Improvement of migraine symptoms with a proprietary supplement containing riboflavin, magnesium and Q10: a randomized, placebo-controlled, double-blind, multicenter trial

Participants and recruitment

Otherwise healthy adults aged 18 to 65 years of either sex were recruited by neurologists practicing in Germany. All participants had migraine with and without aura diagnosed according to the IHS-criteria ICHD-II 1.1 und 1.2 [14]. The age at onset of migraine was less than 50 years of age and diagnosis of migraine was at least one year before study entry. The participants were required to have at least 3 migraine attacks per month in the last 3 months before recruitment and not more than 10 headache days. Patients were excluded if they used migraine prevention (drugs, nutritional supplements or psychotherapy) as well as antipsychotic or antidepressant medication during the last 3 months prior to study entry and throughout the study. Patients with medication overuse were excluded. Patients who had failed to respond to more than 2 different prophylactic agents in the past and patients resistant to all acute migraine drugs were not included.

The study was approved by the Institutional Review Board for each center. All participants gave written informed consent. The study was conducted according to the ethics principles of the Declaration of Helsinki. The trial was registered in the German Clinical Trial Register DRKS00004565.

Interventions

The investigational nutritional product (INP) is a dietary food for special medical purposes (according to EU regulations) containing 400 mg riboflavin (vitamin B2), 600 mg magnesium, 150 mg coenzyme Q10 along with a multivitamin/trace elements combination per 4 capsules (Migravent®, Dolovent®). The amount of additional multivitamin/trace elements per 4 capsules is as follows: 750 μg vitamin A, 200 mg vitamin C, 134 mg vitamin E, 5 mg thiamin, 20 mg niacin, 5 mg vitamin B6, 6 μg vitamin B12, 400 μg folic acid, 5 μg vitamin D, 10 mg pantothenic acid, 165 μg biotin, 0.8 mg iron, 5 mg zinc, 2 mg manganese, 0.5 mg copper, 30 μg chromium, 60 μg molybdenum, 50 μg selenium, 5 mg bioflavonoides. Placebo capsules indistinguishable from verum were used as control. Patients were instructed to take two capsules orally in the morning and two capsules in the evening for 3 months. Treatments for other conditions which may have an effect on migraine prevention were not allowed. These were mainly beta-blockers (e.g. propranolol, bisoprolol, metoprolol), calcium-antagonists (e.g. flunarizine), antiepileptics (e.g. topiramate, valproate), antidepressants (e.g. amitryptiline), supplements containing petasites (butterbur) or tanacetum (feverfew), magnesium, riboflavin or coenzyme Q10 in doses above 50 mg. Non-medical/non-nutritional treatment for migraine prevention like acupuncture or psychotherapy were also not permitted. However, participants were allowed to treat migraine attacks with their usual rescue pain medication and anti-emetics.

Study design

The study was conducted as a randomized, placebo-controlled, parallel-arm, double-blind, prospective multi-center study. After screening, patients underwent a one-month baseline period without treatment to verify that they had more than 3 migraine attacks but not more than 10 days with migraine or non-migraine headaches. The baseline phase also served as a baseline for the evaluation of efficacy parameters. Before entering the baseline phase patients had to meet inclusion and exclusion criteria (see Participants and recruitment). Demographic data, concomitant medication, medical history, migraine diagnosis as well as previous migraine preventive measures were documented by the investigator. Following the baseline phase and provided no inclusion/exclusion criteria were violated, eligible patients were randomized in double-blind fashion to verum or to placebo (1:1). In this follow-up visit patients were also asked to fill in an HIT-6 questionnaire. Randomization was done by computer and randomization lists were prepared. Randomization was done by blocks of four per center. The investigator sequentially allocated the random numbers to patients, starting from the lowest number. A blockwise randomization was used. The sequential order was verified by fax sent to a blinded person at the sponsor and from entries in the screening logs. Both investigator and patient were blinded to the treatments. Treatment with either verum or placebo was for 3 months. Migraine parameters and intake of the investigational products were recorded daily by the patients throughout the baseline and the treatment phase in an electronic diary accessed online via the internet. Compliance (documentation, intake of investigational products) was monitored regularly by the investigator and delegates of the sponsor. Patients were immediately contacted by the investigator if regular documentation was missing for more than a week. At the follow-up visit at the end of the treatment the patients again had to fill in an HIT-6 questionnaire and they were asked to evaluate the tolerability and efficacy of the treatment from their view. Concomitant medication and occurrence of adverse events were checked at each follow-up visit. Compliance was assessed by a pill count of the returned investigational product.

