PACAP‐38 in Cluster Headache: A Prospective, Case–Control Study of a Potential Treatment Target

ABSTRACT

Background

Estimates of pituitary adenylate cyclase‐activating peptide‐38's (PACAP‐38) activity in cluster headache are sparse. We investigated whether plasma levels of PACAP‐38 differ between disease states (i.e., bout, remission, chronic) and compared to headache‐free controls. Secondly, we assessed a possible correlation between plasma levels of PACAP‐38 and Calcitonin gene‐related peptide (CGRP).

Methods

In an observational case–control setup, prospectively collected interictal plasma samples from participants in the Danish Cluster Headache Biobank were analyzed for plasma levels of PACAP‐38 and CGRP. All participants had blood samples drawn; once if chronic cluster headache or controls, and twice if episodic cluster headache (in bout and in remission phase). Plasma levels were measured with validated immunoassays.

Results

Plasma was derived from 205 patients with cluster headache according to ICHD‐3 criteria and 101 sex‐ and age‐matched headache‐free controls. PACAP‐38 plasma levels were significantly higher in all three disease states of cluster headache: compared to controls, they collectively had a 34.3% (95% CI: 20%–49%, p < 0.0001) higher mean PACAP‐38 level. Chronic cluster headache showed the greatest difference by 49.8% (95% CI: 26.7–77.2, p < 0.0001) higher PACAP‐38 levels, while episodic cluster headache in bout and remission showed respectively 39.5% (95% CI: 18.8–63.8, p < 0.0001) and 34.1% (95% CI: 14.2–57.5, p = 0.0005) higher plasma levels. No correlation between plasma levels of PACAP‐38 and CGRP was demonstrated (Spearmans r = 0.08, p = 0.10).

Conclusions

This large‐scale study demonstrated increased PACAP‐38 levels in all disease states of cluster headache compared to headache‐free controls, strengthening the hope of a possible effect of PACAP‐38 targeting treatments in future trials. The lacking correlation between PACAP‐38 and CGRP levels should be interpreted with caution and needs to be investigated in future studies.

PACAP‐38 plasma levels were found elevated in all three disease states of cluster headache compared to headache‐free controls. Findings suggest increased levels of PACAP‐38 as a fundamental pathophysiological dysfunction in cluster headache and strengthen the hope of a possible effect of PACAP‐38 targeted treatment in future trials.

Abbreviations

CGRP

calcitonin gene‐related peptide

CH

cluster headache

ICHD

International classification of headache disorders

PACAP‐38

pituitary adenylate cyclase‐activating polypeptide 38

1. Introduction

Commonly acknowledged as one of the most severe pain conditions, cluster headache (CH) remains a high‐priority research target for improved diagnostics and especially more effective therapy [1]. CH is a primary headache condition existing in two phenotypes; most patients suffer from the episodic variant enduring attack periods (bouts) lasting weeks to months and separated by attack‐free remission periods. The remaining 10%–15% of patients are not relieved with remission periods surpassing three consecutive months, thus belonging to the chronic variant [1, 2, 3]. While the categorization of three different disease states (chronic, episodic in remission, episodic in bout) is clinically logical, the pathophysiological clockwork behind the distinction remains only partly understood.

In the pursuit of pathophysiological mapping, several neuropeptides have within recent years been thoroughly investigated for their role in primary headaches and CH [4, 5].

Calcitonin Gene‐related Peptide (CGRP) has so far been the primary biomarker researched in CH. Unfortunately, CGRP has shown substantial variability in measured levels in patients with CH [6, 7, 8, 9] and so far, limited effect as a treatment target (Galcanezumab [10], Fremanezumab [11, 12], Eptinezumab [13, 14]). Consequently, attention has shifted to another biomarker of rising interest: Pituitary adenylate cyclase‐activating peptide‐38 (PACAP‐38) [15]. PACAP‐38 is a diverse neuropeptide, with a central role in nociception in primary headaches by activity in key structures, including the hypothalamus and the trigeminal ganglion [16, 17, 18, 19, 20]. In migraine, increased levels of PACAP‐38 have been demonstrated, and an early clinical trial of targeted treatment has exhibited effects in prophylactic attack‐reduction [21, 22, 23]. Several arguments support PACAP‐38's implication in CH pathology too; PACAP‐38 and its receptors are expressed in both cranial parasympathetic neurons and the hypothalamus, both regarded as central actors in the distinct features of CH attacks [3, 24]. Secondly, prior provocation studies have reproduced the onset of pain during systemic infusion and subsiding pain levels at the abolishment of PACAP‐38 synthesis [25, 26]. Collectively, these observations may hold promise for PACAP‐38 as a new therapeutic target for CH [15]. Still, the foundation for future treatment trials is limited, as baseline plasma levels of PACAP‐38 in CH have only been systematically studied twice with divergent findings and based on small sample sizes [4, 6, 25].

