Abstract
Background:
Antibody responses to Streptococcus gallolyticus subspecies gallolyticus (SGG) proteins, especially pilus protein Gallo2178, have been consistently associated with colorectal cancer (CRC) risk. Previous case-control studies and prospective studies with up to 8 years of follow-up, however, were unable to decipher the temporality of antibody responses to SGG in the context of the long-term multistep development of CRC. In this study, we analyzed a large US CRC cohort consortium with follow-up beyond 10 years for antibody responses to SGG.
Methods:
We applied multiplex serology to measure antibody responses to 9 SGG proteins in participants of 10 prospective US cohorts (CLUE, CPSII, HPFS, MEC, NHS, NYUWHS, PHS, PLCO, SCCS and WHI) including 4,063 incident CRC cases and 4,063 matched controls. Conditional logistic regression was used to assess whether antibody responses to SGG were associated with CRC risk, overall and by time between blood draw and diagnosis.
Results:
CRC risk was increased among those with antibody responses to Gallo2178, albeit not statistically significant (OR: 1.23; 95%CI: 0.99-1.52). This association was stronger for cases diagnosed <10 years after blood draw (OR: 1.40; 95% CI: 1.09-1.79), but was not found among cases diagnosed ≥10 years after blood draw (OR: 0.79; 95% CI: 0.50-1.24).
Conclusion:
In a large cohort consortium, we reproduced the association of antibody responses to SGG Gallo2178 with CRC risk for individuals diagnosed within 10 years after blood draw.
Impact:
This timing-specific finding suggests that antibody responses to SGG are associated with increased CRC risk only after tumorigenesis has begun.
Introduction
The intestinal commensal Streptococcus gallolyticus subsp. gallolyticus (SGG), formerly known as Streptococcus bovis type I, has been frequently associated with the presence of colorectal adenoma and cancer (CRC) (1). The connection was first identified when cases of SGG-caused infective endocarditis presented with a concomitant adenoma/cancer in the gut (2, 3).
Previous cross-sectional studies have shown that SGG is associated with CRC as well as with precursor lesions. These studies were performed either directly by detecting bacterial DNA in tumor tissue or indirectly by serological methods (4-12). In our laboratory we established a multiplex serology assay (13) to measure antibody responses to up to eleven SGG antigens. These antigens were selected based on either known immunogenicity or predicted localization on the surface of the bacterium (12). Of specific interest were the proteins Gallo2178 and Gallo2179 that build a pilus structure, which has been shown to be important for adhesion and virulence of SGG (14, 15). Our laboratory group has previously applied SGG multiplex serology to analyze two CRC case-control studies conducted in Spain and Germany and found a reproducible association between antibody responses to SGG pilus protein Gallo2178, potentially detecting present as well as past exposure of SGG to the host’s immune system, and the presence of CRC, with significant odds ratios (OR) ranging from 1.58 to 4.13 (11, 12). Our laboratory group further analyzed serum samples from newly diagnosed CRC cases in the European Prospective Investigation into Nutrition and Cancer (EPIC) study and showed that antibody responses to Gallo2178 were statistically significantly associated with a three-fold higher risk of CRC in pre-diagnostic blood samples taken up to 8 years before diagnosis with 2% of controls and 6% of cases affected. Additionally, we identified two other proteins included in SGG multiplex serology (Gallo0272 and Gallo0748) to be significantly associated with CRC risk in the EPIC study, however, with these two proteins the risk was increased by only up to 60% (16).
The cross-sectional findings involving adenomas and the finding with the prospective design demonstrated that SGG infection occurred prior to clinical CRC diagnosis. The limitation, however, of the previous prospective study was a relatively short duration of follow-up (8 years). This is important in the background of the long-term multistep process of genetic and morphological changes in the gut epithelial tissue that lead to CRC development. The process of development from an initial polyp to CRC is estimated to last on average 10-15 years but may vary between 5 and 25 years, dependent on the type of polyp/adenoma and personal risk factor background (17). Thus, for CRC cases diagnosed within 10 years of antibody assessment, carcinogenesis has probably already begun. Current literature suggests that CRC-specific conditions promote colonization of SGG in the gut (18, 19) and that SGG may contribute directly to the process of carcinogenesis (10).
To further support our investigation into the timing of the SGG antibody association with CRC development, we also sought to assess a potential difference in the association by p53 auto-antibody status. Loss-of-function of the tumor suppressor p53 has been found to drive the transition from late adenomas to cancer (20). Missense mutations in the p53 gene lead to inactivation of the protein’s function, and the mutated protein then accumulates for unknown reasons in the cancer cells (21). A minority of these patients (20-40%) develop auto-antibodies against the accumulating p53, which have been shown to be a specific but insensitive marker for presence of CRC (22). For example, a recent prospective study in the Cancer Prevention Study-II (CPS-II) found a statistically significant association between anti-p53 antibodies and CRC risk within 3 years of diagnosis with 13% sero-positive CRC cases and 6% sero-positive controls (23). Thus, assuming that p53 auto-antibodies serve as a surrogate for presence of undiagnosed (pre-)cancerous colorectal lesions, if the association between SGG and CRC is strongest for individuals with detectable p53 auto-antibodies, this would also suggest that the role of SGG in CRC development begins after the initiation of the carcinogenesis process.
In summary, in the present study, we sought to examine the temporality of the association between pre-diagnostic antibody responses to SGG and CRC by analyzing data and samples from a large US cohort consortium with median follow-up times ranging from 4 to 18 years in the participating studies, enabling us to assess how long before CRC diagnosis antibody responses to SGG can be associated with CRC risk. We further explored the interaction of p53 and SGG serology under the assumption that p53 auto-antibodies indicate colorectal lesions already present at baseline blood draw. Both aims seek to address the association of SGG and CRC under the a priori hypothesis that development of antibody responses against SGG is dependent on the presence of a (pre-)cancerous lesion.
