The present study aimed to examine differences in the FC patterns of MwoA-A patients and ascertain whether a disrupted FC may be a predictive marker of anxiety comorbidity. The results showed that the MwoA-A patients had significant regional changes within the visual cortex, which were correlated with anxiety. In addition, dysfunction in cognitive networks was associated with the frequency of headaches. These findings therefore suggested that the anxiety comorbidity is associated with a disrupted FC within many large-scale networks and may provide new insights on the pathophysiology of migraine with psychiatric disorders.
MwoA-A vs MwoA-OA
The MWOA-A patients in this study exhibited some unique functional changes, further supporting the distinct neural substrates of psychiatric disorders in episodic migraineurs. Specifically, the MwoA-A patients exhibited higher ReHo values in the left IPS and decreased FC between the right LG and right IPS, compared to the MwoA-OA participants. However, such abnormalities were not observed in both migraine subtypes compared to the HCs. The IPS is believed to constitute the attention network that can be activated by top-down attention in multiple sensory stimuli, including nociceptive signals[13, 14]. In addition, the findings herein corroborated with those from a previous arterial spin labeling MRI study where alterations in cerebral blood flow in the inferior parietal lobe were shown to have a significant positive correlation with an anxiety disorder in episodic migraine[15]. Moreover, Balderston et al.[16] reported that reducing IPS excitability was sufficient to reduce the physiological arousal related to anxiety, consistent with the findings in the present study. Therefore, this pattern of dysfunction in the IPS may account for the specific psychiatric features observed in migraineurs with anxiety comorbidity.
In addition, the MwoA-A patients exhibited special FC disruption at a large-scale network level within the visual cortex, compared to their MwoA-OA counterparts. Although no study has to data compared migraine patients with and without anxiety comorbidity, previous studies demonstrated that the visual cortex is associated with modulation of pain and psychiatric disorders[8, 10]. For instance, a resting-state study that combined independent component analysis and regions-of-interest FC analysis elicited that enhanced FC between the thalamus and visual cortex was associated with anxiety symptoms[10]. Notably, the thalamus plays a crucial role in the trigeminovascular pathway and relays sensory information to multiple cortical networks[17]. Therefore, dysfunctions in the thalamus were shown to be responsible for disordered thinking and the anxiety network[18] . Moreover, abnormal FC between the amygdala and visual cortex was reported to be significantly correlated with the duration of migraine. The amygdala is a crucial region involved in affective modulation[8]. Overall, functional abnormalities in both the intra- and inter-visual cortex may help in distinctly characterizing migraineurs who are more prone to display overt anxiety behavior.
MwoA-A and MwoA-OA vs HCs
In the present study, both the MwoA-A and MwoA-OA patients exhibited similar levels of regional dysfunctions in the brain, as shown by the decreased ALFF and ReHo values in the right LG, relative to the HCs. More specifically, the neural activity of the right LG in the MwoA-OA patients was more deteriorated and was associated with anxiety. These results therefore suggest that the LG, which is a key node in the visual cortex, might be involved in nociceptive and emotional processing. To support this hypothesis, the involvement of the LG in the modulation of pain and psychiatric disorders has been confirmed through different fMRI methods[8, 19-21]. In addition to shared regional functional deficits, both the MwoA-A and MwoA-OA patients also exhibited high FC between the right LG and the ipsilateral MCC and SFG, which are core regions of the central executive network (CEN) and limbic system, respectively. These results therefore suggest that significant increase in FC between the right LG and the CEN as well as the limbic system could represent fundamental neural substrates for MwoA during the interictal period.
Notably, the CEN, including the dorsolateral prefrontal cortex and lateral posterior parietal cortex is crucial for many high-level neurocognitive functions such as cognitive performance, working memory and decision making[13, 22, 23]. In addition, the MCC is an important component of the cingulate-insular pathway which gates and maintains nociceptive hypersensitivity in the absence of conditioned noxious stimuli[24] and affects the impact of headache in the migraine-free period[25]. Coupled with the aforementioned studies, the current findings suggest that the visual-related patterns of FC with the CEN and limbic system are associated with cognitive function and top-down regulation of pain. Furthermore, a pain-cognition interaction model revealed that chronic migraineurs with no exposure to painful stimuli had lower neural activity within the dorsolateral prefrontal cortex (DLPFC) and MCC compared to the HCs, but had higher neural activation in the presence of painful stimuli compared to pain-free patients[22]. Therefore, the disrupted FC patterns of the LG may direct more focus on feelings of displeasure along with recurrent headache stimuli, and further contribute to the impairment of cognitive networks and the generation of negative emotions.
