Pathogenesis and Preventive Pharmacotherapy of Migraine Aura: A Critical Review of Cortical Spreading Depolarization Mechanisms, Drug Repurposing Evidence, and the Emerging Therapeutic Pipeline

Migraine with aura affects 25-30% of patients with migraine and remains a substantial unmet preventive need. Calcitonin gene-related peptide (CGRP) monoclonal antibodies, large molecules with limited cortical penetration, do not selectively suppress aura, leaving the underlying cortical spreading depolarization (CSD) mechanism therapeutically untouched. This critical review synthesises the pathobiology of CSD and appraises preventive pharmacotherapy specifically for migraine aura, drawing on studies retrieved from PubMed from inception through January 2026, each described by the design that produced it. CSD pathogenesis is organised into four phases: pre-CSD vulnerability, initiation, glial propagation, and neuro-inflammatory transduction. Lamotrigine and memantine target initiation, but their aura evidence is uncontrolled and the network meta-analysis that ranks memantine favourably carries a 95% confidence interval from 1.31 to 23.69, so it constrains the size of a definitive trial only to within an order of magnitude. Tonabersat targets propagation and produced the one randomised trial with aura as a prespecified primary outcome, in 39 patients randomised and 31 analysed per protocol; it was not pursued in migraine and the result stands unreplicated. Repurposing candidates (spironolactone, amiloride) and pipeline agents (selective KATP antagonists, gap-junction modulators, anti-PACAP antibodies) are mapped onto specific phases. Seven falsifiable predictions (H1-H7) operationalise each mechanism for prospective human and preclinical testing. The therapeutic gap for migraine aura prevention reflects correctable methodological choices, not biological intractability: four mechanism-based drug classes (NMDA antagonists, gap-junction modulators, KATP-channel antagonists, and anti-PACAP antibodies) can be tested in aura-enriched populations with biomarker-anchored endpoints to convert the four-phase model into actionable prevention. This version corrects errors in version 1; see the correction notice at the head of the manuscript.