Diet, weight management, physical activity and Ovarian & Breast Cancer Risk in women with BRCA1/2 pathogenic Germline gene variants: systematic review

Introduction Women with pathogenic germline gene variants in BRCA1 and/or BRCA2 are at increased risk of developing ovarian and breast cancer. While surgical and pharmacological approaches are effective for risk-reduction, it is unknown whether lifestyle approaches such as healthful dietary habits, weight management, and physical activity may also contribute to risk-reduction. We conducted a systematic review of evidence related to dietary habits, weight status/change, and physical activity on ovarian and breast cancer risk among women with BRCA1/2 pathogenic variants. Methods We searched Medline, EMBASE, CENTRAL, PubMed, and clinicaltrials.gov up to October 3, 2019. We identified 2775 records and included 21. Results There is limited evidence related to these factors and ovarian cancer risk. For breast cancer risk, evidence suggests higher diet quality, adulthood weight-loss of ≥10 pounds, and activity during adolescence and young-adulthood may be linked with decreased risk. Higher meat intake and higher daily energy intake may be linked with increased risk. Conclusions There is not enough evidence to suggest tailored recommendations for dietary habits or weight management among women with BRCA1/2 pathogenic variants compared to the general population for ovarian and breast cancer risk-reduction, and physical activity recommendations should remain the same.

(PA) may also contribute to cancer risk-reduction among this group of high-risk women.
To date, there are five reviews evaluating the impact of either dietary habits, weight management, or PA, or a combination of only two of these factors (i.e. diet and weight), on ovarian and/or breast cancer risk among women with BRCA1/2 pathogenic germline gene variants from 1997 to 2015 [3][4][5][6][7]. Four of the five previous reviews included women with BRCA1 and BRCA2 pathogenic germline gene variants, but only in the context of breast cancer risk [3,[5][6][7]. And only one review assessed ovarian cancer risk, and this was only in relation to alcohol intake [6]. No studies have exclusively evaluated healthful dietary habits, weight management, and PA together as they relate to both ovarian and breast cancer risk in this high-risk population.
The purpose of the current systematic review was to explore the state of evidence related to these lifestyle factors and ovarian and breast cancer risk among women with BRCA1/2 pathogenic germline gene variants, in order to determine the extent to which lifestyle recommendations should differ compared to the general population.

Search strategy
The search terms and search strategy were developed by four authors (AMC, LG, KBE, KJK), one of whom (KJK) is a medical research librarian specializing in systematic reviews. A systematic search was performed in MED-LINE, EMBASE, Cochrane Library, and ClinicalTrials. gov from inception to October 3, 2019. Search structures, subject headings, and keywords were tailored to each database by KJK. The search was expanded through citation chaining (forward and backward) of included studies. The search terms used can be found in the MEDLINE search strategy [see Additional File 1]. Findings are reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist [8] (Fig. 1). The protocol is registered in PROSPERO(ID:CRD42017060007).

Review process, selection Criteria & Data Extraction
Two authors (AMC, LG) independently screened the titles and abstracts of the articles to identify potentially relevant studies. Studies that passed the title/abstract review were retrieved for full-text review. Disagreements were resolved by consensus and by seeking the opinion of a third author (KBE). Inclusion criteria consisted of studies that: included individuals with BRCA1/2 pathogenic germline gene variants; evaluated weight status, weight change, dietary habits (as defined by dietary patterns, food and beverage intake, multivitamin and mineral supplementation), or physical activity in relation to ovarian or breast cancer risk; published in English; and included human subjects only.

Risk-of-Bias assessment
The Quality Assessment Tool for Quantitative Studies, version 2010, was used for risk-of-bias assessment [9,10]. Six components were evaluated to determine overall study quality: selection bias, design, confounders, blinding, data collection method, withdrawals/dropouts. Quality scores were assigned per criteria described elsewhere [10].

