A
Characterisation and clinical impact of hypertransmission alterations in neovascular age-related macular degeneration using en face OCTA
A
Siqing
Yu
1
Mahnaz
Parian-Scherb
1
Katie
Patel
2
Isabel
Bachmeier
1
Philip
Enders
3
Lebriz
Altay
3
Michael
J.
Koss
4
Siamak
Ansari
Shahrezaei
5
Justus
G.
Garweg
6
Cathy
Cukras
1
Bjoern
Titz
1
Sascha
Fauser
7✉,8
Emailsascha.fauser@roche.com
Roche
Pharma
Research
7,8
Early
Development
7,8
1
F. Hoffmann-La Roche Ltd
Basel
Switzerland
2
Roche Products Ltd
Welwyn Garden City
UK
3
Faculty of Medicine
University Hospital Cologne
Cologne
Germany
4
Eye Center Nymphenburger Höfe
Munich
Germany
5
Karl Landsteiner Institute for Retinal Research and Imaging
Vienna
Austria
6
Department of Ophthalmology, Bern University Hospital
University of Bern, and Berner Augenklinik
Bern
Switzerland
7
Ophthalmology DTA
Roche Innovation Center Basel
124 Grenzacherstrasse
4070
Basel
Switzerland
8
A
Macular degeneration
Imaging techniques, Biomarkers
Siqing Yu
1
*, Mahnaz Parian-Scherb
1
*, Katie Patel
2
, Isabel Bachmeier1, Philip Enders3, Lebriz Altay3, Michael J. Koss4, Siamak Ansari Shahrezaei5, Justus G. Garweg6, Cathy Cukras1, Bjoern Titz1⁑ and Sascha Fauser1⁑
1F. Hoffmann-La Roche Ltd, Basel, Switzerland. 2Roche Products Ltd, Welwyn Garden City, UK. 3Faculty of Medicine and University Hospital Cologne, Cologne, Germany. 4Eye Center Nymphenburger Höfe, Munich, Germany. 5Karl Landsteiner Institute for Retinal Research and Imaging, Vienna, Austria. 6Department of Ophthalmology, Bern University Hospital, University of Bern, and Berner Augenklinik, Bern, Switzerland.
⁑Bjoern Titz and Sascha Fauser contributed equally as co-last authors.
Corresponding author contact details:
Sascha Fauser, Roche Pharma Research and Early Development, Ophthalmology DTA, Roche Innovation Center Basel, 124 Grenzacherstrasse, 4070 Basel, Switzerland.
Email: sascha.fauser@roche.com
Siqing Yu and Mahnaz Parian-Scherb contributed equally to this work.
Meeting presentations
This study has been presented as posters at the Association for Research in Vision and Ophthalmology Annual Meeting, May 5–9, 2024, Seattle, Washington, USA; EURETINA Congress, Sep 19–22, 2024, Barcelona, Spain; Association for Research in Vision and Ophthalmology Annual Meeting, May 3–7, 2025, Salt Lake City, Colorado, USA; EURETINA Congress, Sep 4–7, 2025, Paris, France.
A
Competing interests:Siqing Yu, Mahnaz Parian-Scherb, Isabel Bachmeier, Cathy Cukras, Bjoern Titz and Sascha Fauser are employees of F. Hoffmann-La Roche Ltd. Katie Patel is an employee and stockholder of Roche Products Ltd. Philip Enders reports financial support from Roche. Lebriz Altay reports financial support from Roche and has received lecture honoraria from AbbVie, Apellis, Bayer, Novartis and Roche. Michael J. Koss has received lecture honoraria from LumiThera, and Roche, travelling fees from ARC Laser and Munich Surgical Imaging, and is a stockholder of Ocutrx. Justus G. Garweg reports financial support, consulting fees and lecture honoraria from AbbVie, Bayer, Novartis and Roche. Siamak Ansari Shahrezaei has no potential conflict of interest or financial disclosure. All authors attest that they meet the current ICMJE criteria for authorship.
Running head
Hypertransmission alterations in nAMD
Address for reprints
Sascha Fauser, Roche Pharma Research and Early Development, Roche Innovation Center Basel, 124 Grenzacherstrasse, 4070 Basel, Switzerland
A
This article contains additional online-only material. The following should appear online-only: Figures S1–S6 and Table S1.
Subject terms:
Macular degeneration, Imaging techniques, Biomarkers
A
Abstract
Objectives
To (1) evaluate a method for analysing hypertransmission alterations (HTAs) in eyes with neovascular age-related macular degeneration (nAMD) using en face optical coherence tomography (OCT) images, and (2) investigate associations between HTA phenotypes and best-corrected visual acuity (BCVA).
Methods
HTAs were manually annotated on en face OCT projections (images from eyes with active nAMD from cross-sectional biomarker study) and classified into 3 qualitative phenotypes based on homogeneity, reflectivity, and border delineation. HTA quantitative metrics and associations between HTA phenotypes, metrics, clinical characteristics and BCVA were assessed.
Results
Images from 186 eyes were evaluated. Among gradable eyes (n = 136), HTA lesions were present in 88% (119/136) and classified as homogeneous (19%), heterogeneous (25%), and indeterminate (56%). Heterogeneous HTAs had highest foveal involvement; homogeneous HTAs exhibited higher reflectivity, border delineation, and multifocality; indeterminate HTAs were more common in younger patients with shorter disease duration. Significant associations with lower BCVA (R2 = 0.46) were found for HTA phenotype, larger square root HTA total area (− 5.1 letters/mm, 95% confidence interval [CI] − 7.9, − 2.3) and intraretinal fluid volume (− 1.3 letters/log2-transformed µm3; 95% CI − 1.9, − 0.8). After adjusting for other variables, heterogeneous HTAs were associated with a reduction of 19.0 letters (95% CI 9.8, 28.2; p < 0.001) and 16.8 letters (95% CI 9.3, 24.3; p < 0.001) compared with homogeneous and indeterminate phenotypes, respectively.
Conclusions
HTAs are prevalent in eyes with nAMD and can be objectively classified into distinct phenotypes associated with differences in morphology, demographics, spatial distribution in the macula, and BCVA.
