Hepatocellular carcinoma (HCC) is one of the most common malignant tumours. Its incidence has increased significantly in recent years. The latest statistics show that there are approximately 840,000 new cases of HCC each year worldwide. In addition, HCC is the third leading cause of cancer-related death. In 2018 alone, 780,000 patients worldwide died of HCC [
The onset of HCC is subtle, and there is no practical method for diagnosing HCC at the early stage. Although diagnostic methods and public awareness of the disease have improved in recent years, most patients have reached the late stage when they are diagnosed; thus, surgery is not an option for these patients, and their prognosis is poor [
Although chemotherapy has limited efficacy in advanced HCC, it is still a treatment option in patients who cannot receive transcatheter arterial chemoembolization (TACE) treatment [
China has a high incidence of HCC. The number of new cases and deaths each year is 394,770 and 383,203, respectively, accounting for half of the global number [
Traditional Chinese medicine injection (TCMI) is a convenient and effective new method of administering Chinese medicine. The latest Chinese Society of Clinical Oncology (CSCO) guidelines for the diagnosis and treatment of HCC recommend a variety of TCMIs for the treatment of late-stage HCC [
One approach to create a systematic summary is a network meta-analysis, which is a technique that combines direct evidence and indirect evidence to compare multiple treatment options. This study aims to make various comparisons between a range of TCMIs combined with systemic chemotherapy and systemic chemotherapy alone in the treatment of patients with HCC by using data from available RCTs in a network meta-analysis.
The inclusion criteria include the following: ① research subjects: patients with HCC whose diagnoses were confirmed by imaging (B-ultrasound, CT, or MRI) and alpha-fetoprotein examination or by pathological biopsy; ② intervention measures: TCMIs combined with systemic chemotherapy (intravenous or oral chemotherapy drugs) compared with systemic chemotherapy alone; ③ study type: randomized controlled trial; ④ end point event: objective response rate (abbreviated as ORR, calculated by (CR + PR)/total number) and clinical benefit rate (abbreviated as CBR, calculated by (CR + PR + SD)/total number), both evaluated by RECIST criteria.
The inclusion criteria include the following: ① studies without a precise chemotherapy regimen; ② studies with apparent data errors; and ③ studies without sufficient available data.
A combination of MeSH words and free words was used to develop a search strategy based on the Cochrane Handbook for Systematic Review of Interventions (Version 5.1.0) [
Two researchers independently read the context of the literature, excluded studies that did not meet the inclusion criteria, and cross-checked the results of the included trials. Disagreement was resolved through discussions or a third investigator. The extracted data mainly included the basic information included in the study, the characteristics of the study object, intervention measures, outcomes, and other information. The two researchers independently extracted data according to a predesigned form and evaluated the quality. Repeated reports were combined into one. The bias risk and the quality of RCTs were evaluated according to the “bias risk assessment” tool recommended by Cochrane Handbook 5.1.
The network meta-analysis used the network package in Stata to analyse data and draw a map of the evidence to summarize the direct and indirect comparisons. If the consistency test showed no inconsistency in the comparison of curative effects among intervention methods, the results were analysed by the consistency model. Otherwise, an inconsistency model was used. The maximum likelihood method was used to select the most stable model in multiple comparisons. After 5000 iterations, the final order of the recommended intervention measures was obtained, the pairwise comparison results of the four intervention measures were summarized, and the OR value was used to show the curative effect of different intervention measures. Finally, a funnel chart was used to show publication bias. If most of the studies were located between the dotted lines, publication bias could be ignored.
A total of 1697 articles were retrieved, 198 duplicates were excluded, and 1499 were left. Among them, 1421 articles were excluded by reading the title and abstract. The reasons for exclusion were animal or cell experiments, the research object was not HCC, the use of oral or topical Chinese medicine instead of TCMIs, or the treatment options were arterial interventional chemotherapy or intraperitoneal infusion chemotherapy instead of systemic chemotherapy.
