Abstract
Introduction: The aim of this study was to explore the relationship between choroidal biomarkers and the response to anti-VEGF in PCV eyes. Methods: We conducted a hospital-based retrospective study. We included 54 patients diagnosed with PCV who had received standard 3 monthly anti-VEGF monotherapy and had finished regular follow-ups. Choroidal thickness (CT), three-dimensional choroidal vascularity index (CVI), and the vascular density of choriocapillaris (CCVD) were measured utilizing swept-source optical coherence tomography angiography (SS-OCTA). Effective and poor responders were classified based on the changes in morphologic features. Multivariate linear regression models were performed for the outcomes to determine independent prognostic factors. Receiver operating characteristic (ROC) curves were used to compare the predictive ability of CT and CVI as biomarkers between effective and poor responders. Results: A higher CVI at baseline was the only factor that correlated with the poor response after 3 monthly injections of anti-VEGF (p = 0.038). The greater change of central macular thickness (CMT) was significantly correlated with increased CMT (p = 0.030), decreased CT (p = 0.042), and decreased CVI (p = 0.038) at baseline. Using ROC curves, we found that the CVI value demonstrated superior predictive ability compared to the CT value, with AUC of 0.842 and the best cut-off value of 0.445. Conclusion: A higher three-dimensional CVI using SS-OCTA is a promising biomarker to predict the poor anatomical response to anti-VEGF treatment in PCV patients.
Introduction
Polypoidal choroidal vasculopathy (PCV) is currently considered a subtype of neovascular age-related macular degeneration (nAMD), featured by branching vascular network (BVN) and polypoidal lesions [1]. PCV is more prevalent in Asian populations, with an estimated 22.3%–61.6% of nAMD cases in Asia being subtypes of PCV [2]. However, it is still controversial about whether PCV should be classified under the spectrum of nAMD or pachychoroid diseases, due to significant discrepancies in histopathology, clinical features, and responses to treatment [3, 4]. Many researchers emphasized the importance of a thick choroid (pachychoroid) and choroidal hyperpermeability in PCV [1, 5].
Traditionally, the lesion complex of PCV is typically best detected by indocyanine green angiography (ICGA) [3]. Rapid advancements in imaging technology have enabled the detailed examination of the choroid, especially swept-source optical coherence tomography angiography (SS-OCTA). To quantify the choroidal structure, choroidal thickness (CT) has been proposed as a crucial biomarker of concern. A fair proportion of PCV eyes show significantly increased CT, while other PCV eyes may also have normal or decreased CT, with considerable variation among research findings [6, 7]. An important reason is that both the choroidal vasculature and stroma can be affected by various physiological conditions. Choroidal vascularity index (CVI) is introduced to overcome the limitation of CT, defined as the ratio of vascular area to the total choroidal area, providing a more accurate representation of choroidal characteristics [8].
Despite the unclear role of vascular endothelial growth factor (VEGF) in the pathogenesis of PCV compared to nAMD, anti-VEGF therapy has been considered as the first-line therapeutic option in PCV patients according to published clinical trial data [9]. However, there were variable responses to anti-VEGF according to clinical trial data, indicating potential heterogeneity within PCV eyes [10]. However, there were variable responses to anti-VEGF treatment in published studies, indicating potential heterogeneity within PCV eyes. Differences in treatment outcomes may be associated with variations in choroidal vascular permeability. Multiple investigations reported that PCV eyes with thicker subfoveal CT responded poorly to anti-VEGF treatment [11‒13].
We can optimize the formulation of specific management approaches by identifying advanced imaging and clinical biomarkers. In our study, we investigated the relationship between multiple choroidal biomarkers, clinical parameters, and the response to 3 monthly anti-VEGF therapy in PCV eyes. We also compared the ability of CT and CVI as imaging biomarkers in predicting the anatomical outcomes of PCV eyes.
Methods
The retrospective observational study (NCT06243406) was conducted in the Department of Ophthalmology at Beijing Hospital and adhered to the tenets of the Declaration of Helsinki. Our study was approved by the Institutional Review Board Committee of Beijing Hospital (2022BJYYEC-048-01). We included patients diagnosed with PCV who had received a loading dose of 3 consecutive injections of 0.05 mL anti-VEGF drugs and had finished regular follow-ups from January 2022 to January 2024.
