CGRP Inhibitors for Migraine Prevention

Healthcare · §610

If you get four or more migraine days a month and the older preventives have either failed you or wrecked you with side effects, this is the class of drugs to ask your neurologist about. CGRP-targeting drugs are the first migraine preventives actually designed for migraine — six FDA-approved options, ten years of trial data, and a 2024 American Headache Society statement moving them to first-line. About half of patients cut their monthly attack days in half; a meaningful minority do far better than that.

Decide · Daily Evidence Strong Chapter Healthcare

Best preventive class available. Two fewer migraine days a month on average, half of patients halving their monthly count, occasional remission. Tolerability is in a different league from topiramate or amitriptyline — a monthly self-injection or a daily pill, mostly without side effects, no cognitive fog. The catch is access: list prices near $800 a month, prior-authorisation paperwork on most plans, and a slow-onset cardiovascular safety question that won't fully resolve for another decade.

Every migraine attack involves a surge of a single neuropeptide — calcitonin gene-related peptide, or CGRP — released from the sensory nerves that line the membranes around your brain. CGRP widens blood vessels, irritates the surrounding tissue, and turns up the volume on pain signals to the brain. Block CGRP and you blunt the chain reaction before the attack can build.

That is what this class of drugs does. Four of them are monoclonal antibodies — large protein molecules that float around the bloodstream and grab CGRP (or its receptor) before it can bind. You inject one of these yourself, once a month or once every three months, with a small device that looks like an EpiPen. The other two are gepants — small molecules you swallow as a pill, either every day (atogepant) or every other day (rimegepant). Same target, different shape and timing.

What the trials actually show

The honest summary, across more than a dozen large randomised trials and six approved drugs: you can expect about two fewer migraine days a month than you'd get on placebo, and roughly a one-in-two chance of cutting your monthly attack count in half. A smaller group — maybe one in five — does much better than that, and a meaningful number reach near-remission.

The flagship erenumab trial (STRIVE) enrolled patients averaging 8 migraine days a month. After six months, the high-dose group was averaging 3.7 fewer days versus 1.8 fewer on placebo, with 50% of treated patients halving their monthly count compared with 27% on placebo. Rescue medication use dropped by 1.6 days a month — meaningful for anyone who measures their week by how many triptans they had to take Goadsby et al. 2017. Numbers from the other drugs in the class are similar: fremanezumab in chronic migraine cut headache days by 4.6 versus 2.5 Silberstein et al. 2017; galcanezumab cut episodic migraine days by 4.7 versus 2.8, with a 62% responder rate Stauffer et al. 2018; atogepant 60 mg daily, 4.2 versus 2.5 Ailani et al. 2021; rimegepant every other day, 4.3 versus 3.5 Croop et al. 2021.

Two things to read into those numbers. First, the placebo arms drop a lot too — migraine prevention trials always have big placebo effects, because the act of being in a careful trial changes behaviour and reporting. So the headline "halved my attacks" you might hear from someone on the drug includes a real placebo contribution. Second, the average effect is modest precisely because it is averaged over responders and non-responders together. If you are a responder, you halve your attacks. If you are not, the drug does little, and you should stop after three months and try a different one in the class — switching from a CGRP-receptor drug (erenumab) to a CGRP-ligand drug (the other three antibodies), or to a gepant, recovers response in a real subset of people.

Effect is durable. Open-label extension data out to five years shows no loss of efficacy and no new safety signals beyond what the labels already warn about Reuter et al. 2022.

What changes if it works

Onset depends on which drug. The IV-infused antibody (eptinezumab) can blunt attacks within 24 hours. The self-injected antibodies usually show their hand within the first month and reach full effect by month three. The pills (atogepant, rimegepant) land somewhere in between — meaningful change inside the first few weeks.

For a responder, the first thing that shifts is the calendar. Where you had ten migraine days a month, you have five. Where you had twenty, you have twelve. The days you do still get an attack are often shorter, less severe, or actually answer to a triptan instead of dragging on for two days. You stop counting your week by how many rescue pills you have left in the strip.

The second thing that shifts is the dread. Chronic migraine teaches you to plan around the next attack — you don't book the trip, you don't take the meeting, you don't promise anyone anything because you don't know which days are yours. That anticipation lifts before the attack frequency fully does. Partners notice. Coworkers stop asking if you're okay. The version of you that disappeared into a dark bedroom three afternoons a week stops disappearing.