Materials and Methods
Study population
This CRC cohort consortium comprises 10 prospective US cohorts: Campaign Against Cancer and Stroke (CLUE) (24), Cancer Prevention Study-II (CPSII) (25), Health Professionals Follow-up study (HPFS) (26), Multiethnic Cohort Study (MEC) (27), Nurses’ Health Study (NHS) (28), NYU Women’s Health Study (NYUWHS) (29), Physicians’ Health Study (PHS) (30), Prostate, Lung, Colorectal, and Ovarian Screening Study (PLCO) (31), Southern Community Cohort Study (SCCS) (32) and Women’s Health Initiative (WHI) (33).
Participating cohorts contributed pre-diagnostic blood samples and baseline socio-demographic information from CRC cases and controls for the present study. Cases were defined based on the International Classification of Diseases for Oncology (ICD-O-3) and included all cancers of colon and rectum coded as C180-189, C199, and C209. Controls were randomly selected and matched at a 1:1 ratio to each CRC case by cohort, sex, race, date of birth (± 1 year, relaxed up to ± 5 years for sets without available controls), and date of blood collection (± 1 month, relaxed up to ± 3 months, and further to ± 6 months for sets without available controls). All controls were alive and free of cancer (except non-melanoma skin cancer) at the time of diagnosis of the index case. Study-specific information on cohort size, age at blood draw, number of contributed cases, sex distribution and range of follow-up time is given in Supplementary table S1.
In total, samples from 4,210 CRC cases and their 4,210 matched controls were assayed by multiplex serology. One hundred samples and therefore their 100 matched case or control counterparts had to be excluded due to technical issues, including insufficient volume (n=27), pipetting errors (n=52) or invalid measurements due to insufficient bead counts (n=21). Further, 47 pairs were excluded due to mismatches in race and sex, resulting in a final sample number of 4,063 cases and their respective 4,063 controls.
Multiplex serology
Serum samples were sent on dry ice to the German Cancer Research Center (DKFZ, Heidelberg, Germany) and analyzed in a 1:1000 final serum dilution. Multiplex serology was performed as described previously (12, 13, 16, 34). Briefly, multiplex serology is a fluorescent bead-based assay allowing for analysis of antibody responses to several antigens in one reaction. Antigens were expressed as Glutathione-S-transferase (GST)-tagged fusion proteins and affinity-purified on polystyrene beads (Luminex Corp, Austin, TX, USA) coupled to glutathione-casein. Different antigens were purified on different bead sets as defined by the bead’s internal fluorescence. The antigen-loaded bead sets were mixed and incubated with serum. A Luminex flow cytometer then distinguished between the bead set, and therefore the loaded antigen, as well as quantified the amount of bound serum antibody by a secondary antibody detecting human IgG, IgA, and IgM and a fluorescent reporter conjugate (Streptavidin-R-phycoerythrin). The output was the median fluorescence intensity (MFI) measured on at least 100 beads per set per sample. Net MFI were generated by subtracting two background values resulting from a well containing no serum but antigen-loaded beads and all secondary reagents as well as from a bead set loaded with GST-tag only.
SGG antigens included in the multiplex serology were expressed from strain UCN34. Previous analyses with this assay applied in total eleven SGG antigens (12, 16); however, we excluded two previously non-informative antigens for the present study (Supplementary table S2). P53 serology was applied as described previously (34).
Antigen-specific cut-offs were defined arbitrarily by visual inspection of percentile plots at the approximate inflection point of the curve to dichotomize antibody responses as sero-positive and –negative as previously described for other antigens (35-37) (Supplementary table S2).
Of 82 duplicates within the WHI study set incorporated as blinded quality control samples, correlations for antibody responses (MFI) to SGG antigens ranged from 0.91 to 1.0 indicating a good reproducibility of the measured values.
Statistical analysis
Pearson’s chi-square test was used to assess differences in baseline characteristics between SGG Gallo2178-negative and -positive controls. Conditional logistic regression was applied to analyze the association of each individual SGG protein as well as p53 antibodies with CRC risk and to determine odds ratios (OR) and 95% confidence intervals (CI), overall and by interval between blood draw and diagnosis (<10 years versus ≥10 years). A p-value of below 0.05 was considered statistically significant. All matched case-control sets were exactly matched by study, race and sex but we observed residual confounding by age, which, however, did not affect the strength of the estimates for the exposures of interest. Potential confounders (apart from the matching variables age, sex and race within each cohort) were defined a priori and included education, smoking, BMI, and family history of CRC. To note, among these there was a substantial amount of missing data: 13% of study participants were lacking data on BMI, and 25% on family history of CRC. Among the potential confounders a BMI greater than 30 was associated both with CRC risk and antibody positivity to Gallo2178 (Table 1 and 2). However, adjusting for any of the potential confounding variables, while excluding participants with missing data, did not alter the SGG estimates by more than 10% and therefore results are presented without further adjustment. The observed difference in prevalence of antibody positivity to Gallo2178 by study was accounted for by matching and subsequent conditional logistic regression. However, we also analyzed the association using conditional logistic regression separately by study to see whether these differences potentially affected the association. This analysis was performed separately for cases diagnosed within 10 years after their blood draw and cases diagnosed longer than 10 years after blood draw since studies differed in their follow-up time, which was hypothesized to modify the SGG-CRC association.
