Hostname: page-component-7c8c6479df-995ml Total loading time: 0 Render date: 2024-03-29T01:08:13.346Z Has data issue: false hasContentIssue false

Dietary patterns and risk of oesophageal cancers: a population-based case–control study

Published online by Cambridge University Press:  07 September 2011

Torukiri I. Ibiebele*
Affiliation:
Gynaecological Cancers Group, Queensland Institute of Medical Research, 300 Herston Road, Herston, Brisbane, QLD 4006, Australia
Maria Celia Hughes
Affiliation:
Cancer and Population Studies Group, Queensland Institute of Medical Research, 300 Herston Road, Herston, Brisbane, QLD 4006, Australia
David C. Whiteman
Affiliation:
Cancer Control Group, Queensland Institute of Medical Research, 300 Herston Road, Herston, Brisbane, QLD 4006, Australia
Penelope M. Webb
Affiliation:
Gynaecological Cancers Group, Queensland Institute of Medical Research, 300 Herston Road, Herston, Brisbane, QLD 4006, Australia
*
*Corresponding author: Dr T. I. Ibiebele, fax +61 7 3845 3503, email Torukiri.Ibiebele@qimr.edu.au
Rights & Permissions [Opens in a new window]

Abstract

Epidemiological studies investigating the association between dietary intake and oesophageal cancer have mostly focused on nutrients and food groups instead of dietary patterns. We conducted a population-based case–control study, which included 365 oesophageal adenocarcinoma (OAC), 426 oesophagogastric junction adenocarcinoma (OGJAC) and 303 oesophageal squamous cell carcinoma (OSCC) cases, with frequency matched on age, sex and geographical location to 1580 controls. Data on demographic, lifestyle and dietary factors were collected using self-administered questionnaires. We used principal component analysis to derive three dietary patterns: ‘meat and fat’, ‘pasta and pizza’ and ‘fruit and vegetable’, and unconditional logistic regression models to estimate risks of OAC, OGJAC and OSCC associated with quartiles (Q) of dietary pattern scores. A high score on the meat-and-fat pattern was associated with increased risk of all three cancers: multivariable-adjusted OR 2·12 (95 % CI 1·30, 3·46) for OAC; 1·88 (95 % CI 1·21, 2·94) for OGJAC; 2·84 (95 % CI 1·67, 4·83) for OSCC (P-trend < 0·01 for all three cancers). A high score on the pasta-and-pizza pattern was inversely associated with OSCC risk (OR 0·58, 95 % CI 0·36, 0·96, P for trend = 0·009); and a high score on the fruit-and-vegetable pattern was associated with a borderline significant decreased risk of OGJAC (OR for Q4 v. Q1 0·66, 95 % CI 0·42, 1·04, P = 0·07) and significantly decreased risk of OSCC (OR 0·41, 95 % CI 0·24, 0·70, P for trend = 0·002). High-fat dairy foods appeared to play a dominant role in the association between the meat-and-fat pattern and risk of OAC and OGJAC. Further investigation in prospective studies is needed to confirm these findings.

Type
Full Papers
Copyright
Copyright © The Authors 2011

Oesophageal cancer is the eighth most frequently diagnosed cancer and the sixth leading cause of cancer death worldwide, with an estimated 482 000 new cases and 407 000 deaths in 2008(1, Reference Jemal, Center and DeSantis2). There are two histological types: adenocarcinoma which can occur in the oesophagus (OAC) or the oesophagogastric junction (OGJAC) and oesophageal squamous cell carcinoma (OSCC). The incidence of OAC has been increasing in Western countries(Reference Gallus and La Vecchia3) including Australia(Reference Lord, Law and Ward4, Reference Thomas, Lade and Giles5), with recent figures showing positive annual percentage changes of 4·2 and 4·3 % in the incidence of OAC for men and women, respectively, in the two decades before 2005 in New South Wales, Australia(Reference Stavrou, McElroy and Baker6). These rapid changes in incidence prompted the present study.

Diet is a modifiable risk factor that may influence risk of cancers of the oesophagus(Reference Ziegler7Reference Donaldson9). The majority of epidemiological studies investigating the association between diet and cancer of the oesophagus have focused on individual nutrients(Reference Tzonou, Lipworth and Garidou10Reference Park, Leitzmann and Subar15), individual foods or food groups(Reference Tzonou, Lipworth and Garidou10, Reference Park, Leitzmann and Subar15Reference Navarro Silvera, Mayne and Risch23); however, because of inconsistency among the results, the evidence for a role of diet in the aetiology of oesophageal cancer is not conclusive. Dietary pattern analysis has emerged in recent years as a way to shed more light on the role of diet in modifying disease risk. Unlike the individual nutrient or food group approach, the dietary pattern approach allows the study of foods as they are actually consumed, thus capturing the inter-correlations between foods and nutrients(Reference Schwerin, Stanton and Smith24).

Few epidemiological studies have specifically investigated the association between dietary patterns and risk of OSCC(Reference Bahmanyar and Ye25Reference Hajizadeh, Rashidkhani and Rad28), and only two have investigated OAC(Reference Bahmanyar and Ye25, Reference Chen, Ward and Graubard29). These previous studies have been relatively small in size (185 and 124 OAC cases), and results have been inconsistent. For example, while two studies(Reference Hajizadeh, Rashidkhani and Rad28, Reference De Stefani, Boffetta and Fagundes30) found an increased risk between a ‘Western’ or ‘meat and fat’ dietary pattern and the risk of OSCC, another reported a non-significantly decreased risk(Reference Bahmanyar and Ye25). To further investigate the role of dietary patterns in OAC risk, we have used the information reported on a validated FFQ from a population-based case–control study to identify common food consumption patterns in Australia, and to relate these patterns to the risk of OAC, OGJAC and OSCC.

