Main

Migraine is a common, globally recognized1, neurological disorder characterized by recurrent disabling attacks involving headache and symptoms of brain dysfunction2. The attack has recognized phases: premonitory (prodrome), aura, headache and postdrome, which, although distinctive, can overlap3. The canonical manifestations of an attack—lateralized, throbbing headache with associated sensitivity to light (photophobia), sound (phonophobia) and head movement4—have received considerable attention for the last three decades as treatments aimed at the headache phase of the attack were developed5. The identification of the role of calcitonin gene-related peptide (CGRP)6,7 and the utility of CGRP blockers for both acute and preventive treatment of migraine5 offer the possibility of exploring migraine pathophysiology with new tools.

Migraine symptoms that can occur in the premonitory phase (prodrome) fall broadly into three groups: first, higher center, such as cognitive impairment, manifests as difficulty concentrating or thinking—brain fog as it is often labeled by patients—and fatigue. Second, symptoms broadly reflect homeostatic dysfunction, such as food cravings or polyuria and, third, symptoms usually associated with the headache phase can occur in the premonitory phase, such as sensitivity to light (photophobia) or sound (phonophobia)8. Prodromal symptoms are highly predictive of impending headache9 and are generally considered to be common when enquired after in adults10,11 and children12,13. Functional brain imaging has identified activations in the central nervous system, such as the hypothalamic region14,15, in the premonitory phase. Mapping symptoms to imaging findings suggests central nervous system origins for the migraine attack16, with implications for understanding the pathophysiology and, importantly, where to target therapies.

Therapeutics in migraine have advanced considerably over the last three decades3. It has been argued whether the origin of both the attack and the pain is peripheral or central17, and thus how best to target therapies. Moreover, previous studies of therapeutics in the acute treatment of the premonitory phase (prodrome) have focused on the potential for blocking the onset of headache18,19. The analysis of the prevention of headache in the PRODROME study has demonstrated that ubrogepant given in the prodromal phase prevented moderate or severe headache, when compared with placebo, for the following 48 h (ref. 20). In the present study, we report the effect of ubrogepant, a small-molecule CGRP receptor antagonist21, on the prodromal symptoms commonly reported by the study participants, a prespecified additional endpoint of the PRODROME trial. In addition to the clinical benefit of resolving these symptoms, the findings may offer insights into the role of the central nervous system in the treatment of migraine. The work has been presented in preliminary form at the 75th Annual Meeting of the American Academy of Neurology (Boston, MA, 22–27 April 2023)22.

Results

Participants and baseline characteristics

A total of 1,087 participants were screened, with 518 randomly assigned to double-blind crossover treatment (Fig. 1). The safety and modified intention-to-treat (mITT) populations included 480 and 477 participants, respectively. Participants had a mean age of 42.3 years (Table 1). Most of the participants in the safety population were female (87.7%, n = 421 of 480), with most identifying as White (88.1%, n = 423 or 480) and non-Hispanic (92.7%, n = 445 or 480; Table 1). Of the 518 randomized participants, 84.6% (n = 438) completed the trial and 15.4% (n = 80) discontinued. Lack of treating two ‘qualifying prodrome events’ within the 60-d double-blind treatment period was the most common reason for discontinuation (10.0%, n = 52 of 518). Additional information about PRODROME participants has been previously reported20.

Fig. 1: Participant flow.
figure 1

Purple boxes and arrows indicate the number and flow of qualifying prodrome events treated with ubrogepant. Gray boxes and arrows indicate the number and flow of qualifying prodrome events treated with placebo. mITT, modified intention-to-treat.

Full size image
Table 1 Baseline demographics and migraine characteristics by treatment sequence (safety population)
Full size table

Outcomes

The five most common prodromal symptoms identified at baseline (pre-dose) in the double-blind treatment period for placebo-treated (n = 449) and ubrogepant 100 mg-treated (n = 448) events were photophobia (60.8% and 60.9% of events, respectively), fatigue (50.3% and 50.7%), neck pain or stiffness (40.1% and 40.2%), phonophobia (36.1% and 35.9%) and dizziness (31.0% and 29.0%; Table 2). Although the cognitive impairment symptoms of difficulty concentrating (22.5% and 23.0%) and difficulty thinking (16.5% and 15.4%) were the eighth and eleventh most common prodromal symptoms at baseline before treatment, when considering them as one category ‘either difficulty concentrating or difficulty thinking or both’ was reported in 33.0% of placebo-treated and 32.1% of ubrogepant-treated events, making it the fifth most common prodromal symptom (Table 2). Prodromal symptoms of moderate-to-severe intensity were reported at baseline in 31.7–57.2% of placebo-treated and ubrogepant-treated events: specifically, photophobia (42.9% and 42.1% of events, respectively), fatigue (52.2% and 51.5%), neck pain or stiffness (57.2% and 55.6%), phonophobia (38.9% and 36.0%), dizziness (31.7% and 30.8%), difficulty concentrating (45.5% and 39.8%) and difficulty thinking (45.9% and 42.0%) (Table 2 and Supplementary Table 1).