The inclusion criteria were as follows: PCV eyes confirmed by ICGA; eyes without anti-VEGF in the past 6 months; spherical equivalent refractive errors ranged from −3.0 diopters to +3.0 diopters (mild myopia to mild hyperopia). The exclusion criteria included a history of intraocular surgery or laser treatment including photodynamic therapy (PDT); history of ocular trauma or other ocular disorders; and obscured fundus images due to massive subfoveal hemorrhage or significant cataract. If bilateral eyes both fulfilled the criteria, one eye was randomly selected for analysis. The diagnosis of PCV was confirmed by the presence of polypoidal lesions with BVN detected in ICG, and was verified by two experienced and independent retinal specialists back to back (XBY and SS). Any discrepancies were resolved by discussion. CVH was defined as multifocal choroidal hyperfluorescence with blurred margins during the middle and late phases of the ICGA [14].
We collected patients’ demographics and medical information, including age, gender, history of hypertension and diabetes mellitus, history of ocular diseases, and concomitant therapy. All patients had undergone a comprehensive ophthalmic examination at the initial visit and follow-up, including best corrected visual acuity (BCVA) using LogMAR, slit-lamp biomicroscopy, dilated fundus examination, intraocular pressure (IOP), fluorescein angiography and ICGA (Spectralis HRA + OCT, Heidelberg Engineering, Heidelberg, Germany), as well as SS-OCTA (BM400K, TowardPi Medical Technology Co., Ltd., Beijing, China). Three-dimensional OCTA images of vertical 6 mm × 6 mm × scan depth 3 mm centered on the fovea were obtained using a swept laser source with a wavelength centered at 1,060 nm and a scan rate of 400,000 A-scans per second. All SS-OCTA images had been acquired by one experienced operator between 10:00 AM and 12:00 AM, to minimize the confounding effect of diurnal fluctuations in the CT.
In our study, CT was defined as the distance between the Bruch’s membrane (BM) and the chorio-scleral interface. The segmentation of BM and chorio-scleral interface was detected automatically using the built-in software. The borders of the choriocapillaris layer were delineated extending from the BM to 29 μm below the BM with the built-in algorithm. If necessary, manual modification was performed to correct the erroneous segmentation. Artifacts were minimized by employing volumetric projection artifact removal approaches. The three-dimensional CVI was calculated as the ratio of the choroidal vessel volume to the total choroid volume [15]. The vascular density of choriocapillaris (CCVD) was automatically computed as the ratio of the pixel areas of the vessels to the overall area of the choriocapillaris layer (%). All of the aforementioned indices were automatically measured and computed using the built-in software, as detailed in previously published studies [16, 17]. We selected 3 × 3 grids (supratemporal, superior, supranasal, temporal, central, nasal, inferotemporal, inferior, and inferonasal) centered in the macular fovea, with a total area of 6 mm × 6 mm (shown in Fig. 1), and calculated the average of 9 subfields as the choroidal parameters for analysis. To assess the agreement within and between examiners, the choroid vasculature measurements were segmented and obtained twice by the same examiner (YZ), as well as by two different examiners (JNW and YZ). Intraclass correlation coefficients and coefficients of repeatability were calculated to assess the reliability. The coefficient of repeatability was determined by multiplying the standard deviation of the difference by 1.96.
According to the recommended guidelines, acceptable and poor responders were classified based on the changes in morphologic features at the visit 1 month after the third injection compared to the baseline [18]. In our study, effective responders consist of good responders and partial responders defined in the guidelines, defined as follows: (1) absence of subretinal fluid (SRF), intraretinal fluid (IRF), intraretinal cysts (IRC), or a reduction in central macular thickness (CMT) > 75% of the baseline values (good responders); (2) reduction of CMT of between 25 and 75% of the baseline values, with persistence of SRF, IRF, IRC, or appearance of new IRC, IRF, and SRF (partial responders). Poor responders consist of poor responders and non-response defined in the guidelines, defined as follows: (1) between 0 and 25% reduction in CMT with the persistence of SRF, IRF, IRC, or appearance of new IRC, IRF, and SRF (poor responders); (2) unchanging or increasing CMT, SRF, IRF, and/or PED compared with the baseline (non-response).
Statistical analysis was conducted using the statistical software program R version 4.2.1 (R Foundation for Statistical Computing, Vienna, Austria). For continuous variables, the Shapiro-Wilk test was used to explore the distribution normality. For normal continuous variables, the independent t test was performed to test differences between the effective-responder and poor-responder groups; otherwise, the Wilcoxon rank-sum test was applied. The χ2 test or Fisher’s exact test was used for categorical analysis. A logistic regression model was employed to assess the association of the response status and prognostic factors. Both univariate and multivariate linear regression models were performed for continuous outcomes (the changes of CMT and BCVA after 3 monthly anti-VEGF treatments) to determine the independent prognostic factors. Receiver operating characteristic curves and their area under the curve (AUC) were used to assess and compare the predictive ability of CT and CVI as biomarkers between effective and poor responders. Youden’s index (defined as “sensitivity + specificity – 1”) was utilized to determine the optimal cut-off value. Two-sided p values <0.05 were considered statistically significant.
