Introduction

Pachychoroid spectrum disorders (PSDs) represent a phenotype of age-related macular degeneration (AMD), the leading cause of irreversible vision loss in industrialized nations1,2,3,4,5,6,7. PSD responds differently to anti-vascular endothelial growth factor therapies and photodynamic therapy than does conventional AMD, suggesting its distinct pathogenesis5,7,8,9. Elucidating PSD pathogenesis is critical for facilitating the development of personalized management strategies for AMD.

A pathogenetic hypothesis termed “venous overload choroidopathy” has been proposed to explain one potential cause of PSD10,11,12,13,14. This hypothesis suggests that scleral thickening increases outflow resistance in the vortex veins, consequently inducing stasis in the choroidal venous system. Despite advances in widefield choroidal imaging, current techniques have limitations. Ultrawidefield indocyanine green angiography lacks three-dimensional visualization capabilities15,16,17, whereas widefield optical coherence tomography (OCT) can only visualize the vicinity of the vortex vein ampullae (VVA)13,14,18. These technological constraints have hindered the development of a standardized method for comprehensive VVA structure assessment. Consequently, the morphological characteristics of the VVA in eyes with PSD remain largely unknown, despite their crucial role in PSD pathogenesis19,20,21,22,23,24. Optos Silverstone (Optos, UK) has recently emerged as an innovative imaging system that integrates ultrawidefield scanning laser fundus imaging (200° field of view)-guided swept-source OCT25. The expanded field of view of this device enables OCT imaging of the fundus periphery, where the VVA are located, from a frontal perspective. Such advancements have revealed potential pathogenic mechanisms in various retinal and choroidal diseases18,19,20.

This study aims to elucidate the VVA characteristics in PSD by developing an evaluation method for the three-dimensional structure of VVA using Optos Silverstone.

Results

Patient characteristics

We included 40 patients (46 eyes) with PSD (CSC, 32 eyes [69.6%]; PPE, 9 eyes [19.6%]; PNV, 4 eyes [8.7%]; PCV, 1 eye [2.2%]) and 14 healthy controls (15 eyes). Table 1 shows the detailed patient characteristics. Compared with the healthy group, the PSD group demonstrated significantly greater CCT (P = 0.015) and a lower proportion of symmetrical Haller vessel running patterns (P = 0.010). In the PSD group vs. the healthy group, the patterns were distributed as follows: upper dominant (50.0% vs. 20.0%), lower dominant (21.7% vs. 6.7%), and symmetrical (28.3% vs. 73.3%).

Table 1 Comparisons of patient characteristics between PSD and healthy participants.
Full size table

Comparisons of VVA morphology between PSD eyes and healthy eyes

We observed no significant differences in the supratemporal VVA morphological parameters between PSD eyes and healthy eyes (P ≥ 0.487 for all comparisons, Table 2 and Fig. S1). For infratemporal VVA, eyes with PSD presented a significantly greater mean luminal volume (PSD vs. healthy, 6.9 × 106 ± 1.8 × 106 pixels vs. 4.7 × 106 ± 1.5 × 106 pixels, P < 0.001). The mean vessel diameter and total vessel length of Infratemporal VVA in the PSD were significantly greater in the PSD group than in the control group (P ≤ 0.021 for both comparisons). In contrast, no significant difference in the vessel luminal proportion in the infratemporal VVA region was detected between the groups (P = 0.353).

Table 2 Comparisons of vortex vein ampullae morphological parameters between PSD and healthy participants.
Full size table

After adjusting for potential confounders, the PSD group demonstrated significantly greater infratemporal VVA luminal volume and vessel diameter than the healthy group did (P ≤ 0.011 for both comparisons, Table S1). However, no clear difference was observed in the total vessel length (P = 0.220). Comparison of VVA morphology among PSD subtypes (excluding PCV) revealed no significant differences (Table S2).

Comparisons of VVA morphology among running patterns of the Haller vessel for eyes with PSD

The upper dominant group exhibited the largest supratemporal VVA volume (P = 0.040, Table 3), whereas the lower dominant group presented the largest infratemporal VVA volume (P = 0.111). Compared with the lower dominant group, the upper dominant group presented significantly greater vertical differences in luminal volume (Steel‒Dwass test: upper dominant vs. lower dominant, P = 0.006; symmetry vs. upper dominant, P = 0.476; symmetry vs. lower dominant, P = 0.139).

Table 3 Comparisons of vortex vein ampullae morphological parameters among the running patterns of Haller’s vessel for eyes with PSD.
Full size table

The vertical differences in the luminal proportion and mean vessel diameter demonstrated a similar trend as that for the luminal volume (Steel‒Dwass test: luminal proportion, upper dominant vs. lower dominant, P = 0.003; mean vessel diameter, upper dominant vs. lower dominant, P = 0.001). Figure 1 illustrates a representative case with VVA evaluations. Similar to the PSD group, healthy subjects with upper dominance showed greater supratemporal VVA volume (Table S3).

