Chinese Herbal Medicine as Adjunctive Therapy to Chemotherapy for Breast Cancer: A Systematic Review and Meta-Analysis

Chinese herbal medicine (CHM) has been increasingly employed during therapy for breast cancer, but its efficacy remains a matter of debate. This systematic review examined randomized controlled trials to provide a critical evaluation of this treatment. The results demonstrated that the combined use of CHM with chemotherapy may improve the immediate tumor response and reduce chemotherapy-associated adverse events. Our findings highlight the poor quality of Chinese studies, and additional well-designed randomized controlled trials addressing the role of CHM are warranted. The lack of molecular-based evidence for CHM and Zheng has resulted in a limited understanding and acceptance of CHM and traditional Chinese medicine in Western countries. We believe that researchers should immediately explore a CHM-based cure, and CHM should be applied to routine care as soon as conclusive data are available.


Introduction
Breast cancer is the leading cause of cancer-related deaths in women worldwide [1]. Breast cancer has been one of the fastest growing cancers in China over the past 30 years, with an incidence approaching approximately 96%, which is only slightly lower than that of lung cancer [2].
Chemotherapy plays a key role in the systemic treatment of breast cancer, and it is the cornerstone of therapy for patients who are not candidates for endocrine therapy [3]. Adjuvant chemotherapy can increase the pathological complete response (CR) rate and improve survival in most patients with early stage breast cancer [4][5][6]. The primary objectives of treatment are palliation and improved survival for the vast majority of metastatic breast cancers, which are incurable.
The chemical agents used in chemotherapy are selectively destructive to malignant tissues, but these agents also damage healthy tissue, which results in adverse side effects that negatively impact compliance with cancer treatment. Therefore, there is a clinical need to find an intervention to manage the adverse side effects of chemotherapy and increase patient tolerance and well-being.
Traditional Chinese medicine (TCM) was developed thousands of years ago, long before the advent of modern  (1) Types of Studies. There were randomized controlled trials (RCTs) with two arms, without blinding or language restrictions.

Search Strategy for EMBASE via OvidSP to Identify Potential Articles
(2) Types of Participants. They were female breast cancer patients who were diagnosed using pathological sections and treated with chemotherapy. (3) Types of Interventions. They included chemotherapy protocols of experimental and control groups that were the same or comparable; CHM was used in experimental groups as adjunctive therapy to chemotherapy; placebo or blank controls were eligible; and other treatment was identical in both groups. (4) Type of Outcome.
The following primary outcomes were included: survival; immediate tumor response, defined as CR or partial response (PR) using the World Health Organization scale; control of nausea and vomiting; and improvement in myelosuppression.
The following secondary outcomes were included: performance status evaluated using the Karnofsky performance score (KPS); immune system response, including percent change in T lymphocytes and natural killer (NK) cell activity; QoL; and other side effects, such as alopecia, chemotherapyrelated cardiotoxicity, and cognitive dysfunction.

Exclusion Criteria.
The exclusion criteria were studies of breast cancer patients with other primary cancer; sample size < 30; CHM used in both groups other than experimental medicine and placebo; studies with > 20% withdrawal and/or dropout rates; nonoriginal studies; or duplicate studies.

Study Selection.
Two authors independently screened the trials by first scanning abstracts, titles, and key words to select potential studies based on the inclusion and exclusion criteria. Full articles of the potential studies were obtained for a final determination. If a disagreement occurred, the two authors reviewed the study again, and a third reviewer resolved any disagreement.

Quality
Assessment. Methodological quality was assessed using the 5-point Jadad scale [20]. The Jadad scale includes three domains: randomization, blinding, and withdrawals and dropouts. Only studies with a Jadad score of 3, 4, or 5 were included.
2.6. Assessment of the Risk of Bias. The risk of bias was assessed using the method recommended by the Cochrane Handbook [21]. This tool is a domain-based evaluation in which critical assessments are made separately for the following concepts: randomization, blinding, outcome reporting, and other issues.

Data Management.
Two authors independently extracted the following variables: article title; author(s); journal title; year of publication; study design; sample size; sampling and diagnostic procedures; loss to follow-up; exclusions and reasons; baseline characteristics of patients (e.g., age, breast cancer stage); intervention characteristics (e.g., chemotherapy drugs, CHM patterns, CHM type, duration, and dosage); outcome(s); and conclusions. A third reviewer reviewed the extracted data and stored the original data in a secure computer to avoid changes.

Statistical Analysis.
We performed a meta-analysis using RevMan 5.2 only when sufficient and suitable data were obtained. We conducted a narrative synthesis when there were too few clinically homogeneous studies for a metaanalysis. We calculated the RR and MD separately for dichotomous and continuous variables. A random effects model was used for pooling because of the clinical heterogeneity of CHM, which includes the complexity of ingredients and different therapeutic methods. We analyzed the pre-and posttreatment data separately to avoid biases from estimations of the values of change from baseline. We assessed heterogeneity using the 2 test, with a value < 0.10 indicating a significant difference. Subgroup analyses were conducted to investigate the source of heterogeneity. Funnel plot asymmetry was used to investigate publication bias when there were at least 10 studies. The total effect was tested using the test, and < 0.05 was used to identify significant effects.

