A systematic review and meta-analysis of traditional insect Chinese medicines combined chemotherapy for non-surgical hepatocellular carcinoma therapy

On the background of high morbidity and mortality of hepatocellular carcinoma (HCC) and rapid development of traditional Chinese medicine (TCM), we conducted a systematic review and meta-analysis of randomized clinical trials (RCTs) according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement to assess the clinical effectiveness and safety of traditional insect Chinese medicine and related preparation for non-surgical HCC. RCTs were searched based on standardized searching rules in mainstream medical databases from the inception up to May 2016. Ultimately, a total of 57 articles with 4,651 patients enrolled in this meta-analysis. We found that traditional insect Chinese medicine and related preparation combined chemotherapy show significantly effectiveness and safety in objective response rate (P < 0.001), survival time extension [12 months (P < 0.001); 18 months (P < 0.001); 24 months (P < 0.001); 36 months (P < 0.001)], amelioration for life quality [QoL scores improvement (P < 0.001); KPS improvement (P < 0.001); AFP improvement (P < 0.001)] and reduction of therapeutic toxicity [WBC decrease (P = 0.04); gastrointestinal adverse reactions (P < 0.001)]. In conclusion, traditional insect Chinese medicine and related preparations could be recommended as auxiliary therapy combined chemotherapy for HCC therapy.

(MRI), no further investigation will be required. If not, the biopsy specimen is required 9 . The measurement of alpha-fetoprotein (AFP) may not be useful in clinical practice and will not affect the formulation of final treatment strategy 10 .
Although HCC has a poor prognosis so far, ultrasonography surveillance can diagnose HCC at early stage (single tumors or as many as 3 nodules ≤ 3 cm) when the tumor could be cured by resection, liver transplantation, or ablation, and the 5-year survival rate can exceed 50% 10 . For the patients at the intermediate or end stage of HCC, palliative methods such as transcatheter hepatic arterial chemoembolization (TACE) or intravenous chemotherapy are appropriately approaches 11 . Those methods are limited by their toxicities, drug resistance, and miscellaneous adverse effects for end stage patients, meanwhile the survival benefits are still not proven 12 . Given this limitation of therapy on HCC, complementary and alternative medicine (CAM) has been increasingly applied to it over the past few decades. Physicians are trying to find more adjunctive or auxiliary therapies to improve patients' quality of life or survival time and reduce side effects caused by chemotherapy.
Traditional insect Chinese medicine is an old ancient Chinese medical concept that is still used today in China. It can be recognized as the Chinese animal species medicines, including the dried medical animal bodies, the medical animal bodies without entrails, secretions and excretions of medical animals and the processing products of insects 13 . Owing to the various versions of the translation for TCM, we could not even find a precise translation for traditional insect Chinese medicine from a number of Chinese-English dictionaries of TCM. As a result, we adopted the word "insect" that has the same literary meaning of "animal" to replace it. The traditional insect Chinese medicine is a broad concept, while the narrow definition of it is only confined to the insect or worm drugs. It has been originally recorded in medical book named Fifty-two patients side for the medical interpretation and practice in the Warring States period of China (475-221BC) 14 . And Shennong's herbal classic, one of the earliest Chinese herbal collections of TCM, deeply introduced the effects of traditional insect Chinese medicine 15 . Another Chinese herbal masterpiece, Compendium of materia medica, written by Shizhen Li, collected nearly 500 kinds of traditional insect Chinese medicines and classified them meticulously, promoting traditional insect Chinese medicine to a new peak. The modern experimental researches of traditional Chinese insect medicine have been gradually focussing on anti-tumor properties by inhibiting the proliferation of cancer cells, inducing tumor cell apoptosis and improving the immune function of the human body 16 .
Traditional insect Chinese medicine, as an essential component of TCM, has been widely applied in the treatment of the non-surgical HCC (at the intermediate stage and the advanced stage) for a long time with side effects a rare occurance. Although a great number of published clinical studies have evaluated different kinds of traditional insect Chinese medicines and the related preparations combined chemotherapy for HCC [17][18][19] , the clinical efficacy and pharmacological mechanism remain unclear. No article of systematic review or meta-analysis has been conducted based on considering the traditional insect Chinese medicines as an integral part to evaluate the efficiency and safety for HCC treatment. Therefore, we conducted a systematic review and meta-analysis to investigate the evidence about traditional insect Chinese medicine and the related preparation combined chemotherapy for the clinical effectiveness and safety of non-surgical HCC.

