A Systematic Review of Randomized Controlled Trials on Oral Chinese Herbal Medicine for Prostate Cancer

Background Prostate cancer is the most common malignant tumor associated with male reproductive system. Objective The existing eligible randomized controlled trials (RCTs) were critically appraised for the safety and effectiveness of CHM for prostate cancer. Methods A literature search was conducted by using PubMed, CENTRAL, CNKI, CBM, VIP and Wanfang databases until August 2015. RCTs of CHM or CHM plus conventional medicine for prostate cancer patients were included. The primary outcomes appraised were survival time, time to progression and quality of life. The risk of bias assessment according to the Cochrane Handbook was used to evaluate the methodological quality of the included trials. Revman 5.3 software was used for data analyses. Risk ratio and mean difference (MD) with a 95% confidence interval (CI) were used as effect measures. Meta-analysis was to be used if sufficient trials without obvious clinical or statistical heterogeneity were available. Results A total of 17 RCTs involving 1224 participants were analyzed. One trial was about CHM comparing to no treatment. The remaining 16 trials used CHMs as adjunctive treatment for endocrine therapy. Due to the poor quality of methodologies of most trials, only limited evidence showed that a combination of CHM and endocrine therapy might be more effective in restraining the development of the disease (MD 10.37 months, 95%CI 9.10 to 11.63 months), increasing patients’ survival time (7–15 months) or improving patients’ performance status, when compared to endocrine therapy alone (Karnofsky performance scale average changed 15 scores between groups). No severe adverse event was reported related to CHM. Conclusion Due to the insufficient quality of trials that were analyzed, it is not appropriate to recommend any kind of CHMs in treating prostate cancer at the present time. Well-designed trials with high methodological quality are needed to validate the effect of CHMs for patients with prostate cancer.


Inclusion criteria
RCTs which compared oral administration of CHM (alone or in combination with other treatments) to those without CHM in treating prostate cancer were included in our analysis. The primary outcomes included survival status (such as survival time and survival rate), quality of life, and time to progression (TTP); secondary outcomes included measurements of PSA, patient performance status, prostate volume (PV), serum testosterone, and adverse events. At least one of the above outcomes should be reported in the included trials in our evaluation. Above strategies were adopted for each specific database, and Chinese characters for relevant key words were used when searching Chinese databases.

Study selection
Two authors (Cao H and Mu Y) independently screened the titles and abstracts of the achieved citations from primary searching. Full text of the articles of potential interest were download for further evaluation, those meeting inclusion criteria were included in the final review. Disagreements between the two authors were resolved by discussion and if needed, arbitrated by a third author (Liu JP).

Data extraction and quality assessment
Two authors (Wang Y and Mu Y) extracted the data from the included trials independently on patient characteristics, intervention/control details, clinical outcomes and quality-related information. The discrepancies were resolved through consensus. Missing data were achieved through contacting with authors of the original studies.
Selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other biases (determined according to sample size calculation method, inclusion/exclusion of criteria for patients' recruitment, comparability of baseline data, funding sources, and any other potential methodological flaw that might have influenced the overall assessment) were assessed according to the criteria from the Cochrane Handbook for Systematic Reviews of Intervention [9]. Three potential bias judgments: low risk, high risk, and unclear risk, were determined for each single trial during assessment. A judgment of low risk was made when all the seven items met the criteria as "low risk", a judgment of high risk of bias was made when at least one of the seven items was assessed as "high risk".

Data analysis
Data were summarized using risk ratio (RR) with 95% confidence intervals (CI) for binary outcomes or mean difference (MD) with 95% CI for continuous outcomes. We used Revman 5.3 software from the Cochrane Collaboration for data analyses. Meta-analysis was used if the trials had an acceptable homogeneity on study design, participants, interventions, control, and outcome measures. Statistical heterogeneity was tested by examining I 2 [10], meaning that an I 2 larger than 50% indicated the possibility of statistical heterogeneity. Both fixed effect model and random effect model were used if there was possibility of statistical heterogeneity among trials. Publication bias was explored by funnel plot analysis. Subgroup analyses were conducted to determine the evidence for the different treatment duration if data were sufficient. Sensitivity analyses were used in order to determine whether the conclusions differed if eligibility was restricted to studies without high risk of bias; or if a fixed/random effect model had been applied.

