Introduction
The relation between autonomic nervous system (ANS) dysfunction and common headache disorders (including migraine [
1‐
3], cluster headaches [
4] and tension-type headaches [
5,
6]) has been widely documented. In migraine, a plethora of autonomic symptoms precedes, accompanies and outlasts the headache attacks. These symptoms include, but are not limited to nausea, vomiting, hyperhidrosis, pallor, palpitations, and light-headedness and make an attack that much more intolerable [
7,
8]. An additional clinical significance of ANS dysfunction is the observed increased probability of major cardiovascular disease (CVD - hazard ratio (HR) 1.50, 95% confidence interval (CI) 1.33–1.69), myocardial infarction (odds ratio (OR) 2.2, 95% CI 1.7–2.8), ischemic stroke (OR 1.5, 95% CI 1.2–2.1), and death due to ischemic CVD (HR 1.37, CI 1.02–1.83) shown in patients suffering from migraine with and without auras [
9‐
12]. Schürks and colleagues found that migraine is associated with a twofold increased risk of ischemic stroke, apparent only among people who have migraine with aura [
13]. Thus, to expand on the argument made by Koenig et al. [
9], it is not only important to understand the role of vagally mediated heart rate variability (HRV), but to also better understand overall ANS function among migraine patients and the relationship with cardio- and cerebrovascular comorbidities, using standardized investigations of the ANS.
Research offered molecular explanations for the variety of symptoms seen in migraine patients. One such explanation is the CGRP. The 37-amino acid peptide is a potent vasodilator and plays diverse roles in the human body, influencing blood pressure regulation, angiogenesis, sepsis, arthritis, inflammation and migraine [
14‐
19]. Furthermore, in the central nervous system (CNS), CGRP has been shown to be active in the hippocampus, sets in motion other neuroprotective processes and acts on other brain cells (i.e. astrocytes or oligodendrocytes) [
20]. Conversely it seems to also have an antidepressive effect [
20] and to facilitate the excitotoxic death of hippocampal neurons in a kainic acid seizure model [
21]. Anti-CGRP substances proved effective in providing relief to migraine patients; however, the long-term effects of CGRP modulation are only now beginning to be thoroughly described [
22]. To support this argument, Tringali and Navarra expressed valid concerns in their review, that further long-term observations are required to examine the effects of CGRP-inhibition, as it pertains to autonomic function [
23]. Additionally, clinicians currently have no objective method, with which to evaluate which patients stand to benefit from CGRP modulation.
Researchers investigated ANS function related to migraine and headache disorders since the 1950s. Much of the earlier work, investigating ANS function/dysfunction in migraine patients, was based on the autonomic theory. This idea postulated that much of the pathogenic migraine process could be attributed to the increase in noradrenaline from the nerve endings of the affected blood vessels [
24]. The theory has since been disproven. The resulting ANS function research, however, reported a vast variety of results. Most studies showed reduced sympathetic function in migraine patients; while others reported increased sympathetic function; others still, showed normal sympathetic function. Likewise, the majority of studies reported normal parasympathetic cardiovagal function, while some reported decreased parasympathetic function [
15]. Miglis goes on to describe the variety of methodologies these conclusions were derived from [
15]; ultimately illustrating the need for consistent, standardized testing of the ANS in migraine studies.
In 1985, Ewing and his colleagues suggested a series of tests - which would become the standard for ANS function testing today [
25‐
28]. This series comprises of the deep breathing, Valsalva manoeuvre, orthostatic challenge, and isometric challenge tests. From these, a variety of values can be derived, characterizing autonomic function. Cumulatively, the composite autonomic scoring scale (CASS) combines cardiovagal, sympathetic adrenergic and sudomotor function results into a single score, enabling clinicians to diagnose and monitor disease progression [
26,
28].
Research using standardized ANS testing has aided to evaluate autonomic migraine symptoms – however, there currently exists neither an aggregated, nor a standard set of values, to provide diagnostic or therapeutic evaluation in the clinical or research setting. Koenig and colleagues conducted a meta-analysis of the vagally mediated HRV results in migraine patients versus healthy controls [
9]; meanwhile, Lee and her colleagues conducted a meta-analysis of the electrocardiographic values between the two populations [
29]. Both articles reported differences between migraine patients and healthy controls in their respective investigated parameters. In contrast, there currently exist no meta-analyses which summarize ANS function data gathered using the standardized ANS testing protocol [
28]. To assess current knowledge, we performed a systematic review and meta-analysis of studies comparing ANS function in migraine patients and healthy subjects; focusing on articles which most closely matched the latest standard autonomic testing protocol, described by Novak in 2011 [
28].