Table 1:
Baseline characteristics of the cohorts participating in this study.
| Variable | Total (n=8126) | Controls (n=4063) | Cases (n=4063) |
|---|---|---|---|
| Study, n(%) | |||
| CLUE | 982 (12) | 491 (12) | 491 (12) |
| CPSII | 722 (9) | 361 (9) | 361 (9) |
| HPFS | 302 (4) | 151 (4) | 151 (4) |
| MEC | 1510 (19) | 755 (19) | 755 (19) |
| NHS | 576 (7) | 288 (7) | 288 (7) |
| NYUWHS | 572 (7) | 286 (7) | 286 (7) |
| PHS | 360 (4) | 180 (4) | 180 (4) |
| PLCO | 1240 (15) | 620 (15) | 620 (15) |
| SCCS | 252 (3) | 126 (3) | 126 (3) |
| WHI | 1610 (20) | 805 (20) | 805 (20) |
| Age at blood draw [years] | |||
| Median (range) | 64 (18–89) | 64 (18–88) | 64 (18–89) |
| Sex, n (%) | |||
| Female | 5112 (63) | 2556 (63) | 2556 (63) |
| Male | 3014 (37) | 1507 (37) | 1507 (37) |
| Race/Ethnicity, n (%) | |||
| White | 6134 (75) | 3067 (75) | 3067 (75) |
| African-American | 798 (10) | 399 (10) | 399 (10) |
| Asian-American | 614 (8) | 307 (8) | 307 (8) |
| Latino | 422 (5) | 211 (5) | 211 (5) |
| Other/unknown/multiracial | 158 (2) | 79 (2) | 79 (2) |
| Education, n (%) | |||
| Less than HS | 972 (12) | 468 (12) | 504 (12) |
| Completed HS or GED | 1668 (21) | 823 (20) | 845 (21) |
| Post HS training other than college | 362 (4) | 183 (5) | 179 (4) |
| Some college | 1672 (21) | 845 (21) | 827 (20) |
| College graduate | 1483 (18) | 756 (19) | 727 (18) |
| Graduate school | 1861 (23) | 946 (23) | 915 (23) |
| Missing | 108 (1) | 42 (1) | 66 (2) |
| BMI1 [kg/m2], n (%) | |||
| <30 | 5360 (66) | 2781 (69) | 2579 (64) |
| ≥30 | 1691 (21) | 748 (18) | 943 (23) |
| Missing | 1075 (13) | 534 (13) | 541 (13) |
| Smoking, n (%) | |||
| Never | 3628 (45) | 1853 (46) | 1775 (44) |
| Ever | 4409 (54) | 2168 (53) | 2241 (55) |
| Missing | 89 (1) | 42 (1) | 47 (1) |
| Family history of CRC2, n (%) | |||
| No | 5167 (64) | 2638 (65) | 2529 (62) |
| yes | 907 (11) | 408 (10) | 499 (12) |
| Missing | 2052 (25) | 1017 (25) | 1035 (26) |
all studies except CLUE (variable not available)
all studies except NYUWHS, CLUE (variable not available), SCCS (<75% of variable information available)
Table 2:
Risk factors for antibody positivity to SGG protein Gallo2178 among controls in the cohort consortium.
| Variable | Gallo2178 neg (n=3900) | Gallo2178 pos (n=163) | p-value3 |
|---|---|---|---|
| Study, n (%) | |||
| CLUE | 484 (12) | 7 (4) | |
| CPSII | 352 (9) | 9 (6) | |
| HPFS | 143 (4) | 8 (5) | |
| MEC | 701 (18) | 54 (33) | |
| NHS | 271 (7) | 17 (10) | |
| NYUWHS | 268 (7) | 18 (11) | |
| PHS | 174 (4) | 6 (4) | |
| PLCO | 607 (16) | 13 (8) | |
| SCCS | 116 (3) | 10 (6) | |
| WHI | 784 (20) | 21 (13) | <0.0001 |
| Age at blood draw [years] | |||
| Median (range) | 64 (18–88) | 64 (31–84) | 0.399 |
| Sex, n (%) | |||
| Female | 2447 (63) | 109 (67) | |
| Male | 1453 (37) | 54 (33) | 0.285 |
| Race/Ethnicity, n (%) | |||
| White | 2976 (76) | 91 (56) | |
| African-American | 370 (9) | 29 (18) | |
| Asian-American | 293 (8) | 14 (9) | |
| Latino | 189 (5) | 22 (13) | |
| Other/unknown/multiracial | 72 (2) | 7 (4) | <0.0001 |
| Education, n (%) | |||
| Less than HS | 442 (11) | 26 (16) | |
| Completed HS or GED | 793 (20) | 30 (18) | |
| Post HS training other than college | 178 (5) | 5 (3) | |
| Some college | 814 (21) | 31 (19) | |
| College graduate | 722 (19) | 34 (21) | |
| Graduate school | 913 (23) | 33 (20) | 0.370 |
| Missing | 38 (1) | 4 (2) | |
| BMI1 [kg/m2], n (%) | |||
| <30 | 2670 (68) | 111 (68) | |
| >=30) | 704 (18) | 44 (27) | 0.025 |
| Missing | 526 (13) | 8 (5) | |
| Smoking, n (%) | |||
| Never | 1765 (45) | 88 (54) | |
| Ever | 2096 (54) | 72 (44) | 0.021 |
| Missing | 39 (1) | 3 (2) | |
| Family history of CRC2, n (%) | |||
| No | 2532 (65) | 106 (65) | |
| Yes | 391 (10) | 17 (10) | 0.887 |
| Missing | 977 (25) | 40 (25) | |
all studies except CLUE (variable not available)
all studies except NYUWHS, CLUE (variable not available), SCCS (<75% of variable information available)
Chi-Square test for categorical variables, t-test for continuous variables (age); p-values below 0.05 are considered statistically significant and are marked in bold font
Analysis of the association of antibody responses to SGG with CRC risk stratified by p53 antibody positivity was performed using an unconditional logistic regression model adjusting for matching factors age, sex, study and race/ethnicity.
We further explored the association of SGG with CRC risk separately by different case characteristics. Specifically, we assessed the association in separate models by stage according to TNM classification (early stage I/II and late stage III/IV), site (colon (left, right, not otherwise specified (NOS)), rectum) and age at diagnosis (≤65, 66-75, 76-85 and >85 years).
Results
Cases and controls differed in their baseline characteristics with respect to BMI and family history of CRC: specifically, cases were more likely to be obese and to have a positive CRC family history (Table 1).
