Predictors of 24-month onset of macular fibrosis in type 3 macular neovascularisation

Abstract

AimsTo explore prognostic multimarker models for progression to macular fibrosis (MF) over 24 months specific to type 3 macular neovascularisation (T3 MNV).MethodsThis retrospective, exploratory, single-centre, cohort study comprised 65 eyes of 43 Caucasian patients with treatment naive T3 MNV, all with a 24-month follow-up post anti-VEGF therapy using a strict pro-re-nata (PRN) regimen. Data on demographic features, clinical findings, frequency of intravitreal treatments and optical coherence tomography biomarkers were collected at baseline and after 12 and 24 months of follow-up. Logistic regression models (LRM) and receiver-operating curve (C-index) analyses were performed to evaluate the prognostic ability of the studied biomarkers in discriminating between MF affected and unaffected patients.ResultsAt final follow-up, MF was present in 46.2% of eyes. Subretinal hyper-reflective material (SHRM) and subretinal pigment epithelium multilaminar hyper-reflectivity (multilaminae) emerged as significant predictors for MF, with adjusted odds ratios (OR) of 18.0 (95% CL 13.4 to 24.1) and 11.8 (95% CL 8.66 to 16.0), respectively. Additionally, the presence of multifocal lesions (OR 0.04, 95% CL 0.01 to 0.30) appeared to decrease the likelihood of MF. C-indexes for the selected LRMs ranged between 0.92 and 0.88, indicating a comparably high discriminant ability. Despite consistent treatment schedules between the two groups (MF: median intravitreal treatment (IVT) number=10.5, IQR=7; non-MF: median IVT=10, IQR=6), a decline in best-corrected visual acuity was noted in the group with MF onset over the 24-month follow-up (−13.0 ETDRS letters; 95% CL –22.1 to –3.9; p=0.006).ConclusionOur study identifies SHRM and multilaminae as relevant predictors of 24-month onset of MF in patients with T3 MNV. These findings enrich our understanding of the development of MF in T3 MNV and can guide improved risk prognostication. Future research should consider larger samples and prospective designs to validate these predictors.