Thomasclavelia ramosa and alcohol-related hepatocellular carcinoma: a microbial culturomics study

Hepatocellular carcinoma (HCC) is the most common form of liver cancer (90% of cases), primarily associated with chronic liver disease. Its major risk factors include chronic viral hepatitis infection (HBV, HCV), Metabolic dysfunction-associated Steato-Hepatitis (MASH) and Alcoholic Liver Disease (ALD) [1]. Liver cancer ranks sixth for incidence and third for mortality across diverse cancers, which has caused a great cancer burden globally [2].

The link between alcohol, liver disease, and cancer is well-established. Its mechanism would include direct toxicity of alcohol on the liver, but persistent instrumental factors are suspected, as the evolution of the disease is not reversible upon withdrawal [3]. The gut-liver axis and gut microbiota are potential candidates to explain the persistence of a vicious circle that would explain the persistent excess risk up to 10 years after weaning [4]. Recent studies suggest that alcohol dependence syndrome and ALD are both associated with gut microbiota alteration with distinct features [5, 6]. This indicates that a specific and persistent gut dysbiosis may enhance cirrhosis and hepatocarcinogenesis by the gut-liver axis. Indeed, some data from human studies [7‐12] and animal experimental models [13‐18] indicate that HCC occurrence is related to gut microbiota, and treatment with broad-spectrum antibiotics decreases HCC tumor growth in mice [19].

Several metagenomic studies have investigated gut dysbiosis in patients with ALD based on 16 S rRNA amplicon sequencing [20‐22] or shotgun (whole-genome) sequencing [6]. However, no study described gut dysbiosis in patients with ALD or ALD-associated HCC (ALD-HCC) based on the culturomics approach. It has been shown that the reliability of metagenome assembled genomes (MAGs) obtained through deep sequencing is limited and biased by the production of erroneous MAGs compared to isolation especially for the handful of MAGs that are critical for the study of pathogens [23]. This evidences that culture remains essential and complementary to sequencing when investigating gut pathogens associated with liver cancer.

Culture-based studies focusing on intestinal dysbiosis and liver cancer are scarce. We found a study reporting cultured microbial counts but with minimal microbial taxonomic accuracy and evidencing an Escherichia coli enrichment [7]. Microbial culturomics is a new -omics strategy developed in our center as a high-throughput culture method based on Matrix-Assisted Laser Desorption Ionization Time-Of-Flight Mass Spectrometry (MALDI-TOF MS) and diversified physicochemical culture conditions mimicking the natural microenvironment [24]. Our team applied fungal and bacterial culturomics in the context of liver diseases, demonstrating the importance of ethanol-producing yeast and bacteria in MASH and HBV-associated liver disease [25, 26].

However, to our knowledge, no study has yet characterized microbiota associated with ALD using microbial culturomics. We, therefore, decided to carry out this microbial culturomics study to complete the spectrum of dysbiosis of the gut microbiota, whose instrumental role had been demonstrated in ALD with experimental evidence [27]. Accordingly, this study aims to characterize the microbial signature in patients with ALD and ALD-associated HCC using both culturomics and large-scale sequencing (v3v4 region 16 S rRNA amplicon sequencing).

Full methods are reported in the supplementary data (Supplementary Material 1). Briefly, 59 participants were investigated, including 35 patients with ALD (19 with HCC) and 24 controls (CTL). Liver stiffness measurements in cirrhotic patients were conducted using a FibroScan^® instrument (Echosens, Paris, France). Routine biochemistry, including prothrombin index, platelets count, total bilirubin, serum albumin, alanine aminotransferase, aspartate aminotransferase, gamma-glutamyl transferase, alkaline phosphatase and serum creatinine, were measured. These analyses were performed only among cases. The fecal samples from all participants were collected and analyzed by v3-v4 16 S rRNA amplicon sequencing. We selected 11 samples from patients with ALD and 10 samples from controls for microbial culturomics. For culturomics, 21 samples were selected as this approach is of unequaled value since it is the only one that ascertains viable strains, but it is associated with a huge workload (6 weeks by sample). Statistical analyses were performed using GraphPad Prism Software for Windows (GraphPad Software, San Diego, CA, USA) (version 9.0). P values less than 0.05 were considered significant differences.

