Introduction

Colorectal cancer (CRC) is the second leading cause of cancer-related death and the third most commonly diagnosed cancer in the world, accounting for 1.8 million new cases in 2018 [1]. Accumulating evidence suggests that genetic susceptibility, environmental exposure, metabolic dysfunction, microbiome composition, and breakdown of gut barrier integrity play major roles in CRC etiology [2, 3]. Furthermore, chronic inflammation caused by infections may promote CRC development [4]. Experimental data suggest that microbes damage the gut barrier lining and colonocyte DNA and increase proinflammatory cytokines and oxidative factors [5, 6, 7]. F. nucleatum may contribute to CRC development by the invasion of the colonic mucosa, the recruitment of immune cells, and the generation of an oncogenic microenvironment that facilitates the development of colorectal neoplasia [8]. Increased permeability of the gut barrier leads to extensive translocation of microbes into the lamina propria, such as SGG, which may then contribute to CRC development [9]. Antibodies against SGG and F. nucleatum may serve as markers for microbial invasion due to the presence of colorectal adenoma (CRA) and CRC. Prospective and patient cohort studies across several European populations, using the same multiplex serology employed here, observed a significant association between SGG antibody response and CRC development [10, 11, 12], while the latter study also found an association with CRA presence. While equivalent prospective studies in both Europe and North America based on multiplex serology did not show any association between F. nucleatum antibody response and CRC [13, 14], similar studies have not yet been conducted in precancerous lesions. Thus, due to limited knowledge of the immune response to these bacterial antigens in colorectal neoplastic progression, we assessed whether antibody responses to SGG and F. nucleatum proteins are associated with stages of neoplasia development from small polyps to more advanced adenomas and cancers in a patient case–control study conducted within a CRC screening population.

Materials and Methods

Clinical Characteristics

This study included 292 individuals from Ireland who donated blood samples prior to bowel preparation for colonoscopy following a positive result for the immunochemical Fecal Occult Blood Test (FIT) or prior to surgical resection of a colorectal tumor. Patients with CRC (n = 25), advanced adenoma (n = 82), and polyps (n = 85) were diagnosed at the Departments of Gastroenterology and Surgery, the Adelaide & Meath Hospital, in Dublin, Ireland (most of the patients with CRC were recruited through the surgery department). Controls (n = 100) were individuals with no colorectal neoplasia detected upon colonoscopy (‘colonoscopy-negative’ controls). All CRCs were classified according to the tenth revision of the International Classification of Diseases (ICD-10). Advanced adenoma include adenomas with high-grade dysplasia (HGD), adenomas with at least 20% tubular villous or villous features, all adenomas greater than 10 mm, and the presence of three or more adenomas [15, 16]. The clinical data, including age at diagnosis, sex, pTNM (tumor stage, regional lymph node involvement, and distant metastasis) staging, and primary tumor localization were taken from patient medical records (see Table 1 for the summary of the clinical characteristics of our study cohorts). All patients gave informed consent in accordance with the Helsinki Declaration and all patient samples were pseudonymized to protect participant identity. The study was approved by the Ethical Committee of the St. James’s Hospital and Federated Dublin Voluntary Hospitals Joint Research Ethics Committee (Ireland, reference 2007-37-17).

Table 1 Clinical characteristics of the studied cohort of patients
Full size table

Sample Collection

The blood samples were collected within one day prior to surgery or colonoscopy in 6-mL VACUTAINER® tubes (Cruinn Diagnostics, Dublin, Ireland) with EDTA. Within 4 h of collection, bloods were centrifuged at 2000×g for 10 min to separate the top plasma layer, which was then stored at − 80 °C in cryovials.

Disease and matched normal mucosal tissue samples were collected during resection of primary tumor or by biopsy, before treatment, while all adenoma biopsies were obtained at colonoscopy during a pilot CRC screening program as described previously [17].

