Physical Activity in Young BRCA Carriers and Reduced Risk of Breast Cancer

Abstract

Introduction:

A systematic literature review was conducted to determine if physical activity (PA) levels during adolescent and young adult (AYA) years were associated with reduced lifetime risk of breast cancer among carriers of deleterious mutations in BRCA1 and 2 genes.

Methods:

Ovid/MEDLINE, Embase, Cochrane Central Register of Controlled Trials, Web of Science, and CINAHL were searched for articles including information about AYA PA and breast cancer incidence among women carrying deleterious BRCA1 and 2 gene mutations (search initiated 10/2019; last update and full analyses 03/2021). Independent reviewers screened articles at the title/abstract and full text levels, resolving differences by consensus with lead authors. The NIH Quality Assessment Tools were used to assess sources of bias.

Results:

A total of 1957 unique articles were identified; five met inclusion criteria. Samples size ranged from 68–1185. All studies relied on self-reported AYA PA. One study measured sports involvement; the others measured recreational activity. One large study was null, while four others showed a reduction in breast cancer incidence later in life with higher AYA PA (p ≤ 0.05). However, the protection was limited to pre-menopausal breast cancer in one of the studies (OR= 0.62, 95% CI 0.40–0.96; P-trend = 0.01). Additionally, AYA PA was associated with older age at breast cancer diagnosis in one study (p = 0.03).

Conclusion:

A limited number of studies suggest that AYA PA may reduce or delay risk of breast cancer incidence among carriers of deleterious mutations in BRCA1 and 2 genes.

Introduction

Breast cancer is the most common type of cancer experienced by women.¹ Physical activity (PA) during adulthood,^2–5 as well as in adolescence and young adulthood (AYA),^6–8 has been inversely associated breast cancer incidence among average risk women. The risk reduction may occur via exercise-related adipose, insulin, inflammation, and sex steroid reductions.^{3, 9–11}

Carriage of deleterious BCRA1 and 2 gene mutations significantly increases breast cancer risk.¹² Yet, research evaluating PA and breast cancer risk among BRCA mutation carriers is limited.¹ Given the higher levels of circulating sex steroids among BRCA mutation carriers,¹³ the blunting of estrogen availability with habitual exercise may be particularly beneficial for the mutation carriers.^{14, 15}

To reduce chronic disease risks moderate to vigorous PA is recommended for 60 minutes/day most days per week for all youth (approximately ≥ 20 MET-hours/week), 30–60 minutes/day for adults.¹⁶ However, PA drops precipitously in adolescence and does not rebound, especially among girls.¹⁷ This behavioral shift combined with earlier onset of breast cancer among BRCA mutation carriers^18–21 highlights the need to examine AYA exercise and breast cancer risk in this population specifically.

Methods

The systematic review was performed using PICOT as a structural guide and the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement as the reporting guide.²²

Eligibility Criteria

Studies that included data on PA during AYA years (with or without age stratification) among women, breast cancer incidence, and carriage of deleterious BRCA1 or BRCA2 gene mutations were eligible. Studies not focused on PA, only looking at adult or post-diagnosis PA, or not specific to or stratified for BRCA mutation status, and studies without breast cancer incidence data were not included. Also excluded were: animal studies, clinical trials without published data, conference abstracts, case reports, protocol papers, reviews, systematic reviews, meta-analyses, opinion pieces, letters to the editor, dissertations, book chapters, and studies not available in English.

Search Strategy

A medical librarian initially searched the following databases using controlled vocabulary terms (e.g. MeSH, Emtree) and keywords from the dates of their inception to October 21–22, 2019: Ovid/ MEDLINE; Elsevier/Embase; Cochrane Wiley/Cochrane Central Register of Controlled Trials (CENTRAL); Clarivate/ Web of Science (WOS); and EBSCO/Cumulative Index of Nursing and Allied Health Literature (CINAHL). Searches in WOS were updated on December 21, 2020, in anticipation of institutional loss of database access; the others were updated on March 29, 2021. Search strategies are available in Appendix A. Reference lists of review articles were also examined. Records were exported to EndNote Version X9 (Clarivate Analytics, Philadelphia, PA, USA).

