Anticancer effect of berberine based on experimental animal models of various cancers: a systematic review and meta-analysis

Background Numerous studies have explored the anti-tumor effect of berberine (BBR), but little clinical evidence guides the use of BBR in cancer patients. Objectives Our aim was to investigate the impact of BBR on various cancers in healthy animals to promote the transformation from bench to bed. Search methods PubMed, Embase, Springer, and Cochrane databases were searched from January 2000 to October 2018 for relevant articles. Selection criteria Only published studies focusing on the relationship between BBR and various cancers in vivo were qualified. Two review authors independently assessed the risk of bias for each study, and any disagreement was resolved by discussion or by involving a third assessor. Results A total of 26 studies from 2000 to 2018, focusing on various cancer types, including breast cancer, liver cancer, colorectal cancer, nasopharyngeal carcinoma, lung cancer, gastric cancer, neuroepithelial cancer, endometrial carcinoma, esophageal cancer, tongue cancer, cholangiocarcinoma, and sarcoma were included. Overall, BBR reduced tumor volume (SMD =3.72, 95% CI: 2.89, 4.56, Z = 8.73, p < 0.00001) and tumor weight (SMD =2.35, 95% CI: 1.51, 3.19, Z = 5.50, p < 0.00001) in a linear The dose–response relationship (Pearson r = − 0.6717, p < 0.0001 in tumor volume analysis; Pearson r = − 0.7704, p < 0.0005 in tumor weight analysis). BBR inhibited angiogenesis in tumor tissues (SMD = 4.29, 95% CI: 2.14, 6.44, Z = 3.92, p < 0.00001), but it had no significant effect on the body weight of experimental animals (SMD = 0.11, 95% CI: − 0.70, 0.92, Z = 0.27, p = 0.78). Publication bias was not detected. Conclusion BBR exerted anti-tumor effects in a variety of tumors in vivo, especially breast cancer and lung cancer, and the evidence was still insufficient in colorectal cancer and gastric cancer.


Background
Berberine (BBR) is a natural component purified from the species of the genus Berberis, which has long been used as an anti-diarrheal drug in gastrointestinal disorders in traditional Chinese medicine [1]. At the same time, the anti-tumor effect of BBR has been a hot topic in experimental research in recent years. In the past 3 years, latest studies have shown the anticancer actions of BBR against several high-risk cancers, including lung cancer [2], breast cancer [3], prostate cancer [4], colorectal cancer [5], and gastric cancer [6].
However, little clinical evidence guides the use of BBR in cancer patients. Thus, systematic reviews and metaanalyses of animal studies may help to clarify whether cancer patients could benefit from this approach and promote the transformation of animal studies into humans at the same time [7].
Our aim was to investigate the impact of BBR on cancer growth and its adverse effects in randomized controlled trials in healthy animals.

Identification of studies
From January 2000 to October 2017, relevant literature from PubMed, Embase, Springer, and Cochrane databases was systematically screened. The following Mesh terms and textwords were used: "Neoplasms"[Mesh], "Neoplasia," "Neoplasias," "Neoplasm," "Tumors," "Tumor," "Cancer," "Cancers," "Malignant Neoplasms," "Malignant Neoplasm," "Neoplasm, Malignant," "Neoplasms, Malignant," "Malignancy," "Malignancies," "Berberine"[Mesh]," "Berberine," "Umbellatine," and "BBR." The "AND" or "OR" operator was used to combine these terms in varying combinations. At the same time, references in the articles were also included in the screening. We did not set a language limit during the process. Two authors (Jianhao Xu, Yuming Long) independently reviewed the titles and abstracts identified in the search. In this process, we discussed the articles to incorporate the differences. If problems still could not be resolved, a third assessor (Yusong Zhang) was invited to make a decision. Only published articles were included. No protocol was developed for this review.

Selection criteria
The inclusion criteria were as follows: (1) participants: experimental animals including rodent, mouse, rat, rabbit, guinea pig, dog, horse, sheep, and monkey; (2) invention: BBR only; (3) outcomes: the effects of BBR in animal models after tumor implantation, including tumor volume, tumor weight, tumor vessel density, and body weight; and (4) study design: experiments should be prospectively controlled. The exclusion criteria were as follows: (a) literature published as letters, editorials, abstracts, reviews, and expert opinions; (b) non-animalbased studies; (c) articles with missing data information; (d) similar and repeated studies; and (e) outdated articles with little significance and credibility. Cohen's kappa statistic was used to assess chance-corrected agreement between reviewers (SPSS version 18. 0, SPSS Inc. Chicago, IL) [8].

