Parkinson’s disease (PD) is a common neurodegenerative disease characterized by motor disorders such as bradykinesia, resting tremor, muscular rigidity, and postural instability (Postuma et al., 2015). It has recently been found that PD is also closely related to several psychiatric symptoms, which have a remarkable influence on health-related quality of life (Carod-Artal et al., 2008; Martínez-Martín and Damián, 2010; Gallagher and Schrag, 2012). Anxiety occurs more frequently in PD patients than in patients suffering from other chronic neurodegenerative diseases or in elderly controls with similar disabilities (Pontoon et al., 2009; Martínez-Martín and Damián, 2010). Although PD patients experience anxiety (Carod-Artal et al., 2008), the potential mechanism of anxiety in PD remains a matter of debate.

Anxiety and fear are thought to share a similar neural circuit, called the ‘fear circuit’, where fear as a feeling is experienced in response to existing danger or threat, while anxiety is experienced in response to prospective or imagined future danger or threat (Penninx et al., 2021). The fear circuit is composed of the amygdala, medial prefrontal cortex (mPFC), anterior cingulate cortex (ACC), hippocampus, insula, and striatum (Dresler et al., 2013). As the input and output nuclei of the circuit, the amygdala plays a central role in detecting external threats, transmitting information to other nodes of the fear circuit, and eventually exporting negative emotions such as anxiety and fear (Sun et al., 2020). Other areas of the fear circuit regulate the amygdala through ‘top-down’ cognitive processes, including assessing emotional threats, distracting attention from threats, and learning or extinction fear conditioning memory (Etkin et al., 2011; Mah et al., 2016). In previous studies, dysfunction of the fear circuit has been used to explain the neural mechanism of anxiety in PD (Carey et al., 2021). For example, to date, functional magnetic resonance imaging (fMRI) (Carey et al., 2020; Criaud et al., 2021) and voxel-based morphometry (VBM) (Vriend et al., 2016) studies have reported anatomical or functional alterations in key brain areas of the fear circuit in patients with PD and anxiety. However, the underlying mechanisms of these brain abnormalities remain unclear.

An increasing number of studies have found that aberrant iron metabolism is involved in PD pathogenesis. Excessive iron deposition in the brain, decreased serum iron levels, and increased transferrin levels in the circulation have been observed in patients with PD (Dexter et al., 1991; Yoshihara et al., 2016; Xu et al., 2018; Mochizuki et al., 2020). High-level transferrin transfers iron to the brain, especially the substantia nigra (SN) (Dexter et al., 1991; Costa-Mallen et al., 2017), by translocating iron-loaded transferrin and its receptor to the intracellular compartment (Kaur and Andersen, 2004). Oxidative stress and aggregation of α-synuclein (Ostrerova-Golts et al., 2000) caused by excessive iron deposition in the SN are considered to be involved in the pathogenesis of PD (Ayton and Lei, 2014).

Abnormal iron metabolism has also been shown to be closely associated with anxiety in animal experiments. Increased anxiety-like behaviors have been observed in iron-deficient rats (Beard et al., 2002; Li et al., 2011). After iron treatment, excessive iron deposition was observed in rat brain regions, such as the basal ganglia, frontal cortex, hippocampus, and cerebellum, and was accompanied by increased anxiety behaviors (Maaroufi et al., 2009). In PD patients, low serum iron and high serum transferrin levels are significantly associated with the severity of anxiety (Xu et al., 2018).

Therefore, we assumed that PD patients with anxiety have more severe iron deposition in the brain regions of the fear circuit than PD patients without anxiety or healthy individuals. Furthermore, oxidative stress caused by iron deposition in these brain regions leads to cellular and tissue damage (Nandar et al., 2013), which may be one of the fundamental causes of anxiety in PD. Perhaps because of the invasiveness of previous detection methods, few studies have focused on iron deposition in these brain regions in PD patients with anxiety. However, with the development of MRI technology, noninvasive and accurate quantitative detection of iron deposition in the brain has become possible.

