Introduction

Circadian rhythm (CR) is one of the major determinants of sleep–wake cycle [1, 2], regulated by a “biological clock” localized in the suprachiasmatic nucleus (SCN) [2, 3]. Several physiological and behavioural aspects, including body temperature, hormonal profile and blood pressure are regulated by this biological rhythm. The genetically determined endogenous rhythm is longer than 24 h, but it is synchronized to the daily 24-h by different social and environmental “zeitgebers”. A zeitgeber is an ethologic term used to define any external cue that synchronizes biological rhythms to environmental conditions [4]. Light is the most relevant zeitgeber. In addition to the SCN, retinal photoreceptors and the melatonin secretion by the pineal gland play a prominent role in the circadian rhythm regulation. SCN activation by bright retinal inputs inhibits melatonin production by the pineal gland which, on the contrary, is stimulated by the dark. One of the most relevant expressions of CR deregulation is the sleep–wake cycle alteration that may manifest with sustained levels of vigilance or propensity to inappropriately falling asleep.

Three main typologies of the circadian asset can be recognized: the “morning type” (the so-called “lark”), characterized by an early awakening and bedtime and better performance in the morning, an “evening type” (also defined “owl”), with late wake and bedtime and satisfactory activities in the evening, and an “intermediate typology”. Most subjects are characterized by the intermediate phenotype.

Sleep is able to influence the expression of different clinical conditions, including migraine. A good sleep quality, together with a keep sleep–wake cycle, exerts a protective factor; on the other hand, migraine subjects with severe and frequent attacks often suffer from primary or secondary sleep disturbances [5]. Several studies investigated the relationship between sleep and migraine, but data explaining the influence of circadian typology on migraine characteristics are not exhaustive [6].

The aim of this study was to explore the circadian phenotype in a group of migraine patients and to evaluate the possible influences on different diseases’ characteristics.

Materials and methods

This cross-sectional study was performed in two third-level centres: the Headache centre of the Neurological Clinic, Polytechnic University of Ancona and of the Neurology Unit at Campus Bio-Medico University, Rome. In both centres, expert neurologists in headache diagnosis and management evaluated and enrolled migraine subjects undergoing outpatients’ assessment.

Inclusion criteria were the following: (a) a diagnosis of migraine without aura (MWoA), according of the International Classification of Headache Disorders criteria, 3rd edition (beta version) (ICHD-III beta) [7]; and (b) Italian nationality or good Italian language comprehension and practice. In each patient, a collection of clinical history, comprehensive of drug utilization, was performed in addition to a general and neurological examination. We investigated the presence of other relevant pathologies, especially psychiatric illnesses and related drug assumption. Exclusion criteria were the following: (a) headache determined by brain tumours or other neurological diseases; (b) relevant pathologies as diabetes, renal or hepatic failure or other metabolic chronic disease; (c) night-shift work; (d) long-distance transmeridian flights in the previous month; (e) history of sleep alterations (obstructive sleep apnoea, restless legs syndrome, narcolepsy, etc.); (f) mood or affective disorders, especially seasonal ones; (g) use of melatonin or drugs able to influence endogenous melatonin secretion; (h) chronic use of antidepressants or sleep-inducing drugs; (i) substance and alcohol abuse; (l) relevant neurological or visual deficits; and (m) presence, in the clinical history, of previous preventive therapy for migraine. Mood disorders were evaluated by Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, (DSM V) criteria employing clinical history, Structured Clinical Interview for DSM V (SCID) and 20-item Beck Inventory II (patients with a score > 13 at Beck Inventory II were excluded) [8].

Monthly attacks frequency, years from migraine onset, attacks intensity (mild, moderate or severe), and treatments were investigated.

Patients were asked to fill out the Italian validated version of the Morningness–Eveningness Questionnaire (MEQ-SA) [9, 10]. MEQ-SA is a self-assessment, validated questionnaire to evaluate the circadian typology. It is composed by 19 multiple-choice questions regarding sleep characteristics and preference with a final score ranging from 16 to 86. Subjects are defined as late rising, moderately late rising, intermediate, moderately early rising and early rising when their score ranges from 16 to 30, 31 to 41, 42 to 58, 59 to 69 and 70 to 86, respectively.

Pain intensity was evaluated by the Verbal Rating Scale (VRS), a validated 10-point self-reporting score scale. In order to simplify the management of information, pain intensity was also defined by a four-grade scale: 0 for “no pain”, 1–3 for “mild pain”, 4–6 for “moderate pain”, and 7–10 for “severe pain” [11]. All data were collected during the first visit in one of the two enrolling centre after obtaining by each participant his/her informed written consent according to the Declaration of Helsinki. The ethics committees of Marche Polytechnic University and Campus Bio-Medico University of Rome approved the study.