This study aimed to investigate plasma‐levels of PACAP‐38 in a large sample of patients with CH in the three different disease stages compared to headache‐free controls (HC). We additionally assessed the plasma‐levels of CGRP in the four subgroups intending to examine a possible correlation between plasma‐levels of PACAP‐38 and CGRP. We hypothesized an identification of higher levels of PACAP‐38 in patients with CH as to HCs and a correlation between the PACAP‐38‐ and CGRP‐levels.

2. Methods

This prospective, observational case‐controlled study is a part of the Danish Cluster Headache Biobank, on which 3 prior studies are based [9, 27, 28]. The Biobank consists of prospectively collected blood samples between October 2018 and December 2021 at the Danish Headache Center (DHC), Rigshospitalet‐Glostrup, Denmark, from patients with episodic CH (eCH) in bout and remission, patients with chronic CH (cCH) and HCs.

2.1. Participants

Participants with CH were diagnosed according to current ICHD criteria [1, 29]. Inclusion criteria were an age between 18 and 80 years. Exclusion criteria were chronic headaches other than CH, known current drug abuse, severe somatic or psychiatric co‐morbidity, pregnancy, and breast‐feeding. Use of stable CH prophylactic medication was permitted, while use of steroids (oral and greater occipital nerve blocks) within 30 days from sampling led to exclusion.

Participants with CH were grouped into: cCH, eCH in bout, or eCH in remission. eCH in bout was defined as being episodic and having more than one attack in the preceding week. eCH in remission was defined as being attack‐free for at least three consecutive weeks. Participants recruited in their first bout were followed until remission or a diagnosis of CCH could be made after 1 year.

HCs were recruited through posting on social media and notices posted in the hospital's neighborhood. HCs were matched by sex and age on a 1:1 ratio to the eCH group. HCs had no medical history of primary or secondary headache except infrequent tension‐type headache, delayed alcohol‐induced headache, or headache related to infection such as influenza. Any type of headache was disallowed within 7 days prior to sampling.

2.2. Study‐Visits

Participants were recruited consecutively at DHC and sampled once if being cCH or HCs, and twice if being eCH (in bout and in remission state, respectively). All participants were instructed to fast for at least 8 h and not to drink any liquids 2 h ahead of sampling. On sampling days, a structured interview regarding medicine use, sleep, and physical activity in the preceding 24 h, as well as fasting state, recent viral or bacterial infection, and smoking status (current, past and never) was performed.

Additionally, a baseline semi‐structured interview regarding basic demographic information and clinical features was recorded for all participants on the first day of sampling. For participants with CH, the diagnosis was validated by a physician (A.S., A.S.P., R.J.) or a specially trained medical student (A.F.P.).

2.3. Standard Protocol Approvals, Registrations, and Patient Consents

All participants provided verbal and written informed consent in compliance with the Helsinki Declaration. The study was approved by the Danish Data Protection Agency and the Capital Region Regional Health Research Ethics Committee (H‐16048941).

2.4. CGRP and PACAP‐38‐Measurement

All participants were sampled outside of attacks. Blood was collected from the antecubital vein into standard EDTA‐anticoagulated tubes and inverted multiple times. Upon resting at room temperature for 30 min, samples were centrifuged at 4°C for 10 min at 1409 g. One mL of plasma was transferred to cryotubes (Greiner Bio‐one), where it was kept first at −25°C and then transferred for permanent storage at −80°C pending analysis.

