Data from three large prospective cohort studies and a meta-analysis did not support that oral health plays a major role in the development of colorectal cancer.
Keywords: oral health, tooth loss, tooth decay, periodontal disease, colorectal cancer risk
Abstract
Background
While studies have shown that poor oral health status may increase the risk of cancer, evidence of a specific association with the risk of colorectal cancer (CRC) is inconclusive. We evaluated the association between oral health and CRC risk using data from three large cohorts: the Shanghai Men's Health Study (SMHS), the Shanghai Women's Health Study (SWHS), and the Southern Community Cohort Study (SCCS), and carried out a meta-analysis of results from other relevant published studies.
Patients and methods
This study applied a nested case–control study design and included 825 cases/3298 controls from the SMHS/SWHS and 238 cases/2258 controls from the SCCS. The association between oral health status (i.e. tooth loss/tooth decay) and CRC risk was assessed using conditional logistic regression models. A meta-analysis was carried out based on results from the present study and three published studies.
Results
We found that tooth loss was not associated with increased risk of CRC. ORs and respective 95% CIs associated with loss of 1–5, 6–10, and >10 teeth compared with those with full teeth are 0.87 (0.69–1.10), 0.93 (0.70–1.24), and 0.85 (0.66–1.11) among SMHS/SWHS participants; and 1.13 (0.72–1.79), 0.87 (0.52–1.43), and 1.00 (0.63–1.58) for those with loss of 1–4, 5–10, and >10 teeth among SCCS participants. Data regarding tooth decay were available in the SCCS, but were not associated with CRC risk. Meta-analysis confirmed the null association between tooth loss/periodontal disease and CRC risk (OR 1.05, 95% CI 0.86–1.29).
Conclusion
In this analysis of three cohorts and a meta-analysis, we found no evidence supporting an association between oral health and CRC risk.
introduction
Colorectal cancer (CRC) is the third most common cancer in both men and women worldwide, with an estimated 132 700 and 376 300 new cases, respectively, occurring in the United States in 2015 [1] and in China each year [2]. Inflammatory bowel disease is one established risk factor for CRC [3]. Several other known or suspected risk factors for CRC, such as smoking, dietary fat intake, obesity, and physical inactivity, are known to be associated with systemic inflammation [4], suggesting that inflammation plays a major role in CRC development. Periodontal disease caused by bacterial infections may increase the risk of cancer by the prolonged release of inflammatory mediators [5] and by an increase in the generating of carcinogens (e.g. nitrosamines) [6]. In vitro and animal studies have shown that poor oral health may increase the risk of various cancers [7, 8], including CRC [9]. Multiple tooth loss may be an indicator of periodontal disease, and the number of missing teeth can be viewed as an index of lifetime accumulation of poor oral health, particularly among individuals from a low socioeconomic background (e.g. the vast majority of our US study participants) or populations where preventive dental care is not a standard practice (e.g. the participants of our Chinese study).
Recently, it has been reported that Fusobacterium nucleatum, one of the predominant subgingival microbial species found in chronic periodontitis, presented in overabundance in colorectal carcinoma tissues [9]. Studies have shown that Fusobacterium adhesion A of the F. nucleatum stimulates human CRC cell growth. Fusobacterium nucleatum was also shown to induce tumor multiplicity and selectively recruit tumor-infiltrating myeloid cells to promote tumorigenesis in ApcMin/+ mice [8]. These results suggest a possible direct role of the poor oral health-related microbiome in the pathophysiology of CRC.
Another potential mechanism connecting oral health and CRC is the correlation between the oral and gut microbiomes. Several studies [10–12] and the Human Microbiome Project have shown the overlap between oral and intestinal microbiomes, and that the oral and gut microbiome community types can predict each other. These results suggest that oral health, via its association with the oral microbiome, may serve as a surrogate measurement of gut microbes and be indirectly associated with CRC risk.
