Quantitative Analysis of Choroidal Thickness and Blood Flow in Thyroid-Associated Ophthalmopathy Using Ultra-Widefield Swept-Source OCT Angiography

Abstract

Objective

To quantitatively analyze choroidal thickness and blood flow in thyroid-associated ophthalmopathy (TAO) using ultra-widefield swept-source OCT angiography (UWF SS-OCTA), and to characterize differences among dysthyroid optic neuropathy (DON), non-DON, and normal eyes.

Design

Retrospective cross-sectional study.

Subjects

The study included 65 eyes from 35 TAO patients (15 males, 20 females; mean age 51.8 ± 12.6 years), comprising 5 mild, 40 moderate-to-severe, and 20 DON eyes. A control group of 70 eyes from 38 age- and sex-matched healthy subjects (15 males, 23 females; mean age 52.8 ± 14.9 years) was also included.

Methods

Ultra-widefield SS-OCTA imaging was performed to capture data sets divided into 9 subfields. Automated segmentation was used to measure structural thickness (whole choroid, large-vessel choroidal layer) and vascular density (choriocapillaris, large-vessel choroidal layer). Parameters were compared between TAO and control eyes, and subgroup analyses were conducted between DON and non-DON eyes.

Main Outcome Measures

Choroidal thickness and vascular density in the whole choroid, choriocapillaris, and large-vessel choroidal layer across 9 subfields.

Results

Thyroid-associated ophthalmopathy eyes showed significantly greater whole choroidal thickness (189.9 ± 2.6 μm vs. 182.4 ± 2.3 μm, P < 0.001) and large-vessel choroidal layer thickness (161.4 ± 2.7 μm vs. 149.9 ± 2.3 μm, P < 0.0001) compared with controls, particularly in central and nasal regions. Vascular density was reduced in TAO eyes for the large-vessel choroidal layer (39.86 ± 0.25% vs. 40.54 ± 0.15%, P = 0.014) and showed a nonsignificant decrease in the choriocapillaris (47.57 ± 3.24% vs. 47.89 ± 0.12%, P = 0.067). Subgroup analysis revealed that these differences were primarily driven by DON eyes, whereas non-DON TAO eyes showed no significant changes compared with controls.

Conclusions

Ultra-widefield SS-OCTA effectively identifies choroidal changes in TAO, revealing increased thickness accompanied by reduced vascular density. These findings suggest that choroidal thickening is primarily due to stromal edema rather than vascular engorgement, particularly in DON eyes, highlighting the role of choroidal remodeling in TAO progression.

Financial Disclosure(s)

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Thyroid-associated ophthalmopathy (TAO), also known as thyroid eye disease, is the most prevalent adult orbital disease and the most common extra-thyroidal manifestation of dysthyroid disease.¹ Thyroid-associated ophthalmopathy primarily affects periorbital tissues, with effects including extraocular muscle enlargement, orbital fat hyperplasia, and eyelid and conjunctive congestion, which can cause proptosis, eyelid retraction, diplopia, and optic neuropathy.²

Thyroid-associated ophthalmopathy is also associated with significant vascular abnormalities including disrupted arterial perfusion, congested vein drainage, and impaired blood flow in ocular tissues.^3,4 Previous studies have employed Doppler velocimetry, laser speckle flowgraphy, and OCT angiography (OCTA) to assess orbital and retinal vasculature.5, 6, 7, 8 These studies reported variable findings, such as increased vascular resistance, reduced retinal perfusion, and choroidal thickening, although choroidal vascular alterations and the relationship with structural thickness remain poorly characterized.

Recent advances in ultra-widefield swept-source OCTA (UWF SS-OCTA) have introduced a noninvasive, high-resolution method for evaluating choroidal vascular alterations. With its 6-mm scan depth, UWF SS-OCTA enables detailed visualization of choroidal blood flow, offering particular advantages for assessing choroidal thickening pathologies.^9,10 Although this technology has been extensively utilized in diabetic retinopathy, retinal vein occlusion, and central serous chorioretinopathy,11, 12, 13 its application in TAO has been relatively limited.

