Tobacco and alcohol, known to be involved in the carcinogenesis of esophageal cancer, have been reported to interact in a multiplicative way in the etiology of this neoplasm.1, 2 Areca nut chewing, or betel quid chewing, a common habit in Southeast Asia, has been found recently by epidemiological studies from India, Taiwan and Thailand, to increase the risk of developing esophageal cancer by 4.7–13.3-fold,3, 4, 5 although other exogenous risk factors may be also have contributed to the development of the disease in these studies. In areas where cigarette smoking and alcohol consumption are widespread and the users of these 2 substances are also betel quid chewers, the relationship between the use of these 3 substances and esophageal cancer require study.
Cancer of the esophagus was the sixth leading cause of cancer death among men in Taiwan in 2002.6 The incidence of this tumor increased by >70% between 1990–1999.7 Although the prevalence rates of tobacco use and alcohol intake, two substances associated with risk of esophageal cancer, have slightly decreased in Taiwan in recent years, the use of betel quid has increased and the chewers have become younger.8 Because betel nut use occurs in 33.6% of the smoking population and in 35.9% of those who consume alcohol8 and because, unlike other Southeast Asia countries, betel nut is not chewed with tobacco in Taiwan,9 we present a case-control study investigating the independent and combined effects of alcohol intake, tobacco smoking and betel quid chewing on the development of esophageal cancer and their attributable risks.
Material and methods
Selection of cases and controls
This was a multicenter case-control study conducted in northern and southern Taiwan. There were 3 hospitals in the large collaborative study, including National Taiwan University Hospital (NTUH), located in northern Taiwan, Kaohsiung Medical University Hospital (KMUH) and Kaohsiung Veterans General Hospital (KVGH), both located in tropical southern Taiwan. These hospitals are all highly regarded medical centers in their geographic areas and are accessible to patients from all socioeconomic groups.
A system of rapid case recognition was introduced in our study for the determination of esophageal cancer cases, so that those patients could be identified and selected into our study as soon after diagnosis as possible. In this system, all newly hospitalized subjects who presented at the Department of Chest Surgery in NTUH during July, 1996 to December, 2003, and visited the Department of Chest Surgery and the Department of Gastroenterology at KMUH or the Department of Chest Surgery at KVGH during August, 2000 to December, 2003 suspected of suffering from primary invasive cancer of the esophagus (ICD 9th Revision, code 150) were traced from reports of endoscopy, surgery and radiology, and verified by histological materials. Patients with tumors classified as not mixed or undifferentiated were initially considered eligible and approached for study participation. Only the patients who have consistently histologically confirmed with squamous cell carcinoma of the esophagus by the study pathologists were included. In the process of verification, a standardized criterion for the determination of cancer cases was applied to the 3 attending hospitals. Among the 795 pathology-proved cases, 513 patients (64.5%), 468 men and 45 women (28–89 years of age) were recruited. We excluded 35.5% of cases due to patients who were not mentally alert and coherent, who had been discharged by the time we visited the ward and who refused to be interviewed or lacked blood specimens. All recruited patients were interviewed within 1 week of diagnosis of the disease.
The controls were derived from the same geographic areas as cases. They were chosen from healthy community residents who attended the hospitals for routine physical checkups at the Departments of Preventive Medicine from the same hospitals. All healthy subjects >25 years of age were considered eligible controls. Identified from a developed system for the recognition of background data, the first to third eligible controls were selected within 4 weeks after each esophageal carcinoma patient was identified and matched to case on gender and age (within 3-years). Of the 861 matched controls, 818 subjects (95.0%), 752 men and 66 women (26–89 years of age) were recruited and agreed to be interviewed as the control group. Among them, one control subject was matched to each of the 224 cases, due to only a single match could be found within 4 weeks after the case was identified. All remaining cases were matched to 2 controls, except for 36 cases for each of whom 3 appropriate controls were recruited. The reason we could not approach the refusing controls was that they did not want to be disturbed.
Our study was conducted after the study protocol was reviewed and approved by the Institutional Review Boards of NTUH and KMUH. Both cases and controls gave written, informed consent for the interview and the tracing of their medical records. A pre-designed and pre-tested questionnaire was developed specifically for our study. The interviews were carried out by well-trained interviewers who elicited detailed information on demographic characteristics, substance uses, other beverage consumption and daily diet. The interviewers allowed no one but the case patient or the control to answer the questions.
