Correlation of Cancer Incidence with Diet, Smoking and Socio- Economic Position Across 22 Districts of Tehran in 2008

Variation in incidence of cancer in different geographical locations is partly due to different risk factors that are associated with lifestyle (Parkin et al., 2009; de Vries et al., 2010). Many factors have been identified to be involved in the etiology of cancer such as tobacco, alcohol, occupation, air pollution, water contaminations, diet and nutrition, obesity, physical activity, infectious agents, and solar radiation, which are all probable factors (Schottenfeld and Fraumeni, 2006). Principal risk factors for cancer in western countries consist of smoking, dietary patterns, and reproductive behaviors and in developing countries is infectious agents, but these patterns are changing. For example smoking is increasing in some developing countries (Jemal et al., 2010). Nutritional factors contribute about 35% of cancer cases (Doll and Peto, 1981). The lower incidence of total cancer in Mediterranean countries in comparison with Scandinavian countries, the United Kingdom, and the United States is mostly because of lower incidence cancer of the large bowel, breast, endometrium, and


Introduction
Variation in incidence of cancer in different geographical locations is partly due to different risk factors that are associated with lifestyle de Vries et al., 2010). Many factors have been identified to be involved in the etiology of cancer such as tobacco, alcohol, occupation, air pollution, water contaminations, diet and nutrition, obesity, physical activity, infectious agents, and solar radiation, which are all probable factors (Schottenfeld and Fraumeni, 2006). Principal risk factors for cancer in western countries consist of smoking, dietary patterns, and reproductive behaviors and in developing countries is infectious agents, but these patterns are changing. For example smoking is increasing in some developing countries (Jemal et al., 2010).
Nutritional factors contribute about 35% of cancer cases (Doll and Peto, 1981). The lower incidence of total cancer in Mediterranean countries in comparison with Scandinavian countries, the United Kingdom, and the United States is mostly because of lower incidence cancer of the large bowel, breast, endometrium, and 1 prostate that could be attributable to dietary factors across Mediterranean countries. If the Western people could switch to the Mediterranean diet, some cancers would be decreased (Trichopoulou et al., 2000). The increased colorectal cancer incidence rates in some countries in Asia, Eastern Europe and Spain are associated with changes in dietary and lifestyle factors (Garcia-Alvarez et al., 2007;de Kok et al., 2008;Martin et al., 2008;Center et al., 2009a;2009b).
Many epidemiological studies have been conducted to investigate the role of dietary pattern and various cancers (Trichopoulou et al., 2000;Prentice et al., 2007;Ozasa et al., 2001;Riboli and Norat, 2001;Mohebbi et al., 2011). Meat and processed meat for colorectal malignancy, arsenic for lung and skin cancers, beta carotene supplements for lung cancer, and aflatoxin for liver cancer have been identified in various studies (2007). Convincing evidence support the decreasing effects of vegetables and fruits for mouth, pharynx, larynx, esophagus, stomach and lung cancers, while vegetable alone decrease the risk of colorectal cancer. Red and particularly processed meat increase the risk of colorectal cancer, and processed meat and salt are probable risk factor for stomach cancer (Riboli and Norat, 2001). In a systematic review more studies found no significant relationship between vegetable and fruit intake separately and colorectal cancer risk but the combined consumption of vegetables and fruits reduced the risk in half of the studies. Marques-Vidal reported that relationship between dietary intake and colorectal cancer might be less important than previously reported (Marques-Vidal et al., 2006). There are case control studies suggesting the relation between high consumption of fiber and decreased risk of breast cancer, (Baghurst and Rohan, 1994;La Vecchia et al., 1997) however these finding are not confirmed in longitudinal studies (Verhoeven et al., 1997;Willett et al., 1992).
Review articles point out to conflicting epidemiological studies, which in one hand do not support the overall well known hypotheses such as lack of protective effect of a high fruit-vegetable diet and on the other hand, emphasizing the hazardous effect of fat-intakes in breast cancer (Key et al., 2003;Hanf and Gonder, 2005). A metaanalysis on the association between breast cancer risk and red meat consumption in premenopausal women reflected a summary relative risk as 1.24 (95%CI 1.08-1.42) (Taylor et al., 2009).
