Some observed problems of the evidence presented for zolmitriptan nasal spray in the “Evidence-Based Recommendations of Pharmacological Treatment of Migraine Attacks”

We commend the initiative of developing extensive, evidence-based recommendations for migraine pharmacotherapy (1). The results are appropriately presented with both risk ratios and absolute risk differences (ARDs) (1). In reviewing the triptan section, we noted, however, the unusual recommendation on zolmitriptan 5 mg nasal spray, a drug formulation approved in Europe since 2001 and the USA since 2003, on page 34 in the recommendations: “In subjects with migraine, we suggest against zolmitriptan 5 mg nasal spray for the acute treatment of migraine attacks” (1).

Zolmitriptan nasal spray is the only formulation among 11 triptan formulations not recommended (1). We therefore reviewed the evidence presented for the zolmitriptan 5 mg nasal spray, and we found some problems that should result in a correction of the recommendation.

Review of presented evidence for nasal zolmitriptan 5 mg

Missing pivotal randomized controlled trial

The Forest plot for two-hour headache relief does not include the study by Dodick et al. (2), comprising the largest placebo-controlled trial of zolmitriptan 5 mg nasal spray (n = 1868). Its exclusion most likely underestimates the pooled treatment effect.

Dose misclassification

In the zolmitriptan vs. placebo comparison of the study by Charlesworth et al. (3), the 5 mg headache relief rate (68.9%) seems to have been misreported as the 0.5 mg arm (39.6%). Correcting this restores appropriate weighting in the meta-analysis.

Reporting of wrong outcome

In the zolmitriptan vs. placebo comparison of Gawel et al. (4), the numbers reported (150/461 vs. 38/451) appear to be for two-hour total symptom relief and not for two-hour pain relief.

For two-hour pain relief, the correct numbers (provided by Grünenthal GmbH, Aachen, Germany) are 265/409 vs. 94/394, i.e. a risk ratio of 2.72 (95% confidence interval (CI) = 2.25–3.29) and an ARD of 40.9% (95% CI = 34.1–47.8%)

Of note, the study by Gawel et al. (4) included patients who treated mild migraine attacks. This is methodologically inconsistent with the other studies and could in theory introduce bias.

Recalculated effect estimates

Using the correct numerators for headache relief at two hours for the first treated attack, 162/235 vs. 69/226 for Charlesworth et al. (3), 647/935 vs. 351/933 for Dodick et al. (2) and 265/409 vs. 94/394 for Gawel et al. (4), the combined results are 1074/1579 vs. 514/1553 and result in a risk ratio of 2.06 (95% CI = 1.90–2.22) and an ARD of 34.9% (95% CI = 31.6–38.2%).

Results for pain-free at two hours were not presented in the publications (2,4) but were provided by Grünenthal GmbH (Aachen, Germany). Using the numerators for pain-free at two hours for the first treated attacks, 81/228 vs. 14/218 for Charlesworth et al. (3), 326/ 927 vs. 131/924 for Dodick et al. (2) and 138/409 vs. 33/394 for Gawel et al. (4), the combined results are 545/1564 vs. 178/1536 and result in a risk ratio of 3.01 (95% CI = 2.58–3.51), and an ARD of 23.3% (95% CI = 20.4 −26.1%). Again, please note that the study by Gawel et al. (4) included patients who treated mild migraine attacks. The two-hour pain-free calculation here is based on patients who treated moderate or severe attacks to be consistent with the other two studies.

Suggested implementation

When the full evidence is included and accurately tabulated, the recommendation for nasal zolmitriptan 5 mg warrants a change from “suggest against” to “recommend for” acute treatment of migraine attacks, with “high quality of evidence” and “strong strength of the recommendation”.