Fig. 1
figure 1

Vortex vein ampullae images of a patient with the upper-dominant Haller vessel pattern. (A) This ICGA image shows a 55-year-old male patient with CSC. The axial length was 24.3 mm, and the Haller vessel running pattern was the upper-dominant type. (B,C) Enlarged views of the supratemporal and infratemporal VVA in the ICGA image. (D,E) Three-dimensional structures of the VVA acquired via ultrawidefield scanning laser fundus image-guided swept-source OCT. The luminal volume was 7,578,810 pixels for the supratemporal VVA and 4,705,821 pixels for the infratemporal VVA. CSC central serous chorioretinopathy, ICGA indocyanine green angiography, VVA vortex vein ampulla, OCT optical coherence tomography.

Full size image

Associations between VVA morphology and patient characteristics

The infratemporal VVA luminal volume was significantly positively associated with CCT (β = 0.7 × 10⁴, P = 0.004) and negatively associated with age (β = -6.1 × 10⁴, P = 0.005) but was not significantly associated with axial length or sex (P ≥ 0.101 for both analyses, Fig. 2 and Table S4). The infratemporal VVA mean vessel diameter was significantly positively associated with CCT (P = 0.011) and negatively associated with axial length (P = 0.014) but was not significantly associated with age or sex (P ≥ 0.698 for both analyses). No patient characteristics were significantly associated with the supratemporal VVA luminal volume or mean vessel diameter (P ≥ 0.099 for all analyses). Sex was not significantly associated with any of the VVA morphological parameters (P ≥ 0.088 for all analyses).

Fig. 2
figure 2

Associations between the luminal volume of the VVA and patient characteristics. These scatter plots with regression lines and 95% confidence intervals illustrate the relationships between VVA parameters and patient characteristics. The luminal volume of the infratemporal VVA was positively associated with central choroidal thickness (β = 0.7 × 104, P = 0.004) and negatively associated with age (β = -6.1 × 104, P = 0.005). The associations between the luminal volume of supratemporal VVA and patient characteristics were not statistically significant (P ≤ 0.099). VVA vortex vein ampulla, S supratemporal, I infratemporal.

Full size image

Discussion

This study quantitatively assessed VVA morphology in PSD patients via a developed evaluation technique. Compared with healthy eyes, PSD eyes presented significantly greater vessel luminal volume and mean vessel diameter and no significant differences in supratemporal VVA morphology. Moreover, eyes with PSD presented an enlarged VVA lumina on the dominant side of the macular Haller vessels.

Advances in OCT technology, including faster scan speeds and enhanced depth imaging, have revealed characteristic features of PSD eyes: choroidal thickening, increased choroidal lumen fraction, enlarged vascular diameter, fluid loculation, and asymmetry of the macular Haller’s vessels in the posterior pole3,26,27,28,29. Furthermore, most studies using widefield OCT have demonstrated choroidal thickening and an enlarged vascular diameter extending to the midperiphery in PSD eyes13,14,30. However, Maruko et al., utilizing OCT with the widest field of view, reported minimal differences in peripheral choroidal thickness between PSD eyes and normal eyes31. These findings suggest that venous choroidal overload and interstitial fluid retention extend from the macula toward the VVA, potentially diminishing in the peripheral VVA region13,14,30,31. In eyes with PSD, our study revealed an enlargement of both the lumen and stroma only in the inferior VVA, whereas the superior VVA remained comparable to that of control eyes. If a scleral thickening-induced increase in choroidal outflow resistance could fully explain PSD changes10,11,26, entire choroidal expansion from the VVA to the macula may potentially be observed. We cannot exclude the possibility that scleral properties limited changes in VVA morphology. However, since the sclera blocks OCT laser penetration, we think that intrascleral choroidal vessels are largely excluded from our analysis. Therefore, these findings indicate that impaired choroidal blood drainage alone cannot fully account for the pathogenesis of PSD.

Saito et al. proposed choroidal hyperperfusion as a key feature in the active phase of acute CSC32. Short posterior ciliary arteries, the choroid’s primary inflow vessels, are distributed in the posterior pole and are concentrated around the macula33,34. Choroidal hyperperfusion increases macular choroidal blood flow, elevates venous pressure and causes vascular dilation, while it may promote interstitial fluid accumulation. The selective enlargement of the lumen and interstitium in the inferior VVA region, which is absent in the superior VVA, might indicate gravitational accumulation of excessive macular fluid.

