Long-term effect of photodynamic therapy on choroidal vascularity index versus choroidal thickness as biomarkers for chronic central serous chorioretinopathy

Abstract

Objective

To investigate the long-term impact of half-fluence photodynamic therapy (PDT) on chorioretinal architecture in chronic central serous chorioretinopathy (cCSCR) through novel choroidal vascularity index (CVI) versus previously established subfoveal choroidal thickness (SFCT).

Methods

This post-hoc analysis included prospectively collected swept-source optical coherence tomography (SS-OCT) images of a total of 29 cCSCR and fellow eyes (FE), acquired before, one and 12 months after PDT. CVI, total choroidal area (TCA), luminal area (LA) and stromal area (SA) were calculated using validated binarization technique. Central macular thickness (CMT), SFCT and subretinal fluid (SRF) were assessed. Longitudinal changes following PDT and associations between all parameters were investigated.

Results

CVI, TCA, LA, SFCT and CMT significantly decreased at month 1 (all p < 0.001). From month 1 to 12, CVI (p < 0.001) and LA (p = 0.01) increased but remained significantly lower than at baseline (p = 0.001 and p < 0.001). SFCT and CMT remained stable (all p = 1.0). SA and all FE parameters showed no significant changes at any time point. Comparing affected and FE, there were significant CVI differences at baseline (p < 0.001), month 1 (p = 0.01) and 12 (p < 0.001). SRF was significantly associated with higher CVI (p < 0.001), LA (p = 0.031), SFCT (p = 0.038).

Conclusion

Half-fluence PDT significantly affects choroidal architecture, with sustained CVI and SFCT reductions 12 months post-treatment. Use of CVI as both a treatment response and disease activity marker is merited. There is a large discrepancy between CVI and SFCT, suggesting they each represent different anatomic responses. CVI proves to be a more specific and reliable CSCR biomarker than previously established SFCT.

Introduction

Central serous chorioretinopathy (CSCR) belongs to the pachychoroid disease spectrum, a phenotype characterized by attenuation of the choriocapillaris overlying dilated choroidal veins, and associated with progressive retinal pigment epithelium dysfunction [1]. Typical CSCR findings include a thickened, congested choroid with accumulation of subretinal fluid (SRF) [2]. While acute CSCR is often self-limiting with spontaneous complete fluid reabsorption and good visual recovery within 3 months, SRF and vision distortion persist in 10 to 15%, resulting in chronic disease with associated complications including RPE and photoreceptor atrophy and secondary macular neovascularization (MNV) [3]. Hence treatment intervention is required to prevent irreversible vision loss. Photodynamic therapy (PDT) with verteporfin has proven to be a safe and efficacious treatment option and is considered first-line treatment in the clinical setting. By inducing vascular hypoperfusion, SRF resolution and an increase of visual acuity was reported, regardless of applying the half-fluence or half-dose protocol [4].

The advent of multimodal imaging has considerably improved our understanding of pathophysiologic mechanisms in CSCR. Particularly swept-source optical coherence tomography (SS-OCT) enables detailed high-resolution visualization of the thickened choroid with its individual layers [5]. These technical advances have led to implementation of OCT-based quantitative surrogate markers in an attempt to estimate choroidal health. Historically, subfoveal choroidal thickness (SFCT) was regarded as the primary biomarker for CSCR and adopted in most studies [6]. SFCT was demonstrated to be higher in both affected and fellow eyes (FE), indicating bilaterality of disease, and to reduce with fluid resolution [7]. In recent years, a novel biomarker termed choroidal vascularity index (CVI) has garnered considerable scientific interest. CVI is defined as the ratio of luminal area (LA) and total choroidal area (TCA), corresponding to vascular and both vascular and stromal components, respectively, and is expressed as a percentage [8]. It was reported to be higher in various pathologies, including CSCR [3]. In contrast to prior biomarkers, it allows for demarcation of individual choroidal components, hence it was suggested to more robustly evaluate choroidal status [6].

To date, the influence of PDT on choroidal vasculature has not been widely explored yet. Particularly, studies about long-term treatment effect are very limited. This study was performed to investigate the long-term effect of half-fluence PDT on chorioretinal architecture using a wide range of qualitative and quantitative biomarkers including novel CVI, in a cohort of patients comprised by chronic CSCR (cCSCR).

