Family History of Cluster Headache

Abstract

IMPORTANCE

Genetic and environmental factors are thought to contribute to cluster headache, and cluster headache can affect multiple members of a family. A thorough understanding of its inheritance is critical to understanding the pathogenesis of this debilitating disease.

OBJECTIVE

To systematically review family history rates and inheritance patterns of cluster headache.

EVIDENCE REVIEW

A systematic review was performed in PubMed, Embase, and Cochrane Library. Search criteria were created by a librarian. Articles published between 1985 and 2016, after the publication date of a large review in 1985, were analyzed independently by 2 neurologists to identify family history rates and pedigrees. Pedigrees were analyzed by a genetic counselor.

FINDINGS

A total of 1995 studies were found (1988 through the search criteria and 7 through other means). Forty articles met inclusion criteria: 22 large cohort studies, 1 twin-based study, and 17 case reports or small case series. Across the 22 large cohort studies, the positive family history rate of cluster headache varied between 0%and 22%, with a median of 8.2%. The largest 5 studies, of 1134, 785, 693, 609, and 500 probands each, had a positive family history in 18.0% (numerator not provided), 5.1% (40 of 785 cases), 10.0% (numerator not provided), 2.0% (12 of 609 cases), and 11.2% (56 of 500 cases), respectively. No meta-analysis was performed, given differences in methodologies. Separately, 1 twin-based study examined 37 twin pairs and reported a concordance rate of 5.4% (2 pairs). Finally, 67 pedigrees were identified. Most pedigrees (46 of 67 [69%]) were consistent with an autosomal dominant pattern, but 19 of 67 (28%) were consistent with an autosomal recessive inheritance pattern; 10 pedigrees of probable or atypical cluster headache were identified, and all were consistent with an autosomal dominant inheritance pattern. The sex ratio for cluster headache in identified pedigrees was 1.39 (103:74) in affected men and boys compared with affected women and girls, which is lower than that of the general cluster headache population.

CONCLUSIONS AND RELEVANCE

Cluster headache is an inherited disorder in a subset of families and is associated with multiple hereditary patterns. There is an unexpectedly high preponderance of women and girls with familial cluster headache; genetic subanalyses limited to female participants are necessary to further explore this observation, because these data are otherwise masked by the higher numbers of male participants with cluster headache. Overall, this systematic review supports the notion that familial cluster headache is likely the result of multiple susceptibility genes as well as environmental factors.

Cluster headache is characterized by unilateral pain lasting 15 to 180 minutes that occurs up to 8 times per day, with bouts of headaches lasting for weeks to months punctuated by remission periods. Cluster headaches are associated with restlessness and ipsilateral cranial autonomic features.¹ Pain is severe and located maximally in the orbital, supraorbital, or temporal regions.

There have long been reports of hereditary factors in cluster headache, including physical features, such as facial structure, height, and eye color,^2,3 as well as linkage to human leukocyte antigens.^4–6 More recently, there has been extensive interest in identifying genetic risk factors for cluster headache, with several candidate genes proposed but none consistently verified as causal in multiple patient cohorts.^7,8

Reports of familial cluster headache date back to at least the mid-20th century: a meta-analysis⁹ of the rate of cluster headache between 1947 and 1985 and found 47 first-degree relatives affected in 1182 families. Historically, then, the rate of familial cluster headache was thought to be quite low (47 of 1182 participants [4.1%]), at least in first-degree relatives.

A comprehensive assessment of inheritance in cluster headache would complement recent studies of cluster headache genetics and epigenetics. The objectives of this systematic review are to (1) report on the rate of familial cluster headache and (2) provide the different possible inheritance patterns seen in pedigrees.

