IJC International Journal of Cancer

Transcription

1 IJC International Journal of Cancer Consumption of salted meat and its interactions with alcohol drinking and tobacco smoking on esophageal squamous-cell carcinoma Sihao Lin 1, Xiaorong Wang 1, Chengyu Huang 2, Xudong Liu 1, Jin Zhao 1, Ignatius T.S. Yu 3 and David C. Christiani 4 1 JC School of Public Health and Primary Care, Division of Occupational and Environmental health, The Chinese University of Hong Kong, Hong Kong SAR, China 2 West China School of Public Health, Department of nutrient hygiene, Sichuan University, Chengdu, Sichuan, China 3 Hong Kong Occupational and Environmental Health Academy, Kowloon, Hong Kong, China 4 Department of Environmental Health, Harvard School of Public Health, Boston, MA Etiology of esophageal cancer has not yet been clearly documented, especially in high-risk regions. To evaluate the association between salted meat intake and esophageal squamous-cell carcinoma (ESCC) and to explore its joint effects with alcohol drinking and smoking, a population-based case control study was conducted in a high ESCC risk area in China, including 942 incident ESCC cases and 942 age- and sex-matching controls. A validated food frequency questionnaire was used to collect information on dietary factors, alcohol drinking and tobacco smoking. Conditional logistic regressions were applied to estimate the association between salted meat intake and ESCC and its interactions with alcohol drinking and smoking, with adjustment for other confounders, including total energy intake. Salted meat intake was associated with an increased risk of ESCC, showing an exposure response relationship (p for trend <0.001). Consumption of 50 g salted meat per week was related to an increased risk by 18% (odds ratio , 95% confidence interval: ). Salted meat in combination with either alcohol drinking or smoking had a greater risk than salted meat alone, which was more than additive. The strongest association was seen in the combination of all the three factors, particularly at the highest level of salted meat intake (odds ratio , 95% confidence interval: ). Salted meat intake is strongly associated with ESCC and its interactions with alcohol drinking and/or smoking highlights the significance of reducing salted meat intake among smokers and drinkers with respect to ESCC prevention. Esophageal cancer (EC) is one of the most common cancers in the world, with very high incidence and mortality rates in some developing countries. China is among the countries with the highest incidence rate, contributing to half of the total cases and deaths. 1 In China, >90% of EC occurs in the form of esophageal squamous-cell carcinoma (ESCC). Despite a steady decline in mortality from this cancer over the recent two decades, the incidence and mortality in China remained in the top of the world. Moreover, the 5-year survival rate of ESCC was low in China, ranging 10 30%, because of poor Key words: esophageal cancer, salt-processed meat, alcohol drinking, tobacco smoking, interaction, high-risk individuals Grant sponsor: World Cancer Research Fund International; Grant number: 2010/240 DOI: /ijc History: Received 22 Oct 2014; Accepted 10 Dec 2014; Online 27 Dec 2014 Correspondence to: Prof. Xiaorong Wang, JC School of Public Health and Primary Care, The Chinese University of Hong Kong, 4/F School of Public Health, Prince of Wales Hospital, Shatin NT, Hong Kong SAR, China, Tel.: [ ], Fax: 1[ ], xrwang@cuhk.edu.hk prognosis and lack of effective treatments. 2 Hence, implementation of effective prevention and control of the cancer are of vital importance. Alcohol drinking and tobacco smoking are two wellestablished risk factors for ESCC. 3 7 It is generally recognized that dietary factors play important roles in the development of this cancer However, whether alcohol drinking and smoking interact with diets in the development of ESCC is unclear. Recently, concerns over a high incidence of ESCC have been linked to consumption of processed meat. But, findings from previous studies on the associations between processed meat and ESCC were not consistent A varied definition of processed meat might partially explain the inconsistency of the results. 19 A recent literature review reported that processed meat was significantly associated with esophageal adenocarcinoma, but not with ESCC. 20 Although a number of epidemiologic studies have focused on the association between processed meat and EC, there is a limited number of studies that specified the association between salted meat and ESCC. 21,22 Yanting County, located in the Southwest China, is among the several high ESCC risk areas in the country. The incidence and mortality of EC were 92.81/10 5 and 77.50/10 5, respectively, according to data in Here, >80% of the Int. J. Cancer: 137, (2015) VC 2014 UICC