INTRODUCTION
Optical coherence tomography (OCT) has emerged as a vital tool for evaluating atrophic lesions in non-exudative age-related macular degeneration (AMD), due to its ability to provide high-resolution, cross-sectional images of retinal and subretinal structures [1]. Geographic atrophy (GA) lesions appear on OCT B scans as increased signal transmission into the choroid when the hyperreflective retinal pigment epithelium (RPE) and outer retinal layers, which normally scatter the light, are absent or attenuated [2]. An en face sub-RPE slab presents an opportunity to visualise the area of hypertransmission defect (HyperTD). A hyperTD corresponds to the light hypertransmission into the choroid and appears as a brighter region compared with the surrounding area [3]. Previous studies have shown that the reproducibility of GA lesion measurements on en face OCT imaging is as good as or even better than the measurements obtained with color fundus photography (CFP), fundus autofluorescence (FAF) imaging, or fluorescein angiography (FA) imaging [4, 5]. Compared with conventional imaging modalities, OCT also has advantages in visualising subtle changes of the retinal layers affected by atrophic lesions.
Macular atrophy (MA) refers to atrophy irrespective of cause, either in the context of neovascular AMD (nAMD) or other diseases [7]. Regardless of successful anti-vascular endothelial growth factor (VEGF) treatment, MA develops in most eyes with nAMD and is strongly associated with visual decline [8–10]. Recent studies suggest that the causes of MA development in nAMD can be classified into macular neovascularisation (MNV)-related or MNV-unrelated causes, i.e., due to the progression of nAMD or due to underlying non-exudative AMD [11, 12]. One major challenge in quantifying the outer retinal layer loss on OCT B scans in nAMD is the presence of exudative pathologies associated with MNV, such as subretinal hyperreflective material (SHRM) and fibrovascular pigment epithelial detachment (PED). As a result, automated segmentation of outer retinal layers in eyes with nAMD is at a much earlier stage [13] compared with recent achievements in the automated segmentation algorithm development in the field of non-exudative AMD [14–17].
HyperTD has evolved as an en face OCT imaging biomarker for atrophic lesion measurement in non-exudative AMD [2, 4]. However, our understanding of this biomarker in the context of nAMD remains limited. In this explorative imaging analysis, we introduced the term ‘hypertransmission alteration (HTA)’ for changes observed on en face OCT images of nAMD eyes to differentiate from the hyperTD described in non-exudative AMD. The aim of the study was to test the feasibility of analysing HTA in eyes with nAMD from a cross-sectional dataset; to characterise the observed HTAs; and to analyse the association between HTA phenotypes, quantitative metrics (including homogeneity, reflectivity, border delineation, total area, and foveal involvement), and best-corrected visual acuity (BCVA).
MATERIALS AND METHODS
Study design and patients
A
This was a retrospective analysis of spectral-domain OCT (SD-OCT) images from a prospectively designed, cross-sectional, biomarker study performed at four study sites. The study design was reported in detail previously [18]. A
A
The study protocol was conducted in accordance with the tenets of the Declaration of Helsinki and was approved by the ethics committees of the respective sites. A
Signed informed consent was obtained from each patient before participation.
SD-OCT image process
SD-OCT scans (97-line B scans covering a 20 × 20 degrees area of the macula centered on the fovea) were captured in high-resolution mode using the Spectralis® SD-OCT (Heidelberg Engineering, Heidelberg, Germany). Figure 1 summarises the main steps of imaging analysis. A pre-trained deep-learning model was applied to automatically segment the pathologies including intraretinal fluid (IRF), subretinal fluid (SRF), PED, SHRM, and Bruch’s membrane (BM) [19, 20]. En face OCT projections were generated using a slab located 64–400 µm below the BM, as described previously [2].
Manual grading of HTAs and phenotypes
The foveal centre was manually annotated for each OCT scan by an experienced ophthalmologist (S.Y.). The same ophthalmologist annotated HTAs, which were defined as regions of increased reflectivity compared with the surrounding area on the en face OCT images. Firstly, each HTA patch was annotated on the en face images. Secondly, the HTA was manually classified into 3 phenotypes by considering the major HTA characteristics, including homogeneity, reflectivity, and border delineation (Fig. 2).
Quantitative metrics of HTAs
HTA homogeneity, reflectivity, and border delineation were quantified by analysing the intensity of the pixels within the annotated area of each HTA patch to quantify the differences between HTA phenotypes. The number of HTA patches in each eye was counted, and the total HTA area was quantified. We evaluated foveal proximity and foveal involvement scores to quantify the level of foveal involvement of the HTA lesions. Foveal proximity was defined as the shortest distance from the foveal centre to the nearest pixel of the HTA, based on a previous report [21].
Additional details on imaging, grading and quantitative metrics are provided in the Supplementary Materials.
Statistical analysis
Continuous variables were summarised using mean and SD, or median and interquartile range (IQR) as appropriate, whereas categorical variables were summarised using counts and percentages. To address data skewness, we applied log2 and square root (sqrt) transformations where appropriate. For the log2 transformation, an offset equal to half of the smallest non-zero value was added to the data to accommodate zero values.
Univariate associations with BCVA were evaluated using linear regression models. Statistical significance was defined as a 2-sided Benjamini–Hochberg-adjusted p value of < 0.05. Parameters significantly associated with BCVA in the univariate models were included in a multivariate linear regression model. Collinearity among parameters was assessed using Spearman correlation coefficients and variance inflation factors (VIF). To reduce collinearity, priority was given to factors that simplified model interpretability. An exhaustive search for possible parameter combinations was conducted, and the model with the highest adjusted R² value was selected.
Additional statistical comparisons of feature and clinical characteristics were evaluated with 2-sided t-tests, reporting Holm-adjusted p values. All analyses were performed using R statistical software (R Foundation for Statistical Computing, Vienna, Austria).
RESULTS
Study population
A
A total of 191 eyes of 191 patients with nAMD treated with intravitreal anti-vascular endothelial growth factor were recruited in this study. Three eyes were excluded due to missing SD-OCT images or clinical information. An additional 2 eyes were excluded from the analysis because it was not possible to identify the foveal centre due to severe epiretinal membrane (ERM) (stage 3 or 4 according to ERM OCT staging scheme) [22].The SD-OCT images from the remaining 186 eyes were included in the HTA evaluation. The mean age was 79 ± 6 years (range: 63–95), and 54% of patients were female. Disease duration information for nAMD was available for 162 eyes, and the mean duration was 44 ± 39 months (range: 1–172).
