Abstract
Objectives:
Data support that enterotoxigenic Bacteroides fragilis (ETBF) harbouring the Bacteroides fragilis toxin (bft) gene may promote colorectal tumourigenesis through the serrated neoplasia pathway. We hypothesised that ETBF may be enriched in colorectal carcinoma subtypes with high-level CpG island methylator phenotype (CIMP-high), BRAF mutation, and high-level microsatellite instability (MSI-high).
Methods:
Quantitative polymerase chain reaction assays were designed to quantify DNA amounts of Bacteroides fragilis, ETBF, and each bft gene isotype (bft-1, bft-2, or bft-3) in colorectal carcinomas in the Health Professionals Follow-up Study and Nurses’ Health Study. We used multivariable-adjusted logistic regression models with the inverse probability weighting method.
Results:
We documented 4,476 colorectal cancer cases, including 1,232 cases with available bacterial data. High DNA amounts of Bacteroides fragilis and ETBF were positively associated with BRAF mutation (P ≤0.0003), CIMP-high (P ≤0.0002), and MSI-high (P <0.0001 and P =0.01, respectively). Multivariable-adjusted odds ratios (ORs, with 95% confidence interval) for high Bacteroides fragilis were 1.40 (1.06-1.85) for CIMP-high and 2.14 (1.65-2.77) for MSI-high, but 1.02 (0.78-1.35) for BRAF mutation. Multivariable-adjusted ORs for high ETBF were 2.00 (1.16-3.45) for CIMP-high and 2.86 (1.64-5.00) for BRAF mutation, but 1.09 (0.67-1.76) for MSI-high. Neither Bacteroides fragilis nor ETBF was associated with colorectal cancer-specific or overall survival.
Conclusions:
The tissue abundance of Bacteroides fragilis is associated with CIMP-high and MSI-high, whereas ETBF abundance is associated with CIMP-high and BRAF mutation in colorectal carcinoma. Our findings support the aetiological relevance of Bacteroides fragilis and ETBF in the serrated neoplasia pathway.
Keywords: DNA methylation, dysbiosis, epigenetics, microbiota, molecular pathological epidemiology, pathogen
Introduction
Colorectal neoplasms accumulate the genomic and epigenomic changes under the influences of the intestinal microbiota.(1–3) Enterotoxigenic Bacteroides fragilis (ETBF) harbouring the Bacteroides fragilis toxin (bft) gene appears to play pathogenic roles in colorectal carcinomas.(4–6) ETBF has been linked to colorectal neoplasms in observational human tissue studies.(4–8) In vivo studies showed that ETBF could promote colorectal tumourigenesis through the upregulation of intracellular signalling and/or oncogene expression.(9–12) ETBF may induce MYC overexpression and promote inflammation that contribute to colorectal carcinogenesis.(13) BRAF p.V600E mutant mice with the mucosal colonisation of ETBF develop tumours with serrated-like morphology, high-level CpG island methylator phenotype (CIMP-high), and cytotoxic T cell infiltrates.(14) The CIMP-high status often results in MLH1 promoter CpG island hypermethylation and high-level microsatellite instability (MSI-high) in colorectal carcinomas.(15) Thus, ETBF is a putative oncomicrobe in colorectal cancer and is associated with the serrated neoplasia pathway. In contrast, the presence of non-enterotoxigenic Bacteroides fragilis correlated with the levels of inflammatory cytokines in the mucosa adjacent to colorectal adenomas.(16) Nevertheless, the contribution of non-enterotoxigenic Bacteroides fragilis to colorectal carcinogenesis, if any, might be relatively weaker as compared to that of ETBF.(17)
In this study, we tested the hypothesis that ETBF might be enriched in tumour tissue of colorectal carcinomas with CIMP-high status, BRAF mutation, and MSI-high status, utilising resources of two U.S.-wide prospective cohort studies with available data on colorectal carcinoma incidence, tumour molecular characteristics, and tissue Bacteroides fragilis and ETBF. To our knowledge, this represents the first study to detect and quantify both Bacteroides fragilis and ETBF in more than 1,200 colorectal carcinomas with well-annotated clinical and tumour molecular profiles.
Methods
Use of standardised official symbols
We use HUGO (Human Genome Organisation) Gene Nomenclature Committee (HGNC)-approved official symbols (or root symbols), accompanied by unique HGNC ID where appropriate, for genes and gene products, including BRAF, CACNA1G, CDKN2A, CRABP1, DNMT1, H2AX, IGF2, KRAS, MAPK, MLH1, MYC, NEUROG1, NFKB, PIK3CA, RUNX3, SMOX, and SOCS1, all of which are described at www.genenames.org. The official gene symbols are italicised to differentiate from non-italicised gene product names and non-official colloquial names.
Study population
Detailed methods are described in Figure 1 and Supplementary methods. In brief, we utilized data from two large U.S.-wide prospective cohort studies with participants provided biennial questionnaire responses on lifestyle factors and illnesses, including colorectal carcinoma. Colorectal carcinoma cases were confirmed by physicians through medical record reviews. We collected carcinoma tissue blocks from colorectal cancer patients across the U.S. Informed consent was obtained from all study patients, and the study protocol was approved by institutional review boards. The study adhered to the Declaration of Helsinki.
