Longitudinal Changes in Choroidal Thickness and Choroidal Vascularity Index in Age-Related Macular Degeneration

Age-related macular degeneration (AMD) is a multifactorial disease related to ageing in which inflammation, oxidative stress, and vascular dysfunction of the choroid are key drivers [1, 2]. Therefore, with continuously improving imaging modalities, the study of the choroidal vasculature has been recently met with increasing interest in AMD using enhanced depth imaging spectral domain-optical coherence tomography (OCT) and swept-source OCT [3]. Also, OCT angiography can provide quantitative metrics based on signal decorrelation due to the presence of blood flow, but it is still not widely available and it is more time consuming than OCT.

Several OCT-based choroidal biomarkers have been linked to AMD onset and progression [4, 5], with the two best characterized being choroidal thickness (CT) and Choroidal Vascularity Index (CVI). CT is the most commonly reported, mainly due to the simplicity of obtaining these data, using the built-in caliper available in commercial OCT devices [6]. More recently, CVI has been proposed as an alternative, more reliable, and reproducible parameter compared to CT [7, 8]. It is defined as the ratio of the vascular luminal area to the total choroidal area, obtained after image binarization; thus, it can also provide information about the composition of the choroid. Both CT and CVI appear to be altered in eyes with various retinal diseases compared to healthy controls [9‒11]. In AMD, higher CVI values have been reported after switching to brolucizumab, hypothetically due to a subclinical inflammation caused by the drug [12]. However, how these parameters change over time in eyes with AMD and whether these measurements are influenced by other ocular or non-ocular factors remains unclear. CT and/or CVI can be easily obtained based on routinely captured OCT data and may potentially aid in better characterizing subgroups of AMD patients at higher risk of disease progression. To better understand the association of CT and CVI with AMD progression, we performed a longitudinal observational study to determine the rate of choroidal thinning and changes in CVI in eyes with AMD, adjusted for AMD stage, age, and sex.

In this retrospective, longitudinal, single-center study we reviewed images available in our Heidelberg Eye Explorer (HEYEX^®) database of patients diagnosed with AMD between 2019 and 2023. This study was conducted in accordance with the tenets of the Declaration of Helsinki.

Inclusion criteria were at least 24 months of follow-up (FU), eye-tracked OCT B-scans (with the FU mode on), OCT images acquired with the enhanced depth imaging mode, with a visible scleral/choroidal interface and with an available fundus near-infrared (NIR) image. All OCTs were acquired using the Spectralis HRA-OCT Imaging Platform (Heidelberg Engineering, Heidelberg, Germany) with the high-speed protocol. We included images centered on the fovea with a scan pattern of 30 × 15° (8.6 × 4.3 mm) or 20° × 20° (5.9 × 5.9 mm), a 240–244 μm of separation between B-scans, and with an automatic real-time tracking between 8 and 21 frames. When both eyes of the same patient met study criteria, only one eye was included, chosen randomly. We excluded images with poor image quality (Q-value) <26 dB, with significant image artifacts, and those with evidence of retinal diseases other than AMD, such as diabetic retinopathy, arterial or venous occlusion, retinal dystrophies, vitreomacular interface disorders, or pathologic myopia.

Consecutive OCT B-scans and NIR images were evaluated simultaneously to confirm the presence of subretinal drusenoid deposits (SDDs) and hypertransmission defects corresponding to areas of incomplete or complete retinal pigment epithelium and outer retinal atrophy (cRORA) affecting the central macula. AMD staging was classified at baseline according to OCT biomarkers as follows: central cRORA, exudative AMD (evidence of active exudation with subretinal or intraretinal fluid, the presence of retinal pigment epithelium [RPE] detachment with signs of neovascularization such as heterogeneous reflectivity, choroidal cleft, signs of RPE tear, multilayered and/or hyperreflective scar tissue from previous exudation), and non-advanced AMD (the presence of drusen; SDDs; hyperreflective foci; and no evidence of active or previous exudation, central atrophy, or scar tissue). We also included a control group with no retinal pathology and >50 years of age. In the group with non-advanced AMD at baseline and controls, all available B-scans during the FU period were assessed for the progression to advanced stages (exudative or central cRORA).

From each OCT, the choroid was assessed on a horizontal B-scan centered on the fovea. CT was measured from the RPE-BM complex to the scleral/choroidal interface, using the built-in digital caliper of the HEYEX^® Platform, at three different locations: subfoveal, 2,000 μm nasal, and 2,000 μm temporal to the foveal center (Fig. 1a) [13].

For CVI calculation, the same horizontal B-scans in the 1 × 1 pixel scale for a better resolution [14] were exported from the OCT database and imported into the ImageJ software (version 2.14.0/1.54f; National Institutes of Health, Bethesda, MD, USA). Following the method described by Agrawal et al. [10], the images were first binarized using the Nilblack auto threshold. After setting the scale, a 2,000-μm line centered on the fovea was created, and the choroid was delineated in this area using the polygon tool and stored in the region of interest manager. Finally, the proportion of white pixels per total area in the delineated choroid was calculated with the measurement tool to obtain CVI (Fig. 1b).

