Numerous reports have assessed the relationship between intestinal microbiota (IM) and carcinogenesis in its various forms, including gastric, colorectal and breast cancer^[1-4]. Due to the anatomical and functional connections between the gut and liver, the latter is heavily influenced by intestinal microorganisms and their components^[5]. Moreover, the relationship between hepatocellular carcinoma (HCC) and IM has been discussed extensively in previous reports^[6-8]. Furthermore, IM has been suggested to act as a molecular marker for early non-invasive diagnosis and serves as a therapeutic target in HCC^[8-9].

While the relationship between IM and HCC and the molecular mechanism of IM-induced HCC are well understood, these findings have not been applied to the clinical diagnosis of HCC. In this review, the available information concerning the relationship between IM and HCC and the obstacles to the application of the former to the clinical diagnosis of HCC have been discussed. Moreover, we elucidate the role of IM as a feasible and useful indicator when diagnosing HCC.

1 Relationship between IM and HCC: phenomena and mechanisms

Chronic viral infection and liver cirrhosis are major causes of HCC, and both are closely related to the composition of the gut microbiota^{[5, 7, 10]}. Ren et al.^[8] analyzed 419 samples using high-throughput sequencing of the 16S rRNA gene and found that the fecal microbiota was richer in the phylum Actinobacteria as well as 13 genera, including Gemmiger, Parabacteroides and Paraprevotella, in early HCC as compared with those with liver cirrhosis. Meanwhile, the phylum Verrucomicrobia and 12 genera, including Alistipes, Phascolarctobacterium and Ruminococcus, were significantly decreased in early HCC compared to controls; however, 6 genera, including Klebsiella and Haemophilus, increased. It is worth noting that, although there were significant differences, the intra-group errors were very large^[8]. Ponziani et al.^[11] reported that the fecal microbiota of cirrhotic patients with HCC contained a higher abundance of Enterobacteriaceae and Streptococcus as well as a reduced proportion of Akkermansia compared to controls. Moreover, while Bacteroides and Ruminococcaceae were increased in the former, Bifidobacterium was found to be less abundant. Our previous study comparing HCC patients and healthy controls revealed that Desulfococcus, Enterobacter, Prevotella, Veillonella and many unidentified genera were significantly increased in all stages of HCC, while Cetobacterium was significantly reduced^[12]. These results suggest that the association between the relative abundance of species in IM and HCC remains inconsistent.