Ever drinkers were subjects who had consumed any alcoholic beverage, including beer, wine or distilled spirits, at least once per week for a minimum of 6 months. Ever smokers were subjects who had smoked 10 cigarettes or more for at least 6 months, and ever chewers were subjects who had chewed one betel nut, measured as quid, or more per day for >6 months. The group of ever users could be further categorized into current users and former users. Current users were those who had practiced any of these habits within 1 year before the interview. Former users were those who had stopped any of the habits for at least 1 year before diagnoses or interviews. If the subjects were found to have once used a substance, then the interviewer obtained a detailed history of their drinking, smoking and chewing habits, including when they began and stopped using the substance, how much of the substance they consumed daily, and how long they had used the substance. If they were found to be former users, then they were asked how long it had been since they gave up the use of the substance. Additional information was collected with regard to the kinds of alcohol they consumed and how the smokers inhaled the tobacco. If they were betel quid users, then additional information was collected on the type of materials that were consumed with the betel nut, which in Taiwan are usually inflorescence of Piper Betel Linn, a piece of betel leaf or stem, or mixed. Nevertheless, all types of betel quid products do not use tobacco as an ingredient.
When assessing the risk of total alcohol intake, the types of alcohol beverage, the average frequency and amount of alcohol use for each type of beverage (according to average ethanol content of the beverage and the total amount consumed) was converted to total consumption of alcohol in grams per day. When exploring the effects of cumulative lifetime exposure to these 3 substances on the risk of esophageal cancer, the number of ‘gram/day-years’ and ‘pack-years’ were calculated by multiplying the amount of the substances consumed, ethanol grams per day for drinking, 20-cigarette packs per day for smoking or 10-betel quid packs per day for chewing by the years of the substance used.
Data on the frequency of coffee intake and various types of tea was separately collected. Daily dietary habits were assessed by measuring the consumption of 20 food groups according to 3 time periods (before 20 years, 20–40 years and after 40 years of age). Food intake questions were directed at the frequency and quality for each time period. Only the consumption of the latest period for each patient was used for data analyses.
Odds ratios (OR), as an estimate of the relative risk, and corresponding 95% confidence intervals (CI) were computed for esophageal cancer in relation to uses of the substance by using unconditional logistic regression analyses. All regression models included the matched factors, age (<41, 41–50, 51–60, 61–70, >70 years old), gender and hospital of study as covariates. To control for potential confounding effects, all calculations of OR for esophageal cancer were adjusted for level of education (<7, 7–12, >12 years of education), consumption of fruits (No, 1–7, 8–14, >14 times per week), of vegetables (No, 1–7, 8–14, >14 times per week) and, where appropriate, the use of other substances. Significant tests for linear trend in the OR across exposure strata were calculated by categorizing exposure variables and treating scored variables as continuous. The joint effects of the use of these 3 substances on the risk of esophageal cancer were evaluated on both multiplicative and additive scales. The binary interaction terms calculated by multiplying the indicators for 2 explored risk factors were added to the main effect models, and their significance was tested by the likelihood ratio statistic based on a multiplicative model. The synergism index (SI) proposed by Rothman was computed to assess the empirical deviation from the additive interaction relationship.10 An observed SI value that departed substantially from the expected additive null, i.e., SI not equal to 1, suggested an additive interaction effect. These OR and their variance covariance matrix were used to calculate values for SI and their 95% CI.10 In addition, the proportion of esophageal cancer cases attributable to one or all risk factors considered (population attributable risk percent; PAR%) and corresponding 95% CI were calculated according to the Bruzzi et al.11 and Benichou et al. methods.12 This method provides adjusted PAR% estimates by combining adjusted OR estimates and the observed prevalence of the risk factor among case patients.
As the demographic characteristics found in Table I show, the cases and controls had generally comparable study hospitals, genders and ethnicity. Although the average age of esophageal cancer patients (61.0 ± 12.0 years) was somewhat higher than that of control subjects (59.6 ± 11.4 years), the distribution of age between the 2 groups was similar. Educational level was associated with the risk of esophageal cancer in that patients with higher levels of education, >6 years, were at lower risk of squamous cell carcinoma of the esophagus than the others (OR = 0.1–0.4).