Analysis
Qualitative synthesis of data is provided in narrative form. A meta-analysis was not conducted due to the limited number of studies and heterogeneity in study design and outcome measures. Table 1 provides details of the risk of bias assessment for all studies. Four studies received an overall quality score of strong [11][12][13][14], 16 received moderate [15][16][17][18][19][20][21][22][23][24][25][26][27][28][29], and one received weak [30]. Table 2 provides study characteristics and results of the studies reviewed for both ovarian and breast cancer. Gronwald and colleagues [30] case-control study did not observe a significant association between coffee consumption and ovarian cancer risk (OR 0.7, 95%CI 0.4, 1.3) among 348 women with BRCA1 pathogenic germline gene variant [30]. Information regarding quantity of coffee consumed or statistical adjustments was not provided. No studies assessed PA.

Ovarian Cancer risk
Regarding weight status and weight change, after adjusting for age at menarche, parity, oral contraceptives, height, and hormone replacement therapy, McGee and colleagues [21] case-control study did not observe a significant association between current weight status or weight change throughout adulthood and diagnosis of ovarian cancer among 403 women with BRCA1 and 66 women with BRCA2 pathogenic germline gene variants. Risk was not assessed by BRCA1/2 pathogenic germline gene variants individually. In contrast, Qian and colleagues [28] case-control study observed significant associations between a higher body mass index (BMI) and premenopausal ovarian cancer incidence, for both selfreported BMI (among 102 cases out of 7516 women with BRCA1/2 pathogenic germline gene variants) and a calculated BMI genetic score (BMI-GS; among 967 cases out of 22,588 women with BRCA1/2 pathogenic germline gene variants) based on a Mendelian Randomization approach. Higher self-reported BMI was also associated with increased risk of non-serous ovarian tumors [28]. Significant associations were not observed by BRCA1/2 pathogenic germline gene variants, postmenopausal status, or serous tumor type [28].
In contrast, case-control studies by both Lecarpentier and colleagues [19] (n = 863, adjusted for parity, menopausal status, BRCA1/2 pathogenic germline gene variant, smoking history) and McGuire and colleagues [22] (n = 497, adjusted for age, family history, smoking status, and full-term pregnancies), and a prospective cohort study by Cybulski and colleagues [15] (n = 2498 adjusted for baseline age, gene, menarche age, oral contraceptive use, breast feeding history, mean parity, oophorectomy status, and resident country) did not observe significant associations between alcohol intake and breast cancer risk among women with BRCA1 pathogenic germline gene variant. Similarly, Moorman and colleagues [23] case-only study observed a significant, but weak effect of alcohol intake among 283 breast cancer survivors with  effects were not observed for all alcohol including wine, or wine alone [11]. In contrast, other studies did not observe effects among women with BRCA2 pathogenic germline gene variant [15,19,23].

