Abstract
Background
Previously reported increases in serum levels of vasodilating neuropeptides pituitary adenylate cyclase-activating peptide-38 (PACAP38) and vasoactive intestinal peptide (VIP) during attacks of cluster headache could indicate their involvement in cluster headache attack initiation. We investigated the attack-inducing effects of PACAP38 and vasoactive intestinal peptide in cluster headache, hypothesising that PACAP38, but not vasoactive intestinal peptide, would induce cluster-like attacks in episodic active phase and chronic cluster headache patients.
Methods
In a double-blind crossover study, 14 episodic cluster headache patients in active phase, 15 episodic cluster headache patients in remission phase and 15 chronic cluster headache patients were randomly allocated to receive intravenous infusion of PACAP38 (10 pmol/kg/min) or vasoactive intestinal peptide (8 pmol/kg/min) over 20 min on two study days separated by at least 7 days. We recorded headache intensity, incidence of cluster-like attacks, cranial autonomic symptoms and vital signs using a questionnaire (0–90 min).
Results
In episodic cluster headache active phase, PACAP38 induced cluster-like attacks in 6/14 patients and vasoactive intestinal peptide induced cluster-like attacks in 5/14 patients (
Conclusions
Contrary to our hypothesis, attack induction was lower than expected and roughly equal by PACAP38 and vasoactive intestinal peptide in episodic active phase and chronic cluster headache patients, which contradicts the PAC1-receptor as being solely responsible for attack induction.
Keywords
Introduction
Efforts to elucidate cluster headache (CH) pathophysiology have included investigating signaling molecules suspected to be involved in initiation of CH attacks (1–3). Recently, signaling molecule calcitonin gene-related peptide (CGRP) was shown to induce CH attacks in episodic cluster headache patients in active phase (ECHA patients) and in some chronic cluster headache (CCH) patients, but not in episodic cluster headache patients in remission phase (ECHR patients) (3). Interestingly, CGRP monoclonal antibody galcanezumab reduced attack frequency in episodic but not in chronic CH (4,5).
The neuropeptides pituitary adenylate cyclase-activating peptide-38 (PACAP38) and vasoactive intestinal peptide (VIP) are both members of the VIP-glucagon-growth hormone releasing factor-secretin superfamily (6) and are both essential in parasympathetic communication. Of the three G-protein coupled receptors that PACAP38 activates – PAC1, VPAC1 and VPAC2 – VIP has equal affinity for the latter two (7,8) whereas PACAP38 shows a higher affinity for the PAC1 receptor. PACAP38 and VIP are both vasodilating signaling molecules (9,10) and physiological effects of PACAP38 and VIP infusion include marked facial flushing, heat sensation and increased heart rate (11). PACAP38 infusion, but not VIP infusion, triggers migraine-like attacks in migraine without aura patients (11,12), suggesting VIP is unlikely to play a role in migraine patients. In CH, increased PACAP38 and VIP levels were found in episodic CH patients during attacks (13,14). Interestingly, ECHR patients had lower PACAP38-levels than healthy controls (13). Currently, a PAC1 receptor monoclonal antibody showed promising results in an
Here, we investigated the ability of PACAP38 to induce cluster-like attacks. We hypothesised that PACAP38, not VIP, would induce cluster-like attacks in ECHA and CCH patients. We also hypothesised that neither PACAP38 nor VIP would induce attacks in ECHR patients. Due to similar side effects to PACAP38 and VIP and the lack of migraine induction from VIP (11), we deemed VIP suitable as an “active placebo” in this study.
Methods
The participants were recruited from the Danish Headache Center between December 2017 and August 2019. Participants aged 18–65 and weighing 50–100 kg were eligible for the study if diagnosed with episodic or chronic CH according to the International Classification of Headache Disorders 3-beta (17). Episodic CH patients were defined as being in active phase when having experienced attacks within 30 days and in remission phase when attack free for a minimum of 30 days. CCH were attack free <30 days within the last 12 months. All CH preventive medications, except steroid treatments within 30 days, were allowed with stable dosing. Previous participation in CH provocation studies was allowed. Use of safe contraceptive methods if female of childbearing potential was mandatory. Exclusion criteria were: Other primary headache (except episodic tension-type headache <5 days/month and/or migraine if >12 months since last attack), pregnancy or nursing in females, cardiovascular or cerebrovascular disease, psychiatric disease, or drug misuse. The Regional Health Research Ethics Committee of the Capital Region approved the study (H-17011569). All participants gave written consent to participate after receiving detailed oral and written information. The study was conducted in accordance with the Helsinki II Declaration of 1964, with later revisions. The study was registered at clinicaltrials.gov (identifier NCT03814226) and approved by the Danish Data Protection Agency.