Materials and methods

Study design and participants

We used data from a nationwide case–control study of oesophageal cancer conducted in Australia, the details of which have been described elsewhere(Reference Whiteman, Sadeghi and Pandeya31). In summary, eligible case patients were people aged 18–79 years with a histologically confirmed primary invasive cancer of the oesophagus or oesophagogastric junction diagnosed between 1 July 2002 (1 July 2001, in Queensland) and 30 June 2005, in the mainland states of Australia. Patients were recruited through the major treatment centres and state-based cancer registries. A total of 1577 patients with oesophageal cancer were invited to participate in the study, of whom 1102 patients (858 through clinics and 244 through cancer registries) returned a completed questionnaire (70 % of those invited; 35 % of all patients diagnosed with incident oesophageal cancer during the study period). Details of the histological type and anatomical site of each tumour were abstracted from diagnostic pathology reports by medically qualified investigators. Tumour site was classified according to the WHO classification such that adenocarcinomas that straddled the junction of the oesophagus and stomach were called tumours of the oesophagogastric junction regardless of where the bulk of the tumour lay (OGJAC, ICD-O code 8140/3), while those located entirely above the oesophagogastric junction were considered oesophageal carcinomas(Reference Spechler, Dixon, Genta, Hamilton and Aaltonen32). Eight case patients were deemed ineligible on pathology review and excluded from the analysis, leaving 365 OAC, 426 OGJAC and 303 OSCC patients.

Potential controls were selected at random from the Australian Electoral Roll and frequency matched to the case series by age (5-year age groups), sex and state of residence. We aimed for similar numbers of male cases and controls in each stratum of age and state; female controls were intentionally oversampled at all ages to accommodate their simultaneous enrolment in a parallel study of ovarian cancer(Reference Olsen, Bain and Jordan33). Of the 3258 potentially eligible control participants who were contacted and invited to participate, 216 were excluded (sixteen deceased, sixty-one were too ill, ninety-eight were unable to read or write in English and forty-one were unwilling to participate in the study). Of the 3042 remaining controls, 1580 returned the completed questionnaires (51 % of all potentially eligible controls contacted).

All study participants provided informed written consent. The study was approved by the human research ethics committees of the Queensland Institute of Medical Research and all participating institutions.

Exclusions and final sample size

Of the 1094 cases and 1580 controls who returned the main risk factor questionnaire, 199 participants (152 cases and 47 controls) had no opportunity to complete an FFQ because the nutrition component of the study commenced 6 months after the main study. A further thirty-five cases and five controls omitted responses to 10 % or more of FFQ items, while twenty-seven cases and twenty-one controls had implausible total energy intakes ( < 3360 or >21 000 kJ for men and < 2940 or >16 800 kJ for women). The present analysis included 1507 (98 %) controls and 880 (93 %) cases: 299 with OAC, 336 with OGJAC and 245 with OSCC.

Non-dietary data

Data were collected via a self-administered questionnaire. Information was collected on age, education, smoking history, total lifetime alcohol consumption, use of aspirin or other non-steroidal anti-inflammatory drugs during the past 5 years, height and weight 1 year ago (1 year before diagnosis for cases), frequency of symptoms of gastro-oesophageal reflux, defined as the presence of heartburn (‘a burning pain behind the breastbone after eating’) or acid reflux (‘a sour taste from acid or bile rising up into the mouth or throat’) 10 years before diagnosis, and physical activity.

Dietary data and food grouping

Dietary data were obtained using a 139-item semi-quantitative FFQ, modified from the instrument developed by Willett et al. (Reference Willett, Sampson and Stampfer34), and shown to be valid against weighed food records(Reference Ashton, Marks and Battistutta35Reference Marks, Hughes and van der Pols37) and serum biomarkers(Reference McNaughton, Marks and Gaffney38), and reproducible(Reference Ibiebele, Parekh and Mallitt39) for use in Australia. Respondents recalled how often, on average, they consumed a standard serving size of a specific food item in the previous year (for controls) or in the year before their diagnosis (for cases) for 135 food items. Information on four additional items including the quantity of sugar habitually added to food or beverages and the discretionary use of fat as assessed by the frequency with which visible fat from meat, foods fried at home and fried take-away foods were consumed were obtained. We calculated daily food intake in g by multiplying the frequency of consumption per d by the standard serving size of each food as specified in the FFQ. The foods items were grouped into forty-four predefined food groups based on the similarity of nutrient profiles or culinary usage.

Dietary pattern derivation

Principal component analysis is a data reduction technique that reduces the number of observed variables by creating a small number of factors that account for much of the variance in the food groups(Reference Hatcher40, Reference Kubo, Levin and Block41). We used the PROC FACTOR command in SAS (statistical software version 9.1; SAS Institute, Inc., Cary, NC, USA) to derive dietary patterns using the correlation matrix of the forty-four food groups among the combined group of cases and controls (Table S1, supplementary material for this article can be found at http://www.journals.cambridge.org/bjn). We considered eigenvalues>1, screeplots and interpretability when determining the number of factors to be retained. The preliminary analysis yielded thirteen factors with eigenvalues>1, which together accounted for 53 % of the variance in the forty-four food groups. However, based on the point at which the screeplot of eigenvalues levelled off, we retained three factors that explained 25 % of the total variance in the diet. To facilitate interpretability, the three factors were rotated using varimax rotation to obtain three dietary patterns that were orthogonal and uncorrelated with each other. Factor loadings were calculated for each food group and dietary pattern (Table 1). A positive loading for a food group indicates a direct association with the dietary pattern while a negative loading indicates that the food group contributes inversely to the dietary pattern. For each dietary pattern, a score was calculated for cases and controls by summing up the intake of each food group (in g) weighted by the factor loading of the food groups (Σ((food groupi g/d) × (food groupi factor loading)), where i = food group from 1 to 44)(Reference Nettleton, Diez-Roux and Jenny42). Factor scores for each pattern were then categorised into quartiles using the distribution of the population controls, for further analyses.

