Introduction
Globally, over 2.0 million new colorectal cancer (CRC) cases and 957,147 deaths are estimated to occur in 2025. Among cancers, CRC ranks second both for incidence and for mortality, despite progressive adoption of screening approaches. CRC incidence rates are about 3-fold higher in transitioned versus transitioning countries, although average case fatality remains higher in lower Human Development Index settings (Bray et al., Reference Bray, Ferlay, Soerjomataram, Siegel, Torre and Jemal2018). However, even in Western Countries, up to 20%–30% of cases continue to receive clinical attention only after the onset of symptoms (Runkel et al., Reference Runkel, Schlag, Schwarz and Herfarth2005; Decker et al., Reference Decker, Lambert, Nugent, Biswanger, Samadder and Singh2020), underlining a suboptimal adherence to screening programmes. Intriguingly, CRC, as a multifactorial disease, is associated with a range of individual and environmental risk factors, including patient demographics, genetic background, and lifestyle, all of which variably contribute to the risk of disease development (Robertson & Ladabaum, Reference Robertson and Ladabaum2019).
Genomic landscape of colorectal cancer
Over the past three decades, molecular genetics allowed the dissecting of the main pathways of gene damage implicated in CRC development. Eventually, the advent of next-generation sequencing (NGS) further differentiated CRC into molecular subsets. According to the pre-NGS classification, three main CRC subtypes with different clinical behaviour were defined: chromosomal instability (CIN), microsatellite instability (MSI), and the CpG island methylator phenotype (CIMP). Two major types of genomic instability were initially recognized as alternative mechanisms of CRC carcinogenesis: CIN, which occurs in 80–85% of CRCs (Grady & Carethers, Reference Grady and Carethers2008) and accounts for “the suppressor pathway,” characterized by aneuploidy, chromosome amplifications, and deletions (Vogelstein et al., Reference Vogelstein, Fearon, Kern, Hamilton, Preisinger, Nakamura and White1989; Peinado et al., Reference Peinado, Malkhosyan, Velazquez and Perucho1992; Smith et al., Reference Smith, Carey, Beattie, Wilkie, Lightfoot, Coxhead, Garner, Steele and Wolf2002). The gatekeeper of this pathway is the APC gene, which becomes inactivated according to the classic two hits model, through double somatic events in sporadic cases, or moving by a transmissible germline pathogenic variant in inherited cases in familial adenomatous polyposis coli (i.e. FAP) (Yan et al., Reference Yan, Dobbie, Gruber, Markowitz, Romans, Giardiello, Kinzler and Vogelstein2002). In contrast, MSI CRCs are characterized by defects of the DNA mismatch repair (MMR) system, which lead to the progressive accumulation of unrepaired mutations (Walther et al., Reference Walther, Johnstone, Swanton, Midgley, Tomlinson and Kerr2009). These tumours stay on their own under many respects. As to pathogenesis, approximately 20% arise in the context of Lynch syndrome, an inherited autosomal dominant syndrome due to germ-line mutations in one of the MMR genes (i.e. MLH1, MSH2, MSH6, PMS2, and EPCAM) (Laghi et al., Reference Laghi, Bianchi and Malesci2008; Dal Buono et al., Reference Dal Buono, Gaiani, Poliani, Correale and Laghi2021). The remaining fraction of sporadic MSI CRCs arises through the somatic inactivation of MMR genes, most commonly via MLH1 promoter hypermethylation, thereby accounting for a subset of the cases with the CIMP phenotype (Boland et al., Reference Boland, Thibodeau, Hamilton, Sidransky, Eshleman, Burt, Meltzer, Rodriguez-Bigas, Fodde, Ranzani and Srivastava1998). Phenotypically, they have a preferential proximal location and are marked by a robust immune response. Clinically, their low metastatic potential (Malesci et al., Reference Malesci, Laghi, Bianchi, Delconte, Randolph, Torri, Carnaghi, Doci, Rosati, Montorsi, Roncalli, Gennari and Santoro2007) is coupled with variable responsiveness to chemotherapy but unique responsiveness to immunotherapy, namely to PD-L1 blockade, both in adjuvant and neo-adjuvant settings (Greco et al., Reference Greco, Rubbino, Dal Buono and Laghi2023). Accordingly, they are the first type of solid cancer amenable to non-surgical treatment with curative intent (Taieb et al., Reference Taieb, Svrcek, Cohen, Basile, Tougeron and Phelip2022).
As to precursor lesions, while Microsatellite-Stable (MSS)/CIMP-negative CRC develops through the classical adenoma-carcinoma sequence, MSI and CIMP CRCs may arise through the alternative “serrated pathway” (Jass, Reference Jass2007; Kriegl et al., Reference Kriegl, Vieth, Kirchner and Menssen2012), preferentially taking place in the right colon (Gaiser et al., Reference Gaiser, Meinhardt, Hirsch, Killian, Gaedcke, Jo, Ponsa, Miró, Rüschoff, Seitz, Hu, Camps and Ried2013).
With the advent of NGS, molecular classifications evolved to identify gene expression signatures that can help stratify tumours based on their outcome and responsiveness to treatment. An approach trying to facilitate clinical translation eventually resolved initial classification inconsistencies, leading to a consensus on four main expression signaturesalso known as consensus molecular subtypes (i.e. CMS1 through CMS4). CMS1 features MSI and elicits a strong immune response. CMS2 and CMS4 show high levels of CIN, while CMS3 also includes a fraction of hypermutated tumours. CMS2, or the canonical subtype, displays epithelial features and classic WNT pathway activation. CMS3 also again has epithelial features but is notably defined by marked metabolism alterations. In contrast, CMS4 distinguished by mesenchymal and stromal infiltration signatures, altogether TGF-β activation, and downregulation of the miR-200 family (Guinney et al., Reference Guinney, Dienstmann, Wang, de Reyniès, Schlicker, Soneson, Marisa, Roepman, Nyamundanda, Angelino, Bot, Morris, Simon, Gerster, Fessler, De Sousa, Melo, Missiaglia, Ramay, Barras and Tejpar2015). The latter subtype was also identified by a different approach focused on cancer-cell intrinsic transcriptional features (or CRIS) where it was termed CRIS-B (Cancer Genome Atlas Network, 2012; Guinney et al., Reference Guinney, Dienstmann, Wang, de Reyniès, Schlicker, Soneson, Marisa, Roepman, Nyamundanda, Angelino, Bot, Morris, Simon, Gerster, Fessler, De Sousa, Melo, Missiaglia, Ramay, Barras and Tejpar2015; Isella et al., Reference Isella, Brundu, Bellomo, Galimi, Zanella, Porporato, Petti, Fiori, Orzan, Senetta, Boccaccio, Ficarra, Marchionni, Trusolino, Medico and Bertotti2017).
Notwithstanding, germline mutations are increasingly reported in CRC patients following the widespread use of NGS, approximately 10% of unselected patients (Yurgelun et al., Reference Yurgelun, Kulke, Fuchs, Allen, Uno, Hornick, Ukaegbu, Brais, McNamara, Mayer, Schrag, Meyerhardt, Ng, Kidd, Singh, Hartman, Wenstrup and Syngal2017), and up to 16% of juvenile (i.e. younger than 50 years of age) cases (Pearlman et al., Reference Pearlman, Frankel, Swanson, Zhao, Yilmaz, Miller, Bacher, Bigley, Nelsen, Goodfellow, Goldberg, Paskett, Shields, Freudenheim, Stanich, Lattimer, Arnold, Liyanarachchi, Kalady and Hampel2017; Poliani et al., Reference Poliani, Greco, Barile, Buono, Bianchi, Basso, Giatti, Genuardi, Malesci and Laghi2022). An additional 25% of cases have a positive family history with no clearly identifiable pathogenic variants and are referred to as common familial CRC, which likely recognizes polygenic bases (Kastrinos et al., Reference Kastrinos, Samadder and Burt2020). Altogether, genetic predispositions and increased CRC risks associated with demographics and lifestyle call for a personalization of the screening approach (Jeon et al., Reference Jeon, Du, Schoen, Hoffmeister, Newcomb, Berndt, Caan, Campbell, Chan, Chang-Claude, Giles, Gong, Harrison, Huyghe, Jacobs, Li, Lin, Le Marchand, Potter and Hsu2018; Robertson & Ladabaum, Reference Robertson and Ladabaum2019).
Besides the above-mentioned differences, growing evidence suggests that intestinal microbiota also plays a role in the onset and progression of CRC. The human intestinal microbiome comprises a complex community of bacteria, archaea, viruses, and eukaryotes specific to each individual and stable in healthy individuals (Song et al., Reference Song, Chan and Sun2020; Tiamani et al., Reference Tiamani, Luo, Schulz, Xue, Costa, Khan Mirzaei and Deng2022). Within such complex community, an increasing number of studies has shown that bacteria, chiefly Fusobacterium nucleatum (Fn), may contribute to CRC development through multiple mechanisms, including the interaction with host immune system, the production of cancer-associated metabolites, as well as the release of genotoxic virulence factors (Gholizadeh et al., Reference Gholizadeh, Eslami and Kafil2017; Alexander et al., Reference Alexander, Scott, Pouncey, Marchesi, Kinross and Teare2018; Hashemi Goradel et al., Reference Hashemi Goradel, Heidarzadeh, Jahangiri, Farhood, Mortezaee, Khanlarkhani and Negahdari2019; Justesen et al., Reference Justesen, Nielsen, Jensen, Dessau, Møller, Coia, Andersen, Pedersen and Gradel2022; Genua et al., Reference Genua, Butt, Waterboer and Hughes2023; Kajihara et al., Reference Kajihara, Yahara, Kitamura, Hirabayashi, Hosaka and Sugai2023; Kong et al., Reference Kong, Zhang, Xiang, You, Duan, Zhao, Li, Wu, Zhang, Zhou and Duan2023; Robinson & Allen-Vercoe, Reference Robinson and Allen-Vercoe2023; Russo et al., Reference Russo, Gloria, Nannini, Meoni, Niccolai, Ringressi, Baldi, Fani, Tenori, Taddei, Ramazzotti and Amedei2023). These interactions may also result in specific molecular signatures of CRC (Pleguezuelos-Manzano et al., Reference Pleguezuelos-Manzano, Puschhof, Rosendahl Huber, van Hoeck, Wood, Nomburg, Gurjao, Manders, Dalmasso, Stege, Paganelli, Geurts, Beumer, Mizutani, Miao, van der Linden, van der Elst, Ambrose, Arumugam and Clevers2020). The association of Fn with CRC is currently supported by translational evidence, and its pro-tumourigenic role is addressed by experimental models (Kostic et al., Reference Kostic, Chun, Robertson, Glickman, Gallini, Michaud, Clancy, Chung, Lochhead, Hold, El-Omar, Brenner, Fuchs, Meyerson and Garrett2013; Nakatsu et al., Reference Nakatsu, Li, Zhou, Sheng, Wong, Wu, Ng, Tsoi, Dong, Zhang, He, Kang, Cao, Wang, Zhang, Liang, Yu and Sung2015).
