Gradually shifting clinical phenomics in migraine spectrum: a cross-sectional, multicenter study of 5438 patients

The ratio of MwoA to MwA was 5.9:1 in our study, which is quite consistent with the results of our previous study [20]. The ratio is also similar to the Taipei Area (7.0:1) [21] and Japan (7.4:1) [22], but different from Europe (1.6:1 ~ 3.8:1) [23‐26]. This variation could be due to a genetic predisposition.

Our subjects revealed a significant female preponderance. Another study in a single center in southwestern China found a similar gender distribution from outpatient population [27]. Previous population-based research in Asia and Europe have also revealed that the MwoA group has a high female/male ratio (4.9:1 ~ 2.2:1) [4, 21, 22, 25]. Notably, MwA revealed a closer sex ratio compared to MwoA.

The distribution of age and age at onset showed an excess of female patients with MwoA in reproductive age, and female patients with MwA were more common in younger and older populations, which is consistent with previous reports [4, 27]. Ovarian steroid hormones are also an important trigger in migraine. We observed a clear difference in the percent of headache attacks induced by menstruation or pregnancy between MwoA and MwA patients, which is consistent with prior data [28, 29]. MwoA seemed to alleviate more during pregnancy than MwA, according to Granella et al. [29]. All of these findings suggest that estrogen fluctuation is more important in MwoA than MwA itself. According to previous researches, estrogen affects neuroexcitability and regulates the occurrence of migraine by affecting neurotransmitters such as 5-hydroxytryptamine (5-HT) [30], γ-aminobutyric acid (GABA) [31], calcitonin-gene related peptide (CGRP) [32], opioids [33], and norepinephrine [34], and these neurotransmitters are primarily distributed in the raphe nuclei pontis, periaqueductal gray (PAG), ventral tegmental area (VTA), hypothalamus, and the ventromedial side of the thalamus, which suggests that MwoA is more closely related to the hypothalamus, thalamus, and brainstem than MwA.

MwoA patients were more likely to be connected with triggers than MwA patients, which was consistently reported [35, 36]. Some researchers found that patients with both MwoA and MwA had more triggers for MwoA attacks than WMA attacks [37]. According to previous findings, sleep-wake cycle disturbance [38], changes in mood [39], stress [40], and fears to specific odor [41] are closely related to hypothalamic function, limbic structures and descending modulation of the brainstem. The difference of triggers suggested different activity levels of these structures between the two subtypes. Strong light was the only trigger with a higher incidence in MwA patients in the current study, which is consistent with previous research [4]. This phenomenon may be due to the hyperresponsiveness of visual cortex in MwA [6]. However, some researchers argued that flickering lights were more likely to be a premonitory symptom associated with abnormal brain activity rather than a trigger [42]. Therefore, the high incidence of strong light as a trigger in patients with MwA may correlate with the high incidence of premonitory photophobia.

The PSs are the beginning of alternations in brain function prior to headache attack [42], which suggests that some brain areas are initiatively activated. Based on our results, MwA patients were more likely to be accompanied by PSs, and the most common PSs in MwA and MwoA were different. Neck stiffness, as the most common symptom of MwoA, is associated with activation of the periaqueductal gray, trigeminocervical complex, hypothalamus, and thalamus [43, 44]. Photophobia and phonophobia are more common in MwA and are more likely to be associated with cortical dysfunction. Therefore, the differences in PSs suggest that brain areas involved in MwoA and MwA are not exactly the same, which may provide important insight into the mechanisms of migraine.

Previous researches on migraine chronicity considered migraine as a spectrum disorder, in which the clinical and pathophysiological features of migraine may progress over time [45, 46]. Based on the multidimensional clinical information of distinct migraine subtypes, the current investigation looked into whether migraine was a spectrum condition. According to our results, accompanying symptoms, triggers, and migraine-like headaches were more common in MwoA. Patients with MwoA generally suffered a more severe headache, a higher frequency, and a longer course than MwA. The correlation analysis found that headache characteristics, including accompanying symptoms, triggers, and migraine-like headache, positively correlated with the intensity, frequency, or course of headache in MwA patients. However, these positive correlations did not occur among MwoA patients. Therefore, we speculated that patients with MwA were heterogeneous.

