- Email: [email protected]
Dietary patterns and risk of oral cancer: A factor analysis study of a population in Jakarta, Indonesia
Oral Oncology 45 (2009) e49–e53
Contents lists available at ScienceDirect
Oral Oncology journal homepage: www.elsevier.com/locate/oraloncology
Dietary patterns and risk of oral cancer: A factor analysis study of a population in Jakarta, Indonesia Rahmi Amtha a,b,*, Rosnah Zain b, Ishak Abdul Razak b, Bastaman Basuki c, Boedi Oetomo Roeslan d, Walta Gautama e, Denni Joko Purwanto e a
Oral Medicine Department, Trisakti University, Kyai Tapa, Grogol, Jakarta Barat, Indonesia Oral Cancer Research Coordinating Centers (OCRCC), University of Malaya, Kuala Lumpur, Malaysia c Community Medicine Department, University of Indonesia, Indonesia d Biochemistry Department, Trisakti University, Indonesia e Surgical Oncology Department, Dharmais National Cancer Hospital, Indonesia b
a r t i c l e
i n f o
Article history: Received 22 April 2008 Received in revised form 5 January 2009 Accepted 6 January 2009 Available online 28 February 2009 Keywords: Dietary patterns Oral cancer Factor analysis Jakarta Indonesia
s u m m a r y A matched case-control, hospital-based study of oral cancer was conducted in Jakarta population. The sample included 81 cases and 162 controls. The purpose of this study was to determine the association between dietary pattern and oral cancer in a Jakarta population using factor analysis. Dietary data were collected using food frequency questionnaire and factor analysis was performed on 15 food groups resulting in four principle factors/components being retained. The first factor ‘‘preferred” was characterized by fast food, fermented food, canned food, snacks high in fat and sugar, cooked and raw vegetables, and seafood. The second factor labeled ‘‘combination” was loaded by the intake of dairy product, red meat, white meat and fruits. The third factor labeled ‘‘chemical related was loaded by processed food and monosodium glutamate and the fourth principle component consisted of drinks and grain was labeled as ‘‘traditional”. The conditional logistic regression was done using STATA 8 to obtain the odds ratio (OR) of highest tertile of each component retained from factor analysis and the ORs were then adjusted with risk habits. The consumption the highest tertile of the ‘‘preferred” pattern increased the risk of oral cancer by two-times compared to the lowest tertile of consumption [adjusted odds ratio (aOR) = 2.17; 95% confidence interval (CI) = 1.05–4.50]. The chemical related” pattern showed higher risk of about threefold (aOR = 2.56; 95% CI = 1.18–5.54), while the ‘‘traditional” pattern showed an increased of risk by twofold (aOR = 2.04; 95% CI = 1.01–4.41). In contrast, the ‘‘combination” pattern displayed protective effects in relation to oral cancer (aOR = 0.50; 95% CI = 0.24–1.00). This finding suggests that factor analysis may be useful to determine the diet pattern of a big set of food type and establish the correlation with oral cancer. Ó 2009 Elsevier Ltd. All rights reserved.
Introduction Oral cancer is one of the most common cancers worldwide.1 Tobacco smoking, alcoholic consumption and betel quid chewing have been found to be major risk factors. Besides that, epidemiological studies have shown that diet plays a role in many chronic diseases and various types of cancer including oral cancer.2,3 The American Institute for Cancer Research4 estimated that diet could account for approximately 30% of cancer deaths in United States, similar to the number accounted for by smoking.4 The World Health Organization acknowledges that up to 30% of human cancers are probably related to diet and nutrition.5
* Corresponding author. Address: Oral Medicine Department, Trisakti University, Kyai Tapa, Grogol, Jakarta Barat, Indonesia. Tel.: +62 811184174; fax: +62 215655787. E-mail address: [email protected] (R. Amtha). 1368-8375/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.oraloncology.2009.01.007
Most of the epidemiological studies of diet in relation to oral cancer have been conducted in western countries while studies on food and dietary pattern and risk of oral cancer are scarce in the Asian regions. So far only three studies have been reported in Asian countries regarding the food and diet pattern in association with oral cancer.6,7 The most consistent findings in diet as determinant of cancer risk is the association between consumption of vegetables and fruit in reducing risk of several cancers. About 80% of these studies found a significant protective effect of overall consumption of vegetables and/or fruit or at least of some types of vegetables and fruits.8–12 Studies to determine the relationship between diet and oral cancer presents a challenge due to the complexity of the human diet. Such complexity arose from the fact that food may contain chemical compounds which are well-known for some, while many could not be measured and is still poorly characterized. However, most of these studies are confined to single aspect of food type
e50
R. Amtha et al. / Oral Oncology 45 (2009) e49–e53
where the interaction of each food type may not be taken into account. More recently, several researchers have introduced a new alternative approach in analyzing the dietary pattern and risk of oral cancer.13,14 This approach is called factor analysis which uses the correlation between food and nutrient intake to describe a general dietary pattern which at a later stage may be related to the risk of oral cancer. The factor analysis also values the effect of the diet which is not mediated by one or two specific nutrients, but by nutrients that perhaps operate interactively.15 To date there is no report in the English literature on factor analysis for oral cancer in the Asian region. The purpose of this study was to determine the association between dietary pattern and oral cancer in a Jakarta population using factor analysis. Methods Subjects A case-control study of oral cancer was conducted in five hospitals in Jakarta from January 2005 to April 2006. Cases were subjects with incident cancer, histologically confirmed as squamous cell carcinoma of the oral cavity (C00–C06) excluding salivary gland, pharynx and nasopharynx (C07–C11) (International Classification of Disease 10th Revision). A total of 81 eligible cases aged 23– 74 years (mean age 47.4 years ± 12.4) were included and interviewed using a structured questionnaire which had been validated. The questionnaire consisted of demographic information and risk habits (smoking, alcohol and betel quid chewing) and dietary information using food frequency questionnaire (FFQ) developed for the Indonesian population. One hundred and sixty-two hospital-based matched control subjects aged 22–79 years (mean age 46.9 ± 11.8) were selected for this study. Controls were randomly selected among non-cancer patients who attended the centers for minor aliments. Two control patients were selected for each case on the basis of sex and age within 5 years. Non-Indonesian citizens and those who had cancer, cardiovascular disease and who were undergoing treatment were excluded from this study. A signed informed consent was obtained from all participants and the research protocol study was approved by the Ethics committees of Trisakti University (as main center) and University of Malaya, Kuala Lumpur. Food grouping The FFQ list contained 141 food items which were grouped into 15 food groups as shown in Table 1. Each participant in the study was asked to respond to each item of the FFQ, and the mean daily