Materials and methods

Study design and patient selection

This was a post-hoc analysis of a previous prospective study, conducted at the Karl Landsteiner Institute for Retinal Research and Imaging, Department of Ophthalmology and Optometry of Vienna Health Care Group Clinic Landstraße, Vienna, Austria, after approval from the Ethics Committee of Vienna (EK 17-083-0517). The research was performed in compliance with the tenets of the Declaration of Helsinki and Good Clinical Practice.

This study included eyes with cCSCR, diagnosed between January 2018 and September 2020 and followed for at least 12 months. Diagnosis of cCSCR was made by non-resolving fluid in OCT for at least 12 weeks duration. All eyes showed foveal involvement, confirmed by active leakage in fluorescein angiography and choroidal vascular hyperpermeability in indocyanine green angiography (ICGA), as seen in Fig. 1A, B. Within one month after first presentation and diagnosis at our clinic, all eyes received ICGA-guided PDT according to a modified protocol using a reduced fluence rate of 25 J/cm² and intensity of 300 mW/cm² with full-dose (6 mg/m²) verteporfin once. Treatment-naïve FE were included as controls. Importantly, all FE were further distinguished between inactive, with pachychoroid and no previous CSCR activity, vs. active, with extrafoveal fluid of at least 12 weeks duration, but not requiring treatment (Fig. 1C). Eyes with a history of retinal diseases other than CSCR, previous treatment including PDT, photocoagulation, mineralcorticoid antagonist or anti-vascular endothelial growth factor, and history of systemic or topical corticosteroid use were excluded from analysis to avoid potential confounding effects.

Fig. 1.

Simultaneous fluorescein angiography (A) and indocyanine green angiography (B) of the left eye from a 46-year-old male patient with chronic central serous chorioretinopathy at baseline, showing a focal leakage point and mottled hyperfluorescence in the macular periphery, indicating choroidal hyperpermeability. C Swept-source optical coherence tomography B-scan of the corresponding active fellow eye showing extrafoveal subretinal fluid, not requiring treatment.

Image acquisition

Imaging data were prospectively collected one day before, one and 12 months after PDT. At each visit, scans were recorded using Topcon’s DRI Triton SS-OCT-Angiography (Topcon Corporation, Tokyo, Japan). This device utilizes a 1050 nm wavelength light source, acquires 100.000 A-scans per second and provides a lateral resolution of 20 µm and axial resolution of 2.6 µm (digital) and 8 µm (optical). Further, it is equipped with the motion contrast measure algorithm OCTA Ratio Analysis, optimizing axial resolution, as well as an active eye tracker to minimize motion artifacts [9]. The horizontal single 12-mm subfoveal B-scan was chosen for analysis as it was shown to be representative for vasculature of the entire choroid [10]. The integrated follow-up function ensured exact B-scan location in consecutive visits. All scans were acquired between 9 am to 12 pm to avoid diurnal choroidal changes.

Image grading

Topcon’s integrated analysis software IMAGEnet6 was used for manual annotation and calculation of SFCT, defined as the vertical distance between the outer border of hyperreflective Bruch’s membrane and inner hyperreflective line of choroidal-scleral-interface (CSI); and of central macular thickness (CMT), the distance between anterior RPE border and inner limiting membrane in the central 1000 µm of the fovea [6]. SRF and PED presence was graded. For CVI assessment, all scans were imported into established image analysis software ImageJ. Choroidal segmentation was performed using validated image binarization technique [8]. After binarization using Niblack’s autolocal threshold, the entire choroid was manually annotated using the polygon tool - with the upper border at Bruch’s membrane and lower border at the hyperreflective line of the CSI. The selected area was then converted to a red-green-blue image using colour threshold tool, followed by automatic computation of TCA and LA (dark pixels). Stromal area (SA, light pixels) was calculated by subtracting LA from TCA, CVI by dividing LA by TCA [8] (Fig. 2).

Fig. 2. Swept-source optical coherence tomography subfoveal B-scans and corresponding binarized scans of the choroid from a 48-year-old male patient with chronic central serous chorioretinopathy before and after half-fluence photodynamic therapy.

A, B Baseline scan with central subretinal fluid (SRF), increased subfoveal choroidal thickness (SFCT) of 425 µm and increased choroidal vascularity index (CVI) of 69.88%. C, D At the 1-month follow-up, the scan shows complete SRF resolution and a decrease of SFCT to 341 µm and of CVI to 67.54%. E, F At the final follow-up, the scan exhibits sustained SRF reabsorption, stable SFCT of 349 µm and an increase of CVI to 68.62%.

