Hazard ratios were obtained from Cox regression models with robust estimates for the variance to account for intra-cluster correlations. EVOO indicates extra-virgin olive oil; HR, hazard ratio; MeDiet, Mediterranean diet.
Results obtained from Cox regression models. Adjusted for age, use of hormone therapy, physical activity, body mass index, alcohol consumption, baseline adherence to the Mediterranean diet, age at menopause, total energy intake, smoking status, prevalent diabetes mellitus, family history of cancer, and use of statins. Error bars show 95% confidence intervals. HR indicates hazard ratio.
eTable. Energy, nutrient and supplemental foods intake at baseline and after 3 years among women in the PREDIMED trial by intervention group
eAppendix. Other PREDIMED Investigators
eFigure. Flow-chart of participants in the PREDIMED trial, 2003-2010
Customize your JAMA Network experience by selecting one or more topics from the list below.
Toledo E, Salas-Salvadó J, Donat-Vargas C, et al. Mediterranean Diet and Invasive Breast Cancer Risk Among Women at High Cardiovascular Risk in the PREDIMED Trial: A Randomized Clinical Trial. JAMA Intern Med. 2015;175(11):1752–1760. doi:10.1001/jamainternmed.2015.4838
Breast cancer is the leading cause of female cancer burden, and its incidence has increased by more than 20% worldwide since 2008. Some observational studies have suggested that the Mediterranean diet may reduce the risk of breast cancer.
To evaluate the effect of 2 interventions with Mediterranean diet vs the advice to follow a low-fat diet (control) on breast cancer incidence.
Design, Setting, and Participants
The PREDIMED study is a 1:1:1 randomized, single-blind, controlled field trial conducted at primary health care centers in Spain. From 2003 to 2009, 4282 women aged 60 to 80 years and at high cardiovascular disease risk were recruited after invitation by their primary care physicians.
Participants were randomly allocated to a Mediterranean diet supplemented with extra-virgin olive oil, a Mediterranean diet supplemented with mixed nuts, or a control diet (advice to reduce dietary fat).
Main Outcomes and Measures
Breast cancer incidence was a prespecified secondary outcome of the trial for women without a prior history of breast cancer (n = 4152).
After a median follow-up of 4.8 years, we identified 35 confirmed incident cases of breast cancer. Observed rates (per 1000 person-years) were 1.1 for the Mediterranean diet with extra-virgin olive oil group, 1.8 for the Mediterranean diet with nuts group, and 2.9 for the control group. The multivariable-adjusted hazard ratios vs the control group were 0.31 (95% CI, 0.13-0.77) for the Mediterranean diet with extra-virgin olive oil group and 0.53 (95% CI, 0.23-1.26) for the Mediterranean diet with nuts group. In analyses with yearly cumulative updated dietary exposures, the hazard ratio for each additional 5% of calories from extra-virgin olive oil was 0.72 (95% CI, 0.57-0.90).
Conclusions and Relevance
This is the first randomized trial finding an effect of a long-term dietary intervention on breast cancer incidence. Our results suggest a beneficial effect of a Mediterranean diet supplemented with extra-virgin olive oil in the primary prevention of breast cancer. These results come from a secondary analysis of a previous trial and are based on few incident cases and, therefore, need to be confirmed in longer-term and larger studies.
