The material consisted of 893 study participants (Table 1) diagnosed with CH and subtype by a neurologist according to the criteria of International Classification of Headache Disorders (ICHD), 3rd edition (Olesen 2018). Study participants were recruited from throughout Sweden in collaboration with the neurology clinic at Karolinska University Hospital from 2014 to 2022. The study was approved by the Swedish Ethical Review Authority in Stockholm (diary number 2014/656-31/4). Written informed consent was obtained from all study participants. All experiments were conducted in accordance with the declaration of Helsinki adopted by the World Medical Association in regard to human samples. Upon recruitment to our biobank (described in Steinberg et al. 2018), participants were asked to give a blood sample and fill out a questionnaire involving questions regarding disease characteristics, lifestyle, and family history (Fourier et al. 2023). DNA was extracted from whole blood samples using standard protocols.

Study participants were grouped depending on their self-reported triptan usage. Triptan usage was classified as individuals taking one or multiple of the following: sumatriptan injections, sumatriptan nasal spray, or zolmitriptan nasal spray. The triptan tablet group included individuals who took triptans only in the form of tablets: rizatriptan, sumatriptan, zolmitriptan, and/or eletriptan tablet. Triptan tablets are rarely used in treatment of CH due to their slow-acting pharmacological effect (Brandt et al. 2020). To ensure groups were clearly defined, we excluded tablet-only users from the analysis. Triptan non-users were defined as the remaining individuals who did not take triptans in any form but had answered the survey. Nine of the triptan non-users took ergotamine.

TaqMan® Quantitative Real-Time PCR (qPCR) was used to determine the allele frequency of rs5443, rs1024905, rs6724624, and rs2651899 with TaqMan genotyping assays (Online Resource 1, Table S1) and TaqMan Genotyping MasterMix (Thermo Fischer Scientific, Waltham, USA). qPCRs were conducted using a 7500 Fast Real-Time PCR system (Applied Biosystems, Foster City, CA, USA) according to the recommended protocol with slight modifications; 0.5X of SNP assay, 42 PCR cycles for rs1024905 and rs5443. The 7500 software version 2.0.6 was used for allelic discrimination. Genotype data for rs2651899 from 61.4% of the samples were obtained from a previous publication (Ran et al. 2018).

A polymerase chain reaction (PCR) and restriction fragment length polymorphism method (PCR-RFLP) with BcnI cutting as developed by Schürks et al. (Schürks et al. 2014) was used to genotype the 5-HTTLPR indel (rs4795541) and the accompanying SNP, rs25531. PCR was performed using previously published primers for the 5-HTTLPR variant (Ellerbrock et al. 2021), obtained from Thermo Fisher Scientific. The Mastermix was composed of 0.2 μM forward and reverse primers (Thermo), 1x PCR buffer with (NH4)2SO4 (Thermo), 0.2 mM dNTP (Sigma, Saint Louis, USA), 1 mM Mg2+ (Thermo), and 0.5 U Taq DNA Polymerase recombinant (Thermo) in RNAse free H2O. Each reaction contained 1 μl DNA and 24 μl Mastermix.

The PCR reaction was conducted on a PTC-200 Peltier Thermal Cycler, (Conquer Scientific, San Diego, California) with the following cycling conditions also retrieved from Ellerbrock et al. (Ellerbrock et al. 2021) with slight modifications; 95°C for 10 min, 95°C for 30 s, 60°C for 30 s, and 72°C for 5 min, repeat 35x, elongation step at 72°C for 5 min. The PCR products were run on a 3.5% agarose gel (3.5% agarose (Thermo), 0.008% GelRed DNA Stain (Biotium, Fremont, USA)), at 70 V for 150 min using BioRad PowerPac (Thermo). For each sample, 10 μl of the PCR product was combined with 17 μl RNAse free H2O, 2 μl 10x FastDigest Buffer (Thermo) and 1 μl BcnI enzyme (Thermo), incubated at 37°C for 60 min then at 80°C for 20 min, and ran on a 3% agarose gel at 70 V for 120 min to determine the rs25531 genotype (LA = 126 bp, 62 bp, 341 bp; LG = 126 bp, 62 bp, 174 bp, 167 bp; S = 126 bp, 62 bp, 298 bp).

