The gut microbiome and potential implications for early-onset colorectal cancer

Abstract

Recently, there has been an unexpected trend toward increased incidence of colorectal cancer in younger individuals, particularly distal colon and rectal cancer in those under age 50. There is evidence to suggest that the human gut microbiome may play a role in carcinogenesis. The microbiome is dynamic and varies with age, geography, ethnicity and diet. Certain bacteria such as Fusobacterium nucleatum have been implicated in the development of colorectal and other gastrointestinal cancers. Recent data suggest that bacteria can alter the inflammatory and immune environment, influencing carcinogenesis, lack of treatment response and prognosis. Studies to date focus on older patients. Because the microbiome varies with age, it could be a potential explanation for the rise in early-onset colorectal cancer.

Keywords::

• colon cancer
• colorectal cancer
• early-onset
• Fusobacterium nucleatum
• microbiome
• microbiota
• rectal cancer
• tumor microenvironment

Colorectal cancer (CRC) is the fourth-leading cause of cancer incidence and the second-leading cause of cancer-related death in the USA [1,2]. Since the 1980s, the incidence of CRC in the USA has declined among individuals 50 years and older. Between 1985 and 2013, an estimated decline of 32% was observed, which can be attributed to the increasing use of screening colonoscopies [3,4]. In contrast, there was a 1.6% increase in the incidence of CRC – primarily distal colon and rectal tumors – in patients less than 50 years of age between 2010 and 2013. This trend was consistent among various ethnicities in the USA and was also observed in other high-income countries [3,5–7]. In addition, the distribution of CRC when diagnosed also varies by age. Between proximal, distal and rectal tumors, most CRCs diagnosed in men and women under age 50 are rectal tumors (41% in men and 36% in women), and proximal tumors become more common in older age groups [3].

Several attempts have been made to elucidate the differences between early- and later-onset CRC (LOCRC). Risk factors such as obesity, dietary patterns and lifestyle have been explored, although other studies suggest early-onset CRC (EOCRC) is not associated with typical risk factors such as obesity and diabetes but is rather a phenotypically distinct disease with a unique cancer biology [8–10]. Young CRC patients are more likely than older patients to have a hereditary CRC syndrome, and as many as 35% of patients with CRC under the age of 35 have an identifiable syndrome [11]. Still, the vast majority of patients with EOCRC do not have a hereditary CRC syndrome. It is believed that some factors, such as diet, antibiotics, stress and obesity, are key suspects in the promotion of EOCRC [10]. However, there is a need for well-designed observational epidemiological studies as well as investigations into the microbiome’s role as an independent risk factor or possible potentiator of other risk factors in EOCRC.

In the past few years, researchers have focused increasingly on the microbiome and its implications in various cancers, including CRC. This review will explore our current understanding of the microbiome’s role in CRC, focusing particularly on the bacteria Fusobacterium nucleatum (F. nucleatum), and the potential differences in intratumoral microbiome between EOCRC and LOCRC.

Microbiome overview

The human microbiome is defined as the composite genome of all the microbes in the human body. There are an estimated 10¹³ to 10¹⁴ bacteria (also fungi and viruses) harbored by each human, and most of these organisms reside in the gut [12]. The costs of DNA sequencing decreased sufficiently by the mid-2000s for there to be increased initiatives aimed at characterizing the human microbiome. One such initiative was the Human Microbiome Project at the NIH [13]. The ability to perform metagenomic studies has provided key insight into the composition of the human microbiome and its role in various medical conditions, including depression, diabetes, autoimmune diseases and cancer [14–16].

The human gut contains the majority of the microbiome and is composed of complex networks between bacteria and the human host cells. When in equilibrium, the gut microbiome helps modulate digestion and metabolism, aids in immunity and prevents foreign pathogens from invading. The human microbiota is established after birth and stabilizes during the first 2 years of life as a highly dynamic system dominated by Bifidobacteria. Thereafter, it increases in diversity and richness until peak complexity in the human adult with hundreds of species dominated by the phyla Bacteroidetes and Firmicutes [17–19]. The microbiome varies with nutritional status, exposure to breastfeeding, obesity, antibiotic treatment at a young age and ethnicity [20,21].

Multiple bacteria and specific mechanisms through which the microbiome modulates the immune system have been studied [22]. For example, studies show that commensal intestinal bacteria can modulate regulatory T-cells and gut homeostasis through microbiota fermentation-derived short-chain fatty acids [23]. Studies also suggest that dysbiosis and decreased diversity of the gut microbiome are linked to numerous diseases, including CRC [24,25]. This research has prompted further efforts to understand the composition of the microbiome at various chronological and medical stages in humans to ultimately understand disease biology and propose therapeutic interventions.

Microbiome, colonic inflammation & carcinogenesis

Chronic inflammatory states are a risk factor for many cancers, including CRC. Inflammatory bowel diseases (IBD), such as Crohn’s disease (CD) or ulcerative colitis (UC), especially increase an individual’s risk of developing colitis-related CRC, with a CRC incidence of 2% after 10 years of chronic inflammation, 8% after 20 years and 18% after 30 years [26]. Proposed mediators of inflammation, such as cytokines (TNF-α, IL-6), chemokines (COX-2 and NF-κB) and reactive oxygen species, are associated with carcinogenesis through proliferative cell signaling and enhancement of DNA damage in epithelial cells [27]. An important consideration then is whether the microbiota, which functions close to the gut epithelial cells, can cause these inflammatory states and carcinogenesis.

Early animal models suggest that the microbiota is altered by states of inflammation and may also directly influence carcinogenesis. Arthur et al. studied mouse models and noticed that different microbial clusters, which were varied in diversity but not richness, existed between wild type and colitis/cancer mice [28]. In particular, colitis/cancer mice harbored an expansion of Escherichia coli (E. coli) in the gut. The group also observed bacteria-specific factors such as genotoxins in addition to inflammation that might be required for colitis-related CRC; mice associated with murine adherent-invasive E. coli NC101 developed more invasive cancers, whereas human commensal Enterococcus faecalis (E. faecalis) rarely resulted in tumors in these animals [28].

