In this cross-sectional observational study, we used gold standard, as well as novel multimodal retinal imaging modalities such as OCTA and PS-OCT, to analyze in detail outer retinal and microvascular differences between eyes with and without SF caused by nAMD. Based on published literature, patients with nAMD have a risk of approximately 35% of developing SF within the first year after diagnosis, further increasing thereafter.
5,6,43 Hence, only patients with a minimum history of 12 months of anti-VEGF treatment were consecutively included to provide a number of participants affected by SF that reflects a real-world scenario. In line with the literature, subretinal MNV (type 2 and mixed type MNV) was common in eyes with SF, whereas in the non-SF group, type 1 MNV was more prevalent (
Table 1).
5,6,43,44 All eyes with SF exhibited a preserved microvascular MNV network. Using quantitative metrics, we found that SF is associated with larger lesions and bigger caliber vessels as shown by the larger GVC and GLD, as well as the larger total lesion area and vessel area. Furthermore, the total number of junctions, total vessel length, total number of endpoints, and endpoint density were larger in the SF group.
We found a difference between eyes with and eyes without SF in GLD, as well as total lesion area and total vessels area, suggesting an association between larger MNV complexes and SF development. Bloch et al.
5 investigated the baseline characteristics associated with the development of fibrosis within two years in eyes with nAMD under anti-VEGF treatment. They found that an initial MNV area larger than 5 disk areas was associated with a significantly increased risk by a factor of 4.49 over that of a MNV area of 1.5 disk areas or smaller. However, trend testing did not show a significant overall relationship between treatment-naïve MNV area and the development of SF. In our study we did not compare MNV characteristics at baseline but after a median history of anti-VEGF therapy of more than three years, which has most likely changed the configuration and size of the neovascular complex. There was a significant difference in GVC between the two groups, suggesting vascular maturation and abnormalization resulting in thicker caliber vessels with high flow in SF.
45 The higher total number of vessel junctions and endpoints and the longer total vessel length in the SF group may be expected because of the larger total lesion area and total vessel area. The higher endpoint density indicates a higher number of open-ended vessel segments with fewer connections between the vessels, also indicating vessel abnormalization.
45
PS-OCT was used as an ancillary imaging modality for SF detection and was a sensitive tool particularly in clinically challenging cases with exudate and yellow-whitish material mimicking SF, as shown in previous studies.
17,18,38 Gräfe et al.
38 used PS-OCT in a group of patients with CNV clinically diagnosed with suspected fibrosis, doubtful fibrosis, or lesions suspected not to be fibrotic. The authors found that PS-OCT provided additional information and was useful for identification and quantification, particularly in doubtful cases.
38
In our study ORT was found significantly more often in the SF group than in the non-SF group. Photoreceptors may form ORT in retinal diseases associated with atrophy or degeneration of the RPE, most likely reflecting a survival mechanism of photoreceptors in areas where the photoreceptor support system (i.e., the RPE and choriocapillaris) is missing.
21 There was no difference in the presence of RPD between the SF and non-SF group in our study population, suggesting no particular role for RPD in fibrosis development. HRF had a higher prevalence in the non-SF group than in the SF group, suggesting more intraretinal RPE migration in clinically active MNV and less intraretinal RPE migration once SF has developed. This finding is in line with histopathologic observations, where intraretinal RPE cells were typically found in eyes with CNV and “entombed” RPE cells were seen in areas of fibrovascular scars.
46 We previously used PS-OCT to selectively image birefringent fibrous tissue and facilitate differentiation from depolarizing RPE to identify, when nonfibrous subretinal hyperreflective material would turn into SF.
17,18 We found that presence of SF was associated with loss or porosity of the RPE layer. We observed continuous RPE in eyes with non-fibrotic subretinal material, but observed only disseminated depolarizing particles in areas of SF, presumably remnants of dissociated RPE. This finding fits well the concept of the epithelial-mesenchymal transition, where, as a result of tissue injury, RPE cells undergo a transformation into a myofibroblast cell form, which contributes to collagen deposition and SF formation.
47 RPE migration toward a MNV membrane, regularly observed in type 2 MNV, is commonly believed to be beneficial and may serve a barrier function to prevent extension of the neovascular membrane.
48 Although inhibition of this natural defense mechanism could be detrimental, inhibition of the epithelial-mesenchymal transition may be favorable in eyes with nAMD. The transformation of RPE cells to myofibroblasts has been identified as a possible target for interventional treatment to prevent SF formation and associated functional loss.
10,49 A combination of anti-VEGF with anti-fibrotic therapies may be necessary to simultaneously treat MNV exudation, as well as SF formation.
The authors are aware of certain limitations of the study, including the small number of study participants, as well as the cross-sectional study design. The focus of this study was on differences between eyes with and without SF; however, differences between the groups may have been caused by other underlying conditions such as MNV distribution. Another limitation was the fact that nine eyes from the non-SF group and one eye from the SF group had to be excluded from the quantitative analysis, which may have confounded the results. Despite using a long wavelength SS-OCTA device, allowing for better penetration into tissue, the presence of SF may have caused some shadowing on underlying structures. Furthermore, the software for quantitative OCTA analysis, even with parameter optimization, has limitations in accurate identification of vessels with low intensity, which may not have been perfectly segmented. PS-OCT is an imaging modality that is not commercially available, and, hence, interpretation of the additional tissue-specific imaging B-scans may require some experience. An advantage of the use of PS-OCT, however, was the recently implemented fibrosis detection algorithm, which objectively and automatically identifies SF. Positive aspects include a standardized examination protocol, as well as the use of novel imaging methods such as OCTA or PS-OCT, allowing visualization of different aspects of the MNV lesion that would otherwise not be detected using standard modalities.
In conclusion, using multimodal imaging our study demonstrated in vivo microvascular and morphological differences in eyes with MNV secondary to nAMD with and without SF. Eyes with SF tend to have larger MNV lesions with thicker vessels and are often associated with the presence of ORT.