SEARCH

SEARCH BY CITATION

Keywords:

  • tobacco;
  • alcohol;
  • oral cavity;
  • oropharynx;
  • neoplasia

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Incidence rates of oral and oropharyngeal cancers (oral cancer) in Spain are among the highest in Europe. Spain has a population heavily exposed to various types of tobacco and alcoholic beverages but the role and impact of tobacco type and beverage type in oral carcinogenesis remain controversial. To estimate the independent and joint effects of tobacco smoking and alcohol drinking habits on the risk of developing oral cancer, we carried out a multicenter, hospital-based, case-control study in Spain. Data from 375 patients newly diagnosed with cancer of the oral cavity or oropharynx and 375 matched control subjects were analyzed using multivariate logistic regression procedures. All exposure characteristics of amount, duration and cessation of both tobacco smoking and alcohol drinking were strongly associated with cancer risk following a dose-dependent relationship. At equal intake or duration levels, black-tobacco smoking and drinking of spirits were both associated with a 2- to 4-fold increase in cancer risk compared to blond tobacco smoking or drinking of wine or beer, respectively. While ever exposure to smoking only or drinking only was associated with a moderate and nonsignificant increase in cancer risk, a history of simultaneous exposure to both habits was associated with a 13-fold increase that was compatible with a synergistic effect model (p-value for interaction: 0.008). Exposure to black tobacco smoking and/or drinking of spirits may account for up to 77% of oral cancer occurrence in Spain. Both black tobacco smoking and drinking of spirits place individuals at a very high risk of developing oral cancer. Simultaneous exposure to tobacco and alcohol consumption increases oral cancer risk in a synergistic fashion, even when consumption levels are moderate. These results underline the importance of type of tobacco and alcohol concentration in oral carcinogenesis. © 2003 Wiley-Liss, Inc.

Cancer of the oral cavity and pharynx (oral cancer) is a major and increasing public health problem in Europe. Spain, with a European age-adjusted incidence rate of 31.4 per 100,000 men per year,1 is, after France, the European country with the second highest incidence rate for these cancer sites.

These tumors are prominent among middle-aged adults, and increasing rates are occurring mainly among young and middle-aged men. Population-based screening and early detection programs are rare or nonexistent, and in spite of surgical advances, these cancers remain a disfiguring disease associated with a relatively low survival rate.2, 3

Tobacco smoking and alcohol drinking have long and consistently been identified as the 2 major risk factors for oral cancer in most Western populations.4, 5, 6 Other risk factors that play a role include diets low in fruits and vegetables,7, 8 family history of cancer9, 10 and, to a lesser extent, infection by oncogenic human papillomaviruses (HPV).11, 12, 13

Despite the clear role of tobacco and alcohol consumption in the etiology of oral cancer, these associations have never been assessed in Spain. We hypothesized that highly prevalent black-tobacco smoking and widespread drinking of spirits with a high ethanol content explain the high incidence rates of oral cancer in Spain. To explore the role and impact of tobacco type and beverage type in oral carcinogenesis, we analyzed data from a large, multicenter, hospital-based, case-control study of oral cancer.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The present research was part of an international study on oral cancer coordinated by the International Agency for Research on Cancer (IARC, Lyon, France) and carried out simultaneously in Italy, Spain, Ireland, Poland, Cuba, Canada, India, Sudan and Australia. The aim of the international component was mainly to explore the etiologic role of HPV infection in these tumors, as well as that of other established and potential risk factors such as dietary patterns, oral hygiene, dentition, sexual habits, family history of cancer and the role of several biomarkers of cancer risk susceptibility.14 Partial results from Spain,15, 16 Italy,8, 17 Cuba18 and India19 have already been published.

In Spain, subjects were recruited between November 1996 and July 1999 and they were identified in 4 hospitals: one in Granada, one in Sevilla and 2 in Barcelona.

Research protocols were approved by the IARC and local ethical and/or research committees of each institution. All study subjects signed a written informed consent before performing any study procedures.

Case patients

Case subjects were patients newly and histologically diagnosed in one of the participating hospitals with invasive cancer of the oral cavity and oropharynx, rubrics C01 through C10 of the International Classification for Diseases in Oncology, which include: tongue, base of tongue, gum, floor of mouth, palate, salivary glands, tonsil, other parts of the mouth and oropharynx.20 Patients with invasive tumors originating in the mucosal inner lip were also eligible to participate.

Cases had to be diagnosed in one of the participating hospitals; however, untreated patients diagnosed outside the hospital but referred to the participating hospital for primary therapy were also eligible. All histologic types were accepted. Patients had to be included in our study before undergoing any local or systemic therapy.

A total of 490 cases were identified as eligible for participation in our study. Of those, 74 (15.1%) were excluded because they received surgical treatment within 24 hr after admission, making it impossible to interview the patient before treatment. Eight (1.6%) patients received emergency surgery and 19 (3.9%) could not follow the established protocol because they developed medical complications. Fourteen (2.9%) eligible patients refused to participate in our study. The overall participation rate of eligible cases was 76.5% (375/490): 70.6% in Barcelona, 78.2% in Granada and 90.5% in Sevilla.

A total of 306 cases had oral cancer and a total of 69 cases had oropharyngeal (oropharynx and tonsil) cancer. The site-specific distribution of the tumors was in decreasing frequency: tongue (28%), floor of mouth (15.7%), other parts of the mouth (12%), oropharynx (9.3%), tonsil (9.1%), gum (7.7%), base of tongue (7.2%), palate (5.1%), lip (3.2%) and salivary glands (2.7%). Most (93.4%) tumors were squamous cell carcinomas. The clinical stage was assessed in 364 cases (97.1%) with the following distribution: 20.9% stage I, 18.1% stage II, 16.5% stage III and 44.5% stage IV.

Control patients

Control patients were identified and recruited from the same hospitals from which the cases were identified and they were group-matched to the case subjects by gender and quinquennia of age. Patients with an admission diagnosis related to alcohol or tobacco consumption were not accepted as controls. Patients with oncologic diagnoses were also excluded. If a selected control refused to participate, a replacement with the same matching criteria was used. Overall participation rate of eligible controls was 91%: 90.5% in Sevilla, 91.2% in Granada and 91.1% in Barcelona.

The distribution of the main diagnostic categories was in decreasing frequency: orthopedic conditions and procedures (25.9%), digestive tract diagnoses or procedures (19.2%), lesions and symptoms (12.8%), respiratory tract conditions (9.6%), hematologic and cardiovascular conditions (9.1%), infectious diseases (6.9%), genito-urologic conditions (6.9%) and other conditions (9.6%).

Questionnaire

All consenting cases and controls were interviewed at the corresponding hospital according to a pretested standardized questionnaire administered by trained interviewers. Proxy interviews were not accepted. The questionnaire elicited detailed information on demographic, educational and socio-economic characteristics, on the characteristics of lifetime habits of tobacco smoking and alcohol consumption, as well as on other potential risk factors for oral cancer such as dietary habits, oral hygiene, dentition, sexual habits, history of venereal diseases and family history of cancer.

