Cutrer, F. Michael

Background Migraine is a highly prevalent neurological disorder with many treatment options, both pharmacological and non-pharmacological. Artificial intelligence (AI) has great potential to optimize treatment selection strategies for individual patients. This review provides an overview of AI models and the techniques used to predict migraine treatment outcomes. Methods We conducted a literature search in PubMed and examined studies that reported employing AI models to predict migraine preventive and acute treatment outcomes. We also explored incorporating AI/machine learning to enhance personalized migraine treatment strategies, including forecasting migraine attacks. Additionally, we summarized future research directions, including incorporating multimodality data, using AI frameworks for the discovery of novel treatment targets, and advancing the field with innovative AI techniques such as digital twins, conversational AI and virtual AI agents. Results Studies have employed ML and deep learning on a combination of clinical features and imaging data to predict acute or preventive migraine treatment outcomes with reported success. Continued model optimization, validation, and prospective assessment of the clinical utility of deploying ML models in real-world settings are crucial. Conclusions While AI has demonstrated success in predicting migraine treatment responses, future research incorporating novel AI techniques and diverse data sources could pave the way to advance personalized migraine treatment.

Abstract Objective To develop machine learning models using patient and migraine features that can predict treatment responses to commonly used migraine preventive medications. Background Currently, there is no accurate way to predict response to migraine preventive medications, and the standard trial‐and‐error approach is inefficient. Methods In this cohort study, we analyzed data from the Mayo Clinic Headache database prospectively collected from 2001 to December 2023. Adult patients with migraine completed questionnaires during their initial headache consultation to record detailed clinical features and then at each follow‐up to track preventive medication changes and monthly headache days. We included patients treated with at least one of the following migraine preventive medications: topiramate, beta‐blockers (propranolol, metoprolol, atenolol, nadolol, timolol), tricyclic antidepressants (amitriptyline, nortriptyline), verapamil, gabapentin, onabotulinumtoxinA, and calcitonin gene‐related peptide (CGRP) monoclonal antibodies (mAbs) (erenumab, fremanezumab, galcanezumab, eptinezumab). We pre‐trained a deep neural network, “TabNet,” using 145 variables, then employed TabNet‐embedded data to construct prediction models for each medication to predict binary outcomes (responder vs. non‐responder). A treatment responder was defined as having at least a 30% reduction in monthly headache days from baseline. All model performances were evaluated, and metrics were reported in the held‐out test set (train 85%, test 15%). SHapley Additive exPlanations (SHAP) were conducted to determine variable importance. Results Our final analysis included 4260 patients. The responder rate for each medication ranged from 28.7% to 34.9%, and the mean time to treatment outcome for each medication ranged from 151.3 to 209.5 days. The CGRP mAb prediction model achieved a high area under the receiver operating characteristics curve (AUC) of 0.825 (95% confidence interval [CI] 0.726, 0.920) and an accuracy of 0.80 (95% CI 0.70, 0.88). The AUCs of prediction models for beta‐blockers, tricyclic antidepressants, topiramate, verapamil, gabapentin, and onabotulinumtoxinA were: 0.664 (95% CI 0.579, 0.745), 0.611 (95% CI 0.562, 0.682), 0.605 (95% CI 0.520, 0.688), 0.673 (95% CI 0.569, 0.724), 0.628 (0.533, 0.661), and 0.581 (95% CI 0.550, 0.632), respectively. Baseline monthly headache days, age, body mass index (BMI), duration of migraine attacks, responses to previous medication trials, cranial autonomic symptoms, family history of headache, and migraine attack triggers were among the most important variables across all models. A variable could have different contributions; for example, lower BMI predicts responsiveness to CGRP mAbs and beta‐blockers, while higher BMI predicts responsiveness to onabotulinumtoxinA, topiramate, and gabapentin. Conclusion We developed an accurate prediction model for CGRP mAbs treatment response, leveraging detailed migraine features gathered from a headache questionnaire before starting treatment. Employing the same methods, the model performances for other medications were less impressive, though similar to the machine learning models reported in the literature for other diseases. This may be due to CGRP mAbs being migraine‐specific. Incorporating medical comorbidities, genomic, and imaging factors might enhance the model performance. We demonstrated that migraine characteristics are important in predicting treatment responses and identified the most crucial predictors for each of the seven types of preventive medications. Our results suggest that precision migraine treatment is feasible.