2 Lin et al. 583 What s new? Esophageal cancer has several known risk factors, including smoking, dietary factors, and alcohol consumption. In this paper, the authors asked whether eating lots of salted meat increases cancer risk, either alone or in combination with alcohol and smoking. They assessed the diet, tobacco, and alcohol use of nearly 2,000 cases and controls from a region of China known for high rates of esophageal cancer. Salted meat consumption, they found, accompanied an increase in esophageal cancer risk. And the addition of either alcohol or smoking produced an even larger boost in risk, more than would be expected from a simple additive effect. population was farmers in comparatively lower economic status. The local dietary structure was characterized by high consumption of salt-processed food and tuber crops and little consumption of milk, soy product and fresh fruit. 24 Salted meat, one type of processed meats, was commonly consumed in the area. A recent study conducted in this area reported that salted meat intake more than once a week significantly increased the risk of ESCC. 25 This study was conducted to elucidate the association between salted meat intake and ESCC and explore the interactions of salted meat intake with alcohol drinking and tobacco smoking. Material and Methods Study design and subjects A population-based case control study was conducted in the Yanting County. ESCC cases were recruited between June 2011 and May 2013 from a major local tumor hospital, which is the only esophageal carcinoma specialty hospital in the county. Case inclusion criteria were men or women aged years with newly diagnosed primary incident ESCC cases with a pathologic confirmation (ICD-10 code C15) and living in the area for >15 years. Among the 978 eligible cases, 942 (96%) cases were interviewed successfully. According to the local cancer registry data, the included cases accounted for nearly 70% of all incident ESCC cases in the county during the 2 years. Of those nonrecruited ESCC cases, 98.5% were diagnosed by histopathology. There was no difference in age and gender distribution between the recruited and nonrecruited cases in the county. For each of the cases, one population-based control was selected with the multistage sampling method from the local residents who had lived in the county for at least 15 years. All controls had no cancer at any site. In selecting controls, six townships were randomly selected first from 36 townships of Yanting County. Sampling frame for all residents aged years in the selected townships was available. Then, each control who matched with a case in terms of sex and age (within 2 years) was randomly selected, by using resident identity card number from the potentially eligible residents. If the first randomly selected resident did not agree to participate, the second randomly selected resident was approached. During the same period that cases were recruited, 886 (94%) people from the first round of selection and 56 (6%) from the second selection were recruited as individual matched controls. Finally, 942 pairs of cases and controls were included and analyzed in the study. Data collection Subjects were interviewed face to face by trained interviewers using local dialect. Both subjects and interviewers were unaware of the hypothesis under testing. The interview for the cases was completed at the hospital wards within 1 week after a case was diagnosed. Controls were interviewed at local health clinics. A structured questionnaire was used to collect the following information: (i) basic information, including age, sex, education, marital status, occupation, household income, and residence history; (ii) family history of cancer in first-degree relatives; (iii) tobacco smoking (age started and quit smoking, type of cigarette/tobacco and number of cigarette/tobacco per day); (iv) alcohol drinking habits (age started and quit drinking, type of alcoholic beverage and amount of each beverage consumed each time and drinking frequency of each beverage) and (v) dietary information and habits. The information on (iii), (iv) and (v) referred to 5 years before the cases were diagnosed and 5 years before the controls were interviewed. A food frequency questionnaire (FFQ) with 56 food items was developed and used to elicit dietary information. The food items selected in the FFQ were based on local diets and food availability, which covered more than 97.5% of typical foods in the region. Subjects were first asked whether they ever consumed a certain food. If the response was positive, they were further asked to indicate how often they, on average, consumed the food. Then, how much they consumed each time was asked. A standard bowl with scales was