OUP user menu

Periodontal disease with treatment reduces subsequent cancer risks

Ing-Ming Hwang, Li-Min Sun, Cheng-Li Lin, Chun-Feng Lee, Chia-Hung Kao
DOI: http://dx.doi.org/10.1093/qjmed/hcu078 805-812 First published online: 10 April 2014


Aim: The aim of our study was to evaluate the relationship between routine treatment of periodontal disease (PD) and the subsequent risks for cancers in Taiwan.

Methods: Study participants were selected from the Taiwan National Health Insurance (NHI) system database. The PD with a routine treatment cohort contained 38 902 patients. For each treatment cohort participant, two age- and sex-matched comparison (control) cohort participants were randomly selected. Cox’s proportional hazards regression analysis was used to estimate the effects of PD with treatment on the subsequent risk of cancer.

Results: The overall risk of developing cancer was significantly lower in the treatment cohort than in the patients without treatment (adjusted Hazard ratio = 0.72, 95% confidence interval = 0.68–0.76). The risks of developing most gastrointestinal tract, lung, gynecological and brain malignancies were significantly lower in the treatment cohort than in the comparison cohort. In contrast, the risks of prostate and thyroid cancers were significantly higher in the treatment cohort than in the comparison cohort.

Conclusions: Our findings suggest that PD with treatment is associated with a significantly reduced overall risk of cancer and reduced risks of certain types of cancers.


Periodontal disease (PD) is a chronic inflammatory disease that affects the alveolar bone, periodontal ligament, cementum and gingiva. It is most common in adults, and has traditionally been divided into two categories: gingivitis and periodontitis.1 Gingivitis can progress to periodontitis. PD is caused primarily by Gram-negative bacteria. The chronic bacterial infection in PD results in painful and bleeding gums, alveolar bone destruction and odontoseisis, ultimately stimulating the inflammatory response.2 The worldwide prevalence of the disease varies according to geographic location and race. Estimates of the global prevalence of severe PD generally range from 10% to 15%, and prevalences of gingivitis have been reported to up to 90%.2 One study of 8462 Taiwanese patients aged 35–44 years found that 94.8% of them exhibited signs of PD.3

The associations between PD and systemic manifestations of cardiovascular disease and diabetes are well known.4 Evidence also suggests that people with PD may have a higher risk for cancer of the mouth, upper gastrointestinal tract, breast, kidney and hematopoietic tissues.5,6 The inflammatory cells and mediators produced in response to PD may account for these relationships.6 In addition, the inflammation caused by PD may enhance cellular proliferation and mutagenesis, reduce adaption to oxidative stress, promote angiogenesis and inhibit apoptosis.7 Treatment for periodontal infection can reduce markers of systemic inflammation and endothelial dysfunction within 2–6 months.8 Based on these findings, we hypothesized that routine treatment for PD may reduce the risk of cancer. Therefore, we conducted a population based, retrospective cohort study that investigated the relationship between PD with treatment and subsequent cancer risks in Taiwan.

Materials and methods

Data sources

The participants for our study were selected from the Taiwan National Health Insurance (NHI) system. The NHI program is a universal insurance system that was formed by the Taiwan Department of Health in March 1995 from 13 previously existing insurance-related systems. By the end of 2009, the NHI covered ∼99% of the 23.74 million residents of Taiwan.9 The National Health Research Institutes (NHRI) cooperate with the Bureau of NHI to prepare annual data sets for public use that are based on NHI claims data. The data sets available at the time of our study represented the registry of a randomly sampled cohort of 1 000 000 enrollees in the NHI system with claims from 1996 to 2010. We applied the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) codes to retrieve information based on diagnosis. The database encrypts the patients’ personal information for privacy protection and provides researchers with anonymous identification numbers associated with the relevant claim information, which includes the patient's sex, date of birth, registry of medical services and medication prescriptions. Patient consent is not required for accessing the database. This study was approved by the Institutional Review Board of China Medical University in central Taiwan (CMU-REC-101-012).

Study subjects

Patients who were newly diagnosed and received at least 10 treatments for PD (ICD-9-CM codes 523.4 or 523.5) between 1997 and 2010 were selected as the treatment cohort. The treatments offered to these patients included subgingival curettage (scaling and root planing) and periodontal flap surgery.10 Because the prevalence of PD among the Taiwanese population is ∼95%,3 the comparison cohort consisted of two randomly selected age- and sex-matched people without medical claims for PD care between 1997 and 2010 for each treatment cohort member. The index date was defined as the date of PD diagnosis for follow-up person–years measurements. Patients with a history of cancer (ICD-9-CM codes 140-208) before the index date were excluded from our study.

