Circulating IgA Antibodies Against Fusobacterium nucleatum Amyloid Adhesin FadA are a Potential Biomarker for Colorectal Neoplasia

Fusobacterium nucleatum (Fn) is a gram-negative oral anaerobe and prevalent in colorectal cancer. Fn encodes a unique amyloid-like adhesin, FadA complex (FadAc), consisting of intact pre-FadA and cleaved mature FadA, to promote colorectal cancer tumorigenesis. We aimed to evaluate circulating anti-FadAc antibody levels as a biomarker for colorectal cancer. Circulating anti-FadAc IgA and IgG levels were measured by ELISA in two study populations. In study 1, plasma samples from patients with colorectal cancer (n = 25) and matched healthy controls (n = 25) were obtained from University Hospitals Cleveland Medical Center. Plasma levels of anti-FadAc IgA were significantly increased in patients with colorectal cancer (mean ± SD: 1.48 ± 1.07 μg/mL) compared with matched healthy controls (0.71 ± 0.36 μg/mL; P = 0.001). The increase was significant in both early (stages I and II) and advanced (stages III and IV) colorectal cancer. In study 2, sera from patients with colorectal cancer (n = 50) and patients with advanced colorectal adenomas (n = 50) were obtained from the Weill Cornell Medical Center biobank. Anti-FadAc antibody titers were stratified according to the tumor stage and location. Similar as study 1, serum levels of anti-FadAc IgA were significantly increased in patients with colorectal cancer (2.06 ± 1.47 μg/mL) compared with patients with colorectal adenomas (1.49 ± 0.99 μg/mL; P = 0.025). Significant increase was limited to proximal cancers, but not distal tumors. Anti-FadAc IgG was not increased in either study population, suggesting that Fn likely translocates through the gastrointestinal tract and interact with colonic mucosa. Anti-FadAc IgA, but not IgG, is a potential biomarker for early detection of colorectal neoplasia, especially for proximal tumors. Significance: Fn, an oral anaerobe highly prevalent in colorectal cancer, secretes the amyloid-like FadAc to promote colorectal cancer tumorigenesis. We report that circulating levels of anti-FadAc IgA, but not IgG, are increased in patients with both early and advanced colorectal cancer compared with the healthy controls, and especially in those with proximal colorectal cancer. Anti-FadAc IgA may be developed into a serological biomarker for early detection of colorectal cancer.

FadA is a unique adhesin highly conserved among pathogenic Fusobacterium species and a major virulence factor responsible for bacterial adhesion, induction of inflammation, and colorectal cancer cell proliferation (7,11). The fadA gene levels are 10-to 100-fold higher in colorectal tissues of patients with adenomas or colorectal cancer compared with the normal subjects (7). It is also significantly increased in the fecal microbiome in patients with colorectal cancer compared with the controls (21). FadA consists of two forms, the intact pre-FadA and mature FadA without signal peptide, both of which constitute the active and amyloid-like FadA complex (FadAc; ref. 37). The fadA mRNA is constitutively expressed (38); however, amyloid FadAc is only produced under stress and disease conditions, serving as a molecular switch to convert Fn from a benign commensal to a pathogen (39). Amyloid FadAc binds to extracellular domain of E-cadherin leading to the proliferation of cancer cells via activation of Annexin A1 and Wnt/β-catenin signaling (7,11).
Early diagnosis and treatment of colorectal cancer is the most important determining factor for prognosis and patient survival. Although there are several commercially available serologic biomarkers for colorectal cancer such as carcinoembryonic antigen (CEA), alpha fetoprotein (AFP), and cancer antigen (CA) 19-9 (40-42), it has become clear that these conventional biomarkers have low sensitivity and specificity and are not suited for early diagnosis and predicting prognosis of colorectal cancer (43)(44)(45). Because of the high correlation between Fn and colorectal cancer progression, there have been attempts to use Fn as a biomarker. In this study, we investigate the levels of circulating anti-FadA antibodies in the plasma and serum from two U.S. cohorts to evaluate its applicability as a novel colorectal cancer biomarker.

