Diet diversity and the risk of squamous cell esophageal cancer

Authors


Abstract

A varied diet may have a favorable role against digestive tract cancers. We analyzed the relationship between diet diversity (i.e. measured by the number of different foods consumed at least once per week) and the risk of esophageal cancer. We considered data from a case–control study conducted between 1992 and 1997 in northern Italy on 304 squamous cell esophageal cancer cases below age 78 years and 743 controls admitted to hospital for acute, nonneoplastic conditions, unrelated to tobacco or alcohol consumption. There was a significant inverse association for total diet diversity: the multivariate odds ratio (OR), adjusted for age, sex, area of residence, education, tobacco smoking, alcohol drinking and non-alcohol energy intake was 0.42 (95% confidence interval, CI: 0.25–0.71) for subjects in the highest versus those in the lowest quartile of diversity. Inverse relations were also found for diversity within vegetables (OR = 0.34, 95% CI: 0.21–0.55) and fruits (OR = 0.51, 95% CI: 0.33–0.80). No significant association was found for meat and cereal diversity. These results add epidemiological support to the dietary guidelines recommending a more varied diet, particularly in fruit and vegetables, for esophageal cancer prevention. © 2008 Wiley-Liss, Inc.

In Europe and North America, esophageal squamous cell carcinoma is strongly related to tobacco and alcohol consumption, but various aspects of diet have a role on risk.1 In particular, a high intake of fruit and vegetables has been linked to a lower risk of esophageal cancer.1–3 Most epidemiological studies, however, have addressed this association by considering individual food items, and limited attention has been paid to the analysis of their joint effect by considering dietary patterns.

Food diversity (i.e. the variety of foods consumed) has been recommended to achieve a healthy diet in cancer prevention.4, 5 Diet diversity has been inversely related to all-cause mortality in the First National Health Epidemiologic Follow-up Study.5 Inverse associations were also found with upper respiratory and digestive tract cancers, including oral and pharyngeal,6, 7 laryngeal cancer,8 gastric,9 and colorectal.10–13 No study, to our knowledge, has investigated the relation between diet diversity and esophageal cancer risk.

To assess the role of total and specific food group diversity on the risk of esophageal cancer, we analyzed data from a case–control study conducted in northern Italy.

Material and methods

A case–control study of cancer of esophagus was conducted between 1992 and 1997 in the provinces of Milan, Pordenone and Padova in northern Italy.2 Cases were 304 patients (275 men and 29 women; median age: 60 years, range: 39–77 years) admitted to major teaching and general hospitals in the study areas, with incident, histologically confirmed squamous cell cancer of the esophagus, diagnosed no longer than 1 year before the interview.

Controls were 743 subjects (593 men and 150 women, median age: 60 years, range: 36–77) frequency-matched with cases by 5-year age group, sex, year of interview and area of residence. To compensate for the rarity of esophageal cancer in women a control-to-case ratio of about 5 was chosen for women, as compared to about 2 for men. Controls were selected among patients admitted to the same hospitals as cases for a wide spectrum of acute, non-neoplastic conditions, not related to smoking, alcohol consumption and long-term modification of diet. Twenty-nine percent of the controls were admitted for traumas, 36% for other nontraumatic orthopedic conditions, 12% for acute surgical conditions and 23% for miscellaneous other illnesses, including eye, nose, ear, skin or dental disorders. Less than 5% of both cases and controls who were approached for interview refused to participate.

