Herbal Medicine Formulas for Parkinson's Disease: A Systematic Review and Meta-Analysis of Randomized Double-Blind Placebo-Controlled Clinical Trials

Background: Parkinson's disease (PD) is a debitlitating, chronic, progressive neurodegenerative disorder without modifying therapy. Here, we aimed to evaluate the available evidence of herbal medicine (HM) formulas for patients with PD according to randomized double-blind placebo-controlled clinical trials. Methods: HM formulas for PD were searched in eight main databases from their inception to February 2018. The methodological quality was assessed using Cochrane Collaboration risk of bias tool. Meta-analysis was performed using RevMan 5.3 software. Results: Fourteen trials with Seventeen comparisons comprising 1,311 patients were identified. Compared with placebo groups, HM paratherapy (n = 16 comparisons) showed significant better effects in the assessments of total Unified Parkinson's Disease Rating Scale (UPDRS) (WMD: −5.43, 95% CI:−8.01 to −2.86; P < 0.0001), UPDRS I (WMD: −0.30, 95% CI: −0.54 to −0.06; P = 0.02), UPDRS II (WMD: −2.21, 95% CI: −3.19 to −1.22; P < 0.0001), UPDRS III (WMD: −3.26, 95% CI:−4.36 to −2.16; P < 0.00001), Parkinson's Disease Quality of Life Questionnaire (p < 0.01) and Parkinson's Disease Questionnaire-39 (WMD: −7.65, 95% CI: −11.46 to −3.83; p < 0.0001), Non-motor Symptoms Questionnaire (p < 0.01) and Non-Motor Symptoms Scale (WMD: −9.19, 95% CI: −13.11 to −5.28; P < 0.00001), Parkinson's Disease Sleep Scale (WMD: 10.69, 95% CI: 8.86 to 12.53; P < 0.00001), and Hamilton depression rating scale (WMD: −5.87, 95% CI: −7.06 to −4.68; P < 0.00001). The efficiency of HM monotherapy (n = 1 comparison) was not superior to the placebo according to UPDRS II, UPDRS III and total UPDRS score in PD patients who never received levodopa treatment, all P > 0.05. HM formulas paratherapy were generally safe and well tolerated for PD patients (RR: 0.41, 95% CI: 0.21 to 0.80; P = 0.009). Conclusion: The findings of present study supported the complementary use of HM paratherapy for PD patients, whereas the question on the efficacy of HM monotherapy in alleviating PD symptoms is still open.


INTRODUCTION
Parkinson's disease (PD) is a common chronic neurodegenerative disease characterized by the degeneration of dopaminergic neurons in the substantia nigra (SN) , and presents with non-motor or/and motor syndrome (Rogers et al., 2017). In the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2016, PD was the second leading cause in neurological disorders of years lived with disability (YLDs), contributing to 6.1 million of patients (GBD, 2016 Disease andInjury Incidence andPrevalence Collaborators, 2017). From 2005 to 2015, global deaths due to PD increased by 42.4%, to 117.4 thousands deaths (GBD, 2015 Mortality andCauses of Death Collaborators, 2016), as a result of population aging. With the growing incidence, PD seriously hurt the physical and mental health of the elderly, also produced a heavy economic burden on both families and society. The average annual cost per PD patient was $22,800 in the United States (Kowal et al., 2013) and $36,085 in the UK (Findley et al., 2011). Current conventional treatment for PD is based on the dopamine (DA) replacement therapies and reduction of DA degradation, including levodopa, DA receptor agonists, monoamine oxidase-B inhibitors, catechol-Omethyltransferase inhibitors and other types of drugs (Rogers et al., 2017). However, all the current therapeutic approaches remain palliative and can't inhibit or reverse the progression of PD (Athauda and Foltynie, 2015). Furthermore, frequently with these treatments can lead to obvious adverse events and efficacies diminished, as well as induce therapy-related motor complications such as dyskinesia, choreoathetosis, and fluctuations in motor function (Jenner, 2015). A safer and more effective alternative treatment of PD is increasingly demanded.
The therapy of herbal medicine (HM) for PD is particularly common. In China, HM could be traced in the Huangdi Neijing (Inner Canon of Yellow Emperor) (Zheng, 2009), the earliest existing classics in Chinese herbal medicine (CHM). Up to now, HM is still very popular in the treatment of PD especially in Asian countries (Wang et al., 2011(Wang et al., , 2013. Previous reviews Zhang et al., 2015) found lack of evidence of supporting the use of HM for PD patients because of the generally low-quality studies included. Here, we performed a systematic review and meta-analysis of randomized double-blind placebocontrolled clinical trials of HM formulas for PD patients and further explored the mechanisms of high-frequently used herbs against PD.

METHODS
This systematic review and meta-analysis is conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement (Moher et al., 2010b) and our previous study .

