Glymphatic system dysfunction in patients with cluster headache

Abstract Introduction The aim of this study was to investigate alterations of the glymphatic system function in patients with cluster headache. Methods We enrolled patients with cluster headache and healthy controls, and they underwent brain magnetic resonance imaging (MRI), including diffusion tensor imaging (DTI). We used the MRIcron and DSI studio programs for DTI preprocessing and DTI analysis with perivascular space (DTI‐ALPS) index calculation. Results Fourteen patients with cluster headache and 23 healthy controls were enrolled. The DTI‐ALPS indexes of the groups were significantly different. The DTI‐ALPS index for the patients with cluster headache was lower than that for the healthy controls (1.586 vs. 1.786, p = 0.044). There was a significant negative correlation between the DTI‐ALPS index and age in the patients with cluster headache (r = −0.549, p = 0.042). However, the DTI‐ALPS index was not associated with other clinical characteristics, including disease duration and headache intensity (r = −0.405, p = 0.150; r = −0.048, p = 0.869, respectively). Conclusion Patients with cluster headache had a lower DTI‐ALPS index than the healthy controls; this might indicate glymphatic system dysfunction in the patients with cluster headache. Further research is required to determine whether glymphatic system dysfunction is related to the pathophysiology of cluster headache.

The use of magnetic resonance imaging (MRI) to visualize glymphatic system function is critical for understanding the pathophysiology of various neurodegenerative disorders, and it can be a therapeutic target or be used to determine therapeutic efficacy. MRI has been used to visualize glymphatic system function in several ways, and the use of intrathecal gadolinium-based contrast agents (GBCAs) is the most representative. There have been reports of studies using MRI and patients with NPH, and the intracranial kinetics of GBCAs significantly differ between patients with NPH and controls (Edeklev et al., 2019;Ringstad et al., 2018). However, the method involving the use of intrathecal GBCAs is invasive, and the degree and speed of penetration vary from person to person (Ringstad et al., 2018). As a result, the application of intrathecal GBCAs to investigating glymphatic system function has been challenging and wrought with several limitations.
Diffusion tension imaging (DTI) has been recently utilized to assess glymphatic system function in several studies (Debaker et al., 2020;Yokota et al., 2019). DTI analysis with the perivascular space (DTI-ALPS) calculation method is based on the idea that perivascular ISF movement in the white matter near the lateral ventricles is prominent along the parallel medullary veins (Yokota et al., 2019). The DTI-ALPS method requires no GBCAs, uses clinically common DTI data, and can be performed retrospectively with the use of previous imaging data; therefore, it is highly useful for research.
Cluster headache is the most severe types of headaches, and they are characterized by ipsilateral pain localized to the orbital, supraorbital, and/or temporal areas, as well as associated autonomic processes. Lacrimation (tearing), conjunctiva injection (sclera redness), rhinorrhea, nasal congestion, hyperhidrosis (excessive sweating), and eyelid edema occur ipsilateral to the pain, with only 3% of individuals showing no autonomic symptoms (Sjaastad, 1992). They last for 15−80 min and occur up to eight times a day (Society, 2013). Sleep has been identified as a common trigger for cluster headache. A previous questionnaire-based study involving 275 patients with cluster headache showed that 80% of patients reported that sleep is a trigger for their headache episodes . Patients with cluster headache scored higher on the Pittsburgh Sleep Quality Index than healthy people, indicating a low self-perception of sleep quality. In addition, patients who report that sleep is a trigger for their headache attacks scored higher than those who report sleep is not a trigger factor . Furthermore, the majority of headaches occur at night during sleep, and the greatest issue for patients with cluster headache is sleep inefficiency (Jensen et al., 2007). Cognitive dysfunction does not last beyond headaches, but patients with cluster headache usually experience repeated cognitive deterioration (Favoni et al., 2018). Patients with Alzheimer's dementia also have glymphatic system dysfunction (Nedergaard & Goldman, 2020;Taoka et al., 2017).
Considering all these, and the evidence that the glymphatic system is activated especially during sleep, patients with cluster headache may have decreased glymphatic system function.
This study aimed to investigate alterations of the glymphatic system function in patients with cluster headache. The glymphatic system and F I G U R E 1 Selection process for the patients with cluster headache in this study cluster headache are both linked to sleep, and our research is based on the hypothesis that patients with cluster headache have glymphatic system dysfunction. To our knowledge, there have not been studies on this.