Seventy-eight articles were initially selected, and six articles were manually retrieved according to the references of the selected papers. After reading 84 full texts, 64 articles were further excluded. The reasons for exclusion were basic experiments or reviews (6 articles), clinical studies but not RCTs (10 articles), multicarcinoma studies including HCC patients but no subgroup data of HCC explicitly reported (3 articles), not combined with systemic chemotherapy (25 articles), no reporting of the specific chemotherapy regimen (4 articles), dosage form of the traditional Chinese medicine was not an injection (3 articles), ORR and CBR were not clearly reported (6 articles), apparent data errors (1 article), and the number of articles about a particular Chinese medicine injection was only one (6 articles).
Twenty RCT studies [
Literature screening flowchart.
Network meta-analysis of available comparisons. Line width is proportional to the number of trials including every pair of treatments. Circle size is proportional to the total number of patients for each treatment in the network.
The twenty included studies were published from August 15, 2002, to November 8, 2018, involving a total of 1,418 patients. The number of patients included in the studies varied from 30 to 150. All of the studies reported the patients’ sex, age range, and mean age. Each study claimed that there was no significant difference in sex and age between the combined treatment group and the single treatment group. Sixteen studies reported TNM staging of tumours, one of which enrolled patients without stage limitation, seven of which enrolled patients with stage II and above, and eight of which enrolled patients with stage III and above. Two studies reported the patients’ Child–Pugh score, two reported subtypes of HCC, and five reported the Karnofsky performance score (KPS) at enrolment.
Interventions in the twenty studies were TCMIs combined with systemic chemotherapy, while the single treatment group only received systemic chemotherapy. TCMIs include Aidi injection, compound Kushen injection, and Kanglaite injection. The chemotherapy regimen included 13 combined chemotherapy regimens based on oxaliplatin or cisplatin, three combined chemotherapy regimens based on fluorine drugs, and four single-agent chemotherapy regimens. The dosages of TCMIs were the regular dosages of their respective drug instructions, which were Aidi injection 50–100 ml once a day, compound Kushen injection 10–30 ml once a day, and Kanglaite injection 200 ml once a day. Eighteen studies reported the course of treatment, nine of which had two cycles, eight of which had three to four cycles, and one of which had six cycles. The duration of the treatment ranged from 20 to 126 days, with an average of 61.2 days (for details of the characteristics of the studies, see Table
Details of the characteristics of the studies.
Study ID (author, year) | N (E/C) | M/F | Average age (E/C) | Stage | KPS (E/C) | Treatment (E/C) | Duration | Outcomesc | |
---|---|---|---|---|---|---|---|---|---|
Study 1 (Huang et al., 2002) [ | 30/28 | 43/15 | 57.6/56.2 | II–IV | NR | Aidi Inj+5Fu/5Fu | 28d∗4 | 1, 2, 3, 4, 5 | |