Multiplex Serology

Plasma samples were analyzed for antibody responses against each 11 F. nucleatum and SGG proteins in a final serum sample dilution of 1:100 using multiplex serology in a fluorescent bead-based suspension array, as described previously [18]. Briefly, antigens were recombinantly expressed as Glutathione S-transferase (GST)-tagged fusion proteins and affinity purified on glutathione casein-coupled polystyrene beads (Luminex Corp, Austin, TX, USA) with distinct internal fluorescence [19]. After the pre-incubation step, sera were incubated with the antigen-loaded bead mixture and bound IgG or IgA serum antibodies were labeled separately by biotinylated secondary antibodies (goat anti-Human IgG-Biotin #109-065-098 and goat anti-Human IgA-Biotin #109-065-011, Jackson ImmunoResearch, Westgrove, PA, USA) and a subsequent incubation with Streptavidin-R-Phycoerythrin (MossBio, Pasadena, MD, USA). A Luminex 200 Analyzer (Luminex Corp., Austin, TX, USA) was then used to distinguish the bead sets and their respective antigens as well as to quantify the amount of serum IgG or IgA bound to the antigen. The level of antibody response was given as the median fluorescence intensity (MFI) of at least 100 beads per type measured. Background values against the GST-tag, as well as the bead surface and secondary reagents, were subtracted to generate net MFI values.

No gold standard serological assay was available to validate this multiplex serology. Therefore, antigen-specific cut-offs were defined at the approximate inflection point of frequency distribution curves under the assumption that a sudden rise in the distribution of antibody response over the percentile of sera indicates a cut-off for sero-positivity as we previously described [20]. Thus, for our current study data, MFI values were plotted against the percentage of samples that had at least that MFI value and the cut-off was then set where a higher cut-off would not significantly change the sero-positivity rate.

The multiplex serology assay included 11 proteins from SGG (strain UCN34) and 11 proteins from F. nucleatum (strain ATCC25586) as previously described [11, 13, 21]. Antigen-specific cut-offs with putative protein function are listed in Supplementary Table S1.

DNA Extraction from Colorectal Tissue Biopsies and Quantitative Real-Time Polymerase Chain (qPCR)

To address whether the observed antibody responses to F. nucleatum reflect its presence in the colorectal tract rather than from other potential infection sites, we also correlated the immune responses to the bacterium with existing matched data for 45 subjects on the relative abundance of F. nucleatum in colorectal neoplasia tissue and in the respective normal adjacent mucosa. For the DNA extraction, 20–30 mg of tissue were lysed on ice in 400 μL of lysis buffer (50-mmol/L HEPES pH 7.5, 150-mmol/ L NaCl, 5-mmol/L EDTA) and protease inhibitor (Calbiochem, Hampshire, UK), followed by sonication on ice for 3 × 30 s. Lysates were centrifuged at 10,000g for 10 min at 4 °C. DNA was then extracted using the Norgen All-in-One Purification Kit (cat. no. 24210). DNA was quantified using a NanoDrop 2000c Spectrophotometer (Thermo Scientific, Asheville, NC, USA). DNA extractions were stored at − 80 °C.

Quantitative real-time polymerase chain reaction (qPCR) to quantify the relative abundance of F. nucleatum in both disease and matched normal tissue from patients with CRA or CRC was performed on the Applied Biosystems 7500 Real-Time PCR System (Thermo Fisher Scientific, Dublin, Ireland). Relative quantification (RQ) of F. nucleatum was calculated by 2−ΔCT, where ΔCT is the difference in the copy number threshold (CT) for the test gene (NusG) and reference gene (human prostaglandin transporter, PGT), as described in [17]. Assessment by qPCR of SGG 16 s rRNA relative abundance in the tissue samples provided too few positives (n = 2) to conduct a similar analysis for SGG.

Statistical Analysis

We estimated the association of colorectal neoplasia with respective IgA and IgG sero-positivity to individual F. nucleatum and SGG proteins and combined sero-positivity to both bacteria, using conditional logistic regression models to compute odds ratios (ORs) and 95% confidence intervals (95% CI). Combined sero-positivity was arbitrarily defined as being simultaneously sero-positive to any F. nucleatum and any SGG protein for either IgA or IgG.

Furthermore, we assessed sero-positivity for at least two proteins from a six marker panel subset of SGG antigens (Gallo0272, Gallo0748, Gallo1675, Gallo2018, Gallo2178, and Gallo2179) that have been previously shown to be more strongly associated with CRC risk compared to positivity toward any one of the eleven SGG proteins [11, 12].

To address whether minor inflammatory-related conditions could act as confounders for observed associations, we conducted a sensitivity analysis restricting the control group to those subjects with “no abnormalities detected after colonoscopy” (NAD, n = 37), including hemorrhoids, mild colitis and diverticulosis, or other minor inflammatory conditions.

