Altered functional connectivity of the right caudate nucleus in chronic migraine: a resting-state fMRI study

The CN is a deep cerebrum nucleus with C-shaped structure that consists of the head, body, and tail. It has a direct anatomical connection to the peripheral BG structures and exact fiber projection to the broad regions across the cerebral cortex. These connections engage in the formation of cortico-striato-thalamo-cortical circuits (CSTC) [27]. The striatum is an important part of the CSTC circuits, which consists of the CN and putamen and includes three parts (limbic, associative or cognitive, and sensorimotor areas). Each part creates a unique projection pattern by receiving projections from a different cerebral cortex [2]. The most ventral parts of the CN as one of the parts of the limbic area of the striatum, which mainly receives fibers from limbic and paralimbic cortices [28]. The associative or cognitive area has fiber connections with the frontal, parietal and temporal lobes, containing most of the CN [27, 29]. Similarly, the sensorimotor area receives the projection from the primary motor and somatosensory cortices, containing the dorsolateral edge of the head of the CN [27]. In addition, CN and putamen receive axons from almost all parts of the cortex excluding primary visual, auditory, and olfactory cortices [30]. The CN is involved in the regulation of emotion, cognition (memory, language, visuospatial, executive, computational, and comprehension judgments), movement, and sensation by connecting with these brain regions. Pain is a multidimensional composite with a highly specialized sensory experience [31]. Some evidence suggests an involvement of the CN in pain modulation by affecting the affective and cognitive processing of pain [32] and the CN might be regulated by the bioactive substance (arginine vasopressin, oxytocin, and so on) from the other brain regions, affecting the pain perception [33, 34]. Considering the CN has a complex and diverse function and the brain areas implicated in these projections correlate with the symptoms of CM, we selected the CN as a seed to explore whether it is involved in the pathophysiology of CM.

Cognitive–emotional dysfunction is prevalent in patients with CM, and emotions and different cognitive states (including the aspects of attention and memory) have been shown to modulate pain perception [3]. In our study, CM patients exhibited dysfunction in brain regions belonging to the CSTC pathway compared with HCs. In the CSTC circuits, the cortical axonal fiber projections to the striatum and enters the BG to the thalamus via direct and indirect pathways which recurrently terminates to the brain cortex [35], relating to emotional-cognitive modulation and pain processing.

The ACC and insula are mainly thought to be components of the limbic (emotional) system and are essential in coding the affective and motivational aspects of pain [36]. They are also a part of the somatosensory cortex, which perform the function of cognitive processing [37]. Our results confirmed that both the ACC and the insula affected the CN during pain perception, manifesting as enhanced FC between the CN and these brain regions. Prior findings concluded that the ACC and the CN formed an essential sequential pathway for processing pain avoidance behavior [38], which further implies the ACC and CN are associated with pain modulation. And the ACC is linked to the evaluation and expression of negative emotion [39], it has been found to show structural and functional changes in CM patients. CM patients tend to suffer from negative emotions or irritability when a headache occurs, and the negative emotional state often increases pain. This may explain why recurrent migraines are often combined with depression and anxiety disorders. People with CM have a 3.8 times higher risk of being depressed compared to HCs and approximately 37% of them have a psychiatric disorder [40]. The ventral CN with more interconnection to the limbic system, they are the part of the limbic CSTC that is implicated in affective functions [41]. This explains the mechanisms underlying our results, as the dysfunction of the limbic CSTC circuit could lead to abnormal clinical manifestations in CM. Nevertheless, other brain regions involved in this circuit did not show significant differences in our study. The other complex functions performed by these regions as well as the small sample size may explain it. The CSTC circuits also play a role in Parkinson’s disease and obsessive-compulsive disorders (OCD) [16, 42], and several studies have reported clinical observations regarding the interplay of migraine and the extrapyramidal system. Some findings have demonstrated that patients with BG disorders are more likely to suffer from migraines and suggested that these CSTC circuits can modify the course of migraine [16]. Taken together, these findings suggest that the dysfunction of this system not only leads to the occurrence of motor system disorders, but also chronification of migraine. Further studies need to focus on the role of this circuit in CM.