Study 2 (Hou et al., 2017) [ | 43/43 | 45/41 | 55.7/58.2 | II–IV | 69.27/71.34 | Aidi Inj+5Fu/5Fu | 30d∗1 | 1, 2, 3, 4 | |
Study 3 (Hou, 2018) [ | 47/47 | 57/37 | 54.57/56.24 | II–IV | 62.23/62.41 | Aidi Inj+CAFI/CAFI | 28d∗2 | 1, 2 | |
Study 4 (Hongwen et al., 2012) [ | 36/36 | 48/24 | 52/54 | NR | ≥60/≥60 | Aidi Inj+FOLFOX4/FOLFOX4 | 28d∗4 | 1, 2, 3, 4, 6 | |
Study 5 (Zhang, 2018) [ | 35/35 | 33/37 | 54.93/55.21 | NR | 62.19/62.34 | Aidi Inj+XELOX/XELOX | 14d∗4 | 1, 2 | |
Study 6 (Zhang, 2018) [ | 35/34 | 39/30 | 58.60/58.20 | III–IV | NR | Aidi Inj+XELOX/XELOX | 21d∗4 | 1, 2, 4 | |
Study 7 (Zhang and Wang, 2007) [ | 24/18 | 30/12 | Range: 32–75a | III–IV | ≥60/≥60 | Aidi Inj+DDP+5Fu/DDP+5Fu | 28d∗2 | 1, 2, 3, 4 | |
Study 8 (Shi and Wang, 2011) [ | 35/35 | 36/34 | 51.3/52.9 | III–IV | NR | Aidi Inj+DDP+EPI/DDP+EPI | 21d∗2 | 1, 2, 3, 5 | |
Study 9 (Liu et al., 2013) [ | 30/30 | 33/27 | 55/56 | II–III | NR | Kushen Inj+MAF/MAF | 21d∗3 | 1, 2, 4 | |
Study 10 (Chen and Xun, 2006) [ | 16/14 | 19/11 | 63.4/65.3 | II–III | ≥50/≥50 | Kushen Inj+FAP/FAP | 28d∗2 | 1, 2, 3, 4 | |
Study 11 (Jin, 2016) [ | 36/36 | 42/30 | 54.3a | II–IV | NR | Kushen Inj+GEMOX/GEMOX | 21d∗2 | 1, 2, 4 | |
Study 12 (He, 2018) [ | 43/43 | 49/37 | 60.4/59.2 | II–IV | 64.76/63.98 | Kushen Inj+FOLFOX6/FOLFOX6 | 21d∗6 | 1, 2, 4 | |
Study 13 (Li et al., 2018) [ | 40/40 | 51/29 | 53.8/51.3 | III–IV | NR | Kushen Inj+MAF/MAF | 21d∗3 | 1, 2 | |
Study 14 (Shi, 2011) [ | 30/30 | 35/25 | 52.8/53.4 | IIIb–IV | ≥60/≥60 | Kushen Inj+GP/GP | 30d∗2 | 1, 2, 4 | |
Study 15 (Wu et al., 2002) [ | 30/30 | 53/7 | 42/43 | II–III | NR | Kushen Inj+FUDR/FUDR | 20d∗1 | 1, 2, 3, 4 | |
Study 16 (Guan et al., 2006) [ | 20/18 | NR | Range: 26–65a | III | ≥50/≥50 | Kushen Inj+GEMOXb/GEMOX | 15d∗4 | 1, 2, 3, 5 | |
Study 17 (Yao, 2015) [ | 30/30 | 35/25 | 55/56 | NR | NR | Kushen Inj+GEMOX/GEMOX | 21d∗2 | 1, 2 | |
Study 18 (Xu et al., 2010) [ | 38/37 | 58/17 | 48.5a | II–III | NR | Kanglaite Inj+CAP/CAP | 21d∗2 | 1, 2, 3, 4, 6, 7 | |
Study 19 (Ma et al., 2017) [ | 43/43 | 38/48 | 54.8/53.7 | NR | 56.24/58.49 | Kanglaite Inj+FOLFOX6/FOLFOX6 | 21d∗2 | 1, 2, 8 | |
Study 20 (Li et al., 2014) [ | 75/75 | 79/71 | 53.2/52.1 | I–IV | NR | Kanglaite Inj+ADM/ADM | 30d∗2 | 1, 2, 3, 4, 5, 8 |
N: number; E: experimental group; C: control group; M: male; F: female; NR: not reported; Inj: injection. aMerged data of two groups; bthere were 3 kinds of chemotherapy protocol in this article: GEMOX, GEM+5–Fu+THP, and HCPT+FT207+THP; c1=objective response rate; 2=clinical benefit rate; 3=KPS; 4=adverse reactions; 5=overall survival; 6=time to progress; 7=median survival time; 8=immune function.
Of all the included studies, five studies reported detailed random allocation methods, and the others only mentioned randomness without description; two studies reported the number of patients lost to follow-up; no study mentioned whether the blind method was used or whether the allocation was hidden; all of the studies reported ORR as the primary outcome. The risk of bias and the quality of RCT studies included in this study were acceptable, as shown in Figure
Assessment of risk of bias.