Analyses were adjusted by age and sex and are presented in the text, except where noted, and in main data tables. The results of the unadjusted analysis are included in the supplementary materials (Supplementary Tables S2 to S13).

Point-biserial test was used to evaluate the correlation between F. nucleatum abundance in both colorectal neoplastic and matched normal tissue and antibody response to the bacterium in plasma (in a smaller cohort of patients with available matched data, n = 45).

Multiple-testing adjustment was conducted using the False Discovery Rate (FDR). Given that the p-values are derived from a clear hypothesis-driven approach with a small number of comparisons across two bacteria, we base our interpretation on the observed p-values but, to be cautious, also present the q-values for the multiple-testing correction in Supplementary Tables S2 to S13. P- and q-values < 0.05 were considered statistically significant. All statistical analyses were performed with IBM SPSS Statistic for Windows, version 27.0 (SPSS Inc., Chicago, Ill., USA) and Rstudio, version 4.0.0 (RStudio Team (2020). RStudio: Integrated Development for R. RStudio, PBC, Boston, MA URL http://www.rstudio.com/).

Results

Association of Sero-positivity to F. nucleatum and SGG Proteins with Colorectal Neoplasia Risk

IgG sero-positivity to F. nucleatum protein Fn1426 was associated with CRC (OR = 4.84, 95% CI 1.46–16.0, Table 3). The corresponding IgA sero-positivity to this antigen was also associated with CRC in the crude model (OR = 4.75, 95% CI 1.25–17.9, Supplementary Table S2), which did not retain statistical significance in the adjusted model although there was a similar disease risk point estimate (OR = 3.84, 95% CI 0.96–15.3, Table 2). We found no association of F. nucleatum sero-positivity with detection of polyps or advanced adenomas (Tables 2 and 3).

Table 2 IgA sero-positivity to individual Fusobacterium nucleatum proteins and association with development of polyps, advanced adenoma, and CRC
Full size table
Table 3 IgG sero-positivity to individual Fusobacterium nucleatum proteins and association with development of polyps, advanced adenoma, and CRC
Full size table

For SGG, IgA sero-positivity to Gallo2179 was associated with CRC in the crude model (OR = 3.50, 95% CI 1.17–10.4, Supplementary Table S5). However, this association was not statistically significant in the adjusted model but with little change in the magnitude of the risk estimate (OR = 2.90, 95% CI 0.91–9.19, Table 4). IgA sero-positivity to any of the SGG antigens was associated with advanced adenoma occurrence (OR = 2.02, 95% CI 1.10–3.71, Table 4). IgA sero-positivity to Gallo0272 and Gallo1675 was associated with occurrence of advanced adenoma (OR = 2.67, 95% CI 1.10–6.46 and OR = 6.17, 95% CI 1.61–23.5, respectively, Table 4). Conversely, IgG sero-positivity to Gallo0112A was inversely associated with the presence of polyps (OR = 0.38; 95% CI 0.15–0.92, Table 5). Positivity to two or more proteins of the 6-marker panel of SGG was not associated with colorectal neoplasia (Tables 4 and 5).

Table 4 IgA sero-positivity to individual SGG proteins and association with development of polyps, advanced adenoma, and CRC
Full size table
Table 5 IgG sero-positivity to individual SGG proteins and association with development of polyps, advanced adenoma, and CRC
Full size table

Finally, dual positivity to both F. nucleatum and SGG was not associated with any of the colorectal neoplasia stages assessed, except for IgG dual positivity to these microbes and a decreased risk of polyp development (OR = 0.46, 95% CI 0.24–0.87, Tables 3 and 5).

Sensitivity Analysis Based on the Control Group

A few differences were observed for the analyses conducted in the ‘NAD’ control group compared to the full control group. Firstly, IgG sero-positivity to Fn0387 was inversely associated with the occurrence of polyps (OR = 0.23, 95% CI 0.06–0.79, Supplementary Table S10). Secondly, we did not find any statistically significant association between IgA sero-positivity to Gallo1675 and advanced adenomas (Supplementary Table S12). Lastly, we no longer observed the inverse association between IgG dual positivity to F. nucleatum and SGG with polyps (Supplementary Tables S10 and S13). We confirmed all the other findings for the analysis conducted in the full control group (Supplementary Tables S8, S11, S12, and S13).

