Diapositiva 1

1
EADPH 2009
14th Annual Meeting // September 3-5, 2009 //
Tromsø, Norway
Special Interest Working Group Prevention of
Oral Cancer
ALCOHOL, CARCINOGENIC OR CO-CARCINOGENIC?
Chair Professor Stefano Petti
2
BACKGROUND The IARC Monograph on the evaluation
of carcinogenic risks to humans number 96,
produced by a Working Group of 26 scientific
experts from 15 Countries in Lyon on February
6-13, 2007 updated the previous evaluation of
alcohol drinking undertaken in 1988. Their
conclusions regarding oral cancer were that
regular alcohol consumption is associated with an
increased oral cancer risk. Specifically, daily
consumption of 50g of ethanol (corresponding to 5
drinks) increases the risk for oral cancer by 2
to 3 times. The effect of drinking and smoking
seems to be multiplicative. Alcoholic beverages
and ethanol contained in alcoholic beverages are
classified as carcinogenic to humans. (Seitz HK,
Stickel F. Molecular mechanisms of
alcohol-mediated carcinogenesis. Nature Reviews
Cancer 2007) 3.6 of cancers worldwide derive
from chronic alcohol drinking, including oral
cancer. The main carcinogenic agent is
acetaldehyde, the first ethanol metabolite, due
to its multiple mutagenic effects on DNA.
Lifestyle factors such as smoking, poor oral
hygiene, and malnutrition also increase the risk
of alcohol-associated tumours. (Curado MP,
Hashibe M. Recent changes in the epidemiology of
head and neck cancer. Current Opinions in
Oncology 2009) Oral cancer incidence is
increasing in women and decreasing in men.
Chewing tobacco is a recognized risk factor. HPV
infection is a factor of good prognosis for oral
cavity and oropharynx cancer. The role of tobacco
smoking and alcohol in the genesis of this cancer
is reinforced. (Room T, Babor T, Rehm J. Alcohol
and Public Health. The Lancet 2005) 19 of
overall mouth and oropharynx cancers (22 among
men and 9 among women) are attributable to
alcohol worldwide.
3
BACKGROUND Most of the studies at the basis of
such systematic reviews are case-control or cross
sectional. The effect of smoking and drinking,
used as explanatory variables, on oral cancer has
been assessed by multivariate analyses, while the
multiplicative effect of smoking and drinking at
the same time, is generally obtained by the study
of interaction terms. The proportion of oral
cancer cases generally attributed to drinking and
smoking is as high as 80. However, the
proportions attributable to smoking and drinking
separately are usually set at 80 and 50,
respectively, thus cumulatively exceeding the
highest fraction of attributable cases (that is,
100) (see, for example, Negri E et al. Cancer
Epidemiology Biomarkers Prevention, 1993). This
result suggests that a large fraction of oral
cancer cases develop in individuals who smoke and
drink at the same time and that the relationship
between these lifestyle behaviours and oral
cancer development is more complex than expected.
In fact, a recent meta-analysis of pooled data
regarding the effect of drinking among non
smokers and vice versa reported that drinking
alone would not be associated to oral cancer
(Hashibe et al., Cancer Epidemiology Biomarkers
and prevention, 2009). Therefore, in order to
elucidate the association between alcohol
drinking and oral cancer, a meta-analysis of
pooled data of recent primary studies that
classify subjects according to the smoking AND
the drinking status at the same time, was made.
The two hypotheses on the table are alcohol
drinking is an independent risk factor for oral
cancer (that is, carcinogenic) in this case,
drinking will result associated with oral cancer
among non-smoking individuals. Alcohol acts
prevalently in synergy with smoking in this case
drinking will be associated with oral cancer
solely among smoking individuals.
4
MATERIALS AND METHODS Search of primary
studies MEDLINE search Language English or other
languages but with abstract in English and
available data. Matched terms (i) oral cancer,
mouth cancer, head and neck cancer (ii) alcohol,
drink, beverage, ethanol (iii) risk factor,
epidemiology, odds ratio. Publication Period
last ten years (from 1999 through 2009). Note
primary studies using the same samples and
studies with samples included in broader samples
(analyses of pooled data, systematic reviews,
meta-analyses) were excluded. Number of resulting
studies 342. Primary study selection Outcome
measure. Only studies reporting data regarding
ORAL CANCER, that is, tongue, gingiva, floor of
mouth, palate, buccal mucosa, retromolar area
(ICD-10, C01-06) were selected. Malignant tumours
of major salivary glands (C07-08), tonsils (C09)
and larynx (C12-14) were excluded. When possible,
lip cancer (C00), nasopharynx (C11), oropharynx
(C10), also were excluded. Exposure measure.
Studies providing aggregated information about
drinking and smoking, that is, smoking only
drinking only smoking and drinking neither
smoking nor drinking (INFO studies) were
eligible. Studies providing information about
drinking and smoking separately (NOINFO studies)
were used for the sensitivity analysis (see
later). Dichotomisations non/low (1 drink
daily) drinkers vs. former/high drinkers never
smokers vs. former/current smokers. Data from
all primary studies were pooled into a single
sample
5
Measures of the association between drinking and
oral cancer Odds Ratio (OR) for oral cancer with
95 confidence interval (95CI) among Non smoking
alcohol drinkers Non drinking smokers Smokers and
alcohol drinkers at the same time Relative to non
smokers/non drinkers. OR provides information
regarding the individual risk for oral
cancer. Population Attributable Fraction (PAF)
-also referred to as population attributable risk
or attributable risk- with 95CI. PAF provides
potentially valuable information regarding the
community-level effect of risk factors. The most
transparent approach for estimating a PAF across
multiple risk factors is to use a stratified
model based upon the possible combinations of
risk factors and PAF is estimated for each
combination. More explicitly, PAFDRINKING for the
effect of drinking only, PAFSMOKING for the
effect of smoking only, PAFDRINKINGSMOKING for
the effect of drinking and smoking at the same
time, PAFAGGREGATED for the overall effect of
smoking and drinking. Sensitivity
Analysis Effect of drinking obtained from INFO
studies treating data as if they were not split
for smoking (that is, drinking cases, drinking
controls, non drinking cases and non drinking
controls, independently of smoking status)
compared with effect of drinking obtained from
NOINFO studies. Proportion of exposed, OR and PAF
relative to alcohol drinking were calculated and
compared.
6