Study characteristics and data extraction
A detailed form was designed for data extraction: first author, publication year, country, cancer type, animals' baseline characteristics, intervention, duration, and the data of specific outcomes (tumor volume, tumor weight, tumor vessel density, and body weight). Two review authors extracted the data by using the agreed form.

Quality of evidence and risk of bias
For risk of bias of individual studies, the ARRIVE checklist was used to assess pre-clinical animal studies [9]. For risk of bias among studies, such as publication bias and selective reporting, funnel chart analysis, subgroup analysis, and sensitive analysis were all conducted. Two review authors (Jian−hao Xu, Yuming Long) independently assessed the risk of bias for each study.

Data synthesis and statistical analysis
We carried out statistical analysis by using the Review Manager software (RevMan 5.3) and STATA statistical software package version 12.0 (Stata Corporation, College Station, TX). The primary outcomes were tumor volume, tumor weight, and tumor vessel density of BBR group compared with the control group. The secondary outcome was the change of body weight. Mean value and standard difference (SD) were used as summary statistics. Standard mean difference (SMD) was measured for continuous data. Linear regression and Pearson's correlation analysis were used to study the The doseresponse relationship between BBR and the four outcomes. The heterogeneity among studies was measured by using the I 2 test. The latent publication bias was assessed by using a funnel plot. All statistical tests were two-tailed, and p < 0.05 was considered statistically significant.

Search results
A total of 969 potential articles were identified from the literature search. After selection, 26 studies  matched the inclusion criteria and were suitable for our meta-analysis. The flow diagram in Fig. 1 showed the selection process. A review of the study selection and data extraction indicated excellent agreement between reviewers (k = 0.820).
The dose-response relationship of different cancer types on the relationship between BBR and tumor volume of animals is shown in Fig. 4. For single cancer types, a statistically significant linear relationship in colorectal cancer (Pearson r = − 0.8785, p = 0.0499) and lung cancer (Pearson r = − 0.6718, p = 0.0459) was observed. For total    The colours indicate where the proportion of studies meeting that criteria are less than 25% (red), 25%-50% (pink), 50%-75% (light green) and more than 75% (green).
The colours indicate where the proportion of studies meeting that criteria are less than 25% (red), 25%-50% (pink), 50%-75% (light green) and more than 75% (green) studies, the SMD values of all studies showed a statistically significant decreasing trend with increasing concentration of BBR (Pearson r = − 0.6717, p < 0.0001).
The dose-response relationship of different cancer types on the relationship between BBR and tumor weight of
No statistical heterogeneity was observed (I 2 = 0% for breast cancer).
The dose-response relationship of different cancer types on the relationship between BBR and tumor weight of animals is shown in Fig. 9. For single cancer types, no linear relationship was concluded because of

Body weight
Of the 26 screened articles , 7 [11,16,17,20,22,27,35] reported the relationship between BBR and body weight in animals with breast cancer [11,16], liver cancer [17], colorectal cancer [20,22], gastric cancer [27], and sarcoma [35]. The SMD and the 95%CI in each study are shown in Fig. 10. In view of the obvious heterogeneity(I 2 = 73% for breast cancer; I 2 = 80% for colorectal cancer; I 2 = 52% for liver cancer), we conducted a subgroup analysis of different characteristics mainly on the following aspects: dose of BBR, administration, and cell lines (Fig. 11). I 2 were missing in breast cancer group and liver cancer group due to limited studies. No potential influencing factor was found in colorectal cancer group.
The dose-response relationship of different cancer types on the relationship between BBR and body weight of animals is shown in Fig. 12. For single cancer types, no statistically significant linear relationship was found. For total studies, the SMD values of all studies showed no statistically significant trend(Pearson r = − 0.1440, p = 0.7116).

Publication bias and sensitivity analysis
The publication bias evaluation for the meta-analysis of tumor volume, tumor weight, tumor vessel density, and body weight is shown in Fig. 13. These funnel plots showed that most of the studies are in the upper part of the inverted funnel and approximately symmetrical, suggesting that the publication bias was unobvious.
A sensitivity analysis was performed to assess the stability of our results in terms of tumor volume, tumor weight, tumor vessel density, and body weight. The trim method was used, and the results did not show considerable changes between the previous and new SMDs (Fig. 14). Next, we deleted one individual study at a time, and the results of the rest of the studies were checked for any reversal. The statistical outcomes showed that the pooled SMDs were all still significant although one study was excluded (Fig. 15).

