The present study was part of an EU funded research programme, called NewMood (New Molecules in mood Disorders), which aimed to investigate novel patho-mechanisms of major depression and its comorbid disorders, such as anxiety and migraine (24). Details of the recruitment and responses can be found elsewhere (25, 26). To shortly summarize, participants were recruited in Greater Manchester, UK and Budapest, Hungary by contacting general practices and using advertisements (NewMood website, university advertisements, and newspapers). All willing participants filled out the NewMood booklet, which contained brief standard and validated questionnaires. Altogether n = 2,004 subjects responded at Manchester and n = 1,139 at Budapest by sending back the postal questionnaire and the signed written informed consent. From them, subjects with useful questionnaire data were included in this study regardless of ethnicity or reported medical or psychiatric disorders. In Manchester, data from 1,970 participants with a mean age of 33.50 years, and in Budapest, data from 1,056 participants with a mean age of 31.40 years, were analyzed. The study was approved by local ethics committees and was carried out in accordance with the Declaration of Helsinki.

Background details (e.g., sex, age), information on socioeconomic status, and medical history, including psychiatric disorders and reported migraine, were collected by a brief standard background questionnaire, English and Hungarian versions, respectively.

Reported DEP episodes were derived from a set of questions that were validated in a subset of participants during face-to-face interviews (27). Other pain disorders were determined based on the background questionnaire and coded “yes” if subjects did not report migraine but reported at least one of these conditions: back pain (n = 162), rheumatoid arthritis (n = 31), abdominal pain (n = 43, e.g., irritable bowel syndrome, Crohn’s disease, ulcerative colitis, heartburn), joint pain (n = 16, e.g., arthritis, osteoarthritis), diffuse pain (n = 14, e.g., fibromyalgia, myalgic encephalomyelitis, complex regional pain), or other pain (n = 4, e.g., mastitis, chronic sinusitis).

To identify subjects with migraine(ID) type headache, the ID-migraine questionnaire was used, which is a validated screening tool for migraine (28, 29). Migraine(ID) was coded if the participants experienced at least two symptoms out of nausea, photophobia, and disability during headaches in the previous 3 months (28). Non-migraine headache was coded if only one symptom was experienced, or the subject indicated that they had headaches without these symptoms and therefore not fulfilling the criteria for migraine(ID).

The Big Five Inventory was used to measure five factors of personality (BFI-44), namely, extraversion, agreeableness, conscientiousness, neuroticism, and openness (30). Items are rated on a 5-point Likert-type scale ranging from 1 (disagree strongly) to 5 (agree strongly). For personality factors, continuous weighted dimension scores (sum of item scores divided by the number of items completed) were calculated for the analysis.

SPSS 21.0 for Windows (IBM) was used to carry out statistical analyses. Pearson chi² analysis calculated the difference of migraine(ID), non-migraine headaches, and pain prevalence in subjects with or without DEP. Multivariate ANOVA in the whole population (Budapest and Manchester together) was applied to investigate whether there was an interaction effect of DEP and migraine(ID) on personality factors. Wilk’s lambda statistics and test of between-subject effects were reported. Using univariate ANOVA, post hoc we calculated the significant interaction effects in the subpopulations separately to identify replicable findings, and also tested whether the significant interaction was migraine specific or applied to non-migraine headaches or other pain. Age and sex were covariates in all analyses. In calculations where the whole population was included, study site was added as an independent factor to the analysis to control for cohort differences. Finally, a logistic regression model was built to test the effect of age, sex, the five personality factors, DEP, and any significant interaction between DEP and personality on migraine(ID). All statistical testing adopted a two-tailed p = 0.05 threshold Table 1.

We found important associations between personality traits and both migraine and depression. As previous studies suggested (19, 20), neuroticism appears as a risk factor for both migraine(ID) and DEP in our population study. However, in contrast to previous studies (16, 22), we found that openness increased the risk of migraine, but we also found an important interaction effect in case of openness, indicating higher openness only in migraine(ID) patients without depression.

Openness manifested in a characteristic constellation of cognitive and affective styles, including creativity, curiosity, flexible thinking, preference of novel experience, increased receptiveness for salient stimuli, and absorption in sensory experience (31, 32). Although this is the most controversial trait in the 5-factor model (33, 34), recent studies indicated that it is an important factor in coping with different disorders. Indeed, openness has been reported to play a protective role against depression in longitudinal follow-up studies (35). Similarly, openness was associated with reduced rates of depression among people with somatic conditions such as among hemodialysis patients (36) or in postpartum period (37), also indicating that openness may help to counteract the depressogenic effects of somatic conditions. In addition, openness has been found to be associated with lesser physiological reactivity and higher physiological adaptation, measured by hart rate, blood pressure, and respiratory rhythm changes, to recurrent social evaluative stress, suggesting that openness might be a protective factor against harmful effects of stress (38). Regarding migraine, in line with our results, migraineurs with higher openness are also more flexible and creative in their approach to managing their condition which in turn reduces the impact of migraine on their daily life, which is also reflected in less impaired function in migraineurs with higher openness (39). This effect is especially important in the comorbidity of depression and migraine as migraine and depression have already been reported to have a peculiar relationship reflected by their high comorbidity. In line with our results, this relationship is different from that between depression and other types of headaches (4, 40).