The third thing is the postdrome. Migraineurs know the "migraine hangover" — the foggy, washed-out day that follows the attack, where reading is hard and decisions feel heavy. Fewer attacks means fewer of those days. Focus and energy return on the days you would previously have lost to recovery.

Sleep often gets a smaller bonus. Fewer 4 a.m. attacks pulling you out of bed, fewer evenings spent talking yourself into a triptan instead of dinner, fewer nights staring at the ceiling wondering whether tomorrow is one of the bad days. The drug doesn't act on sleep directly — but the absence of migraine in the background of your night is its own quiet improvement.

None of this is a cure. A meaningful minority of patients does reach near-zero attacks, but most responders go from frequent to occasional, not from frequent to none. Honesty about that is part of the deal — overselling these drugs is the fastest way to be disappointed when the third migraine day of the month still arrives.

What you'd actually take

Six approved drugs, two shapes. You pick one with a neurologist — there is no head-to-head trial telling you the answer, so the choice usually comes down to which delivery suits your life, which one your insurance will pay for, and what other conditions you have.

Give it three months. The American Headache Society defines a fair trial as roughly twelve weeks at the standard dose; before that you're guessing. If you're not at a 30–50% reduction in monthly migraine days by then, switch within the class — receptor-target to ligand-target, or antibody to gepant. Non-response to one is not non-response to all of them.

When to be careful

Pregnancy and planning to conceive. The antibodies hang around in your bloodstream for weeks after you stop — half-lives of three to four weeks, meaningful levels for months. CGRP plays a role in placental blood flow, and pregnant patients were excluded from all the trials. Don't start one if you might be pregnant in the next several months; stop and wait out the washout if you're planning.

Established cardiovascular disease. CGRP is one of the body's vasodilators — it gets called on during ischaemia to open vessels. Trials excluded patients with significant heart disease, recent heart attack, or stroke, so we genuinely don't know how the long-tail safety looks for that group. Healthy hearts are fine; if you have angina, prior MI, or stroke, this is a real conversation with your cardiologist and neurologist, not a shrug.

Raynaud's phenomenon. The same vasodilator role cuts the other way in the fingers and toes: blocking CGRP can worsen the cold-triggered vasospasm of Raynaud's. If your fingers go white and numb in the cold, mention it before starting — it's worth weighing, not a hard stop.

Uncontrolled or borderline hypertension. Erenumab in particular has caused new-onset and worsened high blood pressure in the post-marketing period — most often within a week of the first dose. The FDA added this to the label in 2020 Saely et al. 2021. Check your blood pressure before starting, and again a week and a month in. If you already run high, one of the ligand-targeting antibodies or a gepant may be the safer pick.

Severe constipation history. Erenumab again is the worst offender — about 20% of real-world chronic migraine patients report constipation at six months, with a handful of post-marketing cases needing hospitalisation. Don't stack it on top of opioids or other constipating medications without a plan.

Strong CYP3A4 drug interactions (for the gepants only). Atogepant and rimegepant are processed by the same liver enzyme as many antifungals, HIV drugs, and some anticonvulsants. The antibody injections do not have this problem.

The cost and the paperwork

US list prices in 2026 sit at roughly $700–$815 a month for the injectable antibodies, around $1,150 a month for atogepant, and about $2,000 per quarterly infusion for eptinezumab. Without insurance, that's roughly $8,000–$12,000 a year. With commercial insurance, manufacturer copay programs (BridgeProgram, Ajovy Copay Card, Emgality Savings, the Aimovig Co-Pay Program) generally bring the out-of-pocket cost to near zero — under $5 a month for many patients. Medicare patients are usually shut out of those programs and pay specialty-tier copays unless they qualify for extra-help.

The bigger friction is the prior-authorisation form. Most insurers still want documentation that you have at least four migraine days a month and that you have already tried and failed two classes of older preventives — typically topiramate plus one of propranolol, amitriptyline, or valproate. Chronic-migraine plans sometimes require an onabotulinumtoxinA trial first. The 2024 American Headache Society statement asks insurers to drop this requirement, and about 45% of US commercial plans have, but most still ask Charles et al. 2024.