Although Toffanin et al.^[13] argued that IM and Toll-like receptor 4 (TLR4) do not contribute significantly to tumor initiation, IM plays a vital role in promoting the development of HCC. Increased intestinal permeability, bacterial overgrowth and the impaired ability of Kupffer cells to remove microbial products, can all lead to increased transport of IM and the metabolites thereof into the liver via portal vein circulation^[14]. Increased intestinal permeability is found in many stages of chronic liver disease and cancer and can lead to increased lipopolysaccharide (LPS) levels in the circulatory system^[15-18]. The translocation of bacterial components (such as LPS, peptidoglycan, and flagellin), otherwise known as pathogen-associated molecular patterns (PAMPs), triggers inflammatory responses via TLRs, the latter of which are found in most chronic liver diseases, including HCC^[5]. TLRs are a class of proteins with the ability to recognize structurally conserved molecules derived from microbiota^[19]. TLR4, a receptor of TLRs, is expressed by Kupffer and hepatic stellate cells (HSCs) as well as hepatocytes within the liver and induces inflammatory responses^[20]. These responses ultimately lead to liver fibrosis which, in turn, can develop into HCC^{[10, 21-22]}. Animal model-based studies have shown that an increase in the levels of bacterial LPS in cirrhotic livers activates HSCs and TLR4 in hepatocytes, thereby leading to fibrosis and regulation of epidermal growth factor secretion which, in turn, triggers tumor proliferation^[7]. Moreover, liver inflammation, fibrosis, and HCC formation were inhibited by blocking TLR4 signal transduction in mice^{[15, 23-24]}. Treatment with the TLR4 agonist LPS has been found to promote HCC development in mice, whereas germ-free status and non-absorbable antibiotics reduced hepatic inflammation, fibrosis, and HCC development^[15]. Moreover, Toffanin et al.^[13] found that continuous administration of low doses of LPS increased tumor number and size in conventional wild-type mice; meanwhile, systemic LPS levels and tumor overgrowth were reduced in sterile mice. Dapito et al.^[23] found that the degree of hepatocarcinogenesis in chronically injured livers depended on the composition of the gut microbiota as well as the extent of TLR4 activation in non-bone- marrow-derived resident liver cells, including both hepatocytes and HSCs. These results were different from those of Yu et al.^[15], who found that Kupffer cells were the main targets of LPS/TLR4 signals in the liver, playing a vital role in the induction of tumor necrosis factor α (TNFα) and IL-6. Significant reductions in cytokine production and complementary proliferation in response to diethylnitrosamine have been reported following the inactivation of Kupffer cells. Regarding HCC promotion, another key mechanism of the LPS-TLR4 axis is to prevent NF-κB B-mediated hepatocyte apoptosis. A negative correlation has been reported between tumor occurrence and the apoptosis marker caspase3 in sterile or TLR4-deficient mice^[23]. Activation of the LPS-TLR4 signaling pathway in Kupffer cells has been found to cause TNF- and IL-6-dependent compensatory hepatocyte proliferation, oxidative stress, and a reduction in apoptosis^[15]. Moreover, LPS appeared to activate TLR4 in HCC cell lines, enhancing their invasiveness and inducing epithelial-mesenchymal transition^[25]. Other LTRs, such as TLR2 and TLR9, have also been associated with IM-induced liver injury. TLR2, TLR4, TLR9 and NLP3 receptors were prevented in experimental hepatic fibrosis in knockout mice^[26]. The increase in TLR4 and TLR9 in portal vein circulation promoted the expression of liver TNFα, which has been found to drive the progression of non-alcoholic steatohepatitis^[27]. Overall, intestinal dysbiosis leads to increased intestinal permeability due to the destruction of the intestinal barrier. This increase in intestinal permeability can promote the occurrence of HCC through the PAMP-TLR- mediated signaling pathway. Liver damage reduces the degradation of endotoxin and leads to endotoxin accumulation, thereby causing liver injury. Therefore, intestinal dysbiosis and HCC form a vicious circle: intestinal dysbiosis induces HCC, while HCC further aggravates intestinal dysbiosis (Figure 1).

Yoshimoto et al.^[28] found that mice consuming a high-fat diet experienced altered composition of the gut microbiota, resulting in increased production of deoxycholic acid, secondary bile acid, and metabolic byproduct of the gut microbiota shown to cause DNA damage. The enterohepatic circulation of deoxycholic acid provoked a senescence-associated secretory phenotype in HSCs. This phenotype led to the production of proinflammatory cytokines and tumor-promoting factors that have been found to promote HCC in the liver upon exposure to chemical carcinogens. Additionally, Ma et al.^[29] reported that gut microbiome-mediated bile acid metabolism appeared to regulate liver cancer through hepatic CXCR6⁺ natural killer T cells.

2 The difficulties faced during clinical diagnosis of HCC using IM

Owing to the distribution heterogeneity of IM as well as its sensitivity to external environmental factors, analysis is challenging, and the results are often inconsistent. Although there are several hundred species of bacteria living in the intestinal tract, their distribution is not uniform. The 50 most abundant bacterial cells often account for more than 70% of the total number of bacterial cells^[12]. This heterogeneity makes it difficult to analyze changes in the composition of intestinal bacteria, especially in the species with low abundance. IM is affected by many factors, including diet^[30-33], sex^[34-36], geographical location^[37], health status^[38-39], and stochasticity^[40]. Huang et al.^[36] found that the IM of liver-specific tuberous sclerosis complex 1 knockout mice differed during the development of HCC based on sex. Specifically, female mice developed IM disorder earlier than did male mice. Lam et al.^[41] suggested that diet could affect the composition of the IM, holding the potential to create a stable, healthy environment for the microbiota in the long term. He et al.^[37] reported that regional differences in IM profoundly limited its use in clinical disease diagnosis. While factors such as sex can easily be excluded in clinical diagnosis, this is not true of all factors. For this reason, the patterns observed in animal models are not completely consistent with clinical results, making it difficult to apply such findings to the clinical auxiliary diagnosis of HCC.