Trained interviewers interviewed cases and controls during their hospital stay using a structured questionnaire. This included information on sociodemographic characteristics, lifestyle habits such as tobacco smoking and alcohol drinking, anthropometric measures, personal medical history and family history of cancer. The subjects' usual diet during the 2 years prior to cancer diagnosis or hospital admission (for controls) was investigated through an interviewer-administered food-frequency questionnaire (FFQ), including 78 foods and beverages, as well as a range of recipes, including the most common ones in the Italian diet, grouped into 6 sections: (i) bread and cereal dishes (first courses); (ii) meat and other main dishes (second courses); (iii) vegetables (side dishes); (iv) fruit; (v) sweets, desserts and soft drinks; (vi) milk, hot beverages and sweeteners. The FFQ showed satisfactory validity14 and reproducibility.15 Subjects were asked to indicate the average weekly frequency of consumption of each dietary item. To estimate total energy intake, an Italian food composition database, integrated with other sources when needed, was used.16, 17 A validated section for alcoholic beverages was also included.18

We computed total diversity as the number of different foods consumed at least once per week.9, 13 We also computed diversity within 4 food groups: vegetables (13 items), fruit (12 items), meat (11 items) and cereals (13 items). We categorized subjects into approximate quartiles of total or specific food group diversity based on the distribution of controls.

Odds ratios (OR) of squamous cell esophageal cancer and the corresponding 95% confidence intervals (CI) according to various measures of food diversity were derived from multiple unconditional logistic regression models.19 All models included terms for age, sex, area of residence, education, tobacco smoking, alcohol drinking and non-alcohol energy intake. When variables were entered in the models as continuous terms the unit was set to 1 standard deviation of the controls' distribution. To test for interactions, the differences in −2 × log likelihood of the models with and without an interaction term were compared to the χ2 distribution with 1 degree of freedom.

Results

Table I shows the distribution of 304 cases of squamous cell cancer of the esophagus and 743 controls according to selected variables. By design, the proportion of women was higher in controls than in cases, and the age distribution was similar in cases and controls. Cases reported significantly higher tobacco and alcohol consumption. No association was found with non-alcohol energy intake.

Table I. Distribution of 304 Cases of Squamous Cell Cancer of the Esophagus and 743 Controls According to Sex, Age and other Selected Variables (Italy, 1992–1997)
CharacteristicCasesControls
No. (%)No. (%)
  • 1

    The sum does not add up to the total because of one missing value.

Sex
 Male275 (90.5)593 (79.8)
 Female29 (9.5)150 (20.2)
Age (years)
 <5029 (9.5)78 (10.5)
 50–5447 (15.5)114 (15.3)
 55–5965 (21.4)152 (20.5)
 60–6462 (20.4)146 (19.7)
 65–6957 (18.7)142 (19.1)
 ≥7044 (14.5)111 (14.9)
Smoking status1
 Never smokers33 (10.9)245 (33.0)
 Ex-smokers109 (36.0)287 (38.6)
 Current smokers
  <15 cigarettes/day37 (12.2)86 (11.6)
  15–24 cigarettes/day84 (27.7)92 (12.4)
  ≥25 cigarettes/day40 (13.2)33 (4.4)
Alcohol intake (drinks/week)
 0–2034 (11.2)339 (45.6)
 21–3451 (16.8)182 (24.5)
 35–5562 (20.4)117 (15.7)
 56–8385 (27.9)68 (9.2)
 ≥8472 (23.7)37 (5.0)
Non-alcohol energy intake (kcal/day)
 <1743.8169 (22.7)149 (20.1)
 1743.81–2076.5356 (18.4)148 (19.9)
 2076.54–2448.7561 (20.1)149 (20.1)
 2448.76–2927.2863 (20.7)149 (20.1)
 ≥2927.2955 (18.1)148 (19.9)

Table II gives the ORs of esophageal cancer according to intake quartiles of total, vegetable, fruit, meat and cereal diversity. Significant inverse trends were observed with total diversity (OR for the highest versus the lowest quartile of diversity 0.42, 95% CI: 0.25–0.71), as well as vegetable (OR = 0.34, 95% CI: 0.21–0.55) and fruit diversity (OR = 0.51, 95% CI: 0.33–0.80). No significant association was found for meat and cereal diversity. The correlation coefficients between diet diversity and corresponding food group absolute amount were 0.45 for total diversity, 0.63 for vegetable diversity and 0.63 for fruit diversity. However, further allowance for vegetable and fruit intake in the model did not modify the risk estimates, and the OR for the highest level of total diversity were 0.57 adjusted for vegetables, and 0.47 adjusted for fruit. Corresponding values were 0.48 and 0.35 for vegetable diversity, and 0.60 and 0.53 for fruit diversity.