presented to help the subjects estimate portion size, which was defined as the sizes in which foods were served in one sitting. Reliability and validity of the FFQ were examined, which showed a moderate to strong agreement between 1-monthapart interviews and 24-hr dietary records (Liu XD, Wang XR, Lin SH, et al. Reproducibility and validity of a food frequency questionnaire for a diet-related esophageal cancer study in Chinese population. Preparation for submission). Average salted meat intake per week for each subject was estimated, expressed as the product of the frequency and amount of the consumption. Pack-years was used as a measure of cumulative tobacco smoking. Ever smokers were defined as smoking at least 10 cigarettes or equal amount of

3 584 Salted meat, alcohol drink, smoking and ESCC Table 1. Characteristics of cases and controls Cases (n 5 942) Controls (n 5 942) p-value 1 Age, mean (SD) 60.0 (6.83) 60.2 (6.79) Male, n (%) 672 (71.3) 672 (71.3) Total energy intake (kcal/day), mean (SD) 2,210.2 (768.4) 2,455.1 (814.0) <0.001 Marital status, n (%) Married 884 (93.8) 847 (89.9) Others 2 58 (6.2) 95 (10.1) BMI (kg/m 2 ), mean (SD) 22.1 (2.8) 22.8 (2.5) <0.001 Household income/year, n (%) < ,200 RMB 594 (63.1) 486 (51.6) >1,200 RMB 348 (36.9) 456 (48.4) Family cancer history, 3 n (%) 243 (25.8) 187 (19.9) Alcohol drinking, n (%) Abstainers 333 (35.4) 510 (54.1) <0.001 Ever drinkers 609 (64.6) 432 (45.9) Mean (SD) ethanol intake (g/week) (471.6) 98.5 (166.8) <0.001 Tobacco smoking, n (%) Never smokers 360 (38.2) 569 (60.4) <0.001 Ever smokers 582 (61.8) 373 (39.6) Pack-years, mean (SD) 41.5 (20.6) 36.8 (18.3) <0.001 Salted meat intake, n (%) <0.001 Never consumers 45 (4.8) 107 (11.4) Ever consumers 897 (95.2) 835 (88.6) Average (SD) intake (g/week) (129.8) 95.5 (54.5) <0.001 Average (SD) preserved vegetable intake (g/week) 90.1 (128.3) 52.4 (75.2) <0.001 Average (SD) pickled vegetable intake (g/week) (179.3) (156.4) Frequency of fresh vegetable intake, n (%) <0.001 <Once/day 261 (27.7) 162 (17.2) Once/day 681 (72.3) 780 (82.8) Frequency of fresh fruit intake, n (%) <0.001 <Once/week 756 (53.4) 659 (70.0) Once/week 186 (19.7) 283 (30.0) 1 Pearson s v 2 test for categorical variables and t-test for continuous variables. 2 Including divorced, unmarried, widows and widowers. 3 First-degree relatives with history of cancer at any site. tobacco per week for >6 months. 26 Weekly alcohol intake, expressed as pure ethanol consumption in grams, for each type of beverage, was estimated by multiplying intake frequency per week with drinking volume each time and weighted percentage of ethanol content for each type of beverage. Ever drinkers were defined as consuming any alcoholic beverage, including beer, wine or distilled spirits containing at least 20 g of ethanol, per week for minimum 6 months. 26 Body weight and height 5 years before diagnosis (for the cases) and interviewed (for the controls) were obtained from interview to calculate body mass index (BMI). Total energy intake expressed as kilocalories per day was also estimated from the FFQ according to the Chinese Food Composition tables. 27 The study was approved by the ethics committee of the Chinese University of Hong Kong. A written consent form was obtained from each of the subject. Data analysis We used v 2 or t-test to describe the differences in basic characteristics and relevant factors of interest between the cases and the controls. We applied conditional logistic regression to obtain odds ratios (ORs) for ESCC associated with salted meat, in terms of consumption frequency, portion size and mean intake per week. Never consumers were a reference

4 Lin et al. 585 Table 2. Odds ratios (ORs) for ESCC in relation to salted meat intake Case/control (n) Crude OR (95% CI) Adjusted OR 1 (95% CI) Never intake 45/ Ever intake 897/ ( ) 2.33 ( ) Intake frequency Never intake 45/ <1 time/week 280/ ( ) 1.77 ( ) 1 3 times/week 496/ ( ) 2.40 ( ) 4 times/week 121/ ( ) 7.06 ( ) p for trend <0.001 <0.001 Intake amount each time Never intake 45/ <1/4 bowl 2 70/ ( ) 1.51 ( ) 1/4 to 1/2 bowl 579/ ( ) 1.91 ( ) >1/2 bowl 248/ ( ) 7.28 ( ) p for trend <0.001 <0.001 Mean (SD) intake 3, g/week Never intake 45/ Q1 ( ) 225/ ( ) 1.77 ( ) Q2 ( ) 206/ ( ) 1.78 ( ) Q3 ( ) 186/ ( ) 2.37 ( ) Q4 ( ) 280/ ( ) 5.52 ( ) p for trend <0.001 <0.001 Mean intake, 50 g per week ( ) 1.18 ( ) 1 Adjusted for sex, age, marital status, household income, BMI, family cancer history, smoking, alcohol drinking, intake of pickled and preserved vegetables, fresh vegetable, fruit intake and total energy intake. 2 A bowl commonly used in the local area. 3 Mean intake per week among all consumers was categorized into four levels based on quartiles: Q1 (<22.7 g), Q2 (> g), Q3 (> g) and Q4 (>151.4 g). 