The Mediterranean diet includes a complex of dietary patterns with different components in various regions of the Mediterranean, but is generally characterised by frequent consumption of fruit, vegetables, carbohydrates, pulses and fish, low consumption of meat and cheese, and (as a major common characteristic) abundant use of olive oil for seasoning. The Mediterranean diet has been proposed as a plausible explanation for the longevity of populations from most European countries bounded by the Mediterranean Sea(Reference Trichopoulou, Kouris-Blazos, Wahlqvist, Gnardellis, Lagiou, Polychronopoulos, Vassilakou, Lipworth and Trichopoulos1, 2).
Several peculiar aspects of the Mediterranean diet – as well as the Mediterranean diet-related lifestyle as a whole – have been related to a reduced risk not only of cardiovascular disease, but also of cancer(Reference Bosetti, Gallus, Trichopoulou, Talamini, Franceschi, Negri and La Vecchia3–Reference La Vecchia and Bosetti5). It has been suggested that up to 25 % of colorectal, 15 % of breast and 10 % of prostate, pancreas and endometrial cancers could be prevented by adhering to a Mediterranean diet(Reference Trichopoulou, Lagiou, Kuper and Trichopoulos6).
In the present paper, we review the main findings relating some aspects of the Mediterranean diet (including, in particular, the consumption of olive oil and carbohydrates) to cancer risk using data from an integrated series of multicentric case-control studies conducted in Italy between 1991 and 2007. These include about 10 000 subjects with various common cancers, including 749 oral and pharyngeal, 395 esophageal, 230 gastric, 1225 colon, 728 rectal, 527 laryngeal, 2569 breast, 454 endometrial, 1031 ovarian, 1294 prostate, 767 renal cell cancers and a sum of over 17 000 hospital controls. For all these studies, data collected was based on a reproducible and valid food frequency questionnaire (FFQ)(Reference Decarli, Franceschi, Ferraroni, Gnagnarella, Parpinel, La Vecchia, Negri, Salvini, Falcini and Giacosa7, Reference Franceschi, Barbone and Negri8).
Olive oil – the major common characteristic of diet in various parts of the Mediterranean – is a major source of monounsaturated fats in Mediterranean countries, but also an important source of several micronutrients and food components. It appears to be a favourable indicator of the risk of various common cancers, although the evidence is still open to evaluation.
In our network of case-control studies from Italy, where the subjects’ usual diet was investigated through a validated FFQ, including specific questions aimed at assessing added fat intake patterns, we considered the association of olive oil with breast, ovarian, colorectal and upper digestive and respiratory tract cancers.
With reference to breast cancer, after allowance for demographic and reproductive risk factors, energy intake and, mutually, other types of fats, the continuous odds ratio (OR) for a unit increase intake of olive oil was 0·89 (95 % confidence interval (CI) 0·81, 0·99)(Reference La Vecchia, Negri, Franceschi, Decarli, Giacosa and Lipworth9). This is consistent with results from animal data, ecologic studies and other epidemiological studies conducted in Greece and Spain(Reference Lipworth, Martinez, Angell, Hsieh and Trichopoulos10), which indicate that there is an inverse relation of breast cancer risk with intake of olive oil, but not of butter or margarine.
With reference to ovarian cancer, after allowance for study centre, year at interview, age, education, parity, oral contraceptive use and total energy intake, a reduced risk was observed for high intake of olive oil (OR = 0·68 for the highest quintile of intake compared to the lowest one)(Reference Bosetti, Negri, Franceschi, Talamini, Montella, Conti, Lagiou, Parazzini and La Vecchia11). No significant associations were observed for mixed seed oils, butter and margarine. Thus, this large case-control investigation on diet and ovarian cancer conducted on a population with appreciable heterogeneity of fat intake provided a suggestion for a favourable effect of olive oil, which was only partly accounted for by vegetable intake.
For colorectal cancer risk, the OR for subsequent tertiles of olive oil intake, compared with the lowest one, were 0·87 and 0·83 when colorectal carcinoma was analysed as a whole, 0·82 and 0·81 for colon cancer and 0·96 and 0·88 for rectal cancer(Reference Braga, La Vecchia, Franceschi, Negri, Parpinel, Decarli, Giacosa and Trichopoulos12). Seasoning fats did not appear to increase the risk of colorectal carcinoma, and there was little evidence for a differential effect by fat type. If such a differential effect exists, it could favour olive oil. Likewise, our results did not indicate a relevant role of fried foods on colorectal cancer risk in Italy, and suggested a possible favourable effect of (fried) olive oil on colon cancer risk(Reference Galeone, Talamini, Levi, Pelucchi, Negri, Giacosa, Montella, Franceschi and La Vecchia13).