The questionnaire was designed to include the subject's entire smoking and alcohol histories, breaking down the information into periods of homogeneous consumption. This was done for each type (black air-cured and blond flue-cured) and brand of tobacco smoked, and for each type of alcoholic beverage consumed (beer, wine and spirits) throughout the subject's life. Changes in intensity of consumption and ages at the beginning and end of each period were also recorded.

Statistical analyses

A single data file including all common comparable items of information from each center was compiled and the following tobacco and alcohol consumption related variables were computed: average number of cigarettes smoked per day, average milliliters of pure ethanol consumed per day, average number of drinks consumed per day, duration of each habit, age at starting and stopping each habit, years since stopping each habit, type of tobacco smoked, use of filter cigarettes and type of alcoholic beverage consumed. Milliliters of pure ethanol were converted into drinks by applying a factor of 1/18 (i.e., one drink would approximately contain 18 ml—or 14.4 grams—of pure ethanol). One drink would approximately correspond to one glass of 150 cc of wine, to one beer of 330 cc, or to one drink of hard liquor of 36 cc. Number of pipes and cigars smoked were further converted into cigarettes considering that one cigar and one pipe corresponded to 5 and 4 cigarettes, respectively.

Current smokers and current drinkers were defined as those having the habit at the time of the interview, as well as those stopping the habit within the year before the date of the interview. This precaution was taken to allow for interruption of the habit due to early disease symptoms.

Data were analyzed using unconditional multiple logistic regression models with maximum likelihood estimation of parameter values to obtain odds ratios (OR) as an approximation of the relative risk. Unless otherwise specified in the tables, all logistic-regression models included as covariates: age (in quartiles: ≤ 51 years, 52–59, 60–69, ≥ 70), gender, center, years of schooling (in tertiles: 0, 1–5, 6–8, ≥ 9), average number of cigarettes smoked per day (in quartiles, 0, 1–18, 19–22, 23–34, ≥ 35) and average milliliters of pure ethanol consumed per day (in quartiles: 0, 1–23, 24–72, 73–150, ≥ 151).

To adjust for dietary habits, the total weekly frequency of consumption of specific foods during the last year before cancer diagnosis or hospital admission was elicited and combined into 4 major food groups: starchy foods (including cereal dishes and pulses), animal foods (including meats, fish and cheese), all vegetables (excluding potatoes) and all fruits. Logistic regression models further adjusted for these 4 food groups yielded virtually the same results.

Further adjustment for oral hygiene (as determined by clinical exam) and dentition (number of missing teeth) did not substantially alter the magnitude of the main associations found in these analyses. Similarly, exclusion of cases with tumors with a histology other than squamous-cell carcinoma (n = 30) did not substantially modify the reported associations.

Except for stratified analyses, in which they were excluded, subjects with missing values were included in an additional variable category and considered thus in the analysis.

Tests for trends were performed by including in the logistic regression models the categoric variable of interest treated as a continuous variable and using the likelihood ratio test statistic with one degree of freedom.

The joint effects of tobacco smoking and alcohol consumption variables were assessed by including in the model the 2 main terms and the interaction term between the 2 variables of interest.

Because of the hospital-based design of our study and in order to avoid the assumption that the controls represent a random sample of the population from which the cases were obtained, attributable fractions (AFs) for each exposure or combination of exposures were computed using the distribution of exposures among cases and applying the method described by Bruzzi that allows the use of adjusted estimates of relative risk obtained from logistic regression models (reviewed in Coughlin et al.21). The overall AF for smoking was estimated by adding to the AF for the combined exposure to smoking and alcohol the AF for smoking only. The overall AF for alcohol was obtained by adding to the AF for the combined exposure to smoking and alcohol the AF for alcohol only. The ORs used to compute these estimates are the ones reported in Table VI. The ORs used to compute AFs for type of tobacco and type of alcoholic drink were adjusted for all covariates except alcohol amount and smoking amount, respectively.

Table VI. Joint Effects of Cigarette Smoking and Ethanol Consumption Statuses and Amounts on Risk of Oral Cancer
Drinking status and drinking amountTobacco smoking statusAverage no. of cigarettes smoked per day
Never smokerEver smoker1–1011–20≥21
  1. Top numbers in each cell indicate OR and 95% confidence interval for oral cancer adjusted for age group, gender, center, and years of schooling. Bottom number in each cell indicates the number of cases/controls for that exposure category. Reference group includes subjects who never smoked and never drank.– Figures in bold type indicate statistical significance at the 0.05 level.

Never drinker1.01.572.871.041.85
(reference)(0.53–4.69)(0.56–14.80)(0.18–6.03)(0.31–11.13)
28/537/223/62/82/8
Ever drinker1.6612.665.3115.3918.77
(0.80–3.44)(5.51–29.11)(2.10–13.45)(6.33–37.43)(7.86–44.82)
27/60313/24025/4495/72193/123
1–2 drinks/day2.017.274.6611.098.20
(0.93–4.35)(3.00–17.61)(1.68–12.94)(4.02–30.56)(2.93–22.94)
23/4563/11014/3127/3622/43
3–5 drinks/day1.0721.5532.1826.5522.04
(0.18–6.38)(8.35–55.58)(8.14–127.10)(8.60–81.98)(7.96–61.01)
2/872/5310/622/1540/31
≥ 6 drinks/day6.1939.682.6543.1250.65
(0.98–39.17)(15.67–100.44)(0.26–26.54)(15.02–123.80)(19.11–134.24)
2/6178/771/746/21131/49

Statistical significance was set at the 0.05 level. All p-values were derived from 2-sided statistical tests.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Subjects' characteristics

Table I shows the distribution of study participants according to their case-control status, study center and selected sociodemographic characteristics and main descriptive statistics for tobacco smoking and alcohol drinking. The prevalence of tobacco smoking was higher among case subjects than among control subjects overall and in each of the 3 centers. For alcohol consumption, prevalence of current drinking was also higher among case subjects than among control subjects overall and by center, except in Sevilla, where a similar prevalence was observed. There was no heterogeneity of effects of years of schooling, tobacco smoking and alcohol drinking across studies.

Table I. Distribution of Cases and Controls According to Main Socio-demographic and Exposure Variables
 Total
Cases n (%)Controls n (%)
  • 1

    Mean number of cigarettes smoked per day among ever smokers.

  • 2

    Mean years of tobacco smoking among ever smokers.

  • 3

    Mean number of drinks consumed per day among ever drinkers.

  • 4

    Mean years of alcohol drinking among ever drinkers.