All measurements were performed in all available OCT images separated by at least 1, 2, 3, and 4 years from baseline. For quality control, CVI and CT were measured twice by the same observer on different days and by an independent observer in 10% of the scans for intra-rater and inter-rater analysis.

Baseline characteristics of all study groups were summarized with traditional descriptive measures (mean and standard deviation for continue variables and percentages for dichotomic variables). Kruskal-Wallis test was used to compare continuous baseline characteristics between groups (CT and CVI, age) and Fisher’s exact test for dichotomic variables (sex and the presence of DDS at baseline). Multiple linear regression was used to evaluate the interaction of age, sex, and duration of FU with CT and CVI changes in each group.

In the cohort of eyes with non-advanced AMD, a logistic regression was performed to investigate the relationship between baseline CT and CVI and the development of exudation or central cRORA during FU, adjusted by age, sex, and the presence of SDDs and i/cRORA at baseline, as these are two factors strongly associated with disease progression according to previous studies [15]. Mixed-effect analysis was used to compare CT and CVI at different time points. For the CT, the analysis was repeated for each location: subfoveal, temporal, and nasal to the foveal center. Statistical significance was set at p < 0.05.

The Kendall’s coefficient of concordance (KCC) was used to calculate the intra- and inter-observer agreement. GraphPad Prism (v10.2.1) and RStudio (v2023.12.1) were used for statistical analysis.

We included 439 B-scans from 105 eyes (right eye N = 55, 52.4%) of 105 patients (N = 54, 51.4% female) with a mean age of 77.1 years (range 53–93). Mean FU time was 3.36 years (SD 0.89). Frequencies and baseline characteristics of all the study groups are summarized in Table 1. AMD stage at baseline was n = 46 (43.8%) non-advanced, n = 28 (26.7%) exudative, and n = 5 (4.8%) central cRORA in the study eye; at last FU was n = 17(%) non-advanced, n = 46 exudative (43.8%), and n = 15 (14.3%) central atrophic, with 10 eyes progressing from non-advanced to central atrophic and 18 eyes to exudative AMD.

Baseline CT values differed between groups in all sectors (p < 0.0001 subfoveal, p = 0.0016 temporal, and p = 0.0002 nasal), with healthy eyes having thicker choroids than those with AMD. In eyes with AMD, the thickest choroid was found in the exudative group (Table 2). During FU, the mean values were also different between time points in each study group, with a decreasing trend over time, with eyes with central atrophy having thinner choroids in all sectors (Fig. 2; Table 2). However, after adjustment for age at baseline and sex, the years of FU remained significant in all locations only in the exudative AMD group. With each year of FU, the CT decreased by 5.13% subfoveally (p < 0.0001), 5.28% temporally (p = 0.0003), and 6.27% nasally (p < 0.0001). These values were higher than those observed in healthy eyes, with subfoveal CT decreasing only 3.5% per year (p = 0.001) (Table 3).

In the subgroup of eyes with non-advanced AMD, an OR of 0.97 (95% CI: 0.94–0.99) and 1.02 (95% CI: 1.00–1.04) suggests that there is no strong relationship between subfoveal and nasal CT at baseline and the development of central atrophy during FU, adjusted for age, sex, and the presence of SDDs and RORA at baseline. Temporal CT was nonsignificant (95% CI: 0.99–1.01).

Mean baseline CVI showed different distribution among the four groups (p < 0.0001), with higher values in the control group than in eyes with AMD (Table 4). We observed a decrease in CVI over time, with significant differences in the mixed-effects analysis (Table 4; Fig. 2). However, when adjusted for sex and age at baseline, duration of FU was only significantly associated with CVI changes in the group of eyes presenting with GA at baseline (−1.09% per year, p = 0.022) (Table 3).

In eyes at risk of developing late AMD, CVI at baseline was not associated with the onset of both exudative and central atrophy (95% CI: 0.88–1.64 and 0.97–2.19). The other variables included in the model (age, sex, and SDDs and RORA at baseline) also had no significant effect.

Inter-rater analysis showed high agreement between observers both in CVI calculation (KCC = 0.92, p = 0.0006) and CT measurements (KCC = 0.94, p < 0.0001). Intra-rater agreement was also strong for CVI (KCC = 0.96, p = 0.0002) and CT (KCC = 0.95, p < 0.0001).

In our study, we observed thinner choroids with lower CVI in eyes with AMD compared to healthy eyes. Among all eyes with AMD, those with exudative AMD showed the greatest decrease in CT over time, whereas eyes with central atrophy showed the most significant decrease in CVI. These vascular parameters may aid in further stratification of AMD progression risk.

