A Comprehensive Literature Review
Prepared: 26 October 2026 Subject guideline: Potrebic S, Tanveer S, Becker WJ, et al. Pharmacologic Treatment for Migraine Prevention in Adults Practice Guideline Recommendations: Report of the AAN Guidelines Subcommittee and the American Headache Society. Neurology. 2026;107(7):e214881. PMID 42673560. Companion systematic review: Pringsheim T, Smith DB, Tanveer S, et al. Neurology. 2026;107(4):e218112. PMID 42673559; simultaneous publication in Headache. 2026;66:1971–1990. PMID 42673583. Style: Vancouver numbered citations, numbered in order of first appearance. Every cited PMID was mechanically resolved through NCBI E-utilities esummary on 26 October 2026; unresolved identifiers were excluded. Evidence window: Literature through 6 June 2024 for the guideline’s own systematic review; this review additionally covers trials, meta-analyses, and safety reports published through October 2026.
Abstract
Migraine is among the leading causes of years lived with disability worldwide, affects roughly one in seven adults, and remains substantially undertreated: in a representative US survey, only 13% of people who met criteria for preventive treatment were taking a daily preventive medication [1,2,3]. In August 2026 the American Academy of Neurology (AAN) and the American Headache Society (AHS) published a joint practice guideline on pharmacologic migraine prevention in adults, replacing the 2012 AAN/AHS episodic-migraine guideline and extending scope to chronic migraine, pregnancy and lactation, older adults, sex-specific considerations, comorbidity, and medication overuse [4,5]. The guideline rests on a companion systematic review of 217 studies identified through 6 June 2024, and issues 17 recommendations comprising 52 statements (13 Level A, 34 Level B, 5 Level C according to independent summaries) [4,N3,N4].
The guideline’s central structural decision is to abandon a single ranked drug ladder. It sets a treatment threshold — four or more migraine days per month, four or more moderate-to-severe headache days per month, or substantial migraine-related disability — and then asks which property the patient most needs optimized: efficacy, tolerability, long-term safety, or cost. For efficacy, the named options are atogepant, eptinezumab, erenumab, fremanezumab, galcanezumab, propranolol, topiramate, and valproate in episodic migraine, with onabotulinumtoxinA added and propranolol removed in chronic migraine. For tolerability, the CGRP-targeted agents dominate. For long-term safety, only propranolol and topiramate (episodic) or onabotulinumtoxinA and topiramate (chronic) qualify. The guideline advises waiting 8–12 weeks at a tolerated dose before judging most drugs and 24 weeks before judging onabotulinumtoxinA, and it makes counselling on adverse effects and monitoring them Level A obligations [4,N3,N4].
This review places those recommendations against the primary trial evidence. The CGRP monoclonal antibodies and gepants produce placebo-subtracted reductions of roughly 1–3.5 monthly migraine days, with 50% responder rates commonly 40–65%; head-to-head data remain sparse, but where they exist they favor CGRP-targeted therapy over topiramate on tolerability (TEMPLE: 12% vs 30% discontinuation; HER-MES: 10.6% vs 38.9%) [6,7,8]. Conventional oral drugs retain a role — particularly where cost, comorbidity, or long-term safety dominate — but their effect sizes are smaller and their real-world persistence is poor [9,10]. The most conspicuous evidence-policy tension is candesartan: the guideline found insufficient evidence, while a 457-patient randomized trial published in 2025 — after the guideline’s search cutoff — showed a placebo-subtracted reduction of 1.22 migraine days per month [11]. The review concludes that the 2026 guideline is a defensible, patient-centered synthesis whose main vulnerabilities are the absence of credible long-term comparative data, a shrinking placebo-control margin, unresolved pregnancy safety for CGRP-targeted drugs, and the cost-access gap that determines whether its preferred options are reachable in practice.
1. Introduction: Why This Guideline Matters
Migraine is not a benign recurrent headache. It is a chronic neurological disorder with a genetic and neurovascular substrate, a prodrome that may begin hours before pain, an attack phase mediated largely by calcitonin gene-related peptide (CGRP) signalling within the trigeminovascular system, and an interictal burden that includes cognitive fog, anxiety, and lost productivity [12]. The Global Burden of Disease 2023 analysis estimated that 2.9 billion people — 34.6% of the world’s population on an age-standardized basis — had a headache disorder in 2023, with migraine alone accounting for 487.5 years lived with disability per 100,000 population and a female-to-male disability ratio of more than two to one [1]. Prevalence has been stable for three decades; what has changed is the therapeutic armamentarium [1,13].
That change is the reason a new guideline was necessary. The 2012 AAN/AHS guideline was built on 29 Class I or Class II studies and recommended, at Level A, divalproex sodium, sodium valproate, topiramate, metoprolol, propranolol, and timolol for episodic migraine prevention [5]. It could not address CGRP-targeted therapy, because the first positive phase 2 trial of erenumab was published four years later [14]. Between 2017 and 2024, four CGRP monoclonal antibodies and two oral CGRP receptor antagonists (gepants) were developed for prevention, generating a trial literature large enough to support network meta-analysis and, eventually, a joint specialty-society guideline [4].
The 2026 guideline is therefore both a clinical document and an epistemological one. It answers a question that patients ask constantly — “which preventive should I take?” — in a field where the honest answer is that several drugs work, none works dramatically better than the rest for the average patient, and the differences that matter most are tolerability, safety, cost, and the patient’s own priorities. The panel’s decision to structure recommendations around those priorities rather than around a single efficacy hierarchy is the most important thing about the guideline, and the part most likely to be misread.
This review has four aims. First, to describe what the guideline actually recommends, separating the recommendation statements from the evidence that supports them. Second, to reconstruct the trial evidence base — the CGRP monoclonal antibodies, the gepants, the conventional oral drugs, and onabotulinumtoxinA — with attention to effect sizes, responder rates, and the limits of placebo-controlled designs. Third, to examine comparative effectiveness, safety, adherence, and the special populations the guideline addresses. Fourth, to identify where the guideline’s conclusions are likely to age poorly, either because the search window closed before important trials reported or because the underlying evidence is thinner than the recommendation language implies.
A note on nomenclature. The 2026 document is a joint AAN/AHS guideline. There was no 2021 AAN guideline; the relevant 2021 AHS product was a consensus statement on integrating new migraine treatments into practice, published in Headache in 2021 and corrected in 2022 [15,16]. The 2026 guideline supersedes the 2012 AAN/AHS pharmacologic prevention guideline; the separate 2012 complementary and NSAID guideline is not addressed here [5].
2. Scope, Methods, and the Structure of the 2026 Guideline
2.1 What the panel did
The guideline was developed under the AAN’s 2017 Clinical Practice Guideline Process Manual, a framework that grades both the confidence in the evidence and the strength of the obligation, and that explicitly permits recommendations grounded in indirect evidence, principles of care, and inference when direct trial data are incomplete [4]. The companion systematic review searched MEDLINE, Embase, and ClinicalTrials.gov from database inception through 6 June 2024, and included 217 studies [17,18]. Outcomes were change in monthly headache days, the proportion of patients achieving at least a 50% reduction in monthly migraine days, and validated patient-reported quality-of-life measures [17].
The review’s inclusion criteria were broad — randomized trials of pharmacologic prevention in adults with episodic or chronic migraine — but the evidentiary yield was uneven. For CGRP monoclonal antibodies the review had multiple large, multinational, placebo-controlled trials per drug. For older oral agents it had trials dating from the 1980s and 1990s, many small, many with methodological limitations, and few with modern responder-rate reporting. For some drugs that are widely used in practice — candesartan most conspicuously — the review concluded that the evidence was insufficient to support a recommendation, despite a plausible trial literature [4,11].
2.2 Levels of obligation versus confidence in evidence
The AAN system uses levels of obligation, not the American Heart Association’s Class of Recommendation/Level of Evidence scheme. “Must” corresponds to Level A (very strong), “should” to Level B (strong), and “may” to Level C (weak). Separately, the systematic review grades confidence in the evidence for efficacy as high, moderate, or low [4]. These are different axes: a Level A statement can rest on low-certainty evidence if the obligation is primarily a duty to inform or to avoid harm, and a Level B statement can rest on high-certainty trial data if patient preferences are likely to vary.
This distinction matters for interpretation. The pregnancy recommendations carry the densest concentration of Level A statements precisely because the evidence is weakest — randomized trials of drug safety in pregnancy are not ethically feasible — but the duty to counsel, to avoid known teratogens, and to minimize exposure is categorical [4,N3,N4]. Conversely, the statement that patients should be offered preventive treatment at four or more migraine days per month is Level B, not because the evidence is weak but because the threshold is a value judgment about when the burden of daily medication is worth the benefit.
Table 1. Structure of the 2026 AAN/AHS guideline
| Domain | Highest obligation | Core content |
|---|---|---|
| Deciding to start prevention | Level B | Four-day threshold; disability trigger; inform all patients that effective prevention exists |
| Shared decision making | Level B | Medical/psychiatric history; adverse-effect preferences; oral vs injectable preference |
| Drug choice without comorbidity | Level B | Four priority lists (efficacy, tolerability, long-term safety, cost) plus low-confidence list |
| Pregnancy and planning | Level A | Teratogen avoidance; maximize non-drug care; risk discussion; limited exposure; nifedipine, metoprolol/propranolol, onabotulinumtoxinA (Level C) |
| Lactation | Level A | Counsel on milk transfer using LactMed-type sources |
| Older adults | Level A | Vascular disease, interactions, renal/hepatic function; sedation; hypotension |
| Sex-related factors | Level A | Valproate and PCOS; topiramate and hormonal contraception; urinary retention in older men |
| Comorbidity | Level B/C | Fibromyalgia (amitriptyline); obesity (topiramate); hypertension (enalapril, nifedipine, telmisartan); combined vs independent treatment |
| Medication overuse | Level B | Offer prevention; prefer agents with MOH evidence; prevention alone may suffice |
| Acute-treatment interactions | Level A | Drug interactions and additive side effects |
| Timing of assessment | Level B | 8–12 weeks; 24 weeks for onabotulinumtoxinA; dose optimization |
| How to assess | Level A | Frequency, severity, associated symptoms, QOL, acute use; patient-centric goals |
| Adverse effects | Level A | Counsel before prescribing; monitor at follow-up |
| Stopping | Level B | Relapse risk; discuss tapering after 6 months |
Sources: guideline abstract and recommendation text [4], with obligation levels as reproduced by two independent guideline summaries [N3,N4]. The full text was subscription-only; see Section 19.
2.3 Definitions
The guideline adopts the standard definition of chronic migraine: headache on 15 or more days per month for more than three months, of which at least eight days have migraine features [4,19]. Episodic migraine is everything below that threshold. Medication overuse is defined as regular use of acute or symptomatic medication on more than 9 days per month for prescription migraine-specific drugs, or more than 14 days per month for nonspecific analgesics [4]. These definitions are not neutral: they determine who enters trials, who qualifies for CGRP-targeted therapy under payer rules, and how “response” is counted.
One definitional subtlety deserves emphasis. The guideline’s four-day threshold refers to migraine days or moderate-to-severe headache days, not to “headache days” of any severity. A patient with 20 mild headache days and two migraine days does not automatically meet the threshold, while a patient with four severe migraine days and substantial disability does. The disability clause is the escape valve, and it is the clause most likely to be underused in primary care.
3. The Clinical Problem: Burden, Epidemiology, and the Prevention Gap
3.1 How common, and how disabling
The Global Burden of Disease 2023 headache analysis is the most current global estimate. In 2023, 2.9 billion individuals (95% uncertainty interval 2.6–3.1 billion) were affected by headache disorders, with an age-standardized prevalence of 34.6% (31.6–37.5) and a years-lived-with-disability rate of 541.9 per 100,000 (373.4–739.9); migraine accounted for 487.5 per 100,000 [1]. The female YLD rate was 739.9 per 100,000 versus 346.1 in males [1]. Prevalence has been essentially flat since 1990, which is itself an indictment: the disability has not fallen despite three decades of pharmacologic progress.
A separate narrative synthesis of 357 prevalence studies estimated active headache disorder in 52.0% of the population (95% CI 48.9–55.4), migraine in 14.0% (12.9–15.2), tension-type headache in 26.0% (22.7–29.5), and headache on 15 or more days per month in 4.6% (3.9–5.5) [13]. These figures are higher than GBD’s because they use different case definitions and capture “any headache” rather than moderate-to-severe migraine; the migraine estimate, however, is consistent.
In the United States, the American Migraine Prevalence and Prevention study mailed a validated questionnaire to 120,000 households and analyzed 162,576 individuals. One-year period prevalence of migraine was 11.7% (17.1% in women, 5.6% in men); 31.3% had three or more attacks per month, and 53.7% reported severe impairment or bed rest [3]. Using an expert-panel threshold, 25.7% met criteria for “offer prevention” and a further 13.1% for “consider prevention.” Only 13.0% were taking a daily preventive medication [3]. That gap — roughly a quarter of people with migraine who should be offered prevention, and one in eight actually treated — is the background against which the 2026 guideline’s four-day threshold should be read.
3.2 The chronic migraine problem
Chronic migraine is the expensive, disabling tail of the distribution. It is associated with medication overuse, psychiatric comorbidity, sleep disturbance, lower employment, and higher health-care utilization [19]. The guideline’s decision to issue separate episodic and chronic drug lists reflects a real difference in evidence: onabotulinumtoxinA has a regulatory indication and a large trial program in chronic migraine but failed in episodic migraine [20,21]; the CGRP monoclonal antibodies have positive trials in both, though the chronic-migraine effect sizes are larger in absolute terms because baseline attack frequency is higher [22,23].
Progression from episodic to chronic migraine is not inevitable, and the factors that predict it — attack frequency, medication overuse, obesity, depression, sleep apnoea, low educational attainment — are at least partly modifiable [24]. The International Headache Society’s 2025 position statement argues for earlier preventive treatment, starting at two to four monthly migraine days, on the theory that preventing progression is easier than reversing central sensitization [24]. The 2026 guideline stops short of that, setting the threshold at four days, but its disability clause and its emphasis on shared decision making leave room for earlier treatment in selected patients.
3.3 Why prevention is underused
The reasons are structural and psychological. Patients may normalize their attacks, fear daily medication, or have had a bad experience with an older drug. Clinicians may have limited time, may be uncertain which drug to choose, or may be constrained by formulary rules that require failure of multiple oral agents before CGRP-targeted therapy [15]. Cost is decisive in the United States, where list prices for CGRP monoclonal antibodies exceed $500 per month and gepants are similarly priced; in single-payer systems, prior-authorization criteria ration the same drugs [25]. The guideline cannot solve those problems, but its explicit cost recommendation — offer less expensive options depending on the patient’s formulary and insurance system — at least acknowledges them [4].
4. Pathophysiology and Pharmacologic Targets
4.1 The trigeminovascular system and CGRP
Migraine attacks are generated by activation of trigeminal afferents innervating the dura and large cerebral vessels, with release of CGRP, substance P, and pituitary adenylate cyclase-activating polypeptide (PACAP), neurogenic inflammation, and sensitization of second-order trigeminal neurons [12]. CGRP is the best-validated target: it is elevated in jugular venous blood during attacks, infusion produces migraine-like headache in susceptible individuals, and blockade prevents attacks [12]. The CGRP pathway is therefore both a mechanistic explanation and a therapeutic strategy.
Two pharmacologic approaches dominate. Monoclonal antibodies bind either the CGRP ligand (fremanezumab, galcanezumab, eptinezumab) or the CGRP receptor (erenumab); they are large molecules with half-lives of roughly 2 weeks (erenumab) to 4 weeks (the others), given subcutaneously monthly or quarterly, or intravenously quarterly [26]. Gepants are small-molecule CGRP receptor antagonists, orally available, with half-lives measured in hours; they were developed first for acute treatment and later for prevention, where rimegepant is dosed every other day and atogepant once daily [27,28].
4.2 Why the distinction matters clinically
The pharmacokinetics explain several clinical differences. Monoclonal antibodies are eliminated by proteolytic catabolism rather than hepatic metabolism, so they have few drug–drug interactions and no renal or hepatic dose adjustment [26,29]. They cannot be stopped abruptly without a months-long washout, which matters in pregnancy planning and in suspected adverse reactions. Gepants are hepatically metabolized (atogepant primarily by CYP3A4), have more interaction potential, and wash out quickly [29]. Atogepant causes constipation and modest weight loss; rimegepant is comparatively neutral [28,30]. Neither class carries the cognitive, teratogenic, or metabolic baggage of topiramate and valproate, which is the main reason the guideline places them in the tolerability list.
