Open questions — migraine¶
Last audited: 2026-08-30; OQ-1 and OQ-7 updated 2026-10-13 and OQ-9 updated 2026-10-20 and OQ-12/OQ-16 updated 2026-10-27, pending independent audit. IDs are stable: revise wording or status without renumbering. Tier 1 questions are practice-changing and prospectively testable now; Tier 2 questions need enabling cohorts, methods or longer follow-up.
Tier 1 — practice-changing, designable now¶
OQ-1 — What is the best first preventive strategy?¶
Question. In prevention-eligible adults, how do a conventional oral preventive, a CGRP antibody, a preventive gepant and—where chronic migraine is present—onabotulinumtoxinA compare in a pragmatic, preference-aware sequence?
Why open. Placebo-controlled efficacy is established, and erenumab and atogepant each outperformed topiramate on tolerability and response in direct trials. The published abstract of the 2026 AAN/AHS evidence review describes active-treatment comparisons as limited and generally of low or very low confidence; the multi-class sequence that jointly measures effectiveness, harms, cost and access remains unsettled (Pringsheim 2026, PMID 42673559; Reuter 2022, PMID 34743579; Reuter 2026, PMID 42492556).
OQ-2 — When should prevention start to avert chronic migraine?¶
Question. Among people with rising attack frequency, does early effective prevention produce a sustained reduction in chronic-migraine transition after treatment is withdrawn?
Why open. Frequency, overuse, obesity, mood and sleep predict progression, but predictor modification is not proof of disease modification; current trials mainly measure on-treatment symptom suppression (May 2016, PMID 27389092; Buse 2019, PMID 30589090).
OQ-3 — Withdrawal, prevention or both for MOH?¶
Question. Which drug- and patient-specific strategy—withdrawal first, immediate prevention, or both—maximizes sustained remission while minimizing withdrawal burden and relapse?
Why open. Combined withdrawal and early prevention performed best in one randomized strategy trial, while CGRP trials show improvement without formal detoxification; opioid/barbiturate dependence creates a distinct problem (Carlsen 2020, PMID 32453406; Sirilertmekasakul 2024, PMID 38564060).
OQ-4 — What should happen after the first acute treatment fails?¶
Question. After an adequate NSAID or triptan trial, which within-person sequence best delivers sustained pain freedom and function per cost and adverse event?
Why open. Network evidence ranks licensed oral agents, but cross-trial averages cannot estimate within-person response consistency, route effects or the best switch sequence (Karlsson 2024, PMID 39293828).
OQ-5 — Does combination prevention add enough benefit?¶
Question. In refractory chronic migraine, what is the incremental benefit, harm and cost of CGRP-pathway therapy plus onabotulinumtoxinA versus optimized monotherapy?
Why open. Both classes have efficacy, and a 2026 meta-analysis estimated benefit across observational combination cohorts, but it found no controlled comparative designs and substantial methodological limitations. A recruiting randomized active-comparator trial provides the direct route to an answer (Dodick 2010, PMID 20487038; Sarvari Soltani 2026, PMID 41721358; NCT07040813).
OQ-6 — Can patient-weighted endpoints change treatment choice?¶
Question. Do treatment rankings change when outcomes weight consistency, cognition, interictal burden, family function and patient priorities rather than mean MMD or a single two-hour endpoint?
Why open. Qualitative and measurement studies document priorities not fully represented in registration endpoints, but these outcomes have not been applied across treatments in a common comparative trial (Mangrum 2023, PMID 37140142; Hubig 2022, PMID 35941572).
OQ-7 — Which paediatric prevention package works best?¶
Question. In children and adolescents, how do CBT-first, drug-first and combined stepped-care pathways compare on headache, school participation, family burden and harms?
Why open. CHAMP found amitriptyline and topiramate no better than placebo with more harms, while CBT added benefit when both arms received amitriptyline; the optimal sequence remains unresolved (Powers 2017, PMID 27788026; Powers 2013, PMID 24368463). REACH is recruiting a CBT-with-versus-without-amitriptyline comparison (NCT05889624), while an active, non-recruiting factorial study varies delivery components of a mind–body intervention (NCT04715685).
OQ-8 — Which pregnancy-compatible pathway has the best balance?¶
Question. For people planning pregnancy or breastfeeding, which acute and preventive sequence minimizes maternal disability without avoidable fetal or infant risk?
Why open. Triptan observational data are comparatively mature, but newer CGRP therapies lack adequately powered prospective exposure data, and untreated disease carries its own harms (Marchenko 2015, PMID 25644494; Burch 2020, PMID 31579938).
Tier 2 — enabling science and longer-horizon questions¶
OQ-9 — Can a biomarker predict relative treatment response?¶
Why open. CGRP provocation, imaging, EEG, CSF and polygenic signals are mechanistically informative, but no locked multisite test predicts relative response to one preventive over another; REFORM's negative baseline suPAR result, and a later REFORM null for plasma amylin versus erenumab response, show why prospective candidate testing matters (Al-Khazali 2024, PMID 38859744; Karlsson 2025, PMID 40275185; Domínguez-Moreno 2026, PMID 42867283; Hautakangas 2022, PMID 35115687).