Number of subjects375375
Center  
 Sevilla86 (22.9%)86 (22.9%)
 Granada104 (27.7%)104 (27.7%)
 Barcelona185 (49.3%)185 (49.3%)
Sex  
 Males304 (81.1%)304 (81.1%)
 Females71 (81.9%)71 (18.9%)
Age  
 Mean6060
 ≤5198 (26.1%)94 (25.1%)
 52–5985 (22.7%)88 (23.5%)
 60–69103 (27.5%)103 (27.5%)
 ≥7089 (23.7%)90 (24.0%)
Years of schooling  
 Never schooling43 (11.5%)46 (12.3%)
 1–5110 (29.5%)108 (28.8%)
 6–8109 (29.2%)116 (30.9%)
 ≥9111 (29.8%)105 (28.0%)
 Unknown20
Smoking status, amount and duration  
 Never55 (14.7%)113 (30.1%)
 Ex82 (21.9%)133 (35.5%)
 Current238 (63.5%)129 (34.4%)
 Mean amount12824
 Mean duration24035
Alcohol status, amount and duration  
 Never35 (9.3%)75 (20.0%)
 Ex89 (23.7%)103 (27.5%)
 Current251 (66.9%)197 (52.5%)
 Mean amount385
 Mean duration43835

Effects of tobacco smoking

All measures of tobacco smoking status, amount, duration, cessation and type of tobacco were strongly associated with cancer risk after adjusting for the effects of alcohol consumption (Table II). Compared to current smokers, ex-smokers had a reduced cancer risk regardless of the amount of tobacco smoked and regardless of the duration of the habit. An increased risk for oral cancer was already detected among light smokers (1 to 10 cigarettes per day) and among short-term smokers (1 to 10 years). After excluding ex-smokers, an increasing number of cigarettes smoked per day was linearly associated with increasing risk of oral cancer.

Table II. Odds Ratios (OR) and 95% Confidence Intervals (CI) for Oral Cancer According to Tobacco Smoking-related Characteristics
VARIABLESNo. of casesNo. of controlsOR (95% CI)1
  • 1

    Odds ratios adjusted for center, gender, age group, years of schooling and average daily consumption of pure ethanol.

  • 2

    Among smokers and further adjusted for average amount of cigarettes smoked per day.– Figures in bold type indicate statistical significance at the 0.05 level.

Tobacco smoking status   
 Never smoker551131.00 (reference)
 Ever smoker3202624.10 (2.19–7.69)
 Ex-smoker821332.36 (1.21–4.60)
 Current smoker2381296.41 (3.32–12.37)
 p value for trend test  <0.0001
Smoking amount (average no. of cigs./day)   
 1–1028502.60 (1.23–5.49)
 11–2097804.94 (2.48–9.85)
 21–3087615.23 (2.55–10.74)
 31–4064404.54 (2.11–9.79)
 ≥4144304.35 (1.92–9.82)
 Unknown01 
 p value for trend test  0.003
Smoking amount by smoking status   
 Ex-smoker of 1 to 20 cigs./day40722.68 (1.31–5.50)
 Ex-smoker of ≥21 cigs./day42612.10 (0.99–4.45)
 Current smoker of 1 to 10 cigs./day15213.66 (1.42–9.43)
 Current smoker of 11 to 20 cigs./day70378.48 (3.71–19.37)
 Current smoker of 21 to 30 cigs./day76408.81 (3.82–20.35)
 Current smoker of 31 to 40 cigs./day43217.30 (2.84–18.79)
 Current smoker of ≥41 cigs./day34912.21 (4.08–36.54)
 Unknown01 
 p value for trend test  <0.0001
Years of tobacco smoking   
 1–1010142.23 (0.81–6.16)
 11–2014212.84 (1.09–7.40)
 21–3058573.37 (1.55–7.32)
 31–4073753.79 (1.79–8.03)
 41–5089626.34 (2.95–13.60)
 ≥51763310.19 (4.49–23.13)
 p value for trend test  <0.0001
Smoking duration by smoking status   
 Ex-smoker of 1 to 40 years49912.43 (1.21–4.89)
 Ex-smoker of ≥41 years33422.81 (1.25–6.32)
 Current smoker of of 1 to 40 years106764.40 (2.16–8.95)
 Current smoker of ≥41 years132539.56 (4.62–19.77)
 p value for trend test  <0.0001
Years since quitting smoking   
 Current smoker2381296.78 (3.49–13.17)
 1–330313.88 (1.69–8.91)
 4–819242.53 (1.05–6.12)
 9–1621332.53 (1.09–5.86)
 ≥1712451.16 (0.47–2.85)
 p value for trend test2  <0.0001
Type of tobacco   
 Blond only41672.31 (1.15–4.64)
 Mixed61594.55 (2.12–9.77)
 Black only2151257.72 (3.83–15.55)
 Unknown311 
 p value for trend test2  <0.0001

The relationship between years of tobacco smoking and cancer risk also followed a strong, dose-response relationship overall, and was found among both current and former smokers. Quitting cigarette smoking significantly reduced cancer risk. The reduction was already detectable after quitting for 1 to 3 years. Only after 17 or more years of smoking cessation was the risk close to that of never smokers.

Compared to nonsmokers, increased risks were found for smokers of blond, mixed and black tobacco, but the magnitude of the OR was substantially higher for black tobacco smoking than that for blond or mixed tobacco smoking.

No associations were found with age at start or age at quitting tobacco smoking after adjusting for duration of the habit. Smokers of cigarettes with filter had the same risk as smokers of cigarettes without filter (data not shown).

There were too few exposed women (21 cases and 10 controls) to be able to assess with certain precision the effect of tobacco-related variables among women.

Type of tobacco

We explored in more detail the relationship between type of tobacco smoked and cancer risk (Table III). After allowing for smoking amount and duration, and compared to smokers of only blond tobacco, smokers of only black tobacco had a 3.3-fold increase in the risk of developing oral cancer (data not shown). To minimize potential residual confounding due to smoking amount, we fitted a logistic regression model including a variable that coded categorically the number of cigarettes smoked in single units. This more refined adjustment yielded an even stronger association between black tobacco smoking and oral cancer risk. (OR = 5.4). Further adjustment for years since quitting augmented the OR to 6.2. We also found that the association linked to black tobacco smoking was present in each stratum of smoking amount and smoking duration. Furthermore, within each one of the 3 groups of type of tobacco (blond only, mixed and black only), statistically significant dose-response associations were found with tobacco amount and duration, but these associations were clearly stronger for black tobacco smoking than those for blond or mixed tobacco smoking.

Table III. Odds Ratios (OR) and 95% Confidence Intervals (CI) for the Association Between Type of Tobacco and Risk of Oral Cancer by Levels of Amount and Duration of Tobacco Smoking
 Type of tobacco smokedOR for “Black/Mixed” vs. “blond only”
Blond onlyMixed (blond and black)Black only
Cases/controlsOR (95% CI)1Cases/controlsOR (95% CI)1Cases/controlsOR (95% CI)1OR (95% CI)2
  • Figures in bold type indicate statistical significance at the 0.05 level.