Few human studies have evaluated oral health for its relationship to CRC risk, and results from them have been mixed [13–16]. To systematically and comprehensively evaluate the association between oral health and the risk of CRC, we carried out a case–control study nested in three large cohort studies and carried out a meta-analysis to summarize results from the literature and the current study.
methods
Our nested case–control study used resources from a US cohort (the Southern Community Cohort Study, SCCS) and Chinese cohort studies (the Shanghai Men's Health Study, SMHS, and the Shanghai Women's Health Study, SWHS). Study design and methodologies of these cohort studies have been described previously [17–19]. Briefly, the SCCS recruited ∼86 000 participants between 2002 and 2009 from 12 southeastern states of the United States. Approximately 86% of them were recruited from community health centers (CHCs), institutions providing basic health care and preventative services in underserved areas, resulting in a cohort that includes a large number of individuals of low income and educational status. The remaining 14% of the cohort was recruited through mail-based general population sampling. The SMHS and SWHS recruited 61 480 men and 74 741 women from nine communities in Shanghai, China from 2001 to 2006 and 1996 to 2000, respectively. Participants in the SCCS, SWHS, and SMHS provided written informed consent, and the institutional review boards of all participating institutions approved the study protocols.
Only incident CRC cases occurring after study enrollment were included in the present study. In the SWHS and SMHS, CRC cases were matched with controls in a ratio of 1 : 4 based on age (±2 years) and enrollment in the same interview calendar years. In the SMHS, there was one case with only two controls. In the SCCS, 10 controls were matched with each CRC case based on age (±5 years), recruitment method (CHC or general population), sex, race, recruitment site (site and/or state of enrollment), and smoking status (never, former, and current). In the SCCS, a complete set of 10 controls could not be found for two cases, which resulted in one case having only seven matched controls and the other case, eight matched controls. In total, 309 incident CRC cases and 3085 controls from the SCCS, and 878 cases and 3510 controls from the SMHS/SWHS were selected for the present study. We excluded participants from the study who had no tooth loss data (SCCS: 17 cases and 130 controls; and 157 controls matched with the 17 cases; SWHS/SMHS: none), or had CRC diagnosed within 2 years after enrollment (to minimize the influence of lifestyle or oral health changes related to undiagnosed CRC) and their respective controls (SMHS/SWHS: cases = 53, controls = 212; SCCS: cases = 54, controls = 540). The final study dataset included 825 CRC cases and 3298 controls from the SMHS/SWHS and 238 cases and 2258 controls from the SCCS.
In the SMHS/SWHS, participants were asked about the number of teeth they had lost, using the following categories: ‘none’, ‘1–5 teeth’, ‘6–10 teeth’, or ‘>10 teeth’. In the SCCS, participants were asked the following questions related to oral health: (i) about how many adult teeth have you lost in your lifetime due to tooth decay or gum disease, categorized into ‘none’, ‘1–4’, ‘5–10’, ‘>10 but not all’, and ‘all of them’ and (ii) how many decayed teeth or cavities do you currently have that have not been treated, categorized into ‘none’, ‘1–2’, ‘3–5’, ‘>5’, and ‘no teeth’.
In the SWHS and SMHS, new CRC cases were identified via linkage with the Shanghai Cancer registry. All possible matches were manually checked and were verified through home visits. Medical charts from diagnostic hospitals were reviewed to obtain information on the date and pathologic diagnosis of respective cancer cases [18, 19]. In the SCCS, incident cases of CRC were identified through linkage with 12 state cancer registries and from National Death Index mortality records [17]. CRC and its subsites were defined according to ICD-9 codes (153–154) [20] in the SMHS/SWHS and ICD-10 codes (C18-C21) in the SCCS [21].
In-person interview or mailed surveys (only for non-CHC SCCS) were conducted to collect information on demographics, dietary intake, physical activity, weight, height, smoking and drinking habits, family history of cancer, and other exposures. We estimated body mass index (BMI) as weight in kilograms divided by height in meters squared. Physical activity levels were estimated by multiplying the energy expenditure in metabolic equivalent tasks (METs) measured in hours per week of each activity by hours spent on the activity and summing the values of all activities in METS-hour/week (MET-h/w) [22].
statistical analysis
Cases and control differences in baseline sociodemographic characteristics and suggested risk factors were examined using a χ2 test for categorical variables and t-test for continuous variables. We used a conditional logistic regression model to determine the association between oral health status (i.e. tooth loss/tooth decay—SCCS and tooth loss—SMHS/SWHS) and the risk of CRC in the three cohorts. The tooth loss variable was grouped into four categories: none, 1–5, 6–10, and >10 in the SMHS/SWHS and 1–4, 5–10, and >10 in the SCCS. Tooth decay information was available in the SCCS only and grouped into five categories as follows: none, 1–2, 3–5, >5, and dentures only. Different covariates were included in the logistic regression analyses for the SCCS and SMHS/SWHS data to account for population-specific confounders. In the SCCS, occupation, income, education, and fruit and sweet beverage consumption were included in our conditional logistic regression models. In the SMHS/SWHS, income, education, BMI, exercise, smoking (never, former, and current), red meat consumption, and fruit intake were adjusted. We further carried out stratified analysis by cancer type (colon and rectum), adjusted for similar covariates. All statistical analyses were conducted using SAS, version 9.4 (SAS Institute, Inc.). All tests were two-sided, and P < 0.05 was considered statistically significant.
