Associations of obesity and circulating insulin and glucose with breast cancer risk: a Mendelian randomization analysis

Abstract

Background

In addition to the established association between general obesity and breast cancer risk, central obesity and circulating fasting insulin and glucose have been linked to the development of this common malignancy. Findings from previous studies, however, have been inconsistent, and the nature of the associations is unclear.

Methods

We conducted Mendelian randomization analyses to evaluate the association of breast cancer risk, using genetic instruments, with fasting insulin, fasting glucose, 2-h glucose, body mass index (BMI) and BMI-adjusted waist-hip-ratio (WHR_{adj BMI}). We first confirmed the association of these instruments with type 2 diabetes risk in a large diabetes genome-wide association study consortium. We then investigated their associations with breast cancer risk using individual-level data obtained from 98 842 cases and 83 464 controls of European descent in the Breast Cancer Association Consortium.

Results

All sets of instruments were associated with risk of type 2 diabetes. Associations with breast cancer risk were found for genetically predicted fasting insulin [odds ratio (OR) = 1.71 per standard deviation (SD) increase, 95% confidence interval (CI) = 1.26-2.31, p  =  5.09  ×  10^–4], 2-h glucose (OR = 1.80 per SD increase, 95% CI = 1.3 0-2.49, p  =  4.02  ×  10^–4), BMI (OR = 0.70 per 5-unit increase, 95% CI = 0.65-0.76, p  =  5.05  ×  10^–19) and WHR_{adj BMI} (OR = 0.85, 95% CI = 0.79-0.91, p  =  9.22  ×  10^–6). Stratified analyses showed that genetically predicted fasting insulin was more closely related to risk of estrogen-receptor [ER]-positive cancer, whereas the associations with instruments of 2-h glucose, BMI and WHR_{adj BMI} were consistent regardless of age, menopausal status, estrogen receptor status and family history of breast cancer.

Conclusions

We confirmed the previously reported inverse association of genetically predicted BMI with breast cancer risk, and showed a positive association of genetically predicted fasting insulin and 2-h glucose and an inverse association of WHR_{adj BMI} with breast cancer risk. Our study suggests that genetically determined obesity and glucose/insulin-related traits have an important role in the aetiology of breast cancer.

Breast cancer, insulin, glucose, obesity, genetics, Mendelian randomization analysis

Key Messages

• Mendelian randomization studies eliminate potential influence of reverse causation on study results and are less susceptible to bias and confounding than conventional observational studies. We used this approach to evaluate the association of obesity and glucose/insulin-related traits with breast cancer risk, using the data of a large consortium.

• We found genetically predicted fasting insulin and 2-h glucose levels were positively associated with breast cancer risk, whereas genetically predicted body mass index and waist-hip-ratio with adjustment of BMI were inversely associated with the risk.

• Our study has uncovered complex inter-relations of genetics, obesity, and breast cancer risk, and has provided novel findings regarding roles of circulating glucose and insulin in the risk of this common cancer.

Introduction

General and central obesity have been linked to breast cancer risk in previous studies.^1,2 Body mass index (BMI) and waist-hip-ratio (WHR) are commonly used to measure general and central obesity, respectively. Obesity, particularly central obesity, is a major risk factor for insulin resistance and type 2 diabetes, which are often characterized by elevated fasting insulin and glucose as well as impaired glucose tolerance (usually measured by blood glucose level 2 h after oral glucose challenge).³ Previous studies have linked fasting insulin and glucose levels to increased risks of multiple cancers.^4–6 Proposed mechanisms for these associations include cancer-promoting effects mediated by insulin and insulin-like growth factor (IGF) signalling pathways.⁷ However, the relationship between these biomarkers and breast cancer remains controversial and findings from epidemiological studies are inconsistent.^8,9 Concerns regarding the validity of these observational study findings include potential selection biases, reverse causation, confounding effects, small sample size and differences in assays used to measure the biomarkers of interest.

Mendelian randomization analysis has been used to evaluate potential causal relationships between exposures and disease.^10,11 Genetic variants are used as instrumental variables in the analysis. Random assortment of alleles at the time of gamete formation results in a random assignment of exposures that are related to an allele (or a set of alleles). Thus, Mendelian randomization analyses may reduce potential biases that would afflict conventional observational studies. In the current study, we performed Mendelian randomization analyses to assess associations of obesity (i.e. BMI and WHR) and glucose/insulin-related traits (i.e. fasting glucose, 2-h glucose and fasting insulin) with breast cancer risk, using data from the Breast Cancer Association Consortium (BCAC).

