Herbal Medicine for Behçet’s Disease: A Systematic Review and Meta-Analysis

Patients with Behçet’s disease often use complementary and alternative medicine for treating their symptoms, and herbal medicine is one of the options. This systematic review provides updated clinical evidence of the effectiveness of herbal medicine for the treatment of Behçet’s disease (BD). We searched eleven electronic databases from inception to March 2020. All randomized controlled trials (RCTs) or quasi-RCTs of BD treatment with herbal medicine decoctions were included. We used the Cochrane Handbook for Systematic Reviews of Interventions to assess the risk of bias and the grading of recommendations assessment, development and evaluation (GRADE) approach to assess the certainty of evidence (CoE). Albatross plot was also used to present the direction of effect observed. Eight studies were included. The risk of bias was unclear or low. The methodological quality was low or very low. Seven RCTs showed significant effects of herbal medicine on the total response rate (Risk ratio, RR 1.26, 95% CI 1.09 to 1.45, seven studies, very low CoE). Four RCTs showed favorable effects of herbal medicine on the erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) level compared with drug therapy. Herbal medicine favorably affected the ESR (MD −5.56, 95% CI −9.99 to −1.12, p = 0.01, I2 = 96%, five studies, very low CoE). However, herbal medicine did not have a superior effect on CRP. Two RCTs reported that herbal medicine significantly decreased the recurrence rate after three months of follow-up (RR 0.23, 95% CI 0.09 to 0.63, two studies, low CoE). Our findings suggest that herbal medicine is effective in treating BD. However, the included studies had a poor methodological quality and some limitations. Well-designed clinical trials with large sample sizes are needed.

Two authors (J.H.J. and T.Y.C.) independently searched 11 electronic databases and read all eligible studies in full to determine the extent to which they met the eligibility criteria. Disagreements were resolved by HWL. Two authors (J.H.J. and T.Y.C.) extracted the data from the included studies. The data extraction form collected the first author, year of publication, diagnosis, sample size, duration of treatment, intervention group, control group, the main outcome, results, and AEs. Disagreements were resolved by a third author (H.W.L).

Risk of Bias
Two evaluators (J.H.J. and T.Y.C.) assessed the studies using the risk of bias assessment tool from the Cochrane Handbook for Systematic Reviews of Interventions [18]. The following seven domains were assessed: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other biases. We evaluated the risk of bias as "L" (low risk of bias), "H" (high risk of bias), and "U" (risk of bias is uncertain). Disagreements were resolved by M.S.L. We used the online version of the grading of recommendations assessment, development, and evaluation (GRADE) to assess the certainty of the evidence (CoE). The following seven categories were assessed: (1) number of studies, (2) study design, (3) risk of bias, (4) inconsistency, (5) indirectness, (6) imprecision and (7) other considerations [19].

Data Analysis
Data analyses were performed using Review Manager (Ver. 5.3) software. For dichotomous data, we used the treatment effect as the risk ratio (RR) with a 95% confidence interval (CI). For continuous data, we present the treatment effect as the mean difference (MD) with 95% CI. The chi-squared test and Higgins I 2 test were used to assess heterogeneity. To supplement the results for the meta-analysis of available effects, albatross plots of each included study sample size against respective p-values were used to provide a visual extension of effect direction for the primary and secondary outcomes using STATA/SE v. 16.1 (StataCorp LLC, College Station, TX, USA).

Description of the Included Trials
The searches identified 2036 potentially relevant studies, of which eight [20][21][22][23][24][25][26][27] studies met our inclusion criteria ( Figure 1). The key data from all included RCTs are summarized in Table 1. The RCTs published in Chinese included three master theses. The sample size ranged from 30 to 180. The duration of treatment ranged from three weeks to three months. The included studies used different disease criteria. Five RCTs [21,22,24,26,27] diagnosed BD according to the 1989 ISG criteria, two RCTs [20,23] used the 1989 ISG criteria plus the Standard for Disease in Traditional Chinese Medicine diagnostic criteria, and one RCT [25] used the 2005 ICBD criteria plus the Standard for Disease in Traditional Chinese Medicine diagnostic criteria.    For the control group treatment, three RCTs [20][21][22] used prednisone, two RCTs [23,24] used thalidomide, and three RCTs used prednisone plus thalidomide [25], loxoprofen sodium plus thalidomide [26], and interferon α-2b injection [27]. Seven RCTs used oral administration, and another RCT used injections.
The prescriptions used in the intervention group were different. The constituents of the herbal medicines used in each included study are listed in detail in Table 2. There were 8 prescriptions collected, among which 3 were set prescriptions [20,22,26], 2 were modified set prescriptions [24,25], 2 were pattern identification (PI) prescriptions [23,27], and 1 prescription was formulated based on personal experience [21]. There were 72 herbs in total. The most commonly used herbs for BD were Glycyrrhizae Radix et Rhizoma, Angelicae Sinensis Radix, Paeoniae Radix Alba, Asparagi Radix, and Glycyrrhizae Radix et Rhizoma Praeparata.