meta-analysis
PubMed, Embase, and Web of Science repositories were searched systematically to identify all available cohort and case–control studies that examined the associations between oral health and CRC using the following key words: periodontal diseases OR dental caries OR tooth decay OR tooth loss, AND colorectal neoplasms OR CRC OR colon cancer OR rectal cancer OR colonic neoplasms. The search was completed on 1 May 2015. We limited our search to population-based studies and those in the English language. Relevant study information was extracted from the publications.
Combined OR and 95% CI were used to measure the impact of tooth loss and periodontal disease on the risk of CRC. The heterogeneity across studies was evaluated by the Q test and I2 statistics. The meta-analysis, applying the random-effect model, was carried out using STATA 12.0 software (StataCorp, College Station, TX).
results
characteristics of study participants
Sociodemographic characteristics of the study participants are presented in Table 1. In the SCCS, controls were more likely than cases to be in low or lower-middle income groups (71.5% versus 67.7%, P = 0.02). Participants with CRC in the SWHS/SMHS were more likely than controls to have higher BMI and caloric intake (P < 0.0001 and P = 0.03, separately).
Table 1.
Sociodemographic characteristics of study participants
| Characteristic | SMHS/SWHS |
SCCS |
||||
|---|---|---|---|---|---|---|
| Cases | Controls | P-value | Cases | Controls | P-value | |
| Number (%) | 825 (20.0) | 3298 (80.0) | 238 (9.5) | 2258 (90.5) | ||
| Age (years)a | 59.2 ± 9.2 | 59.5 ± 9.1 | 56.8 ± 8.8 | 56.5 ± 8.6 | 0.54 | |
| 40–49 | 152 (18.4) | 684 (20.7) | 0.53 | 53 (22.3) | 525 (23.3) | 0.92 |
| 50–59 | 239 (29.0) | 939 (28.5) | 93 (39.1) | 906 (40.1) | ||
| 60–69 | 332 (40.2) | 1282 (38.9) | 69 (29.0) | 632 (28.0) | ||
| 70–79 | 102 (12.4) | 393 (11.9) | 23 (9.6) | 195 (8.6) | ||
| Educational level | ||||||
| <High school | 452 (55.4) | 1687 (51.7) | 0.26 | 73 (31.7) | 622 (27.5) | 0.77 |
| High school/vocational school | 212 (26.0) | 886 (27.2) | 79 (33.2) | 802 (35.5) | ||
| Some/completed college | 152 (18.6) | 689 (21.1) | 67 (28.1) | 654 (29.0) | ||
| >College | 19 (8.0) | 180 (8.0) | ||||
| Incomeb | ||||||
| Low | 118 (14.3) | 440 (13.4) | 0.03 | 128 (54.5) | 1126 (50.9) | 0.02 |
| Lower middle | 31 (13.2) | 457 (20.6) | ||||
| Middle | 653 (79.1) | 2546 (77.2) | 45 (19.1) | 310 (14.0) | ||
| Upper middle | 26 (11.1) | 239 (10.8) | ||||
| High | 54 (6.6) | 310 (9.4) | 5 (2.1) | 82 (3.7) | ||
| Smoking statusc | ||||||
| Never | 524 (63.5) | 2201 (66.7) | 0.21 | 86 (36.2) | 814 (36.1) | 0.99 |
| Former | 70 (8.5) | 263 (8.0) | 76 (31.9) | 719 (31.8) | ||
| Current | 231 (28.0) | 833 (25.3) | 76 (31.9) | 725 (32.1) | ||
| BMI (mean ± SD) | 24.5 ± 3.5 | 24.1 ± 3.2 | <0.0001 | 30.8 ± 6.9 | 30.7 ± 7.3 | 0.94 |
| Exercise MET (mean ± SD) | 12.2 ± 6.1 | 12.3 ± 6.7 | 0.82 | 20.7 ± 16.5 | 20.1 ± 17.0 | 0.59 |
| Caloric intake (mean ± SD) | 1798 ± 441 | 1759 ± 442 | 0.03 | 2266 ± 1319 | 2280 ± 1188 | 0.86 |
