After lung, breast cancer is the second most common neoplasm worldwide. In women, it accounts for almost 1 in 4 cancer cases and is the first cause of cancer death in most countries [1
]. Primary prevention through favorable changes in modifiable risk factors is however a major challenge and a public health priority [3
Thus far, only very few modifiable risk factors for breast cancer have been recognized, including overweight/obesity, menopausal hormone replacement therapy (HRT), low physical activity and sedentary behavior, with possible differential associations in the life course of women [4
]. Despite considerable research on the dietary correlates of breast cancer, no consistent evidence exists for specific foods, food groups or nutrients [6
], except for the detrimental role of alcohol drinking [7
]. Nonetheless, healthy dietary patterns, mainly based on frequent intake of plant-based foods, including the Mediterranean diet [8
], have been favorably related to breast cancer risk in several studies [10
In 2007, the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) published the Second Expert Report, in which defined a set of evidence-based public health recommendations on body fatness, physical activity and diet with the aim of reducing the burden of cancer [13
]. In the Third Expert Report, published in 2018, guidelines have been updated according to the latest evidence [14
]. While the 2007 and 2018 conclusions and global recommendations look similar, in the most recent report emphasis was placed on the potential benefit of considering recommendations as an integrated pattern of healthy lifestyle behaviors, which taken together have an impact on cancer prevention [15
Previous studies showed that higher adherence to the 2007 WCRF/AICR nutrition-related recommendations reduced the risk of total cancer and of specific cancers [16
]. An inverse association with breast cancer was reported in most—though not all [17
]—studies, including the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort [19
], the Iowa Women’s Health Study (IWHS) [20
], the Black Women’s Health Study (BWHS) [21
], the Swedish Mammography Cohort (SMC) [22
], the Canadian National Breast Screening Study (NBSS) [23
], and the Vitamins and Lifestyle (VITAL) cohort [24
]. Data are, however, still limited and further quantification is needed. Moreover, no study has yet focused on the most updated version of the WCRF/AICR recommendations.
In the current report, we assessed the relation between adherence to the WCRF/AICR nutrition-related recommendations for cancer prevention and breast cancer risk in a large Mediterranean population, and conducted a meta-analysis of available studies on the topic.
The distribution of selected characteristics in breast cancer cases and controls is shown in Table 1
. Compared to controls, cases were more educated, had lower parity and had more frequently a first-degree relative with breast cancer.
Adherence to the recommendations on physical activity, wholegrains, vegetables, fruit and beans, fast foods and other processed food high in fat starches or sugar, sugar sweetened drinks, and alcohol was associated with a significant lower risk of breast cancer, with significant decreasing trends for increasing adherence (Table 2
). Adherence to the recommendation on red and processed meat was significantly inversely associated with breast cancer in the analyses adjusted for age, centre, and education, while results were marginally significant in fully adjusted analyses. No association was found for adherence to the recommendations on body weight and breastfeeding.
provides the ORs of breast cancer according to the overall 2018 WCRF/AICR score and the diet-specific score. There was an inverse association with the WCRF/AICR adherence score, with an OR of 0.60 (95% CI: 0.51–0.70) for a score ≥5.5 vs. ≤4.25 in the fully adjusted analysis (p for trend < 0.001). The results were consistent according to menopausal status (Supplementary Table S2
). The inverse association was observed in both countries, but was apparently stronger in Switzerland. The ORs for successive score categories were 0.81 (95% CI: 0.69–0.95), 0.83 (95% CI: 0.72–0.97) and 0.74 (95% CI: 0.62–0.88) in Italy, and 0.55 (95% CI: 0.38–0.80), 0.61 (95% CI: 0.44–0.86) and 0.23 (95% CI: 0.15–0.34) in Switzerland. A 1-point increment in the score decreased breast cancer risk by 17% (OR = 0.83, 95% CI: 0.79–0.88). According to the estimated population attributable fraction, 25% of breast cancers could be avoided in our population if all participants would shift towards the highest adherence category. The score based on dietary recommendations only was significantly inversely related to breast cancer risk, too (OR = 0.62, 95% CI: 0.53–0.73 for a score >3.5 vs. <2.25, p for trend < 0.001).
Similar results (but apparently less strong) were observed when analyzing the 2007 version of the score (Supplementary Table S3
). The ORs were 0.67 (95% CI, 0.56–0.79) for the highest vs. the lowest score category (p for trend < 0.001) and 0.85 (95% CI, 0.80–0.90) for a 1-point increment in the score.
In a sensitivity analysis, consistent results were found when omitting each recommendation one by one from the score: the ORs comparing extreme categories varied between 0.54 and 0.69. In particular, the OR was 0.56 (95% CI, 0.44–0.72) when removing from the score the recommendation on breastfeeding.