Plasma‐levels of PACAP‐38 and CGRP were measured by using custom developed sandwich‐Electrochemiluminescence Immunoassays. Assays were internally validated at the operating laboratory, Celerion, Switzerland.

An assay was developed to quantitatively determine PACAP‐38 levels in human EDTA plasma using the MSD S‐PLEX technology. The assay was based on Lundbeck proprietary antibodies with synthetic PACAP‐38 (Bachem, 4031157) as the calibrator. The analytical range of the assay spanned from 50 to 80,000 fg/mL, with a lower limit of quantification (LLOQ) set at 50 fg/mL. To interpolate unknown values, a four‐parameter logistic (4PL) regression model with 1/y ² weighting was utilized. Samples were analyzed in duplicates within the same run and adjusted for the respective dilution factor.

The quantification of CGRP in human EDTA plasma samples was assessed using a CGRP MSD S‐PLEX assay. Like the PACAP‐38 assay, this assay was also based on Lundbeck proprietary antibodies, with synthetic human α‐CGRP (Bachem, 4013281) as the calibrator. The analytical range spanned from 200 to 80,000 fg/mL, with the LLOQ set at 200 fg/mL. A four‐parameter logistic (4PL) regression model with 1/y ² weighting was used for interpolation of unknown values. During the assay, samples were analyzed in duplicates within the same run after an 8‐fold dilution in an assay buffer containing a protease inhibitor solution.

All plasma level results are reported in pg/mL.

2.5. Outcomes

Our primary outcomes were the difference in PACAP‐38 plasma levels between groups (cCH, eCH bout, eCH remission, HC), adjusted for sex, age, co‐morbid migraine, and attack frequency within the last 24 h. Further, we assessed the difference in CGRP plasma levels between groups (cCH, eCH bout, eCH remission, HC), adjusted for the same parameters, and whether there was a correlation between plasma levels of PACAP‐38 and CGRP.

Participants were described by demographic characteristics (biological sex, age, debut‐year of CH, comorbidity, usual attack duration in minutes and usual daily attack‐burden) and clinical features in the preceding 24 h before the day of blood sampling: current headache, presence of CH attacks within the last 24 h or last week, use of abortive medication for the last attack (triptans or oxygen), use of preventive medicine (verapamil or other), time of blood sampling, season of blood sampling (Winter, Spring, Summer, Autumn), sleep duration for the night before sampling (< 6, 6–8, > 8 h), exercise within the last 24 h, infection within the last week (viral or bacterial) and head trauma (within the last month).

2.6. Data Analysis

Sample size for the biobank was determined in an already published study [9]. A flowchart of the inclusion process is presented in Figure 1 and a detailed description is accessible in [9].

FIGURE 1.

Flowchart of inclusion.

Upon measurement, one sample had to be excluded due to hemolysis and one due to insufficient volume. An additional four samples were excluded due to the use of steroids for comorbid diseases (eCH bout n = 1, eCH remission n = 2) or oral transitional treatment of CH (eCH remission n = 1) within 30 days from sampling. An additional 14 samples were discharged due to technical reasons (cCH = 2, eCH bout = 6, eCH remission = 6).

After analysis, one sample with CGRP plasma level outside of the range of validation was excluded. To remove possibly influential and erroneous outliers, we excluded any samples with plasma levels exceeding seven times the IQR from the median value (cCH n = 4, eCH bout n = 1, eCH remission n = 2, HC n = 3).

Samples with plasma levels below quantifiable range (< 0.05 pg/mL for PACAP‐38 (n = 18), < 0.2 pg/mL for CGRP (n = 1)) showed an unequal distribution between the four study groups. Possibly reflecting a systemic tendency, they were included for analysis at a value of 0.049 pg/mL for PACAP‐38 and 0.199 pg/mL for CGRP, respectively.

In order to base results on the highest possible number of samples, we allowed unpaired samples to be included in the unpaired analyses.

2.7. Statistical Methods

All data management and statistical analyses was performed using SAS Statistical software, version 9.4 (SAS Institute Inc., NC, USA) and R software, version 4.3.0 (R Foundation, Vienna, Austria).

Numeric descriptive data are presented as a mean with standard deviations (SD) or as median with inter‐quartile range (IQR) as appropriate. Categorical variables are presented as counts with percentages. The distribution of data was evaluated for normality within each patient group using histograms and QQ plots.