This study aims to systematically evaluate choroidal morphological and blood flow alterations in TAO using UWF SS-OCTA. Through characterization of these vascular changes, the results may provide insights into the pathophysiology of TAO, particularly the choroidal hemodynamic changes underlying TAO progression.

Methods

Subjects

This cross-sectional study retrospectively enrolled patients diagnosed with TAO at Shanghai Tenth People’s Hospital between December 2022 and December 2024. Thyroid-associated ophthalmopathy was diagnosed according to the 2021 European Group on Graves' Orbitopathy clinical practice guidelines.¹⁴ Disease activity was assessed using the Clinical Activity Score (CAS), with patients classified as inactive (CAS <3) or active (CAS ≥3).

Subjects with the history of refractive error higher than ±3 diopters, dense media opacity, uveitis, retinal disease, ocular trauma or tumor, and other orbital diseases were excluded from this study. If the image quality was low (signal strength index <80), these images were also excluded. All participants underwent comprehensive ophthalmic evaluations, including best-corrected visual acuity, intraocular pressure, slit lamp examination, and UWF SS-OCTA. Demographic and clinical data, including gender, age, symptom duration, and the history of thyroid disorders were systematically recorded.

This study was conducted in accordance with the principles of the Declaration of Helsinki, which was approved and followed by the Institutional Review Board of Shanghai Tenth People's Hospital. As a retrospective study, the requirement for informed patient consent was waived by the ethics committee.

Choroidal parameters by UWF SS-OCTA

Choroidal structures were examined using UWF SS-OCTA (BM-400K BMizar, TowardPi Medical Technology). The system combines a long-wavelength (1060 nm) swept-source laser with a 400 kHz A-scan rate, acquiring three-dimensional volumetric data spanning 20 mm (vertical) × 24 mm (horizontal) × 6 mm (depth), with 1536 A-scans per B-scan at 1280 positions. Both Bruch’s membrane and choroid–sclera interface were automatically segmented using the built-in software.

Choroidal thickness was measured as the perpendicular distance between Bruch's membrane and the choroid–scleral interface. The vessel density was quantified automatically as the ratio of vessel-occupied pixels to the total region area. The choriocapillary layer was defined as the 29-μm thick region extending outward from Bruch's membrane. The large-vessel choroidal layer was defined as the region extending outward from the choriocapillary layer to the choroid–scleral interface.

Statistical analysis

All statistical analyses were performed using SPSS version 24.0 (IBM Corp). Descriptions of numerical data were expressed as mean ± standard error. Differences between TAO and control eyes were compared using the independent 2-sample t test or Mann–Whitney U test. P value <0.05 was considered statistically significant.

Results

This study included 35 patients (65 eyes) with TAO, 15 (43%) male and 20 (57%) female, aged between 26 and 69 years (mean age 51.8 ± 12.6 years). Additionally, 38 age-matched (mean age: 52.8 ± 14.9 years, ranging from 24 to 70 years) and sex-matched (15 males, 23 females) healthy controls were included.

In TAO group, 5 eyes were excluded due to media opacities or restrictive strabismus, resulting in 65 eyes included for final analysis. Hertel exophthalmometry measurements ranged from 16 to 29 mm (mean: 22.2 ± 3.4 mm), and the mean intraocular pressure was 22.2 ± 3.5 mmHg (range: 15–26 mmHg).

Based on disease severity, 5 eyes (7.9%) were classified as mild TAO, 40 eyes (63.5%) as moderate TAO, and 20 eyes (28.6%) as severe TAO combined with dysthyroid optic neuropathy (DON). According to CAS value, 5 eyes (3 mild and 2 moderate TAO eyes) were classified as inactive (CAS <3), and the remaining 60 eyes (2 mild, 38 moderate, and 20 DON eyes) were considered active (CAS ≥3).

Figure 1 showed a severe TAO eye, that UWF SS-OCTA enabled comprehensive visualization of choroidal parameters, demonstrating detailed maps of overall choroidal thickness (Fig 1A), choriocapillary density (Fig 1B), large-vessel choroidal layer thickness (Fig 1C), and large-vessel choroidal layer density (Fig 1D). Warmer hues (red/orange) in the images indicated increased choroidal thickness (Fig 1A) and large-vessel choroidal layer thickening (Fig 1C).