Since only antibody responses to SGG protein Gallo2178 were significantly associated with CRC we assessed potential risk factors for Gallo2178 antibody positivity but not the other SGG proteins. Prevalence of antibody positivity to Gallo2178 among controls differed by study, with higher prevalence in MEC and SCCS and lower in WHI, as well as by race and ethnicity, with lower prevalence among whites and higher prevalence in African Americans and Latinos. Additionally, obese individuals and never smokers were more likely to be Gallo2178 antibody positive (Table 2).
Overall, antibody positivity to none of the nine SGG proteins was associated with increased CRC risk. The only suggestion of an increased risk was seen for antibody positivity to Gallo2178, the one protein previously associated with CRC risk, with 4% sero-prevalence among controls compared to 5% among cases, which resulted in a 23% statistically non-significant increase in odds for CRC risk. When exploring the association in individuals diagnosed within 10 years after blood draw, the OR for most of the antigens remained around 1, whereas the association of Gallo2178 with CRC risk was stronger with a statistically significant OR of 1.40 (95% CI: 1.09-1.79, p-value = 0.008). Excluding those individuals diagnosed within two years after blood draw from this subgroup did not alter the observed association of Gallo2178 with CRC risk substantially (OR: 1.38; 95% CI: 1.05-1.82, p-value = 0.020). Among individuals diagnosed more than 10 years after blood draw, the OR for most of the SGG antigens was below 1 and there was no association of Gallo2178 with CRC risk in this subgroup (OR: 0.79, 95% CI: 0.50-1.24, p-value = 0.300) with 4% sero-positive controls compared to 3% sero-positive cases (Table 3).
Table 3:
Antibody positivity to SGG proteins and CRC risk, overall and by time between blood draw and diagnosis (follow-up).
| Overall | <10 years follow-up | ≥10years follow-up | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SGG protein | Controls n=4063 n pos (%) | Cases n=4063 n pos (%) | OR1 | 95% CI | p-value | Controls n=2766 n pos (%) | Cases n=2766 n pos (%) | OR1 | 95% CI | p-value | Controls n=1296 n pos (%) | Cases n=1296 n pos (%) | OR1 | 95% CI | p-value |
| Gallo0112A | 320 (8) | 308 (8) | 0.96 | 0.81-1.13 | 0.610 | 228 (8) | 219 (8) | 0.96 | 0.78-1.17 | 0.647 | 92 (7) | 89 (7) | 0.96 | 0.71-1.31 | 0.815 |
| Gallo0272 | 292 (7) | 302 (7) | 1.04 | 0.88-1.23 | 0.671 | 215 (8) | 237 (9) | 1.11 | 0.92-1.35 | 0.280 | 77 (6) | 65 (5) | 0.84 | 0.60-1.17 | 0.311 |
| Gallo0577 | 246 (6) | 256 (6) | 1.05 | 0.87-1.26 | 0.636 | 175 (6) | 198 (7) | 1.15 | 0.93-1.43 | 0.207 | 71 (5) | 59 (4) | 0.80 | 0.55-1.15 | 0.226 |
| Gallo0748 | 394 (10) | 372 (9) | 0.94 | 0.81-1.09 | 0.402 | 271 (10) | 263 (10) | 0.97 | 0.81-1.16 | 0.717 | 123 (9) | 109 (8) | 0.87 | 0.66-1.15 | 0.325 |
| Gallo1570 | 266 (7) | 259 (6) | 0.97 | 0.81-1.16 | 0.748 | 183 (7) | 191 (7) | 1.05 | 0.85-1.30 | 0.663 | 83 (6) | 68 (5) | 0.80 | 0.57-1.12 | 0.201 |
| Gallo1675 | 352 (9) | 351 (9) | 1.00 | 0.85-1.17 | 0.968 | 233 (8) | 257 (9) | 1.12 | 0.93-1.35 | 0.250 | 119 (9) | 94 (7) | 0.76 | 0.57-1.02 | 0.067 |
| Gallo2018 | 287 (7) | 264 (7) | 0.91 | 0.76-1.09 | 0.298 | 226 (8) | 212 (8) | 0.93 | 0.76-1.14 | 0.476 | 52 (4) | 61 (5) | 0.84 | 0.57-1.24 | 0.371 |
| Gallo2178 | 163 (4) | 197 (5) | 1.23 | 0.99-1.52 | 0.063 | 117 (4) | 160 (6) | 1.40 | 1.09-1.79 | 0.008 | 36 (4) | 37 (3) | 0.79 | 0.50-1.24 | 0.300 |
| Gallo2179 | 209 (5) | 213 (5) | 1.02 | 0.84-1.24 | 0.840 | 149 (5) | 158 (6) | 1.07 | 0.84-1.35 | 0.593 | 60 (5) | 55 (4) | 0.91 | 0.63-1.33 | 0.632 |
Conditional logistic regression model, controls are matched to cases by age, sex and race/ethnicity; Significant associations are marked in bold font (p-value <0.05); Pos = antibody positive
Prevalence of antibody positivity to Gallo2178 differed by study among controls. We examined whether this influenced the association with CRC risk. This analysis was performed separately by time between blood draw and diagnosis to exclude the possibility that observed differences resulted from different follow-up times among studies. Within a follow-up time of less than 10 years the majority of studies found an OR above 1 for the association between Gallo2178 and CRC risk; however, this did not reach statistical significance in any of the individual cohorts alone. The association was strongest in CLUE with 0% sero-positive controls compared to 5% of cases (OR not calculable). In two of the ten cohorts, there was no association suggested between Gallo2178 and CRC risk: NYUWHS (OR: 1.00, 95% CI: 0.25-4.00) and SCCS (OR: 0.89, 95% CI: 0.34-2.30) (Figure 1A). When the time between blood draw and diagnosis was ≥10 years, a result of a statistically insignificant OR of 1 or below was found for 5 out of 8 studies with a sample size of at least 20 cases. In contrast, 3 (PLCO, MEC, and HPFS) of the 8 studies were suggestive of an OR of above 1, although with very wide confidence intervals (Figure 1B).
Figure 1: Forest plot of antibody positivity to Gallo2178 and CRC risk by study, within 10 years of blood draw (A) and with diagnosis after more than 10 years of blood draw.