Figure 1.
Flow diagram of the study population in the Health Professionals Follow-up Study and Nurses’ Health Study.
Tumour tissue analyses
Detailed tissue analyses are described in the Supplementary methods, Supplementary table S1, and Supplementary figures S1–2. Shortly, DNA was extracted from colorectal cancer tissue obtained from sections of formalin-fixed paraffin-embedded (FFPE) blocks. The amount of tissue Bacteroides fragilis and ETBF DNA were measured by a quantitative polymerase chain reaction (PCR) assay and normalized with the reference gene SLCO2A1 as described in Supplementary methods. MSI status, CIMP status, and mutations in KRAS, BRAF, and PIK3CA were determined using established protocols as described in Supplementary methods.
Statistical analysis
Details of statistical analyses were described in Supplementary methods. Briefly, we performed the chi-square or Fisher’s exact tests depending on the patient count in each cell to compare binary variables among ETBF DNA categories. To compare ordinary variables, the Spearman correlation was performed. To assess and adjust for confounding, multivariable-adjusted logistic regression analyses were executed. Survival analyses utilized the Kaplan-Meier method and Cox proportional hazards regression model. The inverse probability weighting (IPW) method was adjusted for selection bias due to tissue data availability.
Results
In 1,232 colorectal carcinoma cases (among the 4,476 incident colorectal cancer cases within the two prospective cohort studies), we conducted quantitative PCR to measure DNA amounts of Bacteroides fragilis, ETBF, and ETBF subtypes (bft-1, bft-2, and bft-3) in tumour tissue. Bacteroides fragilis levels were negative, low, and high in 624 (51%), 304 (25%), and 304 (25%) tumours, respectively. ETBF levels were negative, low, and high in 1,126 (91%), 53 (4.3%), and 53 (4.3%) tumours, respectively.
Table 1 shows clinical, pathological, and molecular characteristics of colorectal cancer cases according to Bacteroides fragilis or ETBF levels. High-level Bacteroides fragilis and ETBF were associated with proximal tumour location (though not statistically significantly at the stringent alpha level of 0.005). Notably, high-level Bacteroides fragilis (referred to as Bacteroides fragilis-high) was significantly associated with CIMP-high (P <0.0001), MSI-high (P <0.0001), and BRAF mutation (P <0.0001), while high-level ETBF (ETBF-high) was associated with CIMP-high (P =0.0002), MSI-high (P =0.01), and BRAF mutation (P =0.0003). Supplementary table S4 shows the combined status of Bacteroides fragilis and ETBF in relation to the clinicopathological features. Bacteroides fragilis-positive ETBF-high tumours showed higher prevalences of MSI-high (32%, 16 of 50 tumours, P =0.0034), CIMP-high (46%, 23 of 50 tumours, P =0.0005), and BRAF mutation (38%, 19 of 50 tumours, P =0.0011), compared to Bacteroides fragilis-negative tumours (14% MSI-high, 21% CIMP-high, 15% BRAF-mutated), Bacteroides fragilis-positive ETBF-negative tumours (19%, 23%, 16%, respectively), and Bacteroides fragilis-positive ETBF-low tumours (10%, 12%, 10%, respectively). Supplementary tables S5–7 show levels of ETBF subtypes (bft-1, bft-2, and bft-3, respectively) in relation to the clinicopathological features. ETBF type 1-high tumours showed a higher prevalence of CIMP-high (46%, 13 of 28 tumours, P =0.014), compared to ETBF type 1-negative tumours (22% CIMP-high). ETBF type 3-high tumours showed higher prevalences of CIMP-high (71%, 5 of 7 tumours, P =0.0066) and BRAF mutation (71%, 5 of 7 tumours, P =0.0037), compared to ETBF type 3-negative tumours (22% CIMP-high and 16% BRAF mutation). In general, case counts of high levels of each ETBF subtype were small, providing limited statistical power.
Table 1.
Clinical, pathological, and molecular characteristics of colorectal cancer cases according to the amount of Bacteroides fragilis and enterotoxigenic Bacteroides fragilis.