Discover Fn in neoplastic lesions of the colon
The seminal reports concerning the enrichment of Fusobacterium genome in CRC lesions appeared as companion papers in Genome Research in 2012. The two papers are similar in their approach, results, and conclusions. The study by Castellarin et al. used RNA-sequencing (after host genome subtraction) to detect an over-representation of Fn sequences in CRC tissues as compared to matched-normal mucosa samples. Thereafter, Fn overabundance in CRC was confirmed by quantitative PCR analysis (Castellarin et al., Reference Castellarin, Warren, Freeman, Dreolini, Krzywinski, Strauss, Barnes, Watson, Allen-Vercoe, Moore and Holt2012). The other paper, by Kostic et al., used whole-genome sequencing followed by a computational subtraction to identify microbial genomes, which detected Fn as the enriched metagenome along with a reduced identification of Bacteroidetes and Firmicutes phyla. Amplification of 16S rDNA and pyrosequencing confirmed the higher load of Fn sequences in tumour specimens, as did in situ hybridization (Kostic et al., Reference Kostic, Gevers, Pedamallu, Michaud, Duke, Earl, Ojesina, Jung, Bass, Tabernero, Baselga, Liu, Shivdasani, Ogino, Birren, Huttenhower, Garrett and Meyerson2012). Both studies also identified a positive association with lymph node metastasis and with the persistence of Fn sequences in distant metastases (in one-fifth of the cases) (Castellarin et al., Reference Castellarin, Warren, Freeman, Dreolini, Krzywinski, Strauss, Barnes, Watson, Allen-Vercoe, Moore and Holt2012; Kostic et al., Reference Kostic, Gevers, Pedamallu, Michaud, Duke, Earl, Ojesina, Jung, Bass, Tabernero, Baselga, Liu, Shivdasani, Ogino, Birren, Huttenhower, Garrett and Meyerson2012).
In a later paper, Kostic et al. confirmed the enrichment of Fusobacteria in CRC as well as in adenomas and stool samples from patients with colonic neoplastic lesions. In parallel, they showed that Fn potentiates gut tumour formation in a mouse model made prone to bowel tumourigenesis through APC engineering, confirming the increase of Fn load in tumour specimens. Fn promotes tumour development by recruiting myeloid-derived suppressor cells (MDSCs), while simultaneously inhibiting T-cell proliferation and inducing their apoptosis within the tumour microenvironment. MDSCs express galectin-9, which binds to T-cell immunoglobulin and mucin domain-containing protein 3 (TIM-3) on lymphocytes, leading to T-cell apoptosis (Dong et al., Reference Dong, Strome, Salomao, Tamura, Hirano, Flies, Roche, Lu, Zhu, Tamada, Lennon, Celis and Chen2002; Kostic et al., Reference Kostic, Chun, Robertson, Glickman, Gallini, Michaud, Clancy, Chung, Lochhead, Hold, El-Omar, Brenner, Fuchs, Meyerson and Garrett2013). Interestingly, the metabolism of gut microbiota drives the transformation of intestinal cells not only in APC-derived models but also in MMR-deficient models (namely MSH2-deficient) (Belcheva et al., Reference Belcheva, Irrazabal, Robertson, Streutker, Maughan, Rubino, Moriyama, Copeland, Surendra, Kumar, Green, Geddes, Pezo, Navarre, Milosevic, Wilson, Girardin, Wolever, Edelmann and Martin2014).
The replication of Castellarin’s work later took place within a broader context of reproducibility in oncological research, yielding negative results (Repass et al., Reference Repass, Maherali and Owen2016; Repass & Reproducibility Project: Cancer Biology, 2018). The identification of microbial genomes comes with obvious challenges, including the methodology, the targeted nucleic acids, and their quantification. In tumour specimens, variability sources embrace the stage and molecular subtypes, which may be associated with different rates of positive findings. The detection of Fn is technically challenging and involves multiple molecular targets, including Fn 16S ribosomal-RNA (rRNA), NusG, FadA, rpoB, and methodological approaches, including amplicon sequencing, RNA sequencing, quantitative PCR, and in situ hybridization, up to the direct recovery of Fn from cancerous lesions. For proper comparison, we provide a list of the employed molecular targets by different PCR approaches and sequencing in Table 1.
Table 1. Primers and probes for quantification of Fusobacterium nucleatum DNA
Despite promising findings over the last 10 years, the role of the microbiome in the natural history of CRC has not yet been clarified (Carethers & Doubeni, Reference Carethers and Doubeni2020), neither the identification of specific bacteria has been translated into clinical practice.
We reviewed the evidence supporting the association between Fn and CRC, with a focus on molecular subtypes and host immune response, as well its potential role as a prognostic and predictive marker. Modifiable factors, such as diet as well as potential exploitation of Fn as a therapeutic target, will also be covered from a translational perspective, foreseeing its potential clinical implications.
Topography of Fusobacterium nucleatum in the gastrointestinal tract
Fusobacterium nucleatum is a Gram-negative anaerobic bacterium belonging to the Bacteroidaceae family. Fusobacteria species differ from other Bacteroidaceae in their ability to produce N-butyrate and modulate canonical Wnt signalling, thereby promoting tumourigenesis (Lazarova et al., Reference Lazarova, Bordonaro, Carbone and Sartorelli2004; Lupton, Reference Lupton2004; Flanagan et al., Reference Flanagan, Schmid, Ebert, Soucek, Kunicka, Liska, Bruha, Neary, Dezeeuw, Tommasino, Jenab, Prehn and Hughes2014). Although Fn is described as an obligate anaerobe, it can grow in up to 6% oxygen (Moore et al., Reference Moore, Holdeman, Smibert, Cato, Burmeister, Palcanis and Ranney1984). Fn represents a small proportion of the commensal microflora of the mouth in humans (Kapatral et al., Reference Kapatral, Anderson, Ivanova, Reznik, Los, Lykidis, Bhattacharyya, Bartman, Gardner, Grechkin, Zhu, Vasieva, Chu, Kogan, Chaga, Goltsman, Bernal, Larsen, D’Souza and Overbeek2002; Brennan & Garrett, Reference Brennan and Garrett2019), and in this milieu, it is symbiotic with other bacteria, but herein and elsewhere, it could also act as an opportunistic pathogen, having been isolated from skin ulcers, peritonsillar abscesses, septic arthritis, and endocarditis (Kapatral et al., Reference Kapatral, Anderson, Ivanova, Reznik, Los, Lykidis, Bhattacharyya, Bartman, Gardner, Grechkin, Zhu, Vasieva, Chu, Kogan, Chaga, Goltsman, Bernal, Larsen, D’Souza and Overbeek2002). It remains unclear whether in these infective contexts it acts like a driver or a passenger.
How Fn colonizes gut mucosa from the oral cavity remains debated (Mesa et al., Reference Mesa, Mesa-López, Egea-Valenzuela, Benavides-Reyes, Nibali, Ide, Mainas, Rizzo and Magan-Fernandez2022; Zhang et al., Reference Zhang, Cai and Ma2018). It has been reported that, although the taxonomic compositions of the oral and gut microbiomes differ, community types observed at these sites can predict each other, showing significant associations between saliva and gut specimens. These results suggest that oral bacterial populations may seed the gut, and by the time they reach the stool, those populations experience the ecological environment of the gut and give rise to consistent community types (Wu et al., Reference Wu, Yang, Zhang, Li, Xiao, Hu, Chen, Yang, Lu, Wang, Luan, Liu, Wang, Xiang, Wang, Zhao, Gao, Wang, Li and Zhu2013; Ding & Schloss, Reference Ding and Schloss2014; Mira-Pascual et al., Reference Mira-Pascual, Cabrera-Rubio, Ocon, Costales, Parra, Suarez, Moris, Rodrigo, Mira and Collado2015; Idrissi Janati et al., Reference Idrissi Janati, Karp, Von Renteln, Bouin, Liu, Tran and Emami2022). A rat model exploring the correlation between the development of apical periodontitis induced by Fn inoculation and related changes in the intestinal flora found that it negatively affects the anti-inflammatory properties of intestinal epithelium and can progress to infect the gut (Haraga et al., Reference Haraga, Sato, Watanabe, Hamada and Tani-Ishii2022).
A meta-genomic analysis by 16S rRNA sequencing and enterotype-based gut microbiota analysis showed that Bacteroides, Proteobacteria, and Firmicutes are the dominant bacterial phyla in healthy controls, as well as in patients with adenomatous polyps and CRC, with minor variations between groups. The relative abundance of Fusobacterium genus was significantly greater in CRC patients relative to normal subjects and to those with adenomas (Yang et al., Reference Yang, Lee, Tu, Huang, Chen, Sun, Tsai, Wang, Chen, Huang, Shiu, Yang, Huang, Chou, Chou, Huang, Sun, Liang, Lin and Lin2019) as well as in tumour samples and stools from CRC patients (Bi et al., Reference Bi, Zhu, Gao, Li, Zhu, Wei, Xie, Cai, Wei and Qin2022).
The infiltration of Fn into mucosal epithelial cells would depend upon an impairment of the intestinal barrier, which allows Fn to adhere and invade gut cells. Surface molecules expressed by Fn include lipopolysaccharides (LPS), adhesin A (FadA), and fusobacterium autotransporter protein 2 (Fap2), which approaches cells expressing Gal-GalNAc and binds to E-cadherin, leading to internalization within epithelial cells. Fn is also capable of releasing RNA into the host cell cytoplasm, where it is subsequently detected by cytosolic sensor retinoic acid-inducible gene 1 (RIG-1). By triggering the activation of β-catenin and NF-kB signalling pathways (Zhou et al., Reference Zhou, Chen, Yao and Hu2018). Through FadA-E-cadherin binding on Toll-like receptor 4 (TLR4), Fn can accelerate colonic carcinogenesis mainly in the presence of pre-existing genetic alterations, while Fap2-TIGIT binding can promote tumour survival by smouldering anti-tumour immunity (Gur et al., Reference Gur, Ibrahim, Isaacson, Yamin, Abed, Gamliel, Enk, Bar-On, Stanietsky-Kaynan, Coppenhagen-Glazer, Shussman, Almogy, Cuapio, Hofer, Mevorach, Tabib, Ortenberg, Markel, Miklić and Mandelboim2015; Borroni et al., Reference Borroni, Qehajaj, Farina, Yiu, Bresalier, Chiriva-Internati, Mirandola, Štifter, Laghi and Grizzi2019) (Figure 1).
Figure 1. The entry of Fn into mucosal epithelial cells relies on surface molecules such as lipopolysaccharides (LPS), adhesin A (FadA), and fusobacterium autotransporter protein 2 (Fap2). Fn targets cells expressing Gal-Gal-Nac via FadA, binds to E-cadherin, and is internalized by epithelial cells. Once inside, Fn releases its RNA into the host cell cytoplasm, which is detected by cytosolic retinoic acid-inducible gene 1 (RIG-1), activating the β-catenin and NF-kB signalling pathways through FadA-E-cadherin binding on TLR4. This FadA-E-cadherin interaction accelerates carcinogenesis in the presence of predisposing mutations. Meanwhile, Fap2-TIGIT binding promotes tumour survival by inhibiting anti-tumour immunity and contributing to chemotherapeutic resistance. Figure created by https://smart.servier.com. Last accessed online on 18 July 2024.