Methods

Eligibility Criteria

Inclusion criteria were established a priori and included all study types that reported both (1) a proband of any age with cluster headaches diagnosed by a neurologist and (2) any attempt to document the presence or absence of cluster headache in family members. Exclusion criteria specified non-English language articles, review articles or other articles that used previously reported data, and articles published before 1985. We chose 1985 as the startingdate because an extensive review of family history, including a meta-analysis, has been reported on articles between 1947 and 1985.⁹

Information Sources and Search

An experienced librarian (A.T.) developed searches for PubMed, the Cochrane Library through Wiley, and Elsevier Embase. Guidelines for the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA)¹⁰ were followed. The main search was constructed in PubMed using a combination of medical subject heading (MeSH) terms selected from core documents supplied by the neurologists (M.W.W. and M.J.B.) and through discussions between the authors. In general, search terms focused on (1) family, epidemiology, inheritance, or transmission and (2) cluster headache or trigeminal autonomic cephalalgia (PubMed search terms are in eTable 1 in the Supplement). These terms were then tested for relevancy, and the main search was finalized in PubMed on December 15, 2016. These search terms were then translated into the Cochrane Library on December 17, 2016, and Embase on December 22, 2016.

Study Selection

Screening was performed independently by 2 neurologists (M.W.W. and M.J.B.). In the screening step, we examined titles and abstracts in Rayyan¹¹ using our inclusion and exclusion criteria. In the eligibility step, we examined full-length articles for inclusion and exclusion criteria. At the end of each stage, disagreements were settled by discussion between the 2 neurologists. In all cases, a consensus was reached.

Data Collection Process, Data Items

Two authors (M.W.W. and M.J.B.) independently extracted data elements from each article. Data extracted included the authors’ names, article title, journal name, publication year, country of authors, country of patients, family history rate of cluster headache, number of probands examined, twin study details, and presence of pedigrees. For quality assessment, these 2 authors independently assessed (1) the criteria for the diagnosis of cluster headache, (2) the method of diagnosis of cluster headache in the probands, and (3) the method of diagnosis of cluster headache in family members.

In the first part of the study, the family history rate of cluster headache was examined and defined as the percentage of probands who have a positive family history of cluster headache. In the second part of our study, pedigrees were analyzed. All articles with pedigrees were examined by a certified genetic counselor (K.J.Q.). Cases of probable cluster headache in 1 reference¹² and so-called atypical cluster headache in another¹³ were analyzed in 2 ways: (1) relatives with probable or atypical cluster headache were considered negative for cluster headache and analyzed as if they were unaffected relatives and (2) relatives with probable or atypical cluster headache were considered positive for cluster headache and analyzed as if they were affected relatives. Both analyses are presented here.

Summary Measures and Synthesis of Results

There were 2 summary measures: family history rate of cluster headache and inheritance pattern. No meta-analysis was performed given the significant differences in methods between studies, in particular the following: (1) the degrees of relatives examined (eg, first-degree relatives only vs first-degree and second-degree relatives), and (2) the method of diagnosis in relatives (ie, interviewing relatives vs obtaining a family history only from the proband).

Statistical Measures

Statistics were performed using JASP version 0.11.1 (University of Amsterdam; https://jasp-stats.org/). The Mann-Whitney U test was used to compare the familial cluster headache rates in different study characteristics. For calculations, categories were created as follows. Sources of participants were categorized as (1) clinics alone or (2) clinics and/or other methods. The number of participants examined was categorized as less than 500 or greater than or equal to 500 individuals. The degrees of relatives examined were categorized as first degree only or more than first degree. The criteria used to diagnose cluster headache were grouped by (1) International Classification of Headache Disorders 1 or International Headache Society Ad Hoc committee criteria or (2) other criteria.^14–17 The methods of diagnosis in the probands were categorized as in person alone or in person and/or another method. The methods of diagnosis in relatives were categorized as from proband alone or from proband and/or another method. A P value less than .05 was considered significant.

Results

Our search terms identified a total of 1988 articles prior to deduplication (553 in PubMed, 94 in Cochrane Library, and 1341 in Embase) (Figure). Through a general review of the literature, we found and included 7 additional articles^18–24 that were not identified by our search criteria. Ultimately, 40 articles were identified that met the predetermined eligibility criteria: 22 large cohort studies,^{13,18–22,24–39} 1 twin-based study,⁴⁰ and 17 case reports and case series.^{12,23,41–55} Large cohort studies were defined as studies that examined at least 20 patients with cluster headache for a possible family history of cluster headache, and case reports or case series were defined as studies with fewer than 10 patients with cluster headache; there were no studies that included between 10 and 20 patients. Also, within the final 40 articles, a total of 70 pedigrees were found^{12,13,26,27,38,41,42,55}: 67 with relatives with a full diagnosis of cluster headache and 3 with relatives that had only probable or atypical cluster headache. In part 1 of our analysis, we examined the family history rates of the large cohort studies and the twin-based study. In part 2, we examined the pedigrees.

