Cancer Therapy Vol 2, 99-106, 2004

1Registro Nacional de Cáncer, Montevideo, Uruguay.

__________________________________________________________________________________

*Correspondence: Dr. Eduardo De Stefani, Avenida Brasil 3080 dep. 402, Montevideo, Uruguay; Tel.: (598) 2 708 23 14; Fax: (598) 2 402 08 10; E-Mail: estefani@adinet.com.uy

Key Words: hypopharyngeal and laryngeal cancers, tobacco smoking and alcohol drinking,

Abbreviations: Age-standardized (World Population) incidence rates, (ASR's); food frequency questionnaire, (FFQ); International Classification of Diseases for Oncology, (ICD-O); odds ratios, (OR's); tobacco-specific nitrosamines, (TSNA)

Cancers of the hypopharynx and larynx has been analized as a sigle entity in several studies (Tuyns et al, 1988; Cattaruzza et al, 1996). According to the International Classification of Diseases for Oncology (ICD-O) (WHO, 1976), hypopharyngeal lesions includes tumors of sinus pyriform, postcricoid region, aryepiglottic fold, posterior wall of hypopharynx and cancers of hypopharynx not elsewhere classified. Accordind to the same classification, laryngeal tumors are divided in glottis lesions, supraglottis tumors, subglottic tumors and lesions of larynx not elsewhere classified (WHO, 1976). Previous studies divided laryngeal cancers in extrinsic and intrinsic (Wynder et al, 1976; Tuyns et al, 1988). Most of the tumors (99.5 %) arising in both sites are squamous cell carcinomas (Austin and Reynolds, 1996).

Hypopharyngeal and laryngeal cancers are very frequent among Uruguayan men (Parkin et al, 1997). According to international comparisons between registries in the Americas, both sites are in first place, followed by Blacks in United States (Parkin et al, 1997). Age-standardized (World Population) incidence rates (ASR's) were of 3.3 per 100,000 men for hypopharyngeal carcinomas, whereas the rate for laryngeal cancer were of 12.2 per 100, 000 men (Parkin et al, 1997).

Tobacco smoking and alcohol drinking are the major risk factors for both sites (Wynder et al, 1976; Tuyns et al, 1988; Austin and Reynolds, 1996; Cattaruzza et al, 1996). Also diet has received particular attention in recent studies (Estève et al, 1996; Riboli et al, 1996; World Cancer Research Fund, 1997). Therefore, we decided to conduct a case-control study in order to compare the risks for smoking and drinking by tumor site in the high-risk population of Uruguay. The a priori hypothesis was that hypopharyngeal and laryngeal are different epidemiological entities. The role of diet will be analyzed in a companion report.

In the time period 1997-2003 all newly diagnosed and microscopically verified squamous cell carcinomas of the hypopharynx and larynx which occurred in men, were considered eligible for this study. Three-hundred and twenty eight (328) of cases were identified in the four major hospitals in Montevideo. Eight patients were excluded from the cases due to phonation problems, leaving a final number of 320 patients (response rate 97.5 %). After careful endoscopic examination by one of the authors (F.O.), the cases we classified as follows: A. hypopharyngeal carcinomas (85 cases, 26.6 %) and B. laryngeal carcinomas (235, 73.4 %). Most of hypopharyngeal tumors were located in the sinus pyriform (78 patients), whereas laryngeal cancers were distributed as follows: glottic lesions (49 cases), supraglottis lesions (67 cases) and transglottis lesions (119 cases). The term transglottis refers to lesions which involved both supraglottis and glottis.