This brief preliminary report, which uses a combined approach of microbial culturomics and 16 S rRNA sequencing, identifies T. ramosa as one of the best candidates for the intestinal pathogenic oncobiont associated with ALD and HCC associated with ALD. T. ramosa is a human gut pathogen associated with several cases of severe infection (bacteremia, infection of aortic aneurysm, osteomyelitis, arthritis, gas gangrene, Fournier’s gangrene, fatal infections) recently associated with human cancer, notably colorectal cancer [33, 34], but also HCC in a 2023 Chinese study [35].

One of the first discovered genotoxin produced by a gut commensal was colibactin from Escherichia coli strains which alkylates DNA, and associated with colon cancer [36]. A another recent study reported that two other gut commensals, Morganella morganii and T. ramosa, exhibited genotoxicity and promoted tumor growth in an experimental model [37]. While indolimines have been characterized as the genotoxic molecules for M. morganii, the small molecules (< 3 kDa) that lead to the genotoxic and pro-cancer role are not known for T. ramosa [37]. Notably, T. ramosa lacks known biosynthetic gene clusters (BGCs) and induces DNA damage via colibactin-independent and indolimines-independent mechanisms [37].

In this study, among the 2 species associated with HCC by sequencing (59 samples), only T. ramosa was identified in culture (13/21 samples, 11/11 in cancer vs. 2/10 in controls). Gemmiger formicilis was not isolated in any sample. Even if we did not succeed in cultivating it here, it cannot be ruled out that G. formicilis is also associated with liver cancer. Indeed, G. formicilis has been associated with colon cancer [38]. Moreover, in a metagenomic meta-analysis on inflammatory bowel disease (IBD), 3 species were associated with IBD: Asaccharobacter celatus, Gemmiger formicilis, and Erysipelatoclostridium ramosum (former name of T. ramosa) [39]. This association of G. formicilis and T. ramosa enriched in a disease associated with an excess risk of cancer (IBD and colon cancer) is consistent with our results.

Limitations of our study included a small sample size and lack of experimental evidence. However, the clear T. ramosa signature (LDA score > 5, p < 0.05) identified here with a limited sample size means that the difference (effect size) is huge, supporting the strength of the association as the first Bradford-Hill criteria for causality. Consistency and reproducibility are fulfilled by two previous studies confirming the association with CRC [33, 34] and the confirmation of the T. ramosa-liver cancer association by another team [35]. The linear discriminant analysis fulfills the biological gradient (the higher the T. ramosa 16 S rRNA number of reads - the higher the risk of HCC). The plausibility and experimental evidence are supported by the recently reported in vitro genotoxicity and in vivo tumor growth-promoting properties of T. ramosa [37].

The robustness of our results is not due to the number of samples or deep sequencing, but to microbial culturomics, the only approach to certify the presence of viable and biologically active microbes with high fidelity taxonomic characterization at the species level thanks to MALDI-TOF MS. Studies with higher statistical power could identify other microbial candidates, and T. ramosa is very likely part of a group of liver cancer oncobionts. Future studies could use deep sequencing and automated microbial culturomics [40] to focus on a larger sample to obtain a better match or at least larger groups, allowing for the stratification of all comorbidities and the stage of liver disease. Robustness is also confirmed by the fact that we found that T. ramosa was the most different in terms of culturomics between ALD and controls after analysis of 21 samples and 32,088 colonies, and the most significantly associated with liver cancer by DNA sequencing after analysis of 59 samples and 2,840,773 reads.

Even if future studies with larger sample size and better representation of participant groups according to the comorbidities, cause, and stage of liver disease are necessary to confirm the association and the potential role of T. ramosa in hepatocellular carcinoma, we believe that our results and recent literature [37] evidenced that T. ramosa could be a potential candidate among the gut commensals favored by alcoholism, and which can then contribute to liver carcinogenesis through its recently demonstrated genotoxic and tumor growth-promoting properties.