Italian mediterranean index and risk of colorectal cancer in the Italian section of the EPIC cohort



Colorectal cancer is among the commonest cancers worldwide. Dietary factors have been linked to colorectal cancer risk, however, few studies have evaluated the relationship between a priori dietary patterns and colorectal cancer risk. We evaluated the effect of adherence to a Mediterranean dietary pattern, as measured by the Italian Mediterranean Index, on the risk of colorectal cancer in the 45,275 participants of the Italian section of the EPIC study who completed a dietary questionnaire. Hazard ratios (HRs) with 95% confidence intervals (CIs) for colorectal cancer in relation to categories of Italian Mediterranean Index score were estimated by multivariate Cox models adjusted for known risk factors, on the whole cohort, on men and women and according to cancer subsite. During a mean follow-up of 11.28 years, 435 colorectal cancer cases were identified. The Italian Mediterranean Index was inversely associated with colorectal cancer risk (HR: 0.50; 95% CI: 0.35–0.71 for the highest category compared to the lowest, P-trend: 0.043). Results did not differ by sex. Highest Italian Mediterranean Index score was also significantly associated with reduced risks of any colon cancer (HR: 0.54, 95% CI: 0.36–0.81), distal colon cancer (HR: 0.44, 95% CI: 0.26–0.75) and rectal cancer (HR: 0.41, 95% CI: 0.20–0.81), but not of proximal colon cancer. These findings suggest that adherence to a Mediterranean diet (as measured by the Italian Mediterranean Index) protects against colorectal cancer in general but not against cancer developing in the proximal colon.

Colorectal cancer is the third most common cancer in men, second most common cancer in women and fourth most important cause of cancer death worldwide.1 Most cases are diagnosed in developed countries, and incidence is higher in men than women.1 The colorectal cancer burden in Italy appears of similar magnitude to that in other resource-rich countries.2 The comprehensive and authoritative review “Food, nutrition and the prevention of cancer: a global perspective” published by the World Cancer Research Fund in 20073 concluded that there was convincing evidence that consumption of red meat, processed meat and alcohol increase the risk of colorectal cancer, while consumption of fiber-containing foods, garlic, milk and calcium was probably protective.

However, foods and nutrients are not consumed in isolation but as part of an overall diet and it is plausible that the effects of individual dietary components on the risk of colorectal cancer depend on overall dietary pattern. In fact dietary pattern approaches are increasingly used in epidemiological studies to investigate effects of diet on disease risk.4 There are two general approaches to dietary patterns: a posteriori approaches in which the patterns are derived from statistical modeling of actual dietary data, and a priori approaches that, relying on evidence of relations between diet and a specific disease, create country or culture-specific dietary recommendations expected to be protective against the disease considered.5

Few studies have evaluated the effects of a priori dietary patterns on colorectal cancer risk. Evidence on the effects of adherence to a Mediterranean-type diet, in particular, is scant and confined to US studies.6–8 The Mediterranean diet reflects the dietary pattern prevalent in the Mediterranean European region up to 1960s9; it was first described by Ancel Keys, who found a lower rates of death and coronary heart disease in countries near the Mediterranean sea, i.e., Italy and Greece.10 Subsequent studies demonstrated a beneficial effect of this kind of diet on cancer risk.11 In our study, we examined the effect of adherence to a Mediterranean diet, as assessed by the Italian Mediterranean Index12 on the risk of colorectal cancer.

Material and Methods

Study subjects

A total of 47,021 volunteers were recruited in 1993–1998 to the Italian section of the European Prospective Investigation into Cancer and Nutrition (EPIC). Volunteers were recruited from five centers: Varese and Turin (Northern Italy), Florence (Central Italy), Naples (Southern Italy, women only) and Ragusa (Southern Italy). In our study, we examined 45,275 participants (14,195 men and 31,080 women) who completed a dietary questionnaire at enrolment, excluding those with cancer at recruitment (except non-melanoma skin cancer) (N = 426); and also those lost to follow-up at baseline; with missing information on diet, anthropometry or lifestyle; and with ratio of total energy intake (determined from the questionnaire) to basal metabolic rate at either extreme of the distribution (cut-offs first and last half-percentiles) (N = 1,320).

Baseline measurements

Dietary variables

The frequency of consumption of food items was investigated by validated13 semiquantitative food-frequency questionnaires designed to capture eating behavior in different regions of Italy: one for northern and central Italy (Varese, Turin and Florence), one for Ragusa and one for Naples. The questionnaires contained questions on 188, 217 and 140 food items, respectively, and were designed to investigate dietary habits in the 12 months before enrolment. The food items were then linked, using specifically designed software,14 to Italian Food Tables15 to obtain estimates of daily intake of energy, and 37 macro- and micro-nutrients.

