Abstract
The association between oral squamous cell carcinoma (OSCC) and periodontitis in large hospital cases with cohort controls has yet to be evaluated. The aim of this study was to investigate the association of periodontitis with OSCC across tumor location and tumor-node-metastasis (TNM) stage among Koreans (N = 424). OSCC cases (n = 146) were recruited from Seoul National University Dental Hospital and matched by age, sex, and smoking to controls (n = 278) from the Yangpyeong health and periodontal cohort in Korea. OSCC was diagnosed through biopsy and radiographs, including computed tomography and magnetic resonance imaging. Tumor location and TNM stage were classified after the surgery. Periodontitis was defined by alveolar bone loss with panoramic radiographs following the guidelines of the Fifth European Workshop in Periodontology. Alcohol intake, education, physical activity, obesity by body mass index, hypertension by blood pressure, diabetes by plasma glucose, and hypercholesterolemia by plasma cholesterol were considered as confounders. Information about age, sex, smoking, alcohol intake, education, and physical activity was obtained through interview; body mass index and blood pressure, through physical examination; and preoperative glucose and cholesterol, through laboratory tests. Bivariate analysis was applied with Fisher’s exact chi-square test. Multivariable conditional logistic regression models were applied to evaluate the adjusted association of periodontitis with OSCC after controlling for confounders. Subgroup analyses were explored by OSCC and periodontitis. Participants with periodontitis were 3.7 times more likely to have OSCC (adjusted odds ratio [aOR] = 3.66, 95% CI = 1.46 to 9.23) than participants without periodontitis. The differences in periodontitis were not statistically significant across TNM stages of OSCC (P > 0.05) and its location (P > 0.05). The link was highlighted among males (aOR = 6.55), elders aged >60 y (aOR = 4.98), and those with more tooth loss (aOR = 9.99). Our data showed that periodontitis was independently associated with OSCC. Thus, the risk of OSCC could be modulated by reducing periodontitis.
Keywords
Introduction
Oral squamous cell carcinoma (OSCC) is the most common oral malignant lesion (Saman 2012). According to the Korea National Cancer Incidence Database, as of January 1, 2015, the crude prevalence rate of lip, oral cavity, and pharynx per 100,000 in both sexes is 38.8 (52.2 in men and 25.3 in women), and the age-standardized prevalence rate per 100,000 in both sexes was 25.0 (35.5 in men and 154.8 in women; Jung et al. 2017). The percentage of oral cancer (OC) in the total cancer count is relatively small; however, the life-threatening characteristic of OC is practically important to the overall wellness of people.
Commonly known risk factors of OC include tobacco and alcohol (Blot et al. 1988). Other risk factors include systemic diseases, such as diabetes (Goutzanis et al. 2007), metabolic syndrome (Chang et al. 2015), and chronic inflammation and infection (Mantovani et al. 2008).
Periodontitis involves chronic inflammation and is the sixth-most prevalent disease worldwide; its global burden increased by 57.3% from 1990 to 2010; and it has been linked to carcinogenesis (Tonetti et al. 2017). However, the association between periodontitis and OSCC is still controversial. From 1990 to August 2018, 18 articles reported on the association between oral health status and head and neck cancer. Of these 18 articles, 15 showed a positive association (Marshall et al. 1992; Bundgaard et al. 1995; Zhang and Yu 1998; Garrote et al. 2001; Rosenquist et al. 2005; Guha et al. 2007; Hiraki et al. 2008; Tezal et al. 2007; Tezal et al. 2009; Divaris et al. 2010; Ansai et al. 2013; de Moraes et al. 2013; Moergel et al. 2013; Wen et al. 2014; Laprise et al. 2016), while the other 3 showed no association (Talamini et al. 2000; Michaud et al. 2008; Eliot et al. 2013). Hitherto, 5 meta-analyses (2013 to August 2018) reported a positive association (odds ratio, 2.0 to 3.21) between periodontal disease and head and neck cancer. There was high heterogeneity in these meta-analyses due to differences in assessment methods for periodontitis, tumor sites, geographic differences, the lack of potential confounders, and the selection of controls. To overcome these limitations, we considered a case-control study encompassing a sufficient number of OSCC cases from the hospital and controls matched to age, sex, and smoking from a community health cohort. This case-control study could reduce the risk of a false-negative finding due to hospital controls and increase the precision of the estimates. Moreover, we applied a definite assessment method for periodontitis using panoramic radiographs to reduce the information bias due to misclassification. Also, confounders—including alcohol intake, education, physical activity, obesity, diabetes, hypertension, and hypercholesterolemia—were considered to reduce false-positive findings due to underadjustments. Specifically, no result was published in Korea on the adjusted association of periodontitis with OSCC according to tumor location and tumor-node-metastasis (TNM) stages.