4.3 Traditional mechanisms
The older drugs were repurposed, not designed. Topiramate blocks voltage-gated sodium channels, potentiates GABA-A currents, inhibits carbonic anhydrase, and antagonizes AMPA/kainate receptors; its mechanism in migraine is unknown but probably multifactorial [31]. Propranolol and other beta-blockers may act through beta-adrenergic blockade of cortical spreading depression, inhibition of nitric oxide synthesis, or modulation of locus coeruleus activity [32]. Amitriptyline is a tricyclic antidepressant with serotonergic and noradrenergic effects, anticholinergic burden, and sodium-channel blockade; its analgesic effect is independent of its antidepressant effect and occurs at lower doses [33]. Valproate increases GABAergic tone and may modulate cortical excitability, but its teratogenicity and metabolic effects have progressively narrowed its role [34]. Candesartan and telmisartan act on angiotensin II type 1 receptors; the mechanism in migraine is uncertain but may involve cerebral vasoregulation or modulation of the renin–angiotensin system in the trigeminovascular unit [35].
Table 2. Pharmacologic targets and practical consequences
| Target/class | Agents | Route/frequency | Key pharmacokinetics | Practical consequences |
|---|---|---|---|---|
| CGRP ligand | Fremanezumab, galcanezumab, eptinezumab | SC monthly/quarterly; IV quarterly | IgG, half-life 27–31 days; proteolytic clearance | No hepatic/renal adjustment; slow washout; minimal drug interactions |
| CGRP receptor | Erenumab | SC monthly | IgG2, half-life ~2 weeks | Same class advantages; constipation signal |
| CGRP receptor (small molecule) | Atogepant, rimegepant | Oral daily or every other day | CYP3A4 metabolism; half-life hours | Interaction potential; rapid washout; GI effects |
| Sodium channel/GABA/CA | Topiramate | Oral daily | Renal excretion; CYP inducer at high dose | Cognitive slowing, paraesthesia, stones, teratogenicity, contraceptive interaction >200 mg/d |
| Beta-adrenergic | Propranolol, metoprolol, timolol | Oral daily | Hepatic (propranolol) or renal (metoprolol) | Bradycardia, hypotension, fatigue; long safety record |
| Serotonin/noradrenaline | Amitriptyline | Oral nightly | Hepatic; anticholinergic | Sedation, dry mouth, constipation, urinary retention, QT |
| GABAergic/other | Valproate, divalproex | Oral daily | Hepatic; protein binding | Teratogenicity, PCOS, weight gain, tremor |
| Angiotensin receptor | Candesartan, telmisartan | Oral daily | Renal/hepatic | Hypotension, hyperkalaemia, dizziness |
Sources: [26,29,31,32,33,34,35].
5. The 2026 Recommendations: What the Guideline Actually Says
5.1 When to start
The first substantive recommendation is that clinicians should inform all patients with migraine that effective preventive treatments exist for frequent attacks [4]. This is not a trivial statement. Many patients have been told, explicitly or implicitly, that nothing can be done beyond acute treatment; the guideline makes that framing a target for correction.
The second is the threshold: offer preventive treatment to patients with four or more migraine days per month, or four or more moderate-to-severe headache days per month, to reduce headache frequency [4,N3,N4]. The third adds substantial disability from migraine as an independent trigger, regardless of day count [4,N3,N4]. All three are Level B.
The threshold is lower than the 2012 guideline’s implicit “frequent or disabling” framing but higher than the IHS 2025 proposal of two to four days [24]. It is also deliberately simple: a day count plus a disability question, usable in a five-minute primary-care visit. The evidence behind it is indirect — no trial randomized patients to “prevention at four days” versus “prevention at eight days” — but the epidemiology of progression and the burden data make it defensible [3,24].
5.2 Shared decision making
Three statements require clinicians to inform patients of appropriate choices in light of medical and psychiatric history, current medications, and contraindications; to discuss and understand preferences about adverse effects; and to discuss and understand preferences about modality — oral versus injectable [4]. These are Level B, and they are the operational heart of the guideline. The panel’s explicit reasoning, as summarized in the recommendation text, is that no single medication has been shown to be clearly superior for reducing attack frequency, and that head-to-head trials have generally not been of sufficient quality to establish superiority [4,N3,N4]. Once that is conceded, the choice necessarily becomes preference-sensitive.
5.3 Choosing a drug: the four lists
Recommendation 3 is the most quoted part of the guideline. It asks what the patient is optimizing for, and provides a separate named list for each priority. The lists differ by migraine type.
Table 3. The guideline’s priority-based drug lists
| Priority | Episodic migraine | Chronic migraine |
|---|---|---|
| Efficacy (high or moderate confidence) | Atogepant, eptinezumab, erenumab, fremanezumab, galcanezumab, propranolol, topiramate, valproate | Atogepant, eptinezumab, erenumab, fremanezumab, galcanezumab, onabotulinumtoxinA, topiramate, valproate |
| Tolerability (fewer side effects) | Atogepant, eptinezumab, erenumab, fremanezumab, galcanezumab | Atogepant, eptinezumab, erenumab, fremanezumab, galcanezumab, onabotulinumtoxinA |
| Long-term safety (widely used, long record) | Propranolol, topiramate | OnabotulinumtoxinA, topiramate |
| Cost | Lower-cost options per formulary/insurance — no named drugs | Lower-cost options per formulary/insurance — no named drugs |
| After higher-confidence options fail (low confidence) | Amitriptyline, flunarizine, metoprolol, pizotifen, rimegepant, telmisartan | Not separately listed in the summary |
Source: recommendation 3 as reproduced by independent guideline summaries [N3,N4]; all statements Level B [N3,N4].
Four features of this table deserve comment.
First, the CGRP-targeted drugs appear in both the efficacy and tolerability lists but not in the long-term-safety list. That is not a judgment that they are unsafe; it is a judgment that they have not been in use long enough for a “widely used for a long period” claim. The oldest CGRP antibody was approved in 2018, giving at most eight years of post-marketing experience — substantial, but shorter than propranolol’s five decades.
Second, topiramate is the only drug that appears in both the efficacy and long-term-safety lists for both migraine types, and it is absent from the tolerability list. This is an uncomfortable but honest placement: topiramate works, it has been used for decades, and it makes many patients miserable. The guideline does not hide that; it simply lets the patient decide which matters more.
Third, valproate appears in the efficacy list for both episodic and chronic migraine but in no other list. Given its teratogenicity, its PCOS risk, and its narrowing role in epilepsy, its presence is best read as “effective but rarely first-choice,” particularly in anyone who could become pregnant.
Fourth, rimegepant appears only in the low-confidence list for episodic migraine, despite being approved for prevention. That reflects the modest placebo-subtracted effect in its pivotal trial (−0.8 migraine days per month) and the limited long-term prevention database at the time of review [27]. Atogepant, with larger effects in both episodic and chronic migraine, sits in the efficacy and tolerability lists [28,36,37].
5.4 Assessing efficacy
The guideline sets a clock. For most preventive medications, clinicians should wait at least 8–12 weeks at the recommended tolerated dose before assessing efficacy; for onabotulinumtoxinA, they should wait 24 weeks [4]. If response is suboptimal by 8 weeks, the dose should be optimized to the maximum tolerated or pre-specified maximum. If there is no efficacy after 8–12 weeks and contributing factors are identified — inconsistent adherence, frequent trigger exposure — shared decision making may justify a longer observation period. If side effects are intolerable, dose reduction or an alternative should be offered. If response remains suboptimal after 12–24 weeks, an alternative preventive should be discussed [17].
The rationale is empirical: in the systematic review, superiority to placebo for oral medications was typically demonstrated between 8 and 12 weeks, most commonly at 12 weeks; onabotulinumtoxinA trials assessed response at 24 weeks; CGRP monoclonal antibody trials assessed at both 12 and 24 weeks [17]. The 24-week botulinum rule is particularly important because the PREEMPT trials showed that a substantial proportion of first-cycle non-responders responded in later cycles [20,38]. Judging botulinum at 12 weeks is a common and avoidable error.
5.5 How to assess
Assessment should document attack frequency; be patient-centric and based on individual goals (Level A); and use frequency, severity, migraine-associated symptoms, quality of life, and acute medication use (Level B) [4]. Clinicians may use a consistent instrument — headache diary, HIT-6, or MIDAS — across visits (Level C) [17]. The hierarchy is deliberate: the patient’s goals outrank the instrument, and the instrument is a memory aid rather than a surrogate endpoint.
5.6 Stopping
Two statements address discontinuation: counsel patients that limited evidence suggests a risk of increased headache days and decreased headache-related quality of life after stopping (Level B), and discuss the potential benefits and risks of tapering after six months of treatment (Level B) [4]. Only two discontinuation studies met the guideline’s design threshold. In one, randomized withdrawal of topiramate 50–200 mg after six months produced a mean increase of 1.19 migraine attack days per four weeks (95% CI 0.71–1.66), more acute medication use, and worse migraine-related quality of life [17,39]. In the other, more than half of galcanezumab month-6 responders lost their response five months after discontinuation [17,40]. The guideline does not mandate lifelong treatment; it mandates a conversation.
5.7 Special-population recommendations
These are summarized in Section 11. In brief: pregnancy carries four Level A statements (teratogen avoidance, maximizing non-drug care, risk discussion, minimizing exposure) and three Level C drug options (nifedipine first; metoprolol or propranolol if nifedipine is unsuitable; onabotulinumtoxinA for chronic migraine only) [N3,N4]. Lactation carries a Level A counselling statement [4]. Older adults carry a Level A statement about discussing hypotension and postural hypotension, plus Level B statements about vascular disease, interactions, renal/hepatic function, sedation, and urinary retention in men [4]. Comorbidity-directed choices are amitriptyline for fibromyalgia, topiramate for increased BMI, and enalapril, nifedipine, or telmisartan for untreated hypertension (the last Level C) [N3,N4]. Medication overuse is an indication to start prevention, not a contraindication, with preference for agents that have evidence in that population [N3,N4].
6. Evidence Base I: The CGRP Monoclonal Antibodies
6.1 Erenumab
Erenumab was among the first CGRP-pathway drugs to report positive phase 3 data. The STRIVE trial randomized 955 adults with episodic migraine to erenumab 70 mg, 140 mg, or placebo monthly for six months. Mean monthly migraine days fell by 3.2 and 3.7 in the two active arms versus 1.8 with placebo (P<0.001 for each), and at least a 50% reduction was achieved by 43.3% and 50.0% versus 26.6% [41]. Acute migraine-specific medication days fell by 1.1 and 1.6 versus 0.2, and both domains of the Migraine Physical Function Impact Diary improved significantly [41]. ARISE, a separate phase 3 trial of 577 patients, found a smaller but consistent effect: −2.9 versus −1.8 monthly migraine days, a treatment difference of −1.0 (95% CI −1.6 to −0.5) [42]. The difference between STRIVE and ARISE is a useful reminder that trial-level effect sizes vary with population, baseline frequency, and placebo response; the confidence intervals overlap.
In chronic migraine, a phase 2 trial of 667 patients randomized to placebo or erenumab 70 or 140 mg showed significantly greater reductions in monthly migraine days than placebo over weeks 9–12 [43]. The LIBERTY trial specifically enrolled patients who had failed two to four prior preventive treatments and showed that erenumab 140 mg produced a greater reduction in monthly migraine days than placebo over 12 weeks [44]. These “difficult-to-treat” populations are the ones most likely to receive CGRP-targeted therapy in practice, and the consistency of benefit across them is clinically important.
6.2 Fremanezumab
Fremanezumab was tested in two pivotal programs. HALO CM randomized 1,130 patients with chronic migraine to quarterly fremanezumab (675 mg at baseline, placebo at weeks 4 and 8), monthly fremanezumab (675 mg loading, then 225 mg), or placebo. Mean monthly headache days fell by 4.3 and 4.6 versus 2.5 (P<0.001 for both), and at least 50% reduction was achieved by 38% and 41% versus 18% [22]. HALO EM randomized 875 patients with episodic migraine; monthly migraine days fell from 8.9 to 4.9 with monthly dosing, from 9.2 to 5.3 with the single higher dose, and from 9.1 to 6.5 with placebo, giving placebo-subtracted differences of −1.5 and −1.3 days [45].
FOCUS is the most clinically informative fremanezumab trial because it enrolled 838 patients with documented failure of two to four preventive medication classes. Monthly migraine days fell by 3.7 with quarterly dosing and 4.1 with monthly dosing versus 0.6 with placebo — differences of −3.1 and −3.5 days, both highly significant [46]. The larger effect in a treatment-resistant population than in the general episodic population is counterintuitive but not new: patients with more frequent attacks have more room to improve, and placebo response is often lower in trials that enroll previously failed patients.
6.3 Galcanezumab
EVOLVE-1 randomized 858 adults with episodic migraine to galcanezumab 120 mg, 240 mg, or placebo monthly for six months. Monthly migraine headache days fell by 4.7 and 4.6 versus 2.8 (P<0.001), with no meaningful difference between doses [47]. EVOLVE-2, a global trial of 915 patients, found reductions of 4.3 and 4.2 versus 2.3, with group differences of −2.0 and −1.9 days [48]. Both trials reported significant improvements in 50%, 75%, and 100% responder rates and in quality-of-life measures, and discontinuation due to adverse events was under 5% across arms [47,48].
CONQUER extended the evidence to patients who had failed two to four preventive categories, showing that galcanezumab 120 mg (with a 240 mg loading dose) reduced monthly migraine days and improved responder rates compared with placebo in a population that included both episodic and chronic migraine [49]. The PERSIST trial added a Japanese phase 3 dataset with consistent results [50]. Across the program, the most common adverse events were injection-site reactions, and the efficacy plateau between 120 mg and 240 mg argues against routine dose escalation.
6.4 Eptinezumab
Eptinezumab is the only intravenously administered CGRP antibody, given every 12 weeks. PROMISE-1 randomized 888 patients with episodic migraine to 30, 100, or 300 mg or placebo. The 100 mg and 300 mg doses met the primary endpoint, with reductions of 3.9 and 4.3 monthly migraine days versus 3.2 with placebo; the 30 mg dose did not separate from placebo [51]. PROMISE-2, in 1,072 patients with chronic migraine, found reductions of 7.7 and 8.2 versus 5.6 over weeks 1–12 (P<0.0001 for both) [23].
DELIVER enrolled 891 patients with episodic or chronic migraine and two to four prior preventive failures. Monthly migraine days fell by 4.8 with 100 mg and 5.3 with 300 mg versus 2.1 with placebo, giving differences of −2.7 and −3.2 [52]. The PREVAIL open-label extension followed patients with chronic migraine for 104 weeks and reported sustained reductions in headache frequency, severity, and disability, with more than half of participants maintaining at least 50% or 75% reductions through multiple dosing intervals [53]. SUNRISE, a predominantly Asian chronic-migraine trial of 978 patients, reported reductions of 7.2 and 7.5 versus 4.8 monthly migraine days, with a between-group difference of −2.4 to −2.7 and a lower proportion of patients with migraine on day 1 [54].
Table 4. Pivotal randomized trials of CGRP monoclonal antibodies for migraine prevention
| Trial (year) | Drug and dose | N | Population | Primary result (monthly migraine days, change vs placebo) | ≥50% responder rate |
|---|---|---|---|---|---|
| STRIVE (2017) [41] | Erenumab 70/140 mg SC monthly | 955 | Episodic | −3.2/−3.7 vs −1.8 | 43.3%/50.0% vs 26.6% |
| ARISE (2018) [42] | Erenumab 70 mg SC monthly | 577 | Episodic | −2.9 vs −1.8 (diff −1.0) | — |
| Erenumab phase 2 (2017) [43] | Erenumab 70/140 mg | 667 | Chronic | Significant reduction vs placebo | — |
| HALO CM (2017) [22] | Fremanezumab 675 mg q3mo or 225 mg monthly | 1,130 | Chronic | Headache days −4.3/−4.6 vs −2.5 | 38%/41% vs 18% |
| HALO EM (2018) [45] | Fremanezumab 225 mg monthly or 675 mg single | 875 | Episodic | −1.5/−1.3 vs placebo (difference) | — |
| FOCUS (2019) [46] | Fremanezumab quarterly or monthly | 838 | Episodic + chronic, 2–4 prior failures | −3.7/−4.1 vs −0.6 | — |
| EVOLVE-1 (2018) [47] | Galcanezumab 120/240 mg monthly | 858 | Episodic | −4.7/−4.6 vs −2.8 | Significant |
| EVOLVE-2 (2018) [48] | Galcanezumab 120/240 mg monthly | 915 | Episodic | −4.3/−4.2 vs −2.3 | Significant |
| CONQUER (2020) [49] | Galcanezumab 120 mg (240 mg load) | — | 2–4 prior failures | Significant reduction vs placebo | Significant |
| PROMISE-1 (2020) [51] | Eptinezumab 100/300 mg IV q12wk | 888 | Episodic | −3.9/−4.3 vs −3.2 | — |
| PROMISE-2 (2020) [23] | Eptinezumab 100/300 mg IV q12wk | 1,072 | Chronic | −7.7/−8.2 vs −5.6 | — |
| DELIVER (2022) [52] | Eptinezumab 100/300 mg IV q12wk | 891 | 2–4 prior failures | −4.8/−5.3 vs −2.1 | — |
Dashes indicate values not extracted from the abstract in a form suitable for tabulation. All trials were randomized, double-blind, placebo-controlled.
