Abstract
Background
Cluster headache (CH) attacks are accompanied by cranial autonomic symptoms indicative of parasympathetic hyperactivity and sympathetic dysfunction ipsilateral to the pain. We aimed to assess cranial autonomic function in CH patients during the remission phase of cluster headache.
Materials and methods
During a remission phase, 38 episodic CH patients underwent the following: dynamic pupillometry, measurement of the superficial temporal artery diameter by ultrasound, and measurement of the retinal vessel diameters from digital retinal photographs. Pupillometry was also performed on 30 age- and sex-matched healthy controls.
Results
Thirty patients were included (27 men, three women, mean age 50.2 years ± 12.6). Seven patients reported occasional side shift of their headache, but with a clear predominating side. Significantly reduced average pupillary constriction velocity and retinal venular diameter on the CH pain side were found. There was no asymmetry of the superficial temporal artery diameters. Compared to healthy controls, cluster patients displayed bilaterally reduced pupillary average and maximum constriction velocities, reduced constriction in percentage and increased latency of the light reflex.
Conclusions
The present findings indicate a bilaterally reduced cranial parasympathetic tone in CH patients in remission phase, with significant lateralization to the CH pain side. This implies a central origin, and a central pathophysiological model of CH is discussed.
Keywords
Introduction
Cluster headache (CH) is a primary headache disorder in which episodes of intense unilateral, periorbital pain are accompanied by ipsilateral cranial autonomic symptoms (1). The autonomic symptoms seem to be both of sympathetic and parasympathetic origin. Lacrimation, conjunctival injection, nasal congestion and rhinorrhea are thought to be symptoms of parasympathetic hyperactivity elicited via activation of the trigemino-parasympathetic reflex (2). The efferent limb of this reflex is the parasympathetic fibers following the seventh (facial) cranial nerve. Secondarily, the parasympathetic activation dilates the internal carotid artery, supposedly resulting in compression of the oculosympathetic fibers running along the vessel wall, giving rise to the Horner-like syndrome of miosis and ptosis on the pain-affected side.
The anatomical basis for this hypothesis is documented (3,4), but the pathophysiological mechanisms remain unclear and physiological examinations during CH attacks have been somewhat contradictory (5,6). Understanding the complex mechanisms of the autonomic symptoms will give valuable insight to the pathophysiology of CH. The basic site of dysfunction is not known, but the hypothalamus has been proposed as the locus in quo (7). An important question is consequently whether the autonomic symptomatology is also of central origin.
We aimed to study the cranial autonomic nervous system in CH remission phase, and wished to do so by using non-invasive methods that did not interfere with normal autonomic function. Measuring the pupillary light reflex by pupillometry is an easy, accessible method to assess several cranial autonomic pathways. Cranial vessel diameters may also be used as an expression for cranial autonomic tone. Broadly speaking, the sympathetic nervous system causes vasoconstriction and the parasympathetic nervous system vasodilation, although there are exceptions to this rule and the parasympathetic fibers innervate only a limited number of vessels. We used ultrasound of the superficial temporal arteries and digital photographs of the retinal arterioles and venules as substrates for measuring vessel diameter on the headache pain side (affected side) versus contralateral (non-affected) side.
Methods
In 2011, we surveyed all episodic CH patients (International Classification of Headache Disorders, second edition (ICDH-2)) registered at the two regional hospitals of Northern Norway between January 1, 2000 and December 31, 2010 (8). The subjects who agreed to participate in further studies, and who were living in Nordland County, were subsequently invited to the outpatient clinic at the Nordland Hospital Trust for examination. The clinic covered travel expenses; otherwise there was no payment for participation. All current medication and previous ophthalmological problems or surgeries were noted, and patients with reported headache attacks less than four weeks prior were excluded. Demographics and data on attack-related autonomic symptoms were recorded. The patients underwent a standardized dynamic pupillometry (performed by HKO) and ultrasound measurement of the luminal diameter of the superficial temporal arteries (KBA), before digital retinal photos were captured for retinal vascular diameter grading (TvH). The latter two researchers were blinded to the CH pain side.