This case-control study aimed to test the hypothesis that the history of periodontitis is associated with OSCC among Koreans, adjusted for confounding variables. Also, we explored whether periodontitis is associated with OSCC according to tumor location and TNM stages.
Materials and Methods
Study Design
This study followed the STROBE guidelines (Strengthening the Reporting of Observational Studies in Epidemiology). This observational hospital case-control study was retrospectively evaluated. The OSCC cases were obtained from patients in the Seoul National University Dental Hospital (SNUDH). The controls were obtained from the Yangpyeong health cohort after matching to age, sex, and smoking.
Ethical Considerations
All participants provided an informed consent statement voluntarily. Ethical approval for this study was reviewed and granted by the SNUDH Institutional Review Board (ERI17014).
Sample Size Estimation
Sample size was estimated via chi-square test based on an alpha error of 0.05 and a beta error of 0.2. In the pilot study, the proportion of periodontitis among controls was 0.3, and that in cases was 0.6. Due to the difficulty in recruiting patients with OSCC, the estimated ratio between the cases and controls was set at 1:2. The estimated sample size was 93 (31 cases and 62 controls). Since 1 confounder increased the sample size by 10%, 10 confounders increased the sample size by 100% to 186 (62 cases and 124 controls). In consideration of stratified analysis with a binary variable, the total sample size doubled to 372 (124 cases and 248 controls).
Selection of Participants
Hospital OSCC cases were patients from SNUDH from 2015 to 2017. The inclusion criteria for cases with a final diagnosis of OSCC were as follows: 1) agreed to be a participant voluntarily, 2) was scheduled for surgery, 3) was not pregnant, and 4) had no inflammatory diseases of the oral and maxillofacial region as well as auto-immune disorders and/or infectious diseases. Controls matched to age (same age), sex, and smoking (yes or no) were randomly selected from the Yangpyeong health and dental cohort, which is part of the Korean Genome Epidemiologic Study, supported by the Korea Centers for Disease Control and Prevention since 2010. For hospital OSCC cases, 153 patients satisfied the inclusion criteria, but only 146 OSCC cases had age-, sex-, and smoking-matched controls from the Yangpyeong cohort. Out of 146 OSCC cases, 137 were matched with controls in a 1:2 ratio and 9 in a 1:1 ratio. As a result, 278 controls were selected. Finally, 424 participants (146 cases and 278 controls) were included in this hospital case-control study for final analysis.
Assessment of OSCC
Initial diagnosis of OSCC (International Classification of Diseases for Oncology codes C02.0 to C06.9) was through oral examination on the first visit by an oral and maxillofacial cancer surgeon (J.H.L.). The initial diagnosis was confirmed by an SNUDH radiologist using panoramic radiographs, enhanced computed tomography, magnetic resonance imaging, and positron emission tomography–computed tomography, and an SNUDH pathologist made the final diagnosis based on histological tissue biopsy. Information on tumor location and TNM stage (Kreppel et al. 2010) was obtained from the gross biopsy of the tumor after the surgery. In this study, carcinoma in situ was noted as stage 0. According to the location, OSCC was classified into tongue (C02.0 to C02.9), floor of the mouth (C04 to C04.9), hard palate (C05.0 to C05.9), buccal mucosa (C06.0), retromolar trigone (C06.2), and alveolar ridge (C06.8 to C06.9; Fritz 2000).