Outcome measures

To measure the incidence of cancer in the treatment and comparison cohorts, the follow-up endpoint was defined as death, withdrawal from the NHI system, a diagnosis of cancer (ICD-9-CM codes 140-208) or the end of 2010. The histories of diabetes (ICD-9-CM code 250), hypertension (ICD-9-CM codes 401-405) and hyperlipidemia (ICD-9-CM code 272) were based on diagnoses made before the index date.

Statistical analysis

The distributions of sociodemographic and clinical characteristics were compared between the treatment and comparison cohorts, and examined using chi-squared analysis. The sex-, age-, occupation- and comorbidity-specific incidence densities of cancer were measured for both cohorts. To assess the relative incidence in each subgroup between the two cohorts, we measured the ratios of incidence rates. Hazard ratios (HRs) and incidences that were based on a 95% confidence interval (95% CIs) were calculated using the Cox proportional hazards model to assess the association between treatment for PD and the risk of developing cancer. Cancer site-specific incidence rates and HRs were also determined. SAS software version 9.1 (SAS Institute, Carey, NC, USA) was used for all data analyses, and two-sided probability values (P) <0.05 were considered statistically significant.


Our retrospective cohort study included 38 902 participants in the PD with treatment cohort and 77 804 participants in the comparison cohort. The mean age and standard deviation of the treatment cohort was 43.1 ± 13.6 years, and 50.6% of the treatment cohort comprised women (Table 1). The comparison cohort had an age and sex distribution similar to that of the treatment cohort. The majority of the treatment cohort participants were white-collar workers, with a greater tendency to have diabetes, hypertension and hyperlipidemia than the comparison cohort. The overall incidence of cancer was 0.74-fold lower in the treatment cohort than in the comparison cohort (41.3 vs. 55.8 per 10 000 person–years; Table 2). Sex- and age-specific analyses showed that the beneficial effects of treatment for PD increased with age, and occurred more frequently in men than in women. The adjusted HR also increased with age, and was higher in men in the treatment cohort than in their female counterparts. The treatment cohort to comparison cohort IRR of cancer was lower for those with diabetes (0.68; 95% CI = 0.60–0.77) and without hypertension (0.71; 95% CI = 0.69–0.74). However, the adjusted HR was higher for those with both diabetes and hypertension.

View this table:
Table 1

Comparisons of demographic characteristics and comorbidities between the PD with treatment cohort and the non-treatment comparison cohort

Periodontal diseaseP-value
No treatmentTreatment
(n = 77 804)(n = 38 902)
Age, mean ± SD (y)43.113.643.113.60.91
Stratify agen%n%
    20–3422 11028.410 95128.20.62
    35–4933 74443.416 97643.6
    50–6415 07519.4758719.5
    Women39 35250.619 67650.60.99
    Men38 45249.419 22649.4
    White-collara36 26346.624 18762.2<0.0001b
    Blue-collarc30 63739.4971525
    Othersd10 90414500012.9
  • SD, standard deviation

  • aWhite-collar: civil services, institution workers, enterprise, business and industrial administration personnel.

  • bChi-square test

  • cBlue-collar: farmers, fishermen, vendors and industrial laborers.

  • dOthers: retired, unemployed and low-income populations.

View this table:
Table 2

Incidence and adjusted HR of cancer stratified by sex, age and comorbidity, compared between the PD with treatment cohort and the non-treatment comparison cohort

Periodontal disease
No treatmentTreatment
EventPYRateaEventPYRateaIRRb (95% CI)Adjusted HRc (95% CI)
Cancer4183749 26155.81846446 53541.30.74 (0.72, 0.77)**0.72 (0.68, 0.76)**
    Female1810390 79046.3851227 66337.40.81 (0.77, 0.85)**1 (Reference)
    Male2373358 47066.2995218 87245.50.69 (0.65, 0.72)**1.28 (1.22, 1.35)**
Age (y)
    20–34213201 26110.6130130 8929.930.94 (0.87, 1.01)1 (Reference)
    35–491497345 80243.3670196 33934.10.79 (0.75, 1.83)**3.71 (3.31, 4.16)**
    50–641471147 28399.958883 13870.70.71 (0.66, 0.76)**8.21 (7.31, 9.23)**
    65+100254 914182.545836 166126.60.69 (0.63, 0.76)**14.5 (12.8, 16.4)**
    White-collar1514339 45244.61008279 84236.00.81 (0.77, 0.84)**1 (Reference)
    Blue-collar2069303 67168.1509110 63146.00.68 (0.63, 0.72)**1.03 (0.98, 1.10)
    Others600106 13756.532956 06258.71.04 (0.95, 1.13)0.99 (0.92, 1.07)
    No3818717 69853.21642420 60739.00.73 (0.71, 0.76)**1 (Reference)
    Yes36531 563115.620425 92878.70.68 (0.60, 0.77)**1.12 (1.02, 1.23)**
    No3251666 31148.81349388 17034.80.71 (0.69, 0.74)**1 (Reference)
    Yes93282 949112.449758 36585.20.76 (0.70, 0.82)**1.12 (1.05, 1.20)**
    No4089737 78755.41766434 71140.60.73 (0.71, 0.76)**1 (Reference)
    Yes9411 47481.98011 82467.70.83 (0.68, 1.01)0.94 (0.80, 1.10)
  • aRate, incidence rate, per 10 000 person–years.