Table I. Distribution and Odds Ratios of Esophageal Cancer Associated With Selected Demographic Factors, Taiwan, 1996–2003
|Study hospital2|| || || |
| NTUH||333 (64.9)||523 (63.9)|| |
| KMUH||80 (15.6)||136 (16.6)|| |
| KVGH||100 (19.5)||159 (19.4)|| |
|Age (years)|| || || |
| <41||20 (3.9)||30 (3.7)|| |
| 41–50||102 (19.9)||180 (22.0)|| |
| 51–60||124 (24.2)||225 (27.5)|| |
| 61–70||141 (27.5)||218 (26.7)|| |
| >70||126 (24.6)||165 (20.2)|| |
|Gender|| || || |
| Female||45 (8.8)||66 (8.1)|| |
| Male||468 (91.2)||752 (91.9)|| |
|Ethnicity|| || || |
| Fukienese||368 (71.7)||608 (74.3)||1.0|
| Aborigines||99 (19.3)||130 (15.9)||1.2 (0.9–1.6)|
| Hakka||22 (4.3)||38 (4.7)||0.9 (0.5–1.6)|
| Others||24 (4.7)||42 (5.1)||0.9 (0.6–1.6)|
|Years of education|| || || |
| <7||283 (55.2)||206 (25.2)||1.0|
| 7–12||196 (38.2)||329 (40.2)||0.4 (0.3–0.5)|
| >12||34 (6.6)||283 (34.6)||0.1 (0.1–0.2)|
Characteristics of alcohol intake and tobacco smoking history associated with risks of esophageal cancer are shown in Tables II and III. Former drinkers and current drinkers experienced, respectively, a 5.5- and 7.6-fold higher risk of esophageal cancer than nondrinkers; however, the elevated risks were only 3.4- and 4.2-fold for former smokers and current smokers. Drinking patients consistently had significantly higher risks for esophageal cancer at each exposure level of drinking duration and quantity. Similar risk patterns were observed among the patients of tobacco smoking. After further adjusting for the amount of a substance consumed or length of time it was consumed, we found that the amount of alcohol consumed had a stronger effect (OR = 2.29 in SD unit change) on the development of esophageal squamous cell carcinoma than the length of time it was consumed (OR = 1.84). In smokers, however, the length of time tobacco had been consumed had a considerably stronger effect on the risk for this cancer than the amount smoked (OR = 1.78 vs. OR = 0.98). The differences in OR of esophageal cancer between intensity and duration of alcohol and of tobacco consumption were strengthened especially when comparing current users with nonusers (intensity compared to duration: OR = 2.57 vs. 0.92 for drinking; OR = 0.90 vs. 2.42 for smoking). Former drinkers and smokers were found to have a significantly lower risk than current users of these substances. Abstinence from the use of alcohol for >10 years and tobacco for >5 years reduced the risk by 70% and 60%, respectively. Drinkers of distilled spirits exhibited the highest esophageal cancer risk among all types of drinkers. Smokers who inhaled the tobacco experienced a higher risk than non-inhaling smokers.