Diet may also contribute to intestinal inflammation and microbiome alterations. Western diets, characterized by a high intake of red meats, processed meats and fats, have been linked to a higher risk of developing CRC [29,30]. Various mechanisms of carcinogenesis have been proposed, such as intestinal DNA damage from animal-derived proteins; microbiota metabolic end-products such as urea, hydrogen sulfide and heme; and carcinogenic effects from by-products of high-temperature cooking [31]. High-fat diets are associated with increased intestinal stem cells, inflammation, bile acid dysregulation, serum leptin levels and cell permeability, all of which may contribute to tumorigenesis [32–35]. The microbiome also varies based on dietary profiles, and different types of meat and cooking methods have been associated with varying stool bacterial profiles, suggesting that diet and food preparations may alter the microbiome and contribute to carcinogenesis [36,37]. For example, studies indicate that commensal sulfate-reducing bacteria (such as Fusobacterium, Desulfovibrio and Bilophila wadsworthia), which produce genotoxic hydrogen sulfide, may contribute to carcinogenesis by inhibiting DNA repair [38]. These bacteria positively correlate with a diet high in fat and animal protein, which is also a known CRC risk factor. Moreover, a recent study found that sulfur-metabolizing bacteria were associated with diets higher in processed meats, low-calorie drinks and low vegetable consumption and were associated with a risk of developing distal colon and rectal cancers [39,40]. In addition, chronic alcohol use is associated with a higher risk of developing CRC. Studies suggest that components of the microbiome, especially certain obligate anaerobes under aerobic conditions, contribute to CRC by metabolizing ethanol to carcinogenic acetaldehyde in the colon and rectum [41–43].

Investigators from Johns Hopkins University reported an association between bacterial biofilms and carcinogenesis [44]. The authors noted IL-6 enrichment and reduced epithelial cell E-cadherin expression in colons with bacterial biofilm formation. Furthermore, the risk of developing CRC was fivefold higher in patients with biofilms compared with those without biofilms, proposing yet another mechanism for carcinogenesis. Interestingly, the vast majority of right-sided tumors had biofilms, and patients with CRCs (right- or left-sided) that were biofilm positive also had normal colonic mucosa coated with biofilm.

Below, we review what appear to be the major bacteria associated with CRCs and summarize the associated mechanisms of oncogenesis proposed in the literature along with prognostic and potential therapeutic considerations (Table 1).

Table 1. Microbiome and specific bacteria studies focusing on association with oncogenesis, risks and mechanisms of carcinogenesis, immunomodulation and therapy responses.