FAMD put participants into three groups, each representing one of three migraine subtypes (MwoA, AWM, and AWNM). The green dots presented AWM patients who were positioned between the yellow AWNM and the red MwoA, and the AWNM and MwoA boundaries were relatively apparent (Fig. 5). These findings suggest that AWM is a compromise between AWNM and MwoA. The differences in basic headache aspects between MwoA and MwA were also noticed when AWM and AWNM were compared. Patients with AWM were more similar to those with MwoA, and the differences between MwoA and AWNM were clearer. And the chaotic MwA data affected the decision tree’s accuracy in identifying MwoA from MwA, according to decision tree analysis. When MwA was separated into AWM and AWNM and compared, the Youden’s index of the decision trees grew dramatically.

These findings suggest that MwA patients should be separated into AWM and AWNM groups, with AWM as a transitional condition. Changes in headache features develop gradually from MwoA to AWM and AWNM. More research is needed to investigate whether patients with MwA should be reclassified in the further diagnostic criteria based on clinical manifestations with vs. without migraine-like headache, as defined in the ICHD-II [19]. Understanding these subtypes could be crucial in guiding future migraine research and improving therapeutic outcomes.

The pathophysiology of aura is commonly acknowledged CSD [47], which is a spontaneous activation of the brain that propagates slowly and continuously. However, the possibility that silent CSD is the underlying etiology of MwoA is much less certain [48]. The hypothalamus, a critical integrator between external stimuli and internal metabolic and endocrinological variables, has been suggested to play a pivotal role in the genesis of headache attacks in MwoA [49, 50]. Internal and external factors may lead to alterations in hypothalamic activity, which may influence pain processing pathways and facilitate trigeminal pain perception, resulting in a headache attack [50‐52].

Our findings showed that MwoA patients had more severe headaches, a stronger link to estrogen, were more easily produced by external causes, and were more likely to have accompanying symptoms and premonitory symptoms. As previously stated, all of these differences suggest that MwoA has a closer relationship to subcortical brain dysfunction, such as the hypothalamus, thalamus, and descending modulation system of the brainstem, than MwA, especially AWNM, and that MwA is more closely aligned with cortical dysfunction. We hypothesize that in MwA and MwoA, the activation degree of different brain regions or neural networks may be different. Different subtypes may have different “predominant regions” in the brain that generate migraine attacks. Notably, the existence of an intermediate type of AWM between MwoA and AWNM suggests that the two generators are not entirely separated, but that a progressive process linkage may exist. Therefore, we tend to consider migraine as a spectrum disorder, with MwoA and AWNM as the two poles.

One strength of this multicenter study is the large number of subjects with multidimensional data on headache characteristics. To our knowledge, this study is the first application of factor analysis of mixed data and decision tree to patients with migraine to distinguish subtypes and investigate the dominant features. Although the probable migraine probably lies on the spectrum of migraine, we excluded the subtype in our study to obtain a more obvious consequence. There are also some limitations in our research. This research was based on outpatient data, not a population-based survey. While patients with more severe symptoms or higher educational attainment are more likely to seek hospital care for their headache and agree to participate the survey. And the study collected the headache data in outpatient clinics without the use of headache diaries, which cannot avoid recall bias. Some patients with both MwA and MwoA, may confound the characters of headache attacks. As we known, a very small percentage of patients have more than one form of headache, per prior studies. Only 9% of Danish migraine patients had both MwA and MwoA, compared to a total of 36% who had MwA only [53]. In comparison to the population of Europe, East Asians have a much lower prevalence of MwA. The ratio of MwoA to MwA was 5.9 ~ 7.4:1 in East Asian population [20‐22]. Therefore, the proportion of comorbidities with both MwA and MwoA in our study should be smaller than that in European population, further minimizing its influence on the study’s findings. In addition, this study was a cross-sectional study that could not track changes in clinical manifestations of migraine patients.