Abstract OBJECTIVE Using individual pharmacogenomic phenotypes to investigate the factors affecting the efficacy and tolerability of verapamil for migraine prevention. BACKGROUND Verapamil has a wide range of dosing in headache disorders without reliable tools to predict the optimal doses for an individual. METHODS This is a retrospective chart review examining adults with existing pharmacogenomic reports at Mayo Clinic who had used verapamil for migraine. Effects of six cytochrome P450 phenotypes on the minimum effective and maximum tolerable verapamil doses for migraine prevention were assessed. RESULTS The mean minimum effective and maximum tolerable doses for the 33 patients in final analysis were 178.2(20–320) mg and 227.9(20–480) mg. A variety of CYP2C9, CYP2D6, and CYP3A5 phenotypes were found, without significant association with the verapamil doses after adjusting for age, sex, body mass index, and smoking status. CONCLUSIONS We demonstrated a wide range of effective and tolerable verapamil doses used for migraine in a cohort with various pharmacogenomic phenotypes.

Abstract Objective To compare the artificial intelligence‐enabled electrocardiogram (AI‐ECG) atrial fibrillation (AF) prediction model output in patients with migraine with aura (MwA) and migraine without aura (MwoA). Background MwA is associated with an approximately twofold risk of ischemic stroke. Longitudinal cohort studies showed that patients with MwA have a higher incidence of developing AF compared to those with MwoA. The Mayo Clinic Cardiology team developed an AI‐ECG algorithm that calculates the probability of concurrent paroxysmal or impending AF in ECGs with normal sinus rhythm (NSR). Methods Adult patients with an MwA or MwoA diagnosis and at least one NSR ECG within the past 20 years at Mayo Clinic were identified. Patients with an ECG‐confirmed diagnosis of AF were excluded. For each patient, the ECG with the highest AF prediction model output was used as the index ECG. Comparisons between MwA and MwoA were conducted in the overall group (including men and women of all ages), women only, and men only in each age range (18 to <35, 35 to <55, 55 to <75, ≥75 years), and adjusted for age, sex, and six common vascular comorbidities that increase risk for AF. Results The final analysis of our cross‐sectional study included 40,002 patients (17,840 with MwA, 22,162 with MwoA). The mean (SD) age at the index ECG was 48.2 (16.0) years for MwA and 45.9 (15.0) years for MwoA ( p < 0.001). The AF prediction model output was significantly higher in the MwA group compared to MwoA (mean [SD] 7.3% [15.0%] vs. 5.6% [12.4%], mean difference [95% CI] 1.7% [1.5%, 2.0%], p < 0.001). After adjusting for vascular comorbidities, the difference between MwA and MwoA remained significant in the overall group (least square means of difference [95% CI] 0.7% [0.4%, 0.9%], p < 0.001), 18 to <35 (0.4% [0.1%, 0.7%], p = 0.022), and 35 to <55 (0.5% [0.2%, 0.8%], p < 0.001), women of all ages (0.6% [0.3%, 0.8%], p < 0.001), men of all ages (1.0% [0.4%, 1.6%], p = 0.002), women 35 to <55 (0.6% [0.3%, 0.9%], p < 0.001), and men 18 to <35 (1.2% [0.3%, 2.1%], p = 0.008). Conclusions Utilizing a novel AI‐ECG algorithm on a large group of patients, we demonstrated that patients with MwA have a significantly higher AF prediction model output, implying a higher probability of concurrent paroxysmal or impending AF, compared to MwoA in both women and men. Our results suggest that MwA is an independent risk factor for AF, especially in patients <55 years old, and that AF‐mediated cardioembolism may play a role in the migraine–stroke association for some patients.

Background: Spontaneous coronary artery dissection (SCAD) is a nonatherosclerotic cause of myocardial infarction (MI) occurring most often in young women without cardiovascular disease (CVD) risk factors. SCAD is associated with fibromuscular dysplasia (FMD), pregnant/postpartum state, and extreme mental stress or physical exertion. Migraine headaches have been reported common in SCAD and may increase CVD risk. The objective was to assess differences in presentation, clinical factors, and outcomes in SCAD patients with migraines (M-SCAD) compared to those without, and determine M-SCAD prevalence. Methods: A retrospective cross-sectional observation study of patients with angiographically confirmed SCAD was conducted using the Mayo Clinic SCAD Registry. Participant-provided data and records were reviewed for migraine history, CVD risk factors, SCAD details, CVD and migraine therapies, and outcomes. Ordinal and nominal variables were compared using Pearson's chi squared test, and continuous variables were compared using Student's t-test. Statistical significance was set at p < 0.05. Results: Among 586 patients with complete questionnaires, 236 had a history of migraines; the lifetime and 1-year prevalence of migraines were 40% and 27% respectively. Migraines are more common in SCAD survivors than a comparable population (40% vs 33%, p=0.0002). M-SCAD patients were more likely to be female (99.6% vs 94%, p=0.001), have SCAD at a younger age (45.3 yrs. vs 47.6 yrs., p=0.004), and report a history of aneurysms (7.6% vs 1.7%, p=0.0004), 46% of which occurred in head or neck vessels. There was a trend toward increased head and neck FMD in M-SCAD (32% vs 23%, p=0.095). Prevalence of depression (27% vs 17%, p=0.005), anxiety (32% vs 24%, p=0.038), recurrent angina (62% vs 44%, p=<0.001) and concern for recurrent SCAD (46% vs 37%, p=0.034) were higher in M-SCAD. Triptans were stopped in 39% of migraineurs at the time of MI. Conclusion: As compared to SCAD survivors without migraines, those with M-SCAD are younger, more often report a history of depression, anxiety, and aneurysms, and have more chest pain after SCAD. Further investigation into this association and development of safe, effective treatments of both headache and angina after SCAD is needed.

