Onabotulinumtoxin-A in Chronic Migraine: Should Timing and Definition of Non-Responder Status Be Revised? Suggestions From a Real-Life Italian Multicenter Experience.

[OBJECTIVE] To clarify whether the clinical response after the first 2 cycles with Onabotulinumtoxin A can accurately predict the long-term response.

[BACKGROUND] Onabotulinumtoxin-A (OBT-A) is an approved preventive treatment option for chronic migraine (CM). Nowadays, it remains to be clarified if the treatment has to be prolonged for at least an entire year (4 injections every 3 months - ie, quarterly, as proposed in the PREEMPT trials) or it can be halted after the second or third injection if not clinically effective.

[DESIGN AND METHODS] We conducted a multicenter observational cohort study based on real-life data on the usage of OBT-A in CM patients from 2 Italian headache centers, Roma Campus Bio-Medico and Milano Besta, adopting the whole 4-injections protocol. We performed a retrospective analysis of medical records of consecutive patients treated in the 2 centers. The main statistical analysis aimed to evaluate longitudinal measures related to headache (monthly headache frequency, monthly number of analgesic drugs, MIDAS). We hypothesized from our clinical practice with OBT-A that only 2 cycles of treatment were not enough to actually define the non-responder status to botulinum toxin A and that probably a longer time of treatment is needed to get the condition of long-term (delayed) responder.

[RESULTS] We considered 115 patients from the 2 centers: 53 in Roma and 62 in Milano. Regarding the main analysis, a clear improvement in each measure was obtained at the 6 months assessment and maintained up to 12 months. Comparing patients with <30% and ≥30% reduction in headache frequency between T0 and T2 or T4 (respectively, "Non-Responders" and "Responders"), we found that the agreement between the classification Responders/Non-Responders at T2 and T4 was not very high (79/104 = 76.0%, with a "moderate" Cohen's Kappa of 0.51), suggesting that the status at T4 is not fully predictable by the status at T2 (λ = 0.47). Responders for headache frequency at T4 were 54.8%. Among Responders at T2, Responders at T4 were 47/62 = 75.8% (95% CI: 64.5%, 85.5%), while among Non-Responders at T2, Responders at T4 were 10/42 = 23.8% (95% CI: 11.9%, 38.1%). Similarly, even when considering the 50% reduction in painkillers consumption or in MIDAS total score between T0 and T2 as possible prognostic factors, the changes occurring at T4 are not strongly predictable by those at T2.

[CONCLUSIONS] A ≥30% reduction in headache frequency at T2 cut-off is not adequate in predicting a late response to treatment: more than a quarter of excluded patients would miss a clinical improvement with an ongoing treatment, while in a similar percentage of Responders the treatment would lose efficacy. Results from our real-life study suggest that we possibly have to postpone the definition of Responder/Non-Responder to OBT-A at least after 1 year of treatment (4 cycles).