We developed the Italian Mediterranean Index by adapting the Greek Mediterranean Index16, 17 to Italian eating behavior, specifically to investigate the effect of a Mediterranean diet on the risk of stroke.12 The Italian Mediterranean Index score is calculated from intake of 11 items: high intakes of six typical Mediterranean foods [pasta, typical Mediterranean vegetables (raw tomatoes, leafy vegetables, onion and garlic, salad, fruiting vegetables), fruit, legumes, olive oil and fish]; low intakes of four “non-Mediterranean” foods (soft drinks, butter, red meat and potatoes); and alcohol. If consumption of typical Mediterranean foods is in the 3rd tertile of the distribution, the person receives 1 point; all other intakes receive 0 points. If consumption of non-Mediterranean foods is in the first tertile of the distribution the person receives 1 point. Alcohol receives 1 point for intake up to 12 g/day; abstainers and persons who consume >12 g/day receive 0. Possible scores range from 0 to 11.

We also calculated the Greek Mediterranean Index16, 17 based on intake of nine items: vegetables, legumes, fruit and nuts, dairy products, cereals, meat and meat products, fish, alcohol and the ratio of monounsaturated to saturated fat. For most of these, consumption above the study median receives 1 point; all other intakes receive 0 points. For dairy products, meat and meat products, consumption below the median receives 1 point. Medians are gender-specific. For ethanol, men who consume 10–50 g/day and women who consume 5–25 g/day receive 1 point; otherwise, the score is 0. The range of possible scores is 0–9.

Lifestyle and anthropometric variables

A standardized lifestyle questionnaire, completed by each participant, collected detailed information on reproductive and medical history, physical activity, alcohol consumption, smoking, education and other socioeconomic variables. Weight and height were measured in light clothes, without shoes.

Identification of colorectal cancer cases and follow-up procedures

In Varese, Turin, Florence and Ragusa, incident cases were identified by study cohort linkage to the databases of the regional cancer registries, which are considered high quality registries with nearly complete cancer registration.18 In Naples, incident cases were identified through linkage to the regional archive of hospital discharges, and by direct telephone contact where necessary.

Colon cancers were primary incident cases, identified by the codes of the International Statistical Classification of Diseases (10th Revision) as follows: proximal (C18.0–18.5); distal (C18.6–C18.7), and overlapping or unspecified sites (C18.8 and C18.9). Rectal cancers were identified by the codes C19 (rectosigmoid junction) and C20 (rectum). Anal cancers were excluded.

Statistical analysis

Follow-up was time from enrolment to cancer diagnosis, death, loss to follow-up or end of follow-up, whichever came first. The end of follow-up varied with centre: December 31, 2006 for Varese, Naples and Florence; December 31, 2008 for Turin and Ragusa.

Multivariate Cox proportional hazard models, with age as primary time variable, were used to assess the association between colorectal cancer risk and intake of each Italian Mediterranean Index component, classified into tertiles with the lowest tertile as reference, and also overall adherence to the Italian Mediterranean Index. We fitted a minimally adjusted model, with non-alcoholic energy intake, age and sex as covariates; and a fully adjusted model, with the additional covariates body mass index (BMI, continuous), smoking status (never, former, current), total physical activity (inactive, moderately inactive, moderately active and active; entered in the model as a continuous variable) and education (≤8 years, >8 years). All models were stratified by centre. Hazard ratios (HRs) with 95% confidence interval (CI) were estimated for the following categories of Italian Mediterranean Index score: 0–1, 2–3, 4–5 and 6–11, with lowest as reference. Linearity of trends across categories was tested by treating each category as a continuous variable in the Cox model.

We ran models for the whole cohort and the following subcategories: men, women, all colon, proximal colon, distal colon and rectal cancer. We performed an additional analysis for women further adjusting for hormone replacement therapy (never/ever); 90 women were excluded from this analysis as information on this confounding variable was unavailable.

We also ran models to assess the effect of the Greek Mediterranean Index on risk of colorectal cancer, using four score categories (0–1, 2–3, 4–5, and 6–9). The analyses were performed with Stata version 11.2 (College Station, TX).


During a mean follow-up of 11.28 years, 435 cases of colorectal cancer were diagnosed in the cohort, of which 326 were colon cancer (126 proximal, 159 distal and 41 overlapping or unspecified) and 109 were rectal cancer. Turin had the highest incidence of colorectal cancer (132 cases per 123,349 person-years), followed by Florence (129 per 139,396 person-years), Varese (106 per 125,121 person-years), Ragusa (41 per 65,527 person-years) and Naples (27 per 57,149 person-years).

