Second messenger signalling bypasses CGRP receptor blockade to provoke migraine attacks in humans

Abstract There are several endogenous molecules that can trigger migraine attacks when administered to humans. Notably, calcitonin gene-related peptide (CGRP) has been identified as a key player in a signalling cascade involved in migraine attacks, acting through the second messenger cyclic adenosine monophosphate (cAMP) in various cells, including intracranial vascular smooth muscle cells. However, it remains unclear whether intracellular cAMP signalling requires CGRP receptor activation during a migraine attack in humans. To address this question, we conducted a randomized, double-blind, placebo-controlled, parallel trial using a human provocation model involving the administration of CGRP and cilostazol in individuals with migraine pretreated with erenumab or placebo. Our study revealed that migraine attacks can be provoked in patients by cAMP-mediated mechanisms using cilostazol, even when the CGRP receptor is blocked by erenumab. Furthermore, the dilation of cranial arteries induced by cilostazol was not influenced by the CGRP receptor blockade. These findings provide clinical evidence that cAMP-evoked migraine attacks do not require CGRP receptor activation. This discovery opens up new possibilities for the development of mechanism-based drugs for the treatment of migraine.


Introduction
Migraine is a highly prevalent disease affecting more than 1 billion people worldwide. 1,2It is characterized by stereotypical attacks of headache accompanied by nausea, vomiting and hypersensitivity to light and sound. 3,4The pathogenesis of migraine attacks involves complex interactions between signalling events within intracranial blood vessels and perivascular trigeminal sensory afferents, denoted as the trigeminovascular system. 3,5 key feature of migraine is that various environmental and endogenous trigger factors can initiate an attack. 6Given that migraine attacks are self-limiting and treatable, this feature provides a unique opportunity to dissect the underlying molecular signalling cascade in vivo in patients.By administering signalling molecules to patients and observing if it triggers an attack, we can gain insights into the underlying mechanisms. 3This is also known as a human provocation model. 6These type of studies, along with others, have provided evidence that calcitonin gene-related peptide (CGRP), a neuropeptide found in abundance in the trigeminovascular system, plays a crucial role. 5When CGRP binds to its receptor, it activates a downstream signalling cascade involving cyclic adenosine monophosphate (cAMP). 5][9][10][11][12][13] This highlights the importance of CGRP and cAMP signalling in migraine pathogenesis.Assuming a single-cell system, it is plausible for a downstream mechanism, such as PDE3 inhibition by cilostazol, to influence an upstream mechanism like CGRP receptor activation through positive or negative feedback loops.Such interactions, where downstream components affect upstream components, are common in human biology and illustrate the complexity of intracellular signalling regulation.However, it is important to recognize that migraine pathogenesis is more intricate than a single-pathway single-cell system.The trigeminovascular system comprises multiple cells from different tissues, adding complexity to the signalling pathways and interactions involved. 3,5Clinical trials have demonstrated that administration of pituitary adenylate cyclase-activating polypeptide (PACAP) and vasoactive intestinal polypeptide (VIP) can also provoke migraine attacks in humans. 3,6,14Similar to CGRP, both PACAP and VIP are able to increase intracellular cAMP levels. 5,15,16oreover, erenumab, a fully human immunoglobulin (Ig) G2 monoclonal antibody (mAb) that targets and blocks the CGRP receptor, was introduced as a migraine preventative drug in 2018. 17,18ther drugs targeting the CGRP signalling pathway have followed. 2owever, not all patients benefit from these treatments.These observations suggest that migraine attacks may not solely depend on the CGRP receptor but could involve a unified downstream cAMP-mediated mechanism.However, this hypothesis has not been directly tested in humans. 3,5hus, it is timely to ascertain whether migraine attack generation requires CGRP receptor activation.This can be established through the use of human provocation models.We hypothesized that treatment with erenumab would mitigate the physiological effects of administration of CGRP, but not those of cilostazol.To test this hypothesis, we conducted a randomized, double-blind, placebo-controlled, parallel trial.

Materials and methods
The protocol was approved by the Regional Health Research Ethics Committee of the Capital Region of Denmark (identifier: H-19073983), the Danish Data Protection Agency (P-2020-652) and the Danish Medicines Agency (identifier: 2020033418).We obtained a signed consent form at the time of screening before any protocol-related procedures or assessments.All study-related procedures complied with the Declaration of Helsinki, with later revisions.The study is registered in ClinicalTrials.gov(identifier: NCT04452929).