Assessment of Periodontitis
Radiographic alveolar bone loss (RABL) of each natural tooth was assessed by 2 trained dentists (Y.S., H.W.C.) using panoramic radiographs (Orthopantomograph OP100; GE Healthcare). RABL, a definitive measure of the history of periodontitis (Armitage 2004), was assessed and recorded by measuring the mesial and distal parts of the remaining teeth from the cementoenamel junction of the tooth up to the highest point of the proximal alveolar bone crest. All mesial and distal parts of the natural teeth, except the third molars, were measured. The reproducibility for the RABL measurements yielded an intraclass correlation coefficient of 0.971 among 415 sites. Classification was categorized following the guidelines of the Fifth European Workshop in Periodontology: “normal,” proximal bone loss <3 mm in <2 nonadjacent teeth; “incipient,” proximal bone loss ≥3 mm in ≥2 nonadjacent teeth; “severe,” proximal bone loss ≥5 mm in ≥30% of teeth (Tonetti et al. 2005). Periodontitis was dichotomized into a periodontitis group for incipient and severe periodontitis and a nonperiodontitis group for normal.
Assessment of Confounders and Effect Modifiers
We included demographic and health-related behavioral factors and systemic factors (Rosenquist et al. 2005; Fitzpatrick and Katz 2010) in the analysis to eliminate the impact of confounders.
All participants were interviewed and screened individually in person by trained personnel for demographic information, such as age (continuous) and sex, as well as health-related behavioral variables, such as smoking, alcohol intake, education level, physical activity, and medical history of systemic diseases (diabetes, hypertension, and hypercholesterolemia). Other health-related behavioral variables were through physical examination (e.g., body mass index, systolic and diastolic blood pressure), and preoperative fasting plasma glucose (FPG) and fasting total cholesterol (FTC) were obtained through serum laboratory tests.
Sex, smoking, alcohol intake, education level, physical activity, obesity, systemic diseases (diabetes, hypertension, and hypercholesterolemia), and tooth loss were dichotomized into male and female, middle school and high school or higher, smoker and nonsmoker, drinker and nondrinker, physical activity and no physical activity, with and without systemic diseases, and more (≥8) and less (0 to 7) tooth loss. Obesity was dichotomized via body mass index: no, <25.0 kg/m2; yes, ≥25.0 kg/m2 (WHO Expert Consultation 2004). Diabetes was validated with preoperative glucose level by categorizing FPG into 2 groups according to the diagnostic criteria of the American Diabetes Association (2014): no, FPG <126 mg/dL; yes, FPG ≥126 mg/dL, insulin shots, antidiabetic medication, and/or diagnosis by physician. Hypertension was validated with preoperative systolic and diastolic blood pressure and categorized into 2 groups according to the criteria of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: no, systolic blood pressure <140 mm Hg or diastolic blood pressure <90 mm Hg; yes, systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or antihypertensive medication (Chobanian et al. 2003). Hypercholesterolemia was validated with preoperative FTC level and categorized into 2 groups according to the criteria of National Cholesterol Education Program’s Adult Treatment Panel III (Expert Panel 2001): no, FTC <240 mg/dL; yes, FTC ≥240 mg/dL or anticholesterol medication.
Statistical Analysis
Characteristics of the variables were derived with mean ± SD values for the continuous variable (age) and with frequency distributions for the categorical variables (all others; Table 1). Chi-square and t tests were used to compare differences in categorical and continuous variables, respectively. For evaluating the specific association of periodontitis with OSCC according to tumor location and TNM stages, a chi-square test was applied among periodontitis, tumor location, and TNM stage. Bivariate analysis was applied with Fisher’s exact chi-square test. Multivariable conditional logistic regression models were used to evaluate the association of periodontitis with OSCC conditional on age, sex, and smoking by calculating the adjusted odds ratios (aORs) and the associated 95% CI after controlling for potential confounders, such as alcohol intake, education level, physical activity, obesity, diabetes, hypertension, and hypercholesterolemia. We constructed the final model (model 2) using the dichotomized periodontitis variables (normal vs. periodontitis). Stratified associations were also analyzed for age, sex, smoking, and tooth loss. Statistical significance was set at P < 0.05. All statistical analyses were performed with SPSS 25.0 (IBM).
