Algometric assessment of sensitization and habituation as predictive markers for OnabotulinumtoxinA response in chronic migraine

BACKGROUND: Chronic migraine is a debilitating disorder characterized by central sensitization and impaired habituation. Although OnabotulinumtoxinA (OnabotA) is an established preventive treatment, its precise mechanism of action and predictors of therapeutic response remain elusive. In this prospective cohort study, we integrated clinical parameters with algometric assessments, a recognized tool for evaluating peripheral sensitization, to develop machine learning-based predictive models to forecast migraine treatment response and effectiveness. METHODS: Seventy-six patients underwent comprehensive clinical evaluations and pressure pain threshold measurements. Key clinical and algometric variables were identified using a feature selection algorithm and incorporated into linear regression models for continuous outcomes (reductions in migraine days, headache days, symptomatic medication use, and triptan use) and an ordinal logistic regression model for categorical treatment response considering percentage of decrease in headache days (three classes: Low (<50%), moderate (50-75%), and high (>75%) responders). RESULTS: The linear regression models yielded significant correlations, exemplified by a reduction in migraine days (ρ = 0.762, RMSE = 5.70) and symptomatic medication days (ρ = 0.503, RMSE = 7.84). Similarly, the ordinal logistic regression model achieved an overall accuracy of 56.6% to predict treatment effectiveness (κ = 0.334 for three classes), as well as 71.1%, 64.5%, and 77.6% accuracies to predict high, moderate, and low OnabotA response. Notably, incorporating algometric data significantly enhanced predictive performance compared to models based solely on clinical variables. CONCLUSIONS: These findings support the potential of combined clinical and algometric evaluation as a practical biomarker for optimizing OnabotA therapy in chronic migraine, warranting further validation in larger, multicenter studies.