High-flow oxygen therapy for treatment of acute migraine: A randomized crossover trial

Abstract

Background

Impaired oxygen utilization and cerebrovascular dysfunction are implicated in migraine. High-flow oxygen is effective in cluster headache and has shown promise in animal models of migraine, but has not been adequately studied in patients with migraine.

Methods

In this randomized, crossover-design, placebo-controlled trial, adult migraineurs self-administered high-flow oxygen or medical air at 10–15 l/min via face mask in blinded fashion starting soon after symptom onset for 30 minutes, for a total of four migraine attacks. Participants recorded the severity of headache, nausea, and visual symptoms on visual analog scales periodically up to 60 minutes.

Results

We enrolled 22 individuals (mean age 36 years, 20 women) who self-treated 64 migraine attacks (33 oxygen, 31 air). The pre-specified primary endpoint (mean decrease in pain score from baseline to 30 minutes) was 1.38 ± 1.42 in oxygen-treated and 1.22 ± 1.61 in air-treated attacks (p = 0.674). Oxygen therapy resulted in relief (severity score 0–1) of pain (24% versus 6%, p = 0.05), nausea (42% versus 23%, p = 0.08) and visual symptoms (36% versus 7%, p = 0.004) at 60 minutes. Exploratory analysis showed that in moderately severe attacks (baseline pain score <6), pain relief was achieved in six of 13 (46%) oxygen versus one of 15 (7%) air (p = 0.02). Gas therapy was used per protocol in 91% of attacks. There were no significant adverse events.

Conclusion

High-flow oxygen may be a feasible and safe strategy to treat acute migraine. Further studies are required to determine if this relatively inexpensive, widely available treatment can be used as an adjunct or alternative migraine therapy.

Introduction

Migraine is a common, disabling condition. Current antimigraine agents (e.g. triptans) are effective; however, their use may be limited by factors such as cost and cardiovascular risks (1). Additional low-cost, safe and effective treatment choices are warranted to optimize migraine management. Oxygen therapy has shown efficacy in the treatment of cluster headache (2), but has not been adequately studied in migraine (3–9). A trial conducted in 1940 was the first to show efficacy of oxygen inhalation via a mask at flow rates of 6 l/min (4). A recent trial suggested efficacy of high-flow inhaled oxygen in patients presenting to the emergency department (ED) with a variety of headache disorders, including migraine (8). Oxygen inhalation did not prove effective in an exploratory trial of glyceryl trinitrate-induced headache (7), which may have a different mechanism from spontaneous migraine, and normobaric oxygen therapy failed to show efficacy against hyperbaric oxygen therapy in two trials (5,6). Oxygen therapy is inexpensive, widely available, has an excellent safety profile, and can be self-administered at home. If truly effective, oxygen therapy may significantly impact the current-day management of migraine around the world (9).

There are many scientific reasons why oxygen might be effective in acute migraine. Abnormal oxygen utilization, tissue hypoxia, and cerebrovascular dysfunction are implicated in the pathogenesis of migraine (10–15). Experimental hypoxia induces migrainous headaches in migraineurs with and without aura, and dilates cerebral arteries in patients with migraine with aura (16–19) whereas hyperoxia has vasoconstrictive effects (3). Although oxygen therapy may not directly suppress cortical spreading depression triggered by strong depolarizing stimuli (20), it may suppress spreading depression triggered by microembolism (21), which is believed to be a migraine trigger (22). Indeed, oxygen therapy inhibits peri-infarct spreading depolarizations (23), reduces inflammation and blood-brain barrier damage in animal models of migraine (24), and may have yet other mechanisms similar to those documented in models of ischemic stroke (15,25).

We designed a pilot study to obtain preliminary data concerning the feasibility, efficacy and safety of high-flow oxygen therapy in acute migraine.

Methods

This clinical trial was approved by the Massachusetts General Hospital (Partners) Human Research Committee and registered at clinicaltrials.gov (NCT01542307).