Rodent models have previously been used to study the associations between gut microbiota and various host diseases^{[36, 42]}. Usually, interfering factors, such as diet^{[2, 30, 32, 43]}, lifestyle^[44-45] and genetic background^[46-48], can be eliminated using animal models. Therefore, a positive result is more easily obtained in animal studies. However, caution is warranted when applying associations between gut microbiota and host diseases in animal models to clinical diagnoses. Clinically, it is very difficult to eliminate the interference of uncontrolled factors. The significant difference between the gut microbiota of rodent models and humans could be attributed to experimental errors^[49].

3 A potential way to use IM for diagnosis of HCC

Using inclusive, comprehensive indicators instead of a single index may be a way to resolve the issues regarding the use of IM in the auxiliary diagnosis of HCC. Previous studies have used alpha diversity indices, the Firmicutes/Bacteroidetes and Bifidobacterium/Enterobacteriaceae ratios, and other indicators to analyze the association between IM and liver diseases. Wong et al.^[50] suggested that the Firmicutes/Bacteroides ratio in feces could be used as a diagnostic indicator of nonalcoholic steatohepatitis. Lu et al.^[51] proposed that the Bifidobacteria/Enterobacteriaceae ratio was indicative of the degree of biological imbalance present in the development of liver diseases. While positive results are indeed possible to obtain, it appears to be difficult to do so consistently. To circumvent this issue, we used a novel comprehensive indicator, namely the degree of dysbiosis, to measure the extent of the IM disorder. We then analyzed the correlation between the degree of dysbiosis and the extent of the deterioration caused by HCC. The results showed that the index was superior to the aforementioned indicators^[12]. However, the indicator requires further refinement using subjects with differing IM bacterial compositions and HCC locations from which representative bacterial taxa can be obtained for the calculation of the degree of dysbiosis.

To enable the establishment of a comprehensive indicator, we offer the following suggestions.

(1) To ensure the universality of the final results, the samples used for the calculation should be selected from as broad a selection of locations as possible.

(2) Because many bacteria in the IM are difficult to culture and may not even have an official scientific name, it is preferable to use DNA sequences instead of species names as molecular markers. While this approach would likely be more troublesome, it is more convenient for the subsequent use of gene chips and other technical methods when investigating the relative abundances of each species.

(3) Although this review focused on the use of IM in the diagnosis of HCC, it is preferable to investigate the molecular markers of IM that possess the ability to distinguish between different cancers during screening to develop a comprehensive indicator, as the patterns of change in IM may vary based on the type of cancer^[52]. The advantage of this approach is that the results obtained can also be used in the auxiliary diagnosis of other cancers.

(4) The formula warrants further refinement. We initially calculated the formula for the degree of dysbiosis as the difference between the sums of the logarithms of the relative abundances of 13 potentially harmful and 7 inherently pathogenic bacterial genera, respectively^[12]. However, the results may be more pronounced when their respective ratios are used. It is generally believed that other metabolic diseases, such as obesity, type 2 diabetes, hypertension, irritable bowel syndrome, and steatohepatitis, are important factors that lead to chronic liver diseases and eventually develop into HCC^[53]. Therefore, we should consider adding the intestinal microorganisms related to these diseases (as we showed in Table 1) into the calculation formula of the comprehensive indicator. This may be more helpful for the early diagnosis of HCC. Among these bacteria, Bifidobacterium and Lactobacilli protect against pathogen-induced NF-κB activation^[54], thus inhibiting the local or systemic inflammatory response induced by endotoxin produced by Gram-negative bacteria, such as Bacteroides, Enterococcus and Escherichia, through the NF-κB pathway^[55-57]. Besides, considering the heterogeneity of relative abundance of bacteria in intestinal microbiota, different molecular markers can be weighted to different extents.

4 Conclusion

There is a close relationship between IM and HCC. The use of a comprehensive indicator of IM will enable the establishment of a unified index for clinical auxiliary diagnosis. However, this indicator offers scope for further refinement in future studies.