OQ-10 — What initiates and terminates the migraine attack?¶
Why open. Trigeminovascular, hypothalamic, brainstem and cortical changes are all documented, but observed activation does not identify the causal ordering or the switch that restores the interictal state (Ashina 2021, PMID 33773610).
OQ-11 — How does cortical spreading depolarization connect to headache?¶
Why open. CSD is the leading aura substrate, yet aura and headache dissociate and the exact junction from cortical event to meningeal/trigeminal pain remains incomplete. A 2026 lamotrigine cohort adds uncontrolled aura-day data but not a controlled CSD-targeted test (Fraser 2019, PMID 31847045; Uzun 2026, PMID 42591364).
OQ-12 — Does migraine treatment modify vascular risk?¶
Why open. Migraine with aura is associated with ischemic stroke. A 2026 active-comparator cohort offers limited reassurance against a large CGRP-antibody hazard relative to onabotulinumtoxinA, but few events and no untreated comparison leave shared causation and treatment-mediated protection unresolved (Zhang 2022, PMID 35451664; Gül 2026, PMID 42390441). Intermediate risk factors are now being measured: a 6,361-patient single-centre EHR cohort found no CGRP-targeting drug raised 12-month systolic BP faster than matched controls, with drug-specific BMI differences, but BP trajectories are not vascular events (Phan 2026, PMID 42928575).
OQ-13 — Is high-frequency gepant exposure free of MOH liability?¶
Why open. The live search through 2026-08-30 found expanding high-frequency and medication-overuse studies but no prospective exposure-threshold study demonstrating whether long-duration frequent gepant use causes, prevents or is neutral for MOH. Current reviews therefore support mechanistic reassurance, not proof of zero liability (Negro 2019, PMID 31081399; Ashina 2023, PMID 36732518).
OQ-14 — Why does migraine become sex-skewed after puberty?¶
Why open. Estrogen fluctuation explains an important attack trigger but not the full prevalence ratio; developmental, genetic, CGRP, social and recognition pathways remain entangled (Nappi 2022, PMID 35456034; Burch 2020, PMID 31579938).
OQ-15 — What is the biological meaning of the chronic threshold?¶
Why open. The ≥15-day line is operational, while disability, genetics and frequency vary continuously and people cross the boundary in both directions (IHS 2018, PMID 29368949; Adams 2015, PMID 25304766).
OQ-16 — Which global care-cascade intervention recovers most disability?¶
Why open. Diagnosis, consultation, acute treatment, prevention and follow-up gaps coexist; comparative implementation trials across resource settings are scarce (Katsarava 2018, PMID 29392600; Steiner 2018, PMID 29445880). Two 2026 high-income claims/record cohorts quantify where the cascade leaks without testing a fix: in England, 57.1% of headache presentations stayed undifferentiated and only 31.3% of propranolol and 17.9% of amitriptyline prescriptions reached recommended doses (Watson 2026, PMID 41513289); in the US, 46.1% of newly diagnosed adults filled an opioid within 3 years (Cohen 2026, PMID 42929214). Both are Pfizer-funded and measure prescriptions, not consumption or outcomes.
Dots not yet connected¶
| # | Dot A | Dot B | The missing junction | Powers |
|---|---|---|---|---|
| 2 | In 139 people given a 20-min CGRP infusion before 24 weeks of erenumab, ≥50% MMD response in weeks 13–24 was 61% (60/99) with a provoked attack versus 52% (13/25) without (p = 0.498); the authors conclude the provocation model cannot predict erenumab response and say it is uncertain whether that null extends to ligand antibodies or gepants (Al-Khazali 2024, PMID 38859744) | Response is not uniform across the pathway: among 252 people who had failed 1–4 CGRP antibodies, 29.7% still reached ≥50% MMD reduction on atogepant at 3 months, and each additional failed antibody lowered the chance of response (RR 0.79, 95% CI 0.64–0.97) (Muñoz-Vendrell 2025, PMID 41315923) | The provocation null result covers one receptor antibody only, while gepant response survives antibody failure in about a third of people. No study has tested whether CGRP provocation predicts gepant response, ligand-antibody response, or response after an antibody-to-gepant switch (2026-10-07 re-run: PubMed 15 on the written query and 40 on a broader provocation query; ClinicalTrials.gov 38; none is that test; queries in LOG) | Precision prescribing (OQ-9) |