• RESULTS
• List of primary info studies
• 1st Author, year of publication, location and
  notes
• Znaor, 2003, India. Lip included, all chewers
  excluded.
• Zavras, 2001, Greece. Pharynx included.
• Subapriya, 2007, India. Lip included, all chewers
  excluded.
• Parker, 2006, US. Lip, oropharynx included, low
  drinkerslt21 drinks weekly.
• Muwonge, 2008, India. All chewers excluded.
• Hashibe, 2009. Multicenter study pooling data
  from 17 primary studies from EU and America,
  extracted from Table 3.
• Pooled Sample (subjects, cases, controls)
• Overall sample (23,919-4,008-19,911). Drinkers
  only (3,881-236-3,645). Smokers only
  (3,521-596-2,925). Drinkers and smokers
  (11,643-2,686-8,957). Non drinkers and non
  smokers (4,874-490-4,384)
• Proportion of drinkers only 16.23
• Proportion of smokers only 14.72
• Proportion of drinkers and smokers 48.68

7
TABLE 1 Pooled estimate of the individual
risk for oral cancer among drinkers only, smokers
only, smokers and drinkers at the same
time OR 95CI ES(lnOR) Drinkers
only 0.58 0.49-0.68 0.0823 Smokers
only 1.82 1.60-2.07 0.0655 Drinkers and
smokers 2.68 2.42-2.97 0.0525
statistically significant at 95 level
8
TABLE 2 Pooled estimate of the population
fraction of oral cancer cases attributable to
drinking only, smoking only, smoking and drinking
at the same time and smoking and/or drinking
(aggregated PAF) PAF low95CI high95CI PA
FDRINKING -3.65 -4.85 -2.51 PAFSMOKING
6.47 5.23 7.64 PAFDRINKINGSMOKING
43.76 40.73 46.49 PAFAGGREGATED
46.58 41.11 51.62
9
TABLE 3 Biased effect of drinking and
smoking EFFECT OF DRINKING NOT SPLIT FOR
SMOKING Proportion exposed 64.90 OR
(95CI) 1.56 (1.45-1.68) PAF (95CI) 23.32 (2
0.08-26.34) EFFECT OF SMOKING NOT SPLIT FOR
DRINKING Proportion exposed 63.40 OR
(95CI) 3.06 (2.81-3.33) PAF (95CI) 42.65 (4
0.80-44.34) EFFECT OF SMOKING AND
DRINKING PAFdrinking 23.32 (20.08-26.34)PAFsm
oking 42.65 (40.80-44.34)PAFdrinkingsmoking
65.97 (60.88-70.68)
10