Table 1 Factor loadings* for the relationship between food groups and factors representing dietary patterns in oesophageal cancer cases and controls

* With orthogonal rotation, the factor loading scores are identical to the correlation coefficients; factor loadings with absolute values < 0·20 are not shown for clarity.

Fruits other than those containing high levels of vitamin A or C.

Vegetables other than red/yellow, leafy green or cruciferous vegetables.

§ An Australian spread made from yeast extract.

Statistical analysis

We used the χ2 test to check for differences in proportions, and ANOVA to check for differences in the distribution of continuous variables across categories of potential risk factors. Unconditional logistic regression was used to calculate OR and 95 % CI as estimates of the relative risks of OAC, OGJAC and OSCC associated with quartiles of dietary pattern scores. We assessed linear trends by ranking factor scores from 1 to 4 (lowest to highest quartile) and modelling this as a continuous variable. We simultaneously adjusted for the potential confounding effects of factors shown to be associated with> oesophageal cancer in our previous studies(Reference Whiteman, Sadeghi and Pandeya31, Reference Pandeya, Williams and Green43, Reference Sadeghi, Bain and Pandeya44). These include age (years); sex (male, female); BMI 1 year previously ( < 25, 25·0–29·9, ≥ 30·0 kg/m2); education (high school only, technical college or diploma, university); frequency of heartburn or acid reflux symptoms in the 10 years before diagnosis (never, < monthly, < weekly, >weekly, daily); pack-years of smoking (0, 1–14·9, 15–29·9, ≥ 30); average lifetime alcohol intake (never, < 1–6, 7–20, ≥ 21 drinks/week); non-steroidal anti-inflammatory drugs use during the past 5 years (never, occasionally, less than weekly, at least weekly) and total energy intake (kJ; log-transformed). Further adjustment for state of residence and physical activity did not alter the effect estimates for the dietary patterns; hence these variables were not included in the final models.

Results

The three retained factors were labelled ‘meat and fat’, ‘pasta and pizza’ and ‘fruit and vegetable’ patterns, on the basis of the food groups with the highest factor loadings (Table 1). The meat-and-fat pattern was characterised by high positive loadings for processed meat, high-fat potato, discretionary fat, red meat, high-fat dairy, poultry with skin on, white bread, sweet snacks and fat spreads; and very low factor loadings for fruits, vegetables and fish. The pasta-and-pizza pattern featured high positive loadings for pasta, tomato-based pasta sauce, wine, pizza, olives or pickled vegetables, white rice and cereal products; and very low intakes of processed meat, high-fat potato, discretionary fat, red meat, and high-fat dairy. The fruit-and-vegetable pattern was characterised by high positive loadings for all types of fruits and vegetables, low-fat potatoes, breakfast cereals and wholemeal bread; and low factor loadings for processed meat, high-fat potato, discretionary fat, red meat, high-fat dairy and alcoholic beverages. Together, the three dietary patterns explained 25·4 % of the total variance in dietary intake (11·7 % for the meat-and-fat, 8·2 % for the pasta-and-pizza and 5·5 % for the fruit-and-vegetable patterns). Among control participants, intake of foods with positive factor loadings increased, and that of foods with negative factor loadings decreased monotonically with increasing dietary pattern score. Mean intake of foods with low factor loadings did not differ significantly across quartiles of dietary pattern scores (Table S2, supplementary material for this article can be found at http://www.journals.cambridge.org/bjn).

In general, cases were more likely to be older ( ≥ 50 years) and to be heavy smokers ( ≥ 30 pack-years) than controls. OAC and OGJAC cases were more likely to be male, obese and to experience symptoms of reflux (>1/month) than OSCC cases and controls, while OSCC cases were more likely to be heavy alcohol consumers ( ≥ 21 standard drinks/week) than OAC and OGJAC cases and controls (Table 2). In comparison with controls in the first quartile of the meat-and-fat dietary pattern, those in the fourth quartile were more likely to be male, more likely to be overweight or obese, more likely to smoke heavily ( ≥ 30 pack-years) or drink heavily ( ≥ 21 standard drinks/week). In contrast, compared with controls in the first quartile of the pasta-and-pizza pattern, those in the fourth quartile were younger, more likely to be female, more likely to have a university education, but less likely to smoke heavily. Compared with controls in the first quartile of the fruit-and-vegetable pattern, those in the fourth quartile were older, more likely to be females and less likely to smoke or drink heavily (Table 3).

Table 2 Characteristics of cases and controls

(Numbers and percentages)

OAC, oesophageal adenocarcinoma; OGJAC, oesophagogastric junction adenocarcinoma; OSCC, oesophageal squamous cell carcinoma.

* From χ2 test (sex) or χ2 test for trend.

Table 3 Characteristics of 1507 control participants according to quartiles (Q) of dietary pattern score

(Numbers, percentages, mean values and standard deviations)

* From χ2 test for trend (BMI, heartburn, smoking, alcohol, non-steroidal anti-inflammatory drugs use) or heterogeneity (sex, education) or ANOVA (age, energy intake).