In the same time period and in the same hospitals, 1235 men which were hospitalized for diseases not related with tobacco smoking, alcohol drinking and without recent changes in their diets were considered eligibles for the study. Thirty five patients refused the interview, leaving a final number of 1200 potential controls (response rate 97.2 %). From this pool of patients, 640 men were frequency matched with cases on age (in ten year intervals) and residence (Montevideo, other counties). The resulting case-control ratio was 1:2. Controls presented the following diseases: abdominal hernia (146 patients, 22.8 %), eye disorders (135, 21.1 %), fractures (70, 10.9 %), injuries (62, 9.7 %), skin diseases (56, 8.8 %), acute appendicitis (52, 8.1 %), hydatid cyst (32, 5.0 %), varicose veins (32, 5.0 %), urinary stones (27, 4.2 %), blood disorders (19, 3.0 %) and osteoarticular diseases (9, 1.4 %).

Both series of participants (cases and controls) were interviewed in the hospitals by two trained social workers. The interviews were performed shortly after admittance to the hospitals. No proxy interviews were accepted. The patients were administered with a questionnaire which included the following sections: (A) sociodemographics, (B) a complete occupational section based in job titles and its duration, (C) history of cancer among first degree relatives, (D) self-reported height and weigh five years before the date of the interview, (E) a complete history of tobacco smoking (age at start, age at quit, number of cigarettes smoked per day, type of tobacco, type of cigarette, inhalation), (F) a complete history of alcohol drinking (age at star, age of quit, number of glasses drunked per day; this was repeated for the main alcoholic beverages consumed in Uruguay: beer, wine and hard liquor), (G) a complete history of maté drinking (maté is the folk name of a local tea prepared by infusion of the herb Ilex paraguariensis; this beverage is usually drunk hot or very hot) and (H) a food frequency questionnaire (FFQ) on 64 food items. This FFQ is considered as representative of the usual diet of Uruguayan population and, although it was not validated, it was tested for reproducibility with reasonably good results. Furthermore, the FFQ allowed the consumption of total energy. All queries on foods referred to the consumption two years before the date of the interview or before the date of the first symptom.

Patients who smoked less than 100 cigarettes in their lifetime were considered never smokers. Smokers who used cigarettes at the time of the interview or who had quitted one year before the interview were defined as current smokers. The remaining patients were defined as former smokers. Regarding type of tobacco, patients were divided into pure smokers of blond or black tobacco when they had smoked each type of tobacco during more than 85 % of their lifetimes. The remaining patients were defined as mixed smokers. Patients who had drunk occasionally or less than monthly were considered never drinkers. Participants who drunk alcohol beverages at the date of the interview or who quitted one year before the interview were defined as current drinkers. The remaining patients were included in the category of former drinkers. Binge drinkers, that is periodic heavy drinkers, were not identified according to the questionnaire. The amount of alcohol drunk was expressed as mililiters of ethanol per day, according to the following calculations: beer-6 % of ethanol per liter, wine-12 % of ethanol per liter and hard liquor-46 % of ethanol per liter. Among types of wine, red wine is almost exclusively consumed by Uruguayan population, in particular the low socioeconomical population (Comisión Honoraria de Lucha contra el Cáncer, 1993). Regarding hard liquor, Uruguayan population consumes grappa (hard liquor derived fron grapes) and caña (hard liquor from sugarcane) (Comisión Honoraria de Lucha contra el Cáncer, 1993).

Relative risk, approximated by the odds ratio (OR) and corresponding ninety five per cent confidence intervals (95 % CI) were estimated by unconditional multiple logistic regression (Breslow and Day, 1980). Comparisons between hypopharyngeal and laryngeal cancers were carried out using multinomial (polytomous) regression (Hosmer and Lemeshow, 1989; Rothman and Greenland, 1998). Comparisons between hypopharyngeal and laryngeal cancers were carried out using multinomial (polytomous) regression (Hosmer and Lemeshow, 1989; Rothman and Greenland, 1998). OR's for tobacco smoking variables were obtained after fitting the following model: age (categorical, 6 strata), residence (ordinal, 2 strata), urban/rural status (ordinal, 2 strata), education (categorical, 3 strata), body mass index (categorical, 4 strata) and alcohol drinking (categorical, 5 strata). The model fitted for alcohol variables was similar, replacing alcohol drinking by tobacco smoking (pack years, categorical, 5 strata).