However, a recent paper by Zepeda-Rivera et al. sheds new light on the colonization capability of the CRC niche by Fn subspecies, confirming considerable strain-to-strain genotypic and phenotypic variations and revealing that a selected clade shows patho-adaptation to CRC. Such group, within Fn subspecies animalis (Fna) was referred to as clade C2 as opposed to C1, the latter representing the main oral pathobiont. Their work moved from pangenomic analysis of Fusobacterium strains obtained from targeted culture from human CRC specimens compared to strains from the oral cavity of non-cancerous subjects. First, at the subspecies level, it emerged that Fna is significantly enriched in CRC specimens, whereas the nucleatum subspecies is predominantly found in the oral niche. Virulence factors such as fp1A, a phospholipase autotransporter binding to host phosphoinosite-signalling lipids, and fadA are not significantly associated with Fna in comparison to other subspecies. Within Fna subspecies, the average nucleotide identity between the clades C1 and C2 ranged from 91.6% to 93.1% (95% being the threshold for species identity). While C1 and C2 have similar conserved core genomes, their accessory genomes vary, C2 having a larger one. The differences for C2 included larger chromosomes, more plasmids, and mobile genetic elements. The clades were also epigenetically distinct, with divergent methylome profiles. While the C2 was the only clade enriched in CRC, the representation of both clades did not differ in the oral cavity of CRC patients. While some virulence factors, such as fadA and fpIA, did not differ between the clades, other adhesins did, with fap2, cmpA, and fusolisin being enriched in Fna C2. Morphologically, FnaC2 cells appear more fusiform and showed a higher level of invasion in HCT116 CRC cell line (noticeably a typical MSI cell-line). Furthermore, C2 possesses enhanced scavenger mechanisms and metabolic potential, which also contribute to patho-adaptation to the intestine and CRC niche. Eventually, in the dextran sodium sulfate-induced colitis ApcMin+/− mouse model of colorectal tumourigenesis, the exposure to (orally introduced) FnaC2 caused a significant increase in adenomas compared to exposure to FnaC1. In confirmatory sequencing analysis of CRC specimens, only FnaC2 was significantly enriched therein (Zepeda-Rivera et al., Reference Zepeda-Rivera, Minot, Bouzek, Wu, Blanco-Míguez, Manghi, Jones, LaCourse, Wu, McMahon, Park, Lim, Kempchinsky, Willis, Cotton, Yost, Sicinska, Kook, Dewhirst and Johnston2024).
A relevant issue concerns the timing and involvement of Fn infection along the development and progression of CRC. It remains to be established whether the infection acts as an initiating event of colorectal tumourigenesis, rather than like a sort of “second-hit” event, accelerating tumour progression (Nakatsu et al., Reference Nakatsu, Li, Zhou, Sheng, Wong, Wu, Ng, Tsoi, Dong, Zhang, He, Kang, Cao, Wang, Zhang, Liang, Yu and Sung2015; Rubinstein et al., Reference Rubinstein, Baik, Lagana, Han, Raab, Sahoo, Dalerba, Wang and Han2019).
Association between Fn and colorectal cancer according to pathological and molecular features
CRC molecular subtypes: microsatellite and methylation status
To assess whether Fn is enriched in specific molecular subtypes of CRC, Tahara et al. assessed its association with MS and methylation status, as well with mutations in BRAF, KRAS, TP53, CHD7, and CHD8 genes (Tahara et al., Reference Tahara, Yamamoto, Suzuki, Maruyama, Chung, Garriga, Jelinek, Yamano, Sugai, An, Shureiqi, Toyota, Kondo, Estécio and Issa2014). Most CRCs (74%) harboured the bacterial genome, which was heavily enriched in 5.4% of the cases. This subset of tumours was more likely to show MSI, with MLH1 hypermethylation, TP53 wild-type, and mutations in CHD7 or CHD8. The shared CIMP+ molecular profile of Fn CRCs supported its pathogenic role in this subset of tumours.
A clear association of Fn with MSI-high, MLH1 methylation, and the CIMP-pathway in CRCs also emerged in the work by Ito et al. They reported a higher Fn content in CRCs (56%) than in pre-malignant lesions (ranging from 30% to 35% in sessile serrated adenomas), in which the bacterium was more frequently detected in CIMP-high lesions than in CIMP-low/zero lesions (Ito et al., Reference Ito, Kanno, Nosho, Sukawa, Mitsuhashi, Kurihara, Igarashi, Takahashi, Tachibana, Takahashi, Yoshii, Takenouchi, Hasegawa, Okita, Hirata, Maruyama, Suzuki, Imai, Yamamoto and Shinomura2015). Fn positivity increased gradually from sigmoid colon to cecum in Sessile Serrated Adenomas (SSAs), but not in CRC. In a later and larger study, the proportion of Fn-high CRCs increased from rectal (2.5%; 4/157) to cecal cancers (11%; 19/178), with a significant linear trend along all subsites, independent of the association with MSI-high and CIMP+ tumours (Tahara et al., Reference Tahara, Yamamoto, Suzuki, Maruyama, Chung, Garriga, Jelinek, Yamano, Sugai, An, Shureiqi, Toyota, Kondo, Estécio and Issa2014). An independent study addressing the abundance of Fn in SSAs confirmed that up to 80% of proximal SSAs harbour invasive Fn, which is three times the frequency encountered in tubular adenomas, irrespectively of their location. While the prevalence of Fn in biofilms might be similar in different colon districts, the prevalence of invasive Fn in right-sided cancers approached 90% versus 42% of distal ones, and no relationship was found between biofilm positivity and tumour invasion by Fn. Interestingly, these authors also reported the presence of Fn in all tested metastatic lymph nodes (Yu et al., Reference Yu, Chen, Fu, Zhou, Peng, Shi, Chen and Wu2016). Data concerning the Fn abundance and CRC molecular subtypes are reported in Table 2.
Table 2. Fn abundance and CRC molecular features
Note: The percentages were calculated in relation to the number of subjects who tested negative/low and high Fn out of the total number of subjects evaluated for each individual study. There are “missed” data with respect to the total negative/low and high Fn. All p-value<0.05 reported in italic format are statistically significant
Abbreviations: NA= Not Available; NS= Not Statistically significant.
a The total number of High positive cases exceeds the value reported in the summary.
b 106/1696 samples with unknown intratumoural presences and tumour characteristics treated as NA and excluded from the statistical testing.
Interestingly, employing NGS to explore the microbiome composition of CRC and its precursor lesions, Liu et al. showed that microbial communities, comprising Fusobacterium (and other CRC-associated pathobions, alike Bacteroides, Parvimonas, and Prevotella) are heterogeneous within a single neoplasia and change along the adenoma-carcinoma progression. Remarkably, taxa significantly enriched in CRC are spread along the whole tumour, although microbial communities do not differ from adjacent normal mucosa. The fold difference in the coefficient of variation for Fusobacterium was similar in adenomas and CRCs, indicating that its abundance is not affected during the transition from adenoma to cancer. However, the authors found differences in the microbiome composition between CRC with and without KRAS mutations, as well as between MSI and MSS ones. In the latter distinction, however, Fn was not among the species enriched in MSI CRCs (Liu et al., Reference Liu, Zhang, Xu, Li, Lau, Chen, Zhang, Zhao, Chen, Sung and Yu2021). By a similar approach, a recent study employing NGS found elevated levels of Fn in CRC classified as CMS1 (which are typically MSI-high), but also in samples classifiable as CMS3 which are metabolic driven. In cell co-cultures, Fn was found to be pro-tumourigenic and pro-invasive, and it altered the metabolic profile. Thus, Fn can speed up the progression of CMS3 CRC, which is enriched in metabolic pathways involving cholesterol and proteoglycan metabolism and is characterized by an upregulation of inflammatory-related signalling pathways, such as IL-8 signalling or Th17 activation (Ternes et al., Reference Ternes, Tsenkova, Pozdeev, Meyers, Koncina, Atatri, Schmitz, Karta, Schmoetten, Heinken, Rodriguez, Delbrouck, Gaigneaux, Ginolhac, Nguyen, Grandmougin, Frachet-Bour, Martin-Gallausiaux, Pacheco and Letellier2022). Additionally, Fn enhanced glutamine metabolism and formate secretion, the latter increasing cancer invasiveness and stemness.
Haruki et al. (Reference Haruki, Kosumi, Hamada, Twombly, Väyrynen, Kim, Masugi, Qian, Mima, Baba, da Silva, Borowsky, Arima, Fujiyoshi, Lau, Li, Guo, Chen, Song and Ogino2020) explored whether the expression of markers of autophagy by CRC cells could be associated with tumour content of Fn. The study assessed the expression of BECN1 (beclin 1), MAP1LC3 (LC3), and SQSTM1 (p62) in tumour cells and their association with the amount of Fn, adjusted for potential molecular confounders (i.e. MSI, CIMP, and BRAF mutations). CRCs with intermediate/high expressed BECN1 are characterized by low-load of Fn, suggesting a possible role for the autophagy in the elimination of this strain (Table 3). However, none of the autophagy proteins was associated with overall survival or with CRC-specific survival.
Table 3. Data by Haruki et al. (Reference Haruki, Kosumi, Hamada, Twombly, Väyrynen, Kim, Masugi, Qian, Mima, Baba, da Silva, Borowsky, Arima, Fujiyoshi, Lau, Li, Guo, Chen, Song and Ogino2020)
a P-trend was calculated by the linear trend across the ordinal categories of tumour BECN1, MAP1LC3, and SQSTM1 immunohistochemical expression level (low, intermediate, and high, as an ordinal predictor variable) in the IPW-adjusted ordinal logistic regression model for the amount of Fn DNA (negative, low, and high, as an ordinal outcome variable). All p<0.05 reported in italic format are statistically significant
The mentioned studies establish that approximately 1 out of 5 CRCs harbours a high copy number of Fn, has been found to cluster in MSI-high cancers, particularly those that are CIMP+ with BRAF mutations. With respect to MSI-high CRC, it would be advisable to stratify the Fn load using a state-of-the-art classification, comprising Lynch syndrome, Lynch-like cases, and sporadic ones. Fn was initially considered a potential environmental factor in the development of elevated MSI CRCs (Liu et al., Reference Liu, Tabung, Zhang, Nowak, Qian, Hamada, Nevo, Bullman, Mima, Kosumi, da Silva, Song, Cao, Twombly, Shi, Liu, Gu, Koh, Li and Giovannucci2018).
Most recently, the work by Joo et al. clarified that both hereditary and sporadic MMR-deficient CRCs are prone to Fn colonization. They confirmed the significant correlation between Fn and BRAF p.V600E somatic mutation and CIMP-high CRC (Table 2), yet the Fn enrichment in hereditary cancers also indicates the importance of the MMR-deficient tumour microenvironment for the colonization, irrespectively of the origin of the defect (Joo et al., Reference Joo, Chu, Georgeson, Walker, Mahmood, Clendenning, Meyers, Como, Joseland, Preston, Diepenhorst, Toner, Ingle, Sherry, Metz, Lynch, Milne, Southey, Hopper and Buchanan2024).