For groupwise comparisons within PACAP‐38 and CGRP, we used paired and unpaired t‐tests for normally distributed data. For non‐normally distributed data, the Wilcoxon‐signed rank test was used for paired data and the Wilcoxon‐Mann–Whitney test for unpaired data. To determine a possible correlation between CGRP in bout and in remission, Pearson's Rank test was applied. Adjusting for multiple comparisons, outcome p‐values were adjusted with the false discovery rate method.

To determine the effect of baseline covariates on PACAP‐38 and CGRP‐levels, we used two models: A generalized linear mixed model with PACAP‐38 as the outcome, assuming a gamma distribution with a logarithmic link function, and a linear mixed model with a logarithmic transformation of CGRP as the outcome. All model assumptions were reasonably met.

We assessed the association between PACAP‐38 and CGRP levels at both the population and grouplevels using scatter plots with LOESS smoothing and Spearman correlation.

Patients with missing data were omitted from the analysis of the examined variable and reported in tables. A level of significance of 5% (p < 0.05, 2‐tailed) was accepted for all tests.

All findings were reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [30].

3. Results

In total, 310 patients with CH were screened, of which 211 were found to be eligible participants. Accordingly, 101 HCs were included.

Upon sorting, 392 samples were included for analysis, deriving from a total of 99 patients with cCH, 106 patients with eCH (103 in remission, 106 in bout, thus 103 full pairs in total) and 101 HCs.

Demographics and clinical characteristics of included participants are presented in Table 1. Matching between eCH and HCs was successful. A summarization of clinical variables on the day of blood sampling is presented in Table 2.

TABLE 1.

Baseline demographics.

	Chronic cluster headache (n = 99)	Episodic cluster headache (n = 106)	Headache‐free controls (n = 101)
Sex
Males, n (%)	64 (63.4)	85.0 (80.2)	81 (80.2)
Females, n (%)	37 (36.6)	21 (19.8)	20 (19.8)
Age (years), mean ± SD	46.9 (14.0)	41.8 (11.8)	41.4 (13.7)
Episodic migraine comorbidity, n (%)	10 (9.9)	10 (9.4)	—
Cluster headache duration (years), median (interquartile range)	12.0 (14.0)	11.5 (15.0)	—
Average attack duration (minutes), mean ± SD	97.8 (61.1)	88.0 (51.5)	—
Attacks per day, mean ± SD	2.3 (1.8)	2.2 (1.3)	—

TABLE 2.

Clinical variables on day of blood sampling.

	Chronic cluster headache (n = 99)	Episodic cluster headache in bout (n = 106)	Episodic cluster headache in remission (n = 103)	Headache‐free controls (n = 101)
Current headache at time of blood sampling, n (%)	36 (36)	25 (24)	8 (8)	0
Cluster headache attack frequency, median (IQR)
Last 24 h	1 (3)	1 (2)	—	—
Last week	10 (18)	8 (12)	—	—
Time since last cluster headache attack, n
< 2 h	12	10	—	—
2 to < 6 h	20	24	—	—
6 to < 12 h	20	14	—	—
12 to < 24 h	15	18	—	—
> 24 h	34	38	—	—
Missing	—	2	—	—
Last attack treated by, n
Oxygen	43	53	—	—
Triptan	21	32	—	—
Preventative medication, n
Verapamil	34	31	13	—
Other	11	3	1	—
Time of blood sampling, n
06:00 a.m. to 08.59 a.m.	24	26	55	65
09:00 a.m. to 10:59 a.m.	50	59	45	33
11:00 a.m. to 12:59 p.m.	21	15	1	1
13.00 p.m. to 14:59 p.m.	6	6	2	1
Season of blood sampling, n
Winter	17	26	9	5
Spring	20	23	28	13
Summer	26	35	26	18
Autumn	38	22	40	64
Sleep last night, n
< 6 h	52	43	36	18
6–8 h	39	52	46	71
> 8 h	10	10	20	11
Missing	0	1	1	1
Infection last week, n
Viral	2	4	12	5
Bacterial	1	1	5	0
Head trauma last month, n	4	0	0	0
Exercise last 24 h, n	19	19	19	31

Abbreviation: IQR, interquartile range.

