Clinical features of cluster headache without cranial autonomic symptoms: results from a prospective multicentre study

Although cranial autonomic symptoms (CAS) are typical in cluster headache (CH), some individuals with CH show no CAS during their headache attacks. Probable cluster headache (PCH) is a subtype of CH that fulfils all but one criterion of CH. This study aimed to investigate the frequency and clinical features of CH and PCH without CAS in comparison to those with CAS. We analysed data from the Korea Cluster Headache Registry, a prospective multicentre registry involving data from 16 hospitals. Of the 216 participants with CH and 26 with PCH, 19 (8.8%) and 7 (26.9%), respectively, did not have CAS. Participants with CH without CAS exhibited less severe anxiety (General Anxiety Disorder-7 score, median [interquartile range], 2.0 [1.0–6.0] vs 8.0 [3.0–12.0], p = 0.001) and depression (Patient Health Questionnaire-9 score, 3.0 [1.0–7.0] vs 7.0 [3.0–11.0], p = 0.042) than those with CAS. Among participants with PCH, headache intensity was less severe in participants without CAS than in those with CAS (numeric rating scale, 8.0 [7.0–8.0] vs 9.5 [8.0–10.0], p = 0.015). In conclusion, a significant proportion of participants with CH and PCH did not have CAS. Some clinical features of CH and PCH differed based on the presence of CAS.

www.nature.com/scientificreports/ was reported to account for approximately 10-20% of CH cases 2,4,5 . Some clinical features of PCH differ from those of CH. Individuals with PCH have a lower incidence of conjunctival injection and forehead sweating than those with CH, but these groups have showed comparable disability 5,6 . Although CAS have been recognised as typical symptoms of CH, 3-7% of individuals with CH were noted to never experience CAS during their CH attacks [7][8][9] . The frequency and clinical features of CH without CAS have been reported in only one instance. A Portuguese study from a single university hospital reported in 2005 that headache intensity was less severe in individuals with CH without CAS than in those with CAS 7 . However, since this study did not distinguish between CH and PCH in its analysis, the frequency and clinical features of CH and PCH without CAS compared to those with CAS remain unclear. Furthermore, this study used data from a single hospital; thus, an additional analysis using data from various settings is needed to validate these findings.
This study aimed to assess (1) the frequencies of CH and PCH without CAS among participants with CH and PCH and (2) the differences in the clinical features of participants with CH and PCH with and without CAS. For this purpose, we used data from the Korean Cluster Headache Registry (KCHR), a prospective, multicentre registry of CH.

Methods
Study design and participants. This descriptive, cross-sectional study aimed to investigate the frequency and clinical features of CH and PCH without CAS using data from the KCHR. The KCHR enrolled consecutive participants with CH aged ≥ 19 years from 16 hospitals (14 university hospitals and two secondary referral general hospitals) in Korea. Participants were enrolled between September 2016 and December 2018.
Detailed information about the KCHR has been described previously 5,10 . In the KCHR, participants fulfilling the available diagnostic criteria for CH and PCH, the third edition beta version of the International Classification of Headache Disorders (ICHD-3 beta), at the time it was created were enrolled 11 . For the present study, we included participants fulfilling the diagnostic criteria for CH and PCH based on ICHD-3 2 . A flow diagram of the participant selection process is presented in Fig. 1.
The KCHR collected data on the following parameters for all participants: sex, age at onset of CH, height, weight, headache intensity on numeric rating scale (from 0 to 10), CH attack frequency per day, mean CH duration since the first cluster period, cluster period duration during the ictal period, total number of cluster periods, smoking status, impact of headache (Headache Impact Test-6 score), circadian and circannual rhythmicity of headache attacks, quality of life (the 3-level version of EuroQol five-dimension scale [EQ-5D-3L]), anxiety (Generalized Anxiety Disorder [GAD-7] score), and depression (Patient's Health Questionnaire-9 [PHQ-9] score) 12,13 . Participants with a GAD-7 score of ≥ 10 were classified as having anxiety, while participants with a PHQ-9 score of ≥ 10 were classified as having depression 12,14 . The previously validated Korean versions of PHQ-9, GAD-7 and EQ-5D-3L were used [14][15][16] . The cluster period was designated as CH attacks occurring in a series for weeks to months, separated by remission periods, usually lasting months or years. The cluster period duration was defined as the duration between the first and last days of each cluster period. The total number of cluster  www.nature.com/scientificreports/ periods was defined as the total number of cluster periods diagnosed as CH. Migraine history was evaluated in all participants. Investigators assessed the circadian and circannual rhythmicity of headache attacks by asking participants whether the CH attacks in cluster periods tended to occur at the same time of the day and year, respectively.