| Characteristic* | All cases (N=1,232) | Amount of Bacteroides fragilis in colorectal cancer tissue |
P† | Amount of enterotoxigenic Bacteroides fragilis in colorectal cancer tissue |
|||||
|---|---|---|---|---|---|---|---|---|---|
| Negative (N=624) | Low (N=304) | High (N=304) | Negative (N=1,126) | Low (N=53) | High (N=53) | P† | |||
| Sex | 0.32 | 0.077 | |||||||
| Male (HPFS) | 539 (44%) | 268 (43%) | 144 (47%) | 127 (42%) | 496 (44%) | 27 (51%) | 16 (30%) | ||
| Female (NHS) | 693 (56%) | 356 (57%) | 160 (53%) | 177 (58%) | 630 (56%) | 26 (49%) | 37 (70%) | ||
| Mean age ± SD (years) | 69.7±9.0 | 69.3±9.1 | 70.4±8.6 | 69.9±9.1 | 0.23 | 69.7±9.0 | 69.2±9.0 | 69.5±9.0 | 0.89 |
| Family history of colorectal cancer in a first-degree relative | 0.91 | 0.16 | |||||||
| Absent | 982 (80%) | 496 (80%) | 241 (80%) | 245 (81%) | 902 (80%) | 43 (81%) | 37 (70%) | ||
| Present | 245 (20%) | 127 (20%) | 60 (20%) | 58 (19%) | 219 (20%) | 10 (19%) | 16 (30%) | ||
| Year of diagnosis | 0.87 | 0.31 | |||||||
| Prior to 1995 | 384 (31%) | 188 (30%) | 101 (33%) | 95 (31%) | 357 (32%) | 16 (30%) | 11 (21%) | ||
| 1996–2000 | 353 (29%) | 191 (31%) | 81 (27%) | 81 (27%) | 319 (28%) | 16 (30%) | 18 (34%) | ||
| 2001–2012 | 495 (40%) | 245 (39%) | 122 (40%) | 128 (42%) | 450 (40%) | 21 (40%) | 24 (45%) | ||
| Tumour location | 0.0063 | 0.021 | |||||||
| Proximal colon | 603 (49%) | 292 (47%) | 132 (44%) | 179 (59%) | 542 (48%) | 22 (42%) | 39 (74%) | ||
| Distal colon | 362 (29%) | 186 (30%) | 103 (34%) | 73 (24%) | 331 (30%) | 22 (42%) | 9 (17%) | ||
| Rectum | 263 (21%) | 144 (23%) | 68 (22%) | 51 (17%) | 249 (22%) | 9 (17%) | 5 (9%) | ||
| Tumour differentiation | 0.0061 | 0.016 | |||||||
| Well to moderate | 1,107 (90%) | 557 (90%) | 287 (94%) | 263 (87%) | 1,005 (89%) | 53 (100%) | 49 (92%) | ||
| Poor | 122 (10%) | 65 (10%) | 17 (6%) | 40 (13%) | 118 (11%) | 0 (0%) | 4 (8%) | ||
| AJCC disease stage | 0.77 | 0.60 | |||||||
| I | 276 (24%) | 144 (25%) | 75 (27%) | 57 (20%) | 253 (24%) | 15 (35%) | 8 (16%) | ||
| II | 365 (32%) | 175 (30%) | 76 (28%) | 114 (41%) | 331 (32%) | 10 (23%) | 24 (48%) | ||
| III | 333 (29%) | 177 (31%) | 79 (29%) | 77 (28%) | 310 (30%) | 11 (26%) | 12 (24%) | ||
| IV | 158 (14%) | 84 (14%) | 43 (16%) | 31 (11%) | 145 (14%) | 7 (16%) | 6 (12%) | ||
| CIMP status | <0.0001 | 0.0002 | |||||||
| Low/negative | 856 (78%) | 440 (79%) | 228 (85%) | 188 (67%) | 786 (78%) | 43 (88%) | 27 (54%) | ||
| High | 247 (22%) | 117 (21%) | 39 (15%) | 91 (33%) | 218 (22%) | 6 (12%) | 23 (46%) | ||
| LINE-1 methylation level | 63.5±10.3 | 63.8±10.4 | 62.7±9.5 | 63.6±10.7 | 0.38 | 63.4±10.4 | 63.8±7.1 | 64.1±9.7 | 0.87 |
| MSI status | <0.0001 | 0.010 | |||||||
| Non-MSI-high | 951 (83%) | 494 (86%) | 250 (89%) | 207 (72%) | 871 (84%) | 46 (90%) | 34 (68%) | ||
| MSI-high | 192 (17%) | 82 (14%) | 31 (11%) | 79 (28%) | 171 (16%) | 5 (10%) | 16 (32%) | ||
| KRAS mutation | 0.32 | 0.90 | |||||||
| Wildtype | 625 (58%) | 322 (58%) | 144 (54%) | 159 (60%) | 566 (58%) | 31 (61%) | 28 (58%) | ||
| Mutant | 458 (42%) | 229 (42%) | 123 (46%) | 106 (40%) | 418 (42%) | 20 (39%) | 20 (42%) | ||
| BRAF mutation | <0.0001 | 0.0003 | |||||||
| Wildtype | 961 (84%) | 488 (85%) | 253 (90%) | 220 (76%) | 884 (84%) | 46 (90%) | 31 (62%) | ||
| Mutant | 188 (16%) | 89 (15%) | 29 (10%) | 70 (24%) | 164 (16%) | 5 (10%) | 19 (38%) | ||
| PIK3CA mutation | 0.093 | 0.51 | |||||||
| Wildtype | 905 (84%) | 468 (85%) | 218 (85%) | 219 (80%) | 826 (83%) | 39 (91%) | 40 (85%) | ||
| Mutant | 175 (16%) | 80 (15%) | 39 (15%) | 56 (20%) | 164 (17%) | 4 (9%) | 7 (15%) | ||
Data are presented as number (%) or as the mean ± SD. Percentage indicates the proportion of cases with a specific clinical, pathologic, or molecular characteristic among all colorectal cancer cases or in strata of Bacteroides fragilis and enterotoxigenic Bacteroides fragilis DNA levels (negative, low, and high).