Choroidal morphology and perfusion in PSD are reportedly regionally imbalanced28,29,35. Using widefield OCT, Ishikura et al. reported that eyes with asymmetrically enlarged macular choroidal vessels also exhibited choroidal thickening near the VVA on the dominant side13. Our study demonstrated VVA enlargement on the dominant side, indicating that asymmetric vascular dilation extends beyond the macula to include the VVA. However, the etiology of this choroidal asymmetry remains elusive.

Takahashi et al. demonstrated an imbalance in perfusion zones and vortex vein systems via postscleral buckling ICGA36. Additionally, Matsumoto et al. reported predominant asymmetric choroidal vessel expansion on the ligated side in a monkey model following vortex vein occlusion37. These findings indicate that choroidal venous outflow obstruction may induce asymmetry in the choroidal vasculature. However, several observations challenge the hypothesis that choroidal asymmetry results from acquired vascular pressure loading: (1) the resolution of asymmetric vessels approximately one month postvortex vein ligation in monkey models;37 (2) a lack of an association between scleral thickness and choroidal vascular asymmetry;38 (3) the presence of choroidal asymmetry in 40–50% of normal eyes;14,29,39,40 and (4) the absence of correlation between age and choroidal asymmetry41. Considering the absence of upper VVA dilation in our study, the effect of impaired choroidal drainage on macular choroidal asymmetry may be limited.

However, while Matsumoto et al. suggest that choroidal venous anastomoses may indicate venous outflow obstruction37, these findings are also observed in normal eyes40. In addition, there is the problem that the method of evaluating choroidal venous anastomoses is subjective. Quantitative assessment of choroidal venous anastomoses, rather than their mere presence, may help clarify their role in pachychoroid-spectrum disorders in future study.

There are several limitations to this study. First, because of its retrospective nature, future longitudinal studies are needed. Second, since the sample size of this study is relatively small, comparisons among PSD subtypes by larger surveys are warranted. Third, the lack of direct measurements of choroidal blood flow restricts the ability to elucidate the relationship between VVA morphological changes and hemodynamics. However, owing to the current lack of instruments capable of measuring peripheral choroidal blood flow, morphological evaluation serves as a necessary surrogate. To mitigate potential diurnal variations in macular choroidal thickness42, which may extend to VVA, examinations were confined to 8:30 am–12:30 pm. The pixel‒to-micrometer ratio for each image is undetermined, potentially complicating comparisons between the quantitative values in this study and those derived from alternative methods. Further research is needed to elucidate the relationship between choroidal venous anastomoses and venous outflow obstruction.

In conclusion, enlargement of both the lumen and stroma was observed exclusively in the infratemporal VVA in eyes with PSD. Furthermore, asymmetry in macular choroidal morphology was observed to extend continuously to the VVA. While VVA morphology is implicated in PSD pathogenesis, impaired choroidal drainage alone may not fully account for the observed VVA alterations.

Methods

This was a single-center case‒control study. This study was conducted as a retrospective observational study. Although informed consent was waived, an opt-out opportunity was provided to patients through the hospital bulletin board and website (The Ethics Committee of Kagoshima University, Kagoshima, Japan). The Ethics Committee approved this study (No. 16012), and all procedures were conducted in accordance with the tenets of the Declaration of Helsinki.

Patients

Consecutive patients with PSD who visited Kagoshima University Hospital between January 2022 and December 2023 and underwent VVA evaluation by Optos Silverstone were included. We also included a healthy group consisting of participants who underwent evaluation of the VVA in the same month as the PSD patients. These control patients were diagnosed with only cataracts and had no intraocular, optic nerve, infectious, or neoplastic diseases. On the basis of previous reports, PSD was defined by the following criteria: (1) the presence of dilated Haller vessels on OCT or indocyanine green angiography (ICGA), (2) choroidal vascular hyperpermeability on ICGA, and (3) the absence of soft drusen in the macular region3,6,20,21. Among eyes fulfilling these PSD characteristics, we classified eyes as follows: pachychoroid neovasculopathy (PNV) for those with macular neovascularization (MNV) without polypoid lesions; pachychoroid polypoidal choroidal vasculopathy (PCV) when MNV was accompanied by polypoid lesions; central serous chorioretinopathy (CSC) in cases of serous retinal detachment without MNV; and pachychoroid pigment epitheliopathy (PPE) when only pigment epithelial detachment (PED) was present4,20,43,44,45. Diagnoses were made by two retina specialists (RF, NM) in consultation, with any disagreements resolved by a senior supervisor (HT).