Statistical analysis

Primary outcomes were possible changes of CVI, TCA, LA, SA, SFCT and CMT during follow-up, which were studied using Repeated-Measures Analysis of Variance (ANOVA). If significant, pairwise comparisons using paired t-tests were conducted. Normal distribution was determined through histograms. Assumption of sphericity was evaluated using Mauchly’s test of sphericity. If the assumption was not met, the model was corrected according to Greenhouse-Geisser (epsilon <0.75) or Huynh-Feldt (epsilon >0.75). Secondary outcomes were associations between all parameters and differences between affected vs. active vs. inactive FE, calculated using t-test and one-way ANOVA with subsequent pairwise comparison. Homogeneity of variance was checked using Levene’s test. All post-hoc pairwise comparisons were adjusted for multiplicity through Bonferroni correction. The statistical significance level was set at α = 0.05. If not stated otherwise, normally distributed data were presented as mean ± standard deviation). All statistical analyses and diagrams were carried out with SPSS version 29 (IBM Corp., New York, USA).

Results

A total number of 85 SS-OCTA scans from 45 visits with a follow-up of 12 months were analysed in this study. 15 PDT-treated eyes and 14 FE were included, 1 FE was excluded due to insufficient image quality. Three (21.43%) FE were classified as active, due to mild extrafoveal SRF not requiring treatment, and 11 (78.57%) FE as inactive. Nine (60%) eyes were from male participants. Mean patient age at baseline was 47.33 ± 13.43 (range, 30–77) years.

Subretinal fluid analysis

Resolution of SRF occurred in 14 out of 15 (93.3%) and of PED in 1 out of 3 (33.33%) treated eyes at final visit. SRF presence was significantly associated with higher mean CVI (69.13 ± 1.85 vs. 67.55 ± 1.52%, p < 0.001), higher mean LA (4.03 ± 0.67 vs. 3.68 ± 0.66 mm², p = 0.031), higher mean CMT (325.36 ± 106.81 vs. 199.25 ± 28.48 µm, p < 0.001) and higher mean SFCT (435.28 ± 84.48 vs. 388.68 ± 95.72 µm, p = 0.038), as compared to SRF absence.

Longitudinal changes in treated eyes

Mean CMT significantly decreased after treatment (p < 0.001). The decrease was significant for baseline vs. month 1 (p < 0.001) and baseline vs. month 12 (p < 0.001). From month 1 to 12, no significant change was recorded (p = 1.0). Table 1 reports detailed data comparing choroidal metrics before and after PDT in affected eyes.

Table 1.

Choroidal anatomic metrics collected from affected and fellow study eyes.