Figure 1.

Representative maps of choroidal thickness. A, choriocapillaris density; B, large-vessel choroidal layer thickness; C, and large-vessel choroidal layer density; D in a TAO eye. TAO = thyroid-associated ophthalmopathy.

TAO eyes exhibited significantly greater mean choroidal thickness compared with controls (189.9 ± 2.6 μm vs. 182.4 ± 2.3 μm, P < 0.001). This increase was most pronounced in the central and nasal regions, with significant differences in the following subfields: upper (210.7 ± 7.6 μm vs. 201.0 ± 5.4 μm, P = 0.0016), central (226.0 ± 8.9 μm vs. 215.2 ± 7.8 μm, P = 0.0402), lower (164.9 ± 7.2 μm vs. 157.2 ± 5.1 μm, P = 0.0018), nasal (188.6 ± 8.9 μm vs. 177.7 ± 6.6 μm, P = 0.0052), and inferonasal (135.6 ± 6.2 μm vs. 124.6 ± 3.7 μm ± 6.8, P < 0.0001). No significant differences were observed in the superotemporal, temporal, inferotemporal, or superonasal subfields (Table 1).

Table 1.

Comparisons of Choroid Parameters between TAO and Control Eyes

Whole Choroidal Thickness (μm), Mean ± SE
	TAO Eyes (%)	Control Eyes (%)	P Value
Superotemporal	211.7 ± 6.67	206.7 ± 6. 27	0.2757
Temporal	195.6 ± 5.91	188.0 ± 5.35	0.1865
Inferotemporal	183.9 ± 6.45	174.8 ± 5.45	0.065
Upper	210.7 ± 7.57	201.0 ± 5.35	0.0016
Central	226.0 ± 5.98	215.2 ± 6.81	0.0402
Lower	164.9 ± 7.15	157.2 ± 5.11	0.0018
Superonasal	196.1 ± 7.42	191.7 ± 6.35	0.0917
Nasal	188.6 ± 8.87	177.7 ± 6.62	0.0052
Inferonasal	135.6 ± 6.18	124.6 ± 3.65	<0.0001
Average	189.9 ± 2.64	182.4 ± 2.29	<0.0001
Choriocapillary Density (%), Mean ± SE
	TAO Eyes (%)	Control Eyes (%)	P Value
Superotemporal	46.85 ± 0.48	47.92 ± 0.37	0.0128
Temporal	46.72 ± 0.21	46.44 ± 0.26	0.1395
Inferotemporal	46.07 ± 0.47	45.75 ± 0.49	0.1727
Upper	48.38 ± 0.19	48.58 ± 0.21	0.5426
Central	48.09 ± 0.19	48.00 ± 0.23	0.5297
Lower	49.00 ± 0.24	49.08 ± 0.21	0.0924
Superonasal	48.53 ± 0.49	48.93 ± 0.43	0.1303
Nasal	47.85 ± 0.27	48.10 ± 0.29	0.2358
Inferonasal	48.13 ± 0.45	48.88 ± 0.13	0.032
Average	47.57 ± 3.24	47.89 ± 0.12	0.0575
Large-Vessel Choroidal Layer Density (%), Mean ± SE
	TAO Eyes (%)	Control Eyes (%)	P Value
Superotemporal	42.00 ± 0.45	42.07 ± 0.41	0.2378
Temporal	38.82 ± 0.44	40.00 ± 0.33	0.0336
Inferotemporal	40.47 ± 0.48	40.24 ± 0.41	0.1128
Upper	41.96 ± 0.39	41.95 ± 0.33	0.3441
Central	39.37 ± 0.41	37.92 ± 0.38	0.0101
Lower	41.15 ± 0.55	41.92 ± 0.37	0.1839
Superonasal	42.47 ± 0.55	42.12 ± 0.42	0.3029
Nasal	40.01 ± 0.73	40.46 ± 0.45	0.1237
Inferonasal	36.44 ± 0.79	38.66 ± 0.53	0.0041
Average	39.86 ± 0.25	40.54 ± 0.15	0.0142
Large-Vessel Choroidal Layer Thickness (μm), Mean ± SE
	TAO Eyes (um)	Control Eyes (um)	P Value
Superotemporal	185.0 ± 6.05	176.4 ± 6.07	0.2374
Temporal	169.2 ± 7.18	163.40 ± 6.03	0.0647
Inferotemporal	154.7 ± 6.04	149.3 ± 6.24	0.8549
Upper	182.3 ± 5.89	171.5 ± 4.54	0.0687
Central	200.1 ± 5.75	181.0 ± 5.13	0.0203
Lower	138.6 ± 5.12	125.1 ± 5.25	0.0184
Superonasal	165.5 ± 4.58	157.6 ± 6.13	0.1170
Nasal	155.4 ± 5.44	141.9 ± 6.83	0.0492
Inferonasal	103.3 ± 4.52	92.14 ± 3.63	0.0371
Average	161.4 ± 2.69	149.92 ± 2.29	0.0014