ORR was the primary outcome, and CBR and quality of life (evaluated by KPS, an improvement was considered when the patient’s KPS rose by more than 10 points) were the secondary outcomes. We used fixed-effect models to analyse them. The statistics showed that all three indicators (ORR, CBR, and KPS) were better in the TCMI combined with the chemotherapy group than in the chemotherapy alone group. In terms of the differences among the three TCMIs, Kanglaite injection tended to be better than the other two in terms of the primary outcome, but it did not show a significant difference (Figures
Forest plots (results of network meta-analysis of 3 kinds of TCMI in treatment of HCC).
Surface under the cumulative ranking curve (SUCRA).
Based on the three indicators, the three different TCMIs were ranked: Kanglaite injections and compound Kushen injections were better than Aidi injections.
Publication bias of the included RCTs was measured by funnel plots (Figure
Funnel plots measuring the publication bias and small-sample effects of included RCTs.
A total of 6 studies reported efficacy indicators other than ORR and CBR, such as time to progression (TTP), overall survival (OS), and median survival time (MST).
Three studies focused on Aidi injection. A study that used 5-FU chemotherapy showed that the 6-month OS in the combined treatment group and the single treatment group was 66% and 28.9%, respectively [
There was only one study on compound Kushen injection. This study, which only included patients with stage III HCC [
There were two studies on Kanglaite injection. A study with capecitabine showed that [
A total of four studies reported the effects of TCMIs on patients' immune systems. Among them, two studies focused on Aidi injection. One study involving 86 patients with HCC [
There were two studies on Kanglaite injection. One study involving 86 patients with HCC [
A total of 13 studies reported the safety and adverse effects of the treatments. During the treatments, adverse reactions were controllable, and no patient discontinued treatment because of adverse reactions. A total of 14 adverse reactions were reported: digestive system reactions, including nausea, vomiting, diarrhoea, abdominal pain, and abnormal liver function; bone marrow suppression, including leukopenia, anaemia, and platelet decline; skin and mucosal reactions, including erythema dermatitis, oral ulcers, and urticaria; and other adverse reactions, including fever, phlebitis, hair loss, and abnormal renal function.
Five articles on Aidi injection showed a significantly lower incidence of adverse reactions in the combined treatment group than in the single treatment group, and the
The incidence of adverse reactions in 5 of 6 articles on compound Kushen injection was significantly lower in the combined treatment group than in the single treatment group. Another article [
Among the two articles of Kanglaite injection, one showed that nausea, vomiting, and hair loss were significantly reduced after the combination of chemotherapy with Chinese medicine. However, there was no significant decrease in the incidence of bone marrow suppression and abdominal pain and diarrhoea. The other article did not show a significant difference in adverse reactions between the two groups.
The rate of early diagnosis of HCC is low, and most patients have lost their chance of surgery at the time of diagnosis. Although there are a large number of interventional therapies for these patients, some patients cannot tolerate invasive procedures. An international multicentre RCT (each study) involving 371 patients with late-stage HCC who were no longer eligible for interventional therapy [
In China, a country with a high incidence of HCC, researchers have developed many different types of TCMIs, which are injections extracted from animal or plant natural medicines that have been widely used as adjuvant treatments for advanced HCC and have been written into the CSCO guidelines [
Many kinds of TCMIs can be selected for the treatment of HCC, but studies on the direct comparison of the relative efficacy between two or more TCMIs are rare. Therefore, it is essential to conduct a network meta-analysis to compare the advantages and disadvantages of different TCMIs and analyse their respective characteristics to make the most suitable choice for patients in clinical practice. Network meta-analysis of TCMIs combined with TACE for the treatment of HCC is common, but a network meta-analysis of TCMIs combined with systemic chemotherapy for HCC has not yet been reported.
This study evaluated the efficacy of three TCMIs combined with systemic chemotherapy in the treatment of HCC, including Aidi injection, compound Kushen injection, and Kanglaite injection. The compositions of the three TCMIs are shown in Table
Compositions of the three TCMIs.
