Olive oil intake is inversely related to cancer prevalence: a systematic review and a meta-analysis of 13800 patients and 23340 controls in 19 observational studies

Dietary fat, both in terms of quantity and quality, has been implicated to cancer development, either positively or negatively. The aim of this work was to evaluate whether olive oil or monounsaturated fat intake was associated with the development of cancer. A systematic search of relevant studies, published in English, between 1990 and March 1, 2011, was performed through a computer-assisted literature tool (i.e., Pubmed). In total 38 studies were initially allocated; of them 19 case-control studies were finally studied (13800 cancer patients and 23340 controls were included). Random effects meta-analysis was applied in order to evaluate the research hypothesis. It was found that compared with the lowest, the highest category of olive oil consumption was associated with lower odds of having any type of cancer (log odds ratio = -0.41, 95%CI -0.53, -0.29, Cohran's Q = 47.52, p = 0.0002, I-sq = 62%); the latter was irrespective of the country of origin (Mediterranean or non-Mediterranean). Moreover, olive oil consumption was associated with lower odds of developing breast cancer (logOR = -0,45 95%CI -0.78 to -0.12), and a cancer of the digestive system (logOR = -0,36 95%CI -0.50 to -0.21), compared with the lowest intake. The strength and consistency of the findings states a hypothesis about the protective role of olive oil intake on cancer risk. However, it is still unclear whether olive oil's monounsaturated fatty acid content or its antioxidant components are responsible for its beneficial effects.


Introduction
Dietary fat, both in terms of quantity and quality, has been implicated to cancer development, either positively or negatively. Monounsaturated and polyunsaturated fats, deriving from olive oil and fish oil respectively, are among those that data are emerging for their relation to certain cancer types [1][2][3]. Concerning olive oil, no answer still exists as to whether the monounsaturated fatty acid content or the antioxidant components of its unsaponifiable fraction are responsible for its beneficial effects. Its fatty acid composition is mainly oleic acid, followed by palmitic and linoleic acids [4]. In addition, extra-virgin olive oil contains phenolic antioxidants, including simple phenols, aldehydic secoiridoids, flavonoids and lignans [5,6]. The high content of oleic acid makes olive oil far less susceptible to oxidation than the polyunsaturated fatty acids, for example. Also, olive oil's most representative phenols are thought to be potent scavengers of superoxide and other reactive species, a possible step for mutagenesis [7].
The effect of olive oil on human health has, till now, mainly been analyzed by studies deriving from Mediterranean populations, where it is consumed in large quantities. Almost all studies do not distinguish between 'plain' olive oil, which is the most used in the world market, and extra-virgin olive oil. Moreover, as it has been already stated, before causally interpreting the usually observed inverse association of olive oil to certain malignancies, residual confounding is an issue, as well as other limitations arising from the fact that associations are reported mainly from case-control studies [8].
To the best of our knowledge, there is no published systematic review and meta-analysis regarding the effect of olive oil consumption on cancer occurrence. Thus, the aim was to evaluate whether olive oil intake is associated with various types of cancer.