Comparisons of whole choroidal thickness, choriocapillary density, large-vessel choroidal layer density, and large-vessel choroidal layer thickness between TAO and control eyes. Data are expressed as mean ± standard error. Vascular density and thickness were compared using independent t tests for parameters with normal distribution and Mann–Whitney U test for parameters with nonnormal distribution.

SE = standard error; TAO = thyroid-associated ophthalmopathy.

To investigate the underlying causes of choroidal thickening, we compared choriocapillary and large-vessel choroidal layer densities between TAO and control eyes. Whereas the mean choriocapillary density was slightly lower in TAO eyes (47.57 ± 3.24% vs. 47.89 ± 0.12%, P = 0.067), the superotemporal (46.85 ± 0.48% vs. 47.92 ± 0.37%, P = 0.0128) and inferonasal (48.13 ± 0.45 vs. 48.88 ± 0.13, P = 0.032) subfields showed a significant reduction in density, and other subfields exhibited no significant differences.

Large-vessel choroidal layer density was significantly reduced in TAO eyes compared with controls (39.86 ± 0.25% vs. 40.54 ± 0.15%, P = 0.0142). This reduction was particularly evident in the following subfields: temporal (38.82 ± 0.44% vs. 40.00 ± 0.33%, P = 0.0067), lower (41.15 ± 0.55% vs. 41.92 ± 0.37%, P = 0.0006), nasal (40.01 ± 0.73% vs. 40.46 ± 0.45%, P < 0.0001) and inferonasal (36.44 ± 0.79% vs. 38.66 ± 0.53%, P = 0.0003) subfields. In contrast, significant increases in density were observed in the central subfield compared with controls (39.37 ± 0.41% vs. 37.92 ± 0.38%, P = 0.011).

The observed increase in choroidal thickness in TAO eyes is likely attributable to stromal expansion, given the concurrent reductions in both choriocapillary and large-vessel choroidal layer densities. Supporting this hypothesis, TAO eyes demonstrated a significantly thicker large-vessel choroidal layer compared with controls (161.4 ± 2.69 μm vs. 149.92 ± 2.29 μm, P < 0.0001), with pronounced differences in the central, lower, nasal, and inferonasal subfields. This thickening was consistent with the observed increase in whole choroidal thickness.

To better characterize these morphological changes, we analyzed spatial distribution patterns using curve profiling. In control eyes, both whole choroidal and large-vessel choroidal layer thickness exhibited a characteristic distribution, highest in the central subfield with secondary peaks in the superotemporal and superonasal subfields, with 3 troughs in the inferonasal, low, and inferotemporal subfields (Fig 2A, B). Although TAO eyes maintained this overall topographic pattern, they demonstrated uniformly increased thickness values across nearly all subfields.

Figure 2.

Regional distributions of vascular parameters in TAO and control eyes: A, overall choroidal thickness, B, large-vessel choroidal layer thickness, C, choriocapillary density, D, large-vessel choroidal layer thickness density. Red dots represent mean values across 9 standardized subfields (nasal, superonasal, upper, superotemporal, temporal, inferotemporal, lower, inferonasal, and central) for each group. LVCL = large-vessel choroidal layer.