equivalent was computed. These grouping were based on the assumption of similarity of nutrient content of the various food items shown in Table 1. The factor analysis was used to identify dietary factors or combinations of foods consumed by the studied sample.14 All variables are considered simultaneously, each one may be related to the others. Factor analysis Kaiser–Meyer–Olkin (KMO) measurement of sample adequacy and Bartlett test of sphericity were initially performed to verify the appropriateness of using factor analysis for this study. When the KMO was more than 0.600 and Bartlett’s test of sphericity was less than 0.05, the sample was considered adequate for factor analysis.16 This was followed by an exploratory factor analysis applied to the 15 food groups to find principle factors/component that would explain the maximum fraction of the variance present in the study groups. A graphical method called the Scree plot would determine the factors that should be retained (the eigenvalues of the factors retained should be more than 1). An orthogonal rotation using varimax method16 was then applied to simplify the principle component and render it easier to interpret. The next step was to name the retained factors. Names were given to reflect the food groups having the highest loadings on that factor. Food groups having a positive loading on a factor/component would contribute directly to that dietary pattern while food groups with negative loading were inversely associated with a given dietary pattern. When a food group was loaded on more than one factors/component, only the factor/component with the highest loading was considered for factor naming. The factor loading of more than 0.4 (for sample size 200–249) using BMDP statistical calculation would be considered as having contributed significantly to the factors. The communality would also be shown which indicates that a particular variable has much in common with other variables taken as a group and loaded in component or factor. Dietary pattern and risk of oral cancer In order to obtain the correlation of each factor/component retained from factor analysis and oral cancer, the 15 food groups were categorized into tertile based on the entire study population and conditional logistic regression (STATA 8 version) was then performed to obtain the risk. The univariate model was undertaken using conditional logistic regression to the confounding effect of each variable. Multivariate analysis model (adjusted odds ratio) was performed for variables which have p > 0.250 in the univariate model.17 A variable was considered to be a confounder if its inclu-
Table 1 Food group classification. Food groups
Food types
1. Grain 2. White meat 3. Red meat 4. Sea food 5. Cooked vegetables 6. Raw vegetables 7. Dairy products 8. Fruits 9. Fermented food 10. Snack 11. Canned food 12. Processed food 13. Fast food 14. Other 15. Drinks
Rice, bread, noodle, sweet potato, cassava, sago, corn Chicken, duck, bird, rabbit Beef and lamb Fish, prawn, squid Leafy green, beans, cruciferous, brinjal, pumpkin, bamboo shoots, and radish Cucumber, others, e.g., petai, jengkol Cheese, margarine, yoghurt, skim and powder milk Banana, papaya, water melon, apple, honeydew, mango, pineapple, jackfruit, guava, orange and fruit juices Salty egg, salty fish, fermented prawn, soya ketchup, fermented fruit, and pickles Fried stuff, desert, sweets and cakes Fish, tomato paste, mushroom Sausages, nugget, and meat/fish balls Chips, hamburger, pizza, instant noodle, ready made spices Monosodium glutamate/MSG Coffee, tea, and canned/carbonated drink
e51
R. Amtha et al. / Oral Oncology 45 (2009) e49–e53
sion in a model changed the OR estimation by more than 10%. The variables that fulfill these criteria as confounders are kept in the models presented with OR estimated by method of maximum likelihood. The 95% confidence intervals (95% CI) were based on the standard error of coefficient estimated and p-value for trend was obtained. Results The sociodemographic characteristics of subjects are as presented in Table 2. This study showed the KMO of 0.671 (more than 0.600) and the Bartlett’s test of sphericity of 0.000 (less than 0.05), making the sample of this study adequate for factor analysis. Four principle factors/components were retained through factor analysis (Table 3), based on KMO and Scree plot where the eigen value was more than 1 (Fig. 1). The first factor/component, which accounted for 27% of the total variance, is labeled as ‘‘preferred”. Fast food, fermented food, canned food, snacks high in fat and sugar, cooked and raw vegetables, and seafood falls into this comTable 2 Sociodemographic characteristic of subjects. Cases
(N = 162)
(N = 81)
N Age group 22–34 35–49 >49 Gender Male Female Marital status Married Single Divorce/widow Risk habits No habit Smoking Alcohol Betel quid chewing Smoking and alcohol Smoking and betel quid Smoking, alcohol, betel quid
5
%
N
%
24 69 69
14.8 42.6 42.6
16 30 35
19.8 37.0 43.2
100 62
61.7 38.3
50 31
61.7 38.3
134 15 13
82.7 9.3 8.02
73 5 3
90.1 6.2 3.07
34 30 1 2 10 3 1
42 37 1.2 2.5 12.3 3.7 1.2
79 65 1 2 14 0 1
48.8 40.1 0.6 1.2 8.6 0 0.6
4
Eigenvalue
Control
ponent. The second factor contributed 11% of the total variance. This factor was loaded by the intake of dairy product, red meat, white meat and fruit labeled ‘‘combination”. The third factor accounted for approximately 9% of the total variance. In this component high loading factor was observed in the processed food and MSG labeled ‘‘chemical related”. The fourth principle component which accounted for approximately 8% of the total variance was loaded by two food group, drinks and grain and this factor has been labeled ‘‘traditional”. The highest communality was shown in ‘‘Fast food” (0.818) and loaded in first factor/component labeled ‘‘preferred”. Table 4 shows that the consumption in the highest tertile of the ‘‘preferred” pattern was found to increase the risk of oral cancer by two-times (adjusted OR 2.17, 95% CI 1.05–4.50, v2 trend 5.446, p < 0.05). The intake of the highest tertile of ‘‘chemical related” pattern was found to be associated with higher risk of about threefold after adjusting for controls (adjusted OR 2.56, 95% CI 1.18–5.54, v2 trend 5.640, p < 0.05). On the other hand, the highest tertile of ‘‘combination” pattern displayed protective effects in relation to oral cancer before and after adjusting for the variables mentioned earlier (adjusted OR 0.50, 95% CI 0.24–1.00, v2 trend 7.335, p < 0.01). Consumption of highest tertile of ‘‘traditional” pattern showed an increased of risk by twofold (OR 2.04, 95% CI 1.01–
3
2
1
0 1
2
3
4
5
6
7
8
9
10 11 12 13 14 15
Component Number Figure 1 Scree plot showing eigenvalues for 15 components (food groups), in the factor extraction of data obtained from food frequency questionnaire.
