OBJECTIVES: Fremanezumab, a humanized monoclonal antibody targeting calcitonin gene-related peptide, is indicated for chronic and episodic migraine (CM, EM) prevention. Using data from the FOCUS randomized controlled trial (RCT; NCT03308968), fremanezumab was found to be cost-effective after inadequate responses to two or more previous classes of migraine preventive treatments from a UK healthcare perspective. This study repeated the cost-effectiveness analysis of fremanezumab vs best supportive care (BSC) using data from PEARL (EUPAS35111), a pan-European, real-world, non‑interventional study.

METHODS: A semi-Markov cost-economic model distributed patients across monthly migraine day (MMD) states (0–28 MMD) (base case scenario). Patients with ≥30% (CM)/50% (EM) reduction in MMDs over 12 weeks continued treatment, and 20% of on-treatment patients positively stopped annually. MMD responder rates were from PEARL or FOCUS, with clinical characteristics from FOCUS. Incremental cost-effectiveness ratios (ICERs) were calculated over a 10-year horizon. Sensitivity analyses were conducted on the base case.

RESULTS: ICERs were £15,093 (FOCUS) and £11,792 (PEARL) per quality-adjusted life-year (QALY) in CM, and £21,610 (FOCUS) and £18,036 (PEARL) per QALY in EM. This shows that cost‑effectiveness improved in the real-world vs RCT setting (~£3.5k reduction). Improvements were driven by greater reductions in headache burden, resulting in QALY gains. In CM, there was a 66.8% greater reduction in incremental MMDs using PEARL (–664.2) vs FOCUS (–398.2), resulting in a similar gain of 61.9% in incremental QALYs (0.799 [PEARL] vs 0.494 [FOCUS]). Similar outcomes were observed in EM (63.4% improvement in incremental MMD reduction; 57.0% improvement in incremental QALYs). Results were robust to sensitivity analysis adjusting variables by ±20%.

CONCLUSIONS: When comparing cost-effectiveness of fremanezumab vs BSC after inadequate responses to two or more classes of preventive migraine treatments, improved ICERs were observed using models populated with real-world vs RCT data, despite more patients remaining on treatment.