Oral Chinese Herbal Medicine as an Adjuvant Treatment for Chemotherapy, or Radiotherapy, Induced Myelosuppression: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

Objective Myelosuppression is a common side effect in cancer patients receiving chemotherapy or radiotherapy. Chinese herbal medicine (CHM) has shown promise in alleviating myelosuppression. Method We searched for randomized controlled trials (RCTs) from seven databases without language restriction. We included RCTs in adults, in which hematological toxicity was measured according to WHO criteria and control group underwent chemotherapy and/or radiotherapy and the treatment group was given oral CHM. Results We searched 1021 articles from the date of databases inception to October 7, 2016. We selected 14 articles for the final analysis. Pooled data showed that CHM significantly decreased the suppression rate of leukocytes, neutrophils, hemoglobin, and platelets compared with the control group, particularly in grade III-IV toxicity (leukocytes: RR = 0.43, 95% CI = 0.33–0.56; neutrophils: RR = 0.39, 95% CI = 0.27–0.58; hemoglobin: RR = 0.33, 95% CI = 0.18–0.61; platelets: RR = 0.61, 95% CI = 0.39–0.95). Conclusions CHM as an adjuvant can alleviate myelosuppression induced by chemotherapy or radiotherapy, reduce grade III-IV toxicity, and maintain therapeutic dose and treatment cycle. However, due to heterogeneity and publication bias, the results should be interpreted with caution and validated by conducting strictly designed multicenter RCTs of high quality and large scale.


Introduction
Myelosuppression, also known as bone marrow (BM) suppression or myelotoxicity, is the major side effect of chemotherapy and radiotherapy. Blood originates in the BM [1] and blood cells have limited life [2]. Once myelosuppression is induced, BM is unable to maintain normal levels of blood cells, which results in a series of complications, such as anemia, infection, and tendency of hemorrhage. Moreover, chemotherapy induces dose-limiting myelosuppression toxicity, but dose reductions would affect treatment efficacy. Therefore, myelosuppression should be addressed while considering chemotherapy or radiotherapy.
It is vital to improve survival rate of cancer patients being treated for anemia, infection, and other complications. However, some treatments could cause adverse effects. For example, G-CSF causes shock or chronic fibrous pneumonia and repetitive platelet transfusion leads to formation of alloantibody [3]. Therefore, it is crucial to find safe agents that can reduce myelosuppression and improve survival rate of patients receiving chemotherapy or radiotherapy. In recent years, Chinese herbal medicine (CHM) has shown promise in this regard.
Studies on CHM preventing myelosuppression have been inconclusive or conflicting. Currently, it is difficult to make the clinical decision of administering CHM as an adjuvant in chemotherapy or radiotherapy. A similar systematic review was published [4], which revealed effects of various CHMs in myelosuppression prevention. To provide better insights for making appropriate clinical decisions, we aimed to review the effects of oral CHM as we believe that the effects vary with the route of administration.  ≥11.0 g/100 ml 9.5-10.9 g/100 ml 8.0-9.4 g/100 ml 6.5-7.9 g/100 ml <6.5 g/100 ml In this study, our objective was to minimize clinical heterogeneity and pool high-quality studies to generate robust evidence regarding the potential therapeutic value of CHM in preventing myelosuppression. This meta-analysis was reported and performed in accordance with PRISMA guidelines (S1 File in Supplementary Material available online at https://doi.org/10.1155/2017/3432750).

Search
Gray articles were identified by searching Chinese Clinical Trial Register (ChiCTR), the American Society of Clinical Oncology (ASCO), and Google. In addition, we manually searched reference lists of relevant articles.