The description of the studies included in the meta-analysis is provided in Table 4
. Based on 13 studies, including the present one, the pooled RR of breast cancer for the highest vs. the lowest category of the WCRF/AICR score was 0.73 (95% CI, 0.65–0.82, I2
= 73.5%) (Figure 1
panel A). Summary estimates were similar for case–control (RR 0.68, 95% CI, 0.51–0.92, p heterogeneity among studies = 0.001, I2
= 81.7%, 4 studies) and cohort studies (RR 0.75, 95% CI, 0.66–0.85, p heterogeneity = 0.004, I2
= 64.8%, 9 studies), and for pre- (RR 0.70, 95% CI, 0.47–1.03, 5 studies) and post-menopausal women (RR 0.67, 95% CI, 0.56–0.80, 8 studies). When we included for the study by Nomura et al. [21
] the results based on the time-varying score rather than those based on the baseline score, the pooled RR became 0.72 (95% CI, 0.64–0.81). In the sensitivity analysis, no single study appreciably influenced the summary results, with the pooled RR comparing the extreme score categories ranging from 0.71 to 0.75 (significant) when omitting individual studies one at a time.
In the linear dose-response meta-analysis, each 1-point increase in the score was associated to a 9% (95% CI, 6%–12%, I2
= 70.2%) RR reduction, based on 16 studies (15 publications) including the present one (Figure 1
, panel B). The pooled RRs were 0.93 (95% CI, 0.91–0.95, p heterogeneity = 0.022, I2
= 49.4%) for cohort and 0.83 (95% CI, 0.80–0.87, p heterogeneity = 0.847, I2
= 0%) for case–control studies. The exclusion of each study in turn did not appreciably influence the summary result, and the estimated RRs were always significant. In particular, the pooled RR was still 0.91 (95% CI, 0.88–0.94) when omitting the pooled analysis by Jancovic et al. [39
], which considered only dietary recommendations, and 0.90 (95% CI, 0.87–0.93) when omitting the study by Van de Brandt [8
], which reported the RR for 1-SD increase in a dietary WCRF/AICR score.
Since all previous studies assessed adherence to the 2007 WCRF/AICR recommendations, as sensitivity analysis, we replaced in the meta-analyses our results for the 2018 score with those for the 2007 score. We obtained very similar summary estimates (RR for the extreme quantile meta-analysis: 0.74, 95% CI, 0.66–0.83, RR for the linear dose-response meta-analysis: 0.91, 95% CI, 0.89–0.94).
No significant publication bias was detected (p = 0.143).
In the present large case–control study, higher adherence to the diet, adiposity, and physical activity recommendations provided by the WCRF/AICR Third Expert Report was associated with reduced breast cancer risk, independently of menopausal status. The results from the sensitivity analysis suggest that the association was not driven by adherence to one specific recommendation, but rather by the combined and synergic effects of the various score components. The meta-analysis indicates that women who highly adhere to the recommendations had an approximately 30% lower risk of breast cancer compared to those with low adherence and that each 1-point increment in the WCRF/AIRC score is associated with a significant 9% reduced risk.
Published studies vary in design, baseline population characteristics, data collection, number and definition of recommendations operationalized in the score, and adjustment factors; the observed heterogeneity is therefore not surprising. In any case, they are largely supportive of the preventing role of following the WCRF/AICR guidelines on breast cancer. Indeed, estimates of the RR of breast cancer comparing extreme score categories were below unity in all but one studies [18
], and were significant in seven out of the 12 previous studies. The favorable role of adherence to the recommendations was evident even when investigating the association continuously, with significantly reduced RRs in nine out of 15 previous studies. RRs were below unity in all the remaining five studies except for one [8
Only a few reports investigated the association between the 2018 release of the WCRF/AICR recommendations and cancer risk [42
]. The present study is the first evaluating the updated recommendations in the prevention of invasive breast cancer. The 2007 and 2018 sets of recommendations were similar, and consequently major differences in terms of associated cancer risk reductions are unlikely. The last release of the guidelines recommends the avoidance of any alcohol (unlike the previous version, which allowed moderate consumption), included avoidance of sugar-sweetened drinks as a separate recommendation, and removed the 2007 recommendation on limiting salt intake and avoid moldy cereals or pulses from the global set of recommendations (moving it into a regional and special circumstances section, based on the fact that salt-preserved food is mostly consumed by people without access to refrigeration and evidence on salt-preserved foods consumption as a risk factor for cancer derived mainly from selected Asian populations). Alcohol has been directly associated to breast cancer even at low amounts [49
]. According to the 2018 alcohol recommendation, we observed a breast cancer risk reduction of 26% among non-drinkers (adherent) vs. women drinking over seven drinks per week. When evaluating the earlier recommendation, the protection associated to adherence (defined as ≤7 drinks/week) was slightly lower, i.e., 21% (OR: 0.79, 95% CI, 0.68–0.92 vs. >14 drinks/week). Thus, adherence to the stricter updated recommendation has the potential to further reduce the incidence of breast cancer. Conversely, since salt-preserved food is uncommon among Western populations and its consumption has been unfavorably related to stomach cancer only, the change in the corresponding recommendation is likely to have a negligible impact on breast cancer risk reduction. The association of sugar-sweetened drink consumption with breast cancer is limited [50
] (although further quantification is needed). It is difficult to assess the added benefit of the updated recommendation on breast cancer in our database, since sugar-sweetened drinks consumption was relatively uncommon in our population at the time of data collection. We found similar results when using scores measuring overall adherence to the 2007 and 2018 guidelines, and our findings are largely in agreement with previous studies relying on recommendations from the earlier WCRF/AICR report. Along this line, recent findings on the association between the 2018 WCRF/AICR recommendation and colorectal cancer risk from the Nurses’ Health Study and Health Professionals Follow-up Study confirmed the results of prior studies based on the 2007 guidelines release [47
Our findings are in agreement with the conclusions of a systematic review [16
] and with studies assessing compliance to the American Cancer Society (ACS) Nutrition and Physical Activity Cancer Prevention Guidelines (which largely overlap with those from the WCRF/AICR), which reported lower risks of breast cancer [23
] and other selected cancers [52
] for higher scores of guideline adherence. Recently, data from the EPIC cohort showed a favorable role of adherence to the 2018 WCRF/AICR recommendations on in situ breast cancer [42
]—a precursor of invasive breast cancer, with which it shares some of the risk factors [54
]—in the subgroup of women enrolled through screening programs and with high screening participation during follow-up.
