FDA-Authorized Non-Invasive Neuromodulation Devices for Migraine and Cluster Headache: Pathways, Indications and Clinical Evidence

FDA-authorized non-invasive neuromodulation devices for migraine and cluster headache: pathways, indications and clinical evidence

FDA has authorized five distinct types of non-invasive neuromodulation device for primary headache disorders. Each type entered the market through its own De Novo classification, which created a Class II regulation with special controls. Later 510(k) clearances then broadened indications, added new populations, moved some devices to over-the-counter (OTC) use and brought in competing devices. The five modalities are:

• transcutaneous supraorbital (trigeminal) electrical stimulation
• single-pulse transcranial magnetic stimulation (sTMS)
• non-invasive vagus nerve stimulation (nVNS)
• remote electrical neuromodulation of the upper arm (REN)
• thermal vestibular stimulation

Key takeaways

• Five De Novo classifications anchor the field. These are Cefaly (DEN120019), Cerena TMS (DEN130022), gammaCore (DEN150048), the ThermoNeuroModulation (TNM) Device (DEN170023) and Nerivio Migra (DEN180059). Each was classified into Class II with device-specific special controls ^{357380268331307}.
• The De Novo evidence came from randomized, sham-controlled trials, and the results were not always clean. FDA granted Cefaly even though one primary comparison (reduction in migraine days) was not statistically significant ³⁶³. The Cerena effectiveness result came from a post-hoc analysis of subjects who treated an attack while in pain ³⁸¹. The gammaCore pivotal ACT1 study missed its overall endpoint but showed benefit in episodic cluster headache ²⁷⁴.
• Most indication expansions went through the 510(k) pathway. For gammaCore, these included acute migraine, migraine prevention, adjunctive cluster prevention, adolescents and hemicrania syndromes ^473453668569. For Nerivio, they included chronic migraine, adolescents, prevention and children aged 8 and older ^705778681754. For SpringTMS, they included prevention and adolescents ⁷³¹⁷⁴³. Some of these expansions had new clinical data behind them. Others relied on extrapolation or real-world data.
• The documented Cefaly Rx-to-OTC switch relied on human factors data, not a new clinical trial. Cefaly Dual's OTC clearance (K201895) stated that no clinical performance testing was done and that a human-factors validation study supported the switch ⁴⁹⁹⁵⁰⁰.

Regulatory framework: the five headache neuromodulation classifications

Note on the gammaCore regulation citation. The gammaCore De Novo decision summary lists 21 CFR 882.8592 ²⁶⁸. A later FDA letter states that it corrects the April 14, 2017 classification order and gives 21 CFR 882.5892 ²⁶⁹. Use the corrected number when citing the classification.

The special controls follow a common pattern across all five types:

• biocompatibility of patient-contacting components
• electromagnetic compatibility (EMC) and electrical, mechanical and thermal safety testing
• software verification, validation and hazard analysis
• characterization of device output
• labeling requirements

Each type also has its own modality-specific controls. The TMS controls require magnetic-output, field-map and sound-level characterization ³⁸⁶. The thermal vestibular controls require waveform and temperature limits and validation of earpiece cleaning ³³⁸. The REN controls require impedance monitoring and electrode performance testing ³¹⁵. The Cefaly and Cerena controls also call for clinical performance data ³⁴⁶³⁸⁶.

Transcutaneous supraorbital stimulation (product code PCC)

Cefaly De Novo (DEN120019)

Cefaly is a TENS device worn on the forehead. Each session lasts 20 minutes, and intensity ramps up over the first 14 minutes ³⁵⁴³⁵⁵. FDA granted it for prophylactic treatment of episodic migraine in adults, as a prescription device that is contraindicated in patients with pacemakers or implanted or wearable defibrillators ³⁵⁶.