Dietary habits-coffee consumption
In Gronwald and colleagues [30] case-control study, no association was observed between coffee consumption and breast cancer risk among 348women with BRCA1 pathogenic germline gene variant (OR 0.8, 95%CI 0.5, 1.1). Amount of coffee consumption and statistical adjustments for the analysis were not specified. In contrast, Nkondjock and colleagues [24] case-control study observed an association between ≥6 cups caffeinated coffee/day among 652women with BRCA1 pathogenic germline gene variant and breast cancer risk (OR 0.25, 95%CI 0.09,0.71). Associations were not observed for women with BRCA2 pathogenic germline gene variant (n = 193). When assessing BRCA1/2 pathogenic germline gene variants collectively, ≥6 cups/day of total coffee (caffeinated and decaffeinated; OR 0.51, 95%CI 0.26, 0.98) and ≥ 6 cups/day of caffeinated coffee (OR 0.31,                 *Data adjusted for use of oral contraceptives, parity, menopausal status, hormone replacement therapy use, age-specific BMI, BMI at age 18, alcohol consumption, occupational activity. Mean METhours/week and mean hours/week also adjusted for number of active years. Number of active years also adjusted for mean METhours/week. OR odds ratio, CI confidence interval, HR hazard ratio, COR case-only odds ratio, IRR interaction risk ratio, # pounds, MET metabolic equivalents 95%CI 0.13,0.71) was associated with lower risk [24]. Associations were not observed for lower levels of coffee consumption or for any level of decaffeinated coffee [24]. All aforementioned results were adjusted for parity, smoking status, oral contraceptive use, alcohol and BMI at 30-years.
Dietary habits-food/nutrient intake Ko and colleagues [12] retrospective cohort assessed dietary intake of vegetables, fruit, meat, seafood, and soybean products, in relation to breast cancer risk among 491 women with BRCA1/2 pathogenic germline gene variants collectively and by pathogenic germline gene variant, and adjusted for menarche, caloric intake, years of education, smoking history, alcohol intake, exercise habits and parity. Regarding women with BRCA1/2 pathogenic germline gene variants collectively, no association was observed with vegetable, fruit, or seafood intake. Intake of 3-10 meat food-items/day was linked with nearly doubling breast cancer risk (HR 1.97, 95%CI 1.13,3.4; p-trend = 0.026) [12]. An inverse relationship was observed for soy and breast cancer risk (HR 0.39, 95%CI 0.19,0.79, p-trend = 0.005) [12]. When assessed by BRCA1/2 pathogenic germline gene variant, a significant positive relationship was observed for meat intake among women with BRCA2 pathogenic germline gene variant (p-trend = 0.027), yet only the highest quartile of meat intake was associated with risk (HR 2.48, 95%CI 1.26,4.89). Moreover, an inverse relationship was observed for soy (p = 0.005), but only the highest quartile of soybean-product intake was associated with reduced risk (HR 0.38, 95%CI 0.16,0.93) [12]. Significant associations were not observed in women with BRCA1 pathogenic germline gene variant. Case-control studies conducted by Nkondjock and colleagues [31] and Kim and colleagues [27] included assessment of nutrients. Nkondjock and colleagues [31] assessed macro/micronutrient intake, alcohol, and coffee and found that among 89 women with BRCA1/2 pathogenic germline gene variants collectively, total energy intake > 2339 kcals/day was associated with nearly tripling breast cancer risk (HR 2.76, 95%CI 1.10,7.02; p-trend = 0.026), when adjusting for age, maximum lifetime BMI, and PA [31]. Analysis was not conducted by variant. Kim and colleagues [27] assessed folic acid, B6, and B12 supplementation and observed that ever-use of a prenatal supplement was associated with reduced likelihood of breast cancer for 400 women with BRCA1/2 pathogenic germline gene variants (OR 0.57, 95%CI 0.34,0.95), when adjusting for age and BRCA1/2 pathogenic germline gene variant. When adjusting for age, BRCA1/2 pathogenic germline gene variant, BMI, parity, alcohol consumption and smoking status, consumption of any folic-acid containing supplement, 8.56-89.29mcg/d of folic acid supplementation, and 0.02-0.34 mcg/d of B12 supplementation, was associated with reduced likelihood of breast cancer among women with BRCA1/2 pathogenic germline gene variants [27]. When stratified by BRCA1/2 pathogenic germline gene variant, significant associations were only revealed for ever-use of any folicacid containing supplement among women with BRCA1 pathogenic germline gene variant [27].
Nkondjock and colleagues [14] conducted a second case-control analysis assessing diet quality among 89 women with BRCA1/2 pathogenic germline gene variants collectively. The four diet quality indexes utilized were indicative of dietary patterns and included the following: Alternative Healthy Eating Index(AHEI), Diet Quality Index-Revised(DQI-R), Alternate Mediterranean Diet Index(aMED), Canadian Healthy Eating Index(-CHEI). An inverse relationship with breast cancer risk was observed for the DQI-R(p-trend = 0.034) and CHEI(p-trend = 0.006); however, only the highest tertile of CHEI was significantly associated with lower breast cancer risk (OR 0.18, 95%CI 0.05,0.68) after adjusting for age, PA and total energy intake [14].