In our case–control study, we observed a significant inverse association between a score reflecting adherence to the WCRF/AIRC dietary recommendations and breast cancer, in agreement with the study by Lavalette et al. [36
], which reported an hazard ratio (HR) of 0.83 (95% CI, 0.74–0.93) for 1-point increment in a WCRF/AICR score without the body fatness and physical activity components (as well as the component relying on breastfeeding). Other studies found non-significant or borderline significant inverse associations [21
] or null associations with dietary WCRF/AICR scores [8
]. Despite the large number of investigations, evidence on specific aspects of diet (single foods/nutrients, food groups, and dietary components) and breast cancer is still open to discussion, with the exception of the widely recognized detrimental role of alcohol [6
]. According to the Continuous Update Project (CUP) by the WCRF/AICR, only limited evidence exists that the consumption of non-starchy vegetables, dairy products, and foods containing carotenoids and calcium decreases the risk of pre-menopausal and/or post-menopausal breast cancer [55
]. Meanwhile, studies investigating healthy dietary patterns, such as the Mediterranean diet [8
], have generally showed inverse associations [10
], pointing to the importance of the overall diet quality as compared to individual foods/food components in breast cancer prevention. Dietary patterns capture the complexity of the diet, account for the interactions among dietary factors, and are more predictive of disease risk when single foods/food components have modest health effects [56
To improve the comparability of our results with those from future studies relying on the 2018 WCRF/AICR guidelines, in the construction of the score, whenever possible, we followed the standard scoring approach recently proposed by a NCI-led collaborative group, which included, among the others, researchers from AICR and WCRF International [31
]. However, (1) we did not include information on waist circumference in the assessment of the recommendation on body weight, as data were incomplete (not available for 173 cases and 381 controls); (2) we adapted the definition of no, partial, and full adherence to the recommendation on physical activity according to the information collected from our questionnaire; (3) we used energy-density as a proxy of the % of total energy intake from ultra-processed foods (we did not collect specific information on ultra-processed foods, as their consumption was relatively infrequent in our population at the time of data collection); and (4) we could not distinguish between exclusive and partial breastfeeding. As for point (1), when we used data on both waist circumference and BMI in the sample of subjects with information on both factors we obtained very similar results (OR for full vs. no adherence to the recommendation on body weight = 1.03, 95% CI, 0.89–1.20, OR for the highest vs. the lowest WCRF/AICR score = 0.56, 95% CI, 0.48–0.66).
The current analysis is based on data collected between 1991 and 2008 while looking at associations with more recent recommendations. Thus, the observation that the protection associated to the adherence to the WCRF/AICR recommendations was found in a population unaware of those guidelines supports the absence of major information bias and hence the validity of the association. The data were collected at multiple centres across two countries. However, similar structured questionnaires and coding manuals were used in all study centres, and interviewers were centrally trained, thus improving the reliability of data collection across study centers. In addition, a comparison across countries indicated a similar distribution of age, menopausal status, and family history of breast cancer, while Swiss women were more educated and more often nulliparae than Italian ones [9
]. A favorable role on breast cancer of adherence to the guidelines was observed in Italian and Swiss centers, although apparently stronger in the latter’s, supporting the consistency of the observed association.
Hospital controls may have different dietary habits from those of the general population, but we minimized bias by excluding from the control series patients admitted to hospital for conditions which may have led to long-term diet modifications. The strengths of the present case–control study lie in the satisfactory reproducibility and validity of the FFQ [27
], the similar catchment area of cases and controls, the almost complete participation rate, and the large sample size. However, measuring usual diet is challenging and some degree of exposure misclassification cannot be excluded. Since weight and height were self-reported, BMIs are likely underestimated. However, information bias is likely to be similar in cases and controls. In addition, although we adjusted our risk estimates for several exposures, including hormone-related factors, diabetes, family history of breast cancer and total energy intake, some residual confounding is still possible.
In conclusion, the present work provides quantitative evidence that greater adherence to the nutrition-related WCRF/AICR recommendations is favorably related to breast cancer risk, and thus suggests that encouraging adherence to the guidelines is a valuable public health strategy for breast cancer prevention.