Studies selection and data extraction
Original research studies published in English between 1990 and March 1, 2011 were selected through a computer-assisted literature search (i.e., Pubmed, http:// ncbi.nlm.nih.gov/PubMed). Combinations of key words relating to the aim of the paper were used, i.e. olive oil, Mediterranean diet, and cancer, or neoplasms. Also, specific searches are included: olive oil and, alternatively, prostate cancer, cancer of the larynx, oral carcinoma, cancer of the pharynx, cancer of the oesophagus, stomach cancer, lung cancer, ovarian cancer, renal cancer, endometrial cancer, pancreatic cancer, breast cancer, colorectal cancer, skin cancer, bladder cancer, haematological malignancies, leukaemia. In addition, the reference list of the retrieved articles was in some cases used to find articles not already allocated. The following information was abstracted according to a fixed protocol: study design, sample size, mean age and gender of participants, follow-up duration, assay methods, effect size measurements and potential confounders. The initial search (01.01.1990-01.03.2011, English language, humans) resulted in 248 entries on olive oil and cancer, 188 entries on olive oil and neoplasms, 297 entries on Mediterranean diet and cancer and 194 entries on Mediterranean diet and neoplasms. From the finally included papers, 3 were prospective, 35 were casecontrols and none was a clinical trial. For the papers that they were not fully available in PubMed the information was retrieved from the abstract. This systematic review follows a cancer-specific structure; all papers referring to olive oil are included, and all papers referring to monounsaturated fats in Mediterranean countries are included too, because a substantial proportion of monounsaturated fat in traditional Mediterranean countries derives from olive oil. Finally, papers referring to the cancer risk examining the monounsaturated to saturated fats ratio were not taken into account, because this ratio is a proxy for olive oil intake, but not a definite evidence for its use and, thus, its beneficial effects. In addition, for the meta-analysis of the selected studies the MOOSE guidelines have been taken into account [9].

Statistical analysis
Results of the studies are presented as odds ratios and their corresponding standard errors (SE) or 95% confidence intervals. From the finally included 38 papers (3 were prospective and 35 were case-controls), the metaanalysis was focused only on studies (n = 19) that solely evaluated raw olive oil intake, in order to eliminate the potential synergistic effect of cooking. Random effects meta-analysis of the selected studies was applied. The effect size measures used for the meta-analysis were the log-odds ratios (and the corresponding SE) of olive oil intake (highest vs. lowest available category) and the endpoints of the selected studies were cancers of above mentioned specified topographies. Heterogeneity was assessed using Cochran's Q and I 2 (I 2 ranges between 0% and 100%, with lower values representing less heterogeneity) and evaluated using the chi-squared test. Sensitivity analyses were performed according to some characteristics of the studies, such as: country of origin (Mediterranean or not) and type of cancer. To assess the presence of publication bias, the "funnel's plot" was tested. All statistical calculations were performed using NCSS 2004 software (Number Cruncher Statistical Systems Co., Utah, USA). Table 1 summarizes studies' characteristics. The majority of the studies that completed the entry criteria for this review were case -control (i.e., 35), while only three were prospective.

Results
In Table 2 the main findings of each study are summarised.

Breast cancer
Several case-control studies, conducted in Greece, Spain, Italy and France suggest that olive oil may be associated with decreased breast cancer. In a Greek hospital-based case-control study of 820 women with breast cancer and 1548 controls, after adjustment for energy intake, more frequent consumption of olive oil was associated with significantly reduced breast cancer risk (odds ratio, OR = 0.75) for more than once a day compared with once a day. This protective association was concentrated in postmenopausal women, but the relevant interaction term was not statistically significant [10]. Evaluating specifically the relation of monounsaturated fat intake with breast cancer, no suggestive association was shown (mutually adjusted RR per quintile 0.97, 95%CI 0.88 to 1.07) [11]. Furthermore in Greece, Trichopoulou et al. [12] studied the relation of conformity to the Mediterranean diet with breast cancer risk in the context of the European Prospective Investigation into Cancer and Nutrition cohort in Greece. Fourteen thousands eight hundred women were followed up for an average of 9.8  Olive oil intake was measured through a food-frequency questionnaire that included 78 specific foods and beverages.
years and identified 240 incident breast cancer cases.
Results showed that although increasing conformity to the Mediterranean diet was not associated with lower breast cancer risk in the entire cohort, however there was a marginally significant inverse association among postmenopausal women (HR = 0.78 for every 2 points; 95%CI 0.62 to 0.98). Moreover, results showed that the increased consumption of olive oil was associated with lower breast cancer risk in the entire cohort (HR = 0.93, 95%CI 0.80 to 1.08) and in the postmenopausal women (HR = 0.85, 95%CI 0.69 to 1.06) respectively, though not statistically significant [12].  Estimates from multiple logistic regression equations are presented, including terms for study center, age, sex, education, alcohol, total energy intake, and simultaneously the various types of oils and fats. After allowance for vegetable intake.
Benito E, et al [29] ORs for the higher available category of monounsaturated fat intake compared for the lowest one is 0.72 Not mentioned in the abstract in PubMed Adjustment for total calorie intake Galeone C, et al [30] OR for fried olive oil, 0.89, for colon cancer 0.82-0.98 Adjusted for age, center, sex, education, body mass index, tobacco smoking, alcohol drinking, non alcohol energy intake, family history, physical activity and red meat intake.