Test for trend were performed after entering categorical variables as ordinal (continuous) in the same model. Departure from the multiplicative model was determined by assessing the likelihood ratio test statistic. An alpha of 0.05 was used as the indicator of statistical significance and, accordingly 95 % Ci s were reported. All Ps were derived from two-sided statistical tests. All the calculations were done with the STATA programme (Stata Reference Manual, 1999).

Distribution of cases and controls by sociodemographic variables, body mass index and total calories are shown in Table 1. As result of the frequency matched design, age, residence and urban/rural status were rather similar. Also education and monthly income were similar. High body mass index was associated with a

Table 1. Distribution of controls and cases by sociodemographic variables and selected risk factors.

1-Education, income, body mass index and total calories are adjusted for each other.

significant reduction in risk (OR 0.3, 95 % CI 0.2-0.5). On the contrary, cases consumed significant higher amounts of energy compared with controls (OR 3.6, 95 % CI 2.3-5.5). It should be taken into account that this estimation includes alcohol energy plus non-alcoholic sources of energy.

Odds ratios of hypopharyngeal and laryngeal carcinomas for tobacco smoking are shown in Table 2. Although the smoking pattern was characterized by higher risks among laryngeal cancers compared with hypopharyngeal carcinomas, the differences between both sites were non significant. Current smokers were associated with an increased risk among laryngeal carcinomas (OR 10.7, 95 % CI 4.3-26.4), whereas the same category of smokers displayed an OR of 6.7 (95 % CI 1.5-30.9, p-value for heterogeneity=0.96). Ex-smokers showed a higher reduction in risk among hypopharyngeal cancers (OR 3.2, 95 % CI 0.6-15.6) compared with laryngeal carcinomas (OR 6.2, 95 % CI 2.5-15.7). Variables which measured the smoking intensity (number of cigarettes/day and number of cigarettes/day among current smokers) were associated with much higher increase in risk among laryngeal tumors compared with hypopharyngeal carcinomas (p-value for heterogeneity=0.12). The same fact was observed for variables which measured smoking duration (years of smoking and years since quit). It should be pointed out that smoking intensity was associated with higher risks compared with smoking duration. This was observed both for hypopharyngeal and laryngeal carcinomas. Cumulative exposure to tobacco smoking (pack years) was associated with an elevated risk in laryngeal carcinomas (OR for heavy smokers 21.3, 95 % CI 8.4-54.3). On the other hand, hypopharyngeal cancers displayed less impressive risk (OR for heavy smokers 9.2, 95 % CI 1.9-44.2). The differences between both locations were close to statistical significance (p-value for heterogeneity=0.07). The effect of type of tobacco yielded non-significant results. More precisely, smokers of black tobacco were associated with a modest increase in risk, more evident in carcinomas of the hypoharynx (reference category: smokers of blond tobacco). On the other hand, smokers of hand-rolled cigarettes were associated with a significant increase of risk of 1.9 (95 % CI 1.3-2.8) among carcinomas of the larynx (reference category: smokers of manufactured cigarettes). The effect of hand-rolling was less marked and not formally significant in squamous cell carcinomas of the hypopharynx. Lifelong users of filter cigarettes were associated with a reduction in risk of 0.6 in both sites and the differences were not significant. Finally, inhalers displayed an increase in risk in hypopharynx and larynx, being the effect higher in laryngeal carcinomas.

Odds ratios of hypopharyngeal and laryngeal carcinomas for alcohol drinking variable are shown in Table 3. Current drinkers displayed an OR of 6.0 (95 % CI 2.0-18.0) for hypoharyngeal cancers, whereas the risk

Table 2. Odds ratios of hypopharyngeal and laryngeal cancers for tobacco smoking variables (1)

1-Adjusted for age (categorical), residence, urban/rural status, education (categorical), body mass index (categorical), and total ml. of ethanol of alcohol drinking (categorical).