One standard drink = 10 g ethanol.

A high score on the meat-and-fat dietary pattern was associated with increased risk of all three oesophageal cancers (OR 2·12, 95 % CI 1·30, 3·46, P for trend = 0·002 for OAC; OR 1·88, 95 % CI 1·21, 2·94, P for trend = 0·002 for OGJAC; OR 2·84, 95 % CI 1·67, 4·83, P for trend < 0·0001 for OSCC; Table 4). A high score on the pasta-and-pizza pattern was associated with a decreased risk of OSCC only (OR 0·58, 95 % CI 0·36, 0·96, P-trend = 0·009), while high scores on the fruit-and-vegetable pattern were associated with a borderline significant decreased risk of OGJAC (OR 0·66, 95 % CI 0·41, 1·04) and a significantly decreased risk of OSCC (OR 0·41, 95 % CI 0·24, 0·70, P for trend = 0·002). We found no association between the pasta-and-pizza or the fruit-and-vegetable dietary patterns and risk of OAC (Table 4).

Table 4 Associations between dietary patterns and risk of oesophageal cancers

(Numbers, percentages, odds ratios and 95 % confidence intervals)

OAC, oesophageal adenocarcinoma; OGJAC, oesophagogastric junction adenocarcinoma; OSCC, oesophageal squamous cell carcinoma; Q, quartiles.

* OR adjusted for age (years); sex; education (school only, technical college or diploma, university); BMI 1 year ago ( < 25, 25·0–29·9, ≥ 30·0 kg/m2); frequency of heartburn or acid reflux 10 years before diagnosis/recruitment (never, < monthly, < weekly, >weekly, daily); pack-years of smoking (0, 1–14·9, 15–29·9, ≥ 30); non-steroidal anti-inflammatory drugs use (never, occasionally, less than weekly, at least weekly); and total energy (kJ, log transformed).

Additionally adjusted for lifetime alcohol intake (never, < 1–6, 7–20, ≥ 21 drinks/week). The ‘meat and fat’ and ‘pasta and pizza’ patterns were not adjusted for alcohol as alcoholic drinks were high loading components of these patterns; however, additional adjustment for alcohol intake did not alter the results.

We then considered whether the observed associations with the dietary patterns could be explained by individual food groups that contributed strongly to that pattern. For each pattern, we selected food groups with a loading ≥ 0·45 (or ≤ − 0·45) and added these one at a time to the model with the dietary pattern variable. Intake of high-fat dairy foods, a major contributor to the meat-and-fat pattern, was significantly associated with increased risk of both OAC and OGJAC. When we included both intake of high-fat dairy foods and the meat-and-fat dietary pattern score in the same model, the association between the meat-and-fat pattern and OAC was attenuated (the OR for quartile (Q) 4 v. Q1 fell from 2·12 to 1·69, 95 % CI 1·00, 2·86) and weaker than that for high-fat dairy foods (OR 2·46, 95 % CI 1·54, 3·94). A similar effect was seen for OGJAC where, in the joint model, there was a strong association with high-fat dairy foods (OR for Q4 v. Q1 = 1·83, 95 % CI 1·17, 2·86) and the association with the meat-and-fat pattern was weakened and non-significant (the OR for Q4 v. Q1 fell from 1·88 to 1·54, 95 % CI 0·95, 2·48). None of the other high-loading food groups appreciably altered the association with the meat-and-fat pattern when included in the model. This suggests that the observed associations with the meat-and-fat pattern were due largely to the intake of high-fat dairy foods and perhaps not a true effect of the overall dietary pattern.

In contrast, while several food groups from each of the dietary patterns were significantly associated with OSCC risk, none appeared to account for a major portion of the associations between the dietary patterns and OSCC suggesting that these were true pattern effects.

When dietary patterns were derived separately for males and females, we observed factor loadings similar to those reported for the total population. We also found no evidence that the observed associations with dietary patterns were modified by BMI, reflux or smoking (data not shown).

Discussion

We identified three dietary patterns in this population-based case–control study, and together they explained 25 % of the variance in dietary intake as measured by an FFQ. The ‘meat and fat’ and ‘fruit and vegetable’ patterns were similar to those found previously in studies of ovarian cancer(Reference Kolahdooz, Ibiebele and van der Pols45), and skin cancer(Reference Ibiebele, van der Pols and Hughes46) using a similar FFQ in an Australian population. Although it had a high positive factor loading for wine and a moderate positive factor loading for spirits, the ‘pasta and pizza’ pattern we observed differed from the ‘alcohol’ or ‘drinker’ dietary patterns described in other studies. Unlike the ‘alcohol’ and ‘drinker’ patterns, this pattern was not characterised by high positive factor loadings on beer, eggs, processed meat, red meat or discretionary fat; instead, it featured moderate loadings on oily fish, nuts, green leafy vegetables, non-oily fish and seafood.

Overall, our results showed that participants with high scores on the meat-and-fat dietary pattern had an almost 2-fold increased risks of OAC and OGJAC, and an almost 3-fold increased risk of OSCC. Participants with high scores on the pasta-and-pizza pattern had a 42 % decreased risk of OSCC, while those with high scores on the fruit-and-vegetable pattern had a borderline significant 34 % decreased risk of OGJAC and a significant 59 % decreased risk of OSCC.