Sample Description
Samples from two existing colorectal cancer sources were used following appropriate ethical guidelines. Study 1 samples were randomly selected from the Kentucky Colon Cancer Genetic Epidemiology Study, a population-base casecontrol study based on the Kentucky Cancer Registry (KCR; refs. 46,47). Recruitment of participants was conducted between April 2003 and December 2010. A total of 1,040 incident colon cancer cases and 1,750 population-based controls completed the study with the collection of comprehensive lifestyle and epidemiological data, pathology information, and fasting blood samples. Cases were defined as individuals diagnosed with histopathologically confirmed incident primary colon cancer (excluding patients with rectal or syndromic cancers) who were invited to participate in the study within 3 months of KCR registration. Cases were eligible to participate if they: (i) had a nonrecurrent diagnosis; (ii) had no known family history or diagnosis of familial adenomatous polyposis (FAP), hereditary nonpolyposis colorectal cancer (HNPCC), Peutz-Jeghers, or Cowden disease; (iii) had no known diagnosis of inflammatory bowel disease such as Crohn disease or ulcerative colitis; (iv) were at least 21 years of age at the time of diagnosis; (v) had contact information listed in the KCR database; (vi) were willing to complete two questionnaires. The majority of participants completed data collection within 12 months (median of 5 months) of their colon cancer diagnosis.
Random digit dialing and friend referrals were utilized to recruit controls representative of the general Kentucky population. Controls consisted of frequency-matched individuals who have never been diagnosed with any cancer except nonmelanoma skin cancer and are over the age of 30, preferably ≥ 50 years old. For cases and controls, self-reported inflammatory bowel disease (e.g., Crohn disease or ulcerative colitis), family history of FAP, and HN-PCC were excluded in the recruitment. The response rate for cases and controls who answered the phone and allowed eligibility determination was 70.8% and 66.7%, respectively. The study was approved by the Institutional Review Boards of the University of Kentucky (Lexington, Kentucky) and University of Virginia (Charlottesville, VA). All participants provided written informed consent.
Eligible cases and controls donated one blood sample and answered selfadministered questionnaires. A two-step approach was used to collect blood samples and lifestyle risk factor data. First, a prepacked phlebotomy kit with detailed written instructions for blood sample collection and written consent forms was sent to each case subject. Participants were instructed to go to their physician offices or adjacent medical facilities for blood draw after overnight fasting. The samples were collected in purple-top (K3EDTA) blood collection tubes and shipped overnight on frozen ice pack. Upon receipt, the blood tubes were spun for 15 minutes at 600 × g and aliquots of plasma and concentrated buffy coat were prepared and frozen at −80°C. Second, a self-administered lifestyle risk factor questionnaire developed by the NCI Colon Cancer Familial Cancer Registry was sent to each participant to collect detailed information on demographics and lifestyle risk factors. The parent Kentucky Colon Cancer Genetic Epidemiology Study is not a matched (or paired) case-control study.
Deidentified plasma samples from 25 randomly selected patients with colorectal cancer and 25 controls matched by gender and age (±1 year) were tested. The patient characteristics are summarized in Table 1.

Study 2 included randomly selected patients with advanced adenoma or adenocarcinoma who underwent surgical treatment at NYP/Weill Cornell Medical
Center from 2013 to 2015. All patients who were seen in the colorectal group were offered participation in the clinical trial. Tissues were collected among those patients who provided consent, and tissue acquisition was feasible. An advanced adenoma was defined as a polyp that was greater than 3 cm in size, and was unable to be removed endoscopically. Deidentified serum samples from 50 patients with adenoma and 50 patients with adenocarcinoma were tested. Tumor stages were classified according to the Union for International Cancer Control classification. The patient characteristics are summarized in Table 1. For tumor locations in study 2, "proximal" include samples labeled as cecum, ascending, ascending/transverse, right colon, or right/cecum, while "distal" include samples labeled as sigmoid, descending, sigmoid/rectum, left colon, rectosigmoid junction, rectum. Samples from patients with mixed proximal and distal tumors were excluded. No tumor location was available for study 1. The study was approved by the Institutional Review Boards of the Weill Cornell Medical Center. All participants provided written informed consent.