What helps in practice: a neurologist who handles a lot of these (rather than primary care), a clear written record of every previous preventive you tried with doses and durations, and patience for the first appeal. Approval usually comes through within a few weeks if the paperwork is clean. Adherence at one year runs around 50–60% — substantially better than the ~25% rate for topiramate over the same horizon, mostly because the side-effect burden is lower.

What people get wrong

"These drugs cure migraine." They don't. They reduce attack frequency. A minority of patients go from frequent to near-zero, but the typical responder goes from frequent to occasional. Plan around that.
"If one didn't work, the class doesn't work for me." Wrong, often. The receptor-binding antibody (erenumab) and the three ligand-binding antibodies (fremanezumab, galcanezumab, eptinezumab) sometimes produce different individual responses; switching across that line, or to one of the gepants, recovers response in a meaningful subset.
"They work like Botox." Both are effective preventives, but they are different drugs aimed at different parts of the system. Botox is a quarterly grid of 31–39 small injections in the head and neck that blocks neurotransmitter release at the nerve ending. A CGRP drug is a single monthly injection (or daily pill) that blocks one specific signal. Many chronic-migraine patients eventually use both together.
"Insurance covers them as first-line now." The guideline says yes; most American insurers still say no. The position statement is a recommendation, not a mandate.
"The trial numbers look small, so the drug must be weak." The average effect is modest because the average pools responders with non-responders. If you are a responder, the personal effect is much larger than two days a month — closer to halving your monthly count.

Where this goes wrong

You don't respond. About a third of patients don't reach a 50% reduction after a fair three-month trial. The fix is to switch drugs within the class, not to abandon the class. If you also don't respond to a second drug with a different target (receptor vs ligand, antibody vs gepant), that's a real signal — go back to your neurologist.
You quietly have medication-overuse headache. If you've been taking a triptan or daily ibuprofen for years, some of your "migraine days" may actually be rebound headache. The CGRP drug helps less than it should until the rescue medication is withdrawn — usually a planned taper alongside starting the preventive.
You stop the drug after it worked. Discontinuing a CGRP antibody after sustained good response usually means your monthly migraine days drift back toward baseline over the next three to six months. Some patients re-respond on restart; some don't fully. This isn't withdrawal, but it isn't a free pause either.
The blood-pressure cuff finds something. A first dose of erenumab that triggers an unexpected jump in blood pressure is uncommon but documented; this is exactly why the early checks matter. Switch to a ligand-binding antibody or a gepant if it happens.
Insurance asks for one more thing. A first denial isn't the answer; it's the opening move. A clean appeal with the prior-preventive history attached is approved more often than not.

Adjacent topics worth a look:

Acute migraine treatment — triptans, ditans, and the acute-use gepants (ubrogepant, rimegepant on-demand, zavegepant nasal). Prevention reduces attacks; rescue treatment ends the ones that still happen.
OnabotulinumtoxinA for chronic migraine — a different mechanism, quarterly injections, often used alongside or before CGRP drugs.
Legacy oral preventives — topiramate, propranolol, amitriptyline, candesartan, valproate — still on the menu, still required by some insurers as the step before a CGRP drug.
Migraine trigger management — sleep regularity, hydration, caffeine pattern, identified food and sensory triggers. None replace prevention drugs at high attack frequency, but they meaningfully shift the floor.
Medication-overuse headache — the often-missed background that limits how well any preventive works.

Related in the handbook

Worsens / aggravates

— CGRP also widens blood vessels, so blocking it can worsen cold-triggered Raynaud's attacks. Mention it if your fingers go white.

Helps

— CGRP inhibitors are the new prevention that halves a typical migraine month — central to women's migraine care.

Alternatives

— CoQ10 is a cheap, low-risk migraine preventive worth trying before or alongside the pricey CGRP drugs.
— Magnesium is a cheap first-line migraine preventive worth trying before stepping up to CGRP drugs.