Efficacy parameters

The primary efficacy endpoint was defined as days with migraine as recorded in the online diary by the patient. Secondary endpoints were maximal pain of migraine headaches as recorded in the online diary, migraine burden as assessed through the HIT-6 questionnaire [15] and subjective patient evaluation of efficacy.

Days with migraine and migraine pain intensity were compared between the one-month baseline period and the last month of the 3-month treatment. A migraine day was defined as a day with at least 4 hours of migraine pain or a day with migraine pain and concomitant intake of pain medication. For each migraine day, pain intensity was rated as mild, moderate or severe by the patient. The HIT-6 questionnaire (headache impact test), which measures the impact of headache on a patient´s life, was filled out by the patient at the start of the treatment (randomization) and at the end of the 3-month treatment. This validated questionnaire consists of 6 questions, each question or item has the following response options: never (6 points), rare (8 points), sometimes (10 points), very often (11 points) and always (13 points) [14]. Headache impact on this scale ranges from 36 (no headache) to 78 (very severe headache). All checked points are added for the analysis. Efficacy of the treatment as a subjective evaluation by each patient was recorded at the end of the study as very good, good, moderate or poor.

Statistical methods

The sample size estimation was performed with the Test of the Ratio of Two Poisson Means module of PASS 11 software (NCSS, LLC. Kaysville, Utah, USA). A reduction by at least 50% in the number of migraine days should be achieved in the verum group because this was considered a clinically relevant improvement and a 30% reduction in the placebo group was assumed as the worst case scenario since placebo effects of around 30% have been seen in trials. Based on these assumptions, the estimated sample size without compensation for drop-out is 39.2 ≈ 40 patients per treatment group, i.e. 80 patients in total. The sample size was increased to 104 patients in total to compensate for overdispersion and a 15% drop-out rate.

The statistical analysis was based on the ICH Topic E9 Note for guidance on statistical principles for clinical trials (CPMP/ICH/363/96). A detailed description of the statistical evaluation was provided in a Statistical Analysis Plan (SAP).

The primary endpoint migraine days was compared statistically in a confirmatory test approach on superiority of Migravent® compared to placebo. The respective statistical test was performed using the generalized linear model in the following form: it was assumed that the number of migraine days during the last month of the 3-month treatment period can be described by Poisson distributions. The migraine day rate depends on treatment and disease severity at baseline, which is defined as the number of migraine days during the one-month baseline period. This was modeled through a Poisson regression with covariate “number of migraine days during the one-month baseline period” and with the factors “treatment” and “center”. The null hypothesis to be tested was whether the rate ratio for treatment (ρ = λ_Placebo/λ_Verum) is smaller or equal to 1 (i.e. log rate ratio is smaller or equal to 0). The type-I error rate was set to α = 0.025 (one-sided). Confounding factors were not controlled for.

The maximal intensity of migraine pain per migraine day during the last month of the 3-month treatment was compared between treatment groups by an ANCOVA with the maximal intensity of migraine pain during the baseline period as baseline covariate.

The burden of disease measured by HIT-6 sum scores for the end of treatment was compared between treatment groups by an ANCOVA with the burden of disease sum scores for the start of treatment as baseline covariate.

Subjective evaluation of efficacy by the patient at the end of therapy was displayed by default descriptive statistics for continuous data as well as for categorical data. The treatment groups were compared by van-Elteren’s test.