Table 3 Loadings factor for the four components identified from factor analysis. Food group
Fast food Cooked vegetables Raw vegetables Fermented food Seafood Canned food Snack high fat and sugar Red meat Dairy product Fruits White meat Processed food MSG Drinks Grain Eigenvalue % Explained variance % Cumulative variance
Dietary pattern
Communality
Preferred food
Combination
Chemical related
Traditional
0.749 0.737 0.709 0.690 0.621 0.572 0.483 0.098 0.084 0.522 0.115 0.015 0.118 0.027 0.418 4.090 27.264 27.264
0.018 0.155 0.015 0.040 0.250 0.082 0.307 0.730 0.722 0.584 0.545 0.082 0.156 0.028 0.077 1.676 11.174 38.438
0.463 0.055 0.152 0.450 0.205 0.260 0.402 0.212 0.149 0.175 0.395 0.588 0.467 0.095 0.127 1.376 9.175 47.613
0.206 0.144 0.039 0.013 0.028 0.261 0.285 0.065 0.173 0.027 0.204 0.040 0.451 0.723 0.518 1.144 7.624 55.237
0.818 0.591 0.528 0.681 0.491 0.469 0.570 0.591 0.580 0.645 0.508 0.354 0.460 0.534 0.465
e52
R. Amtha et al. / Oral Oncology 45 (2009) e49–e53
Table 4 Risk of oral cancer in approximate tertile for food groups defined by factor analysis. Factor Preferred
a
Combinationb
Chemical relatedb
Traditionalb
Tertile of score
Control:cases
Crude OR (95% CI)
Adjusted OR (95% CI)
v2 p trend
1st (5.1) 2nd (8.0) 3rd 1st (1.7) 2nd (3.0) 3rd 1st (2.0) 2nd (2.0) 3rd 1st (4.0) 2nd (5.0) 3rd
58:23 58:23 58:46 42:39 61:19 59:23 67:22 54:24 41:35 60:22 57:23 45:36
1.00 1.04 1.89 1.00 0.35 0.42 1.00 1.43 2.56 1.00 1.15 2.09
1.00 1.05 2.17 1.00 0.36 0.46 1.00 1.48 2.85 1.00 1.22 2.10
5.446*
(0.52–2.07) (0.98–3.64) (0.17–0.70) (0.22–0.81) (0.67–3.07) (1.18–5.54) (0.57–2.26) (1.07–4.06)
(0.53–2.28) (1.05–4.50) 7.335** (0.18–0.75) (0.23–0.91) 5.640* (0.67–3.28) (1.34–6.05) 5.649* (0.59–2.53) (1.02–4.30)
a
Adjusted for smoking status, smoking duration, type of tobacco, number stick of cigarettes, pack-years. Adjusted for smoking, alcohol and betel quid. p < 0.05. ** p < 0.01. b
*
4.41, v2 trend 5.649, p < 0.05) after allowing for ethnic, and dietary intake habit. Discussion In the present study, the dietary pattern was explored in oral cancer subjects from five hospitals in Jakarta, Indonesia and matched controls. Four components named preferred, combination, chemical related and traditional which explained 55% of the variability within the samples was identified. The factors/components identified were then used as co-variables in determining whether the effect of specific nutrient is independent of the dietary pattern. The first factor/component identified as ‘‘preferred food” characterized by the consumption of fast food, cooked and raw vegetables, fermented food, seafood, canned food, and snack high in fat and sugar, was associated with doubling of risk of oral cancer. According to the adjusted risk estimate, it was found that subjects with scores in the highest tertile of the ‘‘preferred food” pattern had greater risk of oral cancer than subjects with lower scores. This finding is consistent with many studies which found that fast food, fermented, canned food and meals that contain high fat and sugar may increase the risk of cancers and other disease such as hypertension and diabetes.5 Fast food (chips, hamburger, pizza, ready made spices and especially instant noodle) represented the greatest communalities in this study, which mean that this food type has much in common with other variables taken as a group and loaded in component which is labeled ‘‘preferred food”. A plausible explanation of the contribution of these foods to oral cancer is the fact that fast food mainly composed of high levels of fat which can produce PAHs during curing (food preparation) using high temperature. The PAHs found in the food have been shown to cause cancer in laboratory animals.11 In addition, the fermented, canned food or processed food in the ‘‘preferred” component may also increase the risk of cancer.5 The world cancer report had also noted that the consumption of salted food as well as preserved and canned food (rich in nitrate performed N-nitroso compounds) is associated with cancer especially gastric cancer. Several biological mechanisms have also been proposed to explain the association between Chinese-style salted fish and nasopharyngeal cancer, including partial fermentation and nitrosamine formation.5 However, the first factor/component also consisted of vegetables (cooked and raw) thus contributing to the causal relationship, although the mechanism of vegetables in increasing the risk of oral cancer is not fully understood. In addition, factor analysis also indicates that single factor alone may not account for disease causation or prevention as other factors such as cultural, social and demo-
graphic characteristics in the same component will interact interchangeably. Considering the varied style of cooking vegetables in Indonesia, there are some plausible reasons for the observation which contradict the previous literature. Astawan in his study found that there are four common styles of cooking vegetables in Indonesia.18 Firstly, the habit of washing the vegetables after being cut, leading to wider surface area of vegetables being exposed to air and water, may result in depletion of some mineral and vitamins (especially C and B). Secondly, the habit of cooking the vegetables with an abundance of water resulting in dissolution of the majority of vitamins, minerals and amino acids11 in the soup and the remaining soup was left uneaten. The third is the habit of boiling the vegetables with water together before the water temperature reaches the boiling point. This may lead to depletion of some of the active constituent of vegetables. The fourth habit is that of overcooking the vegetables as well as reheating them many times. All the above styles may contribute to decreasing the quality of cooked vegetables in Indonesia and if they interact with other risk factors may increase the risk of oral cancer. The second factor/component named ‘‘combination” characterized by meat, dairy product and fruit showed inverse relationship to oral cancer. Fruit and vegetables have been consistently proven as a protective food type in many cancers. In regards to oral cancer, this finding is supported by many studies6,8,10,19,20 who found the strongest protective effects derived from citrus fruit and others which are likelier to be eaten raw, thus pointing to a mechanical cleansing effect of raw fruit and vegetables on the oral cavity and/ or to the beneficial effect of some temperature-sensitive substance(s), such as vitamin C. Meat and dairy product being also constituents of the combination factor would have contributed to the protective effect of this factor and thus would be opposite to some other studies. This other studies conducted in European countries had shown that meat, cheese and milk increased the risk of oral cancer.8,19 However, a study conducted by Franceschi et al. found that major sources of protein and fats in Italian diet (milk, eggs, oil and cheese) and some non-alcoholic beverages did not seem to have a significant influence on development of oral cancer.21 Similarly, Zeng et al. did a study in Beijing, and found that protein and fat intake are related inversely with oral cancer risk.6 Drinking milk showed a moderate inverse association with oral cancer risk with marginally significant OR of 0.5 (95% CI 0.2–1.0). One possibility for the differing finding between this and other European studies is that the food preparation in European food might be different with Asian, especially Indonesian. The most common meat preparation method recorded in this study was by boiling (such as soup, gulai and rendang). Whereas the majority of meat preparations in the European countries such as meat, requires it to be grilled, baked or
R. Amtha et al. / Oral Oncology 45 (2009) e49–e53
roasted and sometimes deep fried. As mentioned earlier the PAHs are carcinogenic and can be found in great amount in food prepared in this manner. Grilled meats are high in PAHs which is formed when the fat drips onto the flame below and it attaches on the meat surface. The broiled and deep fried meats also contain other hazardous agent heterocyclic amines (HAs) such as benzopyrene formed from the burning of amino acids and other substances in the meat under very high temperature.11 This is supported by studies which found the positive associations of oral cancer with intake of nitriterich meat,19 salted meat6 and charchoal-grilled meat.22 However, the findings that meats as part of the ‘‘combination” factor is protective may be explain by the fact that protein which contain of essential amino acid is the basic needs of the body in cell building and defense (immune) mechanism.11 The highest tertile scores of the third food pattern categorized as ‘‘chemical related” food significantly increase the risk of oral cancer. This third component was loaded by processed food and MSG. The chemical contained in this food group has been emphasized as carcinogenic agent to cancer. However, this still need further investigation as some review stated that monosodium glutamate (MSG) is the amino acid which can be metabolized automatically by the body.11 The fourth component which also increased the risk of oral cancer in this study is categorized under ‘‘traditional”. It is characterized by all types of drink (soft drink, canned drink, coffee and tea) and staple food (Indonesia majority rice) which produce high calories. This finding is also supported by Franceschi et al. who found an association between drinks and oral cavity and pharynx cancers.8 The OR of the highest tertile of daily intake of soft drink was 1.6 (95% CI 1.3–2.2) compared to the lowest intake. Carbohydrate intake has also been shown to moderately increase the risk of oral cancer in the study conducted in Beijing.6 In this study, the food items were grouped by the similarity in terms of composition and nutrient value, based on previous studies.6,14,21,23 The four retained factors explain more than half of the total variance. If the patterns fail to explain much of the variance in the food intake as a whole, it is possible that these patterns would not explain much of the variance in a single food or nutrient either, thus limiting their use in nutritional epidemiology.24 Studies in dietary pattern in Indonesia are scarce, and to the best of our knowledge, this is the first study to utilize principal component (factor) analysis in the context of oral cancer. Although the pattern observed in this study provides an indication of the common food consumed the data should not be over-interpreted because of the limitation in sample size. Furthermore, dietary patterns were extracted from data obtained within the study population, therefore, the results cannot be generalized for the Indonesian populations with different dietary habits. Nevertheless, based on these findings, and as a start for the Indonesian population, dietary guides with emphasis on foods and on overall dietary patterns which are health promoting or damaging to the health of the population may be issued. In conclusion this study had shown that meaningful association between dietary pattern and oral cancer was possible using factor analysis, the outcome of which would be used to determine the association with oral cancer. Further research is suggested on the effects of micro- and macro-nutrient composition of food as well as the calories intake and in establishing its link to oral cancer, so that it will be possible to develop nutritional intervention based on these patterns in Indonesia. Conflict of Interest Statement This project has no conflict of interest with any institution that supported my project.