To compare categorical data by Bacteroides fragilis and enterotoxigenic Bacteroides fragilis levels, the chi-square test or Fisher’s exact test was performed. To compare age and LINE-1 methylation level by those levels, the analysis of variance was performed. To compare year of diagnosis by those levels, AJCC disease stage, and tumour location, the Spearman correlation test was performed.
Abbreviations: AJCC, American Joint Committee on Cancer; CIMP, CpG island methylator phenotype; HPFS, Health Professionals Follow-up Study; MSI, microsatellite instability; NHS, Nurses’ Health Study; SD, standard deviation.
We performed multivariable analyses to control for confounding. We used IPW method to reduce selection bias due to the availability of DNA data on Bacteroides fragilis and ETBF. The multivariable-adjusted odds ratios (ORs, with 95% confidence interval) for Bacteroides fragilis-high were 1.40 (1.06-1.85, P =0.018) for CIMP-high and 2.14 (1.65-2.77, P <0.0001) for MSI-high, whereas Bacteroides fragilis was no longer associated with BRAF mutation (Table 2). The multivariable-adjusted ORs for ETBF-high were 2.00 (1.16-3.45, P =0.012) for CIMP-high and 2.86 (1.64-5.00, P =0.0002) for BRAF mutation, whereas ETBF was no longer significantly associated with MSI-high (Table 3).
Table 2.
Logistic regression analysis to assess the association of BRAF, CIMP, and/or MSI status (a predictor variable) with Bacteroides fragilis in colorectal cancer tissue (a binary outcome variable, negative/low vs. high).
| Predictor variable | Univariable OR (95% CI)* | P | Multivariable OR (95% CI)*† | P |
|---|---|---|---|---|
| BRAF mutant (vs. wildtype) | 1.80 (1.48–2.20) | <0.0001 | 1.02 (0.78–1.35) | 0.87 |
| CIMP-high (vs. CIMP-low/negative) | 2.09 (1.74–2.51) | <0.0001 | 1.40 (1.06–1.85) | 0.018 |
| MSI-high (vs. non-MSI-high) | 2.62 (2.15–3.20) | <0.0001 | 2.14 (1.65–2.77) | <0.0001 |
Inverse probability weighting method was applied to adjust for selection bias due to the availability of data on Bacteroides fragilis.
The multivariable logistic regression model initially included sex, age, year of diagnosis, family history of colorectal cancer in a first-degree relative(s), tumour location, MSI, CIMP, long-interspersed nucleotide element-1 methylation level, BRAF, KRAS, PIK3CA. A backward elimination with a threshold P of 0.05 was used to select variables (besides the statuses of BRAF, MSI, and CIMP) in the final model.
Abbreviations: CI, confidence interval; CIMP, CpG island methylator phenotype; MSI, microsatellite instability; OR, odds ratio.
Table 3.
Logistic regression analysis to assess the association of BRAF, CIMP, and/or MSI status (a predictor variable) with enterotoxigenic Bacteroides fragilis in colorectal cancer tissue (a binary outcome variable, negative/low vs. high).
| Predictor variable | Univariable OR (95% CI)* | P | Multivariable OR (95% CI)*† | P |
|---|---|---|---|---|
| BRAF mutant (vs. wildtype) | 3.97 (2.78–5.65) | <0.0001 | 2.86 (1.64–5.00) | 0.0002 |
| CIMP-high (vs. CIMP-low/negative) | 3.59 (2.55–5.07) | <0.0001 | 2.00 (1.16–3.45) | 0.012 |
| MSI-high (vs. non-MSI-high) | 2.70 (1.86–3.93) | <0.0001 | 1.09 (0.67–1.76) | 0.73 |
Inverse probability weighting method was applied to adjust for selection bias due to the availability of data on enterotoxigenic Bacteroides fragilis.
The multivariable logistic regression model initially included sex, age, year of diagnosis, family history of colorectal cancer in a first-degree relative(s), tumour location, MSI, CIMP, long-interspersed nucleotide element-1 methylation level, BRAF, KRAS, PIK3CA. A backward elimination with a threshold P of 0.05 was used to select variables (besides the statuses of BRAF, MSI, and CIMP) in the final model.
Abbreviations: CI, confidence interval; CIMP, CpG island methylator phenotype; MSI, microsatellite instability; OR, odds ratio.
We assessed the prognostic associations of Bacteroides fragilis and ETBF. We performed Kaplan-Meier analyses (Figure 2) and multivariable-adjusted Cox proportional hazards regression analyses (Supplementary tables S8–9). Neither Bacteroides fragilis nor ETBF was associated with survival of colorectal carcinoma patients.
Figure 2.
Kaplan-Meier curves of colorectal cancer-specific (A-B) and overall survival (C-D) of patients according to the DNA amount (negative vs. low-level vs. high-level) of Bacteroides fragilis (A, C) or enterotoxigenic Bacteroides fragilis (B, D). The inverse probability weighting (IPW) method was used to adjust for tissue bacterial data availability.
