Prevalence and Characteristics of Cutaneous Allodynia in Migraine and Probable Migraine: A Population-Based Study

Background: Cutaneous allodynia (CA) is a common feature of migraine and a clinical marker of central sensitization. Probable migraine (PM) is a subtype of migraine that fulls all but one criterion of migraine. The present study aimed to evaluate the prevalence and characteristics of CA and those of migraine in PM. Methods: We used the data of the Korean Sleep-Headache study, which was a nation-wide population-based study on headache and sleep. CA was evaluated using the Allodynia Symptom Checklist-12 (ASC-12) questionnaire with ASC-12 score ≥ 3 classied as CA. Results: Of 2501 participants, the prevalence of migraine and PM were 5.0% and 11.6%, respectively. The prevalence of CA did not signicantly differ between migraine and PM (16.0% vs. 14.5%, respectively, p = 0.701). Individuals with PM with CA reported a higher monthly frequency of headache (3.3 ± 4.3 vs. 1.8 ± 3.6, respectively, p = 0.044), more severe intensity of headache (Visuals Analogue Scale, median and interquartile range, 6.0 [4.0–7.0] vs. 5.0 [3.0–6.0], respectively, p = 0.002), and higher impact of headache (Headache Impact Test-6, 56.3 ± 7.2 vs. 48.3 ± 8.0, respectively, p < 0.001) and disability (Migraine Disability Assessment, 1.00 [0.00– 10.00] vs. 0.00 [0.00–1.00], respectively, p < 0.001) than individuals with PM without CA. Multiple regression analyses revealed that the frequency and intensity of headache, anxiety, and depression were signicant factors of CA in individuals with PM. Conclusions: Approximately one-sixth of individuals with migraine and PM experienced CA in a representative sample of Korea. Anxiety, were signicant CA in individuals with PM. Jeju-do [13]. The estimated sampling error was ± 1.9%. The survey was conducted via door-to-door visits and face-to-face interviews using questionnaires by trained interviewers. All interviewers were employees of Gallup Korea and were not medical personnel. The KSHS survey was conducted between October 2018 and November 2018. KSHS was approved by the Institutional Review Board of Severance Hospital, Yonsei University (Approval No. 2018-1269-001).


Background
Cutaneous allodynia (CA) refers to a condition in which pain is caused by tactile stimulus that generally does not result in pain [1]. The underlying mechanism of CA includes sensitization of the trigeminal nucleus caudalis, which receives afferent input from the meninges and periorbital skin regions [2,3]. Central sensitization is a manifestation of increased excitability of neurons in the central nociceptive pathways, and CA is a clinical marker of central sensitization [2,3]. It has been reported that a signi cant proportion of individuals with migraine experience CA [4][5][6] during episodes of headache. Individuals with migraine combined with CA were reported to have poorer response to acute treatment and a higher rate of progression to chronic migraine (CM) compared to individuals with migraine not combined with CA [7]. High attack frequency, depression, and obesity were reported to be signi cant factors of CA in these individuals; therefore, CA provides clues on the pathophysiology of migraine.
Probable migraine (PM) is a subtype of migraine that ful ls all but one criterion of migraine [8]. It was previously reported that 5-15% of the general population experienced PM during the previous year, and the disability of individuals with PM was similar or lower compared with that in those with migraine [9][10][11][12]. The prevalence of CA in PM was previously reported in an American study, which was slightly lower than that in migraine [6]. Nevertheless, the information on the associated factors and characteristics of CA in individuals with PM in a population-based setting is currently limited. The present study aimed to investigate the prevalence, characteristics, and associated factors of CA in individuals with PM along with those in individuals with migraine.

Survey
We used the data of the Korean Sleep-Headache Study (KSHS), which was a nation-wide population-based cross-sectional survey on headache and sleep. Korea is geographically divided into 15 administrative divisions, which were designated as primary sampling units in the rst stage. In the second stage, we selected representative basic administrative units from each primary sampling unit. Overall, 60 representative basic administrative units were selected for this study. For each representative basic administrative unit, we assigned a target sample size based on the age, sex, and occupation. We targeted 2500 individuals aged ≥ 19 years and sampled using two-stage clustered random sampling methods proportional to the population distribution of all Korean territories, except Jejudo [13]. The estimated sampling error was ± 1.9%. The survey was conducted via door-to-door visits and face-to-face interviews using questionnaires by trained interviewers. All interviewers were employees of Gallup Korea and were not medical personnel. The KSHS survey was conducted between October 2018 and November 2018. KSHS was approved by the Institutional Review Board of Severance Hospital, Yonsei University (Approval No. 2018-1269-001).
Migraine was diagnosed based on the diagnostic criteria for migraine without aura according to the third edition of the International Classi cation of Headache Disorders (ICHD-3; code 1.1) [8]. If the characteristics and accompanying symptoms of a participant's headache ful lled A-D criteria of migraine without aura, the diagnosis of migraine was established. The sensitivity and speci city for the diagnosis of migraine were 75.0% and 88.2%, respectively, compared with the diagnosis in additional telephone interviews by doctors [11].
Probable migraine was diagnosed according to code 1.5 of ICHD-3. If the characteristics and accompanying symptoms of a participant's headache ful lled all but one of the A-D criteria of migraine without aura, the diagnosis of PM was established. We did not separately analyse the individuals according to the presence of aura; therefore, migraine included both migraine with aura (ICHD-3 code 1.2) and migraine without aura (ICHD-3 code 1.1) in the present study. Accordingly, PM included both PM with aura (ICHD-3 code 1.5.2) and PM without aura (ICHD-3 code 1.5.1).