Diagnosis of CH and PCH.
The diagnoses of CH and PCH were based on the ICHD-3 criteria 2 . We further classified participants into one of the following groups based on the cluster period duration: no definitive period, episodic, or chronic. Participants whose first episode of headache attacks fulfilled the CH or PCH criteria were classified, respectively, as having CH or PCH without a definitive period. Participants whose CH or PCH attacks lasted at least 1 year without a remission period or with remissions lasting < 3 months were classified as having chronic CH (CCH) or PCH, respectively. Participants whose CH or PCH attacks occurred in cluster periods, in whom two or more cluster periods lasted from 7 days to 1 year when untreated, and in whom cluster periods were separated by pain-free remission periods of ≥ 3 months were classified as having episodic CH (ECH) or episodic PCH.

Case definitions of CH and PCH without CAS.
According to the ICHD-3, we evaluated CAS in participants with CH and PCH based on the presence of the following conditions: (1) conjunctival injection and/or lacrimation, (2) nasal congestion and/or rhinorrhoea, (3) eyelid oedema, (4) forehead and facial sweating, and (5) miosis and/or ptosis. Since our study started in September 2016, when only the ICHD-3 beta was available, we also assessed the presence of (1) a sensation of fullness in the ear and (2) forehead and facial flushing, which were included as CAS in the ICHD-3 beta but not in the ICHD-3 2,11 .
Ethical considerations. The

Participants.
A total of 251 participants were enrolled. One participant was excluded for not fulfilling the ICHD-3 criteria, and eight were excluded due to incomplete data. Finally, 216 participants with CH and 26 participants with PCH were included in this study (Fig. 1). The reasons for a diagnosis of PCH were as follows: an atypical attack duration (n = 8, 30.7%), where the duration was longer (n = 6) or shorter (n = 1) than typical; missing accompanying symptoms (n = 6, 23.1%); < 5 total number of attacks (n = 5, 19.2%); < 1 attack every other day (n = 3, 11.5%); not a severe headache intensity (n = 2, 7.7%); and a bilateral pain location (n = 2, 7.7%). None of the participants with CH or PCH without CAS had a sensation of fullness in the ear or forehead and facial flushing, which were included as CAS in ICHD-3 beta 11 . The frequencies of CAS in participants with CH and PCH are summarised in Table 1. Conjunctival injection and/or lacrimation was the most common CAS in participants with CH, followed by nasal congestion and/or rhinorrhoea. Sensation of fullness in the ear was the least common CAS, and ptosis or miosis was the second less common CAS. Similar patterns were seen in participants with PCH, with conjunctival injection and/or lacrimation being the most common CAS and ptosis or miosis being the least common.  Table 2. The headache intensity was less severe in participants without CAS than in those with CAS. Anxiety (GAD-7 score ≥ 10) and depression (PHQ-9 score ≥ 10) were less prevalent in participants without CAS than in those with CAS. All other features were similar in participants with and without CAS.

Clinical features of participants with CH based on the presence of CAS.
Of the 216 participants with CH, 19 (8.8%) did not have CAS. Anxiety and depression were less severe in participants with CH without CAS than in those with CAS. Other characteristics were not associated with the presence of CAS (Table 3).
Among the 172 participants with ECH, 15 (8.7%) did not have CAS. Anxiety and depression were less prevalent in participants with ECH without CAS than in those with CAS. The average cluster period duration in months was longer in those with ECH without CAS than in those with CAS. Other clinical features were similar between participants with ECH with and without CAS ( Age (