Participants

Individuals with a diagnosis of frequent (≥1 per month) migraine were screened from the neurology outpatient clinic and neurology consultations performed in the ED from 2012 to 2015, and enrolled after informed consent if they met the following inclusion criteria: age ≥18 years, with frequent attacks of migraine with or without aura according to International Headache Society (IHS) criteria (26). Consistent with inclusion criteria suggested in IHS guidelines (27), all individuals had one to six attacks of episodic migraine per month starting before age 50 years and present for more than one year. Exclusion criteria were nonmigraine primary or secondary headache disorder, chronic obstructive pulmonary disease, oxygen-dependent at baseline, pregnant, active smoker or living with a smoker, inability to self-administer gas therapy in the judgment of the investigator, and (since cylinders were delivered to participants’ homes) residence >50 miles from Massachusetts General Hospital, Boston.

Safety assessment and enrollment procedures

First, an in-hospital safety test was conducted to assess the ability for self-treatment and ensure physiological stability during oxygen inhalation. An increase or decrease in systolic blood pressure >10%, or >10% change in respiratory rate during oxygen inhalation, were considered unsafe. Individuals passing the safety test were considered “fully enrolled” (all participants passed this test). Baseline Migraine Disability Assessment Scale (MIDAS) scores (28), Migraine Assessment of Current Therapy (Migraine-ACT) scores (29), home medications, past medical history, and social history were recorded.

Gas therapy and outcome assessment

Next, a study staff member delivered the following items to participants’ homes: two oxygen and two medical air (containing 21% oxygen) E-cylinders; non-rebreather face masks with gas tubing and connectors; home safety instructions for cylinder storage and the use of oxygen; and four sets (one for each migraine attack) of visual analog scale (VAS) forms, safety questionnaires and stamped return envelopes. Each VAS form included three 0–10 scales—one each for head pain, nausea, and visual symptoms (photophobia, blurred vision, scotomas, etc.). Participants were instructed to complete the VAS forms at baseline (before starting treatment), 15 min, 30 min (end of treatment), and 60 min (post-treatment). Individuals were instructed to mail back the VAS forms and safety questionnaire after each treated attack.

Blinding and randomization

Praxair Inc donated specially designed oxygen and medical air cylinders that looked identical (gray color), except for content labels intended for use during transport as per the United States Department of Transportation regulatory requirements. Once these were transported to participants’ homes, study staff concealed the labels on all cylinders with tamper-resistant tape in order to preserve the blind. At the end of the study, study staff confirmed the position and integrity of the tamper-resistant tape, then removed the tape in order to reveal the original label, and transported the E-cylinders back to a locked research storage area at our hospital.

Randomization was achieved by randomly allocating the labels A–D to E-cylinders (two oxygen, two medical air, in random order). Participants were instructed to use the first cylinder (labeled “A”) at 10–15 l/min for 30 min via a non-rebreather mask as soon as possible at the onset of a migraine attack, and the second, third, and fourth (“B,” “C,” and “D”) cylinders at the same rate and duration for subsequent attacks. The time of migraine onset was determined by the patients based on their usual onset symptoms: head pain, or premonitory symptoms, or the timepoint when they typically consumed antimigraine medications.

Additional study procedures

Participants were requested to delay their usual migraine medications until after gas therapy. The initial delay of ∼30 min for pharmacologic treatment was important to minimize confounding; however, patients were advised to take their antimigraine medications and even interrupt study treatment at any time during the course of the attack if warranted. Individuals had the option of calling the investigators before starting gas therapy to reconfirm study procedures. The study coordinator contacted participants on a monthly basis to inquire about migraine status, general health status, change in medications, use of E-cylinders, problems encountered, and any adverse effects associated with gas therapy. Instructions for the proper use of gas and storage of gas cylinders, and the completion and return of study forms, were reinforced during each call. Since we enrolled people with frequent migraines (at least one attack per month), we expected study participation to last four to six months. Individuals were prematurely terminated from the trial and E-cylinders retrieved if they failed to use at least one cylinder within six months, or if they became pregnant, moved >50 miles away, or withdrew consent.

Study endpoints

The pre-specified primary endpoint analysis was a comparison of the mean change in pain scores from baseline to 30 min between oxygen- and air-treated attacks. Although symptom intensity changes from moderate-severe to mild-moderate may not be equivalent, the VAS was considered a continuous or linear scale for the purposes of this analysis (27). Secondary endpoints were the comparison of mean change in pain, nausea, and visual symptom scores from baseline–15 and baseline–60 min, the percentage of attacks achieving symptom relief (score 0–1), and the number of attacks requiring supplemental medication treatment at 30 min and 60 min in each group. Adverse outcomes and the frequency of new symptoms developing after treatment initiation were monitored routinely by the study physicians by reviewing returned forms and questionnaires and after patient study completion. Feasibility was assessed by inspecting E-cylinder gauges to confirm that gas utilization was as expected.