  • bIRR, incidence rate ratio

  • cAdjusted HR, adjusted hazard ratio, adjusted for age, sex, occupation, type 2 diabetes mellitus, hypertension and hyperlipidemia.

  • *P < 0.05, **P < 0.01

Table 3 shows the specific analyses of cancer types. Patients in the PD with treatment cohort had significantly lower risks for malignant tumors in the gastrointestinal tract, the lungs, the female reproductive organs and the brain. In contrast, the male treatment cohort participants had a significantly higher risk of prostate cancer (HR = 2.11, 95% CI = 1.63–2.73). However, the male treatment cohort had a significantly higher percentage of participants that underwent prostate-specific antigen (PSA) testing than did the men in the comparison cohort, regardless of age, with 1.12% vs. 0.35% for 20–34 years, 7.95% vs. 2.93% for 35–49 years, 20.2% vs. 9.44% for 50–64 years and 39.1% vs. 18.7% for 65 years and older, respectively (Table 4). Treatment cohort participants also had a significantly higher risk of thyroid cancer (HR = 1.54, 95% CI = 1.09–2.09; Table 3), and a significantly higher percentage of treatment cohort participants underwent thyroid diagnostic procedures, compared with the comparison cohort (Table 5).

View this table:
Table 3

Incidence, incidence rate ratio and adjusted HR of subdivision cancers between the periodontal disease (PD) with treatment cohort and the non-treatment comparison cohort

Periodontal disease
No treatmentTreatment
Cancer (ICD-9-CM)EventRateaEventRateaIRRb (95% CI)Adjusted HRc (95% CI)
All418355.8184641.30.74 (0.72, 0.77)**0.72 (0.68, 0.76)**
Hematologic malignancy (200–208)1582.11821.840.87 (0.83, 0.91)**0.84 (0.64, 1.10)
Head and neck (140–149, 161)3905.212295.130.99 (0.95, 1.03)0.97 (0.82, 1.15)
Esophagus (150)1361.82170.380.21 (0.20, 0.22)**0.20 (0.12, 0.34)**
Stomach (151)2032.71621.390.51 (0.49, 0.54)**0.49 (0.37, 0.65)**
Small intestine (152)80.1150.111.05 (0.99, 1.11)1.15 (0.37, 3.53)
Colon/rectum (153–154)5267.022395.350.76 (0.73, 0.79)**0.70 (0.60, 0.82)**
Liver (155)6438.581854.140.48 (0.46, 0.51)**0.44 (0.38, 0.52)**
Pancreas (157)740.99280.630.63 (0.60, 0.67)**0.55 (0.35, 0.85)**
Lung (162)5417.221603.580.50 (0.47, 0.52)**0.45 (0.38, 0.54)**
Skin (173)460.61320.721.17 (1.11, 1.22)**1.05 (0.67, 1.66)
Female breast (174)4215.622736.111.09 (1.05, 1.13)**1.06 (0.90, 1.23)
Gynecology (180–184)3054.071072.400.59 (0.56, 0.62)**0.58 (0.46, 0.72)**
Prostate (185)981.311523.402.60 (2.49, 2.72)**2.11 (1.63, 2.73)**
Bladder (188)1251.67631.410.85 (0.81, 0.89)**0.75 (0.55, 1.01)
Kidney (189)911.21541.211.00 (0.95, 1.04)0.91 (0.65, 1.28)
Brain (191)520.69110.250.35 (0.33, 0.38)**0.35 (0.18, 0.67)**
Thyroid (193)650.87631.411.63 (1.56, 1.70)**1.54 (1.09, 2.09)*
Others3014.02841.880.47 (0.45, 0.49)**0.45 (0.35, 0.58)**
  • aRate, incidence rate, per 10 000 person–years.