Table II. Odds Ratios (OR) and Corresponding 95% Confidence Intervals (CI) Among 304 Cases of Squamous Cell Cancer of the Esophagus and 743 Controls According to Intake Quartiles of Diversity of Selected Food Groups (Italy, 1992–1997)
Diet diversity1CasesControlsOR2 (95% CI)
No.(%)No.(%)
  • 1

    Number of items consumed at least weekly.

  • 2

    Estimates from multiple logistic regression adjusted for age, sex, area of residence, education, tobacco smoking, alcohol drinking and non-alcohol energy intake.

  • 3

    Reference category.

Total diversity
 <23138(45.4)184(24.8)13 
 23 to <2864(21.0)172(23.1)0.42(0.27–0.66)
 28 to <3348(15.8)175(23.6)0.39(0.24–0.63)
 ≥3354(17.8)212(28.5)0.42(0.25–0.71)
 Chi-square trend (p value)    12.2(0.0005)
Vegetable diversity
 <4136(44.7)174(23.4)13 
 4 to <537(12.2)111(14.9)0.49(0.30–0.82)
 5 to <779(26.0)233(31.4)0.48(0.32–0.73)
 ≥752(17.1)225(30.3)0.34(0.21–0.55)
 Chi-square trend (p value)    20.4(<0.0001)
Fruit diversity
 <3111(36.5)133(17.9)13 
 3 to <463(20.7)146(19.7)0.65(0.41–1.03)
 4 to <542(13.8)159(21.4)0.35(0.21–0.58)
 ≥588(29.0)305(41.0)0.51(0.33–0.80)
 Chi-square trend (p value)    11.3(0.0008)
Meat diversity
 <388(29.0)166(22.4)13 
 3 to <452(17.1)142(19.1)0.66(0.41–1.07)
 4 to <549(16.1)134(18.0)0.71(0.42–1.18)
 ≥5115(37.8)301(40.5)0.72(0.47–1.12)
 Chi-square trend (p value)    1.34(0.25)
Cereal diversity
 <443(14.1)99(13.3)13 
 4 to <565(21.4)135(18.2)1.08(0.62–1.87)
 5 to <7124(40.8)267(35.9)1.21(0.73–2.01)
 ≥772(23.7)242(32.6)0.93(0.53–1.62)
 Chi-square trend (p value)    0.05(0.83)

Total, vegetable and fruit diversity were further examined in separate strata of age (<60 and ≥60 years), alcohol drinking (<40 and ≥40 drinks per week) and tobacco smoking (never smoker and smoker) (Table III). The inverse association with total, vegetable and fruit diversity was stronger in never smokers and heavy alcohol drinkers. No substantial heterogeneity was observed across strata of age.

Table III. Odds Ratios (OR) and Corresponding 95% Confidence Intervals (CI) Among 304 Cases of Squamous Cell Cancer of Esophagus and 743 Controls According to Total, Vegetable and Fruit Diversity in Strata of Age, Alcohol Drinking and Tobacco Smoking (Italy, 1992–1997)
Diet diversity1OR2 (95% CI)
OverallAge (years)Alcohol drinking (drinks/week)Tobacco smoking
<60≥60<40≥40Never smoker3Smoker4
  • 1

    Number of items consumed at least weekly.

  • 2

    Estimates from multiple logistic regression adjusted for age, sex, area of residence, education, tobacco smoking, alcohol drinking and non-alcohol energy intake. Continuous OR for an increase of one standard deviation of the control distribution.

  • 3

    Including ex-smokers since 10 or more years.

  • 4

    Including ex-smokers since less than 10 years.

  • 5

    p for interaction <0.0001.

  • 6

    p for interaction <0.05.