Weight status/change
Studies assessing weight management evaluated adulthood and young-adulthood weight status, and adulthood weight change. Moorman and colleagues [23] did not observe effects between BMI one year before diagnosis or at age 18 and breast cancer risk. Among retrospective cohorts, Manders and colleagues [20] observed, among 218 women with BRCA1/2pathogenic germline gene variants, that risk doubled when current weight was ≥72 kg (HR 2.10, 95%CI 1.23,3.59). These findings were observed in a time-varying Cox-proportional hazard model stratified by gene and birth cohort, clustered for family, and adjusted for parity, menopausal status, hormone replacement therapy, and lifetime sports activity [20]. Among 104 Ashkenazi Jewish women, normal BMI at menarche and 21-years significantly delayed age of onset of breast cancer after adjusting for pro-band decade of birth [17]. Nkondjock and colleagues [31] observed that women with BRCA1/2 pathogenic germline gene variants who experienced their maximum BMI at > 43-years were at a nearly 3-fold increased risk of breast cancer ( [29]. For selfreported BMI, when stratified by BRCA1/2 pathogenic germline gene variant, adulthood BMI was inversely associated with breast cancer risk for women with BRCA2 pathogenic germline gene variant, but not BRCA1 pathogenic germline gene variant, and young-adulthood BMI was inversely associated with risk among women with BRCA1 and BRCA2 pathogenic germline gene variants [29]. Similar results were observed when stratified by menopausal status, such that significant inverse associations were observed for adulthood BMI, BMI-GS and premenopausal breast cancer risk, and young adulthood BMI and both pre-and postmenopausal breast cancer risk [29]. Regarding adulthood weight gain, Nkondjock and colleagues [31] observed a positive relationship for weight gain since age 18 (p-trend = 0.011) and 30 (p-trend = 0.013) and breast cancer risk. Women with BRCA1/2 pathogenic germline gene variants who gained 12-35 pounds since age 18 exhibited 3.6-fold increased risk (OR 3.63, 95%CI 1. 18,11.22) and women who gained > 35 pounds exhibited 4.6-fold increased risk (OR 4.64, 95%CI 1.52,14.12) [31]. Since age 30, women who gained 9-20 pounds presented 3.4-fold increased risk (OR 3.43, 95%CI 1. 16,10.14), and women who gained > 20 pounds displayed 4-fold increased risk (OR 4.11, 95%CI 1.46, 11.56) [31]. In contrast, Kotsopoulos and colleagues [18] case-control study did not observe a significant relationship for weight gain between ages 18-30 among 1073 women with BRCA1/2 pathogenic germline gene variants [31]. Additionally, Manders and colleagues [20] did not observe a relationship between adult weight change and pre-or post-menopausal breast cancer risk.
Interestingly, Kotsopoulos and colleagues [18] considered adulthood weight loss between ages 18-30. A significant association was observed between loss ≥10 pounds and decreased breast cancer risk (OR 0.66, 95%CI 0.46,0.93) [18]. When assessed by BRCA1/2 pathogenic germline gene variant, a significant association was observed between weight loss of ≥10 pounds and reduced risk of breast cancer among women with BRCA1 but not BRCA2 pathogenic germline gene variant [18].