Study Selection

The medical librarian pre-screened initial search results to exclude animal studies, case reports, clinical trials with no published results, conference abstracts, obvious opinion pieces, studies about cancers other than breast cancers and those limited to breast cancer in males, protocols, retracted articles, and reviews, including other systematic reviews and meta-analyses (Figure 1). Search results were then equally divided among reviewer pairs for title/abstract and full text review phases. For both phases, each article was screened by each member of the pair independently according to standard procedures. Disagreements were resolved by consensus of reviewers in consultation with the lead authors.

Figure 1. Study Flow Diagram.

Study flowchart of the process of literature search and selection of studies meeting the inclusion criteria

Quality Assessment

Two independent reviewers assessed potential risks for bias using design specific NIH Quality Assessment Tools (Table 1). Similar categories between tools were placed together in the summary; differences in study type or scoring categories were indicated in parentheses.

Table 1.

Bias assessment of studies related to AYA physical activity, carriage of deleterious BRCA mutations, and breast cancer incidence using the NIH Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (OS) and the NIH Quality Assessment of Case-Control Study (CC) criteria

NIH Bias Categories	Kehm 2020 (OS)	King 2003 (OS)	Pijpe 2010 (OS)	Lammert 2018 (CC)	Grill 2017 (OS)
Research question clear	+	+	+	+	+
Study population defined	+	+	+	+	+
Eligible participation ≥50% (OS); randomly selected from eligible if <100% (CC)	+	+	+	−	+
Similar populations recruited (across cohort and cases and controls; Inclusion/ exclusion prespecified	+	+	+	+	+
Processes to select cases/controls valid, reliable, consistent (CC only)	NA	NA	NA	+	NA
Sample size justification a	−	−	−	−	−
Exposure(s) measured occurred prior to outcome(s)	+	+	+	+	+
Follow-up timeframe sufficient (OS only) b	+	+	+	NA	+
Different levels of the exposure related to outcome assessed (OS only)	+	+	+	NA	−
Exposure measures clearly defined, valid, reliable, consistent	+	+	+	+	+
Exposure(s) assessed ≥1 time (cohort); concurrent controls (CC)	−	−	+	+	−
Outcome measures clearly defined, valid, reliable (OS); case-control status differentiation clear (CC)	+	+	+	+	+
Outcome assessors blinded to exposure status (OS); Exposure assessors blinded to case/control status (CC)	−	−	−	−	−
Loss to follow-up after baseline ≤20% (OS only)	+	+	+	NA	+
Key confounders measured and adjusted	+	+	+	+	−

Medium grey + means present; near black − means not present or requirements not met; light grey NA means not applicable/bias category limited to study type indicated in parentheses, not present in the bias instrument for the other study type.

^a.Power – while authors all considered power, none provided sample size justification, necessary effect size, or equations;

^b.sufficient time from exposure to outcome is not applicable because of the retrospective AYA PA questionnaire, but in the observational studies, Grill, Pipje, Kehm, King, there were approximately 28 yrs between the AYA years and breast cancer diagnosis.

Data Extraction

Standard extraction forms were developed and pilot-tested for this systematic review. The following information was extracted: population characteristics, PA assessment, duration of follow-up, risk statistics, and, if available, PA associations with age of diagnosis. Data related to age groups outside AYA were not analyzed.