Preparation of Monoclonal Anti-FadA IgG
The mouse monoclonal (mAb) anti-FadA IgG (5G113G8) was prepared at the Hybridoma Core (Lerner Research Institute) described previously (37) and was used as a standard for determining the concentrations of anti-FadAc IgA and IgG in the samples. This approach is feasible because FadA is highly conserved in Fn (48).

ELISA
The levels of anti-FadAc IgA and IgG were determined by indirect ELISA. The 96-well ELISA plates were coated with 100 μL of purified recombinant FadAc (2 μg/mL) in 0.2 mol/L carbonate/bicarbonate buffer (pH 9.4) and incubated at 4°C overnight. After blocking, the wells were incubated with 100 μL of blood samples serially diluted in blocking buffer for 1 hour at room temperature. To obtain a standard curve, the wells were incubated with 100 μL of purified mouse mAb 5G113G8 at increasing concentrations. Each well was then incubated with 100 μL of goat anti-human IgA-HRP (1:6,000 dilution; PA1-74395, Thermo Fisher Scientific), goat-anti-human IgG-HRP (1:10,000 dilution; A2290, Sigma-Aldrich), or goat anti-mouse IgG-HRP (1:12,000 dilution; 62-6520, Thermo Fisher Scientific) for 1 hour at room temperature, followed by incubation with 100 μL substrate solution (1-Step Ultra TMB-ELISA; Thermo Fisher Scientific) for 30 minutes. The reaction was terminated with 2 mol/L H 2 SO 4 and the absorbance at 405 nm was measured using an automated microplate reader. The levels of anti-FadAc IgA and IgG were determined by testing serial dilutions of the plasma or serum. The values that fell into the linear range of the standard curves generated using mouse mAb 5G113G8 were used to calculate the antibody concentrations. Wells with secondary antibodies alone were used to determine background values. All measurements were performed in duplicate. All assays were performed by one single individual. Intraplate and interplate coefficient of variation (CV) in study 1 and study 2 were as follow: study 1 intraplate CV = 8.2 ± 6.12%, interplate CV = 9.5 ± 0.6%; study 2 intraplate CV = 9.7 ± 6.7%, interplate CV = 10.8 ± 1.7%.

Statistical Analysis
Categorical variables were presented as numbers and percentages and were compared using χ 2 test or Fisher exact test between two groups. Continuous variables were expressed as mean ± SD and were compared using two-tailed Student t test between two groups. The optimal cutoffs of plasma and serum levels of anti-FadAc IgA and IgG were identified with ROC curve analysis using Youden index criterion with the requirement that sensitivity and specificity were at least 0.50. To examine the effect of colorectal cancer status in relation to the levels of anti-FadAc IgA and IgG, a multivariable linear regression model was used to adjust for confounding factor(s) (e.g., gender) with P value

Data Availability Statement
Data are available upon request.

Plasma Levels of Anti-FadAc IgA, but not IgG, are Significantly Increased in Patients with Colorectal Cancer Compared with Healthy Controls
In study 1, the plasma levels of anti-FadAc IgA were significantly increased in patients with colorectal cancer compared with healthy subjects (1.48 ± 1.07 μg/mL vs. 0.71 ± 0.36 μg/mL, P = 0.001; Fig. 1A). The Youden index criterion in the ROC curve analysis yielded the cutoff of 0.81 μg/mL for anti-FadAc IgA levels, with 76% sensitivity and 76% specificity. No difference in anti-FadAc IgG was detected between the two groups (1.20 ± 1.26 μg/mL vs. 1.33 ± 1.07 μg/mL, P = 0.71; Fig. 1B). Furthermore, when the IgA titers were stratified by cancer stage, significant increase was detected in both early (stages I and II) and advanced (stages III and IV) colorectal cancer, compared with the normal controls (Fig. 1C). No tumor location was available for study 1.

