Abstract
Background
Calcitonin gene-related peptide plays a key role in cluster headache pathophysiology. It is released from the trigeminal nerve, which also innervates the eye. In this study, we tested if tear fluid calcitonin gene-related peptide measurement detects elevated calcitonin gene-related peptide levels in cluster headache patients compared to controls.
Methods
Calcitonin gene-related peptide concentration in tear fluid and plasma of 16 active episodic and 11 chronic cluster headache patients (all outside acute attacks) and 60 controls were assessed using ELISA.
Results
Cluster headache patients without use of attack abortive medication in the last 48 h showed significantly elevated tear fluid calcitonin gene-related peptide levels (1.78 ± 1.57 ng/ml, n = 17) compared to healthy controls (0.79 ± 0.74 ng/ml, p = 0.003) and compared to cluster headache patients who had used attack abortive medication in the last 48 h (0.84 ± 1.40 ng/ml, n = 10, p = 0.022). High calcitonin gene-related peptide levels in cluster headache patients were independent of the occurrence of a cluster headache attack in the last 48 hours (no attack: 1.95 ± 1.65 ng/ml, n = 8; attack: 1.63 ± 1.59 ng/ml, n = 9, p = 0.82) as long as no acute medication was used. No significant difference in tear fluid calcitonin gene-related peptide levels between episodic (1.48 ± 1.34 ng/ml) and chronic cluster headache patients (2.21 ± 1.88 ng/ml, p = 0.364) was detected. In contrast to these results in tear fluid, there were no significant group differences in plasma calcitonin gene-related peptide levels.
Conclusion
This study shows that active cluster headache patients have increased calcitonin gene-related peptide levels in tear fluid compared to healthy subjects, which are reduced to control levels after intake of attack abortive medication. Calcitonin gene-related peptide measurement in tear fluid is non-invasive, and has the advantage of allowing direct access to calcitonin gene-related peptide released from the trigeminal nerve.
Keywords
Introduction
Calcitonin gene-related peptide (CGRP) is a 37 amino acid neuropeptide derived via alternative splicing from the calcitonin gene and is expressed in ∼50% of the trigeminal C fibers (1,2). CGRP release from the trigeminal nerve, especially its first division (3,4), is known to play an important role in cluster headache (CH). In episodic CH, external jugular vein blood CGRP concentrations significantly increase during cluster headache attacks and decrease after use of attack abortive medication (5–7). CGRP levels are also significantly elevated outside acute attacks in active episodic CH patients, compared to episodic CH patients in remission (6,8). In addition, intravenous CGRP administration evokes cluster-like headache attacks in CH patients (9). A recent randomised controlled trial showed a preventive effect of the CGRP antibody galcanezumab in episodic CH (10). Unfortunately, the parallel trial in chronic CH missed its primary endpoint (11).
In summary, detection of CGRP levels in CH is crucial to promote research on CH pathophysiology and therapy, and might even have a role in individual prediction of the effect of CH preventive drugs in future. However, existing methods for CGRP determination have important limitations. Detection of CGRP in external jugular blood is invasive, and the reliability of detection in peripheral blood (antecubital vein) has been questioned due to the short half-life of CGRP in blood and strong dilution (12,13). As the outer eye, including the cornea and conjunctiva, is innervated by the first division of the trigeminal nerve (1,14–16), we propose that detection of CGRP in tear fluid may be an alternative, non-invasive and more direct way to detect CGRP release from trigeminal fibers in CH patients. The method has been established in a previous study in migraine patients (17).
The goal of the present study was to establish and validate the detection of CGRP in tear fluid in CH patients. Based on the known data from external jugular vein blood measurements cited above, we hypothesised that active CH patients would exhibit elevated tear fluid CGRP levels compared to healthy controls as long as they have not taken attack abortive medication, and that the detection of CGRP in tear fluid would be more sensitive than in peripheral blood. Thus, we obtained tear fluid and antecubital vein plasma samples from CH patients with and without recent intake of attack abortive medication, and from healthy controls.
Methods
Study population
Participants were recruited from our outpatient headache centre and by advertisements at the University Hospital of the Ludwig-Maximilians-University Munich between April 2016 and November 2019. The study was conducted in accordance with the Declaration of Helsinki and was approved by the ethics committee at the Ludwig-Maximilians-University Munich (526-13). Prior to participation, participants gave written informed consent.
