Abstract

Objective:

This study aimed to investigate white matter microstructural damage, spontaneous brain activity, and functional connectivity alterations in patients with thyroid-associated ophthalmopathy (TAO) and visual impairment using multimodal MRI.

Methods:

Forty-five TAO patients with visual impairment and 32 healthy controls (HCs) underwent diffusion kurtosis imaging (DKI) and resting-state functional MRI (rs-fMRI). Microstructural changes along the visual pathway were quantified using fractional anisotropy (FA), mean kurtosis (MK), and mean diffusivity (MD). Regional spontaneous activity was assessed using the amplitude of low-frequency fluctuation (ALFF) and fractional ALFF (fALFF), and seed-based functional connectivity (FC) analyses were employed. Group differences were examinded using two-sample t-tests with false discovery rate (FDR) correction (p < 0.05). Correlations between imaging parameters and clinical indicators were further analyzed.

Results:

Compared with HCs, the TAO group showed significantly decreased MK, FA, and MD values in the optic radiation, lateral geniculate body, optic tract, and optic nerve, with MK being the most markedly reduced one. Patients also exhibited increased ALFF in the right parahippocampal gyrus and decreased ALFF in the left calcarine fissure, as well as decreased fALFF in the left calcarine fissure and postcentral gyrus. Further analyses revealed decreased FC between the left calcarine fissure and the right lingual gyrus/middle frontal gyrus/bilateral postcentral gyri, and increased FC with the left supplementary motor area/fusiform gyrus. ALFF/fALFF in visual network-related regions correlated with proptosis and extraocular muscle thickening in TAO patients.

Conclusion:

TAO is associated with white matter microstructural damage, aberrant spontaneous neural activity, and functional reorganization. The combination of DKI and rs-fMRI (ALFF/fALFF, FC) may provide comprehensively insights into central neuropathological mechanisms, providing a basis for early diagnosis and targeted intervention.

1 Introduction

Thyroid-associated ophthalmopathy (TAO) is a common orbital disease, which has been ranked as the foremost disorder affecting the orbit (1). The incidence of TAO peaks between the ages of 40 and 60, with a higher prevalence observed in women than in men (2–4). Visual pathway injury is a frequent complication of TAO. This complication often develops insidiously, with early-stage symptoms typically being subtle or even unremarkable; however, in severe cases, it can potentially lead to blindness (5). Therefore, accurate and timely diagnosis, coupled with early intervention, is crucial to reduce TAO-related morbidity.

The pathological mechanisms underlying visual impairment in TAO patients are not yet fully understood. The most widely accepted mechanism is optic nerve compression due to hypertrophy of the extraocular muscles and edema of the apical orbital tissue, leading to neural damage (6). Given that the transmission and processing of visual information engage multiple central nervous system structures,including the optic nerve, optic chiasm, and optic tract—it is plausible that TAO pathophysiology may extend to the central nervous system. Recent neuroimaging studies have showen that TAO affects not only periocular structures but may also induce structure and function alterations in the visual pathway and other brain regions (7–9).

Resting-state functional magnetic resonance imaging (rs-fMRI) captures spontaneous blood oxygen level–dependent (BOLD) signals and provides a unique, noninvasive approach to characterizing brain functional architecture. It has emerged as a valuable tool for assessing the relationship between brain alterations and various ophthalmic diseases (10–13). Among rs-fMRI metrics, the amplitude of low-frequency fluctuation (ALFF) and fractional ALFF (fALFF) can quantify the intensity of regional neuronal activity, whereas functional connectivity (FC) analysis can reveal alterations in functional synchronization across distinct brain regions. These techniques have been gradually applied in investigating the central mechanisms of TAO. For example, Liu et al. (14) demonstrated abnormal spontaneous brain activity in patients with TAO using the ALFF based on rs-fMRI. Li et al. (8) utilized resting-state functional MRI and demonstrated that abnormal brain functional connectivity contributes to impaired mood and cognition in patients with hyperthyroidism. Zhang et al. (6) demonstrated significant morphological and functional alterations in the visual cortex and certain regions of the default mode network in patients with TAO and optic neuropathy (15). Jiang et al. (9), utilizing rs-fMRI, demonstrated alterations in spontaneous neuronal activity and functional connectivity patterns in patients with TAO. Nevertheless, the current understanding of visual pathway injury and associated structural and functional brain changes in TAO remains limited, highlighting a gap in effective methods for early diagnosis and objective assessment of disease severity.