Table II. Odds Ratios of Esophageal Cancer Associated With Alcohol Drinking, Taiwan, 1996–2003
|Former-drinker||114||76||5.5 (3.6–8.6)|| |
|Age at starting drinking (years)|| || || || |
| ≥26||128||100||6.2 (4.0–9.4)|| |
| <26||275||134||7.4 (5.0–10.8)||<0.0001|
|Years of alcohol drinking|| || || || |
| 1–15||45||43||5.5 (3.1–9.9)|| |
| 16–35||203||131||6.1 (4.1–9.2)|| |
| >35||155||60||8.6 (5.5–13.5)||<0.0001|
| SD unit change3|| || ||1.84 (1.55–2.18)4|| |
|Average alcohol intake (g/day)|| || || || |
| 1–20||119||144||3.6 (2.4–5.4)|| |
| 21–40||66||46||6.1 (3.6–10.3)|| |
| >40||218||44||19.5 (12.1–31.2)||<0.0001|
| SD unit change3|| || ||2.29 (1.69–3.08)4|| |
|Type of beverage|| || || || |
| Beer||186||129||5.9 (4.0–8.8)|| |
| Wine||159||76||7.7 (5.0–11.8)|| |
| Spirits||58||29||9.0 (5.0–16.3)|| |
|Years since quitting drinking|| || || || |
| Current-drinker2||289||159||1.0|| |
| 1–5||66||25||1.3 (0.7–2.4)|| |
| 6–10||22||20||0.8 (0.4–1.8)|| |
| >10||26||30||0.3 (0.1–0.6)||0.002|
| Never-drinker||110||584||0.1 (0.1–0.2)|| |
Table III. Odds Ratios of Esophageal Cancer Associated With Cigarette Smoking, Taiwan, 1996–2003
|Former-smoker||116||127||3.4 (2.1–5.3)|| |
|Age at starting smoking (years)|| || || || |
| ≥26||48||84||1.9 (1.1–3.3)|| |
| <26||386||254||4.6 (3.1–6.8)||<0.0001|
|Years of cigarette smoking|| || || || |
| 1–15||15||42||2.3 (1.0–4.9)|| |
| 16–30||134||130||3.8 (2.4–6.1)|| |
| >30||285||166||4.4 (2.9–6.7)||<0.0001|
| SD unit change3|| || ||1.78 (1.30–2.44)4|| |
|Average amount of cigarette|| || || || |
| 1–20||124||131||3.5 (2.2–5.6)|| |
| 21–40||291||199||4.0 (2.7–6.1)|| |
| >40||19||8||4.6 (1.5–13.7)||<0.0001|
| SD unit change3|| || ||0.98 (0.76–1.26)4|| |
|Inhalation of smoking|| || || || |
| No||73||88||2.9 (1.7–4.8)|| |
| Yes||361||256||4.0 (2.7–6.0)||<0.0001|
|Years since quitting smoking|| || || || |
| Current-smoker2||318||211||1.0|| |
| 1–5||71||46||1.4 (0.8–2.3)|| |
| 6–10||12||28||0.4 (0.2–0.9)|| |
| >10||33||53||0.5 (0.3–0.9)||0.020|
| Never-smoker||79||480||0.2 (0.2–0.3)|| |
Table IV shows the risks of contracting esophageal cancer for betel quid chewing. Cancer risks were associated with the years and the amount betel quid had been consumed. A dose-response relationship was found between exposure levels and esophageal cancer risks. Patients who chewed areca nut with a piece of Piper Betel Linn inflorescence experienced the highest risk of esophageal cancer (OR = 2.9). No abstinence effect for betel quid chewing was observed.
Table IV. Odds Ratios of Esophageal Cancer Associated With Betel Quid Chewing, Taiwan, 1996–2003
|Former-chewer||82||29||2.2 (1.3–3.9)|| |
|Age at starting chewing (years)|| || || || |
| ≥30||44||25||1.3 (0.7–2.5)|| |
| <30||160||51||2.7 (1.8–4.3)||<0.0001|
|Years of betel quid chewing|| || || || |
| 1–10||29||20||1.8 (0.9–3.8)|| |
| 11–20||41||23||2.2 (1.1–4.2)|| |
| >20||134||33||3.1 (1.9–5.1)||<0.0001|
| SD unit change3|| || ||0.82 (0.58–1.16)4|| |
|Average amount of chewing|| || || || |
| 1–10||71||45||1.6 (0.9–2.6)|| |
| 11–20||59||18||3.3 (1.7–6.4)|| |
| >20||74||13||3.9 (1.9–7.9)||<0.0001|
| SD unit change3|| || ||1.01 (0.82–1.25)4|| |
|Types of material|| || || || |
| With betel inflorescence||123||43||2.9 (1.8–4.7)|| |
| With betel leaf or stem||13||8||1.2 (0.4–3.7)|| |
| Mixed||68||25||2.1 (1.1–3.9)|| |
|Years since quitting chewing|| || || || |
| Current-chewer2||122||47||1.0|| |
| 1–5||31||10||1.0 (0.4–2.4)|| |
| 6–10||24||5||2.2 (0.7–7.0)|| |
| >10||27||14||0.7 (0.3–1.7)||0.848|
| Never-chewer||309||742||0.5 (0.3–0.8)|| |
The effects of combined exposure to alcohol, tobacco and betel quid on the risk of esophageal cancer are explored in Table V. Smoking patients who did not consume alcohol and betel quid had a 2.0-fold cancer risk compared with those who did not consume any type of substances. The risks were increased substantially for tobacco smokers who also practiced the habits of chewing (OR = 3.8) or drinking (OR = 20.4). Although the role of betel quid chewing on the development of esophageal cancer for patients never exposed to other habits could not be assessed because there were no cancer cases in this category, chewers who had the habit of drinking exhibited a 13.7-fold elevated risk. The strongest effect (OR = 41.2) on risk for squamous cell carcinoma of the esophagus was detected for combined uses of the 3 substances.