Methods

Study population

Included in this analysis are 182 306 participants of European ancestry, whose samples were genotyped using custom Illumina iSelect genotyping arrays: OncoArray (56 762 cases and 43 207 controls) or iCOGS array (42 080 cases and 40 257 controls). Institutional review boards of all involved institutions approved the studies. Selected characteristics of the two datasets are presented in Supplementary Table 1, available as Supplementary data at IJE online. Details of the genotyping protocols in the BCAC are described elsewhere (iCOGS: http://ccge.medschl.cam.ac.uk/research/consortia/icogs/; OncoArray: https://epi.grants.cancer.gov/oncoarray/).^12,13 Genotyping data were imputed using the program IMPUTE2¹⁴ with the 1000 Genomes Project Phase III integrated variant set as the reference panel. Single nucleotide polymorphisms (SNPs) with low imputation quality (imputation r² < 0.5) were excluded. Top principal components (PCs) were included as covariates in regression analysis to address potential population substructure (iCOGS: top eight PCs; OncoArray: top 15 PCs).

Selection of SNPs associated with glucose/insulin-related traits

In December 2016, we searched the National Human Genome Research Institute-European Bioinformatics Institute Catalog of Published Genome-Wide Association Studies and the literature for SNPs associated with the following traits: levels of 2-h glucose (2hrGlu), fasting glucose (FG), fasting insulin (FI), BMI and waist-hip-ratio with adjustment of BMI (WHR_{adj BMI}).^15–19 SNPs associated with any of these traits at the genome-wide significance level (P  <  5  ×  10^–8) in populations of European ancestry were included. For each GWAS-identified locus, a representative SNP with the lowest P-value in the original GWAS publication was selected (linkage disequilibrium r² < 0.1, based on 1000 Genome Phase III CEU data).

Construction of instrumental variables

Weighted polygenic scores for each trait (i.e. wPRS-2hrGlu, wPRS-FG, wPRS-FI, wPRS-BMI and wPRS-WHR_{adj BMI}) were constructed following the formula: wPRS-trait_j$={\sum }_i^{}{\beta }_{i,GX}\mathrm{*}{SNP}_i$⁠, where β_{i, GX}is the beta coefficient of the i^th SNP for the trait of interest from the published GWAS (Supplementary Table 2, available as Supplementary data at IJE online). SNP_i is the imputed dosage of the effect allele in BCAC data (range: 0 to 2). To reduce potential pleiotropic effects, we excluded BMI- and WHR_{adj BMI}-associated SNPs from instruments of 2hrGlu, fasting glucose and insulin (r² < 0.8), and vice versa. The pleiotropic SNPs associated with more than one trait are presented in Supplementary Table 2, available as Supplementary data at IJE online. The F-statistic was taken to indicate whether an instrumental variable was well-powered for Mendelian randomization analysis, with 10 being a commonly used threshold.²⁰ Variance explained (%) and F statistics for a specific trait were calculated following the formulae: ${\sum }_i^{}2\mathrm{*}{\beta }_{i,GX}^2\mathrm{*}{f}_{effect\mathrm{ }allele}\mathrm{*}\frac{\left(1-f_{effect\mathrm{ }allele}\right)}{var\left(X\right)}\mathrm{*}100$and R²*(n-1-k)/(1-R²)/k, respectively, where: R² is percentage of variance explained by used SNPs; f is the frequency of the effect allele reported by GWAS for the trait; var(X) is the variance of trait, see below; n is the sample size of BCAC data; and k is the number of SNPs used in the instrument.²¹

For 2-h glucose, fasting glucose and insulin, β_{i, GX}were further transformed to represent 1 standard deviation (SD) increase with the unit in the original GWAS (2-h glucose: 1 SD = 2 mmol/L, variance = 4; fasting glucose: 1 SD = 0.65 mmol/L, variance = 0.42; fasting insulin: 1 SD = 0.60 ln[pmol/L], variance = 0.36)^17,22 by the formula: β_i,SD = β_i,GX [2*f (SNP_i)(1-f (SNP_i)]  ^0.5/SD. wPRS-BMI and wPRS-WHR_{adj BMI} represented the adjusted 1-SD increase of transformed BMI and WHR_{adj BMI}, as the original GWAS performed the inverse normal transformation for both phenotypes.^18,19,23 We further scaled wPRS-BMI to be equivalent to five units of BMI by performing a linear regression among controls in our dataset: observed BMI ∼ wPRS-BMI + error. Then we calculated the transformed BMI as BMI_wPRS = β₀+ β₁* (wPRS-BMI), where β₀ and β₁ are slope and coefficient from the linear regression model mentioned above, respectively.

Statistical analysis

Given an established association between impaired glucose/insulin traits and type 2 diabetes, an association between constructed instruments and risk of type 2 diabetes is to be expected. We used summary statistics from the DIAbetes Genetics Replication And Meta-analysis (DIAGRAM) Consortium and conducted a Mendelian randomization analysis of our traits using the inverse-variance-weighted two-sample method.^10,24 The Mendelian randomization estimate and standard error were calculated as ${\sum }_i^{}{\beta }_{i,GX}\mathrm{*}{\beta }_{i,GY}\mathrm{*}{\sigma }_{i,GY}^{-2}/\left({\sum }_i^{}{\beta }_{i,GX}^2\mathrm{*}{\sigma }_{i,GY}^{-2}\right)\mathrm{ }$and${\mathrm{ }1/\left({\sum }_i^{}{\beta }_{i,GX}^2\mathrm{*}{\sigma }_{i,GY}^{-2}\right)}^{0.5}$⁠, respectively. GY represents the association between a, SNP and type 2 diabetes risk; thus β_{i, GY} and σ_{i, GY} are beta coefficient and standard error, respectively. The P-value was based on Student’s t distribution, where the degrees of freedom were determined by the number of SNPs included in the instrument for the trait of interest. We calculated Pearson’s correlations between each pair of wPRSs in the control data before and after removal of pleiotropic SNPs. Egger’s regression, as described in Bowden et al.,²⁵ was performed to detect potential pleiotropy of our instruments. We also included all instruments in the same model to evaluate possible independent associations of each instrument with breast cancer risk.