Participants
Adults aged 18-65 years old with a diagnosis of migraine according to the International Classification of Headache Disorders, Third Edition (ICHD-3), 4 were eligible for participation if they had a history of migraine for ≥12 months prior to screening with ≥4 migraine days per month in the 3 months prior to screening.Key exclusion criteria were history of any other primary headache disorder except for tension-type headache, any secondary headache disorder, daily consumption of any medication other than oral contraception and prior treatment with monoclonal antibodies or participation in clinical trials with monoclonal antibodies during the preceding 12 months.Study eligibility was ensured by review of medical records and evaluation by headache specialists at the Danish Headache Center.A full overview of inclusion and exclusion criteria is available on ClinicalTrials.gov(identifier: NCT04452929).

Study design and procedures
We enrolled participants in a randomized, double-blind, placebocontrolled, parallel trial at a single centre in Denmark (Fig. 1).Participants were randomly allocated to a subcutaneous administration of 140 mg of erenumab or placebo (isotonic saline).Erenumab was provided by Novartis Pharma AG and sorted in blinded packaging by Nomeco A/S.Independent pharmacy staff were responsible for randomization and allocation concealment.Seven to 21 days after study drug administration, participants were randomly allocated to receive a continuous intravenous infusion of 1.5 μg/min of CGRP over 20 min or oral intake of 200 mg cilostazol on two experimental study days separated by at least 3 days.Patients were informed that CGRP and cilostazol might induce headache or migraine attack with no additional information on induction frequency, time to onset or features; in addition, patients were informed that erenumab may mitigate the effects of these experimental triggers with no information provided on possible mechanism or site of action.
Patients arrived between 08:00 a.m. and 12:00 p.m. on the experimental study days.Patients were in a supine-position during study-related procedures and assessments.Patients were not allowed to consume any medication within 24 h or four times the plasma half-life of the drug (whichever longest) on the experimental study days except for oral contraception.Furthermore, to remain eligible, participants could not have had a migraine attack during the preceding 48 h nor a headache during the preceding 24 h prior to administration of the experimental trigger.

Headache characteristics and diary
A headache specialist conducted a semi-structured interview at the screening visit.The interview included information on medical history, headache characteristics and frequency, family history of migraine and prior pharmacological treatment; these data were confirmed by review of medical records.
On the experimental study days, investigators recorded data on headache characteristics, vital signs and adverse events every 10 min for 90 min after administration of the experimental trigger.After the in-hospital phase, patients were discharged with a 12-h headache diary for hourly recordings of headache features, accompanying symptoms, intake of any rescue medication and adverse events.

Haemodynamic variables
Patients rested in a quiet room for at least 30 min in a supine position before any measurements.During the in-hospital phase, we conducted electrocardiography and measured the blood pressure and heart rate.Furthermore, we used a high-resolution ultrasonography unit (Dermascan C; Cortex Technology: 20 MHz, bandwidth 5 MHz) as previously described to measure the diameter of the frontal branch of the superficial temporal artery and the radial artery. 19,20

Statistical considerations and analysis
Sample size calculations were based on the difference between two unpaired groups reporting migraine attacks.After administration of CGRP, we assumed that 51% of participants would report migraine attack in the placebo-treatment arm and 20% in the activetreatment arm.After administration of cilostazol, we assumed that 66% of participants would report migraine attack in the placebo-treatment arm and 20% in the active-treatment arm.These assumptions were based on previous migraine provocation studies, in which ∼66% of participants with migraine developed attacks after administration of CGRP and ∼86% after cilostazol, [7][8][9][10][11][12][13] and that treatment prior to a provocation challenge lowers the expected induction rate by ∼20% (0.2 × historical induction rate). 21At 80% power and a 5% level of significance, we estimated that 72 patients would be needed in a two-arm parallel trial.
The primary end point was the incidence of migraine attacks in a 12-h observational period after administration of experimental triggers. 22The secondary end points were difference in area under the curve (AUC) values for headache intensity scores, superficial temporal artery diameter, radial artery diameter, heart rate and mean blood pressure.Pearson's chi-squared test was used to analyse the incidence of migraine attacks and headache as binary categorical data.We made calculations of AUC values in accordance with the trapezium rule. 23We used the Mann-Whitney test to analyse differences in headache AUC scores after administration of CGRP or cilostazol between the active-treatment arm versus placebo-treatment arm.Analyses of haemodynamic AUC values were analysed using unpaired, two-way t-tests.All statistical analyses were performed using R (version 4.1.0). 24A 5% level of significance was accepted for all comparisons.We present data as mean values with standard deviation (SD), except the headache score which is presented as the median.Time to onset of migraine attack, time to intake of rescue medication, time to peak headache intensity and duration of migraine-associated symptoms are presented as median values.Baseline was defined as the time of the start of infusion (T 0 ).