Our finding of a positive association between the meat-and-fat pattern and OAC and OGJAC risks is consistent with the findings of a population-based case–control study from Sweden, where the authors reported a positive association between a high score on a ‘Western’ dietary pattern, similar to our ‘meat and fat’ dietary pattern, and risks of gastric cardia adenocarcinoma and OAC(Reference Bahmanyar and Ye25). Our finding that this association was driven in part by consumption of high-fat dairy foods is consistent with reports from two population-based case–control studies in which positive associations were found between a high-fat dairy food group(Reference Navarro Silvera, Mayne and Risch23), or a high-milk dietary pattern(Reference Chen, Ward and Graubard29) and risk of OAC. An inverse association between total intake of dairy products and risk of OAC has been reported however(Reference Chen, Ward and Graubard29) and in our data, the low-fat dairy food group had a negative factor loading ( − 0·33) in relation to the meat-and-fat pattern, suggesting that participants whose diet followed the meat-and-fat dietary pattern were less likely to consume foods from the low-fat dairy food group. Taken together, it is therefore possible that the fat component of the high-fat dairy food group may be most important in contributing to increased OAC risk.

Our finding of a positive association between the meat-and-fat pattern and OSCC is consistent with findings from other case–control studies from Sweden(Reference Bahmanyar and Ye25), Uruguay(Reference De Stefani, Deneo-Pellegrini and Boffetta27) and Iran(Reference Hajizadeh, Rashidkhani and Rad28). One mechanism postulated to explain this association is that processed meat contains high levels of nitrites and N-nitroso compounds which might exert a carcinogenic effect(Reference Mayne, Risch and Dubrow11).

The inverse association we observed between the pasta-and-pizza dietary pattern and risk of OSCC warrants further exploration, given that this factor had a strong positive loading with wine, and the known strong links between alcohol intake and risk of OSCC. It is, however, consistent with our previous study that found a decreased OSCC risk among those with low to moderate intake of alcohol from wine ( ≤ 90 g/week)(Reference Pandeya, Williams and Green43). In the present analyses, the mean intake of alcohol from wine across all quartiles of the pasta-and-pizza pattern (8·8, 23·2, 45·5, 76·1 g/week) was below this level, indicating that this dietary pattern was characterised by modest wine consumption. In addition to wine, the pasta-and-pizza pattern featured moderate intakes of other foods such as green leafy vegetables, fruit and vegetable juices, nuts, legumes, oily fish and whole grains that contain micronutrients with known antioxidant and anticarcinogenic properties including flavonoids, folates, phytosterols, vitamins A, C, E and dietary fibre. Thus, several aspects of the pasta-and-pizza dietary pattern are similar to the ‘Mediterranean diet’ which has been shown to be associated with decreased risk of cancers of the upper aerodigestive tract(Reference Bosetti, Gallus and Trichopoulou47).

Our finding of an inverse association between the fruit-and-vegetable dietary pattern and risk of OSCC is consistent with the findings of case–control studies from Uruguay(Reference De Stefani, Boffetta and Ronco26, Reference De Stefani, Deneo-Pellegrini and Boffetta27) and Iran(Reference Hajizadeh, Rashidkhani and Rad28). A Swedish study also found an inverse but non-significant association with a ‘healthy’ dietary pattern with components similar to our fruit-and-vegetable pattern(Reference Bahmanyar and Ye25). The World Cancer Research Fund report states that there is evidence to support that fruits probably protect against oesophageal cancer(1). A population-based case–control study(Reference Navarro Silvera, Mayne and Risch23), and a prospective study(Reference Freedman, Park and Subar21) reported inverse association between non-citrus fruits(Reference Navarro Silvera, Mayne and Risch23) or fruits in the Rosaceae botanical group(Reference Freedman, Park and Subar21)and decreased OSCC risk. The component of non-citrus fruits and fruits in the Rosaceae family from both studies are similar to ‘other’ fruits (fruits other than those containing high levels of vitamin C or A) in our study.

The strengths of our present study include its relatively large sample size, the population-based design and our ability to control for a large number of potentially important confounders. Also, ours is one of the few studies to investigate the association between dietary patterns and the three types of oesophageal cancer in the same study. A limitation is the low participation rate among controls and cases. Detailed information on the characteristics of non-participants are not available because of Australian privacy laws; nevertheless, comparison of our study control data to data from the Australian National Health Survey, a representative survey of the Australian adult population conducted in 2004 with a reported 90 % response rate(48), showed similar distributions of key characteristics including educational level, BMI and smoking (ever/never)(Reference Pandeya, Williams and Green49). The strong influence of current diet on recall of past diet raises concerns regarding the possibility of recall bias among cases if their diet changed as a result of their diagnosis, or because they experienced symptoms as a result of the presence of subclinical disease before diagnosis. To minimise this, study participants were asked to report recent changes to their diet in the last year or two. Exclusion of individuals who reported having changed their diet did not alter our findings. Another limitation is the use of FFQ, an instrument prone to measurement error in collecting dietary data; however, the FFQ from which the dietary patterns were derived has shown reasonably good validity when compared with weighed food records(Reference Marks, Hughes and van der Pols36, Reference Marks, Hughes and van der Pols37) and serum biomarkers(Reference McNaughton, Marks and Gaffney38); and any non-differential measurement error is likely to have attenuated the associations that we have observed, thus strengthening the results of our present study. Finally, it is possible that our results could be biased if the association between dietary patterns and oesophageal cancer differs by tumour grade, a strong predictor of survival, and hence study participation. However, we found no evidence that the reported associations differed by tumour grade.