| Saturate fat intake (mean ± SD) | 9.2 ± 4.8 | 9.1 ± 4.7 | 0.94 | 26.1 ± 17.9 | 26.5 ± 16.1 | 0.67 |
| Total fiber intake (mean ± SD) | 11.1 ± 4.0 | 11.1 ± 4.2 | 0.93 | 20.7 ± 13.6 | 20.5 ± 11.4 | 0.77 |
| Family history of any cancer | 256 (31.0) | 941 (28.5) | 0.16 | 130 (58.0) | 1168 (54.6) | 0.33 |
| Family history of CRC | 30 (3.6) | 85 (2.6) | 0.10 | 24 (11.2) | 168 (8.1) | 0.12 |
aIncome levels were as follows: SCCS: low (<$15 000/year per household), lower-middle ($15 000–$24 999/year per household), middle ($25 000–$49 999/year per household), upper-middle ($50 000–$99 999/year per household), and high (≥$100 000/year per household). SMHS: low (<6000 yuan/year per capita), middle (6000–10 000 yuan/year per capita), and high (>10 000 yuan/year per capita). SWHS: low (<4000 yuan/year per capita), middle (4000–8000 yuan/year per capita), and high (>8000 yuan/year per capita).
bIn both SCCS and SWHS/SMHS, age, sex, and race were matching variables.
cIn SCCS, smoking status was matched between cases and controls.
CRC, colorectal cancer; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); MET, metabolic equivalent of task; SD, standard deviation; SCCS, Southern Community Cohort Study; SMHS, Shanghai Men's Health Study; SWHS, Shanghai Women's Health Study.
association of oral health with CRC
Tooth loss was not associated with increased risk of CRC in either the US or Chinese studies [ORs and respective 95% CIs of CRC for 1–5, 6–10, and >10 lost teeth in comparison with none: 0.88 (0.70–1.11), 0.93 (0.69–1.23), and 0.81 (0.63–1.06) in the SMHS/SWHS and 1.13 (0.72–1.79), 0.87 (0.52–1.43), and 1.00 (0.63–1.58) for those with loss of 1–4, 5–10, and >10 teeth in the SCCS; respectively). The ORs and respective 95% CIs between tooth decay and the risk of CRC in the SCCS were 1.21 (0.83–1.77), 1.28 (0.81–2.02), 0.78 (0.41–1.49), and 1.30 (0.87–1.93) for 1–2, 3–5, >5 lost teeth, and dentures only, respectively, in comparison with no lost teeth (Table 2). Further analyses, carried out by assigning the categorical variables to an ordinal value and treating them as continuous variables in the regression model, showed the same null association.
Table 2.
Association between oral health and colorectal cancer risk
| SWHS/SMHS | Cases | Controls | OR (95% CI)a | OR (95% CI)b |
|---|---|---|---|---|
| Teeth lost | ||||
| Overall CRC risk | (n = 825) | (n = 3298) | ||
| None | 146 | 545 | Ref. | Ref. |
| 1–5 | 304 | 1269 | 0.89 (0.71–1.12) | 0.88 (0.70–1.11) |
| 6–10 | 123 | 467 | 0.98 (0.74–1.29) | 0.92 (0.69–1.23) |
| >10 | 252 | 1017 | 0.92 (0.71–1.18) | 0.81 (0.63–1.06) |
| Per category increment | 0.99 (0.91–1.07) P = 0.79 |
0.95 (0.87–1.03) P = 0.19 |
||
| Colon cancer risk | (n = 501) | (n = 2002) | ||
| None | 80 | 293 | Ref. | Ref. |
| 1–5 | 184 | 771 | 0.87 (0.64–1.18) | 0.88 (0.64–1.20) |
| 6–10 | 79 | 296 | 0.97 (0.67–1.40) | 0.95 (0.65–1.38) |
| >10 | 158 | 642 | 0.89 (0.64–1.24) | 0.83 (0.58–1.18) |
| Per category increment | 0.99 (0.89–1.09) P = 0.77 |