Bacterium	Disease, population, (n = population size) and major findings	Ref.
Fusobacterium species	CRC, Human (n = 95 carcinoma/normal colon pairs) CRC is enriched with Fusobacterium species, especially phylotypes similar to F. nucleatum, F. mortiferum and F. necrophorum evidenced by 16S rDNA sequencing in tumors and with FISH probes for Fusobacterium in CRC No correlation noted between F. nucleatum amount and age	[48]
Fusobacterium species F. nucleatum	CRC, Human (n = 99 carcinoma/normal colon pairs) F. nucleatum is over-represented in CRC tumor specimens compared with normal controls by quantitative PCR (p = 2.52 × 10^-6) Subjects with higher proportion of Fusobacterium were more likely to have regional lymph node metastasis (p = 3.5 × 10^-3) No correlation noted between F. nucleatum amount and age	[49]
F. nucleatum	CRC, Human (n = 84) F. nucleatum stimulates proliferation of human CRC cells with FadA (virulence factor) by adhering to and invading E-cadherin-expressing CRC cells (p < 1.0 × 10^-3) FadA complexes activate E-cadherin-mediated cellular signaling and proliferation via β-catenin-regulated transcription (p < 1.0 × 10^-3) CRC and precancerous adenoma patients have elevated FadA gene colonic expression compared with healthy patients (p < 1.0 × 10^-3) Mice xenografts (n = 30) FadA promoted E-cadherin-mediated CRC growth and inflammation in vivo (p < 1.0 × 10^-3)	[56]
F. nucleatum	CRC, Human (n = 86) F. nucleatum is enriched (PCR of stool) in CRC patients (p < 1.0 × 10^-5) and those with adenomas (p < 5.0 × 10^-3) compared with healthy controls F. nucleatum-positive CRC is enriched with proinflammatory gene expression Mice F. nucleatum accelerated tumorigenesis in mice with mutation in one copy of tumor-suppressor gene APC or genetic defects resulting in chronic inflammation (p < 1.0 × 10^-4) (n = 12) F. nucleatum expands myeloid-derived immune cells but not lymphoid immune cells in tumor microenvironment (p < 1.0 × 10^-3) (n = 15)	[60]
Fusobacterium species	CRC, Human (n = 310) Fusobacterium is detectable in 74% (111/149) of CRC tissues and heavily enriched in 9% (14/149) F. nucleatum exhibits 3600-fold and pan-Fusobacterium exhibits 250-fold enrichment in CRC compared with adjacent normal mucosa (both p-values <1.0 × 10^-4) Fusobacterium-high CRC was associated with CIMP positivity (p = 1.0 × 10^-4), TP53 wild-type (p = 1.5 × 10^-3), hMLH1 methylation present (p = 2.8 × 10^-3), MSI (p = 1.8 × 10^-2), CHD7/8 mutation present (p = 2.0 × 10^-3). Data showed high Fusobacterium level had distinct molecular pattern and high rate of mutations overall, thus suggesting pathogenic rather than passenger role Median age of patients of whose CRCs were studied was over 60 years and a significant trend toward F. nucleatum-high status and patient age greater than 70 (p = 0.09) was noted	[50]
F. nucleatum	CRC, Human (n = 598) F. nucleatum-positive CRC were inversely associated with CD3^+ T-cell density (p = 6.0 × 10^-3) No correlation noted between F. nucleatum-status and age (mean age of all patients was 67.2 ± 8.4 years with mean ages of 67.2 ± 8.5, 67.7 ± 7.4 and 6.68 ± 7.3 in F. nucleatum negative, low and high CRC (p = 0.89)	[53]
F. nucleatum	CRC, Human (n = 1069) Higher levels of F. nucleatum in CRC tissue were associated with decreased survival (HR for cancer-specific mortality in F. nucleatum-low vs -high CRC were 1.25 and 1.58, respectively [p = 0.02]) No correlation noted between F. nucleatum-status and age (mean age of all patients was 69.3 ± 8.8 years with mean ages of 69.2 ± 8.8, 71.1 ± 9.0 and 68.8 ± 8.3 in F. nucleatum negative, low and high CRC [p = 0.21])	[62]
F. nucleatum	CRC, Human (n = 98) F. nucleatum activates β-catenin signaling via TLR4/P-PAK1/P-β-catenin S65 cascade (p < 0.05) No correlation noted between F. nucleatum-status and age (mean age: 59.4 vs 58.0 for F. nucleatum-negative vs -positive CRC, respectively [p = 0.281])	[58]
F. nucleatum	CRC, Human CRC cell lines and mice xenografts F. nucleatum promotes CRC proliferation and invasion in CRC cell lines (p < 1.0 × 10^-3 to 0.05) (n = 3) and mouse xenograft animal models (p < 0.01) (n = 15) F. nucleatum promotes tumorigenesis in APC^Min/+ mice (p < 0.05) (n = 10) F. nucleatum promotes tumor proliferation through miRNA-21 and MAPK signaling pathway (p < 0.05) (n = 8 mice) Human (n = 125) Tumors with high amounts of F. nucleatum DNA and miRNA-21 had shorter survival times	[59]
F. nucleatum	CRC, Human (n = 11 primary CRCs and paired liver metastases and n = 201 primary liver tumors) Fusobacterium is viable and migrates with CRC to distant metastasis sites such as liver, suggesting the microbiota are intrinsic and essential to the microenvironment Fusobacterium is rare in primary liver carcinoma and enriched in liver metastases from CRCs compared with primary liver carcinoma (p = 8.0 × 10^-3) Human cell lines and mice xenografts Fusobacterium survives in CRC PDXs through multiple generations (n = 1) Treatment of Fusobacterium-positive PDX model with metronidazole decreases Fusobacterium load and tumor growth (p = 2.0 × 10^-3 for both) (n = 47)	[61]
F. nucleatum	CRC, Human (n = 1019) Prudent diet score was associated with lower risk of F. nucleatum-positive cancers (p = 3.0 × 10^-3) No significant trend noted with prudent diet and F. nucleatum-negative cancer and no significant heterogeneity between F. nucleatum-status CRC in relation to Western dietary pattern scores	[54]
F. nucleatum	CRC, Human (n = 92 in F. nucleatum and recurrence study and n = 173 in validation study for prediction value) F. nucleatum is enriched in recurrent CRC tissue as compared with nonrecurrent CRC (p < 0.01), suggesting it plays a role in recurrence and is an independent predictor of CRC aggressiveness F. nucleatum promotes cancer autophagy activation (p < 0.05) to induce chemotherapy resistance	[51]
F. nucleatum	CRC, Human (n = 1041) Association of F. nucleatum with TIL varies by MSI status In MSI-H CRC, presence of F. nucleatum was negatively associated with TILs (OR: 0.45; CI: 0.22–0.92) In non-MSI-H CRC, presence of F. nucleatum was positively associated with TILs (OR: 1.91; CI: 1.12–3.25)	[92]
F. nucleatum	CRC, Human CRC cell lines F. nucleatum stimulates CRC growth in in vitro model via Annexin A1 (modulator of Wnt/β-catenin signaling) (p < 1.0 × 10^-3 to <0.05) (multiple cell line experiments performed in triplicates and repeated three- to four-times) Expression of Annexin A1 expression in CRC is a predictor of poor prognosis (p < 1.0 × 10^-3) (n = 466) Proposed ‘two-hit’ model where a somatic mutation serves as the first hit and F. nucleatum serves as the second hit based on a positive feedback loop seen between FadA and Annexin A1 that was absent in noncancerous cells	[57]
F. nucleatum	CRC, Human (n = 593) In stage II/III CRC treated with adjuvant therapy, prognostic value of F. nucleatum depends on tumor location and MSI status Disease-free survival was improved in F. nucleatum-high CRC with nonsigmoid CRCs (p = 2.6 × 10^-2) In multivariate analysis, F. nucleatum was an independent favorable prognostic factor for nonsigmoid CRC (HR: 0.42; CI: 0.18–0.97, p = 4.3 × 10^-2) only in non-MSI-H subset (p = 1.4 × 10^-2) No correlation noted between F. nucleatum-status and age (41.2 vs 58.8% of F. nucleatum-high CRCs were younger than 59 vs ≥59 years, respectively, and 45.8 vs 54.2% F. nucleatum-low/-negative CRCs were younger than 59 vs ≥59 years, respectively) (p = 0.286)	[93]
