Prevalence and characteristics of cutaneous allodynia in probable migraine

Cutaneous allodynia (CA) is a pain in response to non-nociceptive stimulation and a marker of central sensitisation. Probable migraine (PM) is a migraine subtype that fulfils all but one criterion of migraine. Headache intensity and the disability of individuals with PM are similar or lower than individuals with migraine. This study compared CA prevalence and characteristics of PM and migraine using a nationally representative sample in Korea. The Allodynia Symptom Checklist-12 (ASC-12) was used to assess CA (ASC-12 score ≥ 3). PM and migraine prevalence were 11.6% and 5.0%, respectively. CA prevalence did not significantly differ between PM and migraine (14.5% vs. 16.0%, p = 0.701). Participants with PM with CA reported a higher monthly headache frequency (3.3 ± 4.3 vs. 1.8 ± 3.6, p = 0.044), more severe headache intensity (Visuals Analogue Scale, 6.0 [4.0–7.0] vs. 5.0 [3.0–6.0], p = 0.002), and higher impact of headache (Headache Impact Test-6, 56.3 ± 7.2 vs. 48.3 ± 8.0, p < 0.001) than those without CA. Multiple regression analyses revealed that headache frequency and intensity, anxiety, and depression were significant factors for CA in participants with PM. In conclusion, CA prevalence among participants with PM and migraine were comparable. Anxiety, depression, and headache frequency and intensity were significant factors for CA in participants with PM.

PM and migraine diagnosis. PM was diagnosed according to the third edition of the International Classification of Headache Disorders (ICHD-3; code 1.5) 13 . If the characteristics and accompanying symptoms of a participant's headache fulfilled all but one of the A-D criteria of migraine without aura, a diagnosis of PM was established.
Migraine was diagnosed based on the diagnostic criteria for migraine without aura according to code 1.1 of ICHD-3 13 . If the characteristics and accompanying symptoms of a participant's headache fulfilled the A-D criteria of migraine without aura, a diagnosis of migraine was established. We previously evaluated the diagnostic ability of our questionnaire for migraine by comparing the diagnosis using the questionnaire and diagnosis made by neurologists with an additional telephone interview 16 . We did not separately analyse participants 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). Impact and disability of headache. The impact of headache was assessed using the Headache Impact Test-6 (HIT-6) 20 . We used the Korean version of HIT-6, which was previously validated in the Korean language 21 . We used the migraine disability assessment (MIDAS) questionnaire to assess headache-related disability 22 . The MIDAS questionnaire is composed of five questions on the loss of or decrease in productive days due to headache during the previous 3 months. The MIDAS questionnaire was previously validated in the Korean language with good sensitivity and specificity 23 .

Assessment of depression and anxiety. Depression was assessed by the Patient Health Questionnaire-9
(PHQ-9) 24 . Participants with PHQ-9 scores of 10 or more were defined as having depression. The Korean version of the PHQ-9 has been evaluated showing a sensitivity of 81.1% and a specificity of 89.9% 25 .
We used the Generalized Anxiety Disorder-7 (GAD-7) questionnaire to evaluate anxiety 26 . A participant giving a positive response to seven or more items was classified as having anxiety. The Korean version of GAD-7 was validated with a sensitivity of 78.1% and a specificity of 74.6% 27 . Ethical approval. This study was approved by the Institutional Review Board of Severance Hospital, Yonsei University (approval No. 2019-1721-001). Written informed consent was obtained from all participants Prior to obtaining written informed consent, all participants were given an explanation on the objective of the study and the data to be collected by interviewers. All procedures involving human participants were in accordance with the ethical standards of the institutional and/or national research committee as well as the tenets of the 1964 Declaration of Helsinki and its later amendments, or comparable ethical standards.
Statistical analyses. The 1-year prevalence of migraine was calculated as the number of cases per 100 persons. Binary and ordinal scales are represented as numbers and percentages. Interval scales are represented as mean and standard deviation or median and interquartile range (IQR) when appropriate. Normality of the ratio of variables was tested with the Kolmogorov-Smirnov test. After normality was confirmed, independent two-tailed t-tests or one-way analyses of variance were used to compare the variables between the groups when appropriate. For non-normally distributed variables, a two-tailed Mann-Whitney U test or Kruskal-Wallis test Since the prevalence of PM and migraine varied significantly according to sex and age, we compared clinical characteristics of PM and migraine according to the presence of CA with age and sex adjustment 16,28 . Logistic regression analyses after age (year, continuous) and sex adjustment were used to compare categorical variables according to the presence of CA in PM and migraine. Linear regression analyses with age and sex adjustment were used to compare numerical variables.