ISRCTN.org Identifier: ISRCTN35739639
Breast cancer, the most frequently diagnosed malignant tumor and the leading cause of cancer death among women, has increasing incidence rates. In 2012, 1.7 million women received a diagnosis of breast cancer. Since the 2008 estimates, breast cancer incidence has increased by more than 20% worldwide, while mortality has increased by 14%.1 In European countries, breast cancer is the most common incident cancer and the first or second (after lung cancer) malignant neoplasm implicated in mortality among women.2
Diet has been extensively studied as a modifiable component of lifestyle that could influence breast cancer development. Epidemiological evidence on the effect of specific dietary factors is still inconsistent, and the only convincing evidence relates to an increased risk in women with high alcohol consumption.3
The inconsistent association between foods or nutrient consumption and breast cancer risk may be partly due to the fact that individuals do not consume foods or nutrients in isolation but mixtures of foods with different nutrient constituents that may interact synergistically to influence biological pathways leading to or protecting from cancer. Thus, assessing diet as a whole, based on overall dietary patterns, provides more useful information on the role of diet in breast cancer risk. The Mediterranean dietary pattern has attracted considerable attention because, historically, breast cancer rates have been lower in Mediterranean countries than in Northern or Central European countries or the United States.4,5 The Mediterranean diet (MeDiet) is characterized by an abundance of plant foods, fish, and especially olive oil.5 In the Lyon Diet Heart Study, participants allocated to a cardioprotective Mediterranean-type diet showed a 61% lower risk of cancer (all subtypes) than those participants allocated to a control diet close to the step 1 American Heart Association prudent diet.6 Recent prospective cohort studies have evaluated the association between adherence to a MeDiet pattern and specifically breast cancer risk.7,8 However, the epidemiological evidence is still limited and conflicting.9,10 Moreover, no randomized trial has ever assessed the effect of the MeDiet on the primary prevention of breast cancer.
To further examine the effects of the MeDiet on breast cancer risk, we have analyzed the effect of the MeDiet supplemented with extra-virgin olive oil (EVOO) or nuts in the randomized intervention of the PREDIMED trial on the incidence of breast cancer.
This study was conducted within the frame of the PREDIMED (Prevención con Dieta Mediterránea) trial (ISRCTN35739639) (http://www.predimed.es).11,12 Briefly, PREDIMED is a large, multicenter, randomized trial designed to test the effects of the traditional MeDiet on the primary prevention of cardiovascular disease (CVD). The protocol (Supplement 2) was approved by the institutional review boards at all study locations. The trial was stopped in December 1, 2010, after a median follow-up of 4.8 years because of evidence of early cardiovascular benefit of both MeDiet groups compared with the control group.
Eligible participants for the PREDIMED trial were men aged 55 to 80 years and women aged 60 to 80 years free of CVD at enrollment, who had either type 2 diabetes mellitus or at least 3 of the following major cardiovascular risk factors: smoking, hypertension, elevated low-density lipoprotein cholesterol level, low high-density lipoprotein cholesterol level, overweight or obesity, or family history of premature coronary heart disease.11 Study candidates were selected from databases of primary health care facilities. Of 8367 candidates meeting enrollment criteria, 89% agreed to participate and provided written informed consent.
During the period October 2003 through June 2009, 7447 participants were enrolled in the PREDIMED trial, of whom 4282 were women. According to the protocol, participants were to be individually randomly allocated in a 1:1:1 ratio to the 3 intervention groups: MeDiet supplemented with EVOO, MeDiet supplemented with mixed nuts, or control diet (advice to reduce dietary fat). The coordinating center constructed a computer-generated randomization table. Allocation was concealed by opaque, sequentially numbered, and sealed envelopes during part of the pilot phase of the study. Nevertheless, envelopes were not used for the remainder of the study. Randomization was stratified by sex and age. Departures from the protocol of individual randomization protocol have been reported in detail elsewhere and affected a small subset of participants in the PREDIMED trial.12 In 1 of the 11 recruitment sites (site D) 245 women, representing 5.9% of the trial were allocated by clusters (clinics) instead of being individually randomized. Also, a small number of women shared the same household of a previously enrolled participant and they were allocated during the duration of the trial to the same group of the trial as their relative. This was the case for 49 women (1.2%) included in this analysis. For the present study, 1 woman was excluded because of a prior diagnosis of breast cancer and 7 other women were excluded because of probable (not confirmed as malignant) breast tumors. Investigators assessing the occurrence of new breast cancer cases were blinded to the intervention.
Participants in the 2 intervention groups were given supplementary foods for free: EVOO (1 L/wk for the participant and their families) or mixed nuts (30 g/d: 15 g walnuts, 7.5 g hazelnuts, and 7.5 g almonds) according to their randomization group. The purpose of supplementation was both to ensure a high consumption of these key components of the traditional MeDiet and to promote a better overall adherence to the intervention.