B. fragilis	CRC, Human cell lines bft activates T-cell factor dependent β-catenin nuclear signaling and c-myc expression resulting in colonic cell transformation suggesting the potential to contribute to oncogenesis (n = 4)	[67]
B. fragilis	CRC, Human (n = 132) First study demonstrating increase of ETBF in CRC patients bft was detected by PCR in stool samples of 38% CRC patients compared with 12% control patients (p = 9.0 × 10^-3)	[65]
B. fragilis	CRC, Mice ETBF activates Stat3 (n = 13 mice) and induces T-helper cell-type 17 inflammatory infiltrates (n = 3–5 mice per group and experiments repeated two- to three-times) and demonstrated a role for endogenous T-cell immune responses in carcinogenesis	[68]
B. fragilis	CRC, Human (n = 98) Mucosa of CRC patients are more often bft-positive than controls (p = 0.04) Late-stage CRC had trend of increased bft positivity compared with early-stage CRC (100 vs 72.7%, p = 9.3 × 10^-2)	[66]
B. fragilis	CRC, Mice (n = 70) Oral administration of low-dose recombinant BFT-2 inhibited carcinogenesis (p < 0.05)	[70]
B. fragilis	CRC, Human cell lines PSA from B. fragilis induces IL-8 via TLR-2 and suppresses migration and invasion of CRC (p < 0.05) (n = 3 experiments)	[71]
S. gallolyticus	CRC, Human (n = 141) S. gallolyticus is presented at higher rates in CRC (20.5% in CRC with bacteremia, 17.3% without bacteremia) compared with control tissues (2%) (p < 0.05) CRC tissue was associated with higher expression (mRNA ratios compared with controls) of IL-1 (1.77), COX-2 (1.63), IL-8 (1.73) (p < 0.05) No correlation noted between detection of S. gallolyticus and age (mean age in controls, CRC without bacteremia and CRC with bacteremia were 57.4 ± 4.7, 59.22 ± 8.13 and 56.6 ± 6.7, respectively) (p = 0.41)	[74]
S. gallolyticus	CRC, Human cell lines S. gallolyticus promotes colon cancer proliferation (p < 0.05) and the effect is dependent on bacterial growth phase (p < 0.05), direct contact between live bacteria and responsive cells (p < 0.05) (cell lines tested with duplicate wells and experiments repeated at least three-times), and β-catenin pathway (p < 0.01) (n = 3 experiments) Mice xenografts S. gallolyticus promoted tumor growth in xenograft models (p < 0.05) (n = 10) Mice (n = 19) S. gallolyticus promotes tumorigenesis in AOM-induced mouse models of CRC (p < 0.05) via β-catenin pathway	[75]
S. gallolyticus	CRC, Mice S. gallolyticus colonization is 1000-fold higher in genetically-prone tumor-bearing mice compared with normal mice (p < 0.01) (n = 17) S. gallolyticus-specific bacteriocin kills commensal gut enterococci in vitro and bile acids enhance this bacteriocin in vivo resulting in increased S. gallolyticus colonization (<0.05). Wnt pathway activation and increased intestinal bile acids create optimal environment for S. gallolyticus colonization	[76]
S. gallolyticus	CRC, Humans (n = 8126) There is a higher risk for CRC in patients who have positive antibody response to S. gallolyticus pilus protein Gallo2178 (OR: 1.23; CI: 0.99–1.52, not statistically significant) with stronger association for cases diagnosed less than 10 years from blood draw (OR: 1.40; CI: 1.09–1.79) compared with those diagnosed more than 10 years from blood draw (OR: 0.79; CI: 0.50–1.24)	[78]
H. pylori	Gastric cancer, Humans (n = 795) Pre-neoplastic gastric lesions regress as a function of the square of H. pylori-negative time Patients who were H. pylori-negative at 12 years experienced 14.8% more regression and 13.7% less progression of pre-neoplastic lesions compared with H. pylori-positive patients at 12 years (p = 1.0 × 10^-3), suggesting patients with pre-neoplastic gastric lesions and H. pylori infection should be treated and cured of infection	[81]
H. pylori	CRC, Humans (n = 384) Meta-analysis of 11 human studies estimated an OR of 1.4 (CI: 1.1–1.8) for association of H. pylori to CRC	[84]
H. pylori	Gastric cancer, Humans (n = 403) H. pylori eradication in precancerous tissue resulted in regression of gastric atrophy (62% regression in antrum and 36% in corpus) when compared with non-H. pylori-eradicated cases (31% regression in antrum and 13% in corpus) (p < 1.0 × 10^-4) over 2.5 years No regression of intestinal metaplasia was seen with H. pylori treatment over 4.5 years, but progression of intestinal metaplasia was more rapid in non-H. pylori-eradicated versus eradicated cases (p < 0.05)	[82]
H. pylori	CRC, Humans (n = 7679) Meta-analysis of nine human studies estimated OR: 0.18 (CI: 0.99–1.40) (p > 0.05) for association between H. pylori infection and CRC risk No statistical association between H. pylori infection and colorectal neoplasm risk was found OR for H. pylori infection and adenomas was 1.8 (CI: 1.35–2.51) (p < 0.01), thus H. pylori may increase risk of developing colorectal adenomas	[85]
General microbiome	PDA, Humans (n = 12) PDA harbors more microbiome with select bacterial profile compared with normal pancreas and normal gut (p < 1.0 × 10^-4) PDA, Mice Germ-free mice and mice with PDA treated with antibiotics were protected against PDA progression (p < 0.01) (n = 8, experiments repeated five-times) Fecal transfer from PDA-bearing mice to genetically susceptible mice hosts accelerated tumorigenesis (p < 0.01) (experiment repeated three-times) Antimicrobial-ablated mice models increased intratumoral immunogenicity with increased CD8:CD4 T-cell ratio and increased expression of PD-1 (p < 0.01–0.05) (n = 10, experiment repeated five-times) Study suggested PDA microbiome programs TAMs via TLR signaling to induce immune tolerance by suppressing T-cell immunity	[87]
General microbiome	PDA, Humans (n = 43 in discovery cohort and n = 25 in validation cohort) PDA microbial diversity was higher in long-term survivors and patients with high diversity had longer survival than those with low diversity (median OS: 9.66 vs 1.66) (p = 1.6 × 10^-4) Long-term survivor PDA were enriched with Pseudoxanthomonas, Saccharopolyspora and Streptomyces Long-term survivors had greater CD3^+, CD8^+ and granzyme B^+-cell expression than short-term survivors (p = 2.73 × 10^-2, p < 1.0 × 10^-4 and p = 0.04, respectively), suggesting microbiome drives antitumor immune responses Mice Gut-tumor microbiome cross-talk exists. Human-to-mouse FMT from human short-term survivors into mice with orthotopic syngeneic tumors showed a small proportion (5%) of mice tumor-harbored human donor microbiome Human-to-mouse FMT from long-term survivors reduced tumor growth and enriched tumors with activated T-cell infiltration relative to FMT from short-term survivors or healthy donors (p < 1.0 × 10^-3)	[88]
General microbiome	Melanoma, Humans (n = 43) Melanoma patients undergoing anti-PD-1 therapy who were responders showed higher alpha diversity (p < 0.01) and relative increase in Ruminococcaceae family (p < 0.01) Mice A ‘favorable’ gut microbiome (e.g., high diversity and abundance of Ruminococcaceae/Faecalibacterium) is associated with increased immune surveillance as seen by reduced tumor growth (p = 0.04) (n = 11), improved responses to anti-PD-L1 therapy (p = 0.20) (n = 7) and higher CD8^+ tumor density (p < 0.05) (n = 6) in long-term survivor human-to-mouse FMT compared with short-term survivor human-to-mouse FMT studies	[89]