Eighteen samples classified as having a LGS genotype (126 bp, 62 bp, 174 bp, 167 bp, and 298 bp) exhibited an extra band (341bp) when running the PCR gel and were therefore sent for Sanger sequencing to verify the genotype at the KIGene facility (Stockholm, Sweden). An additional nine samples with different genotypes were sequenced as positive controls and to verify the correctness of the results.

Clinical features data were obtained from the surveys filled out by patients when recruited to our biobank. For questions regarding attack frequency, attack duration, period duration, and period frequency, some participants filled in multiple answers. For those instances we kept the answer that was most severe (highest attack frequency, longest period duration, etc.). Missing data was not included in the final percentages or analysis. Response rate for clinical data can be found in Tables 1 and 3. Some quantitative variables were grouped evenly in the questionnaire to facilitate better readability for the participants. Heredity was defined as patients having one or more first, second, or third degree relative with CH.

Statistical analysis was conducted using Rstudio 4.1.1 (RStudio Team 2020) and PLINK 1.90 (Chang et al. 2015). Figure 1 was created in GraphPad Prism 5. Categorical data was presented as percentages and numerical data as mean ± standard deviation. Chi-square analyses and Wilcoxon test were used for statistical analysis of phenotypic data. Genetic association was analyzed using logistic regression under an additive model with sex as a covariate. The control group was defined as individuals taking triptans while the triptan non-users were classified as the case group for the logistic regression analysis since our main interest were factors that could lead to patients not using triptans. A two-tailed P-value of 0.05 was deemed significant. Bonferroni correction was applied for genetic testing.

Sex, chronicity, and heredity based on triptan usage in CH patients. The figure shows an increased usage of triptans in males as compared to female CH patients. There is no difference in CH type (episodic vs chronic) and no difference in percentage of individuals with an affected relative (heredity) between triptan users and triptan non-users

A genetic effector score analysis was conducted using a non-weighted additive model. Effector alleles related to triptan non-usage/non-response were identified in our study, as well as in the literature (Online Resource 1, Table S1) (Schürks et al. 2007a, 2014; Christensen et al. 2016; Cargnin et al. 2019; Papasavva et al. 2020). The effector allele was defined as the allele more common in triptan non-users compared to users. For 5-HTTLPR, the S allele was identified as the effector allele while both LG and LA were considered to be the non-effector alleles in a bi-allelic manner. The reported effector allele was equivalent for all SNPs except rs6724624 (Online Resource 1, Table S1); the major allele of rs6724624, C, was more common in non-users in our study, while the minor allele, G, was more common in triptan non-responders with migraine (Cargnin et al. 2019). Since our study included a substantially larger cohort than the Italian migraine study and considering our focus is CH, we conducted the genetic effector score analysis using C as the effector allele for rs6724624. A cumulative effector score for the five variants was attributed to each individual depending on the number of alleles they carried and compared using logistic regression with the effector score as a continuous variable and sex and age as covariates to account for bias. Individuals with missing genotypes for any of the variants were excluded from the analysis (remaining n = 489).

PS Power and Sample Size Calculation program Version 3.0 (Dupont and Plummer 1990) was used for power analysis. With a sample size of 518 CH patients, and the minor allele frequencies (MAFs) of rs1024905, rs6724624, and rs2651899, reported for Europeans in the 1000 Genomes Project Phase 3 and gnomAD exomes database (rs5443) (Ensembl genome browser 2022b), we have 80% power to detect an association with 0.522 < odds ratio (OR) > 1.885 for rs1024905, with 0.340 < OR > 2.082 for rs6724624, with 0.513 < OR > 1.876 for rs2651899 and with 0.466 < OR > 1.901 for rs5443. For the power calculation of 5-HTTLPR we used the MAF (S allele) from an article genotyping a European American population (Odgerel et al. 2013) which gave 80% power to detect 0.513 < OR > 1.876.