Design of the study

This prospective, open-label, multicenter observational study was conducted across nine referral centers in South Korea, targeting migraine patients who presented with headache or dizziness as the primary symptom from January 2020 to May 2024. The patients were assigned to either the nimodipine, topiramate, or combination group at the discretion of the treating physician. The combination group was included to reflect routine prescribing patterns. Analyses were conducted using a per-protocol approach, including only patients who completed the 3-month follow-up assessments and continued their medications throughout the study period. Patients who discontinued the study before the 3-month visit were excluded from the analyses.

This study followed the tenets of the Declaration of Helsinki and was performed according to the guidelines of Institutional Review Board of Pusan National University Hospital (1911-027-085). All patients were fully informed about the study, and written informed consent was obtained from each participant. The trial was registered at cris.nih.go.kr (KCT0010555) and the first registration date was 06/01/20 (https://cris.nih.go.kr/cris/search/detailSearch.do?seq=30149&search_page=L). The study protocol and CONSORT checklist are included as Supporting Information (S1 and S2 Files).

Participants

Patients aged 18–75 years with a diagnosis of migraine, with or without aura, based on the International Classification of Headache Disorders, 3rd edition (ICHD-3), were eligible for inclusion if they had not used preventive migraine medications in the preceding year. Subgroup analysis was also performed for participants meeting the ICHD-3 diagnostic criteria for vestibular migraine. The patients diagnosed as vestibular migraine were evaluated and enrolled by board-certified neurologists with subspecialty training in neuro-otology. To ensure diagnostic accuracy for vestibular migraine, alternative neuro-otologic conditions including Ménière disease, vestibular neuritis, or benign paroxysmal positional vertigo were systematically excluded based on detailed neurological examination, bedside vestibular examination, and/or vestibular function tests when they indicated. Exclusion criteria included; 1) headache or dizziness attributable to another condition, such as central nervous system disorders; 2) presence of significant comorbidities, including liver or kidney diseases, malignancy, or other medical conditions that can hinder participation; 3) requirement for daily analgesic use for severe pain; 4) contraindications to CCBs or topiramate, including pregnancy; 5) use of other preventive migraine medications during the study period; 6) any other condition deemed by the investigators to impede participation or lack of patient consent.

Intermittent uses of analgesics (NSAIDs or acetaminophen) were permitted for acute severe headache. In patients with vestibular migraine, dimenhydrinate and metoclopramide were used only in cases of severe dizziness accompanied by vomiting. Other vestibular suppressants, including betahistine, meclizine, and benzodiazepines, were not used during the study period. Analgesics and dimenhydrinate were not used prophylactically and were administered only intermittently when symptoms were severe.

Intervention

Nimodipine was administered 60 mg per day while topiramate was titrated from 25 mg to 100 mg per day according to individual responses. At enrollment, patients received a printed headache diary to document daily medication use, and headache occurrence and intensity. Patients with vestibular migraine also recorded dizziness frequency and intensity. Diaries were reviewed at each monthly visit when patients completed questionnaires and received prescriptions for the next month. The study terminated 3 months later. Patients were classified as dropouts if they withdrew consent, discontinued medication due to side effects, were considered noncompliant by the investigator, or missed more than one-third of prescribed medications over the month.

Outcomes

The primary endpoint was the change in headache days per week from baseline to 3 months after treatment initiation. The secondary outcomes included headache intensity measured on a 0–10 Wong-Baker pain rating scale (PRS), migraine-related clinical burden assessed using the Headache Impact Test-6 (HIT-6), and disability evaluated with the Migraine Disability Assessment (MIDAS). The PRS and HIT-6 were assessed at the baseline, two and three months later while MIDAS was evaluated at the baseline and three months later. For patients with vestibular migraine, additional outcomes included changes in dizziness days per week and dizziness intensity (0–10 visual analogue scale, VAS). The Dizziness Handicap Inventory (DHI) and UCLA Dizziness Questionnaire (UCLA-DQ) were also assessed at same time.

Statistical analyses

The sample size calculation was primarily conducted for the comparison between the two monotherapy groups. Assuming a power of 80% and a two-sided significance level of 0.05, a mean difference of 0.5 headache days per week, and an estimated standard deviation of 2 days, a minimum of 140 patients per group was required. The assumed effect size was chosen to clinically meaningful difference as the previous studies corresponding to approximately two headache-free days per month [14]. These assumptions were based on pragmatic considerations given the limited availability of prior comparative data in vestibular migraine. Because treatment allocation was investigator driven and non-randomized, with anticipated imbalance in group sizes, the study aimed to enroll a larger overall cohort to allow stable estimation of treatment effects across clinically relevant subgroups. The combination therapy group was included for exploratory purposes and was not incorporated into priori power calculation. Accordingly, the final enrollment target was determined pragmatically, taking into account expected dropout rates of 40–50% observed in routine clinical practice.