Various studies have indicated that incidence of cancer varies across socio-economic position (SEP) groups and health disparities exist between and within countries even within cities (Yiengprugsawan et al., 2007;IAEA, 2011). Social, economical, environmental and political factors have been apparently known as the stem of inequalities in health. Equality of breast cancer incidence in high risk immigrants with the destination society suggests that environmental rather than genetic risk factors play a role of causality (Kelsey and Horn-Ross, 1993;Wu et al., 1996). Understanding the role of socioeconomic determinant of health (SDH) helps implement appropriate interventions to reduce disparities. Socioeconomic characteristics such as income, wealth, education, occupation, social class, and insurance coverage are potential indicators for comparison of different groups based on health and social features (Braveman et al., 2001;Yiengprugsawan et al., 2007;Natale-Pereira et al., 2011). The proportion of cancer incidence across SEP is dissimilar by cancer types; while some cancers are more common in higher socioeconomic positions such as prostate and breast cancers, a variety of cancers prevail in lower SEP groups such as lung and cervical cancers (Yin et al., 2010). Investigation of correlation between SEP and cancer incidence may result in detection of disparities in cancer and particularly socio-economic groups with high burden of cancer. The aim of this study is to investigate correlation of common cancers with dietary groups, total energy, smoking, and SEP.

Materials and Methods
In this ecological study various sources of information were included population data for all 22 districts of Tehran which was obtained from 2006 census of Iran Statistical Center, cancer data from National Cancer Registry in 2008, (Rohani et al., 2011) and dietary pattern from Urban HEART-1 study (Asadi-Lari et al., 2010). Home addresses and phones were extracted from cancer registry database and cancer patients' districts of residence were identified. Age standardized rates (ASR) per 100,000 men and women of Tehran were calculated using the direct method of standardization to new (2000) WHO World Standards (Ahmad et al., 2009). ASR for common cancer types including prostate, colorectal, stomach, skin and bladder cancer in men and breast, colorectal, skin, stomach and ovarian cancer in women were estimated across 22 districts.
Dietary intake, smoking habits and SEP were obtained through Urban Health Equity Assessment and Response Tool (Urban-HEART) project which was conducted in 2008, (Asadi-Lari et al., 2010) where almost 960 households in each 22 districts of Tehran and overall 21,120 households were approached. A smaller sample including 10% of Urban-HEART sample size was investigated for food consumption. A 24-hour food recall questionnaire which referred to the household food consumption during the preceded 24-hour, was completed through a detailed interview usually with the housewife. A specific software was used to transform the food stuff into nutrition categories and energy intake, then the distribution of food groups consumption (mean g/day) across 22 districts were determined. Food groups consisted of bread, grain, vegetable, fruit, meat, egg, dairy products, fat, pastries, drinks and nuts (Asadi-Lari et al., 2010).
To produce SEP groups, a proxy value was calculated for all individuals using principal components analysis (PCA); then a median score for each district was used. Assets and house features consisting of house ownership, room per person, area per capita, having independent bath, kitchen, toilet, owning car, landline, cell phone, freezer, and computer for the household, and the years of education were calculated for individuals over 6 years old (Rohani-Rasaf et al., 2012). Districts were accordingly sorted by these values then categorized into four classes based on population of districts to make a similar population in each class.
We investigated correlation between cancer incidence and food groups, SEP scores and smoking using Pearson correlation; partial correlation was used for better interpretation in smoking levels. To examine the contribution of SEP in common cancers, R square was calculated using regression model. A p-value less than 0.05 was considered as significant.

Results
Although no large variation of the mean of consumption exists between districts we observed lower consumption of bread, fats and vegetables and higher consumption of meat, dairy products and fruit in districts with higher SEP (Table 1).