In conclusion, we have shown that a ‘meat and fat’ dietary pattern was associated with increased risks of OAC, OGJAC and OSCC, a ‘pasta and pizza’ pattern was inversely associated with OSCC risk, while a ‘fruit and vegetable’ pattern was associated with a decreased risk of OSCC and possibly OGJAC but not OAC. However, high-fat dairy foods appear to play a major role in the association between the meat-and-fat dietary pattern and oesophageal cancer. Further analyses using data from prospective studies would be valuable to confirm the role of dietary patterns in oesophageal cancer risk.

Acknowledgements

The Australia Cancer Study was supported by the National Health and Medical Research Council (NHMRC) of Australia (Program No.199600). P. M. W. is supported by a Research Fellowship from the NHMRC of Australia. D. C. W. is supported by a Future Fellowship from the Australian Research Council. The funding bodies played no role in the design or conduct of the study; the collection, management, analysis or interpretation of the data; or preparation, review or approval of the manuscript. T. I. I., M. C. H., D. C. W. and P. M. W. have no personal or financial conflicts of interest. D. C. W. and P. M. W. designed the study; T. I. I. and M. C. H. established the dietary databases, T. I. I. conducted the data analysis; T. I. I., D. C. W. and P. M. W. wrote the paper; and all authors read and approved the final manuscript.

The Australian Cancer Study

Investigators: David C. Whiteman MBBS, PhD, Penelope M. Webb MA, D Phil, Adele C. Green MBBS, PhD, Nicholas K. Hayward PhD, Peter G. Parsons PhD, David M. Purdie PhD.

Clinical collaborators: B. Mark Smithers FRACS, David Gotley FRACS PhD, Andrew Clouston FRACP PhD, Ian Brown FRACP.

Project Manager: Suzanne Moore RN, MPH.

Database: Karen Harrap BIT, Troy Sadkowski BIT.

Research Nurses: Suzanne O'Brien RN MPH, Ellen Minehan RN, Deborah Roffe RN, Sue O'Keefe RN, Suzanne Lipshut RN, Gabby Connor RN, Hayley Berry RN, Frances Walker RN, Teresa Barnes RN, Janine Thomas RN, Linda Terry RN MPH, Michael Connard BSc, Leanne Bowes BSc, MaryRose Malt RN, Jo White RN.

Clinical contributors:

Australian Capital Territory: Charles Mosse FRACS, Noel Tait FRACS.

New South Wales: Chris Bambach FRACS, Andrew Biankan FRACS, Roy Brancatisano FRACS, Max Coleman FRACS, Michael Cox FRACS, Stephen Deane FRACS, Gregory L. Falk FRACS, James Gallagher FRACS, Mike Hollands FRACS, Tom Hugh FRACS, David Hunt FRACS, John Jorgensen FRACS, Christopher Martin FRACS, Mark Richardson FRACS, Garrett Smith FRACS, Ross Smith FRACS, David Storey FRACS.

Queensland: John Avramovic FRACS, John Croese FRACP, Justin D'Arcy FRACS, Stephen Fairley FRACP, John Hansen FRACS, John Masson FRACP, Les Nathanson FRACS, Barry O'Loughlin FRACS, Leigh Rutherford FRACS, Richard Turner FRACS, Morgan Windsor FRACS.

South Australia: Justin Bessell FRACS, Peter Devitt FRACS, Glyn Jamieson FRACS, David Watson FRACS.

Victoria: Stephen Blamey FRACS, Alex Boussioutas FRACP, Richard Cade FRACS, Gary Crosthwaite FRACS, Ian Faragher FRACS, John Gribbin FRACS, Geoff Hebbard FRACP, George Kiroff FRACS, Bruce Mann FRACS, Bob Millar FRACS, Paul O'Brien FRACS, Robert Thomas FRACS, Simon Wood FRACS.

Western Australia: Steve Archer FRACS, Kingsley Faulkner FRACS, Jeff Hamdorf FRACS.