Results
In our study, 54 patients diagnosed with PCV in total were included. The population comprised 29 men (53.7%) and 25 women (46.3%), with a mean age of 68.20 ± 7.82 years. According to the definition, there were twenty-three eyes (42.6%) and thirty-one eyes (57.4%) classified as effective and poor responders to 3 monthly anti-VEGF treatments, respectively.
The comparison of demographic data and clinical characteristics between these two groups is summarized in Table 1. The agreement within and between the examiners was satisfactory as shown in online supplementary material 1 (for all online suppl. material, see https://doi.org/10.1159/000541572). There were no significant differences in age, gender, type of drugs, IOP, the proportion of hypertension and diabetes, the proportion of smoking history, the presence of macular hemorrhage and CVH, BCVA at baseline, CMT, and CCVD between effective- and poor-responder groups at baseline. The CT and CVI at baseline of effective responders were both significantly lower than poor responders (p < 0.01).
Parameters . | Effective responders (N = 23) . | Poor responders (N = 31) . | p value . |
---|---|---|---|
Age, years | 69.04±6.63 | 67.58±9.80 | 0.516 |
Gender, male, n (%) | 14 (60.9) | 15 (48.4) | 0.526 |
Type of drugs, n (%) | |||
Ranibizumab | 11 (47.8) | 13 (41.9) | 0.878 |
Aflibercept | 12 (52.2) | 18 (58.1) | |
IOP, mm Hg | 13.22±3.80 | 13.58±3.99 | 0.735 |
Hypertension, n (%) | 6 (26.1) | 12 (38.7) | 0.496 |
Diabetes, n (%) | 6 (26.1) | 12 (38.7) | 0.496 |
Smoking history, n (%) | 8 (34.8) | 14 (45.2) | 0.626 |
Initial hemorrhage, n (%) | 6 (26.1) | 5 (16.1) | 0.578 |
Initial CVH, n (%) | 7 (30.4) | 13 (41.9) | 0.562 |
Baseline BCVA (LogMAR) | 0.615±0.40 | 0.684±0.39 | 0.530 |
Baseline CMT, μm | 313.22±30.70 | 310.32±31.51 | 0.091 |
Baseline CT, μm | 262.13±24.91 | 295.58±25.18 | <0.01* |
Baseline CVI | 0.370±0.065 | 0.503±0.085 | <0.01* |
Baseline CCVD, % | 53.92±2.37 | 52.20±2.54 | 0.125 |
Parameters . | Effective responders (N = 23) . | Poor responders (N = 31) . | p value . |
---|---|---|---|
Age, years | 69.04±6.63 | 67.58±9.80 | 0.516 |
Gender, male, n (%) | 14 (60.9) | 15 (48.4) | 0.526 |
Type of drugs, n (%) | |||
Ranibizumab | 11 (47.8) | 13 (41.9) | 0.878 |
Aflibercept | 12 (52.2) | 18 (58.1) | |
IOP, mm Hg | 13.22±3.80 | 13.58±3.99 | 0.735 |
Hypertension, n (%) | 6 (26.1) | 12 (38.7) | 0.496 |
Diabetes, n (%) | 6 (26.1) | 12 (38.7) | 0.496 |
Smoking history, n (%) | 8 (34.8) | 14 (45.2) | 0.626 |
Initial hemorrhage, n (%) | 6 (26.1) | 5 (16.1) | 0.578 |
Initial CVH, n (%) | 7 (30.4) | 13 (41.9) | 0.562 |
Baseline BCVA (LogMAR) | 0.615±0.40 | 0.684±0.39 | 0.530 |
Baseline CMT, μm | 313.22±30.70 | 310.32±31.51 | 0.091 |
Baseline CT, μm | 262.13±24.91 | 295.58±25.18 | <0.01* |
Baseline CVI | 0.370±0.065 | 0.503±0.085 | <0.01* |
Baseline CCVD, % | 53.92±2.37 | 52.20±2.54 | 0.125 |
IOP, intraocular pressure; CVH, choroidal vascular hyperpermeability; BCVA, best corrected visual acuity; CMT, central macular thickness; CT, choroidal thickness; CVI, choroidal vascularity index; CCVD, vascular density of choriocapillaris.
*p value <0.05.
Table 2 presents the results of univariate and multivariate logistic regression models of factors affecting the response status. Univariate analysis revealed that a higher CMT (p = 0.038) at baseline was significantly associated with an effective response, while higher CT (p < 0.01) and CVI (p < 0.01) were significantly associated with a poor response. Multivariate analysis revealed that the higher CVI at baseline was the only factor that correlated with the poor response (p = 0.038).