TCMIs | Compositions |
---|---|
Aidi injection | |
Compound Kushen injection | |
Kanglaite injection |
Aidi injection is a compound traditional Chinese medicine preparation; that is, it is a combination of multiple traditional Chinese medicine extracts. The traditional Chinese medicines included in Aidi injection are Astragalus, ginseng, Acanthopanax senticosus, and cantharidin.
Compound Kushen injection, also known as Yanshu injection, is extracted from two kinds of traditional Chinese medicinal herbs: Sophora flavescens and Glabrous Greenbrier Rhizome. Experimental studies have confirmed that compound Kushen injection has an obvious killing effect on tumour cells, such as Hep and H22 cells, in vitro [
Kanglaite injection is an extract from Coix seed, which is a traditional Chinese medicinal herb. The main active ingredient of Kanglaite injection is a triglyceride, which contains four types of fatty acids [
This study showed that the treatment efficacy was significantly enhanced by combining either Aidi injection, compound Kushen injection, or Kanglaite injection with chemotherapy. Although the efficacy of the three different kinds of injections was not significantly different, there seemed to be a tendency that Kanglaite injection was better than the other two in terms of ORR. This suggests that if further research with larger samples is carried out, significant differences may be achieved. These results should provide a reference for the clinical selection of TCMIs in adjuvant treatment assisting chemotherapy for HCC.
It is worth noting that combining TCMIs with chemotherapy could significantly increase the number of CD3+ T lymphocytes, CD4+ T lymphocytes, and the ratio of CD4+/CD8+ T lymphocytes. This showed that TCMIs could enhance the immune function of patients by increasing the activity of T lymphocyte subsets. Systemic chemotherapy often leads to a decline in patients' immune function. This is not only detrimental to tumour control but also brings a potential danger of virus reactivation to patients infected with hepatitis B [
Some studies also reported OS and TTP. However, due to the small amount of data, a network meta-analysis was not carried out.
In terms of treatment safety, according to the results of this study, all kinds of treatment had tolerable adverse reactions, and the addition of TCMIs significantly reduced the incidence of adverse reactions. The safety of these three kinds of TCMIs was satisfactory, and TCMIs could further enhance the safety of chemotherapy and reduce patient suffering.
Our study showed that the TCMIs combined with systemic chemotherapy may be an effective and safe treatment option for patients with HCC. Furthermore, among the three TCMIs, Kanglaite injection has a tendency to outperform the other two injections in terms of ORR. This study laid the foundation for further research in the future. Moreover, our study focused on the role of TCMIs in combination with systemic chemotherapy rather than the interventional chemotherapy, which has not been reported before. Therefore, it is of considerable significance to carry out this study, which provides a basis for the application of systemic chemotherapy combined with TCMIs in patients with advanced HCC.
There are some aspects of this study that need to be improved; especially, the quality of the included literatures is not very satisfactory. Although all the literature included in this study were randomized controlled trials, only 5 of them (25%) described specific methods of random allocation. None of the included studies mentioned whether blindness and random allocation were used. Although ORR and CBR were unlikely to be influenced by lack of blindness, because they were based on image evaluations, KPS might be affected. This potential risk of bias may affect the authenticity and reliability of the results and lead to the reduced power of the test.
In future RCTs of TCMIs, clinical researchers should reduce the bias in the trial process as much as possible to improve the quality of the research evaluating the efficacy of TCMIs in the treatment of HCC. If possible, an RCT directly comparing different kinds of TCMIs should be conducted to evaluate their efficacy.
All data generated and used during the study are available from the corresponding author on request.
Dou Dou and Ze‐yu are considered as co-first authors.
The authors declare that they have no conflicts of interest.
Xiang-gen Zhong was responsible for the design of this study. Dou Dou and Ze-yu Zhang performed the search of related articles, literature screening, data extraction, and quality evaluation. Zhi-yuan Wu and Xu-dong Qiu performed the statistical analysis. Dou Dou, Ze-yu Zhang, and Xiang-gen Zhong performed the manuscript editing. Dou Dou and Ze-yu Zhang contributed equally to this work.
This study was supported by the National Natural Science Foundation of China (no. 81573963).