  • 1

    OR adjusted for age group, gender, center, years of schooling and average daily consumption of pure ethanol.

  • 2

    Among ever smokers and further adjusted for years of tobacco smoking and average number of cigarettes smoked per day. Reference category is smokers of only blond tobacco.

  • 3

    Among ever smokers and further adjusted for years of tobacco smoking and average number of cigarettes smoked per day in single units and entered in the model as categorical. Reference category is smokers of only blond tobacco.

  • 4

    Trend test p < 0.0001.

Average number of cigarettes smoked per day       
 Never smokers55/1131.0 (reference)55/1131.0 (reference)55/1131.0 (reference) 
 Ex14/272.15 (0.86–5.37)15/214.70 (1.41–15.63)53/774.52 (1.74–11.72)2.47 (0.96–6.36)
 Current of 1–2018/232.92 (1.18–7.23)11/125.72 (1.54–21.21)55/2017.90 (6.35–50.43)2.03 (0.69–5.98)
 Current of 21–304/141.83 (0.47–7.18)18/156.94 (2.02–23.77)53/1128.19 (9.22–86.20)3.56 (1.18–10.75)
 Current of ≥315/29.45 (1.40–63.82)17/119.18 (2.57–32.78)54/1716.53 (5.52–49.50)
 p value for trend test .01 .001 <0.0001 
Years of tobacco smoking       
 Ex14/272.18 (0.87–5.51)15/214.70 (1.41–15.73)53/774.81 (1.85–12.51)2.47 (0.96–6.36)
 Current 1–3014/152.66 (0.92–7.68)14/108.47 (2.13–33.70)27/1111.75 (3.48–39.66)1.63 (0.40–6.60)
 Current 31–5010/222.58 (0.88–7.54)25/235.76 (1.81–18.31)84/2619.99 (7.26–55.08)3.94 (1.53–10.19)
 Current ≥513/35.77 (0.90–37.07)7/511.58 (2.34–57.29)51/1129.67 (9.41–93.55)
 p value for trend test .01 .001 <0.0001 
Summary OR341/671.0 (reference)61/591.86 (0.87–4.00)215/1255.42 (2.76–10–63)43.963 (2.09–7.53)

Effects of alcohol drinking

Measures of alcohol drinking status, amount, duration and cessation were also strongly associated with the risk of oral cancer (Table IV). A statistically significant increased risk of oral cancer was detected among subjects drinking as little as one drink per day. Cancer risk increased steadily and markedly with longer duration of the habit with increased risks that were statistically significant after 20 years of alcohol consumption. Quitting alcohol drinking significantly reduced cancer risk after at least 3 years of cessation, but only after 14 years of stopping the habit was the risk close to that of never drinkers.

Table IV. Odds Ratios (OR) and 95% Confidence Interval (CI) for Cancer of the Oral Cavity According to Alcohol Consumption-related Characteristics
VariablesNo. of casesNo. of controlsOR (95% CI)1
  • Figures in bold type indicate statistical significance at the 0.05 level.

  • 1

    Odds ratios adjusted for center, gender, age group, years of schooling and average number of cigarettes smoked per day.

  • 2

    Among drinkers and further adjusted for average daily consumption of pure ethanol.

Alcohol drinking status   
 Never drinker35751.00 (reference)
 Ever drinker3403002.86 (1.59–5.15)
 Ex-drinker891032.12 (1.13–3.99)
 Current drinker2511973.46 (1.88–6.35)
 p value for trend test  <0.0001
Drinking amount (average no. of drinks/day)   
 1591142.00 (1.06–3.77)
 227413.74 (1.62–8.63)
 3–449446.22 (2.82–13.71)
 5–6552810.58 (4.57–24.46)
 7–10683710.29 (4.57–23.17)
 ≥11823513.66 (6.02–30.96)
 Unknown01 
 p value for trend test  <0.0001
Drinking amount by drinking status   
 Ex-drinker of 1 to 4 drinks/day35621.90 (0.95–3.80)
 Ex-drinker of ≥5 drinks/day54405.42 (2.50–11.75)
 Current drinker of 1 to 4 drinks/day1001372.97 (1.56–5.63)
 Current drinker of ≥5 drinks/day1516010.97 (5.31–22.68)
 Unknown01 
 p value for trend test  <0.0001
Years of alcohol drinking   
 1–2027421.37 (0.65–2.91)
 21–3069672.49 (1.22–5.09)
 31–4096843.18 (1.61–6.29)
 41–5088694.00 (1.99–8.02)
 ≥5160375.13 (2.45–10.72)
 Unknown01 
 p value for trend test  <0.0001
Drinking duration by drinking status   
 Ex-drinker of 1 to 40 years56841.58 (0.82–3.07)
 Ex-drinker of ≥41 years33184.53 (2.00–10.27)
 Current drinker of of 1 to 40 years1361093.28 (1.70–6.33)
 Current drinker of ≥41 years115883.43 (1.79–6.59)
 Unknown01 
 p value for trend test  <0.0001
Years since quitting alcohol drinking   
 Current drinker2511973.52 (1.91–6.48)
 1–228183.90 (1.68–9.06)
 3–722301.74 (0.77–3.92)
 8–1320232.28 (0.98–5.32)
 ≥1419311.50 (0.67–3.34)
 Unknown01 
 p value for trend test2  0.003
Type of drink   
 Only beer12361.16 (0.47–2.82)
 Only wine and beer47861.96 (0.96–3.99)
 Only wine32412.71 (1.31–5.60)
 Spirits with or without wine or beer2481377.28 (3.65–14.52)
 Unknown10 
 p value for trend test2  <0.0001
Drinking and meals   
 With meals only69942.22 (1.18–4.15)
 Between meals only90723.58 (1.84–6.98)
 Both, with and in between1781303.71 (1.94–7.08)
 Unknown34 

Each type of alcoholic beverages (beer only, wine and beer, wine only and spirits) was associated with cancer risk, but the association was much stronger for drinking of spirits. The risk increased with increasing ethanol content of each type of drink (p for trend among drinkers: < 0.0001).

Drinking alcoholic beverages only at meals was associated with a reduced risk of oral cancer compared to drinking between meals or always.

No statistically significant associations were observed with age at start or age at quitting alcohol drinking after adjusting for duration.

There were 40 case women and 23 control women ever exposed to alcohol drinking. The overall adjusted OR for ever drinking among women, compared to abstainers, was 2.8 (95% confidence interval [CI] 1.25–6.25) overall, 2.60 for ex-drinkers and 2.94 for current drinkers. Only 3 control women and 9 case women consumed more than one drink per day; thus, associations with alcohol amount, duration and quitting could not be assessed among women.