Table 2 Most common prodromal symptoms identified at baseline (pre-dose) by occurrence (mITT population)
Full size table

Among ubrogepant- and placebo-treated events, respectively, 19.5% and 12.5% were associated with an absence of photophobia starting at hour 2 (OR = 1.72 (95% CI = 1.13–2.61)), 27.3% and 16.8% with an absence of fatigue starting at hour 3 (OR = 1.85 (95% CI = 1.17–2.92)), 28.9% and 15.9% with an absence of neck pain starting at hour 3 (OR = 2.04 (95% CI = 1.25–3.32)), 50.7% and 35.8% with an absence of phonophobia starting at hour 4 (OR = 1.97 (95% CI = 1.38–2.80)) and 88.5% and 82.3% with an absence of dizziness at 24 h post-dose (OR = 1.82 (95% CI = 1.00–3.30)) (Fig. 2 and Supplementary Table 2). An absence of cognitive symptoms was also observed: as early as 1 h post-dose for difficulty concentrating in 8.7% and 2.1% of ubrogepant- and placebo-treated events, respectively (OR = 4.26 (95% CI = 1.17–15.54)) and at 6 h for difficulty thinking in 56.9% and 41.8% (OR = 2.05 (95% CI = 1.14–3.71)) of events (Fig. 2 and Supplementary Table 2).

Fig. 2: Prodromal symptoms of any intensity for the most common prodromal symptoms at timepoints post-dose.
figure 2

The percentage of participants continuing to have prodromal symptoms at timepoints post-dose are shown. a–g, Participants who reported the prodromal symptoms at pre-dose: sensitivity to light (placebo, n = 273 of 449, ubrogepant 100 mg, n = 273 of 448) (a); tired, sleepy or fatigue (placebo, n = 226 of 449, ubrogepant 100 mg, n = 227 of 448) (b); neck pain or stiff neck (placebo, n = 180 of 449, ubrogepant 100 mg, n = 180 of 448) (c); sensitivity to sound (placebo, n = 162 of 449, ubrogepant 100 mg, n = 161 of 448) (d); dizziness, lightheaded, vertigo or imbalance (placebo, n = 139 of 449, ubrogepant 100 mg, n = 130 of 448) (e); difficulty concentrating (placebo, n = 101 of 449, ubrogepant 100 mg, n = 103 of 448) (f); and difficulty thinking (placebo, n = 74 of 449, ubrogepant 100 mg, n = 69 of 448) (g).

Full size image

Analysis censoring data collected after rescue medication use for headache (if needed) yielded similar results (Supplementary Table 3 and Supplementary Fig. 2). Rescue medication use within 24 h post-dose was 21.7% for ubrogepant-treated events compared with 39.4% for placebo-treated events (Supplementary Table 4).

Safety

Treatment-emergent adverse events (TEAEs) were reported for 55 of 462 (12%) qualifying prodrome events treated with placebo and 77 of 456 (17%) events treated with ubrogepant 100 mg within 48 h of administration. Treatment-related TEAEs were identified in 42 of 462 (9%) of placebo-treated events and 60 of 456 (13%) ubrogepant 100 mg-treated events. As previously reported20, there were no adverse events (AEs) leading to study discontinuation and no serious AEs. The most common AEs (≥2%) after placebo and ubrogepant administration were nausea (3% versus 5%), fatigue (2% versus 3%), dizziness (3% versus 2%) and somnolence (1% versus 2%).

Discussion

The results suggest that ubrogepant, a gepant, CGRP receptor antagonist, administered in the premonitory (prodromal) phase of migraine, when headache is absent, may promote the resolution of common prodromal symptoms. Moreover, in this trial, 31.7–57.2% of premonitory symptoms were moderate to severe in intensity and associated with functional disability20,23. As premonitory symptoms can be disabling, their treatment alone is clinically relevant, beyond the consideration that treatment during the prodrome prevents headache onset and improves function over 24–48 h, as demonstrated in the primary analysis of the study20. Greater awareness of the clinical symptomatology of the prodromal phase, as well as the availability of effective treatment, offers a major opportunity to improve the treatment of acute migraine.