Several histopathological studies have shown a decrease in choroidal parameters with age which is accelerated in the presence of AMD [16‒18]. In vivo studies using OCT parameters such as CT and CVI have reproduced these findings [9, 19, 20]. In line with this, we found lower CT and CVI values in AMD eyes compared to healthy eyes, with a tendency to decrease over time [21].

In eyes with exudative AMD, reduction in CT and CVI could be explained by a reduced choroidal permeability caused by intravitreal anti-VEGF injections, which may vary depending on the drug used [22‒25]. In a recent study, Boscia et al. [26] reported a short-term reduction in CVI in patients treated with brolucizumab compared to those treated with aflibercept, which could be explained by the smaller size and higher tissue penetration of brolucizumab. However, there is a transient increase in CVI and CT in exudative AMD during active disease [27], which may explain the nonsignificance and the fluctuation of the CVI changes in the exudative AMD group in our study. Furthermore, we observed that some neovascular AMD OCT biomarkers (i.e., RPE tear or RPE detachment) could affect the CVI calculation by causing a shadow artifact that might disappear at the last FU due to collapse and atrophy. Further prospective studies including exudative AMD eyes are needed to clarify these issues.

In eyes with non-exudative AMD, Sacconi et al. proposed CVI as a biomarker for the onset and progression of geographic atrophy [20], which is supported by our results as we found the greatest CVI decrease in the group of central cRORA (−1.09% per year). Progressive choroidal remodeling with connective tissue replacing previously preexisting vessels could explain this reduction in CVI. Although CVI has been shown to be a more reliable parameter, especially relevant in eyes with atrophic AMD, it is difficult to calculate it as there are still no built-in tools in commercially available OCT platforms. We support its relevance to justify the inclusion of automated CVI calculation in the clinical setting, especially in the evaluation of eyes with central atrophy. Annual changes in choroidal parameters could help to stratify those patients at risk for progression to more advanced disease.

Strengths of our study include a standardized assessment of CT and CVI based on previously published protocols. Moreover, our results were expressed as percentage changes, thus ensuring comparability with existing literature, even when different image processing methods were employed. Additionally, we provide information on repeated measurements for each included eye at different time points, reducing the impact of fluctuations in choroidal perfusion, and an annual rate of change in CT and CVI in both AMD and control eyes.

CT and CVI have good reproducibility between graders [28, 29], which we confirmed with excellent intra- and inter-rater agreement. According to the literature, a normal choroid contains 66% of vascular tissue, which is consistent with our results (CVI = 66.9% in the control group), further validating our applied methods [30].

Limitations of our study include the retrospective design and the irregular distribution of the sample size between groups. Although the most pronounced reduction in CVI was evident in eyes with cRORA, we acknowledge the potential for overestimation bias due to the restricted sample size within this specific subgroup. Second, we did not consider other systemic diseases that may have influenced our results, especially high blood pressure or diabetes, which have been related to lower CT and CVI values in previous studies [31]. Third, CVI measurements may vary with the thresholding method used for image binarization. If a different protocol is used, the absolute values should be interpreted with caution. Fourth, we may have introduced a selection bias toward eyes with thicker choroids at baseline because we did not include eyes in which the scleral/choroidal interface was not clearly visible.

In conclusion, healthy eyes have higher CT and CVI values than eyes with AMD, but in all cases, this decreases over time. In eyes with AMD, thinner choroids may indicate a higher risk for the development of geographic atrophy. In exudative AMD, CVI may fluctuate during FU, but CT showed the greatest reduction per year compared to other stages of AMD. Higher rates of reduction in CVI may help to identify patients who need shorter FU, especially with the advent of geographic atrophy treatments in the future. In patients with non-exudative AMD, the identification of faster changes in CVI over time may facilitate the detection of individuals at risk for atrophy progression. This, in turn, may inform the selection of those who may derive greater benefit from active treatment with complement inhibitors or neuroprotective agents. Further studies in particular with AMD stages at high risk of progression to late-stage AMD are needed to further characterize the value of CT and CVI in stratifying AMD progression risk.

Ethical approval is not required for this study in accordance with local or national guidelines. Written informed consent from participants was not required for the study presented in this article in accordance with local/national guidelines.

L.V.-O. reports receiving travel funding from Bayer. R.P.F. reports receiving grants or research support from Biogen and Bayer and honoraria or other financial support from Apellis, Alimera, Allergan, Astellas, Caterna, Böringer-Ingelheim, Novartis, Ophthea, ODOS, ProGenerika, Roche/Genentech, and Stada Pharm. S.S., E.H.-A., and D.G. have no conflicts of interest to declare.

This study was supported by the DAAD (German Academic Exchange Service, scholarship to L.V.-O.).

Conceptualization: L.V.-O.; methodology and writing – original draft preparation: L.V.-O. and R.P.F.; formal analysis and investigation: L.V.-O., S.S., and E.H.-A.; writing – review and editing: D.G., S.S., E.H.-A., and R.P.F.; supervision: R.P.F. All the authors have read and agreed to the published version of the manuscript.

All the data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.