Risk of Bias
The risk of bias is presented in Figure 2. The risk of bias was assessed using the Cochrane risk of bias tool. Only one RCT [25] used the random number method for random sequence generation. Seven RCTs [20][21][22][23][24]26,27] did not report the random sequence generation method. Among the eight included RCTs [20][21][22][23][24][25][26][27], herbal medicine decoctions, and drug therapy were compared; thus, blinding could not be applied to participants and personnel. None of the RCTs described the method of allocation concealment or blinding of outcome measurement. One RCT [22] did not provide the reasons for patient drop-out and withdrawal. None of the RCTs published or registered their protocol, and they all had an unclear risk of bias with regard to selective outcome reporting.

Risk of Bias
The risk of bias is presented in Figure 2. The risk of bias was assessed using the Cochrane risk of bias tool. Only one RCT [25] used the random number method for random sequence generation. Seven RCTs [20][21][22][23][24]26,27] did not report the random sequence generation method. Among the eight included RCTs [20][21][22][23][24][25][26][27], herbal medicine decoctions, and drug therapy were compared; thus, blinding could not be applied to participants and personnel. None of the RCTs described the method of allocation concealment or blinding of outcome measurement. One RCT [22] did not provide the reasons for patient drop-out and withdrawal. None of the RCTs published or registered their protocol, and they all had an unclear risk of bias with regard to selective outcome reporting.

Certainty of Evidence
The CoE for each outcome was either low or very low (Table 3). CRP: C-reactive protein; CI: confidence interval; ESR: erythrocyte sedimentation rate; RR: risk ratio; MD: mean difference. * The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). a Downgraded by one level: unclear or high risk of bias; b Downgraded by one level: heterogeneity is high; c downgraded by one level: small sample size; d downgraded by two levels: heterogeneity is very high. GRADE Working Group grades of evidence: low certainty ( ): our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect; very low certainty ( ): we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Skin Lesions
Four RCTs [20,21,23,25] assessed the symptom scores for skin lesions. Two RCTs [20,25] reported that the effects of herbal medicine were superior to those of drug therapies, while two RCTs [21,23] showed equivalent effects between the intervention group and the control group. Only three RCTs [20,23,25] were applicable for meta-analysis. The result of the meta-analysis showed a favorable effect of herbal medicine with regard to reducing skin lesions (MD −1.62, CI −2.65 to −0.59, three trials, n = 160, p = 0.002, I 2 = 66%, Figure 4F).

Albatross Plot
The albatross plots for the included studies are shown in Figure 5. Albatross plots were performed by illustrating contours that showed the effect direction and effect sizes range using p-values and given sample sizes. Different plotting colors correspond to the outcome subgroups. Looking at the albatross plot for dichotomous data (Figure 5A), the points scattered across different contour lines. However, most of the points clustered to the right side of the plot, showing that herbal medicine was more favorable for the treatment of BD. For the albatross plot of continuous data ( Figure 5B), the points were less scattered. Although the points were clustered to the right, many points were positioned around the null line, implicating non-significant effects. Both albatross plots had points that were completely isolated, reflecting the possibility of sampling error.

Summary of the Main Results
The aim of this systematic review was to evaluate the effectiveness and safety of herbal medicine for the treatment of BD. Eight studies [20][21][22][23][24][25][26][27] evaluating the effect of herbal medicine on BD showed the following results. Herbal medicine reduces the ESR, reduces the symptom scores (oral ulcers, genital ulcers, eye inflammation, skin lesions), decreases the recurrence rate and improves the total clinical effective rate, although some studies have not provided evidence of the superiority of herbal medicine in terms of the symptom score for arthralgia and the CRP. Three RCTs [21,23,27] reported Es, but these Es were generally mild, and the patients spontaneously recovered. Overall, the results showed that herbal medicine decoctions might be useful in the treatment of BD.

Overall, Completeness and Applicability of the Evidence
This review shows that herbal medicine can be used to improve clinical symptoms in BD patients and that there are fewer Es associated with its use than with drug therapies. Despite the positive results, the included studies had small sample sizes and generally poor methodological quality; furthermore, they were too heterogeneous to allow any firm conclusions to be drawn regarding the different types of herbal prescriptions.

Quality of the Evidence
The CoE was low and very low for all outcomes (Table 3). Among the included RCTs, none reported the randomization methods, allocation concealment, or blinding of information. They did not publish their protocols, and it was not clear whether the planned result indicators were reported accurately. Therefore, they were downgraded one level in the risk of bias domain. The heterogeneity was substantial; thus, they were downgraded one or two levels. The included studies were PICO (patient, intervention, comparison, outcomes) studies, and it was determined that there was insufficient direct evidence of an effect. All outcomes had wide CIs that crossed the assumed threshold of the minimal clinically important difference; thus, they were downgraded one level for imprecision. Furthermore, the number of trials and total sample sizes included in our analysis was not sufficient to enable us to draw firm conclusions.