Figure.

Flow Diagram for Identification, Screening, Eligibility, and Inclusion

Our search criteria appeared to be reasonably inclusive: we identified 20 of 23 articles published after 1984 that were mentioned in 7 previous narrative reviews^9,56–61 on family history of cluster headache. Two of the 3 articles^19,24 that our search failed to identify were added to our analysis, while the third⁶² was excluded because the data were incorporated into the previous meta-analysis from 1947 to 1985 by Russell.⁹

Part 1: Family History Rate

We identified 22 large cohort studies,^{13,18–22,24–39} which examined participants from 14 countries on 3 continents (Asia, Europe, and North America) (Table 1). Most studies (17 of 22 [77%])^{13,18–22,24–29,31–33,36,38} examined at least 100 participants. Recruitment came exclusively from patients seen in clinics in 16 of 22 studies^{13,20–22,24,26,28,29,31–38} (72%), while 5 of 22 studies^{18,19,25,30,39} (25%) recruited participants through other sources, such as national support groups, headache societies, or media postings (in websites, radio, or newspapers). Information on patient recruitment was not available in 1 of 22 studies²⁷ (5%). It should be noted that the largest study,²⁵ which included 1134 participants, was also the only large cohort study that did not use International Classification of Headache Disorders criteria to confirm the diagnosis. Probands were mostly diagnosed exclusively in person (13 of 22 studies^{13,20,24,27–30,32–34,36–38} [59%]), while the remainder may have also been diagnosed by telephone or questionnaire. In most studies ^{18–22,24,25,27,29,30,32,34,36,37,39} a diagnosis of cluster headache in the proban’s relatives was made based on information reported only by the proband (15 of 22 studies [68%]) while in the remainder, relatives were diagnosed in person by telephone or by survey.

Table 1.

Family History Rate of Cluster Headache^a

Source	Year	Countryb	Source of participants	Probands with family history of CH, No. (%)	Total probands examined, No.	Degrees of relatives who were investigated	Criteria used to diagnose CH	Method of diagnosisc
								In probands	In relatives
Large cohort studies
Bahra et al18	2002	United Kingdom	National support group, charity group, clinic	NR (5.0)	230	Part offirst (parent or sibling)	ICHD1	In person or telephone	From probands
Bhargava et al34	2014	India	Clinic	0	30	Unknown	ICHD2	In person	From probands
Cruz et al35	2013	Portugal	Clinic	5 (20.8)d	22	First	ICHD2	Bytelephone	Telephone
Dong et al29	2013	China	Clinic	8 (6.7)	120	Unknown	ICHD2	In person	From probands
Donnet et al36	2007	France	Clinic	6 (5.5)	110	First and second	ICHD1 or ICHD2	In person	From probands
El Amrani et al26	2002	France	Clinic	20 (10.8)	186	First and second	ICHD1	In person or by telephone	In person or by telephone
Haane et al39	2013	The Netherlands	Media (website) and clinic	9 (12.5)	72	First and second	ICHD2e	Survey	From probands
Klapper et al19	2000	Global (online survey)	National support group and media (website)	NR (10.0)	693	Unknown	ICHD1	Survey	From probands
Kudrow and Kudrow27	1994	Not stated, presumed United States	Not stated	NR (8.7)	300	First	IHS Ad Hoc Committee and ICHD1	In person	From probands
Lademann et al20	2015	Germany	Clinic	NR (7.8)	434	Unknown	ICHD2 or ICHD3-beta	In person	From probands
Leone et al33	2001	Italy	Clinic	44 (20.0)	220	First and second	ICHD1	In person	In person, by telephone, or from probands
Lin et al21	2004	Taiwan	Clinic	6 (5.8)	104	Partoffirst degree (parent or sibling)	ICHD1	In person, by telephone, or by survey	From probands
Maytal et al37	1992	United States	Clinic	3 (8.6)	35	Parts of first and second (parent, grandparent, sibling)	ICHD1	In person	From probands
Montagna et al38	1998	Italy	Clinic	5 (2.3)	222	First and second in all families, third in 1 family	ICHD1	In person	Telephone
Rainero et al32	2008	Italy	Clinic	5 (4.6)	109	Unknown	ICHD2	In person	From probands
Riess et al30	1998	Canada	Media (newspaper and radio)	NR (22.0)	51	First	ICHD1	In person	From probands
Rozen and Fishman25	2012	United States	National support group and national headache society	NR (18.0)	1134	Unknown	Diagnosed by neurologist	Survey	From probands
Russell et al31	1996	Denmark	Clinic	25 (6.8)	366	First and second	ICHD1	Bytelephone or survey	Telephone
Sjöstrand et al13	2005	Sweden	Clinic	12 (2.0)	609	Unknown	ICHD2	In person	From probands, questionnaires, and/or personal interviews
Steinberg et al22	2018	Sweden	Clinic	56 (11.2)	500	First, second, and third	ICHD3-beta	Survey	From probands
Taga et al28	2015	Italy	Clinic	40 (5.1)	785	First	ICHD3-beta	In person	Chart reviews
Vikelis and Rapoport24	2016	Greece	Clinic	NR (17.5)	302	Unknown	ICHD3-beta	In person	From probands
Twin-based studies
Ekbom, et al40	2006	Sweden	National registry	2 (5.4)	37	Twins only	ICHD2	Bytelephone	Bytelephone