Table 1 shows baseline characteristics of participants according to category of Italian Mediterranean Index score. Age, BMI and non-alcohol energy intake increased with increasing category of score (increasing adherence to the Italian Mediterranean Index). Subjects in the highest Italian Mediterranean Index score category were more educated, more physically active and smoked less than those in the lowest category. Women adhered to the Italian Mediterranean Index more than men. The centre of Naples adhered to the Italian Mediterranean Index more than the other centers.

Table 1. Baseline characteristics of study participants by categories of Italian Mediterranean Index score
inline image

HRs for developing colorectal cancer according to intake of Italian Mediterranean Index components are shown in Table 2. No component was significantly associated with risk.

Table 2. Hazard ratios (HRs) for developing colorectal cancer in relation to tertiles of intake of the Italian Mediterranean Index components
inline image

Table 3 shows HRs of developing colorectal cancer according to category of Italian Mediterranean Index score in the whole cohort, and in men and women separately. Increasing adherence to the Italian Mediterranean Index was associated with a significantly decreasing risk of colorectal cancer in the whole cohort (P-trend: 0.043) with all three higher categories of score associated with significantly reduced risk of cancer compared to the lowest (reference) category, in both the minimally and fully adjusted models. For men, the linear trend was not significant (P-trend: 0.087, fully adjusted model) but the risk was significantly lower in the highest category compared to the first [HR: 0.54 (95% CI: 0.30–0.96]. For women, all three categories of score above the first (reference), were associated with a significantly lower risk (HR: 0.36 (95% CI: 0.23–0.55); 0.49 (95% CI: 0.33–0.73); 0.46 (95% CI: 0.30-0.72) compared to reference; however P-trend (0.238) was not significant. When the model was further adjusted for hormone replacement therapy, the results were substantially unchanged (data not shown in tables).

Table 3. Hazard ratios (HRs) for developing colorectal cancer in relation to adherence (increasing categories) to the Italian Mediterranean Index
inline image

Table 4 shows the results of the risk analyses for the separate subsites. The risks of all colon cancers, distal colon cancers and rectal cancers were significantly lower than reference for all categories of score above reference, although the trends were not significant [HRs for highest category compared to the first, respectively: 0.54 (95% CI: 0.36–0.81), 0.44 (95% CI: 0.26–0.75) and 0.41 (95% CI: 0.20–0.81)]. For proximal colon cancer, the risk was lower in the fourth category compared to reference, but the reduction was not significant. The numbers of cases in the subtype analysis categories were low (usually <100, Table 4), especially for proximal colon cancer.

Table 4. Hazard ratios (HRs) for developing all colon, proximal colon, distal colon and rectal cancer in relation to adherence (categories of score) to the Italian Mediterranean Index
inline image

Results for the Greek Mediterranean Index were similar to those obtained with the Italian Index, with the following differences: The test for trend for the whole cohort analysis was not significant; for women, adherence to the Greek Mediterranean Index was not associated with lowered colorectal cancer risk while in men adherence produced significant protection; and risk reductions in relation to cancer subtype (both sexes) were significant only for rectal cancer (data not shown).


We have found that adherence to a typical Italian Mediterranean diet, as measured by the Italian Mediterranean Index, is associated with reduced risk of developing colorectal cancer. The reduction in risk was significant in both sexes, but was greater in women. The reduction in risk was also significant for all colon, distal colon and rectal sites considered separately, but non for the proximal colon. No single component of the Index had a significant influence on colorectal cancer risk.

Other studies have investigated the relationship between dietary patterns and risk of colorectal cancer,19 but few have focused on the Mediterranean diet. The PLCO trial found a significant reduction (21%) in colorectal adenoma risk among men who most complied with a Mediterranean dietary pattern, but no significant reduction in women.6 A high score on the Mediterranean Diet Score was associated with reduced risk of colorectal cancer among men only in the US NIH-AARP study, which on subsite analysis was not significant for proximal colon cancer.8 By contrast, adherence to a Mediterranean diet as measured by the Alternate Mediterranean Diet Index, was not associated with reduced colorectal cancer risk, either in men or women.7