Table V. Odds Ratios of Esophageal Cancer Associated With Combined Uses of Alcohol, Tobacco And Betel Quid, Taiwan, 1996-20031
The 2-way synergistic effects of alcohol intake, tobacco smoking and areca nut chewing on esophageal cancer are evaluated in both additive and multiplicative interaction models (Table VI). The risk (OR = 19.7) for alcohol drinkers who smoked were significantly departure from the expected joint risk (OR = 4.4) estimated by a multiplicative interaction model (LRT-χ,2 15.59; p = 0.0001), whereas the joint effects for drinking and chewing as well as smoking and chewing were described well by a non-interaction multiplicative model (LRT-χ2 < 3.84; p > 0.05). Based on the model of additive scale, a significantly additive interaction relationship was found between the simultaneous uses of alcohol and betel quid and esophageal cancer (SI = 2.7; 95% CI = 1.6–4.4).
Table VI. The Interaction Effects of Alcohol Drinking, Cigarette Smoking And Betel Quid Chewing on Esophageal Cancer, Taiwan, 1996–2003
|Alcohol drinking/Cigarette smoking4|| || || || || || || || |
| Never/Never||59||389||1.0|| || || || || |
| Never/Ever||51||195||1.9 (1.2–3.2)|| || || || || |
| Ever/Never||20||91||2.3 (1.2–4.4)|| || || || || |
| Ever/Ever||383||143||19.7 (12.4–31.3)||2.2||8.2 (4.1–16.5)||4.4||15.59||0.0001|
|Alcohol drinking/Betel quid chewing5|| || || || || || || || |
| Never/Never||101||555||1.0|| || || || || |
| Never/Ever||9||29||1.5 (0.6–3.7)|| || || || || |
| Ever/Never||208||187||6.3 (4.4–9.2)|| || || || || |
| Ever/Ever||195||47||16.8 (10.3–27.2)||5.8||2.7 (1.6–4.4)||9.5||1.33||0.248|
|Cigarette smoking/Betel quid chewing6|| || || || || || || || |
| Never/Never||72||470||1.0|| || || || || |
| Never/Ever||7||10||2.7 (0.8–9.5)|| || || || || |
| Ever/Never||237||272||4.0 (2.6–6.0)|| || || || || |
| Ever/Ever||197||66||8.8 (5.2–14.8)||4.7||1.7 (0.7–3.7)||10.8||0.10||0.753|
The adjusted risks of contracting esophageal cancer associated with cumulative lifetime uses of alcohol, tobacco and betel quid, and corresponding population attributable risk proportions are summarized in Table VII. Patients who consumed >900 g/day-year of ethanol, >50 pack-years of cigarettes and >20 pack-years of betel quid separately experienced a 13.9-, 4.2- and 2.8-fold cancer risks. Alcohol intake and tobacco consumption were both the major contributors to the burden of esophageal squamous cell carcinoma in this population, with ever drinking and smoking having a cumulative attributable risk of 66.9% (95% CI = 60.7–73.0%) and 63.4% (95% CI = 53.5–73.2%), respectively. A trend of increasing PAR with increasing cumulative exposure to alcohol was also detected. When one risk factor combined with another was taken into account, the summary PAR for esophageal cancer increased to 69.7–82.6% and, when all 3 factors were considered, to 83.7%.