References

1 World Cancer Research Fund/American Institute for Cancer Research (2007) Food, Nutrition, Physical Activity, and the Prevention of Cancer: A Global Perspective. Washington, DC: AICR.Google Scholar
2 Jemal, A, Center, MM, DeSantis, C, et al. (2010) Global patterns of cancer incidence and mortality rates and trends. Cancer Epidemiol Biomarkers Prev 19, 18931907.CrossRefGoogle ScholarPubMed
3 Gallus, S & La Vecchia, C (2007) Is there a link between diet and esophageal cancer? Nat Clin Pract Gastroenterol Hepatol 4, 23.CrossRefGoogle Scholar
4 Lord, RV, Law, MG, Ward, RL, et al. (1998) Rising incidence of oesophageal adenocarcinoma in men in Australia. J Gastroenterol Hepatol 13, 356362.CrossRefGoogle ScholarPubMed
5 Thomas, RJS, Lade, S, Giles, GG, et al. (1996) Incidence trends in oesophageal and proximal gastric carcinoma in Victoria. Aust N Z J Surg 66, 271275.CrossRefGoogle ScholarPubMed
6 Stavrou, EP, McElroy, HJ, Baker, DF, et al. (2009) Adenocarcinoma of the oesophagus: incidence and survival rates in New South Wales, 1972–2005. Med J Aust 191, 310314.CrossRefGoogle ScholarPubMed
7 Ziegler, RG (1991) Vegetables, fruits, and carotenoids and the risk of cancer. Am J Clin Nutr 53, 251S259S.CrossRefGoogle ScholarPubMed
8 Chainani-Wu, N (2002) Diet and oral, pharyngeal, and esophageal cancer. Nutr Cancer 44, 104126.CrossRefGoogle ScholarPubMed
9 Donaldson, MS (2004) Nutrition and cancer: a review of the evidence for an anti-cancer diet. Nutr J 20, 19.CrossRefGoogle Scholar
10 Tzonou, A, Lipworth, L, Garidou, A, et al. (1996) Diet and risk of esophageal cancer by histologic type in a low-risk population. Int J Cancer 68, 300304.3.0.CO;2-5>CrossRefGoogle Scholar
11 Mayne, ST, Risch, HA, Dubrow, R, et al. (2001) Nutrient intake and risk of subtypes of esophageal and gastric cancer. Cancer Epidemiol Biomarkers Prev 10, 10551062.Google ScholarPubMed
12 Chen, H, Tucker, KL, Graubard, BI, et al. (2002) Nutrient intakes and adenocarcinoma of the esophagus and distal stomach. Nutr Cancer 42, 3340.CrossRefGoogle ScholarPubMed
13 Lu, H, Cai, L, Mu, LN, et al. (2006) Dietary mineral and trace element intake and squamous cell carcinoma of the esophagus in a Chinese population. Nutr Cancer 55, 6370.CrossRefGoogle ScholarPubMed
14 Wu, AH, Tseng, CC, Hankin, J, et al. (2007) Fiber intake and risk of adenocarcinomas of the esophagus and stomach. Cancer Causes Control 18, 713722.CrossRefGoogle ScholarPubMed
15 Park, Y, Leitzmann, MF, Subar, AF, et al. (2009) Dairy food, calcium, and risk of cancer in the NIH-AARP Diet and Health Study. Arch Intern Med 169, 391401.CrossRefGoogle ScholarPubMed
16 Brown, LM, Swanson, CA, Gridley, G, et al. (1995) Adenocarcinoma of the esophagus: role of obesity and diet. J Natl Cancer Inst 87, 104109.CrossRefGoogle ScholarPubMed
17 De Stefani, E, Brennan, P, Boffetta, P, et al. (2000) Vegetables, fruits, related dietary antioxidants, and risk of squamous cell carcinoma of the esophagus: a case–control study in Uruguay. Nutr Cancer 38, 2329.CrossRefGoogle ScholarPubMed
18 De Stefani, E, Deneo-Pellegrini, H, Ronco, AL, et al. (2003) Food groups and risk of squamous cell carcinoma of the oesophagus: a case–control study in Uruguay. Br J Cancer 89, 12091214.CrossRefGoogle ScholarPubMed
19 Gonzalez, CA, Pera, G, Agudo, A, et al. (2006) Fruit and vegetable intake and the risk of stomach and oesophagus adenocarcinoma in the European Prospective Investigation into Cancer and Nutrition (EPIC-EURGAST). Int J Cancer 118, 25592566.CrossRefGoogle ScholarPubMed
20 Yamaji, T, Inoue, M, Sasazuki, S, et al. (2008) Fruit and vegetable consumption and squamous cell carcinoma of the esophagus in Japan: the JPHC study. Int J Cancer 123, 19351940.CrossRefGoogle ScholarPubMed
21 Freedman, ND, Park, Y, Subar, AF, et al. (2007) Fruit and vegetable intake and esophageal cancer in a large prospective cohort study. Int J Cancer 121, 27532760.CrossRefGoogle Scholar
22 Gonzalez, CA, Jakszyn, P, Pera, G, et al. (2006) Meat intake and risk of stomach and esophageal adenocarcinoma within the European Prospective Investigation Into Cancer and Nutrition (EPIC). J Natl Cancer Inst 98, 345354.CrossRefGoogle ScholarPubMed
23 Navarro Silvera, SA, Mayne, ST, Risch, H, et al. (2008) Food group intake and risk of subtypes of esophageal and gastric cancer. Int J Cancer 123, 852860.CrossRefGoogle ScholarPubMed
24 Schwerin, HS, Stanton, JL, Smith, JL, et al. (1982) Food, eating habits, and health: a further examination of the relationship between food eating patterns and nutritional health. Am J Clin Nutr 35, 13191325.CrossRefGoogle Scholar
25 Bahmanyar, S & Ye, W (2006) Dietary patterns and risk of squamous-cell carcinoma and adenocarcinoma of the esophagus and adenocarcinoma of the gastric cardia: a population-based case–control study in Sweden. Nutr Cancer 54, 171178.CrossRefGoogle ScholarPubMed
26 De Stefani, E, Boffetta, P, Ronco, AL, et al. (2008) Exploratory factor analysis of squamous cell carcinoma of the esophagus in Uruguay. Nutr Cancer 60, 188195.CrossRefGoogle ScholarPubMed
27 De Stefani, E, Deneo-Pellegrini, H, Boffetta, P, et al. (2009) Dietary patterns and risk of cancer: a factor analysis in Uruguay. Int J Cancer 124, 13911397.CrossRefGoogle Scholar
28 Hajizadeh, B, Rashidkhani, B, Rad, AH, et al. (2010) Dietary patterns and risk of oesophageal squamous cell carcinoma: a case–control study. Public Health Nutr 13, 11071112.CrossRefGoogle ScholarPubMed