Inclusion criteria for patients were: i) age between 18–65 years and ii) a diagnosis of cluster headache according to ICHD-3 criteria (18), based on a thorough interview with a headache specialist, an unremarkable neurologic examination (except possibly the presence of miosis on the headache side) and a normal cranial MRI. Exclusion criteria for patients were: i) treatment with a CGRP or CGRP-receptor antibody, ii) episodic CH in remission (i.e. outside bout) and iii) presence of a CH attack shortly before sampling, with or without use of acute medication (<3 h).
Inclusion criteria for control subjects were i) age between 18–65 years, ii) less than two mild headache days/month without any migrainous or cluster headache features and iii) no intake of analgesics in the last 48 hours.
Exclusion criteria for all subjects were as previously described (17): i) any ophthalmological condition such as infection or allergic conjunctivitis, ii) use of contact lenses on the day of examination, iii) blood pressure ≥140/90 mmHg at the time of examination or pre-existing arterial hypertension (19), iv) any other severe medical, neurological or psychiatric condition and v) being pregnant or breast-feeding.
In total, 37 patients were assessed for eligibility. Of these, seven patients were excluded because they did not fulfil the inclusion criteria and three patients were excluded due to insufficient amount of tear fluid sampled, leaving 27 patients for final analysis. Seventy-three controls were assessed for eligibility. Eight controls were excluded due to missing inclusion criteria, and in five controls insufficient tear fluid could be obtained; in total, 60 patients were included for final analysis. Subjects were investigated on the same day as their enrolment in the study.
Study procedure
Tear fluid and plasma sampling was performed as previously described (17). In short, sampling was conducted between 9:00 AM and 5:00 PM in a non-fasting condition. We assessed CH attack frequency, intensity and duration in the last week, the attack side, and the presence of CH attacks and/or intake of acute headache medication during 48 hours prior to sampling. Then, participants rested supine for 5 min and blood pressure was measured.
Next, tear fluid was collected separately from the right and left eye. A plastic capillary was used at the lateral canthus as previously described (14) (plastic capillaries (ref. no. 100012), Sanguis, Nümbrecht, Germany). Much care was taken not to irritate the eye during tear fluid collection in order to prevent dilution of samples, and the procedure was immediately stopped if the eye showed signs of irritation with excessive tearing. The amount of tear fluid obtained was measured (range: 1.4–10.0 µl) and the capillary was immediately immersed in a 1.5 ml tube containing 500 µl of tissue protein extractor solution (TPER; Pierce Rockford, IL) and stored at −80°C.
Blood was drawn from the antecubital vein into pre-cooled 5 ml tubes prepared with 15% EDTA and the protease inhibitor aprotinin (Trasylol 10 mg/ml, Sigma-Aldrich, Germany), immediately placed on ice and centrifuged for 10 min at 2000 g at 4°C. Plasma aliquots were stored at −80°C.
CGRP levels in right and left eye tear fluid and plasma were determined using a commercial CGRP sandwich Enzyme-linked Immunosorbent Assay (ELISA) kit (CUSABIO®, Wuhan, China; detection range: 1.56–100 pg/ml, minimal detectable dose: 0.39 pg/ml), following manufacturer’s instructions. Intra-assay precision and inter-assay precision are declared with a coefficient of variation (CV) of <8% and <10%, respectively. Duplicate measurements were performed for each sample. The final CGRP level of each sample was calculated as the average of the two independent measurements.
Endpoints and statistics
Data is presented as mean ± standard deviation. As some variables turned out to be not normally distributed, non-parametric tests were used.
The primary endpoint was the difference in tear fluid CGRP levels in active CH patients without recent (<48 h) intake of acute medication (“unmedicated patients”) compared to healthy controls. This was tested using the Mann-Whitney U test.
Please note that the term “unmedicated” is used in the present study to designate patients that had not used acute (attack abortive) medication in the last 48 h. It does not refer to preventive medication.