We evaluated the odds ratios with 95% confidence interval to determine the factors contributing to CA using univariable and multivariable logistic regression analyses. In individuals with migraine and PM, factors that demonstrated significant differences between those with CA and those without CA were included in the univariable analyses. For multivariable analyses, sociodemographic variables (age, sex, and educational level), headache-related parameters (the intensity and frequency of headache) and psychiatric conditions (anxiety  and depression ) were included to examine the association of CA with migraine and PM. SPSS v24.0 (IBM, Armonk, NY, USA) was used to perform the statistical analyses. Statistical significance was set at two-tailed p < 0.050. The current analysis is the primary analysis for prevalence of CA in PM using KSHS data. Imputation techniques were not used to minimise non-response effects. The approach used in this study will produce slightly larger standard errors in finite samples when compared with imputation techniques 29 .

Results
Survey. Overall, 2501 participants completed the survey. The distribution of sex, age, and educational level among our participants did not significantly differ from that of the total population of Korea (Table 1). Therefore, there was no need to use extracted sampling weights in the statistical analyses.
Of the 2501 participants, 1186 responded positively to the question 'Did you have headache during the previous one year?' . Of the 1186 respondents, 289 (11.6%) and 125 (5.0%) participants had PM and migraine, respectively (Fig. 1). Of the 289 participants with PM, 193 (66.8%) missed the typical duration (criterion B), 94 (32.5%) missed the typical headache characteristics (criterion C), and two (0.7%) missed the typical accompanying symptoms (criterion D) of migraine without aura.
There were no missing data except for PHQ-9, weight, and height. Among all participants, PHQ-9, height and weight data were missing for 3, 7, and 21 participants, respectively. Among participants with PM, the height data was missing for one participant. Among participants with migraine, weight and height data were missing for two and one participant, respectively. No PHQ-9 data was missing for any participants with migraine or PM.
CA in individuals with PM and migraine. The responses to the ASC-12 among participants with PM and migraine are summarised in Table 2. Among participants with PM, 'exposure to cold' was the most common response with less than half of the time or more frequency, and 'resting your face or head on a pillow' was the next most frequent response. The items 'wearing a necklace' and 'wearing earrings' demonstrated the least frequencies. Of 289 participants with PM, 12 (4.1%), 15 (5.2%), and 15 (5.2%) participants were identified with mild, moderate, and severe CA, respectively. The prevalence of CA among participants with PM missing the Table 1. Sociodemographic distribution of survey participants and participants identified as having probable migraine and migraine in the 2018 survey as proportions of the overall population. CI: confidence interval. www.nature.com/scientificreports/ typical duration was significantly higher than that among participants with PM missing the typical characteristics of headache (19.2% vs. 5.3%, respectively, p = 0.002). Among in participants with migraine, CA was most frequently reported with 'exposure to cold' (less than half of the time or half of the time or more), followed by 'combing hair' and 'resting your face or head on a pillow' . The items 'wearing a necklace' and 'wearing earrings' demonstrated the least frequent positive responses. Finally, 11    Fig. 2A). Conversely, there was significant differences in the prevalence of CA among participants with migraine according to headache frequency (p = 0.017). However, post hoc analyses revealed no significant difference among the three groups (Fig. 2B). The overall prevalence of CA was not significantly different between participants with PM and those with migraine (14.5% vs. 16.0%, respectively, p = 0.701).
Clinical characteristics of PM and migraine according to the presence of CA. Headache frequency per month, headache intensity (Visual Analogue Scale), impact of headache (HIT-6), and disability (MIDAS) were significantly higher among participants with PM and migraine combined with CA than among those without CA. Anxiety and depression were more prevalent among both participants with PM and migraine with CA than among those without CA. Considering participants with migraine, anxiety was more severe among participants with migraine with CA than among those without CA. Nevertheless, the severity of depression was not significantly different according to the presence of CA. Photophobia and phonophobia were more prevalent among participants with PM combined with CA; however, there was no significant difference in photophobia and phonophobia according to the presence of CA among participants with migraine (Table 3).