HTA phenotypes and quantitative metrics
On the en face OCT images, HTAs were graded in 73% (136/186) of eyes, with the remaining 27% (50/186) of eyes ungradable due to the presence of a large PED or SHRM (Fig. S1). Among the gradable eyes, HTAs were present in 88% (119/136), of which 25% (30/119) were classified as heterogeneous, 19% (23/119) as homogeneous, and 56% (66/119) as indeterminate phenotypes. Table 1 shows the summary of the descriptive statistics of the main features.
Among the 119 eyes with HTAs, 97 eyes (82%) had single-patch HTAs, and the other 22 (18%) eyes manifested multi-patch HTAs (number of patches ranged from 2 to 12). Among the multi-patch HTAs, homogeneous phenotype was the most observed (64%), followed by heterogeneous (23%), and indeterminate (14%) (Fig. S2).
In total, 181 HTA patches (36 heterogeneous, 76 homogeneous, and 69 indeterminate) were observed in the 119 eyes where HTAs were present. Of these HTA patches, 168 (29 heterogeneous, 74 homogeneous, and 65 indeterminate) were included in the quantitative analysis of homogeneity, reflectivity, and border delineation (all unitless); other patches were excluded as they were not completely captured on the OCT scan. Significant differences between the phenotypes were observed, emphasising the objective differences between the HTA phenotypes (Fig. S3). Homogeneous patches had the highest homogeneity score (0.886 ± 0.199), reflectivity score (144.6 ± 12.7), and border delineation score (0.012 ± 0.041) when compared with heterogeneous (0.649 ± 0.235 [p < 0.001], 134.4 ± 16.1 [p = 0.004], and − 0.014 ± 0.046 [p = 0.009], respectively) and indeterminate patches (0.699 ± 0.178 [p < 0.001], 126.1 ± 7.4 [p < 0.001], and − 0.018 ± 0.017 [p < 0.001], respectively).
Homogeneous HTAs were more likely in older patients (82.5 ± 5.3 years) compared with patients with indeterminate HTAs (77.9 ± 6.5 years, p = 0.001) and patients without HTAs (75.6 ± 5.9 years, p < 0.001). (Fig. S4). Heterogeneous HTAs were associated with longer disease duration (67 ± 48 months) versus indeterminate HTAs, which were associated with shorter disease duration (33 ± 33 months, p = 0.003). Homogeneous HTAs (54 ± 37 months) and eyes without HTAs (40 ± 40 months) showed no significant difference in disease duration (Fig. S4).
Heterogeneous HTAs had the largest mean sqrt total area (3.4 ± 1.3 mm), followed by homogeneous (2.7 ± 1.2 mm) and indeterminate HTAs (2.5 ± 0.9 mm). HTAs had a generally centric distribution in the modified ETDRS grid: HTAs were distributed most commonly in the central subfield, followed by the inner and outer rings, with no clear distribution pattern in the quadrants (Fig. S5).
The median (IQR) foveal proximity and the foveal involvement scores (0.5 mm) for all HTAs were − 0.54 mm (− 1.03, − 0.02) and 100% (61–100%), respectively, suggesting that the majority of HTAs (76%) involved the foveal centre point and half of them affected the entire central foveal subfield. As the involvement scores were measured in larger circles (1.0–2.0 mm), the median scores gradually decreased. Heterogeneous HTAs were associated with lower foveal proximity and higher foveal involvement scores, whereas homogeneous HTAs exhibited the opposite trend (Table 1).
Associations between BCVA and HTAs
A univariate analysis was performed to identify the variables that were significantly associated with BCVA (Table 2). Increasing age (p = 0.019) and female sex (p = 0.003) were significantly associated with lower BCVA. Heterogeneous HTAs were significantly associated with the lowest BCVA (p < 0.001). Compared with homogeneous, indeterminate, and an absence of HTAs, heterogeneous HTAs were associated with a reduction of 17.1, 22.1, and 32.0 letters in BCVA, respectively (p < 0.001). Larger sqrt-transformed total lesion area (p < 0.001), greater foveal involvement (particularly within the 2.0 mm zone [p < 0.001]) and closer proximity to the fovea (p < 0.001) were significantly associated with worse BCVA. Additionally, higher log₂-transformed IRF (p < 0.001) and SHRM volumes (p = 0.016) were negatively associated with BCVA, while greater SRF volume (p = 0.002) was positively associated with better BCVA.
Spearman correlations and VIF were analysed for significant parameters from the univariate analysis. High correlations were found between sqrt total area and foveal involvement scores at 1.5 and 2.0 mm (r = 0.869 and 0.924), and foveal proximity and foveal involvement scores at 0.5 to 1.5 mm (r = 0.920, 0.955, and 0.898). VIF confirmed high collinearity between the sqrt total area and the 4 foveal involvement scores. Consequently, foveal involvement scores were excluded from the multivariate analysis, while the sqrt total area was kept, due to its simpler interpretability (Fig. S6).
The remaining parameters were evaluated for inclusion in a multivariate association model with BCVA. An automated exhaustive search method was employed to identify the model with the highest adjusted R² (0.46), which included age, sex, HTA phenotype, sqrt total area, and log2-transformed IRF volume (Fig. 3).
In this model, female sex was associated with a decrease of 9.0 letters in BCVA (95% confidence interval [CI] − 14.6, − 3.3, p = 0.002). The HTA phenotype was significantly associated with BCVA (p < 0.001). Even after adjusting for other variables, the heterogeneous HTA phenotype remained significantly associated with the lowest BCVA. Specifically, heterogeneous HTAs were associated with a decrease of 19.0 letters (95% CI 9.8, 28.2) compared with homogeneous HTAs, and a reduction of 16.8 letters (95% CI 9.3, 24.3) compared with indeterminate HTAs. Additionally, the sqrt total area was inversely related to BCVA, with each mm increase corresponding to a decrease of 5.1 letters (95% CI − 7.9, − 2.3, p < 0.001). Higher IRF volumes were also associated with lower BCVA, with each log2-transformed µm3 increase resulting in a decrease of 1.3 letters (95% CI − 1.9, − 0.8, p < 0.001).