Statistical analysis

Age, MEQ-SA score and number of attacks per month and years from migraine onset were collected as continuous variables. MEQ-SA score was also coded as a categorical variable, including the different typologies of circadian phenotype: early rising, moderately early rising, intermediate, moderately late rising, and late rising. Attacks intensity was synthesized as a categorical variable based on VRS-10 scale score, with attacks defined as mild, moderate and severe.

We coded patients’ treatments as categorical variables before the visit. Sex was synthesized as a dichotomous variable.

Continuous variables were compared with t test for independent samples, categorical and dichotomous variables were compared with chi-squared test. Differences and trends among MEQ-SA categories of years of migraine, attack intensity and number of attacks per month were analysed first with analysis of variance (ANOVA), then with a multivariate/generalized linear model (GLM/multivariate) considering years of migraine, number of attacks and attacks intensity as outcomes, MEQ-SA categorical variable as the main predictor, age and sex as covariates. Trend among categories was analysed with ANOVA polynomial linear term testing. Statistical analysis was performed with SPSS 13.0 for Windows systems.

Results

During a one-year period, we evaluated 195 migraine subjects. Three subjects refused to participate, and 21 were excluded for the presence of at least one exclusion criterion. Baseline characteristics of the final sample of 171 consecutive patients are synthesized in Table 1. ANOVA showed that the categorical variable containing the five circadian types was able to identify a significantly different trend both for the monthly attacks frequency (p < 0.0001) and for the disease duration (p < 0.0001). No influence was detected for the attacks intensity at ANOVA (p = 0.959), GLM/multivariate (p = 0.885) and chi-squared tests (p = 0.883).

Table 1 Baseline characteristic of the sample
Full size table

The GLM/multivariate test confirmed the previous findings (Figs. 1 and 2), even after correction for age and sex. The GLM/multivariate estimated marginal means are synthesized in Table 2, and pairwise comparisons between the mean number of attacks, mean years from disease onset and mean migraine attacks intensity among each MEQ-SA category are synthesized in Table 3. We observed that the two categories of early-rising subjects had a significantly lower number of migraine attacks but a longer duration of disease if compared to the late-rising subject categories. The chronotype did not seem to affect migraine attacks intensity in this model.

Fig. 1
figure 1

Differences in the number of attacks per month in MEQ-SA categories (GLM/multivariate)

Full size image
Fig. 2
figure 2

Differences in the years of pathology among MEQ-SA categories (GLM/multivariate)

Full size image
Table 2 Estimated marginal means of GLM/multivariate model adjusted for age and sex
Full size table
Table 3 Pairwise comparison of estimated marginal means at GLM/multivariate model
Full size table

Discussion

The findings of the present study suggest that chronotype may influence some clinical aspects of migraine. In particular, we found that the late-rising chronotype is associated to an increased migraine attacks frequency. The correlation between attacks frequency and serotine chronotype may be, at least in part, explained by the fact that subjects with this chronotype are misaligned to the usual social and working habits. Specifically, in morning-type subjects (the larks), circadian rhythms are not disturbed by daily activities and the number of attacks may depend mostly by individual migraine predisposition and other triggers. On the other hand, in serotine subjects (the owls), the interruption of sleep and the consequent lower sleep quality, due to the necessity to engage in social and work activities scheduled in the morning, can trigger migraine attacks. This is particularly true for school activities or employment duties [12].

In our study, serotine chronotype subjects had a lower number of years from migraine onset when they underwent a first evaluation in a headache centre with respect to morning-type subjects. This aspect might be related to the different attacks frequency, because owl subjects with many attacks tend frequently to call on the emergency room or headache-specialized centres. Accordingly, an increased number of disabling attacks could stimulate patients to an earlier search for medical care [13].

Many different medical conditions, ranging from gastrointestinal diseases to autoimmune pathologies, show a clear oscillation in their clinical expression linked to the different phases of the day [14]. The same pattern is also evident in some neurological and psychiatric conditions, including Parkinson and Alzheimer’s diseases, depressive and bipolar disorders and craving for alcohol or heroin. In this respect, the so-called ‘chronodegeneration’ is considered as a primary factor in disease progression [14,15,16]. The knowledge of such patterns is necessary in an attempt to improve patients’ care by modulating timing for therapy administration or managing possible external temporal triggers.