Physical activity
Studies assessing PA, evaluated activity across varying time periods, among women with BRCA1/2 pathogenic germline gene variants collectively. Among Ashkenazi Jewish women, engagement in PA as a teenager was associated with delayed onset of breast cancer [17].
Nkondjock and colleagues [31] did not observe significant associations between PA variables two years before breast cancer diagnosis and breast cancer risk.
Alternatively, Lammert and colleagues [13] and Pijpe and colleagues [25] evaluated PA over longer periods of time. Lammert and colleagues [13] case-control study among 433 women with BRCA1/2 pathogenic germline gene variants assessed PA in adolescence and early adulthood and adjusted analyses for number of children, current BMI, history of oral contraceptive use and/or oophorectomy, and tobacco consumption. Pijpe and colleagues [25] retrospective cohort among 725 women with BRCA1/2 pathogenic germline gene variants assessed lifetime sports activity and adjusted for oral contraceptives, parity, menopausal status, hormone replacement therapy, age-specific BMI, BMI at age 18, alcohol consumption, occupational activity. In certain analyses, mean metabolic equivalent (MET)-hours/week and mean hours/week were adjusted for number of active years, and number of active years were also adjusted for mean MET-hours/week. Lammert and colleagues [13] assessed moderate, vigorous, and total activity in MET-hours/week in adolescence, young adulthood, and overall. Analysis was also stratified by menopausal status at diagnosis. The only association observed was for the highest quartile of moderate activity in adolescence, > 25.88 MET-hours/week, in relation to premenopausal breast cancer risk (HR 0.62, 95%CI 0.40,0.96) [13].
Before age 30, when the lowest sports activity category was used as the reference, 11-22.7 and ≥ 22.7 mean MET-hours/week was associated with a 40% reduction in breast cancer risk [9,25]. Associations were not observed when never-engaging in sports activity was used as the reference category. In contrast, after age 30, everengaging in sports activity was associated with a 37% reduction in breast cancer risk (HR 0.63, 95%CI 0.44,0.91 [25]. When never-engaging in sports activity was the reference category, significant associations were only observed for the lowest category (least amount of activity) of each variable [32]. Table 3 provides a summary of results for diet, weight, and PA in relation to ovarian and breast cancer risk among women with BRCA1/2 pathogenic germline gene variants.