Table VII. Adjusted Odds Ratios And Population Attributable Risk Proportions of Esophageal Cancer Associated With Cumulative Lifetime Uses of Alcohol, Tobacco And Betel Quid, Taiwan, 1996–2003
|Alcohol drinking (g/day-year)|| || || || |
| Never||21.4||71.4||1.0|| |
| 1–300||14.6||14.1||2.8 (1.7–4.3)||9.3 (5.5–13.1)|
| 301–900||16.6||7.1||5.6 (3.5–9.0)||13.6 (10.1–17.1)|
| >900||47.4||7.5||13.9 (9.1–21.2)||44.0 (39.2–48.7)|
| Ever-drinker||78.6||28.6||7.0 (5.0–9.9)||66.9 (60.7–73.0)|
|Cigarette smoking (pack-years)|| || || || |
| Never||15.4||58.7||1.0|| |
| 1–25||25.9||19.2||3.7 (2.4–5.8)||18.9 (13.9–23.9)|
| 26–50||34.7||14.9||4.1 (2.6–6.4)||26.2 (20.6–31.9)|
| >50||24.0||7.2||4.2 (2.5–7.0)||18.2 (13.3–23.2)|
| Ever-smoker||84.6||41.3||4.1 (2.8–6.0)||63.4 (53.5–73.2)|
|Betel quid chewing (pack-years)|| || || || |
| Never||60.2||90.7||1.0|| |
| 1–10||6.4||3.2||1.5 (0.8–3.0)||2.0 (0–5.7)|
| 11–20||3.9||1.2||2.2 (0.9–5.3)||2.0 (0–4.3)|
| >20||29.4||4.9||2.8 (1.7–4.4)||18.7 (12.4–25.1)|
| Ever-chewer||39.8||9.3||2.3 (1.6–3.4)||23.0 (14.9–31.2)|
|PAR for 2 factors combined|| || || || |
| Drinking + smoking|| || || ||82.6 (78.2–87.0)|
| Drinking + chewing|| || || ||70.9 (65.6–76.3)|
| Smoking + chewing|| || || ||69.7 (61.9–77.6)|
|PAR for all factors combined|| || || || |
| Drinking + smoking + chewing|| || || ||83.7 (79.7–87.7)|
Our study shows clearly the independent risks of alcohol intake and tobacco smoking on the development of esophageal cancer, as previously shown in Western countries and some areas of Asia and Africa.1, 2 Several epidemiological studies have also found that duration of use, amount and types of alcohol and tobacco consumed to be strongly associated with elevated risk of esophageal cancer,13, 14, 15 as has our study. Other studies examining the role of alcohol in cancer of the esophagus in non-smokers and of tobacco in non-drinkers have provided further support for the independent effect of using these 2 substances on the occurrence of the disease.16, 17, 18
In conformity with findings from earlier research groups,19, 20, 21 our study shows that the effect of how long the substance was used and how much was used on the occurrence of esophageal cancer differed with regard to alcohol intake and tobacco smoking. After additionally taking how long and how much a substance was consumed in consideration, we found that the risk of contracting cancer of the esophagus strongly depended on the amount of average daily intake of alcohol (intensity) and the years of tobacco had been consumed (duration). The detailed dose- and time-dependent results indicated by prior studies13, 21 that a relatively low risk associated with low intake of ethanol and a short period of time of tobacco consumption; and a relatively high risk associated with high intake of ethanol and long periods of tobacco consumption were also observed in this study. Our findings show that patients who drank high dose of alcohol (>40 g/day) had a 19.8–27.3-fold elevated risk, even they drank alcohol over a period of <16 years. In contrast, patients who smoked tobacco over a long period (>30 years) had a 4.3–4.9-fold elevated risk, even they smoked tobacco <1–20 cigarettes/day (data not shown).