Districts with less-advantaged socioeconomic groups (1) to more-advantaged (4) are listed in Table 1. While smoking in men was positively correlated with SEP (r=0.69 p<0.01), which indicates that smoking rate was higher in affluent than less-advantaged districts, smoking in women had reverse correlation with SEP (r=-0.6 p<0.01). All common cancers in men and women were positively correlated with SEP, which means that cancer incidence increased with departure from less-advantaged to affluent districts except skin and stomach in both genders which showed no significant correlation with SEP.
Stomach cancer in men was higher in more disadvantaged districts in southern Tehran. Higher incidence of breast and colorectal cancers in both sexes were observed in the highest SEP group. Also prostate, colorectal and bladder cancers in men were positively correlated with smoking   rate indicating that higher incidence of cancers prevailed in districts with higher rates of smoking and colorectal cancer in women had reverse correlation with smoking rate at district level. In regression analysis, the standardized coefficient for prostate and breast cancers as depended variables and SEP as predictor were 0.83 and 0.81 respectively. B for colorectal cancers in men and women were 0.71 and 0.69 while stomach cancer in women had a moderate correlation with SEP. No correlation was found between skin cancer in both genders and stomach cancer in men with SEP (Table 3). Fruit, meat and dairy products adjusted by energy were positively correlated with bladder, colorectal, prostate and breast and total cancers in men and women, while these cancers were adversely correlated with bread and fat consumption. Meat consumption was positively correlated with stomach cancer in women and with skin cancer in men (Table 4). After adjustment for SEP, only bladder cancer was correlated with meat (r=0.47) and colorectal cancer with dairy (r=0.5) in men also colorectal cancer was negatively correlated with bread in both sexes.

Discussion
This study showed the distribution of common cancers and their ecological correlation with smoking, food groups and SEP across 22 districts of Tehran in 2008. Food groups and cancers had strong correlation with SEP in Tehran. Breast, prostate, colorectal and ovary cancers clearly were high in affluent districts. It has shown that lower consumption of breads, fats and vegetable and higher consumption of meat, dairy products and fruit and higher smoking significantly was seen in affluent districts.
WHO estimates that common cancer incidence (39 per 100,000 women) and the second leading cause of cancer-related deaths (12.5) was related to breast cancer among women in world in 2008 (GLOBOCAN IARC, 2010) There are an increasing trend in breast cancer (BC) incidence in Iran over the past two decades; according to a population based cancer registry in 5 provinces during 1996-2000 the incidence rate was 16.2 per 100,000 population (Sadjadi et al., 2009) while BC incidence in Tehran was 22.4 in 1998, (Harirchi et al., 2005) (Mohagheghi et al., 2009), and36 in 2008 (current study). Studies inconsistently suggest that fruits and vegetables, low-fat dairy products, fish, monounsaturated and polyunsaturated fatty acids, may reduce the risk of breast cancer, whilst, high intake of meat, poultry, total energy, total fat and saturated fatty acids probably increase the risk (Koo et al., 1997;Holmes et al., 2003;Key et al., 2007;Bissonauth et al., 2008). In this study breast cancer was significantly and positively correlated with meat, dairy and fruit consumption and inversely correlated with bread after adjusting for SEP. In the Netherland cancer incidence low SEP groups were replaced by high SEP groups for increased prostate, breast cancers and BCC so that had notably trend during 1996-2008 (Aarts et al., 2010).
A systematic review on cancer incidence across different SEP groups in 21 countries showed that higher risk of breast cancer in women was seen with high SEP ) and women with lower education level had lower risk of breast cancer, (Spadea et al., 2009) both of which are consistent with our findings, where breast cancer was more seen in affluent districts in higher SEP groups (R=0.82) and have the most strong B between women cancers.