Prostate cancer
Tzonou A, et al [34] Chi-square linear trend adjusted = 0. 44 Adjusted for age, height, Quetelet index, years of schooling and total energy intake. Three case-control studies derive from Spain; in the first one (762 cases -988 controls), for olive intake, the odds ratios for the highest versus the lowest quartile of consumption was 0.66, with a significant dose-response trend [13]. In the second (100 breast cancer cases -100 controls), women in the highest tertile of monounsaturated fat consumption were at lower risk compared with women in the lowest tertile (RR = 0.30; 95%CI 0.1 to 1.08) [14]. At last, in the Canary Islands study, the odds ratio for women in the three upper quintiles of olive oil consumption (≥ 8.8 g/day) was 0.27 [15]. In Italy, in a multicenter case-control study (2564 cases -2588 controls), the odds ratio per unit (30 g) increase of olive oil consumption was 0.89 (95%CI 0.81 to 0.99); the ORs for increasing levels of olive oil consumption, adjusted for total calories, were found as 1.05, 0.99, 0.93 and 0.87 [16]. In the same country, in a prospective study that used nutritional data from 8984 women, an average  [75] Moderate increases in the risk for higher intake of monounsaturated fat were found, which disappeared after correction for saturated fat

Not mentioned in the abstract in PubMed
Adjusted for tobacco smoking and energy intake. follow-up of 9.5 years and 207 incident cases of breast cancer, the participants completed a semiquantitative FFQ, during 1987 to 1992 (ORDET cohort). The "salad vegetables" pattern, principally consisting of raw vegetables and olive oil, was associated with significantly lower (34-35%) breast cancer incidence (RR = 0.66, 95%CI 0.47 to 0.95), comparing the highest to lowest tertile [17]. In south France, a case-control study of 437 breast cancer cases and 922 controls showed an inverse association of olive oil intake with breast cancer risk [18]. In France, in a prospective study, Cottet et al. [19] examined the association between dietary patterns and breast cancer risk. The analyses included 2,381 postmenopausal invasive breast cancer cases diagnosed during a median 9.7-year follow-up period (1993-2005) among 65,374 women from the E3N-EPIC cohort. Results showed that the "healthy/Mediterranean" pattern was negatively associated with breast cancer risk (hazard ratio = 0.85, 95%CI 0.75 to 0.95; P = 0.003 for linear trend), especially when tumors were estrogen receptorpositive/progesterone receptor-negative. Authors concluded that adherence to a dietary pattern including mostly fruits, vegetables, fish, and olive/sunflower oil, along with avoidance of Western-type foods, could result in a substantial reduction in postmenopausal breast cancer risk [19]. Finally, in France, in another study, a positive association of monounsaturated fat intake with breast cancer risk was found [20].
Another study, the EURAMIC study, used adipose biopsies with diverse fat intake patterns from 5 European centres, including southern Europe (Malaga, Spain), to see if oleic acid or other monounsaturates are associated with breast cancer. In 291 postmenopausal incident breast cancer patients and 351 controls the OR (75 th to 25 th percentiles) was 0.40 in Malaga and 1.27 (not statistically significant) in all the other centres pooled. Thus, the strong inverse association between oleic acid concentrations and breast cancer in the Spanish study population was not observed in the study's non-Spanish populations [21].