Assessment of CA
We investigated CA using the 12-item Allodynia Symptom Checklist (ASC-12). ASC-12 measures interictal CA over the previous month using 12 questions on the thermal, mechanical static, and mechanical dynamic symptoms of CA [4]. Participants were asked to rate 12 questions using any of the following responses: 'does not apply to me'; 'never'; 'rarely'; 'less than half the time'; and 'half the time or more'. The rst three responses were scored as 0, while 'less than half of the time' was scored as 1, and 'half the time or more' was scored as 2. If the ASC-12 score was ≥ 3, the participant was identi ed to have CA. ASC-12 scores were further subclassi ed as mild (score 3-5), moderate (score 6-8), and severe (score ≥ 9) CA, respectively [5]. ASC-12 was previously validated in the Korean language for individuals with migraine [14].

Impact and disability of headache
The impact of headache was assessed using the Headache Impact Test-6 (HIT-6). We classi ed the impact of headache based on the HIT-6 score as follows: < 50, little or no impact; 50-55, some impact; 56-60, substantial impact; and ≥ 60, severe impact [15]. We used the Korean version of HIT-6, which was previously validated in the Korean language [16].
We used the migraine disability assessment (MIDAS) questionnaire to assess the disability of headache. The MIDAS questionnaire is composed of ve questions on the loss of or decrease in productive days because of headache during the previous 3 months. We categorized the disability because of migraine based on MIDAS score as follows: < 49, little or no disability; 50-55: mild disability; 56-59: moderate disability; and ≥ 60: severe disability. The MIDAS questionnaire was previously validated in the Korean language with good sensitivity and speci city.

Statistical analyses
The 1-year prevalence of migraine was calculated as the number of cases per 100 persons. For continuous variables, normality was assessed using the Kolmogorov-Smirnov test. Student's t-tests or analyses of variance was used for normally distributed date and Mann-Whitney U test or Kruskal-Wallis test was used for non-normally distributed data, as appropriate. For comparing categorical variables, we used chi-squared tests.
We evaluated the odds ratios with 95% con dence interval to determine the factors contributing to CA using univariable and multivariable logistic regression analyses. In individuals with migraine and PM, factors that demonstrated signi cant differences between those with CA and those without CA were included in the univariable analyses. For multivariable analyses, four models were developed to examine the association of CA with migraine and PM. Model 1 included the sociodemographic variables (age, sex, and educational level) and was used to investigate the association between sociodemographic factors and CA. Model 2, which included features of Model 1 and the intensity and frequency of headache, was used to investigate the association between headache-related parameters and CA. Model 3, which incorporated anxiety (Generalized Anxiety Disorder-7) and depression (Patient Health Questionnaire-9) into Model 1, was used to investigate the association between psychiatric conditions and CA. The last model, Model 4, included the sociodemographic variables, headache-related parameters, and psychiatric conditions; it was used to investigate the association between these features and CA. SPSS v24.0 (IBM, Armonk, NY, USA) was used to perform the statistical analyses. Statistical signi cance was set at p < 0.050. As with most survey studies, missing data resulting from non-responses were present for several variables; the reported data are based on the available data. Imputation techniques were not applied to minimize effects of non-responses [17].

Survey
Overall, 2501 participants completed the survey. The distribution of the sex, age, and educational level of our participants did not signi cantly differ from the total population of Korea (see Supplementary Table,