Study population characteristics
A total of 80 patients provided informed consent from July 2020 to July 2021; 75 patients completed both experimental study days and were included in the final analysis (Fig. 2).One patient discontinued study participation prior to administration of study drug for reasons unrelated to safety; four were not able to complete both experimental study days between Day 7 and Day 21 due to occurrence of headache within 24 h prior to administration of the experimental trigger.Demographic and baseline characteristics are presented in Table 1.Mean age was 33.4 years (SD: 11.1, range: 20-65 years), mean weight was 69.6 kg (SD: 12.8, range: 50-99 kg) and mean body mass index (BMI) was 23.9 (SD: 3.5, range: 19.1-34.8).For headache and medication intake frequency 3 months prior to enrollment, the mean number of monthly headache days was 10.6 days (SD: 4.7, range: 4-24 days), mean number of monthly migraine days was 7.2 days (SD: 3, range: 4-15 days) and mean number of monthly acute medication intake was 6.4 days (SD: 3.2, range: 0-14 days).

CGRP-mediated vasodilation is mitigated by a blockade of the CGRP receptor
We found that erenumab significantly attenuated CGRP-induced dilation of the superficial temporal artery (AUC 0-90min , P < 0.00001)  To study the effects of calcitonin gene-related peptide (CGRP) receptor blockade on experimentally induced migraine attacks, we administered erenumab or placebo to adults with migraine in a randomized, double-blind, placebo-controlled, parallel trial design.

Adverse events
The proportion of patients reporting warm sensations, palpitations and flushing in the erenumab group was lower than in the placebo group (Table 3).2][13] Following administration of cilostazol, the proportion of patients reporting warm sensations and palpitations in the erenumab group was similar to the placebo group (Table 3).

