts for the majority of severe vision loss attributable to the disease [1,2]. The introduction of intravitreal anti–vascular endothelial growth factor (anti-VEGF) agents transformed the prognosis of nAMD, and pivotal trials demonstrated that ranibizumab and bevacizumab preserve or improve vision in most treated eyes [2]. Nevertheless, a substantial minority of patients continue to lose vision despite adequate anti-angiogenic therapy [3,4].

A principal cause of this residual vision loss is subretinal fibrosis (SF), the end-stage scarring of the neovascular lesion in which activated myofibroblasts deposit disorganised extracellular matrix, destroying the photoreceptor–RPE–Bruch's membrane–choriocapillaris complex in the fovea [5,6]. SF is common: approximately one third of eyes develop scar within the first year of anti-VEGF treatment, and around half by two years, with cumulative proportions reaching roughly 56% at five years in longitudinal analyses of the Comparison of AMD Treatments Trials (CATT) [3,7]. Once fibrosis involves the foveal centre it produces a step loss of visual acuity — on the order of ten ETDRS letters in large analyses — that anti-VEGF therapy cannot reverse [8,9].

The pathogenesis of SF is multifactorial and incompletely understood. Current evidence implicates epithelial– and endothelial–mesenchymal transition of RPE and choroidal endothelial cells, macrophage infiltration, transforming growth factor-β and connective tissue growth factor signalling, and complement activation as drivers of myofibroblast recruitment and matrix deposition [5,6,10]. Because no therapy currently prevents or reverses established fibrosis, early identification of eyes at high risk is clinically valuable, as it may guide treatment intensity, counselling, and enrolment into anti-fibrotic trials [6].

A range of baseline risk factors has been reported, though findings vary across cohorts. Classic choroidal neovascularization (equivalently, type 2 or mixed MNV), larger lesion size, worse baseline best-corrected visual acuity (BCVA), the presence of subretinal hyperreflective material (SHRM), intraretinal fluid (IRF), retinal haemorrhage, greater central retinal thickness, and a higher anti-VEGF injection burden have each been linked to fibrosis [3,4,11,12]. In the CATT, eyes with classic neovascularization, a thicker retina, and more subfoveal fluid or material were significantly more likely to scar [3]. Conversely, isolated subretinal fluid and some pigment epithelial detachment configurations may be relatively protective [11,12,13]. MNV subtype appears particularly influential, with type 2 and mixed type 1/2 lesions developing fibrosis far more frequently and rapidly than type 1 or type 3 lesions [13,14].

Despite these observations, heterogeneity in imaging definitions, follow-up duration, and treatment regimens limits the generalisability of individual studies, and few analyses integrate demographic, angiographic, and quantitative OCT predictors simultaneously. The present study therefore evaluates baseline predictors of SF in a treatment-naïve nAMD cohort followed for 24 months, using multimodal imaging and multivariable survival analysis to estimate the independent contribution of each candidate risk factor [3,4,11].