Characteristics of Included
Studies. The initial search identified 2,109 studies published before October 2014 (Figure 1). Review of the titles and abstracts resulted in the inclusion of 243 studies. Of these, 211 studies were excluded after review of the full text because the studies were duplicates and animal experiments, had a small sample size, had a Jadad score < 3, or did not report the clinical stage. Notably, 90 Chinese studies were excluded because of a low Jadad score, which suggests that Chinese studies in this field are of poor quality. A total of 31 studies were finally included in the metaanalysis, of which 28 were retrieved from Chinese databases and 3 from English databases.
Of the total 2,805 patients who were enrolled in these studies, 1,391 received CTC therapy and 1,319 received chemotherapy alone. Ninety-five patients withdrew or dropped out (Table 1).
All trials were described as randomized and had two parallel groups. Concealment of allocation was not reported in any of the Chinese studies. The blinding process was described in the 3 English studies and in only 1 Chinese study. Eight studies reported withdrawals and dropouts. The risk of bias, as assessed using the tool from the Cochrane Collaboration, is shown in Figures 2 and 3.
All recruited patients in the 31 studies were diagnosed using pathological sections, and all studies had a Jadad score of 3, 4, or 5, indicating that the baseline characteristics were comparable among studies (Table 1).

Survival.
One study reported 2-year survival with better survival in the CTC group than in the chemotherapy group, but the difference was not significant (RR = 1.15, 95% CI = 0.86-1.53) [26]. However, the median survival time differed significantly between these groups in another study (WMD = 1.90, 95% CI = 0.77-3.03) [25]. However, neither trial was of high quality.

Decrease in Chemotherapy
Toxicity. Chemotherapyinduced nausea and vomiting (CINV) is one of the most serious and unwanted side effects of chemotherapy [53][54][55]. 1 study was of small size (n < 30); 1 study with high rate of dropout (>20%); 1 study had inauthentic data    between the two arms (WMD = 0.02, 95% CI = −0.29-0.33) [36], and another study reported a significant difference [41]. These two studies were not pooled for analysis because of the different data types.

QoL.
Six studies reported on QoL, but we were not able to pool these results because each study used a different scale and data type. The study by Semiglazov et al. was of high quality and low risk, and the QoL was assessed using 3 FACT-G subscales (physical, emotional, and functional wellbeing) [39]. The intervention group exhibited improvements in the FACT-G total score and in the physical, emotional, and functional well-being scores, and the placebo group had poorer scores. Four studies that reported significantly improved QoL using combined therapy, as assessed using the Chinese version of the FACT-G, were of mixed quality [37,42,50,52]. Two studies reported QoL as assessed using the Chinese version of the EORTC QLQ C-30. Zhong observed improvements in the physical and emotional subscales and the overall health QoL score with CTC therapy [52]. Wang reported significant within-group improvements in some subscales in both groups, but there were no details of comparisons between groups [43].

Other Outcomes.
One high-quality study evaluated CTC therapy for the prevention of chemotherapy-related cognitive dysfunction and reported no significant difference between the two arms [23]. Two studies that reported chemotherapyrelated cardiotoxicity had significant heterogeneity, and CTC Heterogeneity: 2 = 0.01; 2 = 13.27, df = 10 (p = 0.21); I 2 = 25% Test for overall effect: Z = 5.29 (p < 0.00001)