Abbreviations: CH, cluster headache; ICHD, International Classification of Headache Disorders; IHS, International Headache Society; NR, not reported.

^aReferences for the diagnosis of cluster headache include the ICHD editions 1,⁶³ 2,³⁶ 3-beta,³⁵ and 3¹ and HIS Ad Hoc Committee³⁷ criteria.

^bFor country of patients studied, presumed refers to articles in which the country of the patients was not specified, but the patients were personally seen in the clinic of the authors.

^cIn some cases, the method of diagnosis was not specifically mentioned; patients were presumed to be diagnosed in person if the study mentioned that the patient had a neurologic examination completed or a blood sample drawn.

^dOf note, there is an intentional discrepancy in the value for percentage and the value for nominator and denominator for Cruz et al.³⁵ This study reported a positive family history in 20.8% (5 of 24 participants); however, 2 of the 24 patients had probable cluster headache and thus were excluded, and the study did not report if it was these patients who had a positive family history. Therefore, we report the original percentage of 20.8% (5 of 24 participants) instead of 22.7% (5 of 22 participants).

^eFor criteria used to diagnose cluster headache, this letter indicates that maximum attack duration and maximal attack frequency criteria were not examined.

The family history rate of cluster headache in the 22 large cohort studies varied between 0% to 22%, with a median of 8.2% (the data were not normally distributed; therefore we reported the median). The largest 5 studies of 1134,²⁵ 785,²⁸ 693,¹⁹ 609,¹³ and 500 probands²² each, had a positive family history in 18.0% (numerator not provided), 5.1% (40 of 785 cases), 10.0% (numerator not provided), 2.0% (12 of 609 cases), and 11.2% (56 of500 cases), respectively. There was methodologic variability between studies, and we examined variables that might be important, such as direct in-person interviews, studies at headache centers studies that interviewed more closely or more distantly related family members, studies that interviewed relatives directly, or studies with more than 500 probands (Table 2). Each subgroup had a median family history rate between 6.7% and 10.0% and a range similar to the entire cohort. There were no significant differences between the family history rates of cluster headache based on methodological differences in terms of the source of the patient cohorts, the number of probands examined, the degrees of relatives who were investigated, the criteria used to diagnose cluster headache, the study publication date, the method of diagnosis in the proband, or the method of diagnosis in the relatives (Table 3). Because of multiple differences in methodology between studies no meta-analyses were performed.

Table 2.