Our study supported previous accounts suggesting a positive association between depression and migraine; however, DEP had an opposite effect on non-migraine headaches. According to a 2-year follow-up study, preexisting depression increased the risk for developing migraine by 3.4 times and preexisting migraine increased risk for developing depression 5.8-fold (4), while current depression and anxiety was associated with a higher increase in the risk of migraine compared to other types of pains and headaches (40). This specific relationship between depression and migraine, but not other types of headaches, may indicate a common genetic or neurobiological element in the pathophysiology of migraine and depression as suggested previously (6, 8).

The nature of the relationship between painful conditions in general and depression is far from well understood and several factors may play a role in the increased co-occurrence specifically of migraine and depression. There may be a direct causal association, although the direction is unclear. Depression may contribute to increased pain sensitivity, or migraine may lead to depression, through recurrent pain and decreased quality of life (9) or through learned helplessness due to its inescapable nature. Depression may also arise as a neurobiological consequence and side effect of migraine-associated pain. A further possibility is that migraine and depression may be different symptomatic manifestations of the same underlying syndrome (9). Recent studies indicated that common pleiotropic genetic and environmental influences may also account for the increased comorbidity between depression and migraine, including the involvement of common biological pathways such as the dopaminergic and serotonergic pathways, and stress as a major risk factor for both conditions (9).

Although general heritability of big five traits appear to be around 50% reflecting a strong biological background (41), less neurobiological explanations are available for openness compared to other traits such as neuroticism or extraversion (34). Several authors suggest a higher order solution for the 5-factor model with two superfactors, stability (including neuroticism, agreeableness, and conscientiousness) related to the variation in serotonergic function, and plasticity (including extraversion and openness) related to the variation of dopaminergic function (42–44). A key feature of openness is the exploratory tendency on a more abstract, motivational and cognitive level (34). The dopaminergic system modulates the novelty-associated rewarding positive stimuli (34, 45, 46) and though this possibly regulates both the motivational and cognitive aspects of openness (34).

Based on previous observations, openness appears to be associated with prefrontal and anterior cingulate dopaminergic projections (34, 43, 45). In a recent study, there was a positive association between openness and functional connectivity between the right DLPFC and the right midbrain substantia nigra/ventral tegmental area, which is the chief source of dopaminergic inputs during resting state and different tasks (32). Specific association of dopaminergic regulation in the DLPFC with openness is supported by a robust association between neurocognitive tasks reflecting DLPFC function and openness but not extraversion (34). Thus, DLPFC function could be a key mediator of the protective effect of openness on co-occurrence of migraine and depression, as DLPFC has a particular role in both depression and migraine. Decreased DLPFC function was associated with increased vulnerability to depression in the presence of negative stimuli (47). In addition, the DLPFC was found to be involved in pain modulation (48) and was hypothesized to show a constant upregulation in order to enhance the descending modulation of pain (49, 50).

DLPFC was also implicated to play a role in pain processing specifically in migraine, as greater evoked pain-related activity was reported in interictal migraine patients in several areas related to cognitive pain processing, including the DLPFC (51). Further studies that specifically tested pain–cognition interaction in migraineurs demonstrated that DLPFC showed similar cognitive task related deactivation regardless of pain condition (evoked pain or no pain) showing chronic DLPFC engagement in migraineurs across all conditions (49). In addition, reduced cortical gray matter volume in areas including the DLPFC was associated with pain catastrophizing in migraineurs (52).

Thus, openness in migraine may indicate a more optimal manifestation of enhanced DLPFC activity engaging not only with frontal top-down control to reduce responses to negative stimuli (47), including pain (49, 50), but also flexibly increasing receptiveness for novel salient stimuli and thus contributing to less burden and leading to less depression. Furthermore, higher openness to experience in a healthy sample was related to better default mode network (DMN) efficiency (53) indexing the network integrity. Since both depression (54) and chronic pain conditions (55) have been proved to be associated with alterations in DMN connectivity, we speculate that openness might exert its protective effect against depression via better DMN integrity in migraine.

It has been demonstrated that comorbid depression with other somatic disorders is an independent predictor of unfavorable treatment outcome (56). In the case of migraine, a recent study showed that comorbid depression worsens the responsiveness for acute migraine treatment in a large population sample with episodic migraine (57). Thus, understanding the neurobiological background of this exceptionally high and specific comorbidity between migraine and depression may improve our treatment strategies. In this context, it is interesting to note that openness was associated with increased response to placebo analgesia in migraineurs (58) and with better response to both pharmaco- and psychotherapy in depressed people (59). Furthermore, deep transcranial magnetic stimulation of the DLPFC significantly reduced both frequency and intensity of migraine attacks and depressive symptoms (60).

The study has some limitations. Due to our cross-sectional design, we cannot draw any conclusions about the causal relationship between migraine, depression, and openness. In addition, assessment of both DEP and pain and headache was based on self-report in our study. Using ID-Migraine Questionnaire that covers only the last 3 months increased the risk of including some migraineurs with less frequent, lifelong migraine in the non-migraine group However, it is worth noting that ID-Migraine questionnaire is a valid screening tool for migraine with good specificity and sensitivity (61) and the sensitivity and specificity of the ID-migraine questionnaire in our study was in a comparable range (62) with the original validation study (28) and with the meta-analysis results (61). Also, in our study the prevalence of depression was higher in the Manchester cohort than in the Budapest cohort, despite similar recruitment strategies. However, the proportion of migraine(ID) in the subgroups of subjects with and without DEP were very similar in both cohorts, and the significant interaction of DEP and migraine(ID) on openness replicated in both cohorts. Further studies with clinical samples and longitudinal design are needed to reveal the importance of openness on either daily life or on neural activity.