Moreover, fewer studies have investigated white matter microstructural damage in the visual pathway of TAO patients. Diffusion kurtosis imaging (DKI) is an advanced diffusion MRI technique that quantifies the non-Gaussian behavior of water diffusion, enabling more sensitive detection of microstructural heterogeneity—such as axonal density and myelin integrity—in complex biological tissues like the optic nerve (16). Compared to conventional diffusion tensor imaging (DTI), DKI has demonstrated unique advantages in studies of neurodegenerative diseases and optic neuritis (17, 18). Therefore, DKI may be well suited to evaluate visual pathway damage in TAO.

Therefore, to comprehensively elucidate the neuropathological mechanisms of TAO, this study integrated DKI, ALFF/fALFF, and FC techniques to quantify microstructural integrity in various visual pathway regions, spontaneous brain activity, and visual network connectivity patterns in TAO patients and healthy controls (HCs). The study aims to: (1) quantitatively assess microstructural changes in visual pathway regions of TAO patients using DKI parameters; (2) identify key brain regions associated with abnormal cerebral activity in TAO; (3) characteriaze patterns of functional network reorganization; and (4) examine relationships between neuroimaging metrics and clinical phenotypes. The findings may provide novel insights into the central pathological mechanisms of TAO and suggest potential targets for neuromodulatory interventions.

2 Materials and methods2.1 Participants

Patients with visual impairment due to TAO were recruited from the Departments of Ophthalmology and Endocrinology at Shenzhen Traditional Chinese Medical Hospital were enrolled. The inclusion criteria of TAO patients were: (1) Patients who meeting the diagnostic criteria for TAO (based on the Bartley criteria) (19) and presenting with visual impairment; (2) aged between 18–60 years, and (3) no contraindications to MRI.

Exclusion criteria were: (1) Individuals with visual dysfunction not attributable to TAO; (2) unable to complete the facial judgment task; (3) poorly controlled diabetes or severe systemic diseases; (4) a history of traumatic brain injury, brain tumors, or cerebrovascular disease; (5) individuals with a personal or family history of mental disorders; and (6) contraindications to the examination or MR images that did not meet quality requirements.

Finally, 45 TAO patients were included in this study (14 males and 31 females, aged 18 to 60 years, with a mean age of 38.00 ± 6.27 years). In addition, 32HCs were recruited (11 males and 21 females, aged 18 to 60 years, with a mean age of 34.32 ± 7.18 years), and were matched to the TAO group in terms of demographic characteristics such as age, gender, and educational level. Although thyroid function laboratory tests (TSH, FT3, FT4) and orbital MRI were not performed in the healthy control group, all HCs underwent detailed clinical evaluation and medical history collection to exclude: (1) history of thyroid dysfunction (including hyperthyroidism, hypothyroidism, thyroiditis, etc.); (2) history of autoimmune diseases; (3) history of orbital diseases (including proptosis, strabismus, extraocular muscle disorders, etc.); (4) history of ocular surgery or trauma. Furthermore, all HCs received routine physical examination with clinical confirmation of absence of proptosis, extraocular movement disorders, or strabismus signs.

This case–control study was approved by the Medical Ethics Committee of Shenzhen Hospital of Traditional Chinese Medicine (No. S2024-091), and written informed consent was obtained from all participants.

2.2 Imaging data

MRI scanning was performed on a 3.0 T Siemens Prisma scanner with a 64-channel head coil. Soft pads were placed on both sides of the participant’s neck to Results section stabilize the head and minimize motion, and earplugs were inserted into both ears to reduce noise interference. Throughout the scanning procedure, participant s were instructed to remain still with their eyes closed and to stay awake.

The following MRI sequences were acquired in both patients and HCs, including DKI, high-resolution Three-dimensional high-resolution T1-weighted structural imaging (3D-T1WI), and Resting-state blood oxygenation level-dependent functional MRI (rs-fMRI/BOLD).

DKI: Repetition time (TR) = 2,600 ms, echo time (TE) = 62 ms, slice thickness = 3 mm, and slice gap = 0 mm. Diffusion encoding was applied along 30 directions with b-values of 0, 1,000, and 2000 s/mm2.

3D-T1WI: TR = 2,200 ms, TE = 2.45 ms, inversion time (TI) = 900 ms, matrix = 256 × 256, voxel size = 1.0 × 1.0 × 1.0 mm3, and flip angle = 8°.

rs-fMRI(BOLD): TR = 2000 ms, TE = 30 ms, matrix = 240 × 240, 250 time points, slice thickness = 4 mm, slice gap = 0 mm, number of excitations (NEX) = 1, and flip angle = 90°.