Associations of wPRSs with breast cancer risk were evaluated separately in the iCOGs and OncoArray datasets by treating these scores as continuous variables. A logistic regression was performed with age at interview/diagnosis, study site/country and PCs as covariates. The results were then combined using meta-analyses in METAL with a fixed-effects model.²⁶ We performed additional analyses adjusting for certain known breast cancer risk factors listed in Supplementary Table 1, available as Supplementary data at IJE online. Finally, we conducted subanalyses by estrogen receptor (ER) status, age at interview/diagnosis (< 50 versus ≥ 50), menopausal status at interview/breast cancer diagnosis and family history of breast cancer. All statistical analyses were conducted using R statistical software (version 3.1.2).

Results

Approximately 90% of cases included in this study were diagnosed at age 40 or above. A total of 278 SNPs were selected to construct the instruments, for which the number of SNPs for each trait ranged from 4 to 166 (Table 1). The variance of each trait explained by its associated variants ranged from 0.23% for 2-h glucose to 2.89% for BMI (Table 1).

Table 1.

Summary of instrument variables for obesity and glucose/insulin-related traits used in the current study

	All SNPs			After exclusion of pleiotropic SNPs
Traits	No. of variants	Variance explained (%)	F statistics	No. of variants	Variance explained (%)	F statistics
2-h glucose	9	0.56	114.9	4a	0.23	105.7
Fasting glucose	36	2.42	125.7	31b	2.30	107.7
Fasting insulin	18	0.59	60.4	10c	0.27	53.9
BMI	166	2.89	32.6	162d	2.84	32.3
WHRadj BMI	54	1.96	67.5	50d	1.79	65.2

a

Excluding SNPs (or their correlated SNPs with r² > 0.8) associated with fasting glucose, fasting insulin, BMI and WHR_{adj BMI}.

b

Excluding SNPs (or their correlated SNPs with r² > 0.8) associated with levels of 2-h glucose, fasting insulin, BMI and WHR_{adj BMI}.

c

Excluding SNPs (or their correlated SNPs with r² > 0.8) associated with levels of 2-h glucose, fasting glucose, BMI and WHR_{adj BMI}.

d

Excluding SNPs (or their correlated SNPs with r² > 0.8) associated with levels of 2-h glucose, fasting glucose and fasting insulin.

Using data from DIAGRAM, we demonstrated that all genetic instruments were associated with risk of type 2 diabetes in the direction that would be expected (Table 2). The strongest association was observed for the genetic instrument for fasting glucose (OR = 6.37, P  =  5.77  ×  10^–16 and OR = 4.32, P  =  1.12  ×  10^–11 before and after the exclusion of pleiotropic SNPs, respectively).

Table 2.

Associations of obesity and glucose/insulin-related traits with type 2 diabetes using data from DIAGRAM: results from Mendelian randomization analysis

Traits	All SNPs			After exclusion of pleiotropic SNPs
	IV	OR (95% CI)	P	IV	OR (95% CI)	P
2-h glucosea	9	12.0 (6.90–21.0)	2.11 × 10−5	4	21.5 (5.76–80.3)	0.005
Fasting glucosea	36	6.37 (4.87–8.32)	5.77 × 10−16	31	4.32 (3.26–5.73)	1.12 × 10−11
Fasting insulina	18	1.92 (1.10–3.35)	0.024	10	4.62 (1.82–11.7)	0.005
BMI	132	1.92 (1.64–2.25)	2.86 × 10−13	128	2.37 (1.99–2.82)	2.40 × 10−17
WHRadj BMI	53	1.87 (1.53–2.29)	5.93 × 10−8	49	1.99 (1.61–2.46)	3.52 × 10−8

a

ORs calculated based on 1-SD increase in levels of genetically predicted 2-h glucose (2 mmol/L,²²), fasting glucose (0.65 mmol/L,¹⁷) and fasting insulin (0.60 ln[pmol/L]¹⁷).

Table 3.