Materials and Methods
This systematic review and meta-analysis was conducted by order of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement 20  The searching terms were performed as follows: "liver cancer, " "hepatocellular carcinoma, " "liver neoplasm, " "primary liver carcinoma, " "traditional insect Chinese medicine, " "insect drugs, " "cinobufatalin, " "huachansu, " "centipede, " "scorpion, " "leech, " "earthworm, " "cantharides, " "gecko, " "snake venom, " "grand beetle, " "aspongopus, " "gadfly, " "honeycomb, " "chemotherapy, " "interventional therapy" and "random clinical trials". No other restrictions were performed and the free text strategy and Medical Subject Headings (MeSH) terms were conducted in the term searching process. Categories of traditional insect Chinese medicine on literature searching were derived from the classification of MeSH term in CBM database. The searching language in Chinese, English and Japanese was slightly changed based on the situation for different databases adaptation. Study selection. The study selection was conducted by two reviewers (Z.F.Shi, Y.Q.Yan) independently and disagreements were resolved by the common strategy or a third reviewer (J.Xie). Eligible studies that suited for the following criteria were included in this systematic review and meta-analysis: (1) randomized clinical trials (RCTs); (2) patients were diagnosed as non-surgical HCC (TNM stage II or above; BCLC stage B or above) by pathological results or imaging examinations; (3) participants received traditional insect Chinese medicine and related preparation combined chemotherapy in the treatment group, meanwhile chemotherapy alone in the control group; (4) evaluated outcomes including at least one of the following variables: (a) complete response rate (CR), partial response rate (PR) and CR + PR as a proportion of objective response rate; (b) quality of life (QoL) scores or Karnofsky performance scores (KPS); (c) the Child-Pugh scores; (d) major side effects resulting from traditional insect Chinese medicines or chemotherapy; (5) survival time rate (the number of participants in the treatment and the control group who were alive at 6, 12, 18, 24, 36 months). The studies were excluded if they met these following criteria: (a) received the Chinese herbal medicine in the control group; (b) received the irrelevant TCM therapies or western medical methods in experimental group; (c) the clinic pathologic types of liver cancer were not compatible with the HCC criteria; (d) severe clinical illnesses (such as liver or kidney disease) or infections; (e) absence or inconsistency of methods, evaluation criteria or results; (f) non-randomized clinical trials (comments, case reports, reviews, editorials, letters, et al.), articles unrelated with the topic, or duplicated articles.
Particularly, the clinical stage of HCC was identified according to the TNM staging system of the American Joint Committee on Cancer (AJCC) obtained from the National Comprehensive Cancer Network (NCCN) guideline 2015 21 (T refers to the size of the original tumor and whether it has invaded nearby tissue; N represents nearby lymph nodes that are involved; M can be recognized as distant metastasis) or the Barcelona Clinical Liver Cancer (BCLC) system coming from the European Association for the Study of the Liver (EASL) 22 . For the reason of diversity and complexity of TCM therapy, traditional insect Chinese medicine and the related preparation can be obtained from some Chinese herbs that have the synergistic effects without interfering with the major function of traditional insect Chinese medicine. The imaging examination includes the methods of computed tomography (CT) or magnetic resonance imaging (MRI), showing the lesions ≥ 10mm at least. Also, CT or MRI can evaluate the CR or PR: CR means all targeted lesions disappear and pathological lymph nodus must reduce to less than 10mm at least, and PR means targeted lesions decrease 30% in the sum of diameters at least. This solid tumor response was evaluated according to the Response Evaluation Criteria in Solid Tumor (RECIST) guideline (version 1.1), which has been updated and revised by the European Cancer Organization (ECCO) in 2009 23 . The side effects (the chemotherapy related toxicity and TCM adverse effects) included as the follows: bone marrow suppression including leukopenia, thrombocytopenia or anemia; gastrointestinal side effects including nausea, emesis, anorexia, or diarrhea; liver or kidney injury; pyrexia; and pain. The dosage of traditional insect Chinese medicine and chemotherapy in experimental groups and control groups of trials was discrepant. There was no limitation for the dosage in the study searching and including.