2.3 Data processing and analysis2.3.1 DKI data processing and parameter quantification

DKI data were processed using the Syngo.via (https://www.siemens-healthineers.com/digital-health-solutions/syngovia) post-processing workstation. The original DICOM data were first converted to the NIfTI format and then imported into the MRIcron (https://www.nitrc.org/projects/mricron) to generate parametric maps of fractional anisotropy (FA), mean kurtosis (MK), and mean diffusion (MD). Regions of interest (ROIs) were manually delineated on the FA maps along the visual pathway. The assessed structures included the optic radiation, lateral geniculate body, optic tract, and optic nerve. The optic chiasm was excluded from evaluation due to its complex anatomy and the inherent limitations of DKI in clearly delineating this structure. These ROIs were then copied and mirrored to the contralateral side to obtain bilateral FA values. The same ROIs were used to extract MD and MK values in the corresponding regions of the visual pathway. Finally, FA, MK, and MD values were obtained for each structure (optic nerve, optic tract, lateral geniculate body, and optic radiation) and compared between the TAO group and HCs.

2.3.2 fMRI data processing and analysis

The rs-fMRI data were preprocessed using the Data Processing & Analysis for Brain Imaging (DPABI, https://rfmri.org/DPABI). The steps were as follows: (1) removal of the first 10 volumes; (2) slice timing correction; (3) realignment (subjects with head motion exceeding 2 mm in translation or 2° in rotation were excluded. No subjects were excluded in this study due to excessive head motion.); (4) spatial normalization using the Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra (DARTEL) tool; (5) resampling to 3 mm isotropic voxels; and (6) spatial smoothing with a 6 mm full-width at half-maximum (FWHM) Gaussian kernel; (7) Nuisance regression was performed to remove linear and quadratic drift signals, 24 head motion parameters, and signals from white matter and cerebrospinal fluid. For resting-state functional connectivity analysis, three additional steps were applied: (1) global signal regression; (2) temporal band-pass filtering (0.01–0.1 Hz); and (3) scrubbing of volumes with framewise displacement > 0.5, along with the two preceding and one succeeding time point.

Following preprocessing, the ALFF was computed. The preprocessed BOLD time series were transformed into the frequency domain using a Fast Fourier Transform. The square root of the power spectrum was calculated and averaged across the low-frequency range (0.01–0.1 Hz) for each voxel. This ALFF value reflects the strength of regional spontaneous brain activity. fALFF was also calculated as the ratio of the power in the low-frequency range (0.01–0.1 Hz) to that of the entire frequency range (0–0.25 Hz), which helps reduce the influence of high-frequency noise and provides a more specific measure of local brain activity.

Based on the ALFF and fALFF results, the left calcarine sulcus, which showed abnormal activity, was selected as a seed region. The mean time series was extracted from this seed region. Seed-based FC analysis was then performed by computing Pearson’s correlation coefficients between the seed region’s time series and the time series of every other voxel in the brain. The resulting correlation coefficients were transformed to z-values using Fisher’s r-to-z transformation to improve normality. Finally, two-sample t-tests were conducted to compare the FC strength between the TAO patient group and the HCs.

2.4 Statistical analysis

Statistical analysis of demographic characteristics and clinical data was performed using SPSS software (version 22.0). To compare differences between the TAO group and HCs, independent samples t-tests were used for continuous variables with normal distribution, and chi-square tests were applied for categorical variables. For group comparisons of DKI parameters (FA, MK, MD), independent samples t-tests were used with False Discovery Rate (FDR) correction for multiple comparisons. Specifically, correction was applied separately for each parameter in each visual pathway region (optic nerve, optic tract, lateral geniculate body, optic radiation), with significance threshold set at p < 0.05. All reported p-values are FDR-corrected. For whole-brain voxel-wise ALFF and fALFF analyses, Gaussian Random Field (GRF) theory was used for multiple comparisons correction, with voxel-level p < 0.01 and cluster-level p < 0.05. This is standard practice in neuroimaging research. It should be specifically noted that the DPABI data processing software only outputs significant results passing the correction threshold (p < 0.05) without displaying exact p-values, which is a limitation of this software’s standard output format. Multiple comparisons correction: For ALFF and fALFF analyses, multiple comparisons were corrected using the Gaussian Random Field (GRF) theory with a voxel-level p < 0.01 and cluster-level p < 0.05. For functional connectivity analysis, the False Discovery Rate (FDR) correction was applied with a threshold of p < 0.05. An independent two samples t-tests were conducted to compare the differences in amplitude of ALFF, fALFF values, and FC between the TAO group and HCs with respect to clinical disease characteristics. A threshold of p < 0.05 was considered statistically significant.