Results

Figure 1 demonstrates the process of the literature search and study selection. Five studies of the initially identified 3020 met full inclusion criteria: two cohort studies,^{23, 24} two cross sectional analyses,^{25, 26} and one case-control study.²⁷ All studies presented clearly defined research questions, objectives, and population descriptions. Sample size justification and blinding were lacking across studies. Retrospective assessment of AYA PA by self-report was a source of bias across studies but ameliorated potential loss to follow-up. Four studies statistically adjusted for potential confounders, such as BMI; one study was underpowered to do so.²⁶

Study and sample characteristics

Study characteristics are presented in Table 2. Countries of origin varied by study, with race/ethnicity specified in only two studies, described as predominantly white²³ and Ashkenazi Jewish.²⁵ Mean age at enrollment ranged from 40.0–46.3 across studies, while the AYA period assessed ranged from 10–34 years of age. The majority of the articles assessed PA frequency, intensity, and duration,^{23, 24, 26, 27} and computed metabolic equivalent values (METs)^{23, 24, 27} using standard procedures.²⁸ The cut-points for MET comparisons varied across studies. Two studies classified individuals as active versus less- or not active.^{25, 26}

Table 2.

Population characteristics and study designs across studies of AYA physical activity among BRCA1 and BRCA2 deleterious mutation carriers and non-carriers

Citation	Design & Data Source	Baseline Population Characteristics	PA Assessment	AYA PA Timeframe	Follow-up	Analyses
Lammert et al, Breast Cancer Res Treat, 2018	Case-control N=886 (Cases 443; Controls 443) Multi-national (80 centers; 17 countries)	Female, BRCA1 (77.4%) & BRCA2 (22.6%) mutation; Single control/case, matched by mutation, country of residence, birth yr (± 3 yrs); average age controls 50.9 +/− 11.6 years, cases 51.6 +/− 10.9 years; % premenopausal controls 26.6%; cases 8.8%	Questionnaire: MVPA, hrs/wk. Converted to METs: moderate (4.5); vigorous (7.0)	Ages 12–13, 14–17, 18–22, 23–29, 30–34 years	NA	Backwards Stepwise Logistic Regression. Covariates: parity, baseline BMI, tobacco and OCP use, oophorectomy
Kehm, Cancer Research, 2020a	Prospective Cohort N=1185 Multi-national (US, Canada, Australia, New Zealand, & kConFab consortium)	Female, BRCA1 (55.61%) & BRCA2 (44.39%) mutations, aged >18 years, and no risk-reducing surgery; included retrospective and prospective cases; % pre-menopausal 59%	Questionnaire: any PA, hrs/wk, during AYA. Converted to METs.	Ages 12–17 years	annual follow up for a median 10.3 years	Multivariable Cox Proportional Hazards. Covariates: race/ethnicity, education, parity, breastfeeding, tobacco, OCP, HRT and alcohol use, BMI
King, Science, 2003a	Cross-sectional; breast cancer survivor cohort; BRCA1/2 mutation carriers N=104 United States	Females (Ashkenazi Jewish), BRCA1 (65%) & BRCA2 (35%) mutation, aged > 18 years, diagnosed with breast cancer between September 1996 and December 2000; % pre-menopausal NA; 42% diagnosed <50years of age (differed by birth cohort)	Questionnaire: categorized as active vs inactive during AYA	“Teenage years”	NA	Multivariable Cox Proportional Hazards. Covariate: decade of birth
Pijpe, Breast Cancer Res Treat, 2010b	Retrospective Cohort (with prospective follow-up) N=725 Netherlands	Female, BRCA1 (77%) & BRCA2 (23%) mutation, no prior breast cancer history, aged > 18 years; % pre-menopausal cohort 42%, cases 71%	Questionnaire: any PA, hrs/wk, converted to METs; number of years active	Before age 30 years	Up to date of breast cancer incidence; Median age at end of follow-up 45.5 +/− 13.3 yrs	Time varying, Multivariable Cox Proportional Hazards. Covariates: OCP. HRT and alcohol use, parity, menopausal status, BMI at age 18, age specific BMI groups, and occupational activity
Grill, Arch Gynecol Obstet, 2017b	Cross-sectional at RCT baseline N=68 Germany	Female, BRCA1 (61.8%) & BRCA2 (32.2%) mutation, >18 yrs old, enrolled in LIBRE-1; %pre-menopausal NA.	Interviews and Questionnaire: any PA, hrs/wk and sports	Ages 10–19 and 20–29 years	None	Chi Squared Test

HRT, hormonal replacement therapy; hrs, hours; MVPA, moderate to vigorous physical activity; OCP, oral contraceptive pill; RCT, randomized controlled trial; wk, week.