Participant: Patients with cluster headache and healthy controls
This study was conducted retrospectively in a single hospital and (2) Brain MRI including DTI. We excluded patients with the following: (1) previous history of any medical or neurological disease, except cluster headache, (2) structural lesion on brain MRI, or (3) having poor quality DTI for analysis. Figure 1 shows the selection process for patients with cluster headache in this study. We reviewed the medical records to obtain demographic and clinical characteristics, such as age, sex, disease duration (time interval from headache onset to MRI taken), headache intensity, and headache laterality (side with most headaches) for the patients with cluster headache.
We also enrolled age-and sex-matched healthy controls who had no previous history of any medical or neurological disease. All healthy controls underwent brain MRI including DTI, and they had no structural lesions on brain MRI. They had been already recruited from our previous study, which had enrolled a total of 150 healthy participants (Jang et al., 2017). Of the 150 patients, those who did not consent to the use of their data in this study were excluded. Finally, we included F I G U R E 2 Diffusion tensor image analysis with the perivascular space (DTI-ALPS) index calculation. The figure was generated from our previous study (Lee et al., 2021) the remained healthy participants in the control group with similar age and sex as the patients with cluster headache.

Diffusion tensor imaging acquisition
All MRI scans of the participants were obtained using the same MRI scanner, which was a 3 tesla with a 32-channel head coil (AchievaTx; Phillips Healthcare). From March 2018, routine brain MRI protocols at our hospital for patients with cluster headache included DTI. All of the MRI scanning was conducted during out-of-out phase of headache attacks in patients with cluster headache, and MRI scans were performed during the daytime in all participants (before 5 pm). The participants underwent the same brain MRI protocol, which included three-dimensional (3D) FLAIR imaging, coronal T2-weighted imaging, 3D T1-weighted imaging, and DTI. FLAIR and T2-weighed imaging were used to evaluate the structural abnormalities of the brain. The detailed MR parameters for DTI with 32 different diffusion directions have been described in our previous studies (Lee et al., 2021)

Diffusion tensor imaging preprocessing and diffusion tensor image analysis along with the perivascular index calculation
We used the MRIcron and DSI studio programs for DTI preprocessing and DTI-ALPS index calculations. The steps for DTI preprocessing and DTI-ALPS index calculation were described in our previous study (Lee et al., 2021). The preprocessing for DTI included setting up a mask, reconstruction with the DTI method, and fiber tracking. We, subsequently, calculated the DTI-ALPS index (Figure 2) (Lee et al., 2021;Taoka et al., 2017). We drew a rectangular region of interest (ROI) and obtained the fiber orientation and diffusivities of the three directions along the x-, y-, and z-axes as voxel levels at the ROI. Finally, the DTI-ALPS index was obtained using the following formula (Taoka et al., 2017): where Dxxproj is diffusivity along the x-axis in the projection fiber, Dxxassoci is diffusivity along the x-axis in the association fiber, Dyyproj is diffusivity along the y-axis in the projection fiber, and Dzzassoci is diffusivity along the z-axis in the association fiber.

Statistical analysis
A primary analysis of these data was conducted using an a priori test.
No statistical power calculations were performed before the study.

Demographic and clinical characteristics of the patients with cluster headache
We enrolled 14 patients with cluster headache and 23 healthy controls.

Differences between the diffusivities and diffusion tensor image analyses along with the perivascular indexes of the patients with cluster headache and healthy controls
The diffusivities along the x-, y-, and z-axis of the projection, association, and subcortical fibers of the patients with cluster headache and healthy controls were not statistically different ( Table 2).
The DTI-ALPS indexes for the groups were significantly different.
The DTI-ALPS index for patients with cluster headache was lower than that for healthy controls (1.586 vs. 1.786, p = 0.044).

Correlation analysis of diffusion tensor image analysis with the perivascular index and demographic/clinical characteristics
There was a significant negative correlation between the DTI-ALPS index and age in patients with cluster headache (r = −0.549, p = 0.042) ( Figure 3). However, the DTI-ALPS index was not correlated with other clinical characteristics, including disease duration and headache intensity (r = −0.405, p = 0.150; r = −0.048, p = 0.869, respectively).

DISCUSSION
The following were the main findings of this study.
(1) The DTI-ALPS indexes of the patients with cluster headache and healthy controls were different. The DTI-ALPS index was lower for patients with cluster