Variables . | Univariate . | Multivariate . | ||
---|---|---|---|---|
β coefficient . | p value . | β coefficient . | p value . | |
Age | 0.0240 | 0.231 | 0.016 | 0.116 |
Gender (male) | −0.0955 | 0.548 | 0.014 | 0.694 |
Drug (Ranibizumab) | 0.004 | 0.924 | 0.098 | 0.735 |
IOP | −0.0029 | 0.882 | 0.0032 | 0.892 |
Hypertension | −0.0557 | 0.746 | 0.0864 | 0.456 |
Diabetes | −0.167 | 0.329 | −0.072 | 0.593 |
Smoking history | −0.028 | 0.839 | 0.017 | 0.686 |
Initial hemorrhage | 0.154 | 0.444 | −0.126 | 0.417 |
Initial CVH | −0.104 | 0.372 | −0.077 | 0.690 |
Baseline BCVA | −0.156 | 0.413 | −0.035 | 0.723 |
Baseline CMT | 0.00077 | 0.038* | 0.0020 | 0.349 |
Baseline CT | −0.0094 | <0.01* | −0.0051 | 0.143 |
Baseline CVI | −3.343 | <0.01* | −1.989 | 0.038* |
Baseline CCVD | 0.044 | 0.183 | 0.020 | 0.294 |
Variables . | Univariate . | Multivariate . | ||
---|---|---|---|---|
β coefficient . | p value . | β coefficient . | p value . | |
Age | 0.0240 | 0.231 | 0.016 | 0.116 |
Gender (male) | −0.0955 | 0.548 | 0.014 | 0.694 |
Drug (Ranibizumab) | 0.004 | 0.924 | 0.098 | 0.735 |
IOP | −0.0029 | 0.882 | 0.0032 | 0.892 |
Hypertension | −0.0557 | 0.746 | 0.0864 | 0.456 |
Diabetes | −0.167 | 0.329 | −0.072 | 0.593 |
Smoking history | −0.028 | 0.839 | 0.017 | 0.686 |
Initial hemorrhage | 0.154 | 0.444 | −0.126 | 0.417 |
Initial CVH | −0.104 | 0.372 | −0.077 | 0.690 |
Baseline BCVA | −0.156 | 0.413 | −0.035 | 0.723 |
Baseline CMT | 0.00077 | 0.038* | 0.0020 | 0.349 |
Baseline CT | −0.0094 | <0.01* | −0.0051 | 0.143 |
Baseline CVI | −3.343 | <0.01* | −1.989 | 0.038* |
Baseline CCVD | 0.044 | 0.183 | 0.020 | 0.294 |
IOP, intraocular pressure; CVH, choroidal vascular hyperpermeability; BCVA, best corrected visual acuity; CMT, central macular thickness; CT, choroidal thickness; CVI, choroidal vascularity index; CCVD, vascular density of choriocapillaris.
*p value <0.05.
Table 3 displays the results of univariate and multivariate linear regression analysis of factors affecting the change of CMT and BCVA after 3 monthly anti-VEGF treatments. In our research, the change of BCVA was defined as the BCVA after 3 monthly anti-VEGF treatments minus the BCVA at baseline, and the change of CMT was defined as the proportion of the absolute values of the reduction in CMT accounting for the baseline CMT. Univariate analysis showed that only BCVA at baseline was significantly associated with the change of BCVA (p = 0.002), and multivariate analysis showed that none of the factors were associated with the change of BCVA. Univariate analysis showed that the greater change of CMT was significantly correlated with increased CMT at baseline (p < 0.01), decreased CT at baseline (p < 0.01), decreased CVI at baseline (p < 0.01), and increased CCVD at baseline (p = 0.031). Further multivariate analysis revealed that CMT (p = 0.030), CT (p = 0.042) and CVI (p = 0.038) at baseline were biomarkers that remained statistically significant.
