AUTHOR=Yalinay Dikmen Pinar , Ari Cagla , Sahin Erdi , Ertas Mustafa , Mayda Domac Fusun , Ilgaz Aydinlar Elif , Sahin Aysenur , Ozge Aynur , Ozguner Hilal , Karadas Omer , Shafiyev Javid , Vuralli Doga , Aktan Cile , Oguz-Akarsu Emel , Karli Necdet , Zarifoglu Mehmet , Bolay Hayrunisa , Ekizoglu Esme , Kocasoy Orhan Elif , Tasdelen Bahar , Baykan Betul TITLE=Cluster Analysis Revealed Two Hidden Phenotypes of Cluster Headache JOURNAL=Frontiers in Neurology VOLUME=13 YEAR=2022 URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2022.898022 DOI=10.3389/fneur.2022.898022 ISSN=1664-2295 ABSTRACT=Objective

To investigate the possible subgroups of patients with Cluster Headache (CH) by using K-means cluster analysis.

Methods

A total of 209 individuals (mean (SD) age: 39.8 (11.3) years), diagnosed with CH by headache experts, participated in this cross-sectional multi-center study. All patients completed a semi-structured survey either face to face, preferably, or through phone interviews with a physician. The survey was composed of questions that addressed sociodemographic characteristics as well as detailed clinical features and treatment experiences.

Results

Cluster analysis revealed two subgroups. Cluster one patients (n = 81) had younger age at diagnosis (31.04 (9.68) vs. 35.05 (11.02) years; p = 0.009), a higher number of autonomic symptoms (3.28 (1.16) vs. 1.99(0.95); p < 0.001), and showed a better response to triptans (50.00% vs. 28.00; p < 0.001) during attacks, compared with the cluster two subgroup (n = 122). Cluster two patients had higher rates of current smoking (76.0 vs. 33.0%; p=0.002), higher rates of smoking at diagnosis (78.0 vs. 32.0%; p=0.006), higher rates of parental smoking/tobacco exposure during childhood (72.0 vs. 33.0%; p = 0.010), longer duration of attacks with (44.21 (34.44) min. vs. 34.51 (24.97) min; p=0.005) and without (97.50 (63.58) min. vs. (83.95 (49.07) min; p = 0.035) treatment and higher rates of emergency department visits in the last year (81.0 vs. 26.0%; p< 0.001).

Conclusions

Cluster one and cluster two patients had different phenotypic features, possibly indicating different underlying genetic mechanisms. The cluster 1 phenotype may suggest a genetic or biology-based etiology, whereas the cluster two phenotype may be related to epigenetic mechanisms. Toxic exposure to cigarettes, either personally or secondarily, seems to be an important factor in the cluster two subgroup, inducing drug resistance and longer attacks. We need more studies to elaborate the causal relationship and the missing links of neurobiological pathways of cigarette smoking regarding the identified distinct phenotypic classes of patients with CH.