IJC

International Journal of Cancer

Periodontal diseases and risk of oral cancer in Southern India:

Results from the HeNCe Life study
Claudie Laprise1,2, Hameed Puthiyannal Shahul2, Sreenath Arekunnath Madathil2,3, Akhil Soman Thekkepurakkal2,
Geneviève Castonguay2, Ipe Varghese4, Shameena Shiraz5, Paul Allison1,2, Nicolas F. Schlecht6,
Marie-Claude Rousseau2,3, Eduardo L. Franco1,2 and Belinda Nicolau1,2
1
Department of Oncology, Division of Cancer Epidemiology, McGill University, Montreal, QC, Canada
2
Division of Oral Health and Society, Faculty of Dentistry, McGill University, Montreal, QC, Canada
3
Epidemiology and Biostatistics Unit, INRS-Institut Armand-Frappier, Laval, QC, Canada
4
Kerala University of Health Sciences, Medical College PO, Thrissur, Kerala, India
5
Oral Pathology, Government Dental College, Medical College Campus, Kozhikode 8, Kerala, India
6
Department of Epidemiology and Population Health, Albert Einstein College of Medicine, NY
Cancer Epidemiology

Some studies suggest that periodontal diseases increase the risk of oral cancer, but contradictory results also exist. Inad-
equate control of confounders, including life course exposures, may have influenced prior findings. We estimate the extent to
which high levels of periodontal diseases, measured by gingival inflammation and recession, are associated with oral cancer
risk using a comprehensive subset of potential confounders and applying a stringent adjustment approach. In a hospital-
based case-control study, incident oral cancer cases (N 5 350) were recruited from two major referral hospitals in Kerala,
South India, from 2008 to 2012. Controls (N 5 371), frequency-matched by age and sex, were recruited from clinics at the
same hospitals. Structured interviews collected information on several domains of exposure via a detailed life course ques-
tionnaire. Periodontal diseases, as measured by gingival inflammation and gingival recession, were evaluated visually by
qualified dentists following a detailed protocol. The relationship between periodontal diseases and oral cancer risk was
assessed by unconditional logistic regression using a stringent empirical selection of potential confounders corresponding to
a 1% change-in-estimates. Generalized gingival recession was significantly associated with oral cancer risk (Odds
Ratio 5 1.83, 95% Confidence Interval: 1.10–3.04). No significant association was observed between gingival inflammation
and oral cancer. Our findings support the hypothesis that high levels of periodontal diseases increase the risk of oral cancer.

Oral cancer is a major public health concern in India. It

Key words: case-control study, gingivitis, India, mouth neoplasms,
oral cancer, oral health, periodontal diseases, periodontitis
C.L. and H.P.S. contributed equally to this work
Grant sponsor: Canadian Institutes of Health Research; Grant
numbers: MOP 81172, MOP 111207; Grant sponsor: Ministère du
Developpement economique, de l’Innovation et de l’Exportation du
Quebec: Programme de soutien a la recherche (PSR), volet: Soutien
a des initiatives internationales de recherche et d’innovation (SIIRI);
Grant sponsor: Postdoctoral fellowship from the Canadian
Institutes of Health Research (C. Laprise); Grant number: FRN:
135506; Grant sponsor: Scholarship from the Fondation
Universitaire Armand-Frappier de l’INRS and a Psychosocial Oncol-
ogy Research Training (PORT) top-up award (S. Madathil)
DOI: 10.1002/ijc.30201
History: Received 16 Mar 2016; Accepted 2 May 2016; Online 23
May 2016
Correspondence to: Belinda Nicolau, Division of Oral Health and
Society, Faculty of Dentistry, McGill University, 2001 McGill College
Avenue, suite 527, Montreal, QC H3A 1G1, Canada, Tel.: 11-514-
398-7203x094655, Fax: 11-514-398-7220, E-mail: belinda.nicolau@
mcgill.ca
accounts for approximately 77,000 new cases and 52,000
deaths yearly, representing a quarter of all incident cases
worldwide. It is the 2nd and 5th most common cancer
among Indian men and women, respectively.1 Established
risk factors include tobacco smoking, alcohol consumption
and chewing, and specifically in India, paan chewing.2 Other
oral cancer risk factors are human papillomavirus (HPV)
infection, a diet low in nutritional value, low socioeconomic
position (SEP) and poor oral health and behavior.3–6 Indeed,
the association between oral health, more specifically peri-
odontal diseases, and oral cancer has been the subject of
increasing scientific interest.7,8
Periodontal diseases are characterized by chronic inflamma-
tion of the periodontium.9 Dental plaque induced gingivitis,
the mildest and most prevalent form of the disease, is a nondes-
tructive inflammation caused by a community of micro-
organisms found on the teeth as a biofilm.10 It is characterized
by clinical signs of inflammation confined to the gingiva. Gin-
givitis caused by dental plaque is reversible, but if untreated,
can evolve to affect deep tissues, leading to loss of supporting
connective tissue (ligament), alveolar bone and eventually to
tooth loss. This disease, known as periodontitis,9 generally