AOM: Azoxymethane; bft: Entero-toxigen gene; CRC: Colorectal cancer; ETBF: Entero-toxigenic B. fragilis; FMT: Fecal microbiota transplant; HR: Hazard ratio; MSI: Microsatellite instability; MSI-H CRC: Microsatellite instability-high colorectal cancer; OR: Odds ratio; OS: Overall survival; PDA: Pancreatic ductal adenocarcinomal; PDX: Patient-derived xenograft; PSA: Polysaccharide A; TAM: Tumor-associated macrophage; TIL: Tumor-infiltrating lymphocyte; TLR: Toll-like receptor.

Fusobacterium nucleatum

F. nucleatum is a gram-negative obligate anaerobic species belonging to the Bacteroidaceae family. It is a commensal, oral biofilm-forming organism known to contribute to periodontal disease and dental plaque, abscesses of the head and neck and pregnancy complications when translocated from the maternal oral cavity to the intrauterine cavity [45–47]. The intratumoral microbiota of CRC has been characterized by whole genome sequencing and RNA expression specific to F. nucleatum [48,49]. Tahara et al. used molecular PCR studies and revealed that Fusobacterium in cancer tissue was associated with a distinct molecular pattern involving high levels of CpG island methylation and high mutational burden, arguing that bacteria played a pathogenic role rather than a passenger role [50]. The authors also reported an association between high F. nucleatum levels and tumors that were TP53 wild-type, CHD7/8 mutated, MLH1 methylated and microsatellite instability-high (MSI-H). Individuals with CRC over age 70 also had a trend toward F. nucleatum positivity. This research group also found that if normal colorectal tissues were Fusobacterium-enriched, an adjacent CRC was 15-times more likely to be Fusobacterium-enriched compared with tumors adjacent to Fusobacterium-free colorectal tissues. Furthermore, tumors with a high degree of CpG island methylation were more likely to have adjacent normal tissue enriched with Fusobacterium than unmethylated tumors (29.2 vs 6.8%, p = 0.03). Other studies have found F. nucleatum levels are significantly higher in tumor tissue compared with adjacent normal tissue [51–53].

The presence of F. nucleatum also varies with diet, as prudent diets rich in whole grains and fiber are associated with a lower risk of F. nucleatum-positive CRC but not F. nucleatum-negative CRC [54]. Although Western diets were associated with a higher risk of F. nucleatum-positive CRC, this was not a statistically significant association. These data, along with meta-analysis data supporting a higher risk of CRC in Western diets compared with ‘healthy’ diets [55] – characterized by high vegetable, fruit, grains, fish, poultry and low-fat dairy intake – suggest that diet may affect the microenvironment and promote F. nucleatum survival and development of F. nucleatum-positive CRC.

Oncogenesis

Several mechanisms of F. nucleatum-related oncogenesis have been reported in cell lines. Rubinstein et al. reported a mechanism involving E-cadherin/β-catenin signaling through an E-cadherin ligand unique to F. nucleatum, FadA [56]. FadA mediates F. nucleatum binding to CRC and non-CRC tissue. Purified FadA and FadA-positive F. nucleatum stimulated CRC cell line growth but not the growth of noncancerous cell lines, suggesting that the E-cadherin/β-catenin signaling mechanism may promote oncogenesis after a mutation occurs in normal tissue. In 2019, the same group reported another study suggesting that FadA causes upregulation of Annexin A1, a modulator of Wnt/β-catenin signaling. This time, Rubinstein et al. proposed a ‘two-hit’ model of CRC carcinogenesis with the first hit being a somatic mutation and the second hit being F. nucleatum, promoting cancer progression [57].

Chen et al. suggested that invasive F. nucleatum activates β-catenin signaling through the TLR4/P-PAK/P-β-catenin S675 cascade, and this research group proposed TLR4 and PAK1 as potential pharmaceutical targets for treatment [58]. Yang et al. found that F. nucleatum activates TLR4, leading to activation of NF-κB and increased levels of microRNA-21, which functions as an oncogene by inactivating tumor-suppressor genes [59].

Other studies questioned whether F. nucleatum affects the microenvironment immune milieu, thereby allowing CRC to evade the immune system and proliferate. Mima et al. analyzed 598 CRC tumors and found that higher amounts of F. nucleatum in tumor tissue were associated with a lower density of CD3⁺ T-cells (p = 0.012) but not with lower densities of CD8⁺, CD45RO⁺ or FOXP3⁺ T-cells (not statistically significant). This cross-sectional study could not exclude possible reverse causation (i.e., the possibility that higher T-cell density may eradicate F. nucleatum) [53]. However, other studies suggest that F. nucleatum induces proinflammatory cytokine profiles and induces T-cell suppressive activity, thereby downregulating antitumor T-cell-mediated immunity and promoting cancer progression [60].

A study by Bullman et al. reported that F. nucleatum and other bacteria not only reside in primary CRCs but also travel with tumor metastases to distant sites such as the liver [61]. F. nucleatum was not associated with primary liver hepatocellular carcinoma, suggesting that F. nucleatum is intrinsic to the CRC microenvironment. They then treated F. nucleatum-positive human CRC xenografts in murine models with metronidazole and noted a statistically significant decrease in tumor growth and load compared with F. nucleatum-negative human CRC xenograft controls. Bullman et al., along with others, collectively propose causation between F. nucleatum and carcinogenesis and suggest the possibility of therapeutic targeting against F. nucleatum.

Prognosis

Mima et al. subsequently looked at pathology from 1069 CRC tumors in the Nurses’ Health Study and Health Professionals Follow-up Study to determine the prognostic value of F. nucleatum presence in tumors [62]. Of note, the median age in this study of all patients was 69 years. The study found that the amount of F. nucleatum present in CRC was associated with higher CRC-specific mortality (p = 0.023 in univariate and p = 0.20 in multivariable regression analyses). The hazard ratio (HR) for CRC-specific mortality was also increased in F. nucleatum-high cases (HR: 1.58) compared with F. nucleatum-low cases (HR: 1.25) when compared with F. nucleatum-negative cases. This study also reported that F. nucleatum-high tumors were associated with proximal tumor location, higher T stage, poor tumor differentiation, MSI-H status, MLH1 hypermethylation, CIMP-high status and BRAF mutation (p ≤ 0.001). However, unlike the results from Tahara et al., this study did not find an association between tumor F. nucleatum and CIMP-high or BRAF mutation upon multivariate analysis that adjusted for each other.

Bullman et al. reported more F. nucleatum-enriched tumors and metastasis in the cecum and ascending (right-sided) CRCs compared with rectal tumors. Among the right-sided tumors, those with a higher F. nucleatum load had inferior overall survival (OS). These results suggest that F. nucleatum-high cancer represents a more aggressive cancer subtype with a poorer prognosis. The study by Oh et al. in the adjuvant setting demonstrated that F. nucleatum-high tumors were actually associated with improved disease-free survival compared with F. nucleatum-low/negative tumors for nonsigmoid non-MSI-H tumors. However, a trend toward the opposite was found for non-MSI-H sigmoid and rectal cancers. Therefore, the prognostic implications of F. nucleatum-positivity may vary by stage and tumor sidedness.