F I G U R E 3 Correlation analysis of diffusion tensor image analysis
with the perivascular space (DTI-ALPS) index and age in patients with cluster headache. There was a significant negative correlation between the DTI-ALPS index and age in patients with cluster headache headache than for healthy controls.
(2) There was a significant negative correlation between the DTI-ALPS index and age for the patients with cluster headache. (3) The DTI-ALPS index was not correlated with other clinical characteristics such as disease duration or headache intensity for patients with cluster headache.
This was the first study to evaluate glymphatic system function in patients with cluster headache. The lower DTI-ALPS index of patients with cluster headache than that of healthy controls in this study implies that patients with cluster headache have glymphatic system dysfunction. Glymphatic system dysfunction is important in the pathophysiology of neurovascular, neurodegenerative, neuro-inflammatory diseases, and traumatic brain injury, as well as age-related changes in brain functions (Sun et al., 2018). There are also reports of the potential role of the glymphatic system in headache disorders, especially migraine (Messlinger, 2018;Schain et al., 2017;Toriello et al., 2021;Vgontzas & Pavlovic, 2018). Cortical spreading depression, the neural event underlying migraine aura, results in a temporary impairment of glymphatic flow in a mouse model via paravascular space closure for several minutes after inducing cortical spreading depression, followed by a gradual recovery over half an hour (Schain et al., 2017;Toriello et al., 2021). In addition, calcitonin gene-related peptide plays a critical role in the pathophysiology of headaches, particularly migraine. The calcitonin gene-related peptide may be excreted in three TA B L E 2 Differences between the diffusivities and diffusion tensor image analysis with the perivascular space (DTI-ALPS) index in patients with cluster headache and healthy controls compartments: venous blood plasma, cerebrospinal fluid, and possibly the lymphatic system. Thus, glymphatic system dysfunction may be associated with increased levels of calcitonin gene-related peptide, which may produce a headache (Messlinger, 2018;Toriello et al., 2021) Sleep is often associated with cluster headache. According to a polysomnographic study, patients with cluster headache have a lower proportion of sleep stage R (previously referred to as REM sleep), and a long latency for sleep stage R .
Another study also demonstrated longer sleep latency and lower sleep efficiency in patients with cluster headache compared to healthy controls (Lund et al., 2019). Furthermore, numerous sleep disorders have been linked to cluster headaches (Weintraub, 2003). Cluster headaches have been treated with melatonin and other treatments that affect circadian rhythm. Cluster headache patients may develop obstructive sleep apnea (Weintraub, 2003). A study involving mice demonstrated that natural sleep is linked to a 60% increase in the interstitial space of the brain. Because of the high hydraulic resistance in the brain interstitial space during alertness, sleep can increase the convective flow of CSF plus ISF; therefore, there is an increase in the efficiency of the glymphatic system during sleep (Xie et al., 2013).
Several patients with cluster headache have attacks that wake them up, but they usually have sleep difficulties before the attacks, and sleep disorders are common concomitant diseases for patients with cluster headache (Martins, 2015). In addition, cluster headache frequently begins during sleep . Thus, sleep disturbance in patients with cluster headache is thought to diminish the efficacy of glymphatic system function by lowering the flow of CSF and ISF.
Aging is closely related to sleep; therefore, sleep quality deteriorates with age. With aging, insomnia becomes more common, the total duration of sleep becomes shortened, and sleep becomes more disrupted. More importantly, older people rarely reach the deep sleep stage N (previously referred to as non-REM sleep). In adults aged >60 years, the sleep stage N is predominantly shallow, consisting of the shallower sleep stage 1 or 2 (Landolt & Borbély, 2001). As a result, the older brain spends less time in sleep stage 3, potentially resulting in a catastrophic decrease in brain waste clearance, as the efficacy of glymphatic fluid transport is directly proportional to the prevalence of slow-wave activity (Hablitz et al., 2019). These sleep abnormalities in patients with cluster headache are comparable to those observed with age, and this is thought to represent a mechanism of glymphatic system dysfunction. The CSF formation rate and turnover in the rat brain decrease with age, which is consistent with evidence showing that glymphatic clearance decreases with age (Kress et al., 2014). In this study, there was a negative correlation between the DTI-ALPS index and age in patients with cluster headache, which is consistent with the results of previous animal experiments.
The limitations of this study are as follows. First, this was a retrospective study. The study confirmed that alterations in glymphatic system function were identified in patients with cluster headache; however, it is not clear whether this change is the result or cause of cluster headache. Second, there was a limited number of participants; therefore, it is difficult to generalize the results. However, cluster headache is not common, and we only enrolled newly diagnosed patients with cluster headache to exclude medication effects. Further studies with large sample size are needed to confirm our findings.
Third, various factors can influence the DTI-ALPS index, including the imaging plane, head position, and motion-proving gradients (Taoka et al., 2021). However, a previous study found that the results of the DTI-ALPS method were significantly related to glymphatic system function as measured by MRI after intrathecal administration of GBCA (Zhang et al., 2021), and another recent study successfully demonstrated good test-retest reproducibility and robustness (Taoka et al., 2021). Despite these limitations, this study is notable because, to the best of our knowledge, no studies have investigated the link between the glymphatic system and cluster headache.

CONCLUSION
Patients with cluster headache had a lower DTI-ALPS index than the healthy controls. This might indicate glymphatic system dysfunction in patients with cluster headache. Further research is required to determine whether glymphatic system dysfunction is related to the pathophysiology of cluster headache.

ACKNOWLEDGMENTS
This work was supported by a 2019 Inje University research grant.