Evidence. The pivotal PREMICE study was a prospective, multicenter, double-blind, randomized, sham-controlled trial run at five Belgian headache clinics. A one-month run-in was followed by a 90-day treatment period ³⁵⁸³⁵⁹. The trial randomized 67 subjects (34 active, 33 sham), and 59 completed per protocol ³⁶⁰³⁶¹. There were two primary outcomes: change in monthly migraine days, and the proportion of patients with at least a 50% reduction in migraine days ³⁶². FDA's benefit-risk discussion says that the responder finding supported benefit. It also says the between-group comparison for reduction in migraine days was not statistically significant ³⁶³.

There were no serious adverse events in the trial ³⁶⁴. In a European post-market survey of 2,313 users, 4.3% reported complaints, and all were non-serious and reversible ³⁶⁵. FDA concluded that probable benefits outweighed risks while acknowledging the mixed primary results ³⁶⁷³⁶⁸.

510(k) progression for Cefaly

Relivion and other PCC devices

Neurolief's Relivion (K203419) was cleared for acute treatment of migraine in adults, with Cefaly Acute as the predicate. Unlike Cefaly, it stimulates at both the forehead and the occiput ^633635636. This is one of the few PCC 510(k)s with substantial new clinical data:

• RIME study. A sham-controlled trial in 131 patients. Pain relief at 2 hours was 60% with active treatment versus 37% with sham in the modified ITT population (p=0.0180) ⁶⁴³⁶⁴⁴.
• SP-301 study. A randomized, double-blind, placebo-controlled study in 55 patients. Pain-intensity improvement favored active treatment at 1, 2 and 24 hours ⁶³⁹⁶⁴⁰.
• OS-TNS study. Thirty adults. Mean VAS pain change was −79.2% with active treatment versus +14.9% with sham (P=0.0002) ⁶³⁸.

A later Relivion 510(k) (K212106) added a remote physician interface and included no new clinical study ⁷²²⁷²³.

Several other 510(k) devices use Cefaly as the predicate and include no new clinical data:

• ALLIVE (K192773, prevention, Rx) relied on a clinical evaluation report based on the predicate's clinical data ⁷⁹⁶⁸⁰².
• HeadaTerm (K172450, prevention, Rx) listed clinical testing as "Not Applicable" ⁶¹⁶⁶²².
• HeadaTerm 2 (K230782) was cleared OTC on the basis of a 24-participant usability verification ⁸⁶⁸⁸⁷².
• Elexir (K211380) relied on the predicate's clinical data ⁵⁹⁴⁵⁹⁹. Elexir 2.0 (K242719) is a prescription device for acute and preventive treatment ⁸³¹⁸³².
• Milieve (K233751, OTC) ⁵⁸¹⁵⁸² and DOOPANG (K232749, OTC) ⁵¹⁸⁵¹⁹ cite Cefaly Dual OTC as the predicate.
• Migraine TENS Digital Pain Reliever (K210364, OTC) also cites Cefaly Dual OTC ⁸⁵⁸⁸⁶⁰.
• Enso for Migraine (Hinge Health, K254216, cleared 2026-04-16) is an OTC device for acute and preventive treatment. No clinical testing was done. Effectiveness was supported by published literature on the predicate, and a human factors study supported OTC use ^3651656657.

Single-pulse transcranial magnetic stimulation (product code OKP)

Cerena De Novo (DEN130022)

Cerena delivers two brief magnetic pulses. Its device type is defined as directing pulsed magnetic fields at discrete brain regions to induce electrical currents ³⁷⁷³⁷⁸. The original indication was limited to acute treatment of pain associated with migraine with aura ³⁷⁹.

Evidence. The trial enrolled 201 subjects, randomized 1:1 to device or sham. The effectiveness analysis used 113 subjects who treated an attack while in pain (53 active, 60 sham). FDA's summary describes this as a post-hoc analysis ³⁸¹³⁸². In that population:

• pain-free at 2 hours: 37.74% with Cerena versus 16.67% with sham (p=0.0181)
• pain-free at 24 hours: 33.96% versus 10% (p=0.0025) ³⁸³³⁸⁴

Adverse event rates were similar between arms, and no seizures were reported ³⁸⁵. FDA noted that effectiveness on associated migraine symptoms was not demonstrated and that long-term data were not available ³⁸⁸.

