Partial rebreathing is not effective for early treatment of migraine with aura attacks: Results of a double‐blind, randomized, controlled trial ( PAREMA1 )

Objective Evaluate the efficacy and safety of a novel partial rebreathing device for early treatment of acute attacks of migraine with aura. Background Earlier clinical studies have indicated a potential for CO 2 –enriched gas to be effective for acute treatment of migraine with aura, especially when applied during the early part of the aura stage. We developed a partial rebreathing device inducing moderate, steady‐state hypercapnia with normoxia in order to provide a carbon dioxide delivery system combining efficacy, usability, safety, and affordability. Methods This randomized, double‐blind, sham‐controlled, parallel‐group, group‐sequential study was conducted at 15 study sites, nine located in the United States and six in Germany, between March 2023 and February 2025. The study enrolled patients aged 18–65 years with migraine with aura. The study had a sequential two‐stage study design. At the beginning of stage 1, participants were randomized to active or sham and treated up to four attacks. Participants were instructed to treat from the onset of aura and until 5 min after aura cessation. After having reported four attacks in stage 1, participants had the option to continue into stage 2, an open‐label extension in which they could treat up to five attacks with the active device. During stage 1, participants recorded symptom scores in a study diary app at the onset of aura and after 1, 2, 24, and 48 h. Results The study was terminated at the interim analysis point due to the lack of effect, at which point 142 participants had been enrolled (mean age 39.2 years, 81% women [115/142]). Sixty‐seven participants had reported at least one study attack by the time of the study termination. None of the primary or secondary endpoints reached statistical significance. The primary endpoint Absence of Moderate or Severe Pain at 2 hours was 69.7% (46/66) [95.2% confidence interval (CI), 48.5, 90.9] in the sham group and 60.0% (42/70) [95.2% CI, 37.6, 82.4] in the active group ( p = 0.379), whereas Pain Freedom at 2 hours was 18.2% (12/66) [95.2% CI, 1.3, 35.1] in the sham group and 21.4% (15/70) [95.2% CI, 3.6, 39.2] in the active group ( p = 0.717). Conclusion Partial rebreathing inducing moderate hypercapnia with normoxia was not effective for aura‐stage treatment of migraine‐with‐aura attacks. The study was preregistered at ClinicalTrials.gov (registration number NCT05546385).