Data extraction and quality analysis. Two reviewers (Z.F.Shi, Y.Q.Yan) extracted data independently, assessing the inclusion and exclusion criteria before based on the standardized collection. All disagreements were discussed between two reviewers mentioned before, and the final conclusion was reached after deliberation over with a third reviewer (J.Xie). The extracted characteristics comprised the following items: (1) the name of author and the year of publication; (2) sample size; (3) study design; (4) study performed areas; (5) the baseline characteristics of patients; (6) approaches and duration of treatments; (7) outcome evaluation and trials quality assessment.
The quality analysis was performed by two investigators independently (Z.F.Shi, Y.Q.Yan), using the Cochrane Collaboration's tools for assessing the risk of bias 24 . This tool was conducted to evaluate the bias of studies across six domains: (1) the method of random allocation; (2) the concealment of allocation; (3) the blinding method; (4) the integrity of outcome data; (5) the outcome data of selective reports; (6) other bias sources. Every domain mentioned above was assessed by the criterion of "yes", "no" or "unclear". The article had 3 or more "yes" could be recognized as high quality and less than 3 "yes" should be considered as low quality. High quality studies had the low bias risk while the low ones had the high bias risk. The related data was analyzed by the software named the Review Manager (RevMan; version 5.2 the Nordic Cochrane Center, the Cochrane Collaboration, 2012 Copenhagen, Denmark), with the outcomes illustrated by images or tables efficiently and conveniently.
Statistical analysis. The RevMan (version 5.2) was applied to pool and analyze data. The reviewer (Z.F.Shi) calculated the risk ratio (RR) with 95% confidence interval (CI) for the dichotomous outcomes and the standard mean difference (Std.MD) or mean difference (MD) with 95% confidence interval (CI) for the continuous outcomes respectively. Statistical heterogeneity was tested through studies by the method of I 2 statistic. It was a quantitative tool for inconsistency and could provide an estimate of variation resulted from heterogeneity. Results of I 2 statistic between 25 and 50% were regarded as low heterogeneity, 50 and 75% were moderate heterogeneity, and above 75% were high heterogeneity. It should be noted that the Cochrane Handbook (version 5.1.0) indicates the value of I 2 statistic above 50% was considered to have considerable heterogeneity 25 . The subgroup analysis was performed to find the source of heterogeneity. A fixed effects model (applying Mantel-Haenszel method 26 ) was conducted to pool data when the heterogeneity did not exist or moderate, while a random effects model (applying Der Simonian-Laired method 27 ) was performed when there was obvious heterogeneity. Additionally, sensitivity analysis was further conducted to assess the stability of results and evaluate the variation of pooled data. Publication bias was assessed visually by funnel plots and calculated in Egger's test/Begg's test through the software named Stata (version 14.0, StataCrop LP, College Station, US) 28,29 . P values lower than 0.05 were judged as statistically significant for the results, representing that the study has publication bias.