Experimental protocol
In a crossover design, we randomly allocated participants to receive either 10 pmol/kg/min of PACAP38 or 8 pmol/kg/min of VIP by continuous intravenous infusion over 20 min using a time- and volume-controlled infusion pump (Braun Perfusor, Melsungen, Germany). Infusions were performed on two study visits separated by at least 7 days. Allocation of PACAP38 and VIP was balanced.
A brief general and neurological examination and a 12-lead ECG was performed before beginning study day 1. On both study days, females underwent urine pregnancy tests. ECHA and CCH patients were attack-free at least 4 hours before study start to ensure patients were not in a refractory state from a previous attack. ECHR patients were headache free 8 hours before study start. Participants were placed in the supine position and had a venous catheter (Venflon, Braun, Melsungen, Germany) inserted into the antecubital vein for PACAP38 or VIP infusions and for drawing blood. A winged infusion set (SAFETY Blood Collection Set + Holder, Greiner-Bio, Kremsmünster, Austria) was inserted into the contralateral antecubital vein for a single blood draw during infusion and was then immediately disposed (data and blood sampling protocols will be reported in separate publications). After resting for 15 minutes, baseline values of headache intensity and vital signs were measured (T−10 and T0). All participants were discharged with a detailed self-administered headache questionnaire to record any headache 24 h post infusion. This questionnaire covered headache intensity, characteristics, cranial autonomic symptoms (CAS), non-headache symptoms, and use of rescue medication.
Cluster-like attack criteria
In accordance with previous provocation studies (Ekbom, 1968; Bogucki, 1990; Fanciullacci
Cluster-like attack fulfilling either (1) or (2):
Unilateral headache described as mimicking the patient’s usual CH attack (with or without cranial autonomic symptom) Headache fulfilling criteria B and C for CH according to IHS criteria (17) B. Severe unilateral pain lasting 15–180 minutes C. Either or both of the following: At least one CAS ipsilateral to the headache A sense of restlessness or agitation
Headache intensity and questionnaire
We recorded headache intensity of any headache type occurring from baseline (T−10 and T0) and until 90 min after infusion. Intensity was rated on an 11-point numerical scale of 0–10 with 0 representing no headache; 1, a very mild headache (including non-painful sensations of pressing, throbbing, or otherwise altered sensation in the head); 5, headache of moderate intensity and 10, the worst headache imaginable. Participants described whether any headache experienced during the 90 min observation period was similar to a usual CH attack or not. Furthermore, headache characteristics (throbbing, stabbing, pressing), headache localisation, and intake of rescue medication, including intake time and dosage, was recorded.
Cranial autonomic symptoms and non-headache symptoms
We recorded the following cranial autonomic symptoms: Conjunctival injection, lacrimation, rhinorrhea, nasal congestion, eyelid edema, facial sweating, ptosis and miosis. In addition, we recorded restlessness and agitation.
Vital signs
We measured mean arterial blood pressure (MAP) and heart rate using an auto-inflatable cuff (Microlife, Widnau, Switzerland) every 10 min from T−10 until T90 and we monitored ECG (Cardiofax V Nihon Kohden, Tokyo, Japan) continuously in the same interval.
Statistical analyses
Headache intensity scores are presented as medians. Time to onset of cluster-like attack, cranial autonomic symptoms, and restlessness as well as intake time of rescue medication after PACAP or VIP infusion are presented as medians, interquartiles, and range. Heart rate and MAP data are presented as median values.
Sample size was based on previous similar migraine studies showing migraine attack-induction after PACAP38 of 58–73% (11,20,21) and of 0–18% after VIP (11,12) at 5% significance with 80% power. We assumed PACAP38 would provoke attacks in at least 70% of participants versus less than 20% after VIP. Sample size was calculated to be 15 participants in all three groups.