Quantitative susceptibility mapping (QSM) is an emerging MRI technique capable of quantifying susceptibility changes in the brain tissue (Port and Pomper, 2000). Although other metals, such as aluminum and copper, may also result in increased susceptibility measurements, it can still be considered that iron is the most important factor affecting susceptibility differences because the effects of other metals are too mild compared with those of iron in the brain (Becker et al., 2005; Popescu et al., 2009). At present, QSM has been widely used in various aspects of PD research as imaging biomarkers and plays a crucial role in the diagnosis (Xiao et al., 2019), staging (Li et al., 2021), and assessment of motor (Chen et al., 2020) and non-motor (Uchida et al., 2019; Thomas et al., 2020) symptoms in PD. However, no studies to date have used QSM to explore the association between brain iron deposition in the fear circuit and anxiety in PD patients. Furthermore, the distribution of brain iron in patients with PD and anxiety remains unclear.

In this study, we hypothesized that there would be more iron accumulation in the brain regions of the fear circuit in PD patients with anxiety. Here, voxel-based QSM analysis was applied to compare the susceptibility of the whole brain among the three groups, i.e., PD patients with anxiety (PD-A), PD patients without anxiety (PD-NA), and healthy controls (HC). Then, based on the region of interest (ROI)-based analysis, regional QSM values related to the fear circuit were obtained and compared among the three groups, and the correlation between the QSM values of these brain regions and anxiety scores was determined. In this study, we aimed to comprehensively investigate differences in regional magnetic susceptibility among the three groups and explore the role of iron in the potential neural mechanisms of anxiety in PD.

In this study, whole-brain voxel-based QSM analysis and defined ROI-based QSM analysis were combined to identify brain tissue susceptibility alterations related to anxiety in patients with PD. We are the first to report increased brain iron deposition in several areas of the fear circuit in PD patients with anxiety. Specifically, significantly greater mean QSM values in the hippocampus were found in the PD-A group than in the PD-NA group. In addition, compared to the HC group, the PD-A group showed higher QSM values mainly in the fear circuit (mPFC and ACC), SMA, precuneus, angular gyrus, and middle occipital gyrus, while the PD-NA group mainly in the parahippocampal gyrus and superior temporal gyrus. Furthermore, the mean QSM values of the hippocampus, mPFC, and ACC are associated with HAMA scores in the subjects.

In the present study, increased brain iron was observed in the ventral mPFC and ACC in the PD-A group, which was positively correlated with HAMA scores. The ventral mPFC and ACC participate in inhibiting fear conditioning through extinction (Etkin et al., 2011), presumably by promoting adaptation to fearful stimuli to downregulate amygdala activity (Quirk and Mueller, 2008; Mobbs et al., 2009). Moreover, activation of the ventral ACC may be associated with automatic attention disengagement from threats (Vuilleumier et al., 2001). In previous resting fMRI studies, the abnormal amplitude of low-frequency fluctuation (ALFF) activity and functional connectivity (FC) with the amygdala were detected in the ventral mPFC and ACC in PD patients with anxiety (Criaud et al., 2021), and top-down control from the mPFC to the amygdala may be weakened due to dysfunction in these regions in anxiety patients (Kim et al., 2011). In the meantime, as the output nucleus eventually produces anxiety in the fear circuit, abnormal activation of the amygdala was observed in PD patients with anxiety (Dan et al., 2017; Criaud et al., 2021), which may be related to the decreased downregulation of the mPFC and ACC in the amygdala. These neurological dysfunctions may be closely related to dopaminergic system abnormalities in PD. Anxiety symptoms are highly sensitive to dopamine depletion in PD patients (Prediger et al., 2012), which has also been demonstrated in animal studies (Eskow Jaunarajs et al., 2010). Therefore, we considered excessive iron accumulation in the mPFC and ACC may be one of the root causes of anxiety in PD patients.