Type of alcoholic beverage

Table V shows the relationship between type of alcoholic beverage and cancer risk. After allowing for amount and compared to drinkers of only beer, drinkers of spirits (either alone or in combination with wine and/or beer) had a 4.2-fold (95% CI 1.8–9.7) increased risk of developing oral cancer (data not shown). When adjusting more finely for consumption by entering in the logistic regression models a variable coding categorically the number of drinks per day in single units, the OR for spirits remained virtually unchanged (OR = 3.8). We found that, compared to drinkers of wine and/or beer, drinkers of spirits had a consistent increase in cancer risk in each stratum of amount and duration of alcohol consumption. Furthermore, although statistically significant associations were found with amount and duration within each one of the 3 groups of type of drink, the ORs for spirits were consistently higher and stronger than those for the other types of drinks.

Table V. Odds Ratios (OR) and 95% Confidence Interval (CI) for the Association Between Type of Alcoholic Beverage and Risk of Oral Cancer by Levels of Amount and Duration of Alcohol Drinking
 Type of alcoholic beverage consumedOR for “Spirits” vs. “Wine and/or Beer”
Beer onlyWine (without spirits)Spirits (with or without wine or beer)
Cases/controlsOR (95% CI)1Cases/controlsOR (95% CI)1Cases/controlsOR (95% CI)1OR (95% CI)2
  • Figures in bold type indicate statistical significance at the 0.05 level.

  • 1

    OR adjusted for age group, gender, center, years of schooling and average number of cigarettes smoked per day.

  • 2

    Among ever drinkers and further adjusted for years of alcohol drinking and average number of drinks per day. Reference category is drinkers of wine and/or beer.

  • 3

    Among ever drinkers and further adjusted for years of alcohol drinking and average number of drinks per day in single units and entered in the model as categorical.

  • 4

    Trend test p < 0.0001.

Average number of drinks per day       
 Never drinkers35/751.0 (reference)35/751.0 (reference)35/751.0 (reference) 
 1–27/301.33 (0.41–4.37)49/852.58 (1.25–5.32)30/403.00 (1.20–7.51)2.08 (1.02–4.23)
 3–55/626.0 (3.0–221.6)20/293.60 (1.39–9.30)50/306.73 (2.56–17.74)2.07 (0.90–4.78)
 ≥610/124.52 (1.38–14.78)168/6710.54 (4.17–26.60)3.61 (1.52–8.58)
 p value for trend test 0.02 0.005 <0.0001 
Years of alcohol drinking       
 1–305/151.58 (0.39–6.34)18/401.60 (0.67–3.82)73/543.39 (1.58–9.78)2.56 (1.08–6.07)
 31–405/104.94 (0.99–24.7)16/362.65 (1.02–6.89)74/385.93 (2.33–15.1)2.43 (0.96–6.14)
 41–502/110.70 (0.08–5.81)24/304.16 (1.61–10.8)63/326.49 (2.53–16.6)2.19 (0.86–5.60)
 ≥5121/205.23 (1.99–13.8)38/138.58 (2.93–25.1)3.01 (0.92–9.83)
 p value for trend test 0.40 <0.0001 <0.0001 
Summary OR312/361.0 (reference)79/1261.78 (0.77–4.11)248/1373.83 (1.63–8.99)42.32 (1.49–3.63)3

Joint effects of tobacco smoking and alcohol drinking

Table VI shows that ever exposure to only one of the 2 habits was associated with a moderate and not significant increase in cancer risk. However, simultaneous exposure to both habits increased cancer risk by almost 13-fold. The interaction term between tobacco smoking status and alcohol drinking status was highly significant (p = 0.008), suggesting a synergistic joint effect.

The combined effects of smoking and drinking amounts were also compatible with a synergistic model of borderline statistical significance (p-value for interaction term: 0.1). Concomitant consumption of moderate amounts of tobacco (1 to 10 cigarettes per day) and alcohol (1 to 2 drinks per day) was associated with a statistically significant 5-fold increase in the risk of developing oral cancer (not shown). The p-value for the interaction between smoking status and drinking amount was 0.16. The p-value for the interaction between drinking status and smoking amount was 0.05.

No statistically significant interactions were found between type of tobacco and type of alcoholic drink or any other additional joint exposure of tobacco and alcohol consumption characteristics.

Effects of tobacco and alcohol consumption by oral subsites

For all tobacco-related variables, the magnitude of the ORs were slightly larger for oropharyngeal cancer (i.e., oropharynx and tonsil) than that for cancer of the mouth (i.e., tongue, floor of mouth, other parts of the mouth, gum, base of tongue and palate), but their corresponding CIs clearly overlapped. For alcohol-related variables, even though CIs of the ORs also overlapped, associations were clearly stronger for oropharyngeal cancer than for cancer of the mouth (data not shown). For instance, the OR for drinking more than 10 drinks per day was 11.7 for cancer of the mouth and 24.4 for cancer of the oropharynx.

Attributable fractions in the population

Estimated attributable fractions were 77.6% for ever tobacco smoking, 79.7% for ever alcohol consumption and 80.4% for ever exposure to either or both habits. Black tobacco smoking and drinking of spirits explained up to 67.4% and 60.2% of oral cancer incidence, respectively. However, the overall attributable fraction for ever exposure to either black tobacco smoking or spirits drinking was 76.7%.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Our study shows that tobacco smoking and alcohol drinking are strongly associated with oral cancer in Spain. As expected, all relevant measures of tobacco smoking and alcohol drinking were strongly related to cancer risk and there was a risk gradient with increasing intake and longer duration of tobacco and alcohol consumption. Furthermore, removal of these exposures by quitting either of the 2 habits strongly reduced cancer risk, suggesting a true carcinogenic role of these exposures on the oral cavity and the oropharynx. These findings are consistent with those from many other epidemiologic studies carried out in other Western populations in which tobacco smoking and alcohol drinking have also been identified as the prime risk factors for cancers of the upper aerodigestive tract.18, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36

We were, however, particularly interested on the effects of type of tobacco and type of alcoholic beverage on cancer risk, as we hypothesized that the high incidence rates of oral cancer in Spain may be mostly explained by the widespread exposure to black tobacco smoking and drinking of spirits.

Type of tobacco

We found that relative to smokers of blond (flue-cured) tobacco, smokers of black (air-cured) tobacco were at a statistically significant higher risk of developing oral cancer even after allowing for the effects of amount and duration of smoking. The association with black tobacco smoking was particularly strong in our study, and, as shown in our stratified analyses (Table III), the association was observed in each level of smoking intensity and in each level of smoking duration.

Other studies have shown a higher risk of aerodigestive tract cancers linked to black tobacco smoking.22, 30, 37, 38, 39 On the other hand, some did not find a clear indication of a different effect according to the type of tobacco.32, 33 Most previous studies could not rule out a residual effect of intensity and duration; thus, our stratified analyses, in which smoking amount was adjusted in single cigarette units, convincingly show that black tobacco smoking significantly increases cancer risk compared to blond tobacco smoking, and that the association is unlikely to be confounded by the potential residual effects of intensity or duration.