4 Salted meat intake was included in the model as a continuous variable. group in the models. Average intake per week in all ever consumers was categorized into four subgroups based on quartiles. We first evaluated the association between salted meat and ESCC with univariate analysis and then with multivariate models to adjust for potential confounders. We included covariates in the models by using a backward selection procedure: if OR was changed by 10% after elimination from the model or if the distribution between the cases and controls was statistically different (p < 0.05). 28 Meanwhile, we took into account other processed food, including preserved vegetables and pickled vegetables, in the model after ruling out their collinearity with salted meat. We also estimated the association of consuming every 50 g of salted meat per week with ESCC by treating salted meat consumption as a continuous variable in the model. In determining the interactions, we first estimated additive-scale interactions of salted meat consumption with either drinking or smoking on ESCC using Synergy Index (SI) as follows: SI 5 (OR )/[(OR )1(OR )], where OR 41 compared those with the highest salted meat consumption (Q4) in ever smokers or drinkers with the reference. 28 SI >1 indicates potential synergism, whereas SI < 1 suggests potential antagonism. Then, we estimated the interactions by dichotomizing salted meat (low/high intake) and smoking (ever/never) and alcohol drinking (ever/never). The low/high salted meat intake was defined at the midpoint of average intake per week. Trend tests were performed by fitting models with average intake of each category as a continuous variable. We further conducted an analysis on the interactions with smoking and drinking at four levels of salted meat intake (quartiles Q1 Q4). Finally, we assessed the multiplicative interaction using likelihood ratio test among the three factors by including the main terms of salted meat, smoking and drinking and the product term of the three factors in the multivariate model. Statistical analyses were conducted with SPSS version 21.0 for windows. All tests were two-sided, and p-values <0.05 were considered statistically significant. Results Both cases and controls had a mean age of 60 years, and 71% were men. BMI was statistically lower 5 years before

5 586 Salted meat, alcohol drink, smoking and ESCC Table 3. Odds ratios 1 (95% CI) for ESCC in relation to four categories 2 of salted meat intake by drinking and smoking status Alcohol drinking 3 Tobacco smoking 4 Salted meat intake (mean, SD), g/week Abstainers Ever drinkers Never smokers Ever smokers Never intake ( ) ( ) Q1 ( ) 2.42 ( ) 3.93 ( ) 1.91 ( ) 6.24 ( ) Q2 ( ) 2.31 ( ) 4.09 ( ) 2.23 ( ) 5.03 ( ) Q3 ( ) 2.66 ( ) 5.94 ( ) 3.19 ( ) 5.70 ( ) Q4 ( ) 5.82 ( ) ( ) 5.49 ( ) ( ) p for trend <0.001 <0.001 <0.001 Synergy Index ( ) 2.37 ( ) 1 Adjusted for sex, age, marital status, household income, BMI, family history of cancer, intake of pickled and preserved vegetables, fresh vegetable, fresh fruit intake and total energy intake. 2 Mean intake per week among all consumers was categorized into four levels based on quartiles: Q1 (<22.7 g), Q2 (> g), Q3 (> g) and Q4 (>151.4 g). 3 Additional adjustment for smoking (pack-years). 4 Additional adjustment for alcohol drinking (average ethanol intake per week). 5 Trend tests were performed by fitting models with median values of each intake category as a continuous variable. 6 The Synergy Index was for interactions of salted meat intake (Q4) and alcohol drinking/smoking on ESCC. Table 4. Interactions among salted meat intake, alcohol drinking and tobacco smoking on ESCC risk Salted meat intake Alcohol drinking Cigarette smoking Case/control (n) Crude ORs (95% CI) Adjusted ORs (95% CI) 1 Low / Low / ( ) 1.28 ( ) Low / ( ) 2.26 ( ) Low / ( ) 5.58 ( ) High / ( ) 1.98 ( ) High / ( ) 4.62 ( ) High / ( ) 2.74 ( ) High / ( ) ( ) 1 Adjusted for sex, age, marital status, household income, BMI, family history of cancer, intake of pickled and preserved vegetables, fresh vegetables intake, fresh fruit intake and total energy intake. 2 Defined as average intake not more than 90.8 g per week of salted meat (including never consumers). 3 A combination of the low level of salted meat intake with nonsmoking and nondrinking was reference. 