The primary end points were: Difference in incidence of cluster-like attacks 90 min after infusion start between PACAP38 and VIP infusion; difference in area under the curve (AUC) for headache intensity scores (0–90 min) after PACAP38 and VIP; and difference in time to attack onset after PACAP38 and VIP infusion. The secondary end points were difference in AUC of MAP and heart rate 90 min after infusion of PACAP38 and VIP. Incidence of cluster-like attacks and CAS in all three groups was analysed as categorical data using McNemar’s test. We calculated AUC according to the trapezium rule to obtain a summary measure. AUC0–90min for headache scores, MAP, and heart rate were compared using Wilcoxon signed-rank test. Incidence of attacks and AUC0–90min of headache are presented with means and 95% CI.
All analyses were performed with GraphPad Prism Statistics version 8 for Windows (GraphPad Software, Inc., La Jolla, CA, USA). We made no adjustment for multiple analyses. Thus, a level of significance at 0.05 was accepted for each comparison.
Results
We recruited 41 participants (34 men and 7 women) of whom three episodic CH patients took part both in active and remission phase (see Figure 1). Thus, data equivalent to 44 participants was included in final analyses. Mean (standard deviation) age was 38 (11.6) (range 18–60), mean weight 78 kg [12.4] (range 54–100). Enrolment of ECHA stopped before reaching 15 participants because of recruitment difficulties. Of episodic CH patients, 14 were included in active phase and 15 were included in remission, and three patients participated in both phases. Fifteen CCH patients were included. One ECHR patient was excluded before study day 2 due to a flare up in pre-existing sarcoidosis, which was inactive prior to inclusion, and this patient was replaced. Patient descriptions of usual attacks, headache and/or cluster-like attacks on both study days, acute and preventive medications are listed in Table 1.
Flow chart of participant recruitment of episodic and chronic cluster headache patients. One episodic cluster headache remission phase patient was excluded between study day 1 and 2 due to a flare up in a pre-existing sarcoidosis, which happened months after participation in study day 1. The subject was replaced.
Attack characteristics of episodic cluster headache patients in active phase, remission, and chronic cluster headache patients, during spontaneous and provoked attacks after pituitary adenylate cyclase activating peptide and VIP infusion. Cranial autonomic symptoms, restlessness, and acute treatment in episodic cluster headache in active phase (ECHA) patients, chronic cluster headache (CCH) patients, and episodic cluster headache remission phase (ECHR) patients from 0–90 min after 20 min intravenous infusion of pituitary adenylate cyclase activating peptide (PACAP38) (10 pmol/kg/min) or vasoactive intestinal peptide (VIP) (8 pmol/kg/min). Type and dose of preventive medications taken. Patients who developed a cluster-like attack on either study day are marked in red.
1Autonomic symptoms/restlessness: Inj: conjuctival injection; lac: lacrimation; rhi: rhinorrhea; con: nasal congestion; ede: eyelid edema; swe: forehead and facial sweating; pto: ptosis; mio: miosis; res: restlessness.
2Time in minutes.
3Defined according to criteria described in methods.
4Acute therapy: Imi: Sumatriptan 6 mg sc inj; oxy: oxygen; suma: sumatriptan 100 mg orally; SPG: sphenopalatine ganglion neurostimulator; therapy effect: at least 50% reduction in pain intensity over two hours.
*For one 10 min interval the patient was unable to answer either yes or no as to whether pain mimicked usual cluster attack.
Episodic cluster headache active phase
During the 90 min in-hospital phase, six of 14 ECHA patients (43%) developed a cluster-like attack after PACAP38 compared to five of 14 (36%) after VIP (
(a) Numerical Rating Scores of episodic cluster headache active phase patients. Median (purple/green thick line) and individual (thin lines) headache intensity on an 11-point numerical scale from 0 to 10 for 14 episodic cluster headache patients in active phase. There was no difference in AUC0–90 min for headache intensity after PACAP38 compared to VIP (
CAS appeared unilaterally on the pain side during all six PACAP38-induced cluster-like attacks. Median time to onset of attack, onset of autonomic symptoms, onset of restlessness, and median time to intake of medication are shown in Figure 3 (PACAP38) and Figure 4 (VIP).
Onset of cluster-like attack, cephalic autonomic symptoms, restlessness, and intake of rescue medication (medians, interquartiles, and range) from 0–90 min after 20 min intravenous infusion of pituitary adenylate cyclase activating peptide-38 (PACAP38) (10 pmol/kg/min). Red bars: Chronic cluster headache (n = 15). Blue bars: Episodic cluster headache in active phase (n = 14).