Results of the search
The search of 6 databases identified 1657 citations for further evaluation. Full texts of 109 articles were read, and 18 [11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28] trials met our inclusion criteria. However, one of the trial [28] tested a herbal capsule PC-SPES (960mg three times daily), which contains Ganoderma lucidum, Scutellaria baicalensis, Rabdosia rubescens, Isatis indigotica, Dendranthema morifolium, Seronoa repens, Panax pseudoginseng, and Glycyrrhiza uralensis. This herbal product was reported to be contaminated with diethylstilbestrol [29], thus the product may have a similar effect with diethylstilbestrol on inhibiting growth of both androgen-sensitive and androgenindependent prostate cancer cells. Due to the massive controversy of potential contamination of PC-SPES, the drug is no longer available in the U.S. Since it is hard to define whether it still belongs to herbal product, we excluded the trial from this review. Thus, finally 17 trials were included in the review. Details of the study flow were shown in Fig 1. All of the 17 included trials were conducted in China and published in Chinese. Except for one four-arm trial [16], all the remaining trials were parallel two-arm trials. Characteristics of included trials were listed in Table 1.

Narrative of the included studies
Totally 1224 patients were included in this review, with an average of 72 patients per trial. Average age of those participants was 70 years old. The stages of the prostate cancer for majority of the patients were determined as III/IV, which mean the tumor has been infiltrated or beyond the prostate capsule.

Methodology quality assessment
According to the previously mentioned criteria, all included trials were assessed as having high risk of bias (Fig 2).
Only seven trials [14, 15, 17-19, 21, 23] reported the methods of randomization, all of them used a random number table, with one [15] employing statistical software to generate random numbers. Two among the seven trials [14,18] used envelopes to contain the random numbers, and the remaining five trials did not mentioned whether or not they had adequate allocation concealment methods. Sixteen of the included trials used CHM as add-on treatment for endocrine therapy (or combined with chemotherapy); however, placebo was not used in any of them. The remaining trial [14] compared CHM with no treatment without placebo control. Thus, all of the included trials were assessed as high risk of bias on the control factor of blinding to participants. Insufficient information was provided to judge the performance bias induced by lack of blinding to statistician and outcome assessors.
Only three included trials [15,17,23] were assessed as low risk of bias in incomplete outcome data, since they reported the numbers and reason for missing data in each group, and used adequate statistical methods (the last observation carried forward analysis) to deal with the missing data. Another two trials [13,14] only analyzed the data from per-protocol set, and did not interpret results regarding the impact of missing data and reason for drop-out; thus, we evaluated these two trials as having high risk of attrition bias.
Protocols of thirteen of the included trials could not be found to assess the risk of reporting bias of the studies. Three [14,19,20] of the remaining four trials were assessed as low risk of reporting bias, and the other one [24] was evaluated as high risk of reporting bias due to the incomplete reporting of pre-defined outcomes.
No funding issue or methods of sample size calculation were apparent in all of the included trials, and it was difficult to determine whether a study was fraudulent. We evaluated all of the trials as at an unclear risk of other bias.