Substance and claimed effects

CGRP-targeting drugs for migraine prevention are a class of migraine-specific prophylactics that block calcitonin gene-related peptide signalling — a neuropeptide that is released from trigeminal sensory neurons during a migraine attack and that drives vasodilation, neurogenic inflammation, and central sensitisation. The class has two arms: four large-molecule monoclonal antibodies against CGRP itself (fremanezumab, galcanezumab, eptinezumab) or its receptor (erenumab), administered subcutaneously monthly or quarterly (or intravenously quarterly for eptinezumab); and small-molecule oral gepants originally developed for acute use, two of which (atogepant daily, rimegepant every-other-day) are FDA-approved for prevention. Claimed effects covered here: reduction in monthly migraine days and migraine-related disability, reduction in acute (rescue) medication use, conversion from chronic to episodic migraine, and a favourable side-effect profile relative to legacy preventives (topiramate, propranolol, amitriptyline, valproate). Out of scope: acute attack abortion (the original gepant indication; that is a separate decision), non-migraine headache, paediatric use (still poorly studied), and pregnancy.

Evidence by addressing question

Mechanism

CGRP is a 37-amino-acid neuropeptide expressed densely in trigeminal ganglion neurons that innervate the dura. Jugular-venous CGRP rises during spontaneous migraine attacks and falls when an attack aborts with triptans; intravenous CGRP infusion provokes migraine-like headache in migraineurs but not in healthy controls — a chain of human-experimental evidence that established CGRP as causal in migraine pathophysiology rather than merely associated. The class works at two points in that chain: erenumab is a fully human IgG2 that binds the canonical CGRP receptor (the calcitonin-receptor-like-receptor / RAMP1 heterodimer); fremanezumab, galcanezumab, and eptinezumab are humanised IgGs that bind circulating CGRP itself. Gepants (atogepant, rimegepant, ubrogepant, zavegepant) are small-molecule CGRP-receptor antagonists. Pharmacokinetically the two arms are very different: monoclonal antibodies have half-lives of roughly 27–32 days (eptinezumab, fremanezumab) up to ~21 days (erenumab, galcanezumab), are dosed monthly or quarterly, and accumulate to steady state over months; gepants have half-lives of 10–11 hours and reach steady state in days. The mechanism is peripheral in that mAbs do not cross the blood–brain barrier appreciably — yet they prevent attacks, which constrains the older "central origin" model of migraine and supports a peripheral trigeminal initiation step Charles 2024.

Evidence

The class has been tested in more pivotal randomised trials than any previous preventive — every drug has at least one positive phase 3 trial in episodic migraine and several have trials in chronic migraine as well. Effect sizes are broadly consistent across the class.

Erenumab. STRIVE (n=955, episodic migraine, 8.3 mean baseline migraine days/month) showed a placebo-subtracted reduction of 1.4 days at 70 mg and 1.9 days at 140 mg over months 4–6; 50% responder rates were 43% (70 mg) and 50% (140 mg) versus 27% on placebo; acute-medication days dropped by 1.1 and 1.6 days versus 0.2 on placebo Goadsby et al. 2017. In chronic migraine (≥15 headache days/month) erenumab cut migraine days by 6.6 days from a baseline of ~18, versus 4.2 on placebo Tepper et al. 2017. The LIBERTY trial enrolled the harder population — patients who had failed two-to-four previous oral preventives — and still found a 50% responder rate of 30% on erenumab versus 14% on placebo Reuter et al. 2018.

Fremanezumab. In chronic migraine (HALO-CM) the monthly 225 mg arm cut monthly headache days by 4.6 versus 2.5 on placebo; the quarterly 675 mg arm produced similar results Silberstein et al. 2017. The episodic-migraine HALO-EM trial produced a 3.7-day reduction versus 2.2 on placebo Dodick et al. 2018.

Galcanezumab. EVOLVE-1 in episodic migraine: 4.7-day reduction on 120 mg versus 2.8 on placebo; 50% responder rate ~62% versus 39% Stauffer et al. 2018. REGAIN in chronic migraine: 4.8-day reduction versus 2.7 on placebo Detke et al. 2018.

Eptinezumab (the IV-infused mAb). PROMISE-1 in episodic migraine cut migraine days by 3.9 (100 mg) and 4.3 (300 mg) versus 3.2 on placebo over weeks 1–12 Ashina et al. 2020. Onset is the fastest of any preventive — meaningful reduction is detectable on day 1 post-infusion, which is mechanistically explainable by reaching peak plasma levels at the end of infusion rather than over weeks of subcutaneous accumulation.