Accordingly, the main molecular feature of CRC associated with Fn colonization is MSI. One of the key reasons for investigating this relationship is the differential immune responses observed in MSI-H and MSS tumours. MSI-H tumours are known for their strong immune activation, characterized by high levels of tumour-infiltrating lymphocytes (TILs), due to their high mutational burden, which generates numerous neoantigens.
Relationship between Fn and immune response in the tumour microenvironment
The role of Fn in shaping the immune response within the CRC tumour microenvironment has been extensively studied, particularly in relation to T-cell infiltration and macrophage modulation.
The relationship between Fn abundance and the density of infiltrating T-cells in CRC was assessed in a large series of pathological specimens from Nurses’ Health Study and the Health Professionals Follow-up Study. In the study by Mima et al. (Reference Mima, Nishihara, Qian, Cao, Sukawa, Nowak, Yang, Dou, Masugi, Song, Kostic, Giannakis, Bullman, Milner, Baba, Giovannucci, Garraway, Freeman, Dranoff and Ogino2016), 13% (134 out of 1069) of CRCs showed high levels of Fn DNA, more frequently in MSI-high (18%) tumours than in MSI-low or MSS ones (4% of cases). They confirmed the over-representations of MLH1 hypermethylation and CIMP-high status among Fn-high CRCs, together with their right-sidedness, while the association with female sex and BRAF mutation did not reach statistical significance. As to T-cell populations, the amount of Fn was inversely associated with CD3+T-cell density in CRC tissues, while no differences were reported for CRC mortality (Mima et al., Reference Mima, Sukawa, Nishihara, Qian, Yamauchi, Inamura, Kim, Masuda, Nowak, Nosho, Kostic, Giannakis, Watanabe, Bullman, Milner, Harris, Giovannucci, Garraway, Freeman and Ogino2015) (Table 4).
Table 4. Data by Mima et al. (Reference Mima, Sukawa, Nishihara, Qian, Yamauchi, Inamura, Kim, Masuda, Nowak, Nosho, Kostic, Giannakis, Watanabe, Bullman, Milner, Harris, Giovannucci, Garraway, Freeman and Ogino2015)
All p-value<0.05 reported in italic format are statistically significant
Lee et al. (Reference Lee, Yoo, Oh, Jeong, Cho, Kang and Kim2021) reported that, among MSI CRCs, the Fn-high subset was composed of larger tumours with deeper pT stage (as to the prevalence of pT3–pT4 tumours), which harboured less dense FoxP3+ TILs (at tumour centre and invasive margin) but denser CD163+ tumour associated macrophages in the tumour centre (Cavalleri et al., Reference Cavalleri, Bianchi, Basso, Celesti, Grizzi, Bossi, Greco, Pitrone, Valtorta, Mauri, Truini, Dall’Olio, Brandi, Sartore-Bianchi, Ricciardiello, Torri, Rimassa, Siena and Mantovani2019). These findings reveal a different composition of the immune infiltrate (with pro-tumour steering of macrophages) associated with Fn colonization of MSI CRCs.
Due to the well-established association of low densities of CD3+T-cells and poor CRC prognosis (Grizzi et al., Reference Grizzi, Bianchi, Malesci and Laghi2013; Laghi et al., Reference Laghi, Negri, Gaiani, Cavalleri, Grizzi, de’ Angelis and Malesci2020), Mima et al. (Reference Mima, Sukawa, Nishihara, Qian, Yamauchi, Inamura, Kim, Masuda, Nowak, Nosho, Kostic, Giannakis, Watanabe, Bullman, Milner, Harris, Giovannucci, Garraway, Freeman and Ogino2015) later expanded the sample size to assess the association between Fn, T-cell infiltration, and patient outcomes. The study, comprising over 1000 CRCs, confirmed that the amount of Fn was significantly associated only with MSI-high molecular subtype at multivariable statistical analysis, independently of CIMP and BRAF CRC status, which were significant only at univariable statistical analyses. The survival of patients with Fn-high CRC was shorter than that of patients with Fn- cancers, even when stratified by stage. This finding, although in agreement with the low infiltration by T cells, was counterintuitive with respect to the better prognosis of patients with MSI CRC, usually ascribed to their lower metastatic potential (Malesci et al., Reference Malesci, Laghi, Bianchi, Delconte, Randolph, Torri, Carnaghi, Doci, Rosati, Montorsi, Roncalli, Gennari and Santoro2007; Laghi & Malesci, Reference Laghi and Malesci2012).
Consistent with previous findings, Hamada et al. tested whether the association between Fn and the magnitude of the adaptive immune response differs according to CRC MS-status. Accordingly, histopathologic lymphocytic reactions (i.e. the densities of CD3+, CD8+, CD45RO+, and FOXP3+ lymphocytes) were assessed in CRC strata differentiated by MS-status (Hamada et al., Reference Hamada, Zhang, Mima, Bullman, Sukawa, Nowak, Kosumi, Masugi, Twombly, Cao, Song, Liu, da Silva, Shi, Gu, Li, Koh, Nosho, Inamura and Ogino2018) (Table 5). Surprisingly, high loads of Fn were negatively associated with TIL amounts in MSI-high tumours (multivariable OR, 0.45; 95% CI 0.22–0.92), but positively associated with the same variable in MSS tumours (significant interaction, adjusted by molecular confounders). In any case, patients with high loads of Fn in their tumours had an increased risk of disease-specific mortality, irrespective of MS-status (HR=1.27 for non-MSI-high CRCs, HR=2.23 for MSI-high CRCs at multivariable analysis) (Hamada et al., Reference Hamada, Zhang, Mima, Bullman, Sukawa, Nowak, Kosumi, Masugi, Twombly, Cao, Song, Liu, da Silva, Shi, Gu, Li, Koh, Nosho, Inamura and Ogino2018) (Table 6). Currently, it remains unanswered how the bacterium can worsen the survival in the context of tumour groups differing for the type of genetic instability and the amount of immune response. One possible interpretation of this behaviour might lie in a smouldering activity of Fn towards adaptive immune response in MSI-high CRCs with abundant neo-antigens, as opposed to its pro-inflammatory properties in MSS cancers.
Table 5. Data by Hamada et al. (Reference Hamada, Zhang, Mima, Bullman, Sukawa, Nowak, Kosumi, Masugi, Twombly, Cao, Song, Liu, da Silva, Shi, Gu, Li, Koh, Nosho, Inamura and Ogino2018)
Table 6. Data by Hamada et al. (Reference Hamada, Zhang, Mima, Bullman, Sukawa, Nowak, Kosumi, Masugi, Twombly, Cao, Song, Liu, da Silva, Shi, Gu, Li, Koh, Nosho, Inamura and Ogino2018)
However, increasing evidence suggests that other immune cell types, such as natural killer (NK) cells, neutrophils, and dendritic cells (DCs), are also influenced by the presence of Fn in CRC tissues.
NK cells are critical for anti-tumour immunity through their cytotoxic activity and cytokine production. However, Fn has been implicated in immune evasion strategies that impair NK cell function. Studies suggest that Fn may downregulate NK cell activation markers or promote the expression of immune checkpoint molecules, such as TIGIT, which inhibits NK cell-mediated cytotoxicity (Gur et al., Reference Gur, Ibrahim, Isaacson, Yamin, Abed, Gamliel, Enk, Bar-On, Stanietsky-Kaynan, Coppenhagen-Glazer, Shussman, Almogy, Cuapio, Hofer, Mevorach, Tabib, Ortenberg, Markel, Miklić and Mandelboim2015). Additionally, the presence of Fn-derived lipopolysaccharides (LPS) can induce an immunosuppressive cytokine profile, reducing NK cell activation and favouring tumour immune escape (Borroni et al., Reference Borroni, Qehajaj, Farina, Yiu, Bresalier, Chiriva-Internati, Mirandola, Štifter, Laghi and Grizzi2019).
Recent research indicates that Fn can drive the activation of neutrophil extracellular traps (NETs), web-like structures composed of DNA and antimicrobial proteins that neutrophils release in response to infection or tumour signals (Kong et al., Reference Kong, Zhang, Xiang, You, Duan, Zhao, Li, Wu, Zhang, Zhou and Duan2023). In CRC, NETs have been shown to promote EMT, facilitating tumour invasion and metastasis (Salvucci et al., Reference Salvucci, Crawford, Stott, Bullman, Longley and Prehn2022). Furthermore, Fn infection leads to increased recruitment of tumour-associated neutrophils (TANs), which are often associated with poor prognosis in CRC patients (Yan et al., Reference Yan, Liu, Li, Qin and Sun2017).
DCs are key antigen-presenting cells that initiate and regulate adaptive immune responses. However, Fn has been shown to impair DC function, reducing their ability to stimulate anti-tumour T-cell responses (Borroni et al., Reference Borroni, Qehajaj, Farina, Yiu, Bresalier, Chiriva-Internati, Mirandola, Štifter, Laghi and Grizzi2019). In CRC, Fn-colonized tumours exhibit reduced infiltration of mature DCs and an accumulation of tolerogenic DC subsets, which suppress effective immune responses (Gholizadeh et al., Reference Gholizadeh, Eslami and Kafil2017). This may be mediated by Fn-induced activation of the β-catenin and NF-κB pathways, which favour immune suppression over activation (Zhou et al., Reference Zhou, Chen, Yao and Hu2018).
Understanding the role of Fn in modulating NK cells, neutrophils, and dendritic cells provides new insights into its contribution to immune evasion and tumour progression in CRC. Future research should focus on therapeutic strategies targeting these immune interactions, such as modulating neutrophil activation, restoring NK cell cytotoxicity, or enhancing DC antigen presentation. These findings support the broader role of Fn in shaping the immune landscape of CRC beyond T-cell infiltration, offering potential biomarkers and therapeutic targets for improving CRC treatment outcomes.
Environmental factors: diet, cancer, and Fn load
The interplay between environmental factors and CRC risk extends beyond genetic predisposition, with increasing evidence highlighting the influence of diet and microbiome alterations. Among these factors, diet has received particular attention due to its direct role in modulating gut microbiota composition, including Fn colonization levels. Epidemiological studies suggest that diets rich in fibre support beneficial gut bacteria and may reduce Fn load, thereby lowering CRC risk (Mehta et al., Reference Mehta, Nishihara, Cao, Song, Mima, Qian, Nowak, Kosumi, Hamada, Masugi, Bullman, Drew, Kostic, Fung, Garrett, Huttenhower, Wu, Meyerhardt, Zhang and Ogino2017). Conversely, pro-inflammatory diets (e.g. high consumption of processed and red meats) have been associated with an increased abundance of Fn, heightened inflammatory responses, and a greater likelihood of Fn-positive CRC (Liu et al., Reference Liu, Tabung, Zhang, Nowak, Qian, Hamada, Nevo, Bullman, Mima, Kosumi, da Silva, Song, Cao, Twombly, Shi, Liu, Gu, Koh, Li and Giovannucci2018).