In addition, we performed standardized pupillometry on 30 age- and sex-matched controls. The controls were recruited among hospital staff and visiting relatives of inpatients. Exclusion criteria for the control group included a history of diabetes, glaucoma, recurring headache or eye surgery.
Dynamic pupillometry of the light reflex
A handheld NeurOptics PLR-200™ pupillometer was used for recordings. High accuracy of this device with no more than maximum 0.3 mm from infrared photography measurement is documented (9). All measurements were made in a specialized room equipped for the procedure. The patients were adapted to darkness (1 lux ambient illumination) for two minutes before every measurement, recording one pupil at a time starting with the left side. While sitting in a chair, they were asked to focus on a black spot on the wall 5 meters away. The device recorded maximum diameter of the pupil (Max), before a light of 180 micro Watts and 802 milliseconds’ duration flashed. With data sampling at 32 frames per second, the following values were obtained: Minimum diameter of the pupil (Min), constriction (Max-Min/Max) in percentage (Con), onset of constriction in seconds (Lat), average and maximum constriction velocity in millimeter per second (ACV and MCV), average dilation velocity (ADV), and the total time in seconds to recover 75% of pupil resting size after peak constriction (T75).
Figure 1 describes the pupillary reflex pathway. The sympathetic fibers from the superior cervical ganglion mediate pupil dilation, and pupillometric measures of sympathetic function include Max, ADV and T75. The parasympathetic fibers following the third cranial nerve cause pupil constriction, and measures of the parasympathetic function include Con, MCV and ACV. As the T75 increases linearly with the amplitude of the light reflex (10), calculations were made using the T75/A ratio, where A is the amplitude of the reflex in millimeters (Max-Min).
The pupillary light reflex. Light evokes the pupillary light reflex, which is parasympathetically driven (red lines): When light falls on the retina, non-image-forming ganglion cells (intrinsically photosensitive ganglion cells) convey signals to the olivary pretectal nucleus (OPN), which projects to preganglionic parasympathetic neurons in the Edinger-Westphal nucleus (EWN). From here, third-order fibers are carried bilaterally along the oculomotor nerves to the ciliary ganglions (CG). Several short ciliary nerves innervate the iris sphincter muscle, causing constriction of the pupil. Redilation of the pupil seems to be a pure sympathetic (green lines) response. The paraventricular nucleus (PVN) is a major sympathetic premotor nucleus that projects via the intermedio-lateral column (IML) to the ciliospinal center of Budge in the upper thoracic spinal cord. From here second-order sympathetic fibers travel to the superior cervical ganglion (SCG). Third-order sympathetic fibers reach the iris dilatator muscle through the ophthalmic nerve, causing dilation of the pupil. Pupil size reflects the net output of the opposing parasympathetic and sympathetic systems. Locus coeruleus (LC) plays an important role in the pupillary control (black lines), either by contributing to the sympathetic outflow or by attenuating the parasympathetic output by inhibiting the EWN. Several factors like emotions, drugs, arousal and pain influence the activity of the LC and the pupil size. Light also regulates the activity through the complex circadian system (black stippled lines) governed by the suprachiasmatic nucleus (SCN).
Measurement of the luminal diameter of the superficial temporal artery
High-resolution Vivid Cardiovascular Ultrasound Systems (Vivid 7, GE Medical Systems, Little Chalfont, UK) were used. With the individual placed in a supine position, the main trunk of the superficial temporal artery was located by palpation just in front of the ear in the meato-orbital line. A wide band (4.7 13 MHz) linear 1.25 D matrix array transducer was oriented parallel to the artery and a longitudinal section of the main trunk visualized. From a 0.5 cm segment, three diameter measures were obtained (from intima to intima), and the average diameter was used as the reference diameter of the artery.