Onset of cluster-like attack, cephalic autonomic symptoms, restlessness, and intake of rescue medication (medians, interquartiles, and range) from 0–90 min after 20 min intravenous infusion of vasoactive intestinal peptide (VIP) (8 pmol/kg/min). Red bars: Chronic cluster headache (n = 15). Blue bars: Episodic cluster headache in active phase (n = 14).
Episodic cluster headache remission phase
During the 90 min in-hospital phase, none of the ECHR patients experienced cluster-like attacks after PACAP38 or VIP.
During the 90 min phase, three of 15 ECHR patients (20%) experienced headache following both PACAP38 and VIP, although not the same three patients on PACAP and VIP days. We found no difference in AUC0–90 min for headache intensity after PACAP38 compared to VIP (
After PACAP38 infusion, seven of 15 ECHR patients (47%) experienced unilateral CAS compared to three of 15 (20%) after VIP infusion. After PACAP38, median time to onset of CAS was 20 min (range 10–60 min) and median duration was 50 min (range 10–90 min). After VIP, median time to onset of CAS was 30 min (range, 10–40 min) and median duration of symptoms was 30 min (range, 10–110 min).
Chronic cluster headache
During the 90 min in-hospital phase, seven of 15 CCH patients (47%) developed a cluster-like attack after PACAP38 infusion and seven of 15 (47%) developed cluster-like attack after VIP infusion (
Median time to onset of attack, onset of CAS, onset of restlessness, and median time to intake of medication are shown in Figure 3 (PACAP38) and Figure 4 (VIP).
Preventive medications
Nine of 14 ECHA patients (64%) took preventive medication (Table 1). Of the eight ECHA patients who developed cluster-like attacks on one or both study days, six were on preventives and two were not.
Nine of 15 CCH patients (60%) were on preventive medication. Of the nine CCH patients who developed cluster-like attacks on one or both study days, five were on preventives and four were not. One of 15 ECHR patients took preventive medication.
A complete description of type of medication and dose taken for each patient can be found in Table 1.
Mean arterial blood pressure and heart rate
MAP after PACAP38 infusion was not different compared to after VIP infusion in ECHA (MAP AUC0‐90 min,
Heart rate after PACAP38 infusion was higher compared to after VIP infusion in ECHA (heart rate AUC0–90 min,
Discussion
Contrary to our hypothesis, our main finding was that ECHA patients experienced fairly similar incidences of cluster-like attacks from PACAP38 (43%) as from VIP (36%) and that CCH patients experienced exactly the same incidence of cluster like-attacks induced from both PACAP38 (47%) and VIP (47%). Another key finding was that ECHR patients developed zero cluster-like attacks after PACAP38 and VIP infusion. Thus, the primary end points of this study were not met and we must therefore consider both PACAP38 and VIP as weaker provocative agents of cluster headache compared to previously investigated substances.
An overview of previous provocation rates in cluster headache provocation studies can be seen in Table 2. The cluster-like attack induction rates by PACAP38 and VIP in the present study fall within the very wide interval for glyceryl trinitrate (GTN) (not placebo-controlled studies) in ECHA and CCH patients (1,2,18,22–27). Moreover, our induction rates compare to those of CCH patients after CGRP (a placebo-controlled study) (28). However, our provocation rates by both PACAP38 and VIP in ECHA are considerably lower than those by histamine and CGRP (18,28). In the CGRP provocation study, the markedly different attack induction rates by ECHA and CCH indicated possible differing pathophysiological roles of CGRP in those two subtypes of CH. In the present study, both PACAP38 and VIP numerically induce more attacks in CCH than in ECHA; however, the differences in attack induction are small and thus do not clearly indicate a preferential role of attack induction in CCH.
Previous rates of induction of cluster-like attacks in episodic cluster headache patients in active phase (ECHA), episodic cluster headache patients in remission phase and chronic cluster headache (CCH) using histamine, glyceryl trinitrate (GTN), calcitonin gene-related peptide (CGRP), pituitary adenylate-cyclase activating peptide-38 (PACAP38), vasoactive intestinal peptide (VIP) and placebo.