Variables . | Change in BCVAa . | Change in CMTb . | ||||||
---|---|---|---|---|---|---|---|---|
univariate β coefficient . | univariate p value . | multivariate β coefficient . | multivariate p value . | univariate β coefficient . | univariate p value . | multivariate β coefficient . | multivariate p value . | |
Age | 0.0046 | 0.163 | 0.0028 | 0.410 | 0.00035 | 0.647 | 0.00052 | 0.533 |
Gender (male) | −0.115 | 0.102 | −0.0822 | 0.129 | −0.0122 | 0.254 | 0.0108 | 0.297 |
Drug (Ranibizumab) | −0.0667 | 0.348 | −0.0532 | 0.598 | 0.0073 | 0.662 | 0.0048 | 0.831 |
IOP | 0.0072 | 0.208 | 0.0057 | 0.236 | −0.0027 | 0.338 | −0.0011 | 0.576 |
Hypertension | 0.014 | 0.780 | −0.0246 | 0.652 | 0.0063 | 0.589 | −0.00960 | 0.470 |
Diabetes | 0.0862 | 0.281 | 0.0529 | 0.357 | −0.0019 | 0.872 | −0.00402 | 0.753 |
Smoking history | 0.0174 | 0.757 | −0.0023 | 0.906 | −0.0052 | 0.554 | −0.0038 | 0.688 |
Initial hemorrhage | 0.0809 | 0.168 | 0.0278 | 0.841 | −0.0086 | 0.528 | −0.0123 | 0.434 |
Initial CVH | 0.0593 | 0.419 | 0.0424 | 0.485 | −0.0041 | 0.738 | −0.0029 | 0.192 |
Baseline BCVA | −0.1632 | 0.002* | −0.115 | 0.106 | −0.0065 | 0.616 | −0.0133 | 0.323 |
Baseline CMT | 0.00126 | 0.096 | 0.00074 | 0.320 | 0.00051 | <0.01* | 0.00048 | 0.030* |
Baseline CT | −0.00058 | 0.476 | 0.00033 | 0.542 | −0.000087 | <0.01* | −0.00033 | 0.042* |
Baseline CVI | 0.179 | 0.510 | 0.0776 | 0.892 | −0.0217 | <0.01* | −0.0113 | 0.038* |
Baseline CCVD | −0.017 | 0.084 | −0.012 | 0.145 | 0.0025 | 0.031* | 0.0035 | 0.179 |
Variables . | Change in BCVAa . | Change in CMTb . | ||||||
---|---|---|---|---|---|---|---|---|
univariate β coefficient . | univariate p value . | multivariate β coefficient . | multivariate p value . | univariate β coefficient . | univariate p value . | multivariate β coefficient . | multivariate p value . | |
Age | 0.0046 | 0.163 | 0.0028 | 0.410 | 0.00035 | 0.647 | 0.00052 | 0.533 |
Gender (male) | −0.115 | 0.102 | −0.0822 | 0.129 | −0.0122 | 0.254 | 0.0108 | 0.297 |
Drug (Ranibizumab) | −0.0667 | 0.348 | −0.0532 | 0.598 | 0.0073 | 0.662 | 0.0048 | 0.831 |
IOP | 0.0072 | 0.208 | 0.0057 | 0.236 | −0.0027 | 0.338 | −0.0011 | 0.576 |
Hypertension | 0.014 | 0.780 | −0.0246 | 0.652 | 0.0063 | 0.589 | −0.00960 | 0.470 |
Diabetes | 0.0862 | 0.281 | 0.0529 | 0.357 | −0.0019 | 0.872 | −0.00402 | 0.753 |
Smoking history | 0.0174 | 0.757 | −0.0023 | 0.906 | −0.0052 | 0.554 | −0.0038 | 0.688 |
Initial hemorrhage | 0.0809 | 0.168 | 0.0278 | 0.841 | −0.0086 | 0.528 | −0.0123 | 0.434 |
Initial CVH | 0.0593 | 0.419 | 0.0424 | 0.485 | −0.0041 | 0.738 | −0.0029 | 0.192 |
Baseline BCVA | −0.1632 | 0.002* | −0.115 | 0.106 | −0.0065 | 0.616 | −0.0133 | 0.323 |
Baseline CMT | 0.00126 | 0.096 | 0.00074 | 0.320 | 0.00051 | <0.01* | 0.00048 | 0.030* |
Baseline CT | −0.00058 | 0.476 | 0.00033 | 0.542 | −0.000087 | <0.01* | −0.00033 | 0.042* |
Baseline CVI | 0.179 | 0.510 | 0.0776 | 0.892 | −0.0217 | <0.01* | −0.0113 | 0.038* |
Baseline CCVD | −0.017 | 0.084 | −0.012 | 0.145 | 0.0025 | 0.031* | 0.0035 | 0.179 |
IOP, intraocular pressure; CVH, choroidal vascular hyperpermeability; BCVA, best corrected visual acuity; CMT, central macular thickness; CT, choroidal thickness; CVI, choroidal vascularity index; CCVD, vascular density of choriocapillaris.
*p value <0.05.
aChange in BCVA: BCVA after 3 monthly anti-VEGF treatments minus BCVA at baseline.
bChange in CMT: the proportion of the absolute values of the reduction of CMT accounting for the baseline CMT.