e53
Acknowledgement This study was supported by research grants from Trisakti University, Indonesia (no. 0769/Usakti/bsdm/db/10/2004) and University of Malaya (F0150/2005A). We thank Dr. Nugroho Abikusno (Medical Faculty University of Trisakti, Indonesia) and Dr. Cheong Sok Ching (Cancer Research Initiative Foundation, Kuala Lumpur) for their advice. References 1. Parkin MD, Bray F, Ferlay J, Pisani P. Global cancer statistic, 2002. CA A Cancer J Clin 2005;55:74–108. 2. De Stefani E, Deneo-Pellegrini1 H, Ronco AL, Boffetta P, Brennan P, Mun~oz N Castellsaque X, et al. Food groups and risk of squamous cell carcinoma of the oesophagus: a case-control study in Uruguay. Br J Cancer 2003;89: 1209–14. 3. Ronco AL, De Stefani E, Boffetta P, Deneo-Pellegrini H, Acosta G, Mendilaharsu M. Food patterns and risk of breast cancer: a factor analysis study in Uruguay. Int J Cancer 2006;119(7):1672–8. 4. American Institute for Cancer Research, World Cancer Research Fund. Food, nutrition, and the prevention of cancer: a global perspective. Washington, DC: The American Institute for Cancer Research; 1997. 5. Stewart BW, Kleihues P. In: Stewart Bernard W, Kleihues Paul, editors. The cause of cancer in world cancer report. Lyon, France: IARC Press; 2003. p. 22–31. 6. Zheng T, Boyle P, Walter WC, Hu H, Dan J, Evstifeeva TV, et al. A case-control study of oral cancer in Beijing, People’s Republic of China. Association with nutrient intakes, foods and food groups. Oral Oncol Eur J Cancer 1993;9B(1):45–55. 7. Takezaki T, Hirose K, Inoue M, Hamajima N, Kuroishi T, Nakamura S, et al. Tobacco, alcohol and dietary factors associated with the risk of oral cancer among Japanese. Jpn J Cancer Res 1996;87:555–62. 8. Franceschi S, Favero A, Conti E, Talamini R, Volpe R, Negri E, et al. Food groups, oils and butter and cancer of the oral cavity and pharynx. Br J Cancer 1999;80(3–4):614–20. 9. Voorrips LE, Goldbohm A, van Poppel G, Sturmans F, Hermus RJJ, van den Brandt PA. Vegetable and fruit consumption and risks of colon and rectal cancer in a prospective cohort study. The Netherlands cohort study on diet and cancer. Am J Epidemiol 2000;152:1081–92. 10. Soler M, Bosetti C, Franceschi S, Negri E, Zambon P, Talamini R, et al. Fiber intake and the risk of oral, pharyngeal and esophageal cancer. Int J Cancer 2001;91:283–7. 11. Grrosvenor MB, Smolin LA. Nutrition. From science to life. USA: Harcourt College; 2002.. p. 664–725. 12. Cappuccio FP, Rink E, Perkins-Porras L, McKay C, Hilton S, Steptoe A. Estimation of fruit and vegetable intake using a two-item dietary questionnaire: a potential tool for primary health care workers. Nutr Metab Cardivasc Dis 2003;13:12–9. 13. Marchioni DML, de Oliveira Latorre MRD, Eluf-Neto J, Wunsch-Filho V, Fishberg RM. Identification of dietary patterns using factor analysis in an epidemiological study in Sao Paulo. San Paulo Med J 2005;123:124–7. 14. Marchioni DML, Fishberg RM, de Gois Filho, Kowalski Lp, de Carvalho MB, Abrahao M, et al. Dietary patterns and risk of oral cancer: a case-control study in Sao Paulo, Brazil. Rev Saude Publica 2007;41:19–26. 15. Trichopoulous D, Lagiou P. Dietary patterns and mortality. Br J Nutr 2001;85(2):133–4. 16. Pallant J. Factor analysisSPSS survival manual. Sydney: Allen & Unwin; 2005. p. 172–193. 17. Schlesselmen JJ. Case control studies. Design, conduct, analysis. New York: Oxford University Press; 1982. 18. Astawan M. Kandungan gizi aneka bahan makanan. Jakarta: Gramedia; 2004. 19. McLaughlin JK, Gridley G, Block G, Winn DM, Preston-Martin S, Scoenberg JB, et al. Dietary factors in oral and pharyngeal cancer. J Nat Cancer Inst 1988;80:1237–43. 20. Franceschi S, Bidoli E, Barón AE, Barra S, Talamini R, Serraino D, et al. Nutrition and cancer of the oral cavity and pharynx in north-east Italy. Int J Cancer 1991;47(1):20–5. 21. Franceschi S, Barra S, La Vecchia C, Bidoli E, Negri E, Talamini R. Risk factors for cancer of the tongue and the mouth. Cancer 1992;70(9):2227–33. 22. Franco EL, Kowalski LP, Oliveira BV, Curado MP, Pereira RN, Silva ME, et al. Risk factors for oral cancer in Brazil: a case-control study. Int J Cancer 1989;43:992–1000. 23. Franceschi S, Levi F, Conti E, Talamini R, Negri E, Dal Maso L, et al. Energy intake and dietary pattern in cancer of the oral cavity and pharynx. Cancer Causes Control 1999;10:439–44. 24. Schulze MB, Hoffman K, Kroke K, Boieng H. Dietary patterns and their association with food and nutrient intake in the European prospective investigation into cancer and nutrition (EPIC) – postdam study. Br J Nutr 2001;85:363–73.