6.5 Class-level synthesis
Four conclusions are supportable across the monoclonal antibody program.
First, the class works. Each phase 3 program in episodic and chronic migraine met its primary endpoint at the tested dose, with placebo-subtracted reductions typically between 1.0 and 3.5 monthly migraine days and 50% responder rates between 40% and 65% [41,42,43,44,45,46,47,48,49,50,51,52,53,54]. The absolute effects are larger in chronic migraine because baseline frequency is higher, not necessarily because the drugs are more effective there.
Second, the effect is modest by conventional standards but meaningful to patients. A one-to-two-day reduction in monthly migraine days sounds small; a 50% responder rate of 40–50% does not. The apparent paradox is explained by heterogeneity: some patients respond dramatically, some not at all, and the average conceals the distribution. The guideline’s emphasis on individual goals rather than average effect sizes reflects this.
Third, tolerability is the class’s comparative advantage. In head-to-head trials against topiramate, CGRP-targeted therapy produced far fewer discontinuations due to adverse events (Section 10). The most common adverse events are injection-site reactions, constipation (particularly with erenumab and atogepant), and nasopharyngitis; serious adverse events are uncommon [55,56,57].
Fourth, long-term data are reassuring but not definitive. Open-label extensions and meta-analyses consistently show stable efficacy and low rates of adverse-event-related discontinuation over 12–24 months, but the number of patients followed beyond three years is small, and the trials excluded patients with recent cardiovascular or cerebrovascular events [58,59]. The guideline’s refusal to place CGRP-targeted drugs in the long-term-safety list is a statement about the limits of the evidence, not about a known hazard.
7. Evidence Base II: The Gepants
7.1 Rimegepant
Rimegepant was first developed for acute treatment and then tested for prevention at 75 mg every other day. The phase 2/3 prevention trial randomized 747 adults with episodic migraine; 695 were evaluable. Over weeks 9–12, monthly migraine days fell by 4.3 with rimegepant and 3.5 with placebo, a difference of −0.8 days (95% CI −1.46 to −0.20; P=0.0099) [27]. The effect is statistically significant but small, and it is the main reason rimegepant sits in the guideline’s low-confidence list rather than alongside atogepant [4].
The phase 4 trial was designed to answer a more practical question: does rimegepant work in patients who have already failed two to four conventional oral preventives? Among 652 treated patients, monthly migraine days changed by −2.1 with rimegepant and −0.5 with placebo across 12 weeks, a difference of −1.6 days (95% CI −2.1 to −1.2), and the proportion achieving at least 50% reduction in moderate-to-severe migraine days differed by 20.1 percentage points [60]. The larger effect in the treatment-resistant population again reflects lower placebo response and higher baseline burden.
Japanese phase 3 data (484 evaluable patients) met the primary endpoint, and a long-term open-label extension reported sustained safety and effectiveness [61,62]. Rimegepant’s safety profile in prevention trials has been close to placebo, with nausea and urinary tract infection the most common events and discontinuation rates of 1–2% [27,60].
7.2 Atogepant
Atogepant is the most extensively studied gepant for prevention and the only oral gepant that appears in the guideline’s efficacy and tolerability lists for both episodic and chronic migraine [4]. ADVANCE randomized 910 adults with episodic migraine to atogepant 10, 30, or 60 mg daily or placebo for 12 weeks. Monthly migraine days fell by 3.7, 3.9, and 4.2 versus 2.5, giving placebo-subtracted differences of −1.2, −1.4, and −1.7 days (P<0.001 for all) [28]. Constipation occurred in 6.9–7.7% and nausea in 4.4–6.1% of atogepant-treated patients [28]. The 60 mg dose is the recommended adult prevention dose and the one used in subsequent trials.
PROGRESS extended the evidence to chronic migraine. In 778 patients with a mean baseline of 18.6–19.2 monthly migraine days, atogepant 30 mg twice daily and 60 mg once daily reduced monthly migraine days by 7.5 and 6.9 versus 5.1 with placebo, differences of −2.4 and −1.8 [36]. Constipation occurred in 10.9% and 10% versus 3% with placebo, and clinically significant weight decrease (≥7%) occurred in 6% of each atogepant arm versus 2% with placebo [36].
ELEVATE enrolled patients who had failed two to four classes of conventional oral preventives. In 315 randomized patients, monthly migraine days fell by 4.2 with atogepant 60 mg versus 1.9 with placebo, a difference of −2.4 days (95% CI −3.2 to −1.5); constipation occurred in 10% versus 3% [37]. RELEASE, the Japanese phase 2/3 trial, found placebo-subtracted differences of −1.57, −1.90, and −2.10 days for 10, 30, and 60 mg, respectively [63]. The 52-week open-label trial and its extension showed sustained efficacy, with a mean weight loss of 2.16 kg at 52 weeks and 44.8% of participants experiencing at least 5% weight loss at any time [30,64].
Table 5. Pivotal randomized trials of gepants for migraine prevention
| Trial (year) | Drug and dose | N | Population | Primary result (monthly migraine days, change vs placebo) | Key adverse events |
|---|---|---|---|---|---|
| Rimegepant phase 2/3 (2021) [27] | Rimegepant 75 mg every other day | 747 | Episodic | −4.3 vs −3.5 (diff −0.8; P=0.0099) | Nausea; discontinuation 2% vs 1% |
| Rimegepant phase 4 (2025) [60] | Rimegepant 75 mg every other day | 652 | Episodic, 2–4 prior oral failures | −2.1 vs −0.5 (diff −1.6) | Similar to placebo |
| ADVANCE (2021) [28] | Atogepant 10/30/60 mg daily | 910 | Episodic | −3.7/−3.9/−4.2 vs −2.5 | Constipation 6.9–7.7%; nausea 4.4–6.1% |
| PROGRESS (2023) [36] | Atogepant 30 mg BID or 60 mg daily | 778 | Chronic | −7.5/−6.9 vs −5.1 | Constipation 10.9%/10% vs 3%; weight loss |
| ELEVATE (2024) [37] | Atogepant 60 mg daily | 315 | Episodic, 2–4 prior oral failures | −4.2 vs −1.9 (diff −2.4) | Constipation 10% vs 3% |
| RELEASE (2025) [63] | Atogepant 10/30/60 mg daily | 523 | Episodic (Japan) | Differences −1.57/−1.90/−2.10 | Constipation, nasopharyngitis |
7.3 Gepants versus monoclonal antibodies: a mechanistic and practical comparison
The two classes differ in ways that matter beyond efficacy. Monoclonal antibodies are given monthly or quarterly, have minimal drug–drug interactions, and cannot be rapidly discontinued; gepants are oral, can be stopped within days, and are useful when a patient wants to try prevention without committing to an injection. Gepants are also easier to use in patients with needle phobia or anticoagulation-related injection concerns. Against that, gepants require daily or every-other-day adherence, have more interaction potential through CYP3A4, and have less long-term safety data than the antibodies [26,29].
The only direct head-to-head trial compared galcanezumab with rimegepant in 580 patients with episodic migraine. At least 50% reduction was achieved by 62% with galcanezumab and 61% with rimegepant (P=0.70), and the trial was not powered to show superiority on secondary endpoints [65]. The result is best interpreted as evidence of comparable efficacy in a population that was not treatment-resistant, with the choice driven by route, frequency, tolerability, and cost rather than by an efficacy hierarchy.
8. Evidence Base III: Conventional Oral Preventives
8.1 Topiramate
Topiramate has the deepest randomized evidence base of any oral preventive. Two 2004 trials, published months apart, established its efficacy. Brandes and colleagues randomized 483 patients to topiramate 50, 100, or 200 mg daily or placebo for 26 weeks; monthly migraine frequency fell by 2.1 and 2.4 attacks with 100 and 200 mg versus 1.1 with placebo, and 50% responder rates were 49% and 47% versus 23% [31]. Silberstein and colleagues randomized 487 patients and found that monthly migraine frequency fell from 5.4 to 3.3 with 100 mg and from 5.6 to 3.3 with 200 mg, versus 5.6 to 4.6 with placebo; responder rates were 54.0% and 52.3% versus 22.6% [66].
The effect is real but the tolerability burden is substantial: paraesthesia, fatigue, nausea, anorexia, taste perversion, cognitive slowing, word-finding difficulty, kidney stones, metabolic acidosis, and teratogenicity [31,66]. In the guideline’s framework, topiramate is an efficacy option and a long-term-safety option, but not a tolerability option — an unusually honest triangulation. The Canadian Headache Society’s 2024 update went further and downgraded topiramate to a weak recommendation for episodic migraine, citing the balance of benefits and harms [67].
8.2 Beta-blockers
Propranolol is the best-studied beta-blocker. A 1989 trial of long-acting propranolol 160 mg daily in 74 patients reduced monthly attacks by 48% at day 84 compared with placebo [68]. A 1984 crossover trial of 96 patients found headache index scores of 5.71 with timolol, 6.66 with propranolol, and 9.03 with placebo, with no significant difference between the two active drugs [69]. The 2004 Cochrane review pooled 58 trials and 5,072 participants and found clear short-term benefit versus placebo, with no clear differences against calcium antagonists or other beta-blockers [32].
The European Headache Federation’s 2024 re-appraisal is the most rigorous modern synthesis: 20 trials and 1,291 patients, with moderate-certainty evidence that propranolol reduces monthly migraine days by 1.27 (95% CI −2.25 to −0.30) and increases the proportion achieving at least 50% reduction (RR 1.65, 95% CI 1.41–1.93; absolute risk difference 179 per 1,000), alongside a small absolute increase in discontinuation due to adverse events (20 per 1,000) [9]. The authors conclude that propranolol remains a first-line agent because of its long record, low cost, and acceptable tolerability [9]. Metoprolol and timolol share the class effect and are included in the 2012 Level A list [5].
8.3 Amitriptyline
Amitriptyline’s evidence is thinner than its clinical use. The EHF re-appraisal identified only three eligible trials with 622 patients, finding moderate-certainty evidence of a 50% responder benefit (RR 1.60, 95% CI 1.17–2.19) and a small increase in adverse-event discontinuation (50 per 1,000) [33]. The drug is sedating, anticholinergic, and potentially cardiotoxic in overdose; it is best used at low dose (10–50 mg) at bedtime, and it is the guideline’s preferred option for comorbid fibromyalgia [4].
The 2016 melatonin trial is a useful comparator: in 196 patients, melatonin 3 mg reduced headache frequency by 2.7 days, amitriptyline 25 mg by 2.2 days, and placebo by 1.1 days; melatonin was better tolerated than amitriptyline and comparable in efficacy [70]. This is a reminder that “established” does not always mean “best studied.”
8.4 Valproate and divalproex
Divalproex sodium was established in the 1990s. A 1995 trial of 107 patients found a mean of 3.5 migraines per four weeks with divalproex versus 5.7 with placebo, with 48% versus 14% achieving at least 50% reduction [71]. A 1994 triple-blind crossover trial of 43 patients found 3.5 migraine days per four weeks with sodium valproate versus 6.1 with placebo, and 50% versus 18% responders [72]. The 2002 extended-release trial found a modest reduction in four-week migraine rate (1.2 versus 0.6) [73].
Valproate’s problem is not efficacy but safety: neural tube defects and other major malformations, developmental effects, polycystic ovary syndrome, weight gain, tremor, and thrombocytopenia [34]. The 2026 guideline requires clinicians to notify female patients of the PCOS risk (Level A) and to avoid valproate in anyone who could become pregnant if possible [4]. In practice, valproate is now a drug of last resort in women of childbearing potential and a reasonable option in men and in older adults who cannot tolerate other agents.
8.5 Candesartan, telmisartan, and the renin–angiotensin system
Candesartan is the most interesting discordance in the guideline. The 2026 document states that candesartan is widely used but that there is insufficient evidence to support its use, and it does not appear in any recommendation list [N3,N4]. Yet a 457-patient, triple-blind, placebo-controlled trial published in Lancet Neurology in 2025 — after the guideline’s 6 June 2024 search cutoff — found that candesartan 16 mg reduced migraine days by 2.04 versus 0.82 with placebo, a difference of −1.22 days (95% CI −1.75 to −0.70), with dizziness in 30% versus 13% [11]. A 2026 meta-analysis of three trials and 589 participants found a mean reduction of 1.13 monthly migraine days (95% CI −1.51 to −0.76) [74].
The most likely explanation is timing, not disagreement about the data: the guideline’s systematic review could not include a trial published more than a year after its search closed. This is a structural feature of guidelines, not a flaw unique to this one, but it has practical consequences. A clinician reading the guideline in 2026 may conclude that candesartan is unsupported, while the most recent randomized evidence says it works modestly and is cheap. The honest interpretation is that candesartan is probably effective, that the guideline’s statement reflects its evidence window, and that the next update should move it into at least the low-confidence list.
Telmisartan has weaker evidence. A 95-patient trial found a non-significant reduction of 1.65 versus 1.14 migraine days, with a post hoc adjusted analysis showing a 38% versus 15% reduction (P=0.03) and a borderline responder difference (40% versus 25%, P=0.07) [35]. The guideline lists telmisartan, enalapril, and nifedipine as options for comorbid untreated hypertension at Level C [4]. An older crossover trial found candesartan non-inferior to propranolol, with both superior to placebo [75].
8.6 Other oral agents
Flunarizine, pizotifen, and metoprolol appear in the low-confidence list; flunarizine and pizotifen are not available in the United States but are widely used in Europe and Canada [4]. Gabapentin was downgraded to a weak recommendation against use in episodic migraine by the Canadian Headache Society in 2024 [67]. Memantine, levetiracetam, enalapril, and melatonin received weak recommendations for episodic migraine in the same update [67]. A 2025 network meta-analysis of 44 trials and 4,612 participants found significant efficacy for topiramate, valproate, and propranolol and potential benefits for memantine, melatonin, vitamin D3, and combination therapies [76].
Table 6. Selected conventional oral preventive trials and syntheses
| Agent | Key evidence | Effect size | Tolerability notes |
|---|---|---|---|
| Topiramate | Two 26-week RCTs, n=483 and 487 [31,66] | 50% responder 47–54% vs 23%; migraine frequency −2.1 to −2.4 vs −1.1 | Paraesthesia, cognitive slowing, stones, teratogenicity |
| Propranolol | Cochrane 58 trials/5,072 [32]; EHF 20 trials/1,291 [9] | MMD −1.27 vs placebo; 50% responder RR 1.65 | Fatigue, bradycardia, hypotension |
| Metoprolol/timolol | 1984 crossover, n=96 [69]; 2012 Level A [5] | Headache index 5.71–6.66 vs 9.03 | Class effects |
| Amitriptyline | EHF 3 trials/622 [33]; melatonin RCT n=196 [70] | 50% responder RR 1.60 | Sedation, anticholinergic, urinary retention |
| Divalproex/valproate | 1995 RCT n=107 [71]; 1994 crossover n=43 [72] | 48% vs 14% and 50% vs 18% responders | Teratogenicity, PCOS, weight gain |
| Candesartan | 2025 RCT n=457 [11]; 2026 MA n=589 [74] | MMD −1.22 vs placebo; MA −1.13 | Dizziness 30% vs 13% |
| Telmisartan | 2009 RCT n=95 [35] | −1.65 vs −1.14 (NS); post hoc 38% vs 15% | Hypotension, hyperkalaemia |
9. Evidence Base IV: OnabotulinumtoxinA
OnabotulinumtoxinA is approved for chronic migraine and is the only preventive with a distinctly different population indication. The PREEMPT program randomized 1,384 patients with chronic migraine to 155–195 U every 12 weeks or placebo in a 24-week double-blind phase followed by a 32-week open-label phase. In the pooled analysis, 56.1% of onabotulinumtoxinA-treated patients versus 49.1% of placebo-treated patients had fewer than 15 headache days per month during weeks 13–24 (P=0.010); the proportion rose to 81.1% by weeks 53–56, and 79.4% achieved fewer than 15 headache days during any consecutive 12-week period [20].
The medication-overuse subgroup is clinically important. Among 904 patients with chronic migraine and acute medication overuse, onabotulinumtoxinA reduced headache days by 8.2 versus 6.2 with placebo at week 24 (P<0.001), with significant improvements in migraine days, moderate-to-severe headache days, cumulative headache hours, and HIT-6 category [77]. The per-cycle responder analysis showed that a meaningful proportion of first-cycle non-responders responded in the second or third cycle, which is the evidentiary basis for the guideline’s 24-week assessment rule [38].