Systematic Review Results From Table 1, Grouped by Categories^a

Characteristic	No. of studies	Family history rate of cluster headache, median (range), %
All population studies	22	8.2 (0–22)
Studies
With diagnosis in the proband made exclusively by in-person interview	13	6.7 (0–22)
Performed at headache centersa	16	6.8 (2–22)
Limited to first-degree relatives	6	7.3 (5–22)
With interviewswith relatives instead of relying on proband information	6	8.8 (2.0–20.8)
With inclusion of second-degree and/or third-degree relatives	8	9.7 (2.3–20)
With >500 participants	5	10 (2–18)

^aStudies performed at headache centers.^{13,18,20–22,24,26,28–33,36–38}

Table 3.

Lack of Significant Differences Between Studies Based on Collection Characteristics^a

Category	Comparison	P value
Study publication date	≤2004 vs >2004	.64
Sources of participants	Clinic vs nonclinic	.12
No. of probands examined	<500 vs ≥500	>.99
Degrees of relatives	First vs >first	.68
Criteria used to diagnose cluster headache	International Classification of Headache Disorders 1 vs other	.64
Methods of diagnosis
In probands	In-person vs not in-person	.15
In relatives	Proband recall vs interview	.63

^aComparisons were performed using the Mann-Whitney U test for all categories. A P value less than .05 was considered significant.

In addition to the large cohort studies, we identified 1 population study⁴⁰ that exclusively examined twins using a national registry-which reported a concordance rate of 5.4% (2 of 37 twin pairs). The 2 concordant pairs were both monozygotic, and the 35 discordant pairs consisted of 10 monozygotic and 25dizygotic sets of twins (with the same sex in 13 pairs and different sexes in 12 pairs).

We identified another 17 studies^{12,23,41–55} that examined fewer than 10 probands with cluster headache each (eTable 2 in the Supplement). These studies were primarily case reports or case series and had methods similar to the larger studies. We deemed these studies too small or focused to be included in the aggregate calculation of family history rate of cluster headache. These studies are included here for completeness, should other researchers want to perform an analysis of all studies between 1985 and 2016 with a positive family history of cluster headache identified with our methodology.

Part 2: Pedigrees

A total of 71 pedigrees in 8 studies^{12,13,26,27,38,41,42,55} were identified, with 1 pedigree⁴¹ removed because it was a duplicate of a pedigree from another article.⁴² Some of the studies^12,13 identified patients with probable cluster headache or atypical cluster headache (defined as cases that “did not fulfil the diagnostic criteria for CH [cluster headache] but had clinical symptoms with more resemblance to CH than to migraine or other trigeminal autonomic cephalgia syndromes”^13(p1068)). Atypical cluster headache cases did not meet criteria either because the duration of their headache lasted for longer than 180 minutes or the frequency of headaches was less than every other day. In analysis A, we considered relatives with probable or atypical cluster headache to be unaffected relatives; in this analysis, there were 67 pedigrees.^{12,13,26,27,38,41,42,55} In analysis B (eTable 3 in the Supplement), we reanalyzed the pedigrees considering both relatives with probable or atypical cluster headache and relatives with cluster headache to be affected individuals; in this analysis, there were an additional 3 pedigrees.^12,13

Analysis A, Omitting Probable and Atypical Cluster Headache

Across 67 pedigrees (Table 4),^{12,13,26,27,38,41,42,55} there were 177 affected and 788 unaffected individuals, with a sex ratio (affected men and boys divided by affected women and girls) of 1.39 (103:74). All 67 pedigrees were consistent with autosomal dominant or autosomal recessive patterns of inheritance. Autosomal dominant patterns were found in 46 of 67 pedigrees (69%) and autosomal recessive patterns in 19 of 67 (28%), with insufficient information to distinguish between autosomal dominant or autosomal recessive in the remaining 2 of 67 (3%).

Table 4.