At baseline and quarterly thereafter, dieticians ran individual and group sessions, with up to 20 participants, separately for each group. In the appropriate individual sessions, a 14-item dietary screening questionnaire was used to assess adherence to either of the MeDiets, and a 9-item dietary screening questionnaire was used to assess adherence to the control diet. The answers to the questionnaires were used as a tool to personalize the intervention for each participant and to negotiate changes to upgrade adherence to either the MeDiet or the control diet.
Participants in the control group also received dietary training at the baseline visit and completed the 14-item dietary screener used to assess baseline adherence to the MeDiet. Thereafter, during the first 3 years of the trial, they received a leaflet explaining the low-fat diet on a yearly basis. However, the realization that the more infrequent visit schedule and less intense support for the control group might be limitations of the trial prompted us to amend the protocol in October 2006. Thereafter, participants assigned to the control diet received personalized advice and were invited to group sessions with the same frequency and intensity as those in the MeDiet groups, with the use of a separate 9-item dietary screener. During the study, participants in the control group received gifts of nonfood items as incentives. Attained changes in diet are shown in the eTable in Supplement 1.
Energy restriction was not specifically advised, nor was physical activity promoted in any group. The intervention did not target drug prescriptions; thus, it was implemented within the regular medical care of the participants.
Cases were defined as the first invasive breast cancer (International Classification of Diseases for Oncology codes C50.1-C50.9). Availability of results from a cytological or histological examination was considered as confirmation. Even though information on biological parameters was not requested for a case to be accepted, medical records were reviewed to extract this information. Incident cases through December 1, 2010, were identified from 2 sources: review of all the medical records of each participant by a panel of physicians (masked to the intervention), both at the primary health care level and at the hospital level, and death certificates (International Classification of Diseases, Ninth Revision, code 174 or International Classification of Diseases, 10th Revision, code C50). A clinical events committee blinded to the intervention and the dietary information of participants adjudicated all end points using prespecified criteria.
Cancer incidence was defined as a secondary outcome in the original study protocol. Five specific cancer locations were always included as relevant outcomes in all interim analyses and in all reports prepared every year for the Data and Safety Monitoring Board of the PREDIMED trial: breast cancer, lung cancer, prostate cancer, colorectal cancer, and gastric cancer. These results on breast cancer are the first results for any cancer that have been analyzed and submitted for publication in the PREDIMED trial.
Follow-up ended at the time of diagnosis of an invasive breast cancer, death, last follow-up contact, or December 1, 2010, whichever occurred first.
At baseline and once yearly during follow-up, a validated 14-item MeDiet screener,13 a general medical questionnaire, a 137-item validated food frequency questionnaire,14 and the Minnesota Leisure-Time Physical Activity Questionnaire15,16 were administered. Information from the food frequency questionnaire was used to calculate intake of energy and nutrients. Other lifestyle-related variables such as smoking, health conditions, and sociodemographic variables were assessed by a 47-item general questionnaire.12 In addition, trained study personnel directly measured weight, height, and waist circumference.
Sample size was estimated for the primary end point, namely, CVD. It was reassessed in 2008 and set at 7400 participants with the assumption of a 6-year follow-up and underlying CVD event rates of 8.8% and 6.6% in the control and intervention groups, respectively.12
Our main analyses were performed on an intention-to-treat basis. We used Cox regression models to assess the effect of the intervention on malignant breast cancer incidence. First, we fitted a crude model, and then we adjusted for age (3 groups: ≤60, >60 to 70, and >70 years), baseline body mass index (calculated as weight in kilograms divided by height in meters squared, continuous and quadratic), waist-to-height ratio (dichotomous), use of hormone therapy, leisure-time physical activity (categorized into quartiles), total energy intake (categorized into quartiles), alcohol consumption (dichotomous), age at menopause (dichotomous), smoking habit, type-2 diabetes, use of statins, family history of cancer, baseline adherence to the MeDiet (high vs low) and propensity scores that used 29 baseline variables to estimate the probability of allocation to each intervention group. Multivariable Cox models used robust estimates for the variance to account for intra-cluster correlations in participants who were not individually randomized, considering as clusters the households and each of the clinics of Site D that were allocated in clusters. Models were stratified for recruitment center and educational level. In an ancillary analysis, we merged both MeDiet groups and assessed their effect compared with the control group. For the primary analysis, we excluded 7 women with a non–pathologically confirmed incident breast cancer. In sensitivity analyses, we included these women as cases or as noncases. We repeated our analyses after excluding women who received a diagnosis of malignant breast cancer during the first year of follow-up, considering only malignant neoplasms positive for estrogen receptors (ERs) and excluding all women from site D (n=346) and second members of the same household. We did subgroup analyses stratifying by age, smoking status, alcohol intake, prevalent type 2 diabetes, obesity, use of hormone therapy, family history of cancer, and baseline adherence to the Mediterranean diet. However, the small number of cases in some of the strata precluded fitting the models for some of these subgroups. Analyses were repeated with Poisson regression models with robust estimates for the variance. Finally, we also completed a per-protocol analysis in which we used time-dependent Cox models to assess the association between attained consumption of EVOO during follow-up (cumulative mean across all the available food frequency questionnaires) and subsequent incidence of breast cancer.