0.96 (0.86–1.06) P = 0.41 |
||
| Rectum cancer risk | (n = 324) | (n = 1296) | ||
| None | 66 | 252 | Ref. | Ref. |
| 1–5 | 120 | 498 | 0.92 (0.65–1.29) | 0.89 (0.63–1.27) |
| 6–10 | 44 | 171 | 0.98 (0.63–1.53) | 0.85 (0.54–1.36) |
| >10 | 94 | 375 | 0.96 (0.64–1.42) | 0.80 (0.53–1.21) |
| Per category increment | 1.00 (0.88–1.13) P = 0.95 |
0.93 (0.81–1.06) P = 0.29 |
||
| SCCS | OR (95% CI)a | OR (95% CI)c | ||
| Teeth lost | ||||
| Overall CRC risk | (n = 238) | (n = 2258) | ||
| None | 31 | 308 | Ref. | Ref. |
| 1–4 | 67 | 570 | 1.15 (0.74–1.81) | 1.13 (0.72–1.79) |
| 5–10 | 44 | 493 | 0.88 (0.54–1.44) | 0.87 (0.52–1.43) |
| >10 | 96 | 887 | 1.05 (0.67–1.63) | 1.00 (0.63–1.58) |
| Per category increment | 0.99 (0.87–1.13) P = 0.85 |
0.97 (0.85–1.16) P = 0.69 |
||
| Colon cancer risk | (n = 172) | (n = 1626) | ||
| None | 20 | 217 | Ref. | Ref. |
| 1–4 | 53 | 406 | 1.40 (0.81–2.40) | 1.32 (0.75–2.30) |
| 5–10 | 34 | 357 | 1.02 (0.57–1.85) | 0.98 (0.54–1.80) |
| >10 | 65 | 646 | 1.05 (0.61–1.80) | 0.99 (0.56–1.73) |
| Per category increment | 0.95 (0.82–1.11) P = 0.53 |
0.94 (0.80–1.11) P = 0.47 |
||
| Rectum cancer risk | (n = 56) | (n = 533) | ||
| None | 10 | 80 | Ref. | Ref. |
| 1–4 | 12 | 139 | 0.70 (0.29–1.68) | 0.67 (0.27–1.67) |
| 5–10 | 8 | 114 | 0.58 (0.21–1.55) | 0.53 (0.19–1.45) |
| >10 | 26 | 200 | 1.07 (0.47–2.44) | 0.98 (0.41–2.34) |
| Per category increment | 1.08 (0.82–1.41) P = 0.61 |
1.05 (0.78–1.39) P = 0.76 |
||
| Tooth decay | ||||
| Overall CRC risk | (n = 236) | (n = 2214) | ||
| None | 91 | 956 | Ref. | Ref. |
| 1–2 | 50 | 447 | 1.17 (0.81–1.69) | 1.21 (0.83–1.77) |
| 3–5 | 31 | 260 | 1.26 (0.81–1.96) | 1.28 (0.81–2.02) |
| >5 | 12 | 151 | 0.84 (0.45–1.58) | 0.78 (0.41–1.49) |
| Only dentures | 52 | 400 | 1.36 (0.93–1.98) | 1.30 (0.87–1.93) |
| Per category increment | 1.06 (0.97–1.16) P = 0.20 |
1.05 (0.95–1.15) P = 0.35 |
||
| Colon cancer risk | (n = 170) | (n = 1590) | ||
| None | 71 | 700 | Ref. | Ref. |
| 1–2 | 36 | 314 | 1.10 (0.72–1.69) | 1.13 (0.72–1.76) |
| 3–5 | 19 | 186 | 0.99 (0.57–1.69) | 0.99 (0.56–1.74) |
| >5 | 5 | 102 | 0.48 (0.19–1.21) | 0.43 (0.17–1.12) |
| Only dentures | 39 | 288 | 1.31 (0.85–2.02) | 1.28 (0.80–2.04) |
| Per category increment | 1.03 (0.93–1.15) P = 0.54 |
1.03 (0.92–1.15) P = 0.67 |
||
| Rectum cancer risk | (n = 56) | (n = 527) | ||
| None | 17 | 217 | Ref. | Ref. |
| 1–2 | 11 | 114 | 1.34 (0.59–3.04) | 1.30 (0.55–3.05) |
| 3–5 | 10 | 59 | 2.40 (0.99–5.78) | 2.31 (0.91–5.85) |
| >5 | 7 | 45 | 2.17 (0.83–5.66) | 1.94 (0.71–5.28) |
| Only dentures | 11 | 92 | 1.52 (0.67–3.49) | 1.45 (0.61–3.43) |
| Per category increment | 1.15 (0.96–1.38) P = 0.13 |
1.13 (0.94–1.37) P = 0.20 |
||
aConditional logistic analysis with no adjustment.
bIn SMHS/SWHS, models were adjusted for income, education, BMI, exercise, smoking (never, former, and current), and red meat and fruit consumption.
cIn SCCS, models were adjusted for occupation, income, education, and fruit and sweet beverage consumption.