Parameter Mean ±  Standard deviation		Affected Eye (A)	Active FE (B)	Inactive FE (C)	One-Way ANOVA Post-hoc test	Repeated Measures ANOVA for (A) Post-hoc test
CVI (%)	BSL	69.91 ± 0.83	69.75 ± 0.62	66.91 ± 1.4	p < 0.001*** A vs B, p = 1.0 A vs C, p < 0.001*** B vs C, p = 0.002**	p < 0.001*** BSL vs MO1 p < 0.001*** BSL vs MO12 p = 0.002** MO1 vs MO12 p < 0.001***
	MO1	66.52 ± 1.45	69.59 ± 0.5	66.95 ± 1.59	p < 0.001*** A vs B, p = 0.008** A vs C, p = 1.0 B vs C, p = 0.002**
	MO12	69.02 ± 0.92	68.62 ± 0.62	67.0 ± 1.17	p < 0.001*** A vs B, p = 1.0 A vs C, p < 0.001*** B vs C, p = 0.026*
TCA (mm²)	BSL	5.90 ± 0.75	6.31 ± 0.96	5.26 ± 1.02	p = 0.112 A vs B, p = 1.0 A vs C, p = 0.251 B vs C, p = 0.240	p < 0.001*** BSL vs MO1 p < 0.001*** BSL vs MO12 p = 0.002** MO1 vs MO12 p = 0.264
	MO1	5.43 ± 0.78	6.22 ± 1.10	5.19 ± 1.05	p = 0.255 A vs B, p = 0.561 A vs C, p = 1.0 B vs C, p = 0.305
	MO12	5.60 ± 0.23	6.12 ± 0.51	5.21 ± 0.27	p = 0.216 A vs B, p = 0.917 A vs C, p = 0.797 B vs C, p = 0.298
LA (mm²)	BSL	4.13 ± 0.55	4.4 ± 0.7	3.53 ± 0.73	p < 0.001*** A vs B, p = 1.0 A vs C, p = 0.084 B vs C, p = 0.136	p < 0.001*** BSL vs MO1 p < 0.001*** BSL vs MO12 p < 0.001*** MO1 vs MO12 p = 0.01**
	MO1	3.62 ± 0.56	4.33 ± 0.79	3.49 ± 0.76	p = 0.168 A vs B, p = 0.301 A vs C, p = 1.0 B vs C, p = 0.190
	MO12	3.87 ± 0.56	4.24 ± 0.77	3.49 ± 0.66	p = 0.097 A vs B, p = 1.0 A vs C, p = 0.235 B vs C, p = 0.214
SA (mm²)	BSL	1.77 ± 0.21	1.9 ± 0.26	1.73 ± 0.3	p = 0.042* A vs B, p = 1.0 A vs C, p = 1.0 B vs C, p = 0.911	p = 0.125 BSL vs MO1 p = 1.0 BSL vs MO12 p = 0.32 MO1 vs MO12 p = 0.136
	MO1	1.8 ± 0.25	1.89 ± 0.31	1.71 ± 0.3	p = 0.542 A vs B, p = 1.0 A vs C, p = 1.0 B vs C, p = 0.975
	MO12	1.73 ± 0.23	1.94 ± 0.39	1.71 ± 0.3	p = 0.412 A vs B, p = 0.646 A vs C, p = 1.0 B vs C, p = 0.609
CMT (µm)	BSL	379.53 ± 100.45	234 ± 36.37	203 ± 28.58	p < 0.001*** a A vs B, p = 0.02* A vs C, p < 0.001*** B vs C, p = 1.0	p < 0.001*** BSL vs MO1 p < 0.001*** BSL vs MO12 p < 0.001*** MO1 vs MO12 p = 1.0
	MO1	200.67 ± 44.41	259 ± 72.55	200.4 ± 29.48	p = 0.166 A vs B, p = 0.124 A vs C, p = 1.0 B vs C, p = 0.145
	MO12	198.52 ± 24.68	218 ± 12.53	203.82 ± 31.37	p = 0.513 A vs B, p = 0.779 A vs C, p = 1.0 B vs C, p = 1.0
SFCT (µm)	BSL	453.33 ± 80.57	450 ± 26.46	360.5 ± 117.85	p = 0.06 A vs B, p = 1.0 A vs C, p = 0.066 B vs C, p = 0.472	p < 0.001*** BSL vs MO1 p < 0.001*** BSL vs MO12 p = 0.001*** MO1 vs MO12 p = 1.0
	MO1	407 ± 86.89	442 ± 40.71	368.1 ± 114.41	p = 0.428 A vs B, p = 1.0 A vs C, p = 0.98 B vs C, p = 0.749
	MO12	399.53 ± 74.94	443.33 ± 30.62	390.9 ± 89.1	p = 0.076 A vs B, p = 1.0 A vs C, p = 1.0 B vs C, p = 0.56

Comparison of choroidal anatomic metrics between eyes with chronic central serous chorioretinopathy (CSCR), treated with half-fluence photodynamic therapy (A), and their untreated active (B) and inactive (C) fellow eyes (column 6). Comparison of choroidal markers before and after treatment in eyes affected by CSCR (column 7).

*p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. P values are adjusted for multiplicity through Bonferroni correction for all post-hoc tests.

FE fellow eye, ANOVA analysis of variance, BSL baseline, MO1 month 1, MO12 month 12, vs versus, CVI choroidal vascularity index, TCA total choroidal area, LA luminal area, SA stromal area, CMT central macular thickness, SFCT subfoveal choroidal thickness.

^aHomogeneity of Variances was not met, instead Brown-Forsythe test was used.

Mean SFCT was significantly decreased after PDT (p < 0.001). Mean reductions were significant for baseline vs. month 1 (p < 0.001) and baseline vs. month 12 (p = 0.001). SFCT remained stable between month 1 and 12 (p = 1.0) (Table 1, Fig. 3 top).