Int. J. Cancer: 139, 1512–1519 (2016) V

C 2016 UICC
Laprise et al.
What’s new?
Can gum disease predict the onset of oral cancer? As with many such questions, there’s data supporting both sides. A new
report comes down on the side of yes—periodontal disease is associated with risk of oral cancer. It’s a complex question,
because both gum disease and oral cancer share certain established risk factors, such as smoking. These authors applied
stringent criteria to correct for potential confounders. They evaluated periodontal disease by measuring inflammation and
recession of the gums and determined that gingival recession—but not inflammation—did significantly associate with oral can-
cer risk.
results from a complex interplay between bacterial infection
(e.g., composition of the biofilm) and the host response (e.g.,
genetics, immunity), often modified by behavioral factors (e.g.,
tobacco and alcohol consumption, poor oral hygiene).9
The link between inflammation and cancer has long been
suspected but the critical role that inflammation plays in
tumorigenesis has only recently been understood and some
of the underlying molecular mechanisms are still being inves-
tigated.11–13 For example, inflammation produced by both
bacterial and viral infections increases cancer risk and it has
been estimated that approximately 15% of incident cancers
worldwide are caused by persistent infections with specific
viral, bacterial and eukaryotic agents.14 Inflammatory bowel
diseases such as ulcerative colitis and Crohn’s disease are
associated with an increased risk of colorectal cancer.15
Chronic irritation and subsequent inflammation produced by
damaging environmental and lifestyle factors (e.g., exposure
to cigarette smoke, asbestos, silica) are also key factors in
carcinogenesis.16
Several authors have investigated whether periodontal dis-
eases may be a risk factor for oral cancer and their overall
results support this hypothesis.17–22 However, there are stud-
ies reporting modest or no significant associations.23,24 Diver-
gent findings could be due to a focus on limited numbers of
confounders that are not measured across the individual’s life
span.17 Although the above literature supports the biological
plausibility of an association between periodontal inflamma-
tion and oral cancer, the exact mechanism involved is still
unknown.17 Moreover, the incidence of periodontal diseases
and oral cancer is high in India25 and most of the oral cancer
is explained by paan chewing habits.26 The unraveling of
these complex associations is further complicated by the fact
that both diseases share common established risk factors
(e.g., smoking).27
In this study, we estimate the extent to which high levels
of periodontal diseases, measured by gingival inflammation
and recession, are associated with oral cancer risk in a sam-
ple of subjects living in Kerala, Southern India, using a com-
prehensive adjustment approach for confounding with a large
set of life course variables.
Material and Methods
Study population and data collection
The HeNCe Life study is an international hospital-based
case-control study investigating head and neck cancer aetiol-
Int. J. Cancer: 139, 1512–1519 (2016) V
C 2016 UICC
1513
ogy in three countries, Brazil, Canada and India. We followed
a rigorous research protocol including pilot studies at each
site to validate and adapt the instruments to each culture,
and applied several quality controls procedures throughout
the data collection. Ethical approval for the study was
Cancer Epidemiology
obtained from the IRBs of all institutions involved. Data to
test the current hypothesis were drawn from the Indian site,
for which recruitment was conducted from September 2008
to October 2012 in two major teaching hospitals: Govern-
ment Dental and Medical Colleges, Kerala, Southern India.
Specifically in India, the consent forms and research instru-
ments were translated using the back translation method into
the local language (English–Malayalam–English) and the den-
tists performing the data collection were native speakers. All
participants signed an informed consent form, and the study
protocol was approved by Research Ethics Offices at the Gov-
ernment Dental and Medical Colleges, India, and McGill
University, Montreal, Canada.
Incident cases of histologically confirmed oral squamous
cell carcinoma corresponding to International Classification
of Diseases (ICD)210 codes C00–C0628 were identified. Con-
trol subjects were recruited at eight outpatient clinics in the
same hospitals, were frequency-matched to cases by age (5-
year categories) and sex and presented with non-neoplastic
diseases unrelated to tobacco or alcohol. The eligibility crite-
ria were to be born in India, be an adult 18 years or older,
have no history of cancer, HIV infection or mental disorders,
and live within 150 km of the hospital area. The main socio-
demographic characteristics distribution in the control sub-
jects was similar to the findings of a recent survey conducted
in the source population (data not shown).29
Data collection, performed by three professionally trained
dentists, included a detailed oral clinical examination, oral
brush biopsy and mouthwash sampling for HPV and genetic
analysis, and a questionnaire using a life grid technique that
has been shown to improve recall.30 Individual semi-
structured interviews were conducted to collect information
on sociodemographic, environmental and behavioral factors
during three periods of the participants’ life (16 years, 17–
30 years and > 30 years).
Periodontal disease ascertainment
Periodontal health was assessed by experienced dentists in a
dental office located at the recruitment hospitals. The partici-
pants were seated in a semi-supine position on a dental chair
 1514
and the examination was performed with the help of a halo-
gen light and mouth mirror. Assessment followed a standard
sequence, starting from the upper right quadrant and finish-
ing in the lower right quadrant. First, gingiva was dried using
cotton swabs to avoid reflections from the salivary coating