Total diversity0.70 (0.57–0.86)0.70 (0.54–0.91)0.69 (0.54–0.89)0.76 (0.61–0.94)0.64 (0.51–0.80)0.575 (0.45–0.72)0.795 (0.64–0.97)
Vegetable diversity0.74 (0.62–0.88)0.69 (0.54–0.88)0.78 (0.62–0.98)0.79 (0.64–0.97)0.68 (0.54–0.85)0.586 (0.46–0.74)0.836 (0.68–1.00)
Fruit diversity0.76 (0.64–0.90)0.68 (0.53–0.87)0.83 (0.66–1.05)0.906 (0.73–1.11)0.636 (0.51–0.79)0.606 (0.47–0.75)0.876 (0.72–1.06)

Discussion

The results of our study indicate that total, vegetable and fruit diversity are inversely related to squamous cell cancer of the esophagus.

The collection of extensive dietary information using a reproducible and valid FFQ14, 15 is the major strengths of our study. Further strengths include the comparable catchment areas of cases and controls, the high participation rate and the possibility of allowance for energy intake and several other covariates in the analyses. Still, as in most case–control studies, selection and recall biases are possible, as a recent cancer diagnosis may influence recall of past diet in cases. However, we investigated dietary habits in the 2 years before diagnosis. Moreover, the hypothesis of dietary variety in esophageal cancer etiology was unknown to the interviewers and to the Italian public opinion. Dietary habits of hospital controls may differ from those of the general population. However, we excluded from the control group any subject admitted for conditions associated to long-term modifications of diet. Further, the reproducibility of any diet information of hospital and out-of-hospital subjects was satisfactory,15, 20 and interviewing subjects in the same hospital setting has improved the comparability of information between cases and controls.

The findings of our study are in agreement with those of other investigations on cancers of the aerodigestive tract,6–9, 11, 13 suggesting a beneficial effect of a more varied diet, including, in particular, a large variety of fruit and vegetables. Meat and cereal diversities were not associated to esophageal cancer risk, probably reflecting the fact that food items in these categories have not been consistently implicated in esophageal cancer etiology.2, 3

The inverse relation between total, vegetable and fruit diversity was stronger in never smokers and heavy alcohol drinkers. Heavy drinkers, in particular, tend to have a poorer diet.21 However, the absence of consistent findings across the 2 major recognized risk factors for squamous cell esophageal cancer suggests that this apparent difference can be due, largely or totally, to chance.

Total diversity may simply be an indicator of more favorable dietary or lifestyle correlates of esophageal cancer. When we divided our control subjects into 2 strata of diversity, 28% of those below the median of diversity were heavy drinkers (≥40 drinks/week) vs. 23% of those above the median, and 42% vs. 36% were smokers (including ex-smokers since less than 10 years), respectively. However, major identified potential confounding factors, including not also alcohol and tobacco consumption, but also indicators of socioeconomic status, were allowed in the analysis. Further, the association with vegetable and fruit diversity was slightly attenuated after allowance for total energy as well as vegetable and fruit intake, and was independent and of the same magnitude of those of vegetable and fruit quantity.2

Vegetables and fruit are rich in several micronutrients and other food compounds such as carotenoids, vitamins C and E, fibers, flavonoids and other plant sterols, which display both complementary and overlapping mechanisms of action, including antioxidant effects, binding and dilution of carcinogens in the digestive tract, and alteration of hormone metabolism.22, 23 It is conceivable that these overlapping mechanisms are favored by a varied diet including several different types of fruit and vegetables. This can explain the beneficial effect of a diet varied—not only rich—in vegetables and fruit on esophageal cancer, although it remains possible that a frequent consumption of vegetables and fruit is a non-specific indicator of a more affluent and better-planned diet.

In conclusion, the present findings indicate that variety in the diet (particularly in terms of vegetables and fruit) has a beneficial effect on esophageal carcinogenesis, and further support the dietary guidelines suggesting the adoption of a more varied diet for cancer prevention.24

Acknowledgements

This work was undertaken while C.L.V. was a senior fellow at the International Agency for Research on Cancer. The authors thank Ms. I. Garimoldi for editorial assistance.

Ancillary