PRECLUDE, a 775-patient phase 3 trial in episodic migraine, was negative: neither 155 U nor 195 U reduced monthly migraine days significantly more than placebo, and secondary endpoints including headache days and 50% responder rates were also non-significant [21]. The contrast between chronic and episodic migraine is one of the clearest examples of disease-state-dependent efficacy in the field, and it justifies the guideline’s decision to restrict onabotulinumtoxinA to chronic migraine in its efficacy, tolerability, and long-term-safety lists [4].
10. Comparative Effectiveness and Network Meta-Analysis
10.1 Head-to-head trials
Four head-to-head or active-comparator trials are informative, and all four point in the same direction: CGRP-targeted therapy is better tolerated than conventional oral prevention, with equal or greater efficacy. A fifth trial, in vestibular migraine, provides proof-of-concept in a distinct population.
HER-MES randomized 777 patients to erenumab or topiramate in a double-dummy design. Discontinuation due to adverse events occurred in 10.6% with erenumab and 38.9% with topiramate (OR 0.19, 95% CI 0.13–0.27), and at least 50% reduction in monthly migraine days was achieved by 55.4% versus 31.2% (OR 2.76, 95% CI 2.06–3.71) [7]. The trial is the strongest single argument for choosing a CGRP antibody over topiramate when both are accessible.
APPRAISE compared erenumab with nonspecific oral preventives in 621 patients who had failed one or two prior treatments. At 12 months, 56.2% of erenumab-treated patients versus 16.8% of oral-preventive patients completed the assigned treatment and achieved at least 50% reduction (OR 6.48, 95% CI 4.28–9.82); 2.9% versus 23.3% discontinued due to adverse events, and 2.2% versus 34.6% switched medication [8]. The composite primary endpoint — persistence plus response — captures the real-world problem better than a pure efficacy endpoint.
TEMPLE compared atogepant 60 mg with topiramate (50–100 mg) in 545 patients over 24 weeks. Discontinuation due to treatment-emergent adverse events was 12% with atogepant versus 30% with topiramate (RR 0.4, 95% CI 0.3–0.6); at least 50% reduction was achieved by 64% versus 39% (RR 1.6, 95% CI 1.4–2.0); and monthly migraine days fell by 6.3 versus 4.5 (difference −1.8, 95% CI −2.5 to −1.0) [6]. TEMPLE is the first trial to demonstrate both better tolerability and better efficacy for a gepant against topiramate, and it substantially strengthens the case for atogepant as a first-line oral option.
Galcanezumab versus rimegepant was a proof-of-concept comparison in 580 patients with episodic migraine; 50% responder rates were 62% and 61% (P=0.70), with no significant differences in adverse events [65]. The trial was not designed to show equivalence, and its confidence interval may conceal a clinically relevant difference, but it is the only direct comparison between a monoclonal antibody and a gepant.
Table 7. Selected comparative and proof-of-concept trials
| Trial (year) | Comparison | N | Primary outcome | Result |
|---|---|---|---|---|
| HER-MES (2022) [7] | Erenumab vs topiramate | 777 | Discontinuation due to AE | 10.6% vs 38.9% (OR 0.19); ≥50% response 55.4% vs 31.2% |
| APPRAISE (2024) [8] | Erenumab vs nonspecific oral preventives | 621 | Persistence + ≥50% response at 12 months | 56.2% vs 16.8% (OR 6.48); AE discontinuation 2.9% vs 23.3% |
| TEMPLE (2026) [6] | Atogepant vs topiramate | 545 | AE discontinuation over 24 weeks | 12% vs 30% (RR 0.4); ≥50% response 64% vs 39%; MMD −6.3 vs −4.5 |
| Galcanezumab vs rimegepant (2024) [65] | Monthly galcanezumab vs every-other-day rimegepant | 580 | ≥50% response | 62% vs 61% (P=0.70) |
| INVESTMENT (2024) [78] | Galcanezumab vs placebo in vestibular migraine | 40 | VM-PATHI change | −14.8 vs −5.1 (difference −9.6; P=0.044) |
10.2 Network meta-analyses
Network meta-analysis is the only way to compare drugs that have never been tested head-to-head, but it inherits the limitations of the included trials: heterogeneous populations, different placebo responses, different dose regimens, and inconsistent outcome definitions. Several recent syntheses are worth separating.
The American College of Physicians review (2025) included 61 trials and 20,680 patients evaluating 16 treatments for episodic migraine. It found moderate-certainty evidence that CGRP monoclonal antibodies resulted in fewer discontinuations due to adverse events than topiramate (risk difference −16.2%, 95% CI −18.4% to −12.8%), and low-certainty evidence that they reduced migraine-related disability and improved quality of life compared with gepants. For other comparisons, the evidence was moderate- or low-certainty of no clinically important difference, uncertain, or absent. The authors concluded that no high-certainty evidence favored one preventive over another [79].
The BMJ Evidence-Based Medicine network meta-analysis (2026) included 199 trials and 47,420 participants. Compared with placebo, calcium channel blockers (MD −1.78, moderate certainty), CGRP-targeted therapies (MD −1.69, high certainty), and beta-blockers (MD −1.50, moderate certainty) were the most effective for reducing monthly migraine days. Beta-blockers, calcium channel blockers, anticonvulsants, and CGRP-targeted therapies probably reduced monthly attack frequency. However, calcium channel blockers (RR 1.40) and anticonvulsants (RR 1.14) were more likely to be discontinued than placebo, whereas CGRP-targeted therapies were not [80]. The synthesis supports the guideline’s tolerability hierarchy while reminding us that calcium channel blockers (flunarizine, cinnarizine) are effective where available.
The prior-failure network meta-analysis (2026) included 18 trials and 7,281 patients with a history of preventive failure. Compared with placebo, low-certainty evidence suggested that fremanezumab (MD −3.30), eptinezumab (−3.35), galcanezumab (−2.73), atogepant (−2.30), and erenumab (−2.20) may be among the most effective for reducing monthly migraine headache days [81]. The overlapping confidence intervals and low certainty ratings mean that ranking should not be used to choose among them; the class effect is more robust than any individual drug effect.
A 2024 network meta-analysis of chronic migraine (24 trials, 8,789 patients) ranked onabotulinumtoxinA highest for reducing monthly migraine days and topiramate highest for 50% response, but the effect estimates were imprecise and some were implausibly large [82]. A 2024 Bayesian network meta-analysis of anti-CGRP antibodies (10 trials) ranked galcanezumab 120 mg highest for mean migraine-day reduction and fremanezumab quarterly highest for responder rate, with eptinezumab safest [83]. A 2026 episodic-migraine network meta-analysis (16 trials, 9,123 patients) ranked galcanezumab 240 mg highest for efficacy and fremanezumab 225 mg highest for 50% response, with erenumab having the best safety profile [84]. These rankings should be read as hypothesis-generating: the trials differ in baseline frequency, prior treatment exposure, and endpoint definitions, and network meta-analysis cannot fully adjust for those differences.
10.3 What the comparative evidence actually supports
Three conclusions are defensible.
First, CGRP-targeted therapy is better tolerated than topiramate, and probably better tolerated than most conventional oral preventives. The evidence is strongest for erenumab versus topiramate and atogepant versus topiramate, and it is consistent across network meta-analyses [6,7,79,80].
Second, there is no reliable efficacy hierarchy within the CGRP class. The head-to-head trial of galcanezumab versus rimegepant was negative for superiority, and network rankings are unstable across analyses [65,79,80,81,82,83,84]. Choosing among them should be driven by dosing interval, route, adverse-effect profile, pregnancy plans, and cost.
Third, the older drugs are not obsolete. Propranolol and topiramate have effect sizes that overlap with the CGRP class in some trials, decades of safety data, and low cost. The guideline’s long-term-safety list exists for a reason: a 25-year-old who plans pregnancy in five years, or a patient in a health system that cannot fund a monoclonal antibody, may reasonably choose propranolol first [9,17].
11. Special Populations
11.1 Pregnancy and lactation
Pregnancy is the guideline’s most prescriptive section and its most honest about uncertainty. Four Level A statements require clinicians to make anyone of childbearing potential aware of fetal risk in an unplanned pregnancy and to avoid known teratogens — divalproex sodium and topiramate — where possible; to maximize nonpharmacologic approaches in pregnancy or planned pregnancy; to use pharmacologic prevention only after a thorough risk-and-alternatives discussion with joint decision making; and, if prevention is necessary, to develop a strategy that limits overall exposure while maintaining control [N3,N4].
The named drug options are all Level C. Nifedipine is offered first, on the basis that the American College of Obstetricians and Gynecologists lists no potential associated risks, even though the AAN’s confidence in its efficacy is low [4]. If nifedipine is not an option or is ineffective, metoprolol or propranolol may be offered, with explicit balancing of cardiovascular abnormalities, cleft lip or palate, neural tube defects, and fetal growth restriction [4]. For chronic migraine only, onabotulinumtoxinA may be offered, with the caveat that outcome data are extremely limited and mainly pertain to preconception and first-trimester exposure; a Class III study of 397 exposed women found an overall fetal defect prevalence of 2.6%, consistent with background rates [4,85].
The CGRP antibodies are not recommended in pregnancy, and the evidence for concern has strengthened. A 2026 retrospective cohort using the AM-PREGNANT database analyzed 7,579 singleton pregnancies in 7,119 individuals with pregestational migraine. Early pregnancy exposure to an anti-CGRP monoclonal antibody was associated with an increased risk of spontaneous abortion (adjusted HR 1.42, 95% CI 1.12–1.80), while propranolol exposure was not (aHR 1.02, 95% CI 0.71–1.47) [86]. The absolute risk remains uncertain, and confounding by indication is possible — people who take a CGRP antibody may differ from those who do not — but the signal is consistent with the biological role of CGRP in placental development. Galcanezumab’s long half-life means that subtherapeutic concentrations can persist into the second trimester in a substantial proportion of maternal exposures [87].
Lactation carries a single Level A statement: clinicians must counsel that preventives pass into breast milk in varying amounts and can affect breastfed infants, using evidence-based sources such as LactMed [4]. Data are scarce for the newer drugs. A 2026 narrative review reported that phase 1 pharmacokinetic studies show minimal gepant transfer into breast milk (relative infant doses 0.04–0.19%, well below the 10% safety threshold), but no human lactation data exist for CGRP monoclonal antibodies [88].
11.2 Older adults
Three statements address older adults. Clinicians should assess vascular disease, drug interactions, and reduced renal or hepatic function depending on the drug (Level B); should discuss sedation and confusion with sedating preventives — amitriptyline, valproate, topiramate, and pizotifen — and consider lower starting doses (Level B); and must discuss hypotension or postural hypotension, avoiding drugs that significantly lower blood pressure or starting low (Level A) [4]. The last statement is supported by trials of telmisartan, propranolol, and metoprolol that showed significant blood-pressure reductions [4].
The evidence base in older adults is thin. A pooled analysis of fremanezumab trials in participants aged 60 or older found efficacy and safety consistent with younger patients [89], and a pooled analysis of erenumab trials across age groups found no new safety signals [90]. But these are subgroup analyses, not trials designed for older adults, and they exclude frail patients and those with multimorbidity. The guideline’s advice is therefore cautious and largely extrapolated.
A practical corollary not stated explicitly in the guideline: in older adults with polypharmacy, a monthly injectable with no hepatic or renal adjustment and few drug interactions may be safer than adding a fourth oral drug, even if the oral drug is cheaper. The guideline’s structure permits that reasoning under the tolerability and long-term-safety priorities, but it does not spell it out.
11.3 Sex-related factors
Two Level A statements address sex-specific harms. Clinicians prescribing valproic acid must notify female patients that it can increase the risk of polycystic ovary syndrome; in women with epilepsy, PCOS incidence was higher with valproate (OR 3.04, 95% CI 2.09–4.43) [4,91]. Clinicians prescribing topiramate must notify female patients of childbearing potential that it can make hormonal contraception less effective at doses above 200 mg per day [4]. The guideline notes that doses of 200 mg or less produce minimal or no reduction in estrogen and norethindrone, that doses above 200 mg can reduce etonogestrel concentrations in implant users, and that low-dose topiramate as used in migraine prevention has not been shown to increase unintended pregnancy compared with other oral preventives [4,92,93].
A third statement addresses older men: clinicians should evaluate the risk of urinary retention and either avoid anticholinergic drugs such as amitriptyline or discuss the possibility; tricyclics were significantly more likely than placebo to cause urinary retention (OR 2.52, 95% CI 1.29–4.90) [4].
11.4 Comorbidity-directed choices
The guideline’s comorbidity recommendations are among its most useful practical contributions. For fibromyalgia, amitriptyline is recommended (Level B); the supporting evidence includes reductions in sleep disturbance (SMD −0.97), fatigue (SMD −0.64), and improvement in quality of life (SMD −0.80) in fibromyalgia trials [4]. For increased body mass index, topiramate is recommended when an oral preventive is being considered (Level B), because it produces significant weight loss at 50–200 mg; the guideline also lists amitriptyline, propranolol, pizotifen, and flunarizine as drugs that promote weight gain [4].
For untreated hypertension, clinicians may inform patients that enalapril, nifedipine, and telmisartan can treat both conditions (Level C); candesartan is explicitly excluded for insufficient evidence [N3,N4]. The hypertension recommendation is logically attractive but evidentially weak — telmisartan’s trial was marginal, and enalapril and nifedipine have limited migraine-specific data [35]. The candesartan exclusion, as discussed in Section 8.5, is likely to be temporary.
For depression and anxiety, the guideline does not make a specific drug recommendation beyond the general principle of treating both conditions and monitoring closely. The UNITE trial showed that fremanezumab reduced monthly migraine days by 5.1 versus 2.9 and improved Hamilton Depression Rating Scale scores by 6.0 versus 4.6 in patients with migraine and major depressive disorder [94]. A 2026 meta-analysis of 12 studies and 3,603 patients found consistent reductions in depressive and anxiety symptoms during CGRP-targeted prevention [95]. These data support the use of CGRP-targeted therapy in patients with comorbid mood disorders, but they do not establish that it is superior to a conventional preventive plus an antidepressant.
11.5 Medication overuse
The guideline reframes medication overuse as an indication to start prevention rather than a barrier. Three Level B statements: offer preventive medication to people with migraine who meet criteria for medication overuse; offer it to those with medication-overuse headache; and prefer preventives with evidence in that population — CGRP monoclonal antibodies, atogepant, onabotulinumtoxinA, and topiramate — over drugs lacking such evidence [N3,N4].
The evidence behind the reframing is genuinely interesting. Studies suggest that starting a preventive alone is approximately as effective as starting a preventive plus acute-medication withdrawal, at least in the populations studied; the guideline cautions that those studies included only small numbers of people overusing opioids or barbiturates, so the conclusion should not be extrapolated to those drugs [4]. The RESOLUTION trial, published in 2026, randomized 608 patients with chronic migraine and medication-overuse headache to eptinezumab plus brief patient education or placebo plus education. Eptinezumab reduced monthly migraine days by 6.9 versus 3.7 in weeks 1–4, a difference of −3.2 (95% CI −4.2 to −2.2) [96]. The erenumab phase 4 program in chronic migraine with medication overuse showed that a substantial proportion achieved remission at month 6, with low relapse at one year [97,98]. The PREEMPT medication-overuse subgroup showed a similar pattern for onabotulinumtoxinA [77].
The practical implication is that clinicians should not withhold prevention until a patient “detoxifies.” That advice was common in the 2000s and is now contradicted by randomized evidence. The caveat about opioids and barbiturates is important, because those patients may still require supervised withdrawal.
12. Safety: What to Counsel, What to Monitor
12.1 CGRP monoclonal antibodies
The class is generally well tolerated. In pooled phase 3 data for fremanezumab involving more than 2,800 patients, overall adverse events occurred in 58–65% of patients and serious adverse events in less than 1–2%, with cardio- and cerebrovascular events in 2–6% regardless of baseline vascular history [55]. Pooled eptinezumab data from five trials and 2,867 adults found treatment-emergent adverse events in 54.8% versus 52.3% with placebo, infusion-site reactions in 1.3% versus 0.9%, and nasopharyngitis as the most common event in the 300 mg group [56].
Three safety questions deserve specific attention.
Hypertension. CGRP is a vasodilator, and theoretical concern about raising blood pressure is plausible. In pooled erenumab trials, hypertension adverse events occurred in 0.9% of placebo, 0.8% of 70 mg, and 0.2% of 140 mg patients; in the postmarketing database, 362 hypertension adverse events were reported over more than 245,000 patient-years of exposure, 26.2% serious, for an exposure-adjusted incidence of 0.144 per 100 patient-years [99]. The signal is small but real, and it justifies blood-pressure monitoring in patients with existing hypertension.
