Oral Chinese Herbal Medicine as Prophylactic Treatment for Episodic Migraine in Adults: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

Background The prophylactic effects of Chinese herbal medicine (CHM) for migraine were examined in numerous clinical trials. This review aimed to analyze the effectiveness and safety of CHM as prophylactic treatment of migraine compared to flunarizine. Methods Nine databases were searched for randomized controlled trials (RCTs) that evaluated effects of CHM for episodic migraine prophylaxis compared to flunarizine, published before March 2019. Results Thirty-five RCTs with 2,840 participants met the inclusion criteria, and 31 of them were included in meta-analyses. The overall meta-analysis indicated that, when compared to flunarizine, CHM reduced the frequency of migraine attacks at the end of treatment (EoT) (21 studies, mean difference (MD) −1.23, 95% confidence interval (CI) (−1.69, −0.76)) and at the end of follow-up (EoFU) (five studies, MD −0.96, 95% CI (−1.70, −0.21)). Subgroup analyses based on the treatment duration, follow-up duration, and the dosage of flunarizine showed that CHM was superior to or comparable with flunarizine in reducing migraine frequency. Similar results were also found for secondary outcomes such as the pain visual analogue scale, migraine duration, responder rate, and acute medication usage. In particular, the studies that used CHM containing herb pairs (Chuan Xiong plus Bai Zhi and Chuan Xiong plus Tian Ma) showed promising results. However, the certainty of this evidence was evaluated as “low” or “very low” using the Grading of Recommendations, Assessment, Development and Evaluations approach. Conclusion CHM appeared to be comparable with flunarizine in reducing the frequency of episodic migraine attacks in adults at EoT and EoFU and well-tolerated by participants, regardless of the treatment duration, follow-up duration, and dosage of flunarizine. Due to the low certainty of the evidence, the suggested promising prophylactic outcomes require higher quality evidence from further rigorous RCTs.


Introduction
Migraine is a primary headache disorder, manifesting in episodic headache attacks which usually lasts for 4-72 hours. e typical characteristics of a migraine headache include unilateral, pulsating pain with moderate to severe intensity, aggravation by routine physical activity, and association with nausea and/or photophobia and phonophobia [1].
Migraine has an estimated global prevalence of 14.7% [2]. According to the Global Burden of Disease published in 2018, migraine was ranked as the seventh most disabling disease and the third leading cause of disability of people aged 15-49 years [3]. Furthermore, the total financial burden of the disease on individuals and society, as reported in 2010, equated to over three billion pounds a year in the United Kingdom [4].
Migraine can be subdivided into episodic and chronic migraine. e former refers to headache attacks occurring less than 15 days per month, while the latter refers to 15 or more headache days per month. Episodic migraine accounts for the majority of migraineurs [1], but the condition can progress to chronic migraine, if not properly managed [5]. It was estimated that approximately 2.5% of episodic migraine cases develop into chronic migraine annually [6].
e clinical management of migraine involves pain rescue and prophylactic treatment. Generally, migraineurs are advised to be on continuous prophylactic treatment to reduce the frequency and severity of attacks. However, it was estimated that more than half of the migraineurs were unsatisfied with prophylactic pharmacotherapy due to insufficient improvements and unbearable side effects [7][8][9]. Flunarizine is a first-line medication recommended by clinical guidelines for migraine prophylaxis [10][11][12][13]. Its effects of preventing migraine attacks in adults have been confirmed by recently published systematic reviews [14,15], clinical trials [16][17][18], and experimental studies [19][20][21][22][23]. However, unwanted adverse effects such as tiredness, mood swings, weight gain, and depression limit its use in clinical practice [16,24].
Chinese herbal medicine (CHM) has been widely used in clinical practice for thousands of years in China, usually in the form of herbal formulas consisting of a group of individual herbs and often involves herb pairs [25]. CHM has been gradually gaining acceptance worldwide [26][27][28][29]. ere have been a number of laboratory experiments [30][31][32], clinical trials [33], and systematic reviews with meta-analyses [34][35][36][37] evidence supporting CHM as a potential alternative therapy for migraine.
Four systematic reviews showed that oral CHM was more effective than placebo or conventional pharmacotherapies for migraine management [35][36][37][38]. However, none of these reviews provided evidence of prolonged treatment effects, which is important for migraine prophylaxis. In addition, the reviews accepted a range of medication used in the control groups [35][36][37]. Furthermore, most of the studies included in these reviews did not fulfil requirement that an effective migraine prophylaxis should be taken for no less than four weeks [10-12, 39, 40].
To provide more precise evidence to support the use of oral CHM for preventing episodic migraine in adults, this systematic review evaluates the clinical effectiveness and safety of CHM comparing with a first-line medication (flunarizine) in randomized controlled trials (RCTs).