From October 2003 through June 2009, 4282 women were randomly assigned to 1 of the 3 intervention groups (eFigure in Supplement 1). Their baseline characteristics are presented in Table 1. The mean (SD) age of participants was was 67.7 (5.8) years, and mean (SD) body mass index was 30.4 (4.1). Most women underwent menopause before 55 years, and less than 3% used hormone therapy. Baseline characteristics were well balanced in the 3 groups.
During a median (SD) follow-up time of 4.8 (1.7) years, we identified 35 confirmed incident cases of malignant breast cancer. Among them, 33 had available information on ER status and 31 were positive. Of 27 cases with information on progesterone receptor status, 21 were positive, and of 21 with information on ERBB2 receptors, 12 were positive. For 122 participants, no information for breast cancer incidence during follow-up was available.
Women allocated to the MeDiet supplemented with EVOO showed a 62% relatively lower risk of malignant breast cancer than those allocated to the control diet (95% CI, 0.16-0.91) (Figure 1). Participants in the MeDiet supplemented with nuts showed a nonsignificant risk reduction compared with women in the control group (hazard ratio [HR], 0.63 [95% CI, 0.28-1.41]). When both MeDiet groups were merged together, we observed a 51% relative risk reduction (95% CI, 0.24-0.98) (Table 2). When we excluded women who received a diagnosis of malignant breast cancer during the first year after enrollment, the results hardly changed. Similarly, the results did not substantially change after including women with breast cancer with no cytological or histological confirmation either as cases or as noncases or when we considered only ER-positive malignant neoplasms. In the stratified analyses, all but 2 point estimates (for MedDiet+nuts versus control among participants with high body mass index and for those with high baseline adherence to MedDiet) showed an inverse association between the MeDiet plus EVOO intervention and the incidence of breast cancer (Table 3).
When we assessed the 3 trial groups together in a per-protocol analysis, participants who attained a higher EVOO consumption during follow-up exhibited the lowest risk (HR for 5th vs 1st quintile, 0.18 [95% CI, 0.06-0.57]) (Figure 2). In these analyses with yearly cumulative updated dietary exposures, the HR was 0.72 (95% CI, 0.57-0.90) for each additional 5% of calories from EVOO.
In this secondary analysis of the PREDIMED trial, we found a significant inverse association between consumption of a MeDiet supplemented with EVOO and breast cancer incidence. A high consumption of EVOO (≥15% of total energy intake) seems to be instrumental for obtaining this significant protection. A nonsignificant risk reduction was observed with the MeDiet supplemented with nuts.