Discussion
We investigated the amounts of tissue Bacteroides fragilis DNA and ETBF DNA in relation to key colorectal tumour molecular features associated with the serrated neoplasia pathway, namely the CIMP-high, BRAF mutation, and MSI-high statuses which are known as the molecular pathological features of the consensus molecular subtype 1 (CMS1).(18) Fusobacterium nucleatum (19,20) and pks+ Escherichia coli (21) are among the most extensively studied bacteria implicated in colorectal carcinogenesis. There is limited understanding of the potential pathogenic effect of Bacteroides fragilis and enterotoxigenic Bacteroides fragilis (ETBF). Therefore, this study focused on Bacteroides fragilis and ETBF. Leveraging the well-annotated colorectal cancer cases in our molecular pathological epidemiology (22) database, we found the relationship of Bacteroides fragilis level with CIMP-high and MSI-high as well as that of ETBF level with CIMP-high and BRAF mutation.
Colorectal carcinomas represent a group of molecularly heterogeneous tumours that develop under the effect of the intestinal microbiota and host immune system.(22) It is a considerable challenge to determine pathogenic effect of certain microbial species detected in stool specimens, which harbour an enormous number of microorganisms. Targeted analyses of specific tissue microbes such as Bacteroides fragilis and ETBF have advantages with their presence within the tumour microenvironment containing the relatively low amount of other commensal microbial species. Previous studies showed that ETBF (DNA or bacteria isolate) was present in human colorectal carcinoma tissues (4,6,7) and premalignant neoplasms such as tubular adenoma and serrated lesions (5) and that the amount of ETBF was generally higher in neoplastic tissue than in normal colorectal mucosa and higher in proximal colorectal tumours than in distal tumours.(6) Nevertheless, all of these prior studies (4,6–8,23,24) had small sample sizes (all n ≤96) and hence robust statistical analyses of the relationships of ETBF with various clinical, prognostic, pathological, and molecular features were not possible in contrast to our current study.
The pathogenic roles of ETBF have been topics of active investigations. Studies showed that ETBF might promote tumour growth through MYC upregulation and the activation of the NFKB and MAPK signalling pathways.(13,25) Furthermore, ETBF might upregulate reactive oxygen species (ROS) through SMOX (HGNC: 15862; spermine oxidase), resulting in the phosphorylation of H2AX (HGNC: 4739; H2A.X variant histone) (the phosphorylated form being termed “gamma-H2AX”) in DNA break sites.(26) If ROS-induced DNA damage remains inadequately repaired by gamma-H2AX or if the genome stability moderated by ROS-induced DNMT1 (HGNC: 2976;DNA methyltransferase 1) is compromised,(27) genetic mutations might accumulate and potentially contribute to tumourigenesis. Elevated levels of ROS and inflammatory tissue condition play an important role in increased cancer susceptibility.(28) Previous studies showed that ETBF-induced inflammation leads to hypermethylation in promoter CpG islands of multiple tumour suppressor genes in the MinApcΔ716+/− mouse colorectal tumour model,(29) and that BRAF mutation might synergize with ETBF to accelerate promoter CpG island hypermethylation, resulting in tumours with serrated-like histology.(14) Our findings support the hypothesis that ETBF might be one of the pivotal bacterial factors for colorectal tumourigenesis related to the serrated pathway leading to CIMP-high colorectal carcinomas. Taken together, evidence suggests that ETBF may contribute to colorectal tumourigenesis through epigenetic aberrations and serrated pathological changes.
We found that non-toxigenic Bacteroides fragilis abundance was associated with the tumour MSI-high status, consistent with a previous report.(30) We further showed that this relationship persisted after adjusting for the CIMP and BRAF mutation statuses and was not driven by ETBF. The MSI-high status is primarily caused by MLH1 promoter hypermethylation (most often in the setting of CIMP-high tumours) or by germline and/or somatic alterations of one of mismatch repair genes.(15) A previous study (31) has shown that Bacteroides fragilis (which is most commonly non-toxigenic) might exhibit the immunoinhibitory effect via its lipopolysaccharide binding in colorectal carcinoma classified as CMS1, which is associated with CIMP-high, MSI-high, and BRAF mutation. Our findings suggest that Bacteroides fragilis was enriched in colorectal tumours with CIMP-high and MSI-high, the latter of which is characterised by brisk immune responses. Taken together, it is conceivable that Bacteroides fragilis (even without bft toxigenic property) might related to colorectal carcinogenesis through its immunosuppressive effect. Further research is needed to clarify the biological roles of non-toxigenic Bacteroides fragilis in colorectal carcinomas.
The prognostic role of Bacteroides fragilis and ETBF in colorectal carcinoma tissue has been studied. Prior studies showed that higher DNA levels of Bacteroides fragilis in colorectal carcinomas (n =180) were associated worse patient survival (32) and that higher DNA levels of ETBF in colorectal carcinomas (n =96) were correlated with worse overall survival,(23) in contrast to the findings of the current study using over 1,200 patients. It is not uncommon to observe variable results in different studies for a number of reasons, including study populations, sample sizes, confounders, and analysis methods. Further research is needed for accurate estimations of prognostic roles of both Bacteroides fragilis and ETBF.