Search Strategy
Randomized double-blind placebo-controlled clinical trials of HM formulas for PD were searched in eight databases from their inception to February 2018. They are PubMed, EMBASE, Cochrane Central Register of Controlled Trials (CENTRAL), Web of science, Chinese National Knowledge Infrastructure (CNKI), Chinese VIP Information, Wanfang database and Chinese Biological Medical Literature Database (CBM). Moreover, we hand searched additional relevant studies using the reference list of previous reviews. The search strategy of PubMed was as follows, and was modified to suit other English or Chinese databases. PubMed

Study Selection
Two authors (CS-S and H-FZ) independently engaged in the selection of studies by reading study titles, abstracts and full texts. The disagreement was resolved by the corresponding author (GZ) or repeated discussion.

Inclusion Criteria
Type of study: the articles were randomized double-blind placebo-controlled clinical trials.
Type of participants: participants were of any age or sex with a confirmed diagnosis of PD according to the UK Brain Bank criteria (Hughes et al., 1992), Chinese National Diagnosis Standard (CNDS) for PD in 1984 (Wang, 1985), CNDS updated version in 2006 for PD (Zhang, 2006) or other formal comparable criteria. Type of intervention: Analyzed interventions were HM formulas or HM formulas plus western conventional medicine (WCM) according to PD treatment guidelines, 2 regardless of the form of the drug, dosage, frequency or duration of the treatment. Comparator interventions were placebo or placebo plus WCM. Type of outcome measures: the primary outcomes were total Unified Parkinson's Disease Rating Scale (UPDRS) score, UPDRS I (Mental Score), UPDRS II (Activities of Daily Life), UPDRS III (Motor Score), and UPDRS IV (Complications of treatment). The secondary outcomes were: (1) Parkinson's Disease Quality of Life Questionnaire (PDQL) and Parkinson's Disease Questionnaire-39 (PDQ-39); (2) Non-motor Symptoms Questionnaire (NMSQuest) and Non-Motor Symptoms Scale (NMSS); (3) Parkinson's Disease Sleep Scale (PDSS); (4) Hamilton depression rating scale (HAMD); (5) Adverse events.

Exclusion Criteria
Studies were excluded if they were any one of the followings: (1) clinical trials evaluating the other alternative and complementary medicines mixed in the experimental group or control group in the treatment of PD; (2) single herb, herbal extracts and herbal components; (3) case series, reviews, comments and protocols; (4) animal studies and in vitro studies; (5) duplicated publications.

Quality Assessment
The methodological quality was evaluated by using the Cochrane Collaboration's risk of bias tool (Higgins et al., 2011). The quality of each study was assessed by following seven biases: adequate sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data addressed (ITT analysis), free of selective reporting and other bias. Each domain can be rated as "+" (low risk of bias), "-" (high risk of bias), or "?" (unclear risk of bias), which were the three categories for the degree of each potential bias.

Data Extraction
Two authors (CS-S and HF-Z) independently extracted the data according to predefined extraction form as follows: (1) General information: the first author's name, publication year, and publication language; (2) Participants: diagnostic criteria, study design, total number and number in comparison groups, gender and mean age; (3) Intervention: herbal preparations, dose, frequency, course of treatment, followup; (4) Outcome measures. If the study had multiple comparison groups, we chosen the most relevant groups for analysis. The original authors were contacted if further information was needed. Disagreements were resolved through discussing with corresponding author (GZ).
The constituent of HM formulas for PD in each included study was recorded. The herbs with cumulative frequencies over 50% are documented and ranked.

Data Synthesis and Statistical Analysis
We synthesized all data and performed meta-analyses on RevMan 5.3 software. Continuous outcomes were using weighted mean differences (WMD) or standardized mean differences (SMD) with 95% confidence intervals (CIs), while dichotomous outcomes were summarized using risk ratio (RR) with 95% confidence intervals (CIs). Heterogeneity among studies was detected by I 2 and Chi 2 tests. If substantial statistical heterogeneity existed (I 2 ≥ 50%, P < 0.10), a random-effects model was used. If there was no observed heterogeneity (I 2 < 50%, P > 0.10), a fixed-effect model was applied. Possible sources of heterogeneity were explored by subsequent sensitivity analyses. If more than ten trials were identified in each outcome, publication bias was detected by funnel plot analyses and Egger's test.

Description of the Screening Process
The detailed screening process was summarized in the PRISMA flow diagram (Figure 1). A total of 7,521 potentially relevant hits were initially yielded from the eight databases and other sources, in which 6,570 records were remained after removal of duplicates.

HM Monotherapy vs. Placebo
One study (Zhao et al., 2009) showed that the efficacy of HM monotherapy was similar to placebo according to UPDRS II (P > 0.05), UPDRS III (P > 0.05) and total UPDRS score (P > 0.05) in PD patients who never received levodopa treatment.

HM Monotherapy vs. Placebo
In the only one study (Zhao et al., 2009), neither the experimental group nor the control group provide any information about adverse events.

Publication Bias
We did not performed the Funnel plot and Egger's test because the number of studies in each meta-analysis was less than ten.