  • bIRR, incidence rate ratio, rate of case group divided by the rate of comparison group.

  • cAdjusted HR, adjusted hazard ratio, adjusted for age, sex, occupation, type 2 diabetes mellitus, hypertension and hyperlipidemia.

  • *P < 0.05, **P < 0.01

View this table:
Table 4

The distribution of PSA between the PD with treatment cohort and the non-treatment comparison cohort

Periodontal diseaseP-value
VariableNo treatmentTreatment
Age (y)
  • PSA, prostate-specific antigen

  • aChi-square test

View this table:
Table 5

The distribution of thyroid diagnostic procedures between the PD with treatment cohort and the non-treatment comparison cohort

Periodontal diseaseP-value
VariableNo treatmentTreatment
Thyroid procedures
Age (y)
  • aChi-square test


Cancer has been the leading cause of death in the general population in Taiwan since 1982, and the age-adjusted incidence rate has increased steadily, reaching 276 new cases per 100 000 in 2008.11 However, different trends have been reported based on surveillance epidemiology and end results data, which showed that the overall incidence rates of cancer for all racial and ethnic groups combined decreased by 0.7% per year during 1999–2006.12 Because of such inconsistencies among the results of previous studies and the challenge that rising cancer rates pose to the public health care system, the government of Taiwan has encouraged further population-based investigations of cancer epidemiology and prevention. The Taiwan NHI record databases represent valuable resources for such populated-based studies. We previously used NHI data to evaluate the risk of cancer for patients with Parkinson’s disease, and found that Taiwanese patients with Parkinson’s disease have a lower risk of developing colorectal and lung cancers.13 Because PD is highly prevalent in Taiwan, any significant finding may have a significant impact on efforts to improve public health, and our current study used a similar study design to identify relationships between PD with treatment and the risks of different types of cancer.

We showed that PD with treatment significantly reduced the subsequent risks of certain types of cancer, including malignancies of the gastrointestinal tract, the lungs, the female reproductive organs and the brain. However, prostate and thyroid cancer risks were significantly higher in the PD with treatment group. A relationship between PD and an increased risk of cancer has been proposed in previous studies that investigated both overall and site-specific cancer rates.6 The scientific rationale behind the potential association is that inflammation is a major factor in both PD and cancer,14 and that an increase in systemic circulatory C-reactive protein and other inflammatory markers may contribute to such a relationship.15 This link is also supported by higher incidences of cancer in patients with chronic inflammatory conditions.7 The presence of inflammatory cells and mediators associated with tumors are key indicators of the progression of cancer,16 and associations between certain types of cancer and chronic inflammation have been shown for inflammatory bowel disease and colon cancer, hepatitis B and C infections and liver cancer, Heliobacter pylori-associated ulcers and gastric cancer, human papillomavirus infection and cervical cancer, Epstein–Barr virus infection and Hodgkin's lymphoma, and Burkitt's lymphoma and nasopharyngeal carcinoma.16,17

Previous studies on the association of PD and the overall risk of cancer have been relatively inconclusive, with one study reporting a small increase (odds ratio = 1.14) in the overall risk of cancer,18 and other investigators finding no significant association.19 Previous studies have shown that PD is linked to higher risks of oral cavity, esophagus, stomach, pancreas, lung, breast, kidney and hematopoietic malignancies.18 However, two studies suggested an inverse association between the risk of prostate cancer and the number of teeth lost.18,20 Our results showed that PD with treatment is associated with a significantly lower overall risk of cancer, and reduced risks for malignancies of the digestive system (esophagus, stomach, colon, liver and pancreas), the lungs, the female reproductive organs and the brain were observed separately. Most of these types of cancer were observed to be related to PD in earlier studies. Conversely, our results suggest that PD with routine treatment plays a protective role regarding the development of certain types of cancer. Studies have shown that subgingival scaling in patients with widespread periodontitis reduced inflammatory markers,21 and the efficacy of anti-inflammatory medications in preventing some cancers have been widely discussed,22–24 with most studies suggesting that regular use of aspirin and other non-steroidal anti-inflammatory drugs may reduce the long-term risk of colorectal, esophageal, gastric, biliary, breast and genitourinary cancers. Our results may also suggest that anti-inflammation therapies can effectively reduce cancer risks.

