Phase Specic Pain Localization in Cluster Headache Patients.

Background and Objective: Applying local treatments like neuromodulation or injections for cluster headache, requires exact knowledge of the anatomical structures and pain topography. However studies with emphasis on exact pain localization are rare although local treatments are increasingly used for patients in whom systemic pharmacotherapy is ineffective or contraindicated. Here, survey results with emphasis on exact pain location in cluster headache attacks for onset of pain, peak pain and radiation of pain, are presented. Methods: Data from 631 respondents were collected for 23 months using an online survey composed of 117 questions on pain location, epidemiology, and clinical features. 5260 datapoints on 44 pain locations were analyzed.

Differences in the patterns of pain location might help identify further target structures for (local) treatments.
Patient satisfaction with available treatments is low.

Background And Aim
Although cluster headache has been known for many years, major CH surveys with emphasis on pain localization are rare (1)(2)(3)(4). Results from previous international cross-sectional surveys on CH describe demography (2,5,6), clinical characteristics including symptoms (2,4,6,7), and diagnostic and therapeutic challenges (8)(9)(10). Current pharmaceutical management of CH(11) − (12) including preventive treatment involves substances like triptans, prednisolone, lithium, verapamil and others which, especially when used by patients in high doses over long periods of time, have the potential for side effects like cardiac events (13), fatigue, nausea, tremor, depression, increased serum glucose and others (12).
Neuromodulatory approaches like non-invasive vagus nerve stimulations (14), Sphenopalatine Ganglion Stimulation (15), Percutaneous Bioelectric Current Stimulation (16) or Greater Occipital Nerve blocks (17) are increasingly used for patients in whom systemic pharmacotherapy is ineffective or contraindicated (18). Applying neuromodulatory treatments requires exact knowledge of the anatomical structures and pain topography relevant for CH. However, few surveys evaluate exact topographic pain locations in CH patients. This article highlights frequencies and distributions of 44 pain locations during the onset, peak and radiation phases of cluster headache attacks in order to analyze phase-speci c pain locations. This is to contribute to a better understanding of pain topography of CH in order to identify further targets and helping to improve effectiveness for local treatments.

Methods And Materials
The data were collected from October 2015 to August 2017 using a German online cross-sectional survey.
Respondents were invited to the survey by email from German headache centers and by links placed in online self-help groups.
The questionnaire was developed by the authors with the participation of patients from a CH self-help forum. The survey was composed of 117 questions of different types: multiple choice, multi-select, ll-in questions, drop-down selections, numeric rating scales and image maps (see Additional le 1). In addition to pain localization, data on epidemiological aspects, diagnosis, clinical features and treatment were collected.

Collection of pain location data
Patients were asked to choose pain locations via multiple response questions for onset ("which area hurts rst?"), peak ("during the attack: where is the most severe pain?") and radiation ("where does the pain radiate to?").
In order to achieve a high resolution of the pain topography, patients had to select up to 5 of 44 points per phase distributed on a realistic representation of the ventral and dorsal head including neck and shoulder region (Fig. 1).
These points were de ned by consensus among the authors and involved feedback from CH patients. Selection criteria were the authors' clinical experience, peripheral sensory innervation of the head including neck and shoulder regions (Nn.V1-V3, C2-C4), and known localization points for occipital nerve blocks (17), botulinumtoxin injections (19) and acupuncture points for trigeminal headache (20).

Inclusion criteria
Not all survey questions had to be answered mandatorily. However, incomplete questionnaires were excluded. Further inclusion criteria consisted of gender, patients > 18 years old and physician-con rmed cluster headache diagnosis date (Fig. 2).

Duplicates and data integrity
Respondents were only able to complete one survey. Duplicate entries were technically prevented. The authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Ethics committee vote
The cross-sectional survey was anonymized; therefore the investigation was released by the responsible authority without the need for an ethics vote (Ethics Committee of the North Rhine Medical Association,

Pain localization during stages of CH attacks
For pain location, patients could select up to 5 points; however, mean responses per phase  Most pain locations at onset (as well as in peak pain) are reported periorbital. However, during onset, occipital (point 40) and nuchal (point 35) pain locations differ signi cantly in frequency compared to during peak pain and radiation. (Fig. 4).

General observations on pain location
Retroorbital pain is reported by 86% (n: 545/631) of patients. Periorbital pain localization can be further discriminated into the medial (5) and lateral (19) corners of the eye as well as supra- (11,18) and infraorbital (13,20) regions.
Patients reported fewer locations and less peri-orbital involvement during pain radiation (n: 1418) (Fig. 7) compared to onset (n: 1893) and peak pain (n: 1949), but dorsal pain locations (points 32-44) were reported more frequently during pain radiation and onset versus peak pain.