A number of previous studies have reported that 10 or more years of cessation of smoking led to a decline in risk of esophageal cancer.13, 15, 21, 22 For drinking cessation no clear favorable effect was found, in those studies former drinkers were at a lower,13, 21 similar or even higher risk than current drinkers.15, 22 In our present study, quitting alcohol for 10 or more years and tobacco for 5 or more years were separately associated with a 70% and 60% risk reduction. Although the risks for long-term quitter (>10 years) of alcohol and tobacco were both decreased toward the risks for never users, the cancer risks of such former uses still remained higher than those risks for never drinkers (OR = 2.3, p < 0.05) and for never smokers (OR = 2.4, p < 0.05). From a biological viewpoint, an apparent duration effect together with reduction of elevated risk after a relatively short time period of smoking cessation supports the hypothesis that tobacco contains both early-stage and late-stage carcinogens, as evaluated by IARC Monographs.1 In contrast, the risk reduction after a long term of drinking cessation, as well as the inconsistent findings provided by earlier studies,13, 15, 21, 22 suggest that alcohol may not merely involve in a late-stage carcinogenesis of esophageal cancer, as proposed by IARC Monographs.2
The practice of chewing of betel quid varies from country to country. In India and certain parts of Southeast Asia, where case-control studies show an association between betel quid use and esophageal cancer,3, 5, 15 tobacco is used frequently as an additional ingredient in betel quid products. In our study, a 2.9- and 2.2-fold elevated risk of contracting esophageal cancer was found to relate to areca nut chewing with a piece of Piper Betel Linn inflorescence and with betel leaf/stem and inflorescence mixed, respectively; both types of products do not use tobacco as an ingredient in Taiwan.4, 9 Comparable results were reported by studies conducted in 2 areas of India,15, 23 in those studies a 1.6–2.2-fold increase risk for esophageal cancer was observed among chewers without tobacco. These findings suggest that chewing products without tobacco was an independent risk factor for squamous cell carcinoma of the esophagus. Arecoline, a major component of betel nut, is its most abundant alkaloid. Evidence from in vitro studies have shown that arecoline can produce 2 N-nitrosamine carcinogens.24 The genotoxic and cytotoxic effects on various types of cells have also been detected.25 Ramchandani et al.26 further indicated that habitual use of pan masala, a common chewing product used with areca nut in India, may exert a carcinogenic and co-carcinogenic influence in the stomach and esophagus.
Epidemiological studies available in the literature for the assessment of joint effects on risk of esophageal cancer for alcohol drinking, tobacco smoking and betel quid chewing in 3-way combination are limited. Studies focused on this issue were conducted in 3 regions of India by methods of case-control.3, 15, 27 The highest elevated risks of esophageal cancer were all identified among persons who simultaneously practiced these 3 habits. One study further demonstrated that the combined risk for using alcohol, tobacco and betel quid was significant departure from risk-product multiplicatively.15 Another 2 studies also illustrated a synergistic effect for the combined habits of alcohol drinking with smoking and chewing.3, 27 In our study, although an especially strong risk (OR = 41.2) of esophageal cancer was noticed among patients who concomitantly consumed the 3 substances, no sufficient data can be used for the exploration of 3-way interaction because there were no cancer cases in the category of non-drinking and non-smoking chewers. Owing to areca nut chewing had a higher risk for contracting oral cancer28 and certain oral precancers, such as oral leukoplakia and submucous fibrosis9 than esophageal cancer. The competing causes of tumor occurrence from these diseases might explain the absence of this type of case in this population.
The interaction between alcohol intake and cigarette smoking has been studied in some detail in recent years. A number of studies have shown that alcohol and tobacco interact in a multiplicative way.1, 14, 15, 29 Nevertheless, our study, as well as a large-scale of study established in South America provided evidence that simultaneously used the 2 substances have a multiplicative interaction effect on risk of esophageal cancer.13 Castellsagué et al. further indicated that light alcohol drinking among non-smokers and light cigarette smoking among non-drinkers both had little or no effect on the cancer risk; however, combined exposure to the same moderate consumption amounts increased the odds ratio to a 12.1–18.9-fold statistically significant risk in either gender.13
The evidence for 2-way interaction of betel quid chewing with drinking and smoking was also examined in our study. Adjusted for the third habit and possible confounding variables, the joint risks for chewing and drinking as well as chewing and smoking were fitted well by a multiplicative model, a finding that is supported by earlier studies that have shown that a substantially higher cancer risk was marked among areca nut chewers who also drank alcohol3, 23, 27 or smoked tobacco.3, 23 Considering chewing with or without tobacco as a category, one multicenter case-control study in Southern India found a multiplicative interaction effect between chewing and drinking as well as chewing and smoking.15 Although areca nut is chewed without tobacco in Taiwan, our results demonstrate that betel quid chewing modified the risk of drinking in determining the development of esophageal cancer based on an additive interaction model. From the viewpoint of translating epidemiological findings into public health practice, such data indicates that prevention or elimination of betel quid chewing in those practicing, compared to those not practicing the habit of drinking could lead to a substantial reduction of the occurrence of esophageal cancer in this population.