The spatial distribution of choriocapillary density in control eyes revealed an inverse relationship with thickness patterns, showing maximal density in the lower subfield and relatively high values in superotemporal and nasal regions (Fig 2C). Thyroid-associated ophthalmopathy eyes displayed an altered density profile, with significant reductions in the superotemporal and nasal subfields.

Analysis of large-vessel choroidal layer density in control eyes identified 3 distinct peaks corresponding to the superotemporal, upper and superonasal subfields, a pattern that paralleled the distribution of overall choroidal thickness (Fig 2D). Although TAO eyes preserved similar density values at these peak locations, significant regional differences showed marked reductions in temporal and inferonasal subfields, and elevated density in the central subfield.

To evaluate differences among TAO subgroups, patients were stratified into non-DON group (45 eyes: 5 mild and 40 moderate TAO) and DON group (20 severe TAO eyes). Both groups were compared against control eyes. Our analysis revealed that the DON group exhibited significantly greater choroidal thickness compared with both non-DON and control groups (P < 0.01), whereas non-DON eyes showed only minimal, nonsignificant thickness increases relative to controls (Table 2).

Table 2.

Comparisons of Choroid Parameters among Non-DON, DON, and Control Eyes

	Non-DON Eyes	DON Eyes	Control Eyes	P value1	P value2	P value3
Average whole choroidal thickness (um)	186.00 ± 2.74	208.64 ± 6.23	182.4 ± 2.29	<0.0001	0.5039	0.0019
Average choriocapillary density (%)	47.64 ± 0.14	47.86 ± 0.24	47.89 ± 0.12	0.4263	0.1130	0.7991
Average LVCL density (%)	39.57 ± 0.24	40.72 ± 0.32	40.54 ± 0.15	0.0370	0.0416	0.2492
Average LVCL thickness (um)	155.33 ± 2.84	179.22 ± 6.17	149.92 ± 2.29	<0.0001	0.0447	<0.0001

Comparative analyses of choroidal parameters were performed including: average whole choroidal thickness, choriocapillary density, large-vessel choroidal layer density, and large-vessel choroidal layer thickness between TAO subgroups (non-DON and DON) and control eyes. Statistical significance was evaluated for 3 pairwise comparisons: (1) non-DON versus DON eyes (P value1), (2) non-DON versus control eyes (P value2), and (3) DON versus control eyes (P value3).

DON = dysthyroid optic neuropathy; LVCL = large-vessel choroidal layer.

Both TAO subgroups exhibited slightly reduced choriocapillary density compared with controls, though this difference was not statistically significant. Notably, large-vessel choroidal layer density followed a distinct pattern, being highest in DON eyes, intermediate in controls, and lowest in non-DON eyes, although the difference between DON and control eyes did not reach statistical significance.

Large-vessel choroidal layer thickness showed marked increases in both TAO subgroups relative to controls (DON: 179.22 ± 6.17 μm vs. non-DON: 155.33 ± 2.84 μm vs. control: 149.92 ± 2.29 μm; both P < 0.001). These progressive choroidal vascular changes, particularly the severity-dependent thickening of large-vessel choroidal layer suggest a strong association between vascular remodeling and TAO progression, with the most dramatic alterations occurring in DON cases.

Discussion

In this study, we employed UWF SS-OCTA to evaluate choroidal structural thickness and blood flow changes in TAO. Before this investigation, we proposed that choroidal vascular dilation might be a primary mechanism that contributes to choroidal thickening in TAO because of the congested vein drainage. However, our findings revealed significant thickening of both the full choroid and large-vessel choroidal layers, whereas vascular density in both the choriocapillary and large-vessel choroidal layers showed inverse decrease, although these decreases did not reach statistical significance in most subfields.

This apparent paradox suggests that choroidal thickening in TAO may be driven primarily by nonvascular stromal expansion rather than vascular engorgement. To analyze this discrepancy between structural thickening and blood flow decrease, we should analyze the potential mechanisms from a pathological perspective.