Discussion
This systematic review did not find cohesive evidence supporting the need for tailored recommendations regarding dietary habits, weight management and PA for ovarian and breast cancer risk-reduction among women with BRCA1 or BRCA2 pathogenic germline gene variants. Regarding ovarian cancer risk, there was limited evidence supporting relationships between dietary habits and • For Ashkenazi Jewish women, engagement in physical activity as teenager associated with delayed onset breast cancer (King, 2003) lb pound; BMI body mass index a Seven studies assessed alcohol intake (6 exclusive to alcohol, 1 included alcohol with nutrient intake), two assessed coffee intake, one assessed supplement use (folic acid, B6, B12), one assessed food group intake, one assessed nutrient intake (and included alcohol), one assessed diet quality b Both BRCA1 and BRCA2 pathogenic germline gene variants combined in the analysis c Only BRCA2 pathogenic germline gene variant in the analysis d Folic acid:8.56-≤ 89.29mcg/d; B12:0.02-≤ 0.34mcg/d e Only BRCA1pathogenic germline gene variant in the analysis f One study observed no association (Gronwald, 2006) and one study observed OR0.51(0.26,0.98) for total coffee consumption in relation to breast cancer risk  g Three studies observed no association between alcohol intake and breast cancer risk in BRCA1/2 variants collectively (Cybulski, 2015;Nkondjock, Robidoux, 2006; Lecarpentier 2011), one study observed an association in BRCA1 but not BRCA2 when tobacco use was included as an interaction (Lecarpentier, 2011), one study observed an association in BRCA1 but not BRCA2(Dennis, 2010), one study observed a weak effect of alcohol when comparing breast cancer survivors compared to survivors without BRCA, no effect was observed for BRCA2(Moorman, 2010), one study observed an association in BRCA2 but not BRCA1 (McGuire, 2006), one study observed an effect for alcohol when comparing survivors with BRCA2 to survivors without BRCA, but an effect was not observed in BRCA1 (Dennis, 2011) h One study (King, 2003) assessed weight status and physical activity i Association applies to pre-and post-menopausal breast cancer risk j One study observed a significant association with weight gain since age 18 and 30 and increased breast cancer risk for BRCA1/2 variants (Nkondjock, Robidoux 2006), one study did not observe a significant association with 10-20 or > 20 lb. weight gain between the ages of 18 and 30 for BRCA1/2 variants collectively and by variant, and when age at diagnosis was between 30 and 40 years or > 40 years (Kotsopoulos, 2005) k One study observed a significant inverse association between breast cancer risk and self-reported adulthood overweight/obesity and genetically scored overweight/obesity (Qian, 2019), one study observed a significant positive association between breast cancer risk and adulthood overweight/obesity(Abba, 2019), one study observed a significant positive association between breast cancer risk and adulthood overweight/obesity beyond age 43 (Nkondjock, Robidoux, 2006), one study observed a significant positive association with postmenopausal breast cancer risk and adulthood body weight ≥ 72 kg (Manders, 2011) ovarian cancer incidence. The limited findings related to weight management and premenopausal ovarian cancer risk are similar to findings observed in the general population, which suggests probable relationship between body fatness and increased risk of ovarian cancer [33].
Among the general population, evidence is probable and convincing that alcohol intake increases pre-and postmenopausal breast cancer risk, respectively [34]. Thus current cancer prevention guidelines recommend limiting alcohol [34]. Among women with BRCA1/2 pathogenic germline gene variants collectively and by BRCA1/2 pathogenic variant, evidence is mixed [11,15,16,19,22,23]. Notably, some studies demonstrated no association between alcohol intake and breast cancer risk among women with BRCA1/2 pathogenic germline gene variants collectively [15,19,31], while others observed a reduction in risk with alcohol intake in women with BRCA1 pathogenic germline gene variants [16,23] and BRCA2 pathogenic germline gene variants [22]. Considering the known potential harms associated with alcohol, stronger and more consistent evidence is needed to support more liberal guidelines for alcohol use in women with BRCA1/2 pathogenic germline gene variants.
Evidence is limited related to food/nutrient intake and breast cancer risk among women with BRCA1/2 pathogenic germline gene variants [12,14,27,31]. An association was not observed with vegetable intake [12], and findings for micronutrients are mixed, pending the nutrient [27,31]. Among the general population, evidence is limited and inconclusive regarding the relationship between nonstarchy vegetables, nutrients, and breast cancer risk [34].
Evidence is also mixed for adulthood weight gain [18,20] and adulthood weight status [20,26,29,31]. Among the general population, evidence is probable that overweight/obesity in young adulthood decreases pre-and postmenopausal breast cancer risk, and overweight/ obesity in adulthood increases postmenopausal breast cancer risk [34]. Whether weight management recommendations should differ for women with BRCA1/2 pathogenic germline gene variants remains elusive.
Regarding PA, activity in adolescence and lifetime activity appear to have some association with breast cancer risk-reduction among women with BRCA1/2 pathogenic germline gene variants collectively [13,25]. This notion is supported by evidence from the general population, such that it is probable that PA, regardless of intensity, reduces postmenopausal risk and vigorous-intensity activity reduces premenopausal risk [34]. Thus, activity recommendations should remain consistent with recommendations for the general population. This is the first study to our knowledge to systematically evaluate whether tailored recommendations related to dietary habits, weight management and PA may be effective in reducing ovarian and breast cancer risk among women with BRCA1/2 pathogenic germline gene variants. We consider the following factors limitations of the current state of evidence: small number of studies for both ovarian and breast cancer risk in this high risk population, especially when considering the available large epidemiological studies that have established associations of lifestyle factors among the general population; heterogeneity in methods to evaluate lifestyle factors; inconsistent confounding factors; no data evaluating hormone receptor status; limited data evaluating by gene variant and menopausal status. Considering these limitations, notably the heterogeneity of the current evidence, inability to separate analyses by BRCA1/2 pathogenic germline gene variant, and retrospective nature of the majority of studies conducted, it is difficult to determine the extent of which recommendations for lifestyle factors should differ for this higher risk population. Future observational studies should address these limitations, specifically prospective, larger cohort studies enabling one to assess risk for these factors by gene variant.