Contents lists available at ScienceDirect
Oral Oncology journal homepage: www.elsevier.com/locate/oraloncology
Dietary patterns and risk of oral cancer: A factor analysis study of a population in Jakarta, Indonesia Rahmi Amtha a,b,*, Rosnah Zain b, Ishak Abdul Razak b, Bastaman Basuki c, Boedi Oetomo Roeslan d, Walta Gautama e, Denni Joko Purwanto e a
Oral Medicine Department, Trisakti University, Kyai Tapa, Grogol, Jakarta Barat, Indonesia Oral Cancer Research Coordinating Centers (OCRCC), University of Malaya, Kuala Lumpur, Malaysia c Community Medicine Department, University of Indonesia, Indonesia d Biochemistry Department, Trisakti University, Indonesia e Surgical Oncology Department, Dharmais National Cancer Hospital, Indonesia b
a r t i c l e
i n f o
Article history: Received 22 April 2008 Received in revised form 5 January 2009 Accepted 6 January 2009 Available online 28 February 2009 Keywords: Dietary patterns Oral cancer Factor analysis Jakarta Indonesia
s u m m a r y A matched case-control, hospital-based study of oral cancer was conducted in Jakarta population. The sample included 81 cases and 162 controls. The purpose of this study was to determine the association between dietary pattern and oral cancer in a Jakarta population using factor analysis. Dietary data were collected using food frequency questionnaire and factor analysis was performed on 15 food groups resulting in four principle factors/components being retained. The first factor ‘‘preferred” was characterized by fast food, fermented food, canned food, snacks high in fat and sugar, cooked and raw vegetables, and seafood. The second factor labeled ‘‘combination” was loaded by the intake of dairy product, red meat, white meat and fruits. The third factor labeled ‘‘chemical related was loaded by processed food and monosodium glutamate and the fourth principle component consisted of drinks and grain was labeled as ‘‘traditional”. The conditional logistic regression was done using STATA 8 to obtain the odds ratio (OR) of highest tertile of each component retained from factor analysis and the ORs were then adjusted with risk habits. The consumption the highest tertile of the ‘‘preferred” pattern increased the risk of oral cancer by two-times compared to the lowest tertile of consumption [adjusted odds ratio (aOR) = 2.17; 95% confidence interval (CI) = 1.05–4.50]. The chemical related” pattern showed higher risk of about threefold (aOR = 2.56; 95% CI = 1.18–5.54), while the ‘‘traditional” pattern showed an increased of risk by twofold (aOR = 2.04; 95% CI = 1.01–4.41). In contrast, the ‘‘combination” pattern displayed protective effects in relation to oral cancer (aOR = 0.50; 95% CI = 0.24–1.00). This finding suggests that factor analysis may be useful to determine the diet pattern of a big set of food type and establish the correlation with oral cancer. Ó 2009 Elsevier Ltd. All rights reserved.
Introduction Oral cancer is one of the most common cancers worldwide.1 Tobacco smoking, alcoholic consumption and betel quid chewing have been found to be major risk factors. Besides that, epidemiological studies have shown that diet plays a role in many chronic diseases and various types of cancer including oral cancer.2,3 The American Institute for Cancer Research4 estimated that diet could account for approximately 30% of cancer deaths in United States, similar to the number accounted for by smoking.4 The World Health Organization acknowledges that up to 30% of human cancers are probably related to diet and nutrition.5
* Corresponding author. Address: Oral Medicine Department, Trisakti University, Kyai Tapa, Grogol, Jakarta Barat, Indonesia. Tel.: +62 811184174; fax: +62 215655787. E-mail address: [email protected] (R. Amtha). 1368-8375/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.oraloncology.2009.01.007
Most of the epidemiological studies of diet in relation to oral cancer have been conducted in western countries while studies on food and dietary pattern and risk of oral cancer are scarce in the Asian regions. So far only three studies have been reported in Asian countries regarding the food and diet pattern in association with oral cancer.6,7 The most consistent findings in diet as determinant of cancer risk is the association between consumption of vegetables and fruit in reducing risk of several cancers. About 80% of these studies found a significant protective effect of overall consumption of vegetables and/or fruit or at least of some types of vegetables and fruits.8–12 Studies to determine the relationship between diet and oral cancer presents a challenge due to the complexity of the human diet. Such complexity arose from the fact that food may contain chemical compounds which are well-known for some, while many could not be measured and is still poorly characterized. However, most of these studies are confined to single aspect of food type
e50
R. Amtha et al. / Oral Oncology 45 (2009) e49–e53
where the interaction of each food type may not be taken into account. More recently, several researchers have introduced a new alternative approach in analyzing the dietary pattern and risk of oral cancer.13,14 This approach is called factor analysis which uses the correlation between food and nutrient intake to describe a general dietary pattern which at a later stage may be related to the risk of oral cancer. The factor analysis also values the effect of the diet which is not mediated by one or two specific nutrients, but by nutrients that perhaps operate interactively.15 To date there is no report in the English literature on factor analysis for oral cancer in the Asian region. The purpose of this study was to determine the association between dietary pattern and oral cancer in a Jakarta population using factor analysis. Methods Subjects A case-control study of oral cancer was conducted in five hospitals in Jakarta from January 2005 to April 2006. Cases were subjects with incident cancer, histologically confirmed as squamous cell carcinoma of the oral cavity (C00–C06) excluding salivary gland, pharynx and nasopharynx (C07–C11) (International Classification of Disease 10th Revision). A total of 81 eligible cases aged 23– 74 years (mean age 47.4 years ± 12.4) were included and interviewed using a structured questionnaire which had been validated. The questionnaire consisted of demographic information and risk habits (smoking, alcohol and betel quid chewing) and dietary information using food frequency questionnaire (FFQ) developed for the Indonesian population. One hundred and sixty-two hospital-based matched control subjects aged 22–79 years (mean age 46.9 ± 11.8) were selected for this study. Controls were randomly selected among non-cancer patients who attended the centers for minor aliments. Two control patients were selected for each case on the basis of sex and age within 5 years. Non-Indonesian citizens and those who had cancer, cardiovascular disease and who were undergoing treatment were excluded from this study. A signed informed consent was obtained from all participants and the research protocol study was approved by the Ethics committees of Trisakti University (as main center) and University of Malaya, Kuala Lumpur. Food grouping The FFQ list contained 141 food items which were grouped into 15 food groups as shown in Table 1. Each participant in the study was asked to respond to each item of the FFQ, and the mean daily