Estimate effects of CHM for prostate cancer patients
Primary outcome measurements. Survival time/rate. Three trials [12,15,22] reported survival rates of participants with up to 8 years follow up duration. One trial [12] found no difference between CHM plus castration and castration alone on increasing survival rate (RR 1.08, 95%CI 0.86 to 1.36, 60 participants). The second trial [22] also showed no difference between CHM plus MAB and MAB alone on increasing survival rate (RR 1.04, 95%CI 0.92 to 1.18, 295 participants). The remaining trial [15] reported a small effect of CHM as an add-on treatment on increasing survival rate (RR 1.51, 95%CI 1.05 to 2.18, 59 participants), compared with MAB plus chemotherapy.
Two trials compared CHM plus MAB (and chemo-therapy) versus MAB (and chemo-therapy) alone and reported survival time of participants. However, one of them [27] did not report sufficient data for further analysis, which showed median survival time for CHM group was 25.2 months, and an obvious difference (p<0.01) was found between groups (18.9 months for MAB group). The other trial [15] showed significant difference between groups on this outcome (MD 15.87 months, 95%CI 11.90 to 19.84 months, 59 participants).
All of the three trials [11,21,25] which mentioned the scores of FACT-P (pre and post treatment)in each subscale, showed that the herbal product (Yangyin Yishen Decoction, Qianlie Xiaozheng Decoction, and 1st Qianliexianai Decoction) as an adjunctive therapy had a better effect on improving patients' quality of life compared with MAB alone (p<0.05). Similar results were found in two other trials [13,17] for comparison of a combination of herbal medicine and MAB versus MAB alone when assessing patients' quality of life through QLQ-C30 scale. 75% of patients in combination group (12 participants) got more than 6 scores changed after treatment, but in MAB group only 42.9% patients (6 participants) reported a similar improvement.
The remaining one trial [12] did not clarify which scales it employed for quality of life assessment; however, the result showed the combination of Yiqi Jiedu Quyu Decoction and MAB were better than MAB alone on improving patients' quality of life (MD -16.00, 95%CI -19.11 to -12.89, 44 participants).
The progress of the disease. Only two trials reported time to progression (TTP, months) as an outcome measurement. One of them [16] was a four-arm trial which assessed the adjunctive effect of CHM for MAB or combination of MAB and chemo-therapy. Another trial [15] only compared CHM plus MAB and chemo-therapy versus MAB and chemo-therapy alone. The results showed a large effect of CHM as an add-on treatment for MAB (MD 6.60 months, 95% CI 5.50 to 7.70 months, 1 trial, 25 participants) or MAB plus chemo-therapy (MD 10.37 months, 95%CI 9.10 to 11.63 months, 2 trials, 84 participants, I 2 = 0%, fixed-effect model) in restraining the development of the disease (including the radiographic progression and the progression to metastatic disease).
Secondary outcome measurements. PSA. Fourteen out of seventeen included trials reported PSA (ug/l or ng/ml) pre and post treatment; baseline PSA levels among trials were different and the treatment duration varied from 3 months to 15 months. Meta analysis was not conducted due to the obvious clinical and statistical heterogeneity among trials. One trial [12] showed a better effect of castration (prostatectomy, bilateral orcheictomy and diethylstilbestrol) than CHM (Juzao Pills) plus castration on reducing PSA (MD 4.13ug/l, 95%CI 2.37 to 5.89, 60 participants). Eight trials [11, 14, 18-21, 23, 25] showed a benefit of CHM for this outcome (at least 2.38 ug/l PSA lower in CHM group), and the remaining five trials [13,17,24,26,27] showed no difference between groups on controlling serum PSA in prostate cancer patients. Details of the results from each single trial were listed in Table 3.
Performance status. Within the five trials which reported KPS scores, three of them [15,18,19] showed a superior effect of herbal product as add-on therapy compared to MAB (MD 10.00, 95%CI 7.19 to 12.81, 60 participants), MAB plus chemotherapy (MD 19.12, 95%CI 15.27 to 22.97, 59 participants), or other endocrine therapy alone (MD 12.00, 95%CI 11.01 to 12.99, 90 participants) on improving patients' performance status. One trial [23] found a similar number of patients whose KPS scores increased more than 10 between herbal medicine (Fuzheng Huayu Decoction) plus MAB group and MAB single applied group (RR 1.55, 95%CI 0.84 to 2.87, 25 participants). The remaining trial [20] only reported the KPS post treatment in both groups (52.79 in Qianlie Yuai Decoction plus control group, 50.54 in anti-androgen mono-therapy group) without the standard deviation or 95%CI of this outcome, thus we failed to interpret the data.
Serum testosterone. Only one trial [19] reported the change of serum testosterone after treatment, the result showed no difference between a combination of herbal decoction and MAB (Goserelin acetate sustained-release depot treatment plus Bicalutamide oral administration) and MAB alone for this outcome (MD -0.10 ng/ml, 95%CI -0.21 to 0.01, 60 participants) after six months of the treatment.
Prostate Volume (PV). Two trials reported the patients' PV before and after treatment; both of them showed that the average PV of the participants before the treatment was over 50 ml. However, one trial [12] found a significant benefit of herbal medicine (Juzao Pills) as addon therapy for castration (Prostatectomy, bilateral orchiectomy and diethylstilbestrol) on reducing the volume of prostate (MD -5.20ml, 95%CI -5.67 ml to -4.73 ml, 60 participants), but the other trial [13] showed no difference between MAB therapies with or without herbal decoction administration (MD -2.61ml, 95%CI -6.62 ml to 1.40 ml, 30 participants) on PV after treatment.
Adverse events. Five trials mentioned adverse events in the herbal medicine group. Four of them [12,[14][15][16] reported there was no adverse event during the treatment duration. The remaining one trial [23] reported mild or moderate adverse events in Fuzheng Huayu Decoction groups. Symptoms of blood vessel constriction occurred in 31% of patients (9 cases), post-castration syndrome occurred in 34.5% of patients (10 cases), 27.6% of patients were reported to have anemia (8 cases), and 3.4% of patients had liver function damage (1 case) in combination of Fuzheng Huayu Decoction and MAB group. However, compared with control group (MAB), the combination group still showed lower incidence rate of adverse events (p<0.05). Details of reporting of adverse events were shown in Table 2.
Assessment for publication bias. Since none of the meta-analysis in this review included more than 10 studies, funnel plot was not conducted to assess the potential publication bias.