Correlation Between F. nucleatum Tissue Levels and the Antibody Response in Plasma

The relative abundance of F. nucleatum in CRC and CRA disease tissue and the respective normal adjacent mucosa, as previously ascertained by qPCR for 45 patients, showed little correlation with the immune responses. The only exception was a significant positive correlation between F. nucleatum abundance in the normal mucosa and levels of the IgA antibody response to the Fn1426 antigen (Correlation coefficient (r) = 0.38, p < 0.01, Supplementary Table S14).

Discussion

In this study we attempted to detect mucosal and systemic antibody responses to SGG and F. nucleatum by separate detection of IgA and IgG. We observed that sero-prevalence to some of the bacterial antigens varied significantly between cases with colorectal neoplasia and control groups and that some of these differences were associated with colorectal neoplasia across the major developmental stages from polyps to tumors.

It remains to be clarified whether these microbes infect healthy colon tissue prior to the carcinogenic process or if they only infect a developing neoplasia due to bacterial translocation across the impaired gut barrier. In both hypotheses, the bacteria may induce mucosal and systemic antibody responses and potentially cause pro-carcinogenic effects, as previously hypothesized for genotoxic Escherichia coli species and Enterotoxigenic Bacteroides fragilis (ETBF) [20].

Regarding F. nucleatum, IgG and IgA sero-posivitiy to Fn1426 were associated with CRC, although caution is strongly advised in interpreting the large point estimates (4.84 and 4.75, respectively) due to the wide CIs and modest samples numbers for CRC. However, IgA sero-positivity to Fn1426 did not retain significance upon adjustment. Two large prospective studies—one from the European Prospective Investigation into Cancer and Nutrition (EPIC) and another from a cohort consortium in the USA suggested that pre-diagnostic antibody responses to F. nucleatum proteins were not associated with CRC risk [13, 14]. Conversely, a study conducted by Wu et al. showed that patients with CRC infected with F. nucleatum produced higher levels of IgA and IgG than the control groups [22]. Taken together with the results presented here, these studies support the hypothesis that F. nucleatum might act as a “passenger bacterium” increasing in abundance due to favorable growth conditions with dysplastic progression. A similar conclusion on a non-pathogenic reverse causality rationale for the presence of this bacteria was reached by a large German CRC screening study—conducted in stool samples and based on 16S rRNA analysis—where F. nucleatum abundance was strongly associated with CRC but not with adenomas (either advanced or or non-advanced) [23]. However, it remains possible from these observations that the bacterium is involved in the cancer transition from advanced adenomas without having a causative role at the early stage of carcinogenesis.

Concerning SGG, IgA sero-positivity to Gallo2179 was associated with a 3.5-fold odds of CRC development compared to controls, although the statistical significance was not retained upon adjustment. A similar result was found in a Spanish case–control study, based on the same methodology employed here, where the antibody response to Gallo2179, alone and in combination with Gallo2178, was significantly associated with CRC risk [24]. Gallo2178 and Gallo2179 are pilus proteins that are assumed to be virulence factors in SGG-induced infective endocarditis but also in invasion of CRC tissue [9, 25]. The pattern of antibody responses to SGG proteins with precancerous lesions differed. Firstly, positivity for the IgA responses to any of the SGG antigens or separately to Gallo0272 and Gallo1675 were associated with the presence of advanced adenoma. Finally, IgA sero-positivity to Gallo0112A was negatively associated with the presence of polyps. So far, functions of Gallo0272, Gallo1675, and Gallo2018 were only predicted by amino acid similarities to proteins of other bacterial species: Gallo0272 is homologous to one of the agglutinin receptors in the oral bacterium Streptococcus gordonii, which mediates binding to host cell and bacterial receptors representing an important virulence factor [26]. Gallo1675 is a cell wall protein with unknown function [27], while Gallo2018 is putatively involved in bacteriocin synthesis [27].