It is interesting to examine the time course and magnitude of efficacy of ubrogepant for the different prodromal symptoms. The most common symptom, photophobia, which was present in approximately 61% of qualifying prodromal events, appeared to resolve after treatment with ubrogepant as early as 2 h post-dose, with a consistent effect across the first 8 h and an effect still evident at 48 h post-dose. Although being reported by only approximately 22% of participants, difficulty concentrating may also benefit from ubrogepant treatment, with resolution as early as 1 h post-dose and up to 6 h, and the trend continuing to 24 h post-dose. In contrast, ubrogepant appeared to have limited impact on the resolution of dizziness, which was present in approximately 30% of qualifying prodromal events, and these effects were observed primarily at the 24-h timepoint. Given the varied pathophysiology of the different symptoms, one would not expect ubrogepant to be equally effective or have a similar time course for all prodromal symptoms. Still, similar trends were observed overall for prodromal symptom resolution, which appeared to occur more frequently with ubrogepant than placebo.

The symptoms of the premonitory (prodromal) phase of a migraine attack have been recognized for more than a century24. In a case series of 50 patients, 17 identified symptoms on the day before their headache25, whereas a larger series of 530 reported 160 patients identifying premonitory (prodromal) symptoms and speculated on a role of the hypothalamus in the pathophysiology26. Studies based on clinic populations in adults10,11,27 and children12,13 and on large cross-sectional material28 have consistently identified premonitory symptoms, such as fatigue and cognitive dysfunction. In patients who recognize these symptoms, they are highly predictive of impending headache9, although the reliability of prediction remains an issue29.

Many of these common premonitory symptoms, such as photophobia, fatigue, phonophobia, cognitive impairment—difficulty concentrating or difficulty thinking, or both—and dizziness, may well originate in the brain. It has long been argued whether migraine is primarily a disease of the brain or of peripheral, specifically vascular, origin30. The new data firmly support a brain origin for migraine attacks. Both symptoms of the premonitory phase, such as photophobia and thirst or food cravings, and outcomes from functional imaging studies implicate brain regions in the earliest phase of a migraine attack31. As an example, laboratory studies implicating neuropeptide Y in the modulation of trigeminovascular nociceptive traffic32, and its localization in the hypothalamus33, or studies of ventral tegmental area reward mechanisms and their interaction with trigeminovascular nociception34 offer support to neuroimaging studies that demonstrate activation of these regions in the premonitory phase of an attack14,15,35. Photophobia, a symptom classically associated with the headache phase of migraine4, may also be reported in the premonitory (prodromal) phase and is indeed the most commonly reported symptom in the present study. Brain imaging has shown a signature in the visual cortex for premonitory phase photophobia36, demonstrating a central nervous system-associated biological change. Similarly, brainstem sites, such as the rostral ventromedial medulla, are implicated in trigeminovascular nociceptive modulation37 with variation over the cycling of migraine38.

There remains considerable debate around the site of action of anti-migraine treatments. The triptans, serotonin 5-HT1B/1D receptor agonists, were developed on the inchoate basis that they constricted dilated cranial blood vessels which were the cause of the pain39. Several findings, including dissociation of the timing of a vascular effect and the pain response40, and the neural effect of the drugs41, have led to the broad adoption of a view that they act primarily through a neural target42. Although sumatriptan can be measured in the cerebrospinal fluid43, and the exclusively neurally acting ditans, 5-HT1F receptor agonists44 are both effective45 and have brain access46, migraine therapies are widely considered to act outside the brain. A positron emission tomography study of telcagepant showed low receptor occupancy after a less than optimal dose of 140 mg47,48, leaving unresolved the issue of whether there was potential for gepants to have effects in the brain. Although it may be argued that the effects of ubrogepant in the present study on photophobia and phonophobia or neck discomfort reflect a primary peripheral site of action with brain consequences, it is difficult to argue that position for cognitive dysfunction, leaving a brain site of action as a more attractive, overall hypothesis. Broadly, the findings of the clinical trial support imaging studies that have identified central nervous system sites as the locus of initiation of a migraine attack.

Limitations

The participants studied could identify prodromal symptoms that were reliably followed by headache as determined via a rigorous screening procedure20. Clinical experience suggests that it is not complex to point out these prodromal symptoms to patients and offer the possibility of the association, which patients can then recognize. Nevertheless, widespread adoption of this approach will require further real-world experience. Whether gepant intervention alters migraine aura or influences the postdrome49 was not rigorously explored in the present study and offers interesting avenues for insights and treatment opportunities in the future.