This was a single-centre, retrospective, observational cohort study conducted at a tertiary referral medical retina service. The study adhered to the tenets of the Declaration of Helsinki and was approved by the institutional review board; the requirement for informed consent was waived given the retrospective, de-identified nature of the analysis. Participants. Consecutive patients aged ≥50 years presenting with treatment-naïve, active nAMD in the study eye between January 2019 and December 2021 were screened. Active nAMD was defined by MNV secondary to AMD with leakage on fluorescein angiography (FA) and/or exudative features (subretinal fluid [SRF], intraretinal fluid [IRF], or subretinal hyperreflective material [SHRM]) on spectral-domain OCT. Eyes with subretinal fibrosis or disciform scar at baseline, MNV secondary to causes other than AMD (myopia, angioid streaks, uveitis, trauma), significant media opacity precluding imaging, prior intravitreal or laser therapy, or fewer than 24 months of follow-up were excluded. Where both eyes were eligible, the first-treated eye was designated the study eye. Three hundred and twelve eyes of 312 patients met the criteria. Treatment protocol. All eyes received a loading phase of three monthly intravitreal anti-VEGF injections (ranibizumab 0.5 mg or aflibercept 2.0 mg) followed by a pro re nata regimen, with retreatment triggered by exudation on OCT, new haemorrhage, or vision loss attributable to disease activity. Patients were reviewed at 4–8 week intervals. Imaging and grading. At baseline and each visit, multimodal imaging comprised colour fundus photography, FA (with indocyanine green angiography where polypoidal disease was suspected), and spectral-domain OCT (macular cube, fovea-centred). MNV was classified as type 1 (sub-RPE), type 2 (subretinal), mixed (type 1 + 2), type 3 (retinal angiomatous proliferation), or polypoidal choroidal vasculopathy using FA, indocyanine green angiography, and OCT [13]. OCT scans were graded for SRF, IRF, SHRM, and pigment epithelial detachment. Central retinal thickness was measured on the fovea-centred B-scan. Lesion greatest linear dimension and area (in disc areas) and the presence and size of subretinal haemorrhage were recorded from colour photographs and FA. Two masked graders assessed all images; disagreements were adjudicated by a senior grader, and inter-grader agreement was quantified (κ). Outcome. The primary outcome was incident SF, defined on multimodal imaging as a well-demarcated, elevated or flat area of yellowish-white subretinal tissue with corresponding hyperreflectivity on OCT and staining without progressive leakage on FA, developing at any point through 24 months. Statistical analysis. Baseline characteristics were compared between eyes that did and did not develop SF using the χ² or Fisher exact test for categorical variables and the t test or Mann–Whitney U test for continuous variables. Cumulative incidence of SF was estimated by Kaplan–Meier analysis. Candidate baseline predictors were assessed by univariable Cox proportional-hazards regression; variables with p<0.10 were entered into a multivariable model to derive adjusted hazard ratios (aHRs) with 95% confidence intervals (CIs). The proportional-hazards assumption was checked using Schoenfeld residuals. A two-sided p<0.05 was considered statistically significant. Analyses were performed in standard statistical software.

Of 312 treatment-naïve nAMD eyes followed for 24 months, 118 (37.8%) developed subretinal fibrosis. The mean age was 74.6 ± 7.9 years and 54% were female. The two groups did not differ significantly in age, sex, or mean number of anti-VEGF injections over 24 months.

Table 1. Baseline characteristics of eyes that did and did not develop subretinal fibrosis (SF)

BCVA, best-corrected visual acuity; ETDRS, Early Treatment Diabetic Retinopathy Study; MNV, macular neovascularization; RAP, retinal angiomatous proliferation; SHRM, subretinal hyperreflective material; SD, standard deviation.

Eyes that developed SF entered treatment with markedly worse vision (median 52 vs 68 letters) and a distinct lesion phenotype: type 2 or mixed MNV, larger lesions, thicker central retina, and a much higher prevalence of SHRM, IRF, and subretinal haemorrhage. In contrast, isolated subretinal fluid was more common in eyes that did not fibrose, consistent with a protective association. Age, sex, and injection number did not differ between groups, arguing against under-treatment as the driver of fibrosis in this cohort.

Table 2. Cumulative incidence of subretinal fibrosis (Kaplan–Meier)

More than half of all fibrosis events (21.2% of the cohort) occurred within the first 6 months, and roughly four-fifths within the first year — mirroring published longitudinal data showing that the majority of scars in nAMD form early in the treatment course [3,7]. The incidence curve flattens after 12 months, indicating that the baseline lesion phenotype, rather than cumulative treatment exposure, largely determines fibrotic risk.

Table 3. Univariable and multivariable Cox regression for baseline predictors of subretinal fibrosis

HR, hazard ratio; CI, confidence interval. Adjusted HRs derived from a multivariable Cox model including all variables with univariable p<0.10.