Secondary endpoints were the differences between tear fluid CGRP levels in active CH patients with intake of acute medication (“medicated patients”) compared to unmedicated CH patients with and without a recent attack (between 3 and 28 h before sampling) and healthy controls (assessed by Kruskal Wallis ANOVA, using the Mann-Whitney U test as a post-hoc test). The same analyses were performed for plasma CGRP levels, also as secondary endpoints.
For comparison of age and gender between groups, Kruskal-Wallis ANOVA and chi-square tests were used. Spearman’s rho was used to test for correlations. Wilcoxon’s test was used to compare CGRP levels between eyes or between tear fluid and plasma. Statistical analysis was performed with SPSS Statistics 25 (IBM Corporation, Armonk, NY, USA). Significance was accepted at p <0.05 (two-tailed).
Results
Study population
Sixteen episodic CH patients during an active episode, 11 chronic CH patients and 60 healthy control subjects were included (see Table 1 for a description of the study population).
Description of study population. Values are mean ± standard deviation or numbers of subjects and proportions.
1Patients’ retrospective report (average of the last week).
2Patients who reported a CH attack in the last 48 h, but did not take any acute medication.
3Patients who reported a CH attack and acute headache medication intake in the last 48 h before sampling.
4A relatively large number of CH attacks were unmedicated (without intake of attack abortive medication). This was due to either previously undiagnosed CH (these patients were enrolled at the time of their first visit at our outpatient headache center) or due to unavailability of acute medication because of prescription restrictions in primary care.
5Medication included a triptan in every case.
6Patients who reported no CH attack and no intake of acute medication in the last 48 h.
Over all subjects, there was no significant difference in CGRP levels in tear fluid between eyes (right eye: 1.09 ± 1.43 ng/ml, left eye: 0.91 ± 1.10 ng/ml, Z = −0.87, p = 0.383). There was a significant correlation between CGRP levels from right and left eye (rho = 0.555, p = 0.01). Where not indicated otherwise, results of tear fluid CGRP measurement from the left and right eye were averaged for the remainder of the study.
Over all subjects, average tear fluid CGRP levels were 0.99 ± 1.10 ng/ml and average plasma CGRP levels were 6.22 ± 4.00 pg/ml. Therefore, CGRP levels in tear fluid were 159 times higher than in plasma. Tear fluid and plasma CGRP levels were significantly correlated (rho = 0.383, p = 0.01).
There were significant group differences in age (see Table 1, Kruskal Wallis ANOVA, χ2[2] = 10.2, p = 0.006) and gender (χ2[2] = 12.3, p = 0.002). Pairwise tests revealed significant age differences between controls and both CH groups (episodic: U = 264, Z = −2.755, p = 0.006; chronic: U = 197, Z = −2.108, p = 0.035) but no difference between episodic and chronic CH (U = 69, Z = −0.939, p = 0.368). Similarly, there were significant gender differences between episodic CH patients and both controls and chronic CH patients (controls: χ2[1] = 11.7, p = 0.001; chronic CH: χ2[1] = 6.0, p = 0.014) but not between controls and chronic CH patients (χ2[1] = 0.005, p = 0.943). However, age and gender did not significantly influence CGRP levels, as there were no significant correlations between age and CGRP levels in any of the three groups in tear fluid (controls: rho = 0.069; p = 0.601, episodic: rho = 0.056, p = 0.837; chronic: rho = −0.515, p = 0.105) or plasma (controls: rho = 0.230, p = 0.077; episodic: rho = 0.316, p = 0.232; chronic: rho = −0.337, p = 0.311).
CGRP tear fluid levels are elevated in unmedicated CH patients compared to healthy controls
Tear fluid CGRP levels in unmedicated CH patients were significantly elevated compared to healthy controls (CH patients: 1.78 ± 1.57 ng/ml, n = 17, controls: 0.79 ± 0.74 ng/ml, n = 60, U = 268, Z = −2.972, n = 60, p = 0.003, Figure 1). In contrast, there was no significant difference between unmedicated CH patients and controls in plasma CGRP levels (CH patients: 7.30 ± 4.00 pg/ml, controls: 6.10 ± 4.17 pg/ml, U = 394, Z = −1.425, p = 0.154). There was no significant difference in tear fluid or plasma CGRP levels between unmedicated episodic and chronic CH patients (tear fluid episodic: 1.48 ± 1.34 ng/ml, n = 10, tear fluid chronic: 2.21 ± 1.88 ng/ml, n = 7, U = 25, Z = −0.976, p = 0.364; plasma episodic: 6.03 ± 3.93 pg/ml, plasma chronic: 9.13 ± 3.59 pg/ml, U = 16, Z = −1.854, p = 0.070).