equivalent was computed. These grouping were based on the assumption of similarity of nutrient content of the various food items shown in Table 1. The factor analysis was used to identify dietary factors or combinations of foods consumed by the studied sample.14 All variables are considered simultaneously, each one may be related to the others. Factor analysis Kaiser–Meyer–Olkin (KMO) measurement of sample adequacy and Bartlett test of sphericity were initially performed to verify the appropriateness of using factor analysis for this study. When the KMO was more than 0.600 and Bartlett’s test of sphericity was less than 0.05, the sample was considered adequate for factor analysis.16 This was followed by an exploratory factor analysis applied to the 15 food groups to find principle factors/component that would explain the maximum fraction of the variance present in the study groups. A graphical method called the Scree plot would determine the factors that should be retained (the eigenvalues of the factors retained should be more than 1). An orthogonal rotation using varimax method16 was then applied to simplify the principle component and render it easier to interpret. The next step was to name the retained factors. Names were given to reflect the food groups having the highest loadings on that factor. Food groups having a positive loading on a factor/component would contribute directly to that dietary pattern while food groups with negative loading were inversely associated with a given dietary pattern. When a food group was loaded on more than one factors/component, only the factor/component with the highest loading was considered for factor naming. The factor loading of more than 0.4 (for sample size 200–249) using BMDP statistical calculation would be considered as having contributed significantly to the factors. The communality would also be shown which indicates that a particular variable has much in common with other variables taken as a group and loaded in component or factor. Dietary pattern and risk of oral cancer In order to obtain the correlation of each factor/component retained from factor analysis and oral cancer, the 15 food groups were categorized into tertile based on the entire study population and conditional logistic regression (STATA 8 version) was then performed to obtain the risk. The univariate model was undertaken using conditional logistic regression to the confounding effect of each variable. Multivariate analysis model (adjusted odds ratio) was performed for variables which have p > 0.250 in the univariate model.17 A variable was considered to be a confounder if its inclu-
Table 1 Food group classification. Food groups
Food types
1. Grain 2. White meat 3. Red meat 4. Sea food 5. Cooked vegetables 6. Raw vegetables 7. Dairy products 8. Fruits 9. Fermented food 10. Snack 11. Canned food 12. Processed food 13. Fast food 14. Other 15. Drinks
Rice, bread, noodle, sweet potato, cassava, sago, corn Chicken, duck, bird, rabbit Beef and lamb Fish, prawn, squid Leafy green, beans, cruciferous, brinjal, pumpkin, bamboo shoots, and radish Cucumber, others, e.g., petai, jengkol Cheese, margarine, yoghurt, skim and powder milk Banana, papaya, water melon, apple, honeydew, mango, pineapple, jackfruit, guava, orange and fruit juices Salty egg, salty fish, fermented prawn, soya ketchup, fermented fruit, and pickles Fried stuff, desert, sweets and cakes Fish, tomato paste, mushroom Sausages, nugget, and meat/fish balls Chips, hamburger, pizza, instant noodle, ready made spices Monosodium glutamate/MSG Coffee, tea, and canned/carbonated drink
e51
R. Amtha et al. / Oral Oncology 45 (2009) e49–e53
sion in a model changed the OR estimation by more than 10%. The variables that fulfill these criteria as confounders are kept in the models presented with OR estimated by method of maximum likelihood. The 95% confidence intervals (95% CI) were based on the standard error of coefficient estimated and p-value for trend was obtained. Results The sociodemographic characteristics of subjects are as presented in Table 2. This study showed the KMO of 0.671 (more than 0.600) and the Bartlett’s test of sphericity of 0.000 (less than 0.05), making the sample of this study adequate for factor analysis. Four principle factors/components were retained through factor analysis (Table 3), based on KMO and Scree plot where the eigen value was more than 1 (Fig. 1). The first factor/component, which accounted for 27% of the total variance, is labeled as ‘‘preferred”. Fast food, fermented food, canned food, snacks high in fat and sugar, cooked and raw vegetables, and seafood falls into this comTable 2 Sociodemographic characteristic of subjects. Cases
(N = 162)
(N = 81)
N Age group 22–34 35–49 >49 Gender Male Female Marital status Married Single Divorce/widow Risk habits No habit Smoking Alcohol Betel quid chewing Smoking and alcohol Smoking and betel quid Smoking, alcohol, betel quid
5
%
N
%
24 69 69
14.8 42.6 42.6
16 30 35
19.8 37.0 43.2
100 62
61.7 38.3
50 31
61.7 38.3
134 15 13
82.7 9.3 8.02
73 5 3
90.1 6.2 3.07
34 30 1 2 10 3 1
42 37 1.2 2.5 12.3 3.7 1.2
79 65 1 2 14 0 1
48.8 40.1 0.6 1.2 8.6 0 0.6
4
Eigenvalue
Control
ponent. The second factor contributed 11% of the total variance. This factor was loaded by the intake of dairy product, red meat, white meat and fruit labeled ‘‘combination”. The third factor accounted for approximately 9% of the total variance. In this component high loading factor was observed in the processed food and MSG labeled ‘‘chemical related”. The fourth principle component which accounted for approximately 8% of the total variance was loaded by two food group, drinks and grain and this factor has been labeled ‘‘traditional”. The highest communality was shown in ‘‘Fast food” (0.818) and loaded in first factor/component labeled ‘‘preferred”. Table 4 shows that the consumption in the highest tertile of the ‘‘preferred” pattern was found to increase the risk of oral cancer by two-times (adjusted OR 2.17, 95% CI 1.05–4.50, v2 trend 5.446, p < 0.05). The intake of the highest tertile of ‘‘chemical related” pattern was found to be associated with higher risk of about threefold after adjusting for controls (adjusted OR 2.56, 95% CI 1.18–5.54, v2 trend 5.640, p < 0.05). On the other hand, the highest tertile of ‘‘combination” pattern displayed protective effects in relation to oral cancer before and after adjusting for the variables mentioned earlier (adjusted OR 0.50, 95% CI 0.24–1.00, v2 trend 7.335, p < 0.01). Consumption of highest tertile of ‘‘traditional” pattern showed an increased of risk by twofold (OR 2.04, 95% CI 1.01–
3
2
1
0 1
2
3
4
5
6
7
8
9
10 11 12 13 14 15
Component Number Figure 1 Scree plot showing eigenvalues for 15 components (food groups), in the factor extraction of data obtained from food frequency questionnaire.