Discussion
The present meta-analysis shows that interictal differences in ANS function can be observed in migraine patients compared to healthy controls – that is, all ANS function test values were found to be significantly lower in migraine patients. Meta-analysis revealed a significant main effect with respect to sympathetic adrenergic function (MD
Valsalva manoeuvre = -0.17; Hodge’s g
orthostatic challenge = -0.28; Hodge’s g
isometric challenge = -0.55) in migraine patients – implying that the sympathetic and baroreceptor signalling in these patients was disrupted compared to their healthy peers. Furthermore, a larger main effect was shown for cardiovagal function (Hodge’s g
deep breathing = -0.32; MD
Valsalva manoeuvre = -0.17); involving the vagal nerve in the manifestation of migraine episodes [
25‐
28]. Considered together, the data suggest that ANS homeostasis in migraine patients is lower – compared to healthy individuals – reacting to changes in ANS signalling levels (such as CGRP) with increased sensitivity.
This phenomenon may be due to the increased quantities of circulating autonomic signalling molecules such as CGRP [
58‐
61]. CGRP’s effects outside of the blood-brain-barrier (BBB) have been well documented [
14‐
19]; however, within the BBB (that is, centrally) there is room for discussion. The mere fact that “CGRP and/or its receptor have been found in the cortex, hippocampus, thalamus, hypothalamus, pituitary, striatum, amygdala, cerebellum, and such migraine-relevant sites in the brainstem as the locus ceruleus, raphe nuclei, and the trigeminal nucleus caudalis” [
20] shows the remaining potential to learn about migraine’s pathophysiology. From an ANS perspective, many of these neuroanatomical sites correlate with the autonomic central network [
62]. We are unfortunately limited to speculation at this point, with respect to addressing “cause-and-effect”, as CGRP’s half-life causes sampling difficulties peripherally [
63‐
67] – while CSF testing by means of lumbar puncture has not yet been published. This makes correlating CGRP levels and autonomic function very difficult. Moreover, the paroxysmal autonomic symptomatology, which manifests during peri-ictal migraine phase of the cycling episodes [
68], may represent a point below which ANS function drops – possibly due to CGRP overproduction inherent to the migraine phenotype or possibly due to overproduction or overflow of other neurotransmitters [
58,
59,
69‐
74]. Consequently, this overflow may tip the nervous system into the well-described pre-ictal, ictal and post-ictal phases of migraine [
75‐
77].
These findings bring into question the roles of other prophylactic migraine medication and its influence on the ANS of migraine patients. One possible explanation could be that, due to their lipophilicity, beta-blockers (e.g. propranolol, bisoprolol and metoprolol) [
78,
79] could actually contribute to antagonization of other adrenergic and noradrenergic signalling pathways of the central autonomic network (such as in the insular cortex) [
80]. Further pharmacological effects of beta-blockers and angiotensin antagonists [
81], in terms of their prophylactic roles, was not investigated with respect to the ANS, in the available literature. Controlled migraine trials focused relatively miopically on outcomes such as headache-free days or acute-medication consumption, without necessarily accounting for all of the symptoms which accompany migraine - autonomic symptoms such as drowsiness, nausea, or changes in appetite [
77]. Collecting information regarding autonomic symptoms in migraine should further advance our understanding of this disease as a whole.
Our systematic review identified two articles which conducted their ANS testing during the ictal phase of migraine [
51,
82]. These found no statistically significant difference between the ictal and interictal values; however, the ictal and healthy control values differed statistically in one of the articles [
51]. We hypothesize, based on the data available [
51,
82], that ANS function in the peri-ictal phase of migraine may be even lower than the aggregated values reported in our meta-analysis of interictal data.