Measurement of retinal arteriolar and venular diameter
The intraocular pressure was recorded bilaterally using Tono-Pen XL (Medtronic Solan, Jacksonville, FL, USA). Both pupils were dilated with one drop of Tropicamide 0.5% (Chauvin Pharmaceuticals Ltd., Kingston upon Thames, Surrey, England, UK). An FF450 plus IR Fundus Camera (Carl Zeiss Meditec, Jena, Germany) captured optic nerve-centered retinal photos of both eyes. Retinal vascular diameters were measured by computer assistance with IVAN, the updated version of Retinal Analysis software (University of Wisconsin, Madison, WI, USA) (11). All vessels with diameter more than 35–40 µm coursing through the area of one-half to one disc diameter from the optic disc margin were measured, and the six largest of each vessel type were summarized as the central retinal artery equivalent (CRAE) and the central retinal vein equivalent (CRVE) (12). The protocol for grader-interaction on the automated measures was in accordance with previously validated protocols (11) with minor modifications according to the Retinal Vascular Imaging Centre (RetVIC) (Centre for Eye Research Australia, University of Melbourne, Australia).
Hypothesis and data analysis
The null hypothesis was that there is no difference between any of the measurements obtained from the headache affected side compared to the non-affected side in CH patients. Nor is there any difference in pupillary light reflex measured by pupillometry on either side between CH patients and healthy controls. As there was no significant difference between left and right side in the control group, the means of the two were used for calculations. A paired t-test was used for comparing the means between the affected and non-affected side, and independent samples t-test was used when comparing means between patients and controls. In addition, the pupillometry data were also analyzed with repeated-measures analysis of variance (ANOVA), where side was set as the within-subjects factor (affected vs. non-affected side in patients, right vs. left side in controls) and patients vs. control was set as the between-subjects factor. The probability value (alpha) was considered consistent with the null hypothesis when larger than 0.05.
The data were analyzed using the SPSS Software package for Windows version 20 (SPSS Inc, Chicago, IL, USA). The Regional Ethics Committee of Northern Norway reviewed the research protocol and recommended the study. The Norwegian Data Protection Authority approved data registration.
Results
Demographics, clinical characteristics and medication of the cluster headache patients included in the study.
y: years; CA: cluster attacks.
Self-reported accompanying symptoms during cluster attacks.
Ten people were diagnosed with hypertension, and one used beta-adrenergic blocking agents. The rest used calcium channel blockers, diuretics or angiotensin-II receptor antagonists. Five reported permanent use of Verapamil. Another individual reported use of selective α1-blocker in treatment of benign prostate hyperplasia. The controls were the same age and gender as the CH patients (27 men, three women, mean age 50.3 years ± 13.6). Five controls reported diagnosis of hypertension, and two used beta-blockers.
Pupillary response to light on affected and non-affected side in 30 episodic cluster headache patients, and the mean values of pupillometric measures in 30 healthy controls.
aAffected vs. non-affected side, paired samples t-test. bCluster headache patients vs. controls, independent samples t-test. T75: 75% of pupil resting size after peak constriction; A: amplitude of reflex in millimeters; df: degree of freedom.
Temporal artery diameter and retinal vascular diameters on affected and non-affected side in 30 episodic cluster headache patients.
aPaired samples t-test.
There were no differences in temporal artery diameters. Retinal venular diameter (CRVE) was significantly smaller on the affected vs. non-affected side. Retinal arteriolar diameter (CRAE) were also smaller on the affected side, but this was not statistically significant. There was no significant correlation between retinal vascular diameters and ACV measures.
Discussion
Attenuation of the light reflex
This study documents significantly attenuated pupillary light reflexes in CH patients in remission phase compared to healthy controls. The reduction is bilateral, but more pronounced on the headache-affected side, and must be caused by reduced parasympathetic outflow over the third cranial nerve. Reduced light reflexes within 48 hours of spontaneous migraine attacks have also been shown in a previous study (13). However, in that study repeated examination after one week no longer showed any difference between migraine patients and healthy controls. Attenuated light reflexes may therefore be a general, possibly transient, phenomenon following episodic headache. As we did not record exact time elapsed from last attack in our patients, we are unable to correlate this to the pupillometric findings and assess whether the light reflex normalized with time.