Colorectal cancer
A number of case-control and prospective studies indicate that there is no appreciable association between intake of total, saturated, monounsaturated, or polyunsaturated fat intakes, in general, with the risk of colorectal cancer, or that the existing results are unconvincing [22][23][24].
Examining the association of olive oil and monounsaturated fat intake mainly in Mediterranean populations, with colorectal cancer in case-control studies, olive oil has been found to have a slight protective effect, while monounsaturated fat intake appeared uninfluential. In an Italian study of 1953 patients with colorectal carcinoma (1225 colon and 728 rectum) and 4154 controls, the ORs for successive tertiles, compared with the lowest one, were 0.87 (95%CI 0.75 to 1.01) and 0.83 (95%CI 0.70 to 0.99) (p = 0.03) when colorectal carcinoma was analyzed as a whole, 0.82 (95%CI 0.68 to 0.98) and 0.81 (95%CI, 0.66 to 0.99) (p = 0.04) for colon carcinoma, and 0.96 (95%CI 0.77 to 1.19) and 0.88 (95% CI 0.66 to 1.12) for rectal carcinoma. After allowance for vegetable intake, results were in the same direction, not statistically significant [25]. In the same sample [26,27] monounsaturated fat intakes appeared uninfluential [28]. Results towards the same direction were presented by Benito et al. [29] in 1991 [4,29]. When, patients from Switzerland were examined together (recruited from 1992 to 2000, 1394 colon cancer cases, 886 rectal cancer cases and 4765 controls), the OR for an increment of one portion of fried foods per week was 0.89 for colon cancer, 0.97 for rectum and 0.93 for colorectum, for the use of olive oil [30].
Other studies examining the role of monounsaturated fat in colorectal cancer outside of the Mediterranean basin were not taken into consideration, because in these countries monounsaturated fat may derive from sources other than olive oil [31][32][33].

Prostate cancer
In three case-control studies, neither monounsaturated lipids deriving mainly from olive oil nor olive oil per se were found to be associated with prostate cancer risk. In a case-control study in Greece, 320 patients with histologically confirmed incident prostate cancer and 246 controls were included [34]. Among added lipids, after adjustment for total energy intake among others, olive oil was unrelated to the risk (p = 0.66), though monounsaturated fats, largely deriving from meat or olive oil, were positively associated (OR 1.05, 95%CI 0.75 to 1.48, p = 0.77). As it is stated, this contradiction could in part be explained by the high content of vitamin E in olive oil, which was significantly inversely related to prostate cancer risk (OR 0.53, 95%CI 0.30 to 0.94, p = 0.03). In another population-based case-control study in New Zealand, after adjusting for energy intake, 317 prostate cancer cases and 480 controls completed a food-frequency questionnaire, in which for monounsaturated fatty acids-rich vegetable oils, a single question was asked about usual consumption of olive oil, canola or peanut oil, collectively, whereas biomarkers for fatty acids were measured in erythrocytes [35]. Increased consumption of MUFA-rich vegetables oils (not total MUFA, or MUFA of animal origin) was inversely associated with prostate cancer risk. Thus, according to the authors, a specific cancer-protective effect could be attributed to non-MUFA components of vegetable oils, as the antioxidant components of olive oil. In another population-based case-control study in Australia in 858 men aged (< 70 years) with histologically confirmed prostate cancer (Gleason grade > 5) and 905 age-frequency-matched men, higher consumption of olive oil (> 0.25 and < 0.25 l/month compared to non-consumption) had an OR = 0.8 (p for trend = 0.12) [36]. There was evidence for an inverse association of oleic acid with prostate risk with a marginal non-significance, which seems difficult to interpret, because olive oil only supplied 8% of oleic acid consumed.
Monounsaturated fat intakes and their association with prostate cancer in different countries than Mediterranean were not taken into consideration, since a major source of monounsaturated fat in North America, for example, is meat [37][38][39].