Allodynia in individuals with migraine and PM
The responses to ASC-12 in individuals with migraine and PM are summarized in Table 1. The responses to ASC-12 were scored as described above. In individuals with migraine, CA was most frequently reported with 'exposure to cold', followed by 'combing hairs' and 'resting your face or head on a pillow'. The items 'wearing a necklace' and 'wearing earrings' demonstrated the least frequently positive responses. Finally, 11 (8.8%), seven (5.6%), and two (1.6%) individuals with migraine were identi ed with mild, moderate, and severe CA, respectively. The prevalence of CA in individuals with migraine and PM with ≥ 15 headaches per month was signi cantly higher in those with < 1 headache per month (Fig. 2). In individuals with PM, 'exposure to cold' was the most common response, and 'resting your face or head on a pillow' was the next frequent response. The items 'wearing a necklace' and 'wearing earrings' demonstrated the least frequencies. Of 289 individuals with PM, 12 (4.1%), 15 (5.2%), and 15 (5.2%) individuals with migraine were identi ed with mild, moderate, and severe CA, respectively. The prevalence of CA in individuals with PM missing the typical duration was signi cantly higher than that in those with missing the typical characteristics of headache (19.2% vs. 5.3%, respectively, p = 0.002). The overall prevalence of CA was not signi cantly different between individuals with migraine and those with PM (16.0% vs. 14.5%, respectively, p = 0.701).
Clinical characteristics of migraine and PM according to the presence of CA Headache frequency per month, headache intensity (Visual Analogue Scale), impact of headache (HIT-6), and disability (MIDAS) were signi cantly higher in individuals with migraine and PM combined with CA than in those without CA. Anxiety and depression were more prevalent in both individuals with migraine and PM with CA than in those without CA. Photophobia and phonophobia were more prevalent in individuals with PM; however, they were not signi cantly different according to the presence of CA in individuals with migraine ( Table 2).    In individuals with PM, univariable analyses revealed that moderate and severe headache intensity, anxiety, and depression were signi cantly associated with CA. In Model 1, no factor was signi cantly associated with CA. Model 2 revealed that moderate and severe headache intensity were signi cantly associated with CA. In Model 3, depression was signi cantly associated with CA. Model 4 demonstrated that moderate and severe headache intensity and anxiety were signi cantly associated with CA (Table 4). OR: odds ratio, CI, con dence interval, GAD-7: Generalized Anxiety Disorder-7, PHQ-9: Patient Health Questionnaire-9. REF: Reference.