4 Defined as average intake >90.8 g per week of salted meat. (2), never; (1), ever. diagnosis in the cases than in the controls. More cases had a family cancer history compared with the controls. Smoking and alcohol drinking were more common in the cases, who consumed more tobacco and ethanol compared with the controls. Meanwhile, the cases consumed 176 g of salted meat every week, on average, contrast to 96 g in the controls. The cases also consumed more preserved and pickled vegetables, and less frequently consumed fresh vegetables and fruits (Table 1). Salted meat intake was significantly associated with an increased risk of ESCC, in terms of intake frequency, portion size and mean intake per week (Table 2). Compared with never consumers, ever consumers had a 2.3-fold increased risk. There was a clear exposure response relationship in individual mean intake, with adjusted ORs from 1.8 at the lowest to 5.5 at the highest level. Furthermore, the consumption of 50 g salted meat per week was associated with an increased risk by 18% (OR , 95% confidence interval [CI]: ). In comparison with those who never consumed salted meat and nonsmoking or nondrinking, all other subgroups had significantly higher risks (Table 3). The risk tended to be higher with increasing consumption of salted meat (p for trend <0.01), regardless of smoking or drinking status. Those at the highest level of salted meat intake (Q4) with either smoking or drinking had a greater risk, compared with others at lower salted meat intake levels. SI for Q4 of salted meat intake with alcohol drinking was 1.91 (95% CI: ) and with smoking was 2.37 ( ), implying a significant interaction of the high consumption of salted meat with alcohol drinking or smoking. The effects of different combinations of the three factors were further examined (Table 4), where salted meat consumption was classified as either high or low level. In the absence

6 Lin et al. 587 Figure 1. Interactions (odds ratio) with the combination of alcohol drinking, tobacco smoking and salted meat intake by four categories (Q1 Q4) on ESCC. Interactions were adjusted for sex, age, marital status, household income, BMI, family history of cancer, intake of pickled and preserved vegetables, intake of fresh vegetables, fresh fruits and total energy intake. Mean intake of salted meat per week among all consumers was categorized into four levels based on quartiles: Q1 (<22.7 g), Q2 (>22.7 to 90.8 g), Q3 (>90.8 to g) and Q4 (>151.4 g). Never intook salted meat, nonsmoking and nondrinking was reference. Overall, no matter with or without drinking and smoking, salted meat intake showed increasing risks from Q1 to Q4. The combinations of the three factors showed greatest risks on ESCC, particularly at the highest level of salted meat consumption. of smoking and drinking, high consumption of salted meat was associated with a nearly twofold risk (OR , 95% CI: ). The risk increased in presence of alcohol drinking (OR , 95% CI: ) or smoking (OR , 95% CI: ). The strongest association was seen in the combination of high salted meat intake with both smoking and drinking (OR , 95% CI: ). A further analysis of the interactions was performed, where quartiles of salted meat intake was used (Fig. 1). Never consumption of salted meat, nondrinking and nonsmoking was a reference group. Overall, no matter with or without drinking and smoking, salted meat intake showed increasing risks from Q1 to Q4. The risk was greater when salted meat intake coexisted with either drinking or smoking, also tending to increase from Q1 to Q4. However, the further greater risks were observed when all of the three factors were present. The greatest risk (OR , 95% CI: ) was showed in the combination of smoking and alcohol drinking at Q4 of salted meat intake, though the product term of the three factors in the model did not reach significance (p ). Discussion In this study, we found that consumption of salted meat, one of the common dietary practices in the study area, was positively associated with the risk of ESCC, showing an exposure response relationship. Moreover, we observed the interactions of salted meat intake with alcohol drinking and tobacco smoking. Salted meat intake, no matter measured in the frequency, portion size or average intake per week, was associated with an increased risk of ESCC. The observed association was not confounded by other salt-processed foods, such as preserved vegetables and pickled vegetables, as these foods were adjusted in the multivariate models. Processed meat usually consists of meat items preserved by nitrite/salt treatment, smoking or fermentation, including all types of sausages, bacon, ham and salted meat. As summarized in a recent meta-analysis, the association between meat and EC varied with meat type and histologic type of EC. 29 The definition and components of processed meat might be different among studies, which yielded inconsistent results on processed meat and EC. 19 In this study, salted meat was almost the only type of processed meat consumed in the study area. Our findings were consistent with studies conducted in the Uruguay population, which focused specifically on salted meat consumption. 