Second common cancer incidence in men is prostate cancer (GLOBOCAN IARC, 2010). Incidence of prostate cancer in this study was higher than estimation of WHO for Iran in 2008 (19.5 per 100,000 v.s 11.6). Many risk factors such as aging, race, family history, gonadal steroids and Insulin-like growth factor I (IGF-I), processed red meat, dairy products and fat (as provoking factors) (Koo et al., 1997;Ngo et al., 2003;Tseng et al., 2004;Schottenfeld and Fraumeni, 2006;Grant, 2010) and cereals, nuts, oilseeds, fish, (Hebert et al., 1998) vitamin E (Schottenfeld and Fraumeni, 2006) soy products (Hebert et al., 1998) and cholesterol (Hu et al., 2012) (as protective factors) have been identified for prostate cancer; in our study dairy products and meat correlated positively and bread and fat negatively with prostate cancer. In a prospective study, prostate cancer was not associated with the vegetablefruit or red meat-starch pattern, but higher intake of the Southern pattern characterized by cornbread, grits, sweet potatoes, okra, beans, and rice showed a significant reduced risk (Tseng et al., 2004). On the other hand, in districts with low education level, prostate cancer was less common, which is in line with other studies (Steenland et al., 2004;Spadea et al., 2009) and might be due to more PSA tests, early detection or more access to healthcare services.
Colorectal cancer (CRC) was estimated at global level in 2008 to be the second and third common cancer in women and men respectively (GLOBOCAN IARC, 2010). CRC incidence in men (16.4) and women (12.2) were higher than WHO estimation for Iran, while CRC incidence for Tehrani men was higher than neighboring countries such as Pakistan and Turkey (IAEA, February 2011, Jemal et al., 2011. Trichopoulou proposed that approximately up to 25% of the incidence of CRC, could be prevented if the population diet could transfer to the Mediterranean diet (Trichopoulou et al., 2000). The combined consumption of vegetables and fruit, high intake of fiber, fish, unsaturated fat, vitamins and calcium decrease the risk and high consumption of red or processed meat, smoked or salted fish actually appears to be harmful, which increase CRC risk (Jacobs et al., 2003;Marques-Vidal et al., 2006;Bravi et al., 2010;Gonzalez and Riboli, 2010;Randi et al., 2010;Chan et al., 2011). Our results indicate that CRC was correlated with meat, dairy products and fruits adjusted for total energy and reverse correlation with bread and fat. CRC was more common in higher SEP groups in 21 countries in 1997,  which was consistent with our findings with a correlation coefficient of 0.69-0.71. Higher CRC incidence among high SEP group recommended a higher prevalence of risk factor such as smoking in this group in Tehran but in 2012 the overall incidence of CRC after controlling for other risk factors was significantly higher among low educational level or low-SEP groups where prevalence of adverse health behaviors along with BMI explained 36% of the association of local SEP with risk of colorectal cancer (Doubeni et al., 2012a;2012b). Evidence indicates a upward slope between CRC incidence and socioeconomic deprivation in recent decade which SEP pattern in CRC has changed direction (Kee et al., 1996;Pollock and Vickers, 1997;Oliphant et al., 2011).
Skin cancer is the most common cancers in Iran, notably that basal cell carcinoma (BCC) was more frequent in the current study. It is suggested that fat intake, residence in high solar radiation and sun exposure are associated with an increased risk of BCC (van Dam et al., 1999) while monounsaturated fat consumption is associated with a lower risk (van Dam et al., 2000). In a prospective study, high meat and fat intake increased SCC particularly in patients with a family history, (Ibiebele et al., 2007) which was in line with our findings only in men. In a multinational study across 21 countries excessive risk of melanoma was observed in higher SEP,  and in another study this association was attributed to education, (Harrison et al., 1998) contrary to our findings which indicated prevailed skin cancer in district 20 with lower SEP.
Stomach cancer incidence in Iranian men was estimated higher than the world statistics in 2008 (GLOBOCAN IARC, 2010), while our findings depict a lower incidence than WHO estimates, regarding the fact that our data are based on pathology reports which may underestimate the highly fatal stomach cancer (Zendehdel et al., 2010). Various risk factors have been recognized such as helicobacter pylori (H pylori) infection, cured meat, smoking, and alcohol while Mediterranean diet, vegetables and fruits, vitamin C, retinol and high intake of cereal fiber decrease the risk of stomach cancer (Kolonel et al., 1981;Demirer et al., 1990;Schottenfeld and Fraumeni, 2006;Wang, 2009;Gonzalez and Riboli, 2010). An ecologic analysis in Brazil suggested that differences in living standards among populations probably have an important role in incidence differences (Koifman and Koifman, 1997). The high level of education in men and higher SEP were associated with lower stomach cancer risk and for education exist even after age adjustment (RR highest to lowest level=0.54) which was diluted by variables such as smoking, intake of vitamin C, beta carotene, alcohol, coffee and family history of stomach cancer van Loon et al., 1998). Our results showed that stomach cancer only in women was correlated with meat and correlation of stomach cancer and SEP in both sexes was inconsistent so that higher incidence in men was in low SEP districts, while in women it prevailed in high SEP districts.