3 Results3.1 Comparison of demographic and clinical characteristics

No statistically significant differences were observed in gender, age, or years of education between the TAO group and the HCs (all p > 0.05). Best-corrected visual acuity was significantly lower in the TAO as compared to HCs (p < 0.05) (Table 1). The TAO group showed greater bilateral exophthalmos and thicker extraocular muscles than HCs (all p < 0.05) (Table 2).

ParametersTAO group (n = 45)Healthy controls (n = 32)t/χ2 valuepAge (years)a37.00 ± 7.1433.55 ± 7.081.230.226Gender (Male/Female)b12/2014/80.980.252Years of education (years)a14.27 ± 4.5615.65 ± 4.430.590.561Best corrected visual acuitya0.73 ± 0.160.98 ± 0.05−8.280.011*Free triiodothyronine (FT3, pmol/L)6.06 ± 3.2///Free thyroxine (FT4, pmol/L)14.89 ± 8.27///Total triiodothyronine (TT3, pmol/L)2.64 ± 2.8///Total thyroxine (TT4, pmol/L)128.88 ± 24.16///Thyroid-stimulating hormone (TSH, mIU/L)3.18 ± 4.46///TSH receptor antibody (TRAb, U/L)13.25 ± 8.47///

Comparison of general clinical characteristics.

aFor two sample t tests, b for X2, and *p < 0.05 was considered statistically significant. “/” indicates data not obtained or not applicable. Thyroid function tests (FT3, FT4, TT3, TT4, TSH, TRAb) were not performed in healthy controls as they were not clinically indicated for this group. Bold values indicate p < 0.05, statistically significant.

ParametersTAO group (n = 45)Healthy controls(n = 32)t valuepExophthalmos (mm)Right exophthalmos20.24 ± 1.7815.64 ± 1.568.95<0.001*Left exophthalmos20.47 ± 1.8315.33 ± 1.539.23<0.001*Extraocular muscle thickness (mm)Right lateral rectus5.72 ± 3.273.19 ± 0.243.120.005*Left lateral rectus6.11 ± 3.153.12 ± 0.544.24<0.001*Right medial rectus4.67 ± 2.312.89 ± 0.563.420.008*Left medial rectus4.54 ± 2.252.95 ± 0.583.310.009*Right superior rectus4.76 ± 3.123.42 ± 0.632.830.005*Left superior rectus4.89 ± 3.243.23 ± 0.692.690.007*Right inferior rectus6.78 ± 4.673.58 ± 0.744.56<0.001*Left inferior rectus6.54 ± 4.213.55 ± 0.874.32<0.001*

Comparison of exophthalmos and extraocular muscle thickness.

Two sample t-tests, and *p < 0.05 was considered statistically significant. Bold values indicate p < 0.05, statistically significant.

3.2 Results of DKI analysis

Measurement and analysis of the MK, FA, and MD revealed varying degrees of reduction in these metrics across seveal regions of the visual pathway (including the optic radiation, lateral geniculate body, optic tract, and optic nerve) in the TAO group compared with HCs (Figure 1). Among all the metrics, MK showed the most pronounced decrease. These findings indicate the presence of impairments of varying severity in the visual pathways of the patients (Figure 2 and Table 3).

The FA, MK, and MD maps of the optic nerve, optic tract, lateral geniculate body, and optic radiation planes. FA, fractional anisotropy; MK, mean kurtosis; and MD, mean diffusion.

Altered DKI measurements in the optic nerve, optic tract, lateral geniculate body, and optic radiation in the TAO as compared to HCs. FA, fractional anisotropy; MK, mean kurtosis, and MD, mean diffusion; *indicates p < 0.05.

Anatomical regionParametersTAO patients (n = 45)Healthy controls (n = 32)p (FDR corrected)Optic nerveFA value (×10−3 mm2/s)0.17 ± 0.050.38 ± 0.090.021*MK value0.55 ± 0.180.82 ± 0.250.008*MD value0.60 ± 0.190.88 ± 0.260.042*Optic tractFA value (×10−3 mm2/s)0.20 ± 0.060.40 ± 0.100.005*MK value0.41 ± 0.140.67 ± 0.180.003*MD value0.65 ± 0.210.97 ± 0.290.543Lateral geniculate bodyFA value (×10−3 mm2/s)0.18 ± 0.060.43 ± 0.100.012*MK value0.58 ± 0.190.85 ± 0.260.005*MD value0.63 ± 0.200.92 ± 0.280.724Optic radiationFA value (×10−3 mm2/s)0.27 ± 0.090.48 ± 0.120.026*MK value0.65 ± 0.220.89 ± 0.270.007*MD value0.80 ± 0.250.93 ± 0.270.745

Results of DKI measurements in the optic nerve, optic tract, lateral geniculate body, and optic radiation.

