The cold pressor test in interictal migraine patients – different parasympathetic pupillary response indicates dysbalance of the cranial autonomic nervous system

A hallmark of migraine attacks is a concomitant variety of vegetative symptoms, such as loss of appetite, nausea or vomiting, with some patients even showing signs of activation of the cranial autonomic nervous system (e.g. parasympathetic system), such as lacrimation, sweating, rhinorrhea or nasal congestion [1‐3]. Furthermore, there is increasing evidence that modulation of the parasympathetic nervous system might be useful in the prevention of, or the cessation of migraine attacks [4, 5]. Otherwise, previous studies of autonomic function in migraine showed inconclusive and even conflicting results regarding the role and interaction of the sympathetic and parasympathetic system.

The advantages of pupillometric testing are that it a) assesses both the sympathetic and parasympathetic innervation simultaneously concerning the pupillary reflex, and that it is b) a well-established method to evaluate autonomic function in the innervation area of the cranial nerves for various conditions [6‐8]. The analyses of heart rate and blood pressure allow for at least a rough and clinically relevant evaluation of the cardiovascular autonomic nervous system (ANS). In the present study, we directly tested cranial and cardiovascular autonomic responses of migraineurs in the interictal phase during sustained sympathetic stimulation by the cold pressor test (Additional file 1). We thereby specifically tested the hypothesis whether migraine patients, when compared to age- and gender-matched controls, show different autonomic responses of the cranial ANS in the interictal phase as measured by pupillary response.

There were no adverse effects in the control group. In the migraine group one female suffered from a syncope shortly after T2. The data are summarized in Table 1.

Pupillary size and changes in pupillary size depend on many factors (e.g. time, light, environment, sleepiness, emotional state etc.), but reflect in general the balance between the sympathetic (primarily dilatation) and parasympathetic (primarily constriction) nervous system tonus.

At baseline (T0) there was no significant difference in the pupillary and cardiovascular parameters between the migraineurs and the controls, indicating that there are no profound changes in the ANS of migraine patients under normal circumstances. This is in line with a recently published study by Cambron et al., who did not find differences of pupil parameters in migraine patients, neither in the interictal phase nor during migraine attacks [7]. However, at T2 (i.e. five minutes after sympathetic stimulation), the constriction velocity was significantly higher in the migraine patients. This might indicate that the ANS is at least slightly dysregulated in migraine patients also in the interictal, non-headache phase and that sympathetic stimulation can unravel this difference in ANS thresholds. However, the results of previous studies on the ANS thresholds and changes in migraine patients are inconclusive and partially conflictive. At first glance, in great contrast to our data, Mylius et al. showed a significantly slower constriction velocity and a smaller amplitude of pupil constriction within two days after an attack in migraine patients, thus inferring parasympathetic hypofunction [16]. However, for comparison with our data, one has to recognize that the time points of ANS-measurements were different in both studies. While they made their measurements within two days after a migraine attack, this time period was an exclusion criterion for our study, where measurements only more than two days after an attack were recorded. Thus, the data might be conclusive, since migraine patients might suffer parasympathetic dysregulation in the following way: a) lower parasympathetic thresholds under normal circumstances with activation by sympathetic stimulation, as we have shown; b) obvious parasympathetic hyperactivation during migraine attacks possibly triggered by pain in the attack, or vice versa, as a fundamental condition in the pathophysiology of migraine headache [16]; and c) parasympathetic hypofunction directly postictal after the attack, as shown by Mylius [16]. Moreover, Drummond et al. also argued for an increase of the parasympathetic tone during a migraine attack directly related to trigeminal-parasympathetic reflexes, when observing the dilatation of dermal blood vessels during attacks [17].

Corresponding to our findings, a previous study by Tassorelli et al. (15) demonstrated a miotic phase with a maximum at five minutes during the cold pressor test after an initial very short mydriasis in healthy volunteers. Our dataset implies that this physiological parasympathetic pupillary response to the cold pressor-test is more pronounced in migraine patients. This might indeed be an indirect correlate of at least slight parasympathetic dysregulation in migraine.

Pupil dilatation to baseline directly follows pupil constriction. This redilatation process can be divided into two phases: the initial and rapid redilatation phase is rather an effect of withdrawal of the parasympathetic tone than sympathetic activation, whereas the later and slower dilatation phase seems to be an active process induced by peripheral sympathetic innervation [6]. Altogether, we did not record any significant differences between the migraineurs and Ctr in this two-staged pupil dilatation process. However, analyzing the time course of the dilatation process more precisely, there was a slight delay in reaching the maximum dilatation velocity in the migraine group, while velocity itself was unchanged. The migraine group reached the highest dilatation velocity at T2, whereas the Ctr did so at T1, which may be interpreted in terms of a slight dysbalance towards the parasympathetic nervous system (PSNS) in migraine patients. Taken together our findings and the results of the previous studies, there is sufficient evidence of slight dysregulation of the parasympathetic cranial ANS in migraine patients.