Table 3 Loadings factor for the four components identified from factor analysis. Food group
Fast food Cooked vegetables Raw vegetables Fermented food Seafood Canned food Snack high fat and sugar Red meat Dairy product Fruits White meat Processed food MSG Drinks Grain Eigenvalue % Explained variance % Cumulative variance
Dietary pattern
Communality
Preferred food
Combination
Chemical related
Traditional
0.749 0.737 0.709 0.690 0.621 0.572 0.483 0.098 0.084 0.522 0.115 0.015 0.118 0.027 0.418 4.090 27.264 27.264
0.018 0.155 0.015 0.040 0.250 0.082 0.307 0.730 0.722 0.584 0.545 0.082 0.156 0.028 0.077 1.676 11.174 38.438
0.463 0.055 0.152 0.450 0.205 0.260 0.402 0.212 0.149 0.175 0.395 0.588 0.467 0.095 0.127 1.376 9.175 47.613
0.206 0.144 0.039 0.013 0.028 0.261 0.285 0.065 0.173 0.027 0.204 0.040 0.451 0.723 0.518 1.144 7.624 55.237
0.818 0.591 0.528 0.681 0.491 0.469 0.570 0.591 0.580 0.645 0.508 0.354 0.460 0.534 0.465
e52
R. Amtha et al. / Oral Oncology 45 (2009) e49–e53
Table 4 Risk of oral cancer in approximate tertile for food groups defined by factor analysis. Factor Preferred
a
Combinationb
Chemical relatedb
Traditionalb
Tertile of score
Control:cases
Crude OR (95% CI)
Adjusted OR (95% CI)
v2 p trend
1st (5.1) 2nd (8.0) 3rd 1st (1.7) 2nd (3.0) 3rd 1st (2.0) 2nd (2.0) 3rd 1st (4.0) 2nd (5.0) 3rd
58:23 58:23 58:46 42:39 61:19 59:23 67:22 54:24 41:35 60:22 57:23 45:36
1.00 1.04 1.89 1.00 0.35 0.42 1.00 1.43 2.56 1.00 1.15 2.09
1.00 1.05 2.17 1.00 0.36 0.46 1.00 1.48 2.85 1.00 1.22 2.10
5.446*
(0.52–2.07) (0.98–3.64) (0.17–0.70) (0.22–0.81) (0.67–3.07) (1.18–5.54) (0.57–2.26) (1.07–4.06)
(0.53–2.28) (1.05–4.50) 7.335** (0.18–0.75) (0.23–0.91) 5.640* (0.67–3.28) (1.34–6.05) 5.649* (0.59–2.53) (1.02–4.30)
a
Adjusted for smoking status, smoking duration, type of tobacco, number stick of cigarettes, pack-years. Adjusted for smoking, alcohol and betel quid. p < 0.05. ** p < 0.01. b
*
4.41, v2 trend 5.649, p < 0.05) after allowing for ethnic, and dietary intake habit. Discussion In the present study, the dietary pattern was explored in oral cancer subjects from five hospitals in Jakarta, Indonesia and matched controls. Four components named preferred, combination, chemical related and traditional which explained 55% of the variability within the samples was identified. The factors/components identified were then used as co-variables in determining whether the effect of specific nutrient is independent of the dietary pattern. The first factor/component identified as ‘‘preferred food” characterized by the consumption of fast food, cooked and raw vegetables, fermented food, seafood, canned food, and snack high in fat and sugar, was associated with doubling of risk of oral cancer. According to the adjusted risk estimate, it was found that subjects with scores in the highest tertile of the ‘‘preferred food” pattern had greater risk of oral cancer than subjects with lower scores. This finding is consistent with many studies which found that fast food, fermented, canned food and meals that contain high fat and sugar may increase the risk of cancers and other disease such as hypertension and diabetes.5 Fast food (chips, hamburger, pizza, ready made spices and especially instant noodle) represented the greatest communalities in this study, which mean that this food type has much in common with other variables taken as a group and loaded in component which is labeled ‘‘preferred food”. A plausible explanation of the contribution of these foods to oral cancer is the fact that fast food mainly composed of high levels of fat which can produce PAHs during curing (food preparation) using high temperature. The PAHs found in the food have been shown to cause cancer in laboratory animals.11 In addition, the fermented, canned food or processed food in the ‘‘preferred” component may also increase the risk of cancer.5 The world cancer report had also noted that the consumption of salted food as well as preserved and canned food (rich in nitrate performed N-nitroso compounds) is associated with cancer especially gastric cancer. Several biological mechanisms have also been proposed to explain the association between Chinese-style salted fish and nasopharyngeal cancer, including partial fermentation and nitrosamine formation.5 However, the first factor/component also consisted of vegetables (cooked and raw) thus contributing to the causal relationship, although the mechanism of vegetables in increasing the risk of oral cancer is not fully understood. In addition, factor analysis also indicates that single factor alone may not account for disease causation or prevention as other factors such as cultural, social and demo-
graphic characteristics in the same component will interact interchangeably. Considering the varied style of cooking vegetables in Indonesia, there are some plausible reasons for the observation which contradict the previous literature. Astawan in his study found that there are four common styles of cooking vegetables in Indonesia.18 Firstly, the habit of washing the vegetables after being cut, leading to wider surface area of vegetables being exposed to air and water, may result in depletion of some mineral and vitamins (especially C and B). Secondly, the habit of cooking the vegetables with an abundance of water resulting in dissolution of the majority of vitamins, minerals and amino acids11 in the soup and the remaining soup was left uneaten. The third is the habit of boiling the vegetables with water together before the water temperature reaches the boiling point. This may lead to depletion of some of the active constituent of vegetables. The fourth habit is that of overcooking the vegetables as well as reheating them many times. All the above styles may contribute to decreasing the quality of cooked vegetables in Indonesia and if they interact with other risk factors may increase the risk of oral cancer. The second factor/component named ‘‘combination” characterized by meat, dairy product and fruit showed inverse relationship to oral cancer. Fruit and vegetables have been consistently proven as a protective food type in many cancers. In regards to oral cancer, this finding is supported by many studies6,8,10,19,20 who found the strongest protective effects derived from citrus fruit and others which are likelier to be eaten raw, thus pointing to a mechanical cleansing effect of raw fruit and vegetables on the oral cavity and/ or to the beneficial effect of some temperature-sensitive substance(s), such as vitamin C. Meat and dairy product being also constituents of the combination factor would have contributed to the protective effect of this factor and thus would be opposite to some other studies. This other studies conducted in European countries had shown that meat, cheese and milk increased the risk of oral cancer.8,19 However, a study conducted by Franceschi et al. found that major sources of protein and fats in Italian diet (milk, eggs, oil and cheese) and some non-alcoholic beverages did not seem to have a significant influence on development of oral cancer.21 Similarly, Zeng et al. did a study in Beijing, and found that protein and fat intake are related inversely with oral cancer risk.6 Drinking milk showed a moderate inverse association with oral cancer risk with marginally significant OR of 0.5 (95% CI 0.2–1.0). One possibility for the differing finding between this and other European studies is that the food preparation in European food might be different with Asian, especially Indonesian. The most common meat preparation method recorded in this study was by boiling (such as soup, gulai and rendang). Whereas the majority of meat preparations in the European countries such as meat, requires it to be grilled, baked or
R. Amtha et al. / Oral Oncology 45 (2009) e49–e53
roasted and sometimes deep fried. As mentioned earlier the PAHs are carcinogenic and can be found in great amount in food prepared in this manner. Grilled meats are high in PAHs which is formed when the fat drips onto the flame below and it attaches on the meat surface. The broiled and deep fried meats also contain other hazardous agent heterocyclic amines (HAs) such as benzopyrene formed from the burning of amino acids and other substances in the meat under very high temperature.11 This is supported by studies which found the positive associations of oral cancer with intake of nitriterich meat,19 salted meat6 and charchoal-grilled meat.22 However, the findings that meats as part of the ‘‘combination” factor is protective may be explain by the fact that protein which contain of essential amino acid is the basic needs of the body in cell building and defense (immune) mechanism.11 The highest tertile scores of the third food pattern categorized as ‘‘chemical related” food significantly increase the risk of oral cancer. This third component was loaded by processed food and MSG. The chemical contained in this food group has been emphasized as carcinogenic agent to cancer. However, this still need further investigation as some review stated that monosodium glutamate (MSG) is the amino acid which can be metabolized automatically by the body.11 The fourth component which also increased the risk of oral cancer in this study is categorized under ‘‘traditional”. It is characterized by all types of drink (soft drink, canned drink, coffee and tea) and staple food (Indonesia majority rice) which produce high calories. This finding is also supported by Franceschi et al. who found an association between drinks and oral cavity and pharynx cancers.8 The OR of the highest tertile of daily intake of soft drink was 1.6 (95% CI 1.3–2.2) compared to the lowest intake. Carbohydrate intake has also been shown to moderately increase the risk of oral cancer in the study conducted in Beijing.6 In this study, the food items were grouped by the similarity in terms of composition and nutrient value, based on previous studies.6,14,21,23 The four retained factors explain more than half of the total variance. If the patterns fail to explain much of the variance in the food intake as a whole, it is possible that these patterns would not explain much of the variance in a single food or nutrient either, thus limiting their use in nutritional epidemiology.24 Studies in dietary pattern in Indonesia are scarce, and to the best of our knowledge, this is the first study to utilize principal component (factor) analysis in the context of oral cancer. Although the pattern observed in this study provides an indication of the common food consumed the data should not be over-interpreted because of the limitation in sample size. Furthermore, dietary patterns were extracted from data obtained within the study population, therefore, the results cannot be generalized for the Indonesian populations with different dietary habits. Nevertheless, based on these findings, and as a start for the Indonesian population, dietary guides with emphasis on foods and on overall dietary patterns which are health promoting or damaging to the health of the population may be issued. In conclusion this study had shown that meaningful association between dietary pattern and oral cancer was possible using factor analysis, the outcome of which would be used to determine the association with oral cancer. Further research is suggested on the effects of micro- and macro-nutrient composition of food as well as the calories intake and in establishing its link to oral cancer, so that it will be possible to develop nutritional intervention based on these patterns in Indonesia. Conflict of Interest Statement This project has no conflict of interest with any institution that supported my project.