1ECHA: Studies using doses of both 0.9 mg and 1 mg sublingual glyceryl trinitrate. CCH: Studies using doses of both 0.9 mg and 1 mg sublingual glyceryl trinitrate and inhalation of 1.0–1.2 mg glyceryl trinitrate.
2From Vollesen et al. (28), a randomised, double-blind, placebo-controlled cross-over study where infusion of placebo (saline) and 1.5 µg/min calcitonin gene-related peptide (CGRP) were investigated.
Our finding of zero attacks in ECHR correspond to findings from GTN and CGRP provocation of CH (1,2,18,22–28). Thus, there are repeated data on the lack of cluster-like attack induction in ECHR patients. Furthermore, a remarkably low placebo attack induction rate was seen in the active phase patients in the CGRP vs. placebo study (11% in ECHA, 0% in CCH) (28). This underscores that cluster-like attacks are not readily provoked in patients where we would not expect to see them (i.e. patients in remission phase or patients receiving placebo intervention). Taken together, while the attack induction rates by PACAP38 were lower than expected and while VIP failed as an active placebo, we argue that both signaling molecules do actually provoke cluster-like attacks but are perhaps less powerful drivers of these attacks.
The fact that cluster-like attacks were induced in the present study (although at a modest rate) poses the question as to how the infusion of PACAP38 or VIP induces attacks in CH patients. In CH, increased PACAP (13) and VIP (14) plasma levels during spontaneous CH attacks support their possible involvement in initiating attacks. The localisation of CH pain to the ophthalmic division of the trigeminal nerve and the co-occurrence of CAS point to the trigeminal autonomic reflex (TAR) as being involved – although it is less obvious if CAS result from the pain or are, in fact, driving the pain (29). The TAR comprises afferent trigeminal nerve input and parasympathetic output through the superior salivatory nucleus to the facial nerve via the SPG (29). The TAR presumably interacts with the hypothalamus in CH (30); however, the precise mechanism and order of events is not fully understood.
PACAP38 and its receptors are strategically well located at both afferent and efferent ends of the TAR (31,32). At the afferent end, only PACAP38, not VIP, is found in the trigeminal ganglion (33), which points against the trigeminal ganglion as the main attack-triggering site. However, since VPAC1−2 receptors, to which VIP has equal affinity as PACAP38,
Attack initiation at the efferent end of the TAR (via the SPG) is also theoretically possible (29) due to the location of PACAP38, VIP and the VPAC1-2 and PAC1-receptors in the SPG (32). Interestingly, we found median time to onset of CAS after PACAP38 infusion preceded onset of attacks. Also, after VIP infusion in ECHA patients, onset of CAS preceded onset of attack. However, in a previous sham-controlled crossover study (35), parasympathetic activation by low frequency SPG stimulation in 20 CH patients implanted with an SPG neurostimulator did not trigger more attacks than sham stimulation (35% vs. 25%), although low-frequency stimulation induced CAS in 80% of patients. This suggests that parasympathetic stimulations are insufficient to trigger attacks. Additionally, in our group of ECHR patients, we showed that triggering parasympathetic outflow (i.e. CAS in 47% of patients after PACAP38 infusion and 20% after VIP) did not suffice to induce attacks.
Our ECHR patients also illustrate that a susceptibility – which is presumably centrally controlled – is necessary for triggering substances to elicit attacks. VIP passes the blood-brain-barrier to a very small degree (36,37). Exogenous infusion of PACAP38 is also not considered to cross the blood-brain barrier in a manner leading to changes in brain function (38). Thus, the present study suggests that infusion of peripherally acting substances can only trigger cluster-like attacks when activating a system already in a permissive state. Previously, the hypothalamus (3) was named a potential modulator of this permissiveness, though this remains speculative at present.
Which receptor(s) could be responsible for PACAP38- and VIP-induced cluster-like attacks is unknown. In migraine patients, the lack of migraine-like attack induction by VIP implicates the PAC1-receptor in PACAP38-induced migraine. Supporting this, PAC1 knockout mice show reduced pain responses (39,40) and in rats, a centrally given PAC1-receptor antagonist (41) and a monoclonal PAC1-receptor antibody (15) inhibited dural nociceptive-evoked activation of central trigeminovascular neurons. Interestingly, a PAC1-receptor antibody recently failed its primary end points in a phase II migraine prevention study (35,41,42); however, issues with dosing, potency or alternative receptor splice variants (43) cannot be excluded as reasons for the failure. A recently developed high-affinity and selective antibody for PACAP (both PACAP27 and PACAP38) could clarify whether targeting the PACAP molecule holds therapeutic potential (44). Given our similar cluster-like attack induction by PACAP38 and VIP, it seems unlikely the PAC1-receptor is solely responsible for PACAP38- and VIP-induced attacks.
