Association of a Cyclical Migraine Phenotype With Disease Progression A 1-Year Time Series Analysis

Background and Objectives Longitudinal studies assessing cyclic fluctuations of migraine attacks using time series analysis are scarce. Here, we analyze headache frequency fluctuations over a year in a cohort of patients with migraine, and we then evaluate how this behavior has an effect on clinical evolution. Methods Monthly headache frequency was prospectively collected using an electronic diary. Prognosis after 1 year was calculated as the headache frequency change rate after 12 months (HCR-M12) as a dependent variable. Monthly headache time series was decomposed into all the possible sum of sinusoids through a fast Fourier transform (FFT) algorithm, and the frequencies with the highest power were used to define the patient's cyclic phenotype during 1 year (patient's number of cycles per year, c/y). Patients with a cyclic phenotype were those with >2 c/y. Finally, we studied how this cyclic phenotype was associated with HCR-M12 using generalized linear models (GLMs). Results A total of 142 patients were included (85.2% female; mean age 48.0 +/- 9.7 years), 50.0% fulfilled the International Classification of Headache Disorders, third edition, criteria for chronic migraine (CM). After 1 year, a 50.7% (72/142) of patients changed their initial diagnosis, and progression (frequency worsening) was observed in 14.1% (10/71) of patients with episodic migraine (EM). After applying an FFT, 45.1% (64/142) of patients fitted into a cyclic phenotype. In GLM, statistically significant main effects associated with HCR-M12 were the use of preventive therapy (beta [SE]: 74.1 [34.6]; p = 0.034) and cyclic phenotype (beta [SE]: 158.33 [55.1]; p = 0.005). A post hoc analysis found that patients with EM with cyclic phenotype without adequate preventive therapy were statistically significantly associated with progression. Discussion Monthly headache frequency data can be fitted by sinusoidal models. Having a cyclic phenotype has an effect on clinical evolution and has been statistically significantly associated with migraine progression after 1 year. Particularly, patients with EM with cyclic phenotype tend to increase their headache frequency over time. Preventive treatment seems to play a fundamental role in modulating this cyclic behavior, especially in patients with low-frequency EM.

Automated summary

This study analyzed cyclic fluctuations of migraine attacks using time series analysis and found that patients with a cyclic phenotype have a significant impact on clinical evolution. Particularly, patients with episodic migraine and cyclic phenotype tend to increase their headache frequency over time.