29 Chen, H, Ward, MH, Graubard, BI, et al. (2002) Dietary patterns and adenocarcinoma of the esophagus and distal stomach. Am J Clin Nutr 75, 137144.CrossRefGoogle ScholarPubMed
30 De Stefani, E, Boffetta, P, Fagundes, RB, et al. (2008) Nutrient patterns and risk of squamous cell carcinoma of the esophagus: a factor analysis in Uruguay. Anticancer Res 28, 24992506.Google ScholarPubMed
31 Whiteman, DC, Sadeghi, S, Pandeya, N, et al. (2008) Combined effects of obesity, acid reflux and smoking on the risk of adenocarcinomas of the oesophagus. Gut 57, 173180.CrossRefGoogle ScholarPubMed
32 Spechler, SJ, Dixon, MF, Genta, R, et al. (2000) Adenocarcinoma of the oesophago-gastric junction. In World Health Organization Classiciation of Tumours: Pathology and Genetics of Tumours of the Digestive System [Hamilton, SR and Aaltonen, LA, editors]. Lyon: IARC Press.Google Scholar
33 Olsen, CM, Bain, CJ, Jordan, SJ, et al. (2007) Recreational physical activity and epithelial ovarian cancer: a case–control study, systematic review, and meta-analysis. Cancer Epidemiol Biomarkers Prev 16, 23212330.CrossRefGoogle ScholarPubMed
34 Willett, W, Sampson, L, Stampfer, MJ, et al. (1985) Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 122, 5165.CrossRefGoogle ScholarPubMed
35 Ashton, BA, Marks, GC, Battistutta, D, et al. (1996) Under reporting of energy intake in two methods of dietary assessment in the Nambour trial. Aust J Nutr Diet 53, 5360.Google Scholar
36 Marks, GC, Hughes, MC & van der Pols, JC (2006) The effect of personal characteristics on the validity of nutrient intake estimates using a food-frequency questionnaire. Public Health Nutr 9, 394402.CrossRefGoogle ScholarPubMed
37 Marks, GC, Hughes, MC & van der Pols, JC (2006) Relative validity of food intake estimates using a food frequency questionnaire is associated with sex, age, and other personal characteristics. J Nutr 136, 459465.CrossRefGoogle ScholarPubMed
38 McNaughton, SA, Marks, GC, Gaffney, P, et al. (2005) Validation of a food-frequency questionnaire assessment of carotenoid and vitamin E intake using weighed food records and plasma biomarkers: the method of triads model. Eur J Clin Nutr 59, 211218.CrossRefGoogle ScholarPubMed
39 Ibiebele, TI, Parekh, S, Mallitt, KA, et al. (2009) Reproducibility of food and nutrient intake estimates using a semi-quantitative FFQ in Australian adults. Public Health Nutr 12, 23592365.CrossRefGoogle ScholarPubMed
40 Hatcher, L (1994) A Step-by-Step Approach to Using SAS for Factor Analysis and Structural Equation Modeling. Cary, NC: SAS Publishing, SAS Institute, Inc.Google Scholar
41 Kubo, A, Levin, TR, Block, G, et al. (2008) Dietary patterns and the risk of Barrett's esophagus. Am J Epidemiol 167, 839846.CrossRefGoogle ScholarPubMed
42 Nettleton, JA, Diez-Roux, A, Jenny, NS, et al. (2008) Dietary patterns, food groups, and telomere length in the Multi-Ethnic Study of Atherosclerosis (MESA). Am J Clin Nutr 88, 14051412.CrossRefGoogle ScholarPubMed
43 Pandeya, N, Williams, G, Green, AC, et al. (2009) Alcohol consumption and the risks of adenocarcinoma and squamous cell carcinoma of the esophagus. Gastroenterology 136, 12151224, e1–e2.CrossRefGoogle ScholarPubMed
44 Sadeghi, S, Bain, CJ, Pandeya, N, et al. (2008) Aspirin, nonsteroidal anti-inflammatory drugs, and the risks of cancers of the esophagus. Cancer Epidemiol Biomarkers Prev 17, 11691178.CrossRefGoogle ScholarPubMed
45 Kolahdooz, F, Ibiebele, TI, van der Pols, JC, et al. (2009) Dietary patterns and ovarian cancer risk. Am J Clin Nutr 89, 297304.CrossRefGoogle ScholarPubMed
46 Ibiebele, TI, van der Pols, JC, Hughes, MC, et al. (2007) Dietary pattern in association with squamous cell carcinoma of the skin: a prospective study. Am J Clin Nutr 85, 14011408.CrossRefGoogle ScholarPubMed
47 Bosetti, C, Gallus, S, Trichopoulou, A, et al. (2003) Influence of the Mediterranean diet on the risk of cancers of the upper aerodigestive tract. Cancer Epidemiol Biomarkers Prev 12, 10911094.Google ScholarPubMed
48 Australian Bureau of Statistics (2006) National Health Survey: Summary of Results 2004–05 sVol. cat. no. 4364·0, Canberra.Google Scholar
49 Pandeya, N, Williams, GM, Green, AC, et al. (2009) Do low control response rates always affect the findings? Assessment of smoking and obesity in two Australian case–control studies of cancer. Aust N Z Public Health 33, 312319.CrossRefGoogle ScholarPubMed
Figure 0

Table 1 Factor loadings* for the relationship between food groups and factors representing dietary patterns in oesophageal cancer cases and controls

Figure 1

Table 2 Characteristics of cases and controls(Numbers and percentages)

Figure 2

Table 3 Characteristics of 1507 control participants according to quartiles (Q) of dietary pattern score(Numbers, percentages, mean values and standard deviations)

Figure 3

Table 4 Associations between dietary patterns and risk of oesophageal cancers(Numbers, percentages, odds ratios and 95 % confidence intervals)

Supplementary material: File

Ibiebele supplementary table 1

Ibiebele supplementary table 1

Download Ibiebele supplementary table 1(File)
File 30.7 KB
Supplementary material: File

Ibiebele supplementary table 2

Ibiebele supplementary table 2

Download Ibiebele supplementary table 2(File)
File 64 KB