DISCUSSION
In this study, we present the prevalence and characteristics of HTAs in eyes with nAMD. These HTAs, which differ from the hypertransmission defects associated with GA, were commonly large, mostly fovea-involving and manifested as 3 distinct phenotypes that had significant associations with BCVA and exhibited pronounced differences in quantified homogeneity, reflectivity, and border delineation.. Heterogeneous phenotype and large sqrt total area were found to have the most significant association with BCVA. Our study addresses a critical gap in knowledge by clarifying the characteristics of HTAs in eyes with nAMD.
A
Although anti-VEGF treatments effectively control exudation caused by neovascularisation, the development of MA in nAMD is a key factor contributing to a significant decline in vision despite prolonged therapy [8–10]. The MA observed in nAMD may be multifactorial in origin. The AREDS2 study (NCT00345176) found atrophy in 40% of the eyes at the incidence of MNV and, in half of these eyes, MA was attributable to pre-existing GA [23, 24]. Recent studies suggest that the causes of MA development in nAMD can be classified as MNV-related (mechanical damage to the RPE, collapse of serous PED, and fibrosis) or MNV-unrelated (drusen-related) [11, 12]. It remains unclear whether MA has a similar appearance to GA in AMD [4]. In eyes with non-exudative AMD, no distinct hyperTD phenotypes have been reported [2, 4]. The different phenotypes of HTAs in nAMD described in this study correspond to MA of different appearances and, therefore, may suggest different origins for the MA observed in nAMD.The HTAs detected in eyes with nAMD in this study were categorised as homogeneous, heterogeneous, and indeterminate phenotypes. Homogeneous HTAs, observed in 19% of the gradable eyes, showed the highest similarity to the hyperTDs observed in GA secondary to AMD; these HTAs were observed in older patients and appeared as lesions of mostly homogeneous hyperreflectivity with well-defined boundaries [30]. On OCT B scans, homogeneous HTAs were associated with continuous choroidal hypertransmission due to complete atrophy of ONL, photoreceptor layers, and RPE layers. Compared with heterogeneous or indeterminate phenotypes, homogeneous HTAs less frequently involved the fovea and appeared more frequently as multi-patch formations, indicating an underlying GA aetiology.
A
Given that neovascularisation can lead to rapid, sustained, and irreversible, severe vision loss, it has been the primary target of most therapeutic interventions. When neovascularisation occurs in the context of non-exudative AMD, the underlying non-exudative AMD often continues to progress unmonitored [4]. Heterogeneous HTAs were found in 25% of the gradable images and were associated with the lowest BCVA after adjusting for other vision-relevant variables. In a previous study, we evaluated the severity of the outer retinal band disruption at the foveal centre in eyes showing fovea-involving HTAs (the majority were heterogeneous or indeterminate phenotypes) [26]. The eyes showing heterogeneous HTA at the foveal centre were associated with more severe outer retinal band disruption and worse BCVA than non-foveal-involving HTAs, suggesting more advanced stages of MA [26]. The heterogeneous phenotype appeared as lesions associated with non-continuous choroidal hypertransmission on OCT B scans due to patches of complete atrophy of the ONL, photoreceptor layers, and RPE layers, which are often located above SHRM or fibrovascular PED. Such co-localisation suggests that heterogeneous HTAs correspond to MA development due to MNV-related causes, such as mechanical damage to the RPE from the intrusion of neovessels and the blockage of oxygen and nutrients from the choroid to the outer retinal layers due to fibrotic tissue [11].Our findings on both homogeneous and heterogeneous HTAs provide valuable insights into the complex pathology of MA in nAMD, which might need to be addressed through different pathways. Homogeneous HTAs may aid in identifying underlying non-exudative AMD, whereas heterogeneous HTAs could serve as indicators of MA development associated with MNV.
Indeterminate HTAs were the most commonly observed phenotype in around half of the study eyes and were associated with higher BCVA compared with the other 2 phenotypes. They appeared more often in younger patients with shorter disease duration. On OCT B scans, indeterminate HTAs were associated with noncontinuous choroidal hypertransmission due to an attenuation of the RPE and disruption of photoreceptor layers. Therefore, we hypothesise that indeterminate HTAs correspond to lesions in an earlier stage of MA development. In non-exudative AMD, choroidal hypertransmission is exclusively associated with RPE disruption [27]. In contrast, the indeterminate phenotype in nAMD was observed in the absence of complete RPE disruption but with attenuated or disrupted ONL and/or photoreceptor layers. A longitudinal analysis to correlate changes in outer retinal layers observed on OCT B scans with HTA appearance on en face imaging will enhance our ability to recognise indeterminate HTAs, potentially allowing for an earlier diagnosis of MA in nAMD. Early identification may also help tailor treatment strategies to specific disease stages, thereby improving patient outcomes.
We demonstrated distinct differences between phenotypes in objectively quantified pixel intensity metrics, including homogeneity, reflectivity, and border delineation. Our findings suggest that HTA phenotypes in nAMD can be objectively classified and may be amenable to automated classification by artificial intelligence. The development of such a tool would facilitate more precise and unbiased diagnoses on a larger scale, further enhancing our understanding of disease progression in nAMD.
In addition to HTA phenotype, sqrt total area, foveal proximity, and foveal involvement scores were also significantly associated with BCVA, in line with what is considered relevant in non-exudative AMD progression [4, 25]. Most of the HTAs in this study involved the foveal centre point, with half affecting the entire central foveal subfield. Additionally, heterogeneous HTAs exhibited the highest degree of foveal involvement, whereas homogeneous HTAs demonstrated the opposite trend. Therefore, in the multivariate model, where HTA phenotypes were considered, foveal proximity was not selected by the exhaustive search approach, as it did not add additional value in association with BCVA in the multivariate model.
