Open Access

Background: Monoclonal antibodies (mAbs) targeting the calcitonin gene-related peptide (CGRP) pathway have shown good efficacy in migraine prophylaxis. However, a subset of patients does not respond to the first mAb treatment and switches among the available mAbs. The goal of this study is to characterize the switching pattern of migraine patients treated with anti-CGRP(-receptor, -R) mAbs, and to describe the headache burden of those who did not switch, switched once, and switched twice. Methods: This study used real world data from the NeuroTransData Cohort, a registry of migraine patients treated at outpatient neurology clinics across Germany. Patients who had received at least one anti-CGRP(-R) mAb were included. Headache diaries were collected at baseline and during treatment, along with quality of life measures every three months. Results were summarized for the subgroups of patients who did not switch and those with one and two switches. Results: Of the 655 eligible patients, 479 did not switch, 135 switched once, 35 twice, and 6 three or more times. The ≥ 50% response rates for monthly migraine days were 64.7%, 50.7%, and 25.0% for the no switch, one switch, and two switches groups in their last treatment cycles, respectively. Quality of life measures improved for the no switch and one switch groups, but not for the two switches group. Conclusion: Patients who switched among anti-CGRP(-R) mAbs during the course of their treatment still benefited overall but to a lesser extent than those who did not switch. Treatment response in patients who switched twice was markedly lower compared to the no switch and one switch subgroup. Keywords: Calcitonin-gene-related peptide; Migraine; Monoclonal antibody; Prophylaxis; Real-world experience.

A database, PubMedMA (PubMed Meta-Analyses), was developed, which is based on the analysis of biomedical research papers and summaries using a Large Language Model (LLaMa3). The aim is to collect medical data in a structured database as medical records and make them usable for meta-analyses. By using Large Language Models, the analysis and extraction of data from the PubMed database for medical papers and summaries is implemented and automated as a pipeline. The generated medical records include information on diseases, drugs, medical groups and outcomes as well as results and are stored in the PubMedMA database. This enables comprehensive medical meta-analyses and promotes research through easy accessibility and structured query options that ensure clear access to the data. The evaluation of the results confirmed that the accuracy of data extraction and processing was generally good. The database proves to be a potentially valuable tool for medical researchers and the pharmaceutical industry to accelerate research and development processes and improve treatment methods.