Bacteroides fragilis

Bacteroides fragilis (B. fragilis) is a well-characterized human colonic anaerobic bacterium that colonizes colorectal mucosa but only makes up 0.5–1.0% of the gut microbiome [63]. B. fragilis consists of two molecular subtypes: non-toxigenic B. fragilis and entero-toxigenic B. fragilis (ETBF). ETBF secretes B. fragilis toxin (BFT), which is encoded by the bft gene and causes acute diarrheal illness. There are limited data to suggest an association with ETBF and IBD [64]. Early studies showed the bft gene was detected more frequently in the stool of patients with a history of CRC compared with normal controls, suggesting an increased prevalence of ETBF in patients with CRC [65]. The gene was also more frequent in both right- and left-sided human colonic tumors compared with normal control mucosa, with increased gene expression in late-stage CRC patients [66]. Such findings suggest the bft gene and ETBF are associated with CRC pathogenesis.

Wu et al. studied human colonic epithelial cell clones and reported that BFT activated T-cell factor-dependent β-catenin signaling, with subsequent c-myc transcription resulting in cellular proliferation [67]. A later study by the group showed a Stat3- and T-helper-cell type-17-dependent mechanism for inflammation-induced oncogenesis [68]. Another study suggested a mechanism where ETBF mediates carcinogenesis through long noncoding RNA, BFAL1. BFAL1 was expressed more in tumor cells compared with adjacent tissue and was seen to activate the mTOR pathway. The authors suggested that ETBF and BFAL1 levels may have prognostic value because patients with higher levels of both had a worse prognosis. Dejea et al. studied the mucosa of patients with familial adenomatous polyposis (FAP), a hereditary condition caused by germline mutations in APC tumor-suppressor genes, which promotes polyp and CRC development [69]. The research group noted that FAP mucosa showed patchy areas of bacterial mucus invasion made up predominantly of E. coli and B. fragilis. Study results suggested that co-colonization with E. coli and B. fragilis promoted carcinogenesis through toxin-mediated colonic epithelial damage and IL-17 induction in FAP mucosa and sporadic CRC with APC loss.

Interestingly, a study by Lv et al., which initially set out to prove how BFT-2 promotes CRC, ultimately showed that low-dose oral recombinant BFT-2 actually inhibited the formation of CRC in mice [70]. The proposed mechanism was that of enhancement of the immune system, inhibition of cell proliferation and promotion of apoptosis. A later study published in 2018 revealed polysaccharide A (PSA) from B. fragilis induced cytokines such as IL-8, inhibited CRC proliferation and impaired CRC cell migration via toll-like receptor-2 signaling, thus exhibiting antitumor properties [71]. Further studies to delineate mechanisms of carcinogenesis versus tumor suppression are required to better characterize the likely multifaceted role B. fragilis plays in CRC development and suppression. The genes and mediators of carcinogenesis proposed above, such as bft and long noncoding RNAs, can serve as potential therapeutic targets. At the same time, BFTs and PSA might be considered as possible treatments for CRC in the future and are worthy of further study.

S. gallolyticus

Streptococcus bovis (S. bovis), a Lancefield group D gram-positive organism, has been linked to colon cancer and infective endocarditis. In 2011, a review of 52 case reports and 31 case series showed 60% of S. bovis-infected patients also had concomitant adenomas and carcinomas [72]. Further meta-analysis showed S. bovis biotype I infections with Streptococcus gallolyticus (S. gallolyticus) had a sevenfold higher risk associated with CRC than biotype II-infected patients. S. gallolyticus has a higher affinity for collagen I and IV and can form biofilms [73]. Its predilection for heart valves and colonic mucosa is likely related to the richness in collagen type I and IV in these tissues, respectfully.

In one study, researchers could isolate S. gallolyticus at higher rates in tumor and nontumorous tissue from patients with CRC compared with controls. This bacterium was associated with elevated inflammatory markers, such as IL-1, COX-2 and IL-8, suggesting inflammation-based pathogenesis [74]. Kumar et al. reported that S. gallolyticus promoted the proliferation of tumor cells by targeting the β-catenin pathway and, for the first time, proposed a tumor-promoting role of S. gallolyticus [75]. S. gallolyticus has also been reported as a passenger bacterium and can secrete specific toxins that kill other bacteria and promote its colonization of colonic tumors [76,77].

These proposed mechanisms of action can potentially assist in the discovery of targets for antibody testing for screening, prognostic and therapeutic purposes. For instance, one study suggested that antibodies against S. gallolyticus, particularly against the protein Gallo2178 that forms a complex with other proteins to form a pilus necessary to biofilm formation and virulence, are associated with a 40% increased risk of CRC development in individuals diagnosed with CRC within 10 years of antibody testing [78]. Although there are currently no guidelines, further studies in S. gallolyticus detection and its association with different stages of CRC may contribute to future screening tools for CRC.

Helicobacter pylori

Helicobacter pylori (H. pylori), a gram-negative bacillus, is frequently associated with gastric cancer. Studies have shown that it causes oxidative stress, induces gastritis and promotes inflammation and carcinogenesis in gastric cancer through various mechanisms [79,80]. Some of these mechanisms involve remodeling the microenvironment, making it suitable for colonization. Others involve the expression of proteins such as VacA and CagA, which promote inflammation. Studies have also shown that eradicating H. pylori infections reduces the incidence of precancerous lesions but does not reduce the risk of gastric cancer in patients who already have metaplasia or dysplasia, suggesting an early pathogenic role [81,82].

However, the role of H. pylori in CRC development remains controversial. A nationwide cohort study in Taiwan showed a higher incidence of CRC in H. pylori-infected individuals compared with noninfected individuals (HR: 1.87; 95% CI: 1.37–2.7) [83]. A meta-analysis of 11 studies estimated an odds ratio (OR) of 1.4 (95% CI: 1.1–1.8) for the association between H. pylori infection and CRC risk, although the authors report the possibility of publication bias [84]. Another large meta-analysis of the East Asian population reported no association between H. pylori infections and CRC risk, except in the Japanese population upon subgroup analysis, conflicting with prior studies [85]. The mechanisms of carcinogenesis and exact mechanisms related to gastric cancer and possible CRC development have yet to be elucidated, and further studies are needed to evaluate the role of H. pylori in CRC carcinogenesis.

