Profiling of the tumor-associated microbiome in patients with hepatocellular carcinoma

Hepatocellular carcinoma (HCC) accounts for 75–85% of malignant liver tumors and is the third leading cause of cancer death. In the complex sequential cascade of hepatocarcinogenesis, which mostly evolves from chronic inflammation with cirrhotic transformation of the liver parenchyma, an important role has more recently been assigned to the gut microbiota and its detection in the tumor micromilieu [1‐4]. The gut microbiota refers to the diverse community of microorganisms including bacteria and fungi, residing in the gastrointestinal tract [5]. This community of microorganisms supports health by modulating (among others) the digestion, the immune system function, and the defense against pathogens. It also produces compounds like short-chain fatty acids that strengthen the gut barrier and reduce inflammation. Diet, medications, and environment shape its composition, making it vital for overall health [6, 7]. Microbiota can nevertheless contribute also to tumorigenesis through several direct or indirect mechanisms. It has been shown that a microbial dysbiosis can drive cellular senescence [8], chronic inflammation, induce the production of carcinogenic metabolites [9], influence the tumor microenvironment [10] and contribute to drug resistance [11].

The potential routes of entry for the gut microbiota into the liver are the portal vein, through the draining of the intestine, and the bile duct system. Under normal conditions, bacteria modulate tissue regeneration and repair, contributing to maintaining liver homeostasis via a process known as the gut‒liver axis [12, 13]. However, microbiota dysbiosis can play a role in the inhibition of normal tissue regeneration and the induction of liver failure and cancer [14‐17]. In the latter case, the presence of micriobiota in the tumor microenvironment (TME) can affect the mechanism of immunosurveillance [18], affecting the immune cell recruitment [19], suppressing the anti-tumor activity of the infiltrating cells and enabling the relase of pro-tumorigenic cytokines [10]. The presence of microbiota in the TME may facilitate disease progression, promote metastasis development and affect the response to therapy [20, 21]. Bacteria can drive tumor progression through various pathways, including DNA damage, the activation of oncogenic signals and the modulation of the immune response [22‐27]. Recent research has also demontsrated the role of microbiome in cellular senescence [8]. One example is given by Helicobacter pylori which upregulates pathways such as NF-kB and p53-p21, accelerating cellular senescence. The release of a pro-inflammatory secretome by senescent cells has been shown to play a pro-tumorigenic effect of non-senescent cells. Also Enterococcus is recognized to be associated with tumorigenesis, especially through the activation of ROS production and the activation of protumorigenic pathways such as EGFR [4]. The characterization of the bacteria in patients with HCC could offer valuable insights into the role of the microbiota in this malignancy [21, 28‐32]. The presence of bacteria colonizing the human gastrointestinal tract, such as Helicobacter, Enterococcus, Bacteroides or Fusobacterium, has already been shown in fresh HCC tissues, as well as in fresh adjacent tumor tissue [33]. However, the bacterial communities found in formalin fixed-paraffin embedded (FFPE) tissues differ considerably from those found in fresh samples, revealing that bacteria in FFPE samples are not frequently found in the human gastrointestinal tract [31, 34, 35]. Despite increasing evidence that the tumor microbiota may influence HCC progression and outcomes, significant knowledge gaps remain. Most studies have focused on gut or blood samples, while direct analysis of bacterial communities in HCC tumor tissue—especially using FFPE samples—has been limited by technical challenges and incomplete reference databases. Additionally, the heterogeneity of tumor-associated microbiota and unclear functional roles further complicate biomarker discovery. Therefore, systematic studies are needed to precisely characterize the tumor microbiota in relation to HCC stage and survival, ultimately aiding in the identification of clinically relevant microbial biomarkers. The aim of our explorative study was to characterize the bacterial microbiota in the tumor tissue of patients with HCC according to stage and overall survival (OS) as well as in nontumor FFPE tissue where available. We developed a systematic analysis pipeline incorporating detailed taxonomic annotation and a comprehensive database to enable precise identification of taxa in FFPE samples and aid in detecting potential biomarkers in HCC patients.

