Clinical predictors for efficacy of erenumab for migraine: a Registry for Migraine (REFORM) study

Erenumab has proven effective for migraine prevention; however, a substantial proportion of people with migraine do not benefit from treatment, and among those who do, there is considerable variability in response. This study aimed to identify clinical predictors of therapeutic response to erenumab and evaluate their predictive value in a large cohort of people with migraine. We conducted a single-centre, prospective, longitudinal cohort study of adults with migraine, experiencing ≥4 monthly migraine days. All participants received erenumab 140 mg monthly for 24 weeks and recorded their response in a headache diary with daily entries. A semi-structured interview was conducted at enrolment, and patient-reported outcome measures were collected before and after treatment. Treatment responders were classified as participants achieving a reduction from baseline of ≥50% in average monthly migraine days across weeks 13 through 24. Clinical predictors were analysed using logistic regression analysis. In total, 570 participants with migraine provided data eligible for analysis. Of these, 298 (52.3%) participants were classified as treatment responders, and the remaining 272 (47.7%) were non-responders. Independent predictors associated with a lower likelihood of response to erenumab were chronic migraine (odds ratio 0.63, 95% confidence interval 0.43-0.91; <i>P</i> = 0.030), daily headache (odds ratio 0.41, 95% confidence interval 0.24-0.67; <i>P</i> = 0.003) and previous failure of ≥3 preventive migraine medications (odds ratio 0.54, 95% confidence interval 0.37-0.77; <i>P</i> = 0.005). Conversely, better outcomes were observed with higher age (10-year increase: odds ratio 1.22, 95% confidence interval 1.06-1.41; <i>P</i> = 0.017). Multivariate model area under curve was 64.6% (60.0-69.2%). Participants with an early response to erenumab (≥50% reduction within weeks 1-12) were less likely than late responders to have chronic migraine [119/217 (57.1%) versus 61/79 (77.2%); <i>P</i> < 0.001], had lower Migraine Disability Assessment Scores [median (IQR): 52 (30-85) versus 65 (35-120); <i>P</i> = 0.029], more often had unilateral headache [193 (88.9%) versus 63/79 (79.7%); <i>P</i> = 0.041], and experienced less ictal allodynia measured by Allodynia Symptom Checklist-12 scores [median (IQR): 4 (0-8) versus 6 (2-8) versus; <i>P</i> = 0.024]. In conclusion, chronic migraine, experiencing daily headache, and having ≥3 preventive medication failures were independently associated with a lower likelihood of response to erenumab. Moreover, patients with a more severe clinical phenotype were more likely to respond later. Prediction of treatment responses might be improved by incorporating machine learning models and multimodal biomarkers, facilitating a shift towards personalized medicine.