Systolic blood pressure is positively correlated with headache frequency: A 3-month follow-up study on 441 migraine patients

Newly diagnosed migraine patients without anti-hypertensive or migraine preventive treatment were prospectively enrolled. Standardized BP measurement was done at each visit. Patients were dichotomized by the overall arithmetic mean of baseline systolic BPs (SBPs). Correlations between baseline SBPs and monthly headache days (MHDs), based on prospective headache diaries, were determined. Participants were treated as usual for 3 months. The associations between changes (Δ) in BP and in MHD at 3 months were examined. improvement odds ratio=1.70, 95% confidence interval=1.05-2.76, p=0.032). SBP was associated with headache frequency in migraine patients and the association was stronger in those with higher SBP. The improvement in headache frequency paralleled that in SBP, although the underlying mechanisms remain to be determined.


Introduction
The association between blood pressure (BP) and migraine has been an issue of debate in the literature and also in the daily clinical practice, especially when it comes to the association between increased risks for cerebrovascular and cardiovascular events and migraine (1)(2)(3)(4)(5). The association between high BP and headache was first reported in the early twentieth century (6), although subsequent reports were inconsistent (7)(8)(9)(10)(11)(12)(13)(14)(15). Some studies reported that migraine was associated with an increased risk not only for co-morbid hypertension (8,9), but also for developing hypertension during follow-up (10,15). On the other hand, it was also reported that patients with migraine had lower systolic and pulse pressures, but higher diastolic pressure when compared with those without (7), and migraine was more common among normotensive individuals than hypertensive patients (16). However, most of these reports examined the prevalence or incidence of migraine within groups with or without hypertension. Whether BPs per se could have an impact on headache frequency in patients with migraine, or vice versa, has rarely been investigated.
Whether the outcomes of BP and headache in migraine are correlated with each other is uncertain. Based on cross-sectional studies, hypertension was believed to be a risk factor for migraine chronification (17)(18)(19), although longitudinal data have been scarce. However, some insights could be derived from migraine preventive or anti-hypertensive treatment. Some anti-hypertensive agents are also effective in migraine prophylaxis (20)(21)(22)(23)(24). Besides, a meta-analysis of clinical trials on anti-hypertensive agents reported a lower incidence of headache reported as an adverse event in the treatment group than in the placebo group (25). These findings suggest that the outcomes of headache and BP could be correlated, although further clarification is necessary.
The objectives of the present study were to investigate the association between headache frequency and BP per se in migraine patients, as well as the association between the outcomes of headache and BP at 3 months.

Participants
This was a prospective study enrolling newly diagnosed migraine patients without prior treatment with migraine preventive or anti-hypertensive agents from the Headache Clinic of Taipei Veterans General Hospital. The diagnosis of migraine was made based on the criteria of the third edition of the International Classification of Headache Disorders (26). Patients were eligible at their first visits if they were (a) willing to participate in the study, (b) aged 20 or above, (c) diagnosed as migraine without aura or migraine with aura, and (d) occurrence of migraine before age 50.
Patients were excluded if they had (a) an acute headache with a history of <1 month, (b) a secondary headache disorder, (c) a history of use of migraine preventive medications or anti-hypertensive drugs within 3 months before enrollment, and (d) physical or mental disorders precluding the completion of detailed history taking.
Eligible patients were enrolled in the study and were asked to keep a headache diary, and a headache day was defined as a calendar day with a headache lasting ≥ 4 hours.
They received preventive treatment at the discretion of the treating physicians.
Patients were included in the final analyses if they had at least one follow-up clinic visit within 3 months. The study protocol was approved by the Institutional Review Board at the Taipei Veterans General Hospital (IRB-TPEVGH No. 2018-10-008A).
All participants provided informed consent prior to participation.

Blood pressure measurement
Blood pressures (BPs) were measured at each visit using a standard protocol. After having a rest in a quiet room for >5 minutes, two BP readings were obtained by a research assistant using a non-invasive stand-alone BP monitor (WatchBP Office Central, Microlife Corp., Taipei, Taiwan). The averaged values of the two BP readings, including systolic BPs (SBP), diastolic BPs (DBPs), and mean arterial pressures (MAPs), were used for analyses. The patients were categorized as having above-or below-average SBPs if their baseline SBPs were higher or lower, respectively, than the arithmetic mean value of the baseline SBPs of the entire study population. Identical procedures were used to measure BP at the follow-up visit at 3 months.