The receiver operating characteristic curves of response status versus CT and CVI values at baseline are shown in Figure 2. The AUC of CT was 0.813 (range, 0.697–0.928), with an overall sensitivity of 0.826 and a specificity of 0.677. The AUC of CVI was 0.879 (range, 0.782–0.976), with an overall sensitivity of 0.870 and a specificity of 0.742. Through comparison, the AUC of CVI was significantly higher than that of CT (p = 0.033). Based on the Youden index, the optimal cut-off point of CT for predicting a poor response to anti-VEGF treatment was 280.5, and the Youden index was 0.503. The optimal cut-off point of CVI was 0.445 with the Youden index of 0.612. It revealed that the CVI value showed a better ability for predicting the response status than the CT value at baseline.
We used the cut-off point of CVI as an acceptable biomarker for dividing all PCV eyes into high-CVI and low-CVI subgroups accordingly. The representative SS-OCTA images of high-CVI and low-CVI PCV eyes at baseline and the visit 1 month after the third injection are displayed in online supplementary material 2. The comparison of parameters between high-CVI and low-CVI PCV eyes is shown in Table 4. High-CVI PCV eyes showed a significantly higher proportion of CVH, higher CT values, and lower CCVD values compared to low-CVI PCV eyes. Meanwhile, there were significantly more effective responders in low-CVI PCV eyes than in high-CVI PCV eyes.
Parameters . | High-CVI (N = 26) . | Low-CVI (N = 28) . | p value . |
---|---|---|---|
Age, years | 68.28±7.33 | 68.13±8.04 | 0.781 |
Gender, male, n (%) | 17 (27.4) | 12 (42.9) | 0.166 |
Type of drugs, n (%) | |||
Ranibizumab | 13 (50.0) | 11 (39.3) | 0.605 |
Aflibercept | 13 (50.0) | 17 (60.7) | |
IOP (mm Hg) | 12.89±3.37 | 13.82±3.74 | 0.384 |
Hypertension, n (%) | 8 (30.8) | 10 (35.7) | 0.923 |
Diabetes, n (%) | 7 (26.9) | 11 (39.3) | 0.500 |
Smoking history, n (%) | 9 (34.6) | 13 (46.4) | 0.555 |
Initial hemorrhage, n (%) | 5 (19.2) | 6 (21.4) | 1.00 |
Initial CVH, n (%) | 15 (57.7) | 5 (17.9) | 0.006* |
Baseline BCVA (LogMAR) | 0.639±0.43 | 0.655±0.38 | 0.735 |
Baseline CMT, μm | 311.59±288.16 | 312.08±31.73 | 0.639 |
Baseline CT, μm | 304.65±26.84 | 253.27±25.30 | <0.01* |
Baseline CCVD, % | 51.7±2.06 | 54.3±2.21 | 0.022* |
Effective response, n (%) | 6 (23.1) | 17 (60.7) | 0.012* |
Parameters . | High-CVI (N = 26) . | Low-CVI (N = 28) . | p value . |
---|---|---|---|
Age, years | 68.28±7.33 | 68.13±8.04 | 0.781 |
Gender, male, n (%) | 17 (27.4) | 12 (42.9) | 0.166 |
Type of drugs, n (%) | |||
Ranibizumab | 13 (50.0) | 11 (39.3) | 0.605 |
Aflibercept | 13 (50.0) | 17 (60.7) | |
IOP (mm Hg) | 12.89±3.37 | 13.82±3.74 | 0.384 |
Hypertension, n (%) | 8 (30.8) | 10 (35.7) | 0.923 |
Diabetes, n (%) | 7 (26.9) | 11 (39.3) | 0.500 |
Smoking history, n (%) | 9 (34.6) | 13 (46.4) | 0.555 |
Initial hemorrhage, n (%) | 5 (19.2) | 6 (21.4) | 1.00 |
Initial CVH, n (%) | 15 (57.7) | 5 (17.9) | 0.006* |
Baseline BCVA (LogMAR) | 0.639±0.43 | 0.655±0.38 | 0.735 |
Baseline CMT, μm | 311.59±288.16 | 312.08±31.73 | 0.639 |
Baseline CT, μm | 304.65±26.84 | 253.27±25.30 | <0.01* |
Baseline CCVD, % | 51.7±2.06 | 54.3±2.21 | 0.022* |
Effective response, n (%) | 6 (23.1) | 17 (60.7) | 0.012* |
CVI, choroidal vascularity index; PCV, polypoidal choroidal vasculopathy; IOP, intraocular pressure; CVH, choroidal vascular hyperpermeability; BCVA, best corrected visual acuity; CMT, central macular thickness; CT, choroidal thickness; CCVD, vascular density of choriocapillaris.