Summary of findings
With concerns on 16 of the 17 trials, we have no confidence on determining the add-on effect of oral administration of herbal medicine for increasing survival rate and improving main symptoms (including PSA and PV) of patients with prostate cancer. Due to the poor methodological quality of the included trials and the insufficient number of trial participants, only limited evidence showed combination of herbal medicine and MAB may have better effect on increasing patients' survival time (7-15 months) or improving patients' performance status compared with MAB alone (KPS scores average changed 15 between groups). The remaining trial [14] found an herbal decoction, Chuanlong Yiai Decoction, may have an effect on reducing PSA (MD -51.07ug/l) of patients with prostate cancer compared with no treatment. However, a conclusion could not be drawn due to the small sample size and potential high risk of bias in the trial.
Safety of herbal medicine was rarely reported in the included trials; thus, we could not make any confirmed conclusions on this outcome.

Agreement and disagreement with other studies
We found one review [30] assessed the complementary and alternative medicine (CAM) for prostate cancer, in which herbal preparations were involved. However, only PC-SPES and Zyflamend (an anti-inflammatory proprietary blend of 10 standardized herbal extracts including turmeric, holy basil, green tea, huzhang, ginger, Chinese goldthread, oregano, Scutellaria baicalensis, and barberry) were mentioned in that review, and no data from a complete clinical trial could be used for evaluating those herbal preparations. The author demonstrated that the available data provided insufficient evidence to support the use of CAMs (including herbal preparations) for the treatment of prostate cancer patients.
Compared to the above review, our study included 17 trials with different types of herbal preparations. Though the strength of the evidence was low, we found potential effect of oral herbal preparations for patients with prostate cancer on some key symptoms' improvement.

Limitations of the review
Low levels of evidence in this review were mainly caused by the poor quality and small sample size of original included trials. The inconsistency of findings of herbal medicine's effect on improving main symptoms of prostate cancer among these trials further reduced the internal validity of the evidence.
Although we searched both Chinese and English databases, all of the included trials were retrieved from Chinese literature, which may have introduced potential selection bias and limited the external generalization of the evidence.
Types of clinical stages of prostate cancer determine the details of treatment. In this review, almost half of the included trials did not report or classify clinical stages of the participants, which made the subgroup analysis regarding to the different clinical stages of the disease unavailable.