Statistical analysis
Baseline characteristics between the above-and below-average SBP groups were compared by using student's t test for continuous variables or Mann-Whitney U test if the variables were not normally distributed, and by using chi-square test for categorical variables. The changes in SBP (ΔSBP) and the change in the number of monthly headache day (MHD) (ΔMHD) at 3 months were determined, and missing data were imputed by using the last-observation-carried-forward (LOCF) method.
Paired t test was used to compare the differences between baseline and at 3 months.
As the null hypothesis of normality of our SBP and MHD data was rejected by the Shapiro-Wilk test, Spearman's rho statistics was used to examine the correlations between SBPs and the numbers of MHDs at baseline and between the changes in SBPs (ΔSBP) and the changes in MHDs (ΔMHDs) at 3 months. The patients in the above-and below-average SBP groups were further divided into higher and lower MHDs, as well as SBPs, by the medians within that group. Logistic regression modeling was carried out to compare the odds of having higher SBPs between patients with higher and lower MHDs at baseline, and to estimate the odds ratios (ORs), as well the 95% confidence intervals (CIs), for having 5-mmHg improvement in SBP (27) at 3 months between patients with and without ≥50% reduction in MHD.
Potential confounders were controlled, including age, sex, body mass index (BMI), monthly days of analgesic use at baseline (days/month), headache severity at the time of BP measurement (0-10/10 on numerical rating scale), analgesic use on the day of clinical visit, baseline BP status (<mean/≥mean), and preventive treatment during the follow-up period. All statistical analyses were carried out by using IBM SPSS Statistics for Windows, version 24.0 (IBM Corp., Armonk, NY, USA. Statistical significance was defined as a two-sided p of <0.05.

Correlations between blood pressures and headache frequency at baseline
Overall, headache frequency increased with baseline SBP (figure 1) in the entire study population. There was a positively correlation between MHD and SBP in patients with above-average baseline SBPs (Spearman's rho=0.181, p=0.012), but not in those with below-average baseline SBPs (Spearman's rho=0.029, p=0.646) ( figure   1A, table 2). On the other hand, there was no correlation between SBP and headache severity at the time of BP measurement at baseline (Spearman's rho=0.051, p=0.290).
The correlations were similar but weaker in DBPs and MAPs (table 2). Patients with higher MHDs were more likely to have higher SBPs after controlling for potential confounders, although the association was present only in patients with above-average SBPs (OR=2.31, 95% CI=1.17-4.56, p=0.017), but not in those with below-average SBPs (OR=1.21, 95% CI=0.66-2.21, p=0.534) ( Figure 1B).

Treatment and clinical outcomes at 3 months
The average interval from baseline to 3-month follow-up was 2.1±0.9 months.

Association between changes in systolic blood pressures and headache frequencies at 3 months
Overall, the extent of MHD improvement was greater in patients with higher baseline SBPs ( figure 2A). There was a statistically significant correlation between ΔMHDs and ΔSBP (Spearman's rho=0.114, p=0.016), which was derived from patients with above-average SBPs at baseline (Spearman's rho=0.257, p<0.001), but not from those with below-average SBPs (Spearman's rho=0.052, p=0.410) (figures In the entire study population, patients with a ≥ 50% decrease in MHDs were more likely to have SBP reductions by 5 mmHg (OR=1.63 [1.05-2.55], p=0.031) when compared with those without, after controlling for potential confounders (table   3). Besides, the findings were consistent even when baseline SBP status (<mean/≥mean) (model 2) or headache severity at the time of BP measurement (model 3) were further controlled (table 3). The findings were consistent for 3-mmHg and 7-mmHg reductions, and were more pronounced in those treated with preventive medications in exploratory analysis (supplementary table).