Results
The flowchart of article search and selection in the meta-analysis is presented in Fig. 1. Of a total of 358 potentially relevant articles identified from 15 different electronic databases, one hundred and ninety-one studies of them were ruled out for the reason of duplication. Ninety-three publications were further excluded, after detailed screening and analyzing, for the following reasons: (a) thirty-five studies were confirmed unrelated to article topics or could not find the full-text papers even after contacting with authors; (b) ten studies were found to be systematic reviews and meta-analyses; (c) Thirty-two studies were non-clinical trials or non-randomized clinical trials; (d) Sixteen articles were unrelated to hepatocellular carcinoma or traditional insect Chinese medicine. Seventy-four full-text articles were eligible and then reviewed independently. Of 17 publications further excluded, 11 publications were excluded because the inappropriate criteria of experimental or control group, for instance, the control group contains TCM therapy. Three publications missed sufficient data (i.e. Control group), which made the report outcomes untrustworthy. Three publications obtained irrelevant observation criteria, which seemed inconsistent with the criteria mentioned before. Overall, a total of 57 articles with 4,651 patients were enrolled in this meta-analysis  . (Table 1) Included articles characteristics and quality evaluation. A total of 57 publications met with the final eligibility criteria. All studies were randomized clinical trials (RCTs) and carried out in the hospitals in China. All trials were single-center studies except one performed in 3 different areas' hospitals in China 45 . Published year of articles was between 1999 and 2015. These trials were conducted among men and women between 18 and 85 years old. There were eighteen articles clearly introduce the stage of HCC 30, 32, 35, 39, 40, 45, 46, 55-57, 59, 60, 65, 68, 69, 74, 80, 85 . The most common stage of HCC was the stage II-III based on the National Comprehensive Cancer Network (NCCN) guidelines 2015 21 except for 2 publications including stage I hepatocellular carcinoma 45,74 . (Table 1) Seventeen articles clearly reported the Child-Pugh grades included class A and B 31, 32, 37, 39, 41, 44, 45, 49-51, 53, 58, 64-67, 73 . However, Child-Pugh class C was extra included in 5 articles mentioned before 39,44,50,51,64 and Child-Pugh class A was only included in 1 article 73 . The traditional insect Chinese medicine and the related preparation are diversified in included articles: the major insect related medicines contained as follow: cinobufotalin injection, Aidi injection, sodium cantharidinate and vitamin B6 injection, sodium norcantharidate injection, jinlong capsule, and compound cantharides capsule. (Table 1) The major chemotherapy plan in studies was transcatheter hepatic arterial chemoembolization (TACE), while some articles reported additional chemotherapy regimens, which included FAP (5-fluorouracil + doxorubicin + cisplatin), FAM (5-fluorouracil + doxorubicin + mitomycin), FOLFOX4 (oxaliplatin + calcium folinate + 5-fluorouracil) and 5-fluorouracil 30 For the included articles, all of them expect 7 studies (12%) were only mentioned the allocation sequence generation without showing the specific random method. However, of the 7 articles, three used shuffling envelopes method 36,45,74 and 4 used the random number table method 31,32,48,49 . Only 3 articles (5%) mentioned the allocation concealment from the included studies: the sealed envelope method was the way they chosen 36,45,74 . All articles did not report the blinding method, excepted 1 article which was a single-blind randomized clinical trial 70 . Most of studies (92%) were ranked as low risk of outcome integrity data, while 5 articles (8%) reported participants' dropout reason 51,63,67,82,84 . None of the articles clearly illustrated the reporting bias and only 2 studies (3%) were rated as low risk for no other bias 31,32 . (Table 2, Figs 2, 3). We conducted subgroup analysis on the condition of the pervasive low quality (high bias risk) of included studies. We extracted the high quality articles (5 papers, 510 participants) from the 46 papers based on the Cochrane Collaboration's tools for assessing risk of bias mentioned before 31,32,36,45,74 , making meta-analysis to compare with the rest low quality articles (41 papers, 3,149 participants). The meta-analysis reported that the objective response rate (CR + PR) of high quality group, low quality group and overall group all revealed traditional insect Chinese medicine combined chemotherapy was superior to single chemotherapy method for the non-surgical HCC (Low quality: RR = 1.21, 95% CI = 1.01 to 1.45, P = 0.04 < 0.05; High quality: RR = 1.32, 95% CI = 1.23 to 1.42, P < 0.001; Overall: RR = 1.31, 95% CI = 1.22 to 1.40, P < 0.001). The results showed that the objective response rate for the treatment group and the control group has statistical difference. The heterogeneity did not exist in the trials (Low quality: P = 0.92, I 2 = 0%; High quality: P = 0.78, I 2 = 0%; Overall: P = 0.88, I 2 = 0%) and the fixed effects model was performed to calculate combined data by Mantel-Haenszel test.

Meta-analysis. Objective response rates (CR + PR).