21,22 Salted meat preparation and consumption were a long tradition in the Yanting area. Preserving meat and food with salt was a main way because there were no refrigerators available in such rural areas until the recent decade. Meat was usually preserved with much salt and then dried in the wind and under the sun. Such procedures may lead to formation of N-nitroso compounds from the reaction of nitrosamine and nitrite, which were proven to be animal carcinogens and possible human carcinogens. 30,31 Moreover, salted meat may also contain other carcinogens, such as heterocyclic amines, polycyclic aromatic hydrocarbons, particularly being cooked or processed at high temperature. 19 In addition to the carcinogens, a high concentration of salt in the processed meat may increase the risk of EC, although salt itself is not a carcinogen. The mechanism may be that salt can directly damage esophageal mucosa, causing susceptibility to esophagitis and increasing the risk of EC. 32 Another major finding from this study is that salted meat intake might have an interaction with alcohol drinking and smoking. Alcohol drinking and smoking are two established risk factors for ESCC, and the two factors interact with each other on the development of EC. 3,7,33 However, few studies have examined the interactions of smoking and alcohol drinking with other factors. 26,34 Wu et al. reported a significant interaction on ESCC among smoking, alcohol drinking and betel nuts chewing in Taiwan. 34 An adjusted OR tremendously increased to (95% CI: ) when the three factors were present simultaneously. 34 Lee et al. also reported a synergic effect among alcohol drinking, smoking and betel quid on the risk of EC in Taiwan. 26 An earlier study from Switzerland reported that the association between diets and the risk of EC was stronger in heavy alcohol drinkers, indicating an interaction between poor diet and alcohol drinking on EC. 35 In this study, we detected the interactions of salted meat intake with smoking and

7 588 Salted meat, alcohol drink, smoking and ESCC alcohol drinking. When other potential confounders were adjusted, the interactions of salted meat with drinking or with smoking were significantly more than additive effects. The result implied that smoking and/or alcohol drinking possibly modified the association of salted meat intake with the risk of ESCC, particularly when salted meat was consumed at high level. In the study area, it is a common practice that local residents usually use salted meat as a dish going with alcohol drinking, and some of them even smoke simultaneously. Given the high prevalence of smoking and alcohol drinking in China, 36 it definitely would be an important public health problem in the regions where salted meat is commonly consumed. There might be plausible mechanisms for the interactive effects. When alcohol is consumed with salted meat, the former may function as a solvent, enhancing penetration of those carcinogenic molecules in salted meat into mucosal cells. 19 Meanwhile, tobacco may induce specific mutations in DNA that are less efficiently repaired in the presence of alcohol. 7 Salted meat usually contains much saturated fat. A significant association between high saturated fat intake and ESCC was reported from a study in a high-risk area of Iran. 37 Bioassays and epidemiologic studies indicated that tobacco smoking might modify the effect of dietary fat on the risk of cancers In this case, tobacco smoking might enhance the carcinogenic potency of salted meat and exerted a synergic effect on the risk of ESCC when these two risk factors were present together. Moreover, N-nitroso compounds contained in salted meat and the nitrosation of nicotine during tobacco processing plus the tobacco-specific nitrosamines resulted from cigarette smoking 41 spontaneously increased the total consumption of N-nitroso compounds, which increased the risk of ESCC when they were present concurrently. To the best of our knowledge, this is the first study to investigate the interactions of salted meat with smoking and alcohol drinking on ESCC. There are several major strengths of this study. First, on the basis of population-based design, our study subjects could be relatively representative of the study area. We recruited about 70% of all incident ESCC cases in the study area during 2 years, and the controls were randomly selected from the same population as the cases, which were reasonably representative samples of ESCC cases and local residents in the study area. Second, by recruiting incident cases and conducting the interview within a week after a case was diagnosed, we were able to reduce recall errors and avoid possible reverse causality because of a change in subjects diets after being diagnosed with the disease. Third, the sample size was comparatively big, which allowed us to examine the interactions by combining