The strengths of this study are its randomized design, the achieved changes in the participants’ dietary habits according to the intervention,17 little residual confounding with almost no changes in estimates after adjustment for many potential confounders, and the thorough and blind revision of medical information to assess outcomes. The adjudication committee, whose members were blinded to the intervention group, assessed the events with specific criteria, dispelling potential misclassification biases. We also acknowledge some limitations. First, breast cancer was not the primary end point of the PREDIMED trial. Thus, the present work is only a prespecified secondary analysis of a large nutritional intervention trial and we cannot warrant that all women had mammograms free from suggestive findings at baseline. However, the randomization was able to yield well-balanced and comparable groups, and, given the large sample size, a balance in other characteristics can be safely assumed. Second, the number of observed breast cancer cases was small. The potential for missing some incident breast cancer cases is basically null regarding clinically relevant events. In any case, this possibility will affect only women lost to follow-up, and most of them belonged to the control group. Therefore, undetected cases of breast cancer would more likely have increased even further the rate in the control group. Accordingly, our results would tend to underestimate the beneficial effect of the intervention. The low rate of breast cancer among women in the PREDIMED trial should not be surprising. If the MeDiet is actually protective against breast cancer, a low incidence is to be expected in a study with these characteristics, especially when overall adherence to the MeDiet was good already at baseline. Third, we do not have information on an individual basis on whether and when women in our trial underwent mammography. Potentially, cancers could be missed without mammograms. However, because of the randomized design and the large sample size, we believe that we can safely assume an even distribution of subclinical cases in the 3 groups under the null hypothesis. Also, we prioritized specificity in our protocol for case ascertainment and we believe that our protocol for confirmation of cases ensures a high degree of specificity.18(p359) Fourth, our participants were white postmenopausal women at high cardiovascular risk. Thus, our results may not be generalizable to other age groups or ethnicities. Fifth, information on reproductive factors known to be associated with breast cancer risk was not available for further adjustment. Nevertheless, because of the randomized allocation of participants, it is not likely that these factors may have introduced substantial confounding. Fifth, our study cannot disentangle whether the observed beneficial effect was attributable mainly to EVOO or to its consumption within the context of the traditional MeDiet. Sixth, according to the study event definitions, we collected information on malignant tumors. Thus, we did not register noninvasive tumors such as in situ tumors. Therefore, we cannot include noninvasive cases in our analyses. Seventh, up to October 2006 when the study protocol was amended,12 the intervention in the control group was less intense than in the intervention group. Consequently, some differences in social support, positive expectations, and empowerment could have existed between the intervention and the control groups. Nevertheless, only 5 cases of breast cancer had been identified up to that date. In addition, it seems unlikely that the magnitude of the risk reduction in breast cancer can be explained only in terms of increased social support.
No prior nutrition intervention trial has addressed the effect of the MeDiet specifically on breast cancer. In the Lyon Diet Heart Study, a randomized trial, a protective effect of a cardioprotective Mediterranean-type diet against overall cancer incidence was observed, supporting the hypothesis of an anticancer effect of the MeDiet.6 The potential beneficial effect of the MeDiet may be explained by several mechanisms,19 for example, a reduction in DNA oxidative damage.20 Specifically for breast cancer, results from observational studies have been inconsistent. A recent meta-analysis21 reported no association between adherence to the MeDiet and breast cancer incidence in cohort studies, revealing a pooled estimate risk ratio of 1.01 (95% CI, 0.88-1.16), whereas results from case-control studies suggested an 18% risk reduction (95% CI, 0.69-0.97). However, most cohort studies8-10,22 included in this meta-analysis were conducted outside the Mediterranean geographical area and it cannot be assumed that a proper MeDiet was followed outside this region. The EPIC study is the only large cohort study that has included countries from the Mediterranean area.7 In that study, the HR for postmenopausal women was 0.93 (95% CI, 0.87-0.99) when comparing high (10-16 points) vs low (0-5 points) adherence to the MeDiet. If this information (as a prior) were integrated with our present results using a simple Bayesian approach recommended in epidemiology,18 the posterior relative risk would be 0.92 (95% CI, 0.87-0.98). The differential effects of the 2 MeDiet interventions on breast cancer may be attributed to a higher consumption of EVOO among participants allocated to the MeDiet supplemented with EVOO as our ancillary analyses showed (Figure 2). Consumption of EVOO accounted for 22% of total caloric intake in the MeDiet supplemented with EVOO, whereas nuts represented 10% of the total calories in the MeDiet supplemented with nuts. The stronger inverse association with EVOO consumption may also be ascribed to its high polyphenol content.