3.1. Higher Plasma Levels of PACAP‐38 in All CH Disease States Compared to HCs

Increased levels of PACAP‐38 were identified in all CH states (cCH, eCH bout and eCH remission) as compared to HCs (p = < 0.0001–0.003) (Table 3, Figure 2).

TABLE 3.

Plasma levels of pituitary adenylate cyclase‐activating polypeptide‐38 (PACAP‐38) and calcitonin gene‐related peptide (CGRP) and comparisons within the four groups.

	Chronic cluster headache	Episodic cluster headache in bout	Episodic cluster headache in remission	Headache‐free controls	p
PACAP‐38
Participants, n	99	105	102	98	—
Plasma level (pg/mL), mean ± SD	0.18 (0.13)	0.16 (0.12)	0.15 (0.1)	0.11 (0.06)	—
cCH vs. HC a					< 0.0001
eCH bout vs. HC a					0.0004
eCH remission vs. HC a					0.003
eCH bout vs. eCH remission b					0.27
eCH bout vs. cCH a					0.28
eCH remission vs. cCH a					0.18
CGRP
Participants, n	99	106	102	101	—
Plasma level (pg/mL), mean ± SD	1.03 (0.32)	0.93 (0.36)	1.02 (0.42)	0.95 (0.38)	—
cCH vs. HC c					0.20
eCH bout vs. HC c					0.77
eCH remission vs. HC c					0.28
eCH bout vs. eCH remission d					0.09
eCH bout vs. cCH c					0.11
eCH remission vs. cCH c					0.84

Note: All cluster headache groups had significantly higher PACAP‐38 plasma‐levels as to HCs. No significant difference was found between groups in CGRP‐plasma‐levels.

Abbreviations: cCH, Chronic cluster headache; CGRP, Calcitonin gene‐related peptide; eCH, Episodic cluster headache; HC, Headache‐free control; PACAP‐38, Pituitary adenylate cyclase‐activating polypeptide‐38.

^a Wilcoxon–Mann Whitney test.

^b Wilcoxon‐Rank test.

^c Students t‐test.

^d Paired t‐test.

FIGURE 2.

Comparison of pituitary adenylate cyclase‐activating polypeptide‐38 (PACAP‐38) plasma levels within the four study groups. Increased PACAP‐38 plasma levels were found in all three disease states (episodic cluster headache in bout, episodic cluster headache in remission, chronic cluster headache) as to headache‐free controls. Significant differences between groups are indicated with asterisks. Lower limit of quantification = 0.05 pg/mL, indicated with dashed line. **Episodic cluster headache in remission vs. headache‐free controls: p = 0.003. ***Episodic cluster headache in bout vs. headache‐free controls: p = 0.0004. ****Chronic cluster headache vs. headache‐free controls: p = < 0.0001.

Parameter estimates for the differences between the disease groups and HCs were provided in a generalized mixed model. It showed that collectively all CH patients had a mean PACAP‐38 level 34.3% (95% CI: 20.1%–48.6%, p < 0.0001) higher than HCs, adjusted for sex, age, comorbid episodic migraine, and attack frequency within the last 24 h. By subgroup, cCH showed the greatest difference by 49.8% (95% CI: 26.7–77.2, p < 0.0001) higher PACAP‐38 levels as to HCs, while eCH bout and eCH remission showed respectively 39.5% (95% CI: 18.8–63.8, p < 0.0001) and 34.1% (95% CI: 14.2–57.5, p = 0.0005) higher plasma levels.

At comparison of PACAP‐38 levels within the three disease states, levels were numerically, but not statistically different (p = 0.2–0.3) with mean ± SD values being respectively 0.18 ± 0.13 pg/mL for cCH, 0.16 ± 0.12 pg/mL for eCH bout, and 0.15 ± 0.1 pg/mL for eCH remission (Table 3, Figure 2).

The generalized mixed model showed no effect of sex (p = 0.68), age (p = 0.10), comorbid episodic migraine (p = 0.07) or attack frequency within the last 24 h (p = 0.77) on the estimates.