Cancer of the larynx
Concerning laryngeal cancer, a protective effect of olive oil intake was found in cross-sectional studies. An analysis was performed in the combined dataset from two case-control studies (1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000) conducted in northern Italy and Switzerland for 68 women less than 79 years, with histologically confirmed cancer of the larynx and 340 controls. After multivariate adjustment, an inverse association was found between high versus low intake of olive oil and laryngeal cancer (OR = 0.28, 95% CI 0.09 to 0.89) [40]. Olive oil has been related to a better prognosis to male laryngeal cancer patients, whereas, the same study group published a study in which the occurrence of new primaries among male laryngeal cancer patients was lowered by one-third with higher intake of monounsaturated fatty acids [41,42]. In another study conducted in the same countries, with 527 histologically confirmed laryngeal cancer cases and 1297 controls, an inverse association of the risk was observed with olive oil (OR = 0.4 for the highest compared to the lowest quintile of intake, p = 0.003), not statistically significant when controlling for total vegetable consumption (OR = 0.66, p = 0.45) [43]. Finally, in an analysis where cancers of the oral cavity, pharynx other than nasopharynx, larynx and oesophagus were collectively assessed [44], olive oil used in salads and/or in cooking, was significantly inversely associated (OR contrasting frequency of consumption above versus below median = 0.78 with 95% CI 0.67 to 0.90).

Cancer of the pharynx and the oral cavity
As it is stated in the paper by Garavello et al. [45] analysing the few prospective and case-control studies available, monounsaturated fats (and olive oil) seem to be inversely related to oral and pharyngeal cancer risk. In a case-control analysis in Italy (1992-1997), with 598 cases and 1491 controls, risk was approximately halved in the highest compared to the lowest quintile of olive oil (OR = 0.4, 95%CI 0.3 to 0.7), which was slightly attenuated by allowance for vegetable intake [46]. Moreover, in a recently conducted case-control study in Servia which included 45 cases with histopathological diagnosis of undifferentiated carcinoma of nasopharyngeal type (UCNT), and 90 controls, frequent/moderate consumption of olive oil was significantly negatively associated with UCNT compared with rare or never consumption of olive oil (OR = 0.42, 95%CI 0.19 to 0.91, p = 0.03) [47].
For oral cancer exclusively, in a case-control study in Greece with 106 patients with oral carcinoma and an equal number of controls, added lipids, which in Greece are represented overwhelmingly with olive oil, were found inversely significantly associated with oral carcinoma risk (p = 0.04) [48]. The joint analysis of Lagiou et al. [44], for oral, pharyngeal, laryngeal and oesophageal carcinoma, confirmed the same results. Finally, in a recently conducted case-control study in Greece in the context of the European alcohol-related cancers and genetic susceptibility in Europe project (239 incident upper aerodigestive tract, UADT, cases and 194 hospital controls), authors concluded that stricter adherence to the Mediterranean diet was associated with a substantial and significant decrease in UADT cancer risk (30% for a two-unit increase in score), whereas after mutual adjustment, no individual dietary component of this diet was significantly associated with this risk [49].

Oesophageal cancer
In three case-control studies, one yields inconclusive results for the association of monounsaturated lipids with oesophageal cancer and two others show a significant inverse association with olive oil intake. In Greece, 56 cases of adenocarcinoma and 43 cases of squamous cell carcinoma were compared with 200 controls. Neither for adenocarcinoma (OR = 1.07, 95%CI 0.72 to 1.60 between an increment of a marginal quintile in the intake of monounsaturated lipids), nor for squamous cell carcinoma (OR = 0.74, 95%CI 0.49 to 1.11 for the same increment), an association with monounsaturated lipids (mainly olive oil) was established [50,4].
In the second study (1992 to 1997) in northern Italy, 304 histologically confirmed squamous oesophageal carcinoma cases and 743 controls were interviewed, after correction for multiple confounders, olive oil intake showed a significant reduction of cancer risk (OR = 0.4). Without adjustment for total vegetable consumption, olive oil also showed a significant reduction of cancer risk, with no monotonic exposure association, that is, even consumers of a minimal quantity of olive oil appeared to be at a reduced risk for the disease, as consumers of a maximal quantity (OR = 0.3, 95% CI = 0.1 to 0.5), for the highest versus the lowest quintile of olive oil intake [51] and the same significant reduction was also found for monounsaturated fatty acids [52]. In a third case-control study in France (208 cases and 399 controls), all males, for olive oil, the OR for consumers versus non-consumers was 0.70 (95% CI 0.54 to 0.90) [53].