Survival time. Five kinds of survival time were assessed in the meta-analysis including 6 months, 12 months, 18 months, 24 months and 36 months. (Fig. 5). Twelve articles including 1,206 participants analyzed the 6 months survival time 38,41,44,45,50,51,54,63,71,76,82,84 . The meta-analysis showed the 6 months survival time for the experimental group (used traditional insect Chinese medicine combined chemotherapy) and the control group (used chemotherapy alone) was irrelevant (RR = 1.04, 95% CI = 0.99 to 1.10, P = 0.14å 0.05). The heterogeneity was moderately high (P = 0.14, I 2 = 31%), which might be due to a particular trial 54 included in the meta-analysis. So we performed the sensitivity analysis and if this trial was excluded, the I 2 value would be down to 0%. The fixed effects model was performed to pool data. Twenty articles including 1,831 participants assessed the 12 months survival time 38, 41, 42, 44, 45, 47, 50-52, 54, 61, 63, 64, 67, 71, 75, 76, 78, 82, 84 . The meta-analysis illustrated that the experimental group improved the 12 months survival time compared with the control group (RR = 1.42, 95% CI = 1.32 to 1.53, P < 0.001). The heterogeneity was moderately high (P = 0.02, I 2 = 43%) while the I 2 value was lower than 50%. The fixed effects  The meta-analysis illustrated the experimental group significantly improved the 24 months survival time compared with the control group (RR = 1.70, 95% CI = 1.50 to 1.91, P < 0.001). The heterogeneity was low among the included articles (P = 0.24, I 2 = 17%) and the fixed effects model was performed to pool data. Twelve articles including 1,381 participants analyzed the 36 months survival time 42,44,52,61,64,67,75,76,78,82,84 . The meta-analysis illustrated the experimental group significantly improved the 36 months survival time compared with the control group (RR = 1.93, 95% CI = 1.55 to 2.39, P < 0.001). There was no heterogeneity in these 4 articles (P = 0.86, I 2 = 0%) and the fixed effects model was performed to pool data. We did not find any valid articles that had analyzed the 5 years survival time, and only one article analyzed 3 months survival time 68 .
Side effects. This part of meta-analysis included 6 kinds of side effects, which contained bone marrow depression, gastrointestinal adverse reactions, liver damage, kidney damage, fever and pain. As for bone marrow depression, two laboratory indexes including white blood cell (WBC) decrease (13 articles, 1058 patients) 30, 32, 37, 40, 43, 46, 51, 53, 63-65, 67, 68 and hemoglobin (HB) decrease (7 articles, 563 patients) 30-32, 37, 53, 64, 67 were applied as random effects model (the heterogeneity of WBC decrease: P < 0.001, I 2 = 85%; HB decrease: P = 0.04, I 2 = 54%); one laboratory index named the platelets (PLT) decrease (only 3 articles, 193 patients) 32,37,53 was applied fixed effects model (heterogeneity: P = 0.6, I 2 = 0%). The subgroup analysis (seen in the Supplementary Information of the article) of WBC decrease was performed based on the drugs classification, literature publication years and experimental areas, but the source of heterogeneity was still unclear. However, the subgroup analysis of HB decrease found that the heterogeneity came from experimental areas (North of China: P = 0.27, I 2 = 23%; South of China: P = 0.13, I 2 = 46%). The incidence rates of HB decrease and PLT decrease in experimental group were irrelevant to control group  with corresponding Risk Ratio (RR) and 95% CI were (0.84, 0.62to 1.12, P = 0.23å 0.05), and (0.97, 0.75 to 1.26, P = 0.84å 0.05), respectively. The incidence rate of WBC decrease in the experiment group was lower than in control: (0.74, 0.55 to 0.99, P = 0.04 < 0.05). (Fig. 7). Red blood cell (RBC) decrease, another important component of bone marrow depression, was excluded owing to only one trial analyzing it 30 . Twenty articles including 1,529 patients reported data on gastrointestinal adverse reactions 31-34, 40, 46, 53, 58, 59, 64, 65, 67, 69, 75, 78-81, 83, 86 . The random effects model was conducted to pool data because the heterogeneity (P < 0.001, I 2 = 62%) was obvious. We made the subgroup analysis based on the drugs categories, publication years and experimental areas. The outcome of subgroup analysis showed that the variation among the different experimental areas was the source of heterogeneity (North of China: P = 0.34, I 2 = 11%; South of China: P = 0.18, I 2 = 28%). The meta-analysis showed the gastrointestinal adverse reactions in the experimental group were significantly fewer than the control group (RR = 0.63, 95% CI = 0.51 to 0.77, P < 0.001). Eleven articles including 815 patients and 5 articles including 340 participants gathered information on liver damage 33,34,40,51,53,67,73,78,80,81,83 and kidney damage 53,67,78,80,83 respectively. The random effects model was performed to collect data for the liver damage group because the heterogeneity was high (P < 0.001, I 2 = 92%). The subgroup analysis based on the drugs categories, publication years and experimental areas didn't reveal the source of heterogeneity. The fixed effects model was conducted to collect data for kidney damage group because no obvious heterogeneity was found in the included articles (P = 0.97, I 2 = 0%). The meta-analysis demonstrated that the liver damage and kidney damage in the experimental group were irrelevant to control group (liver damage: RR = 0.56, 95% CI = 0.31 to 1.02, P = 0.06 > 0.05; kidney damage: RR = 0.55, 95% CI = 0.24 to 1.27, P = 0.16 > 0.05).