3.2. No Difference in Plasma‐Levels of CGRP Between CH and HCs

We did not identify any difference in plasma levels of CGRP between HCs and either cCH, eCH bout, or eCH remission (p = 0.2–0.8) (Table 3, Figure S1).

A comparison of CGRP levels within the three disease states did not show any difference between cCH and eCH bout (p = 0.11) and eCH remission (p = 0.84), respectively. A paired comparison of plasma levels in eCH bout to eCH remission did not detect an apparent difference either (p = 0.09). Still, numerically, plasma levels showed rising CGRP levels in remission as to bout, confirmed by a Pearson's correlation (r = 0.52, p < 0.0001).

In a linear mixed model, CGRP‐levels were unaffected by sex (p = 0.07), comorbidity with migraine (p = 0.12) and attack‐frequency within the last 24 h (p = 0.75). Only age displayed an effect, with each one‐year increase associated with a 0.79% increase in CGRP levels (95% CI: 0.50%–1.08%, p < 0.001).

3.3. No Correlation Between PACAP‐38 and CGRP‐Levels

Plasma levels of PACAP‐38 and CGRP did not show a correlation, neither on the level for the entire study population (Spearmans correlation; r = 0.08, p = 0.10) nor within each group (Spearmans correlation; eCH remission; r = 0.05, p = 0.59, eCH bout; r = 0.15, p = 0.12, cCH; r = −0.002, p = 0.99, HC; r = 0.10, p = 0.32) (Figure 3).

FIGURE 3.

Correlation between pituitary adenylate cyclase‐activating polypeptide‐38 (PACAP‐38) and calcitonin gene‐related peptide (CGRP) plasma levels. Correlation visualized by scatter‐plot with smooth LOESS‐curves. No correlation between plasma levels of Pituitary adenylate cyclase‐activating polypeptide‐38 (PACAP‐38) and Calcitonin gene‐related peptide (CGRP) was identified, neither for the whole study population nor within the four subgroups. Lower limit of quantification for PACAP‐38 = 0.5 pg/mL indicated with dashed line.

4. Discussion

We present a large‐scale assessment of systemic PACAP‐38 plasma levels in CH, showing markedly increased levels in all three disease states separately compared to HCs. PACAP‐38 levels were increased by nearly 50% in cCH compared to HCs, followed by 36% in eCH bout and 34% in eCH remission. No such difference in CGRP plasma levels was identified. Lastly, we did not identify an apparent correlation of CGRP and PACAP‐38 levels.

4.1. PACAP‐38

Confirming our initial hypothesis, PACAP‐38 showed increased levels in all CH disease stages in comparison to HCs. This objective has only been explored in two prior studies [4, 6, 25]. The first study reported lower PACAP‐38 levels in eCH in remission as to HCs and higher levels during attacks in participants with eCH compared to in remission [25]. The other study reported higher levels in eCH bout as to cCH but no difference compared to levels in HCs [6]. Thus, both studies present findings unaligned with ours, possibly explained by the much lower sample sizes (n = 9 and n = 31, respectively). The use of only historic controls in the last‐mentioned study constitutes an additional argument for our prospective, large‐scale study to provide a more certain estimate of PACAP‐38 levels in CH. Our findings are in line with the reported increased levels of PACAP‐38 in migraineurs as to healthy controls [22, 23], which support PACAP‐38 being a central actor in the pathology of both migraine and CH.