SpringTMS lineage

• K140094 (SpringTMS). Same indication as Cerena, with hardware and use-authorization changes. The summary cites the Cerena predicate as K130556 ^606608609.
• K161663 (sTMS mini). A smaller device with an internal battery. Supported by non-clinical testing only ^430434438.
• K162797 (SpringTMS). Expanded the indication to "acute and prophylactic treatment of migraine headache," dropping the restriction to migraine with aura ⁷³¹⁷³⁴. Support came from a prospective, single-arm, multicenter observational study measured against a performance goal. In the full analysis set (n=132), mean monthly migraine headache days fell by 2.8 from a baseline of 9.1 (P<0.0001) ⁷³⁵⁷³⁶. Of 217 subjects, 28.57% reported at least one adverse event, and none were serious ⁷³⁷.
• K182976 (SpringTMS). Added adolescents aged 12 and older. No new clinical testing was done, and the summary refers back to the predicate's single-arm ESPOUSE study ^743746747.
• K230358 (SAVI Dual). Same indications as SpringTMS, adding a cellular use-authorization option. Supported by non-clinical testing ^818822823811.

The prevention claim for sTMS is the clearest example in this field of a new indication cleared through a 510(k) on single-arm data compared against a performance goal, without a sham control.

Non-invasive vagus nerve stimulation (product code PKR)

gammaCore De Novo (DEN150048)

gammaCore stimulates the cervical vagus nerve through the skin on the side of the neck. It was granted for acute treatment of pain associated with episodic cluster headache in adults ²⁶⁵. The labeling states that safety and effectiveness in chronic cluster headache have not been established ²⁶⁷. The decision summary cites animal data on reduced trigeminal nucleus caudalis glutamate as a potential mechanism ²⁶⁶.

Evidence. Two randomized, double-blind, sham-controlled trials supported the grant ²⁷¹.

• ACT1 (pivotal). The endpoint was response at 15 minutes for the first attack. Overall, 26.7% of active subjects responded versus 15.1% with sham (P=0.10). In the episodic cluster headache subgroup, the rates were 34.2% versus 10.6% (P<0.01) ²⁷⁴.
• ACT2. The endpoint was the proportion of attacks pain-free at 15 minutes. Overall, the rates were 13.5% versus 11.5% (P=0.71). In episodic cluster headache, they were 47.5% versus 6.2% (P<0.01) ²⁷⁶²⁷⁷.

There were no device-related serious adverse events ²⁷⁸. FDA noted uncertainty in the size of clinical improvement and no demonstrated benefit in chronic cluster headache ²⁸⁶.

Indication expansion through 510(k)

gammaCore shows how far a single De Novo classification can be extended through 510(k)s. Several of the added indications rested on trials that missed their prespecified primary endpoints, on post-hoc analyses, or on case series.

Remote electrical neuromodulation (product code QGT)

Nerivio Migra De Novo (DEN180059)

Nerivio is a smartphone-controlled stimulator worn on the upper arm. It treats headache by stimulating at a site away from the head and neck ³⁰²³⁰⁵. It was granted for acute treatment of migraine in adults without chronic migraine ³⁰⁶.

Evidence. The pivotal trial was a 252-subject, multicenter, randomized, double-blind, sham-controlled study ³⁰⁸. In the modified ITT population:

• pain relief at 2 hours: 66.7% with active treatment versus 38.8% with sham, a gain of 27.9 percentage points (p<0.0001)
• pain freedom at 2 hours: 37.4% versus 18.4% (p=0.0036)
• relief of the most bothersome symptom: 46.3% versus 22.2% ³¹¹³¹²

Device-related adverse events occurred in 4.8% of active and 2.4% of sham subjects. All were mild, and none were serious ³¹³.