Serum Levels of Anti-FadAc IgA, But not IgG, are Significantly Increased in Patients with Colorectal Cancer Compared with Advanced Precancerous Polyps
Similarly, serum anti-FadAc IgA titers from study 2 were also found to be significantly increased in patients with colorectal cancer compared with those with advanced adenomas (2.06 ± 1.47 μg/mL vs. 1.49 ± 0.99 μg/mL, P = 0.025; Fig.   2A). There was a significant gender difference between the adenoma and colorectal cancer groups in study 2 (Table 1). After adjusting for gender, anti-FadAc IgA levels remain significantly higher in colorectal cancer compared with the advanced adenoma group (estimate = 0.592 μg/mL, SE = 0.262, P = 0.026).
When the results were stratified with tumor stage and location, the anti-FadAc IgA levels were only increased in advanced colorectal cancer compared to the advanced adenoma group (Fig. 2C). However, for patients with proximal tumors, the IgA levels were significantly increased in both early and advanced stages of colorectal cancer compared with the advanced adenoma group (Fig. 2D).

Discussion
In this investigation, we found that levels of circulating anti-FadAc IgA were significantly increased in patients with colorectal cancer compared with healthy controls or to patient with advanced adenomas. The anti-FadAc IgA levels were elevated in both early and advanced colorectal cancer suggesting that it may be useful for not only advanced-stage but also early-stage diagnosis. This is consistent with our previous finding that the fadA genes levels are increased stepwise from normal to adenomas, and from adenomas to carcinomas (7). The adenoma group in study 2 exhibited higher titers than the normal subjects of cohort 1. However, because different specimens (plasma vs. serum) were used for these two studies, the results are not directly comparable. The elevated anti-FadAc IgA is closely associated with proximal tumors, consistent with previous reports of enriched Fn in proximal colorectal cancer (30,31). Interestingly, we did not observe changes in IgG levels. These findings suggest that Fn likely translocates through the digestive tract inducing mucosal immune responses.
Serologic diagnosis is desirable due to the noninvasive nature, as well as easy attainment of the specimens. Previous studies reported colorectal cancer diagnosis using serum CEA, AFP, and CA19-9, achieving sensitivities of 80.43%, 73.91%, and 69.57%, and specificities of 75.00%, 69.44%, and 61.11%, respectively (41). Using anti-FadAc IgA, we achieved sensitivity and specificity both at 76%, demonstrating its superiority over AFP and CA 19-9 as potential markers. Using whole bacteria Fn as antigen, a previous study tested anti-Fn IgA in colorectal cancer, with a low sensitivity of 36.43%. Even when combined with CEA, the sensitivity was still only 53.10% (49). This may be due to the presence of common bacterial components conserved among most bacterial species such as outer membrane proteins and lipopolysaccharides making it difficult to predict colorectal cancer. Given that FadA is uniquely conserved in Fn, anti-FadAc antibody is a more specific biomarker. Furthermore, we used the colorectal cancer-promoting form of FadA, that is, the amyloid FadAc (39), thus achieving significantly improved sensitivity. We used the Youden index with the requirement of both sensitivity and specificity greater than 0.5 for the selection of the cut-off points. By doing so, it guarantees the cutoff yields a better approach than a random decision with a fair coin toss for potential clinical diagnosis. We anticipate that combining circulating anti-FadAc IgA with additional biomarkers may further improve the diagnostic sensitivity and specificity.
Our study had a few limitations. First, our sample size was limited, especially for those with clearly defined tumor locations. Future studies with larger sample sizes and external validations are warranted to confirm the findings, and to compare with the above-mentioned serologic parameters to stratify to additional patient and tumor characteristics, such as smoking, tumor histology, and molecular types. Second, as a proof-of-concept study, we performed retrospective analysis using samples available to us. As a result, different specimens were used in the two studies, one with plasma and the other with serum, rendering the results from the two studies incomparable. There was no adenoma group in study 1 and no healthy controls in study 2; therefore, we do not know whether anti-FadAc IgA can also be used for early detection of advanced precancerous adenomas. Third, given the retrospective nature of study, we cannot dismiss the possibility of reverse causality. As such, caution must be exercised for causal interpretation of our results. Prospective cohort study is warranted.
In summary, our study sheds novel light on using specific bacterial virulence factors as biomarkers for detection of colorectal neoplasia. Anti-FadAc IgA is a potential novel biomarker for colorectal cancer, especially for the tumors located in the proximal colon.
Conceptualization, supervision, funding acquisition, writing-original draft, writing-review and editing.