Hereditary Patterns in 67 Pedigrees of Cluster Headache^a

Pedigree	Affected participants, No./total No.
	Individuals	Generations
Autosomal dominant
Bordini et al12
2	2/18	2/4
3	2/10	2/3
El Amrani et al26
1	3/14	3/4
5	3/19	2/4
9	3/12	3/3
Kudrow and Kudrow27
K118	3/4	2/2
K119	3/9	3/4
K120	3/7	3/3
K121	4/7	3/3
K122	4/6	2/2
Sjöstrand et al13
2	2/9	2/3
9	4/17	3/3
Zarrilli et al41
A	2/11	2/3
E	4/12	3/3
G	2/20	2/3
Autosomal dominant with incomplete penetrance
El Amrani et al,26 16	3/14	2/3
Montagna et al38
B	2/18	2/4
E	4/14	3/3
Autosomal dominant or in an X-linked pattern
El Amrani et al26
2	3/17	2/3
4	2/8	2/3
6	2/16	2/3
7	4/11	2/4
8	3/10	2/3
12	2/14	2/3
Kudrow and Kudrow27
K100	2/9	2/3
K101	2/7	2/2
K103	2/5	2/3
K104	2/6	2/3
K105	2/7	2/3
K106	2/9	2/3
K110	2/11	2/3
K111	2/10	2/3
K112	2/7	2/3
K113	2/11	2/3
K114	3/10	3/4
K115	3/8	3/3
K116	3/7	2/3
K117	3/7	3/3
Sjöstrand et al,13 6	2/8	2/3
Autosomal dominant or X-linked with incomplete penetrance
Haan et al55	3/26	2/5
Montagna et al,38 A	3/18	2/5
Sjöstrand et al13
1	3/12	2/3
5	3/13	2/3
Zarrilli et al,41 B	3/22	2/4
Autosomal dominant, X-linked, or polygenic
El Amrani et al,26 3	3/12	2/3
Autosomal dominant or X-linked with incomplete penetrance or polygenic
Kudrow and Kudrow,27 K123	5/22	3/4
Autosomal dominant with incomplete penetrance or autosomal recessive
El Amrani et al,26 20	3/21	1/4
Zarrilli et al,41 F	3/22	1/3
Autosomal recessive
Bordini et al,12 1	2/21	1/3
El Amrani et al26
10	2/11	1/3
11	2/9	1/3
13	2/20	1/4
14	2/13	1/3
15	2/8	1/2
17	2/14	1/3
18	2/15	1/3
19	2/15	1/3
Kudrow and Kudrow27
K102	2/7	1/3
K107	2/4	1/2
K108	2/15	1/3
K109	2/5	1/2
Montagna et al38
C	2/11	1/3
D	2/7	1/3
Sjöstrand et al,13 3	2/10	1/3
Zarrilli et al41
C	2/24	1/4
D	3/18	1/3
Autosomal recessive or polygenic
De Simone et al42	8/151	2/4

^aResults were analyzed for the most likely inheritance pattern, although it should be noted that incomplete penetrance and multifactorial and/or polygenic inheritance patterns cannot be ruled out. In this analysis, we ignored the diagnosis of atypical cluster headache; relatives identified by the original studies as having atypical cluster headache were considered to be unaffected relatives. Numbers in the leftmost column refer to the figure numbers (for Bordini et al¹²) or pedigree number (for El Amrani et al,²⁶ Kudrow and Kudrow,²⁷ and Sjöstrand et al¹³). Letters in the leftmost column refer to pedigree letters (for Zarrilli et al⁴¹ and Montagna et al³⁸)

Of the 46 autosomal dominant pedigrees,^{13,26,27,38,41,55} 28 (61%) had insufficient information to exclude an X-linked pattern of inheritance, because there were no cases of male-to-male transmission of cluster headache within the families. Nine of 46 cases (20%) displayed incomplete penetrance.^{13,26,27,38,41,55} Two of 46 autosomal dominant cases (4%) involved complex pedigrees that could alternatively be explained by polygenic inheritance.^26,27 One pedigree depicts 2 parents with cluster headache whose child was also affected.²⁶ This could represent polygenic inheritance in which each parent carries the same susceptibility gene or different ones. Alternatively, this could be a case of autosomal dominant or X-linked inheritance in which 1 parent represents a sporadic phenocopy within the family. Neither parent had any additional known family history. The second pedigree²⁷ depicts 5 affected individuals from 2 independent families in which not all 5 affected individuals are blood related. Therefore, this family could represent a complex polygenic pattern of inheritance in which susceptibility genes are present in both intermarrying families. Alternatively, 1 or more of these individuals may represent a sporadic phenocopy within the pedigree.