Hypertransmission has been recognised as a key OCT anatomic biomarker of atrophy [27]. Although it was acknowledged that hypertransmission may not always penetrate to the underlying choroid in eyes with tall PEDs [27], its prevalence in nAMD remains unclear. In this cross-sectional study, we identified that large PED or SHRM interfered with the reliable assessment of HTAs in a quarter of eyes, highlighting a limitation of our HTA analysis approach in advanced nAMD. Further research is needed to explore whether newer OCT technology with improved penetration might more reliably evaluate hypertransmission in these cases, such as swept-source OCT or SD-OCT with enhanced depth imaging.
A significant limitation of this study is the cross-sectional design, which precludes longitudinal results to support our hypothesis that different HTA phenotypes in nAMD may be associated with different stages of MA. This hypothesis was informed by observations of atrophic severity in the outer retinal layer on corresponding B scans and differences in disease duration and age among the phenotypes. However, data on disease duration information were not available for all eyes due to the retrospective nature of the study. The cross-sectional study design also prevented us from reliably assessing the MNV subtypes that have been shown to be associated with MA development [11, 12]. Conducting an HTA analysis within a longitudinal dataset with extended follow-up in treatment-naïve eyes will be the next step in our research.
In summary, the present study demonstrated that HTAs were frequently observed in eyes with nAMD. We identified three distinct HTA phenotypes, each exhibiting unique differences in appearance, visual associations, and demographics. These results suggest that HTAs in eyes with nAMD may serve as valuable biomarkers for MA of varying causes and stages. Consequently, they could enhance the accuracy of diagnoses and prognosis and inform more comprehensive, individualised treatment strategies to address the multifaceted nature of this pathology.
Supplementary information is available on Eye’s website.
SUMMARY
What was known before
Hypertransmission defects are biomarkers for atrophic lesions in non-exudative AMD
Hypertransmission into the choroid can be observed in OCT B scans of eyes with non-exudative or neovascular AMD (nAMD), and is associated with outer retinal layer loss
What this study adds
A
A
A
Introduction of the term hypertransmission alteration (HTA) for changes in hypertransmission observed on en face OCT images from eyes with nAMD
Characterisation of three HTA phenotypes significantly associated with visual function, morphology, demographics and spatial distribution in the macula
Identification of associations between HTA phenotype and clinical features:
Indeterminate HTAs were more common in younger patients with shorter disease duration
Heterogeneous HTAs may correspond to atrophic lesions associated with MNV
Homogeneous HTAs may correspond to underlying non-exudative AMD
A
Data Availability
For eligible studies, qualified researchers may request access to individual patient-level clinical data through a data request platform. At the time of writing, this request platform is Vivli: https://vivli.org/ourmember/roche/. For up-to-date details on Roche’s Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents, go to https://go.roche.com/data_sharing. Anonymized records for individual patients across more than one data source external to Roche cannot, and should not, be linked due to a potential increase in risk of patient re-identification.
Fig. S1
Representative example of ungradable en face OCT image for HTA, and comparisons of BCVA and exudative pathologies between gradable and ungradable en face OCT images.
A
Fig. 1
Diagram illustrating the OCT volume segmentation, en face image generation, and HTA analysis.
A OCT segmentation of BM on the volume scan and the extraction of the sub-BM slab for en face image generation. B The process of en face image generation before the HTA grading and analysis. Histogram matching and median normalisation were applied to remove dark line artefacts and to improve the uniformity of measures for different features, respectively. C The method of HTA analysis. Measurements using the HTA mask include the number of patches, area, reflectivity, homogeneity, and border delineation. Measurements using HTA (outlined in green) and foveal centre annotation (red point) include foveal proximity (red line) and foveal involvement scores of 4 foveal-centered circles with a radius of 0.5 mm, 1.0 mm, 1.5 mm, and 2.0 mm, respectively (orange circles).
BM Bruch’s membrane; HTA hypertransmission alteration; OCT optical coherence tomography.
A
Fig. 2
Representative example images of different HTA phenotypes.
Top row: A heterogeneous HTA appearing on the en face OCT image (histogram matched and median normalised) as a heterogeneous patch with mostly ill-defined boundaries. The NIR image shows a lesion of a similar shape to that observed on the en face OCT image. On the cross-sectional OCT B scan, a barcode-shaped, non-continuous choroidal hypertransmission located above SHRM through the area of HTA (location indicated by the green line on the NIR image) is seen due to a complete atrophy of the ONL, photoreceptor layers, and RPE layers. Middle row: A homogeneous HTA appearing on the en face OCT image as a predominantly homogeneously hyperreflective patch with mostly well-defined boundaries. The NIR image shows a lesion of a similar shape to that observed on the en face OCT image. On the OCT B scan, continuous choroidal hypertransmission (the white bar) is seen due to complete atrophy of the ONL, photoreceptor layers, and RPE layers. Bottom row: An indeterminate HTA appearing on the en face OCT image as a lesion of lower hyperreflectivity with ill-defined boundaries, which shows a different appearance from the abnormality on the NIR image. On the OCT B scan, an attenuated RPE and partially disrupted photoreceptor layers above SHRM are observed.
HTA hypertransmission alteration; NIR near infrared reflectance; OCT optical coherence tomography; ONL outer nuclear layer; RPE retinal pigment epithelial; SHRM subretinal hyperreflective material.
A
Fig. 2
Representative example images of different HTA phenotypes.
Top row: A heterogeneous HTA appearing on the en face OCT image (histogram matched and median normalised) as a heterogeneous patch with mostly ill-defined boundaries. The NIR image shows a lesion of a similar shape to that observed on the en face OCT image. On the cross-sectional OCT B scan, a barcode-shaped, non-continuous choroidal hypertransmission located above SHRM through the area of HTA (location indicated by the green line on the NIR image) is seen due to a complete atrophy of the ONL, photoreceptor layers, and RPE layers. Middle row: A homogeneous HTA appearing on the en face OCT image as a predominantly homogeneously hyperreflective patch with mostly well-defined boundaries. The NIR image shows a lesion of a similar shape to that observed on the en face OCT image. On the OCT B scan, continuous choroidal hypertransmission (the white bar) is seen due to complete atrophy of the ONL, photoreceptor layers, and RPE layers. Bottom row: An indeterminate HTA appearing on the en face OCT image as a lesion of lower hyperreflectivity with ill-defined boundaries, which shows a different appearance from the abnormality on the NIR image. On the OCT B scan, an attenuated RPE and partially disrupted photoreceptor layers above SHRM are observed.