Discussion
In this large prospective clinic-based cohort (n=441), the relationship between SBPs and headache frequencies was demonstrated in newly diagnosed preventiveand anti-hypertensive-naïve migraine patients. At baseline, SBPs were correlated with headache frequencies (Spearman's rho=0.181, p=0.012), and patients with higher MHDs were more likely to have higher SBPs (OR=2.31, 95% CI=1.17-4.56, p=0.017). At 3-month follow-up, the decrease in headache frequency paralleled that in SBP. The findings were more pronounced in patients with above-average SBPs, i.e. ≥120.1 mmHg. More importantly, the findings remained after controlling for potential confounders, including headache severity at the time of BP measurement and presence or absence of analgesic use.
One of the strengths of the present study is the relatively large sample size. In the present study, SBP per se and headache frequency, rather than the presence or absence of hypertension or migraine, were collected and analyzed. We assessed the association between a physiological parameter and attack frequency, not merely the association between two diagnostic categories. Such in-depth analysis would not be feasible without such data abundance. Secondly, the data are of high quality. Migraine diagnoses were made by in-person clinical evaluations by headache specialists. BP was collected prospectively using a standard protocol. The number of MHDs were assessed with prospective daily diaries. In addition, the prospective and longitudinal nature of the present study reduced recall bias, which could be common for retrospective or cross-sectional designs.
In the baseline analysis of the present study, headache frequencies were correlated with SBPs in migraine patients, consistent with some prior reports, which provided evidence in favor of a positive association between migraine and hypertension (8-10, 13-15, 27-30). Some prior studies have reported an inverse association (7,11,12,31) or even the absence of association (32). The discrepancies could be, at least in part, attributed to the differences in study populations and methodologies. Most of the published studies were population-based studies, and investigated the prevalence or incidence of migraine and hypertension diagnoses rather than the disease severities. In contrast, the current study examined the association between headache frequency and SBP per se in migraine patients using both cross-sectional and longitudinal data, and patients with above-and belowaverage baseline SBPs were analyzed separately. Otherwise it would not be easy to pick up the correlation between SBP and headache frequency as the overall association in the entire study sample was not significant. On the other hand, even though theoretically acute medications could have a role, the proportions of patients using NSAIDs or triptans were comparable between patients with above-and belowaverage baseline SBPs (table 1). Besides, the association remained after controlling for potential confounders (supplementary figure). However, the associations were not strong. It could not be excluded that part of the associations was derived from patients with SBPs much higher than the rest of the study population. Nevertheless, patients with very high SBPs, i.e. ≥ 160 mmHg constituted only 3.4% (n=15) of the study population, and might not have a substantial contribution to the finding. Therefore, it is likely that the association was present in patients with above-average SBPs, and could be more pronounced in those with very high SBPs. Further studies are needed to provide evidence for or against such an association.
During follow-up, there was a trend toward greater improvement in headache frequency in those with higher baseline SBPs ( figure 2A). There was a positive correlation between ΔSBP and ΔMHD, especially in patients with above-average SBPs at baseline (figures 2B & 2C). Our findings seemed incongruent with those of the HUNT studies, which concluded with an inverse relationship between BPs and migraine (12,31). In the HUNT studies, elevated SBPs and pulse pressures were associated with decreased odds of having migraine at 11-years' follow-up. Differences in study design (clinic-based vs. population-based), outcome measure (change in headache frequency in migraine patients vs. change in migraine prevalence), and follow-up duration (3 months vs. 11 years) could account for these differences. More importantly, a substantial proportion (50.6%) of patients in the present study were treated with preventive medications. Therefore, the finding of the present study could not be compared with those of the HUNT studies directly. However, our findings could have a more practical impact for headache clinicians as the results were derived from a clinical setting.
There was accumulating evidence for increased risks of cardiovascular and cerebrovascular events in migraine patients when compared to healthy controls (1)(2)(3)(4)(5), and elevated BP might be a potential missing link between migraine and vascular events. In the present study, a ≥50% improvement in headache frequency was associated with an increased likelihood of SBP reduction at 3 months in the present study. On the other hand, even though migraine with aura was commonly mentioned as a risk factor for vascular events (4), the proportions of patients with aura were comparable between above-and below-average baseline SBP groups, which could suggest that BP might not have a role in the association between migraine aura and vascular risks. Even though SBP was selected as the primary BP parameter in the present study, the relative significance of different BP parameters as to the underlying pathophysiology remains unknown. Recently, it was reported that there was a positive correlation between central SBP and the volume of cerebral white matter hyperintensities (28), which could suggest that SBP could play a more important role.
On the other hand, some studies suggested the potential roles of endothelial dysfunction (35-37) and systemic inflammation (38-41). Endothelial dysfunction has long been believed to participate in the pathophysiology of hypertension and vascular diseases (35), and its role in migraine is drawing increasing attention in recent years (36). In fact, according to a pilot study from our group, migraine patients had increased circulating endothelial-specific microRNAs (37), two of which, i.e. miR-155 and miR-126, were related to a number of chronic pain disorders (42).