Discussion
The primary ndings of the present study were the following: (1) approximately one-sixth of individuals with migraine and PM experienced CA, and the prevalence of CA was not signi cantly different between those with migraine and those with PM; (2) individuals with migraine and PM with CA experienced more severe symptoms and higher impact of headache and disability than those without CA; and (3) headache intensity, anxiety, and depression were signi cant factors of CA in individuals with PM. In those with migraines, anxiety was a signi cant factor of CA.
The present study found that 16.0% and 14.5% of individuals with migraine and PM, respectively, had CA. Our results are similar to those of a previous Korean clinic-based study, which reported that CA was observed in 14.5% of patients with migraine [14]. These values were lower than those reported in previous studies from the Western countries. American Migraine Prevalence and Prevention (AMPP) study, a large population-based in USA, reported that 62% of individuals with migraine had CA [6]. Migraine in America Symptoms and Treatment study, another American large population-based study, reported that the prevalence of CA in those with migraine was 40% [5]. A Dutch cohort study revealed that CA was present in 70% of participants with migraine [7]. One possible explanation for the lower prevalence of CA in the present study is the difference in the migraine symptoms in the Asian countries. The symptoms are milder in the Asian countries than in the Western countries. Moderate headache intensity was reported in 30-65% of individuals with migraines in Asian countries [18,19]. In the Western countries, 80-85% of individuals with migraine reported severe headache intensity [20,21]. Photophobia was reported in 40-65% of individuals with migraine in Asian countries and 75-85% of individuals with migraine in Western countries [18,[21][22][23]. Headache intensity and photophobia were reported as signi cant predictors of CA in individuals with migraine [5]. Therefore, milder headache intensity and lower photophobia might result in lower prevalence of CA. Another possible explanation is the difference in the body mass index (BMI), which has been reported to be lower in Asian populations than in Western populations [24]. High BMI was reported to be a signi cant factor for CA in the AMPP study [4].
That study found that obese (BMI, 30-40 kg/m 2 ) and morbidly obese (BMI, ≥ 40 kg/m 2 ) individuals had higher risk of CA. In the present study, there was no signi cant difference in BMI in individuals with migraine and PM according to the presence of CA. This discrepancy might be because of differences in BMI. Only two individuals with migraine and four individuals with PM were obese.
Furthermore, none of the individuals with migraine and PM quali ed for morbid obesity. Ethnic differences could be another possible explanation. It has been reported that pain sensitivity varies among ethnic groups [25]. Further studies in various migraine populations are required for a better understanding of the prevalence and contributing factors of CA.
In the present study, 'exposure to cold' and 'resting your face or head on pillow' were most frequently positively responded items in both individuals with migraine and PM. 'Combing hairs' and 'pulling your hair back' followed the next frequency. Allodynia is classi ed as mechanical dynamic, mechanical static, and thermal allodynia. They differ in terms of the transmission nerve bres and nociceptors [26,27]. Each item of ASC-12 complied three types of CA. 'Exposure to cold' and 'resting your face or head on pillow' corresponded to thermal allodynia, and 'combing hairs' and 'pulling your hair back' items corresponded to mechanical dynamic allodynia [4]. High positive response rate to items of thermal and mechanical dynamic allodynia in individuals with migraine was previously reported in a Brazilian study [28]. The present study is the rst to identify a high positive response rate to items corresponding to thermal and mechanical dynamic allodynia in individuals with PM, which is similar to that in individuals with migraines.
The headache frequency, headache intensity, disability, and impact of headache were higher in those with migraine and PM combined with CA than in those without CA in the present study. Further, CA was more prevalent in those with the chronic form (≥ 15 episodes per month) of migraine and PM than in those with low headache frequency (< 1 episode per month) (Fig. 2). The close associations of CA with symptom severity and chronicity have been previously reported in migraine [4,6]. The present study provides evidence that such an association is also present between CA and PM.
In the present study, anxiety and depression were identi ed as signi cant factors of CA in individuals with PM. The signi cant association of anxiety and depression with CA has been reported previously. Kao et al. reported that anxiety was a signi cant factor of CA using multivariable regression analyses. Furthermore, comorbid anxiety and depression were also associated with the severity of CA [29]. Louter et al. reported that CA was associated with higher prevalence of depression in individuals with migraine [30]. CA, anxiety, and depression a signi cant risk factor of CM transformation from episodic migraine (EM) [7,31]. CM has a higher prevalence in the presence of anxiety, depression, and CA than EM [6]. Therefore, our ndings added an evidence for the signi cant association of anxiety and depression with CA and suggested sharing pathophysiological mechanisms of CA with anxiety and depression. Biogenic amines might be involved in a possible shared mechanism. Allodynia is a characteristic of FM, which is a chronic condition of widespread pain [32]. In an animal model of bromyalgia, decreased tactile threshold was correlated with depressive behaviours [33]. The animal model demonstrated a decreased level of biogenic amines including dopamine, 5hydroxytryptoptamine, and norepinephrine in the spinal cord, thalamus, and prefrontal cortex.
The prevalence of migraine in the present study was lower than that in previous Western studies. The prevalence of migraine in Asian countries is 3-10%, which is lower than that in Western countries where it is 11-18% [34]. Therefore, migraine prevalence in the present study was similar to those in previous Asian studies. The reported prevalence of PM ranges widely (USA, 4.5%; Singapore, 6.2%; France, 10.0%; Korea. 11.5%; England: 14.6%) [9, 10, 12, 35]; therefore, the prevalence of PM in the present study was broadly similar to those in previous studies. The similarities in the prevalence of migraine and PM between the present and previous studies suggest that appropriate evaluation of migraine and PM in the current study.
The present study has some limitations. First, we used ASC-12 in the evaluation of CA. The gold standard of assessing CA is quantitative sensory testing (QST); it requires specialized equipment and is di cult to conduct in clinical practice and epidemiological studies. ASC-12 was previously validated in comparison with QST [7]. It was also validated in Korean individuals with migraine [14].
Second, we did not investigate the disease durations of migraine and PM. Disease duration was reported to be a signi cant factor of CA in patients with migraine. Since medical consultation and awareness of migraine diagnosis is not high in Korea, it would be di cult to know participants' exact disease duration of migraine. We believed that the assessments of disease duration were less feasible and, therefore, did not include them in the analyses. Finally, we did not evaluate the use of medications in the participants. Some medications for migraine prevention, such as serotonin-norepinephrine reuptake inhibitors and anticonvulsants, might relieve CA [36][37][38]. Further studies on the use of medications are required to provide accurate information of CA in patients with migraine and PM.
The present study includes several strengths. First, we used a two-stage clustered random sampling method proportional to the distribution of the total population of Korea. Furthermore, the estimated sampling error was low. This approach allowed us to successfully assess CA in individuals with migraine and PM in a population-based setting. Second, in the present study, the responses of 12 items in addition to the total score of ASC-12 were analysed. We found that 'exposure to cold', 'resting your far or head on a pillow', and 'combing hair' were the most frequently responses both in individuals with migraine and PM. Third, our study used questionnaires which were specialized validated in Korea language for assessing migraine, anxiety and depression. Such process enabled us to accurately evaluate migraine, PM, anxiety and depression in the present study.

Conclusions
Approximately one-sixth of individuals with migraine and PM experienced CA in a general-population-based sample in Korea. The prevalence of CA was not signi cantly different between those with migraine and those with PM. Individuals with migraine and those with PM combined with CA had more severe symptoms than those not combined with CA. Anxiety, depression, and high frequency of Written consent was obtained from all participants before the survey interviews.

Consent for publication
Not applicable Availability of data and materials The data used in this study are available from the corresponding author on reasonable request.