Which pathophysiological mechanisms besides a primary cranial autonomic dysregulation might also contribute to the observed differences in cranial autonomic response between the migraine patients and healthy controls: First, it could be due to a difference of peripheral sensory perception and/or central pain processing. Previous studies could demonstrate cutaneous allodynia (CA) for usually not painful sensory stimuli, particularly thermal stimuli, in more than half of patients with episodic migraine during a migraine episode [18, 19]. One study even could show such changes in migraine patients prior to an episode [20]. It is generally accepted that such cutaneous allodynia is a consequence of central sensitization of pain processing pathways and an impairment of the descending pain inhibitory pathways [18, 21‐23]. In fact, these mechanisms can lead to a vicious circle in that sense that recurrent migraine attacks can promote central sensitization, which in turn impairs diffuse noxious inhibitory control (DNIC) [23]. Thus, changes of central sensitization and the descending inhibitory pathways could contribute to the observed differences between migraineurs and the healthy controls by perceiving the cold stimulus during the CPT “more painful”. However, comparable pain rating scores between both groups argue against that hypothesis but cannot definitely exclude it.

Secondly, habituation of sensory stimuli, which is mainly a thalamo-neocortical process, can play also a role. Previous studies have shown that migraineurs have deficits in sensory habituation after repeated stimuli of different sensory modalities (i.e. visual, somatosensory) even in the interictal phase [24, 25]. Coppola et al. [26] were able to show, that CPT can significantly change habituation of visually evoked potentials in healthy controls, but not in migraineurs indicating less plasticity of sensory cortical areas. This could result in a faster habituation of the cold stimulus by the CPT in healthy subjects as compared to the migraineurs thus successfully preventing a further continuous increase of pupillary constriction velocity, as shown by our study. However, one would expect that such a habituation deficit is not that specific affecting only constriction velocity, while dilatation velocity not.

Regarding the higher lifetime rate of syncopes (migraineurs: 46% vs. Ctr: 31%) and particularly a higher lifetime risk for repeated syncopes (migraineurs: 13% vs Ctr: 5%) in migraine patients, changes in the cardiovascular autonomic system should be expected [27]. Since we focused on the cranial ANS of migraine patients in that study, we only performed basic cardiovascular monitoring by measuring blood pressure and heart rate at different time points; however, we did not explicitly apply continuous blood pressure measurements and also did not perform analysis of the heart rate variability. The obtained basic cardiovascular responses (i.e. blood pressure and heart rate) to the cold pressor test were comparable in migraine patients and Ctr. The diastole was slightly (not significantly) increased in migraine patients compared to Ctr. Further subclassifying by headache disability by the MIDAS, we observed an only marginally higher resting state diastolic blood pressure in more disabled migraineurs. Shechter et al. explicitly compared three different groups (i.e. disabling migraine, non-disabling migraine and healthy controls) and also did not find significant differences between the three groups when comparing blood pressure response to a psychological stressor.

Cortelli et al. did not find any impairment of the autonomic control of the cardiovascular system in migraineurs interictally [28]. Domingues et al. used two different protocols to provoke a cardiovascular autonomic response, one by mental stress and one by CPT. The latter one was quite similar to our scheme and they also could not find a difference in heart rate and blood pressure after CPT in migraineurs compared to healthy controls [29]. Daluwatte et al. [30] addressed this question of coupling the cranial with cardiovascular ANS in a cohort of healthy children. They also did not find any correlation between PLR and heart rate variability (HRV), despite significant changes in HRV during the different PLR phases [30].

In our study migraine patients and Ctr did not exhibit any differences in sympathetic or parasympathetic regulation of the cardiovascular response at any timepoint (T0-T2) during the CPT. But as already mentioned, we did not apply the necessary gold standard measurements (heart rate variability and continuous blood pressure measurements) therefore to really make clear statements about that issue.

In summary, our data indicate different activation thresholds of the cranial ANS in migraineurs in the interictal phase during sympathetic stimulation. Based on our findings, the role of upper brainstem parasympathetic dysregulation in the pathophysiology of migraine should be further examined and elucidated in more detail. Particularly, the underlying differences in somatosensory processing on the level of cerebral networks and neuronal ensembles might be a target for future therapy strategies.