It is relevant to note that the included ANS function values (with one exception [
57]) were initially measured in the late 1980s and early 1990s [
4,
51‐
54,
56], when the “autonomic theory” of the pathophysiology of migraine was among the main hypotheses suggested to explain migraine. As the theory was disproven, these ANS function values remained unaccounted for. The discovery, as well as clarification of the physiological roles, of CGRP and other relevant neurotransmitters (such as pituitary adenylate cyclase-activating polypeptide (PACAP), glyceryl trinitrate (producing nitric oxide), etc.) [
58,
59,
69‐
74] allowed researchers to correlate neurotransmitter levels with ANS dysfunction in migraine. Furtherstill, a genome-wide association study of migraine patients found 38 distinct genome loci associated with 44 independent susceptibility markers for forms of migraine [
83]. Among these was the NGF gene (nerve growth factor) which was shown to be associated with hereditary sensory and autonomic neuropathy, type 5 [
84]. Many of the other loci identified have roles either in the structures of the brain where ANS signalling takes place or in human vasculature. In summation, ANS function testing may have a new supporting role as a biomarker of migraine.
Agreements and disagreements with other studies or reviews
Three recent autonomic function, case-control studies were not included in our final analysis due to a lack of data; which were unattainable after attempting to contact the corresponding authors [
36‐
38]. All three studies were conducted after Novak published the standardized version of the ANS testing protocol [
28]. One of these studies postulated that there exists an impairment of the primary autonomic system and/or neurotransmitter function in migraine patients [
38]. Meanwhile the other two studies suggested that there exists an increased vasomotor reactivity in patients with migraine [
36,
37]. These conclusions would appear to agree with the data we aggregated. Other studies looked at autonomic function in migraine patients, either with isolated autonomic tests (exclusive HRV analysis through electrocardiography - ECG) or parts of ANS function testing protocols (HRV using the head-up tilt-table test). Miglis [
15] comprehensively reviewed these ANS investigations conducted in migraine patients – therefore, it was not the aim of this paper to repeat his findings. Rather, we aimed to supplement his work, by accumulating the published ANS values, albeit, for individual tests most similar to the internationally accepted quantitative autonomic testing protocol described by Novak [
28].
Searching Pubmed for meta-analyses investigating ANS function and migraine produced only a handful of results. Of these, none analysed publications which used the protocols suggested by Ewing et al. or Novak [
25‐
28]. The meta-analysis by Lee et al. looked at ECG findings in migraine patients. The initial problem in this study is that two ECG recording methods (24-hour ambulatory vs. short duration) were analysed together – yielding different amounts of autonomic data for analysis [
29]. Further still, the authors analysed certain cardiac autonomic results, excluding other results describing autonomic function in the studied populations (i.e. not using tilt-table test values from the Mosek et al. study [
85]). The meta-analysis by Koenig et al. aimed to analyse the HRV in headache patients vs. controls – using various methodology to arrive at HRV results. While HRV is the beat-to-beat variation of heart rate, the methods ranged from measurements over five minutes to those over 48 h [
9]. Therefore, there currently exist no meta-analyses which summarize ANS function data gathered using the standardized ANS testing protocol.
Potential biases in the review process
Our systematic review faced several potential limitations. We employed a specific set of criteria, to reduce bias; however, these criteria limited the publications which were included – namely, from only three research groups. Additional publications met the inclusion criteria [
36‐
38,
82,
85‐
90]; however, these did not report the required values and the corresponding authors were unreachable, so that the missing information could be obtained. The meta-analysis reviewed studies which measured the ANS function parameters, examined in the latest guidelines to autonomic testing [
28]. Unfortunately, none of the included studies followed these guidelines, nor used the composite autonomic severity score. Furthermore, a high risk of bias (Table
1) could be seen in all but one study, as the peri-ictal ANS function values may differ even more significantly from those of healthy controls, with respect to the length of the shortest prevalence period for the parameter of interest [
53]. That is, interictal ANS testing was conducted once per patient and there exists a high chance that ANS functions may differ when averaged throughout an entire month. Furthermore, perhaps the ictal measurements also differ in relation to when in the migraine cycle, the ANS testing was conducted (pre-ictal vs. ictal vs. post-ictal).