Cardiovascular and cerebrovascular events. A EudraVigilance analysis of 9,441 individual case safety reports found that cardiovascular reports accounted for 12.8% of the total, with disproportionate reporting of hypertension for erenumab (ROR 1.45, 95% CI 1.14–1.85), atrial fibrillation (2.36) and myocardial infarction (2.21) for galcanezumab, and pallor, deep vein thrombosis, hot flush, and palpitations for fremanezumab [100]. Spontaneous reporting cannot establish causality, and the denominator is uncertain. A pooled analysis of four eptinezumab trials found no meaningful differences in blood pressure or heart rate over 56 weeks [101]. A 2026 registry of 17 patients with prior cerebrovascular or cardiovascular disease treated with CGRP pathway inhibitors for a total of 31 patient-years found no recurrent events or new ischaemic lesions, but the sample is far too small to exclude a clinically important risk [59].
Long-term tolerability. A 2026 meta-analysis of 11 studies with at least 12 months of follow-up found a pooled discontinuation rate of 23% for any reason and 3% for adverse events, with no emergent safety signal and stable adverse-event profiles beyond the first year [58]. The included studies were judged at severe risk of bias, so the estimates should be read as reassuring rather than definitive.
12.2 Gepants
Atogepant’s characteristic adverse effects are constipation (roughly 7–11%) and nausea (4–6%), with a dose-related increase in clinically significant weight loss (6% with ≥7% loss in PROGRESS versus 2% placebo) [28,36]. Long-term extension data show a mean weight loss of 2.16 kg at 52 weeks and at least 5% loss in 44.8% of participants at any time, with higher baseline BMI predicting greater loss [30]. Rimegepant’s prevention trials reported adverse-event rates similar to placebo, with nausea the most common event and discontinuation around 1–2% [27,60].
Hepatic safety is a class consideration for gepants because of CYP3A4 metabolism. Atogepant is contraindicated in severe hepatic impairment and requires dose adjustment with strong CYP3A4 inhibitors or inducers; rimegepant has similar interaction concerns [29]. No signal of clinically important hepatotoxicity has emerged in the prevention trials, but post-marketing surveillance is shorter than for the antibodies.
12.3 Conventional oral preventives
The adverse-effect burden of the older drugs is the main reason the guideline’s tolerability list excludes them.
Topiramate causes paraesthesia (often the most bothersome), fatigue, nausea, anorexia, taste perversion, cognitive slowing, word-finding difficulty, mood change, kidney stones, metabolic acidosis, and acute angle-closure glaucoma (rare). It is teratogenic (oral clefts, hypospadias, small-for-gestational-age) and reduces hormonal contraceptive efficacy at doses above 200 mg/day [17,31,66,92]. Baseline and periodic electrolytes and renal function are reasonable in patients with risk factors.
Valproate causes weight gain, tremor, thrombocytopenia, hyperammonaemia, pancreatitis, and polycystic ovary syndrome; it is among the most teratogenic drugs in neurology, with neural tube defects and neurodevelopmental effects [34,91]. The guideline’s Level A statement on PCOS and its advice to avoid valproate in pregnancy are the minimum standard of care.
Amitriptyline causes sedation, dry mouth, constipation, blurred vision, urinary retention, orthostatic hypotension, and QT prolongation; it is dangerous in overdose [33]. It should be used cautiously in older men with prostatic enlargement and in patients with cardiac disease.
Beta-blockers cause fatigue, bradycardia, hypotension, cold extremities, sleep disturbance, and bronchospasm in susceptible patients; they can mask hypoglycaemia symptoms in diabetes [9,32]. They are inexpensive and have the longest safety record of any preventive class.
Candesartan and telmisartan cause dizziness, hypotension, hyperkalaemia, and renal impairment, particularly in patients with chronic kidney disease or on potassium-sparing diuretics [11,35]. The 30% dizziness rate with candesartan 16 mg in the 2025 trial is higher than many clinicians expect and is a reason to start low.
OnabotulinumtoxinA causes injection-site pain, neck pain, headache, ptosis, and, rarely, distant spread of toxin effects; neutralizing antibodies can reduce response with repeated dosing [20,21]. The PREEMPT program’s safety profile was consistent with long-standing cosmetic and neurologic use.
Table 8. Adverse effects and monitoring by drug class
| Class | Common adverse effects | Serious/uncommon | Monitoring |
|---|---|---|---|
| CGRP mAbs | Injection-site reactions, nasopharyngitis, constipation | Hypertension, hypersensitivity, possible CV events | Blood pressure at baseline and follow-up; pregnancy status |
| Gepants | Constipation, nausea, weight loss (atogepant) | Hepatic enzyme rise (theoretical), drug interactions | Hepatic function if risk factors; CYP3A4 interactions |
| Topiramate | Paraesthesia, cognitive slowing, fatigue, stones | Acidosis, glaucoma, teratogenicity | Electrolytes, renal function, pregnancy counselling |
| Valproate | Weight gain, tremor, GI upset | Teratogenicity, PCOS, pancreatitis, thrombocytopenia | LFTs, CBC, pregnancy prevention |
| Amitriptyline | Sedation, dry mouth, constipation, urinary retention | QT prolongation, overdose cardiotoxicity | ECG if cardiac risk; urinary symptoms |
| Beta-blockers | Fatigue, bradycardia, hypotension, cold extremities | Bronchospasm, heart block | Heart rate, blood pressure |
| ARBs | Dizziness, hypotension | Hyperkalaemia, renal impairment | Blood pressure, potassium, creatinine |
| OnabotulinumtoxinA | Neck pain, headache, ptosis | Distant spread, antibody-mediated loss of effect | Clinical response per cycle |
12.4 Pregnancy and lactation safety
Section 11.1 covers the guideline’s recommendations. From a safety-monitoring standpoint, three points are worth restating. First, topiramate and valproate are the two preventives with unequivocal teratogenic signals, and both should be avoided in anyone who could become pregnant unless there is no alternative [17,34]. Second, the CGRP antibodies are large immunoglobulins that cross the placenta increasingly after the first trimester, and the 2026 spontaneous-abortion signal, while confounded, argues for contraception counselling and pregnancy testing before initiation in people of childbearing potential [86,87]. Third, lactation data are reassuring for gepants but almost absent for monoclonal antibodies; the guideline’s LactMed-based counselling is the practical standard [17,88].
13. Adherence, Persistence, and Real-World Effectiveness
13.1 The persistence problem
Randomized trials measure efficacy under idealized conditions: motivated patients, frequent visits, free drug, and exclusion of comorbidity. Real-world persistence measures something closer to effectiveness, and the contrast is stark. A 2026 meta-analysis of 53 real-world studies found pooled persistence of 93.3% at 3 months, 84.2% at 6 months, and 76.2% at 12 months, with substantial heterogeneity (I² 82–93%). Among 1,242 documented discontinuation events, lack of efficacy was the leading reason (63.7%), followed by adverse events or poor tolerability (15.1%) [10]. Atogepant had lower 3-month persistence than monoclonal antibodies (83.5% versus 94.2%, P=0.013), although the atogepant data were limited to early follow-up [10].
The German Pain e-Registry comparison is even more striking. At six months, persistence was 89.4% for CGRP monoclonal antibodies, 43.0% for low-evidence oral preventives, and 34.0% for high-evidence oral preventives (all P<0.001). Adverse-drug-reaction discontinuation occurred in 7.0% with CGRP antibodies versus 35.7% and 44.5%, and inefficacy discontinuation in 3.6% versus 21.3% and 21.5% [102]. Within the high-evidence oral cohort, tricyclic antidepressants had the steepest early attrition, followed by beta-blockers, with somewhat better but still suboptimal persistence for topiramate and flunarizine [102].
These data reframe the guideline’s tolerability list. The list is not a soft preference; it predicts whether a patient will still be taking the drug six months later. A drug that reduces migraine days by two but is discontinued by half of patients within six months has a smaller population-level effect than a drug that reduces days by 1.5 and is continued by 85%.
13.2 Real-world effectiveness
Real-world studies generally reproduce trial efficacy in broader populations, with some attenuation. A narrative review of more than six years of fremanezumab data identified 14 publications from 10 randomized trials and 34 publications from 23 real-world studies; across both, fremanezumab was associated with improvements in monthly migraine and headache days from as early as four weeks and sustained for up to 52 weeks [103]. A prospective single-center study of 89 patients with high-frequency episodic or chronic migraine found a mean reduction from 21.1 to 12.4 headache days over three months, with 61.8% achieving at least 50% response; six of ten patients refractory to erenumab responded to fremanezumab [104].
The “switch” question is clinically important and under-studied. A retrospective German study of 196 patients found that 50% response to erenumab was 35.7% in patients with no more than one prior preventive failure and 30.7% in those with two to six failures — not significantly different (adjusted OR 1.39, 95% CI 0.53–3.69) [105]. This suggests that prior failure count is a weak predictor of response to a CGRP antibody, and that a trial of therapy may be warranted even in heavily pre-treated patients.
Long-term persistence with eptinezumab has been reported out to four years in a cohort of 112 patients with treatment-resistant chronic migraine, with sustained reductions in monthly headache days and disability among continuers [106]. The two-year EUREkA cohort reported persistence factors for anti-CGRP antibodies, although the abstract does not provide extractable effect estimates [107]. These are useful signals but not substitutes for registry data with complete follow-up.
13.3 Why patients stop
The dominance of “lack of efficacy” over “adverse events” in discontinuation statistics is important. It suggests that many patients abandon treatment before an adequate trial, or that their expectations are not met by a one-to-two-day average reduction. The guideline’s 8–12-week rule and its emphasis on patient-centric goals are direct responses to this problem [4]. So is the recommendation to optimize the dose at eight weeks rather than abandoning the drug [4].
A practical implication: clinicians should set expectations explicitly at initiation. A 50% reduction is a good outcome, not a minimum; a 30% reduction with improved function may be worth continuing; and a patient who has no response at 12 weeks at a tolerated dose should be switched rather than continued indefinitely.
14. Cost, Access, and Equity
14.1 The cost problem
CGRP monoclonal antibodies and gepants are expensive. In the United States, list prices for the antibodies and gepants are at specialty-drug levels, and net prices after rebates are lower but opaque. In single-payer systems, the drugs are negotiated at lower prices but restricted by prior-authorization criteria. NICE’s technology appraisals position the CGRP monoclonal antibodies for adults who have failed at least three oral preventive drug treatments, which is a narrower population than the 2026 AAN/AHS efficacy list implies [N1,N2]. Rimegepant was recommended by NICE for acute treatment in 2023, not for prevention [108].
The guideline’s cost recommendation is deliberately non-specific: offer less expensive options depending on the patient’s formulary and insurance system, with no named drugs [4]. That is honest — cost is local — but it leaves clinicians without a practical algorithm. A reasonable interpretation is that propranolol, topiramate, amitriptyline, and candesartan should be tried first when cost is the dominant constraint, and that CGRP-targeted therapy should be reserved for patients who fail or cannot tolerate them, unless the patient’s formulary permits earlier access.
14.2 Cost-effectiveness
Formal economic evaluations generally find CGRP-targeted therapy cost-effective in patients with high-frequency or chronic migraine and prior treatment failure, and less clearly cost-effective in low-frequency episodic migraine. A Dutch analysis of fremanezumab versus best supportive care in chronic migraine with inadequate response to topiramate or valproate and onabotulinumtoxinA found that fremanezumab was cost-saving (€2,514 saved per patient) and gained 1.45 quality-adjusted life-years over a lifetime horizon; in a broader population with two to four prior failures, the incremental cost-effectiveness ratio was €2,547 per QALY gained [25]. Those figures are far below conventional willingness-to-pay thresholds, but they depend on drug price, real-world persistence, and the assumption that responders remain on treatment.
14.3 Equity
The equity dimension is less studied than the cost dimension but no less important. Migraine prevalence is highest in women and in working-age adults, and disability is highest in women [1,3]. Prevalence estimates are also sensitive to how and where cases are ascertained, which complicates cross-country comparisons. CGRP-targeted therapy is least accessible in exactly the settings where the burden is greatest, and the trials that established it enrolled predominantly White, female, otherwise healthy participants. The 2026 guideline does not address equity directly, and that is a gap. A health system that makes CGRP antibodies available only after three oral failures, while oral preventives are poorly tolerated and often discontinued, may reproduce the very undertreatment the guideline seeks to correct [10,102].
15. International Guideline Comparison
The 2026 AAN/AHS guideline is one of several contemporary documents, and the differences among them are instructive.
Table 9. Comparison of major migraine prevention guidelines
| Guideline (year) | Scope | First-line/strong recommendations | CGRP-targeted therapy positioning | Notable differences |
|---|---|---|---|---|
| AAN/AHS (2012) [5] | Episodic migraine prevention | Level A: divalproex, valproate, topiramate, metoprolol, propranolol, timolol | Not available | No chronic migraine; no CGRP agents |
| AHS consensus (2021/2022) [15,16] | Integration of new treatments | Not graded like a guideline | CGRP agents as options; eptinezumab included | Consensus, not systematic review; practical focus |
| EHF (2019) [109] | CGRP monoclonal antibodies | All four recommended | Erenumab, fremanezumab, galcanezumab, eptinezumab recommended; evidence low–high | European access and reimbursement context |
| Canadian Headache Society (2024) [67] | Episodic and chronic migraine | CGRP agents, propranolol, topiramate, onabotulinumtoxinA; topiramate downgraded to weak | CGRP agents approved by Health Canada 2018–2024 | Downgraded topiramate; weak against gabapentin; weak for memantine, levetiracetam, enalapril, melatonin |
| NICE (2022–2023) [108,N1,N2] | Health technology appraisal | CGRP mAbs after ≥3 oral preventive failures | Restricted access; rimegepant acute only (2023) | Cost-effectiveness-driven; narrower than AAN/AHS |
| AAN/AHS (2026) [4] | Episodic and chronic migraine, special populations | Four priority lists; CGRP agents in efficacy and tolerability lists | CGRP agents first-line for efficacy/tolerability | Patient-priority framework; pregnancy, older adults, comorbidity, MOH |
The EHF 2019 guideline was the first to recommend all four monoclonal antibodies, finding low-to-high quality evidence in episodic migraine and medium-to-high quality evidence in chronic migraine [109]. It relied on expert opinion for several clinical questions, including switching and discontinuation, because the trial evidence did not address them [109]. The 2026 AAN/AHS guideline is more conservative in some respects — it does not recommend CGRP antibodies in pregnancy, and it places them outside the long-term-safety list — and more expansive in others, particularly its special-population recommendations.
The Canadian Headache Society’s 2024 update is the closest comparator in methodology: a systematic review of 61 new studies, meta-analysis where possible, and a modified Delphi process. It downgraded topiramate to a weak recommendation for episodic migraine and added weak recommendations for memantine, levetiracetam, enalapril, and melatonin [67]. The divergence from the AAN/AHS guideline on topiramate is instructive: the Canadian panel weighted tolerability and long-term harms more heavily, while the AAN/AHS panel retained topiramate in the efficacy and long-term-safety lists. Both are defensible; the difference reflects the values of the panels, not a contradiction in the data.
NICE’s position is the most restrictive. By requiring failure of at least three oral preventive classes before funding a CGRP antibody, NICE prioritizes cost-effectiveness over population-level access [N1,N2]. The 2026 AAN/AHS guideline, by placing CGRP agents in the efficacy and tolerability lists without a prior-failure requirement, takes a different view. The gap between the two is a major source of international variation in migraine prevention practice.
16. What the Guideline Changes in Practice — and What It Does Not
16.1 What changes
A numeric threshold replaces a vague one. “Four or more migraine days or four or more moderate-to-severe headache days, or substantial disability” is easy to apply and easy to audit. It should increase the proportion of patients offered prevention, particularly in primary care [3,4].
CGRP-targeted therapy becomes a first-line option, not a last resort. The guideline places atogepant, eptinezumab, erenumab, fremanezumab, and galcanezumab in the efficacy and tolerability lists for both episodic and chronic migraine. That is a direct challenge to step-therapy policies that require failure of multiple oral drugs first [4,17,N1].
Tolerability becomes a legitimate primary criterion. The guideline does not treat tolerability as a secondary consideration to be traded off against efficacy; it gives it its own list. That legitimizes choosing a CGRP antibody over topiramate for a patient who prioritizes feeling well over a marginally larger average effect [6,7,17].
Medication overuse is destigmatized. The recommendation to offer prevention to patients with medication overuse, and to prefer agents with evidence in that population, replaces the older “detoxify first” reflex [17,77,96].
The assessment clock is standardized. Eight to twelve weeks for most drugs and 24 weeks for onabotulinumtoxinA gives clinicians and patients a shared expectation and a defensible stopping rule [17,38].
Pregnancy counselling becomes a Level A obligation. The four Level A statements on pregnancy are the strongest language in the document and should change how every prescriber discusses teratogenicity, contraception, and planning [N3,N4].
16.2 What does not change
There is still no efficacy hierarchy. The guideline explicitly declines to rank drugs by efficacy and cites the inadequacy of head-to-head trials. That is honest, but it leaves clinicians without guidance on sequencing after a first CGRP agent fails. Should they switch within class, switch to a gepant, or add onabotulinumtoxinA? The guideline does not say, and the evidence does not either [65,79].