HTA hypertransmission alteration; NIR near infrared reflectance; OCT optical coherence tomography; ONL outer nuclear layer; RPE retinal pigment epithelial; SHRM subretinal hyperreflective material.
A
Fig. 3
Multivariate analysis results of BCVA (ETDRS letters) associations with evaluated features.
A Multivariate model results in a forest plot. B BCVA comparison between eyes with different HTA phenotypes (individual data points overlaid, significant comparisons are marked with their BH-adjusted p value indicated). The difference between the absence of HTA and the other groups has been explained by the coefficient of sqrt HTA total area, and therefore, no significance is found here.
BCVA best-corrected visual acuity; BH Benjamini–Hochberg; CI confidence interval; ETDRS Early Treatment Diabetic Retinopathy Study; HTA hypertransmission alteration; IRF intraretinal fluid; sqrt square root.
aVolume of IRF was analysed in the foveal central subfield, and log2 transformation was applied to reduce the skewness of the data.
A
Fig. 3
Multivariate analysis results of BCVA (ETDRS letters) associations with evaluated features.
A Multivariate model results in a forest plot. B BCVA comparison between eyes with different HTA phenotypes (individual data points overlaid, significant comparisons are marked with their BH-adjusted p value indicated). The difference between the absence of HTA and the other groups has been explained by the coefficient of sqrt HTA total area, and therefore, no significance is found here.
BCVA best-corrected visual acuity; BH Benjamini–Hochberg; CI confidence interval; ETDRS Early Treatment Diabetic Retinopathy Study; HTA hypertransmission alteration; IRF intraretinal fluid; sqrt square root.
aVolume of IRF was analysed in the foveal central subfield, and log2 transformation was applied to reduce the skewness of the data.
|
Evaluated features
|
HTA absent
(n = 17)
|
All HTAs
(n = 119)
|
HTA present (n = 119)
|
||
|---|---|---|---|---|---|
|
Heterogeneous
(n = 30)
|
Homogeneous
(n = 23)
|
Indeterminate
(n = 66)
|
|||
|
Mean (SD) BCVA, letters
|
79 (6)
|
62 (22)
|
47 (18)
|
64 (14)
|
69 (17)
|
|
Mean (SD) age, years
|
76 (6)
|
79 (7)
|
79 (8)
|
82 (5)
|
78 (6)
|
|
Female sex, n (%)
|
10 (59)
|
64 (54)
|
15 (50)
|
14 (61)
|
35 (53)
|
|
Mean (SD) disease duration, months
|
40 (40)
|
45 (40)
|
67 (48)
|
54 (37)
|
33 (33)
|
|
ERM presence, n (%)
|
4 (24)
|
10 (8)
|
5 (17)
|
0 (0)
|
5 (8)
|
|
Number of patches, median (IQR)
|
NA
|
1 (1–1)
|
1 (1–1)
|
2 (1–4)
|
1 (1–1)
|
|
Mean (SD) sqrt of total HTA
area, mm
|
NA
|
2.73 (1.12)
|
3.39 (1.30)
|
2.69 (1.17)
|
2.45 (0.90)
|
|
Median (IQR) foveal proximity, mm
|
NA
|
−0.54
(− 1.03, − 0.02)
|
−0.66
(− 1.44, − 0.32)
|
−0.23
(− 0.90, − 0.09)
|
−0.49
(− 0.91, − 0.04)
|
|
Median (IQR) foveal involvement score at 0.5 mm, %
|
NA
|
100
(61–100)
|
100
(88–100)
|
91
(31–100)
|
100
(52–100)
|
|
Median (IQR) foveal involvement score at 1.0 mm, %
|
NA
|
83
(49–100)
|
98
(70–100)
|
78
(41–96)
|
81
(47–98)
|
|
Median (IQR) foveal involvement score at 1.5 mm, %
|
NA
|
58
(38–91)
|
84
(49–100)
|
54
(39–87)
|
58
(35–78)
|
|
Median (IQR) foveal involvement score at 2.0 mm, %
|
NA
|
41
(27–70)
|
64
(37–97)
|
35
(24–72)
|
41
(26–62)
|
|
Median (IQR) IRF volume, µm³a
|
12.0
(12.0–12.0)
|
12.0
(12.0–20.4)
|
12.0
(12.0–19.9)
|
19.5
(13.2–22.6)
|
12.0
(12.0–20.0)
|
|
Median (IQR) SRF volume, µm³a
|
18.7
(16.0–23.8)
|
16.0
(16.0–23.6)
|
16.0
(16.0–21.2)
|
16.0
(16.0–16.0)
|
20.9
(16.0–25.1)
|
|
Median (IQR) PED volume, µm³a
|
24.2
(23.0–25.0)
|
25.2
(24.1–25.8)
|
25.4
(23.9–25.7)
|
24.4
(23.1–25.2)
|
25.4
(24.5–26.1)
|
|
Median (IQR) SHRM volume, µm³a
|
17.6
(15.4–21.7)
|
19.5
(13.4–22.2)
|
21.2
(18.5–22.9)
|
15.0
(13.4–21.2)
|
19.2
(13.4–22.3)
|
| BCVA best-corrected visual acuity; ERM epiretinal membrane; HTA hypertransmission alteration; IQR interquartile range; IRF intraretinal fluid; NA not applicable; PED pigment epithelial detachment; SD standard deviation; SHRM subretinal hyperreflective material; sqrt square root; SRF subretinal fluid. | |||||
| aThe volumes of IRF, SRF, PED, and SHRM were analysed in the central foveal subfield. | |||||
|
Parameter
|
Category
|
Coefficient
(95% CI)
|
R2
|
p value
|
Adjusted p value
|
|---|---|---|---|---|---|
|
Age, years
|
−0.64 (− 1.17, − 0.11)
|
0.041
|
0.019
|
0.025
|
|
|
Sex
|
Male
|
Reference
|
|||
|
Female
|
−10.68
(− 17.78, − 3.58)
|
0.063
|
0.003
|
0.005
|
|
|
Disease duration, days
|
−0.04 (− 0.13, 0.06)
|
0.005
|
0.455
|
0.506
|
|
|
Presence of ERM
|
−4.25 (− 16.16, 7.65)
|
0.004
|
0.481
|
0.513
|
|
|
HTA phenotype
|
Heterogeneous
|
Reference
|
0.230
|
< 0.001
|
< 0.001
|
|
Homogeneous
|
17.14 (6.79, 27.49)