CGRP levels in tear fluid (a) and plasma (b) in unmedicated episodic (n = 10) and chronic (n = 7) CH patients and healthy controls (n = 60). All CH patients in this part of the analysis had restrained from taking acute medication (triptans or oxygen) in the last 48 h (called “unmedicated” for the purpose of the present study). (a) Significantly elevated CGRP tear fluid levels were detected in CH patients compared to healthy controls (p = 0.003), but there was no significant difference in tear fluid CGRP between episodic and chronic CH patients (p = 0.364). (b) There were no significant group differences in plasma CGRP levels (p = 0.154). Values are mean ± standard error.
The average number of cluster attacks per day was not related to tear fluid CGRP levels (rho = −0.378, p = 0.135, n = 17) or plasma CGRP levels (rho = 0.299, p = 0.244).
CGRP tear fluid levels are similar to controls after intake of acute headache medication
In a next step, we performed a more detailed subgroup analysis of CH patients. CH patients were divided into i) those that had been free of attacks and acute medication in the last 48 h (“unmedicated without attack”, n = 8), ii) those that had experienced an attack in the last 48 hours but not taken acute medication within the last 48 h (“unmedicated with attack”, n = 9) and iii) those that had experienced an attack and had taken acute medication in the last 48 hours (“medicated with attack”, n = 10, the acute medication included a triptan in all cases).
When CGRP tear fluid levels in these three patient groups and the healthy control group were compared, there was a significant main effect of group (χ2[3] = 11.0, p = 0.012). Pairwise comparisons were used to further test for group differences (Figure 2). The medicated group (0.84 ± 1.40 ng/ml) showed significantly lower tear fluid CGRP levels than both unmedicated patient groups (unmedicated without attack: 1.95 ± 1.65 ng/ml, U = 15, Z = −2.21, p = 0.027; unmedicated with attack: 1.63 ± 1.59 ng/ml, U = 17, Z = −2.286, p = 0.022), but was comparable to the control group (U = 236, Z = −1.074, p = 0.283). The two unmedicated groups were not significantly different from each other (U = 33, Z = −0.289, p = 0.815), but both unmedicated groups showed significantly higher tear fluid CGRP levels than the control group (unmedicated without attack: U = 120, Z = −2.284, p = 0.022; unmedicated with attack: U = 148, Z = −2.174, p = 0.030). In contrast, there were no group differences in plasma CGRP levels (χ2[3] = 3.81, p = 0.283).
CGRP levels in tear fluid (a) and plasma (b) in subgroups of CH patients and healthy controls. Subgroups were defined as follows: Medicated with attack (CH patients who had experienced a CH attack and taken acute medication in the last 48 h), unmedicated with attack (CH patients who had experienced an attack in the last 48 h but not taken acute medication) and unmedicated without attack (no attack or acute medication in the last 48 h). (a) Tear fluid CGRP levels in medicated patients were undistinguishable from those of healthy controls (p = 0.283). There was also no difference between unmedicated patients with and without attack in the last 48 h (p = 0.815). However, tear fluid CGRP levels were significantly higher in both groups of unmedicated patients compared to those of both the medicated patients and the control group (all p < 0.05, see Results section for details). (b) No significant group differences were detected in plasma CGRP levels (p = 0.283). Values are mean ± standard error.
No difference in CGRP tear fluid levels ipsi- and contralateral to the usual attack side
In unmedicated CH patients, tear fluid CGRP levels measured ipsi- and contralateral to the usual attack side did not differ (ipsilateral: 1.81 ± 1.50 ng/ml, contralateral: 1.78 ± 2.21 ng/ml, Z = −0.54, p = 0.586, n = 17).
Discussion
Main findings of this study are that tear fluid CGRP concentrations in active CH patients are significantly elevated compared to controls, as long as CH patients have not used acute medication. This did not depend on the recent occurrence of an attack, or on the diagnosis of episodic versus chronic CH. No significant group differences were found in plasma CGRP levels.