from obscuring details. Further, both labial and lingual gingi-
val surfaces were assessed for color, size, contour, consis-
tency, surface texture, position and spontaneous bleeding.
The degree of gingival inflammation was recorded, from no
sign of inflammation to severe gingival inflammation, the latter
corresponding to the presence of marked redness and swelling.
Generalized and localized gingival recession were assessed vis-
ually by looking at the displacement of the free gingival margin
from the cement–enamel junction. This qualitative informa-
tion was then coded by two of the dentists (SHKP, SAM) and
Cancer Epidemiology
two indicators of periodontal diseases—presence of inflamma-
tion and of recession—were created and coded as follows:
none, mild, moderate and severe inflammation, and normal
attachment, localized, and generalized gingival recession. We
considered recession to be generalized when more than one
quadrant was involved. All the steps in the coding process,
including any disagreement between the two dentists who
coded the data, were discussed with the research team.
Statistical analysis
Univariate and multivariate unconditional logistic regression
analysis was used to assess the association of the two main
exposures (gingival inflammation and recession) with oral
cancer by estimating odds ratios (OR) and respective 95%
confidence intervals (CIs). Stata 13.1 (Stata Corp., College
Station, TX) was used for statistical analysis. Covariates
included lifetime history of paan chewing, bidi and cigarette
smoking, alcohol consumption, age, sex and years of educa-
tion.31 In brief, the lifetime consumption of paan was calcu-
lated for participants who reported chewing paan for at least
one year at any point in their lives, classified as ever chewers.
To obtain a comprehensive history of paan chewing, we
asked the subjects to describe their habit by dividing their
lifetime consumption into periods based on the homogeneity
of paan composition and frequency of use. For each period,
the following information was collected: i) age at start; ii) age
at end; iii) frequency of use (quids per day/month/year); iv)
minutes of use each time they chewed and v) paan composi-
tion. The duration of each chewing period was calculated as
the difference between age at beginning and end of the
period in years. Similarly to the calculation of pack-years in
the tobacco literature, we translated “duration” and
“frequency” of paan chewing into a lifetime paan exposure
measurement called “chew-years”. Cumulative lifetime expo-
sure was also computed for cigarette and bidi smoking (in
pack-years), as well as for alcohol consumption (in number
of standard drinks—an equivalent of 18 ml of pure ethanol—
per week).
First, we estimated the associations of the two main expo-
sures with oral cancer using a strategy of adjustment for risk
Periodontal diseases and oral cancer
factors known a priori, i.e., paan chewing, smoking, alcohol
consumption, age, sex and years of education, to control for
their confounding effects. In addition, for biological plausibil-
ity we further tested these associations by restricting the anal-
ysis to the anatomical sub-sites gum and buccal mucosa,
which are directly affected by periodontal diseases.
Second, because residual confounding is a major issue in
assessing the association between periodontal diseases and
oral cancer, we used an extensive adjustment approach for
confounding using all variables available in the HeNCe Life
database. Our assumption was that there could be antece-
dents to periodontal diseases that could have also mediated
oral cancer risk, and thus their distal influence would have to
be controlled for. We retained variables as empirical con-
founders if their inclusion in the model caused a change of
1% or larger in either direction (above or below the pread-
justed estimate) in the estimate for the measure of association
(i.e., the OR) between the main exposure (severe gingival
inflammation or generalized gingival recession) and oral can-
cer. This cut-off was arbitrarily defined in the interest of con-
servatism, to select all variables that were even mild
empirical confounders or mediators of the association
between the periodontal health indicators and oral cancer.
Demographic and socioeconomic characteristics (e.g., educa-
tion, occupation, housing conditions, employment history),
behavioral factors (e.g., smoking, chewing, drinking, diet, sex-
ual habits), family environment and social support (e.g.,
parents–participant relationships, marriage) at three stages of
the individual’s life (childhood, early adulthood and late
adulthood) as well as family history of cancer and oral health
status were considered candidate empirical confounders. Var-
iables were modeled in a continuous, ordinal or categorical
form as appropriate.
Results
A total of 426 oral cancer cases and 865 control subjects
were invited to participate in the HeNCe Life study’s Indian
site and 350 and 371 accepted the invitation, representing a
response rate of 82.2 and 42.9%, respectively. A further 87
patients were excluded: 6 because they were oropharynx can-
cer cases, 80 because of an edentulous condition and 1 due
to lack of information on dental clinical examination
(unavailable periodontal disease assessment).
Table 1 presents the distribution of cases (N 5 306) and
controls (N 5 328) according to selected characteristics. The
most common oral cancer sub-site was buccal mucosa (men:
38.2%, women: 42.1%), followed by gum (men: 23.7%,
women: 36.1%), tongue (men: 19.7%, women: 13.5%) and
floor of the mouth (men: 9.3%, women: 1.5%). Cases tended
to have a lower level of education and to consume fewer
fruits and vegetables than controls. As expected, the fre-
quency and duration of bidi smoking, alcohol drinking and
paan chewing was higher among cases. Regarding oral health
variables, the proportions of individuals who brushed their
teeth less than twice a day, had severe gingival inflammation
Int. J. Cancer: 139, 1512–1519 (2016) V
C 2016 UICC
Laprise et al. 1515