Conditional logistic regression models were applied to determine OR (diamonds) and 95% CI (horizontal lines), controls are matched to cases by study, age, sex and race/ethnicity. A) No OR is given for CLUE since the denominator was 0. B) no values are given for CPSII and SCCS since total case numbers are below 20. Weight shows the contribution of each study to the overall number of participants in %. The vertical line at an OR of 1 serves as a reference for null association. Pos = antibody positive.
Previous studies have shown that auto-antibodies to p53 are a marker for presence of CRC (38). In this CRC cohort consortium, we found antibody positivity to p53 was statistically significantly associated with CRC risk among those individuals diagnosed within 10 years of their blood draw (OR: 1.53, 95% CI: 1.23-1.89) but not diagnosed after 10 years from blood draw (OR: 0.87, 95% CI: 0.50-1.33). Assuming that p53 auto-antibodies serve as a surrogate for presence of undiagnosed colorectal lesions present at baseline blood collection, we performed a case-control analysis of antibody responses to Gallo2178 with CRC stratified by p53 auto-antibody positivity. In line with results presented above by time between blood draw and diagnosis, association of antibody responses to Gallo2178 with CRC risk was stronger among p53-auto-antibody-positive individuals in the overall study (OR: 2.74, 95% CI: 1.09-6.87) than in p53-auto-antibody-negative individuals (OR: 1.15; 95% CI: 0.92-1.43, p(interaction)=0.073) (Table 4). The difference between p53-auto-antibody positives and negatives in the association of antibodies to Gallo2178 with CRC risk was smaller within 10 years after blood draw. Here, there was a statistically significant association of antibody responses to Gallo2178 with CRC risk observable also among p53-auto-antibody negatives (OR: 1.33, 95% CI: 1.02-1.72). In contrast, the association among p53-auto-antibody positives was slightly weaker than in the overall cohort and not statistically significant (OR: 2.37, 95% CI: 0.92-6.09, p (interaction)=0.315). Among those cases who were diagnosed ≥10 years after blood draw, the prevalence of Gallo2178 antibodies was higher among cases than controls only among p53-auto-antibody positives (5% positive cases compared to 0% positive controls, OR not calculable) but not among p53-auto-antibody negatives (OR: 0.74, 95% CI: 0.47-1.16; p(interaction) not calculable) (Table 4).
Table 4:
Antibody positivity to Gallo2178 and CRC risk by p53 auto-antibody positivity and time between blood draw and diagnosis.
| N | N (%) Gallo2178 pos | ||||||
|---|---|---|---|---|---|---|---|
| Follow-up time | p53 auto-antibody | controls/cases | Controls | Cases | OR1 | 95% CI | P interaction1 |
| Overall | Neg | 3865/3797 | 156 (4) | 174 (5) | 1.15 | 0.92-1.43 | |
| Pos | 198/266 | 7 (4) | 23 (9) | 2.74 | 1.09-6.87 | 0.073 | |
| <10 years | Neg | 2614/2540 | 110 (4) | 139 (5) | 1.33 | 1.02-1.72 | |
| Pos | 152/226 | 7 (5) | 21 (9) | 2.37 | 0.92-6.09 | 0.315 | |
| ≥10 years | Neg | 1250/1256 | 46 (4) | 35 (3) | 0.74 | 0.47-1.16 | |
| Pos | 46/40 | 0 (0) | 2 (5) | - | - | - | |
unconditional logistic regression model with adjustment for study, sex, race and age; Significant associations (p-value<0.05) are marked in bold font; Neg = negative; Pos = positive
Finally, we addressed potential differences in the association of Gallo2178 and CRC risk by characteristics of CRC and did not identify a substantial difference in the association by stage, site, or age at diagnosis (Supplementary table S3).
Discussion
In this CRC cohort consortium, we found that antibody responses to SGG protein Gallo2178 were statistically significantly associated with a 40% increase in CRC risk among individuals diagnosed within 10 years of their blood draw, and that there was no association with antibody responses to SGG for individuals diagnosed 10 or more years after blood draw. Further, the association of antibody responses to Gallo2178 with CRC risk was more pronounced among p53-auto-antibody-positive cases, a surrogate for the presence of (pre-)cancerous lesions at baseline. These data support the hypothesis that SGG infection of gut epithelial tissue after an initial precursor lesion has formed may act as a cancer promotor increasing CRC risk once tumorigenesis has already begun. Because this was a serological study, however, it could not be assessed whether SGG might have colonized the gut lumen before an antibody response in serum was detectable and whether this colonization might have had increased the risk of developing CRC already. Another interpretation of our findings would be simply that individuals with (pre-)cancerous lesions are more likely to harbor SGG antibodies than individuals farther away from cancer development.
Gallo2178 is expressed from the pil1 operon together with proteins Gallo2177 and Gallo2179. These proteins build a SGG pilus that was shown to be important for bacterial adhesion to collagen as well as for biofilm formation, which consequently relates to SGG’s virulence in infective endocarditis and infection of colorectal tissue (14, 15). Gallo2177 is a sortase that assembles Gallo2178 and Gallo2179 into a pilus structure (15). Gallo2179 was shown to have collagen-binding abilities, preferentially collagen type I, present for example in damaged heart valves, followed by collagen type IV, found in the basal lamina of epithelial tissues. This collagen-binding domain is similar to collagen-binding proteins of other bacteria, e.g. Staphylococcus aureus. Gallo2178 has been shown to be the major pilin, the backbone structure of the pilus (14, 15). Why only antibody responses to Gallo2178 but not to other SGG proteins, including Gallo2179, were significantly associated reproducibly in this and previous studies (11, 12, 16) remains unclear. Possible factors affecting this involve host immune responses recognizing antigenic epitopes differently between individuals as well as bacterial strains expressing different sets of proteins. Supporting the latter, the prevalence of antibody positivity to Gallo2178 among controls was lower than to any of the other analyzed SGG proteins. The underlying reason could be a higher specificity of this protein due to lower similarities to proteins of other bacteria and consequently less cross-reactivity. For example, there is no homologue of Gallo2178 encoded in Staphylococcus aureus, which is not true for Gallo2179 (39). This could make Gallo2178 a more specific marker than the other proteins included in SGG multiplex serology leading to a stronger association with CRC risk.