Associations of genetically predicted obesity and glucose/insulin-related traits with breast cancer risk: results from Mendelian randomization analysis

	All SNPs				After exclusion of pleiotropic SNPs
Traits	OR	95% CI	P	Phet	OR	95% CI	P	Phet
2-h glucosea	1.50	1.21–1.86	2.13 10−4	0.608	1.80	1.30–2.49	4.02 × 10−4	0.566
Fasting glucosea	1.06	0.95–1.17	0.291	0.543	1.02	0.91–1.14	0.749	0.357
Fasting insulina	1.16	0.96–1.41	0.128	0.939	1.71	1.26–2.31	5.09 × 10−4	0.442
BMI
per five-unitb	0.70	0.65–0.76	5.25 × 10−22	0.042	0.70	0.66–0.77	5.05 × 10−19	0.086
per SDa	0.76	0.72–0.80	5.25 × 10−22	0.042	0.77	0.73–0.82	5.05 × 10−19	0.086
WHRadj BMIa	0.85	0.79–0.91	4.48  ×  10−6	0.132	0.85	0.79–0.91	9.22 × 10−6	0.152

a

ORs calculated based on 1-SD increase in levels of genetically predicted 2-h glucose (2 mmol/L,²²), fasting glucose (0.65 mmol/L,¹⁷), fasting insulin (0.60 ln[pmol/L],¹⁷), BMI and WHR_{adj BMI}.

b

ORs calculated based on 5-unit increase of genetically predicted BMI (see Methods).

Removing pleiotropic SNPs did not appreciably change the associations of instruments with breast cancer risk (Table 3). A 1-SD increase in genetically predicted 2-h glucose levels was associated with an 80% increased risk of breast cancer (OR = 1.80, 95% CI = 1.30-2.49, p  =  4.02  ×  10^–4). An inverse association was observed for both genetically predicted BMI and WHR_{adj BMI} (per five units of BMI increase: OR = 0.70, 95% CI = 0.66–0.77, P  =  5.05  ×  10^–19; per unit increase of genetic risk score for WHR_{adj BMI}: OR = 0.85, 95% CI = 0.79-0.91, P  =  9.22  ×  10^–6). The association of breast cancer risk with genetically predicted fasting insulin became significant after excluding pleiotropic SNPs (OR = 1.71, 95% CI = 1.26-2.31, P  =  5.09  ×  10^–4). No association was observed for genetically predicted fasting glucose. Results of iCOGS and OncoArray are shown separately in Supplementary Table 3, available as Supplementary data at IJE online.

Genetically predicted fasting insulin was correlated with both genetically predicted 2-h glucose and WHR_{adj BMI} (Supplementary Table 4, available as Supplementary data at IJE online). Exclusion of pleiotropic SNPs attenuated these correlations. Mutual adjustment of all instruments did not materially change the observed associations with breast cancer risk described above (Supplementary Table 5, available as Supplementary data at IJE online). We evaluated the associations of genetically predicted obesity and glucose/insulin-related traits with traditional risk factors for breast cancer and found that genetically predicted fasting insulin and WHR_{adj BMI} were associated with BMI in controls. Further, genetically predicted BMI was correlated with age at menarche, age at first live birth and breastfeeding history (Supplementary Table 6, available as Supplementary data at IJE online). Adjusting for these covariates did not materially change the observed associations of genetically predicted fasting insulin, BMI and WHR_adjBMI with breast cancer risk (Supplementary Table 7, available as Supplementary data at IJE online). Finally, using Egger’s regression, we found that the intercept in the model was noticeable for genetically predicted 2-h glucose, BMI and WHR_{adj BMI,} indicating a strong pleiotropic effect for these instruments (P  <  0.005 for β₀, Supplementary Table 8, available as Supplementary data at IJE online).²⁵ No apparent pleiotropy was found for genetically predicted fasting insulin. The Mendelian randomization estimates from Egger’s regression remained significant after accounting for detected pleiotropy for genetically predicted BMI and WHR_{adj BMI} (Supplementary Table 8, available as Supplementary data at IJE online).

Stratified analysis was performed by age, menopausal status, ER status and family history of breast cancer. Genetically predicted 2-h glucose, BMI and WHR_{adj BMI} were consistently associated with breast cancer across all strata (Figure 1A, C andD, P_het > 0.05, exclusion of pleiotropic SNPs). The association with genetically predicted fasting insulin was restricted to ER(+) cancer (Figure 1B, P_het 0.007, exclusion of pleiotropic SNPs). The results of stratified analysis are shown for other sets of instrumental variables in Supplementary Figures 1 (inclusion of pleiotropic SNPs) and 2 (fasting glucose, exclusion of pleiotropic SNPs), available as Supplementary data at IJE online.

Figure 1.

Associations of genetically predicted obesity and levels of circulating glucose and insulin with overall breast cancer risk: stratified analysis. The P_{heterogeneity} was obtained from heterogeneity test across strata.