In this explorative study of tumor tissues obtained from patients with early- and advanced-stage HCC, we identified a heterogeneous microbiota profile, where the predominant bacteria in both tumor and adjacent tissues were those typically found in the UGI tract [43]. Although the small cohort size did not permit the use of more robust statistical methods, the application of PCA and indicator species analysis allowed the identification of several genera and species that may be associated with the variables under investigation. The results were consistent with those of prior studies [21] and highlight marked dysfunction in the gut‒liver axis, especially the UGI‒liver connection. Bacilli was the most relevant genus identified. In this class and across all species, the most notable and unexpected findings were four species belonging to the genus Streptococcus: S. mitis, S. oralis, S. thermophilus and S. anginosus. All of them are considered viridans group streptococci (VGS), with S. mitis and S. oralis being the predominant species belonging to Streptococcus, which causes severe clinical disease in cancer patients [51]. Furthermore, a recent report indicated that the secretion of antigens by S. anginosus is a proinflammatory mechanism contributing to the pathogenesis of several cancers, including HCC [52]. It should be noted that while immunohistological examination of the tissues revealed the presence of both Gram-positive and Gram-negative bacteria, our dataset indicates that the majority of Gram-negative bacteria were ultimately classified as contaminants (e.g., Acinetobacter lwoffii, which is commonly found on the skin). It is important to acknowledge the potential limitations of immunohistochemical identification of bacterial components in FFPE tissues. In particular, the use of anti-LPS antibodies to identify Gram-negative bacteria may yield false-positive results due to the persistence of LPS molecules in tissue. LPS is a structurally stable component of the outer membrane of Gram-negative bacteria and can remain in tissues long after the bacteria themselves have died and lysed. Consequently, positive LPS staining does not necessarily indicate the presence of intact or viable bacteria but may instead reflect residual bacterial debris. This biological persistence may account for the observed discrepancy between broad IHC detection of Gram-negative bacteria and their classification as likely contaminants in sequencing data. Furthermore, non-specific binding of antibodies, especially in inflamed or necrotic tissue regions, may contribute to background staining. To address these challenges and more accurately interpret bacterial presence in FFPE samples, multimodal validation using complementary techniques such as fluorescence in situ hybridization (FISH) or 16 S rRNA-targeted probes would be advisable to confirm the identity and spatial distribution of bacteria and reduce the risk of overinterpretation based on IHC alone.