Methods
is review applies the methods recommended by the Cochrane Handbook of Systematic Reviews of Interventions 5.1.0 [41]. e review protocol was registered in PROSPERO (CRD42019123039).

Database Search and Study Screening.
A rigorous electronic search was initially conducted in five English databases-PubMed, Excerpta Medica Database (EMBASE), Cumulative Index of Nursing and Allied Health Literature (CINAHL), Cochrane Central Register of Controlled Trials (including the Cochrane Library), and the Allied and Complementary Medicine Database (AMED), and four Chinese databases-Biomedical Literature, China National Knowledge Infrastructure (CNKI), Chongqing VIP (CQVIP), and Wanfang database, from their inceptions to November 2018. An updated search was conducted in March 2019. e search strategy was designed according to three groups of search terms: participant condition (migraine), intervention (Chinese medicine, CHM, and related terms), and control (flunarizine). Reference lists of previously published reviews were screened for eligibility.

Study Selection.
e inclusion criteria for this review were as follows: (1) participants aged between 18 and 75 years; (2) diagnosis of episodic migraine, with or without aura, according to clinical guidelines [1,39,40,42]; (3) treatment intervention of orally administered CHM; (4) utilized only flunarizine as the control intervention; studies which allowed acute pain medications were included if the same medications were used in both the intervention and control groups; and (5) evaluated at least one of the following outcomes: migraine frequency, number of migraine days per month, responder rate, headache pain severity, average duration of attack, acute medication usage, and health-related quality of life. e exclusion criteria were as follows: (1) studies which focused on acute migraine attack management; (2) combination of CHM with other types of Chinese medicine therapy or pharmacotherapy; (3) treatment duration of less than four weeks; and (4) different acute pain rescue medications applied in the intervention and control groups.

Data Extraction.
After screening titles and abstracts, full texts were obtained and checked for eligibility by two authors (SL and CSZ). Data from eligible studies were extracted and cross-checked by two research assistants (YX and LL) using the EpiData software (EpiData Association, Odense, Denmark). Information of authors, publication year, title, journal, setting, study design, diagnostic criteria, sample size, dropout, age, gender, duration of migraine, CHM formula names and ingredients, dosage of flunarizine, treatment duration, follow-up duration, outcome measures, and adverse events (AEs) were extracted. Disagreements were discussed and resolved by the reviewers (SL and CSZ). Where there were missing, conflicting, or unclear data, we contacted the authors of the respective studies for clarification.

Risk of Bias Assessment.
e methodological quality of included studies was assessed by two authors (SL and CSZ) using the Cochrane Risk of Bias Tool [41]. Trials were judged as "low," "unclear," or "high" risk of bias for the domains of sequence generation, allocation concealment, blinding of participants, blinding of personnel and outcome assessors, incomplete outcome data, selective reporting, and other 2 Evidence-Based Complementary and Alternative Medicine forms of bias such as conflicts of interest. Discrepancies were discussed with a third reviewer (XG).

Publication Bias Assessment.
Publication bias was assessed by funnel plots and Egger's test using the Stata 12 software (StataCorp LLC, Texas, USA), where more than 10 RCTs were included in the meta-analysis.
2.6. Certainty of Evidence Assessment. e certainty of evidence, referring to the strength or reliability of study findings, was evaluated using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach [43,44]. e GRADE approach classifies the certainty of evidence in four levels (high, moderate, low, and very low) based on five factors: risk of bias, imprecision, inconsistency, indirectness, and publication bias.

Data Analyses.
Available data were merged for metaanalyses in RevMan 5.3.0 to evaluate the effects of CHM. e primary outcome measure was the frequency of migraine, and the secondary outcomes included days of migraine, pain visual analogue scale (VAS), duration of migraine attack, responder rate, acute medication usage, quality of life scores, and AEs. Treatment effects were evaluated at two time points: at the end of treatment (EoT) and at the end of follow-up (EoFU), where possible. Frequency analyses were conducted on CHM formulas and individual herbs. Mean difference (MD) and 95% confidence intervals (CI) were used for continuous data, while risk ratios (RR) with 95% CIs were for dichotomous data. Statistical heterogeneity was assessed using the I 2 statistic. e random-effects model was selected for the meta-analyses presenting high heterogeneity with unknown reason; otherwise, the fixed-effects model was employed [41]. Where possible, subgroup analyses were performed to explore heterogeneity based on variables including the treatment duration, follow-up duration, and the dosage of flunarizine. Subgroup analyses were also conducted based on RCTs which applied the same CHM formulas and common herb pairs. AEs were summarized, and the frequencies were compared between groups.

Results
e original comprehensive electronic database search (until November 2018) identified 4,958 citations, and the updated search conducted in March 2019 yielded another 50 citations. In total, 35 RCTs met the inclusion criteria, with 31 RCTs included in the meta-analyses ( Figure 1).