Importantly, the articles published by Havanka-Kanniainen et al. [
51‐
54] did not disclose whether the same study population was used throughout their publications. They cite their previous studies [
51‐
53] in the final article [
54]; allowing us to believe that the data in the final article is original. A test of exclusion found that the three articles did not uniformly affect the significance of the individual tests; moreover, only Valsalva ratio was shown to cross the zero-line upon exclusion of the earlier three results. (Fig.
6) An additional argument for inclusion of all four articles is that the final article [
54] summarized ANS function in 273 migraine patients interictally, while the other articles [
51‐
53] investigated other hypotheses.
Quality of the evidence
The body of evidence concerning ANS function testing in migraine patients is not negligible; however, the structure with which it was conducted (i.e., methodology, reported results) is heterogeneous. We were able to include seven articles, although an additional ten qualified based on respective methodologies [
36‐
38,
82,
85‐
90]. Of the seven articles included, the biggest variation – and thus limitation – was in the results reported. For example, for the isometric challenge, one group reported the mean difference in diastolic blood pressure [
4], the second group reported the maximum change in diastolic blood pressure [
51‐
54], while the third group decided to measure “the average R-R interval during the 15 seconds preceding the contraction … divided by the minimal R-R interval during the contraction period” [
56]. The last group decided to measure BP “before the grip and at the one-minute intervals during handgrip“ [
57]. All four of these variations are correlates of cardiovagal function, but exemplify the inconsistency of autonomic testing at its infancy. Moreover, the meta-analysis of these data required calculating the standard mean differences, due to the inconsistency in scales used by the individual groups. Therefore, this is a large limiting factor of the results published at this time and, by extension, of our meta-analysis.
Overall completeness and applicability of evidence
This meta-analysis offers a complete and systematic overview of the published ANS function tests which are relevant to examine in migraine patients. The paper presents the values expected in this patient population. In the composite autonomic severity score (CASS), initially suggested by Low [
26], sudomotor function testing is also one of the three main components. This, however, was not initially part of Ewing’s suggested testing methodology [
25] and, therefore, it was impossible to conduct a meta-analysis using the CASS. Moreover, articles citing the latest autonomic testing protocol (later than Novak’s 2011 article [
28]) in their methodology, did not include all the values required for our meta-analysis nor sudomotor function results [
36‐
38].
Conclusion
This systematic review and meta-analysis shows – with the limited data available – that ANS function is significantly impaired in migraine patients. The ANS values included in this meta-analysis were gathered during the interictal phase of the patients’ migraine cycles – more precisely, without paroxysmal autonomic symptoms associated with the peri-ictal migraine phase. The data suggest, ANS function in migraine patients operates at a lower threshold of homeostasis during the interictal phase of the migraine cycle.
Implications for Methodological Research
The impact of autonomic migraine symptoms – as well as increased likelihoods of cardio- and cerebrovascular events – go underappreciated in daily clinical practice. The aggregated results from the meta-analysis allow future research questions to have a reference for ANS function in the migraine population.
Even though autonomic nervous system dysfunction cannot lead to migraine diagnosis, more attention on ANS dysfunction may help to further elucidate its role as a biomarker of migraine and improve the management of migraine patients. Future research using smartphone headache diaries would also benefit from gathering the autonomic prodromal symptom data, to build upon our presented findings and further elucidate the pathophysiology of individual migraine attack. This should help establish earlier warning signs, which ultimately can benefit patient guidance, regarding administration of abortive migraine medication – such as triptans – which show greater effect when administered earlier in the migraine attack phase. Additionally, ANS testing offers an extra method with which researchers can quantify the effect of increased presence of CGRP – or perhaps other neurotransmitters – found in migraine patients [
14,
16,
58‐
61,
69‐
74,
91,
92].
In light of the growing use and effectiveness of anti-CGRP-mAb therapy, this meta-analysis should offer a foundation upon which further ANS function research – as well as clinical trial research – can create future experimental methodologies, which more closely observe (in addition to the standardized side-effect and severe adverse event reporting) the effects of these new and rapidly developing therapies.
Contributions of Authors.
ARP and CW conceived the study and developed the protocol with KZ. ARP was responsible for data collection and statistical analysis supported by KZ and CW. The manuscript was drafted by ARP and revised by KZ and CW. All authors approved the final version.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.