Unmedicated cluster headache patients vs. healthy controls
CH patients who had not taken acute medication in the last 48 hours (unmedicated patients) had significantly elevated CGRP levels in tear fluid compared to healthy controls. This further corroborates the role of CGRP in CH pathophysiology and demonstrates that elevated CGRP levels in CH patients can be detected in tear fluid, most likely corresponding to CGRP released from the first division of the trigeminal nerve. Tear fluid CGRP levels in unmedicated CH patients in the present study were nominally higher than those obtained for interictal migraine patients in our previous study (1.78 ± 1.57 ng/ml vs. 1.10 ± 1.27 ng/ml) (17), but data from different studies cannot be compared directly. There was no difference in CGRP tear fluid levels in unmedicated CH patients with or without a CH attack in the last 48 h, and both were significantly elevated respective to controls. This has two implications. First, it shows that active cluster headache patients have elevated CGRP levels even if they had no attack in the previous 48 hours. This suggests that baseline CGRP release from trigeminal fibers is elevated compared to controls in these patients. One hypothesis would be that this reflects susceptibility to development of an attack. Similarly, plasma CGRP levels in active episodic CH patients outside attacks have been shown to be elevated compared to CH patients in remission (6,7). Elevated interictal CGRP levels have also been shown in migraine, both in blood (20,21) and in tear fluid (17). Second, tear fluid CGRP levels in CH patients with an attack in the last 48 hours were not further elevated beyond those of CH patients without attack. It must be considered that none of the patients were measured during an acute attack, and the last attack had occurred between 3 and 48 hours before sampling. Within a full-blown, unmedicated attack, CGRP levels would be expected to be further elevated as previously shown in jugular vein blood (6,7). Assessing the time course of tear fluid CGRP levels over the course of CH attacks would be an interesting topic for a follow-up study.
Apart from the intake of acute medication (see below), there are several factors that could be hypothesised to influence tear fluid CGRP levels in CH patients outside acute attacks. Higher levels in patients with more frequent attacks might be expected. In the present study, no such correlation was found; however, the attack frequency was based on retrospective report only. Similarly, shorter time since the last attack might be associated with higher tear fluid CGRP levels. Future studies using more detailed attack calendars will be able to address these points. In principle, the use of cluster headache preventive medication might also influence CGRP levels between attacks, as some preventive medications are thought to act by reduction of CGRP release (8). Larger studies will be needed to answer this question, as numbers of patients with and without preventive medication in the present study were too small for this analysis.
Episodic vs. chronic cluster headache
Tear fluid CGRP levels were not significantly different between active episodic and chronic unmedicated CH patients, although it has to be mentioned that they were nominally higher in chronic CH patients. There is only one previous study comparing CGRP levels (in this case obtained from the antecubital vein) between chronic and active episodic CH patients, also finding no significant difference (22). These results would support the view that active episodic CH and chronic CH have a similar activation of the trigemino-vascular system, resulting in similar CGRP levels. It would be a very interesting follow-up project to look at tear fluid CGRP values in episodic CH patients in remission. If CGRP is a marker of CH activity, one would expect CGRP levels to be lower in episodic CH patients in remission compared to active CH patients, as has been shown in jugular vein blood (6,8). In contrast, the study cited above (22) found more strongly elevated CGRP levels in episodic CH patients in remission compared to those of chronic CH patients in antecubital blood, and the authors proposed that this might be due to CGRP depletion by frequent attacks in chronic CH. One drawback of the study is that intake of acute headache medication was not taken into account (22), which from our and previous results seems to have a large impact on CGRP levels (6,7).
Medicated vs. unmedicated cluster headache
In patients who had taken acute medication in the previous 48 hours, tear fluid CGRP was significantly reduced compared to unmedicated patients, and undistinguishable from controls. This is consistent with previous results showing that use of sumatriptan significantly reduces CGRP levels in CH patients (6,7) and in line with the presumed mechanism of triptans in CH (reduction of CGRP release from trigeminal fibers) (23,24). It is also similar to our results in migraine, where intake of acute medication in the previous 48 hours resulted in tear fluid CGRP levels similar to controls, while unmedicated patients had elevated levels (17). In the present study, a time window of 48 h was used to exclude any residual action of acute medication. However, it would certainly be interesting to investigate the time course of action of acute medication on tear fluid CGRP levels in CH patients. This could be performed within a study investigating the time course of tear fluid CGRP over the acute CH attack as outlined above.