Risk of Bias.
All studies mentioned randomization, however, only 13 RCTs (37.1%) [46,48,49,51,55,59,60,62,67,69,76,78,79] were assessed to have "low risk" of bias in terms of sequence generation; four studies (11.4%) [45,50,54,74] were considered as "high risk" in this domain as participants were allocated based on the order of enrollment. All included studies were assessed as "unclear risk" of bias for allocation concealment due to lack of adequate information. In terms of blinding of participants, research personnel, and outcome assessors, all studies were judged as "high risk" of bias because no adequate blinding methods were employed despite the different types of intervention between groups. Most of the studies were at "low risk" of bias for incomplete outcome data; only one [48] was assessed as "high risk" in this domain due to a high and unbalanced dropout rate.
irty-one studies (88.6%) were assessed as "unclear risk" of bias regarding selective outcome reporting due to the lack of registered protocols, while four RCTs were assessed as "high risk" because they did not report AEs [53] or outcome data of the follow-up phase [56][57][58] ( Figure 2). Studies included in meta-analysis (n = 31)
Based on the dosage of flunarizine, it was found that CHM showed equivalent effects when compared to a dose of 5 mg flunarizine at EoFU (two studies, MD: −1.29, 95%

Assessment Using GRADE.
e certainty of evidence obtained from meta-analyses on the primary outcome measures is presented in Table 4. Oral CHM was more effective than flunarizine for reducing migraine frequency at EoT and EoFU, but the certainty of this evidence was "low" and "very low," respectively.

Adverse Events.
AEs were categorized and calculated to assess the safety of the treatments. Based on the available data from 15 studies that reported detailed information of AEs, the number of AEs in the CHM group was less than that in the flunarizine group (34 vs 50). As shown in Table 5, most AEs were mild and did not require additional medical management. None of the participants dropped out due to AEs. Gastrointestinal symptoms (i.e., nausea, stomach discomfort, diarrheal, and abdominal distension) and other symptoms (including drowsiness, dizziness, fatigue, and insomnia) were commonly seen in both CHM and flunarizine groups. Other AEs reported in the CHM group were three cases of irregular menstruation [57], one acne [48], and one slight decrease in platelet count [55], while the flunarizine group had common reports of weight gain [57,76], some extrapyramidal symptoms such as ataxic [57,72] and involuntary movements [47], as well as one case of moderate liver function impairment [76].

Summary of Results.
is systematic review provides evidence showing that oral CHM is more effective than flunarizine for episodic migraine prophylaxis based on these outcome measures: migraine frequency, pain VAS, migraine attack duration, responder rate, and acute  medication usage at EoT. Oral CHM also showed better effects than flunarizine for migraine frequency and acute medication usage at EoFU. In addition, there was no difference between CHM and flunarizine in migraine days (at both EoT and EoFU) and pain VAS and migraine attack duration at EoFU. Nevertheless, the overall methodological quality of the included studies was low, and the certainty of evidence was "low" or "very low" based on GRADE assessment. As substantial heterogeneity existed in most meta-analyses, subgroup meta-analyses were conducted on the primary outcome measures, based on the treatment duration, follow-up duration, and dosage of flunarizine. ese subgroup analyses showed that CHM produced superior or equivalent effects as flunarizine. However, heterogeneity remained considerable within the subgroup analyses. e possible causes of heterogeneity are the uses of different CHM formulas, disease severity, and duration across trials. It should be noted that two studies reported migraine frequency at EoT with exceptional results [51,75]; therefore, they were excluded from the metaanalysis of this outcome. is study shows that CHM is well-tolerated when compared to flunarizine. Fifteen studies provided detailed information on AEs; all of them were mild or moderate and did not require specific management. Most of the common complaints in both CHM and flunarizine groups such as fatigue, insomnia, and digestive problems could be the associated symptoms of migraine [1] rather than side effects caused by treatments. One case of decreased platelet count was reported, but there was not a confirmed association between this event and CHM [55]. Weight gain and ataxia were only reported in the flunarizine group, which agrees with previous research results on flunarizine [15,16,18,24,81,82]. Considering that aging [81,[83][84][85] and increased dosage [11,40,86,87] of flunarizine and the predisposing factors for its side effects, CHM could be an alternative for elderly patients and those who suffer from common side effects of flunarizine.
As shown in the meta-analyses, CHM has an advantage in reducing acute pain medication usage, indicating its potential of preventing medication overuse, which is a common concern in headache [10] and chronic migraine [88] treatment.
Previous research suggested that CHM is an effective add-on therapy for migraine [36]. It is known that flunarizine should be taken for several weeks to show its full effects in migraine prevention [11,40,83,84]; during this period, patients may be unsatisfied with its treatment effects. e subgroup analysis in our review demonstrated that oral CHM was superior to flunarizine when they were used for four weeks and equivalent to flunarizine when they were used for eight weeks. Hence, our results support the use of CHM as a potential adjuvant therapy to increase the effectiveness of flunarizine. However, the drug interactions between flunarizine and Chinese medicine herbs or formulas have not been well investigated, and this is an area which requires further research. 14 Evidence-Based Complementary and Alternative Medicine