|
||||
|
Indeterminate
|
22.07 (13.82, 30.31)
|
||||
|
Absent
|
32.00 (20.19, 43.81)
|
||||
|
Number of HTA patches
|
−1.17
(− 3.81, 1.47)
|
0.007
|
0.383
|
0.471
|
|
|
Sqrt of HTA size, mm
|
−7.76
(− 11.02, − 4.49)
|
0.161
|
< 0.001
|
< 0.001
|
|
|
Foveal proximity, mm
|
−9.32
(− 14.13, − 4.50)
|
0.112
|
< 0.001
|
< 0.001
|
|
|
Foveal involvement score at 0.5 mm, %
|
−0.17
(− 0.28, − 0.06)
|
0.077
|
0.002
|
0.004
|
|
|
Foveal involvement score at 1.0 mm, %
|
−0.20
(− 0.32, − 0.07)
|
0.078
|
0.002
|
0.004
|
|
|
Foveal involvement score at 1.5 mm, %
|
−0.23
(− 0.36, − 0.10)
|
0.094
|
< 0.001
|
0.002
|
|
|
Foveal involvement score at 2.0 mm, %
|
−0.28
(− 0.41, − 0.14)
|
0.127
|
< 0.001
|
< 0.001
|
|
|
IRF volume, µm³a
|
−1.45
(− 2.13, − 0.77)
|
0.119
|
< 0.001
|
< 0.001
|
|
|
SRF volume, µm³a
|
1.35
(0.50, 2.21)
|
0.069
|
0.002
|
0.004
|
|
|
PED volume, µm³a
|
−0.25
(− 2.30, 1.81)
|
0.000
|
0.812
|
0.812
|
|
|
SHRM volume, µm³a
|
−1.06
(− 1.92, − 0.20)
|
0.043
|
0.016
|
0.023
|
|
| BCVA best-corrected visual acuity; CI confidence interval; ERM epiretinal membrane; ETDRS Early Treatment Diabetic Retinopathy Study; HTA hypertransmission alteration; IRF intraretinal fluid; PED pigment epithelial detachment; SHRM subretinal hyperreflective material; sqrt square root; SRF subretinal fluid. | |||||
| aThe volumes of IRF, SRF, PED, and SHRM were analysed in the foveal central subfield, and log2 transformation was applied to reduce the skewness of the data. | |||||
Electronic Supplementary Material
Below is the link to the electronic supplementary material
References
1.
Holz FG, Sadda SR, Staurenghi G, Lindner M, Bird AC, Blodi BB et al. Imaging protocols in clinical studies in advanced age-related macular degeneration: recommendations from Classification of Atrophy Consensus Meetings. Ophthalmology. 124, 464–478 (2017).
2.
Shi Y, Yang J, Feuer W, Gregori G & Rosenfeld PJ. Persistent hypertransmission defects on en face OCT imaging as a stand-alone precursor for the future formation of geographic atrophy. Ophthalmol Retina. 5, 1214–1225 (2021).
3.
Liu J, Shen M, Laiginhas R, Herrera G, Li J, Shi Y, Hiya F et al. Onset and progression of persistent choroidal hypertransmission defects in intermediate age-related macular degeneration: a novel clinical trial endpoint. Am J Ophthalmol. 254, 11–22 (2023).
4.
Schaal KB, Rosenfeld PJ, Gregori G, Yehoshua Z & Feuer WJ. Anatomic clinical trial endpoints for nonexudative age-related macular degeneration. Ophthalmology. 123, 1060–1079 (2016).
5.
Lujan BJ, Rosenfeld PJ, Gregori G, Wang F, Knighton RW, Feuer WJ et al. Spectral domain optical coherence tomographic imaging of geographic atrophy. Ophthalmic Surg Lasers Imaging. 40, 96–101 (2009).
6.
Wu Z, Luu CD, Ayton LN, Goh JIK, Lucci LM, Hubbard WC et al. Optical coherence tomography-defined changes preceding the development of drusen-associated atrophy in age-related macular degeneration. Ophthalmology. 121, 2415–2422 (2014).
7.
Dolz-Marco R, Balaratnasingam C, Messinger J, Li M, Ferrera D, Freund KB et al. The border of macular atrophy in age-related macular degeneration: a clinicopathologic correlation. Am J Ophthalmol. 193, 166–177 (2018).
8.
Bhisitkul RB, Mendes TS, Rofagha S, Enanoria W, Boyer DS, Sada SR et al. Macular atrophy progression and 7-year vision outcomes in subjects from the ANCHOR, MARINA, and HORIZON studies: the SEVEN-UP study. Am J Ophthalmol. 159, 915–924 (2015).
9.
Gune S, Abdelfattah NS, Karamat A, Balasubramanian S, Marion KM, Morgenthien E et al. Spectral-domain OCT-based prevalence and progression of macular atrophy in the HARBOR study for neovascular age-related macular degeneration. Ophthalmology. 127, 523–532 (2020).
10.
Chandra S, Arpa C, Menon D, Khalid H, Hamilton R, Nicholson L et al. Ten-year outcomes of antivascular endothelial growth factor therapy in neovascular age-related macular degeneration. Eye (Lond). 34, 1888–1896 (2020).
11.
Borrelli E, Barresi C, Ricardi F, Berni A, Grosso D, Viggiano P et al. Distinct pathways of macular atrophy in type 3 macular neovascularization associated with AMD. Invest Ophthalmol Vis Sci. 65, 18 (2024).
12.
Finn AP, Pistilli M, Tai V, Daniel E, Ying G-S, Maguire MG et al. Localized optical coherence tomography precursors of macular atrophy and fibrotic scar in the comparison of age-related macular degeneration treatments trials. Am J Ophthalmol. 223, 338–347 (2021).
13.
Ehlers JP, Lunasco LM, Yordi S, Cetin H, Le TK, Sarici K et al. Compartmental exudative dynamics in neovascular age-related macular degeneration: volumetric outcomes and impact of volatility in a Phase III clinical trial. Ophthalmol Retina. 8, 765–777 (2024).