*p value <0.05.
Discussion
This retrospective study in the Asian population diagnosed with PCV evaluated multiple parameters correlated with treatment response. The higher CVI at baseline was the only factor that correlated with the poor response after 3 monthly injections of anti-VEGF. We innovatively applied the three-dimensional CVI to predict the worse response to anti-VEGF treatment and found that the CVI value showed superior predictive ability compared to the CT value. In our results, the best cut-off point of CT value was 280.5 μm, slightly higher than the cut-off point reported by Chang et al. (267.5 μm) and Jiménez-Santos et al. (257 μm) [12, 13]. After using the optimal cut-off point of CVI (0.445) as a biomarker, high-CVI PCV eyes exhibited a significantly higher proportion of choroidal vascular hyperpermeability, higher CT values, lower CCVD values, and poorer response to anti-VEGF treatment.
Currently, the therapeutic options for PCV mainly evolved from typical nAMD. Based on the EVEREST II study, compared to ranibizumab alone, ranibizumab combined with PDT achieved better visual improvements and a higher rate of PL closure [19]. The Asia-Pacific Vitreo-retina Society recommended treat-and-extend (T&E) dosing regimens as efficacious and safe choices for nAMD and PCV patients [20]. In clinical practice, our findings may contribute to developing treatment strategies for PCV. For pachychoroid PCV eyes, only anti-VEGF monotherapy may be insufficient, and the combination of anti-VEGF with PDT has been revealed to be a better strategy. PDT could remodel the vasculatures of the choroid and reduce excessive permeability of the choroidal vasculature. Sakurada et al. found that combination therapy could achieve better visual outcomes and less need for retreatment in PCV eyes with greater subfoveal CT [21].
According to recently published literature, PCV is considered a type overlapping between AMD and pachychoroid disease. It has been reported that several novel metrics for quantifying the choroid are associated with treatment outcomes of PCV with inconsistent conclusions. Chang et al. set the SFCT value of 267.5 µm as a cut-off line to divide PCV patients into pachychoroid and nonpachychoroid subgroups [13]. The pachychoroid PCV eyes had significantly younger age, fewer AMD-like features, and less response to 3 monthly anti-VEGF treatments (27.8 vs. 83.3%). Fenner et al. found that in a PCV monotherapy treatment trial, a lower CVI at baseline was significantly associated with reduced disease activity at month 12 [22]. Kim et al. also reported that a thinner subfoveal choroid at baseline significantly correlated with favorable treatment response (complete resolution of macular exudation) [11]. On the other hand, the EVEREST II cohort study showed that there was no significant relationship between central CT and anatomical or functional outcomes [23].
To solve the diversity of the published results, CVI is proposed as a more robust and reliable indicator than the CT value. It has been demonstrated that CT could vary obviously with age, gender, and refractive error and showed variability and poor reproducibility to segmentation errors [24]. At the same time, the choroid consists of the stroma and vessels; hence, CT values cannot exactly reveal the anomalous change of choroidal structure. Since the introduction of CVI, CVI has been reported a lower covariance and was resistant to physiological parameters, including age, systolic blood pressure, axial length or IOP, as well as the type of OCT machine and the scanning area [25]. Vyas et al. [26] investigated the predictive value of several biomarkers, including volumetric CVI on visual and anatomical outcomes in eyes with PCV following anti-VEGF therapy, and demonstrated a positive correlation between volumetric CVI calculated through OCTA and disease activity at month 12 [26]. In our study, we used automated algorithms to evaluate the three-dimensional volume of choroidal vessels based on the entire scan, which is more accurate compared to two-dimensional images from single B-scans. Moreover, our analysis demonstrated that three-dimensional CVI was a more accurate biomarker than CT for the prediction of anatomical outcomes.
In the previous study, PCV eyes with thick and thin choroids resembled central serous chorioretinopathy and AMD, respectively [27]. PCV eyes with pachychoroid phenotype exhibit dilated large choroidal vessels and choriocapillary attenuation [28]. We provided a potential biomarker to explain the inconsistency within PCV eyes. After adjusting for age, gender, IOP, presence of hypertension, diabetes and macular hemorrhage, and other OCTA measurements in our multivariate analysis, it was shown that increased CVI was still a significant predictor for the poor response to anti-VEGF treatment. For one reason, several studies have demonstrated that the functions of VEGF in the genesis and stimulation of type 1 macular neovascularization (MNV) are different. MNV with pachychoroid phenotype is typically characterized by large caliber vessels and lack of capillaries within the lesions, which is primarily related to arteriogenesis [29, 30]. The mechanism of arteriogenesis is the dilation of preexisting mature vessels rather than the germination of new capillaries, and this process is mainly driven by platelet-derived growth factor instead of VEGF. This theory might explain the lower levels of VEGF in pachychoroid neovasculopathy and inadequate reaction to anti-VEGF treatment of pachychoroid PCV eyes [30]. However, further evaluation of the morphological characteristics of MNV in these two subtypes is required to validate this hypothesis.