Fig. 3. Comparison of longitudinal changes of subfoveal choroidal thickness (top) and choroidal vascularity index (bottom) in eyes compromised by chronic central serous chorioretinopathy before and up to one year after treatment with half-fluence photodynamic therapy.

Numbers next to boxplots indicate median values. **p ≤ 0.01, ***p ≤ 0.001. P values are adjusted for multiplicity through Bonferroni correction for all post-hoc tests.

Mean CVI significantly changed after treatment (p < 0.001). A significant decrease from baseline to month 1 (p < 0.001), followed by a significant increase from month 1 to 12 (p < 0.001) was recorded. At final visit, mean CVI still remained significantly lower than at baseline (p = 0.002) (Table 1, Fig. 3 bottom).

Mean TCA significantly changed after PDT (p < 0.001), with a decrease from baseline to month 1 (p < 0.001), followed by a nonsignificant increase from month 1 to 12 (p = 0.264). At final visit, mean TCA continued to be significantly lower than at baseline (p = 0.002) (Table 1, Fig. 4 top).

Fig. 4. Longitudinal changes of total choroidal, luminal and stromal area in eyes compromised by chronic central serous chorioretinopathy before and up to one year after treatment with half-fluence photodynamic therapy (top).

Comparison of mean choroidal vascularity index between eyes with chronic central serous chorioretinopathy, treated with half-fluence photodynamic therapy, and their untreated active and inactive fellow eyes before and after treatment (bottom). TCA total choroidal area, LA luminal area, SA stromal area, BSL baseline, MO1 month 1, MO12 month 12, PDT photodynamic therapy. *p ≤ 0.05, **p ≤ 0.01; ***p ≤ 0.001. P values are adjusted for multiplicity through Bonferroni correction for all post-hoc tests.

Mean LA was significantly affected by treatment (p < 0.001). LA shrank from baseline to month 1 (p < 0.001), followed by a significant increase from month 1 to 12 (p = 0.01). At final visit, mean LA remained significantly lower than at baseline (p < 0.001). Mean SA was not influenced by PDT (p = 0.125) (Table 1, Fig. 4 top).

Longitudinal changes in untreated fellow eyes

The FE control groups did not show any significant changes at any time point for CMT (p = 0.806 active, p = 0.59 inactive), SFCT (p = 1.0 active, p = 0.417 inactive), CVI (p = 0.261 active, p = 0.706 inactive), TCA (p = 0.73 active, p = 0.487 inactive), LA (p = 0.287 active, p = 0.524 inactive) and SA (p = 0.242 active, p = 0.417 inactive).

Comparison between affected and fellow eyes

When mean CVI was compared between the three groups, there were significant differences at baseline (p < 0.001), month 1 (p = 0.01) and month 12 (p < 0.001). At baseline, CVI was significantly higher in affected vs. inactive FE (p < 0.001) and active vs inactive FE (p = 0.002). After the substantial CVI reduction at month 1, previous difference between affected and inactive FE disappeared (p = 1.0). Treated eyes now exhibited a significantly lower mean CVI compared to active FE (p = 0.008). At final visit, mean CVI became significantly higher than inactive FE (p < 0.001). Detailed data are presented in Table 1 and Fig. 4 (bottom).

Contrarily, no significant differences were found for mean SFCT between affected and FE at any time point. At baseline, there was a trend of higher SFCT in affected vs. inactive FE (p = 0.066) but did not reach statistical significance (Table 1).

Mean CMT at baseline was significantly higher in affected vs. active FE (p = 0.02) and affected vs. inactive FE (p < 0.001). For TCA, LA, SA no differences were detected between all groups at any time point (Table 1).

Discussion

In this paper, we studied a set of eyes with cCSCR and their untreated FE over the course of 12 months. The main objective was to investigate choroidal tissue remodelling after treatment with half-fluence PDT.

Current studies most commonly use spectral domain (SD)-OCT for choroidal imaging, however it was shown that CVI is not affected by OCT machines [11]. In fact, there are even advantages to SS-OCT as it was reported to be the superior modality to measure SFCT [12], detect CSI [13] and have lower signal-to-noise ratio at greater scanning depths compared to SD-OCT [5], which proves ideal to visualize the choroid. Further, we analysed the single foveal scan as Agrawal et al. [10] demonstrated CVI to be highly reproducible regardless of using foveal, central or total macular scans, concluding that foveal scans are representative for total macular CVI. Regarding the region of CVI measurement, most studies have used only the central 1500 µm. However, subfoveal CVI seems to be higher than in the macular periphery [7], which is why we analysed the entire 12-mm foveal scan aiming to avoid any confounding effects and report representative values for the entire choroid.