It is known that the LC exerts an inhibitory influence on the pupillary light reflex via inhibition of the Edinger-Westphal nucleus (EWN) (14). The LC also contributes to oculosympathetic outflow via excitatory projections to preganglionic sympathetic neurons, and therefore exerts a dual effect on pupillary control. Noxious stimuli dilate the pupils (reflex dilation) without affecting the light reflex, most probably via activation of the LC. However, anxiety and conditioned fear (where a neutral stimulus produces anticipation of a noxious stimulus) produce both reflex dilation and attenuation of the light reflex, suggesting activation both of sympathetic and parasympathetic premotor neurons of the LC. The latter probably involve activation of LC via the amygdala. In a recent functional magnetic resonance imaging (fMRI) study of a CH patient, brainstem activation during pain was interpreted as a behavioral “fight-or flight” response, in accordance with a state of “increased readiness” (15). The remission phase of CH may represent a state similar to conditioned fear.
The LC is the main noradrenergic nucleus of the brain. It is connected to a number of regions in the hypothalamus and the forebrain, and plays part in a number of vital functions including wakefulness, responses to stress, regulation of emotion and modulation of pain (14). The LC densely innervates the neurons of the somatosensory trigeminal nucleus, and activation of the LC inhibits the neurons of the trigeminal nucleus involved in pain perception (16). Little is known about the mechanisms that terminate headache attacks, but LC may play an important role. In a recent study, migraine patients were shown to have increased hypothalamic functional connectivity with a number of brain regions involved in autonomic functions, including the LC (17).
Most previous pupillometry studies of CH have recorded pupillary responses to topically induced pharmaceuticals, acting on the autonomic pathways in the eye. These studies have mainly found oculosympathetic hypofunction similar to a third-order Horner syndrome on the headache-affected side, both in active and inactive phase (18,19). In the present study, there was no statistically significant oculosympathetic dysfunction. The light reflex may be a less potent trigger of sympathetic response than pharmacological testing, but most of these studies have used the contralateral eye as control and not a control group, which may have caused them to miss both bilateral and parasympathetic dysfunction. Drummond recorded pupillary responses to light in 30 CH patients immediately before, during and within one hour after attack (20), and found increased constriction velocity and decreased dilation velocity on the symptomatic side both during and in between attacks. We have documented the opposite in our patients, but the findings are not readily contradictory. Drummond’s patients were evaluated in an active phase, whereas ours were in a remission phase and thus in a different autonomic state. As expected, the increased parasympathetic activation found during attacks in Drummond’s patients was less pronounced one hour later. Our findings may indicate a chronic downregulation to avoid further attacks.
Superficial temporal artery diameter
We did not find any differences in superficial temporal artery diameter between the affected and non-affected side in the CH patients in remission phase. This is in accordance with a previous study of nine CH patients, investigating the luminal diameter of the superficial temporal artery both during and in between attacks (21). The authors found no asymmetry outside attacks, whereas a significant decrease in diameter on the non-affected side was observed during attacks. This was interpreted as a general pain-induced arterial vasoconstriction with relative arterial dilation on the headache side, due to ipsilateral trigemino-parasympathetic activation. An alternative explanation would be an underlying sympathetic dysfunction on the pain-affected side.
Retinal arteriolar and venular diameter
The ocular blood supply comes from the ophthalmic artery, of which 10% supplies the retinal arterioles via the central retinal artery, and 90% supplies the choroidal and ciliary vascular networks (22). We found significantly smaller retinal venular diameter on the headache-affected side compared to the non-affected side in CH patients. The vessels of the retina are autoregulated, and not innervated by autonomic fibers (23). However, the choroidal and extra-ocular vessels, including the central retinal artery up to the lamina cribrosa, have shown a rich supply of autonomic vasoactive nerves (24). In addition, both α- and β-adrenergic, muscarinic and angiotensin II receptors have been located in the retinal vessels (25–27).