Table 1. Sociodemographic, oral health, lifestyle and tumor characteristics of study participants
Male Female
Cases (N 5 173) Controls (N 5 182) Cases (N 5 133) Controls (N 5 146)
Variables N % N % N % N %
Age
Mean (SD) 60.0 (10.8) 59.2 (11.3) 59.8 (11.5) 59.9 (12.1)
Religion
Hindu 109 63.0 103 56.6 93 69.9 98 67.1
Non Hindu 64 37.0 79 43.4 40 30.1 48 32.9
Educational level1
Low 112 64.7 80 44.0 123 92.5 77 52.7
High 61 35.3 102 56.0 10 7.5 69 47.3
Gingival inflammation

Cancer Epidemiology
No 85 49.1 126 69.2 63 47.4 97 66.4
Mild/moderate 61 35.3 44 24.2 47 35.3 42 28.8
Severe 27 15.6 12 6.6 23 17.3 7 4.8
Gingival recession
No 92 53.2 122 67.0 68 51.1 103 70.6
Local 8 4.6 17 9.3 6 4.5 12 8.2
Generalized 73 42.2 43 23.6 59 44.4 31 21.2
Missing teeth
Median (IQR) 6 (2–13) 3.5 (1–10) 8 (3–17) 6 (3–11)
Brushing teeth frequency
Twice a day 34 19.7 90 49.5 35 26.3 74 50.7
Once a day 139 81.3 92 50.5 98 73.7 72 49.3
Dentist visits
Never 25 14.5 17 9.3 27 20.3 19 13.0
Only when painful 134 77.5 148 81.3 96 72.2 112 76.7
Regularly 14 8.1 17 9.3 10 7.5 15 10.3
HPV prevalence 0 0 0 0
Smoking bidi
Mean pack-years (SD) 15.4 (21.4) 8.2 (19.7) 0.12 (1.0) 0.10 (0.9)
Smoking cigarettes
Mean pack-years (SD) 10.3 (18.6) 17.7 (31.2) 0.02 (0.2) 0.8 (93.0)
Paan chewing
Mean chew-years (SD) 168.7 (207.3) 22.3 (83.7) 232.9 (200.7) 54.2 (148.4)
Alcohol drinking
Mean drink/week (SD) 12.2 (47.2) 6.5 (53.1) 0.1 (0.6) 0.1 (0.8)
Vegetable consumption
Mean portion/week (SD) 12.6 (3.9) 12.6 (4.2) 11.9 (3.6) 12.8 (3.9)
Fruit consumption
Mean portion/week (SD) 2.7 (1.8) 3.1 (2.3) 2.3 (0.9) 2.7 (1.6)
Tumor site (ICD-10 codes)
Tongue (C02) 34 19.7 18 13.5
Floor of the mouth (C04) 16 9.3 2 1.5
Gum (C03) 41 23.7 48 36.1

Int. J. Cancer: 139, 1512–1519 (2016) V

C 2016 UICC
 1516 Periodontal diseases and oral cancer

Table 1. Sociodemographic, oral health, lifestyle and tumor characteristics of study participants (Continued)
Male Female
Cases (N 5 173) Controls (N 5 182) Cases (N 5 133) Controls (N 5 146)
Variables N % N % N % N %
Buccal mucosa (C06) 66 38.2 56 42.1
Other2 16 9.3 9 6.8
Global TNM stage
Stage I 18 10.4 13 9.8
Stage II 14 8.1 14 10.5
Stage III 90 52.0 64 48.1
Stage IV 51 29.5 42 31.6
1
Number of educational years adjusted for the time period of formal education (before and after 1950).
2
Oral cancer lesions spanning more than one anatomical sub-site.
Cancer Epidemiology

Abbreviations: HPV: human papillomavirus; ICD: International Classification of Diseases; IQR: interquartile range; SD: standard deviation; TNM:
tumor node metastasis.