Of the 19 autosomal recessive pedigrees, 1 (5%) involved a complex pedigree that could alternatively be explained by polygenic inheritance.⁴² In this case, the pedigree depicts 8 affected individuals within a kindred group of 4 independent families from the Naples region of southern Italy that had intermarried.⁴² For true autosomal recessive inheritance, all 4 families would be required to carry recessive mutations in the same gene. Such a high carrier frequency may be possible in some local populations. Alternatively, the affected individuals in this pedigree may be the result of multiple shared susceptibility genes, or 1 or more of these individuals may represent a sporadic phenocopy within the family. The remaining 2 families could be consistent with either an autosomal dominant pattern with incomplete penetrance or an autosomal recessive pattern of inheritance. For both of these families, autosomal recessive inheritance would require that 2 siblings each have children with a partner who is a carrier for a recessive mutation in the same causative gene (ie, the recessive gene is carried by 3 independent families).^26,41 Alternatively, an autosomal dominant mutation with incomplete penetrance could also explain the pattern of affected individuals observed in these families.

Analysis B Including Probable or Atypical Cluster Headache

Pedigree analyses were repeated to include cases of probable or atypical cluster headache as affected individuals in these families, thereby including3 pedigrees that were previously excluded (full results in eTable 3 in the Supplement). All 10 pedigrees^12,13 were consistent with autosomal dominant patterns of inheritance. When incorporating these cases into the familial cluster headache pedigrees, 51 of 70 families (73%) had pedigrees consistent with an autosomal dominant pattern of inheritance, while 17 of 70 families (24%) had pedigrees consistent with an autosomal recessive pattern.

Discussion

In a systematic review of studies published between 1985 and 2016, the family history rate of cluster headache varied between 0% and 22%, with a median of 8.2%. When limiting our data to subgroups of studies that performed broader investigations—specifically subgroups with either more than 500 patients, second-degree and third-degree relatives, or direct interviews with relatives (Table 2)—the median was between 8.8% and 10%, with a range of 2%to 20.8%. Thus, the familial rate of cluster headache may be as common as 1 in 10 or even 1 in 5 individuals. In analyzing pedigrees, the most common inheritance pattern was autosomal dominant (69%), followed by autosomal recessive (28%). The inclusion of probable and atypical cluster headache diagnoses into the pedigree analysis did not substantially alter the distribution of inheritance patterns.

When comparing our results with a meta-analysis by Russell,⁹ who found a family history rate of 4.1% (47 of 1182 individuals) for first degree-relatives in studies published between 1947 and 1985, every study of first-degree relatives between 1985 and 2016 found a higher rate of positive family history, ranging from 5% to 22% (Table 2). The difference in family history rate of first-degree relatives between these 2 periods is unclear but may reflect differences in clinical practice, patient awareness, or family dynamics.

The proportion of women with familial cluster headache was higher than expected in our data set. In the pedigrees with a positive family history of cluster headache, the sex ratio (affected men and boys divided by affected women and girls) is 1.39 (103:74), whereas in the general cluster headache literature, it is 4.3.⁶⁴ Some of the large cohort studies from Table 1 had similar sex ratios, of 2.3 (28 men and boys:12 women and girls),²⁸ 2.0 (4:2),²⁹ 2.0 (6:3),³⁰ 1.9 (17:9),³¹ and 1.5 (12:8).²⁶ This difference suggests that factors predisposing men to develop sporadic cluster headache may be different than those that play a role in the pathogenesis of familial cluster headache. Additional investigation is needed to determine potential genetic and environmental factors involved in the sex ratio of cluster headache: the sex ratio in the general cluster headache population has decreased over the decades,⁶⁵ and recent studies of more than 1000 participants with cluster headache have found sex ratios of2.6 (816:318),²⁵ 2.2 (1104:497),⁶⁶ and 1.3 (4356:3233).⁶⁷ There also may be systematic differences in sex between the data collected from researchers examining single families and researchers examining populations. Nevertheless, this finding warrants increased attention in future genetic analyses of cluster headache. As cluster headache is more common in men, data from male study participants have predominated genetic studies thus far and likely mask relevant data from female participants. One potential conclusion of this analysis is that women and girls are more similarly affected by cluster headache compared with men and boys in familial cluster headache. A subset analysis limited to female participants is required to explore why this apparent sex difference exists, and if so, whether inherited factors drive differential susceptibility to cluster headache between different sexes.