Statistical inference was prespecified for the primary endpoint only. Secondary and exploratory outcomes were interpreted descriptively. Where post-hoc pairwise comparisons were performed, p-values were adjusted using the Bonferroni method for reference.

For continuous variables, mean, SD, and confidence intervals (CI) were presented, and statistical analyses were performed using the t-test and Wilcoxon rank-sum test. For categorical variables, the chi-squared test or Fisher’s exact test was used. To determine interaction effects in repeated measures data, linear mixed-effects models (LMMs) were applied using restricted maximum likelihood estimation. The models included age, visit, group, and a group-by-visit interaction as fixed effects, and a subject-specific random intercept to account for repeated measures. Post-hoc comparisons were conducted to evaluate between-group differences at each visit and within-group changes over time. Bonferroni-adjusted significance thresholds were applied for inferential interpretation. Missing data were imputed using the Last Observation Carried Forward (LOCF) method to provide conservative estimates in this non-randomized setting.

Statistical analyses were performed using R version 4.3.3 (R Core Team, 2024; https://cran.r-project.org), with linear mixed-effects models fitted using the lme4 and lmerTest packages. A two-sided p-value < 0.05 was considered statistically significant.

Demographics and clinical features

A total of 850 patients (687 (81%) women, mean age ± SD = 41 ± 12) had been recruited with the first visit in February 2020 and the last in May 2024 (Fig 1). Of these, 465 patients (55%) completed the study. Among the 385 patients (45%) lost to follow-up, 301 (35%) withdrew their consent voluntarily, 70 (8%) were dropped out due to poor medication adherence (missing more than one-third of the prescribed medications during the preceding month), and 14 (2%) discontinued the study due to adverse events.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study flow diagram of the migraine cohort.

https://doi.org/10.1371/journal.pone.0344948.g001

Adverse events leading to study discontinuation were reported in 14 patients (14/850, 2%), and included paresthesia in 9 (64%), gastrointestinal disturbances in 3 (21%), rash in one (7%), and other side effect in the remaining one (7%). The nimodipine group showed a lower tendency of side effects (2 in the nimodipine group, 8 in the topiramate group, 4 in the combination group; p = 0.056).

No significant differences were observed between the patients included and excluded regarding the age, sex, and proportion of vestibular migraine (S1 Table). In addition, the participants were evenly allocated to the three treatment groups (p = 0.085). The baseline scores of PRS, MIDAS, and HIT-6 did not differ between the patients included and excluded, but the frequency of headache was higher in the patients excluded (p = 0.013) (S1 Table).

Analyses for migraine (total cohort)

Of the 465 patients who completed the follow-ups, 205 (44%) belonged to the nimodipine group, 160 (34%) to the topiramate group, and 100 (24%) to the combination group. Patients in the combination group were older (p < 0.001) and had more headache days (p = 0.008, Table 1). The scores of PRS, MIDAS, and HIT-6 were higher in the topiramate group (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Clinical characteristics and baseline scores in patients with migraine and vestibular migraine.

https://doi.org/10.1371/journal.pone.0344948.t001

The primary endpoint, the headache days decreased by approximately 2 days per week in all groups with a mean decrease at 1.9 in the nimodipine group, 1.8 in the topiramate group and 1.9 in the combination group (p-values < 0.001 for all groups, S1 Fig A, Table 2). Significant reductions of secondary outcomes were also observed in PRS (2.4–2.6 points), MIDAS (23–40 points), and HIT-6 (9–12 points) scores in all three groups (S1 Fig B–D and Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Primary and secondary outcomes after three months.

https://doi.org/10.1371/journal.pone.0344948.t002

Primary inter-group analyses focused on comparisons between the two monotherapy groups. No significant differences were observed between nimodipine and topiramate groups in the mean reduction of headache days (mean difference: −0.1, p = 0.691) or PRS scores (mean difference: 0.2, p = 0.543) (Table 3). Similary, no significant group differences were detected in these outcomes when all groups were compared (p = 0.945 and 0.368, respectively, Fig 2A and 1B). However, the decrease in MIDAS and HIT-6 scores differed among three groups (p = 0.004 and 0.040, respectively, Fig 2C and 1D, S2 Table). Post-hoc analyses (Table 3) showed greater reductions in MIDAS (16 points) and HIT-6 (3 points) scores in the topiramate group compared to the nimodipine group (p = 0.005 and 0.012, Table 3). The topiramate group also showed a greater HIT-6 reduction than the combination group (3 points, p = 0.038, Table 3). Temporal changes from baseline to the final visit were significant, while age did not significantly influence these changes (S2 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Mean differences between the groups between-group pairwise comparisons.

https://doi.org/10.1371/journal.pone.0344948.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Changes (Δ) of the scores among the nimodipine, topiramate, and combination groups analyzed using a linear mixed model in the patients with migraine.