Pediatric migraine is out of scope. The guideline applies to adults 18 and older. The CHAMP trial, which found no benefit of amitriptyline or topiramate over placebo in children and adolescents, is a reminder that adult evidence does not extrapolate [110]. Pediatric prevention remains an evidence desert.
Cost is acknowledged but not solved. The cost list has no named drugs, and the guideline does not address prior authorization, patient assistance, or health-system design [4].
Long-term comparative safety is unresolved. The guideline’s long-term-safety list is a proxy for “we have known these drugs for decades,” not a claim that they are safer than CGRP-targeted therapy. The absence of a long-term comparative trial is the largest evidence gap in the field [58,59].
17. Controversies and Evidence Gaps
17.1 The search-cutoff problem
The guideline’s systematic review closed on 6 June 2024. Several important trials reported afterward, including the 2025 candesartan trial [11], the TEMPLE trial of atogepant versus topiramate (published September 2026) [6], the rimegepant phase 4 prevention trial (2025) [60], and the RESOLUTION trial of eptinezumab in medication-overuse headache (2026) [96]. TEMPLE is particularly consequential: it is the first trial to show a gepant superior to topiramate on both tolerability and efficacy, and it directly addresses the guideline’s claim that head-to-head evidence is insufficient to establish superiority [6]. The guideline’s conclusions are not wrong given its evidence window, but they are already partly outdated.
This is a structural problem with guidelines in fast-moving fields, not a unique failure. The practical response is to read the guideline as a snapshot through mid-2024 and to check for subsequent trials before applying its drug lists rigidly.
17.2 The shrinking placebo response margin
Placebo response in migraine prevention trials has increased over time, narrowing the placebo-subtracted effect of active drugs. A 2026 commentary argues that the success of modern therapeutics has forced a re-evaluation of placebo-controlled trials, because the accumulating evidence base and changing patient expectations inflate placebo response and make historical effect sizes incomparable [111]. The implication is that network meta-analyses spanning decades may underestimate the relative benefit of newer drugs or overestimate the benefit of older ones, depending on when their trials were conducted.
17.3 Pregnancy and CGRP safety
The 2026 spontaneous-abortion signal is the most important safety question in the field. The adjusted hazard ratio of 1.42 (95% CI 1.12–1.80) is moderate and could be confounded by migraine severity, but it is consistent with the biology of CGRP in placentation [86]. The guideline’s advice to avoid CGRP antibodies in pregnancy is prudent, but it leaves patients with few options: nifedipine, propranolol, metoprolol, and onabotulinumtoxinA for chronic migraine, all with limited evidence [4]. Prospective pregnancy registries are urgently needed.
17.4 Long-term safety and the “unknown unknowns”
The CGRP class has been in widespread use for less than a decade. The long-term safety meta-analysis found low adverse-event-related discontinuation but was based on 11 observational studies at severe risk of bias [58]. The EudraVigilance signals for hypertension, atrial fibrillation, and myocardial infarction are hypothesis-generating [100]. The registry of patients with prior cardiovascular disease is reassuring but includes only 17 patients [59]. A patient who takes a CGRP antibody for 20 years is, in evidence terms, participating in an uncontrolled experiment.
17.5 Medication overuse and withdrawal
The guideline’s position that prevention alone may suffice is supported by trials in which withdrawal was not required, but the evidence is limited by small numbers of opioid and barbiturate users [4]. The RESOLUTION trial adds support for eptinezumab plus brief education in chronic migraine with medication-overuse headache, but it does not answer whether withdrawal adds value in patients who respond to prevention [96]. The interaction between prevention, withdrawal, and patient preference remains under-studied.
17.6 Generalizability
The trials that established CGRP-targeted therapy enrolled predominantly female, White, otherwise healthy adults with few comorbidities. The 2026 guideline’s special-population recommendations are largely extrapolations. Older adults, patients with renal or hepatic impairment, people with severe mental illness, and pregnant people are underrepresented [89,90]. Real-world registries partially fill the gap, but they are subject to selection bias and incomplete follow-up [10,102].
17.7 The “low-confidence” list and rimegepant
Rimegepant’s placement in the low-confidence prevention list is defensible on the basis of its pivotal trial, which showed a placebo-subtracted reduction of only 0.8 migraine days [27]. But the phase 4 trial in patients with prior oral failures showed a larger effect (−1.6 days), and the drug is approved for prevention in several jurisdictions [60]. The guideline’s categorization may therefore understate its value in treatment-resistant patients, particularly those who prefer an oral option and cannot tolerate atogepant.
17.8 What about METEORIC?
The seed brief for this review listed “METEORIC” among the pivotal trials to snowball. A search of ClinicalTrials.gov for “METEORIC” returned no migraine prevention trial; the only registered study with that acronym is METEORIC-HF, a heart-failure trial of omecamtiv mecarbil (NCT03759392). It is possible that the seed conflated METEORIC with another trial acronym, or that a migraine trial named METEORIC was registered outside ClinicalTrials.gov. No such trial is cited in the 2026 guideline’s reference list or in the systematic review’s reference list. This review therefore treats METEORIC as unverified and does not cite it.
18. Future Directions and Ongoing Trials
Several ongoing studies will shape the next guideline update.
The APT Comparison Study (NCT06972056) is a phase 4 randomized trial of approximately 1,335 patients comparing atogepant with topiramate and propranolol on effectiveness and tolerability. It is one of the first large, pragmatic, three-arm comparisons of a gepant against two conventional oral preventives, and it should provide the head-to-head evidence the 2026 guideline found lacking.
Rimegepant for chronic migraine prevention (NCT07729462) is a phase 3 trial of once-daily rimegepant versus placebo in approximately 400 adults with chronic migraine. If positive, it would extend gepant prevention from episodic to chronic migraine and strengthen rimegepant’s position in future guidelines.
Rimegepant for intermittent prevention of menstrual migraine (NCT06641466) is a phase 3 trial of peri-menstrual dosing in approximately 723 women. Menstrual migraine is under-studied and disproportionately affects young women, and a positive trial would fill a genuine gap.
Atogepant added to onabotulinumtoxinA in chronic migraine has been studied in a 24-week open-label trial of 75 patients, with safety and exploratory efficacy results [112]. A randomized combination trial would be more informative, particularly for patients who partial-respond to botulinum.
Real-world registries — including the German Pain e-Registry, the EUREkA cohort, and national headache registries — will provide the long-term safety and persistence data that randomized trials cannot [102,106,107].
Pregnancy registries for CGRP antibodies and gepants are essential. The 2026 signal will not be resolved by claims data alone; prospective exposure registries with standardized outcome ascertainment are needed [86,87].
Biomarker and precision-medicine research aims to predict who will respond to which preventive. No validated biomarker is currently available, and the guideline does not recommend one. The heterogeneity of response — some patients achieve 100% reduction, others none — is the strongest argument for continued research in this area.
19. LIMITATIONS of This Review
This review has several important limitations, and readers should weigh them before acting on its conclusions.
Access to the primary guideline text. The 2026 AAN/AHS guideline and its companion systematic review are subscription-only. This review relied on the published abstracts [4,17,18], the AAN press release, and two independent, detailed guideline summaries [N3,N4] for the recommendation statements, statement counts, obligation levels, and supporting rationales. The summaries agree with each other and with the press release on the core recommendations, but they are secondary sources. Exact wording, the full evidence tables, and the GRADE certainty ratings per outcome could not be verified against the published PDF. Where this review reports “13 Level A, 34 Level B, 5 Level C” or specific rationales (for example, the topiramate withdrawal effect size), those details come from the secondary summaries and should be checked against the primary document before being cited in derivative work.
Search strategy. This is a narrative, citation-based review, not a systematic review. It used 50 PubMed E-utilities queries, a curated set of 112 references, EuropePMC reference-list snowballing of the guideline and its companion systematic review, and targeted ClinicalTrials.gov queries. It did not follow PRISMA, did not dual-screen records, did not perform risk-of-bias assessment, and did not pool data. Publication bias and selective outcome reporting are therefore unaddressed.
Abstract-level extraction. For most trials, effect sizes and responder rates were extracted from structured abstracts rather than full texts. Some abstracts do not report the endpoint of interest, and the tables in this review mark those cells with dashes rather than inferring values. Where full texts were available through open access, they were used; where they were not, the abstract was the ceiling.
Citation verification. Every PMID in the reference list was mechanically resolved through NCBI E-utilities esummary on 26 October 2026. Verification confirms that the identifier exists and that the title, journal, and year match the cited record; it does not confirm that the cited claim is an accurate representation of the paper. That is a human interpretive step, and errors of interpretation remain possible.
Guideline comparison. The international comparison in Section 15 relies on the published guidelines and, for NICE, on technology appraisal documents and a BMJ news report. NICE guidance changes frequently; readers should check the current NICE website before applying the access criteria described here.
Regulatory statements. Statements about FDA and Health Canada approvals are descriptive and are not supported by numbered PMID citations. They are based on the trial publications, the AHS consensus statement, and the Canadian Headache Society guideline’s statement that CGRP agents were approved by Health Canada between 2018 and 2024 [15,67]. No regulatory document was formally verified for this review.
Unverified seed item. The seed brief listed “METEORIC” among the pivotal trials. No registered migraine prevention trial with that name was identified (Section 17.8). This is recorded as a gap rather than papered over.
Scope. The review addresses pharmacologic prevention only. It does not cover acute treatment, neuromodulation, behavioral therapy, dietary interventions, or procedural treatments except where they bear directly on prevention (for example, onabotulinumtoxinA). It does not address pediatric migraine except as a contrast. It does not provide dosing tables or prescribing algorithms, which are outside the scope of a literature review.
Currency. The literature search for this review extended to October 2026, but the field moves quickly. Trials reported after that date, and post-marketing safety signals that emerge later, are not captured.
20. Conclusions
The 2026 AAN/AHS guideline is one of the most comprehensive syntheses of migraine prevention evidence yet produced, and its central move — replacing a ranked drug ladder with a patient-priority framework — is the right one for a field in which several drugs work, none dominates, and the differences that matter most are tolerability, safety, cost, and fit with the patient’s life.
Its recommendations are defensible on the evidence through mid-2024. The four-day threshold is simple and actionable; the four priority lists are honest about trade-offs; the 8–12-week and 24-week assessment rules are grounded in trial timing; the medication-overuse reframing is supported by randomized data; and the pregnancy recommendations, though based on weak evidence, reflect a strong ethical obligation. The guideline’s willingness to say that head-to-head evidence is insufficient to rank drugs is a strength, not a failure.
Its vulnerabilities are equally clear. The search cutoff excludes trials that have already changed practice, most notably the 2025 candesartan trial and the 2026 TEMPLE trial. The long-term safety of CGRP-targeted therapy remains an open question. The pregnancy signal for CGRP antibodies requires prospective study. The cost-access gap means that the guideline’s preferred options are unavailable to many of the patients who need them most. And the absence of pediatric evidence, comparative sequencing data, and validated response biomarkers leaves important clinical questions unanswered.
For practicing clinicians, the practical synthesis is this: offer prevention at four or more migraine days or when disability is substantial; choose the drug that fits the patient’s priority — efficacy, tolerability, long-term safety, or cost — rather than a universal first-line; give it a fair trial of 8–12 weeks at a tolerated dose, or 24 weeks for onabotulinumtoxinA; counsel on adverse effects and pregnancy risk before prescribing; reassess with the patient’s goals rather than a single number; and do not withhold prevention from patients with medication overuse. Those steps, more than any particular drug choice, are what the guideline changes.
References
References are numbered in order of first appearance. All PMIDs were verified via NCBI E-utilities esummary on 26 October 2026.
- GBD 2023 Headache Collaborators. Global, regional, and national burden of headache disorders, 1990-2023: a systematic analysis for the Global Burden of Disease Study 2023. Lancet Neurol. 2025;24(12):1005-1015. PMID: 41240916. doi:10.1016/S1474-4422(25)00402-8.
- Stovner LJ, Hagen K, Linde M, Steiner TJ. The global prevalence of headache: an update, with analysis of the influences of methodological factors on prevalence estimates. J Headache Pain. 2022;23(1):34. PMID: 35410119. doi:10.1186/s10194-022-01402-2.
- Lipton RB, Bigal ME, Diamond M, Freitag F, Reed ML, Stewart WF. Migraine prevalence, disease burden, and the need for preventive therapy. Neurology. 2007;68(5):343-9. PMID: 17261680. doi:10.1212/01.wnl.0000252808.97649.21.
- Potrebic S, Tanveer S, Becker WJ, Burch R, Cooke LJ, Fenton T, et al. Pharmacologic Treatment for Migraine Prevention in Adults Practice Guideline Recommendations: Report of the AAN Guidelines Subcommittee and the American Headache Society. Neurology. 2026;107(7):e214881. PMID: 42673560. doi:10.1212/WNL.0000000000214881.
- Silberstein SD, Holland S, Freitag F, Dodick DW, Argoff C, Ashman E. Evidence-based guideline update: pharmacologic treatment for episodic migraine prevention in adults: report of the Quality Standards Subcommittee of the American Academy of Neurology and the American Headache Society. Neurology. 2012;78(17):1337-45. PMID: 22529202. doi:10.1212/WNL.0b013e3182535d20.
- Reuter U, Dycke AV, Versijpt J, Holle-Lee D, Farkas MK, Tepper SJ, et al. Tolerability, safety, and efficacy of atogepant versus topiramate in adults with migraine (TEMPLE): a randomised, head-to-head, phase 3b trial. Lancet Neurol. 2026;25(9):806-817. PMID: 42492556. doi:10.1016/S1474-4422(26)00214-0.
- Reuter U, Ehrlich M, Gendolla A, Heinze A, Klatt J, Wen S, et al. Erenumab versus topiramate for the prevention of migraine – a randomised, double-blind, active-controlled phase 4 trial. Cephalalgia. 2022;42(2):108-118. PMID: 34743579. doi:10.1177/03331024211053571.
- Pozo-Rosich P, Dolezil D, Paemeleire K, Stepien A, Stude P, Snellman J, et al. Early Use of Erenumab vs Nonspecific Oral Migraine Preventives: The APPRAISE Randomized Clinical Trial. JAMA Neurol. 2024;81(5):461-470. PMID: 38526461. doi:10.1001/jamaneurol.2024.0368.
- Versijpt J, Deligianni C, Hussain M, Amin F, Reuter U, Sanchez-Del-Rio M, et al. European Headache Federation (EHF) critical re-appraisal and meta-analysis of oral drugs in migraine prevention – part 4: propranolol. J Headache Pain. 2024;25(1):119. PMID: 39044170. doi:10.1186/s10194-024-01826-y.
- Pellesi L, Yangjeh A, Hajjaj I, Onan D, Waliszewska-Prosół M, Martelletti P. Persistence and Reasons for Treatment Discontinuation of Anti-CGRP Therapies in Migraine Prophylaxis: A Systematic Review and Meta-analysis of Real-World Studies. Pain Ther. 2026. PMID: 42814345. doi:10.1007/s40122-026-00895-y.
- Øie LR, Wergeland T, Salvesen Ø, Gravdahl GRB, Aschehoug I, Gulati S, et al. Candesartan versus placebo for migraine prevention in patients with episodic migraine: a randomised, triple-blind, placebo-controlled, phase 2 trial. Lancet Neurol. 2025;24(10):817-827. PMID: 40975098. doi:10.1016/S1474-4422(25)00269-8.
- Lu W, Zhang Y, Shi C, Tao Y, Li S, Sun Y, et al. Migraine: Epidemiology, Risk Factors, Pathophysiology, and Treatment. MedComm (2020). 2026;7(9):e70933. PMID: 42676555. doi:10.1002/mco2.70933.
- Ran Y, Wang X, Dong Z. Global burden of headache disorders: innovations, disparities, and priorities for action. Lancet Neurol. 2025;24(12):988-989. PMID: 41240922. doi:10.1016/S1474-4422(25)00388-6.
- Sun H, Dodick DW, Silberstein S, Goadsby PJ, Reuter U, Ashina M, et al. Safety and efficacy of AMG 334 for prevention of episodic migraine: a randomised, double-blind, placebo-controlled, phase 2 trial. Lancet Neurol. 2016;15(4):382-90. PMID: 26879279. doi:10.1016/S1474-4422(16)00019-3.
- Ailani J, Burch RC, Robbins MS. The American Headache Society Consensus Statement: Update on integrating new migraine treatments into clinical practice. Headache. 2021;61(7):1021-1039. PMID: 34160823. doi:10.1111/head.14153.
- Burch RC, Ailani J, Robbins MS. The American Headache Society Consensus Statement: Update on integrating new migraine treatments into clinical practice. Headache. 2022;62(1):111-112. PMID: 34873692. doi:10.1111/head.14245.