Characteristics of Variables according to Oral Squamous Cell Carcinoma Cases and Cohort Controls (N = 424).
Values are presented as n (%) or mean ± SD. The ratio between cases and controls matched for age, sex, and smoking was 1:2, except for 9 cases of 1:1. Bold denotes statistical significance (P < 0.05).
Periodontitis: normal, proximal bone loss <3 mm in <2 nonadjacent teeth; incipient, proximal bone loss ≥3 mm in ≥2 nonadjacent teeth; severe, proximal bone loss ≥5 mm in ≥30% of teeth present.
Smoking: no, never smoked; yes, past smoker and currently smoking.
Alcohol intake: no, never drank; yes, past drinker and currently drinking.
Physical activity: no, no exercise; yes, any form of exercise >30 min daily.
Obesity, as body mass index: no, <25.0 kg/m2; yes, ≥25.0 kg/m2.
Diabetes: no, fasting plasma glucose <126 mg/dL; yes, fasting plasma glucose ≥126 mg/dL, insulin shots, antidiabetic medication, and/or diagnosed by the physician.
Hypertension: no, systolic blood pressure <140 mm Hg or diastolic blood pressure <90 mm Hg; yes, systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or antihypertensive medication.
Hypercholesterolemia: no, fasting total cholesterol <240 mg/dL; yes, fasting total cholesterol ≥240 mg/dL.
Results
Characteristics of Participants
This hospital case-control study consisted of 424 participants: 146 OSCC cases (34.43%; 90 men and 56 women; mean age, 63.8 y) and 278 controls (65.57%; 167 men and 111 women; mean age, 64.4 y; Table 1).
As compared with controls, cases included more patients with periodontitis, fewer alcohol drinkers, more high school graduates or higher, more patients with diabetes, and more patients with hypercholesterolemia (P < 0.05). However, hypertension, physical activity, obesity, and tooth loss were not different between cases and controls (P > 0.05).
Distribution of OSCC according to Tumor Location, TNM Stage, and Periodontitis
OSCC showed the highest prevalence in the alveolar ridge (44.5%), followed by the tongue (21.2%), buccal mucosa (17.1%), hard palate (6.8%), retromolar trigone (6.2%), and floor of the mouth (4.1%; Table 2). Stage IV cancers were the most representative (39.7%) and carcinoma in situ the least (2.7%).
Distribution of Oral Squamous Cell Carcinoma according to Tumor Location, TNM Stage, and Periodontitis (N = 146).
TNM, tumor-node-metastasis.
Row percentage.
Column percentage.
Periodontitis: incipient and severe periodontitis.
In terms of TNM stages of OSCC, stage IV showed the highest proportion in the periodontitis group (40.9%), whereas it was stage II for the nonperiodontitis group (44.4%). However, the prevalence of periodontitis was not significantly different across the TNM stages of OSCC (Fig. 1A).
Prevalence of periodontitis across (
In terms of the location of OSCC, the alveolar ridge had the highest number among other locations (45.3%) in the periodontitis group, in contrast to the tongue (55.6%) for the nonperiodontitis group. Tongue cancer showed less prevalence of periodontitis, which was not statistically significant (Fig. 1B).
Association between Periodontitis and OSCC
Compared with patients having normal status, those with incipient periodontitis were 3.5 times more likely to have OSCC (aOR = 3.46, 95% CI = 1.35 to 8.90), and those with severe periodontitis were 4.1 times more likely (aOR = 4.07, 95% CI = 1.50 to 11.03), thereby showing a significant dose-response relationship (trend, P = 0.003; Table 3, model 1). The group encompassing incipient and severe periodontitis also showed a significant association with OSCC (aOR = 3.66, 95% CI = 1.46 to 9.23; Table 3, model 2).