Microbiome & CRC therapy effects

Recently, studies of non-CRC malignancies suggested that the microbiome can affect treatment responses and patient prognoses. Pancreatic ductal adenocarcinoma (PDAC) mouse model studies suggest that certain bacteria in PDAC metabolize drugs, such as gemcitabine, into inactive metabolites, which promote drug resistance [86]. Other murine studies report treatment with antibiotics to eradicate bacteria in tumors modulates the microenvironment and induces T-cell activation, reversing intratumoral immune tolerance and increasing sensitivity to immunotherapy [86,87]. Results from another preclinical study suggest that PDAC microbiome diversity and microbiome signatures predict long-term survival independent of treatment, and fecal microbial transplants in mouse models can modulate tumor immunosuppression and growth [88].

Similar findings were reported in a prospective melanoma study. Gopalakrishnan et al. proposed that highly diverse and favorable gut microbiomes (containing an abundance of Ruminococcaceae and Faecalibacterium) promote more robust immune responses against melanoma when treated with anti-PD-1 immunotherapy compared with low-diversity microbiomes [89]. Stool transplantation from immunotherapy-responding mice into nonresponding mice caused a robust T-cell immune response in recipient mice. Thus, a favorable gut microbiome may be causal to immune checkpoint response.

We know that MSI-H colon cancers are associated with more robust T-cell-mediated antitumor activity due to higher mutational frequencies and neoantigen presentation. Thus, MSI-H is associated with better clinical outcomes and response to immunotherapy [90,91]. As previously discussed, F. nucleatum-rich CRCs are associated with lower tumor-infiltrating lymphocytes (TILs). Hamada et al. investigated the differences in TIL patterns and MSI status associated with F. nucleatum and found that in MSI-H tumors, F. nucleatum presence was negatively associated with TILs [92]. These results generate hypotheses requiring further study. One such hypothesis is that F. nucleatum promotes immune evasion in MSI-H tumors. Interestingly, in non-MSI-H tumors, F. nucleatum was positively associated with TILs, raising the question as to whether the proinflammatory effects of F. nucleatum on non-MSI-H tumors outweigh any T-cell-suppression effects.

Oh et al. retrospectively studied the prognostic impact of F. nucleatum on stage II and III CRC treated with oxaliplatin-based adjuvant therapy. Study investigators reported that high levels of intratumoral F. nucleatum were a favorable prognostic factor for nonsigmoid colon cancers (HR: 0.42; p = 0.043) but that this prognostic effect was observed only in the non-MSI-H subset of nonsigmoid colon cancers [93]. These studies together suggest that proximal tumor location, MSI-status and F. nucleatum load, among other factors, may be important in determining prognosis and response to treatments. However, further large-population studies are required to make prognostic and predictive conclusions. In another recent study, Yu et al. found that F. nucleatum was associated with a higher risk of CRC recurrence and promoted chemotherapy resistance [51]. It will be interesting to study these variables in prospective studies to better understand their prognostic implications.

Potential therapeutic targeting of the microbiome

There may be a therapeutic role for antibiotics to modulate the CRC microbiome and decrease tumor growth, as previously described in the Bullman study [61]. However, results from a large case-controlled observational study reviewing 23,980 CRC cases and 137,077 controls suggested that prior antibiotic use over the 10 years before the diagnosis of cancer increased the risk of colon cancer and decreased the risk of rectal cancer [94]. Specifically, penicillins increased the risk of colon cancer, and tetracyclines decreased the risk of rectal cancer. These observational data suggest that prior antibiotic use may predispose an individual to CRC. In contrast, preclinical studies indicate there may be a therapeutic role for antibiotics by their modulation of the microenvironment [61]. Further studies aimed at better understanding the relationship between antibiotics and CRC development, including how these therapies modulate the microbiome, are required to understand better the risks and therapeutic potential of different classes of antibiotics.

Fecal microbiota transplantation (FMT) is an effective treatment for C. difficile infections and may be a treatment option in other diseases such as irritable bowel syndrome and IBD. However, more adequately powered trials are required to assess efficacy and safety more fully [95–97]. Although there is no evidence of FMT outcomes and CRC in human clinical trials, mouse models have suggested potential benefits through the reconstitution of a beneficial microbiome, thereby limiting inflammation and neoplastic development [98].

Probiotics are live bacteria that have demonstrated an ability to confer health benefits to a host. The most common bacteria used in probiotics in the food industry are Lactobacilli and Bifidobacterium. Probiotics appear to change the composition of the microbiome and exhibit protective, anti-inflammatory properties such as enhancing epithelial barriers, increasing anticarcinogenic metabolites and reducing inflammation [99]. They also inhibit the growth of other pathogenic bacteria. In murine models, probiotics have been shown to decrease precancerous changes in the colon [100–102]. An in vitro study demonstrated the antitumor effect of Lactobacillus delbrueckii on human tumor CRC cell lines by inducing apoptosis [103]. In a randomized, double-blind study including patients with CRC and adenomatous polyps, probiotic administration resulted in fecal microbiome changes, improved epithelial barrier function, decreased colorectal proliferation, decreased genotoxin levels and decreased levels of inflammatory markers such as IL-2 [104]. Proinflammatory cytokines are also reduced when probiotics are administered after surgery [105]. Other studies suggest that probiotics reduce the incidence of gastrointestinal toxicities and infections when administered with chemotherapy or radiation therapy [99]. Modulating the microbiome through an easy and relatively safe means using probiotics remains an exciting new research territory that requires further randomized prospective study before potential benefits regarding treatment or prevention of CRC can be elucidated.

Based on prior studies that have been detailed earlier in this review, the microbiome milieu impacts the degree of immune surveillance, potential for carcinogenesis, tumor progression and also tumor response to chemotherapy and immunotherapy. Thus, modulation of the microbiome using antibiotics, probiotics or FMT to decrease the risk of tumor development, progression or nonresponsiveness to therapy poses potential research opportunities for future consideration.

Microbiome & EOCRC

Although numerous studies have explored associations between the microbiome and CRC, there is still a paucity of data specifically looking at microbiomes of young individuals with EOCRC. EOCRC and distal CRC are not associated with the typical risk factors of CRC, and their rising incidence might suggest that these cases differ from the general population in their gut microbiome composition. Understanding these differences may elucidate new genetic or environmental risk factors that predispose patients to CRC development and hopefully provide targets for screening and intervention.