Discussion

In this large study, we found that genetically predicted obesity, 2-h glucose and fasting insulin were associated with breast cancer risk. Measured BMI has been well established to be positively associated with breast cancer risk in postmenopausal women but inversely related to the risk in premenopausal women. Results from epidemiological studies investigating the association of breast cancer risk with WHR, fasting insulin and glucose have been inconsistent. Traditional epidemiological studies are prone to biases, including confounding, selection biases, recall biases and reverse causality. Mendelian randomization analyses take advantage of the random assignment of genetic alleles during gamete formation to minimize the biases commonly encountered in traditional epidemiological studies. When instrumental variables are not associated with any potential confounders and are not linked to the outcome via any alternative pathway, Mendelian randomization analysis using such instrumental variables resemble randomized clinical trials, and thus could provide strong results for causal inference for the exposure of interest.¹⁰

We found that the risk of breast cancer increased approximately 70% for each 1-SD increase of genetically predicted fasting insulin levels. Previous epidemiological studies were unable to reach a conclusion regarding the association between fasting insulin and breast cancer risk. A meta-analysis reported a null association for fasting insulin.⁸ However the I², an indicator of heterogeneity across studies, was considerable. Our results provide strong evidence to support a positive association. Insulin is an important growth factor with cancer-promoting features such as stimulating cell mitosis and migration and inhibiting apoptosis. Its mitogenic effects involve the activation of Ras and the mitogen-activated protein kinase pathway,²⁷ of which the role in cancer development has been recognized.²⁸ Further, insulin may inhibit the production of sex hormone-binding globulin and lead to elevated bioavailable estrogen levels.²⁹ It also has been shown that knockdown of insulin and IGF-1 receptors inhibits hormone-dependent growth of ER(+) breast cancer cells.³⁰ This may explain the association of fasting insulin with ER(+) breast cancer observed in this study.

Previous epidemiological studies have suggested that fasting glucose may be a risk factor for breast cancer, but few have assessed 2-h glucose levels, as the latter are difficult to investigate in large prospective cohort studies. Overall, a meta-analysis of prospective studies showed no strong evidence to support an association of fasting glucose levels and risk of breast cancer in non-diabetic women.⁹ In the current study, we found a positive association with breast cancer for genetically predicted 2-h glucose levels but not for fasting glucose. Although fasting glucose and 2-h glucose are closely correlated,³¹ they represent different biological processes. The genetically determined fasting glucose levels primarily reflect the glycogenolysis activity in liver and hepatic insulin sensitivity.³² On the other hand, the levels of post-challenge glucose are mainly determined by the amount and pace of insulin released into blood stream in response to the challenge as well as by the glucose uptake in skeletal muscle cells (in other words, it primarily reflects beta cell function and skeletal muscle insulin sensitivity³³). The reasons why genetically predicted 2-h glucose, but not fasting glucose, is associated with increased risk of breast cancer are not clear. One animal study has provided evidence that transgenic mice with inactivated insulin and IGF-1 receptors in skeletal muscles (impaired skeletal muscle insulin sensitivity) can manifest hyperinsulinaemia and an accelerated development of breast cancer.³⁴ Since genetically predicted 2-h glucose is correlated with instruments for other traits, we cannot completely rule out the possibility that the association of 2-h glucose may be mediated by other insulin-related traits; even these traits were carefully adjusted, and pleiotropic SNPs were excluded in our analyses.

We reported previously that genetically predicted BMI was inversely associated with breast cancer risk in both pre- and postmenopausal women.³⁵ We have now confirmed this finding with a much larger sample size and more BMI-associated SNPs. Whereas our finding for premenopausal breast cancer is consistent with previous observational studies, the inverse association observed in our study between genetically predicted BMI and postmenopausal breast cancer risk contradicts previous findings based on measured BMI. Multiple lines of evidence suggest that early life body size may be inversely associated with both premenopausal and postmenopausal breast cancer risk.^36,37 It has been speculated that reduced serum estradiol and progesterone levels, due to an increased frequency of anovulation, play a role. In addition, the association is further supported by the observation that early life fatness was inversely correlated with IGF-1 levels measured in later adulthood.³⁸ We hypothesize that genetically predicted BMI may be more closely correlated to early life body weight, and obesity determined using measured BMI later in life may be more closely related to environmental and lifestyle factors that are associated with breast cancer risk. Indeed, one previous study found that a BMI-genetic score was positively associated with weight gain before reaching middle age but inversely associated with weight gain after reaching middle age.³⁹ If the hypothesis is correct, our study may provide additional support for preventing weight gain later in life to reduce the risk of breast cancer.

Results from previous studies regarding the association of WHR with breast cancer risk have been inconsistent. Although several previous studies reported that measured WHR was associated with breast cancer risk,⁴⁰ we recently found that this association was substantially attenuated after adjusting for BMI using data from a large prospective cohort study conducted among Chinese women.⁴¹ In the current study, we observed an inverse association between genetically predicted WHR_{adj BMI} and breast cancer risk in both pre- and postmenopausal women. This finding was unexpected, given the close association of measured WHR with type 2 diabetes.⁴² As discussed previously for the BMI findings, we hypothesize that genetically predicted WHR_{adj BMI} may reflect visceral adipose tissue level in early life, whereas measured WHR in late adulthood may reflect accumulation of visceral fats later in life. Additional research is needed to understand the inter-relationship of genetically predicted WHR, measured WHR and breast cancer risk.