The primary objective of the PRODROME trial was to evaluate the efficacy of ubrogepant on the occurrence of headache after administration during the prodrome phase of a migraine attack. The trial was not primarily designed to evaluate the impact of ubrogepant on resolution of prodromal symptoms; these endpoints were prospectively defined as additional endpoints in the protocol. These analyses were outside the hierarchical gatekeeping procedure to control for type-1 error and were therefore not controlled for multiple comparisons. Although our results suggest a consistent amelioration of common prodromal symptoms after ubrogepant compared with placebo, additional studies specifically designed to evaluate the effect of acute treatment on prodromal symptoms are warranted.

Additional limitations resulting from the study design include participants being instructed to administer the study drug when they felt confident that a headache would follow within 1–6 h, and not at the onset, or first recognition, of prodromal symptoms. Furthermore, we did not analyze differences in symptom resolution between those who did and those who did not develop headache. Analyses censoring data collected after rescue medication use for headache (if needed) showed results similar to the uncensored data, suggesting that a difference in rescue medication use between ubrogepant and placebo treatment groups was probably not a confounding factor (Supplementary Table 3 and Supplementary Fig. 2).

Conclusion

Based on this double-blind, placebo-controlled trial, ubrogepant 100 mg, when administered in the premonitory (prodromal) phase of migraine, before a headache has commenced, may treat symptoms such as photophobia, phonophobia and cognitive dysfunction. The clinical phenotype of the premonitory phase, its functional neuroimaging findings and the corroborating experimental neurobiology all point to a pathophysiology in the central nervous system. Reversing these symptoms with a gepant emphasizes the importance of the brain in migraine and offers the promethean possibility that targeting central nervous system mechanisms will be a fruitful path to new therapeutics in this common and disabling disorder.

Methods

Trial design

The trial design for the PRODROME study has been previously reported20. PRODROME (NCT04492020) was a phase 3, multicenter, randomized, double-blind, placebo-controlled, crossover trial conducted from 21 August 2020 through 19 April 2022, at 75 research centers and headache clinics in the United States of America (Supplementary Table 5). After a 60-d screening period, participants with migraine, who could reliably identify migraine attacks with prodromal symptoms that were reliably followed by headache within 1–6 h at least 75% of the time, were randomized to sequence A or B in a 1:1 ratio and entered the double-blind treatment period, during which they were tasked to treat two qualifying prodrome events within a 60-d period (Supplementary Fig. 1). A qualifying prodrome event was defined by the presence of prodromal symptoms that the participant was confident would be followed by a headache within 1–6 h. Headache should not be present at the time of a qualifying prodrome event and the participant should not have had a headache in the previous 48 h. Acute treatment should not have been used within 48 h before the qualifying prodrome event. Defining a qualifying prodrome event as 1–6 h before the expected headache onset was done to provide time for the acute treatment to take effect and aid in differentiating aura symptoms from prodromal symptoms.

Participants in sequence A received placebo to treat the first qualifying prodrome event and ubrogepant 100 mg to treat the second qualifying event, whereas participants in sequence B received ubrogepant 100 mg to treat the first qualifying prodrome event and placebo to treat the second qualifying event. The two treatments had at least 7 d of washout in between.

During the double-blind study period, after taking the study drug, rescue medication was permitted only after headache onset. For a mild headache, rescue medication was permitted 24 h after headache onset. For a headache of moderate or severe intensity, rescue medication was permitted at any time after headache onset. No rescue medication was permitted during the qualifying prodrome event if a headache had not yet occurred. Permitted rescue medications included nonsteroidal anti-inflammatory drugs, acetaminophen, triptans, ditans, ergots, analgesics or combination analgesics, opioids or antiemetics.

During screening, participants identified and confirmed prodromal symptoms from a predetermined list of 30 symptoms. This list included common and uncommon prodromal symptoms, as well as the options ‘other’ and ‘a feeling, not otherwise described or difficult to describe’.

An automated interactive web response system was used to manage randomization, which occurred at visit 2. To maintain the blind, identical blister cards were dispensed at visit 2 and the visit after treatment of the first qualifying prodrome event (visit 3). Each blister card contained two tablets of either placebo or ubrogepant 50 mg. Participants in treatment sequence A received a card with placebo at visit 2 and ubrogepant 100 mg at visit 3, whereas participants in treatment sequence B received a card with ubrogepant 100 mg at visit 2 and placebo at visit 3. During the double-blind treatment period, each participant was instructed to take the study drug for each qualifying prodrome event as soon as they were confident that a headache would inevitably follow within 1–6 h.