The retinal autoregulation is mediated through arteriolar caliber changes and capillary recruitment, and involves myogenic responses, metabolic signals and endothelial function. The overall goal for autoregulation is to maintain optimal retinal blood flow under conditions of varying perfusion pressure and metabolic needs (23). The mean ocular perfusion pressure equals the mean blood pressure in the ophthalmic artery minus the pressure in the veins leaving the eye, and the venous pressure is close to the intraocular pressure (IOP).
During CH attacks, conjunctival injection, dilation of the ophthalmic artery, and increased ocular pulsatile flow imply that the orbital circulation on the headache-affected side is subject to vasodilation and increased blood flow (6,22). In line with this, one study of changes in blood flow velocities of the ophthalmic artery in CH patients during the Valsalva maneuver concluded with an ipsilateral tonic vasodilatory state during cluster periods (28). In the remission phase they found increased peripheral vascular reactivity, both compared to the non-affected side and to controls, reflecting a low resistance vascular bed on the headache-affected side with enhanced vasodilatory responses to stimuli. Another study found bilaterally reduced IOP and ocular pulsatile flow in the remission phase compared to controls (22), and the authors argued that the most reasonable explanation for this was increased pre-ocular vascular resistance in the ophthalmic or internal carotid arteries.
In the present study, both arteriolar and venular diameters were reduced on the headache-affected side, although only statistically significant for the venular diameter. As retinal vessels are autoregulated, our results are most consistent with reduced total blood flow in the orbital area. What causes this flow reduction is unclear—could it be upstream or downstream influences, and is it of neural, vascular smooth muscle or endothelial origin? As the knowledge in general on this topic is limited, we can only speculate on what causes the asymmetry in our subjects. Repeated parasympathetic storms during CH attacks, or the pain itself, may produce local vasoactive substances or environmental changes that affect the retinal vessels after the bout has passed. Alternatively, the findings may represent a state of altered autonomic tone, as was found in the Valsalva maneuver study (28). The central retinal artery is innervated both by sympathetic and parasympathetic nerve fibers (24), and the reduced extra-ocular flow could result either from a relative increase in upstream sympathetically derived vasoconstriction, or a relative decrease in upstream parasympathetically derived vasodilation. As we have not found any signs of increased oculosympathetic activity, the former seems less likely.
A central origin of symptoms?
The parasympathetic symptoms of CH attacks are mediated via the rostral, ventrolateral subregion of the superior salivatory nucleus (SSN), whose neurons project to the lacrimal glands, nasopalatine mucosa and the cerebral and choroidal vasculature via the pterygopalatine ganglion (29). The LC inhibits the parasympathetic activity of these neurons via inhibitory α2-adrenoceptors in the same way it inhibits the EWN. In consequence, activation of the LC may contribute to both our main findings. Could this represent an upregulated activity of the LC during a CH remission phase?
The striking periodicity of CH attacks has pointed at the hypothalamic suprachiasmatic nucleus (SCN) as a possible attack trigger. The cells of the SCN display an intrinsic rhythmicity, which synchronizes with the external dark-light cycle conveyed from the retina through the retinohypothalamic tract, and control the biological rhythms of the body (30). Activation of the SCN causes reduction of melatonin synthesis, and plasma melatonin has been reported to be reduced in CH patients, particularly in active cluster periods (31). When light hits the retina, SCN inhibits the sympathetic premotor neurons of the paraventricular nucleus of the hypothalamus (PVN), resulting in reduced sympathetic output to the pupillary dilator muscle, and the pupil constricts. At the same time, connections to the LC activate premotor sympathetic neurons, working to redilate the pupils. The PVN is the main descending autonomic nucleus of the hypothalamus, and like the LC, it is also involved in pain modulation (32). We suggest that the interconnections between these three nuclei are involved in CH pathophysiology.