We showed that genetically predicted obesity and circulating insulin and glucose levels were positively correlated with risk of type 2 diabetes. Epidemiological studies have shown that a previous diagnosis of type 2 diabetes was related to an elevated risk of breast cancer risk, although the association was weak to moderate.⁴³ However, in a previous study, we found a null association between a polygenetic risk score for type 2 diabetes and breast cancer risk.⁴⁴ It is possible that lifestyle changes after diabetes diagnosis and/or diabetes treatment may have confounded this association. Given the significant association we found in this study for breast cancer risk with genetically predicted fasting insulin and 2-h glucose, two factors that are strongly associated with type 2 diabetes risk, we suggest that type 2 diabetes may be associated with breast cancer risk.

The sample size of our study is very large, providing us sufficient statistical power for Mendelian randomization analyses of multiple obesity, glucose/insulin-related traits and breast cancer risk. Our ability to perform Mendelian randomization analysis is limited by the genetic variants identified to date in GWAS, and the variance explained by these genetic variants for some traits is small. We used 10 instruments in our main analysis, which could lead to false-positive findings due to multiple comparisons. However, the associations reported in this study for 2-h glucose, fasting insulin, BMI and WHR_{adj BMI} were robust, reaching the stringent Bonferroni corrected significance level (P  <  0.05/10 = 0.005). Pleiotropy was found for the associations of obesity, but it is not likely that the observed associations can be primarily explained by pleiotropic effects.

In summary, this study provided new evidence that genetically predicted fasting insulin, 2-h glucose, BMI and WHR_{adj BMI} are associated with breast cancer risk in women. Further research into the complex association of genetics, obesity, glucose/insulin-related traits and breast cancer risk will help to improve the understanding of underlying biological mechanisms for the associations observed in this study and may provide tools to reduce breast cancer risk.

Funding

This work at Vanderbilt University Medical Center was supported in part by the National Cancer Institute at the National Institutes of Health [grant numbers R01CA158473, R01CA148677], as well as funds from Anne Potter Wilson endowment to W.Z. Genotyping of the OncoArray was principally funded from three sources: the PERSPECTIVE project, funded from the Government of Canada through Genome Canada and the Canadian Institutes of Health Research, the Ministère de l’Économie, de la Science et de l’Innovation du Québec through Genome Québec, and the Quebec Breast Cancer Foundation; the National Cancer Institute Genetic Associations and Mechanisms in Oncology (GAME-ON) initiative and Discovery, Biology and Risk of Inherited Variants in Breast Cancer (DRIVE) project [grant numbers U19CA148065, X01HG007492); and Cancer Research UK [grant numbers C1287/A10118, C1287/A16563]. BCAC is funded by Cancer Research UK [grant number C1287/A16563], the European Community’s Seventh Framework Programme [grant number 223175 (HEALTH-F2-2009–223175) (COGS)] and by the European Union’s Horizon 2020 Research and Innovation Programme [grant agreements 633784 (B-CAST) and 634935 (BRIDGES)]. Genotyping of the iCOGS array was funded by the European Union [HEALTH-F2-2009–223175], Cancer Research UK [C1287/A10710], the Canadian Institutes of Health Research for the ‘CIHR Team in Familial Risks of Breast Cancer’ program, and the Ministry of Economic Development, Innovation and Export Trade of Quebec [PSR-SIIRI-701]. Combining the GWAS data was supported in part by the National Institute of Health (NIH) Cancer Post-Cancer GWAS initiative [grant number U19CA148065; DRIVE, part of the GAME-ON initiative]. For a full description of funding and acknowledgments, see Supplementary Note, available as Supplementary data at IJE online. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

Acknowledgements

The authors thank Jirong Long, Wanqing Wen, Yingchang Lu and Kim Kreth of Vanderbilt Epidemiology Center for their help with this study. The authors also wish to thank all the individuals who took part in these studies and all the researchers, clinicians, technicians and administrative staff who have enabled this work to be carried out.