The most consistent evidence of a favourable role of olive oil came for upper digestive and respiratory tract cancers. With reference to oral and pharyngeal cancer, high intake of olive oil was associated with significantly lowered risk (OR = 0·4 for the highest v. the lowest quintile of intake) (Table 1)(Reference Franceschi, Favero, Conti, Talamini, Volpe, Negri, Barzan and La Vecchia14). The beneficial effect of olive oil was attenuated by the introduction of vegetable consumption in the model. Mixed seed oils and margarine were not related to cancer risk, whereas a strong positive association emerged for butter (OR = 2·3). Fat intake patterns exerted a stronger influence on the risk of cancer of the oral cavity and pharynx than on colorectal cancer or breast cancer in companion Italian studies.
Likewise, in a case-control study of oesophageal cancer, olive oil intake showed a significant reduction of risk (OR = 0·3), while butter consumption was directly associated with oesophageal cancer risk (OR = 2·2) (Table 1)(Reference Bosetti, La Vecchia, Talamini, Simonato, Zambon, Negri, Trichopoulos, Lagiou, Bardini and Franceschi15). No significant associations emerged with consumption of specific seed oils, mixed seed oils or margarine. The introduction into the model of total vegetable consumption slightly reduced the association with olive oil, which remained, however, statistically significant.
In an analysis of the role of various seasoning fats in relation to laryngeal cancer risk, a significant reduction of cancer risk was observed for olive oil (OR = 0·4) and specific seed oils (OR = 0·6), while mixed seed oils were directly associated with laryngeal cancer risk (OR = 2·2) (Table 1)(Reference Bosetti, La Vecchia, Talamini, Negri, Levi, Dal Maso and Franceschi16). No significant associations emerged for consumption of butter and margarine.
In conclusion, large multricentric Italian studies showed that olive oil is a favourable indicator of breast, ovarian, colorectal, but mostly of upper digestive and respiratory tract cancers. For these neoplasms, the relative risk difference between extreme levels of olive oil v. butter consumption reached a factor 4 to 5, pointing to olive oil as a relevant factor of Mediterranean diet on cancer risk.
The beneficial effect of olive oil could be explained by antioxidant properties, due both to oleic acid itself and to the presence of other antioxidants, such as vitamin E and polyphenols(Reference Owen, Giacosa, Hull, Haubner, Wurtele, Spiegelhalder and Bartsch17). It is, however, possible that the observed associations with olive oil are not due to its specific components, but to the fact that higher consumption of olive oil is an indicator of healthier dietary habits, with a more frequent consumption of vegetables, and possibly of other beneficial lifestyle factors.
Cereals and fibres
In previous studies from northern Italy, refined grain intake was associated with an increased risk of stomach, colorectal, upper aero-digestive tract and thyroid cancers(Reference Chatenoud, La Vecchia, Franceschi, Tavani, Jacobs, Parpinel, Soler and Negri18), while frequent consumption of whole grains significantly decreased the risk of most cancers analysed, including those of the upper aero-digestive tract, stomach, colorectum, liver, breast, ovary, bladder and kidney(Reference Chatenoud, Tavani, La Vecchia, Jacobs, Negri, Levi and Franceschi19).
With reference to our data, the OR of colon and rectal cancers were below unity for high intakes of most types of dietary fibres, and no appreciable differences emerged between the two sites(Reference Negri, Franceschi, Parpinel and La Vecchia20). A strong protective effect of various types of dietary fibres was also found for upper aero-digestive tract neoplasms (OR were 0·4 for oral/pharyngeal and esophageal and 0·3 for laryngeal cancer)(Reference Pelucchi, Talamini, Levi, Bosetti, La Vecchia, Negri, Parpinel and Franceschi21, Reference Soler, Bosetti, Franceschi, Negri, Zambon, Talamini, Conti and La Vecchia22). A moderate inverse association between selected fibres, particularly cellulose and soluble non-cellulose polysaccharides, was found for hormone-related cancers, i.e. breast, ovary and prostate(Reference La Vecchia, Ferraroni, Franceschi, Mezzetti, Decarli and Negri23–Reference Pelucchi, La Vecchia, Chatenoud, Negri, Conti, Montella, Calza, Dal Maso and Franceschi25), while no overall association was found for renal cell cancer(Reference Galeone, Pelucchi, Talamini, Negri, Montella, Ramazzotti, Zucchetto, Dal Maso, Franceschi and La Vecchia26).
Thus, in this Italian population, high frequency of consumption of whole grains is an indicator of reduced cancer risk, whereas high intake of refined grains is directly associated with cancer.
Dietary fibres may act on cancer risk by reducing glycaemic load (GL) and improving insulin sensitivity, favourably influencing insulin-like growth factor 1 (IGF-1), i.e. a promoter of the process of carcinogenesis at various sites(Reference Yu and Rohan27), and by other mechanisms specific to each site. For example, fibres might protect from colorectal cancer by increasing stool bulk, thus reducing transit time and contact of carcinogens with the colonic mucosa, and by preventing the conversion of primary to secondary bile acids(Reference Bingham28). Also, it was suggested that fibres might influence the levels and availability of estrogens and other steroid hormones in breast carcinogenesis.
