Date of Award

Fall 2022

Document Type

Open Access Dissertation


Health Services and Policy Management

First Advisor

Sudha Xirasagar


Background: Periodontal disease, a chronic bacterial infection of the gums, if untreated, causes chronic inflammation over years and eventual tooth loss. The associations between periodontal disease (measured by missing teeth) and certain sites of cancer have been established, in epidemiological studies, and attributed to systemic inflammatory impacts causing carcinogenic change over decades. Because periodontal exudates are toxic, containing locally bioactive bacteria and toxins, this study hypothesized and studied the potential higher cancer risk of anatomic sites exposed to oral exudates (exudate-exposed site cancers (EEC); head and neck, oral, pharynx, larynx, esophagus, stomach). The study studied the risk of exudate exposed sites and the cancer risk of remote gastrointestinal (GI) sites (colon, rectal, and pancreas), sites that are primarily affected by the systemic inflammation pathways. The study also assesses the relative risks of different site groups, EEC vs. remote (GI) cancers, and in turn, remote (GI) cancers vs. other cancers (excluding EEC) among the US population reporting tooth loss, representing chronic periodontal disease. To mitigate the likelihood of spurious findings just by chance, the study also assesses the risk of a comparator disease, cardiovascular disease using the same dataset, a disease that is well known to be associated with chronic periodontitis.

Methods: Pooled data were used, from the United States’ Behavioral Risk Factor Surveillance Survey (BRFSS) for 2016, 2018, and 2020, which included dental questions. A total of 305,882 participants from 31 states that fielded the cancer survivorship module in these years were included in the final analysis. Prior periodontitis was identified by the number of missing teeth not lost due to trauma or orthodontic extraction. Severity of periodontitis was categorized as 0 to 5 missing teeth; 6 to 31 missing teeth; and 32 (all) missing teeth. EEC, remote GI cancers, and CVD were our outcome variables for regression analyses, using hierarchical logistic regression. EEC was present if the individual reported being diagnosed with one of the above-mentioned EEC malignancies and not present, if they had no cancer diagnosis. The coding for remote GI cancer, and CVD variables were similar. Cancer type was our outcome variable for the multinomial regression to assess the relative risk of EEC vs. remote GI cancer, and remote GI cancer vs. other cancer. We accounted for the known cancer risk factors; demographics (age, sex, race); SES (education, marital status, health insurance, rurality, region); lifestyle factors (current or past smoking/tobacco use vs. never, and current heavy alcohol use vs. not); and metabolic disease (diabetes).

Results: Of total 305,882 study-eligible respondents aged over 18 years; 250,589 persons had 0-5 missing teeth, 34,850 had 6-31 missing, and 20,443 had all teeth missing. Respondents reported a total of 709 EEC cancers; 1,879 remote GI cancers; 40,167 other cancers; and 263,127 respondents had no cancer diagnosis. Relative to persons with 0 to 5 missing, the adjusted odds of EEC for persons with 6 to 31 missing teeth, and with all missing teeth were 1.43 (95% CI: 1.17-1.76) and 2.08 (95% CI: 1.66-2.59), respectively. The corresponding odds ratios for remote GI cancers were 1.34 (95% CI: 1.19-1.52) and 1.49 (95% CI: 1.30-1.72), respectively. The adjusted odds of EEC, relative to remote GI cancer, were 1.07 (95% CI: 0.84–1.35) among the 6-31 missing teeth category and 1.39 (95% CI: 1.07–1.80) among the category with all teeth missing, using 0-5 missing teeth as the reference group. The adjusted odds ratios for remote GI cancer relative to other cancers (excluding EEC) were 1.22 (95% CI: 1.07-1.38) for the 6-31 missing teeth category, and 1.48 (95% CI: 1.28-1.71) for all teeth missing, using 0-5 missing teeth as the reference group. For all analyses stated above, subgroup analysis was conducted, among the smoking/tobacco user stratum and non-tobacco user stratum, with similar results in each sub-group. Sensitivity analysis, limiting the analytic sample to adults aged 30 or more, and separately to adults aged 50 or higher, showed no change in the results. For the comparator disease, CVD, the corresponding odds ratios for missing teeth categories were 1.64 (95% CI: 1.58–1.70) and 1.88 (95% CI: 1.80–1.96), respectively.

Conclusion: Prior oral disease represented by missing teeth is associated with a high likelihood of EEC with a trend of dose-response relationship. It is also associated with a lower but statistically significant risk of remote GI cancers. The relative risk of EEC is significantly greater than remote GI cancers particularly among the severe oral disease group (all teeth missing). Remote GI cancer risk, in turn, is greater than other cancer risk. The consistent observation of a dose-response relationship reflects increasing risk with increasing severity of periodontitis. The confirmation of the well-documented CVD-periodontitis association in these data, mitigates the likelihood of spurious findings and concerns regarding use of cross sectional, self-reported data. It provides additional validation of a true association between chronic oral disease and exudate exposed cancers, an association that is of higher effect size than the association with remote GI cancers, despite our data being cross-sectional, self-reported. Our findings warrant further research to explore the biological pathways involved, as well as more definitive causal validations using prospective studies. More research of these types will be needed to enable far-reaching policy and practice changes. Despite the limitations, our study, makes a substantial contribution to the established body of evidence, reinforcing a conclusion that programs and policies promoting oral self-care and oral health coverage, particularly among working-age populations and senior adults, may enable substantial reductions in the incidence of these cancers. Key study limitations include self-reported data, cross-sectional data, and absence of data on human papilloma virus disease, an important causal factor in head and neck and upper GI cancers.

Available for download on Sunday, December 15, 2024