Table 2. Odds ratio (OR) for the associations of inflammation and recession with oral cancer stratified by cancer site
OR (95%CI)
Gingiva and buccal mucosa
All oral sites cancer sites only
Periodontal health indicators Crude Multivariate1 Crude Multivariate1
Gingival inflammation
No 1.00 1.00 1.00 1.00
Mild/moderate 1.89 (1.33–2.69) 1.45 (0.96–2.20) 2.01 (1.36) 1.46 (0.90–2.38)
Severe 3.97 (2.25–7.00) 2.28 (1.18–4.38) 4.43 (2.43–2.97) 2.97 (1.45–6.10)
Gingival recession
No/localized 1.00 1.00 1.00 1.00
Generalized 2.60 (1.85–3.67) 1.74 (1.15–2.62) 3.09 (2.13–4.50) 1.91 (1.20–3.03)
1
Adjusted for paan chewing, bidi and cigarette smoking, alcohol consumption, age, gender and number of educational years.

and generalized recession were higher among the cases com-
pared to controls. Cases also had a higher mean number of
missing teeth than controls. Extensive molecular testing for
HPV DNA yielded no positive results among both cases and
controls, irrespective of oral specimen type.32
Table 2 summarizes ORs for the associations of gingival
inflammation and recession with oral cancer sites combined
and for gingival and buccal mucosa cancers adjusted for
paan chewing, smoking, alcohol consumption, age, sex and
years of education. There was a statistically significant
increased risk of oral cancer (OR 5 2.28, 95%CI: 1.18–4.38)
and gingival and buccal mucosa cancers (OR 5 2.97; 95%CI:
1.45–6.10) with severe levels of inflammation. Similarly, gen-
eralized recession was associated with an increased risk of
oral cancer (OR 5 1.74, 95%CI: 1.15–2.62) and gingival and
buccal mucosa cancers (OR 5 1.91, 95%CI: 1.20–3.03).
A similar pattern of associations was observed after
adjustment for all empirical confounders and using the con-
servative 1% change-in-estimate cut-off (Table 3). The risk of
oral cancer was consistently higher among those with gener-
alized recession (after stringent adjustment: OR 5 1.83,
95%CI: 1.10–3.04). Similarly, severe gingival inflammation
seems to increase the risk of oral cancer when adjusting for
known risk factors (OR 5 1.99, 95%CI: 1.05–3.78). Although
in the direction of an increased risk, this association was no
longer statistically significant after stringent adjustment for
confounding (OR5 2.02, 95%CI: 0.90–4.55).
Discussion
In this study, we investigated the association between peri-
odontal diseases and oral cancer risk. We observed that gen-
eralized gingival recession, an indicator of past history of
periodontal diseases, is associated with nearly a doubling in
oral cancer risk after exhaustively controlling for a compre-
hensive list of potential confounders. Our results for general-
ized recession are in accordance with a recent meta-analysis
that reported a pooled OR adjusted for known risk factors
(mostly smoking and alcohol consumption) for the associa-
tion between periodontal diseases and head and neck cancer
of 2.23 (95%CI: 1.46–3.43).20
The relationship between periodontal diseases and oral
cancer is not well understood, owing to the intricacy of the

Int. J. Cancer: 139, 1512–1519 (2016) V

C 2016 UICC
Laprise et al. 1517

Table 3. Odds ratio (OR) for the associations of severe inflammation and generalized gingival recession with oral cancer after extensive
adjustment approach for confounding
Odds ratio (95%CI)
Periodontal health indicators Crude OR Adjusted OR1 Adjusted OR2
Gingival inflammation
No/moderate 1.00 (ref.) 1.00 (ref.) 1.00 (ref.)
Severe 3.18 (1.83–5.52) 1.99 (1.05–3.78) 2.02 (0.90–4.55)3
Gingival recession
No/localized 1.00 (ref.) 1.00 (ref.) 1.00 (ref.)
Generalized 2.60 (1.85–3.67) 1.74 (1.15–2.62) 1.83 (1.10–3.04)4
1
Adjusted for paan chewing, bidi and cigarette smoking, alcohol consumption, age, gender, # educational years.
2
Adjusted for all empirical confounders changing estimates by 61%.
3
Adjusted for paan chewing, bidi and cigarette smoking, alcohol consumption, # educational years, living area (CH and AH), job employment, # sib-
lings, civil status, # children, family support, FHC, LT housing conditions; fruits consumption, # meals/day (CH and AH), chile, red chile, coriander,

Cancer Epidemiology
fish, milk, pulses, turmeric, ginger, coffee and tea consumption (AH), fish, milk sugar and tuber consumption (CH), height (CH), weight, PA at work,
home and as leisure, LT measles, mumps, whocought, nose/throat diseases, diabetes and buccal candida, no consensual sex and oral sex practice.
4
Adjusted for paan chewing, bidi smoking, alcohol consumption, age, # educational years, living area, address stability, religion, civil status, # sib-
lings, # children, FHC, LT housing conditions, fruits consumption, # meals/day (CH and AH), red chile, ginger and tea consumption; fish, sugar oil
and tuber consumption (CH), milk consumption (AH), cooking oil, height (CH and AH), weight, PA at work, LT measles, mumps, chicken pox, who
cough, hepatitis, asthma, nose/throat diseases, depression, diabetes, buccal candida and oral sex practice.
Abbreviations: #: number of; AH: adulthood; CH: childhood; FHC: family history of cancer; LT: lifetime; PA: physical activity; SEP: socioeconomic
position.