Direct comparisons between tumor and adjacent nontumor tissues revealed the presence of H. pylori exclusively within the tumoral compartment. Although the liver is as atypical niche for H. pylori [53], the detection of DNA isolated from this species in both palliative and curative HCC patient groups aligns with previous findings [54, 55]. Additionally, the comparison between curative and palliative patients’ group identified a larger abudance of S. aureus in the first group. Moreover, the S. aureus abundance was greater in patients whose OS exceeded 16 months. S. aureus can be linked to chronic inflammation, which can foster a pro-tumorigenic microenvironment, as well as to acute inflammation which on the contrary promotes an anti-tumorigenic microenvironment [56]. Our results in HCC need therefore further clarification. In few patients we were able to analyse the microbiome in tumoral and adjacent tissues. Notably, bacteria were detected in both compartments, suggesting damage to the entire organ [51, 57‐72]. The comparison between tumor and nontumor tissues also highlights tissue tropism [73]. For example, Campylobacteria and Bacteroida were exclusively identified within tumor tissues, whereas Alphaproteobacteria were detected only in nontumor regions. Additionally, some tissues were colonized by a single species, whereas others contained multiple species. Given the small cohort size, we must consider this result as preliminary. However, tissue tropism likely plays a role in liver pathogenesis, and a larger cohort study could clarify its potential correlation with clinical outcome. The data used in this study were obtained from archival tissues. Sequencing and taxonomic annotation revealed numerous sequences unlikely to represent genuine liver colonizers, as these species are absent in human UGI/LGI tracts. To further analyze this discrepancy, we developed a novel methodology, yielding two key findings: in certain FFPEs (e.g., SOR75), species richness remained stable, implying limited infiltration; in others (e.g., SOR43, SOR95), species richness increased significantly, suggesting either multi-bacterial invasion or possible contamination. Currently, there are insufficient available data to distinguish between these two results. Although FFPE samples are among the largest and most valuable resources in biobanking, it is important to acknowledge that, in most cases, they were collected and preserved under non-sterile conditions. As a result, microbiome analysis of these samples does not follow a straightforward pipeline, and careful consideration must be given to the potential origin of the detected bacterial species. Furthermore, due to the preservation and DNA extraction protocols, some underestimation of high GC content DNA sequences cannot be entirely excluded [74]. Taking the previous considerations into account, it is still noteworthy that both SOR43 and SOR95, each with a richness of approximately 35, were associated with necrotic liver tissue (data not shown). Further research is needed to explore more comprehensively the possibility of an association between bacterial richness and necrosis. Moreover, new models are required to elucidate whether bacteria are the cause or the consequence of hepatocellular carcinoma linking the gut-liver axis, similar to the zebrafish model [75], which has recently been proposed as a model for investigating the gut–brain axis [75, 76]. Recent research has highlighted the significant impact of microbial communities on liver health and disease progression. The influence of certain bacteria on the liver microenvironment is increasingly well understood. For instance, S. mitis [77] has been implicated in the development of pyogenic liver abscesses, which can compromise local immune defenses and create a milieu more susceptible to malignant transformation. By weakening immune surveillance, these abscesses may facilitate the emergence and progression of cancerous lesions. Additionally, H. pylori has been shown to induce significant liver injury by triggering robust inflammatory responses [55]. In the context of chronic liver diseases such as cirrhosis, this persistent inflammation can accelerate cancer progression and worsen patient outcomes. Collectively, these findings underscore the pivotal role of microbial factors in shaping the liver microenvironment and influencing the trajectory of liver disease and hepatocarcinogenesis. On another front, although dysbiosis has been strongly associated with carcinogenesis, microbiome-based therapies have demonstrated considerable potential in cancer treatment [5]. For example, fecal microbiota transplantation and the administration of live probiotics are well-established strategies in oncology [78]. In HCC, however, few studies have focused specifically on probiotic interventions. A retrospective study involving 1,267 patients with hepatitis B virus (HBV) infection undergoing antiviral therapy showed that probiotic use was associated with a reduced risk of developing HCC [79]. Furthermore, preclinical studies in animal models have demonstrated that probiotics can inhibit tumor growth and reduce tumor size by modulating the immune response [80]. Overall, probiotics exert beneficial effects by modulating gut-liver axis signaling, suppressing pro-tumorigenic inflammation, and reducing HCC risk in a dose-dependent manner among high-risk populations. While further clinical validation is required, probiotics represent a safe, cost-effective adjunct to standard HCC therapies. We acknowledge some limitations in the current study, which also highlight important opportunities for future improvement. The size of the cohort was small; therefore, validation studies are necessary to confirm these preliminary results. The nature of FFPE samples presents inherent challenges, as they typically yield low DNA content. In our case, successful PCR amplification was achieved in approximately 30% of the total samples. Then, the FFPE samples are not usually collected or preserved under sterile conditions, which may influence microbial profiles. Nonetheless, the ability to extract and analyze microbial DNA from such archived material demonstrates the feasibility of microbiome studies in FFPE tissues and opens the door for future methodological optimization and broader applications, even in retrospective clinical cohorts.

Understanding the influence of the “dysbiotic” microbiome in liver carcinogenesis is crucial and presents an opportunity to leverage microbiome profiling as a predictive tool for evaluating response to oncological immunotherapy. Furthermore it may direct research towards the modulation of microbiota in the therapy of malignant liver disease.