Molecular pathways and proteins
Among these included studies, a wide range of molecular targets, which are essential for the anti-cancer effect of BBR, was revealed. Except for three articles [15,33,35] that did not involve the discussion of molecular mechanisms, the remaining 23 articles [10-14, 16-32, 34] analyzed the anti-tumor mechanism of BBR. The pharmacological effects of BBR was summarized into five aspects: proliferation(including apoptosis, autophagy, cell cycle arrest, and others), intracellular oxidative stress, inflammation, angiogenesis, and migration. Table 3 shows how BBR works in different scenarios of multiple types of cancers. In addition, in order to understand the anticancer mechanism more clearly and deeply, Table 4 shows the clustering analysis of the common molecular pathways and target proteins between studies. Fig. 9 The dose-response relationship of BBR and tumor vessel density The most frequently studied pathways were on cell proliferation and 19 articles focused on this mechanism. Seven of these studies involved tumor cell apoptosis pathways (breast cancer [9,12], liver cancer [17,18], lung cancer [26], gastric cancer [27], esophageal cancer [32]), one involved autophagy pathways (neuroepithelial cancer [30]), and five involved cell cycle arrest pathways (colon cancer [22], lung cance [25], gastric cancer [27], esophageal cancer [32], cholangiocarcinoma [34]). The second most common frequently studied pathways were on cell migration. Four articles in three cancers studied the relationship between BBR and tumor cell migration (breast cancer [13,16], liver cancer [19], endometrial carcinoma [31]). There was only one study reported the relationship between BBR and intracellular oxidative stress (breast cancer [14]), inflammation (breast cancer [14]), and angiogenesis (liver cancer [19]) respectively.

Discussion
We performed a systematic review and meta-analysis to systematically evaluate the efficacy and adverse effect of BBR on various cancers. The results showed that BBR could inhibit the growth of a variety of cancers in vivo, In the past 3 years, numerous studies have attempted to elucidate the relationship between BBR and breast/lung cancer. By using molecular modeling and in vitro studies, BBR significantly reduced EGFR and AKT phosphorylation and may be a useful alternative to lapatinib, an EGFR inhibitor which can cause acquired drug resistance in breast cancer patients [36]. BBR lowers blood sugar, increases insulin sensitivity, and corrects lipid metabolism disorders; it may reduce the incidence of breast cancer [37]. Single-drug BBR has an obvious inhibitory effect on lung cancer cells; BBR can inhibit doxorubicin (DOX)-mediated STAT3 activation and sensitize lung cancer cells to the cytotoxic effects of DOX treatment. Given the widespread clinical application of BBR and its low toxicity, our findings are important for the development of a new combination of BBR and DOX for the treatment of lung cancer [38]. In addition to medical treatment, BBR has protective effects on radiation-induced lung injury via intercellular adhesion molecular-1 and transforming growth factorbeta-1 in patients with lung cancer [39].    Although, in the present study, the therapeutic effect of BBR in colorectal and gastric cancer required more evidence, numerous studies have confirmed the gain effect of BBR combined with chemotherapy in recent years. Latest research shows that the combination of the second generation Hsp90 inhibitor NVP-AUY922 and BBR therapy could inhibit a variety of oncogenic signaling pathways of colorectal cancer [40]. Another study showed that BBR as an adjunctive therapeutic agent could attenuate chemical resistance during gastric cancer treatment. The combination of 5-FU and BBR showed synergistic inhibition of survivin and STAT3 levels, thereby enhancing the death of gastric cancer cells [41]. In addition to the 5-FU-based chemotherapy regimen, BBR treatment reduced cisplatin resistance in gastric cancer cells by modulating the miR-203/Bcl-w apoptotic axis [42].
In the present study, body weight index was used to evaluate the growth of experimental animals to indirectly evaluate the adverse effects of BBR. However, studies have shown that BBR could induce weight loss in rodents [43,44] and humans [45,46]. In recent years, research has reported that BBR affected body weight by upregulating AMPK and UCP3 expression to control energy expenditure [47]. Therefore, the toxic side effects of BBR cannot be objectively and accurately evaluated by the change of body weight alone.

Conclusion
BBR exerted anti-tumor effects in a variety of tumors in vivo, especially for breast cancer and lung cancer. However, evidence was still insufficient in colorectal cancer and gastric cancer. One of its anti-tumor mechanisms was anti-angiogenesis. There was a dose-response relationship in the anti-tumor effects. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.