process by which chronic infections and inflammation may
result in carcinogenesis. The main hypothesis to explain this
association centers on the oxidative damage to cellular DNA
caused by inflammation as the host immune system interacts
with infectious agents, which results in permanent genomic
alterations (e.g., point mutations, deletions or rearrange-
ments). Indeed, the frequency of p53 mutations is similar in
tumors and in other chronic inflammatory diseases (e.g.,
rheumatoid arthritis and inflammatory bowel disease).33 Our
results and prior knowledge on chronic inflammation and
risk of cancer support the hypothesis that periodontal dis-
eases are a risk factor for oral cancer. Conceivably, as one
could hypothesize that periodontal diseases and oral cancer
could both be independent outcomes of the same risk factors,
we decided to control for a large number of potential con-
founders and mediators. Because the association between
periodontal diseases and oral cancer remained even after
exhaustively controlling for these variables, we concluded
that the association is real, and thus represents a potential
target for oral cancer prevention.
Our study has some methodological limitations that need
to be considered. First, to attain validity our adjusted regres-
sion models included a considerable number of covariates
that may have resulted in a lower precision of estimates, thus
adding even more conservatism to our findings. Even with
this overly conservative estimation, however, the persistent
associations of generalized recession and severe inflammation
with gingival and bucco-mucosal cancer suggest that peri-
odontal diseases increase the risk of oral cancer. Also, we did
not consider oral health variables for empirical control
(brushing frequency, dental floss use and dentist visit). We
suspected that they were distal variables in the causal path-
way from periodontal diseases to oral cancer. Indeed,
although poor oral hygiene is a risk factor for oral cancer, we
postulated that periodontal diseases were a consequence of it.
We carried out sensitivity analyses by adding oral health vari-
ables in all final models; no important change in estimates
was observed and results remained significant (data not
shown).
Another concern could be exposure misclassification. Our
assessment of periodontal diseases was visual. We could not
collect data on pocket depth, clinical attachment loss and
alveolar bone height on radiographs, which are the gold
standard measurements for periodontal epidemiological
research. We observed in our pilot study that patients could
not endure a complete dental exam; most of them were frail,
had large lesions in their mouth, and all had to go through a
1-hr interview. Therefore, our decision to choose a visual
assessment of periodontal diseases, a less invasive and painful
method than others (e.g., pocket depth, loss of attachment),
was on ethical grounds. However, there is evidence that self-
reported periodontal disease correlates well with the objective
measure of disease.34–36 In addition, the assessment of peri-
odontal disease was performed by trained dentists in a dental
office using a halogen lamp and a dental mirror. Further-
more, our results were stronger for generalized recession,
which is easily visualized and thus less likely to be misclassi-
fied by examiners than inflammation. Indeed, gingival
inflammation may be transitory, whereas clinical attachment
loss, of which recession is a large part, is among the most
common clinical measures of periodontal diseases37–39
because it is a valid representation of a person’s history of
periodontitis.39,40 Therefore, we believe that our measures
adequately represent the periodontal status of the partici-
pants. Finally, we cannot rule out the possibility of reverse
causality in the association between periodontal diseases and

Int. J. Cancer: 139, 1512–1519 (2016) V

C 2016 UICC
 1518 Periodontal diseases and oral cancer

oral cancer. However, the positive associations observed
between oral cancers and generalized gingival recession, a
marker of past history of periodontal diseases, make this
explanation less plausible.
The scope and quality control measures of this study are
significant strengths. Cases consisted of histologically con-
firmed primary incident cancers and the sample was relatively
large for this disease. We collected data on several domains of
exposures along the individual life span and used the life grid
technique to reduce measurement errors. In comparison, the
majority of prior evidence is limited by small sample sizes and
lack of control for confounding. In our study, we were able to
comprehensively adjust for several exposures throughout the
individual’s entire life span, which dealt with any confounding
in the association between periodontal diseases and oral cancer
Cancer Epidemiology
associations. To our knowledge, this is the first study using var-
iables informative of several exposures over the life span and a
stringent change-in-estimate strategy, resulting in a very strict
control of confounding, reinforcing the idea that periodontal
diseases are significantly associated with oral cancer. In addi-
tion, we presented new evidence regarding oral cancer sub-
sites and compared the risk across sites within the oral cavity.
More importantly, our findings add information in a field of
research where the available evidence is controversial. The fact
that in this population HPV infection is not a risk factor for
oral cancer (none of our cases and controls was HPV positive
via a sensitive assay in multiple oral samples) obviates concerns
that any possible effects of inflammation would imply the
interplay of HPV effects on carcinogenesis.