Atogepant (oral gepant for prevention). ADVANCE (n=873, episodic migraine, baseline ~7.5 monthly migraine days) showed reductions of 3.7 (10 mg), 3.9 (30 mg), and 4.2 (60 mg) migraine days versus 2.5 on placebo over 12 weeks Ailani et al. 2021. PROGRESS extended this to chronic migraine: 7.5-day reduction (60 mg daily) versus 5.1 on placebo from a baseline of ~19 days Pozo-Rosich et al. 2023.

Rimegepant. The same 75 mg oral tablet that aborts acute attacks reduced monthly migraine days by 4.3 versus 3.5 on placebo when dosed every other day for 12 weeks; a meaningful proportion of patients achieved a 50% reduction in moderate-to-severe migraine days Croop et al. 2021. The dual acute-and-preventive label is unique in the class.

Across the class, the placebo-subtracted effect is roughly 1.5–2.5 fewer migraine days per month in episodic migraine and 2–3 fewer in chronic migraine — modest in absolute terms (placebo arms also drop substantially, a known feature of preventive-migraine trials), but the responder analyses tell the more important story: roughly 40–60% of patients achieve a ≥50% reduction in migraine days, and a meaningful minority (~20%) achieve ≥75% reduction or near-complete remission. Long-term open-label extensions out to 3–5 years show sustained efficacy without tolerance development Reuter et al. 2022.

Practice / clinical consensus

The American Headache Society 2024 position update reclassified CGRP-targeting therapies as first-line for migraine prevention — explicitly removing the prior requirement that patients first fail two classes of older oral preventives before insurance would approve a CGRP drug Charles et al. 2024. The rationale named in the statement: every prior first-line preventive (topiramate, propranolol, valproate, amitriptyline, candesartan) was originally developed for a different indication and adopted off-label for migraine; tolerability is poor and adherence is correspondingly low. CGRP-targeting therapies are the first class designed for migraine. The statement also flagged that real-world response is often better when CGRP drugs are started earlier rather than after multiple failed prior preventives. International guideline bodies (EHF, NICE) have generally moved in the same direction over 2022–2024, though access criteria vary by country.

Protocol

Dosing varies by drug: erenumab 70 mg or 140 mg subcutaneous monthly; fremanezumab 225 mg monthly or 675 mg quarterly subcutaneous; galcanezumab 240 mg loading dose then 120 mg monthly subcutaneous; eptinezumab 100 mg or 300 mg IV every 3 months; atogepant 10/30/60 mg oral daily (60 mg is standard for chronic migraine); rimegepant 75 mg oral every other day for prevention (or as needed for acute use, max 18 doses/month). Subcutaneous mAbs are self-administered with a prefilled syringe or auto-injector at home. The expected onset varies: eptinezumab IV produces measurable effect within 24 hours; subcutaneous mAbs typically show benefit within the first month, with maximal effect by month 3; gepants have a similar 1–4 week onset window. The American Headache Society guidance is that a 3-month trial is needed to assess response, and that non-responders to one CGRP-targeting drug may respond to another (switching within the class is reasonable, including switching ligand-vs-receptor target).

Contraindications

The package inserts and post-marketing data identify several concerns. Hypertension: post-marketing surveillance of erenumab identified new-onset and worsened hypertension, most commonly within seven days of the first dose; the FDA added a warning in 2020 Saely et al. 2021. This is a class concern in theory (CGRP is a vasodilator; blocking it could raise blood pressure) but the post-marketing signal has been most prominent for erenumab. Cardiovascular disease: phase 3 trials excluded patients with significant cardiovascular disease; CGRP plays a protective vasodilatory role during ischaemia in animal models, so the class is treated cautiously in patients with stable angina, prior MI, or stroke. Long-term cardiovascular outcome trials are not yet complete. Constipation: most prominent with erenumab (~3% in trials, ~21% in real-world cohorts); a small number of post-marketing cases of severe complications including hospitalisation and surgery prompted a label update. Gepants also cause some constipation and nausea but at lower rates. Pregnancy and lactation: mAbs have half-lives of 3–4 weeks and would persist into a pregnancy, with theoretical concerns about CGRP's role in placental and uteroplacental blood flow; trials excluded pregnant patients and the drugs are not recommended in pregnancy. Pregnancy registries (Aimovig pregnancy registry, MotherToBaby) are accumulating data. Hypersensitivity: rare angioedema and anaphylactoid reactions reported post-marketing. Drug interactions are minimal for mAbs (large molecules; not CYP-metabolised). Gepants are CYP3A4 substrates, so atogepant and rimegepant dosing is reduced or contraindicated with strong CYP3A4 inhibitors (ketoconazole, ritonavir) or inducers (rifampin, certain anticonvulsants).