Ovarian cancer incidence in Iran is lower than similar world statistics and much lower than USA in all age groups, (Arab et al., 2010;GLOBOCAN IARC, 2010) but it is one of the most common cancers in women in Tehran. Consumption of eggs, lactose intake and dairy foods increase the risk of ovarian cancer and inconsistent studies suggested that fruits and vegetables intake and plant-based diet may decrease the risk (Kushi et al., 1999;McCann et al., 2003;Larsson et al., 2006). In the current study there was a positive correlation between meat and dairy products and ovarian cancer and a reverse correlation with fat intake. SEP explained 32% of ovarian cancer, which justifies a positive and strong inference, likewise other studies in the literature which propose higher SEP as a predictor of ovarian cancer Liu et al., 1998).
Bladder cancer incidence in Iranian men (11.2) has been estimated higher than the similar figures in Eastern Mediterranean Region and the global incidence. (GLOBOCAN IARC, 2010) Studies suggest that pork, barbecued meats, fat, butter, soy, eggs and excessive coffee consumption are associated with increased risk (Radosavljevic et al., 2005;Keszei et al., 2010;Silberstein and Parsons, 2010) and vegetables, fruits, and cereal are associated with decreased risk of bladder cancer. Bladder cancer incidence is correlated with SEP (Buntinx et al., 2003), while when adjusted for smoking the relative risk of bladder cancer has been decreased (Goy et al., 2008). This study, nevertheless showed that consuming meat, dairy products, and fruit and SEP were related to high incidence of bladder cancer while districts with higher intake of bread and fat were correlated with reduced bladder cancer.
Various limitations existed in this study; first ecologic fallacy was the main obstacle which may bias the inferences and any generalized conclusion should be therefore avoided. As explained in the method, two sources of data were compared in this study, one came from a large population based household survey, which sought for inequalities in social determinants of health (Urban HEART) (Asadi-Lari et al., 2010) and one arose from mandatory cancer registry. Amalgamating these two sets of information may deviate the inferences due to the fact that there is no evidence of dietary pattern of cancer patients. Second, diet has a long term effect on cancer which has been estimated between 5-20 years; (Colditz et al., 2012) therefore correlating two sets of data at the same year may also deviate the conclusions. Our main assumption, for this inference, was unremarkable changes in dietary pattern at least in short-term, which is the least incubation period for cancer. Third, using a single 24-hour-recall may inherently jeopardize the inferences, however is not an problem if the goal is to estimate a mean intake for a population this method will introduce no bias (Rothman et al., 2008). Fourth, cancer registry in Tehran is based on pathology reports, which inherently faces with criticisms that stigmatize the data as incomprehensive and incomplete particularly for stomach cancer with high fatality rate, although the national coverage of cancer registry is estimated at 81% in 2008 (Rohani et al., 2011).
In conclusion, there is a wide range of age standard cancer incidence in 22 districts of Tehran which is correlated to SEP and diet. Socioeconomic patterns in incidence varied for specific cancers. Higher SEP was associated with higher rates of breast, CRC, prostate, bladder and ovarian cancers. There was lower consumption of bread, fat and vegetables and higher consumption of meat, dairy products and fruit in affluent districts. Consumption of fruit, meat and dairy products adjusted by energy were positively correlated with bladder, colorectal, prostate and breast, while these cancers were adversely correlated with bread and fat intake. However, majority of associations with food groups disappeared after adjustment for SEP. Whether what trait in SEP groups leads to clustering warrants further research to include individual data.