The increased risk of black compared to blond tobacco is consistent with the ecologic studies showing a relatively high incidence of tumors of upper aerodigestive tract in southern Europe and Latin America, where this kind of tobacco is mainly consumed.40 This is in agreement with biochemic and experimental studies that show that higher levels of some carcinogenic compounds such as N-nitrosamines are found in smoke of black tobacco than in smoke of blond tobacco.4 Furthermore, levels of hemoglobin adducts as well as urine mutagenicity in experimental systems are also increased among black tobacco smokers.

Type of alcoholic beverage

Our study found an independent effect of type of alcoholic beverage on cancer risk. For each level of alcohol intake or duration, drinkers of spirits were between 2 and 3 times more likely to develop oral cancer than drinkers of only wine or beer. As with smoking, adjustment by number of daily drinks in single units did not substantially alter the results. Thus, the increased risk attributed to drinking of spirits cannot be explained by differences in amount or duration of pure ethanol consumed. In Spain, the 2 most frequently consumed spirits among adults are brandy (13% of drinkers) and whisky (9% of drinkers).

The effect of type of beverage on the risk of developing cancer of the upper aerodigestive tract remains controversial in the literature. Although some studies report an association with type of beverage, many did not properly allow for the effect of drinking intensity and duration. This is particularly important, as reported associations with type of beverage might be confounded by the fact that subjects drinking spirits were probably those drinking heavily or for a long time. Similarly, some authors claim that the most prevalent alcoholic beverage in a given population is also the one most heavily consumed and consequently the one in which an increased risk is most likely to be detected.18, 26, 31, 34, 41, 42, 43

For cancers of the upper aerodigestive tract, some, but not all, studies found an effect with consumption of spirits. In Brazil, the excess risks for oral cancer were observed with increased consumption of distilled sugarcane spirit.24 Studies in Italy showed an excess risk for wine,44 alone or with other alcoholic beverages,45 but the data were too sparse to study similar effects among heavy drinkers of other types of beverages. In France, while a report showed that consumption of hot Calvados explained about half the risk of esophageal cancer, another study failed to demonstrate risk differences by specific types of drinks. In contrast, consistent with our results, a population-based prospective study of oropharyngeal and esophageal cancer conducted in Denmark, in which the effects were also stratified by consumption levels, did find an elevated risk for consumption of spirits and, to a lesser extent, beer, but not for wine.26 While other studies have shown a reduced cancer risk for consumption of wine compared to other alcoholic beverages,23, 33, 35 other studies have found a stronger effect from wine.41, 42 Consistent with our results, 2 recent reports of studies conducted in Brazil43 and Puerto Rico46 show that alcohol concentration is a risk factor for oral cancer regardless of the total amount consumed.

The mechanisms by which alcoholic beverages induce oral cancer are not clear. The most likely explanation is that alcohol or its metabolites are human carcinogens. Despite the lack of clear experimental evidences regarding carcinogenicity of ethanol, a clear association with oral cancer risk appears for all kind of alcoholic beverages, all of which have ethanol as a common ingredient. Another possibility is that the effect of alcoholic beverages is due to the carcinogenic effect of acetaldehyde, the first metabolite of ethanol.47 In addition to a systemic effect, it has been suggested that ethanol can be converted to acetaldehyde in saliva, acting directly to damage the mucosa of the upper gastrointestinal tract. On the other hand, it has been proposed that alcohol might act as cocarcinogen by several pathways or mechanisms: (i) by increasing permeability of the mucosa to other carcinogens; (ii) by induction of enzymes involved in the metabolism of tobacco smoke carcinogens both systemically and locally and (iii) by direct toxic or oxidative effect on the epithelial mucosa. Consumers of alcohol may also differ in other aspects of lifestyle, such as diet: alcohol drinking may be associated with poor fruit and vegetable intake, and furthermore, it may have a role as a folate antagonist.6 In our data, adjustment by dietary variables did not substantially alter the risk estimates (data not shown).

Finally, another hypothesis to be considered is that besides ethanol, alcoholic beverages contain other substances with potential carcinogenic effect, such as N-nitrosamines and urethane,5 as well as substances with anticarcinogenic properties. The former could, to some extend, help to explain the differential effect of alcoholic beverages: N-nitrosamines are found in liquors, although they are also present in beer. Urethane, a substance with carcinogenic potential in experimental studies, may be found in liquor but not in beer and wine, while most of the antioxidant compounds found in wine and to some extent in beer, are almost completely lacking in spirits.5

Joint effects of tobacco and alcohol consumption

We found a statistically significant synergism between the effect of smoking and the effect of drinking on the risk of developing oral cancer. Thus, while cancer risk was only moderately increased in never smokers who ever drank (OR = 1.66) and among never drinkers who ever smoked (OR = 1.57), simultaneous exposure to both tobacco and alcohol consumption increased the risk almost 13-fold compared to subjects never exposed to these habits. The interaction between smoking and drinking status was highly significant (p = 0.008), indicating that the individual risks interact synergistically. These findings are relevant as most previous studies exploring the modeling of the joint risks did not find a significant departure from the multiplicative model.22, 26, 27, 35, 44 However, at least 2 studies, one in Brazil and one in Cuba, also found that the effects of smoking and drinking were compatible with a supramultiplicative model.18, 24

Impact of tobacco and alcohol consumption on cancer incidence

The impact of tobacco smoking was similar to that of alcohol drinking, as AFs were 77.6% and 79.7%, respectively. Up to 80.4% of these cancers can be attributed to the joint exposure of both habits. This is due to the relatively high prevalence of individuals simultaneously exposed to both habits in the population (64%) and to the strong association found with cancer risk in these individuals. More importantly, exposure to black tobacco smoking and/or drinking of spirits may explain up to 76.7% of oral cancer occurrence in Spain. Based on these estimates, it can be concluded that the high prevalence in the population of subjects exposed to smoking and drinking, and among these the high prevalence of black tobacco smoking and spirits drinking, largely explain the high incidence rates of oral cancer in Spain.

Study limitations

Case-control studies have some important limitations and are subject to bias, and our study is no exception. Selection bias might be an issue but this potential bias is somehow minimized by the fact that: (i) controls were selected from the same hospitals as the cases; (ii) only acute conditions were accepted; (iii) control patients hospitalized for diagnoses related to alcohol and tobacco consumption were carefully excluded; (iv) participation rates overall and by center were very high and (v) Spain has universal health coverage with public hospitals that provide quality cancer care. Although some oral cancer cases in the population might be attending other hospitals in and out of the influence area of the participating centers, patients' referral patterns should not be related to tobacco and alcohol consumption. Other biases such as recall bias and interview bias may influence the data but adjustment by interviewer, quality and duration of interview and the multivariate approach of the statistical analyses reassure a small impact if any on the reported findings.