Discussion
A novel finding of our study is that migraine attacks induced by upregulation of intracellular cAMP levels are not affected by blocking the CGRP receptor.Our work provides clinical evidence that cAMP-evoked migraine attacks do not require CGRP receptor activation.We want to highlight three important observations in this context: (i) patients who were pretreated with erenumab still experienced migraine attacks when cilostazol was administered; (ii) patients who were susceptible to migraine attacks following CGRP administration also had migraine attacks induced by cilostazol; and (iii) most participants who did not experience migraine attacks after receiving CGRP did have migraine attacks when cilostazol was administered.
Cilostazol, a PDE-3-inhibitor, blocks both isoforms of PDE3, which is responsible for metabolizing cAMP. 25As a result, cilostazol leads to accumulation of intracellular cAMP.It is important to note that cilostazol does not directly increase cAMP generation but inhibits its degradation, thus relying on upstream regulators such as the CGRP receptor or other factors.Notably, the observation that cilostazol induces migraine attacks even in the presence of a CGRP receptor blockade suggests the existence of a CGRP-independent pathway for migraine attacks in humans.A preclinical study demonstrated that PACAP-induced hypersensitivity in a mouse model of migraine-like pain is independent of CGRP. 26However, it is important to interpret these findings cautiously as CGRP receptor antagonism was still able to inhibit pain induced by cilostazol in the same mouse model. 27This suggests that the translatability of the model may have limitations.Nonetheless, the PACAP signalling pathway is of interest as both the ligand and its receptors are expressed within the trigeminovascular system. 5In humans, PACAP binds to three different receptors (VPAC1, VPAC2 and PAC1), and experimental data have shown that activation of these receptors can increase intracellular levels of cAMP. 5Furthermore, a recent clinical trial demonstrated that Lu AG09222, an anti-PACAP mAb, was effective in inhibiting PACAP38-induced cephalic vasodilation and reducing concomitant headache in healthy volunteers. 28ilostazol primarily inhibits the isoform PDE3A, which is predominantly found in vascular smooth muscle cells. 29Its inhibition of PDE3 leads to dilation cranial arteries in both healthy volunteers and individuals with migraine. 30,31Of note, cilostazol is used to relieve symptoms of intermittent claudication.Partly because of its vasodilatory effect. 32Our findings support a vascular site of action, as we observed a minor increase in arterial diameter during the 90 min in-hospital monitoring phase.However, further observation over a longer period is necessary to confirm these results, as previous studies with longer observational periods have shown a more pronounced increase in arterial diameter over time. 30,31These results align with the observation that stimulation of downstream targets of cilostazol, such as ATP-sensitive potassium (K ATP ) channels and large-conductance calcium-activated potassium (BK Ca ) 1 (0-7) 1 (0-6) 1 (0-7) BMI = body mass index; SD = standard deviation.a High frequency episodic migraine defined as <15 monthly headache days with ≥8 monthly migraine days.
b Data were collected retrospectively at baseline through a semi-structured interview.
4][35] Interestingly, cilostazol can bind directly to the inner surface of BK Ca channels, stimulating their activity. 36Sumatriptan (an anti-migraine drug) inhibits cAMP/ PKA-signalling [37][38][39][40] and constricts the superficial temporal artery and middle meningeal artery in parallel with pain relief during cilostazol-induced migraine attacks.This suggests a perivascular or peripheral site of action. 30Given that cilostazol is a lipophilic compound, 41,42 we cannot exclude additional peripheral or central neuronal sites of action.Although both isoforms of PDE3 are colocalized with CGRP in the trigeminal ganglion, 43 PDE3 activity does not regulate CGRP release. 27However, we cannot exclude a direct effect of cilostazol on the trigeminal ganglion as the PDE3A isoform is present in non-neuronal cells there. 44Furthermore, activation of the cAMP-PKA second messenger pathway can sensitize dural mechanonociceptors in rats, potentially independent of a  Calcitonin gene-related peptide (CGRP): There was a significant difference in the incidence of migraine attacks following administration of CGRP in individuals with migraine randomized to erenumab compared to placebo (P = 0.024).Cilostazol: There was no difference in the incidence of migraine attacks following administration of cilostazol in individuals with migraine randomized to erenumab compared to placebo (P = 0.533).vascular involvement.However, further data are needed to elucidate the potential effects of cilostazol within the nervous system, considering its limited ability to cross the blood-brain barrier in several species, 45 which is consistent with the rarity of adverse events related to the CNS for cilostazol. 32e found that CGRP receptor blockade did not completely prevent CGRP-induced migraine attacks.One-quarter of patients in the erenumab group still experienced a migraine attack following administration of CGRP, which exceeds the expected rate of a nocebo response [8.1% (95% CI, 2.5-15.5%)]. 46This finding is clinically interesting as one-third of migraine patients do not report a substantial reduction in migraine frequency after treatment with an anti-CGRP receptor mAbs. 18,47,48It is worth mentioning that CGRP receptor blockade did partially mitigate CGRP-induced vasodilation, but there was still a residual vascular response.This suggests that CGRP signalling, although to a lesser extent, may also occur through receptors other than the canonical CGRP receptor.This residual effect could provide a partial explanation for why breakthrough migraine attacks can still occur despite treatment with anti-CGRP receptor mAbs, albeit at a lower frequency.][51][52][53] Supposedly, circulating CGRP could bind to the AMY1 receptor, leading to mild vasodilation, [49][50][51][52][53] as observed in our study.0][51] Further data are needed to elucidate the role of the AMY1 receptor and its interaction with CGRP signalling in migraine pathophysiology.
Although we applied rigorous criteria during recruitment and standardized experimental procedures, we could not rule out various environmental factors (e.g.stress, food intake) after discharge from the laboratory and the subsequent outpatient 12-h observation phase.However, alternative 12-h inpatient monitoring would have limited the feasibility of the study.In addition, the migraine attack induction rates for CGRP and cilostazol in the placebo arm were comparable to previously reported studies with these substances. 11,12,31,54Differences in T max between participants are possible after oral administration of cilostazol.Nonetheless, the induction rate and median time to onset of cilostazol-induced migraine attacks is comparable to previous reports. 11,12,31,54As mentioned earlier, we noticed a slight increase in the diameter of the The increase in the diameter of the superficial temporal artery in individuals randomized to erenumab was significantly less compared to placebo (P < 0.00001).Erenumab, however, did not completely inhibit the increase in diameter as there was a small increase compared to historical data with saline infusion in individuals with migraine without preventive medication.(B) Mean change in diameter of the radial artery relative to baseline.The increase in the diameter of the radial artery in individuals randomized to erenumab was significantly less compared to placebo (P < 0.00001).Erenumab, however, did not appear to completely inhibit the increase in diameter as there was a small increase compared to historical data with saline infusion in individuals with migraine without preventive medication.(C) Mean heart rate in absolute values.There was a significant increase in the heart rate in individuals randomized to placebo compared to erenumab (P < 0.00001).(D) Mean arterial pressure in absolute values.There was a significant decrease in the mean arterial pressure in individuals randomized to placebo compared to erenumab (P < 0.01).Error bars represent standard deviation.Historical data, A 33 and B. 35 superficial temporary artery in both study arms after administering cilostazol.However, a longer observation period would have given us more information about potential vascular effects, as previous studies have shown a progressive increase in arterial diameter over time. 30,31We cannot exclude differences in bioavailability of erenumab between individuals.6][57] Experimental study days took place between Day 7 and Day 21 after randomization, enabling onset and maintenance of efficacy.