CRC, colorectal cancer; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); SCCS, Southern Community Cohort Study; SMHS, Shanghai Men's Health Study; SWHS, Shanghai Women's Health Study.
Furthermore, we found that the association between CRC and tooth loss (in the three cohorts) and tooth decay (in the SCCS) did not vary by gender (data not shown) or cancer site (colon or rectum; Table 2). We also found no effect modification from total energy intake, BMI, or exercise (data not shown).
meta-analysis
We identified 275 unique records in PubMed, Web of Science, and Embase based on our primary search key words. Among them, 271 were excluded based on a review of their abstracts because they were not population-based. The four remaining articles that described five studies [13–16] (i.e. two investigating the association between tooth loss and CRC [13, 14] and three others examining the association between periodontal disease and CRC risk/death [14–16]) were included in the meta-analysis (see supplementary Figure S1, available at Annals of Oncology online for the flow chart for the meta-analysis study selection process). The characteristics of the included studies are given in Table 3.
Table 3.
Characteristics of studies included in the meta-analysis
| Study | Source of participants | Follow-up duration (years) | No. of subjects (case/control or event/cohort members) | Exposure of interest and definition of reference (group and dental status) | Periodontal disease (subjects/controls) | Odds ratio or relative risk (95% CI) | Covariates adjusted |
|---|---|---|---|---|---|---|---|
| Hujoel et al. [16] | NHANES I, United States | 21 | 98/11 328 | Periodontal disease (yes/no) | No (n = 3962) Yes (n = 2092) |
1.00 (Ref.) 0.91(0.49–1.70) |
Age and gender |
| Michaud et al., [14] | HPFS, United States | 16 | 1043/47 332 | Periodontal disease (yes/no) | No periodontal disease (n = 40 512 (828 CRC)) Periodontal disease [n = 7863 (215 CRC)] |
1.00 (Ref.) 1.05 (0.90–1.23) |
Age, race, physical activity, history of diabetes, alcohol, BMI, geographic location, height, calcium intake, total caloric intake, red meat intake, fruit and vegetable intake, vitamin D score, smoking history and pack-years |
| Ahn et al., [15] | NHANES III, United States | 6 | 39 (death with periodontal disease)/12 605 | Periodontal disease (yes/no) | No (n = 10 400) Yes (n = 1826 moderate periodontitis, n = 379 severe periodontitis) |
1.0 (Ref.) 3.58 (1.15–11.16) |
Age and sex and after further adjustment for smoking, education, race/ethnicity and BMI |
All three studies were cohorts based in the United States.
BMI, body mass index; HPFS, Health Professionals Follow-up Study; NHANES, National Health and Nutrition Examination Survey.
Indicators of poor oral health were not significantly associated with risk of CRC when data from three of the published studies and our current study were meta-analyzed. No significant heterogeneity was found. The pooled ORs and respective 95% CIs for the association between total tooth loss (uppermost versus no tooth loss) or periodontal diseases (with versus without disease status) and CRC risk were 1.00 (0.997–1.01) and 1.05 (0.86–1.29) (Figure 1).
Figure 1.
Forest plot of oral health associated with colorectal cancer in overall analysis.
There was no heterogeneity among studies in overall analyses for periodontal disease (P = 0.10). To examine the sensitivity of the present meta-analysis, we repeated the meta-analysis sequentially, excluding each individual study. The results remained largely unchanged, with OR (95% CI) changing from 1.07 (0.85–1.35, when removing Hujoel's study, P = 0.046), 1.10 (0.78–1.56, when removing Michaud's study, P = 0.047), 1.01 (0.90–1.14, when removing Ahn's study, P = 0.326), 1.02 (0.79–1.31, when removing SCCS, P = 0.078) to 1.09 (0.95–1.25, when removing SMHS/SWHS, P = 0.157). In a bias test using a funnel plot and Begg's test, no significant publication bias was found (P = 0.31).
discussion
In the current analysis of 1063 CRC cases and 5556 controls from three cohort studies, two conducted among Chinese men and women in the SMHS and SWHS, and the other carried out among black and white men and women in the southeastern United States in the SCCS, we did not find a significant association between tooth loss (all three cohorts) or tooth decay (SCCS only) and the risk of CRC.