e53
Acknowledgement This study was supported by research grants from Trisakti University, Indonesia (no. 0769/Usakti/bsdm/db/10/2004) and University of Malaya (F0150/2005A). We thank Dr. Nugroho Abikusno (Medical Faculty University of Trisakti, Indonesia) and Dr. Cheong Sok Ching (Cancer Research Initiative Foundation, Kuala Lumpur) for their advice. References 1. Parkin MD, Bray F, Ferlay J, Pisani P. Global cancer statistic, 2002. CA A Cancer J Clin 2005;55:74–108. 2. De Stefani E, Deneo-Pellegrini1 H, Ronco AL, Boffetta P, Brennan P, Mun~oz N Castellsaque X, et al. Food groups and risk of squamous cell carcinoma of the oesophagus: a case-control study in Uruguay. Br J Cancer 2003;89: 1209–14. 3. Ronco AL, De Stefani E, Boffetta P, Deneo-Pellegrini H, Acosta G, Mendilaharsu M. Food patterns and risk of breast cancer: a factor analysis study in Uruguay. Int J Cancer 2006;119(7):1672–8. 4. American Institute for Cancer Research, World Cancer Research Fund. Food, nutrition, and the prevention of cancer: a global perspective. Washington, DC: The American Institute for Cancer Research; 1997. 5. Stewart BW, Kleihues P. In: Stewart Bernard W, Kleihues Paul, editors. The cause of cancer in world cancer report. Lyon, France: IARC Press; 2003. p. 22–31. 6. Zheng T, Boyle P, Walter WC, Hu H, Dan J, Evstifeeva TV, et al. A case-control study of oral cancer in Beijing, People’s Republic of China. Association with nutrient intakes, foods and food groups. Oral Oncol Eur J Cancer 1993;9B(1):45–55. 7. Takezaki T, Hirose K, Inoue M, Hamajima N, Kuroishi T, Nakamura S, et al. Tobacco, alcohol and dietary factors associated with the risk of oral cancer among Japanese. Jpn J Cancer Res 1996;87:555–62. 8. Franceschi S, Favero A, Conti E, Talamini R, Volpe R, Negri E, et al. Food groups, oils and butter and cancer of the oral cavity and pharynx. Br J Cancer 1999;80(3–4):614–20. 9. Voorrips LE, Goldbohm A, van Poppel G, Sturmans F, Hermus RJJ, van den Brandt PA. Vegetable and fruit consumption and risks of colon and rectal cancer in a prospective cohort study. The Netherlands cohort study on diet and cancer. Am J Epidemiol 2000;152:1081–92. 10. Soler M, Bosetti C, Franceschi S, Negri E, Zambon P, Talamini R, et al. Fiber intake and the risk of oral, pharyngeal and esophageal cancer. Int J Cancer 2001;91:283–7. 11. Grrosvenor MB, Smolin LA. Nutrition. From science to life. USA: Harcourt College; 2002.. p. 664–725. 12. Cappuccio FP, Rink E, Perkins-Porras L, McKay C, Hilton S, Steptoe A. Estimation of fruit and vegetable intake using a two-item dietary questionnaire: a potential tool for primary health care workers. Nutr Metab Cardivasc Dis 2003;13:12–9. 13. Marchioni DML, de Oliveira Latorre MRD, Eluf-Neto J, Wunsch-Filho V, Fishberg RM. Identification of dietary patterns using factor analysis in an epidemiological study in Sao Paulo. San Paulo Med J 2005;123:124–7. 14. Marchioni DML, Fishberg RM, de Gois Filho, Kowalski Lp, de Carvalho MB, Abrahao M, et al. Dietary patterns and risk of oral cancer: a case-control study in Sao Paulo, Brazil. Rev Saude Publica 2007;41:19–26. 15. Trichopoulous D, Lagiou P. Dietary patterns and mortality. Br J Nutr 2001;85(2):133–4. 16. Pallant J. Factor analysisSPSS survival manual. Sydney: Allen & Unwin; 2005. p. 172–193. 17. Schlesselmen JJ. Case control studies. Design, conduct, analysis. New York: Oxford University Press; 1982. 18. Astawan M. Kandungan gizi aneka bahan makanan. Jakarta: Gramedia; 2004. 19. McLaughlin JK, Gridley G, Block G, Winn DM, Preston-Martin S, Scoenberg JB, et al. Dietary factors in oral and pharyngeal cancer. J Nat Cancer Inst 1988;80:1237–43. 20. Franceschi S, Bidoli E, Barón AE, Barra S, Talamini R, Serraino D, et al. Nutrition and cancer of the oral cavity and pharynx in north-east Italy. Int J Cancer 1991;47(1):20–5. 21. Franceschi S, Barra S, La Vecchia C, Bidoli E, Negri E, Talamini R. Risk factors for cancer of the tongue and the mouth. Cancer 1992;70(9):2227–33. 22. Franco EL, Kowalski LP, Oliveira BV, Curado MP, Pereira RN, Silva ME, et al. Risk factors for oral cancer in Brazil: a case-control study. Int J Cancer 1989;43:992–1000. 23. Franceschi S, Levi F, Conti E, Talamini R, Negri E, Dal Maso L, et al. Energy intake and dietary pattern in cancer of the oral cavity and pharynx. Cancer Causes Control 1999;10:439–44. 24. Schulze MB, Hoffman K, Kroke K, Boieng H. Dietary patterns and their association with food and nutrient intake in the European prospective investigation into cancer and nutrition (EPIC) – postdam study. Br J Nutr 2001;85:363–73.