Data collection

We conducted comprehensive ophthalmic examinations and multimodal imaging at the initial visit. Examinations included best-corrected visual acuity (BCVA) measurement (decimal visual acuity), intraocular pressure assessment, refraction (RM8900, Topcon, Japan), axial length measurement (OA-2000, Tomey, Japan), and slit-lamp biomicroscopy. Multimodal imaging comprised spectral-domain optical coherence tomography (SD-OCT; Spectralis, Heidelberg Engineering, Germany), optical coherence tomography angiography (OCTA; PLEX Elite 9000, Carl Zeiss Meditec, Germany), fluorescein angiography (FA) and indocyanine green angiography (ICGA; Spectralis HRA and Optos California), color fundus photography (DRI OCT Triton, Topcon, Japan), and fundus autofluorescence (FAF; Spectralis HRA, Heidelberg Engineering). Furthermore, we performed VVA evaluations via ultrawidefield scanning laser fundus imaging-guided swept-source OCT (Optos Silverstone).

VVA evaluation

This study focused on evaluating the superior and inferior VVA located in the temporal quadrants of the fundus. The VVA evaluations were conducted as follows using Optos Silverstone (Fig. 3): (1) Guided by ultrawidefield scanning laser ophthalmoscope images, we performed 3–5 OCT volume scans (6 mm × 6 mm) at strategic locations encompassing and surrounding each VVA (Fig. S2). (2) We acquired a series of choroidal en-face OCT images, each representing a single-pixel depth from the retinal pigment epithelium (RPE) to the sclera (Supplemental video). The RPE was used as the reference boundary, and sequential en-face images were obtained at one-pixel intervals throughout the choroidal depth. To determine the choroidoscleral interface, we analyzed luminance changes along the depth axis at each measurement point. The location of steep luminance increase was identified for every point, and the deepest of these locations was designated as the overall depth. (3) We extracted luminal structures from choroidal en-face images derived from multiple OCT volume scans (Fig. S3). (4) The extracted luminal structures from multiple scans were spatially aligned using the position information recorded during image acquisition. (5) To reduce any misalignments, we implemented an automated algorithm that montaging at maximized the overlap of luminal structures across multiple scans. To mitigate the overestimation of blood vessel volume, we implemented superpixel segmentation using the simple linear iterative clustering (SLIC) method (Fig. 3)46. This process was initially applied to the layer with the largest blood vessel area and subsequently extended to all layers. For each corresponding set of superpixels across layers, we selected the one with the largest volume. (6) For vessel skeletonization, we automatically extracted the centerlines of the planar lumina. We subsequently manually corrected any skeletons that deviated significantly from the apparent lumen orientation. We then implemented an automated algorithm to adjust the centerline of the lumen structure along the depth axis. This process estimates the true center of the lumen throughout its depth, allowing us to obtain accurate skeletal information on the vessel structure. (7) On the basis of previously reported methods, we identified the center of each vortex vein ampulla (VVA)20. We then analyzed a defined region surrounding this central point to calculate and quantify VVA morphological parameters.

Fig. 3
figure 3

Process of generating a composite image of the vortex vein ampulla. Multiple en-face optical coherence tomography images acquired in the region of the vortex vein ampullae are aligned on the basis of their positional information. We modified the images to maximize the overlapping areas of the luminal structure and used the superpixel segmentation method to correct for image misalignment.

Full size image

We expanded the analysis range, starting from an initial diameter of 100 pixels and incrementally increasing it by 100 pixels up to a maximum diameter of 600 pixels. For the interstitial analysis, we focused solely on the areas immediately adjacent to the vessel lumina. We analyzed the following morphological parameters: luminal volume, luminal proportion, total vessel length, and mean vessel diameter.

Statistical analysis

In this study, VVA evaluations were performed twice at different time points for 19 patients and 20 eyes (28 locations total). The reproducibility between these evaluations was assessed using intraclass correlation coefficient (ICC) with a two-way random effects model. A single examiner performed all the examinations, while a single analyst conducted all the analyses. We selected a 400-pixel analysis range for this study based on ICC values. This range demonstrated an ICC > 0.800 for all the items, with the highest ICC for two of the four outcomes and the second highest for the remaining two (Table S5). Participant characteristics and VVA morphological parameters were compared via the Mann‒Whitney U test for continuous variables and Fisher’s exact test for categorical variables. To compare the VVA morphological parameters across Haller’s vessel running patterns, we employed the Kruskal‒Wallis test, followed by post hoc Steel‒Dwass tests for statistically significant items, adjusting for multiple comparisons.

Associations between the central choroidal thickness (CCT), its related factors (age and axial length)47,48, and VVA morphological parameters were examined using linear regression analysis. Moreover, despite unknown potential confounders influencing VVA morphology, we compared the PSD and control groups via multiple linear regression, adjusting for factors associated with CCT: age, axial length, sex, and Haller vessel running patterns14,39,47,48. P value = 0.05 was set as the statistical cutoff, and we performed all analyses using R software (version 4.3.2).

Meeting presentation

This material was presented at the 39th Annual Meeting of the Japanese Society for Ocular Circulation, Nara, Japan, in 2023.