We planned to enroll 40 patients to obtain sample size estimates for a future trial. Due to funding limitations we terminated the trial in September 2015. All accrued data were analyzed and the full results are presented herein.

Statistical analysis

Mann-Whitney U test, Student t test, and Fisher exact tests were used as appropriate. Repeated-measures analysis of variance (ANOVA) and Wilcoxon were used to analyze temporal changes in VAS scores. Data are presented as percentages, mean (±SD) or odds ratios (ORs) with 95% confidence intervals (CIs), as appropriate. A value of p < 0.05 was considered significant. SPSS version 21 was used for analyses. All analyses were based on the intention-to-treat principle.

Results

Participant characteristics are shown in Table 1. We enrolled 22 patients who generated data for 64 migraine attacks (33 self-treated with high-flow oxygen, 31 with medical air). Patients had relatively severe, disabling, and inadequately controlled migraine as evidenced by their frequent ED visits, high MIDAS scores (81% had Grade III–IV), high Migraine-ACT scores, and use of multiple prophylactic and abortive medications (Table 1).

Table 1.

Participant characteristics.

Variable	n = 22
Age	36 ± 10
Female	91%
Migraine with aura	32%
Caucasian	91%
Non-Hispanic ethnicity	73%
Employed full time	77%
Education (college or higher)	91%
Marital status (married)	45%
Ex-smoker	9%
Hours of exercise per week	3.8 ± 4
Family history of migraine	59%
Vascular risk factors (e.g. hypertension)	9%
Proportion with ≥1 lifetime ED visit	77%
Proportion with ≥1 ED visit in prior year	50%
Number of lifetime ED visits	7.8 ± 14
Number of ED visits in prior year	1.1 ± 2
Active acute migraine medications
Triptans	82%
NSAIDs or acetaminophen	100%
Opioids	27%
Active prophylactic migraine medications
NSAIDs or acetaminophen	23%
Anti-epileptics	46%
Calcium channel blockers	23%
Beta-blockers	36%
Botulinum toxin	18%
MIDAS score (mean ± SD, median)	42 ± 40, 28
MIDAS groups
1 (Little or no disability)	14%
2 (Mild disability)	5%
3 (Moderate disability)	18%
4 (Severe disability)	63%
Migraine-ACT score
0 (no need to make a change)	23%
1	23%
2	27%
3	9%
4	18%

ED: emergency department; NSAIDs: nonsteroidal anti-inflammatory drugs; MIDAS: Migraine Disability Assessment Scale; Migraine-ACT: Migraine Assessment of Current Therapy; SD: standard deviation.

Compliance and feasibility

A total of 14 patients (64%) self-treated four attacks; one patient, three attacks; two patients, two attacks; three patients, one attack; and two patients did not treat any attack. All forms were duly completed and mailed back, except for one set from a patient who treated four attacks. The reasons for study noncompletion included premature termination by the investigator because of infrequent migraine attacks or the inability to access the cylinder in time (e.g. migraine occurred while traveling) (n = 5), moved out of state (n = 1), and interval pregnancy (n = 1). One participant withdrew consent.

Blinding procedures proved successful (i.e. no evidence that the tamper-resistant tape had been removed by the patient). No participant called to re-clarify procedures. Only two individuals (9%) inhaled less than the expected quantity of gas, for a total of six migraine attacks (four intended for oxygen and two intended for air treatment).

Outcomes

There were no significant differences between oxygen- and air-treated attacks in baseline pain scores (5.88 ± 1.89 versus 5.87 ± 1.82, p = 0.99), baseline nausea scores (3.29 ± 2.73 versus 3.84 ± 2.82, p = 0.43), or baseline visual symptom scores (2.71 ± 2.61 versus 3.03 ± 3.04, p = 0.65). No patient used medications during the 30-min gas therapy.