Several studies have indicated the presence of SGG already in early colorectal lesions, including polyps and adenoma, although antibody responses to SGG were assessed by ELISA and Western blot, respectively [28, 29]. Both studies suggest that SGG may drive the transition from the normal epithelium to the early stages of colorectal neoplasia and thus, CRC. Our results align with these findings as we observed that sero-positivity to SGG proteins were more associated with colorectal adenomas rather than CRC, although, this may reflect the lower power with the smaller numbers of cancers analyzed. The findings that antibody responses to SGG appear before cancer diagnosis or in precancerous lesions suggests that SGG infection may be a potential etiological factor in the transition of a polyp to malignant disease and its detection could help to identify precursors that may more likely progress to cancer [24].

Dual positivity to both F. nucleatum and SGG was not associated with any of the colorectal neoplasia stages assessed except for an IgG dual positivity to these microbes and a decreased risk of polyp development. However, as there were limited group numbers dual positive to these bacterial species (33/85 patients with polyps, 39/85 with advanced adenoma and 10/25 with CRC), larger patient cohort studies are needed to confirm or refute a protective role of the combined sero-positivity to both bacteria in the early stages of carcinogenesis. We did not find any association between positivity to two or more proteins of the six markers SGG panel and colorectal neoplasia, in contrast with the findings in a German case–control study, where positivity to at least two proteins from this panel was significantly associated with a 1.81- and 2.98-fold higher risk of CRC and non-advanced adenoma, respectively [12]. However, this result may be due to the small number of positives for at least two proteins of the panel for both IgA (CRC = 2 and polyps = 8) and IgG antibody responses. In our sensitivity analysis, there were two notable deviations from the adjusted analyses using the full control group. These observations, while limited due to small number of NAD subjects (n = 37), warrants consideration of the control group characteristics in future studies, where a strength here was employing colonoscopy findings to dichotomise the controls into subjects that presented with minor inflammatory-related conditions and those with no pathologies.

Since serology is an indirect systemic of past and/or current infection, it is unknown whether the observed antibody responses result from other infection sites than the colorectum. To partially address this, we correlated the relative quantification of F. nucleatum, ascertained by qPCR of its DNA in disease and matched normal mucosa tissues of 45 study participants, with levels of the IgA and IgG response to F. nucleatum. We observed that levels of IgA antibody response to Fn1426 were positively correlated with F. nucleatum abundance in matched normal mucosa tissue of patients with CRA and CRC. Sero-positivity to this antigen was associated with occurrence of CRC in this study, which indicates that the observed association might truly result from an infection in the gut with F. nucleatum.

Our study has several notable strengths. Firstly, we analyzed the antibody response to F. nucleatum and SGG in different stages of colorectal neoplasia within  a similar demographic cohort and we also conducted an exploratory analysis to assess whether the nature of the control groups may act as a potential confounder. Another strength was the application of multiplex serology to detect different Ig classes and to assess a more detailed immune response to the bacterium beyond just overall sero-positivity. As numerous antigens and two Ig classes were assessed, correcting our findings for multiple comparisons may conceivably be warranted. While we acknowledge that the application of FDR correction would remove significance of our results, this may be considered over-stringent due to the correlated nature of the tested antigens.

A main limitation in this study is the modest sample size across the neoplasia groups, especially for the CRC cases (n = 25). While it was an added strength for this study to have matched immune responses and measured colorectal disease and surrounding tissue abundances of F. nucleatum, this novel analysis was limited by the modest number of matched samples and the lack of measurements of SGG abundance in the tissue. Future prospective studies, specifically detecting F. nucleatum and SGG in stool or tissue biopsies in larger cohorts within different study settings, are needed to help clarify whether the antibody response originates from infections in the colorectal tissue or other sites of the body. Reverse causality is a highly possible factor underlying our results particularly considering a lengthy immune response during the generally long period of neoplastic development. Finally, residual confounding cannot be ruled out, as we could only control for age and sex among relevant covariates (as there were no data on, for example, BMI, smoking, diet, or antibiotic use).

Conclusion

In this study we found that sero-positivity to certain SGG and F. nucleatum proteins were associated with the presence of advanced stages of colorectal neoplasia, including CRC. Thus, the evaluation of antibody response to bacteria may be a useful resource to identify individuals at increased risk for developing CRC or to detect the presence of CRC at the early stages. These findings need to be validated in other settings with increased samples sizes to assess F. nucleatum and SGG serology as a potential biomarker of the immune response to bacterial agents in the developing colorectal neoplasia.