Supplementary Material for: Mediation analysis of inflammatory biomarkers in the association between migraine and stroke among postmenopausal women: the Women’s Health Initiative

Introduction: The mechanisms linking migraine to stroke are unclear. Systemic inflammation may contribute, but the mediating role of inflammatory biomarkers is unknown. We evaluated whether systemic inflammatory biomarkers mediate the migraine-stroke association in postmenopausal women.
Methods We analyzed data from 147,730 postmenopausal women without baseline stroke in the Women’s Health Initiative, a large U.S. prospective cohort of clinical trials and an observational study. Baseline levels of C-reactive protein (CRP; n = 43,754), tumor necrosis factor-alpha (TNF-α; n = 10,610), and interleukin-6 (IL-6; n = 19,637) were assessed as potential mediators. Stabilized inverse probability weighting addressed biomarker subsampling. We used Cox proportional hazards models to estimate associations of migraine history and stroke and of quartile-categorized biomarkers and stroke, linear regression models to estimate associations between migraine history and log-transformed biomarkers, and spline-based Cox models assessed stroke risk across continuous biomarker levels.
Results Migraine history was associated with increased risks of total stroke (adjusted HR [aHR]: 1.10; 95% CI: 1.02–1.19) and ischemic stroke (aHR: 1.14; 95% CI: 1.05–1.25), but not hemorrhagic stroke. Higher CRP and IL-6 levels were consistently related to increased total and ischemic stroke risks. Compared with the lowest quartile (Q1, log-transformed), aHRs (95% CIs) for total stroke across Q2–Q4 were 1.11 (1.00–1.22), 1.18 (1.07–1.30), and 1.50 (1.35–1.65) for CRP, and 2.23 (1.87–2.65), 2.63 (2.22–3.12), and 2.31 (1.94–2.76) for IL-6. In contrast, TNF-α showed a different pattern, with an elevated risk at moderate levels and attenuation at higher concentrations. None of the biomarkers were associated with hemorrhagic stroke. No significant associations were observed between migraine history and any of the three biomarkers. The estimated indirect effects of migraine on ischemic stroke through CRP, TNF-α, and IL-6 were small and not statistically significant (all p > 0.05). The proportions mediated were 0.65% for CRP, –2.82% for TNF-α, and 3.98% for IL-6, indicating minimal contribution of these biomarkers to the migraine–stroke association.
Conclusion Inflammatory biomarkers did not serve as a mediator in the relationship between migraine and stroke in postmenopausal women, suggesting that the association is likely driven by other biological pathways.