In the few prior research studies, antibody responses to Gallo2178 have been consistently associated with CRC. A small study by Boleij et al. analyzed early stage CRC (n=44) and asymptomatic controls (n=47) for antibody responses to Gallo2178 and found 9% positive cases at a pre-defined specificity of 100% (6). We previously reported antibody responses to Gallo2178 associated with CRC with up to 4-fold increased odds in two independent case-control studies (11, 12). The new data from the prospective analysis presented here are in line with an independent prospective study from Europe: antibody responses to Gallo2178 were associated with a 3-fold increased risk for CRC in blood samples taken up to 8 years before diagnosis (16). In both prospective studies, antibody responses to Gallo2178 were found in only 6% of CRC cases, making it a rare exposure. PCR data from Lopes et al. support this as they found SGG DNA in only 5 rectal swab specimens (11%) out of 54 individuals undergoing colonoscopy (40).
The novelty of this current study, however, is that we were able to analyze blood samples with a longer follow-up time than in the study described above (16), i.e. taken more than 8 years before diagnosis. It is estimated that progression from a colorectal polyp to CRC lasts 10-15 years (17). Antibody responses to Gallo2178 were not associated with increased CRC risk when the blood was sampled 10 or more years prior to diagnosis. These results were in line with analysis stratified by antibody positivity to tumor suppressor p53, a putative marker for prevalent undiagnosed CRC and adenoma (38): CRC risk was increased almost 3-fold with antibody responses to Gallo2178 only among p53 antibody-positive individuals, i.e. those with a suspected prevalent lesion in the gut. When regarding this interaction separately by follow-up time, antibody responses to Gallo2178 were associated with CRC risk also in the p53 auto-antibody negatives within 10 years of diagnosis but not when follow-up was equal to or more than 10 years. Individuals diagnosed within 10 years of their blood draw were much more likely to have already had a precursor lesion at the time of blood draw than those individuals with a longer time span between blood sampling and diagnosis as supported by the p53-auto-antibody finding. As proposed by Tjalsma et al., SGG could be considered a so-called passenger bacterium, a pathogen-turned-commensal that is able to infect the epithelium upon decrease of the gut epithelial integrity after tumor formation, which then might or might not act as a carcinogenesis-promoting agent (41). This hypothesis is supported by several mechanistic studies including a study by Boleij et al., where it was shown that SGG growth is stimulated in vitro under metabolic conditions of the CRC microenvironment (19). Aymeric et al. reproduced this finding and showed that SGG colonization in the colon is promoted by CRC-specific conditions: increased secondary bile-acids in the oncogenic context were shown to induce bacteriocin synthesis in SGG, which led to killing of other bacterial species creating a new niche for SGG colonization in the gut (18). Infection of the gut epithelial tissue by SGG was shown to be enabled by the presence of collagen-rich surfaces as present in (pre-)cancerous lesions with diminished epithelial integrity but not by adhesion to or internalization by normal epithelial cells themselves (14).
A promoting effect of SGG infection on carcinogenesis and thus on the formation of malignant cancer out of a precursor lesion cannot directly be inferred from the data presented here: we showed only that antibody responses to SGG within 10 years of diagnosis were associated with an increased risk of developing CRC. Thus, these antibodies were more frequent in individuals later diagnosed with malignant disease. It is instructive that a cancer-promoting effect of SGG was suggested by a mechanistic study by Kumar et al., who showed that SGG treatment increased proliferation of CRC cell lines but also tumor burden in an azoxymethane-induced mouse model of CRC (10). Further studies are needed to confirm these findings and identify the underlying mechanisms.
This study does have several limitations. First, assessment of SGG infection by serological analysis provided data on a systemic marker for past, acute or chronic SGG infection. Thus, this method is not able to determine whether there exists a local acute or persistent infection at the site of interest. However, the advantage is that a blood draw is less invasive and easier to conduct on a large scale than a biopsy in the gut, especially in asymptomatic individuals. Unfortunately, though, the natural history of antibody responses to SGG proteins is unknown, including at what exact event antibody responses first occur (e.g., colonization of the epithelium, entry into blood stream) and how stable these antibody responses are. Serial samples from the same individual over time could give insight into these questions. Furthermore, the cut-off for sero-positivity to SGG was set arbitrarily due to a lack of reference samples because no serological gold standard for SGG diagnostics are available. The reproducibility of the association of antibody responses to Gallo2178 with CRC risk in independent studies (11, 12, 16), however, supports the strength of the findings presented here. This reproducibility also argues against a potential mis-interpretation of the results as being just a chance finding. A Bonferroni-correction for multiple testing in this study would have required a p-value of 0.006 for significance with 9 SGG antigens analyzed. The p-value for the association of antibody responses to Gallo2178 with CRC risk within 10 years after blood draw was 0.008 and thus would not have been regarded as a statistically significant finding outside the already published context (11, 12, 16). A further limitation of the study was missing information in variables of potentially high interest, such as family history of CRC and BMI, which might have led to residual confounding. However, it should be emphasized that the large sample size of this consortium offered the possibility to comprehensively assess the association of this rare exposure with CRC risk. Finally, a more detailed analysis by follow-up time beyond 10 years might have provided a more elaborate view on the temporality of the association of SGG with CRC risk, however, we were not powered for such further stratified analyses due to the rarity of the exposure.
In conclusion, we reproduced the finding that antibody responses to SGG Gallo2178 were associated with a 1.4-fold higher risk for CRC development among individuals who were diagnosed within 10 years of blood draw. This temporality of the association together with an interaction with auto-antibodies to p53, a putative marker for undiagnosed (pre-)cancerous lesions at baseline, supports the hypothesis that SGG infection of gut epithelial tissue in individuals for whom tumorigenesis has already begun may promote carcinogenesis.