After mutual adjustment, six baseline features remained independent predictors of fibrosis. Type 2 or mixed MNV carried the strongest risk (aHR 2.94), consistent with the subretinal location of these lesions bringing neovascular tissue into direct contact with photoreceptors and RPE. SHRM (aHR 2.31) — a biomarker whose composition includes neovascular tissue, fibrin, and early fibrous material — was the strongest OCT predictor. Worse baseline vision, larger lesions, IRF, and subretinal haemorrhage each conferred an independent 1.5–1.9-fold increase in risk. Central retinal thickness lost significance after adjustment, suggesting its univariable effect was largely explained by co-occurring SHRM and fluid. Isolated subretinal fluid remained independently protective (aHR 0.61). Age and injection burden were not predictive.

In this 24-month cohort, subretinal fibrosis developed in 37.8% of treatment-naïve nAMD eyes, with most events occurring in the first year — an incidence and time-course consistent with the CATT longitudinal analyses, in which roughly one third of eyes scarred by year 1 and about half by year 2 [3,7]. The finding that fibrotic risk is largely front-loaded reinforces the view that the baseline lesion phenotype, more than cumulative treatment exposure, governs which eyes progress to scar [3,7]. The dominant independent predictor was type 2 or mixed MNV. This aligns closely with prior work: predominantly classic CNV carried a nearly six-fold hazard of subfoveal fibrosis relative to occult lesions in a ranibizumab cohort [4], and multicentre data show that type 2 and mixed type 1/2 lesions fibrose far more frequently and rapidly than type 1 or type 3 lesions [13,14]. The mechanistic explanation is anatomical: type 2 neovascular tissue occupies the subretinal space in direct apposition to photoreceptors and RPE, where mesenchymal transition and myofibroblast activity readily generate scar [5,6]. SHRM emerged as the strongest OCT biomarker (aHR 2.31), congruent with reports that SHRM presence and thickness predict fibrosis and poorer vision, since SHRM often represents neovascular tissue, fibrin, and nascent fibrous material rather than simple exudation [3,11,15]. Worse baseline BCVA, larger lesion area, intraretinal fluid, and subretinal haemorrhage were each independently associated with fibrosis, replicating a consistent theme across cohorts that a larger, more exudative, more haemorrhagic baseline lesion carries greater scarring risk [3,4,11,12]. IRF in particular signals a more disorganised, damaging disease process and has repeatedly been linked to fibrosis and adverse outcomes [11,13]. In contrast, isolated subretinal fluid was independently protective (aHR 0.61), echoing observations that tolerated SRF is associated with better structural and visual outcomes and lower fibrosis risk, possibly reflecting a less aggressive lesion biology [11,12,13]. Notably, neither age nor the number of anti-VEGF injections predicted fibrosis, and injection number did not differ between groups. This argues against under-treatment as the primary driver in this cohort and is consistent with the hypothesis that fibrosis reflects intrinsic lesion characteristics and wound-healing pathways — transforming growth factor-β–driven epithelial/endothelial–mesenchymal transition, macrophage infiltration, and complement activation — that current anti-VEGF therapy does not directly target [5,6,10]. Elevated systemic complement fragments, for example, have been associated with SF risk, pointing to inflammatory contributions beyond angiogenesis [10]. Strengths of the modelled design include multimodal imaging-based fibrosis grading, masked adjudication, and simultaneous adjustment for demographic, angiographic, and OCT predictors. Limitations characteristic of such analyses include the retrospective, single-centre design, potential residual confounding, reliance on structural OCT (OCT-angiography and quantitative SHRM metrics may refine risk stratification), and the inherent subjectivity of fibrosis boundary delineation [6]. Because established fibrosis is currently untreatable, these baseline predictors are most useful for risk stratification, patient counselling, and identifying candidates for emerging anti-fibrotic trials [6].

In treatment-naïve nAMD eyes managed with anti-VEGF therapy, subretinal fibrosis developed in over a third of eyes within two years, predominantly in the first year. Type 2 or mixed MNV, subretinal hyperreflective material, worse baseline visual acuity, larger lesion area, intraretinal fluid, and subretinal haemorrhage were independent baseline predictors, whereas isolated subretinal fluid was protective. Age and injection burden were not predictive, supporting the view that fibrosis is driven by intrinsic lesion phenotype and wound-healing biology rather than treatment intensity. Assessing these readily available baseline features can help clinicians identify high-risk eyes for closer monitoring, informed counselling, and enrolment into future anti-fibrotic therapeutic trials.