^a.Total population of Kehm study 15,550 and King, 1008. The populations listed in table are the stratified groupings of BRCA 1 & 2 deleterious mutations;

^b.Additional age groups outside the AYA timeframe of interest not presented.

AYA Physical Activity Associations with Breast Cancer Incidence

Four of five studies showed a significant reduction in breast cancer risk, up to 40%, with higher AYA PA^24–27 (N=3 lifetime risk;^24–26 N=1 limited to pre-menopausal breast cancer and adolescent activity²⁷). A dose-response relationship was not demonstrated. In contrast, the largest study was null,²³ though the direction of AYA PA effect for BRCA2 mutation carriers alone was protective.²³ One study additionally demonstrated an association between AYA PA and older age at breast cancer diagnosis among BRCA mutation carriers (p = 0.03).²⁵

Discussion

The majority of the studies reviewed support a potential role for AYA PA in reducing or delaying breast cancer incidence among deleterious BRCA variant carriers, which expands upon previous breast cancer risk reductions associated with adult^2–5 and AYA PA⁸ among average risk women. Importantly, effect sizes herein align with the breast cancer risk reduction estimates with AYA PA in the average risk population.⁸ Taken together, these data suggest that beneficial effects of AYA PA may not be overwhelmed by genetic predisposition to breast cancer, though the dose of PA required has yet to be established.

While one may posit that the data are equivocal, based on the null study sample size versus sample in the remaining studies showing benefit, it is important to factor in the value of replication across sites. The collective data also suggest that the relation between AYA PA and breast cancer risk among BRCA mutation carriers may be nuanced and potentially more evident for BRCA2, adolescent versus young adult PA, and pre- versus postmenopausal cancers. Lower age of diagnosis and higher risk rates among BRCA 2 mutation carriers^18–21 emphasize the need to stratify along these lines. Additionally, all studies utilized retrospective self-report PA, which is prone to bias that may vary across regions and contribute to inconsistent results. Lastly, study designs and PA cut-points varied and limited studies were identified with the following: prospective follow-up;^{23, 24} alignment of AYA age groups;^{23, 27} stratification by age groups,^{26, 27} pre- versus postmenopausal cancer,²⁷ and BRCA mutation type;²³ and adequate accounting of confounders, like BMI and hormone use,^{23, 24, 27} which may differentially impact pre- versus postmenopausal risk.

Future studies with prospective, standardized, and repeated AYA PA assessment and more diverse populations among BRCA mutation carriers are needed. Nevertheless, the benefits of PA far outweigh risks of sedentarism across chronic conditions,¹⁶ so a physically active lifestyle across the lifespan is recommended regardless of BRCA status.

Table 3.

Adolescent and young adult physical activity and risk of breast cancer among BRCA1 and BRCA2 deleterious mutation carriers versus non-carriers