In conclusion, whole grain foods – that have a higher content of dietary fibres – should replace refined cereal foods. This is of specific relevance in countries like Italy, where white bread and pasta are major components of the diet.
Glycaemic index and glycaemic load
The roles of refined and whole grain foods and of dietary fibres on cancer are connected to their dietary glycaemic index (GI) (an indicator of the rate of adsorption of carbohydrates and, hence, a measure of insulin demand) – and GL (which combines the quality as well as the quantity of carbohydrate consumed). GI, and in particular GL, have been suggested to be relevant factors in the aetiology of several chronic diseases, including cancers(Reference Barclay, Petocz, McMillan-Price, Flood, Prvan, Mitchell and Brand-Miller29, Reference Jenkins, Kendall, Augustin, Franceschi, Hamidi, Marchie, Jenkins and Axelsen30). This points to a potential role of insulin and, hence, IGF in cancer promotion.
Several investigations from our network of case-control studies have considered the relation between GI and GL and cancer risk, providing a large amount of data from a population with frequent consumption of carbohydrates (Table 2). The OR of upper aero-digestive tract neoplasms for the highest as compared to the lowest level of dietary GI and GL were 1·5 and 1·8, respectively(Reference Augustin, Gallus, Franceschi, Negri, Jenkins, Kendall, Dal Maso, Talamini and La Vecchia31). The associations were similar for all cancer sites considered, including those of the oral cavity/pharynx, oesophagus and larynx.
With reference to digestive tract cancers, there was a direct association between dietary GL and gastric cancer (OR = 1·9), but no consistent pattern of risk was found with GI(Reference Augustin, Gallus, Negri and La Vecchia32). An increased risk of colorectal cancer emerged for high GI (OR = 1·7) and GL (OR = 1·8) after allowance for several covariates, including dietary fibres(Reference Franceschi, Dal Maso, Augustin, Negri, Parpinel, Boyle, Jenkins and La Vecchia33). The OR were higher for colon than rectal cancer.
Considering female hormone-related neoplasms, direct associations with breast, ovarian and endometrial cancers were found in our studies for both GI (OR between 1·4 and 2·1) and GL (OR between 1·3 and 2·7)(Reference Augustin, Dal Maso, La Vecchia, Parpinel, Negri, Vaccarella, Kendall, Jenkins and Franceschi34–Reference Augustin, Polesel and Bosetti36). The associations were observed in pre- and postmenopausal women, and were consistent across strata of major covariates identified. Furthermore, a prospective investigation conducted in northern Italy between 1987 and 1992, including 289 women with breast cancer found increased risks for women with high dietary GI (relative risk (RR) = 1·57 for the highest v. lowest quintile) and GL (RR = 2·53)(Reference Sieri, Pala and Brighenti37). Therefore, these studies support the hypothesis of a possible role of hyperinsulinaemia/insulin resistance in selected female cancers.
With reference to the study of prostate cancer, the OR were 1·23, 1·24, 1·47 and 1·57 for increasing levels of GI as compared to the lowest quintile(Reference Augustin, Galeone and Dal Maso38). The corresponding values for GL were 0·91, 1·00, 1·20 and 1·41. Also, GI and GL appeared to have a similar influence at different stages of the disease, measured by the Gleason score.
High dietary levels of GI and GL were associated with thyroid cancer, too(Reference Randi, Ferraroni, Talamini, Garavello, Deandrea, Decarli, Franceschi and La Vecchia39). Compared with the lowest tertile, the OR in subsequent tertiles were 1·68 and 1·73 for GI, and 1·76 and 2·17 for GL.