Potential Biases in the Review Process
This review has several limitations. First, the included studies used different herbal prescriptions, the effectiveness of which for the treatment of BD was not well known. Therefore, future studies should analyze studies using similar herbal prescriptions. Second, the evidence of improvements in symptoms varied according to the herbal decoction, possibly due to the varying compositions and dosages of the herbs. This review shows that herbal medicine has effects that are superior to those of drug therapy according to the composition of herbs, but the effects of dosages were unclear. Therefore, future studies should focus on the detailed composition and dosages of herbs. Third, all included studies were conducted in China, where no negative studies have been reported [28]. Furthermore, the albatross plots also showed scattered points across contour lines, with a few points being isolated from the other point clusters. As the sample sizes of the included studies were relatively small, this would likely reflect possible sampling bias.

Agreements and Disagreements with Other Studies or Review
We found three previous systematic reviews [12][13][14] on the use of herbal medicine for BD. These studies reported that herbal medicine was better than drug therapy for the treatment of BD. We identified two new RCTs [24,25] and extracted evidence from them. The results and evidence levels were similar to those of the studies included in the three previous systematic reviews. Moreover, the authors of those reviews expressed concern regarding the small sample sizes and the poor quality of the included studies. Future well-designed RCTs with large sample sizes are thus warranted.

Potential Mechanism of Action
In spite of the comparative absence of compelling evidence toward herbal medicines for BD, the potential features that may be related point towards benefits. These properties include anti-inflammation, immunoregulation, and antioxidation with chronic autoimmune disease and the studies focusing on the fusion of medicinal plants and cytokine activity effects. Since the disparity in the expression of innate immunity-related cytokines cannot only play a crucial role in BD pathogenesis but can also be pivotal in the level of severity of the disease [29]. Further, the disease activity score and clinical activity index may also be influenced by the levels of anti-inflammatory cytokines [30,31]. The herbal prescriptions mostly used Glycyrrhizae Radix et Rhizoma and Angelicae Sinensis Radix, which have anti-inflammatory effects [32,33]. The biological of herbal medicines linked with BD were not focused on in the present review. However, herbal medicines' properties used to treat BD must be researched further.
The therapeutic effects of herbal medicine possibly rely on the obtainability and quantity of the different components in the production. The applicable data are sometimes not included in many publications. The daily prescribed quantity of the trials included in the study is diverse across the included studies-in their condition severity and traditional diagnosis type. However, no research has been conducted on the prime dose to reduce BD symptoms. In the present study, when we analyzed the result direction and dosage, treatment time, dosage and time, and the type of results, no direct links with dose and the treatment time for relevant changes of various results. The variety in trials does not present clear relations. Different herbal medicines were compared and examined with drug therapies using several methods. The contrasts in significance may result from the type of herbal medicines and dosage in treatments. The quantity and frequency of herbal medicines utilized in the trials included in the study may be inadequate to create a noteworthy effect in biochemical variables. Thus, it is required to conduct studies ranging in doses and comparing a variety of herbal medicines to various outcomes to answer such questions.

Implications for Nutrients
The applicability of the present review could be questioned in the fields of nutrients. Herbal materials are derived from plants, and many such substances are part of both food supplements and nutraceuticals as well as medicinal products [34][35][36][37]. In the US, regulatory bodies such as FDA governing plant-based medicines usually regard them as dietary supplements [35]. Dietary recommendations refer to herb usage as an outstanding source of antioxidants in Australia [34,38]. Furthermore, in Traditional East Asian Medicine, such medicines are consumed in the form of herbal tea or supplemental food. With the increase of the presence of herbs in diets owing to their health advantages, utilization of herbal medicines for BD may be possible in the capacity of a herbal supplement, in addition to the main diet containing functional foods or nutraceuticals per respective regulatory systems across countries [35,[39][40][41]. However, such supplements for BD should ensure avoidance of any side effects by undergoing quality testing and safety protocols taken for dietary supplements and functional foods.

Implications for Practice
BD is a chronic inflammatory disease in which ulceration occurs repeatedly. Herbal medicine is associated with a lower rate of recurrence and is relatively safer than drug therapies. It appears useful in clinical practice. However, the included studies had only short-term treatment periods and used various forms of prescriptions. Thus, long-term clinical research and standardized prescriptions should be implemented in future studies.

Implications for Research
This review has several limitations with regard to the research process. First, the risk of bias in this review was unclear. The majority of trials did not report randomization procedures, and all of them lacked information on blinding. None of them reported the randomization and allocation methods, published their protocols, or registered at PROSPERO. Thus, future studies should be described in detail or registered at PROSPERO. Second, rigorous RCTs should be carried out to analyze the effectiveness of herbal medicine for the treatment of BD. Adequate data on the clinical outcomes of BD treated with herbal medicine could guide clinical decision-making. Future studies should be comprehensively reported according to the CONSORT reporting guidelines [42].

Conclusions
This review showed that herbal medicine decoctions might be useful in the treatment of BD. However, the quality of the current evidence was low, the small effect size reduced the clinical significance, and the small number of rigorous studies prevented us from drawing firm conclusions. Well-designed RCTs are needed to determine whether herbal medicine is a viable option for the treatment of BD.

Institutional Review Board Statement:
Not applicable for studies not involving humans or animals.