In addition, we found that excessive iron accumulation also existed in the hippocampus in the PD-A group compared with the other groups, according to the ROI-based QSM analysis. The hippocampus plays an important role in the regulation of emotion during fear conditioning. Along with the amygdala and mPFC, the hippocampus mediates fear conditioning and extinction and is also used to regulate stress responses (Anacker and Hen, 2017). The hippocampus is responsible for encoding and retrieving contextual information or cues related to threats (Bannerman et al., 2014). Decreased control of the hippocampus on contextual retrieval, leading to the recovery of previously vanished fear conditioning, is thought to be one of the mechanisms by which anxiety disorders develop (Rauch et al., 2006). In addition, the hippocampus is involved in assessing the threat level of stimulation (Mah et al., 2016). For example, a recent animal study showed that the hippocampus is involved in the response to acute stress (von Ziegler et al., 2022), suggesting that changes in the hippocampus may lead to an abnormal threat response. Several anatomical MRI studies have found that hippocampal volume is negatively correlated with anxiety scores in patients with anxiety (Irle et al., 2010). In contrast, no morphological changes in the hippocampus were found in the PD-A group, suggesting that iron deposition in the hippocampus of the PD-A group had increased even before morphological changes occurred, supporting the idea that QSM was more sensitive to earlier alterations of brain tissue than conventional anatomical MRI measurements in PD patients with anxiety. Thus, excessive iron accumulation in the hippocampus may be closely linked to anxiety in PD.

Anxiety may be caused by an imbalance in the fear circuit (Mah et al., 2016). In other words, whether the ventral neural system (amygdala) is highly active or the dorsal nervous system (mPFC, ACC, and hippocampus) control of the amygdala is weakened, this may cause anxiety. Our findings seem to support the hypothesis that damage to the dorsal nervous system may be caused by excessive iron accumulation. Pathologically, iron tends to accumulate in Lewy bodies (Castellani et al., 2000), which is a gradual neurodegenerative process that leads to the redistribution of iron in the brain in PD. Excessive iron accumulation has been suggested to cause oxidative stress in brain cells, resulting in DNA damage, mitochondrial damage, and lipid oxidative damage in dopaminergic neurons, ultimately leading to their death (Ostrerova-Golts et al., 2000; Sian-Hülsmann et al., 2011). Then, excessive iron accumulation in the dorsal nervous system disrupts the downregulation of the ventral nervous system rather than spontaneous overactivation of the ventral nervous system, which causes anxiety in PD patients. Therefore, we thought that the root cause of the high lifetime risk of anxiety in PD patients may be the redistribution of brain iron due to neurodegeneration.

Incidentally, widespread increased susceptibility alterations in the default mode network (DMN), including the precuneus, angular, mPFC, and ACC, were also found in this study. Decreased functional connectivity within the DMN has been reported in PD patients with anxiety (De Micco et al., 2021), but this contrasts with another study on social anxiety (Arnold Anteraper et al., 2014). The damage caused by the abnormal susceptibility alterations within the DMN may explain the difficulty of anxiety patients in focusing attention on the self and away from goal-oriented stimuli (De Micco et al., 2021). In addition, although striatal abnormalities have been discovered in many previous neuroimaging studies of anxiety in PD, whether morphological (Oosterwijk et al., 2018), functional (Wang et al., 2017), or metabolic (Ceravolo et al., 2013; Bayram et al., 2020), no related susceptibility changes were observed in this study. These striatal abnormalities may be related to secondary dopamine depletion resulting from damage to the mPFC and ACC.

The increased QSM value of the SMA may be associated with postural instability in PD (Pinson et al., 2022). The association between anxiety and motor symptoms in PD remains controversial (Prediger et al., 2012). Freezing of gait seems to be more strongly associated with anxiety in PD (Ehgoetz Martens et al., 2014; Gilat et al., 2018). However, in other studies of PD patients with anxiety, no similar relationship between anxiety and motor symptoms was found (Arnold Anteraper et al., 2014; Carey et al., 2020). Statistically, our results support the idea that anxiety has no influence on motor symptoms. We speculated that this difference may stem from the timing of anxiety symptoms in PD: anxiety as a precursor symptom before PD and anxiety that progresses continuously after PD may have different neural mechanisms (Dissanayaka et al., 2014).