The null results found in our cohort analyses are generally in line with the results of previous studies. In a hospital-based case–control study involving 662 cases and 1324 age- and sex-matched controls [13], ORs and 95% CIs were 1.22 (0.97–1.52), 1.11 (0.82–1.50), and 0.92 (0.56–1.51) for those subjects with 9–20 teeth, 1–8 teeth, and no teeth, compared with those having >20 teeth. This study [13] was not included in our meta-analysis due to the lack of binary exposure information that was used in the meta-analysis. Similarly, the analysis conducted in the Health Professionals Follow-Up Study [14] also found no significant association between tooth loss and CRC, with ORs and 95% CIs being 0.93 (0.78–1.12) and 1.10 (0.87–1.37) for participants with 17–24 teeth and 0–16 teeth compared with those with >24 teeth. This study also found no significant association between periodontal disease and CRC risk, with OR (95% CI) being 1.05 (0.90–1.23) [14]. In a study of 11 328 adults, including 2092 with periodontitis, 2603 with gingivitis, 3962 without any teeth, and 2671 with healthy periodontium, using data from the National Health and Nutrition Examination Survey, no association was observed between periodontitis, gingivitis, or edentulism and CRC, with respective ORs and 95% CIs being 0.91 (0.49–1.70), 1.27 (0.69–2.34), and 1.07 (0.62–1.84) [13].
However, our results are not in complete agreement with a report from the National Health and Nutrition Examination Survey III [15]. This study followed a cohort of 12 605 individuals for an average of 6 years, and found that periodontal disease was associated with increased CRC mortality (relative risk of 3.58, 95% CI 1.15–11.16). Because mortality is more prone to the influence of socioeconomic status, which is also associated with oral health, residual confounding is possible. Our meta-analysis, which included all but one of these published studies, and our nested case–control studies revealed consistently null results. Given that the upper bounds of the 95% CI for the OR of CRC associated with high tooth loss in the two Shanghai cohorts, the SCCS, and the meta-analysis, were, respectively, 1.18, 1.58, and 1.01, any large increase in risk of this cancer due to poor oral health can be ruled out.
Our study included a large sample size across multiple races/ethnicities and a comprehensive evaluation that also took into consideration possible modifications from diet, socioeconomic characteristics, and other lifestyle factors. However, several limitations of the study should also be acknowledged. First, the statistical power for some stratified analyses is low, resulting in non-precise point estimates. Second, information on oral health was self-reported and potentially prone to recall errors. In addition, information on tooth loss was missing for 4.3% of SCCS participants. Because oral health information was collected at the baseline survey before CRC occurrence, the exposure misclassification is likely to be non-differential, i.e. independent of case–control status, and thus may bias the results toward null. The limited information available on oral health and heterogeneity in exposure assessment are also drawbacks to our study. Finally, our study is limited by the lack of measurements of oral microbiome composition and of chronic systematic inflammation, prohibiting us from directly investigating potential mechanisms that have been suggested to link oral health and CRC risk.
In conclusion, data from three large cohorts and a meta-analysis did not support the hypothesis that oral health is associated with the development of CRC. Future study on this topic should include objective and direct measurements of oral health and may consider incorporating microbiome profiling, evaluation of biomarkers of systematic inflammation, and measurement of immune responses to chronic inflammation.
funding
This work was supported by grants from the US National Cancer Institute: R37 CA070867 (Principal Investigator: WZ), R01 CA082729, UM1 CA173640, and R25 CA160056 (Principal Investigator: X-OS), and R01 CA092447 (Principal Investigators: WJB and WZ).
disclosure
The authors have declared no conflicts of interest.