Supplementary Material
Acknowledgements
The National Cancer Institute funds: this consortium (R01 CA190428, PI: Epplein); the Southern Community Cohort Study (U01 CA202979, PI: Blot); the NYU Women’s Health Study; the NHS/HPFS (U01 CA167552; UM1 CA186107; P01 CA087969; UM1 CA167552); the PHS (R01 CA097193; R01 CA040360; R01 HL034595); and the MEC (U01 CA164973, PI: Loic Le Marchand).
R.M. Peek is supported by R01 DK058587 and R01 CA077955.
T.L. Cover is supported by NIH R01 AI039657, R01 AI118932, P01 CA116087 and the Department of Veterans Affairs BX000627.
The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services through contracts, HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C.
The development of H. pylori multiplex serology was funded in part by the Joint Initiative for Innovation and Research of the German Helmholtz Association.
CLUE thank the participants and staff for their contributions, as well as the Maryland Cancer Registry, Center for Cancer Surveillance and Control, Department of Health and Mental Hygiene, 201 W. Preston Street, Room 400, Baltimore, MD 21201, http://phpa.dhmh.maryland.gov/cancer, 410-767-4055.
Footnotes
Conflict of interest statement:
The authors declare no potential conflicts of interest.
References
- 1.Boleij A, van Gelder MM, Swinkels DW, Tjalsma H. Clinical Importance of Streptococcus gallolyticus infection among colorectal cancer patients: systematic review and meta-analysis. Clin Infect Dis 2011;53:870–8. [DOI] [PubMed] [Google Scholar]
- 2.Klein RS, Recco RA, Catalano MT, Edberg SC, Casey JI, Steigbigel NH. Association of Streptococcus bovis with carcinoma of the colon. N Engl J Med 1977;297:800–2. [DOI] [PubMed] [Google Scholar]
- 3.Ruoff KL, Miller SI, Garner CV, Ferraro MJ, Calderwood SB. Bacteremia with Streptococcus bovis and Streptococcus salivarius: clinical correlates of more accurate identification of isolates. Journal of Clinical Microbiology. 1989;27:305–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Abdulamir AS, Hafidh RR, Mahdi LK, Al-jeboori T, Abubaker F. Investigation into the controversial association of Streptococcus gallolyticus with colorectal cancer and adenoma. BMC Cancer. 2009;9:403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Abdulamir AS, Hafidh RR, Bakar FA. Molecular detection, quantification, and isolation of Streptococcus gallolyticus bacteria colonizing colorectal tumors: inflammation-driven potential of carcinogenesis via IL-1, COX-2, and IL-8. Mol Cancer. 2010;9:249. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Boleij A, Roelofs R, Danne C, Bellais S, Dramsi S, Kato I, et al. Selective antibody response to Streptococcus gallolyticus pilus proteins in colorectal cancer patients. Cancer Prev Res (Phila). 2012;5:260–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Boleij A, Roelofs R, Schaeps RM, Schulin T, Glaser P, Swinkels DW, et al. Increased exposure to bacterial antigen RpL7/L12 in early stage colorectal cancer patients. Cancer. 2010;116:4014–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Garza-Gonzalez E, Rios M, Bosques-Padilla FJ, Francois F, Cho I, Gonzalez GM, et al. Immune response against Streptococcus gallolyticus in patients with adenomatous polyps in colon. Int J Cancer. 2012;131:2294–9. [DOI] [PubMed] [Google Scholar]
- 9.Paritsky M, Pastukh N, Brodsky D, Isakovich N, Peretz A. Association of Streptococcus bovis presence in colonic content with advanced colonic lesion. World J Gastroenterol 2015;21:5663–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kumar R, Herold JL, Schady D, Davis J, Kopetz S, Martinez-Moczygemba M, et al. Streptococcus gallolyticus subsp. gallolyticus promotes colorectal tumor development. PLoS Pathog 2017;13:e1006440. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Butt J, Romero-Hernandez B, Perez-Gomez B, Willhauck-Fleckenstein M, Holzinger D, Martin V, et al. Association of Streptococcus gallolyticus subspecies gallolyticus with colorectal cancer: Serological evidence. Int J Cancer. 2016;138:1670–9. [DOI] [PubMed] [Google Scholar]
- 12.Butt J, Werner S, Willhauck-Fleckenstein M, Michel A, Waterboer T, Zornig I, et al. Serology of Streptococcus gallolyticus subspecies gallolyticus and its association with colorectal cancer and precursors. Int J Cancer. 2017;141:897–904. [DOI] [PubMed] [Google Scholar]
- 13.Waterboer T, Sehr P, Michael KM, Franceschi S, Nieland JD, Joos TO, et al. Multiplex human papillomavirus serology based on in situ-purified glutathione s-transferase fusion proteins. Clin Chem 2005;51:1845–53. [DOI] [PubMed] [Google Scholar]
- 14.Boleij A, Muytjens CM, Bukhari SI, Cayet N, Glaser P, Hermans PW, et al. Novel clues on the specific association of Streptococcus gallolyticus subsp gallolyticus with colorectal cancer. Journal of Infectious Diseases. 2011;203:1101–9. [DOI] [PubMed] [Google Scholar]
- 15.Danne C, Entenza JM, Mallet A, Briandet R, Debarbouille M, Nato F, et al. Molecular characterization of a Streptococcus gallolyticus genomic island encoding a pilus involved in endocarditis. Journal of Infectious Diseases. 2011;204:1960–70. [DOI] [PubMed] [Google Scholar]
- 16.Butt J, Jenab M, Willhauck-Fleckenstein M, Michel A, Pawlita M, Kyro C, et al. Prospective evaluation of antibody response to Streptococcus gallolyticus and risk of colorectal cancer. Int J Cancer. 2018. [DOI] [PubMed] [Google Scholar]