Migraine is a complex clinical condition with different aspects still to be fully defined [17]. According to one pathophysiologic hypothesis, migraine is considered a sort of “oscillopathy” of the pathway between thalamus and cortex. Transition between wakefulness and sleep is regulated by a thalamocortical oscillator that produces slow and synchronized oscillations (theta and delta rhythms). During sleep, the oscillator is activated, while in wakefulness, the oscillator is not functioning and fast rhythms are dominating [18]. Different studies hypothesized that thalamocortical dysrhythmia may be responsible for the altered synchronicity in migraine [19, 20]. Accordingly, a recent study showed that patients with migraine with aura have a peculiar connectivity pattern with an increased cortical activation after visual stimulation. A reduced cortical activation (susceptibility) is, on the contrary, present in patients not experiencing aura symptoms [21].

The relationship between sleep and headache is complex. While hypnic headache or paroxysmal hemicrania during REM sleep are strictly related to nocturnal rest [22], the correlation between sleep and migraine is less clear. Many studies reported that the majority of attacks occur in the morning, but this finding is still controversial [20, 23]. Sleep deprivation or marked alteration of bedtime could predispose to migraine [5, 24]. More in general, sleep alterations tend to increase all the typologies of morning headaches [25]. This is the reason for our choice to not include in the study night-shift work because these people presented a well-documented alteration of sleep cycles and a predisposition towards different types of headaches [26]. At the same time, we decided to avoid the inclusion of subjects with a positive anamnesis for sleep alterations [27, 28].

Migraine attacks may be characterized by different temporal patterns, with several triggering factors determining diurnal or awakening onset, so configuring different phenotypes. Recent studies support the hypothesis of a central role of hypothalamus in migraine pathogenesis. Schulte et al. sustained that functional changes in hypothalamo-brainstem connectivity could be the real driver of migraine attacks [29].

Circadian rhythm sleep disorders (CRSDs) are described as chronic alterations of sleep–wake rhythm secondary to alterations of the circadian timing system or to disagreement between endogenous circadian rhythm and the sleep–wake times ruled by school or work activities [2]. According to our data, we could hypothesize that migraine is a sort of CRSD generated by an alteration of circadian rhythm imposed by living and working activities in predisposed subjects. On the contrary, when the sleep–wake rhythm is similar to the endogenous rhythm, migraine subjects may present a low frequency of severe headaches. According to some authors, migraine could be interpreted as the result of a main effort to maintain homeostasis contrasting the desynchronization between circadian phenotype and lifestyle [30].

Therapeutic implications of circadian rhythms in neurological diseases presenting with paroxysmal events as epilepsy have been largely investigated. Different studies, evaluating the influence of circadian typologies, showed that different time of drug administration could modify the effectiveness and tolerability of treatment [31, 32]. Migraine and epilepsy share different pathophysiologic aspects including an alteration in the brain electric connections. Both present paroxysmal crisis and share some treatment approaches. Based on these considerations, a “chronotherapy” could also be taken into consideration for migraine patients. Accordingly, a personalized time for drug administration could improve treatment effectiveness.

Different studies showed low melatonin levels in subjects with chronic migraine [33]. In migraine patients, higher levels of melatonin are dosed in the days free from headache [34, 35]. Based on these data, some trials evaluating melatonin as a possible therapeutic approach for migraine were performed, but the results were controversial. Only a dose of 3 mg/day was significantly associated to a positive effect [34] but a more rational employment of melatonin therapy, based on patients’ chronotype, should be considered to fully evaluate the potentiality of this therapeutic approach.

Study limitations

The present study has several limitations. The most relevant one is related to the fact that we did not associate the MEQ-SA score to any other instrumental evaluation as actigraphy, body temperature rhythm, melatonin secretion rhythm or polysomnography. However, some studies show that MEQ-SA score is a reliable instrument to define ideal sleep time, which is not related to sleep phase but to sleep quality [12]. Moreover, we did not perform a follow-up because our study had a cross-sectional design. This approach did not allow to fully determine the temporal ordering of the various items and the exact causality in the relationship between chronotype and migraine characteristics. Moreover, by excluding patients with sleep alterations reduced the possibility to generalize our results to all migraine subjects. In this respect, we decided to include only patients affected by migraine without aura to obtain a homogeneous sample.

Conclusions

In conclusion, our findings suggest that chronotype may be considered an influencing factor on some, relevant migraine characteristics and could represent a significant aspect to evaluate and consider in the attempt to increase the possibilities to optimize the therapeutic management of patients.