Stomach cancer
For stomach cancer, in Italy, 83 stomach cancer cases of negative microsatellite instability (MSI) (a distinctive molecular pathway of carcinogenesis) were dietary assessed. Olive oil had a statistically significant protective effect (OR = 0.6, 95%CI 0.3 to 1.00, for the highest versus the lowest tertile, p for trend = 0.05), whereas for MSI + stomach cancer cases, olive oil had also a protective effect, not statistically significant across tertiles (OR = 0.5, 95%CI 0.2 to 1.1, for the highest versus the lowest tertile, p for trend = 0.07) [54]. A protective effect of olive oil to stomach cancer was also detected in another Italian case-control study [55].

Lung cancer
Evidence is scarce. In a hospital-based, case-control study of lung cancer in Italy (342 with primary lung cancer and 292 controls), use of olive oil found to offer a protection towards lung cancer (OR = 0.67, 95%CI 0.45 to 0.99) [56]. Prospective studies on the association of lung cancer with dietary fat were not taken into consideration in countries, such as USA and Norway, because monounsaturated fat intake was mainly from sources other than olive oil [57][58][59].

Ovarian cancer
Evidence concerning olive oil in take in relation to ovarian cancer risk is limited; in two case-control studies, olive oil and monounsaturated fat (mainly olive oil) intake had a protective effect. Firstly, in Greece (189 epithelial ovarian cancer cases and 200 controls), an inverse relation of monounsaturated fat intake and risk for ovarian cancer was found (OR = 0.80, 95%CI 0.65 to 0.99 for 1SD increase in consumption on a daily basis) [60]. In a more recent case-control in Italy (1992 to 1999) 1031 cases and 2411 controls were included. After correction for multiple confounders, a reduced risk of ovarian cancer was observed for high intake of olive oil (OR = 0.68, 95%CI 0.50 to 0.93, for the highest quintile, compared with the lowest one), as well for higher intake of monounsaturated fat and oleic acid [61,62]. Other studies of diet and ovarian cancer were not taken into consideration, since the information was collected in non-Mediterranean countries [63,64].

Endometrial cancer
In three epidemiologic studies, there is a suggestion of a protective effect. In a case-control in Switzerland and northern Italy (274 endometrial cancer cases and 572 controls), after correction for energy intake, more frequent consumption of olive oil was associated with a decreased risk for endometrial cancer, though results were not statistically significant (OR = 0.82 for highest vs. lowest tertile) [65,4]. Two hospital case-control studies have been published from Greece. In the first one (145 cases and 298 controls), the only statistically suggestive association was the inverse one with monounsaturated fats (OR = 0.74, 95%CI 0.54 to 1.03) [66]. In the second one (84 cases and 84 controls), a protective effect of added lipids, which in the Greek diet are primarily represented by olive oil, was highly suggestive [67].