Beyond diet, other environmental factors may also contribute to Fn enrichment in the gut microbiome. Studies indicate that smoking, alcohol consumption, and antibiotic use can significantly alter gut bacterial diversity and create an environment favouring pathogenic bacteria like Fn (Gholizadeh et al., Reference Gholizadeh, Eslami and Kafil2017). Additionally, urbanization and industrial pollutants may impact gut homeostasis, potentially increasing susceptibility to Fn-associated tumourigenesis (Alexander et al., Reference Alexander, Scott, Pouncey, Marchesi, Kinross and Teare2018). Recent research has also proposed a role for Fn-derived bacterial metabolites, such as riboflavin, in shaping the tumour microenvironment by activating mucosal-associated invariant T (MAIT) cells, which drive pro-inflammatory signalling linked to CRC progression (Li et al., Reference Li, Simoni, Becht, Loh, Li, Lachance, Koo, Lim, Tan, Mathew, Nguyen, Golovato, Berkson, Prlic, Lee, Minot, Nagarajan, Dey, Tan and Newell2020).
Besides the link between high Fn loads and CRC molecular profiles, extrinsic factors contributing to CRC such as diet composition may also be associated with tumour bacterial content. Indirect evidence for such an association came from two studies. The first one showed that adherence to a prudent diet (enriched in whole grains and dietary fibre) rather than to a Western-type diet was associated with a reduced risk of developing Fn + cancers, irrespectively of tumour location, in U.S. prospective cohorts (Mehta et al., Reference Mehta, Nishihara, Cao, Song, Mima, Qian, Nowak, Kosumi, Hamada, Masugi, Bullman, Drew, Kostic, Fung, Garrett, Huttenhower, Wu, Meyerhardt, Zhang and Ogino2017). The authors suggested that the cancer preventive effect of a diet enriched in dietary fibres may be modulated by the microbiota, as assessed through the load of Fn (Table 7). The second study evaluated the association between the intake of foods sustaining the release of inflammatory cytokines (IL-6 and TNF receptor superfamily member 1B) and high levels of C-reactive protein with the risk of developing CRC stratified by Fn content (Liu et al., Reference Liu, Tabung, Zhang, Nowak, Qian, Hamada, Nevo, Bullman, Mima, Kosumi, da Silva, Song, Cao, Twombly, Shi, Liu, Gu, Koh, Li and Giovannucci2018). Inflammatory effects of the diet were estimated by empirical dietary inflammatory pattern score, a system in which high scores correspond to an inflammatory diet and high plasma levels of IL-6, TNF receptor superfamily member 1B, and C-reactive protein. Higher scores were associated with an increased risk of Fn + CRC, again in the proximal colon, supporting an interaction between an inflammatory diet and the microbiota. As a component of the gut ecosystem, Fn residency could be influenced by alimentary habits in driving colonic carcinogenesis (Liu et al., Reference Liu, Tabung, Zhang, Nowak, Qian, Hamada, Nevo, Bullman, Mima, Kosumi, da Silva, Song, Cao, Twombly, Shi, Liu, Gu, Koh, Li and Giovannucci2018) (Table 8).
Table 7. Data by Mehta et al. (Reference Mehta, Nishihara, Cao, Song, Mima, Qian, Nowak, Kosumi, Hamada, Masugi, Bullman, Drew, Kostic, Fung, Garrett, Huttenhower, Wu, Meyerhardt, Zhang and Ogino2017)
All p-value<0.05 reported in italic format are statistically significant. Abbreviations: CI, Confidence Interval; HR, Hazard Ratio.
a Empirical dietary inflammatory pattern (EDIP) scores.
b Tests for trend were conducted using the median value of each quartile category as a continuous variable.
c We tested for heterogeneity by using a likelihood ratio test, comparing a model that allows separate associations for the two CRC.
d Stratified by age, calendar years, and gender and adjusted for total caloric intake (kcal/day), family history of colorectal cancer in any first-degree relative, history of previous endoscopy, pack-years of smoking (never, 0–4, 5–19, 20–39, or >40), body mass index (kg/m2), physical activity (MET-hours/week), and regular aspirin or NSAID use (≥2 tablets/week).
Table 8. Data by Liu et al. (Reference Liu, Tabung, Zhang, Nowak, Qian, Hamada, Nevo, Bullman, Mima, Kosumi, da Silva, Song, Cao, Twombly, Shi, Liu, Gu, Koh, Li and Giovannucci2018)
All p-value<0.05 reported in italic format are statistically significant. Abbreviations: CI, Confidence Interval; HR, Hazard Ratio.
a Empirical dietary inflammatory pattern (EDIP) scores.
The association between various dietary factors and faecal Fn was evaluated in a cross-sectional Japanese study involving healthy adults without a history of colorectal cancer or precancerous lesions. A high intake of dairy products in healthy adults may reduce Fn and prevent colorectal cancer (Narii et al., Reference Narii, Zha, Sobue, Kitamura, Shiba, Mizutani, Yamada and Yachida2023). Although within its limitations, the study by Shimomura et al. found that the abundance of Fn had a significant natural indirect effect on CRC risk based on their highest fibre intake compared to the lowest fibre intake (Shimomura et al., Reference Shimomura, Zha, Komukai, Narii, Sobue, Kitamura, Shiba, Mizutani, Yamada, Sawada and Yachida2023).
Recently, a new model of CRC pathogenesis was developed involving the bacterial riboflavin synthase from the enriched colorectal bacteria, i.e. Fn. Its role in CRC development may be attributed to microbe-derived riboflavin metabolites activating mucosal-associated invariant T-Cell (MAIT) (Li et al., Reference Li, Simoni, Becht, Loh, Li, Lachance, Koo, Lim, Tan, Mathew, Nguyen, Golovato, Berkson, Prlic, Lee, Minot, Nagarajan, Dey, Tan and Newell2020), which produce proinflammatory cytokines and cytotoxic molecules integral to the pathological process of CRC (Wang et al., Reference Wang, Zhang, Wu, Qin and He2022).
Sikavi et al. hypothesized that the association between a diet linked to a greater abundance of sulfur-metabolizing bacteria and distal CRC is moved toward specific molecular subtypes and/or relative enrichment (or depletion) of intra-tumoural CRC-associated microbes. Their data did not provide conclusive evidence to support their hypothesis, and they were unable to draw significant conclusions about the relationship between dietary content and molecular tumour status in CRC. The only emerging difference was the negative association with Bifidobacterium spp. of distal CRC in men (Sikavi et al., Reference Sikavi, Nguyen, Haruki, Ugai, Ma, Wang, Thompson, Yan, Branck, Wilkinson, Akimoto, Zhong, Lau, Mima, Kosumi, Morikawa, Rimm, Garrett, Izard and Chan2021).
While this section primarily examines dietary influences, it is crucial to recognize that a combination of environmental exposures likely contributes to Fn colonization, gut dysbiosis, and CRC development. Future research should investigate how various environmental factors interact with dietary patterns to shape Fn-mediated tumourigenesis. A deeper understanding of these relationships could guide preventive strategies, including dietary modifications and microbiome-targeted interventions, to mitigate the risk of Fn-associated CRC.
Fn and outcome of CRC patients
Pathological features of Fn + tumours and their role as prognostic biomarkers
When considering the association with patient demographics, no significant associations emerged as to the relationship of Fn amount with patient age or sex. Besides the above-mentioned association with right-sided MSI CRC, in a model adjusted for patient age, sex, CRC location, and BMI, patients with high faecal Fn abundance had a 3-fold increased likelihood of being diagnosed with rectal compared with colon tumours, and 5-fold increased risk compared with right-sided CRC (Eisele et al., Reference Eisele, Mallea, Gigic, Stephens, Warby, Buhrke, Lin, Boehm, Schrotz-King, Hardikar, Huang, Pickron, Scaife, Viskochil, Koelsch, Peoples, Pletneva, Bronner, Schneider and Ose2021).
As to tumour pathological features, the rate of Fn + usually increases with the depth of tumour invasion (T), poor tumour differentiation (de Carvalho et al., Reference de Carvalho, de Mattos Pereira, Datorre, Dos Santos, Berardinelli, Matsushita, Oliveira, Durães, Guimarães and Reis2019; Haruki et al., Reference Haruki, Kosumi, Hamada, Twombly, Väyrynen, Kim, Masugi, Qian, Mima, Baba, da Silva, Borowsky, Arima, Fujiyoshi, Lau, Li, Guo, Chen, Song and Ogino2020; Joo et al., Reference Joo, Chu, Georgeson, Walker, Mahmood, Clendenning, Meyers, Como, Joseland, Preston, Diepenhorst, Toner, Ingle, Sherry, Metz, Lynch, Milne, Southey, Hopper and Buchanan2024; Liu et al., Reference Liu, Tabung, Zhang, Nowak, Qian, Hamada, Nevo, Bullman, Mima, Kosumi, da Silva, Song, Cao, Twombly, Shi, Liu, Gu, Koh, Li and Giovannucci2018; Mima et al., Reference Mima, Sukawa, Nishihara, Qian, Yamauchi, Inamura, Kim, Masuda, Nowak, Nosho, Kostic, Giannakis, Watanabe, Bullman, Milner, Harris, Giovannucci, Garraway, Freeman and Ogino2015; Sun et al., Reference Sun, An, Tian, Wang, Guan, Dong, Zhao and Hao2016; Wei et al., Reference Wei, Cao, Liu, Yao, Sun, Li, Li, Zhang and Zhou2016), neural invasion, and with the presence of nodal (N) (Castellarin et al., Reference Castellarin, Warren, Freeman, Dreolini, Krzywinski, Strauss, Barnes, Watson, Allen-Vercoe, Moore and Holt2012; Yamaoka et al., Reference Yamaoka, Suehiro, Hashimoto, Hoshida, Fujimoto, Watanabe, Imanaga, Sakai, Matsumoto, Nishioka, Takami, Suzuki, Hazama, Nagano, Sakaida and Yamasaki2018; Yan et al., Reference Yan, Liu, Li, Qin and Sun2017), and distant metastasis (M) (Chen et al., Reference Chen, Zhang, Zhang, Yue, Wang, Pan, Zhang, Liu and Zhang2022), although such results have not been entirely confirmed (Tahara et al., Reference Tahara, Yamamoto, Suzuki, Maruyama, Chung, Garriga, Jelinek, Yamano, Sugai, An, Shureiqi, Toyota, Kondo, Estécio and Issa2014; Ito et al., Reference Ito, Kanno, Nosho, Sukawa, Mitsuhashi, Kurihara, Igarashi, Takahashi, Tachibana, Takahashi, Yoshii, Takenouchi, Hasegawa, Okita, Hirata, Maruyama, Suzuki, Imai, Yamamoto and Shinomura2015; Chen et al., Reference Chen, Lu, Ke and Li2019, Reference Chen, Zhang, Zhang, Yue, Wang, Pan, Zhang, Liu and Zhang2022; Eisele et al., Reference Eisele, Mallea, Gigic, Stephens, Warby, Buhrke, Lin, Boehm, Schrotz-King, Hardikar, Huang, Pickron, Scaife, Viskochil, Koelsch, Peoples, Pletneva, Bronner, Schneider and Ose2021). Contrarily, it is worth noting that Nakatsu et al., investigating the microbial communities at different stages of carcinogenesis, found that Fn was predominantly enriched in the early stage (i.e. stage I–II) of CRC (Nakatsu et al., Reference Nakatsu, Li, Zhou, Sheng, Wong, Wu, Ng, Tsoi, Dong, Zhang, He, Kang, Cao, Wang, Zhang, Liang, Yu and Sung2015). Data related to Fn abundances, demographics, and pathological features in CRC are reported in Table 9.