The majority of studies on the microbiome in CRC include a mean age of patients around or above 60 years. Many studies, particularly those investigating F. nucleatum in CRC, did not find statistically significant correlations between patient age and the amount of F. nucleatum present in tumor [48,49,58,93]. For example, in the 2015 study by Mima et al., which included CRCs from a patient population that had a mean age of 67 ± 8.4 years, no statistically significant differences in the amount of F. nucleatum with age were reported (p = 0.89) [53]. A subsequent study also by Mima et al., which included 1069 CRC patients, similarly showed no statistically significant difference in age and F. nucleatum-status; the mean patient age was 69.3 ± 8.8 years [62]. In their study of patients with stage II/III CRCs, Oh et al. reported a higher percentage of F. nucleatum-containing CRCs in patients 59 years of age or older [93]. Still, there were no statistically significant differences between F. nucleatum-status and age. However, in the study by Tahara et al., although the median age of the patient population was over 60 years, the investigators noticed a significant trend toward F. nucleatum-high status and patient age of greater than 70 years (p = 0.09) [50]. In a study of S. gallolyticus in CRC by Abdulamir et al., the mean ages of controls, subjects having CRCs without bacteremia and those having CRCs with bacteremia were 57.4 ± 4.7, 59.22 ± 8.13 and 56.6 ± 6.7 years, respectively. Age was not associated with S. gallolyticus detection in CRC (p = 0.41) [74]. Although these studies showed no consistent trends, they were not designed to compare the microbiome in EOCRC versus LOCRC and generally consisted of older patients.

A recent large meta-analysis, studying over 2500 individuals whose ages ranged from 20 to 89 years and who had a range of diseases, including CRC, sought to better understand if differences in microbiomes existed between age groups [106]. The study reported that microbiomes varied across different age groups with ‘young’ comprising subjects aged 20–39 years, ‘middle-age’ comprising subjects aged 40–59 years and ‘elderly’ comprising subjects aged 60 years and over. One of the findings suggested that younger age groups tend to gain more microbial taxa related to diseases (including F. nucleatum), while the elderly tend to lose microbes that are associated with a ‘healthy gut.’ This study and others suggest that the microbiome does vary with age, and EOCRC is likely associated with a distinct microbiome. Further studies are required to understand why the microbiome changes with age and its consequences regarding disease risk and course.

Our team is currently trying to fill this knowledge gap. We have an ongoing study in which we extract archived tumor DNA from patients diagnosed with CRC below the age of 45 and above age 65 years [107]. Intratumoral DNA is analyzed using a 16S ribosomal gene sequence, and preliminary results show a higher incidence of certain bacteria such as Moraxella osloensis and trends toward higher rates of F. nucleatum in younger patients than previously thought. F. nucleatum was detected in 28% of younger patients (5/18) and 23% of older patients (3/13). The study is still recruiting patients with the goal of obtaining larger sample sizes so that we can get a clearer picture of microbiome differences between older and younger patients and whether increased levels of F. nucleatum, for example, can explain the rising incidence of EOCRC patterns.

Conclusion

 The trend towards increased incidences of EOCRC cannot solely be explained by hereditary syndromes, and other contributing factors need to be considered. We believe the microbiome’s impact on CRC risk and carcinogenesis is a growing field of research that deserves more attention. While many studies have elucidated pathogenesis through bacterial biofilms, toxins, and inflammation, further studies are required to elucidate possible markers that could be used for CRC screening or targets for pharmaceutical intervention in the clinical setting. As the microbiome is modulated by factors that are associated with CRC risk, such as (but not limited to) diet, age, and antibiotics, the microbiome’s role in CRC development is likely more important than we previously imagined. Additional studies are especially required to determine the microbiome’s role in EOCRC.

Future perspective

The microbiome plays a significant role in many disease processes, and recent burgeoning CRC studies suggest that the microbiome may play an even more prominent role in carcinogenesis and treatment outcomes than previously imagined. Numerous bacteria are implicated in CRC carcinogenesis and metastasis, and there is an increasing number of treatment-potentiating targets to consider. While we have some evidence that the microbiome varies with age, we also know that it is influenced by a variety of other factors such as medications (metformin, proton pump inhibitors, antibiotics, etc.), diet, stress and diseases such as diabetes and obesity. These considerations perhaps make age itself irrelevant when considering the rise of EOCRC [9,10,108]. However, the microbiome differences in young versus EOCRC remain a topic with sparse data worth further exploration. We believe that future studies such as ours will help elucidate the tumor microbiome and microenvironment in EOCRC and ultimately lead to the development of adaptive screening modalities, a reduction in the risk of cancer development in younger people and an improvement in outcomes for all patients with CRC.

Executive summary

• Over the last decade, there has been a noticeable trend toward increased early-onset colorectal cancer (EOCRC), primarily involving distal and rectal tumors, which cannot be completely attributed to hereditary disorders or inflammatory bowel disease.

• In addition to hereditary causes, clinicians should consider other possible risk factors for EOCRC, such as diet, co-morbidities, prior antibiotic exposures and the gut microbiome.

• Observational and correlative studies suggest that the gut microbiome varies with age, ethnicity, diet, disease, breast feeding, inflammatory gastrointestinal disorders, prior antibiotic use and other factors.

• Colorectal cancers (CRCs) are enriched with Fusobacterium, and preclinical and clinical studies suggest that the bacterium is intrinsic to the tumor microenvironment, contributes to carcinogenesis and confers poorer prognosis. Other bacteria, such as B. fragilis, S. gallolyticus and H. pylori, may also contribute to CRC carcinogenesis.

• With further study, we hope to more thoroughly elucidate microbiome changes due to diet, medications and co-morbidities among other factors, understand the resulting effect on carcinogenesis and ultimately identify high-risk patients that would benefit from earlier screening and counseling accordingly.

• We need more studies to determine if specific bacterial biomarkers or tumor microbiome sequencing can be used for CRC screening, prognostication and prediction of responses to different chemotherapy and immunotherapies.

• While other epidemiologic risk factors have been explored, there is a lack of data on the differences between microbiomes from patients with EOCRC and microbiomes from patients with later-onset CRC. We believe the microbiome may play a role independently or in combination with other factors in the development of EOCRC, and this field warrants further research.

Acknowledgments

The authors acknowledge ML Hartley, science writer for clinical research at The Ruesch Center for the Cure of Gastrointestinal Cancers, Lombardi Comprehensive Cancer Center, Georgetown University (DC, USA) for her edits and suggestions during the composition of this manuscript.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.