Adjusted Association between Periodontitis and Oral Squamous Cell Carcinoma (N = 424).
Bold denotes statistical significance (P < 0.05).
Model 1: Adjusted association of periodontitis (3 groups) by multivariable conditional logistic regression model, conditional on age, sex, and smoking, adjusted for alcohol intake, education level, physical activity, obesity, diabetes, hypertension, and hypercholesterolemia.
Model 2: Adjusted association of periodontitis (2 groups) by multivariable conditional logistic regression model, same as Model 1.
Stratified Association between Periodontitis and OSCC
There was an effect modification by age, sex, smoking, and tooth loss on the association of periodontitis (normal vs. incipient and severe) with OSCC. The link was more highlighted among males (aOR = 6.5, 95% CI = 1.12 to 38.22), elders aged >60 y (aOR = 5.0, 95% CI = 1.37 to 18.11), and those with more tooth loss (aOR = 10.0, 95% CI = 1.22 to 81.52), but it was less highlighted among nonsmokers (aOR = 3.2, 95% CI = 1.07 to 9.37; Table 4).
Association of Periodontitis with Oral Squamous Cell Carcinoma Stratified by Age, Sex, and Smoking.
Multivariable conditional logistic regression model, conditional on age, sex, and smoking, adjusted for variables in Table 3. Bold denotes odds ratio >3.664 and statistical significance (P < 0.05).
Periodontitis: incipient and severe periodontitis.
Discussion
This well-designed hospital case-cohort study showed that among Koreans, those with periodontitis are 3.7 times more likely than those without periodontitis to have OSCC, after controlling for various confounders. Our results confirmed previous studies showing the positive association of periodontitis with OC. For comparison across countries, international case-cohort programs must be organized as soon as possible for future research.
There was an effect modification of age, sex, smoking, and more tooth loss. The link in our data highlighted an increased prevalence of 5 times for those >60 y, 6.5 times for males, and 10.0 times for those with more tooth loss. A previous study also showed a predilection of the link to older adults and to males (Rosenquist et al. 2005). Our data also support previous results that more tooth loss has given impact on the association with OC (Marshall et al. 1992; Bundgaard et al. 1995; Rosenquist et al. 2005; Ansai et al. 2013). In addition, the prevalence of smokers in our study was 44.5%, which is close to the prevalence of 44.8% among the representative Koreans matched by age and sex to the participants of our study (Park et al. 2018). Moreover, our data showed that alcohol drinking was negatively associated with OC. However, in 3 international cohorts, moderate drinking (1 to 2 drinks/day) showed an increased incidence in oral and pharynx cancer, and light drinking (≤1 drink/day) showed no association (Choi et al. 2018). Hence, future studies including information about alcohol exposure over time will clarify these discrepancies.
Our data showed that the prevalence of periodontitis was high in OSCC but not different across TNM stages of OSCC. Moreover, the severity of periodontitis did not show any differences across early and severe types of OC. We speculate that chronic periodontal inflammation does not have a critical role in the progression of OSCC. Some previous studies reported that OSCC was usually detected in advanced stages in relation to periodontitis (Rosenquist et al. 2005). The discrepancies between our data and previous results indicate further studies for clarification.
The novelties of this study as compared with previous reports are as follows. First, hospital cases were based on Koreans with OSCC, while the controls were from the general Korean population, which reduced the bias associated with errors resulting from hospital controls. Second, despite the rarity among cancers, a sufficient number of cases were included, which reduced the uncertainty due to small numbers. Third, controls were selected by matching well-known risk factors, such as age, sex, and smoking, to cases which adjusted the influence of these factors on the association. Fourth, history of periodontitis was assessed with panoramic radiographs, which reduced information bias due to the misclassification of periodontitis. Fifth, definitive confounders—including alcohol intake, education level, physical activity, obesity, diabetes, hypertension, and hypercholesterolemia—were added in the model for the adjustments. Finally, stratified analyses for the link were done to evaluate the effect modification of age, sex, smoking, and tooth loss.