Supplementary Material
acknowledgements
We thank the research members and participants of the Shanghai Men's Health Study (SMHS), the Shanghai Women's Health Study (SWHS), and the Southern Community Cohort Study (SCCS). We also thank Nan Kennedy for editing this manuscript.
references
- 1.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2015. CA Cancer J Clin 2015; 65(1): 5–29. [DOI] [PubMed] [Google Scholar]
- 2.Chen W, Zheng R, Baade PD et al. Cancer statistics in China, 2015. CA Cancer J Clin 2016; 66(2): 115–132. [DOI] [PubMed] [Google Scholar]
- 3.Beaugerie L, Itzkowitz SH. Cancers complicating inflammatory bowel disease. N Engl J Med 2015; 372(15): 1441–1452. [DOI] [PubMed] [Google Scholar]
- 4.Corrêa Lima MP, Gomes-da-Silva MHG. Colorectal cancer: lifestyle and dietary factors. Nutr Hosp 2005; 20(4): 235–241. [PubMed] [Google Scholar]
- 5.Correa P. Bacterial infections as a cause of cancer. J Natl Cancer Inst 2003; 95(7): E3. [DOI] [PubMed] [Google Scholar]
- 6.Shapiro KB, Hotchkiss JH, Roe DA. Quantitative relationship between oral nitrate-reducing activity and the endogenous formation of N-nitrosoamino acids in humans. Food Chem Toxicol Int J Publ Br Ind Biol Res Assoc 1991; 29(11): 751–755. [DOI] [PubMed] [Google Scholar]
- 7.Soory M. Oxidative stress induced mechanisms in the progression of periodontal diseases and cancer: a common approach to redox homeostasis? Cancers 2010; 2(2): 670–692. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Kostic AD, Chun E, Robertson L et al. Fusobacterium nucleatum potentiates intestinal tumorigenesis and modulates the tumor-immune microenvironment. Cell Host Microbe 2013; 14(2): 207–215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Ray K. Colorectal cancer: Fusobacterium nucleatum found in colon cancer tissue—could an infection cause colorectal cancer? Nat Rev Gastroenterol Hepatol 2011; 8(12): 662. [DOI] [PubMed] [Google Scholar]
- 10.Seville LA, Patterson AJ, Scott KP et al. Distribution of tetracycline and erythromycin resistance genes among human oral and fecal metagenomic DNA. Microb Drug Resist Larchmt N 2009; 15(3): 159–166. [DOI] [PubMed] [Google Scholar]
- 11.Dal Bello F, Hertel C. Oral cavity as natural reservoir for intestinal lactobacilli. Syst Appl Microbiol 2006; 29(1): 69–76. [DOI] [PubMed] [Google Scholar]
- 12.Maukonen J, Mättö J, Suihko M-L, Saarela M. Intra-individual diversity and similarity of salivary and faecal microbiota. J Med Microbiol 2008; 57(Pt 12): 1560–1568. [DOI] [PubMed] [Google Scholar]
- 13.Hiraki A, Matsuo K, Suzuki T et al. Teeth loss and risk of cancer at 14 common sites in Japanese. Cancer Epidemiol Biomark Prev 2008; 17(5): 1222–1227. [DOI] [PubMed] [Google Scholar]
- 14.Michaud DS, Liu Y, Meyer M et al. Periodontal disease, tooth loss, and cancer risk in male health professionals: a prospective cohort study. Lancet Oncol 2008; 9(6): 550–558. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ahn J, Segers S, Hayes RB. Periodontal disease, Porphyromonas gingivalis serum antibody levels and orodigestive cancer mortality. Carcinogenesis 2012; 33(5): 1055–1058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hujoel PP, Drangsholt M, Spiekerman C, Weiss NS. An exploration of the periodontitis—cancer association. Ann Epidemiol 2003; 13(5): 312–316. [DOI] [PubMed] [Google Scholar]
- 17.Signorello LB, Hargreaves MK, Blot WJ. The Southern Community Cohort Study: investigating health disparities. J Health Care Poor Underserved 2010; 21(1 Suppl): 26–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Shu X-O, Li H, Yang G et al. Cohort profile: the Shanghai Men's Health Study. Int J Epidemiol 2015; 44(3): 810–818. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Zheng W, Chow W-H, Yang G et al. The Shanghai Women's Health Study: rationale, study design, and baseline characteristics. Am J Epidemiol 2005; 162(11): 1123–1131. [DOI] [PubMed] [Google Scholar]
- 20.Zhang X, Giovannucci EL, Wu K et al. Associations of self-reported sleep duration and snoring with colorectal cancer risk in men and women. Sleep 2013; 36(5): 681–688. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.International Statistical Classification of Diseases, Tenth Revision. Geneva, Switzerland: World Health Organization, 1992. [Google Scholar]
- 22.Muraki I, Imamura F, Manson JE et al. Fruit consumption and risk of type 2 diabetes: results from three prospective longitudinal cohort studies. BMJ 2013. doi:10.1136/bmj.f5001. [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.