Audience

Indication: adults with episodic or chronic migraine for whom preventive therapy is appropriate — typically defined as ≥4 monthly migraine days, or fewer but with severe disability, or frequent rescue-medication use that risks medication-overuse headache. Migraine is roughly 3:1 female-to-male; the trials reflect this skew. Older adults and adolescents are under-studied; an atogepant paediatric trial is in progress. Patients with frequent medication-overuse headache benefit substantially, often with concurrent withdrawal of the overused acute drug.

Misconceptions

(a) "CGRP drugs cure migraine." They do not. They reduce attack frequency by a meaningful margin; complete remission is uncommon but does occur in a minority. (b) "All four mAbs are interchangeable." The ligand-binders (fremanezumab, galcanezumab, eptinezumab) and the receptor-binder (erenumab) sometimes produce different individual responses; non-response to one is not non-response to the class, and switching is a reasonable clinical step. (c) "They work like Botox." OnabotulinumtoxinA is mechanistically and pragmatically different — quarterly injections of 31–39 sites in head and neck, narrower mechanism (SNARE-mediated reduction of neuropeptide release including but not limited to CGRP), and approved only for chronic migraine. Both are effective and often used sequentially or in combination. (d) "Insurance now covers them as first-line." The AHS 2024 statement is a recommendation, not a binding rule. Coverage is improving (~45% of US commercial plans drop step-therapy requirements as of mid-2024) but most patients still face prior authorisation and a step-therapy trial of older drugs.

Practicalities

US list prices in 2024–2026: roughly $700–$815 per month for the injectable mAbs, ~$1,150/month for atogepant, and ~$2,000 per quarterly eptinezumab infusion. Without insurance the annual cost is roughly $8,000–$12,000. With commercial insurance most patients pay a specialty-tier copay ranging from $0 (with manufacturer copay assistance) to several hundred dollars per month. Manufacturer patient-assistance programs (BridgeProgram, Ajovy Pay-As-Little-As, Emgality Savings, Aimovig Co-Pay) bring the out-of-pocket cost to ~$0–$5/month for most commercially-insured patients; Medicare patients are typically ineligible for these and face higher out-of-pocket costs unless on extra-help. Prior authorisation is the dominant access barrier in practice: insurers typically require documentation of ≥4 monthly migraine days plus failure of 2+ older preventives, despite the 2024 guideline. Botox-first requirements persist on some chronic-migraine plans. Real-world adherence at 12 months is roughly 50–60% for CGRP mAbs (substantially better than ~25% for topiramate at the same horizon), with discontinuations mostly for non-response rather than side effects.

Failure modes

(a) Non-response. Roughly 30–40% of patients do not achieve a 50% reduction by month 3; switching within the class (especially from a receptor-binder to a ligand-binder or vice versa) recovers response in a meaningful subset. (b) Insurance denial. The most common failure mode in the US. Documenting prior preventive failures (specific drugs, doses, durations, tolerability issues) before requesting a CGRP drug shortens the appeals process. (c) Medication-overuse headache: patients on daily NSAIDs or weekly triptans need concurrent acute-medication withdrawal; the CGRP drug masks but does not fix the rebound pattern by itself. (d) Discontinuation rebound. Stopping a CGRP mAb after sustained good response usually leads to a return toward baseline frequency over 3–6 months; some patients re-respond on restart, some have diminished response.

Stakes

Untreated frequent migraine carries a measurable disability burden — migraine is the second-leading cause of years lived with disability in adults under 50 in Global Burden of Disease estimates. Chronic migraine (≥15 headache days/month) is associated with depression, anxiety, sleep disturbance, occupational impairment, and increased healthcare utilisation. Medication-overuse headache from rescue-drug overuse is a downstream complication of inadequate prevention. The expected delta from successful CGRP prevention is fewer than half the prior monthly migraine days for ~half of patients — a substantial change in lived functional capacity, not a marginal one.