Author	N	PA Age	Mutation	PA	Pre-men Dx	95% CI	Post-men Dx	95% CI	Lifetime	95% CI
Odds Ratios, 95% CIs
				MVPA
Lammert, 2018a	443	12yr – 17yr	BRCA1/2	Q1	1		1		1
				Q2	1.04	0.70–1.53	1.25	0.71–2.18	1.09	0.76–1.56
				Q3	1.48	0.94–2.32	1.39	0.71–2.73	1.48	0.97–2.27
				Q4	0.62	0.40–0.96	1.53	0.87–2.71	0.82	0.56–1.20
		18yr – 34yr	BRCA1/2	Q1	1		1		1
				Q2	1.53	0.99–2.37	0.94	0.51–1.74	1.35	0.90–2.02
				Q3	1.11	0.70–1.74	0.70	0.36–1.36	0.99	0.65–1.50
				Q4	0.99	0.65–1.49	1.15	0.67–1.98	1.03	0.71–1.49
				TOTAL
		12yr – 17yr	BRCA1/2	Q1	1		1		1
				Q2	0.83	0.53–1.31	0.89	0.48–1.67	0.84	0.56–1.27
				Q3	1.4	0.92–2.12	1.18	0.65–2.14	1.33	0.9–1.96
				Q4	0.89	0.59–1.33	1.1	0.63–1.91	0.94	0.65–1.36
		18yr – 34yr	BRCA1/2	Q1	1		1		1
				Q2	1.57	0.99–2.49	0.94	0.48–1.84	1.41	0.92–2.15
				Q3	1.41	0.89–2.24	1.55	0.84–2.87	1.44	0.95–2.19
				Q4	1.08	0.71–1.66	0.99	0.55–1.78	1.07	0.73–1.57
Hazards Ratios, 95% CIs
Kehm, 2020b	1185	12yr – 17yr	BRCA1	MVPA Q2–Q5 vs. Q1	NA		NA		1.01	0.76–1.36
			BRCA2	Q2–Q5 vs. Q1	NA		NA		0.88	0.68–1.14
King, 2003c	104	12yr – 17yr	BRCA1/2	Active /very active vs. inactive	NA		NA		0.63	NA
Pijpe, 2010d	725	<30yr	BRCA1/2	Low	NA		NA		1.0
				Med	NA		NA		0.60	0.38–0.96
				High	NA		NA		0.58	0.35–0.94
Chi Squared Test									Cancer N (%)	No Cancer N (%)
Grill, 2017e	68	10yr – 19yr	BRCA1/2	Active	NA		NA		27(59%)	19 (86%)
				Less Active	NA		NA		3(14%)	19 (41%)

[Boldface indicates statistical significance (p<0.05).]

^a.MVPA adolescent Q1 ≤6.75, Q2 >8.55 and ≤15.75, Q3 >15.75 and ≤27.56, Q4 > 25.88 MET-hrs/wk and young adult Q1 ≤6.75, Q2 >8.55 and ≤15.75, Q3 15.75 and ≤ 29.25, Q4 .29.25. Total recreational adolescent PA Q1≤24.25, Q2 >24.25 ≤ 40.25, Q3 >40.25 and ≤69.25, Q4 >69.25 and young adult Q1 ≤20.33, Q2 >20.33 and ≤ 38.58, Q3 38.58 and ≤ 63, >63 MET-hrs/wk;

^b.MVPA Q1 0–14 METs/wk vs combined Q2 to Q5 ranging 15–121 METs/wk, reported prospective and retrospective combined data – all adolescent PA was retrospective report regardless of f/u for cases;

^c.active or very active in recreational, dance, and sports activities versus inactive;

^d.low <10.6, medium 10.6–21.7, high ≥ 21.7 MET-hrs/wk presented;

^e.active versus less active compared to peers at age 10–19 years.

Samples presented are BRCA mutation carriers only; Data presented herein for Grill 2017 were obtained via personal communication with the lead author (original article presented the data as Figure 1). King and Kehm also included non-carriers and total sample was 1,008 and 15,550 respectively (total sample data not presented); The outcome for cox proportional hazards analyses were breast cancer incidence for Kehm and Pijpe studies, but age of diagnosis for King et al. CI, confidence interval; hrs, hours; MET, metabolic equivalents; MVPA, moderate to vigorous physical activity; PA, physical activity; pre-men, premenopausal cancer; post-men, postmenopausal cancer; Q; quantile.