Discussion
The main findings of the present study were as follows: (1) Approximately one-eleventh of participants with CH and a quarter of those with PCH did not have CAS; (2) Anxiety and depression were less severe in participants with CH without CAS than in those with CAS; and (3) Headache intensity was milder in participants with PCH without CAS than in those with CAS. Other clinical features of CH and PCH did not differ between participants with and without CAS. CH has been characterised by recurrent attacks of severe unilateral headache and ipsilateral CAS 1,17 . CAS were considered a key characteristic of CH and included as a diagnostic criterion since the publishing of the first edition of the ICHD 8,18 . Nevertheless, it has been reported that some individuals with CH-like headaches did not experience CAS during headache attacks 9,19 . The second edition of the ICHD, published in 2004, included a sense of restlessness or agitation as an accompanying symptom of CH, in addition to CAS, and a diagnosis of CH without CAS became possible 4 . The subsequent editions of diagnostic criteria for CH retained restlessness or agitation as an accompanying symptom along with CAS 2,11 . Therefore, CH and PCH without CAS are currently included in the ICHD-3, and our study enrolled participants based on these definitions.
We found that 8.8% of participants with CH did not have CAS. This frequency was similar or somewhat higher than that found in previous studies. A case series of 163 patients with CH in Sweden found that 3.1% of the patients did not have CAS 19 , while an Italian clinic-based study of 251 patients with CH found that 2.8% of the patients did not have CAS 8 . Another Italian study found that 7.5% of patients with CH did not have CAS 9 . In a clinic-based study in Portugal, not having CAS was reported in 6.1% of patients with CH or PCH 7 . Possible causes for the discrepancy between the findings from our study and those from previous studies include differences in the diagnostic criteria, ethnicity, and study setting. The abovementioned Swedish study used three criteria for CH diagnosis (World Federation of Neurology, Ekbom, and ICHD-1) 18,20,21 . The two Italian studies used the ICHD-1, which did not include restlessness and agitation as accompanying symptoms. All four studies were conducted in European countries, while our study was conducted in Korea. Individuals with CH in Asian Age (   www.nature.com/scientificreports/ countries showed lower attack frequencies and shorter attack durations than those in Western countries 22 . All four previous studies used data from a single hospital, while the present study used data from the KCHR, which contained data from 16 hospitals. It has been consistently reported that individuals with CH have a higher frequency of anxiety and depression than those with migraine or without headache 10,[23][24][25][26][27] . Additionally, it has been demonstrated that individuals with CCH are more likely to be affected by anxiety and depression than those with ECH 10 . High frequencies of anxiety and depression were observed in the CH and PCH groups in the present study. Furthermore, to the best of our knowledge, this study is the first to report that anxiety and depression were less severe in participants with CH without CAS than in those with CAS. What is the possible mechanism underlying the association of anxiety and depression with CAS? One possible explanation is the role of shared anatomical substrates between the affective symptoms and CH in the pathogenesis of CAS. Neuroimaging findings have revealed an altered metabolism of the pain matrix and hypothalamus in patients with CH and affective disorders [28][29][30] . Therefore, decreased frequencies of anxiety and depression in participants with CH without CAS may be related to the roles of the pain matrix and hypothalamus in the pathogenesis of CAS. This hypothesis could be evaluated by comparing the neuroimaging findings between individuals with CH with and without CAS. Parasympathetic activation mediated by the trigeminal-autonomic reflex has been ascertained as the mechanism for the presentation of CAS in CH 31 . The trigeminal-autonomic reflex is a brainstem connection between the trigeminal nerve and facial cranial parasympathetic nerve outflow, which is activated by the stimulation of the trigeminovascular system 32 . The hypothalamus has been hypothesised to play a role in initiating CH attacks and causing the activation of the trigeminovascular system 33,34 . Once triggered, this system stimulates the trigeminal-autonomic reflex and results in the occurrence of CAS 35 . The present study found that approximately 9% of individuals with CH did not have CAS; these individuals had less anxiety and depression, which are closely related with the hypothalamus, compared to those with CAS. These findings suggest that some individuals with CH may have a weaker signal from the hypothalamus, which causes headache attacks but does not produce CAS. The weaker signal induces less anxiety and depression in individuals with CH. Our findings also suggest that the activation of the trigeminal-autonomic reflex is not an essential part of CH attacks. www.nature.com/scientificreports/ Conjunctival injection and/or lacrimation has been consistently reported to be the most common CAS in individuals with CH 8,[36][37][38] . In the present study, conjunctival injection and/or lacrimation, as the most frequent CAS, was observed in 86.6% of participants with CH. Conjunctival injection and/or lacrimation was the most frequent CAS in Asian as well as Western countries 6,22,39,40 . The frequency of CAS in the present study was similar to that found in previous studies from Asian countries. The frequency of ptosis or miosis in the present study (22.2%) was lower than that found in studies from Western countries. A prospective clinical study in the UK found that 76% of individuals with CH had ptosis 38 . A Danish study reported that ptosis was present in 44.8% of patients with CH 37 . In Asian studies, a lower frequency of ptosis or miosis has been observed-a Japanese study showed a ptosis frequency of 8.1% 6 , while a study at a Chinese tertiary headache centre revealed that ptosis or miosis was present in 16.7% of patients with CH 40 and a Korean multicentre study reported a rate of 8.5% 39 . The similarity in CAS frequencies mentioned above suggests that this parameter was correctly evaluated in the present study.
Our study had some limitations. First, it was conducted as a multicentre prospective study, but it did not represent the whole CH and PCH population. Therefore, our findings should be validated using another dataset to enhance the generalisability. Second, we included relatively small samples of patients with PCH and CCH. Although we tried to enrol all eligible patients with PCH and CCH over the 3-year study period, the number of relevant cases may have been too small for some subgroup analyses. In other words, the statistical power was weakened by the limited sample size. Further studies that include a sufficient number of participants with PCH and CCH are needed to verify our findings. Third, we used the GAD-7 and PHQ-9 to assess anxiety and depression, respectively. However, these instruments only indicate a state of anxiety and depression and cannot be used to confirm the diagnosis. Additional analyses diagnosing anxiety and depression according to the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders will confirm the relationship of anxiety and depression with CAS in CH.
In conclusion, nearly 9% of participants with CH did not have CAS. The prevalence of anxiety and depression was lower in these patients than in those with CAS. Other clinical features were similar between participants with and without CAS. One-quarter of the participants with PCH did not have CAS and headache intensity  www.nature.com/scientificreports/ was less severe in these participants than in those with CAS. The frequency of anxiety and depression did not differ between participants with PCH with and without CAS. Our study proposed that some clinical features of CH and PCH differed based on the presence of CAS. The findings of the present study may help enhance the understanding of the pathophysiology of CH.

Data availability
The data used in the present study are available from the corresponding author on reasonable request.