• SENSITIVITY ANALYSIS
• List of NOINFO studies
• Anaya-Saavedra, 2008, Mexico.
• Pintos, 2008, Canada. Oropharynx included.
• Moreno-Lopez, 2000, Spain.
• Marques, 2008, Brazil. Pharynx included.
• Marchioni, 2007, Brazil. Oropharynx included.
• Llewellyn, 2004, UK. Agelt45yrs, low drinklt2AU
  females, lt3AU males.
• Guneri, 2005, Turkey.
• Balaram, 2002, India. Bidi smoking included in
  smoking.
• Applebaum, 2007, US.
• Pooled Sample (subjects, cases, controls)
• Overall sample (3,969-1,512-2,457)
• Drinkers (2,483-1,031-1,452). Non drinkers
  (1,486 -481-1,005)
• Smokers (2,727-1,102-1,625). Non smokers
  (1,242-410-832)

11
TABLE 4 Comparison of the estimated effects of
drinking, unadjusted for smoking, at individual
and population levels as obtained from NOINFO and
INFO studies 95CI ES(lnOR) NOINFO
STUDIES OR 1.48 1.30-1.70 0.0688 PAF 2
0.40 14.33-25.72 INFO STUDIES OR 1.56
1.45-1.68 PAF 23.32 20.08-26.34 statisti
cally significant at 95 level statistically
non significant at 95 level
12
BIOLOGICAL PLAUSIBILITY Synergistic carcinogenic
activity of ethanol and tobacco smoking
(1). Alcohol increases the permeability of human
oral mucosa, not acting as lipid solvent, but
disrupting the polar head groups of the lipid
bilayers thus facilitating the diffusion of polar
molecules including tobacco carcinogens, such as
N-nitrosonornicotine or benzo(a)pyrene. Synergist
ic carcinogenic activity of ethanol and tobacco
smoking (2). Chronic alcohol consumption leads to
an induction of cytochrome P-4502E1 (CYP2E1),
which metabolizes ethanol to acetaldehyde. This
cytochrome enzyme is also involved in the
metabolism of various xenobiotics, including
procarcinogens (nitrosamines, aflatoxin,
vinylchloride, polycyclic hydrocarbons,
hydrazines) to their ultimate carcinogens. For
example, the microsomal activation of
nitrosopyrrolidine, present in tobacco smoke, to
its ultimate carcinogen is significantly enhanced
in the oesophagus after alcohol ingestion in
rats. Glutathione S-transferases (GST)
polymorphism. Among smoking low drinking (1.5-3
drinks daily) individuals, generally at low oral
cancer risk, deletion of GST (enzymes directly
catalyzing the conjugation of reactive tobacco
and alcohol metabolites with glutathione) -a
condition present in 50 (GSTM1) and 20 (GSTT1)
of Caucasian population-, causes an increase in
oral cancer risk by 3-4 times.
13
COMMENTS The proportion of oral cancer cases
attributable to alcohol drinking and smoking is
confirmed high and close to 50. However, such
proportion is lower than the reported and
generally accepted value of 80. This result
reinforces the role of other oral cancer risk
factors, such as betel quid/areca nut chewing,
HPV infection, low fruit/vegetable consumption,
genetic predisposition, etc. and suggests that
the interaction between lifestyle, environmental
and genetic factors leading to oral cancer
development is very complex. In terms of oral
cancer prevention, this is discouraging result,
as it implies that effective population-based
primary prevention policies must focus on several
aspects and not merely to smoking and
drinking. Most oral cancer cases attributable to
alcohol drinking are due to the synergistic
effect of smoking and drinking. This result has
important consequences in designing oral cancer
awareness campaigns, as the goal of preventing
smoking and drinking at the same time could be
more easily achievable than alcohol banning or
quitting.