Our study has limitations. First, we used FFPE tissue as fresh frozen tissue was unavailable. FFPE tissue has been known to contain lower quality DNA compared to fresh frozen tissue. In addition, RNA specimens were not available in our study. Hence, we measured the presence (amount) but not functional statuses of Bacteroides fragilis and ETBF. Nonetheless, FFPE tissue has been widely available in cancer epidemiology studies, which enabled us to conduct large-scale research on Bacteroides fragilis and ETBF. Second, measurement errors exist in any laboratory analyses including our microbial assays. Nonetheless, we have described our assay validation procedures and shown high precision and wide dynamic ranges of our bacterial assays. We also demonstrated high precision of our MethyLight assay (33) and high sensitivity and specificity of each CIMP marker for the detection of CIMP-high status (34). A small fraction of ETBF-positive patients with detectable bft were not classifiable for bft subtypes, possibly due to mutations in primer binding sites in bft subtyping assays. Third, in our analyses of the associations between ETBF subtypes and clinicopathological features, a small number of ETBF subtype-positive cases resulted in limited statistical power. Further studies are needed to clarify the associations between ETBF subtypes and clinicopathological features. Fourth, tumour tissue was not available in all incident colorectal cancer cases that had occurred in the cohorts. Tumour tissue availability depends on certain clinicopathological factors and was not random. Therefore, we implemented the IPW method to address this selection bias. We did not see any substantial difference in results with and without IPW. Fifth, a vast majority of our cohort participants were non-Hispanic Whites. A study (35) suggested a racial difference of colorectal tumour-resident microbes (Bacteroides and Fusobacterium species) between Blacks and Whites. Therefore, our hypothesis needs to be tested in different populations.
The present study has notable strengths. First, our molecular pathological epidemiology (22) database that integrated clinicopathological characteristics and tumour molecular/microbial data in a large number of tumours allowed us to examine the relationship of Bacteroides fragilis and ETBF with tumour molecular features while adjusting for potential confounders and selection bias. Second, quality control and quality assurance efforts were made at each step of specimen collection and processing to ensure the scientific rigour of our study. Third, we developed and validated robust quantitative PCR assays for Bacteroides fragilis and ETBF. All of the other tissue assays had also been extensively validated. Fourth, our patient subjects were derived from the well-known background population of the cohort studies and drawn from numerous hospitals around the U.S., which increases the generalisability of our findings.
In conclusion, our findings support the aetiological relevance of Bacteroides fragilis and ETBF in the serrated colorectal carcinogenesis pathway and will inform further investigations of Bacteroides fragilis and ETBF in colorectal carcinoma.
Supplementary Material
Acknowledgments
The authors would like to acknowledge the contribution to this study from central cancer registries supported through the Centers for Disease Control and Prevention’s National Program of Cancer Registries (NPCR) and/or the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program. Central registries may also be supported by state agencies, universities, and cancer centers. Participating central cancer registries include the following: Alabama, Alaska, Arizona, Arkansas, California, Delaware, Colorado, Connecticut, Florida, Georgia, Hawaii, Idaho, Indiana, Iowa, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Mississippi, Montana, Nebraska, Nevada, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Puerto Rico, Rhode Island, Seattle SEER Registry, South Carolina, Tennessee, Texas, Utah, Virginia, West Virginia, Wyoming. We would like to acknowledge Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts, U.S. The authors assume full responsibility for analyses and interpretation of these data.
Funding
This work was supported by U.S. National Institutes of Health (NIH) grants (P01 CA87969 to Meir J Stampfer; UM1 CA186107 to Meir J Stampfer; P01 CA55075 to Walter C Willett; UM1 CA167552 to Walter C Willett; U01 CA167552 to Lorelei A Mucci and Walter C Willett; R35 CA197735 to S.O.; R01 CA151993 to S.O.; R21 CA230873 to S.O.; R50CA274122 to T.U.); by Cancer Research UK Grand Challenge Award (C10674/A27140, to M.G., W.S.G., C.L.S., and S.O.); by an ASCO Conquer Cancer Foundation Career Development Award (to M.G.); by a grant from Prevent Cancer Foundation and Harvard T.H. Chan School of Public Health (to T.U.); and by the Stand Up to Cancer Colorectal Cancer Dream Team Translational Research Grant (SU2C-AACR-DT22-17 to M.G.), administered by the American Association for Cancer Research, a scientific partner of SU2C. The content is solely the responsibility of the authors and does not necessarily represent the official views of NIH. H.K. and S.U. were supported by fellowship grants from the Uehara Memorial Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Conflict of interest