Two sample t-tests, and *p < 0.05 was considered statistically significant. Bold values indicate p < 0.05, statistically significant.

3.3 Results of the ALFF analysis

In the statistical analysis, with age, gender, and framewise displacement (a head motion parameter) included as covariates, comparisons were conducted between the TAO group and HCs. Multiple comparisons were corrected using the GRF (voxel-level p < 0.01, cluster-level p < 0.05). Compared with HCs, the TAO group exhibited a significant increase in ALFF in the right parahippocampal gyrus and a significant decrease in the ALFF in the left calcarine sulcus (Figure 3A and Table 4). In addition, the fractional ALFF (fALFF) was significantly decreased in the left calcarine sulcus and the left postcentral gyrus (Figure 3B and Table 4).

(A) Brain regions with ALFF differences between TAO patients and HCs; (B) Brain regions with fALFF differences between TAO patients and HCs. CAL. L, left calcarine sulus; PHG. R, right parahippocampal gyrus; and PoCG. L, left postcentral gyrus; *indicates p < 0.05.

MetricClusterBrain regionMNI coordinates (x, y, z)Cluster size (voxels)TALFFCluster 1Right parahippocampal gyrus(33,-39,-12)2425.2684Cluster 2Left calcarine sulcus(0,-87,3)139−4.1924fALFFCluster 1Left calcarine sulcus(−6,-81,3)109−4.0680Cluster 2Left postcentral gyrus(−51,-21,45)109−3.8297

Brain regions with ALFF/fALFF differences between TAO patients and HCs.

Montreal Neurological Institute (MNI) space; coordinates (X, Y, Z) of the peak activation in MNI space; peak (T-value).

3.4 Results of functional connectivity analysis

In the FC analysis, the false discovery rate (FDR) correction was applied for multiple comparisons, with the statistical significance set at p < 0.05. Since the primary visual cortex is located in the cortical area surrounding the calcarine sulcus, the two previously identified calcarine sulcus regions showing significant differences were used as seed masks for further seed-based FC analysis. This analysis aimed to compare differences in functional connectivity between these visual brain regions (which exhibited alterations in spontaneous brain activity in patients) and other brain regions. Compared to HCs, the TAO group exhibited increased FC between the left calcarine sulcus and the left supplementary motor area (SMA) and the left fusiform gyrus, as well as decreased FC between the left calcarine sulcus and the right lingual gyrus, right middle frontal gyrus, and bilateral postcentral gyri (Figure 4 and Table 5).

Comparison between patients with TAO and healthy controls (HCs). (A) Represents the brain regions showing differences in whole-brain functional connectivity using the calcarine sulcus as a seed region based on the ALFF metric. (B) Represents the brain regions showing differences in whole-brain functional connectivity using the calcarine sulcus as a seed region based on the fALFF metric. L (left); R (right); CAL, calcarine sulcus; FFG, fusiform gyrus; LING, lingual gyrus; MFG, middle frontal gyrus; PoCG, postcentral gyrus; and SMA, supplementary motor area; *indicates p < 0.05; **indicates p < 0.01; ***indicates p < 0.001.

Seed RegionConnected brain regionMNI coordinates (x, y, z)Cluster size (voxels)Peak Tp (FDR corrected)Left calcarine sulcus (ALFF)Right lingual gyrus(−24, −12, −39)2234.64950.002Right middle frontal gyrus(−42, −36, 63)273−4.40050.003Left supplementary motor area(42, −27, 48)354−4.46110.002Left calcarine sulcus (fALFF)Left fusiform gyrus(15, −48, −3)252−4.742< 0.001Left postcentral gyrus(54, −9, 51)421−4.36070.001Right postcentral gyrus(−9, 6, 57)2384.55090.002

Detailed results of seed-based functional connectivity analysis.

Positive T-values indicate increased FC in TAO patients compared to HCs; negative T-values indicate decreased FC; Seed region: left calcarine sulcus selected based on ALFF/fALFF analysis results (regions showing abnormal spontaneous activity); FDR-corrected p-values were estimated based on peak T-values and degrees of freedom (df = 75) using the Benjamini-Hochberg procedure. DPABI software does not output exact p-values for FC analysis; these values represent conservative estimates derived from post-hoc calculations. All reported connections survived FDR correction at p < 0.05; *p < 0.05 w