Meat and fish
The traditional Mediterranean diet included a limited intake of red meat that has been associated to increased risks of colorectal and other cancers(Reference Trichopoulou, Lagiou, Kuper and Trichopoulos6). In our integrated series of studies from Italy, the multivariate OR for the highest v. lowest level of red meat intake were 1·2 for cancer of the oral cavity/pharynx(Reference Franceschi, Favero, Conti, Talamini, Volpe, Negri, Barzan and La Vecchia14), 1·9 for oesophagus(Reference Bosetti, La Vecchia, Talamini, Simonato, Zambon, Negri, Trichopoulos, Lagiou, Bardini and Franceschi15), 1·2 for stomach(Reference Lucenteforte, Scita, Bosetti, Negri and La Vecchia40), 1·1 for colorectum(Reference Franceschi, Favero, La Vecchia, Negri, Conti, Montella, Giacosa, Nanni and Decarli41), 2·1 for liver(Reference Talamini, Polesel, Montella, Dal Maso, Crispo, Tommasi, Izzo, Crovatto, La Vecchia and Franceschi42), 3·1 for larynx(Reference Bosetti, Negri and Kolonel43), 1·1 for breast(Reference Franceschi, Favero, La Vecchia, Negri, Dal Maso, Salvini, Decarli and Giacosa44), 1·8 for endometrium, 1·5 for ovary(Reference Bosetti, Negri, Franceschi, Pelucchi, Talamini, Montella, Conti and La Vecchia45), 1·0 for prostate(Reference Bosetti, Micelotta, Dal Maso, Talamini, Montella, Negri, Conti, Franceschi and La Vecchia46) and 0·8 for renal cell cancer(Reference Bravi, Bosetti, Scotti, Talamini, Montella, Ramazzotti, Negri, Franceschi and La Vecchia47). This and previous reports(Reference Tavani, La Vecchia, Gallus, Lagiou, Trichopoulos, Levi and Negri48) point to red meat as another important factor (after vegetables and fruits) in the nutritional aetiology of selected human cancers.
Fish intake (a major source of n-3 polyunsaturated fatty acids, PUFA) has been shown to be a favourable indicator for several common cancers(Reference Fernandez, Chatenoud, La Vecchia, Negri and Franceschi49). In our datasets, the continuous OR for an increase of 1 g/week of n-3 PUFA were below unity for all cancers considered, except prostate: 0·7 for oral/pharyngeal and oesophageal cancer, and 0·9 for colon, rectal, breast and ovarian cancer(Reference Tavani, Pelucchi, Negri, Bertuzzi and La Vecchia50). Together with the favourable effect of fish and n-3 PUFA intake on the cardiovascular system(51, Reference Marckmann and Gronbaek52), the results from these data indicate that fish is a preferable substitute for meat intake in the Mediterranean diet.
A Mediterranean diet score
In a prospective investigation conducted in Greece including more than 22 000 adults, adherence to the Mediterranean diet was assessed using a ten-point scale(Reference Trichopoulou, Costacou, Bamia and Trichopoulos4). The present study found that a higher degree of adherence to the Mediterranean diet was associated with a reduction in total mortality and mortality from cancer (OR = 0·76 for a two-point increment in the Mediterranean-diet score).
We applied a similar score to define a dietary pattern reflecting favourable aspects of diet in our Italian data. This simple and intuitive score summarised eight of the major characteristics of the Mediterranean diet, i.e. high consumption of cereals, legumes, fruit, vegetables, low consumption of meat, milk and dairy products, high monounsaturated/saturated fat ratio and moderate alcohol intake(Reference Bosetti, Gallus, Trichopoulou, Talamini, Franceschi, Negri and La Vecchia3). The relation between this a priori defined nutritional pattern and the risk of cancers of the upper aero-digestive tract has been evaluated. For all cancers considered, a significant reduced risk was found for increasing levels of the Mediterranean score: the multivariate OR for subjects with ≥6 Mediterranean characteristics, compared with those with <3, were 0·40 for oral and pharyngeal, 0·26 for oesophageal and 0·23 for laryngeal cancer.
Findings from our network of case-control studies, together with other reports from Mediterranean and non-Mediterranean countries(Reference Trichopoulou, Costacou, Bamia and Trichopoulos4, Reference Trichopoulou, Lagiou, Kuper and Trichopoulos6, Reference Mitrou, Kipnis and Thiébaut53), provide additional evidence that a number of peculiar characteristics of the Mediterranean diet favourably affect cancer risk. In particular, olive oil appears to have a beneficial effect on cancer risk as compared to butter and other added fats. Moreover, whole grain carbohydrates, which have a higher content of dietary fibres and lower GI, should be preferred to refined ones, for the prevention of several common cancers. A low-risk diet for cancer would also imply reducing red meat and increasing consumption of fish. Recent surveys in Italy also indicated that overweight and obesity have not been rising over the last two decades as in the USA and several European countries(Reference Gallus, Colombo, Scarpino, Zuccaro, Negri, Apolone and La Vecchia54). The impact on body weight is therefore an additional favourable effect of the Mediterranean diet, not only for its implication on cardiovascular diseases, but also on cancer risk(Reference Calle, Rodriguez, Walker-Thurmond and Thun55, Reference Reeves, Pirie, Beral, Green, Spencer and Bull56).
These data supporting the Mediterranean diet and its favourable effects on cancer are of particular importance at a public health level, given the tendency towards globalisation of diet in most areas of the world.
The present study was conducted with the contribution of the Italian Association for Cancer Research and the Italian League Against Cancer. The work in the present paper was undertaken while C.L.V. was a senior fellow at the International Agency for Research on Cancer. The authors thank Mrs Ivana Garimoldi for editorial assistance.