For another possible reason, in pachychoroid PCV eyes with persistent fluid accumulation, we can speculate that the observed exudative fluid may not result from associated MNV lesions [30]. It might result from the chronic impairment of the retinal pigment epithelial (RPE)-BM complex caused by long-standing vascular hyperpermeability and choroidal ischemia, leading to RPE pump malfunction [31]. The disruption of the balance between RPE pumping ability and the choroidal hydrostatic pressure could cause persistent accumulation of fluid. Choroidal vasculature exposed to such chronic pressures may develop aneurismal dilatations or polypoidal lesions [32]. Furthermore, the morphology of non-exudative MNV is featured by the long-standing presence of massive neovascular trunks and fewer capillaries, indicating the underlying compensatory effect of non-exudative MNV in chronic pachychoroid disorders. Our research observed that PCV eyes with a high CVI showed significantly lower CCVD than PCV eyes with a low CVI. We reconfirmed that pachy-vessels might cause a sequence of choriocapillaris dysfunction and structural attenuation followed by subsequent RPE complications such as MNV. The possible explanations are the inward displacement of the Haller's vessels and the disintegration of the RPE barrier.
Advanced imaging technologies and extensive literature have led to the discovery of various morphological biomarkers that could potentially forecast treatment response and recurrence risk. There is evidence suggesting that pachydrusen, irregular RPE elevation, and CVH may be predictive biomarkers for progression [33]. Bo et al. found that the development of nonexudative BVN and the progression of polypoidal lesions were associated with exudative recurrences [34]. Although the major strength of our study is the volumetric analysis of the choroidal vascular by integrating multiple scans, only the parameters of choroidal characteristics are insufficient for comprehensively predicting the response. Additional parameters still need investigation, especially the phenotype of PCV lesions, including the total lesion areas, total vessel length, junction density, and vessel density. Further evaluation of the association between the choroidal features and the lesion characterization will also contribute to explore the mechanisms of different subtypes of PCV. Moving forward, it is expected to establish a predictive model based on multimodal imaging parameters.
Indeed, there exist several limitations in our study. First, one major limitation is the retrospective design and the relatively small sample size. Secondly, we only used the morphological changes as the indication of response to treatment, and morphological and functional responses may not correlate. Thirdly, our research included PCV patients who received the treatment of different anti-VEGF agents, which might increase variations in the evaluation of treatment effects. Nevertheless, current available evidence suggests that efficacy and safety are comparable across different anti-VEGF agents [35]. Fourthly, greater axial length was associated with attenuated choroidal circulation, and refractive changes can affect the choroid vascularity index according to recently published studies [36, 37]. However, the measurement of axial length was not routinely performed in patients in our study. Although we limited the refractive status of included eyes to minimize the potential influence of the axial length on choroidal measurements, it might also bring potential selection bias. Lastly, we only analyzed the response status and related factors after the initial 3 monthly treatment, without tracking the patients for their long-term response and recurrence status.
In conclusion, our study revealed that three-dimensional CVI using SS-OCTA was a promising biomarker to predict the response to anti-VEGF treatment in PCV patients. This biomarker could also be a useful tool to identify the population at risk of an inadequate response to anti-VEGF, and more frequent follow-up visits and supplemental therapy like PDT could be recommended. PCV eyes with a high CVI tended to have a higher proportion of CVH, higher CT values, lower CCVD values, and to have poorer responses to 3 monthly anti-VEGF monotherapy. The notable distinctions may broaden our comprehension of the pathogenesis of this disease.
Statement of Ethics
This study protocol was reviewed and approved by the Institutional Review Board Committee of Beijing Hospital, Approval No. [2022BJYYEC-048-01]. Written informed consent was not required. The need for informed consent was waived by the Institutional Review Board Committee of Beijing Hospital, Approval No. [2022BJYYEC-048-01].
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
Funding Sources
This work was supported by Central High Level Hospital Clinical Research Funding (BJ-2024-089) and National High Level Hospital Clinical Research Funding (BJ-2022-104).
Author Contributions
Y.Z. analyzed the data and conceptualized and drafted the manuscript. J.W., J.L., S.S., and X.G. contributed to the acquisition of data. X.Y. provided comments and revised the manuscript. All authors read and approved the final version of the manuscript.
Data Availability Statement
The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from Xiaobing Yu upon reasonable request.