C.L.S. has a grant from Janssen on enterotoxigenic Bacteroides fragilis. M.G. has the research funding of Janssen, Honorarium and AstraZeneca, and is advisory board of Chroma code. This study was not funded by any of these commercial entities.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Advani SM, Swartz MD, Loree J, Davis JS, Sarsashek AM, Lam M, et al. Epidemiology and Molecular-Pathologic Characteristics of CpG Island Methylator Phenotype (CIMP) in Colorectal Cancer. Clin Colorectal Cancer. 2021;20(2):137–47.e1. 10.1016/j.clcc.2020.09.007. [DOI] [PubMed] [Google Scholar]
- 2.Kumar A, Ali A, Kapardar RK, Dar GM, Nimisha Apurva, et al. Implication of gut microbes and its metabolites in colorectal cancer. J Cancer Res Clin Oncol.2023;149(1):441–65. 10.1007/s00432-022-04422-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lin Y, Kong DX, Zhang YN. Does the Microbiota Composition Influence the Efficacy of Colorectal Cancer Immunotherapy? Front Oncol. 2022;12:852194. 10.3389/fonc.2022.852194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Viljoen KS, Dakshinamurthy A, Goldberg P, Blackburn JM. Quantitative profiling of colorectal cancer-associated bacteria reveals associations between Fusobacterium spp., enterotoxigenic Bacteroides fragilis (ETBF) and clinicopathological features of colorectal cancer. PLoS One. 2015;10(3):e0119462. 10.1371/journalpone.0119462. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Purcell RV, Pearson J, Aitchison A, Dixon L, Frizelle FA, Keenan JI. Colonization with enterotoxigenic Bacteroides fragilis is associated with early-stage colorectal neoplasia. PLoS One. 2017;12(2):e0171602. 10.1371/journalpone.0171602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Boleij A, Hechenbleikner EM, Goodwin AC, Badani R, Stein EM, Lazarev MG, et al. The Bacteroides fragilis toxin gene is prevalent in the colon mucosa of colorectal cancer patients. Clin Infect Dis. 2015;60(2):208–15. 10.1093/cid/ciu787. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Zhou Y, He H, Xu H, Li Y, Li Z, Du Y, et al. Association of oncogenic bacteria with colorectal cancer in South China. Oncotarget. 2016;7(49):80794–802. 10.18632/oncotarget.13094. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Zamani S, Taslimi R, Sarabi A, Jasemi S, Sechi LA, Feizabadi MM. Enterotoxigenic Bacteroides fragilis: A Possible Etiological Candidate for Bacterially-Induced Colorectal Precancerous and Cancerous Lesions. Front Cell Infect Microbiol. 2019;9:449. 10.3389/fcimb.2019.00449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Wu S, Rhee KJ, Albesiano E, Rabizadeh S, Wu X, Yen HR, et al. A human colonic commensal promotes colon tumorigenesis via activation of T helper type 17 T cell responses. Nat Med. 2009;15(9):1016–22. 10.1038/nm.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Liu QQ, Li CM, Fu LN, Wang HL, Tan J, Wang YQ, et al. Enterotoxigenic Bacteroides fragilis induces the stemness in colorectal cancer via upregulating histone demethylase JMJD2B. Gut Microbes. 2020;12(1):1788900. 10.1080/19490976.2020.1788900. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Hwang S, Lee CG, Jo M, Park CO, Gwon SY, Yi HC, et al. Enterotoxigenic Bacteroides fragilis infection exacerbates tumorigenesis in AOM/DSS mouse model. Int J Med Sci. 2020;17(2):145–52. 10.7150/ijms.38371. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chung L, Thiele Orberg E, Geis AL, Chan JL, Fu K, DeStefano Shields CE, et al. Bacteroides fragilis Toxin Coordinates a Pro-carcinogenic Inflammatory Cascade via Targeting of Colonic Epithelial Cells. Cell Host Microbe. 2018;23(2):203–14.e5. 10.1016/j.chom.2018.01.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rhee KJ, Wu S, Wu X, Huso DL, Karim B, Franco AA, et al. Induction of persistent colitis by a human commensal, enterotoxigenic Bacteroides fragilis, in wild-type C57BL/6 mice. Infect Immun. 2009;77(4):1708–18. 10.1128/IAI.00814-08. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.DeStefano Shields CE, White JR, Chung L, Wenzel A, Hicks JL, Tam AJ, et al. Bacterial-Driven Inflammation and Mutant BRAF Expression Combine to Promote Murine Colon Tumorigenesis That Is Sensitive to Immune Checkpoint Therapy. Cancer Discov. 2021;11(7):1792–807. 10.1158/2159-8290.CD-20-0770. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hinoue T, Weisenberger DJ, Lange CP, Shen H, Byun HM, Van Den Berg D, et al. Genome-scale analysis of aberrant DNA methylation in colorectal cancer. Genome Res. 2012;22(2):271–82. 10.1101/gr.117523.110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kordahi MC, Stanaway IB, Avril M, Chac D, Blanc MP, Ross B, et al. Genomic and functional characterization of a mucosal symbiont involved in early-stage colorectal cancer. Cell Host Microbe. 2021;29(10):1589–98.e6. 10.1016/j.chom.2021.08.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Chan JL, Wu S, Geis AL, Chan GV, Gomes TAM, Beck SE, et al. Non-toxigenic Bacteroides fragilis (NTBF) administration reduces bacteria-driven chronic colitis and tumor development independent of polysaccharide A. Mucosal Immunol. 2019;12(1):164–77. 