Recently, anatomical studies have documented direct descending projections from the PVN to the spinal trigeminal nucleus caudalis (Sp5C), which receives trigeminal afferent sensory input (32). These projections are bilateral, but with a clear ipsilateral predominance, and most of them project to the ophthalmic (V1) area of the Sp5C. Stimulation of the PVN at the spinal level produces antinociception, whereas lesions facilitate nociception, and it probably modulates trigeminal pain in the same way (33). Hypothetically, a downregulation of the premotor sympathetic activity of the PVN, caused by activation of the SCN or other endogenous or exogenous factors, may result in reduced damping of trigeminal somatosensory inflow, producing bilateral, but predominantly ipsilateral, disposition to trigeminal pain. The PVN may therefore be involved in the initiation of the attack, introducing pain and oculosympathetic hypofunction, whereas the LC may be involved in the termination of the attack by reducing trigeminal firing, inhibiting cranial parasympathetic output and increasing the oculosympathetic output. In other words, CH may represent a condition in which chronic or transient downregulation of PVN is compensated for by an upregulation of the LC. The evidence of a third-order oculosympathetic dysfunction during attacks does not contradict this hypothesis, as the assumption of an attack-induced internal carotid artery dilation may still be true. A subtle central sympathetic dysfunction may even help explain why minor changes in the carotid vessel wall should cause a Horner-like syndrome, and why not all pupillometric studies have found a classical postganglionic ocular sympathetic dysfunction.
Strengths and limitations
The number of participants included in this study is small, and consequently conclusions must be drawn with caution. On the strong side, the CH diagnoses of the patients included were well validated, and the recordings were made with standardized methods by the same researcher each time. The researcher was blinded to the headache pain side when measuring the temporal artery and retinal vessels, which strengthens the findings. The pupillometry measurements were purely computer produced and not prone to subjective influence. The greatest weakness of the study is that we did not record systemic blood pressure of the patients during retinal photographing, and that we did not have healthy controls to whom we could compare the retinal vascular diameter findings. Several of the participants reported use of vasoactive drugs, which may affect both IOP and retinal vessel diameters. However, such medication should affect both eyes to the same extent, and cannot explain the asymmetry of the present findings. Regarding the attenuated pupillary light reflex in CH patients, this could be caused by a higher level of anxiety during testing compared to controls. To compensate for this, one could have tested pupillary reflexes not only under resting conditions, but also during different states of arousal and pain, when the LC is likely to be active.
Conclusion
The pupillometric findings in this study document bilaterally reduced pupillary parasympathetic tone in CH patients in a remission phase, but with lateralization to the headache-affected side. Asymmetry of the retinal vasculature may also be interpreted as reduced parasympathetic tone in the orbital circulation ipsilateral to the pain.
Clinical implications
Cluster headache (CH) attacks are accompanied by autonomic symptoms involving both the sympathetic and parasympathetic nervous systems. Thirty episodic CH patients (27 males, three females) in remission phase underwent pupillometry of the light reflex, measurement of the superficial temporal artery diameter by ultrasound, and measurement of the retinal vessel diameters from digital retinal photos. The patients displayed significantly attenuated pupillary light reflexes in both eyes compared to healthy controls. The findings were lateralized to the headache-affected side. The retinal venules were significantly smaller on the headache-affected side compared to the contralateral side, indicating a reduced total blood flow to the orbital area. The findings imply reduced parasympathetic function over the third and seventh cranial nerves in the pain-free state of CH.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or non-profit sectors.
Conflict of interest
None declared.
Acknowledgments
The authors would like to thank Professor Tom Wilsgaard at the Department of Community Medicine, UiT the Arctic University of Norway, for help with the statistical analysis.