Part of the established pathophysiology of CH is the release of vasodilatory neuropeptides (including PACAP‐38) during a CH attack, consequent to activation of the trigeminovascular pathway [16, 31]. Concordantly, one would expect increased PACAP‐38 levels in patients with CH as compared to HCs, as demonstrated in our results. Moreover, we would hypothesize increasing levels with increased attack activity. While attack frequency within the last 24 h from sampling did not affect PACAP‐38 levels, we found the highest PACAP‐38 levels in our cCH population, who often, but not exclusively, suffer the most continuous attack activity of the CH disease groups. Collectively, this may indicate that regular rather than recent attack activity induces increased PACAP‐38 levels. Furthermore, we would presume diminishing PACAP‐38 levels outside periods of attack activity, for example, in the remission phase. While not showing a statistically significant difference, we found plasma levels to be numerically lower in eCH remission compared to both eCH in bout and cCH. We suggest this pattern could represent a latently increased PACAP‐38 level in all patients with CH, including during remission, permanently lowering the threshold for activation of the trigeminovascular pathway and trigeminal autonomic reflex compared to HCs. Adding to prior observations of PACAP‐38 infusion inducing attacks in CH and targeted treatment displaying promising effects in migraine [21], these findings build hope of PACAP‐38 as a possible therapeutic target for CH [15]. We recognize that some aspects challenge the direct transferability of PACAP‐38's implication in migraine to CH: First, the previous studies in migraine have encompassed ictal measurements, while our sampling in cCH and eCH was interictal. It is possible that PACAP‐38 levels may be modulated by attack activity, as implicated by the previous demonstration of increased ictal plasma levels by Tuka et al. [25] Second, the difference in PACAP‐38 plasma levels in migraneurs compared to healthy controls was greater compared to our findings [32]. Third, provocation studies of PACAP‐38 have demonstrated a higher rate of attack induction in migraine compared to CH. In spite of these discrepancies, this study substantiates PACAP‐38 dysregulation as a shared pathophysiological feature of migraine and CH, and thus an evident, pursuable treatment target. Correspondingly, other shared pathophysiological players in migraine and CH, such as vasoactive intestinal peptide, intracellular mechanisms, and ion channels, constitute additional interesting prospects for potential targeting [3, 33].

4.2. CGRP

By now, studies have repeatedly shown great variance in CGRP levels, which consequently must be considered an unstable biomarker [4, 9]. Possible explanations for the decidedly inconsistent results have been debated thoroughly [4], highlighting the source of the sample, sampling location, and cycles of sample thawing as contributing factors. Further, the use of different assays for CGRP measurement and their specificity to either the α‐ or β‐isoform have resulted in highly heterogeneous, bordering on incomparable estimates. Lastly, the suggested sensibility of CGRP levels to prolonged storage surpassing 6 months adds to the incomparability of separate studies [34].

Our study identified similar levels of CGRP in CH as to HCs, irrespective of disease state. Four published studies have previously compared plasma levels of CGRP in patients with CH as to HCs, all with divergent results [6, 8, 9, 31]. Of greatest interest is the comparison to the results of the most recently published study by Petersen et al.; a study of CGRP levels in the same study cohort as ours, emanating from the Danish Cluster Headache Biobank [9]. Presenting reduced levels of CGRP in all three disease states as to HCs, these results are unlike the present study's, despite the collected plasma material being from the exactly same cohort. Possible causes for the discrepancy include the longer sample storage for the current study and the use of different assays with highly variable sensitivity, constituting a topic of intense debate [34]. Still, we find the diverging results within the same cohort an evident argument for CGRP being profoundly unstable, an unsuitable biomarker for CH and possibly partly explaining the limited treatment effect in prior trials of CGRP mAbs [10, 11, 12].

4.3. PACAP‐38–CGRP Interaction

None of our analyses indicated an apparent relationship between plasma levels of PACAP‐38 and CGRP, neither at the population level nor within the different diagnostic groups, contradicting our initial hypothesis. The lacking correlation may be attributed to the CGRP measurements, based on an assay perhaps subjected to uncertainty. Thus, the lack of correlation should be interpreted with caution. No prior studies have performed a correlation‐assessment; therefore, a reproduction in a similar study set‐up is desirable to confirm whether there is really no interaction between these two important neuropeptides.

Collectively, our findings give an impression of two peptides without a synchronous pattern. From a pathophysiological angle, the PACAP‐38 and CGRP pathways are indeed two separate entities with separate receptors and molecular downwardcascades, in spite of their identical end effect: vasodilation [35]. Thus, it is possible that the two systems are activated independently or unequally in patients with CH, explaining a more pronounced increase of PACAP‐38 as to CGRP levels. Another plausible explanation is the theory of CGRP being released as a function of pain, rather than being the cause. This would explain why only samples taken ictal or immediately after an attack have displayed consistently increased CGRP levels [4]. Oppositely, our interictal sample material indicates continuously elevated levels of PACAP‐38 in individuals with CH, suggesting increased PACAP‐38 as a fundamental pathophysiological dysfunction and causative contributor to CH pathology.