Pancreatic cancer
In two case-control studies, one yields inconclusive results for the association of monounsaturated lipids with pancreatic cancer, the second one shows a significant inverse association with olive oil. In a hospital case-control study conducted in Greece (181 cases and 181 hospital-181 hospital visitor controls), for a 1SD increase of the energy-adjusted residual of monounsaturated fat OR was 1.04 (95%CI 0.86 to 1.25) when compared with hospital controls and for 1SD increase in monounsaturated consumption OR was 0.97 (95%CI 0.80 to 1.17) compared to visitor controls, with no statistically significant association [68]. In northern Italy, in a case-control study conducted between 1983-1992 (362 cases with histologically confirmed, pancreatic cancer risk and 1502 controls), cancer risk was inversely associated with consumption of olive oil (OR = 0.58 for subsequent tertiles of intake) after allowance for sociodemographic factors and tobacco smoking (ORs were 0.76 for intermediate, and 0.60 for highest score of intake and the risk was significant) [69,70]. Monounsaturated fat intakes and oleic acid intakes in different countries than Mediterranean are not taken into consideration, since major sources of monounsaturated fat in North America is, for example, meat, and major sources of oleic acid in these countries could be canola, codliver, coconut, soybean, and almond oils [71][72][73].

Bladder cancer
In a case-control study in Belgium (200 cases and 386 controls) results showed that there was a statistically significant inverse association between olive oil intake and bladder cancer consistent with a linear dose-response relationship: middle versus the lowest tertile (OR = 0.62, 95%CI 0.39 to 0.99) and the highest versus the lowest tertile (OR = 0.47, 95%CI 0.28 to 0.78; p-trend = 0.002) [74]. However, findings from another study conducted by Riboli and colleagues [75], showed that in a Spanish population with an average dietary pattern typical of Mediterranean populations, monounsaturated fat intake was associated with a moderate increase at risk, which disappeared after correction for saturated fat intake.

Meta-analysis of the studies
As mentioned above for the meta-analysis 19 studies that evaluated solely olive oil intake were taken into account (Table 3). In total, data from 13800 cancer patients and 23340 controls were analysed in the present work. The combined effect of the highest percentile of olive oil intake compared with the lowest was highly significant (p < 0.001). Particularly, people in the highest group of olive oil consumption had 0.66-times lower odds of having any type of cancer (logOR = -0.41 95%CI -0.53 to -0.29) ( Table 3) or 34% lower likelihood of having any type of cancer. The effect-equality test that was applied revealed that the study-specific effect size measures were heterogeneous (Cohran's Q = 47.52, I 2 = 62%, p = 0.0002), but all to the same protective direction ( Figure 1).
Then, the studies were categorized by type of cancer, i.e., breast (n = 5 studies), digestive (i.e., colorectal, oral cavity, pharynx and oesophagus, pancreatic) (n = 8 studies) and other. Compared with low intake, high olive oil consumption was associated with lower odds of developing breast cancer (logOR = -0.45, 95%CI -0.78 to -0.12) and lower odds of developing a cancer of the digestive system (logOR = -0.36 95%CI -0.50 to -0.21) ( Table 3). As regards the type of origin, in both Mediterranean and non-Mediterranean people that reported olive oil intake were less likely to have developed any type of cancer ( Table 3). Heterogeneity of the effect-size measures was observed in studies performed in the Mediterranean region (n = 15, p = 0.0002), whereas no heterogeneity was observed in non-Mediterranean studies (n = 4, p = 0.12).

