To identify data-driven migraine phenotypes and compare serum 25-hydroxyvitamin D levels across these phenotypes.
Approach:
Study Design: Retrospective chart review of 95 migraine patients with complete data on various clinical features and serum vitamin D levels.
Cluster Analysis: K-means clustering (k = 4) was applied to sixteen standardized clinical features to derive migraine phenotypes.
Statistical Analysis: Vitamin D levels were compared across clusters using Kruskal–Wallis and ANOVA tests; logistic regression assessed vitamin D's predictive value for severe phenotype membership.
Key Findings:
Four migraine phenotypes were identified: left-sided aura-rich cluster, right-sided moderate cluster, bilateral high-disability cluster, and bilateral lower-disability cluster.
Vitamin D deficiency was prevalent (mean 17.01 ± 9.71 ng/mL; 67.4% below 20 ng/mL) but did not significantly differ across phenotypes.
Female sex was a significant predictor of severe phenotype membership (OR 3.37, 95% CI 1.20–9.46, p = 0.021).
Interpretation:
Clinically interpretable migraine phenotypes can be derived from routine variables, but they represent a descriptive partition rather than discrete biological subtypes. Vitamin D deficiency was common but not significantly different among the identified phenotypes.
Limitations:
The study is retrospective and based on a small sample size.
Vitamin D levels were measured at a single time point, which may not reflect long-term status.
The study did not include non-migraine comparison groups.
Conclusion:
While vitamin D deficiency was highly prevalent among migraine patients, it did not differ significantly across the identified phenotypes, suggesting further investigation is warranted in larger cohorts.