Epidemiological studies on the association between EVOO consumption and breast cancer incidence are scarce. A meta-analysis of case-control studies concluded that olive oil consumption, including not only EVOO but also other common types of olive oil (with a lower content of bioactive polyphenols), was inversely associated with breast cancer incidence.23 These case-control studies have been conducted in Mediterranean countries, and they consistently found an inverse association between olive oil consumption and breast cancer risk. This finding, however, was not replicated in the EPIC cohort.24 Nonetheless, it is noteworthy that none of these studies differentiated between types of olive oil. Several biological mechanisms could explain the putative anticarcinogenic properties of EVOO. All types of olive oil provide a high supply of monounsaturated fatty acids, mainly oleic acid, as well as squalene, whereas EVOO also contains various biologically active compounds, such as the polyphenols oleocanthal, oleuropein, hydroxytyrosol, and lignans. In vitro studies have suggested that oleic acid has an antiproliferative effect by affecting the expression of human oncogenes.25 The hydrocarbon squalene has been reported to exert a beneficial effect on intracellular oxidative stress and DNA oxidative damage in mammary epithelial cells.26 Olive oil polyphenols may have a potential role in breast cancer prevention.27 Oleocanthal has been associated with inhibition of tumor growth and proliferation, migration, and invasiveness of breast cancer cells in in vitro or in vivo breast cancer models.28 Oleuropein has been associated with increased apoptosis of cultured breast cancer cells through different pathways.29,30 Also, hydroxytyrosol has been reported to reduce intracellular reactive oxygen species in human breast epithelial cells and to prevent oxidative DNA damage in both human breast epithelial cells and human breast cancer cells.31 Lignans are phytoestrogens whose consumption has been associated with a lower risk of breast cancer in postmenopausal women.32
In the PREDIMED trial, participants in the control group did not reduce their total fat intake substantially—albeit their saturated fat intake stayed less than 10% during follow-up—even though they were advised to follow a low-fat diet. This result can be ascribed to the rooted tradition of adherence to the MeDiet, particularly among older people. On the other hand, several prospective studies have suggested that higher fat intakes, especially animal fat, may be associated with a higher risk of breast cancer.33-36 Moreover, in the Women’s Health Initiative study, total fat consumption was associated with a higher risk of breast cancer.37 Also, in the Women’s Health Initiative study, women who reported the highest levels of fat intake at baseline and therefore may have achieved the greatest reduction in fat intake showed a significantly lower risk of breast cancer.38 Among women with early-stage breast cancer in the Women's Intervention Nutrition Study, lower fat intakes were associated with lower estrogen-negative breast cancer recurrence.39 Taking all this evidence into account, greater reductions in the incidence of breast cancer could have been observed in the control group had these women followed a truly low-fat diet.
The results of the PREDIMED trial suggest a beneficial effect of a MeDiet supplemented with EVOO in the primary prevention of breast cancer. Preventive strategies represent the most sensible approach against cancer. The intervention paradigm implemented in the PREDIMED trial provides a useful scenario for breast cancer prevention because it is conducted in primary health care centers and also offers beneficial effects on a wide variety of health outcomes.40 Nevertheless, these results need confirmation by long-term studies with a higher number of incident cases.
Correction: This articles was corrected on November 5, 2018, to correct protocol deviations and present reanalyses of the PREDIMED Trial, and was corrected on November 2, 2015, for an error in the units for physical activity in Table 1.
Corresponding Author: Miguel A. Martínez-González, MD, Department of Preventive Medicine and Public Health, University of Navarra–School of Medicine, Edificio de Investigación, 2a planta, C/ Irunlarrea, 1, 31008 Pamplona, Navarra, Spain (firstname.lastname@example.org).
Published Online: September 14, 2015. doi:10.1001/jamainternmed.2015.4838
Author Contributions: Drs Toledo and Martínez-González had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Salas-Salvadó, Buil-Cosiales, Estruch, Ros, Corella, Hu, Serra-Majem, Basora, Sorlí, Martínez-González.