Conclusion
Nutritional factors play a major role in cancer initiation and development [76]. The present systematic review and meta-analysis of observational studies revealed that, overall, olive oil consumption was associated with lower odds of cancer development. Most prominent results were observed for breast cancer and cancers of the digestive system, while the aforementioned relationship was similar to studies performed in Mediterranean as well as non-Mediterranean countries. Meta-analyses have several weaknesses, due to inherent biases and differences in study designs (different control for residual confounding, different socio-demographic and other lifestyle characteristics that may alter food habits), publication bias, etc [9]. The use of only case-control studies may also limit the interpretation of the results as causal. Moreover, in 2 studies no exact information about the effect size was reported in the retrieved abstract; however, the direction of the effect was similar to the other studies. Therefore, it could be speculated that the inability to include these studies in the meta-analysis would not alter the combined results. At this point it should also be noticed that although it is acceptable to consider Cancer Combined Individual Figure 1 Forest plot of studies that evaluated the association between olive oil intake on cancer development (data are presented as log Odds Ratios and the corresponding 95%CI).
studying the risk associated with extreme categories, such as the highest vs. lowest, the use of an ordered variable, like the effect sizes per incremental quintile would be preferable; however, the available results did not allow for such an analysis. Moreover, the analysis of studies was focused only on those that evaluated raw olive oil intake, in order to avoid the synergistic effect of cooking. Nevertheless, the large number of studies enrolled, makes the present work one of the few that systematically evaluated the cancer-monousaturated lipid (especially olive oil) associations. Ecologic comparisons and meta-analysis of prospective cohort studies suggest that cancer morbidity and mortality are lower in Mediterranean countries, where olive oil represents a substantial fraction of dietary fat [77]. Prospective studies show evidence that higher degree of adherence to the Mediterranean diet is associated with a reduced mortality for cancer of all types, examined in one model [78]. According to a review, in Western countries, approximately 25% of the incidence of colorectal cancer, 15% of breast cancer, and 10% of prostate, pancreas, and endometrial cancer could be prevented if traditional Mediterranean dietary patterns were followed [79]. However, ecologic comparisons are characterized by complex forms of confounding, thus, are difficult to interpret [4], in addition, though the protective role of Mediterranean diet has been advocated by several studies, mutual confounding among food groups and nutrients, as plant foods and olive oil, have not been addressed in depth, and, it is probable that available Mediterranean dietary scales cannot capture all the interrelations among a multitude of foods consumed traditionally.
From ecologic comparison of 28 countries, 76% of the inter-country variation in colorectal cancer incidence rates could be attributed to three dietary factors: meat, fish and olive oil, in combination; meat and fish were found positively associated, whereas olive oil was negatively associated. The authors stated that olive oil could influence secondary bile acid patterns in the colon that, in turn, might influence polyamine metabolism in colonic cells reducing possibility to progression from normal mucosa to adenoma and, eventually, carcinoma [80]. Also, olive oil seems to be inversely associated with breast cancer risk, although there are no robust data to consider. Less experimental and epidemiologic evidence has been collected for the other types of cancer, such as gynecological, urological and respiratory cancers, and there are no available data for haemopoietic malignancies. As it is stated previously, most available studies till today are case-control ones and in the majority of them total energy intake is controlled for. Most of these studies are examining the role of a series of different nutrients and are not examining the specific role of added fats, monounsaturated fats (in Mediterranean countries) and in particular olive oil, to the above mentioned different types of cancer.
In addition, a recent meta-analysis of biomarkers of dietary fatty acid intake and risk of breast cancer concluded that there is no significant association between oleic acid and breast cancer risk in case-control studies, whereas in cohort studies there was significant higher risk for women with high levels of this fatty acid [81]. These papers referred were not taken into consideration, because it was not clear from the studies that oleic acid was deriving from olive oil only or from other dietary sources as well. Also, selection and nutrient-specific information biases are reported to be minimized in some of the studies taken into consideration, while others have also provided additional results after taking into consideration the mutual confounding by different food groups and nutrients, such as of olive oil and vegetables. Finally, though there is not enough data for skin cancer, it could be stated that due to the high concentrations of squalene in olive oil, which is transferred to the skin, olive oil intake could be searched for a protective effect against skin cancer [5].
Evidence to support that olive oil conveys protection against occurrence of different types of cancer necessitates more epidemiological studies, especially prospective ones, specifically designed to address these issues, which would all adjust for total energy intake. The ongoing and future well designed cohort studies will help to further examine the association and questions arising, such as, firstly if olive oil intake facilitates more vegetable intake, thus, maximising its beneficial effects to cancer prevention, and secondly, if the possible beneficial effects of olive oil are attributed to its monounsaturated content or to its other components, could be addressed [1,8,82]. Large enough and well conducted trials, both for countries with high as with low intake of olive oil, could be necessary, for the purpose of generalisability. and had responsibility for its final content. DP analyzed data, wrote parts of the paper and had responsibility for its final content. All authors read and approved the final manuscript.