Table 9. Fn abundance, demographics and pathological features in CRC
All p-value<0.05 reported in italic format are statistically significant. Abbreviations: NA= Not Available; NS= Not Statistically significant.
a The percentages were calculated in relation to the number of subjects who tested negative/low and high Fn out of the total number of subjects evaluated for each individual study. There are “missed” data with respect to the total negative/low and high Fn.
b 106/1696 samples with unknown intratumoural presences and tumour characteristics treated as NA and excluded from the statistical testing.
A study from China also reported a significant association with TNM components, including metastases (Sun et al., Reference Sun, An, Tian, Wang, Guan, Dong, Zhao and Hao2016).
Data from a small European cohort (Kunzmann et al., Reference Kunzmann, Proença, Jordao, Jiraskova, Schneiderova, Levy, Liska, Buchler, Vodickova, Vymetalkova, Silva, Vodicka and Hughes2019) found an increased risk of death (overall survival; HR 1.68; 95% CI, 1.02–2.77; p=0.04) only after adjusting for age, TNM stage and adjuvant treatments; interestingly, the inclusion of MS-status in the model led to lose statistical significance (HR 1.80; 95% CI 0.97–3.28, p = 0.06). Differently, a North European (Bundgaard-Nielsen et al., Reference Bundgaard-Nielsen, Baandrup, Nielsen and Sørensen2019) study on a smaller cohort failed to find any significant association between Fn and CRC or adenoma (Table 10) and their outcome at 5 years. Eventually, in a study from South America, Fn-high content was also associated with a higher TNM stage (Table 11) and worse patient CRC-specific survival (de Carvalho et al., Reference de Carvalho, de Mattos Pereira, Datorre, Dos Santos, Berardinelli, Matsushita, Oliveira, Durães, Guimarães and Reis2019).
Table 10. Data by Bundgaard-Nielsen et al. (Reference Bundgaard-Nielsen, Baandrup, Nielsen and Sørensen2019)
Note: P-value, Not significant between each neoplasia and paired normal tissue.
Table 11. Data by de Carvalho et al. (Reference de Carvalho, de Mattos Pereira, Datorre, Dos Santos, Berardinelli, Matsushita, Oliveira, Durães, Guimarães and Reis2019)
All p-value<0.05 reported in italic format are statistically significant. Abbreviations: MSI, microsatellite instability; CI, Confidence Interval; OR, Odds ratio; NE, Not Evaluated.
In the largest study, high loads of Fn had a negative association with CRC-specific mortality in a stage-stratified Cox model, indicating that Fn would act as a negative prognostic factor across stages, independently of other biomarkers. The prognostic associations of Fn with CRC outcome have also been the matter of two recent meta-analyses, both supporting the association of Fn with both shorter survival and higher CRC stage (Colov et al., Reference Colov, Degett, Raskov and Gögenur2020; Gethings-Behncke et al., Reference Gethings-Behncke, Coleman, Jordao, Longley, Crawford, Murray and Kunzmann2020).
Besides the match with more advanced stage at diagnosis supporting an accelerating role for Fn in tumour progression, an interesting study evaluating the features of patients with/out metachronous adenomas after polypectomy detected a higher abundance of Fn in patients with metachronous adenomas than in those without. Multivariable analysis showed that a high abundance of Fn, male gender, and age were associated with metachronous adenomas (Xue et al., Reference Xue, Xie, Zou, Qian, Kang, Zhou, Pan, Xia, Chen and Fang2021). Such translational results are intriguing as they also would support a role for Fn in tumour development.
Overall, studies point to the association of Fn with worse outcomes and advanced CRC stage, yet it remains to be firmly established whether such association is stage-independent. If the prognostic association with worse CRC outcomes holds true across stages, Fn may act as a disease modifier, accelerating tumour progression and worsening prognosis (Table 12).
Table 12. Fn actionability as modifier in disease progression
All p-value<0.05 reported in italic format are statistically significant. Abbreviations: LARC, locally advanced rectal cancer; NA, Not Available; OS, Overall Survival; DFS, Disease Free Survival; DSS, Disease Short Survival; HR, Hazard Ratio.
a Independent of MS-status at multivariate analysis.
b Cumulative data from cohorts 2 (n=92) and 3 (n=173); p-value at multivariate independent from TNM stage (II vs III).
Fn as a potential driver of mechanisms of tumour invasion
Similarly, it remains unclear whether Fn can also trigger epithelial-to-mesenchymal transition (EMT) in CRC cells, favouring their spread. EMT comprises the loss of E-cadherin accompanied by the expression of mesenchymal drivers, alike TWIST1 (Celesti et al., Reference Celesti, Di Caro, Bianchi, Grizzi, Basso, Marchesi, Doni, Marra, Roncalli, Mantovani, Malesci and Laghi2013) and enhanced invasiveness. Ma et al. (Reference Ma, Luo, Gao, Tang and Chen2018) demonstrated that Fn promotes the proliferation and invasion of normal colon epithelial cells (NCM460) by interacting with E-cadherin without affecting its expression. In their study on stage III and IV CRCs, Yan et al. reported that the expression of EMT-related markers (N-cadherin) was associated with Fn loads, being shifted towards mesenchymal features in Fn-high CRCs, altogether with the expression of markers of stemness (Yan et al., Reference Yan, Liu, Li, Qin and Sun2017). Most recently, in vitro experiments by Kong et al. demonstrated that Fn-induced NETs indirectly accelerated malignant tumour growth through angiopoiesis and facilitate tumour metastasis via EMT-related cell migration, matrix metalloproteinase (MMP)-mediated degradation of basement membrane proteins, and the subsequent trapping and dissemination of CRC cells (Kong et al., Reference Kong, Zhang, Xiang, You, Duan, Zhao, Li, Wu, Zhang, Zhou and Duan2023). Salvucci et al. observed a dysregulation of MAPK signalling, inducing EMT at the protein level, when comparing Fusobacteriales-high and Fusobacteriales-low patients in the TCGA-COAD-READ cohort. Their bioinformatic analysis reported that patients with mesenchymal tumours and a high prevalence of Fn have worse prognosis in CRC (Salvucci et al., Reference Salvucci, Crawford, Stott, Bullman, Longley and Prehn2022).
Zhang et al. found that, mechanistically, Fn activated the TGF-β1/SMAD signalling pathway to promote EMT via the miR-122-5p/FUT8 axis, stimulating CRC cells to excrete exosome-wrapped miR-122-5p, and activating the FUT8/TGF-b1/Smads axis to promote metastasis (Zhang et al., Reference Zhang, Wang, Yu, Zhang, Wang, Shang, Xin, Li, Ning, Zhang and Zhang2023).
Recently, new findings revealed that Fn bacterial ferritins that protect DNA from oxidative stress (Fn- DNA hunger/stationary phase protective proteins [Dps]) is a novel multifunctional Fn virulence factor that lyses and disrupts erythrocytes, enhances intracellular survival of Fn in macrophages, and promotes the migration of CRC cells via the CCL2/CCL7-induced EMT and CRC metastasis. A high level of serum anti-Fn-Dps antibody was found to be prevalent in populations, and elevated anti-I-Dps antibody levels were observed in CRC patients (Wu et al., Reference Wu, Guo, Chen, Li, Huang, Liu and Zhang2023).
The mechanistic link between Fn and the enhancement of the EMT process as a route accelerating CRC progression is intriguing. However, it should be reconciled with the association with MSI CRC, which is characterized by a better outcome and consistently by a molecular profile with poor EMT features (Celesti et al., Reference Celesti, Di Caro, Bianchi, Grizzi, Basso, Marchesi, Doni, Marra, Roncalli, Mantovani, Malesci and Laghi2013).
Fn as a candidate predictive biomarker: impact on CRC chemotherapy. Resistance mechanisms and therapeutic implications
Currently, the behaviour of CRC largely depends upon administered treatments, either surgical or medical, the latter both in adjuvant and neo-adjuvant settings.
Studies have shown that Fn can promote chemo-resistance through multiple mechanisms, including the activation of TLR4/MYD88 signalling, leading to autophagy-mediated survival of cancer cells (Yu et al., Reference Yu, Guo, Yu, Sun, Ma, Han, Qian, Kryczek, Sun, Nagarsheth, Chen, Chen, Hong, Zou and Fang2017). This is particularly relevant in patients undergoing 5-fluorouracil (5-FU) or oxaliplatin-based chemotherapy, where high Fn loads have been correlated with higher recurrence rates and reduced treatment efficacy. More recently, La Course et al. found that 5-FU has potent antibacterial activity against Fn CRC tumour isolates, suggesting that this treatment could inhibit the growth of these bacteria within tumours. Their study highlights the potential dual benefits of 5-FU as both an effective cancer treatment and a potent antimicrobial agent within the tumour microenvironment (LaCourse et al., Reference LaCourse, Zepeda-Rivera, Kempchinsky, Baryiames, Minot, Johnston and Bullman2022). Further research is needed to unravel the impact of 5-FU on the microbiota within tumour lesions and its implications for cancer therapy.
Accordingly, data concerning the association between patient outcome and Fn should be evaluated considering clinical settings involving the administration of chemotherapy. In Korean patients with either high-risk stage II or stage III CRC receiving adjuvant therapy (either FOLFOX or CAPOX), the abundance of Fn did not differentiate survival. However, subgroup analyses showed that in patients with cancer proximal to the sigmoid colon (i.e. cecal, ascending, transverse, and descending colon), those with Fn-high CRC had a better disease-free survival (Oh et al., Reference Oh, Kim, Bae, Kim, Cho and Kang2019). In this subset, Fn was an independent prognostic factor at multivariable analysis (HR, 0.42; 95% CI, 0.18 to 0.97; p = 0.043). This favourable effect on patient outcome was restricted to non-MSI-high cancers. In another Korean study including patients who received chemotherapy (sample size, 246 patients), either with adjuvant or palliative intent, Fn-high was associated with poor overall survival in the palliative cohort (p = 0.042), also at multivariable analysis (adjusted HR 1.69 [95% CI 1.04–2.75], p = 0.034), while its amount was not associated with recurrence in the adjuvant cohort (Lee et al., Reference Lee, Han, Kang, Bae, Kim, Won, Jeong, Park, Kang and Kim2018). One study from China (Chen et al., Reference Chen, Lu, Ke and Li2019), including high-risk stage II and stage III patients, reported an association between high Fn levels and worse outcomes in CRC patients (Table 3). The distribution of Fn + CRC was widespread along the colon, with a slightly more frequent in the descending segments. A similar worsening behaviour associated with abundance of Fn on the overall survival in patients with stage IV CRC was also reported in a small Japanese study (Yamaoka et al., Reference Yamaoka, Suehiro, Hashimoto, Hoshida, Fujimoto, Watanabe, Imanaga, Sakai, Matsumoto, Nishioka, Takami, Suzuki, Hazama, Nagano, Sakaida and Yamasaki2018). Yan et al., by studying stage III and IV CRCs, found that high loads of Fn correlated with local and nodal invasion, as well with distant metastasis, and discriminated a worse CRC-specific survival, irrespectively of TNM staging (Yan et al., Reference Yan, Liu, Li, Qin and Sun2017). Interestingly, patients with low Fn levels had a better disease-free survival than those with high Fn levels, if treated with adjuvant therapy (with significant interaction).