Misclassification of study exposures might distort the true associations between these factors and oral cancer risk. However, only differential misclassification (cases differentially over- or under-reporting exposure with regard to controls) might be responsible for an artificial effect. However, we think this is quite unlikely given the magnitude and statistical significance of the associations and the internal consistency of the results (i.e., positive associations were found for intensity and duration and inverse associations were found for quitting).

Given the strength of the associations and the refined statistical adjustments performed, the issue of residual confounding, although always possible, would need to be exerted by a risk factor very strongly related to both the exposures of interest and to cancer status in order to explain the strong reported associations. We adjusted for most of the relevant risk factors reported in the literature and we further performed stratified analyses excluding groups of cases and/or controls that could potentially distort the results, and the findings were systematically minimally altered.

Implications

Our study shows that smoking and drinking patterns, especially the type of tobacco smoked, the type of alcoholic beverage consumed and the concomitant exposure to both habits (even in moderate amounts) modulate the risk of developing oral cancer. Attention should be given to these aspects of smoking and drinking habits. Although it has been suggested that moderate consumption of alcohol could be beneficial for some diseases, it does not seem to apply to cancers of the upper aerodigestive tract. Our data suggest that consumption of spirits and hard liquors must be avoided, especially among smokers and subjects with a history of smoking. Our findings regarding oral cancer, along with those for other cancer sites, underscore the importance of public health strategies targeted to prevent and reduce smoking and drinking exposures. The public should also be aware of the high risk of oral cancer induced by consumption of beverages with high ethanol content, as well as that induced by black tobacco smoking. Given the relatively poor survival rates of patients diagnosed with oral cancer, these strategies remain the best and probably only available measure to effectively prevent and control oral cancer.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We thank Mr. J. Muñoz and Ms. G. Albero (Institut Català d'Oncologia, Spain) for data management and statistical analyses. Ms. C. Rajo was responsible for the secretarial workload and her efforts are greatly appreciated. The authors are indebted to the surgeons, oncologists and supervisors of the fieldwork, as well as to the many persons who volunteered to participate in the studies.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  • 1
    Ferlay J, Bray F, Sankila R, Parkin DM. EUCAN: cancer incidence, mortality and prevalence in the European Union 1996. version 3.1. IARC CancerBase No. 4. Lyon, France: IARC Press, 1999.
  • 2
    Berrino F, Gatta G, EUROCARE Working Group. Variation in survival of patients with head and neck cancer in Europe by site of origin of the tumours. Eur J Cancer 1998; 14: 215461.
  • 3
    Forastiere A, Koch W, Trotti A, Sidransky D. Head and neck cancer. N Engl J Med 2001; 345: 1890900.
  • 4
    IARC monographs on the evaluation of the carcinogenic risk of chemicals to humans. vol 38. Tobacco smoking. Lyons, France: International Agency for Research on Cancer, 1986.
  • 5
    IARC monographs on the evaluation of the carcinogenic risk of chemicals to humans. vol 44. Alcohol drinking. Lyons, France: International Agency for Research on Cancer, 1988.
  • 6
    Blot WJ. Invited commentary: more evidence of increased risks of cancer among alcohol drinkers. Am J Epidemiol 1999; 150: 113840.
  • 7
    Franceschi S, Favero A, Conti E, Talamini R, Volpe R, Negri E, Barzan L, La Vecchia C. Food groups, oils and butter, and cancer of the oral cavity and pharynx. Br J Cancer 1999; 80: 61420.
  • 8
    Tavani A, Gallus S, La Vecchi C, Talamini R, Barbone F, Herrero R, Franceschi S. Diet and risk of oral and pharyngeal cancer. An Italian case-control study. Eur J Cancer Prev 2001; 10: 1915.
  • 9
    Foulkes WD, Brunet JS, Sieh W, Black MJ, Shenouda G, Narod SA. Familial risks of squamous cell carcinoma of the head and neck: retrospective case-control study. BMJ 1996; 313: 71621.
  • 10
    Foulkes WD, Brunet JS, Kowalski LP, Narod SA, Franco EL. Family history of cancer is a risk factor for squamous cell carcinoma of the head and neck in Brazil: a case-control study. Int J Cancer 1995; 63: 76973.
  • 11
    Mork J, Lie AK, Glattre E, Hallmans G, Jellum E, Koskela P, Moller B, Pukkala E, Schiller JT, Youngman L, Lehtinen M, Dillner J. Human papillomavirus infection as a risk factor for squamous-cell carcinoma of the head and neck. N Engl J Med 2001; 344: 12531.
  • 12
    Gillison ML, Koch WM, Capone RB, Spafford M, Westra WH, Wu L, Zahurak ML, Daniel RW, Viglione M, Symer DE, Shah KV, Sidransky D. Evidence for a causal association between human papillomavirus and a subset of head and neck cancers. J Natl Cancer Inst 2000; 92: 70920.
  • 13
    Franceschi S, Munoz N, Bosch XF, Snijders PJ, Walboomers JM. Human papillomavirus and cancers of the upper aerodigestive tract: a review of epidemiological and experimental evidence. Cancer Epidemiol Biomarkers Prev 1996; 5: 56775.
  • 14
    Castellsagué X, Herrero R, Muñoz N, Franceschi S, Bosch FX, Walboomers JMM. Human papilloma virus (HPV) and oral cancer: an international study of molecular epidemiology. In: VicentJJA, ed. 1st World Congress on Head and Neck Oncology. Bologna, Italy: Monduzzi Editore S.p.A., 1998. p 1058.
  • 15
    Sanchez MJ, Martinez C, Nieto A, Castellsague X, Quintana MJ, Bosch FX, Muñoz N, Herrero R, Franceschi S. Oral and oropharyngeal cancer in Spain: influence of dietary patterns. Eur J Cancer Prev 2003; 12: 4956.
  • 16
    Nieto A, Sanchez MJ, Quintana MJ, Castellsague X, Martinez C, Muñoz J, Bosch FX, Muñoz N, Herrero R, Franceschi S. BMI throughout life, intake of vitamin supplements and oral cancer in Spain. IARC Sci Publ 2002; 156: 25961.
