Candesartan is recommended for migraine prevention, but real-world data on effectiveness, persistence of response, and titration tolerability across episodic migraine (EM) and chronic migraine (CM) and treatment lines are limited.

We analyzed consecutive adults starting candesartan for migraine prevention in the prospective French FHU InovPain registry. Visits occurred at month 3 (M3), month 9 (M9), and month 15 (M15). The primary endpoint was the M3 ≥ 50% responder rate, defined as ≥ 50% reduction from baseline in monthly migraine days (MMD). The main analysis used a modified intention-to-treat/last-observation-carried-forward approach (modified ITT/LOCF; n = 183), including patients who discontinued candesartan before M3 because of treatment-emergent adverse events as non-responders. Observed-case analyses were supportive. Exploratory analyses assessed EM versus CM, CM-to-EM conversion, and outcomes according to prior preventive exposure.

In the modified ITT/LOCF analysis, 77/183 patients achieved a ≥ 50% MMD reduction at M3 (42.1 [95% CI 35.2;49.3]%). In a fully conservative analysis including all candesartan initiators, the M3 ≥ 50% responder rate was 77/207 (37.2 [30.9;44.0]%). In the observed-case analysis, 77/163 patients achieved ≥ 50% response (47.2 [39.7;54.9]%). Mean MMD decreased from 14.1 to 9.7 and MHD from 17.3 to 11.2 at M3 (both P < 0.001). Among patients remaining on treatment and under follow-up, observed-case ≥ 50% responder rates were 75.6% at M9 and 78.9% at M15; these estimates reflect persistence of response among treatment continuers rather than population-level response rates. Corresponding modified ITT/LOCF ≥ 50% responder rates were 44.8% and 43.2%. At M3, ≥ 50% responder rates were higher in EM than CM (66.2% vs. 31.5%), while 43.8% of baseline CM patients no longer met CM criteria. Outcomes were similar in prophylaxis-naive and previously treated patients. Treatment-attributed adverse events occurred in 29.0% and led to discontinuation in 15.8%, mainly during early titration. Systolic blood pressure decreased modestly, whereas diastolic blood pressure remained stable.

In this real-world observational cohort, candesartan was associated with clinically meaningful reductions in migraine burden by M3, including in conservative modified ITT/LOCF analyses. Longer-term observed-case data suggested persistence of response among treatment continuers, but should not be interpreted as population-level effectiveness. These findings support candesartan as a pragmatic oral preventive option across migraine phenotypes, while highlighting the need for cautious interpretation of observational data and for prospective comparative studies.