Out of scope

Acute migraine abortion (triptans, ditans, acute gepants — ubrogepant, rimegepant on-demand, zavegepant nasal); onabotulinumtoxinA for chronic migraine; legacy oral preventives (topiramate, propranolol, amitriptyline, candesartan, valproate); lifestyle / trigger management (sleep regularity, hydration, caffeine pattern, identified food and sensory triggers); paediatric migraine prevention; pregnancy-specific migraine management.

Credibility range

Optimist case

The first true migraine-specific preventive class, validated by direct experimental human evidence linking CGRP to attacks. Six FDA-approved drugs, dozens of phase 3 trials with consistent results across episodic and chronic migraine, real-world cohorts confirming trial efficacy, long-term safety out to 5+ years with no emergent adverse signals beyond the labelled hypertension/constipation concerns. Onset within weeks (immediately for eptinezumab IV); tolerability orders of magnitude better than topiramate or amitriptyline; minimal drug interactions for mAbs. The AHS first-line recommendation reflects field consensus that these are the best preventives available. The most compelling individual datum: ~50% of patients in pivotal trials achieved ≥50% reduction in migraine days — a class of magnitude rarely seen in CNS prophylaxis.

Skeptic case

Placebo-subtracted effect sizes are 1.5–2.5 days per month — clinically meaningful but not transformative on average. Trial placebo arms also dropped 2–3 migraine days, suggesting a large regression-to-the-mean and placebo component baked into the apparent benefit; the marginal CGRP-specific contribution is smaller than the headline numbers suggest. A 2025 cost-effectiveness analysis found no clear advantage over topiramate or onabotulinumtoxinA when both treatments and acute-care offsets were considered. Long-term cardiovascular outcomes are unknown — CGRP is a protective vasodilator during ischaemia, and the trial-excluded populations (significant CV disease) are exactly those who might be at risk. Pregnancy registries are still small. List prices are 10–50× legacy preventives; the broader public-health calculus depends on whether real-world responders are over-represented among trial enrollees.

Author's call

The class is genuinely effective, well-tolerated, and a meaningful upgrade over legacy preventives — strong enough to merit the AHS first-line classification. The honest framing is that the average effect is "two fewer migraine days a month" and the responder framing is "half of patients halve their attacks"; neither is a cure and neither is hype. Cost and access remain the dominant friction. Cardiovascular long-term safety is an open question worth tracking but not a current contraindication for patients without baseline disease. The right action verb is decide — this is a clinician-collaboration decision, not a self-start supplement.

Stakeholder and incentive map

Pharma manufacturers (Amgen/Novartis: erenumab; Teva: fremanezumab; Eli Lilly: galcanezumab; Lundbeck: eptinezumab; AbbVie/Allergan: atogepant; Pfizer/Biohaven: rimegepant) drive direct-to-consumer advertising and physician education; commercial incentive obvious.
Headache specialists and the American Headache Society have moved aggressively to first-line status; some critics note overlap between guideline authors and industry consulting relationships, though the data backing the recommendation is independent of that relationship.
Insurers and PBMs push back through step-therapy and prior authorisation; step-therapy requirements typically demand topiramate and one of {propranolol, amitriptyline} or onabotulinumtoxinA-first for chronic migraine.
Patient advocacy (American Migraine Foundation, Global Healthy Living Foundation, Coalition for Headache and Migraine Patients) pushes for first-line coverage; advocacy is well-funded and effective but downstream-aligned with pharma incentives.
Primary care remains under-educated on the class; most prescribing happens through headache specialists or neurology, creating an access bottleneck independent of cost.

Population variability

Sex skew: trials are ~85% female, mirroring real-world migraine prevalence. Response rate is not strongly sex-modulated in published analyses. Age: trials enrolled adults 18–65 mostly; older adults are under-represented but no clear safety signal in the 65+ subgroup data. Race: trials are predominantly white (75–85%); efficacy in Black and Hispanic populations is not strongly differentiated in subgroup analyses but power is limited. Migraine subtype: aura and non-aura migraineurs respond similarly; chronic migraineurs (≥15 days/month) achieve larger absolute day reductions but smaller percent reductions. Prior preventive failure: harder-to-treat patients (2+ failed preventives) respond at lower rates but still meaningfully above placebo Reuter et al. 2018. Medication-overuse headache patients respond well, especially when acute drugs are concurrently withdrawn. BMI, age of migraine onset, and aura status are not robust response predictors.