Summary of Evidence
This is first systematic review of randomized double-blind placebo-controlled clinical trials to assess the efficacy and safety of HM formulas for PD. Fourteen high-quality randomized controlled trials (Pan et al., 2009(Pan et al., , 2011(Pan et al., , 2013Zhao et al., 2009Zhao et al., , 2013Guo, 2010;Kum et al., 2011;Chen M. Y. et al., 2014;Guo et al., 2014;Wen et al., 2015;Li et al., 2016;Yu, 2016;Cai et al., 2017;Yang, 2017) involving 1,316 patients suffering from PD were identified. HM paratherapy was significant for improving motor symptoms and non-motor functions, whereas there was a negative result of complications of treatment. One trail (Zhao et al., 2009) indicated that HM monotherapy was not superior to the placebo. Eleven out of fourteen studies (Guo, 2010;Kum et al., 2011;Pan et al., 2011Pan et al., , 2013Zhao et al., 2013;Chen M. Y. et al., 2014;Guo et al., 2014;Li et al., 2016;Yu, 2016;Cai et al., 2017;Yang, 2017) reported no serious adverse events relevant with HM formulas, indicating that HM formulas were generally safe and well tolerated for PD patients. Thus, the findings of present study supported the complementary use of HM paratherapy for PD patients, whereas HM monotherapy for PD is still lack of evidence.

LIMITATIONS
First, the members of the International Committee of Medical Journal Editors published a statement requiring that all clinical trials must be registered in order to be considered for publication (DeAngelis et al., 2004). However, most of included studies didn't formally register. Protocols were not available to confirm free of selective reporting. Thus, further clinical trials must register prospectively in international clinical trials registry platform. Second, although we included randomized double-blind placebocontrolled trials, some inherent and methodological weaknesses still existed in the primary studies: (1) An adequate sample size is crucial to the design of RCTs (Lewis, 1999), but only 4 trials (Kum et al., 2011;Zhao et al., 2013;Li et al., 2016;Yu, 2016) applied pre-trial sample size estimation; (2) PD is a chronic degenerative disease. Long-term efficacy and safety are important assessments to decide the clinical usefulness of an agent in treatment, but only one trial  had the long-term duration of follow-up at 6 month; (3) Intention-totreat (ITT) analysis could avoid bias and false-positive results, which is the recommended standard approach to analyse data from RCTs (Abraha et al., 2017). However, only two studies (Kum et al., 2011;Li et al., 2016) adopted ITT analysis. (4) In the present study, only 6 trials conducted assessor blinding. Considering the characteristics of outcome measurement of PD patients (e.g., UPDRS), assessor blinding successfully eliminates assessment bias and increases the accuracy and objectivity of outcomes results. Triple blindness is needed in further PD trials. Thus, CONSORT 2010 statement (Moher et al., 2010a) and CONSORT Extension for Chinese Herbal Medicine Formulas 2017 (Cheng et al., 2017) should be applied in trial reporting and publication. Third, the herbal composition, drug formulation and dose of the intervention were not exact same, which would lead to clinical heterogeneity. To assess the efficacy and safety of HMs in a clinical trial, all subjects should be given exactly the same intervention in terms of product identity, purity, dosage, and formulation. Fourth, our study only included trials published in the English and Chinese languages and all the included studies were conducted in China, which may affect the generalizability of present findings. In the further studies, the international coliaboration is needed in order to get more qualified stuidies. Finally, different types and stages of PD can influence disease progression and response to treatment (Reinoso et al., 2015). It is difficult to differentiate the effectiveness of  HM formulas targeting these subgroups due to insufficient data of primary studies. The pertinent research should be conducted in future clinical trials, which would contribute significantly to explore the responsiveness of specific PD subgroup to interventions.

Implications
Up to now, several systematic reviews of traditional medicine for PD (Kim et al., 2012;Wang et al., 2012;Zhang et al., 2014Zhang et al., , 2015 have been performed. However, low-quality of included primary studies hindered our conclusions. For example, the two articles written by Kim et al. (2012) and Zhang et al. (2015) belong to high-quality systematic reviews; however, the inherent limitations existed in the included low-quality primary studies. The present study only included randomized double-blind placebo-controlled trials, which remains the gold standard of trial design (Athauda and Foltynie, 2016). These trials reported the detailed randomized methods; placebocontrolled group accounts for the placebo effects that don't depend on the treatment itself , and thus increasing the reliability of experiment results. The present study provided the evidence to support HM paratherapy for PD, whereas there is still lack of available evidence for HM monotherapy for PD. However, it should be remembered that a lack of scientific evidence does not necessarily mean that the treatment is ineffective (Kotsirilos, 2005). To explore the efficacy of HM monotherapy for PD is needed in the future.
Currently, most available PD therapies are mainly aimed at motor symptoms . Non-motor symptoms (NMS) are common in PD patients across all disease stages and are a key determinant of QOL (Martinez-Martin et al., 2012). However, NMS have received limited attention and targeted treatments remain a challenge (Kulisevsky et al., 2018). The present systematic review provided the sportive evidence for the effectiveness and safety of HM paratherapy for NMS of PD patients. Thus, it is worthy of further studies.