In their innovative paper, Yu et al. detected abundant Fn in CRCs of patients with recurrence after chemotherapy as compared to patients without. The load of Fn above a cut-off identified by ROC-curve analysis could predict recurrence (Yu et al., Reference Yu, Guo, Yu, Sun, Ma, Han, Qian, Kryczek, Sun, Nagarsheth, Chen, Chen, Hong, Zou and Fang2017), yet their analyses were adjusted for stage but not for adjuvant therapy. Noticeably, in cellular and xenograft models, Fn led to the development of chemo-resistance. Mechanistically, Fn enhanced the expression of TLR4 and MYD88 transcripts, inducing a selective loss of miR-18a* and miR-4802 expression, ultimately activating autophagy (Yu et al., Reference Yu, Guo, Yu, Sun, Ma, Han, Qian, Kryczek, Sun, Nagarsheth, Chen, Chen, Hong, Zou and Fang2017). Meanwhile, Bullmann et al. showed in another fascinating paper that Fn is maintained in distant metastases, documenting the stability of the microbiome between primary and secondary tumour lesions. As viable bacteria identical to those isolated from the primary cancer were retrieved and cultured from metastatic lesions, the persistence of Fusobacterium species remains clonal through the establishment of metastases. In an in vivo model, Fn persisted in patient-derived xenografts through multiple passages, and such bacteria were invasive once incubated with CRC cells. Eventually, treatment of the xenografts with metronidazole reduced the Fusobacterium load, cell proliferation, and tumour growth. However, the authors did not confirm the association between Fusobacterium load and CRC recurrence (Bullman et al., Reference Bullman, Pedamallu, Sicinska, Clancy, Zhang, Cai, Neuberg, Huang, Guevara, Nelson, Chipashvili, Hagan, Walker, Ramachandran, Diosdado, Serna, Mulet, Landolfi, Ramon Y Cajal and Meyerson2017).
Moreover, emerging evidence highlights the role of Fn-derived exosomal miRNAs (e.g. miR-1246, miR-92b-3p, and miR-27a-3p) in promoting EMT and metastasis. Guo et al. recently found that exosomes (tiny particles released by cells) from CRC cells infected with Fn help cancer spread by carrying specific miRNAs. When CRC cells (HCT116, MSI; SW480, MSS) were exposed to these exosomes, they moved more easily and changed shape, like an EMT process, which helps cancer spread. The miRNA content in these exosomes was different, with high levels of miR-1246, miR-92b-3p, and miR-27a-3p. These miRNAs increased cell movement by lowering GSK3b levels and reducing E-Cadherin while raising Vimentin levels. The exosomes also increased CXCL16 levels, further promoting migration. In both lab and animal models, exposure to these exosomes led to more tumour growth and liver metastases. In CRC patients, higher levels of exosomal CXCL16 and miR-1246/92b-3p/27a-3p were linked to more Fn and advanced cancer stages (Guo et al., Reference Guo, Chen, Chen, Zeng, Liu and Zhang2021). These miRNAs may serve as potential biomarkers for identifying patients at higher risk of therapy resistance, suggesting a need for microbiome-based stratification in CRC treatment plans.
In summary, Fn plays a dual role in CRC chemotherapy responses both as a driver of resistance and as a potentially targetable component of treatment. Future studies should explore how modulating Fn levels through antibiotics, microbiome interventions, or targeted therapies could enhance chemotherapy efficacy and improve patient outcomes.
Locally advanced rectal cancers treated with neo-adjuvant therapy
A translational paper investigated by RNA in situ hybridization (RNA-ISH), digital image analysis, and qPCR, the abundance of Fn in tumours from patients with locally advanced rectal cancer (LARC) treated with neoadjuvant chemotherapy (Serna et al., Reference Serna, Ruiz-Pace, Hernando, Alonso, Fasani, Landolfi, Comas, Jimenez, Elez, Bullman, Tabernero, Capdevila, Dienstmann and Nuciforo2020), having a cohort of untreated patients as control. The concordance between the two approaches (qPCR and RNA-ISH) was high (agreement rate, 86%). Fn was mainly located at the luminal surface of the cancers, and its density was significantly higher in untreated cancer samples than in tumour specimens collected after treatment, with a positivity rate of 57% and 25%, respectively. Although Fn abundance did predict the responsiveness to treatment, the difference between the rates of responsiveness in Fn + (34%) and negative (13%) tumours only approached significance p=0.08. However, considering Fn status in specimens collected after treatment, 59% of patients with Fn + tumours relapsed, as compared to only 11% among Fn- ones (OR 11.6, 95% CI 3.2–43.3; p<0.001). Noteworthy, Fn drop after therapy was associated with better outcomes. Data on CD3+ and CD8+ TILs were also available for a subset of patients. While no difference was reported in TIL densities related to Fn status before surgery, CD8+ cells appeared to be higher in Fn-negative patients before and after treatment, suggesting that Fn persistence after neo-adjuvant therapy for LARC is associated with its relapse, possibly related to reduced immune cytotoxicity (Serna et al., Reference Serna, Ruiz-Pace, Hernando, Alonso, Fasani, Landolfi, Comas, Jimenez, Elez, Bullman, Tabernero, Capdevila, Dienstmann and Nuciforo2020).
Shortly following the first report on the relevance of Fn as to the outcome of LARC treated with neoadjuvant therapy, another study explored the microbiome as a predictor of responsiveness to neoadjuvant therapy in LARC patients. In their prospective longitudinal study, Yi et al. also report several associations of the microbiome profile with LARC as well as with its responsiveness to neoadjuvant therapy. These almost parallel findings are rather surprising in light of the previously strengthened associations with right-sided, MSI colon cancers. In LARC, the composition of the microbiome investigated by NGS was different in patients and controls, mainly for the contribution of the pathobions in a subset of patients, referred to as having type 1 LARC. These patients also showed a significantly lower fraction of responders to neoadjuvant therapy than patients with type 2 LARC. Although Fn was significantly reduced post-therapy, the network of microbiome composition between pre- and post-therapy was associated with responsiveness, butyrate-producing bacteria being associated with a better response than that observed in patients colonized by Coriobacteriaceae, Granulicatella, R. pickettii, and E. tayi other than Fusobacterium. Accordingly, the authors developed a receiver operating characteristic (ROC) curve analysis for microbiome composition to predict therapy responsiveness, with valuable performances in both training and validation settings. Although numbers were small with respect to the number of analyzed variables, this work supports a new biological variable, possibly estimating the outcome of the patients with LARC treated with neoadjuvant therapy (Yi et al., Reference Yi, Shen, Shi, Xia, Zhang, Wang, Zhang, Wang, Sun, Zhang, Zou, Yang, Zhang, Zhu, Goel, Ma and Zhang2021).
Potential exploitation of Fn for the diagnosis of CRC
There are high expectations on the increasing exploitation of biomarkers, detectable in faeces or blood, that might help to improve the screening for CRC (Eklöf et al., Reference Eklöf, Löfgren-Burström, Zingmark, Edin, Larsson, Karling, Alexeyev, Rutegård, Wikberg and Palmqvist2017; Xie et al., Reference Xie, Gao, Cai, Sun, Zou, Chen, Yu, Qiu, Gu, Chen, Cui, Sun, Liu, Cai, Xu, Chen and Fang2017; Chung et al., Reference Chung, Gray, Singh, Issaka, Raymond, Eagle, Hu, Chudova, Talasaz, Greenson, Sinicrope, Gupta and Grady2024; Imperiale et al., Reference Imperiale, Porter, Zella, Gagrat, Olson, Statz, Garces, Lavin, Aguilar, Brinberg, Berkelhammer, Kisiel and Limburg2024; Komaroff, Reference Komaroff2024). Several studies have reported the potential value of Fn in implementing CRC diagnosis, suggesting that it might act as a diagnostic biomarker. Wong et al. reported that the quantification (by real-time PCR) of faecal Fn combined with a faecal immunochemical test (FIT) could improve the diagnosis of advanced adenoma and CRC. Specifically, Fn increased the sensitivity of FIT for the detection of both CRC (from 73% to 92%) and advanced adenoma (from 15.5% to 38.6%). In terms of copy number, the relative abundance of faecal Fn DNA in CRC and advanced adenoma groups was 132 and 3.8 folds higher than in the control group, respectively (Wong et al., Reference Wong, Kwong, Chow, Luk, Dai, Nakatsu, Lam, Zhang, Wu, Chan, Ng, Wong, Ng, Wu, Yu and Sung2017) (Table 13). Similarly, Liang et al. found that with faecal Fn alone, the sensitivity and specificity of CRC diagnosis was 82.0% and 80.7%, respectively, while the sensitivity increased to 92.8% when faecal Fn was used in combination with FIT plus testing for the DNA of other faecal bacteria (i.e. Bacteroides clarus, Roseburia intestinalis, Clostridium hathewayi, and one undefined species, labelled as m7) (Liang et al., Reference Liang, Chiu, Chen, Huang, Higashimori, Fang, Brim, Ashktorab, Ng, Ng, Zheng, Chan, Sung and Yu2017) (Table 13).
Table 13. Data by Liang et al. (Reference Liang, Chiu, Chen, Huang, Higashimori, Fang, Brim, Ashktorab, Ng, Ng, Zheng, Chan, Sung and Yu2017), Wong et al. (Reference Wong, Kwong, Chow, Luk, Dai, Nakatsu, Lam, Zhang, Wu, Chan, Ng, Wong, Ng, Wu, Yu and Sung2017)
All p-value<0.05 reported in italic format are statistically significant. Abbreviations: CI, Confidence Interval.
a plus, other faecal bacteria (i.e. Bacteroides clarus, Roseburia intestinalis, and Clostridium hathewayi) Sensitivity = 82.4%.
b plus, other faecal bacteria (i.e. Bacteroides clarus, Roseburia intestinalis, and Clostridium hathewayi) Specificity = 81.5%.
Conversely, Aitchison et al. assessed the faecal amount of Fn in a cohort study involving 185 patients referred for FIT compared to CRC patients and age-matched controls (both n=57). The rate of positivity was higher in patients undergoing FIT and in CRC patients (47%) than in controls (7%; p<0.001), but no association was found between the carriage of Fn and FIT positivity (p=0.59). However, the presence of Fn in the stools was associated with an enhanced risk of finding colonic neoplastic lesions at colonoscopy (O.R. 3.1; p=0.02) (Aitchison et al., Reference Aitchison, Pearson, Purcell, Frizelle and Keenan2022).