Discussion
Hepatocellular carcinoma (HCC) is a worldwide serious disease, causing serious health problems to human beings. The treating methods of western medicine, which include liver transplantation, liver resection, radiofrequency ablation (RFA), transcatheter hepatic arterial chemoembolization (TACE) and molecular targeted drugs, have already been used on HCC therapy for a long time. As for non-surgical hepatocellular carcinoma, the effective clinical evidence is insufficient and the heated debate also was aroused for the selection of optimal treating methods 88,89 . The current chemotherapy performed to end-stage HCC has been proved many side effects: bone    marrow depression, gastrointestinal adverse reactions, liver damage, kidney damage, fever, pain, hair loss, dizziness, et al. These complications also become a huge burden for HCC patients. Based on this serious context, various complementary and alternative medicine (CAM) treatments have been administered for HCC in clinical practice. The traditional insect Chinese medicine, as an essential part of TCM, has been utilized to treat HCC as an auxiliary method (or CAM methods) for a long time. The effective and safe evaluation of traditional insect Chinese medicine for non-surgical HCC treatment is urgent and necessary. To the best of our knowledge, this article is the first PRISMA compliant systemic reviews and meta-analysis to evaluate the effects and safety of traditional insect Chinese medicine combined chemotherapy for non-surgical HCC. Fifty-seven articles, all of which were conducted in China, included 4,651 participants in the meta-analysis. Even with strong cultural and population bias, seven obvious advantages still need to be emphasized: (a) fourteen medical databases were utilized and the publication language was English, Chinese and Japanese; (b) the article quality was strictly accessed and evaluated by three cooperated reviewers (T.B.Song, Y.Wan and L. Shang); (c) we performed a comprehensive analysis for different categories of the including traditional insect Chinese medicine and the subgroup analysis and sensitivity analysis was both well conducted to seek the source of heterogeneity; (d) we contacted authors on whether the included articles could receive on full-text; (e) Egger's test, Begg's test and funnel plot was performed to confirm whether the publication bias existed in the studies; (f) the contribution network plot graph was strictly designed to evaluate the mutual effectiveness among the included traditional insect Chinese medicines as clearly exhibited as graphs; (g) the Cochrane risk of bias tool was performed to assess the study quality. The studies' outcomes showed good qualities were still insufficient for the reason of the high risk on blinding methods bias, selective reporting bias, and other bias.
The meta-analysis was carried out in 46 articles to evaluate the objective response rate (CR + PR), showing the significantly effectiveness in the experimental group (traditional insect Chinese medicine combined chemotherapy) compared with the control group (chemotherapy) alone. The subgroup analysis based on the articles' quality difference also provided the same result. These results suggest that the short-term effects of traditional insect Chinese medicine as auxiliary method combined chemotherapy are superior to chemotherapy alone. It is important to note that the publication bias on objective response rate indicates inconformity between funnel plot in visual appearance and Egger's/Begg's test results. Involved studies showed that experimental groups significantly improve the survival time at 12 months, 18 months, 24 months and 36 months compared with control groups, revealing that traditional insect Chinese medicine combined chemotherapy might prolong the survival time of non-surgical HCC patients. However, no statistical difference was found in 6 months survival time. The meta-analysis contained quality of life (QoL) scores, Karnofsky performance scores (KPS) and AFP improvement, all showing significant improvement in the experimental group compared with the control group. We did not observe large heterogeneity among these studies and the results indicated an excellent improvement of patients' life quality for the the utilization oftraditional insect Chinese medicine. The meta-analysis evaluated the side effects ratio after therapy, clearly showing the obvious reduction of gastrointestinal adverse reaction and WBC decrease in the experimental group compared to the control group. The liver damage, kidney damage, fever, pain, HB decrease and PLT decrease did not show significantly statistically difference although all of them indicated the same conclusion that side effects ratio of traditional insect Chinese medicine combined chemotherapy lower than chemotherapy alone. Although we found that the obvious heterogeneity in WBC decrease, HB decrease, gastrointestinal adverse reactions, liver damage, fever and pain was obvious, most of them did not show the source of heterogeneity except the HB decrease and gastrointestinal adverse reactions. The heterogeneity of HB decrease and gastrointestinal adverse reactions both came from the variation among the different experimental areas. The meta-analysis showed that the improvement of immune function in the experimental group was higher than the control group although the heterogeneity was obvious. We did not find the source of heterogeneity of them. So we could not give the firmly conclusion on the effectiveness of immune function improvement except we could include more relevant articles in the future meta-analysis. For forty-six included trials focused on objective response ratio, all experimental groups utilized traditional insect Chinese medicine related preparation as auxiliary therapy combined chemotherapy for the advanced HCC therapy.