Previous studies have shown that CM patients with higher subjective and objective cognitive impairment mainly manifested the most striking deficits in memory/delayed recall (65.3%) [40]. In our study, CM patients showed enhanced FC between cognition-related brain regions and the right CN compared with HCs. These brain regions include the ACC, insular cortex, superior and inferior parietal gyrus, precuneus, and orbitofrontal cortex (OFC). These brain areas are jointly involved in cognitive regulation through projections to the CN and are also included in the cognition-related CSTC circuit. Damage to the CN has been known to cause cognitive impairment [43]. The parietal lobe was thought to have an association with the function of basic attention, language, and social cognition [44]. The precuneus belongs to the default mode network (DMN) that regulates higher-order cognitive function [14]. The OFC belonged to the frontostriatal circuits and importantly influences higher cognitive brain function [45]. A recent study has reported decreased intrinsic rs-FC in the cognitive networks such as the DMN, salience network, and central executive network (CEN) of CM patients [14]. In contrast, our results showed enhanced FC between the CN and cognition-related brain areas. Such inconsistent findings are likely due to the variations in sample recruitment, the proportion of gender, selection of seed, and the analysis methods. Our results suggest that the synchronous changes in BOLD signals between these brain regions may cause cognitive dysfunction in CM.

Our study found high FC values between the CN and brain regions related to sensory perception (including the calcarine cortex, lingual gyrus, temporal gyrus, temporal pole of the superior temporal gyrus, and inferior parietal gyrus) in CM patients compared to HCs. Previous MRI studies of CM patients have illustrated that corresponding brain networks with dysfunction primarily include the sensory networks (auditory network, AN; and visual network, VN). Enhanced FC in brain areas within the VN and AN implies an increased visual or auditory stimuli may potentiate the prominence of pain inputs, causing increased pain sensation [46]. In addition, neurons sensitive to visual, auditory, and somatosensory modalities have been identified in the CN and the lesions of CN can induce behavioral disturbances and over-reactivity, suggesting a failure of the inhibitory modulation of sensory inputs [36]. Visual signals transmitted through the retina are transmitted progressively along the occipito-parietal and occipito-temporal pathways [47]. The calcarine cortex and lingual gyrus as the pivotal areas for the visual pathway are associated with the perception of phosphenes [48], visual processing, and spatial memory [49]. Enhanced FC of the CN with the calcarine cortex and lingual gyrus has been presented among CM patients in our study, which may reflect disrupted visuomotor function and the failure of the CN in modulating sensory inputs. These were consistent with certain findings that impairments of visuomotor speed processing were observed in migraine patients with a higher frequency of headache attacks and a longer course of the disease [50]. The temporal pole is involved in comprehensive multisensory processing and integrates visual, auditory, olfactory, and somatosensory stimuli [51]. The region ranging from the inferior temporal visual cortex to the tail of the CN forms the visual corticostriatal loop that processes visual information and contributes to working memory and cognitive strategy selection [44]. As described in the diagnostic criteria, photophobia and phonophobia are two key non-head pain symptoms that are indicative of migraine [52]. The visual pathway and some related regions may lead to the symptoms of migraine. Therefore, our findings suggest that abnormal FC among the sensory-related regions reflects deficits in sensory processes in CM patients.