Verteporfin PDT has been widely proven as an efficacious treatment for cCSCR. It targets the pathologic processes associated with choroidal hyperpermeability and induces vascular hypoperfusion. Mechanism of action is based on production of free radicals which induce endothelial cell destruction and lead to vascular occlusion, remodelling and constriction of congested vessels [3, 14]. Previously, complete SRF resolution rates between 71 [15] and 91% [16] were reported following half-fluence PDT. Our results showed SRF resolution in 93.3% of treated eyes at month 1 without recurrence at month 12, along with a significant decrease of CMT 1 and 12 months after PDT (both p < 0.001), supporting the efficacy of PDT.

In order to objectively measure choroidal activity, several imaging biomarkers have been proposed. Diameter of the largest intrachoroidal vessel is supposed to estimate vascular engorgement degree [6]. However, a single vessel may not adequately represent the entire vasculature. Traditionally, SFCT was considered the primary marker for choroidal health. Yet, SFCT was documented to significantly vary with age, axial length, intraocular pressure, systolic blood pressure, and to be prone to segmentation errors, resulting in poor reproducibility [6]. Furthermore, SFCT does not pinpoint exact anatomic components affected by a disease of interest. First introduced by Agrawal et al. [8], the novel OCT-marker CVI has gained substantial scientific interest recently. In contrast to SFCT, it is independent of above-mentioned patient factors, shows greater reproducibility and importantly, enables demarcation and quantification of individual choroidal components. In this study, we aimed to explore choroidal response to half-fluence PDT, with a focus on comparing SFCT and CVI.

We noted a decreased SFCT 1-month-post-PDT (p < 0.001), which has been previously described [3, 4, 17, 18]. However, our SFCT remained stable from month 1 to 12. This is in contrast to previous reports of SFCT rebounding to baseline values at month 6 [19] or further decreasing until month 6 [17].

Numerous groups examining CVI after PDT could not find any significant changes after treatment [3, 4, 18, 20]. Christou et al. concluded that CVI only changes if either LA or TCA varies more than the other, otherwise the ratio remains the same [4]. Van Rijssen et al. suggested that treatment effect relies on processes in the vessel walls and permeability rather than luminal density [20]. Our study, however, demonstrated that CVI is significantly affected by PDT. CVI was reduced at one and 12 months post-PDT (p < 0.001 and p = 0.002, respectively). This is supported by Sevik et al. [17], Park et al. [21] and Izumi et al. [22], who described a CVI reduction up until month 6. Notably, we observed CVI to increase again until final visit, but without returning to baseline levels. Importantly, all examined parameters did not change in the FE groups, suggesting a solely local treatment effect.

Interestingly, while SFCT remained stable until final visit, there was a simultaneous increase in CVI. This indicates CVI’s capability to detect certain changes in choroidal architecture that SFCT lacks. CVI may even detect subclinical activity, which may be missed by SFCT. We consider this discrepancy between CVI and SFCT highly relevant new information, since they currently represent the two main biomarkers for choroidal architecture. Consequently, they should not be treated as interchangeable. These results suggest that they each represent different anatomic responses. SFCT may express the reduction of TCA and absorption of interstitial fluid rather than vessel decongestion [4]. This concept seems plausible considering that in our set of eyes, TCA also shrank significantly at month 1 and final visit (p < 0.001 and p = 0.002, respectively) and remained steady after month 1 (p = 0.264). On the contrary, CVI increase reflected an increase in LA. The substantial changes in LA also confirm the direct remodelling effect of PDT on choroidal vessels as the main treatment targets [4]. As LA seems to be the main driver for CVI, this raises the question whether simply using LA rather than CVI is merited. LA provides a direct measurement of vasculature and hence is useful for investigating and understanding involvement of vascular structures in disease pathology. However, LA by itself does not reveal the proportion of the choroid that is made up of vascular components. Further, interpretation of LA is challenging in situations, in which total choroidal area is increased or reduced, such as with inflammation or high myopia [23]. By comparison, CVI provides a contextual understanding of the overall choroidal structure and thus offers a more precise assessment of choroidal health.