Study Selection and Data
Extraction. Studies were selected by two researchers (BNH, RL). First, titles were read and irrelevant or duplicate articles were removed. Second, editorials, opinions, or case reports were excluded through abstract screening. Third, after reading the full text, on the basis of inclusion and exclusion criteria, articles were filtered. BNH and RL selected articles independently and any disagreement was resolved by discussion. If they still disagreed, the third reviewer (SQH) intervened to resolve the disagreement.
Data were extracted by two researchers (DL, ZQ) separately, including authors, years, cancer type, number of patients (including number of dropouts or withdrawals), intervention, outcome, detailed content on CHM, and Jadad score. The extracted data were verified by the third author (QW). Disagreements were resolved by face-to-face discussion among the three authors. Studies included in quantitative synthesis (meta-analysis) (n = 14)

Statistical
Analysis. The statistical analysis was performed using Stata 14.0. For dichotomous outcomes, relative risk (RR) and 95% confidence interval (CI) were used. The model of random effects or fixed effects was applied according to heterogeneity of pooled data. Heterogeneity was assessed by Cochran's Q and the 2 statistics [22]. In random effects model, 2 of ≥50% or value of ≤0.10 indicated significant heterogeneity, whereas, in fixed effects model, 2 of ≤50% and value of ≥0.10 indicated homogeneity. If heterogeneity was significant, we performed sensitivity analysis to identify the possible sources.

Characteristics of Articles Included.
In total, 1275 patients (14 articles) were considered, 65 dropped out or withdrew [5-7, 9, 11-18], 71 lacked myelosuppression data [8], and 62 took CHM at different time points [18]. Finally, 1077 patients were included. Table 2 shows the characteristics of the articles included. Cancers included ovarian cancer (3 articles), breast cancer (3 articles), throat cancer (1 article), non-small-cell lung cancer (6 articles), and colorectal cancer (1 article). All articles were graded at least 3 on Jadad scale and one was graded 5. CHM was administered to patients twice a day during chemotherapy and radiotherapy.

Publication Bias and Sensitivity
Analysis. There were 4 meta-analyses including no less than 10 articles; of them, one showed the presence of a publication bias ( Table 3). The remaining articles had no evidence to indicate the presence of potential publication bias.
We analyzed sensitivity by changing the analysis model ( Table 4). The RR of pooling data was robust.