There are at least 2 potential explanations for a familial pattern of cluster headache: genetics and shared environmental factors. From a genetics standpoint, this systematic review supports the notion of complex genetic heterogeneity in cluster headache,^7,58,68,69 with the identification of multiple inheritance patterns in pedigrees. Possible susceptibility genes for cluster headache include the hypocretin/orexin receptor type 2 (HCRTR2),^32,70–74 alcohol dehydrogenase 4 (ADH4),^74–77 G protein beta 3 (GNB3),^77,78 pituitary adenylate cyclase-activating polypeptide type I receptor (ADCYAP1R1),^79,80 and membrane met alloendopeptidase (MME) genes.^79,80 We have limited data on genetic anticipation that was reported in 2 studies^13,81; in both studies, there was a statistically significant difference of 9 years between the first generation and the second and/or third generations in 17 parent-child pairs⁸¹ and 29 first-generation relatives with typical and atypical cluster headache vs 25 second-generation and third-generation relatives.¹³ Future studies are needed to examine this genetic aspect of familial cluster headache.

In terms of shared environmental factors, there has been recent interest in the epigenetic mechanisms in cluster headache, with reports on altered gene expression associated with inflammation,^6,82 circadian patterns,^83,84 and exposure to toxins, such as cadmium in tobacco.⁸⁵ All of these factors could be part of a shared environment. Tobacco use is of particular interest because it is seen at a much higher rate in patients with cluster headache than the general population,^25,86,87 and secondhand smoke exposure appears to be common in patients with cluster headache.⁸⁸ Environmental factors alone are unlikely to explain a strong familial aggregation,^59,89,90 as seen in some of the pedigrees, but familial cluster headache could be the result of multifactorial inheritance or the combination of both susceptibility genes as well as environmental factors. Our conclusions were based on 70 individual families; larger studies on cluster headache inheritance patterns are needed to explore this potentially multifactorial inheritance.

Limitations

This study has several limitations. First, in most of the studies, the diagnosis of cluster headache in family members was made indirectly, using information provided by the probands. There is an inherent verification bias, in that individual probands have varying abilities to distinguish cluster headache from other forms of headache in family members. Second, familial cluster headache may be underestimated in many studies because of a variety of factors, including an inability to contact certain family members (death and refusal to participate were both listed as reasons) and the fact that some studies only investigated first-degree relatives. Third, familial cluster headache may be overestimated in some studies because most did not report if their probands could have been related to each other. In 2 of our studies, there were reports of 6 affected patients from 4 families²¹ and 12 affected patients from 5 families¹³; if multiple members of the same family all report a family history of cluster headache, it could overestimate the proportion with a positive family history. Finally, our search criteria failed to find 7 of the final 40 articles that reported a family history of cluster headache; these 7 articles were identified through other sources. Six of these 7 articles mentioned a family history of cluster headaches only in the results section. Because we first screened by title and abstract, it is possible that we missed other articles. A more comprehensive search might be performed in the future if the search is applied not only to the title and abstract but to the entire article.

Conclusions

In this systematic review, cluster headache had a family history rate of 0% to 22%, with a median of 8.2%. The inheritance pattern of cluster headache across 67 pedigrees was consistent with an autosomal dominant pattern in 69% and an autosomal recessive pattern in 28%. Future genetic studies of familial cluster headache should direct attention to sex differences and headache age at headache onset.

Key Points.

Question

What is the family history rate and inheritance pattern of cluster headache?

Findings

In a systematic review of articles published between 1985 and 2016, which included 22 large cohort studies, cluster headache has a family history rate of 0% to 22%, with a median of 8.2%. A total of 67 pedigrees were identified, and the inheritance pattern of cluster headache was consistent with an autosomal dominant pattern in 69% of participants and an autosomal recessive pattern in 28%.

Meaning

Per this systematic review, cluster headache is an inherited disorder in a subset of patients, and the pattern of inheritance could be the result of multiple susceptibility genes as well as environmental factors.