Recent researches connect PD-causing mouth bacteria to tumor growth in the colon and reveal possible treatments that may prevent colon cancer.25,26 Prior work in mouse models has shown that gut bacteria, which include species found in the mouth, can promote the development of colon tumors. A series of genetic studies on human colon biopsies revealed that one family of microbes—Fusobacteria—is present in healthy tissue adjacent to colorectal cancer. Although Fusobacteria start off in the mouth and are frequently associated with PDs, they can migrate through blood vessels to far reach of the colon. Finally, the strain of mouth bacteria that causes PD may have a more lethal health consequence of colon cancer. Because colon cancer is one the most common cancers in the world, therefore our findings based on a national-wide population-based study suggest this bacterial mouth-to-colon relationship is the public health manifestations within Taiwan and globalization.

We also observed significantly higher risks for prostate and thyroid cancers following PD with treatment. However, further analyses revealed significantly higher rates of PSA testing and thyroid diagnostic procedures in the PD with treatment cohort. More regular dental care may provide patients with increased health education, thus encouraging more frequent cancer screenings. The Taiwanese government provides free screening for prostate, cervical, breast and colorectal cancers for all patients with risk indicators, which may have contributed to the more frequent diagnosis of prostate cancer patients in our treatment cohort. Cervical cancer is the most common gynecological cancer in Taiwan,11 and more frequent Papanicolaou smear examinations may result in the identification of a greater number of cervical CIS/CIN/dysplasia cases. CIN and CIS lesions are known to precede invasive tumors in the progression of cervical cancer. Patients often receive treatment before the disease progresses to invasive cancer. However, we did not include CIN or CIS as a study end-point regarding cervical cancer, which may have otherwise led to a decrease in the relative number of cervical cancer cases in our treatment cohort, as well as the total number of gynecological cancer cases. Breast cancer also occurred more frequently in our treatment cohort, but the difference did not reach statistical significance. In contrast, colorectal cancer occurred at a significantly lower frequency, which we did not expect based on our assumption that the treatment cohort received more frequent routine health check-ups. Thus, for colorectal cancer, the anti-inflammatory effect of PD treatment may have overcome the screening effect. Likewise, the anti-inflammatory effects of PD treatment may have counteracted the effect of increased screening for oral cavity, head and neck cancers, which may partially explain the lack of a significant difference in the risks of cancers in these locations between the treatment and comparison cohorts. Regarding the higher incidence of thyroid cancer, however, it is likely that incidental observations of the head and neck regions during routine dental examinations of the oral cavity may have increased opportunities for more prominent neck masses, such as thyroid masses, to be detected.

The strength of our findings lies in the population-based study design, allowing for greater generalization of our findings. However, our study has several limitations. First, information on the lifestyle or behavior of patients is lacking in the NHI database, making it impossible to adjust for health-related factors, such as smoking and alcohol consumption. Second, there are clear evidences that less educated, lower income and in general lower social class patients with higher incidence rate for cancer. However, the above information is also lacking or incorrect in the NHI database. Third, the treatment fee for PD is only partially covered by the NHI, and out-of-pocket costs are high for some PD treatments, resulting in limited treatment options and subsequent negative outcomes for patients who are unable to pay. We attempted to control for this possible confounder using adjustments based on patient occupation. Finally, the evidence derived from a cohort study is generally of a lower methodological quality than that obtained from randomized trials because a cohort study design is subject to many biases related to adjustment for confounders. Despite our meticulous efforts for adequate control of confounding factors, biases may have remained because of immeasurable or unknown confounders. Nonetheless, the data we obtained on PD and cancer diagnoses were highly reliable.

In conclusion, our population based, retrospective cohort study indicated that PD with treatment may reduce the risks of certain types of cancers, as well as the overall risk of cancer. The potential anti-inflammatory effects of routine PD treatment may have contributed to these observations. Further studies are warranted to elucidate the relationships among cancer, PD and effects of PD treatment.


The study was supported in part by the study projects (DMR-103-018 and DMR-103-020) in our hospital; Taiwan Ministry of Health and Welfare Clinical Trial and Research Center for Excellence (DOH102-TD-B-111-004), Taiwan Ministry of Health and Welfare Cancer Research Center for Excellence (MOHW103-TD-B-111-03); and International Research-Intensive Centers of Excellence in Taiwan (I-RiCE) (NSC101-2911-I-002-303). The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

Conflict of interest: None declared.


Conception/design: I.-M. Hwang, L.-M. Sun and C.-H. Kao; provision of study material or patients: C.-F. Lee and C.-H. Kao; collection and/or assembly of data: C.-L. Lin; and data analysis and interpretation/manuscript writing/final approval of manuscript: all authors.


  • *I.-M. Hwang and L.-M. Sun are contributed equally for this work.


View Abstract