4.4. Strengths and Limitations

This study investigated PACAP‐38 in a large sample of patients with CH. Methodological strengths are a thoroughly selected cohort of clinically well‐characterized participants and systematic collection, handling, storage, and analysis of samples. Limitations are, firstly, the inevitable selection bias of recruiting participants from a tertiary headache center. Secondly, our choice of CGRP‐assay constitutes a possible limitation, due to the proposed variance of assay sensitivity. No prior literature documents the same variance of PACAP‐38 assays, but this cannot be ruled out. Thirdly, patients were allowed to use preventive and abortive medication due to ethical considerations and further to have episodic headaches other than CH, possibly affecting measured plasma levels of the two neuropeptides. Fourth, this study only encompassed interictal sampling due to feasibility; thus, a demonstration of both ictal PACAP‐38 levels and changes over time remains a desirable research prospect. Furthermore, the inclusion of a positive control group, in terms of a cohort with another chronic headache disorder, for example, chronic migraine or new daily persistent headache, may aid in determining whether PACAP‐38 increasement may be associated with a general chronic headache status and not CH specifically.

5. Conclusion

In summary, this prospective study presents a large‐scale assessment of PACAP‐38 plasma levels in patients with CH. PACAP‐38 plasma levels were found elevated in all three CH disease states compared to headache‐free controls, holding promise for a future possible response to PACAP‐38‐targeted treatments in a profoundly burdened patient group.

Author Contributions

Marie‐Louise K. Søborg: conceptualization, writing – original draft, methodology, writing – review and editing, formal analysis, data curation, investigation. Nunu Lund: conceptualization, writing – review and editing. Agneta Snoer: conceptualization, writing – review and editing. Mads Barloese: writing – review and editing, supervision. Rigmor Højland Jensen: supervision, writing – review and editing, conceptualization. Anja Sofie Petersen: conceptualization, investigation, writing – review and editing.

Conflicts of Interest

The study was funded partly by an investigator‐initiated grant from Lundbeck Pharma and partly by The Capital Region of Denmark's Research Foundation and the Research Foundation of Rigshospitalet. Lundbeck Pharma paid for the analyses of the samples. Payments were made directly to Celerion, Switzerland. Lundbeck Pharma did not influence the design, conduct, or interpretation of this study. M.‐L.K.S. and A.S.P. both report being sub‐investigators in clinical trials and received a restricted research grant from Lundbeck Pharma (paid to the institution), a manufacturer of Eptinezumab (a CGRP‐targeting monoclonal antibody under investigation as a preventive treatment for cluster headache). A.S.P. has given lectures for Pfizer. N.L. has received a restricted research grant from The Capital Region of Denmark's Research Foundation and Lundbeck Pharma and has given lectures for Pfizer, Viatris, and Dagens Medicin. A.S. was working as a medical doctor in the DHC at the time of conceptualizing this project but is currently employed by H. Lundbeck A/S, Copenhagen. M.B. reports no disclosures. R.H.J. reports lectures for Eli‐Lilly, which manufactures Galcanezumab (a monoclonal antibody targeting CGRP, FDA approved as a preventive treatment for episodic cluster headache). R.H.J. is a principal investigator in clinical trials for Lundbeck Pharma and received funding as a restricted and an unrestricted research grant (paid to the institution) from Lundbeck Pharma, a manufacturer of Eptinezumab (a CGRP‐targeting monoclonal antibody under investigation as a preventive treatment for cluster headache).

Supporting information

Figure S1: Title: Comparison of calcitonin gene‐related peptide (CGRP) plasma‐levels in the four study groups. No significant difference in calcitonin gene‐related peptide (CGRP) plasma‐level was found between the four study groups.

Søborg M.‐L. K., Lund N., Snoer A., Barloese M., Jensen R. H., and Petersen A. S., “ PACAP‐38 in Cluster Headache: A Prospective, Case–Control Study of a Potential Treatment Target,” European Journal of Neurology 32, no. 9 (2025): e70341, 10.1111/ene.70341.

Data Availability Statement

The Danish Cluster Headache Biobank contains sensitive information and can consequently not be shared in full form according to Danish data protection law. De‐identified data that underlie the results of this article can be shared with qualified researchers upon request.