14.
Pramil V, de Sisternes L, Omlor L, Lewis W, Sheikh H, Chu Z et al. A deep learning model for automated segmentation of geographic atrophy imaged using swept-source OCT. Ophthalmol Retina. 7, 127–141 (2023).
15.
Chiang JN, Corradetti G, Nittala MG, Corvi F, Rakocz N, Rudas A et al. Automated identification of incomplete and complete retinal epithelial pigment and outer retinal atrophy using machine learning. Ophthalmol Retina. 7, 118–1126 (2023).
16.
Mai J, Lachinov D, Riedl S, Reiter GS, Vogl W-D, Bogunovic H et al. Clinical validation for automated geographic atrophy monitoring on OCT under complement inhibitory treatment. Sci Rep. 2023;13(1):7028.
17.
Derradji Y, Mosinska A, Apostolopoulos S, Ciller C, De Zanet S & Mantel I. Fully-automated atrophy segmentation in dry age-related macular degeneration in optical coherence tomography. Sci Rep. 11, 21893 (2021).
18.
Titz B, Siebourg-Polster J, Bartolo F, Lavergne V, Jiang Z, Gayan J et al. Implications of ocular confounding factors for aqueous humor proteomic and metabolomic analyses in retinal diseases. Transl Vis Sci Technol. 13, 17 (2024).
19.
Lu HX & Maunz A. Optical coherence tomography segmentation of retinal fluids using deep learning. Invest Ophthalmol Vis Sci. 64, 1124 (2023).
20.
Li YY, Bachmeier I, Yu S, Albrecht T, Lloyd Jones I, Maunz A, et al. Retinal layer (RL) segmentation on spectral-domain optical coherence tomography (SD-OCT) in eyes with neovascular age-related macular degeneration (nAMD) and diabetic macular edema (DME) using Deep Learning (DL). Abstracts of the ARVO Imaging in the Eye Conference May 3, 2025. Invest Ophthalmol Vis Sci. 2025;66:PB008.20.
21.
Keenan TDL, Agron E, Keane PA, Domalpally A, Chew EY, Age-Related Eye Disease Study Research Group et al. Oral antioxidant and lutein/zeaxanthin supplements slow geographic atrophy progression to the fovea in age-related macular degeneration. Ophthalmology. 132, S0161-6420(24)00425-1 (2024).
22.
Govetto A, Lalane RA, 3rd, Sarraf D, Figueroa MS & Hubschman JP. Insights into epiretinal membranes: presence of ectopic inner foveal layers and a new optical coherence tomography staging scheme. Am J Ophthalmol. 175, 99–113 (2017).
23.
Domalpally A, Danis RP, Trane R, Blodi BA, Clemons TE, Chew EY et al. Atrophy in neovascular age-related macular degeneration: age-related eye disease study 2 report number 15. Ophthalmol Retina. 2, 1021–1027 (2018).
24.
Zanzottera EC, Ach T, Huisingh C, Messinger JD, Spaide RF & Curcio CA. Visualizing retinal pigment epithelium phenotypes in the transition to geographic atrophy in age-related macular degeneration. Retina. 36, S12-S25 (2016).
25.
Schmitz-Valckenberg S, Sadda S, Staurenghi G, Chew EY, Fleckenstein M, Holz FG et al. Geographic atrophy: semantic considerations and literature review. Retina. 36, 2250–2564 (2016).
26.
Yu S, Titz B, Parian-Scherb M, Bachmeier I, Enders P, Altay L et al. Hypertransmission defect phenotypes and functional associations in eyes with neovascular age-related macular degeneration. Inv Ophthalmol Vis Sci. 65, PB0082 (2024).
27.
Sadda SR, Guymer R, Holz FG, Schmitz-Valckenberg S, Curcio CA, Bird AC et al. Consensus definition for atrophy associated with age-related macular degeneration on OCT: classification of Atrophy Report 3. Ophthalmology. 125, 537–548 (2018).
ACKNOWLEDGEMENTS
The authors gratefully acknowledge Prof. Christine A. Curcio for reviewing the results of this work and for providing constructive suggestions that significantly improved the manuscript. Medical writing support was provided by Natasha Hornby, PhD, and Anne Nunn, PhD, of Helix, OPEN Health Communications, and funded by F. Hoffmann-La Roche Ltd, in accordance with Good Publication Practice guidelines (www.ismpp.org/gpp- 2022).
A
AUTHOR CONTRIBUTIONS
SY and MPS participated in the conception and design of the work, the acquisition, analysis, and interpretation of data, and drafting and critically revising the manuscript. KP and IB participated in the analysis and interpretation of data and manuscript review and revision. PE, LA, MJK, SAS, and JGG participated in the acquisition and interpretation of data and manuscript review and revision. CC participated in the conception and design of the work, interpretation of data, and manuscript review and revision. BT and SF participated in the conception and design of the work, acquisition, analysis, and interpretation of data, and manuscript review and revision. All authors gave final approval of the version to be published and agree to be accountable for all aspects of the work.
A
FUNDING
This study was funded by F. Hoffmann-La Roche Ltd., Basel.
A
The funder participated in the design, conduction, data collection, analysis, and writing of this manuscript.COMPETING INTERESTS
SY, MP-S, IB, CC, BT and SF are employees of F. Hoffmann-La Roche Ltd. KP is an employee and stockholder of Roche Products Ltd. PE reports financial support from Roche. LA reports financial support from Roche and has received lecture honoraria from AbbVie, Apellis, Bayer, Novartis and Roche. MJK has received lecture honoraria from LumiThera, and Roche, travelling fees from ARC Laser and Munich Surgical Imaging, and is a stockholder of Ocutrx. JGG reports financial support, consulting fees and lecture honoraria from AbbVie, Bayer, Novartis and Roche. SAS has no potential conflict of interest or financial disclosure. All authors attest that they meet the current ICMJE criteria for authorship.
Supplementary information is available at EYE’s website: https//www.nature.com/eye:
Supplementary methods
additional details on imaging, grading and quantitative metrics.
Table S1. Percentage of complete captures of each ETDRS subfield on the included OCT scans.