It should be emphasized that, in this study, significantly increased iron deposition in the hippocampus was shown using the ROI-based QSM analysis, but no corresponding changes were found in the whole-brain voxel-based QSM analysis. We attributed this difference to the technical limitations of the voxel-based analysis: on one hand, the voxel-based analysis may lack the ability to reliably assess too small brain regions such as the SN and hippocampus; on the other hand, possible errors in spatial normalization may lead to some extent of spatial uncertainty in the results, although spatial smoothing had been applied to minimize this spatial uncertainty (Uchida et al., 2019). Therefore, while voxel-based QSM analysis is a comprehensive and convenient method of analysis, ROI-based QSM analysis can more sensitively and accurately assess susceptibility alterations in brain ROIs. Although we detected significant susceptibility differences in the mPFC and ACC between the PD-A and HC groups, no difference was observed between the PD-A and PD-NA groups. This may be because several subjects in the PD-NA group had HAMA scores close to the critical value of 12. In other words, the subjects in the PD-NA group could not be classified as the PD-A group but could possibly progress to anxiety according to the grouping criteria in this study, which may lead to reduced statistical differences. This conformed to the theory that the lifetime risk of anxiety is approximately 60% (Martínez-Martín and Damián, 2010) and is a possible explanation for the longer duration of PD in the PD-A group than in the PD-NA group in this study, which was consistent with previous research (Dissanayaka et al., 2014). Thus, we believe that the role of increased brain iron in the mPFC and ACC in the neural mechanisms of anxiety in PD should be valued.

This study has some limitations in addition to the ones listed above. First, limited by the comparatively small sample size, this study was only a preliminary exploration of the possible neural mechanisms of anxiety in PD, while a comparison between subtypes of anxiety in PD was impossible. Second, regarding previous studies (Wang et al., 2017), only the HAMA scale was used in the assessment of anxiety in this study, which was too simple to comprehensively assess anxiety severity. Future studies should use a combination of multiple anxiety scales. Third, although screenings were conducted and patients with obvious depression were excluded from the PD clinic, the HAMD score of PD-A subjects was statistically superior to that of other subjects, which was consistent with the high comorbidity of anxiety and depression in PD (Martínez-Martín and Damián, 2010). This may be related to some overlap in scoring rules between the HAMD and HAMA scales. Despite taking the HAMD score as a covariate, its effects on the results could not be completely ruled out. Fourth, the serum iron and transferrin levels of the subjects were not collected, which makes it impossible to study the relationship between serum iron and transferrin levels with iron deposition in the brain. In the future, we would add the collection of these clinical indicators in the enrollment of subjects. Fifth, due to the limitation of 3.0 T MRI in studying small structures in the brain (Wang et al., 2019), the application of 7.0 T MRI in this study may provide more information in the future. Finally, this was a cross-sectional study in which the follow-up of the subjects was not completed. Therefore, the longitudinal development of anxiety symptoms in patients with PD, especially PD-NA subjects close to the anxiety grouping criteria, could not be investigated at present. We would make a great effort to further study the neural mechanism of the occurrence and development of anxiety in PD after completing follow-up work.

Based on a review of the literature, this study, most likely for the first time, revealed the distribution pattern of brain iron and excessive iron accumulation in the fear circuit of PD patients with anxiety, providing a new approach to explaining the neural mechanism of anxiety in PD. However, the hypothesis proposed in this study may only be used to explain progressive anxiety after PD, and further studies are needed to explain the similarities and differences between the neural mechanisms of precursor anxiety and progressive anxiety in PD.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Ethics Committee of Wuxi People’s Hospital. The patients/participants provided their written informed consent to participate in this study.

XF, FW, XH, KaC, and LZ: study concept and design. KaC, LZ, HM, KeC, and YS: data collection. KaC: data analysis and manuscript writing. XF, XH, LZ, and XM: project development and manuscript revision. All authors contributed to the article and approved the submitted version.

This study was supported by the Jiangsu Province Natural Science Foundation (No. BK20191143, XF), the National Natural Science Foundation of China (No. 81271629, XF), the Medical Expert Team Program of Wuxi Taihu Talent Plan 2021, and the Top Talent Support Program for young and middle-aged people of Wuxi Health Committee (No. BJ2020015, XH).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as potential conflicts of interest.The reviewer YY declared a shared parent affiliation with the authors to the handling editor at the time of review.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.