The Mediterranean dietary pattern is characterized by high levels of components found protective against colorectal cancer and by low levels of components associated with increased risk of developing this disease. Thus a Mediterranean diet contains high proportions of fruit, vegetables and pulses, which are rich in dietary fiber, associated with risk reduction.3 The diet also contains a high proportion of pasta, which is a rich source of low glycemic index carbohydrates—associated with reduced cancer risk through action on the IGF axis.20 Furthermore, although the WCRF report3 found only a limited evidence of reduced colorectal cancer risk associated with high consumption of fruit and non-starchy vegetables, a recent meta-analysis by Aune et al. found that high vs. low consumption of both fruit and vegetables was associated with a modest but significant reduction in colon (not rectal) cancer risk.21 A Mediterranean diet is also characterized by low red and processed meat consumption, both of which have been convincingly associated with increased risk of colorectal cancer.3 These foods may promote colorectal carcinogenesis in several ways. Nitrates and nitrites added as preservatives to processed meats contribute to the production of N-nitroso compounds in the gut, several of which are known human and animal carcinogens.22 Meat cooked at high temperatures can contain heterocyclic amines and polycyclic aromatic hydrocarbons which are also carcinogens.22 Meat also contains hem iron, which promotes the formation of N-nitroso compounds and can lead to production of free radicals.23 Our Italian Mediterranean Index negatively scores alcohol consumption above 12 g/day, for both men and women; while in the Greek Mediterranean Index, the upper limits of “moderate” alcohol consumption are higher (25 g/day for women and 50 g/day for men). Alcohol consumption is convincingly associated with increased colorectal cancer risk with convincing and probable evidence for men and women, respectively.3 Acetaldehyde is a highly reactive metabolite of alcohol that can be carcinogenic.24 Alcohol may also function as a solvent, enhancing the penetration of other carcinogenic molecules into mucosal cells. In addition, alcohol may stimulate regenerative cell growth by various cytotoxic mechanisms including the excess production of oxygen free radicals.24

The lack of risk reduction for proximal colon may reflect the different biological characteristics of the mucosa of this part of the colon compared to distal colonic mucosa, that are acquired in embryonic or postnatal development. Furthermore cancer in the ascending colon has epidemiological and clinicopathological characteristics that differ from that at other colonic sites. These differences could determine differing responses to hypothesized environmental factors.25

Regarding the weaker risk reduction in men that women, this could be mainly due to the low number of cases in the individual categories. In both sexes, however, the effect of the Italian Mediterranean diet was in the same direction—protective against colorectal cancer.

To our knowledge, our study is the first to investigate the relation between adherence to a Mediterranean diet and risk of colorectal cancer in Europe. The instrument we used—the Italian Mediterranean Index—is relatively new and was developed by us to investigate the relationship between the typical Italian Mediterranean diet and risks of stroke, ischemic stroke and hemorrhagic stroke in the Italian EPICOR cohort study.12 In that study, we in fact assessed four a priori dietary indices: Healthy Eating Index 2005,26 the DASH diet,27 the Greek Mediterranean Index16, 17 and the Italian Mediterranean Index. We found that adherence to any one of these dietary indices was protective against at least one type of stroke but that a high score on the Italian Mediterranean Index was associated with the greatest risk reduction.

In our article, we investigated the effect of adherence to the Greek Mediterranean Index as well as the Italian Mediterranean Index, finding that those with a high score on the Greek Mediterranean Index had reduced risk of colorectal cancer; however, this protection disappeared when women alone were considered and only persisted for the rectal cancer subsite in the subsite analyses.

That adherence to these both dietary indices was found to be protective against colorectal cancer is not surprising, since both give points for high intake of fish, legumes and vegetables (typical Mediterranean vegetables in the Italian Mediterranean Index, all vegetables except potatoes in the Greek Mediterranean Index); and also give points for low intake of red and processed meat. However, the Italian Mediterranean Index was more sensitive in identifying reduced risk than the Greek Mediterranean Index, especially among women. Several differences between the two indices are likely to be responsible for this: (a) more restrictive definition of “good” or “healthy” dietary behavior (upper or lower tertile of the distribution in the Italian Mediterranean Index as opposed to the median cut-off in the Greek Mediterranean Index); and (b) lower cut-offs in the Italian Mediterranean Index to separate “healthy” from “unhealthy” alcohol consumption (12 g/day vs. 50 g/day in men and 25 g/day in women) and (c) more specific food items contributing to the Italian Mediterranean Index (pasta vs. potatoes, olive oil vs. butter and moderate alcohol vs. soft drinks). For these reasons, the Italian Mediterranean Index appears more suitable as well as more sensitive for capturing aspects of the diet that are healthy (protective against colorectal cancer and stroke), at least among Italians.

Strengths of our study are its prospective design, the use of detailed dietary questionnaires, and small number of participants lost to follow-up, limiting the possibility of selection bias. The study also exploited the information available on several non-dietary variables allowing us to control for their supposed confounding effect in the analyses.

A limitation of the study is that consumption estimates were based on a single dietary assessment in which participants were asked about eating habits over the preceding year, although dietary patterns are more reliable than single foods/nutrients as indicators of long-term usual diet.28 Furthermore, we cannot rule out confounding by factors that we were not able to estimate or estimated in a sub-optimal way in our questionnaires.


The authors thank all the participants in the Italian section of the EPIC study, A. Evangelista and D. Del Sette for technical support, Don Ward for help with English, Compagnia di San Paolo and Programma Integrato Oncologia, Regione Toscana, Ministero della Salute for financial support.