Table 3. Odds ratios (and 95 % CI) of hypopharyngeal and laryngeal cancers for alcohol drinking

1-Adjusted for age (categorical), residence, urban/rural status, education (categorical), body mass index (categorical), and tobacco smoking (pack years categorical).

2-Adjusted for age (categorical), residence, urban/rural status, education (categorical), body mass index (categorical), tobacco smoking (pack years categorical), wine (categorical) and hard liquor (categorical).

3-Adjusted for age (categorical), residence, urban/rural status, education (categorical), body mass index (categorical), tobacco smoking (pack years categorical), beer (categorical) and hard liquor (categorical).

4-Adjusted for age (categorical), residence, urban/rural status, education (categorical), body mass index (categorical), tobacco smoking (pack years categorical), beer (categorical) and wine (categorical) .

among laryngeal carcinomas was only of 1.6 (95 % CI 0.9-2.5, p-value for heterogeneity 0.02). The number of beer drinkers among Uruguayan males is extremelly low (10 cases among hypopharynx lesions and 30 among laryngeal cancers). Thus, it is not surprising that resulting OR's were non-significant. On the other hand, red wine is the preferred alcoholic beverage in Uruguay. Although red wine intake was associated with higher OR´s among patients with hypopharyngeal cancers compared with cancers of the larynx, the differences were non-significant. Hard liquor consumption displayed an increased risk among hypopharyngeal carcinomas (OR 3.3, 95 % CI 1.6-6.8), whereas laryngeal cancers showed a modest elevation in risk of 1.5. The test for heterogeneity was statistical significant (p-value=0.03). Total alcohol consumption, years of drinking alcohol, years of cessation alcohol drinking and lifelong consumption of alcohol (alcohol years) were directly associated with elevated risk for patients with hypopharyngeal carcinomas (OR for heavy drinkers of alcohol 12.8, 95 % CI 4.0-41.2, p-value for trend <0.0001). On the other hand, cancers of the larynx displayed less impressive effects (OR for heavy drinkers of alcohol 2.5, 95 % CI 1.4-4.5). All the variables above mentioned showed significant differences between both sites (p-value for heterogeneity=0.03 for total alcohol drinking). Finally, pattern of consumption of alcoholic beverages displayed an OR of 6.9 for mixed drinkers (95 % CI 2.2-21.3) among hypopharyngeal lesions compared with 1.7 (95 % CI 1.0-2.9, p-value for heterogeneity=0.05) in the same category among laryngeal carcinomas.

Joint effects of tobacco smoking (cigarettes/day) and alcohol drinking for hypopharyngeal and laryngeal cancers are shown in Table 4. Both sites displayed independent effect for tobacco smoking and alcohol drinking (see marginals). Whereas the effect of alcohol drinking was much higher in patients with hypopharyngeal lesions, tobacco smoking displayed higher effect in laryngeal cancer. These differences by site were statistical significant (p-value for heterogeneity=0.04 for tobacco and 0.02 for alcohol drinking). Curiously joint effects for heavy smokers and heavy drinkers were rather similar in both sites (OR ~43). Whereas this high risk was due to alcohol in hypopharyngeal cancers, joint OR's should be attributed to tobacco smoking in laryngeal lesions. The results followed a multiplicative model.

According to our results, tobacco smoking is a major risk factor for laryngeal cancer, whereas alcohol drinking displayed significant increases in risk among hypopharyngeal carcinomas. Moreover, whereas the effect of tobacco smoking is not significant different between both tumor sites (although cigarettes per day and pack years were much higher in laryngeal carcinoma compared with hypopharyngeal lesions), alcohol drinking displayed significant heterogeneity. These results replicates the findings of Tuyns et al. in the large multicenter study of IARC (Tuyns et al, 1988). In this study, results for hypopharynx and epilarynx carcinomas are compared with

endolaryngeal lesions (Tuyns et al, 1988). These authors strongly suggested that 68 % of the hypopharyngeal cancers versus 28 % of the endolaryngeal cancers are attributable to alcohol drinking. Other studies (Brugere et al, 1986; Barón et al, 1993) reported an elevated risk of hypopharyngeal carcinomas for alcohol drinking, after controlling for tobacco smoking.