In conclusion, our results suggest that periodontal diseases

from variables that we measured in the study. The selection of are a genuine and independent risk factor for the occurrence
covariates to control in nonexperimental studies is usually of oral cancers in a sample of subjects from Kerala, in South-
based on the understanding of the relationship between the
ern India. India is a developing country and access to health
exposure and the outcome, or empirically on statistical associa-
care and awareness of proper oral care are limited in most
tions.41,42 The use of a priori knowledge and causal graph
parts of the country. Our study reveals the importance of
theory is strongly recommended when there is sufficient
implementing various primary preventive measures at a pop-
knowledge of the causal structure of the variables, but the
ulation level. Proper health education, regular oral prophy-
causal pathway between periodontal diseases and cancer is still
laxis and the promotion of research can help to tackle oral
not fully understood. The strategy we employed can be useful
cancer. Above all, in light of previous and current findings,
to identify new confounders.42 There is controversy on how we
periodontal diseases should be considered as possible risk fac-
should control for confounding in epidemiological studies and
tors for oral cancer.
many strategies exist. One strategy often used to identify con-
founders is an empirical one using a change-in-estimate
method criterion43,44 with a cut-off at 10%. We opted for a
Acknowledgements
more stringent cut-off to minimize the possibility of residual We would like to thank our collaborators at the Indian site and the HeNCe
confounding exerted by variables that could independently Life study team. B. Nicolau holds a Canada Research Chair in Life Course
exert little confounding but collectively could distort the target Oral Epidemiology.

References
1. Ferlay JSI, Ervik M, Dikshit R, et al. , Cancer
incidence and mortality worldwide: IARC Can-
cerBase No. 11 [Internet], Version 1.0. Lyon,
France: International Agency for Research on
Cancer, 2013. Available at: http://globocan.iarc.fr.
Accessed on January 26, 2016.
2. Gupta B, Ariyawardana A, Johnson NW. Oral
cancer in India continues in epidemic propor-
tions: evidence base and policy initiatives. Int
Dent J 2013;63:12–25.
3. Chaturvedi AK. Epidemiology and clinical aspects
of HPV in head and neck cancers. Head Neck
Pathol 2012;6:S16–24.
4. Tezal M, Sullivan MA, Hyland A, et al.
Chronic periodontitis and the incidence of
head and neck squamous cell carcinoma. Can-
cer Epidemiol Biomarkers Prevent 2009;18:
2406–12.
5. Krishna Rao SV, Mejia G, Roberts-Thomson K,
et al. Epidemiology of oral cancer in Asia in the
past decade—an update (2000–2012). Asian Pac J
Cancer Prev 2013;14:5567–77.
6. Ahrens W, Pohlabeln H, Foraita R, et al. Oral
health, dental care and mouthwash associated
with upper aerodigestive tract cancer risk in
Europe: the ARCAGE study. Oral Oncol 2014;50:
616–25.
7. Galvao-Moreira LV, da Cruz MC. Oral micro-
biome, periodontitis and risk of head and neck
cancer. Oral Oncol 2016;53:17–9.
8. Atanasova KR, Yilmaz O. Looking in the Por-
phyromonas gingivalis cabinet of curiosities: the
microbium, the host and cancer association. Mol
Oral Microbiol 2014;29:55–66.
9. Pihlstrom BL, Michalowicz BS, Johnson NW.
Periodontal diseases. Lancet 2005;366:1809–20.
10. Marsh PD. Dental plaque as a microbial biofilm.
Caries Res 2004;38:204–11.
11. Coussens LM, Werb Z. Inflammation and cancer.
Nature 2002;420:860–7.
12. Grivennikov SI, Greten FR, Karin M. Immunity,
inflammation, and cancer. Cell 2010;140:883–99.
13. Coussens LM, Zitvogel L, Palucka AK. Neutraliz-
ing tumor-promoting chronic inflammation: a
magic bullet? Science 2013;339:286–91.
14. Pisani P, Parkin DM, Munoz N, et al. Cancer
and infection: estimates of the attributable frac-
tion in 1990. Cancer Epidemiol Biomarkers Pre-
vent 1997;6:387–400.
15. Ekbom A, Helmick C, Zack M, et al. Increased
risk of large-bowel cancer in Crohn’s disease with
colonic involvement. Lancet 1990;336:357–9.
16. Balkwill F, Mantovani A. Inflammation and can-
cer: back to Virchow? Lancet 2001;357:539–45.
17. Meyer MS, Joshipura K, Giovannucci E, et al. A
review of the relationship between tooth loss,
periodontal disease, and cancer. Cancer Causes
Control 2008;19:895–907.
18. Javed F, Warnakulasuriya S. Is there a relation-
ship between periodontal disease and oral cancer?
A systematic review of currently available evi-
dence. Crit Rev Oncol Hematol 2015;
19. Yao QW, Zhou DS, Peng HJ, et al. Association of
periodontal disease with oral cancer: a meta-anal-
ysis. Tumour Biol 2014;35:7073
20. Zeng XT, Deng AP, Li C, et al. Periodontal dis-
ease and risk of head and neck cancer: a meta-
analysis of observational studies. PloS One 2013;8:
e79017
21. Michaud DS, Kelsey KT, Papathanasiou E, Genco
CA, Giovannucci E. Periodontal disease and risk
of all cancers among male never smokers: an
updated analysis of the Health Professionals Fol-
low-up Study. Ann Oncol 2016;27:941–7.
22. Fitzpatrick SG, Katz J. The association between
periodontal disease and cancer: a review of the
literature. J Dent 2010;38:83–95.
23. Divaris K, Olshan AF, Smith J, et al. Oral health
and risk for head and neck squamous cell carci-
noma: the Carolina head and neck cancer study.
Cancer Causes Control 2010;21:567–75.