Periodontal disease is a chronic inflammatory disease that is significantly associated with an increased risk of OC (Michaud et al. 2008) as compared with gastric, pancreatic, lung, prostate, hematologic, and other cancers (Fitzpatrick and Katz 2010). In addition, deep periodontal pockets were suggested as a niche for viral infections, such as human papillomavirus (Hormia et al. 2005) and Epstein-Barr virus and cytomegalovirus (Saygun et al. 2005). Our data did not include these as confounders, which may have led to overestimation of our results. The mechanism of chronic inflammatory processes in OC, including the aforementioned variables, should be indicated as future research.
Moreover, poor oral hygiene and premalignant lesions such as leukoplakia and lichen planus were reported as related factors resulting in chronic inflammation, which are also associated with head and neck cancer (Zheng et al. 1990). Since bacterial burden is essential to periodontitis, the association of bacterial burden and precancerous lesions with OC could lead to hypotheses for new research. The connection to chronic inflammation is made in 2 ways: through infectious conditions and through activation of oncogenes. Both ways release chemical mediators that induce carcinogenesis (Mantovani et al. 2008). Based on the fourth Korea National Health and Nutrition Examination Survey (2007 to 2009), the prevalence rate of periodontitis was 32.0% (Lee et al. 2017). Due to the high prevalence rate of periodontitis, the advantageous effect of periodontal care and therapy will eventually lessen the risk of OC. Prevention and treatment of periodontitis could thus serve as preventive care for OC, which could be of great clinical relevance. Hence, it is speculated that the risk of OSCC could be modulated through the reduction of periodontal inflammation.
Further research must focus on detecting OSCC at an earlier stage through systemic influences such as salivary and serum biomarkers. Salivary proteomics and blood exosome analysis may be used to identify diagnostic markers in the detection of OSCC in association with chronic periodontitis. Saliva has been an interesting alternative diagnostic fluid that is reproducible and noninvasive, although its composition is easily affected by environmental factors (Wong 2006). Moreover, an innovative salivary diagnostic tool that is easily accessible to people should be developed.
This study has some limitations. First, a case-control study design has inherent selection bias for controls. However, it is the most appropriate study design for cancer research because OC is a rare disease among other cancers. Second, information was limited on other potential risk factors, such as viral infections, precancerous lesions, and dietary information. These limitations could lead to overestimation of the association. Notwithstanding these limitations, this study is valid enough to evaluate the association of periodontitis with OSCC.
Conclusion
Overall, our data showed that periodontitis was independently associated with OSCC. Thus, periodontitis could be a risk factor of OSCC and its progress. Moreover, the high-risk groups were elders and males. As such, the risk of OSCC could be modulated by reducing periodontitis.
Author Contributions
Y.J. Shin, J.H. Lee, H.D. Kim, contributed to conception, design, data acquisition, analysis, and interpretation, drafted and critically revised the manuscript; H.W. Choung, contributed to data analysis and interpretation, drafted and critically revised the manuscript; I.C. Rhyu, contributed to conception and design, drafted and critically revised the manuscript. All authors gave final approval and agree to be accountable for all aspects of the work.
Footnotes
Acknowledgements
The authors appreciate the efforts of all health professionals and staff who took their time to gather information for the Yangpyeong cohort study and the Seoul National University cancer-periodontitis study. The authors also give special thanks to all the participants in these studies.
This study was supported by the Bio & Medical Technology Development Program of the National Research Foundation funded by the Korean Ministry of Science and ICT (NRF-2017M3A9B6062984) and by a grant from the Seoul National University Dental Hospital (No. 07-2017-0006).
The authors declare no potential conflicts of interest with respect to the authorship and/or publication of this article.