Trial participants

Eligible participants were adults (aged 18–75 years), female or male (based on self-report) with at least a 1-year history of migraine with or without aura consistent with a diagnosis according to the International Classification of Headache Disorders, 3rd edn4 and a history of two to eight migraine attacks per month with moderate-to-severe headache in each of the 3 months before screening. At the screening visit, participants were asked whether they could identify migraine attacks in which prodromal symptoms were present and likely to be followed by headache within 1–6 h at least 75% of the time. If participants answered in the affirmative, participants received an eDiary to record all their qualifying prodrome events during the 60-d screening period. To be eligible for randomization, participants were required to record 3–16 qualifying prodrome events during the 60-d screening period, with at least 75% followed by a headache, of any intensity, within 1–6 h.

Exclusion criteria have been reported previously20. Briefly, exclusion criteria included: history of clinically important hematological, endocrine, pulmonary, renal, hepatic, gastrointestinal or neurological disease that has not been stable for at least 1 year, history of cardiovascular, cerebrovascular or neurological disease, confounding psychiatric conditions including dementia or epilepsy, difficulty distinguishing migraine from other tension type or other headache types, chronic migraine, trigeminal autonomic cephalalgia, painful cranial neuropathy, overuse of medication for migraine (opioids or barbiturates >2 d per month, triptans or ergots ≥10 d per month, simple analgesics ≥15 d per month) within the last 3 months or previous CGRP injectable use within the previous 3 months.

Outcome measures

The primary endpoint of the PRODROME trial was absence of moderate or severe intensity headache within 24 h post-dose. Prespecified additional endpoints (that is, exploratory endpoints) were the absence of the five most common prodromal symptoms, of any intensity (mild, moderate or severe), at each of the prespecified timepoints (1, 2, 3, 4, 6, 8, 24 and 48 h), regardless of the presence of headache. Additional post-hoc analyses were performed on cognitive impairment symptoms, difficulty thinking or difficulty concentrating, or both. After identifying a qualifying prodrome event in which the participant was confident that a headache would follow within 1–6 h, the participant entered their premonitory (prodrome) symptoms in the eDiary and reported the presence or absence of those prodromal symptoms (up to six patient-identified symptoms were preprogrammed into the eDiary for ease) at the prespecified timepoints. Entries could not be altered retrospectively. After treating the qualifying prodrome event with study medication and completing their initial eDiary entry, participants were required to report the absence or presence of a headache and prodromal symptoms at each timepoint post-dose. Safety assessments included the incidence of adverse events, clinical laboratory tests, electrocardiograms, vital signs, physical examination and the Columbia Suicide Severity Rating Scale. Adverse events were collected from the time of informed consent until 30 d after the last dose. An adverse event was considered to be a TEAE if the adverse event began or worsened (increased in severity or became serious) on or after the date of the first dose of study drug. The safety population included all treated participants who took at least one administration of study drug.

Statistical analysis

All efficacy analyses used the mITT population, defined as all randomized participants with one or more assessments of headache occurrence within 24 h after taking a double-blind study drug for one or more qualifying prodromal events during the double-blind treatment period. The mITT population is defined in the protocol and is equivalent to a full analysis set consistent with the International Council of Harmonisation (ICH) guidelines50. The safety population included all treated participants who took one or more administrations of study drug. Determination of sample size was previously reported; 480 participants in the mITT population were estimated to provide 95% power to determine a 16-point treatment difference in response rate for the primary endpoint, using a two-sided 5% significance level20.

For each of the five most common prodromal symptoms, as well as difficulty concentrating and difficulty thinking, the absence of symptoms of any intensity at each timepoint was analyzed using a generalized linear mixed model (GLMM) in the observed binary response variable. ORs (95% CIs) were based on the GLMM with treatment group, treatment period and pre-dose baseline prodromal symptom intensity as categorical fixed effects. An unstructured covariance matrix was selected for the covariance matrix of the residual effects for the repeated measurements, corresponding to the two qualifying prodrome events, within a participant. Covariance structure of compound symmetry was used when the model did not converge. Similar analyses were performed by censoring data collected after rescue medication. All symptom efficacy analyses were done using SAS software, v.9·4 or newer (SAS Institute, Inc.).

Standard protocol approvals, registration and patient consents

The Independent Ethics Committee or Institutional Review Board (IRB; that is, Advarra or University of Utah IRB) at each study site approved the study protocol, informed consent forms and recruitment materials before patient enrollment. The studies were conducted in accordance with the ICH guidelines, applicable regulations and the Declaration of Helsinki. All patients provided written informed consent before screening. The study is registered with ClinicalTrials.gov (NCT04492020). The study protocol and statistical analysis plan have been published1.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.