CGRP drugs removed 0.7 to 3.8 migraine days a month in the trials
The older oral preventives managed 0.44 to 1.4. The larger change is that the new drugs work in people the old ones failed, and are tolerated well enough to stay taken.
| Group | Value (days) |
|---|---|
| Topiramate | 1.4 |
| Beta-blockers | 1.27 |
| Amitriptyline | 1.2 |
| Flunarizine | 0.44 |
The drugs that target calcitonin gene-related peptide reduce monthly migraine days by between 0.7 and 3.8 compared with placebo, depending on the agent and the population studied [s1]. The oral preventives they were introduced against — beta-blockers, topiramate, amitriptyline, flunarizine — manage between 0.44 and 1.4 days [s1]. On the headline number, that is an improvement, not a transformation. The substantive change is elsewhere: the new drugs work in people for whom the old ones did not, and they are tolerated well enough that patients keep taking them.
A review summarising the International Headache Society's guidance on migraine treatment and prophylaxis, together with meta-analyses of the newer agents, lays out the comparison directly [s1]. In episodic migraine, in patients averaging eight migraine or headache days a month at baseline, the oral prophylactics lowered migraine days per month against placebo by 1.27 (beta-blockers), 1.4 (topiramate), 1.2 (amitriptyline) and 0.44 (flunarizine) [s1].
What the newer agents deliver
The four monoclonal antibodies against CGRP or its receptor are effective in both episodic and chronic migraine, with reductions in monthly migraine days of 0.7 to 3.2 for eptinezumab, 1.3 to 3.8 for fremanezumab, 1.1 to 3.7 for galcanezumab and 1.0 to 2.5 for erenumab [s1]. The small-molecule CGRP receptor antagonist atogepant is supported by high-level evidence for 0.7 to 2.4 fewer migraine days per month, again across both episodic and chronic migraine [s1].
Those are ranges rather than point estimates because they span different trials and populations, and the ranges overlap the old drugs' numbers at the bottom end. A reader comparing 1.4 against 1.0 to 2.5 should not conclude that the newer class is several times better on average effect size. It is not.
Where the class does something the old drugs could not
Three findings in the review carry more clinical weight than the day counts.
The monoclonal antibodies, atogepant and onabotulinumtoxinA have been found effective in patients for whom previous oral migraine prophylactics were ineffective [s1]. That is a different claim from a larger average effect: it means the failure of one class does not predict failure of the other, which changes what happens after a first preventive does not work.
They are also effective in chronic migraine both with and without acute drug overuse [s1] — historically a population where preventives performed poorly and where the overuse itself was often treated as something to be resolved first.
And they are well tolerated [s1]. Tolerability is not a soft outcome in migraine prevention. A drug that works in a trial and is stopped after six weeks because of cognitive fog, weight change or sedation does not prevent migraine in a life. The review's own framing is that the newer prophylactics are especially useful for patients in whom the traditional oral drugs and onabotulinumtoxinA are ineffective, not tolerated, or contraindicated [s1].
The safety picture from post-marketing data
Trials establish efficacy in selected populations over defined periods. Signals that are rare, or that emerge in people trials excluded, show up afterwards in spontaneous reporting databases. A systematic review of 30 pharmacovigilance studies drawing on the FDA's Adverse Event Reporting System, the WHO's VigiBase and EudraVigilance, covering publications up to September 2025, assembled what those databases currently show for erenumab, galcanezumab, fremanezumab, eptinezumab, rimegepant, ubrogepant, atogepant and zavegepant [s2].
For the antibodies, consistent disproportionate reporting appeared for injection site reactions, alopecia — fremanezumab reporting odds ratios ranging from 2.73 to 6.9 — and constipation, primarily with erenumab, at reporting odds ratios from 4.92 to 17.94 [s2]. For the gepants, common signals were nausea and fatigue or somnolence, with highly specific signals for severe constipation with atogepant (ROR025 = 19.99) and dysgeusia with zavegepant (ROR025 = 212.07), the latter linked to its nasal route of administration [s2].
The signals that matter most are rarer. Reversible cerebral vasoconstriction syndrome was reported disproportionately for erenumab (ROR 9.43, 95% CI 4.5–19.8) and cervical artery dissection for galcanezumab (ROR 14.0, 95% CI 6.22–31.4) [s2]. No comparable signals have been detected for the gepants to date, though the review is explicit that this distinction needs confirmation as real-world exposure grows [s2]. At class level, a signal for cerebrovascular disease as a whole was identified for CGRP inhibitors as a group, at a reporting odds ratio of 1.22 [s2].
How to read a disproportionality signal
These numbers are not risk estimates. A reporting odds ratio measures how often an event appears alongside a drug in a voluntary reporting database relative to all other drugs in the same database. It is sensitive to publicity, to litigation, to how closely a drug is being watched, and to the fact that people who report are not a random sample of people who take the drug. It cannot tell you how many people out of a thousand are affected, and it cannot establish that the drug caused the event.
What it can do is flag where to look. CGRP is a vasodilator, so a cerebrovascular signal in a class that blocks it is biologically coherent rather than arbitrary, which is why the review treats the divergence between antibodies and gepants as a question rather than a settled difference [s2].
What to watch
Two things. Whether the antibody–gepant divergence on cerebrovascular signals survives as gepant exposure catches up [s2]. And whether the effect sizes hold in unselected populations: trial ranges reported above were measured in people who met entry criteria and attended follow-up, and prevention outside a trial is a longer, messier proposition.
This article is informational and is not medical advice. It does not describe doses or recommend any drug; those decisions belong with a reader and their clinician.
Sources
- Drugs for Migraine Prophylaxis — Deutsches Ärzteblatt International, 2026-02-26
- Post-marketing safety of CGRP monoclonal antibodies and gepants: A systematic review of spontaneous reporting system data — Headache, 2026-03-15
Sources
- Drugs for Migraine Prophylaxis — Deutsches Ärzteblatt International , February 26, 2026
- Post-marketing safety of CGRP monoclonal antibodies and gepants: A systematic review of spontaneous reporting system data — Headache , March 15, 2026
More on
Chocolate, cheese and caffeine are blamed for migraine far more often than tested
A pooled analysis puts patient-reported rates at 13.9%, 8.0% and 12.0%. All of it is self-report, the heterogeneity is enormous, and the reviewers stop short of recommending avoidance.
Among GLP-1 users, oxycodone carried the highest short-term gut risk
A 411,000-patient claims study found adults on GLP-1 drugs who started oxycodone had more severe constipation and bowel obstruction at 30 days than those starting hydrocodone or tramadol.
507 digital therapeutics are approved in four countries. Effect sizes aren't comparable
A new structured dataset finally makes the approvals countable. Two trials published weeks apart show why counting them says little: 0.86 fewer migraine days in one, a 20-point symptom shift in the other.
Acetaminophen is the top US cause of acute liver failure, nearly all from overdose
At recommended doses it is one of the best-tolerated painkillers. The liver harm concentrates at high doses, and half of the failure cases are unintentional, often from combining products.