- 17.Morson B President’s address. The polyp-cancer sequence in the large bowel. Proc R Soc Med 1974;67:451–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Aymeric L, Donnadieu F, Mulet C, du Merle L, Nigro G, Saffarian A, et al. Colorectal cancer specific conditions promote Streptococcus gallolyticus gut colonization. Proc Natl Acad Sci U S A. 2018;115:E283–E91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Boleij A, Dutilh BE, Kortman GA, Roelofs R, Laarakkers CM, Engelke UF, et al. Bacterial responses to a simulated colon tumor microenvironment. Mol Cell Proteomics. 2012;11:851–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Frank SA. Dynamics of Cancer: Incidence, Inheritance, and Evolution. Princeton (NJ)2007. [PubMed] [Google Scholar]
- 21.Brosh R, Rotter V. When mutants gain new powers: news from the mutant p53 field. Nat Rev Cancer. 2009;9:701–13. [DOI] [PubMed] [Google Scholar]
- 22.Soussi T p53 Antibodies in the sera of patients with various types of cancer: a review. Cancer Res 2000;60:1777–88. [PubMed] [Google Scholar]
- 23.Teras LR, Gapstur SM, Maliniak ML, Jacobs EJ, Gansler T, Michel A, et al. Prediagnostic Antibodies to Serum p53 and Subsequent Colorectal Cancer. Cancer Epidemiol Biomarkers Prev 2017. [DOI] [PubMed] [Google Scholar]
- 24.Huang HY, Alberg AJ, Norkus EP, Hoffman SC, Comstock GW, Helzlsouer KJ. Prospective study of antioxidant micronutrients in the blood and the risk of developing prostate cancer. Am J Epidemiol 2003;157:335–44. [DOI] [PubMed] [Google Scholar]
- 25.Calle EE, Rodriguez C, Jacobs EJ, Almon ML, Chao A, McCullough ML, et al. The American Cancer Society Cancer Prevention Study II Nutrition Cohort: rationale, study design, and baseline characteristics. Cancer. 2002;94:500–11. [DOI] [PubMed] [Google Scholar]
- 26.Wei EK, Giovannucci E, Fuchs CS, Willett WC, Mantzoros CS. Low plasma adiponectin levels and risk of colorectal cancer in men: a prospective study. J Natl Cancer Inst 2005;97:1688–94. [DOI] [PubMed] [Google Scholar]
- 27.Kolonel LN, Henderson BE, Hankin JH, Nomura AM, Wilkens LR, Pike MC, et al. A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol 2000;151:346–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Colditz GA, Hankinson SE. The Nurses’ Health Study: lifestyle and health among women. Nat Rev Cancer. 2005;5:388–96. [DOI] [PubMed] [Google Scholar]
- 29.Toniolo PG, Levitz M, Zeleniuch-Jacquotte A, Banerjee S, Koenig KL, Shore RE, et al. A prospective study of endogenous estrogens and breast cancer in postmenopausal women. J Natl Cancer Inst 1995;87:190–7. [DOI] [PubMed] [Google Scholar]
- 30.Lee JE, Wei EK, Fuchs CS, Hunter DJ, Lee IM, Selhub J, et al. Plasma folate, methylenetetrahydrofolate reductase (MTHFR), and colorectal cancer risk in three large nested case-control studies. Cancer Causes Control. 2012;23:537–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Hayes RB, Reding D, Kopp W, Subar AF, Bhat N, Rothman N, et al. Etiologic and early marker studies in the prostate, lung, colorectal and ovarian (PLCO) cancer screening trial. Control Clin Trials. 2000;21:349S–55S. [DOI] [PubMed] [Google Scholar]
- 32.Signorello LB, Hargreaves MK, Steinwandel MD, Zheng W, Cai Q, Schlundt DG, et al. Southern community cohort study: establishing a cohort to investigate health disparities. J Natl Med Assoc 2005;97:972–9. [PMC free article] [PubMed] [Google Scholar]
- 33.Design of the Women’s Health Initiative clinical trial and observational study. The Women’s Health Initiative Study Group. Control Clin Trials. 1998;19:61–109. [DOI] [PubMed] [Google Scholar]
- 34.Reuschenbach M, Waterboer T, Wallin KL, Einenkel J, Dillner J, Hamsikova E, et al. Characterization of humoral immune responses against p16, p53, HPV16 E6 and HPV16 E7 in patients with HPV-associated cancers. Int J Cancer. 2008;123:2626–31. [DOI] [PubMed] [Google Scholar]
- 35.Rollison DE, Giuliano AR, Messina JL, Fenske NA, Cherpelis BS, Sondak VK, et al. Case-control study of Merkel cell polyomavirus infection and cutaneous squamous cell carcinoma. Cancer Epidemiol Biomarkers Prev 2012;21:74–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Michael KM, Waterboer T, Sehr P, Rother A, Reidel U, Boeing H, et al. Seroprevalence of 34 human papillomavirus types in the German general population. PLoS Pathog 2008;4:e1000091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Migchelsen SJ, Martin DL, Southisombath K, Turyaguma P, Heggen A, Rubangakene PP, et al. Defining Seropositivity Thresholds for Use in Trachoma Elimination Studies. PLoS Negl Trop Dis 2017;11:e0005230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Suppiah A, Greenman J. Clinical utility of anti-p53 auto-antibody: systematic review and focus on colorectal cancer. World J Gastroenterol 2013;19:4651–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol 1990;215:403–10. [DOI] [PubMed] [Google Scholar]
- 40.Lopes PG, Cantarelli VV, Agnes G, Costabeber AM, d’Azevedo PA. Novel real-time PCR assays using TaqMan minor groove binder probes for identification of fecal carriage of Streptococcus bovis/Streptococcus equinus complex from rectal swab specimens. Journal of Clinical Microbiology. 2014;52:974–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Tjalsma H, Boleij A, Marchesi JR, Dutilh BE. A bacterial driver-passenger model for colorectal cancer: beyond the usual suspects. Nat Rev Microbiol 2012;10:575–82. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.