FC values between the CN and the ACC, MCC, lingual gyrus, and precuneus were also higher among CM patients than in EM patients. Patients with CM reported more somatic complaints—including fatigue, sleep disturbances, and nausea—compared with EM patients [53]. Moreover, CM patients are more susceptible to exhibiting comorbid psychiatric disorders, such as depression, anxiety, and OCD [40]. Our results provide further support for these conclusions. However, unlike in previous studies, we failed to find any marked differences in the CN-based FC between EM patients and HCs. This may be related to our seed selection. The seeds used in previous studies were mostly spherical regions in the CN. The neurons of these seeds were responsible for specific functional aspects in the EM population and showed significant differences compared with those in HCs. However, we used the entire CN based on the AAL template as the seed, representing the total function of the CN. Therefore, differences in CN-based FC values were only apparent between CM patients and HCs, not between EM patients and HCs. Similar patterns have been reported in a previous study showing that lower levels of GABA/Water and GABA/Cr were only observed in CM vs. HC comparisons, and not in EM vs. HC comparisons [25].

Several imaging studies have reported specific brain changes in MOH patients that help differentiate MOH from other types of headaches [54]. However, an increasing number of researchers consider MOH to be a complication of other headaches as well [6‐9]. CM patients with a high frequency of migraine attacks tend to overuse acute analgesics, which increases the risk of MOH. MOH may also contribute to the progression of CM. Our results indicated that the combined MOH may not lead to significant changes in CN-based FC in CM. An FCD study reported that compared with EM patients, MOH patients showed lower FCD in the right CN with a fronto-temporal-parietal distribution pattern [21]. The results of the FCD analysis suggested that the FC in the CN decreased in the MOH population without CM but increased in the EM population. In our study, the combined MOH was based on the CM diagnosis that means MOH developed from CM. We did not include the other primary headache and a subgroup of patients with MOH evolve from other headache types. Therefore, we can only conclude that MOH that developed from CM may not influence the FC of the CN. Some evidences showed that CM patients with MOH had improved FC between the prefrontal cortex and regions of the CN and frontal cortex within the CEN after treatment[55]. A magnetic resonance spectroscopy study suggested that there was no difference in neurochemical levels between CM patients with and without MOH [25]. Our findings further support that MOH is only comorbidity of CM and the presence of MOH would not change certain intrinsic properties of CM.

Correlations between FC values of the CN and the characteristics of CM remain unclear. Our results showed that FC between the bilateral calcarine cortex, bilateral lingual gyrus, and CN were positively correlated with BMI. One study found that a higher BMI was associated with lower FC in the caudate-dorsolateral prefrontal cortex and higher FC in the caudate-medial temporal lobe [56]. The CN is a part of the reward system that is closely connected to BMI, and a higher BMI is considered to be associated with poorer cognitive flexibility. Previous studies have observed higher rs-FC values between the sensorimotor network and DMN in individuals with obesity [57]. Moreover, obesity is also a critical risk factor for migraine progression [58, 59]. Based on these findings, we speculate that the dysfunction of right CN-based FC with the visual cortex can damage the processing of visual information and cause abnormal sensory and cognitive impairments in CM patients, and these dysfunctions may be related to obesity. Susceptible populations with higher BMI scores may progress to CM earlier and exhibit more severe symptoms.

The VAS scores for headaches were inversely correlated with right CN-related FC in the left calcarine cortex and right lingual gyrus. The CN has been reported to play an essential role in sensory processing and pain inhibition [60]. The visual cortex can independently encode and differentiate visual cues related to pain anticipation [61]. And the pain in turn can alter visual cortex excitability that might reflect defensive strategies against pain [62]. Based on this, we believe that the functional network of the visual cortex and CN is involved in pain perception and the regulation of CM.

This study has three principal limitations that need to be addressed in future studies. First, the sample size was relatively small, and the included population may not fully represent the characteristics of the entire patient population. We seek to enroll a larger population in future studies to further explore the mechanisms of CM. Second, we selected the bilateral CNs as the region of interests. The CN is an elongated C-shaped nucleus that consists of three subregions: the head, body, and tail. Different subregions of the CN may have unique functions; however, we have not discussed the FC of each part separately. Third, this was a cross-sectional study, and the results did not indicate any specific causal relationships between the altered FC of the CN and the formation of CM. In the future, we will conduct follow-up visits, connect better clinical information, and implement a better study design to address these limitations.