In conclusion, we found that CVI and SFCT both represent different morphologic components of the choroid and provide relevant information when used in conjunction rather than substitutively for one another.

In healthy eyes, CVI was reported to vary from 49 [7] to 65% [8] and shown to be higher in eyes with acute CSCR, ranging from 53 [7] to 71% [8]. Hence CVI was suggested as a diagnostic marker. We reported a mean CVI of 69.91 ± 0.83% in cCSCR before treatment. However, it is uncertain if CVI also differs between affected eyes and their FE. Yang et al. examined three-dimensional CVI and found no difference between affected and FE [24]. Sahoo et al. mapped CVI according to the Early Treatment of Diabetic Retinopathy grid and failed to reveal differences in any zones when comparing affected eyes and FE [25]. Iovino et al. conducted a multicentre study with 64 eyes and also did not detect any difference [3]. These findings made CVI’s role as an activity criterion questionable. Our analysis, however, showed CVI of affected eyes to be significantly higher than of inactive FE (p < 0.001) before treatment. This may be explained by our distinction between inactive and active FE, with the latter exhibiting extrafoveal SRF not requiring treatment. Other studies may have not revealed such difference due to inclusion of active FE, since it is known that eyes with SRF show higher CVI than those with resolved SRF or inactive FE [8]. Therefore, we demonstrated that CVI qualifies not only as a diagnostic but also as a disease activity marker.

A big concern following standard PDT protocol was choroidal ischemia. In an attempt to minimize complications, safety-enhanced protocols using half-fluence and/or half-dose PDT were implemented [18]. The CVI and LA increase after month 1 in our series suggests a reperfusion of choroidal vessels. This may also be interpreted as a sign of disease recurrence, however the lack of return to baseline levels at final visit and more importantly, the absence of SRF recurrence, leads us to support the theory of choroidal reperfusion. Nonetheless, studies with even longer follow-up are needed to elucidate this topic. In summary, these findings indicate not only a successful but also a safe treatment effect by inducing choroidal hypoperfusion in the short-term without causing ischemia in the long-term, thus preserving choroidal health.

The limitations of this study include the single-centre nature and small sample size. Nonetheless, even in this set of eyes we were able to detect highly significant associations and differences in the investigated parameters. However, studies with larger samples and CVI assessment of the entire choroid are undoubtedly warranted to validate these findings. The main strengths of this study include the prospectively collected data along with a standardised imaging protocol, using SS-OCT which allows for excellent visualization of choroidal components, and the rigorous standardised image analysis using validated binarization procedure. The FE-controlled design and the further distinction between active and inactive FE allowed for elimination of interindividual variability and confounding effects. Importantly, our patients were followed over the course of 12 months. The limited number of studies on this subject have mostly compared CSCR eyes only before and after PDT in a single follow-up visit, rather than longitudinally. To the best of our knowledge, this study represents the longest follow-up of cCSCR eyes after PDT to date, exhibiting natural disease course after treatment intervention.

In conclusion, we found PDT to significantly impact choroidal architecture, with sustained CVI and SFCT reductions one year after treatment. Further, we demonstrated CVI’s value as both a treatment response and disease activity marker. Importantly, we detected a large discrepancy between SFCT and CVI, arguing for CVI to be a more specific and reliable biomarker. By reporting an increase in choroidal perfusion in the long-term without disease recurrence, we showed preservation of choroidal health. Finally, our findings add to the growing recognition of CVI as an important new biomarker and deepen our understanding of the complex multifactorial nature of cCSCR.

Summary

What was known before:

• Subfoveal choroidal thickness and choroidal vascularity index represent the two main biomarkers of the choroid in eyes with chronic central serous chorioretinopathy.

• Half-fluence photodynamic therapy significantly affects subfoveal choroidal thickness.

What this study adds:

• Half-fluence photodynamic therapy significantly affects choroidal vascularity index, both in the short- and longterm.

• There is a large discrepancy between choroidal vascularity index and previously established subfoveal choroidal thickness.

• Use of choroidal vascularity index as both a treatment response and disease activity marker is merited.

Author contributions

SAS and PTAB designed the research; PTAB, MS and DAB acquired and analysed the data; SAS, KK, MS and SK contributed to critical revision of the article; SAS provided resources and supervised the project; PTAB wrote the article.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.