Acquisition, analysis, or interpretation of data: Toledo, Salas-Salvadó, Donat-Vargas, Buil-Cosiales, Estruch, Corella, Fitó, Hu, Arós, Gómez-Gracia, Romaguera, Ortega-Calvo, Serra-Majem, Pintó, Schröder, Bulló, Serra-Mir, Martínez-González.
Drafting of the manuscript: Toledo, Donat-Vargas.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Toledo, Donat-Vargas, Romaguera, Serra-Majem, Martínez-González.
Obtained funding: Salas-Salvadó, Estruch, Ros, Corella, Gómez-Gracia, Serra-Majem, Martínez-González.
Administrative, technical, or material support: Salas-Salvadó, Estruch, Corella, Fitó, Hu, Serra-Majem, Schröder, Bulló, Serra-Mir, Martínez-González.
Study supervision: Salas-Salvadó, Estruch, Corella, Fitó, Gómez-Gracia, Ortega-Calvo, Pintó, Sorlí, Martínez-González.
Conflict of Interest Disclosures: Dr Salas-Salvadó received grants from Instituto de Salud Carlos III and the International Nut and Dried Fruit Foundation during the conduct of the study and has received consultancy fees from Danone Research and Eroski outside the present work; he is also a nonpaid member of the Scientific Committee of the International Nut and Dried Fruit Foundation. Dr Estruch, outside the present work, has received grants from the California Walnut Commission, nonfinancial support from Patrimonio Comunal Olivarero (Spain), La Morella Nuts (Spain), and Borges SA (Spain) and grants from Novartis Farmaceutica SA, Cerveceros De España, Sanofi, FIVIN-Spain, Instituto Cervantes Albuquerque, the Harvard School of Public Health, the Culinary Institute of America, the International Family Doctors Association, and Instituto Cervantes (Milan, Italy). Dr Ros, during the conduct of the study, received grants from Instituto de Salud Carlos III and nonfinancial support from the California Walnut Commission, Patrimonio Comunal Olivarero, La Morella Nuts, and Borges SA. Dr Ros, outside the present work, has received grants from the California Walnut Commission and nonfinancial support from Nuts for Life, La Asturiana SA, the International Nut and Dried Fruit Council, and the California Walnut Commission and personal fees and other compensation from Nuts for Life and La Asturiana SA. Dr Hu, outside the present work, has received grants from the California Walnut Commission and Metagenics. Dr Fitó, outside the present work, has received personal fees from Menarini and AstraZeneca. No other disclosures are reported.
Funding/Support: The supplemental foods used in the study were generously donated by Patrimonio Comunal Olivarero and Hojiblanca, Spain (EVOO); the California Walnut Commission, Sacramento, California (walnuts); and Borges SA (almonds) and La Morella Nuts (hazelnuts), both from Reus, Spain. The PREDIMED trial was supported by the official funding agency for biomedical research of the Spanish government (Instituto de Salud Carlos III) through grants provided to research networks specifically developed for the trial: RTIC G03/140 (Coordinator: R.E.) and RTIC RD 06/0045 (Coordinator: M.A.M.-G.). All investigators of the PREDIMED trial belong to Centro de Investigación Biomédica en Red (CIBER), an initiative of Instituto de Salud Carlos III. We also acknowledge grants from the National Institutes of Health (1R01HL118264-01 and 1R01DK102896); Fondo de Investigación Sanitaria–Fondo Europeo de Desarrollo Regional (PI04/0233, PI05/0976, PI07/0240, PI10/01407, PI10/02658, PI11/00049, PI11/02505 and AGL2010-22319-C03-03); Consejería de Salud de la Junta de Andalucía (PI0105/2007); and the Generalitat Valenciana, Spain (ACOMP/2013/165 and ACOMP/2013/159).
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We thank the participants in the trial for their enthusiastic and continuous collaboration, and the Independent Data and Safety Monitoring Board (F. Xavier Pi-Sunyer, MD, Columbia University; Carlos Alberto González, MD, Catalan Insitute of Oncology; Joan Sabaté, MD, DrPH, Loma Linda University), and the collaborators at the different study sites. None of them received any compensation.