- Pringsheim T, Smith DB, Tanveer S, Becker WJ, Burch R, Cooke LJ, et al. Systematic Review of Pharmacologic Treatment for Migraine Prevention in Adults: Report of the AAN Guidelines Subcommittee and the American Headache Society. Neurology. 2026;107(7):e218112. PMID: 42673559. doi:10.1212/WNL.0000000000218112.
- Pringsheim T, Smith DB, Tanveer S, Becker WJ, Burch R, Cooke LJ, et al. Systematic Review of Pharmacologic Treatment for Migraine Prevention in Adults: Report of the AAN Guidelines Subcommittee and the American Headache Society. Headache. 2026;66(9):1971-1990. PMID: 42673583. doi:10.1111/head.70200.
- Lipton RB, Silberstein SD. Episodic and chronic migraine headache: breaking down barriers to optimal treatment and prevention. Headache. 2015;55 Suppl 2:103-22; quiz 123-6. PMID: 25662743. doi:10.1111/head.12505_2.
- Silberstein SD, Diener HC, Dodick DW, Sommer K, Lipton RB. Sustained benefits of onabotulinumtoxinA treatment in chronic migraine: An analysis of the pooled Phase 3 REsearch Evaluating Migraine Prophylaxis Therapy (PREEMPT) randomized controlled trials. Headache. 2024;64(7):838-848. PMID: 38982666. doi:10.1111/head.14743.
- Pozo-Rosich P, Blumenfeld AM, Lipton RB, DeGryse RE, Li B, Adams AM, et al. OnabotulinumtoxinA for the preventive treatment of episodic migraine: Results from the phase 3, multicenter randomized, double-blind, placebo-controlled phase of the PRECLUDE trial. Cephalalgia. 2025;45(10):3331024251370769. PMID: 41091731. doi:10.1177/03331024251370769.
- Silberstein SD, Dodick DW, Bigal ME, Yeung PP, Goadsby PJ, Blankenbiller T, et al. Fremanezumab for the Preventive Treatment of Chronic Migraine. N Engl J Med. 2017;377(22):2113-2122. PMID: 29171818. doi:10.1056/NEJMoa1709038.
- Lipton RB, Goadsby PJ, Smith J, Schaeffler BA, Biondi DM, Hirman J, et al. Efficacy and safety of eptinezumab in patients with chronic migraine: PROMISE-2. Neurology. 2020;94(13):e1365-e1377. PMID: 32209650. doi:10.1212/WNL.0000000000009169.
- Pozo-Rosich P, Caronna E, Sacco S, Peres MFP, Ashina S, Özge A, et al. Early treatment in migraine – A call to shift prevention from attacks to disease progression: A position statement from the International Headache Society. Cephalalgia. 2025;45(10):3331024251387721. PMID: 41134822. doi:10.1177/03331024251387721.
- Wolters S, Carpay JA, Pronk MH, Zuurbier KWM, Driessen MT, Lyras L, et al. A Dutch cost-effectiveness analysis of fremanezumab versus best supportive care in patients with chronic migraine and inadequate response to prior preventive therapy. BMC Neurol. 2024;24(1):214. PMID: 38914929. doi:10.1186/s12883-024-03697-x.
- Janković SM, Janković SV. Anti-calcitonin Gene-Related Peptide Monoclonal Antibodies in Migraine: Focus on Clinical Pharmacokinetics. Eur J Drug Metab Pharmacokinet. 2024;49(3):277-293. PMID: 38461486. doi:10.1007/s13318-024-00885-5.
- Croop R, Lipton RB, Kudrow D, Stock DA, Kamen L, Conway CM, et al. Oral rimegepant for preventive treatment of migraine: a phase 2/3, randomised, double-blind, placebo-controlled trial. Lancet. 2021;397(10268):51-60. PMID: 33338437. doi:10.1016/S0140-6736(20)32544-7.
- Ailani J, Lipton RB, Goadsby PJ, Guo H, Miceli R, Severt L, et al. Atogepant for the Preventive Treatment of Migraine. N Engl J Med. 2021;385(8):695-706. PMID: 34407343. doi:10.1056/NEJMoa2035908.
- Janković SM, Janković SV. Anti-Calcitonin Gene-Related Peptide Monoclonal Antibodies in Migraine: Focus on Drug Interactions. Eur J Drug Metab Pharmacokinet. 2024;49(3):263-275. PMID: 38457093. doi:10.1007/s13318-024-00887-3.
- Bond DS, Smith JH, Sinclair A, Goadsby PJ, Ailani J, Liu Y, et al. Weight loss with atogepant in the long-term treatment of migraine: An interim analysis of a safety endpoint from a phase 3, multicenter, open-label, 156-week extension study. Headache. 2026;66(8):1730-1740. PMID: 42396885. doi:10.1111/head.70162.
- Brandes JL, Saper JR, Diamond M, Couch JR, Lewis DW, Schmitt J, et al. Topiramate for migraine prevention: a randomized controlled trial. JAMA. 2004;291(8):965-73. PMID: 14982912. doi:10.1001/jama.291.8.965.
- Linde K, Rossnagel K. Propranolol for migraine prophylaxis. Cochrane Database Syst Rev. 2004:CD003225. PMID: 15106196. doi:10.1002/14651858.CD003225.pub2.
- Lampl C, Versijpt J, Amin FM, Deligianni CI, Gil-Gouveia R, Jassal T, et al. European Headache Federation (EHF) critical re-appraisal and meta-analysis of oral drugs in migraine prevention-part 1: amitriptyline. J Headache Pain. 2023;24(1):39. PMID: 37038134. doi:10.1186/s10194-023-01573-6.
- Chan AYL, Wong ICK, Man KKC. Improving evidence-based antiseizure medication prescribing during pregnancy: strengthening research and clinical practice. Expert Rev Clin Pharmacol. 2026;19(7):615-618. PMID: 42464699. doi:10.1080/17512433.2026.2705845.
- Diener HC, Gendolla A, Feuersenger A, Evers S, Straube A, Schumacher H, et al. Telmisartan in migraine prophylaxis: a randomized, placebo-controlled trial. Cephalalgia. 2009;29(9):921-7. PMID: 19250283. doi:10.1111/j.1468-2982.2008.01825.x.
- Pozo-Rosich P, Ailani J, Ashina M, Goadsby PJ, Lipton RB, Reuter U, et al. Atogepant for the preventive treatment of chronic migraine (PROGRESS): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. 2023;402(10404):775-785. PMID: 37516125. doi:10.1016/S0140-6736(23)01049-8.
- Tassorelli C, Nagy K, Pozo-Rosich P, Lanteri-Minet M, Sacco S, Nežádal T, et al. Safety and efficacy of atogepant for the preventive treatment of episodic migraine in adults for whom conventional oral preventive treatments have failed (ELEVATE): a randomised, placebo-controlled, phase 3b trial. Lancet Neurol. 2024;23(4):382-392. PMID: 38364831. doi:10.1016/S1474-4422(24)00025-5.
- Silberstein SD, Dodick DW, Aurora SK, Diener HC, DeGryse RE, Lipton RB, et al. Per cent of patients with chronic migraine who responded per onabotulinumtoxinA treatment cycle: PREEMPT. J Neurol Neurosurg Psychiatry. 2015;86(9):996-1001. PMID: 25500317. doi:10.1136/jnnp-2013-307149.
- Diener HC, Agosti R, Allais G, Bergmans P, Bussone G, Davies B, et al. Cessation versus continuation of 6-month migraine preventive therapy with topiramate (PROMPT): a randomised, double-blind, placebo-controlled trial. Lancet Neurol. 2007;6(12):1054-62. PMID: 17988947. doi:10.1016/S1474-4422(07)70272-7.
- Stauffer VL, Wang S, Voulgaropoulos M, Skljarevski V, Kovacik A, Aurora SK. Effect of Galcanezumab Following Treatment Cessation in Patients With Migraine: Results From 2 Randomized Phase 3 Trials. Headache. 2019;59(6):834-847. PMID: 30942898. doi:10.1111/head.13508.
- Goadsby PJ, Reuter U, Hallström Y, Broessner G, Bonner JH, Zhang F, et al. A Controlled Trial of Erenumab for Episodic Migraine. N Engl J Med. 2017;377(22):2123-2132. PMID: 29171821. doi:10.1056/NEJMoa1705848.
- Dodick DW, Ashina M, Brandes JL, Kudrow D, Lanteri-Minet M, Osipova V, et al. ARISE: A Phase 3 randomized trial of erenumab for episodic migraine. Cephalalgia. 2018;38(6):1026-1037. PMID: 29471679. doi:10.1177/0333102418759786.
- Tepper S, Ashina M, Reuter U, Brandes JL, Doležil D, Silberstein S, et al. Safety and efficacy of erenumab for preventive treatment of chronic migraine: a randomised, double-blind, placebo-controlled phase 2 trial. Lancet Neurol. 2017;16(6):425-434. PMID: 28460892. doi:10.1016/S1474-4422(17)30083-2.
- Reuter U, Goadsby PJ, Lanteri-Minet M, Wen S, Hours-Zesiger P, Ferrari MD, et al. Efficacy and tolerability of erenumab in patients with episodic migraine in whom two-to-four previous preventive treatments were unsuccessful: a randomised, double-blind, placebo-controlled, phase 3b study. Lancet. 2018;392(10161):2280-2287. PMID: 30360965. doi:10.1016/S0140-6736(18)32534-0.
- Dodick DW, Silberstein SD, Bigal ME, Yeung PP, Goadsby PJ, Blankenbiller T, et al. Effect of Fremanezumab Compared With Placebo for Prevention of Episodic Migraine: A Randomized Clinical Trial. JAMA. 2018;319(19):1999-2008. PMID: 29800211. doi:10.1001/jama.2018.4853.
- Ferrari MD, Diener HC, Ning X, Galic M, Cohen JM, Yang R, et al. Fremanezumab versus placebo for migraine prevention in patients with documented failure to up to four migraine preventive medication classes (FOCUS): a randomised, double-blind, placebo-controlled, phase 3b trial. Lancet. 2019;394(10203):1030-1040. PMID: 31427046. doi:10.1016/S0140-6736(19)31946-4.
- Stauffer VL, Dodick DW, Zhang Q, Carter JN, Ailani J, Conley RR. Evaluation of Galcanezumab for the Prevention of Episodic Migraine: The EVOLVE-1 Randomized Clinical Trial. JAMA Neurol. 2018;75(9):1080-1088. PMID: 29813147. doi:10.1001/jamaneurol.2018.1212.
- Skljarevski V, Matharu M, Millen BA, Ossipov MH, Kim BK, Yang JY. Efficacy and safety of galcanezumab for the prevention of episodic migraine: Results of the EVOLVE-2 Phase 3 randomized controlled clinical trial. Cephalalgia. 2018;38(8):1442-1454. PMID: 29848108. doi:10.1177/0333102418779543.
- Mulleners WM, Kim BK, Láinez MJA, Lanteri-Minet M, Pozo-Rosich P, Wang S, et al. Safety and efficacy of galcanezumab in patients for whom previous migraine preventive medication from two to four categories had failed (CONQUER): a multicentre, randomised, double-blind, placebo-controlled, phase 3b trial. Lancet Neurol. 2020;19(10):814-825. PMID: 32949542. doi:10.1016/S1474-4422(20)30279-9.
- Hu B, Li G, Li X, Wu S, Yu T, Li X, et al. Galcanezumab in episodic migraine: the phase 3, randomized, double-blind, placebo-controlled PERSIST study. J Headache Pain. 2022;23(1):90. PMID: 35896988. doi:10.1186/s10194-022-01458-0.
- Ashina M, Saper J, Cady R, Schaeffler BA, Biondi DM, Hirman J, et al. Eptinezumab in episodic migraine: A randomized, double-blind, placebo-controlled study (PROMISE-1). Cephalalgia. 2020;40(3):241-254. PMID: 32075406. doi:10.1177/0333102420905132.
- Ashina M, Lanteri-Minet M, Pozo-Rosich P, Ettrup A, Christoffersen CL, Josiassen MK, et al. Safety and efficacy of eptinezumab for migraine prevention in patients with two-to-four previous preventive treatment failures (DELIVER): a multi-arm, randomised, double-blind, placebo-controlled, phase 3b trial. Lancet Neurol. 2022;21(7):597-607. PMID: 35716692. doi:10.1016/S1474-4422(22)00185-5.
- Starling AJ, Kudrow D, Kapur N, Awad SF, Grossman SW, Patel F, et al. Long-Term Reductions in Headache Frequency, Severity, and Disability with Eptinezumab in Adults with Chronic Migraine: Results from the PREVAIL Trial. Pain Ther. 2026;15(3):785-801. PMID: 41996064. doi:10.1007/s40122-026-00831-0.
- Yu S, Matsumori Y, Kim BK, Gryglas-Dworak A, Giorgadze G, Pozo-Rosich P, et al. Efficacy and safety of eptinezumab in a predominantly Asian population with chronic migraine: Results of the randomized, double-blind, placebo-controlled SUNRISE trial. Cephalalgia. 2025;45(10):3331024251386095. PMID: 41091746. doi:10.1177/03331024251386095.
- Diener HC, McAllister P, Jürgens TP, Kessler Y, Ning X, Cohen JM, et al. Safety and tolerability of fremanezumab in patients with episodic and chronic migraine: a pooled analysis of phase 3 studies. Cephalalgia. 2022;42(8):769-780. PMID: 35331009. doi:10.1177/03331024221076485.
- Smith TR, Spierings ELH, Cady R, Hirman J, Schaeffler B, Shen V, et al. Safety and tolerability of eptinezumab in patients with migraine: a pooled analysis of 5 clinical trials. J Headache Pain. 2021;22(1):16. PMID: 33781209. doi:10.1186/s10194-021-01227-5.
- Rizzoli P, Marmura MJ, Robblee J, McVige J, Sacco S, Nahas SJ, et al. Safety and tolerability of atogepant for the preventive treatment of migraine: a post hoc analysis of pooled data from four clinical trials. J Headache Pain. 2024;25(1):35. PMID: 38462625. doi:10.1186/s10194-024-01736-z.
- Hoehne CL, Overeem LH, Sanchez-Del-Rio M, Deligianni C, Gil-Gouveia R, Versijpt J, et al. Decoding the long-term safety of anti-CGRP (receptor) mAbs: a meta-analysis and systematic review. J Headache Pain. 2026;27(1):10. PMID: 41484941. doi:10.1186/s10194-025-02256-0.
- Bellotti A, Noseda R, Masciullo R, Staedler C, Gobbi C, Zecca C. Long-term safety of CGRP pathway inhibitors in migraine patients with prior cerebrovascular or cardiovascular disease. J Neurol. 2026;273(10). PMID: 42742761. doi:10.1007/s00415-026-14099-z.
- Pozo-Rosich P, López JAG, Lisewski P, Aslan AN, Seehra H, Thiry A, et al. A phase 4, randomized, double-blind, placebo-controlled trial evaluating the efficacy and tolerability of rimegepant for the prevention of episodic migraine in adults with a history of inadequate response to traditional oral preventive medications. Cephalalgia. 2025;45(11):3331024251391378. PMID: 41255098. doi:10.1177/03331024251391378.
- Kitamura S, Matsumori Y, Yamamoto T, Ishikawa T, Hoshino Y, Yoshimatsu H, et al. Efficacy and safety of rimegepant for the preventive treatment of migraine in Japan: A double-blind, randomized controlled trial. Headache. 2025;65(8):1403-1412. PMID: 40542538. doi:10.1111/head.14995.
- Matsumori Y, Kitamura S, Yamamoto T, Ishikawa T, Hoshino Y, Yoshimatsu H, et al. Long-term safety and effectiveness of rimegepant for the preventive treatment of migraine in Japan: an open-label extension of a phase 3 study. J Headache Pain. 2026;27(1). PMID: 41761058. doi:10.1186/s10194-026-02290-6.
- Matsumori Y, Yamada H, Nagaseki Y, Shimizu K, Nagy K, Matsuzawa R, et al. Atogepant for the preventive treatment of episodic migraine in Japanese participants: A phase 2/3, randomized, double-blind, placebo-controlled trial with an active treatment extension (RELEASE). Cephalalgia. 2025;45(9):3331024251374569. PMID: 41002192. doi:10.1177/03331024251374569.
- Ashina M, Tepper SJ, Reuter U, Blumenfeld AM, Hutchinson S, Xia J, et al. Once-daily oral atogepant for the long-term preventive treatment of migraine: Findings from a multicenter, randomized, open-label, phase 3 trial. Headache. 2023;63(1):79-88. PMID: 36651532. doi:10.1111/head.14439.
- Schwedt TJ, Myers Oakes TM, Martinez JM, Vargas BB, Pandey H, Pearlman EM, et al. Comparing the Efficacy and Safety of Galcanezumab Versus Rimegepant for Prevention of Episodic Migraine: Results from a Randomized, Controlled Clinical Trial. Neurol Ther. 2024;13(1):85-105. PMID: 37948006. doi:10.1007/s40120-023-00562-w.