Concerning the mechanisms of tobacco smoking, it is clear that this risk factor is a rich source of carcinogens (IARC, 1986). In particular, tobacco contains high amounts of tobacco-specific nitrosamines (TSNA). These compounds have been implicated in the carcinogenesis of lung, oral cavity, pharynx, larynx, esophagus and urinary bladder (Hecht, 2002). In particular, air-cured tobacco (black tobacco) has greater concentrations of TSNA compared with flue-cured tobacco (blond tobacco). In discordance with previous studies (Benhamou et al, 1985; De Stefani et al, 1987; De Stefani et al, 1988, 1993; Boffetta, 1993; Sancho-Garnier and Theobald, 1993), the present study failed to show a significant increase in risk among smokers of black tobacco. This could be the results of the decline in the sales of black tobacco in the Uruguayan market (De Stefani et al, 1994). On the other hand, in the present study, hand-rolled cigarettes were associated with a significant increase in risk for squamous cell laryngeal carcinomas, and, in a lesser degree for hypopharyngeal carcinoma. Our results replicates findings from previous studies (De Stefani et al, 1992, 1993; Launoy et al, 2000). Since hand-made cigarettes are also filled with blond tobacco, there is uncertaintly concerning about the chemical composition of the blond tobacco used for fill hand-rolled cigarettes. More precisely, hand-made blond tobacco could contain higher amounts of carcinogenic chemicals compared with the flue-cured tobacco used for manufactured cigarettes. Further studies are needed in order to clarify this issue.

At difference with the mechanisms of action of tobacco, alcohol drinking has been the subject of considerable debate (IARC, 1988; Blot, 1999). Recent reviews suggested that alcohol drinking acts through ethanol or its major metabolite (acetaldehyde) (World Cancer Research Fund, 1997, Blot, 1999). On the other hand, beer, wine and hard liquor could contain carcinogenic substances (Schlecht et al, 2001). Also, it is possible that alcohol could be a solvent for tobacco carcinogens or facilitate the action of these carcinogens inducing injury of the mucosa (Blot, 1999). Since hypopharyngeal mucosa is in direct contact with alcohol, this mechanism could fit with our findings. On the other hand, laryngeal mucosa is mainly related with the inhaled tobacco.

As other case-control studies, our study has a several limitations. Perhaps the most important drawback is the potential for selection bias. This bias is almost impossible to exclude. We tried to minimize selection bias by frequency matching cases and controls on age, residence and urban/rural status. Another important bias is classification bias. Since is widely known that tobacco smoking and alcohol drinking are the main determinants of these malignancies, both patients and interviewers could induce differential reponse of the cases. This could result in results close to the null. Our study has also strengths. The precise validation of the lesion by an expert endoscopist is a strength. Also, the high response rate (both for cases and controls) is another important strength.

In summary, we conducted a case-control study in Uruguay in order to compare OR's for hypopharyngeal and laryngeal carcinomas. Tobacco smoking was associated with higher risks in lesions of the larynx, but

Table 4. Joint effects of tobacco smoking and alcohol drinking in hypopharyngeal and laryngeal cancers (1)

1-Adjusted for age, residence, urban/rural status, education, body mass index and for each other.

the differences with hypopharyngeal squamous cell carcinomas of the hypopharynx did not reach statistical significance. Most of the variables related with alcohol drinking displayed significantly higher OR's among hypopharyngeal carcinomas compared with laryngeal cancers. These results suggest that, concerning smoking and drinking, hypopharyngeal and laryngeal could be distinct epidemiological entities. Therefore, these tumors sites should not be joined as a single disease in future diseases.

Supported by a grant from International Agency for Research on Cancer.