Factors associated with CA in PM and migraine. For participants with PM, univariable analyses revealed that moderate and severe headache intensity, anxiety, and depression were significantly associated with CA. In multivariable analyses, moderate and severe headache intensity, anxiety, and depression were significantly associated with CA (Table 4). Univariable analyses in participants with migraine indicated that ≥ 15 headaches per month, moderate and severe headache intensity, anxiety, and depression were significantly associated with CA. Multivariable analyses revealed that only anxiety was significantly associated with CA (Table 4).

Discussion
The primary findings of the present study were the following: (1) approximately one-sixth of participants with PM and migraine experienced CA, and the prevalence of CA was not significantly different between those with PM and those with migraine; (2) participants with PM and migraine with CA experienced more severe symptoms and a greater impact of headache and disability than those without CA; and (3) headache intensity, anxiety, and depression were significant factors of CA in participants with PM. In participants with migraine, anxiety was a significant factor of CA.
Allodynia is classified as mechanical dynamic, mechanical static, and thermal allodynia. These types differ in terms of the transmission nerve fibres and nociceptors 30,31 . Each item of the ASC-12 comprised three types of CA. 'Exposure to cold' and 'resting your face or head on pillow' corresponded to thermal allodynia, and 'combing hair' and 'pulling your hair back' items corresponded to mechanical dynamic allodynia 5 . High positive response rate to items of thermal and mechanical dynamic allodynia among individuals with migraine was previously reported in a Brazilian study 32 . The present study found that 'exposure to cold' and 'resting your face or head on pillow' were the most frequently reported positive items among both individuals with PM and migraine. 'Combing hair' and 'pulling your hair back' was the next most frequent positive response. The present study is the first to identify a high for positive response rate for items corresponding to thermal and mechanical dynamic allodynia in individuals with PM, which is comparable to that in individuals with migraines.
We included participants with PM and migraine who had at least one headache attack during the preceding year. Nevertheless, some individuals with PM and migraine had a headache frequency of < 1 attack per year 33 . These individuals may not have been classified as having migraine and PM in our study. Participants with migraine and PM with a headache frequency of < 1 attack per year may have different prevalence and clinical   www.nature.com/scientificreports/ features of CA compared to those with a frequency of ≥ 1 attack per year. Although most individuals with migraine had a headache frequency of ≥ 1 attack per year, the findings in the present study were only applicable to PM and migraine, with headache a headache frequency of ≥ 1 attack per year 33 . Previous studies have shown that a significant proportion of individuals with migraine had CA. The American Migraine Prevalence and Prevention (AMPP) study, a large population-based study in the USA, reported that 62% of individuals with migraine had CA 7 . Migraine in America Symptoms and Treatment (MAST) study, another American large population-based study, reported that the prevalence of CA in individuals with migraine was 40% 6 . A Dutch cohort study revealed that CA was present in 70% of individuals with migraine 8 . The present study found that 16.0% of individuals with migraine had CA. These values were lower than those reported in previous studies in Western countries but results were comparable to those reported in a Korean clinic-based study, which found that 14.5% of individuals with migraine had CA 19 . One possible explanation for the lower prevalence of CA in the present study is the difference in migraine symptoms in Asian countries. Symptoms are milder in Asian countries than in Western countries. Moderate headache intensity has been reported in 30-65% of individuals with migraines in Asian countries 34,35 . In Western countries, 80-85% of individuals with migraine reported severe headache intensity 36,37 . Photophobia has been reported in 40-65% of individuals with migraine in Asian countries and 75-85% of individuals with migraine in Western countries 34,[37][38][39] . Headache intensity and photophobia were significant predictors of CA in individuals with migraine 6 . Another possible explanation is the difference in body mass index (BMI), which is lower in Asian populations than in Western populations 40 . High BMI was reported to be a significant factor for CA in the AMPP study 5 with obese (BMI, 30-40 kg/m 2 ) and morbidly obese (BMI, ≥ 40 kg/m 2 ) individuals having a higher risk of CA. In the present study, only four participants with PM and two participants with migraine were obese. Furthermore, none of the participants with migraine and PM qualified for morbid obesity. The 1-year based diagnostic strategies for PM and migraine in our study could be another possible reason for the discrepancy in the prevalence of CA between our study and previous reports. The Dutch and Brazilian studies enrolled individuals with migraine regardless of headache frequency and reported a high prevalence of CA 32,41 . The AMPP study recruited individuals who had at least one severe headache in the previous year and reported a comparable prevalence of CA 5 . The MAST study, which included individuals with migraine with monthly headache frequency, reported a slightly lower prevalence of CA than that in other studies but higher than CA prevalence found in our study 42 . The present study identified participants with PM and migraine with a headache frequency of ≥ 1 attack per year and reported a lower prevalence of CA compared to previous studies. Ethnic differences could be another possible explanation. It has been reported that pain sensitivity varies among ethnic groups 43 . Further studies in different migraine populations are required for a better understanding of the prevalence and contributing factors of CA.