- Silberstein SD, Neto W, Schmitt J, Jacobs D. Topiramate in migraine prevention: results of a large controlled trial. Arch Neurol. 2004;61(4):490-5. PMID: 15096395. doi:10.1001/archneur.61.4.490.
- Medrea I, Cooper P, Langman M, Sandoe CH, Amoozegar F, Hussain WM, et al. Updated Canadian Headache Society Migraine Prevention Guideline with Systematic Review and Meta-analysis. Can J Neurol Sci. 2024:1-23. PMID: 39506371. doi:10.1017/cjn.2024.285.
- Pradalier A, Serratrice G, Collard M, Hirsch E, Feve J, Masson M, et al. Long-acting propranolol in migraine prophylaxis: results of a double-blind, placebo-controlled study. Cephalalgia. 1989;9(4):247-53. PMID: 2692838. doi:10.1046/j.1468-2982.1989.0904247.x.
- Tfelt-Hansen P, Standnes B, Kangasneimi P, Hakkarainen H, Olesen J. Timolol vs propranolol vs placebo in common migraine prophylaxis: a double-blind multicenter trial. Acta Neurol Scand. 1984;69(1):1-8. PMID: 6367336. doi:10.1111/j.1600-0404.1984.tb07772.x.
- Gonçalves AL, Martini Ferreira A, Ribeiro RT, Zukerman E, Cipolla-Neto J, Peres MF. Randomised clinical trial comparing melatonin 3 mg, amitriptyline 25 mg and placebo for migraine prevention. J Neurol Neurosurg Psychiatry. 2016;87(10):1127-32. PMID: 27165014. doi:10.1136/jnnp-2016-313458.
- Mathew NT, Saper JR, Silberstein SD, Rankin L, Markley HG, Solomon S, et al. Migraine prophylaxis with divalproex. Arch Neurol. 1995;52(3):281-6. PMID: 7872882. doi:10.1001/archneur.1995.00540270077022.
- Jensen R, Brinck T, Olesen J. Sodium valproate has a prophylactic effect in migraine without aura: a triple-blind, placebo-controlled crossover study. Neurology. 1994;44(4):647-51. PMID: 8164818. doi:10.1212/wnl.44.4.647.
- Freitag FG, Collins SD, Carlson HA, Goldstein J, Saper J, Silberstein S, et al. A randomized trial of divalproex sodium extended-release tablets in migraine prophylaxis. Neurology. 2002;58(11):1652-9. PMID: 12058094. doi:10.1212/wnl.58.11.1652.
- Abdeldayem MR, Elsherbeeny A. Candesartan for Migraine Prophylaxis: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. Clin Neuropharmacol. 2026. PMID: 42715430. doi:10.1097/WNF.0000000000000694.
- Stovner LJ, Linde M, Gravdahl GB, Tronvik E, Aamodt AH, Sand T, et al. A comparative study of candesartan versus propranolol for migraine prophylaxis: A randomised, triple-blind, placebo-controlled, double cross-over study. Cephalalgia. 2014;34(7):523-32. PMID: 24335848. doi:10.1177/0333102413515348.
- Wu J, Wu J, Zhang J, Liu B, Song X, Wu Y, et al. Oral preventive medications for migraine in adults aged 18-65: a network meta-analysis. Front Pharmacol. 2025;16:1620887. PMID: 40978493. doi:10.3389/fphar.2025.1620887.
- Silberstein SD, Blumenfeld AM, Cady RK, Turner IM, Lipton RB, Diener HC, et al. OnabotulinumtoxinA for treatment of chronic migraine: PREEMPT 24-week pooled subgroup analysis of patients who had acute headache medication overuse at baseline. J Neurol Sci. 2013;331(1-2):48-56. PMID: 23790235. doi:10.1016/j.jns.2013.05.003.
- Sharon JD, Krauter R, Chae R, Gardi A, Hum M, Allen I, et al. A placebo controlled, randomized clinical trial of galcanezumab for vestibular migraine: The INVESTMENT study. Headache. 2024;64(10):1264-1272. PMID: 39344988. doi:10.1111/head.14835.
- Damen JAA, Yang B, Idema DL, Vernooij RWM, Huis In ’t Veld L, Kusters M, et al. Comparative Effectiveness of Pharmacologic Treatments for the Prevention of Episodic Migraine Headache: A Systematic Review and Network Meta-analysis for the American College of Physicians. Ann Intern Med. 2025;178(3):369-380. PMID: 39899873. doi:10.7326/ANNALS-24-00315.
- Khalili M, Haghdoost F, Liaghatdar A, Torabiardakani K, Mahdian F, Levit T, et al. Effectiveness and tolerability of pharmacological prophylaxis for migraine headaches: a systematic review and network meta-analysis of randomised controlled trials. BMJ Evid Based Med. 2026. PMID: 42236149. doi:10.1136/bmjebm-2025-114222.
- Khalili M, Haghdoost F, Liaghatdar A, Torabiardakani K, Mahdian F, Atkin-Jones T, et al. Effectiveness and tolerability of pharmacological prophylaxis in migraine patients with prior preventive treatment failure: A systematic review and network meta-analysis of randomized controlled trials. Cephalalgia. 2026;46(4):3331024261441287. PMID: 42011527. doi:10.1177/03331024261441287.
- Zhao C, Li C, Yu X, Dai X, Zou W. Effectiveness and safety of pharmacological prophylaxis for chronic migraine: a systematic review and network meta-analysis. J Neurol. 2024;271(9):5762-5777. PMID: 38910144. doi:10.1007/s00415-024-12512-z.
- Haridas MP, Tripathy A, Maiti R, Srinivasan A. Efficacy and Safety of Anti-CGRP Monoclonal Antibodies in Prevention of Chronic Migraine: A Bayesian Network Meta-analysis. Clin Psychopharmacol Neurosci. 2024;22(1):23-32. PMID: 38247409. doi:10.9758/cpn.23.1109.
- Shakir I, Shahzad F, Shabbir A, Shahabuddin S, Shabbir H, Karamat S, et al. Efficacy and safety of CGRP monoclonal antibodies for migraine prevention in episodic migraine; a network Meta-analysis. Eur J Clin Pharmacol. 2026;82(2):27. PMID: 41546694. doi:10.1007/s00228-025-03934-3.
- Brin MF, Kirby RS, Slavotinek A, Adams AM, Parker L, Ukah A, et al. Pregnancy Outcomes in Patients Exposed to OnabotulinumtoxinA Treatment: A Cumulative 29-Year Safety Update. Neurology. 2023;101(2):e103-e113. PMID: 37137724. doi:10.1212/WNL.0000000000207375.
- Flatman LK, Leal LF, Leroux E, Lipton RB, Bérard A. Early Pregnancy Use of Anti-CGRP Monoclonal Antibodies and Propranolol and Spontaneous Abortion Risk. JAMA Neurol. 2026. PMID: 42862583. doi:10.1001/jamaneurol.2026.3573.
- Boz I, Brandes JL, Robinson C, Scantamburlo Fernandes MF, Lipsius S, Morford L, et al. Pregnancy outcomes following exposure to galcanezumab: Analysis of cases from the galcanezumab clinical trials and postmarketing database. Headache. 2026. PMID: 42856055. doi:10.1111/head.70242.
- Pejic J, Anandakumar J, Marks LA, Halker Singh RB, Ailani J. Anti-calcitonin gene-related peptide treatments in women’s health: Gaps and clinical evidence in menstrual-related migraine, pregnancy, lactation, and hormonal contraception: A narrative review. Headache. 2026. PMID: 42808257. doi:10.1111/head.70195.
- Nahas SJ, Naegel S, Cohen JM, Ning X, Janka L, Campos VR, et al. Efficacy and safety of fremanezumab in clinical trial participants aged ≥60 years with episodic or chronic migraine: pooled results from 3 randomized, double-blind, placebo-controlled phase 3 studies. J Headache Pain. 2021;22(1):141. PMID: 34819017. doi:10.1186/s10194-021-01351-2.
- Lampl C, Kraus V, Lehner K, Loop B, Chehrenama M, Maczynska Z, et al. Safety and tolerability of erenumab in individuals with episodic or chronic migraine across age groups: a pooled analysis of placebo-controlled trials. J Headache Pain. 2022;23(1):104. PMID: 35978286. doi:10.1186/s10194-022-01470-4.
- Hu X, Wang J, Dong W, Fang Q, Hu L, Liu C. A meta-analysis of polycystic ovary syndrome in women taking valproate for epilepsy. Epilepsy Res. 2011;97(1-2):73-82. PMID: 21820873. doi:10.1016/j.eplepsyres.2011.07.006.
- Sarayani A, Winterstein A, Cristofoletti R, Vozmediano V, Schmidt S, Brown J. Real-world effect of a potential drug-drug interaction between topiramate and oral contraceptives on unintended pregnancy outcomes. Contraception. 2023;120:109953. PMID: 36641095. doi:10.1016/j.contraception.2023.109953.
- Lazorwitz A, Pena M, Sheeder J, Teal S. Effect of Topiramate on Serum Etonogestrel Concentrations Among Contraceptive Implant Users. Obstet Gynecol. 2022;139(4):579-587. PMID: 35594123. doi:10.1097/AOG.0000000000004697.
- Lipton RB, Ramirez Campos V, Roth-Ben Arie Z, Galic M, Mitsikostas D, Tassorelli C, et al. Fremanezumab for the Treatment of Patients With Migraine and Comorbid Major Depressive Disorder: The UNITE Randomized Clinical Trial. JAMA Neurol. 2025;82(6):560-569. PMID: 40323613. doi:10.1001/jamaneurol.2025.0806.
- Al-Sayegh Z, Al-Khazali S, Al-Khazali HM, Christensen RH, Videbech P, Lipton RB, et al. Depressive and anxiety symptoms during CGRP-targeted migraine prevention: A systematic review and meta-analysis. Cephalalgia. 2026;46(9):3331024261484898. PMID: 42731077. doi:10.1177/03331024261484898.
- Jensen RH, Lundqvist C, Schytz HW, Tassorelli C, Vernieri F, Lanteri-Minet M, et al. Eptinezumab With Patient Education for Chronic Migraine and Medication-Overuse Headache: The Randomized, Placebo-Controlled RESOLUTION Trial. Neurology. 2026;106(8):e214863. PMID: 41886713. doi:10.1212/WNL.0000000000214863.
- Tepper SJ, Dodick DW, Lanteri-Minet M, Dolezil D, Gil-Gouveia R, Lucas C, et al. Efficacy and Safety of Erenumab for Nonopioid Medication Overuse Headache in Chronic Migraine: A Phase 4, Randomized, Placebo-Controlled Trial. JAMA Neurol. 2024;81(11):1140-9. PMID: 39283627. doi:10.1001/jamaneurol.2024.3043.
- Tepper SJ, Dodick DW, Lanteri-Minet M, Dolezil D, Gil-Gouveia R, Lucas C, et al. Efficacy and Safety of Erenumab in Adults With Medication Overuse Headache: Final Results From a Phase 4 Randomized Placebo-Controlled Study. Eur J Neurol. 2025;32(8):e70328. PMID: 40838472. doi:10.1111/ene.70328.
- Dodick DW, Tepper SJ, Ailani J, Pannacciulli N, Navetta MS, Loop B, et al. Risk of hypertension in erenumab-treated patients with migraine: Analyses of clinical trial and postmarketing data. Headache. 2021;61(9):1411-1420. PMID: 34591982. doi:10.1111/head.14208.
- Sorbara EE, Barbieri MA, Russo G, Cicala G, Spina E. Cardiovascular Adverse Drug Reactions of Anti-Calcitonin Gene-Related Peptide Monoclonal Antibodies for Migraine Prevention: An Analysis from the European Spontaneous Adverse Event Reporting System. BioDrugs. 2024;38(2):275-285. PMID: 38402495. doi:10.1007/s40259-024-00651-8.
- Smith TR, Spierings ELH, Cady R, Hirman J, Ettrup A, Shen V. Cardiovascular outcomes in adults with migraine treated with eptinezumab for migraine prevention: pooled data from four randomized, double-blind, placebo-controlled studies. J Headache Pain. 2021;22(1):143. PMID: 34823467. doi:10.1186/s10194-021-01360-1.
- Überall MA, Müller-Schwefe PCG, Küster MA, Jansen JP. Treatment Persistence in Migraine Prophylaxis Comparing CGRP Monoclonal Antibodies vs. High-/Low-Evidence Conventional Oral Preventives-A Comparative Real-World Evidence Study of Depersonalized Data of the German Pain e-Registry. J Clin Med. 2026;15(5). PMID: 41827402. doi:10.3390/jcm15051985.
- Barbanti P, Smith T, Ramirez Campos V, Guevara Morel AE, Ling C, Bhogal P, et al. Evaluating the Complementary Nature of Randomized Controlled Trials and Real-World Evidence Using Over 5 Years of Experience with Fremanezumab: A Narrative Review. Adv Ther. 2026;43(9):3737-3755. PMID: 42310163. doi:10.1007/s12325-026-03647-8.
- Polanco M, Gárate G, Sánchez-Gudín J, Madera J, Pascual J, González-Quintanilla V. Real-World Lessons with Fremanezumab as the Third Available CGRP Monoclonal Antibody in a Third-Level Hospital: Focus on the Factors Predicting Response. J Clin Med. 2025;14(4). PMID: 40004586. doi:10.3390/jcm14041054.
- Schmidt C, Becker L, Lindner D, Kleinschnitz C, Becker AW, Holle D, et al. Real-world short-term effectiveness of erenumab after single versus multiple prior preventive treatment failures. BMC Neurol. 2026;26(1). PMID: 42661173. doi:10.1186/s12883-026-05311-8.
- Hsieh WJ, Nicol K, Ng CC, Hou TW, Phan P, Keith SW, et al. Eptinezumab in treatment-resistant chronic migraine: Four-year real-world effectiveness and treatment persistence. Cephalalgia. 2026;46(5):3331024261451427. PMID: 42153799. doi:10.1177/03331024261451427.
- Caronna E, Mas-de-Les-Valls R, Egeo G, Vázquez MM, Nieves-Castellanos C, Portocarrero L, et al. Long-Term Effectiveness and Persistence Factors of Anti-CGRP Monoclonal Antibodies in Migraine: 2-Year Results From the EUREkA Cohort. Neurology. 2026;106(12):e214945. PMID: 42202240. doi:10.1212/WNL.0000000000214945.
- Iacobucci G. NICE recommends rimegepant for treating acute migraine in adults. BMJ. 2023;382:2104. PMID: 37709356. doi:10.1136/bmj.p2104.
- Sacco S, Bendtsen L, Ashina M, Reuter U, Terwindt G, Mitsikostas DD, et al. European headache federation guideline on the use of monoclonal antibodies acting on the calcitonin gene related peptide or its receptor for migraine prevention. J Headache Pain. 2019;20(1):6. PMID: 30651064. doi:10.1186/s10194-018-0955-y.
- Powers SW, Coffey CS, Chamberlin LA, Ecklund DJ, Klingner EA, Yankey JW, et al. Trial of Amitriptyline, Topiramate, and Placebo for Pediatric Migraine. N Engl J Med. 2017;376(2):115-124. PMID: 27788026. doi:10.1056/NEJMoa1610384.
- Lev N, Borovoi L, Elefant E, Domany E, Daniel O, Ifergane G. A victim of its own progress: why the success of modern therapeutics forces a re-evaluation of placebo-controlled trials in migraine prevention. Front Neurol. 2026;17:1871028. PMID: 42494401. doi:10.3389/fneur.2026.1871028.
- Rothrock J, Najib U, Ailani J, Ashina S, Bao J, Smith JH, et al. Safety, tolerability, and efficacy of atogepant added to onabotulinumtoxinA for the preventive treatment of chronic migraine: A phase 3, multicenter, 24-week, open-label study. Cephalalgia. 2026;46(4):3331024261429118. PMID: 41944477. doi:10.1177/03331024261429118.
Non-indexed sources cited in text
N1. National Institute for Health and Care Excellence. Eptinezumab for preventing migraine. Technology appraisal guidance TA871. London: NICE; 2023. https://www.nice.org.uk/guidance/TA871 N2. National Institute for Health and Care Excellence. Fremanezumab for preventing migraine. Technology appraisal guidance TA764. London: NICE; 2022. https://www.nice.org.uk/guidance/ta764 N3. Guideline Central. 2026 AAN/AHS pharmacologic treatment for migraine prevention in adults guideline now available. 1 September 2026. https://www.guidelinecentral.com/insights/sep-2026-aanahs-pharmacologic-treatment-migraine-prevention-guideline-spotlight/ N4. Cerebral Torque. The new migraine prevention guideline: what the AAN and AHS recommend. 2 September 2026 (updated 4 October 2026). https://www.cerebraltorque.com/blogs/migrainescience/aan-ahs-migraine-prevention-guideline