Implications for the clinical practice
First aim of the treatment for patients with prostate cancer is to strengthen their self immunefunction to alleviate the symptoms and get better quality of life. According to this review, herbs or herbal prescriptions with the function of strengthening the spleen and kidney "qi" are commonly used to achieve this objective. Top five of most frequently used herbs were Astragalus mongholicus (in 12 trials), Rhizoma atractylodis macrocephalae (in 9 trials), Codonopsis pilosula (in 9 trials), Herbah edyotis (in 9 trials), and Poriacocos (in 7 trials). The first three and Poriacocs were commonly used for tonifying (taking a tonic or nourishing food to build up) one's qi and enhancing body immunity. Herba Hedyotis was reported to be used in several studies [31][32][33], which showed effects on enhancing specific or non-specific immunity and was recommended to be provided to cancer patients.
Herbal medicine may have effects on alleviating the symptoms or reducing the incidence rate of adverse events caused by endocrine therapy; however, insufficient evidence was seen supporting this conclusion.

Implications for the future research
First of all, the protocol of the clinical trial should be pre-defined according to the research question. Ideally, preclinical data should be provided to support the design of the clinical trials. Authors are encouraged to register their study protocols before trial implementation to ensure the research can be conducted according to the pre-defined standard. In the protocol, the primary and secondary outcomes should be defined for the research. While one can argue whether PSA is a good marker for prostate cancer, it is the only approved marker to indicate the development of prostate cancer, and testing of serum PSA has been standardized. Therefore, PSA values should be included as a parameter to evaluate the outcome of prostate cancer trials. If possible, PSA doubling time during treatments should be determined. A documentation of a decrease of PSA could be more informative than a single measurement of PSA value. Secondly, there should be unified scales for assessing the quality of life of patients with prostate cancer (such as EORTC QLQ-PR25 and FACT-P). Therefore, those scales should be used and interpreted appropriately. Furthermore, in order to observe the long-term efficacy of treatment with assessment of patients' survival time or time to progression, follow-up with adequate duration should be considered during clinical study.
According to this review, for outcomes like patients' survival time or change of PSA after treatment, a combination of herbal medicine and endocrine therapy did not show advantages compared with endocrine therapies themselves. Thus, the sample size of future studies should be calculated carefully to make sure that data could achieve enough power for statistical analysis. Meanwhile, factors that may influence or downgrade the methodological quality of studies should be controlled during the research. Adequate methods are requested for random number generation, allocation concealment, missing data handling, avoiding performance bias and other bias. Reporting of the clinical study should follow the international standard or criteria, such as CONSORT statement [34].
In addition to treatment trials, prostate cancer prevention trials involving CHM should not be overlooked. Although it is not easy to carry out primary prevention trials, it is possible that the use of CHM could prevent or delay of the recurrence of prostate cancer. One of the authors (Shiuan Chen) has recently completed a trial to determine the effects of white button mushrooms (WBM) against biochemically recurrent prostate cancer [35]. Therapy with WBM appears to both impact PSA levels and modulate the biology of biochemically recurrent prostate cancer by decreasing immune-suppressive factors, such as myeloid-derived suppressor cells (MDSCs). This could be one way to evaluate anti-prostate cancer properties of CHM.
CHM has been applied in China for more than 2,000 years and shown to be useful in many ways. CHM is complex in nature and used mainly for symptom management. Our analysis was not able to reach a clear conclusion mainly due to inconsistency of the trial design, variable outcome measurements, and the complexity of CHM regimens. Furthermore, patient population might not be well controlled. It is also possible that during the trials, participants also took various supplements without telling the doctors. Through a careful and well-designed trial with a definitive outcome measurement, it is more likely that conclusions addressing the usage of CHM in prostate cancer treatment will be generated.

Conclusions
Due to the insufficient quality of trials that were analyzed, it is not appropriate to recommend any kind of CHMs in treating prostate cancer at the present time. Well designed trials with high methodological quality are needed to validate the effect of CHMs for patients with prostate cancer.
Supporting Information S1