M-OA vs HCs
Migraine and anxiety are both affected by sensitization of the central nervous system and are more sensitive to extrinsic stimuli. In addition, previous studies proved the hypothesis that migraine and psychiatric disorders may share common central nervous circuits, such as the neurolimbic network[26] and the cortico-limbic circuit[27],underlying vulnerably to nociceptive and affective stimuli[28]. Given the high comorbidity of anxiety and depression in migraine patients, it would be reasonable to consider that certain features of migraine may influence the association with anxiety. A common feature of chronic neuropathic pain is the appearance of cutaneous allodynia, which was shown to be associated with an increase in self-reported anxiety symptoms[29] and an independent risk factor of clinical anxiety[30]. Additionally, cutaneous allodynia is characterized by central hypersensitization and exacerbation in response to innocuous somatosensory stimuli[31]. Notably, some resting-state FC analysis[31, 32] showed that cutaneous allodynia was involved in the descending pain modulation pathway, including the pons, thalamus, posterior cingulate cortex, precuneus, inferior temporal cortex, middle frontal cortex and occipital cortex, in MwoA patients. These results corroborate our findings that MwoA-OA patients exhibited increased visual-related FC with the right PCC/PCUN, left MFG and left ITG compared with HCs, and make our study more stable and convincing clinically.
Moreover, the PCC/PCUN and MFG are core regions in the default mode network (DMN) and CEN, respectively, involved in self-monitoring and task processing. The DMN appears to play a dominant role during endogenous neural activity and is deactivated during the processing of exogenous cognitive tasks. Tseng[33] showed that interindividual differences in anxiety were associated with dysfunctions in the DMN especially during the encoding of painful processes. This suggested that the DMN may not be a neural substrate for the modulation of pain, but also involves a strong affective component. Additionally, Zou et al.[34] revealed that an altered DMN function could represent the relief of clinical symptoms, such as the intensity and frequency of headaches following acupuncture, consistent with previous findings by our research group[25]. Furthermore, a shift in DMN connectivity was illustrated to modulate the pain effect and evaluate the therapeutic effect. Moreover, the MFG is located in the DLPFC and plays an important role in preventing pain, through top-down inhibition pathway. Schwedt et al.[35]also revealed that enhanced pain-induced activation of DLPFC in migraineurs was positively correlated with the frequency of headaches, contrary to the results obtained in the present study. The discrepant results might be attributed to patient heterogeneity, migraine phases and study methodology. Additionally, the regional neural activity of the MFG was reported to be effective in predicting the treatment efficacy of pain and anxiety through drug therapy[36] and transcranial magnetic stimulation[37].
In this study, the visual FC imbalance with the DMN and CEN detected in MwoA-OA patients exhibited significant correlations with the frequency of headaches, which may indicate the functional deficits to compensate recurrent stimuli. In addition, the ITG is considered to be part of the DMN[38] and has been shown to be associated with the deterioration of pain, consistent with the PCC/PCUN[39, 40]. According to the above mentioned studies, the abnormal visual-related FC patterns may account for some clinical and psychiatric symptoms in MwoA patients, such as the frequency of headaches, cutaneous allodynia and anxiety comorbidity. They may also be involved in pain-inhibiting and anxiety-relieving treatments.
The present study has several limitations that should be considered. First, given the inadequate and subjective measurement of anxiety scores, the possibility of a classification bias cannot be ruled out. Second, the present study did not include cognitive function assessments and other psychiatric disorders. Future studies should therefore investigate the relationship between the cognitive deficits and psychiatric disorders in migraineurs. More studies are also required to assess the reciprocal effects of cognitive function and psychiatric disorders in migraine. Finally, there was no difference in the level of education between the two migraine groups, while the HCs showed a higher level of education than the other two groups. Although the main findings were obtained after controlling for the level of education, the potential confounding effects of participants' education on the results can not be ruled out completely.