Pain

The prespecified primary endpoint, the mean decrease in pain score from baseline to 30 min, was 1.38 ± 1.42 (range: −2 to 4) in oxygen-treated and 1.22 ± 1.61 (range: −1 to 5) in air-treated attacks (p = 0.674). The mean change in pain scores at 15 and 60 min were also not significantly different (Table 2). Multiple prespecified secondary endpoints measured at 60 min (Table 2) suggest efficacy of oxygen therapy. Complete resolution of headache, defined as pain score 0–1, tended to be higher with oxygen compared to medical air (24% versus 6%, p = 0.05). The proportion of attacks with notable improvement in pain, defined as pain score decrease by ≥3 points, tended to be higher with oxygen (42% versus 23%, p = 0.08). Significant headache relief, defined as pain score 0–1 or an improvement by ≥4 points, was higher with oxygen treatment (45% versus 23%, p = 0.048). In exploratory analysis, attacks with mild-to-moderate pain, defined as baseline score <6, oxygen treatment afforded significant relief (score 0–1 achieved in six of 13 (46%) oxygen versus one of 15 (7%) air, p = 0.023; and score 0–1 or improvement ≥4 points achieved in seven of 13 (54%) oxygen versus two of 15 (13%) air, p = 0.029). Oxygen-treated attacks resulted in a numerically lower (but non-significant) requirement of rescue medications (36% versus 52%, p = 0.16).

Table 2.

Trial endpoints.

Endpoints	Oxygen (n = 33)	Medical air (n = 31)	p value	OR (95% CI)
Primary (0–30 minutes)
decrease in pain score (mean, SD)	1.38 ± 1.42	1.22 ± 1.61	0.67	–
Secondary (at 15 minutes)
decrease in pain score (mean, SD)	0.77 ± 1.09	0.44 ± 0.92	0.18
Secondary (at 60 minutes)
Pain
decrease in pain score (mean, SD)	2.05 ± 2.23	1.57 ± 1.94	0.38
Final severity score 0–1	24%	6%	0.05	4.64 (0.90–23.90)
Score improved ≥3points	42%	23%	0.08	2.56 (0.85–7.51)
Score 0–1, or improved ≥3 points	45%	23%	0.05	2.86 (0.97–8.46)
Post-gas therapy medication use	36%	52%	0.16	1.87 (0.69–5.07)
Visual symptoms a
Final severity score 0–1	36%	6%	0.004	8.29 (1.68–41.0)
Nauseaa
Final severity score 0–1	64%	65%	1.00	0.96 (0.35–2.67)

OR: odds ratio; CI: confidence interval.

a

Visual symptoms and nausea could not be assessed for ≥3-point improvement since very few attacks had baseline severity scores exceeding four points for these symptoms.

Nausea and visual symptoms

The mean decrease in VAS nausea and VAS visual symptom scores from baseline to 60 min were not significantly different between oxygen- and air-treated attacks (nausea, 1.58 ± 2.41 versus 1.92 ± 2.52, p = 0.579; visual, 1.36 ± 1.64 versus 1.10 ± 1.78, p = 0.967). Complete resolution of nausea, defined as nausea score 0–1, was not significantly different (64% versus 65%, p = 1.0); however, complete resolution of visual symptoms, defined as visual score 0–1, was significantly greater with oxygen as compared to medical air (36% versus 6%, p = 0.004).

Adverse events

There were no significant adverse events. One oxygen-treated attack was associated with dry mouth, and one with transient finger tingling. Two air-treated attacks were associated with significant worsening of usual migraine symptoms, requiring premature treatment cessation in one. Analysis of VAS scores showed that symptom worsening during gas inhalation was not significantly different between oxygen- and air-treated attacks (pain, 12% versus 14%; nausea, 12% versus 7%; visual symptoms, 6% versus 7%). No patient required rescue antimigraine medications during the 30-min gas therapy.

Discussion

The results of this randomized, blinded, crossover pilot clinical trial indicate that the prompt self-administration of high-flow oxygen inhalation is feasible, safe, and possibly beneficial as an acute migraine therapy. Despite the long time interval between enrollment and self-treatment, no patient called to clarify procedures during an attack and there were few instances of incomplete therapy. There were no significant adverse events, and no patient reported discomfort while inhaling the gas. Preservation of the study blind was satisfactory. While a clinically important difference between oxygen and air breathing for our primary outcome was not apparent, some secondary outcomes (Table 2) suggest that high-flow oxygen may afford relief of headache and visual symptoms.