In the present study, headache intensity, disability, and impact of headache were higher among participants with PM and migraine combined with CA than among those without CA. The close associations of CA with symptom severity and chronicity have been previously reported in migraine 5,7 . The present study provides evidence that such an association is also present between CA and PM.
In the present study, anxiety and depression were identified as significant factors for CA in individuals with PM. A significant association of anxiety and depression with CA has been reported previously. Kao et al. reported that anxiety was a significant factor of CA using multivariable regression analyses. Furthermore, comorbid anxiety and depression were also associated with CA severity 44 . Louter et al. reported that CA was associated with higher prevalence of depression among individuals with migraine 41 . CA, anxiety, and depression were significant risk factors of CM transformation from episodic migraine (EM) 8,45 . CM has a higher prevalence in the presence of anxiety, depression, and CA than EM 7 . Therefore, our findings corroborate existing evidence that supports the significant association of anxiety and depression with CA, suggesting shared pathophysiological mechanisms with anxiety and depression. Biogenic amines may be involved in a possible common mechanism 46 . Allodynia is a characteristic of FM, which is a chronic condition of widespread pain 47 . In an animal model of FM, decreased tactile threshold correlated with depressive behaviors 48 . The animal model demonstrated a decreased level of biogenic amines including dopamine, 5-hydroxytryptoptamine, and norepinephrine in the spinal cord, thalamus, and prefrontal cortex 48 .
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 reported as 11-18% 28 . Therefore, migraine prevalence in the present study was similar to that 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%) 14,15,17,49 ; therefore, the prevalence of PM in the present study was broadly similar to that reported in previous studies.
The present study has several limitations. Firstly, we used the ASC-12 in the evaluation of CA. The gold standard for the assessment of CA is QST; however, this requires specialised equipment and is difficult to conduct in clinical practice and epidemiological studies. The ASC-12 was previously validated in comparison with QST 8 . It was originally written in English and has been translated into Portuguese, Spanish and Korean 32,50,51 . The Korean translation was constructed through forward and backward translation by a translator and reliability was assessed by calculating internal consistency of items. Portuguese and Spanish translations of the ASC-12 were not validated in comparison with QST, but studies using these two translations revealed similar prevalence of CA among individuals with migraine compared to studies using the English version ASC-12 32,50 . Using a Korean translation of the ASC-12, which was not fully validated, could be a limitation of our study 52 . Secondly, we did not evaluate cooperation rate in our survey. We conducted a headache survey in 2009 using the same sampling method, except for target sample size, and involving the same social research company, Gallup Korea 16 . Cooperation in the 2009 survey was 38.6% and we assume that the cooperation rate in this study may be comparable. The distribution of age, sex and educational level in this study was similar to that of the total population of Korea in the 2009 survey and KSHS. Furthermore, the one-year prevalence of migraine in the 2009 Scientific Reports | (2021) 11:2467 | https://doi.org/10.1038/s41598-021-82080-z www.nature.com/scientificreports/ survey (6.1%) and the present study (5.0%) was similar to the reported prevalence in Asian countries 53 . Finally, we did not evaluate the use of medication among participants. Some types of medication for migraine prevention, such as serotonin-norepinephrine reuptake inhibitors and anticonvulsants, may relieve CA [54][55][56] . Further studies including medication use are required to provide accurate information of CA in patients with migraine and PM. The present study has several strengths. Firstly, 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. Secondly, in the present study, we analysed the responses of 12 items in addition to the total ASC-12 score. We found that 'exposure to cold' , 'resting your face or head on a pillow' , and 'combing hair' were the most frequent responses given by participants with migraine and PM. Thirdly, our study used questionnaires that were specialized validated in the Korean language for assessing migraine, anxiety and depression. This process allowed us to accurately evaluate migraine, PM, anxiety and depression.
In conclusion, approximately one-sixth of individuals with PM experienced CA in a general-populationbased sample in Korea. The prevalence of CA was not significantly different between participants with PM and those with migraine. Anxiety, depression, and high frequency of headaches were significant factors of CA among participants with PM.

Data availability
The data used in this study are available from the corresponding author on reasonable request.