Figure 1
Figure 1 Study design.Participants were enrolled in a randomized, double-blind, placebo-controlled, parallel trial.Participants were randomly allocated to a subcutaneous administration of 140 mg of erenumab or placebo (isotonic saline).Seven to 21 days after study drug administration, participants were randomly allocated to receive a continuous intravenous infusion of 1.5 μg/min of calcitonin gene-related peptide (CGRP) over 20 min or oral intake of 200 mg cilostazol on separate experimental study days.

Figure 2
Figure 2 Flow chart of participant recruitment of individuals with migraine.To study the effects of calcitonin gene-related peptide (CGRP) receptor blockade on experimentally induced migraine attacks, we administered erenumab or placebo to adults with migraine in a randomized, double-blind, placebo-controlled, parallel trial design.

Figure 3
Figure 3 Headache characteristics after administration of calcitonin gene-related peptide (CGRP) in individuals with migraine randomized to erenumab or placebo.Participants were randomized to erenumab or placebo 7-21 days prior to administration of CGRP.Headache intensity was rated on an 11-point numeric rating scale from 0 to 10 (median, thick lines).The median time to onset of migraine attack was 240 min (range: 30-540 min) in the erenumab group and 40 min (range: 10-480 min) in the placebo group.The median time to intake of rescue medication was 240 min (range: 120-660 min) in the erenumab group and 150 min (range: 40-480 min) in the placebo group.Square markers represent individual reported headache intensity.

Figure 4
Figure 4 Headache characteristics after administration of cilostazol in individuals with migraine randomized to erenumab or placebo.Participants were randomized to erenumab or placebo 7-21 days prior to administration of cilostazol.Headache intensity was rated on an 11-point numeric rating scale from 0 to 10 (median, thick lines).The median time to onset of migraine attack was 240 min (range: 10-600 min) in the erenumab group and 300 min (range: 20-600 min) in the placebo group.The median time to intake of rescue medication was 300 min (range: 120-720 min) in the erenumab group and 390 min (range: 70-600 min) in the placebo group.Square markers represent individual reported headache intensity.

Figure 5
Figure 5 Haemodynamic changes following administration of calcitonin gene-related peptide (CGRP) in individuals with migraine randomized to erenumab or placebo.(A) Mean change in diameter of the superficial temporal artery relative to baseline.The increase in the diameter of the superficial temporal artery in individuals randomized to erenumab was significantly less compared to placebo (P < 0.00001).Erenumab, however, did not completely inhibit the increase in diameter as there was a small increase compared to historical data with saline infusion in individuals with migraine without preventive medication.(B) Mean change in diameter of the radial artery relative to baseline.The increase in the diameter of the radial artery in individuals randomized to erenumab was significantly less compared to placebo (P < 0.00001).Erenumab, however, did not appear to completely inhibit the increase in diameter as there was a small increase compared to historical data with saline infusion in individuals with migraine without preventive medication.(C) Mean heart rate in absolute values.There was a significant increase in the heart rate in individuals randomized to placebo compared to erenumab (P < 0.00001).(D) Mean arterial pressure in absolute values.There was a significant decrease in the mean arterial pressure in individuals randomized to placebo compared to erenumab (P < 0.01).Error bars represent standard deviation.Historical data, A33 and B.35

Figure 6
Figure 6 Haemodynamic changes following administration of cilostazol in individuals with migraine randomized to erenumab or placebo.(A) Mean change in diameter of the superficial temporal artery relative to baseline.There was no difference in the diameter of the superficial temporal artery in individuals randomized to erenumab compared to placebo (P = 0.238).(B) Mean change in diameter of the radial artery relative to baseline.There was no difference in the diameter of the radial artery in individuals randomized to erenumab compared to placebo (P = 0.984).(C) Mean heart rate in absolute values.There was no difference in the heart rate in individuals randomized to erenumab compared to placebo (P = 0.659).(D) Mean arterial pressure in absolute values.There was no difference in the mean arterial pressure in individuals randomized to placebo compared to erenumab (P = 0.480).Error bars represent standard deviation.