The publication bias for objective response rate did exist in this meta-analysis after Egger's and Begg's test. It might be derived from (a) some authors tended to deliver positive results of articles to editors while prejudiced negative results 90 ; (b) some magazine editors or reviewers had a preference to positive results of articles while caviled to negative results to some extent 87 ; (c) trials that received government funding had more possibility to publish on some magazines than received private or company funding 91 . The meta-analysis would overstate the degree of association between treating effects and risk factors because of the publication bias, which would bring mistakes for clinical patient's therapy or health decision-making.
Limitation. There are several potential limitations of current data and results that should be interpreted with caution. Firstly, included clinical studies all conducted in different centers in China, which may bring the regional and cultural bias and influence the widely application of traditional insect Chinese medicine in the future. Secondly, for the reason of varied medical development in different regions of China, clinical centers in China conducting the trials have different clinical levels of tumor diagnosis and treatment, which could result in a bias of the reported incidence rate. Thirdly, the included randomized clinical trials have many flaws caused by human baseline risk factors (all patients were Chinese), incomplete methodological design of trials, small sample trials (some studies have less than 30 patients per group) and index criteria deficiency (the information of the clinical stage of HCC or Child-Pugh scores deficiency). Fourthly, some results showed significantly heterogeneity among the included studies, which may be due to the sample size, different experimental region in China, medicine application and dose, publication years, as well as duration of treatment. However, we need to understand that establishing the random-effects model to pool data cannot give exact and stable conclusion in this situation. Fifthly, the application of traditional insect Chinese medical preparations varied in included articles. For example, some insect medical preparations were used through the way of injection (i.e. cinobufotalin injection, sodium cantharidinate and vitamin B6 injection, Aidi injection, et al.) and some were administrated through the oral way (i.e. Jinlong capsule, compound cantharis capsule, et al.). The time of applying medicines was unsynchronized, which also might exert potential time bias.

Conclusion
This system review and meta-analysis offers a practical and beneficial result for the effectiveness and safety of traditional insect Chinese medicine combined chemotherapy on non-surgical hepatocellular carcinoma (HCC). This combined therapy can provide assistance for improving the objective response rate (CR + PR) of HCC, prolonging long term survival time (12, 18, 24, 36 months), enhancing life quality (QoL scores and KPS) of patients, increasing some laboratory index (AFP) and immune functions (CD3 + , CD4 + , CD4 + /CD8 + , NK) and reducing some adverse effects (WBC decrease, gastrointestinal adverse effects). However, some results (6 months survival time, PLT decrease, HB decrease, liver damage, kidney damage, pain, fever and CD8 + ) cannot support a convincing conclusion because they cannot obtain the statistical difference. We should make the conclusion cautiously that this combined therapy can be applied as an auxiliary clinical treating method on non-surgical HCC treatment because many included publications' low quality increases risk and bias and affects the reliability of study to some extent. So the traditional insect Chinese medicine and related preparation still need further standard, multicenter, double-blind randomized clinical trials (RCTs) to provide more clinical evidences in the future. Data Availability Statement. All data in this article are fully available without restriction.