Knowledge gaps

Long-term cardiovascular outcomes (10+ years) are unknown — the post-marketing surveillance period is now ~7 years (erenumab approved May 2018) but cardiovascular events accumulate slowly and the patient population is young. Pregnancy outcomes are accumulating in registries but the sample is still small. Head-to-head trials between CGRP drugs are largely absent — drug choice within the class is empirical. Mechanistic predictors of responder vs non-responder status are not yet validated for clinical use. Combination with onabotulinumtoxinA is widely practised but trial evidence is limited to small studies. Paediatric efficacy and safety data are emerging but not yet sufficient for routine adolescent use. The optimal duration of treatment after sustained response — whether to attempt discontinuation, taper to lower frequency, or continue indefinitely — has no rigorous evidence base.

Scope vs brief. The brief named migraine day frequency, severity, acute medication use, side-effect profile, and access. The article covers all five end to end, with frequency carrying the evidence section, severity and acute-medication use folded into the payoff narrative, side effects in contraindications and failure-modes, and access in practicalities. No silent narrowing.

Category placement. Chose medical over supplements or mental: this is prescription-only, requires a neurologist, and sits in the medical-care decision flow. Pairs with action: decide.

Rating difficulties.

energy / focus / sleep / mood were the hardest calls — every effect is indirect through reduced attack burden rather than primary action. Settled on 2 / 2 / 1 / 2 to reflect "real but mediated"; resisted the urge to score these higher because the average migraineur is not the average reader, and the indirect effect only materialises for responders.
cost_burden 4 vs 5: list prices put this in the $8K–$12K/year range, which is the 4 band ($2K–$10K) overlapping into 5 (>$10K). Landed on 4 because (a) commercial-insured patients with copay assistance typically pay near zero, and (b) the upper end of list pricing is mostly hit by Medicare patients. If the catalogue's reader is the median US adult on commercial insurance, 4 reflects real out-of-pocket; 5 would over-weight uninsured cost.
controversy 2: efficacy itself is not contested. The 2 captures cost-effectiveness debate and the open long-term cardiovascular safety question — both real, neither dominant. Not 3 because there is no field-level disagreement about whether the drugs work.
evidence 5: meets the §6 bar (2+ rigorous trials per drug, guideline-backed, broad consensus on efficacy). The open CV question doesn't dent evidence — that's a controversy axis concern.
cadence daily: forced choice. The class spans monthly/quarterly injections, every-other-day pills, and daily pills. Treated as a lifelong-maintenance medication, which the spec's daily definition explicitly includes.

Contraindications. Included pregnancy, cardiac-condition, uncontrolled-hypertension. Skipped breastfeeding — data is genuinely mixed across the class and the call is "discuss with prescriber" rather than a blanket avoid; this didn't fit the closed-vocabulary signal. Skipped blood-thinners — no significant interaction.

Drug-by-drug detail. Article names all six drugs by generic and brand once each in the protocol section; the evidence section anchors numbers to specific trials. Resisted writing a separate section per drug — that would have been a buyer's guide, not a prevention overview.

Excluded.

Acute-use gepants (ubrogepant, zavegepant nasal, rimegepant when used on-demand) — separate decision flow, pointed at in out-of-scope.
OnabotulinumtoxinA — adjacent class; flagged in out-of-scope as a future entry.
Paediatric migraine prevention — under-studied; not a body-handbook reader audience.
Detailed head-to-head drug comparisons within the class — head-to-head trials largely don't exist, so a buyer's guide would be speculation.

Separate-entry candidates.

Onabotulinumtoxin A for chronic migraine — different mechanism, similar decision flow, worth its own entry.
Medication-overuse headache — common, often missed, sits underneath every migraine-prevention failure case. Strong candidate.
Triptans for acute migraine — the natural pair to a preventive entry.
Migraine trigger management — sleep / hydration / caffeine pattern — would pull together a behavioural-prevention entry.

Future links. Wire related ids once these companion entries exist. Left related empty for now rather than pointing at unrelated entries.

Open question for review. Whether to flag erenumab's specific safety profile (hypertension, constipation) as the contraindications focus and downgrade other drugs in the class. Current draft does this in contraindications and failure-modes, but the class label is unified. If reviewer prefers a per-drug split, that would mean restructuring the contraindications callout.