The main characteristic of TAO is the increased orbital tissue volume, including the extraocular muscles, extracellular matrix, and intraorbital fat. The elevated intraorbital pressure significantly imbeds blood flow, affecting both arterial supply and venous drainage. The choroidal vascular supply derives mainly from the posterior ciliary arteries, which branch from the ophthalmic artery.¹⁵ Venous drainage occurs through the superior ophthalmic vein (SOV), which normally demonstrates anterograde anterior-to-posterior flow.¹⁶ In TAO, elevated orbital pressure might disrupt this delicate hemodynamic balance through compromised arterial supply and congested venous drainage.

In TAO, increased vascular resistance in the ophthalmic and central retinal arteries have been reported,⁶ as well as the disrupted venous blood drainage, even reversed or arterialized venous flow in severe cases.¹⁷ However, these changes manifest differently across disease phases. In inactive TAO phases, the persistently elevated vascular resistance in ophthalmic artery and central retinal artery could induce reduced retinal and choroidal perfusion.⁶ Some studies used OCTA to detect decreased retinal capillary density,⁷ or an initial increase in microvascular perfusion in mild TAO followed by a decline in the severe stage.⁸

In contrast, active TAO presents with a hyperdynamic circulatory state characterized by macrovascular dilation and compensatory perfusion increases. This alteration manifests as elevated blood flow velocities and reduced resistance in ophthalmic and central retinal arteries,¹⁸ along with enhanced capillary density in both superficial and deep macular layers.^19,20 This vasodilatory response might be mediated by inflammatory factors, such as thrombomodulin that inhibits vascular contractility and promotes orbital perfusion.^21,22

However, with the progresses of TAO, sustained inflammation and tissue expansion generate progressively elevated orbital pressures, reaching up to 40 mmHg in severe cases.²³ This mechanical compression ultimately leads to vascular collapse, manifesting reduced retinal capillary density,²⁴ choroidal peripapillary capillary density,²⁵ and the increase of resistance indexes in ophthalmic artery.⁶

In TAO, the elevated orbital pressure by increased intraorbital volume could impede venous outflow,²³ which may lead to vascular dilation and choroidal thickening. Previous studies reported decreased SOV blood flow in TAO,^26,27 with SOV enlargement observed in 90% of DON eyes.²⁸ These changes correlate with clinical severity, as active TAO eyes exhibit more pronounced SOV flow reduction than inactive cases,¹⁸ and moderate-to-severe Graves’ orbitopathy shows significantly lower flow than mild cases.⁵ Some studies have even documented reversed SOV flow in TAO, particularly in DON eyes,²⁹ indicating severe venous stasis. However, with the implementation of orbital decompression surgery, SOV blood flow got significantly improved.³⁰

Although it is well accepted that blood flow decreases in the ophthalmic artery and SOV in TAO eyes, the changes in choroidal vasculature remain controversial. Previously, the increased choroidal thickness in TAO is mainly attributed to impaired venous outflow,³¹ which causes choroidal thickening due to vessel dilation.

However, our results demonstrated significantly greater choroidal thickness in TAO patients that are consistent with previous reports,31, 32, 33 but reduced vascular density both in choriocapillary and large-vessel choroidal layers. The earlier studies were limited by discrete point measurements at the nasal and temporal regions or solely at the subfoveal area. In contrast, UWF SS-OCTA provides a larger scanning area (24 mm × 20 mm) and enables average choroidal thickness and the vascular density measurements for each subfield, thus the values obtained in our study will improve the data precision.

Notably, we observed significant choroidal thickening in the superior, inferior, and nasal regions, aligning with previous findings.³⁴ These localized changes may be linked to extraocular muscle involvement, as muscle thickening predominantly affects the inferior and medial rectus muscles, followed by the superior and lateral rectus muscles.³⁵

Apart from structural assessments, some prior OCTA studies reported reduced choriocapillaris flow but increased deeper choroidal flow compared with healthy subjects.³⁶ Corresponding to this, another study reported higher luminal area mainly in large choroidal vessels, but insignificant change of subfoveal choroidal thickness in TAO patients.³⁷ Nonetheless, our results showed no significant difference in choriocapillary density between TAO and control eyes, whereas there was reduced large-vessel choroidal density, particularly in the inferonasal subfield. These discrepancies may stem from variations in sample size, measurement heterogeneity, and inconsistent TAO staging. As in DON eyes, higher large-vessel choroidal density was detected compared with non-DON and control eyes.