10.1038/s41385-018-0085-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Guinney J, Dienstmann R, Wang X, de Reyniès A, Schlicker A, Soneson C, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. 2015;21(11):1350–6. 10.1038/nm.3967. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mima K, Sukawa Y, Nishihara R, Qian ZR, Yamauchi M, Inamura K, et al. Fusobacterium nucleatum and T Cells in Colorectal Carcinoma. JAMA Oncol. 2015;1(5):653–61. 10.1001/jamaoncol.2015.1377. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Mima K, Nishihara R, Qian ZR, Cao Y, Sukawa Y, Nowak JA, et al. Fusobacterium nucleatum in colorectal carcinoma tissue and patient prognosis. Gut. 2016;65(12):1973–80. 10.1136/gutjnl-2015-310101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Arima K, Zhong R, Ugai T, Zhao M, Haruki K, Akimoto N, et al. Western-Style Diet, pks Island-Carrying Escherichia coli, and Colorectal Cancer: Analyses From Two Large Prospective Cohort Studies. Gastroenterology. 2022;163(4):862–74. 10.1053/j.gastro.2022.06.054. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ogino S, Nowak JA, Hamada T, Milner DA Jr., Nishihara R. Insights into Pathogenic Interactions Among Environment, Host, and Tumor at the Crossroads of Molecular Pathology and Epidemiology. Annu Rev Pathol. 2019;14:83–103. 10.1146/annurev-pathmechdis-012418-012818. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Bao Y, Tang J, Qian Y, Sun T, Chen H, Chen Z, et al. Long noncoding RNA BFAL1 mediates enterotoxigenic Bacteroides fragilis-related carcinogenesis in colorectal cancer via the RHEB/mTOR pathway. Cell Death Dis. 2019;10(9):675. 10.1038/S41419-019-1925-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Purcell RV, Permain J, Keenan JI. Enterotoxigenic Bacteroides fragilis activates IL-8 expression through Stat3 in colorectal cancer cells. Gut Pathog. 2022;14(1):16. 10.1186/s13099-022-00489-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Wu S, Powell J, Mathioudakis N, Kane S, Fernandez E, Sears CL. Bacteroides fragilis enterotoxin induces intestinal epithelial cell secretion of interleukin-8 through mitogen-activated protein kinases and a tyrosine kinase-regulated nuclear factor-kappaB pathway. Infect Immun. 2004;72(10):5832–9. 10.1128/IAI72.10.5832-5839.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Goodwin AC, Destefano Shields CE, Wu S, Huso DL, Wu X, Murray-Stewart TR, et al. Polyamine catabolism contributes to enterotoxigenic Bacteroides fragilis-induced colon tumorigenesis. Proc Natl Acad Sci U S A. 2011;108(37):15354–9. 10.1073/pnas.1010203108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.O’Hagan HM, Wang W, Sen S, Destefano Shields C, Lee SS, Zhang YW, et al. Oxidative damage targets complexes containing DNA methyltransferases, SIRT1, and polycomb members to promoter CpG Islands. Cancer Cell. 2011;20(5):606–19. 10.1016/j.ccr.2011.09.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Federico A, Morgillo F, Tuccillo C, Ciardiello F, Loguercio C. Chronic inflammation and oxidative stress in human carcinogenesis. Int J Cancer. 2007;121(11):2381–6. 10.1002/ijc.23192. [DOI] [PubMed] [Google Scholar]
- 29.Maiuri AR, Peng M, Podicheti R, Sriramkumar S, Kamplain CM, Rusch DB, et al. Mismatch Repair Proteins Initiate Epigenetic Alterations during Inflammation-Driven Tumorigenesis. Cancer Res. 2017;77(13):3467–78. 10.1158/0008-5472CAN-17-0056. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hale VL, Jeraldo P, Chen J, Mundy M, Yao J, Priya S, et al. Distinct microbes, metabolites, and ecologies define the microbiome in deficient and proficient mismatch repair colorectal cancers. Genome Med. 2018;10(1):78. 10.1186/s13073-018-0586-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Sulit AK, Daigneault M, Allen-Vercoe E, Silander OK, Hock B, McKenzie J, et al. Bacterial lipopolysaccharide modulates immune response in the colorectal tumor microenvironment. NPJ Biofilms Microbiomes. 2023;9(1):59. 10.1038/S41522-023-00429-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Wei Z, Cao S, Liu S, Yao Z, Sun T, Li Y, et al. Could gut microbiota serve as prognostic biomarker associated with colorectal cancer patients’ survival? A pilot study on relevant mechanism. Oncotarget. 2016;7(29):46158–72. 10.18632/oncotarget10064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Ogino S, Kawasaki T, Brahmandam M, Cantor M, Kirkner GJ, et al. Precison and performance characteristics of bisulfite conversion and real-time PCR (MethyLight) for quantitative DNA methylation analysis. J Mol Diagn. 2006;8(2):209–17. 10.2353/jmoldx.2006.050135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Ogino S, Kawasaki T, Kirkner GJ, Kraft P, Loda M, Fuchs CS. Evaluation of markers for CpG island methylator phenotype (CIMP) in colorectal cancer by a large population-based sample. J Mol Diagn. 2007;9(3):305–14. 10.2353/jmoldx.2007.060170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Farhana L, Antaki F, Murshed F, Mahmud H, Judd SL, Nangia-Makker P, et al. Gut microbiome profiling and colorectal cancer in African Americans and Caucasian Americans. World J Gastrointest Pathophysiol. 2018;9(2):47–58. 10.4291/wjgp.v9.i2.47. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.