Discussion
Treatment of breast cancer using CHM has been described in Chinese medical texts for more than 2,000 years. Accepting TCM as science rather than myth remains a challenge in Western countries despite the recent increase in the use of CHM. This review has several limitations. First, we did not identify studies in languages other than Chinese and English. CAM use is reportedly high in East Asia, where CHM originated, with use rates of 29-83% in South Korea and 50% in one study in Japan [56,57]. Therefore, additional studies  should be identified from or conducted in these areas to further investigate the efficacy of CHM. Second, all of the included Chinese studies had relatively small sample sizes, ranging from 40 to 101 participants. None of these studies reported the details of sample size calculation. Third, the Chinese trials did not clearly report allocation concealment or blinding, and none of the Chinese studies were placebo controlled or double blinded, which could have resulted in bias and an overestimation of CTC efficacy [58]. Publication bias may also have existed. The asymmetry of the funnel plot may be the result of an insufficient number of trials and significant statistical heterogeneity (Figures 5, 7, 15, and 17). There were also different data types and assessment methods for outcomes, which may have resulted in statistical heterogeneity.
We also cannot ignore the low quality of the included trials; however, that may not be a sound reason to exclude a systematic review. A systematic review embraces the features of  systematization and comprehensiveness, which differentiate it from a normal review. In addition, the CHM used differed significantly among trials. Inevitably, the pharmacological actions of these treatments would not be the same. A random effects model was used for pooling because of the clinical heterogeneity. Because of this limitation, we cannot draw a convincing conclusion. Nevertheless, the problems with the current studies identified in this review are significant, and a great deal of work needs to be done to evaluate the efficacy of CAM using a modern and rigorous methodology. Only 3 studies evaluated the Zheng TCM pattern, which is another key limitation of the included studies [37,42,52]. Zheng, also known as syndrome or pattern, is the core concept in TCM, and it describes the entire physiological and/or pathological pattern of the patient [59]. Zheng is usually evaluated through a comprehensive analysis of clinical signs and symptoms. TCM practitioners collect the signs by inspection, auscultation, olfaction, inquiry, pulse, and palpation. TCM practitioners in clinical practice prescribe CHM based on Zheng.
CHM therapy is more efficacious when based on the correct judgment of the Zheng classification according to the Chinese medical system. One clinical study found that the therapeutic effect of CHM for the treatment of irritable bowel syndrome was more sustainable when based on the TCM pattern than on standard treatment [60]. The key role of Zheng in TCM should not be ignored despite the controversial results reported by other clinical studies, which indicate that the efficacy of Zheng-based treatment is not advantageous over standard treatment [61][62][63]. Patients are not administered the same CHM for a long period of time in real practice, and the treatments reported in clinical trials did not follow a pattern that is commonly used in actual clinical practice because Zheng is dynamic during the treatment course. The biggest challenge in the exploration of Zheng-based CHM therapy using an RCT is the standardization of Zheng. Currently, the process of Zheng is highly subjective, and a nationwide and objective process is needed to improve its use. Randomized, multicenter trials should be conducted for this purpose. Analyses of Zheng at the molecular level may also enable acceptance of TCM on a scientific basis for the West. Table 2 lists the herbal medicines that were commonly used for the treatment of breast cancer in the identified studies in this review. The pooled data in this review demonstrated that the adjunctive use of CHM with chemotherapy may improve immediate tumor response and performance status and reduce the occurrence of adverse events associated with chemotherapy. We were unable to verify whether CHM helped stimulate the immune system, as measured using CD3 + , CD4 + , CD8 + , and CD4 + /CD8 + cells, because of the mixed quality and significance of the included studies. The evidence is too limited to make any confident conclusions.  Events  23  39  31  25  23  26  25  28  38  48  26  21  25   378   Total  30  44  37  30  30  30  27  30  39  51  30  24  31   433   Events  15  31  18  18  15  18  18  16  24  40  16  13  15   257   Total  30  44  29  30  30  30  27  27  39  49  30  Heterogeneity: 2 = 0.00; 2 = 11.05, df = 12 (p = 0.52); I 2 = 0% These results suggest that combined therapy has potential benefits for breast cancer patients. The finding of CHM efficacy as an adjunctive therapy for breast cancer is similar to the findings of other reviews for hepatocellular carcinoma, nonsmall-cell lung cancer, colorectal cancer, and nasopharyngeal carcinoma [58,[64][65][66]. A recent systematic review involving 8 RCTs showed that CHM combined with conventional therapy for breast cancer was efficacious in improving QoL and decreasing hot flashes, but this study did not identify as many clinical trials as it could have [67]. Breast cancer patients undergoing chemotherapy and/or endocrine therapy were included in that review, and the effect of CHM for breast cancer should have been examined separately in those two groups. In addition, the review focused on the effects on QoL and hot flashes but did not evaluate other cancerrelated symptoms. Finally, the reviewers only presented a narrative synthesis without a meta-analysis, which made the conclusion unconvincing. GRADE should be applied to judge the evidence and make recommendations regarding the application of CHM in the treatment of breast cancer. The present study suggests that recommendations for CHM combined with chemotherapy could be made for breast cancer, but TCM may be too complex to be immediately adopted by physicians in Western countries. The most fundamental and often-overlooked challenge is the lack of a 1 : 1 correlation between modern allopathic and Chinese holistic medical approaches [68]. We cannot make specific recommendations despite the rapid increase in the use of CHM and reported potential benefits because of the complexity of this system and the variable data.
Current evidence on the use of CHM as an adjunctive treatment with chemotherapy for breast cancer remains equivocal. Our findings highlight the poor quality of Chinese   Heterogeneity: 2 = 1.79; 2 = 2.21, df = 1 (p = 0.14); I 2 = 55% Test for overall effect: Z = 1.81 (p = 0.07)   studies, and additional well-designed RCTs addressing the role of CHM are warranted. The lack of molecular-based evidence for CHM and Zheng has resulted in a limited understanding and acceptance of CHM and TCM in Western countries. We believe that researchers should immediately explore a CHM-based cure, and CHM should be applied to routine care as soon as conclusive data are available [69]. For researchers devoted to the promotion of TCM or CHM, numerous barriers need to be addressed, including the standardizations of the Zheng classification and herbal agents, appropriate study designs, and the identification of the mechanisms of CHM at the molecular level.  Test for subgroup differences: 2 = 115.59, df = 2 (p < 0.00001); I 2 = 98.3%