In this exploratory proof-of-concept study we did not anticipate benefit with only 64 treated attacks. The trial design and outcome measure selection was generally consistent with published recommendations for the design of migraine clinical trials (30). However, several exceptions were considered acceptable. Unlike oral or injectable treatments, oxygen is an inhaled treatment strategy with a short half-life expected to have transient or adjunctive benefit. The primary endpoint (mean change in pain score at 30 min) was selected on the assumption that patients would respond promptly and maximally during gas treatment. However, significant benefit at 60 min was achieved on multiple outcome measures, suggesting that the effects may be more gradual, sustained, and substantial. This information would be relevant in the design of future trials. Patients were permitted to take rescue medications at any time, although they were requested to wait 60 min or longer if possible as per guideline recommendations. Secondary outcomes were assessed at one hour and not the recommended two hours given the short half-life of oxygen. Because of the home design, we could not capture the details of migraine; the endpoints were based on self-report. The dose and method of delivery (10–15 l/min via face mask) was pragmatic, designed to deliver the maximum concentration (100%) of oxygen, and the duration of therapy (30 min), though twice that of other trials (2,8), was chosen based on the capacity of the E-cylinders and the expectation that longer treatment duration would have greater effect.

Oxygen’s mechanisms warrant further investigation. We observed a more favorable effect on visual symptoms as compared to pain and nausea, suggesting greater effect on the occipital cortical regions than the deeper structures believed to underlie the pain and gastrointestinal accompaniments of migraine. Whether these visual effects result from inhibition of cortical spreading depression or other mechanisms is not known. Further studies are required to investigate oxygen’s effects on different neuronal subpopulations and understand whether it involves the activation of parasympathetic as opposed to trigeminovascular pathways, as observed in studies of cluster headache (31). Finally, our results may be explained by the unexpectedly high proportion of women who were enrolled, although in the case of cluster headache it is believed to have greater benefit in men (32). Gender-dependent effects should be considered in future trials.

In this small pilot trial the primary endpoint was negative. However, the positive results on multiple secondary endpoints, and the blinded crossover trial design that has been successfully used in similar trials (2), suggest the need for further trials. Our study has several strengths and limitations. Participants started therapy as soon as possible after migraine symptom onset, so an imbalance in onset-to-treatment time is unlikely. Baseline scores were balanced despite the relatively small number of attacks; however, there was a heterogeneous mix of migraine with and without aura, with individuals using different medications. The attacks were treated promptly and at home, so our results may not be generalizable to the more severe or treatment-resistant migraine attacks in patients presenting to the ED. We cannot determine if every attack after enrollment fulfilled criteria for migraine since patients did not maintain headache diaries. In this pilot trial we did not adjust for multiple comparisons or perform multivariable analysis. Given self-reported assessment and outcomes after each migraine attack, we were unable to distinguish the response to each visual symptom (photophobia, scotomas) or effects on subgroups (e.g. autonomic symptoms). We cannot confirm that all attacks received the same “dose” of treatment, and cannot comment on the effects of oxygen on the aura of migraine. We were careful with participant selection, so safety still needs to be confirmed in the broader migraine population.

Supplemental home oxygen is widely used and known to be safe. In a trial of 204 ED patients with primary headache disorders (47% tension, 27% migraine, 1% cluster, and 25% undifferentiated headaches), oxygen therapy resulted in a significant improvement in pain scores, and had a lower requirement for rescue analgesics (8). A Cochrane analysis of 11 clinical trials involving 209 participants with migraine or cluster headache found that hyperbaric oxygen therapy was effective in relieving pain, but its high cost and poor availability are major limitations (9). The encouraging results of our trial should prompt further trials to determine the efficacy of high-flow oxygen inhalation (a relatively low-cost and widely available treatment) in patients with acute migraine.

Clinical implications

Home-based oxygen therapy appears feasible, safe, and potentially effective for patients with episodic migraine with and without aura.

Additional studies are warranted to investigate hyperoxia’s therapeutic potential and mechanisms in migraine.

If confirmed, inhaled oxygen may become a low-cost and widely available alternative or adjunctive treatment for migraine.