It should be critically considered whether these promising results support the implementation of the molecular detection of Fn DNA in stool-based tests to be comparatively assessed with other molecular approaches improving the diagnostic yield of FIT (Imperiale et al., Reference Imperiale, Ransohoff, Itzkowitz, Levin, Lavin, Lidgard, Ahlquist and Berger2014, Reference Imperiale, Porter, Zella, Gagrat, Olson, Statz, Garces, Lavin, Aguilar, Brinberg, Berkelhammer, Kisiel and Limburg2024; Komaroff, Reference Komaroff2024).
However, published meta-analyses do not support unequivocal conclusions, as two would support Fn as a good diagnostic marker (Amitay et al., Reference Amitay, Werner, Vital, Pieper, Höfler, Gierse, Butt, Balavarca, Cuk and Brenner2017; Yu et al., Reference Yu, Guo, Yu, Sun, Ma, Han, Qian, Kryczek, Sun, Nagarsheth, Chen, Chen, Hong, Zou and Fang2017; Peng et al., Reference Peng, Cao, Li, Zhou, Zhang, Nie, Cao and Li2018), and another found data less encouraging (Huang et al., Reference Huang, Peng and Xie2018), and a third one found comparative metrics (Sze & Schloss, Reference Sze and Schloss2018) discouraging.
Most recently, and although with limitations, Zhang et al. suggested salivary Fn DNA as a non-invasive potential biomarker for the detection and prognosis of CRC. They showed that the relative level of Fn DNA was increased in the saliva of CRC patients compared to subjects with clean colonoscopies, hyperplastic polyps, or adenomas. Besides, Fn DNA had a performance in CRC diagnosis superior to carcinoembryonic antigen and carbohydrate antigen 19-9, its levels also being associated with overall and disease-free survival of the patients (Zhang et al., Reference Zhang, Zhang, Gui, Zhang, Zhang, Chen, Zhang, Wang, Zhang, Shang, Xin and Zhang2022).
However, to properly address these issues requires very large numbers, avoiding population bias, and additional translational efforts are needed before embarking on such large studies.
Fn as a potential molecular target for CRC treatment
In gastroenterology, phage therapy has been addressed mainly in infectious diseases, such as cholera. Currently, it is being explored in the eradication of Fn in CRC. Intestinal microbiota undergoes significant modifications in CRC (Kannen et al., Reference Kannen, Parry and Martin2019). Zheng et al. observed detrimental over-population of Fn in mice and patients, suppressing the beneficial butyrate-producing Clostridium butyricum. In human saliva, the authors isolated a temperate (i.e. lysogenic) phage capable of targeting and killing Fn without impacting the Clostridium butyricum population, thus reporting the occurrence of a natural system selectively controlling bacterial proliferation, which may turn into a possible therapeutic strategy. Additionally, they developed a phage-guided biotic–abiotic hybrid nano-system that could increase the chemotherapeutic potency of Irinotecan against CRC cells, whilst also selectively killing the Fn population, thus allowing at the same time the expansion of butyrate-producing bacteria (Zheng et al., Reference Zheng, Dong, Pan, Chen, Fan, Cheng and Zhang2019). Another report showed that Fn can modulate signalling pathways and activate autophagy, which may play a key role in mediating CRC chemoresistance. As mentioned, the authors demonstrated that Fn was significantly more represented in the CRC tissue of patients who had a post-chemotherapy recurrence of disease compared to those without recurrence. Such a contribution of Fn to chemoresistance can be mechanistically explained by interference with the TLR4 receptor, MYD88 signalling, and autophagy activation involving miRNAs. Among the latter, the reduction of miR-4802 and of miR-18a* is implicated in the accumulation of autophagosomes, indicating that Fn could activate autophagy through the miRNA modulation (Yu et al., Reference Yu, Feng, Wong, Zhang, Liang, Qin, Tang, Zhao, Stenvang, Li, Wang, Xu, Chen, Wu, Al-Aama, Nielsen, Kiilerich, Jensen, Yau and Wang2017).
In HCT116 MSI CRC cells, Fn was able to stimulate proliferation and migration and up-regulated the expression of c-MET, which in turn, if knocked-down, reduced such stimulation. Either endogenous or exogenous miR-139–5p weakened the effects of Fn in this cancer cell system, similar to c-MET knockdown, indicating that miRNAs are involved in mediating Fn-induced effects (Zhao et al., Reference Zhao, Tao, Li, Zheng, Liu and Liang2020). These different modulations in the response to therapy suggest Fn as a potential molecular target.
Accordingly, ongoing clinical trials are investigating therapeutic strategies to target this bacterium. A Phase 2 clinical trial involves the use of oral metronidazole to reduce Fn levels in CRC tissues, with the goal of assessing its impact on tumour cells and the surrounding microenvironment (Oncology Institute of Southern Switzerland, 2024).
Another ongoing study explores the potential benefits of combining metronidazole with postoperative chemotherapy in patients with high levels of Fn. Researchers aim to determine whether reducing this bacterium can enhance the effectiveness of chemotherapy in CRC treatment (Fang, Reference Fang2020).
These studies highlight the growing interest in targeting Fn as a potential strategy to improve colorectal cancer treatment outcomes.
Concluding remarks
Among the components of the gut microbiome, Fn has gained attention for its association with colonic neoplasia and with given molecular subtypes, particularly MSI-high CRC. The parallel associations with BRAF-mutated and CIMP+ tumours suggested that it may be involved in the development of sporadic MMR defects. This hypothesis has been refuted by suggesting a possible contribution of the molecular pathway of both hereditary and sporadic MMRd CRC subtypes in Fn colonization, independently of its role.
Fn high loads have also been associated with advanced stages and worse prognosis, suggesting that Fn can act as a disease modifier, enhancing CRC progression and reducing patient survival. Several data point to its capability of smouldering anti-tumour immunity while eliciting pro-tumoural inflammatory responses, which would contribute to its involvement in tumour progression. However, it remains to be established whether it acts as a driver of carcinogenesis or as an accelerator of the process, particularly in specific molecular settings. On the one hand, Fusobacteria could be detected in the biofilm covering colonic mucosa of patients with genetically determined FAP and the enrichment of Fn occurs already in early-stage carcinoma (Nakatsu et al., Reference Nakatsu, Li, Zhou, Sheng, Wong, Wu, Ng, Tsoi, Dong, Zhang, He, Kang, Cao, Wang, Zhang, Liang, Yu and Sung2015). On the other side, the rescue of bacterial DNA, as well as of live bacteria from metastatic lesions, resembles the maintenance of a clonal alteration conferring a selective advantage. Accordingly, treatment with antibacterial drugs inhibited the growth of Fn+ xenografts, suggesting its participation in the maintenance of clonal expansion. In addition, it has been shown that Fn can orchestrate the network of Toll-like receptors (TLRs), miRNAs, and ultimately autophagy, contributing to the emergence of chemoresistance. In a future perspective, the confirmation of these data might yield valuable insights to improve CRC clinical management.
In conclusion, although we are far from understanding the involvement of the components of the intestinal microbiome in the development and progression of CRC, Fn is paving the way by representing the first exogenous feature contributing to CRC molecular profiling.
Abbreviations
- 16S
-
Subunit 16
- 5-FU
-
5-fluorouracil
- APC
-
Adenomatous Polyposis Coli
- BECN1
-
Beclin 1
- BMI
-
Body Mass Index
- BRAF
-
B-Raf Proto-Oncogene, Serine/Threonine Kinase
- CAPOX
-
Capecitabine plus oxaliplatin
- CD3
-
Cluster of Differentiation 3
- CD45RO
-
or UCHL1, ubiquitin carboxyl terminal esterase L1
- CD8
-
Cluster of Differentiation 8
- CHD7
-
Chromodomain Helicase DNA Binding Protein 7
- CHD8
-
Chromodomain Helicase DNA Binding Protein 8
- CIMP
-
CpG island methylator phenotype
- CIN
-
Chromosomal Instability
- CRC
-
Colorectal cancer
- CRIS A–E
-
CRC intrinsic subtypes A–E
- CXCL16
-
C-X-C Motif Chemokine Ligand 16
- EMT
-
Epithelial-to-Mesenchymal Transition
- EPCAM
-
Epithelial Cell Adhesion Molecule
- FadA
-
Adhesin A
- FAP
-
Familial adenomatous polyposis
- Fap2
-
Fusobacterium autotransporter protein 2
- FIT
-
Faecal Immunochemical Test
- Fn
-
Fusobacterium Nucleatum
- FOLFOX
-
Leucovorin (folinic acid), fluorouracil, plus oxaliplatin
- FOXP3
-
Forkhead Box P3
- Gal-GalNAc
-
D-galactose N-Acetyl-D-galactosamine
- HR
-
Hazard-ratio
- IL-6
-
Interleukin 6
- KRAS
-
KRAS Proto-Oncogene, GTPase
- LARC
-
Locally Advanced Rectal Cancer
- LPS
-
Lipopolysaccharides
- MAP1LC3 (LC3)
-
Microtubule Associated Protein 1 Light Chain 3 Alpha
- MDSCs
-
Myeloid-derived suppressor cells
- miRNAs
-
Micro RiboNucleic Acid
- MLH1
-
MutL Homolog 1
- MMR
-
DNA Mismatch Repair
- MMRd
-
DNA mismatch repair deficiency
- mRNA
-
Messenger Ribonucleic Acid
- MSH2
-
MutS Homolog 2
- MSH6
-
MutS Homolog 6
- MSI
-
Microsatellite Instability
- MSS
-
Microsatellite Stable
- MUTYH
-
MutY DNA Glycosylase
- MYD88
-
Myeloid differentiation primary response 88
- NF-kB
-
Nuclear factor-κB
- NGS
-
Next Generation Sequencing
- NusG
-
Transcription termination factor NusG
- OR
-
Odds-ratio
- PCR
-
Polymerase Chain Reaction
- PMS2
-
PMS1 Homolog 2
- qPCR
-
Quantitative Polymerase Chain Reaction
- rDNA
-
Ribosomal Desoxy- (or Deoxy-) riboNucleic Acid
- RIG-1
-
Retinoic acid-inducible gene 1
- RNA
-
RiboNucleic Acid
- RNA-ISH
-
RNA in situ hybridization
- rpoB
-
RNA polymerase subunit beta
- SQSTM1 (p62)
-
Sequestosome 1
- SSAs
-
Sessile Serrated Adenomas
- TIGIT
-
T Cell Immunoreceptor With Ig And ITIM Domains
- TIL
-
Tumour-infiltrating lymphocyte
- TLRs
-
Toll-like receptors
- TLR4
-
Toll-like receptor 4
- TNF
-
Tumour necrosis factor
- TNM
-
tumour (T), nodes (N), and metastases (M)
- TP53
-
Tumour Protein P53
- TWIST1
-
Twist Family BHLH Transcription Factor 1
- Wnt
-
Wingless-related integration site
Author contribution
Conceptualization, G.L., and L.L.; Writing Original Draft, G.L., and L.L.; Writing Original Draft Support, R.F., F.C., M.C., G.F., and M.M; Writing Review and Editing, G.L., L.L., and R.F.; Supervision, G.L., L.L., B.F., M.A., and R.L.
Funding
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Disclosure statement
The authors declare none.
Open access