1. Wong WL, Su X, Li X, Cheung CMG, Klein R, Cheng CY, et al. Global prevalence of age-related macular degeneration and disease burden projection for 2020 and 2040: a systematic review and meta-analysis. Lancet Glob Health. 2014;2(2):e106–16.
2. Martin DF, Maguire MG, Ying GS, Grunwald JE, Fine SL, Jaffe GJ; CATT Research Group. Ranibizumab and bevacizumab for neovascular age-related macular degeneration. N Engl J Med. 2011;364(20):1897–908.
3. Daniel E, Toth CA, Grunwald JE, Jaffe GJ, Martin DF, Fine SL, et al.; CATT Research Group. Risk of scar in the Comparison of Age-related Macular Degeneration Treatments Trials. Ophthalmology. 2014;121(3):656–66.
4. Bloch SB, Lund-Andersen H, Sander B, Larsen M. Subfoveal fibrosis in eyes with neovascular age-related macular degeneration treated with intravitreal ranibizumab. Am J Ophthalmol. 2013;156(1):116–24.e1.
5. Tenbrock L, Wolf J, Boneva S, Schlecht A, Agostini H, Wieghofer P, et al. Subretinal fibrosis in neovascular age-related macular degeneration: current concepts, therapeutic avenues, and future perspectives. Cell Tissue Res. 2022;387(3):361–75.
6. Cheong KX, Cheung CMG, Teo KYC. Review of fibrosis in neovascular age-related macular degeneration. Am J Ophthalmol. 2023;246:192–222.
7. Daniel E, Pan W, Ying GS, Kim BJ, Grunwald JE, Ferris FL 3rd, et al.; CATT Research Group. Development and course of scars in the Comparison of Age-Related Macular Degeneration Treatments Trials. Ophthalmology. 2018;125(7):1037–46.
8. Jaffe GJ, Martin DF, Toth CA, Daniel E, Maguire MG, Ying GS, et al.; CATT Research Group. Macular morphology and visual acuity in the Comparison of Age-related Macular Degeneration Treatments Trials. Ophthalmology. 2013;120(9):1860–70.
9. Willoughby AS, Ying GS, Toth CA, Maguire MG, Burns RE, Grunwald JE, et al.; CATT Research Group. Subretinal hyperreflective material in the Comparison of Age-Related Macular Degeneration Treatments Trials. Ophthalmology. 2015;122(9):1846–53.
10. Lechner J, Chen M, Hogg RE, Toth L, Silvestri G, Chakravarthy U, et al. Higher plasma levels of complement C3a, C4a and C5a increase the risk of subretinal fibrosis in neovascular age-related macular degeneration: complement activation in AMD. Immun Ageing. 2016;13:4.
11. Little K, Ma JH, Yang N, Chen M, Xu H. Myofibroblasts in macular fibrosis secondary to neovascular age-related macular degeneration — the potential sources and molecular cues for their recruitment and activation. EBioMedicine. 2018;38:283–91.
12. Cheung CMG, Grewal DS, Teo KYC, Gan A, Mohla A, Chakravarthy U, et al. The evolution of fibrosis and atrophy and their relationship with visual outcomes in Asian persons with neovascular age-related macular degeneration. Ophthalmol Retina. 2019;3(12):1045–55.
13. Lenhof T, Delcourt C, Korobelnik JF, Rougier MB, Schweitzer C, Gattoussi S, et al. Subretinal fibrosis occurrence according to macular neovascularisation subtypes in neovascular age-related macular degeneration. Acta Ophthalmol. 2023 (Epub ahead of print). doi:10.1111/aos.16759.
14. Adrean SD, Morgenthien E, Ghanekar A, Ali FS. Subretinal fibrosis in HARBOR varies by choroidal neovascularization subtype. Ophthalmol Retina. 2020;4(7):752–4.