three factors in different ways. Finally, information on many other factors, such as family history, lifestyles and other preserved food and dietary factors, was available, which minimized potential confounding effects of these factors on the association of interest. On the other hand, the interpretation of the results should be cautious because of some limitations of the study. The first concern is information bias, which is inherent in any case control study. Collecting accurate dietary data from years ago is a challenge. We introduced the study to subjects as a general health survey rather than an EC study. We also presented a commonly used bowl with scales and food pictures during the interviews to help the subjects recall and estimate consumed amount of a certain food. Despite that, the misclassification of exposures was possible. However, the potential exposure misclassification was likely to be nondifferential because: first, we used identical methods and manner to collect and quantify exposure data in the cases and the controls. In the data collection, the study hypotheses and objectives were blinded to field interviewers and study subjects. Second, the association between salted meat and ESCC was consistently seen in terms of the frequency, portion size and average intake amount. Third, we detected the exposure response relationship between salted meat intake and ESCC, indicating that the exposure estimates reflected the real exposures to a large extent. Thus, the nondifferential misclassification might have caused an underestimation of the associations concerned. Then, selection bias, particularly a control group, might be another concern for a case control study. The cases were recruited from a major tumor hospital in the area, where nearly 70% of local EC patients sought service. Moreover, community-based controls were selected from same communities of the cases. According to a previous diet survey in this area, the daily consumption of salted meat was about 16 g per person, 24 which was comparable with the amount found in the current control group (14 g/day), implying that the selected controls were reasonably representative. In addition, although the overall sample size was comparatively big, the number of subjects in some subgroup analysis was not adequate. For example, there was a small number in the subgroups of either only smoking or only drinking, which made the risk estimates for these subgroups unstable. In conclusion, the study provides additional evidence for a strong association between salted meat intake and ESCC. There was a possible interaction of salted meat intake with alcohol drinking and/or tobacco smoking. It is important to educate people to quit smoking and alcohol drinking. Meanwhile, reducing salted meat consumption may add additional benefits in control and prevention of ESCC, particularly among alcohol drinkers and smokers. Our findings suggest targeting individuals with these three habits for screening and preventing ESCC. Additional studies with prospective design are needed to confirm the interactions observed in this study and to elucidate possible mechanisms underlying the interaction. Acknowledgements The authors thank Dr. Li Jun, Dr. Wang Dong and the team in Yanting Tumor Hospital for their cooperation in data collection.

8 Lin et al. 589 References 1. Ferlay J, Shin HR, Bray F, et al. Estimates of worldwide burden of cancer in 2008: GLOBO- CAN Int J Cancer 2010; 127: Guo P, Huang ZL, Yu P, et al. Trends in cancer mortality in China: an update. Ann Oncol 2012; 23: Wu M, Zhao JK, Zhang ZF, et al. Smoking and alcohol drinking increased the risk of esophageal cancer among Chinese men but not women in a high-risk population. Cancer Causes Control 2011; 22: Brown LM, Blot WJ, Schuman SH, et al. Environmental factors and high risk of esophageal cancer among men in coastal South Carolina. J Natl Cancer Inst 1988; 80: Steevens J, Schouten LJ, Goldbohm RA, et al. Alcohol consumption, cigarette smoking and risk of subtypes of oesophageal and gastric cancer: a prospective cohort study. Gut 2010; 59: Ishiguro S, Sasazuki S, Inoue M, et al. Effect of alcohol consumption, cigarette smoking and flushing response on esophageal cancer risk: a population-based cohort study (JPHC study). Cancer Letters 2009; 275: Castellsague X, Mu~noz N, De Stefani E, et al. 