Thyroid-associated ophthalmopathy (TAO) represents a multi-disciplinary and multi-system orbital autoimmune disease (Bartalena et al., 2008). Orbital fibroblasts result in orbital connective tissue volume expansion, inflammation and remodeling in TAO patients (Neag and Smith, 2021). The clinical manifestations include exophthalmos, diplopia, eyelid retraction, dysthyroid optic neuropathy (DON) and exposed keratitis (Bartalena et al., 2008; Bartalena et al., 2021; Hodgson and Rajaii, 2019), which can lead to disfigurement, vision loss and even blindness. Besides, patients with TAO often suffer from emotional and psychological abnormalities, including memory deficits, emotional lability, depression and anxiety (Bahn, 2010; Baxter et al., 2015; Tramunt et al., 2018). Disfiguring changes in the appearance represent one of the causes of emotional and psychological abnormalities (Farid et al., 2005; Hodgson and Rajaii, 2019). Previous studies (Kahaly et al., 2005; Ponto et al., 2011; Wiersinga, 2012) have shown that the quality of life (QoL) is affected in patients with TAO, and as the severity of the disease increases, the quality of life score also decreases (Park et al., 2003). Since this disease is related to the quality of life, it affects the mental health of patients and has a negative impact on social stability (Kahaly, 2019). Therefore, exploring the TAO-related brain structural alterations can enhance our understanding of the pathophysiological mechanisms underlying TAO.

Previous neuroimaging studies have shown extensive structural changes in the white matter (WM) and grey matter (GM) in TAO patients (Luo et al., 2022; Silkiss and Wade, 2016; Wu et al., 2020), involving the emotional, cognitive regulation and vision functional areas. Two recent studies (Li et al., 2022; Liu et al., 2022) applied diffusion tensor imaging (DTI) to evaluate the microstructural changes of the optic nerve and visual pathway in TAO patients, and found that moderate-severe TAO patients showed significantly decreased fractional anisotropy (FA) compared with controls in the lateral geniculate (LG) and optic radiation (OR). However, the brain structural abnormalities beyond visual pathway remain unclear. Wu and colleagues (Wu et al., 2020) revealed significantly decreased FA, and increased mean, axial and radial diffusivities in the right middle occipital gyrus, cuneus and superior occipital gyrus in TAO patients. Notably, our recent study (Luo et al., 2022) found significantly increased GM volume in the right middle temporal gyrus in TAO patients, which was significantly positively correlated with depression and anxiety scores of the patients. However, the aforementioned studies only focused on the structural changes in distinct brain regions, ignoring the interregional interactions between anatomically separated structures, which may play a vital role in terms of the brain information transference and processing.

Recent studies have shown that the brain structural networks can be constructed using DTI and probabilistic tractography (Wen et al., 2017; Xu et al., 2016). Graph theory analysis is the most commonly used technology in neuroimaging and brain network research, which can effectively reveal the topological structure of the brain network. Brain structural networks exhibit many topological properties inherent to naturally occurring biological networks, including highly connected hubs and small-worldness (Wen et al., 2017). Using tractography and graph theory analysis, brain structural networks in various diseases have been extensively studied, including the attention-deficit/hyperactivity disorder (Cao et al., 2013), Tourette syndrome (Wen et al., 2017), subcortical ischemic vascular cognitive impairment with no dementia (SIVCIND) (Sang et al., 2020), systemic lupus erythematosus (Xu et al., 2016) and others. However, there is relatively limited research on the brain structural network, especially the network topological abnormalities related to TAO.

Graph theory analysis was used by Wu and colleagues (Wu et al., 2021) to assess the alterations of brain structural networks in TAO patients compared with healthy controls, including the global and nodal topological metrics and network-based statistics. They found that TAO patients showed preserved global network parameters but significantly altered nodal properties, and the nodal properties also exhibited significant correlations with neuropsychological performances and visual acuity. However, their study did not take into account the different phases of TAO. Two phases in the development of TAO have been identified and widely recognize: the inactive phase and the active phase (Bartalena et al., 2008; Bartalena et al., 2021; Hodgson and Rajaii, 2019; Mourits, 1989). In the active phase, ocular edema, congestion, pain and other symptoms were observed in patients due to orbital inflammation. Meanwhile, in the inactive phase, the orbital inflammatory reaction subsided, fibrotic changes appeared, and the clinical symptoms of the patients were alleviated or subsided (Mourits et al., 1989) (external eye photography shown in supplementary Fig. S1). Furthermore, TAO patients at these two phases have obviously different emotional, psychological and cognitive differences (Farid et al., 2005). While the patients in the active phase often have increased psychological pressure due to factors such as the appearance changes, visual function effects and unsatisfactory treatment effects, patients in the inactive phase have alleviated or tolerated appearance changes are alleviated or tolerated and improved visual function, so the psychological stress is also reduced. In consequence, it is valuable to use the diffusion tractography and graph theory analysis to explore the potential changes of whole brain structural networks in different phases of TAO patients, which may provide a basis to reveal the neurobiological mechanism of TAO.

In this study, we first applied diffusion probabilistic tractography to construct the whole brain white matter (WM) structural networks in inactive and active TAO patients as well as healthy controls. Then, we used graph theory to characterize the network topological properties. Our hypothesis is that different global and nodal topological metrics will be present at different phases of TAO, and the topological changes of brain structure could correlate with the emotional, psychological and ophthalmic parameters of TAO patients.