Discussion
We performed meta-analysis of CHM as an adjuvant treatment for chemotherapy-or radiotherapy-induced myelosuppression. The results suggested that oral CHM is effective in protecting WBC, neutrophils, HB, and PLT. Begg, Egger, and Harbord tests revealed the presence of publication bias in WBC (toxic grades I-IV) analysis. The sensitivity analysis indicated that all analyses were robust with no evident publication bias.
In this meta-analysis, the curative effect for grades III-IV was better than that for I-IV. The myelosuppression rate at toxic grades I-II showed no significant difference among the CHM treatment, no-CHM treatment, and placebo treatment groups (supplementary document). Result suggested that toxic grades I-II were responsible for the difference between toxic grades III-IV and toxic grades I-IV. We further analyzed the results for toxic grades I-II. In general, toxic grades III-IV were the index for the clinical treatment of patients showing myelosuppression, in whom treatment for toxic grade II or lower mainly focuses on symptomatic therapy. These results were similar to those in published articles about treating myelosuppression inpatients mostly at toxic grades III-IV. Source data from the 14 articles was insufficient to confirm our hypothesis about toxic grades I-II. We did not analyze toxic grades I, II, III, IV, and II-IV, as they provided no extra information.
The main finding of our meta-analysis was different from that of the previous one [4], which indicated that CHM significantly protected peripheral WBC, but not peripheral HB or PLT. We found that CHM can protect not only peripheral WBC, but also HB, PLT, and neutrophils. The previous study analyzed continuous data from 8 articles but did not pool data for neutrophils. Our meta-analysis pooled ranked data for WBC, neutrophils, HB, and PLT from 14 articles. Moreover, the referenced articles were different. We did not include the articles in the previous meta-analysis, as seven of eight articles did not meet our inclusion criteria Evidence-Based Complementary and Alternative Medicine 5  and the remaining one was of low methodological quality (Jadad scale < 3 score). In addition, different data style may induce different results. Ranked data reflects the curative effect of treating various patients under serious conditions, rather than a simple summary that would conceal effective treatment, such as continuous data.
To obtain robust results in the present meta-analysis, we included high-quality RCTs using Jadad scale to screen     the methodological quality of articles and excluded studies that treated patients with acupuncture, intravenous CHM, CHM granules, patented CHM drug, or CHM extract, to reduce clinical heterogeneity among articles for comparison with previous meta-analyses. According to the theory for TCM, the herbs used in the 14 articles belong to yiqijiedu (tonifying qi and disintoxication) treatment. Our metaanalysis only included oral CHM; the results showed low heterogeneity (WBC, HB, and neutrophils at toxic grades I-IV) and homogeneity, which suggested that the yiqi-jiedu treatment could be the same treatment. To validate our hypothesis, we performed subgroup analysis of WBC at toxic grades III-IV (only WBC at toxic grades III-IV was included in all 14 articles). The 14 articles were divided by the herb into 4 subgroups, which means that yiqi-jiedu treatment was also subdivided into 4 treatments. The results were similar. Moreover, sensitivity analysis based on change of analysis model did not fundamentally alter most of the pooled results. In Asia, CHM has long been used in the treatment of chronic diseases and cancer as a primary therapy or adjuvant therapy. Nevertheless, one may assume that CHM played the role of placebo. Two of the 14 articles reported that the control group received placebo [5,13], which was indistinguishable in terms of taste and appearance from the study medication. There was no significant placebo effect. Furthermore, several high-quality articles suggested that CHM was not a potential placebo [26,27].
Some articles indicated that there was a positive interaction between the administered CHM and concomitant chemotherapy or radiotherapy [28,29]. The administration of CHM along with chemotherapy may offer some immediate benefits to patients such as improvement of tumor response and quality of life. However, in myelosuppression, there was a negative interaction between the two therapies. The therapeutic mechanism of CHM as an adjuvant treatment for myelosuppression remains unclear. The dominant opinion about the effectiveness of CHM is that it alleviates myelosuppression by multitarget treatments [30,31]. Neutropenia is a common dose-limiting toxicity that is difficult to treat. Our meta-analysis indicated that oral CHM as an adjuvant can prevent the decrease in neutrophils during chemotherapy and radiotherapy, suggesting that CHM maintains the therapeutic dose and treatment cycle. Furthermore, CHM can prevent the decrease in WBC, HB, and PLT. Compared with the other treatments, the advantage of CHM is that it evidently improves the curative effect of chemotherapy and radiotherapy.
Our meta-analysis has some limitations. First, CHM treatment is an individualized therapy. According to the theory for TCM, the difference in individualized therapy should be based on different diseases or symptoms. In our meta-analysis, the treatments in 14 articles belonged to yiqi-jiedu. Furthermore, our results show low heterogeneity and homogeneity and the results of subgroup analysis are similar. Although clinical heterogeneity is low, division into 4 subgroups could improve the robustness of the results. However, as the present research is limited, it was difficult to perform further analysis. Second, our meta-analysis did not include results for lymphocytes. Neutrophils: lymphocyte ratio is also an important hematological toxicity predictive index. However, there are no relevant articles meeting our inclusion criteria. Third, we included 5 types of cancers, with different drug sensitivity. Fourth, only two articles mention the concrete blind method. Fifth, we could not assess the impact of CHM treatment on other clinically meaningful endpoints, such as overall survival and quality of life, due to the limited number of reports available.
Future studies could focus on the following points. First, drugs for tonifying qi and disintoxication play an important role in the treatment. Therefore, these types of drugs should be studied, as further subdividing treatment could provide more robust results, new therapy ideas, or drugs against myelosuppression. Second, better blind methods should be designed to improve methodological quality. Third, further studies can focus on other clinical endpoints, such as quality of life, survival, and length of stay, as there is a lack of highquality studies demonstrating the correlation between CHM and these endpoints.

Conclusions
Evidence obtained from this study suggests that CHM can be used as an adjuvant to alleviate myelosuppression induced by chemotherapy or radiotherapy and, in particular, reduce grade III-IV toxicity. However, due to heterogeneity and publication bias in the partial results of this meta-analysis (WBC at toxic grades I-IV, neutrophils at toxic grades I-IV), the results should be interpreted with caution and validated by conducting strictly designed multicenter RCTs of high quality and large scale.

Disclosure
Bonan Hou and Rui Liu are co-first authors.