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: Fan Y, Yuan JM, Wang R, et al. Alcohol, tobacco, and diet in relation to esophageal cancer: the Shanghai cohort study. Nutr Cancer 2008; 60: Lucenteforte E, Garavello W, Bosetti C, et al. Diet diversity and the risk of squamous cell esophageal cancer. Int J Cancer 2008; 123: Hung HC, Huang MC, Lee JM, et al. Association between diet and esophageal cancer in Taiwan. J Gastroenterol Hepatol 2004; 19: Gao YT, McLaughlin JK, Gridley G, et al. Risk factors for esophageal cancer in Shanghai, China. II. Role of diet and nutrition. Int J Cancer 1994; 58: Ziegler RG, Morris LE, Blot WJ, et al. Esophageal cancer among black men in Washington, D.C. II. Role of nutrition. J Natl Cancer Inst 1981; 67: Keszei AP, Schouten LJ, Goldbohm RA, et al. Red and processed meat consumption and the risk of esophageal and gastric cancer subtypes in the Netherlands cohort study. Ann Oncol 2012; 23: De Stefani E, Boffetta P, Ronco AL, et al. Processed meat consumption and risk of cancer: a multisite case-control study in Uruguay. Br J Cancer 2012; 107: Levi F, Pasche C, Lucchini F, et al. Processed meat and the risk of selected digestive tract and laryngeal neoplasms in Switzerland. Ann Oncol 2004; 15: Cross AJ, Freedman ND, Ren J, et al. Meat consumption and risk of esophageal and gastric cancer in a large prospective study. Am J Gastroenterol 2011; 106: Li K, Yu P. Food groups and risk of esophageal cancer in Chaoshan region of China: a high-risk area of esophageal cancer. Cancer Invest 2003; 21: Choi Y, Song S, Song Y, et al. Consumption of red and processed meat and esophageal cancer risk: meta-analysis. World J Gastroenterol 2013; 19: World Cancer Research Fund/American Institute for Cancer Research. Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Washington, DC: AICR, Salehi M, Moradi-Lakeh M, Salehi MH, et al. Meat, fish, and esophageal cancer risk: a systematic review and dose-response meta-analysis. Nutr Rev 2013; 71: De Stefani E, Aune D, Boffetta P, et al. Salted meat consumption and the risk of cancer: a multisite case-control study in Uruguay. Asian Pac J Cancer Prev 2009; 10: De Stefani E, Deneo-Pellegrini H, Boffetta P, et al. Meat intake and risk of squamous cell esophageal cancer: a case-control study in Uruguay. Int J Cancer 1999; 82: Wang X, Fan JC, Wang AR, et al. of esophageal cancer in Yanting regional report of a national screening programme in China. Asian Pac J Cancer Prev 2013; 14: Xiao P, Tao DM, Huang CY, et al. A survey of diet structure in a high risk area of esophageal cancer (in Chinese). Modern Prev Med 2006; 33: Song Q, Wang X, Yu IT, et al. Processed food consumption and risk of esophageal squamous cell carcinoma: a case-control study in a high risk area. Cancer Sci 2012; 103: Lee CH, Lee JM, Wu DC, et al. Independent andô combined effects of alcohol intake, tobacco smoking and betel quid chewing on the risk of esophageal cancer in Taiwan. Int J Cancer 2005; 113: [PMC][ /ijc.20619] [ ] 27. Yang YX, Wang GY, Pan XC. China Food Composition table, 2nd edn. Beijing: Peking University Medical Press, Rothman KJ, Greenland S. Modern epidemiology. Philadelphia, PA: Lippincott Williams & Wilkins, Zhu HC, Yang X, Xu LP, et al. Meat consumption is associated with esophageal cancer risk in a meat- and cancer-histological-type dependent manner. Dig Dis Sci 2014; 59: Scanlan RA. Formation and occurrence of nitrosamines in food. Cancer Res 1983; 43:2435S 40S. 31. Tricker AR, Preussmann R. Carcinogenic N- nitrosamines in the diet: occurrence, formation, mechanisms and carcinogenic potential. Mutat Res 1991; 259: Onuk MD, Oztopuz A, Memik F. Risk factors for esophageal cancer in Eastern Anatolia. Hepatogastroenterology 2002; 49: Sakata K, Hoshiyama Y, Morioka S, et al. Smoking, alcohol drinking and esophageal cancer: findings from the JACC study. J Epidemiol 2005; 15(Suppl 2):S Wu IC, Lu CY, Kuo FC, et al. Interaction between cigarette, alcohol and betel nut use on esophageal cancer risk in Taiwan. Eur J Clin Invest 2006; 36: Levi F, Pasche C, Lucchini F, et al. Food groups and esophageal cancer risk in Vaud, Switzerland. Eur J Cancer Prev 2000; 9: Li Q, Hsia J, Yang G. Prevalence of smokingô inô China in N Engl J Med 2011; 364: Jessri M, Rashidkhani B, Hajizadeh B, et al. Macronutrients, vitamins and minerals intake and risk of esophageal squamous cell carcinoma: a case-control study in Iran. Nutr J 2011; 10: Hoffmann D, Rivenson A, Abbi R, et al. A study of tobacco carcinogenesis: effect of the fat content of the diet on the carcinogenic activity of 4- (methylnitrosamino)-1-(3-pyridyl)-1-butanone in F344 rats. Cancer Res 1993; 53: Hebert JR, Kabat GC. Distribution of smoking and its association with lung cancer: implications for studies on the association of fat with cancer. J Natl Cancer Inst 1991; 83: Taioli E, Nicolosi A, Wynder EL. Possible role of diet as a host factor in the etiology of tobaccoinduced lung-cancer an ecological study in Southern and Northern Italy. Int J Epidemiol 1991; 20: Hoffmann D, Adams JD, Piade JJ, et al. Chemicalô studies on tobacco smoke. LXVIII. Analysis of volatile and tobacco-specific nitrosamines in tobacco products. IARC scientific publications, vol. 31. Lyon, France: IARC,