Esophageal cancer is one of the most fatal forms of carcinoma. Because improvements in prognosis have been dissatisfying,30 primary prevention and intervention are important in the control of the disease. Our study shows that 66.9%, 63.4% and 23.0% of the etiological fraction of esophageal cancer were, respectively, attributable to alcohol intake, tobacco smoking and betel quid chewing, with 65.8%, 28.7% and 81.3% that explained separately by each highest exposure level. These data indicate that a substantial health benefit to esophageal cancer could be expected by efforts of reducing the prevalence of drinking and smoking, especially for the prevalence of more than 900 g/day-year alcohol consumption (PAR = 44.0%). Similar health benefit could be anticipated in population of North American,31 in which the estimated PAR for drinking (67.7%) and smoking (65.0%) in relation to squamous cell carcinoma of the esophagus were comparable to estimates in our study. Considering the intake of alcohol together with the use of tobacco, the combined population attributable risk proportion increased to 82.6%, which was similar to findings obtained from populations in South America and Northern Italy.13, 32 Combining with the additional consumption of betel quid, the summary population attributable fraction rose to 83.7%. Although the 3 agents were the important risk factors for contracting esophageal cancer at individual level, their impact at the population level should not be overlooked. As reported in our previous studies, smoking not only explains about 62.0% of lung cancer in men in Taiwan,33 but combining with betel quid chewing, the 2 factors account for 86.5% of oral precancer (oral leukoplakia or submucous fibrosis).9 Avoidance or cessation of use of these substances would greatly benefit the health of the residents in Taiwan.
Although approximately 36% of cancer cases in our study were excluded from the final data analyses, similar patterns for selected demographic factors, such as age and gender between recruited and non-recruited patients were detected in each geographic area. Education in the present study was found to strongly relate to risk of esophageal caner. The potential confounding effect has been well controlled, however, and the main findings for alcohol drinking, tobacco smoking and betel quid chewing remained when their effects were examined by stratifying education level. Under-representation of smokers and drinkers among control subjects due to high education level was improbable because the proportions of tobacco and alcohol users found in our study were comparable with those reported from 2 large-scale studies designed by the methods of cross-sectional and cohort study.8, 34 Although the proportion of chewers (10.1%) in men was somewhat lower than those estimated from the 2 studies (12.2–16.5%), the underestimate should not substantially alter the results for chewing of betel quid. Due to an in-personal interview administered by well-trained interviewers and the social acceptability of uses of the 3 substances in Taiwan, the extent to which the recall bias arose in our study should be limited.
In our present study, all the ever-chewers were tobacco smokers or alcohol drinkers, and chewing added just 1.1% to smoking and drinking in the computation of the attributable fraction of contracting esophageal cancer for the combination of the 3 substances, it is worthy to notice whether chewing could be just an indicator of smoking or drinking. In our study, a clear dose-response relationship was found between the duration and intensity of consumption of betel quid and esophageal cancer risk. Epidemiological findings from Znaor et al.15 showed that non-drinking and non-smoking chewers had a 3.3-fold elevated risk of esophageal cancer. Hoffmann et al.24in vitro studies found 2 N-nitrosamine carcinogens in arecoline, the genotoxic and cytotoxic effects of areca nut-related compounds have also been detected by Sundqvist et al.,25 as discussed previously. Although we have no data for examining the effect of this substance use among non-drinkers and non-smokers, the carcinogenic role of chewing should not be dismissed.
In summary, the effect of the amount of substance consumed and the number of years it was used on the etiology of esophageal cancer differs between alcohol and tobacco. Alcohol intake is the major contributor to the burden of this site of cancer in this population, and it also interacts with tobacco smoking in a supra-multiplicative way and with chewing of betel quid in a supra-additive way.