Abstract Background: OnabotulinumtoxinA has been widely used for control of chronic migraine. The aim of the current study was to evaluate the efficacy of different doses of the onabotulinumtoxinA therapy in patients with chronic migraine.Methods: This is a retrospective paired comparison study on patients with chronic migraine who received at least 3 rounds of 150 units of onabotulinumtoxinA followed by at least 3 rounds of 200 units of onabotulinumtoxinA. The data from the patient-reported questionnaires about headache days, severe headache days and wearing off periods were reviewed. Results: A total of 175 patients were included in this study. The headache days and severe headache days decreased from 13.62±10.79 and 5.88±6.73 to 11.02±10.61and 4.01±4.89 days, after increase in the onabotulinumtoxinA dose, respectively (P<0.001 for both comparisons). The favorable effect of the 200 units compared to the 150 units of the onabotulinumtoxinA, was independent from the headache location and the duration of the onabotulinumtoxinA 150 units therapy; and persisted as patients continued to receive the higher dose of onabotulinumtoxinA. Increase in the onabotulinumtoxinA dose was also associated with a decreased wearing off period (P<0.05). Conclusion: Increase in the onabotulinumtoxinA dose is associated with better headache control and independently improves the numbers of the headache and severe headache days.

Objective “Pain interference” and “headache impact” refer to negative consequences that pain and headache have on one’s life. This study investigated determinants of these negative impacts in a large patient cohort who have chronic migraine with medication overuse. Methods Six hundred and eleven adults were enrolled from 34 headache, neurology, and primary care clinics. Negative consequences of chronic migraine with medication overuse were determined using the Patient-Reported Outcomes Measurement Information System (PROMIS) Pain Interference 6b questionnaire and the Headache Impact Test 6. Relationships between PROMIS-6b and Headache Impact Test 6 scores with demographics, headache characteristics, medication use, anxiety symptoms, and depression symptoms were assessed with linear regression. Elastic Net regression was used to develop a multiple regression model. Results PROMIS-6b T-Scores averaged 65.2 (SD 5.4) and Headache Impact Test 6 scores averaged 65.0 (SD 5.3), indicating severe negative consequences of chronic migraine with medication overuse. Chronic migraine with medication overuse interfered with enjoyment of life, concentration, daily activities, doing tasks away from home, and socializing. Depression symptom severity had the strongest relationship with pain interference and headache impact. Moderate-to-severe headache frequency, headache intensity, and anxiety symptoms were also associated with pain interference and headache impact. Conclusions Chronic migraine with medication overuse is associated with substantial negative consequences, the extent of which is most strongly related to depression symptoms.

AbstractBackgroundCurrently, there is no biologically based rationale for drug selection in migraine prophylactic treatment.MethodsTo investigate the genetic variation underlying treatment response to verapamil prophylaxis, we selected 225 patients from a longitudinally established, deeply phenotyped migraine database (N = 5983), and collected uninterrupted quantitated verapamil treatment response data and DNA for these 225 cases. We recorded the number of headache days in the four weeks preceding treatment with verapamil and for four weeks, following completion of a treatment period with verapamil lasting at least five weeks. Whole‐exome sequencing (WES) was applied to a discovery cohort consisting of 21 definitive responders and 14 definitive non‐responders, and the identified single nucleotide polymorphisms (SNPs) showing significant association were genotyped in a separate confirmation cohort (185 verapamil treated patients). Statistical analysis of the WES data from the discovery cohort identified 524 SNPs associated with verapamil responsiveness (p < 0.01); among them, 39 SNPs were validated in the confirmatory cohort (n = 185) which included the full range of response to verapamil from highly responsive to not responsive.ResultsFourteen SNPs were confirmed by both percentage and arithmetic statistical approaches. Pathway and protein network analysis implicated myo‐inositol biosynthetic and phospholipase‐C second messenger pathways in verapamil responsiveness, emphasizing the earlier pathogenic understanding of migraine. No association was found between genetic variation in verapamil metabolic enzymes and treatment response.ConclusionOur findings demonstrate that genetic analysis in well‐characterized subpopulations can yield important pharmacogenetic information pertaining to the mechanism of anti‐migraine prophylactic medications.