Int. J. Cancer: 139, 1512–1519 (2016) V

C 2016 UICC
Laprise et al.
24. 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;9:550–8.
25. Agarwal V, Khatri M, Singh G, et al. Prevalence
of periodontal diseases in India. J Oral Health
Commun Dent 2010;4(Spl):7–16.
26. Sankaranarayanan R, Duffy SW,
Padmakumary G, et al. Risk factors for can-
cer of the buccal and labial mucosa in Ker-
ala, southern India. J Epidemiol Community
Health 1990;44:286–92.
27. Sheiham A, Watt RG. The common risk factor
approach: a rational basis for promoting oral health.
Community Dent Oral Epidemiol 2000;28:399–406.
28. World Healh Organization. Malignant neoplasms
of lip, oral cavity and pharynx (C00-C14) ICD-
10, 2010. Available at: http://apps.who.int/classifi-
cations/icd10/browse/2010/en#/C00-C14.
Accessed on December 10, 2014.
29. IIPS. International Institute for population
sciences. Global Adult Tobacco Survey India
2009–2010. Available at: http://www.who.int/
tobacco/surveillance/survey/gats/gats_india_
report.pdf. Accessed on February 27, 2015.
30. Blane DB. Collecting retrospective data: develop-
ment of a reliable method and a pilot study of its
use. Soc Sci Med 1996;42:751–7.
Int. J. Cancer: 139, 1512–1519 (2016) V
C 2016 UICC
31. Madathil SA, Rousseau MC, Allison P, et al. Mater-
nal and paternal contribution to intergenerational
psychosocial transmission of paan chewing. Com-
munity Dent Oral Epidemiol 2015;43:289–97.
32. Laprise C, Madathil SA, Allison P, Abraham
P, Raghavendran A, Shahul HP,
ThekkePurakkal AS, Castonguay G, Coutlee F,
Schlecht NF, Rousseau MC, Franco EL, et al.
No role for human papillomavirus infection in
oral cancers in a region in southern India. Int
J Cancer J Int du Cancer 2016;138:912–7.
33. Yamanishi Y, Boyle DL, Rosengren S, et al.
Regional analysis of p53 mutations in rheumatoid
arthritis synovium. Proc Natl Acad Sci U S A
2002;99:10025–30.
34. Blicher B, Joshipura K, Eke P. Validation of self-
reported periodontal disease: a systematic review.
J Dent Res 2005;84:881–90.
35. Dietrich T, Stosch U, Dietrich D, et al. Prediction
of periodontal disease from multiple self-reported
items in a German practice-based sample.
J Periodontol 2007;78:1421–8.
36. Eke PI, Dye BA, Wei L, et al. Self-reported measures
for surveillance of periodontitis. J Dent Res 2013;92:
1041–7.
37. Goodson JM, Haffajee AD, Socransky SS. The rela-
tionship between attachment level loss and alveolar
bone loss. J Clin Periodontol 1984;11:348–59.
1519
38. Yoneyama T, Okamoto H, Lindhe J, et al. Prob-
ing depth, attachment loss and gingival recession.
J Clin Periodontol 1988;15:581–91.
39. Machtei EE, Dunford R, Hausmann E, et al. Lon-
gitudinal study of prognostic factors in estab-
lished periodontitis patients. J Clin Periodontol
1997;24:102–9.
40. Armitage GC. Periodontal diagnoses and classifi-
cation of periodontal diseases. Periodontology
2004;34:9–21.
41. Pourhoseingholi MA, Baghestani AR, Vahedi M.
How to control confounding effects by statistical
analysis. Gastroenterol Hepatol Bed Bench 2012;5:
79–83.
42. Velentgas P, Dreyer NA, Nourjah P, Smith SR,
Torchia MM, eds. Developing a Protocol for
Observational Comparative Effectiveness
Research: A User’s Guide. AHRQ Publication No.
12(13)-EHC099. Rockville, MD: Agency for
Cancer Epidemiology
Healthcare Research and Quality; January 2013.
www.effectivehealthcare.ahrq.gov/Methods-OCER.
cfm.
43. Mickey RM, Greenland S. The impact of
confounder selection criteria on effect
estimation. Am J Epidemiol 1989;129:125–37.
44. McNamee R. Regression modelling and other
methods to control confounding. Occup Environ
Med 2005;62:500–6, 472.