Nerivio 510(k) expansions

• K201824. Removed the chronic migraine exclusion ⁷⁰⁵⁷⁰⁸. In the supporting data, 57.1% of participants had pain relief in at least half of their treated attacks ⁷⁰⁹.
• K203181. Lowered the minimum age to 12 ⁷⁷⁸⁷⁸¹. In the adolescent study (final analysis set n=39), pain relief at 2 hours was 71.8% and pain freedom was 35.9% ⁷⁸⁴.
• K223169. Added preventive treatment, with the device used every other day ⁶⁸¹⁶⁸⁴. In a double-blind active-versus-sham trial, monthly migraine days fell by 3.97 with active treatment versus 1.28 with sham, a gain of −2.69 days (p<0.001) ⁶⁸⁵. Adolescent real-world app data were also submitted ⁶⁸⁸.
• K232152 (Nerivio Infinity). Rechargeable version. No new clinical trials ⁴⁸⁷⁴⁸⁸.
• K241756. Lowered the minimum age to 8 ⁷⁵⁴⁷⁵⁷. Support came from a retrospective analysis of real-world data on 293 children aged 6 to 11, which reported no adverse events ⁷⁵⁹⁷⁶⁰.
• K250405. Labeling updates on long-term use and use in pregnancy, supported by post-market real-world data and a published survey study ⁵¹⁰⁵¹¹.

Nerivio is the clearest example in this field of real-world data supporting pediatric expansion through 510(k).

Thermal vestibular stimulation (product code QAR)

The TNM Device (DEN170023) delivers controlled thermal waveforms through earpieces placed in the ear canals to stimulate the vestibular system ³²⁸³²⁹. It was granted for prophylactic treatment of episodic migraine in patients aged 12 and older ³³⁰.

Evidence. The supporting study was a multicenter, triple-blinded, randomized, placebo-controlled study of adjunctive treatment in adults aged 18 to 65. It randomized 100 subjects: 50 active, 30 placebo and 20 partially active placebo ³³²³³³. The primary endpoint was the within-group reduction in monthly migraine days in the active group from baseline to month 3 ⁸⁴⁷. Excluding subjects with more than 15 headaches per month, the results were:

• per protocol: −3.6 days with active treatment versus −0.9 with placebo (p=0.0142)
• ITT: −3.3 versus −0.9 (p=0.0246) ^849850851

The ≥50% responder rates (57.1% versus 33.3% per protocol) were not statistically significant ⁸⁵²⁸⁵³. Use in patients aged 12 to 17 was supported by extrapolation, since the trial enrolled only adults ³⁴⁰. No serious adverse events occurred ³³⁶.

Cross-cutting regulatory observations

1. De Novo with sham-controlled RCTs set the bar for each new mechanism. Every founding classification relied on a randomized sham or placebo comparison. FDA accepted mixed or post-hoc primary results where risks were low and could be mitigated by special controls ^363381286.
2. 510(k)s have carried new indications, not only design changes. Prevention claims for sTMS, nVNS and Nerivio ^734456684, and new disorders for gammaCore ⁴⁷³⁵⁷², were cleared by 510(k) under the same regulation, with FDA treating the added claim as the same intended use even when the predicate indication was narrower.
3. The evidence behind pediatric expansions varies. It ranged from no new clinical data (SpringTMS K182976) ⁷⁴⁷, to adult extrapolation (gammaCore K203546, TNM) ⁶⁷¹³⁴⁰, to prospective adolescent studies and real-world data (Nerivio K203181, K241756) ⁷⁸⁴⁷⁵⁹.
4. OTC conversions in the PCC class depend on usability evidence. Cefaly Dual OTC, HeadaTerm 2 and Enso were supported by human-factors or usability work, not new effectiveness trials ^500872657.
5. Follow-on PCC devices increasingly cite literature or the predicate's clinical data. Examples include ALLIVE, Elexir and Enso ^802599656. This keeps the bar to entry for supraorbital TENS competitors low.

Open questions worth pursuing

• How CMS coverage policy and payer evidence reviews treat these devices compared with their FDA indications.
• Postmarket safety signals for each product code.
• Whether any newer De Novo requests for headache (for example, non-neuromodulation digital or optical therapies) are building on these special controls.
• How EU MDR notified-body certification compares with the US pathway for the same devices.