The headache days (Headache freq, A) and Wong-Baker pain rating scale (PRS, B) decreased over three months in all groups (p < 0.001 for both) without differences among the groups (p = 0.945 and 0.368, respectively). The Migraine Disability Assessment (MIDAS, C) and Headache Impact Test-6 (HIT-6, D) scores showed reductions over three months in all groups (p < 0.001 for both), but significant differences were observed between-group differences (p = 0.004 and 0.040, respectively, S2 Table). (Dots (Dots represent model-estimated means; bars indicate 95% confidence intervals. Red = nimodipine, Blue = topiramate, Green = combination group).

https://doi.org/10.1371/journal.pone.0344948.g002

Analyses for vestibular migraine

Of the 131 patients with vestibular migraine (131/456, 28%) who completed the follow-ups, 61 (47%) belonged to the nimodipine group, 38 (29%) to the topiramate group, and 32 (24%) to the combination group (Table 2). Baseline comparisons showed younger patients in the topiramate group (p = 0.003) and lower PRS scores in the nimodipine group (p = 0.046, Table 2).

Headache-related outcomes improved significantly in all three groups, with reductions in the headache days per week (1.2–2 days/week), and PRS (1–3 points) and HIT-6 scores (9–12 points) (S1 Fig E, F, and H, Table 2). The MIDAS scores also improved significantly in the nimodipine and topiramate groups (p < 0.001 for both), but not in the combination group (p = 0.128, S1 Fig G, Table 2). The dizziness-related scores, including the dizziness days per week, VAS, DHI, and UCLA-DQ, showed significant reductions in all three groups (p < 0.001, Table 2 and S1 Fig I–L). Consistent with the primary analytic strategy, inter-group comparisons focused on the two monotherapy groups. Inter-group comparisons showed no significant differences in reductions of headache days, PRS, or MIDAS scores among the groups (Fig 3A-3C). However, the topiramate group showed a greater HIT-6 (5 points, p = 0.013) and PRS (1 points, p = 0.021) reductions than the nimodipine (Table 3). For dizziness-related outcomes, no significant differences were observed among the groups (Fig 3E–3H) even though the post-hoc pairwise comparisons indicated a greater VAS reduction (1 points) in the topiramate group than in the combination group (p = 0.039, Table 3). Temporal changes from baseline to the final visit were significant, while age did not significantly influence these changes (S3 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Changes (Δ) of the scores among the nimodipine, topiramate, and combination groups analyzed using a linear mixed model in the patients with vestibular migraine.

For headache-related measures (A-D), significant reductions were observed for the headache days (A), PRS (B), MIDAS (C), and HIT-6 (D) scores (p < 0.001 for all). Between-group comparisons showed no significant differences for headache days, PRS, and MIDAS (p = 0.544, 0.052, and 0.179, respectively, S3 Table), while HIT-6 scores significantly differed between groups (p = 0.041, S3 Table). For dizziness-related measures (E – H), including dizziness frequency (E), visual analogue scale for dizziness intensity (VAS, F), Dizziness Handicap Inventory (DHI, G), and UCLA Dizziness Questionnaire (UCLA-DQ, H) scores, all groups demonstrated significant reductions over the 3-month period (p < 0.001 for all, S3 Table), with no significant differences among groups (S3 Table). (Dots represent model-estimated means; bars indicate 95% confidence intervals. Red = nimodipine, Blue = topiramate, Green = combination group).

https://doi.org/10.1371/journal.pone.0344948.g003

Imputation analysis

An additional analysis using LOCF method was performed to address missing data from patients lost to follow-up (n = 385, 45%). Overall, results were largely consistent with the primary complete-case linear mixed model analysis for most outcomes based on percent changes from baseline

Among the 850 patients with migraine, percent changes in headache frequency, PRS, and HIT-6 scores decreased significantly across all treatment groups (all p < 0.001; S2 Fig A, B, and D). In contrast, percent changes in MIDAS scores did not show significant improvement following imputation, and no significant between-group differences were detected (p > 0.05; S2 Fig C).

Similarly, among the 255 patients with vestibular migraine, the percent changes in headache frequency, PRS, and HIT-6 scores significantly decreased in all groups (all p < 0.001; S3 Fig A, B, and D). In contrast, the percent change in MIDAS scores showed no improvement from the baseline (p = 0.931; S3 Fig C). No significant differences were observed among the treatment groups for any of the measured outcomes (S3 Fig 3).