Mechanisms of Herbs.
ere has been increasing experimental research on the active compounds of CHM in attempts to elucidate their potential mechanisms for migraine. For example, one of the main compounds found in Chuan Xiong (Ligusticum chuanxiong), senkyunolide I, was proved to reduce migraine pain by adjusting the levels and turnover rates of monoamine neurotransmitters and decrease nitric oxide (NO) levels in the blood and brain [32]. Bai Zhi (Angelicae Dahuricae Radix) was reported to have antimigraine actions by modulating the levels of vasoactive substances such as NO, calcitonin gene-related peptides, and endothelin [89,90]. Tian Ma (Gastrodia elata) contains the ⊕○○○ Very low a,b,c GRADE working group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: we are moderately confident in the effect estimate. e true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: our confidence in the effect estimate is limited. e true effect may be substantially different from the estimate of the effect Very low certainty: we have very little confidence in the effect estimate. e true effect is likely to be substantially different from the estimate of the effect Explanation: a High risk of bias in blinding may limit the certainty of the results; b high heterogeneity may limit the certainty of the results; c small sample size may limit the certainty of results  active ingredient, gastrodin, which has been found to demonstrate antimigraine, antihyperalgesic, and antinociceptive effects, possibly by inhibiting trigeminal nerve activation at central sites and also inhibiting the peripheral release of calcitonin gene-related peptides following the NO scavenging effect [91,92].
is review has also provided meta-analysis evidence supporting the use of two herb pairs in migraine, namely, Chuan Xiong plus Bai Zhi and Chuan Xiong plus Tian Ma. Herb pairs form the basis of CHM formulation and are believed to result in synergistic effects or reduced side effects/toxicity [93]. It has been suggested that herb pairs are potential research entry-point for research on CHM mechanisms [93]. e two herb pairs evidenced in this review are also CHM formulas that have been traditionally used for the treatment of headaches/migraines [80]. e combination of Chuan Xiong and Bai Zhi is a formula known as Du Liang Wan with experimental studies showing the function of adjusting the level of neurotransmitters and vasoactive substances to relieve neurogenic inflammation [94,95]. e other herb pair (Chuan Xiong plus Tian Ma) is known as Da Chuan Xiong Wan, which has been proved to reduce inflammatory mediators through inhibition of the NF-kappaB pathway [31].

Strengths and Limitations of is Study.
One strength of this review is that the prolonged effects of oral CHM has been evaluated, which has been highlighted by clinical guidelines as an important outcome assessment of prophylactic treatments of migraine [10,11,96,97]. Furthermore, the active comparator in this review was restricted to flunarizine with a treatment duration of at least four weeks; this is consistent with the recommendations of clinical guidelines for migraine prophylaxis [10-12, 39, 40]. is allows for more targeted evaluations and reduces variables regarding the different types and doses of conventional migraine prophylaxis treatment.
is review also conducted subgroup analyses based on the treatment duration, follow-up duration, and dosage of flunarizine. e major limitations of this systematic review include the low methodological quality of included studies and high heterogeneity across most meta-analyses and subgroup analyses, reducing the certainty of evidence. Future studies need to adopt more rigorous designs to ensure appropriate sequence generation, allocation concealment, and blinding procedures. It was noted that the CHM formulas were administered with different forms including decoction, pills, and granule. However, this systematic review failed to conduct subgroup meta-analysis based on CHM forms due to the small number of studies which applied same CHM formulation and reported same outcome measures. Differences of effectiveness among diverse CHM forms with the same formulation could be explored in future studies. Safety evaluation of treatments should also be given more attention in future clinical studies so that clinicians and patients will be able to make more informed decisions.

Conclusion
Cautiously, the oral CHM has the potential to act as an alternative prophylactic treatment of migraine. e results from this review show that the effects of oral CHM are, at least, equivalent to flunarizine in preventing migraine attacks in adults at EoT and EoFU, well-tolerated by participants, regardless of the treatment duration, follow-up duration, and dosage of flunarizine. However, these results need to be interpreted with caution due to the low certainty of evidence. Future studies with more rigorous designs are needed to provide more concrete evidence for stronger conclusions. is review also provides evidence for two herb pairs, Chuan Xiong plus Bai Zhi and Chuan Xiong plus Tian Ma for migraine prophylaxis. In addition, this review draws attention to the potential and need to evaluate oral CHM as an adjunct treatment to flunarizine in the prophylactic treatment of migraines.