Thus, our results suggest that the choroidal stromal edema is the primary driver of choroidal thickening rather than vascular engorgement. Supporting this, we detected increased average large-vessel choroidal layer thickness in TAO eyes, particularly in the central, inferior, and nasal regions. Notably, these changes in the large-vessel layer align with the observed overall choroidal thickening.

Additionally, previous studies have reported a positive correlation between increased choroidal thickness and CAS,^31,33 that higher choroidal thickness was observed in active TAO compared with stable cases.^34,36,38 However, no significant differences were found between moderate to severe TAO patients and healthy controls.³⁴ In contrast, the primary focus of the present study was on disease severity (mild, moderate-to-severe, and DON) rather than on activity. Notably, the vast majority of our TAO eyes (92.3%) were in the active phase (CAS ≥3), which limited a direct activity-based comparison. However, investigating the relationship between choroidal vascular parameters and disease activity using longitudinal UWF SS-OCTA represents an important direction for our future research.

Since most of our TAO subjects were in the active phase, we performed stage-specific analyses. These revealed significant choroidal changes in DON eyes, whereas no notable differences were observed between non-DON and control eyes (P > 0.05). Our findings align with previous reports demonstrating substantially greater submacular choroidal thickness in DON eyes than non-DON and control eyes.³⁶ Beyond thickness measurements, Wu et al²⁵ additionally reported significantly reduced peripapillary choroidal capillary density in DON eyes compare with non-DON cases.

These progressive vascular changes, especially the severity-dependent thickening of large-vessel choroidal layer suggests a strong association between choroidal remodeling and TAO progression. However, the relative contributions of disease activity versus severity to choroidal thickness and hemodynamic alterations require further investigation.

Thus, the discrepancy of increased choroidal thickness and decreased blood flow may be explained by several underlying processes: (1) extracellular matrix remodeling and stromal edema contributing to choroidal expansion independent of vascular dilations; (2) vascular compression resulting from elevated orbital pressure and stromal edema, which might lead to reduced luminal flow, despite compromised arterial supply and impaired venous drainage; (3) altered autoregulatory mechanisms in orbital vessels may disrupt the typical relationship between vascular morphology and perfusion. However, the precise regulation of choroidal blood flow in TAO remains under explored.

Our investigation has several important limitations that should be considered when interpreting the results. First, the relatively modest sample size, particularly the mild TAO subgroup (only 5 eyes) may impact the statistical power of our findings. Therefore, larger cohort studies are needed to validate these preliminary findings. Second, the automated segmentation protocol (with a fixed choriocapillaris layer thickness of 29 μm beneath Bruch's membrane) may introduce systematic bias in choriocapillary density measurements. Third, the predominant active TAO cases in our cohort precluded meaningful analysis of disease stage-dependent effects.

Conclusions

Our UWF SS-OCTA study provides novel, high-resolution characterization of choroidal structural and vascular alterations in TAO, revealing that choroidal thickening was prominently driven by stromal edema rather than vascular engorgement, which is most pronounced in DON cases. These choroidal changes reflect a multifactorial pathophysiology process involving mechanical compression, inflammatory mediated stromal remodeling and impaired vascular dynamics. Critically, our findings underscore the necessity to fully understand the dynamic process of hemodynamic changes among arterial perfusion, microvascular remodeling, and venous outflow during TAO progression.

Therefore, future longitudinal studies employing UWF SS-OCTA combined with Doppler ultrasound at defined disease stages (eg, at presentation, during active inflammation, and after immunosuppressive) are warranted. Such studies could clarify whether choroidal thickening and reduced vascular density precede clinical deterioration, and whether these parameters are reversible, which will potentially establish the causal relationships between choroidal remodeling and TAO progression.

Manuscript no. XOPS-D-25-01050R2.