Discussion
The survey is the largest conducted in Germany among cluster headache patients to date and the only article describing phase-speci c pain dynamics in CH patients. Comparable survey-based studies with a different focus have been conducted in the US (2), Germany(4,21) France(1), Denmark (7,22) and the Netherlands (8), but this study on pain location patterns is novel as it highlights differences and commonalities with respect to frequencies and distribution of pain locations during onset, peak pain and pain radiation. Different pain locations can be observed during pain onset and peak of a cluster attack.
pain is mainly present during onset and radiation phases of an attack and less frequent during the peak phase. During onset and peak there is a peri-orbital pain focus that can be further discriminated into medial and lateral corners the eye as well as supra-and infraorbital regions.
Data on demographics and epidemiology (male to female ratio, eCH to cCH ratio, diagnostic delay) as well as clinical characteristics (pain features and accompanying. vegetative symptoms) are in line with results of other published CH surveys. It has to be taken into account that adverting the survey in headache centers and among the members of the self-help groups might explain the high proportion of chronic cluster headache patients in terms of a selection bias.
Furthermore, patients report a dissatisfaction with current management of cluster headache especially with neuromodulatory methods.
We were able to investigate further pain locations in addition to the locations described by diagnostic criteria (severe or very severe unilateral orbital, supraorbital and/or temporal) in the IHS ICHD-3 classi cation (23). Our ndings of pain location are in line with previous publications (3,4); however, by explicitly asking patients to discriminate between onset, peak and radiation, we were able to show differences on a per-point basis (pain locations 1-44) thus making it possible to resolve those pain regions into more detail. This might have been overlooked in the past because peri-orbital pain is so dominant that other pain locations were not mentioned if they were not speci cally asked for.
Patients report a simultaneous occurrence or radiation of ventral (Nn. V1-V3, locations 1-31) and dorsal pain (dorsal: area of distribution from Nn. C1-C4, locations 32-44), which is also described in cervicogenic headache (24). The fact that the cervical and occipital innervation is connected to the trigeminal nerve system through various structures (e.g. trigeminocervical complex(25)) supports our ndings on ventral and dorsal pain from a neurophysiological perspective and may help in identifying further target structures for treatment.
In recent years, along with pharmacotherapy, local stimulation procedures have evolved (26). While topographic pain patterns may be of secondary signi cance for drug treatment, the described pain patterns and locations might serve as possible targets for non-pharmaceutical approaches especially when the CH is drug-resistant. Given our ndings, it could be important or even mandatory to carefully examine the exact pain locations before selecting the appropriate neuromodulating approach. For example, a cluster headache that develops in the dorsal parts of the head or the occipital/nuchal area might respond differently to locally applied treatments (like electrical stimulation or nerve blockades) than a CH that starts from periorbital areas. For the dorsal cluster, a local nerve block or implantation of an ONS stimulator might be more promising, whereas in a cluster exclusively around the eye, stimulation of the pterygopalatine ganglion (SPG) (27) might be more suitable.

Page 11/18
CH is a disease with a low prevalence compared to other forms of headache. To reach as many patients as possible, a survey conducted online is well accepted. In respect to demographics and major clinical features, our ndings are in line with other published studies, thereby con rming the external validity of our data. As the number of questions increased, the number of answers decreased for some questions.
This may have distorted the results, especially in the areas of data on patient satisfaction with current management of CH. While all cases were reportedly diagnosed by a doctor, the method does not allow to verify whether cluster headache or another form of (trigeminal autonomic) headache was present. We considered IHS ICHD-3 criteria but could not validate the diagnostic quality of the questionnaire. In order to minimize recall/information bias for questions related to the past, we explicitly asked the patients to relate those questions only to observations of the last two years Conclusions Analysis of the pain location data shows phase speci c frequencies and distributions of pain location during the three stages of a cluster headache attack. Single pain spots differ signi cantly in frequency during the three attack phases. Dorsal pain is more frequent during onset and radiation, compared to peak. Extra-orbital pain locations are more frequent during pain radiation. These ndings will help to better understand CH and might help to identify further target structures for local treatments.

Declarations
Ethics approval and consent to participate The data was collected anonymously. For this reason, according to the Ethics Committee of the North Rhine Medical Association, Düsseldorf, Germany (application 272/2015), no ethics vote was necessary, and the study was therefore exempt from oversight.
Within the survey, patients gave us online consent to process their data.

Consent
Availability of data and materials The datasets supporting the conclusions of this article are included within the article and its additional les.

Competing interests
Philipp Schröder is shareholder of the medical device company "Columbus Health Products GmbH", Germany.
Within the past three years, Charly Gaul has received honoraria from Allergan, TEVA-Ratiopharm, Boehringer Ingelheim, Lilly, Novartis, Desitin Arzneimittel, Cerbotec, Bayer Vital, Hormosan, Grunenthal, Reckitt Benckiser. He has no stocks or ownership interests in any pharmaceutical or medical device companies.
Albrecht Molsberger is CEO and shareholder of the medical device company "Columbus Health Products GmbH", Germany.

Funding
The author(s) received no nancial support for the research, authorship, and/or publication of this article.
Authors' contributions PS conducted the data analysis, interpretation and visualization of the data. The questionnaire was developed by PS, AM and CG. AD contributed to the statistical analysis. AM and CG supervised the survey. All authors were involved in survey development, and in discussing, writing and approving the nal version.  Reported frequencies per point for onset, peak and radiation pain Figure 4 Bubble chart for onset-pain location frequencies. Points 1-44 are mapped to X,Y coordinates of the chart.
Size of bubble = Z-value: diameter of bubble corresponds to pain location frequency [percentage, Table   Page 17/18 1]. Bubbles are sized proportionally.

Figure 5
Bubble chart for peak-pain location frequencies. Points 1-44 are mapped to X,Y coordinates of the chart. Size of bubble = Z-value: diameter of bubble corresponds to pain location frequency [percentage, Table   1]. Bubbles are sized proportionally.

Figure 6
Bubble chart for radiation-pain location frequencies. Points 1-44 are mapped to X,Y coordinates of the chart. Size of bubble = Z-value: diameter of bubble corresponds to pain location frequency [percentage,