| 5 | Interictal burden tracks attack frequency only loosely: MIBS-4 correlated with monthly migraine headache days at rs 0.21 at baseline and 0.32 at month 6 in 462 CONQUER participants (Lampl 2026, PMID 42327799); of 200 people treated with a CGRP antibody, 81 (40.5%) had fewer than 7 attacks in month 6, and 22 of those 81 (27.2%) still had MIBS-4 ≥3; monthly headache days did not differ between that residual-burden subgroup and the other 59 (4.11 vs 3.25, p = 0.127) (Silvestro 2026, PMID 42029417) | Two placebo-controlled RCTs now report MIBS-4: galcanezumab −1.9 vs placebo −0.8 at month 3 (Lipton 2023, PMID 36797223); every-other-day rimegepant −0.9 points vs placebo at week 12 (95% CI −1.4 to −0.4) (Pozo-Rosich 2025, PMID 41255098) | Neither trial abstract decomposes the MIBS-4 change into the part explained by fewer migraine days, and no PubMed record is that mediation analysis. No completed randomized active-comparator trial reports MIBS-4. NCT07535723 (not yet recruiting; onabotulinumtoxinA, a greater occipital nerve block, or both) lists MIBS-4 as a secondary outcome. The observational TRIUMPH Japan interim lists MIBS-4 among patient-reported outcomes and does not report a MIBS-4 contrast (Takeshima 2026, PMID 42018233) (2026-10-07 re-run: written PubMed query 4 records; ClinicalTrials.gov MIBS-4 or interictal burden 18 records; queries in LOG) | Patient-centred treatment ranking (OQ-6) |
| 7 | In 121 neurology-clinic adults, stigma (SSCI-8) added 6% unique variance to MIDAS severe disability and 5% to MSQ Emotional Function beyond headache frequency, allodynia, catastrophizing and mood, and mediated the frequency–disability relationship (Seng 2022, PMID 36321956) | An uncontrolled single-company workplace headache-education programme reached 73,432 employees: 82.9% of those without headache said they would change their attitude toward colleagues with headache, with an estimated 14.7 full-productivity days/year gained per employee with headache (Sakai 2023, PMID 36967710) | Stigma predicts disability, and education shifts attitudes, but no randomized trial has tested a stigma-targeted intervention against clinical outcomes such as disability, care seeking or consultation (2026-10-07: PubMed 3 records, ClinicalTrials.gov 1 record, none qualifying; queries in LOG) | Care seeking and accommodation |
| 8 | In CaMEO's 4,022 migraine–partner dyads, 48.2–57.4% of people with migraine reported reduced family participation at least monthly, and 71.7% with chronic migraine felt they would be better parents without headaches (Buse 2016, PMID 27132088); the 12-item IMPAC scale was derived from 13,064 respondents to quantify partner and child impact (Lipton 2017, PMID 28185239) | Family outcomes have been measured only in uncontrolled treatment cohorts. In 27 galcanezumab patient–caregiver dyads, the caregiver Relatives' Stress Scale fell 6.5 points over 6 months (95% CI 0.8–12.2) (Fofi 2022, PMID 36205100), and the 173-patient SQUARE erenumab cohort recorded IMPAC, with 87% at grade III–IV at baseline (Gantenbein 2025, PMID 40304557) | No randomized trial has used a family-spillover measure as an endpoint, so caregiver change cannot be separated from regression to the mean and expectation (2026-10-07: PubMed 2 records, none qualifying; ClinicalTrials.gov 3 records, only the single-arm NCT04968093 and the observational NCT07029126 measure caregivers; queries in LOG) | Household-level benefit |
| 10 | Short-term perimenstrual prevention beats placebo: 6 days of frovatriptan 2.5 mg twice daily cut the incidence of menstrually associated migraine from 67% to 41% in a 546-woman crossover (Silberstein 2004, PMID 15277618), and a 14-RCT, 4,601-patient network meta-analysis ranked it the best short-term option (OR 1.87, 95% CI 1.48–2.38 vs placebo) (Zhang 2023, PMID 37400775) | Phase 3 CGRP-pathway trials in menstrual migraine are short perimenstrual courses versus placebo or acute care, not continuous daily prevention: atogepant 60 mg or placebo once daily for 7 days starting 3 days before menses, for 3 cycles (NCT06806293, n=468, active, not recruiting; that schedule is in jRCT2051240268 and is not stated on the ClinicalTrials.gov record); 7-day perimenstrual rimegepant versus as-needed rimegepant, placebo, or standard acute care (NCT06641466, n=723, recruiting); ubrogepant once daily for 7 days starting 3 days before menses versus placebo (NCT06417775, n=496, active, not recruiting) | Each of those trials is against placebo or acute-only care. No trial randomizes perimenstrual short-term prevention against continuous daily prevention and counts total treatment days and acute-medication days. The nearest design puts everyone on an extended-cycle contraceptive and randomizes frovatriptan against placebo only during the hormone-free interval (Coffee 2014, PMID 24450487) (2026-10-07 re-run: written PubMed query 17; broader hormone query 19; ClinicalTrials.gov menstrual migraine 39; none a strategy head-to-head; queries in LOG) | Minimize total treatment and overuse days |
Governance¶
- Close an OQ only when direct evidence answers the stated comparison, population and outcome; adjacent efficacy evidence is insufficient.
- Preserve IDs when a question is narrowed or moved between tiers.
- Promote a dot to a numbered OQ when a feasible design and decision consequence are explicit.