Tear fluid CGRP levels ipsi- and contralateral to the side of cluster headache
Cluster headache is unilateral, and most patients have their attacks always on the same side (3). Therefore, one would expect that trigeminal activation and consequent CGRP release from trigeminal fibers might also be unilateral, at least during the acute attack. To the best of our knowledge, this question has not been investigated before. Detection of CGRP in tear fluid offers the opportunity to separately assess tear fluid from the left and right eye, supposedly reflecting CGRP release from the left and right trigeminal nerve. In the present study, there were no differences in CGRP levels measured in tear fluid taken from the eye ipsi- versus contralateral to the usual headache side. As discussed above, we did not measure patients during attacks. Maybe the elevated baseline CGRP levels in CH patients are due to bilateral CGRP release, while additional release during an attack could be ipsilateral to the side of the attack. This is another question that can be answered by measuring tear fluid CGRP levels over the course of an attack.
Tear fluid vs. plasma CGRP levels in cluster headache
Tear fluid CGRP levels were 159 times higher than those found in plasma in the present study, similar to our previous results (17). We propose that this is due to direct innervation of the eye by the trigeminal nerve and consequently less dilution. Different degradation in tear fluid compared to plasma might also play a role, but to the best of our knowledge this has not been studied up to now.
While significant group differences were found for tear fluid CGRP, no significant differences in CGRP plasma levels obtained from peripheral blood were detected in CH patients compared to controls or between medicated and unmedicated CH patients. This is similar to our results in migraine patients (17).
Two earlier studies investigated CGRP levels in CH patients in antecubital vein blood, finding elevated CGRP levels in episodic CH patients in remission compared to chronic CH patients (22) and significantly higher CGRP levels in migraine patients compared to interictal episodic CH patients (25). As these comparisons were not made in the present study, results cannot be directly compared. It must be mentioned that the feasibility of CGRP measurement in peripheral blood (vs. blood taken from the jugular vein) has been discussed. At least in migraine, there are several studies that were not successful in detecting elevated CGRP levels in blood taken from the antecubital vein (13,26).
Limitations
Several limitations should be considered. The short half-life of CGRP (approximately 7 min in plasma) leads to fast degradation (27). Much care was taken to minimise this problem by rapid processing and freezing of the samples including a protease inhibitor in the vials. Some degradation may have happened nonetheless. Another limitation might be individual variability in tear fluid production. For further studies, we plan to assess total tear fluid protein and normalise tear fluid CGRP to this value. Another limitation is the small sample sizes, especially in some of the subgroup analyses. However, our sample size is comparable to previous studies of CGRP in cluster headache (6–8,22) and results were significant and consistent with the existing literature. Nonetheless, results should be confirmed and expanded in further studies, preferentially including larger samples. Especially, this would allow analysis of the effect of preventive cluster headache medication on tear fluid CGRP levels.
Active cluster headache patients without recent intake of attack abortive medication showed increased baseline tear fluid CGRP levels (measured outside acute attacks) compared to healthy controls. In active cluster headache patients who recently used acute medication, tear fluid CGRP levels were undistinguishable from those found in healthy controls. Tear fluid CGRP levels obtained from the eye ipsi- vs. contralateral to the usual cluster headache attack side were not significantly different from each other. Together with our previous results in migraine patients, the present data suggest that CGRP detection in tear fluid is a valid, sensitive and non-invasive method to assess CGRP released from the trigeminal nerve in headache patients.
Clinical implications
Active cluster headache patients (outside acute attack) show increased baseline tear fluid CGRP levels. Elevated tear fluid CGRP levels might be rather a sign of active cluster headache than of its episodic or chronic course. After use of acute medication, tear fluid CGRP levels in cluster headache patients were reduced to those found in controls.
Footnotes
Acknowledgements
We thank all patients and healthy controls for participating in this study and Sigrid Langer for technical support.
Declaration of conflicting interests
The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