A significant limitation of our study is the lack of a pure placebo arm. To ensure blinding of both participants and staff, we chose VIP infusion since its side effect profile is very similar to that of PACAP38 (11,12,20,45). VIP had not previously been investigated in CH. Optimally, we would have included a third arm – a placebo infusion of saline. However, this would have increased the risk of drop out either from increasing the burden of participation or due to patients with episodic CH changing between disease phases before completing all three study days. Previous experience from CGRP provocation versus placebo of CH (3) in an otherwise identical study design showed remarkably small placebo response to provocation – only one of 32 CH patients had an attack on placebo day. Therefore, to compromise between feasibility and optimal study design, we chose only the two PACAP and VIP study arms. To minimise expectation bias, patients were informed that either infusion could possibly induce headache but the type, onset or duration of headache and whether it would mimic their usual attacks was not disclosed. Another possible limitation was allowing patients to use CH preventive medication. However, in ECHA patients, cluster-like attacks occurred in six of nine (67%) on preventives and in two of five (40%) not on preventives. In CCH patients, cluster-like attacks occurred in five of nine (55%) on preventives and in four of six (67%) not on preventives. Thus, intake of existing preventives does not hinder PACAP38- or VIP-induced attacks. The major strengths of our study are our experimental human model, large sample size of 44 CH patients and robust crossover design.
In conclusion, the present study demonstrated for the first time that CH patients in active phase (i.e. ECHA and CCH) are susceptible to cluster-like attack induction by PACAP38 and VIP, although with lower attack induction rates than by CGRP provocation of CH (3). Unlike in migraine, VIP induced an attack incidence similar to PACAP38, which speaks against the PAC1-receptor as being solely responsible for cluster-like attack induction. Finally, our results agree with CGRP provocation of CH in that attack induction was not possible in ECHR patients. Following the failure of a PAC1-receptor antibody in a Phase II migraine prevention trial and the recent development of a PACAP antibody for testing in migraine prevention, our findings support the rationale of testing a PACAP antibody in CH patients.
Article highlights
PACAP38 provokes cluster-like attacks in episodic active phase and chronic cluster headache and not in episodic cluster headache remission phase. Induction rates were lower than expected and comparable to the hypothesised “active placebo” drug also used, VIP. Cluster-like attack initiation by PACAP38 elucidates the rationale of extending to cluster headache testing of an anti-PACAP monoclonal antibody currently under development for migraine prevention.
Footnotes
Acknowledgements
We would like to thank laboratory assistants Lene Elkjær and Winnie Grønning for assistance with supplies and equipment and MD PhD student Lanfranco Pellesi for help with blood samples.
Declaration of conflicting interests
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: ALHV, AS, BC, ASP, AHJ and RHJ report no competing interests. JH has consulted for and/or served on advisory boards for Allergan, Autonomic Technologies Inc. (ATI), Chordate Medical AB, Eli Lilly, Hormosan Pharma, Novartis and Teva. He has received honoraria for speaking from Allergan, Autonomic Technologies Inc. (ATI), Chordate Medical AB, Novartis and Teva. He also received honoraria from Sage Publishing and Quintessence Publishing for serving on editorial boards and reviewing manuscripts and from Springer Healthcare for writing publications. JH receives research support from Celgene, the Migraine Trust and the Medical Research Council (MRC). These activities have no relation to the present article. He serves as associate editor for
Ethics or Institutional Review Board Approval
All participants provided written consent to participate after receiving written and oral information in accordance with the Declaration of Helsinki of 1964, with subsequent revisions. The study was approved by the Ethics Committee of the Capital Region of Denmark (H-17011569). The study was registered retrospectively at ClinicalTrials.gov (identifier NCT03814226).
Funding
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: The study was funded by grants from Lundbeckfonden (grant R252-2017-1317) and Research Foundation of Rigshospitalet.