  • 17
    Talamini R, Vaccarella S, Barbone F, Tavani A, La Vecchia C, Herrero R, Muñoz N, Franceschi S. Oral hygiene, dentition, sexual habits and risk of oral cancer. Br J Cancer 2000; 83: 123842.
  • 18
    Garrote LF, Herrero R, Reyes RM, Vaccarella S, Anta JL, Ferbeye L, Muñoz N, Franceschi S. Risk factors for cancer of the oral cavity and oro-pharynx in Cuba. Br J Cancer 2001; 85: 4654.
  • 19
    Balaram P, Sridhar H, Rajkumar T, Vaccarella S, Herrero R, Nandakumar A, Ravichandran K, Ramdas K, Sankaranarayanan R, Gajalakshmi V, Muñoz N, Franceschi S. Oral cancer in southern India: the influence of smoking, drinking, paan-chewing and oral hygiene. Int J Cancer 2002; 98: 4405.
  • 20
    Percy C, Van Holten V, Muir CS, Percy C, Van Holten V, Muir CS. International Classification of Disease for Oncology (ICD-O). 2nd ed. Geneva: WHO, 1990.
  • 21
    Coughlin SS, Benichou J, Weed DL. Attributable risk estimation in case-control studies. Epidemiol Rev 1994; 16: 5164.
  • 22
    Tuyns AJ, Esteve J, Raymond L, Berrino F, Benhamou E, Blanchet F, Boffetta P, Crosignani P, del Moral A, Lehmann W. 1988. Cancer of the larynx/hypopharynx, tobacco and alcohol: IARC international case-control study in Turin and Varese (Italy), Zaragoza and Navarra (Spain), Geneva (Switzerland) and Calvados (France). Int J Cancer 1988; 41: 48391.
  • 23
    Blot WJ, McLaughlin JK, Winn DM, Austin DF, Greenberg RS, Preston-Martin S, Bernstein L, Schoenberg JB, Stemhagen A, FraumeniJrJF. Smoking and drinking in relation to oral and pharyngeal cancer. Cancer Res 1988; 48: 32827.
  • 24
    Franco EL, Kowalski LP, Oliveira BV, Curado MP, Pereira RN, Silva ME, Fava AS, Torloni H. Risk factors for oral cancer in Brazil: a case-control study. Int J Cancer 1989; 43: 9921000.
  • 25
    Franceschi S, Bidoli E, Negri E, Barbone F, La Vecchia C. Alcohol and cancers of the upper aerodigestive tract in men and women. Cancer Epidemiol Biomarkers Prev 1994; 3: 299304.
  • 26
    Gronbaek M, Becker U, Johansen D, Tonnesen H, Jensen G, Sorensen TI. Population based cohort study of the association between alcohol intake and cancer of the upper digestive tract. BMJ 1998; 317: 8447.
  • 27
    Franceschi S, Levi F, La Vecchia C, Conti E, Dal Maso L, Barzan L, Talamini R. Comparison of the effect of smoking and alcohol drinking between oral and pharyngeal cancer. Int J Cancer 1999; 83: 14.
  • 28
    Muñoz N, Castellsagué X. Epidemiology of oesophageal cancer. Eur J Gastroent Hepatol 1994; 6: 64955.
  • 29
    Castellsague X, Munoz N. Cancer of the oral cavity and pharynx in nonsmokers who drink alcohol and in nondrinkers who smoke tobacco. J Natl Cancer Inst 1999; 91: 13368.
  • 30
    Castellsagué X, Muñoz N, De Stefani E, Victora CG, Castelletto R, Rolón PA, Quintana MJ. Independent and joint effects of tobacco smoking and alcohol drinking on the risk of esophageal cancer in men and women. Int J Cancer 1999; 82: 65764.
  • 31
    Kjaerheim K, Gaard M, Andersen A. The role of alcohol, tobacco, and dietary factors in upper aerogastric tract cancers: a prospective study of 10,900 Norwegian men. Cancer Causes Control 1998; 9: 99108.
  • 32
    Schlecht NF, Franco EL, Pintos J, Negassa A, Kowalski LP, Oliveira BV, Curado MP. Interaction between tobacco and alcohol consumption and the risk of cancers of the upper aero-digestive tract in Brazil. Am J Epidemiol 1999; 150: 112937.
  • 33
    Merletti F, Boffetta P, Ciccone G, Mashberg A, Terracini B. Role of tobacco and alcoholic beverages in the etiology of cancer of the oral cavity/oropharynx in Torino, Italy. Cancer Res 1989; 49: 491924.
  • 34
    Franceschi S, Barra S, La Vecchia C, Bidoli E, Negri E, Talamini R. Risk factors for cancer of the tongue and the mouth. A case-control study from northern Italy. Cancer 1992; 70: 222733.
  • 35
    Bundgaard T, Wildt J, Frydenberg M, Elbrond O, Nielsen JE. Case-control study of squamous cell cancer of the oral cavity in Denmark. Cancer Causes Control 1995; 6: 5767.
  • 36
    Hayes RB, Bravo-Otero E, Kleinman DV, Brown LM, FraumeniJrJF, Harty LC, Winn DM. Tobacco and alcohol use and oral cancer in Puerto Rico. Cancer Causes Control 1999; 10: 2733.
  • 37
    De Stefani E, Correa P, Oreggia F, Leiva J, Rivero S, Fernandez G, Deneo-Pellegrini H, Zavala D, Fontham E. Risk factors for laryngeal cancer. Cancer 1987; 60: 308791.
  • 38
    Rolón PA, Castellsagué X, Benz M, Muñoz N. Hot and cold mate drinking and esophageal cancer in Paraguay. Cancer Epidemiol Biomark Prev 1995; 4: 595605.
  • 39
    Castelletto R, Castellsagué X, Muñoz N, Iscovich J, Chopita N, Jmelnitsky A. Alcohol, tobacco, diet, mate drinking and oesophageal cancer in Argentina. Cancer Epidemiol Biomark Prev 1994; 3: 55764.
  • 40
    Bosch FX, Cardis E. Black tobacco and cancer: introducing an epidemiological review. Eur J Cancer 1991; 27: 13458.
  • 41
    Leclerc A, Brugere J, Luce D, Point D, Guenel P. Type of alcoholic beverage and cancer of the upper respiratory and digestive tract. Eur J Cancer Clin Oncol 1987; 23: 52934.
  • 42
    Barra S, Franceschi S, Negri E, Talamini R, La Vecchia C. Type of alcoholic beverage and cancer of the oral cavity, pharynx and oesophagus in an Italian area with high wine consumption. Int J Cancer 1990; 46: 101720.
  • 43
    Schlecht NF, Pintos J, Kowalski LP, Franco EL. Effect of type of alcoholic beverage on the risks of upper aerodigestive tract cancers in Brazil. Cancer Causes Control 2001; 12: 57987.
  • 44
    Zambon P, Talamini R, La Vecchia C, Dal Maso L, Negri E, Tognazzo S, Simonato L, Franceschi S. Smoking, type of alcoholic beverage and squamous-cell oesophageal cancer in northern Italy. Int J Cancer 2000; 86: 1449.
  • 45
    La Vecchia C, Franceschi S, Favero A, Talamini R, Negri E. Alcohol intake and cancer of the upper digestive tract. Pattern of risk in Italy is different from that in Denmark. BMJ 1999; 318: 128990.
  • 46
    Huang W-Y, Winn DM, Brown LM, Gridley G, Bravo-Otero W, Diehl SR, FraumeniJrJF, Hayes RB. Alcohol concentration and risk of oral cancer in Puerto Rico. Am J Epidemiol 2003; 157: 8817.
  • 47
    Harris EL. Association of oral cancers with alcohol consumption: exploring mechanisms. J Natl Cancer Inst 1997; 89: 16567.