Other authors from the Breast Cancer Association Consortium: Beeghly-Fadiel J Alicia,¹ Hoda Anton-Culver,¹⁵ Natalia N Antonenkova,¹⁶ Volker Arndt,¹⁷ Kristan J Aronson,¹⁸ Paul L Auer,^2,190 Myrto Barrdahl,²¹ Caroline Baynes,⁹ Laura E Beane Freeman,²² Matthias W Beckmann,²³ Sabine Behrens,²¹ Javier Benitez,^2,245 Marina Bermisheva,²⁶ Carl Blomqvist,²⁷ Natalia V Bogdanova,^2,16,289 Stig E Bojesen,^30–32 Hiltrud Brauch,^33-35 Hermann Brenner,^3,17,356 Louise Brinton,³⁷ Per Broberg,³⁸ Sara Y Brucker,³⁹ Thomas Brüning,⁴⁰ Barbara Burwinkel,^4,412 Qiuyin Cai,¹ Trinidad Caldés,⁴³ Federico Canzian,⁴⁴ Brian D Carter,⁴⁵ Jose E Castelao,⁴⁶ Jenny Chang-Claude,^{21, 47} Georgia Chenevix-Trench,⁴⁸ Ting-Yuan David Cheng,⁴⁹ Christine L Clarke,⁵⁰ NBCS Collaborators,^51–65 Don M Conroy,⁹ Fergus J Couch,⁶⁶ David G Cox,^{67, 68} Angela Cox,⁶⁹ Simon S Cross,⁷⁰ Julie M Cunningham,⁶⁶ Kamila Czene,⁷¹ Mary B Daly,⁷² Kimberly F Doheny,⁷³ Thilo Dörk,²⁹ Isabel dos-Santos-Silva,⁷⁴ Martine Dumont,¹⁴ Alison M Dunning,⁹ Miriam Dwek,⁷⁵ H Shelton Earp,⁷⁶ Diana M Eccles,⁷⁷ A Heather Eliassen,^{13, 78} Christoph Engel,^79,80 Mikael Eriksson,⁷¹ D Gareth Evans,^81,82 Laura Fachal,⁹ Peter A Fasching,^23,83 Jonine Figueroa,^{37,84, 85} Olivia Fletcher,⁸⁶ Henrik Flyger,⁸⁷ Lin Fritschi,⁸⁸ Marike Gabrielson,⁷¹ Manuela Gago-Dominguez,^89,90 Susan M Gapstur,⁴⁵ Montserrat García-Closas,³⁷ Mia M Gaudet,⁴⁵ Maya Ghoussaini,⁹ Graham G Giles,^5,6 Mark S Goldberg,^91,92 David E Goldgar,⁹³ Anna González-Neira,²⁴ Pascal Guénel,⁹⁴ Eric Hahnen,^95–97 Christopher A Haiman,⁹⁸ Niclas Håkansson,⁹⁹ Per Hall,⁷¹ Emily Hallberg,¹⁰⁰ Ute Hamann,¹⁰¹ Patricia Harrington,⁹ Wei He,⁷¹ Alexander Hein,²³ Belynda Hicks,¹⁰² Peter Hillemanns,²⁹ Frans B Hogervorst,¹⁰³ Antoinette Hollestelle,¹⁰⁴ Robert N Hoover,³⁷ John L Hopper,⁶ Anthony Howell,¹⁰⁵ Guanmengqian Huang,¹⁰¹ Anna Jakubowska,¹⁰⁶ Wolfgang Janni,¹⁰⁷ Esther M John,^108–110 Nichola Johnson,⁸⁶ Kristine Jones,¹⁰² Audrey Jung,²¹ Rudolf Kaaks,²¹ Maria Kabisch,¹⁰¹ Michael J Kerin,¹¹¹ Elza Khusnutdinova,^26,112 Cari M Kitahara,³⁷ Veli-Matti Kosma,^113–115 Stella Koutros,²² Peter Kraft,^12,13 Vessela N Kristensen,^55,59,60 Diether Lambrechts,^116,117 Loic Le Marchand,¹¹⁸ Sara Lindström,^12,119 Martha S Linet,³⁷ Jolanta Lissowska,¹²⁰ Sibylle Loibl,¹²¹ Jan Lubinski,¹⁰⁶ Craig Luccarini,⁹ Michael P Lux,²³ Tom Maishman,^77,122 Ivana Maleva Kostovska,¹²³ Arto Mannermaa,^113–115 Siranoush Manoukian,¹²⁴ JoAnn E Manson,^13,125 Sara Margolin,¹²⁶ Dimitrios Mavroudis,¹²⁷ Hanne Meijers-Heijboer,¹²⁸ Alfons Meindl,¹²⁹ Usha Menon,¹³⁰ Jeffery Meyer,⁶⁶ Anna Marie Mulligan,^131,132 Susan L Neuhausen,¹³³ Heli Nevanlinna,¹³⁴ Patrick Neven,¹³⁵ William T Newman,^81,82 Sune F Nielsen,^3,301 Børge G Nordestgaard,^30–32 Olufunmilayo I Olopade,¹³⁶ Andrew F Olshan,¹³⁷ Janet E Olson,¹⁰⁰ Håkan Olsson,³⁸ Curtis Olswold,¹⁰⁰ Nick Orr,⁸⁶ Charles M Perou,¹³⁸ Julian Peto,⁷⁴ Dijana Plaseska-Karanfilska,¹²³ Ross Prentice,¹⁹ Nadege Presneau,⁷⁵ Katri Pylkäs,^139,140 Brigitte Rack,¹⁴¹ Paolo Radice,¹⁴² Nazneen Rahman,¹⁴³ Gadi Rennert,¹⁴⁴ Hedy S Rennert,¹⁴⁴ Atocha Romero,^43,145 Jane Romm,⁷³ Emmanouil Saloustros,¹⁴⁶ Dale P Sandler,¹⁴⁷ Elinor J Sawyer,¹⁴⁸ Rita K Schmutzler,^95–97 Andreas Schneeweiss,^41,149 Rodney J Scott,^150,151 Christopher Scott,¹⁰⁰ Sheila Seal,¹⁴³ Caroline Seynaeve,¹⁰⁴ Ann Smeets,¹³⁵ Melissa C Southey,¹⁵² John J Spinelli,^153,154 Jennifer Stone,^6,155 Harald Surowy,^4,412 Anthony J Swerdlow,^143,156 Rulla Tamimi,^12,13,78 William Tapper,⁷⁷ Jack A Taylor,^147,157 Mary Beth Terry,¹⁵⁸ Daniel C Tessier,¹⁵⁹ Kathrin Thöne,¹⁶⁰ Rob AEM Tollenaar,¹⁶¹ Diana Torres,^101,162 Melissa A Troester,¹³⁷ Thérèse Truong,⁹⁴ Michael Untch,¹⁶³ Celine Vachon,¹⁰⁰ David Van Den Berg,⁹⁸ Ans MW van den Ouweland,¹⁶⁴ Elke M van Veen,^81,82 Daniel Vincent,¹⁵⁹ Quinten Waisfisz,¹²⁸ Clarice R Weinberg,¹⁶⁵ Camilla Wendt,¹²⁶ Alice S Whittemore,^109,110 Hans Wildiers,¹³⁵ Robert Winqvist,^139,140 Alicja Wolk,⁹⁹ Lucy Xia,⁹⁸ Xiaohong R Yang,³⁷ Argyrios Ziogas¹⁵ and Elad Ziv.¹⁶⁶ Member affiliations:

¹Division of Epidemiology, Department of Medicine, Vanderbilt Epidemiology Center, Vanderbilt-Ingram Cancer Center, Vanderbilt University School of Medicine, Nashville, TN, USA, ²Division of Cardiovascular Medicine, Vanderbilt University Medical Center, Nashville, TN, USA, ³Centre for Cancer Genetic Epidemiology, Department of Public Health and Primary Care, University of Cambridge, Cambridge, UK, ⁴Department of Electron Microscopy/Molecular Pathology, The Cyprus Institute of Neurology and Genetics, Nicosia, Cyprus, ⁵Cancer Epidemiology & Intelligence Division, Cancer Council Victoria, Melbourne, Victoria, Australia, ⁶Centre for Epidemiology and Biostatistics, Melbourne School of Population and Global Health, The University of Melbourne, Melbourne, Victoria, Australia, ⁷Division of Molecular Pathology, The Netherlands Cancer Institute - Antoni van Leeuwenhoek Hospital, Amsterdam, The Netherlands, ⁸Division of Psychosocial Research and Epidemiology, The Netherlands Cancer Institute - Antoni van Leeuwenhoek hospital, Amsterdam, The Netherlands, ⁹Centre for Cancer Genetic Epidemiology, Department of Oncology, University of Cambridge, Cambridge, UK, ¹⁰Fred A, Litwin Center for Cancer Genetics, Lunenfeld-Tanenbaum Research Institute of Mount Sinai Hospital, Toronto, ON, Canada, ¹¹Department of Molecular Genetics, University of Toronto, Toronto, ON, Canada, ¹²Department of Epidemiology, University of California Irvine, Irvine, CA, USA, ¹³N,N, Alexandrov Research Institute of Oncology and Medical Radiology, Minsk, Belarus, ¹⁴Division of Clinical Epidemiology and Aging Research, German Cancer Research Center (DKFZ), Heidelberg, Germany, ¹⁵Department of Public Health Sciences, and Cancer Research Institute, Queen’s University, Kingston, ON, Canada, ¹⁶Cancer Prevention Program, Fred Hutchinson Cancer Research Center, Seattle, WA, USA, ¹⁷Zilber School of Public Health, University of Wisconsin-Milwaukee, Milwaukee, WI, USA, ¹⁸Division of Cancer Epidemiology, German Cancer Research Center (DKFZ), Heidelberg, Germany, ¹⁹Department of Health and Human Services, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA, ²⁰Department of Gynaecology and Obstetrics, University Hospital Erlangen, Friedrich-Alexander University Erlangen-Nuremberg, Comprehensive Cancer Center Erlangen-EMN, Erlangen, Germany, ²¹Human Cancer Genetics Program, Spanish National Cancer Research Centre, Madrid, Spain, ²²Centro de Investigación en Red de Enfermedades Raras (CIBERER), Valencia, Spain, ²³Institute of Biochemistry and Genetics, Ufa Scientific Center of Russian Academy of Sciences, Ufa, Russia, ²⁴Department of Oncology, Helsinki University Hospital, University of Helsinki, Helsinki, Finland, ²⁵Department of Radiation Oncology, Hannover Medical School, Hannover, Germany, ²⁶Gynaecology Research Unit, Hannover Medical School, Hannover, Germany, ²⁷Copenhagen General Population Study, Herlev and Gentofte Hospital, Copenhagen University Hospital, Herlev, Denmark, ²⁸Department of Clinical Biochemistry, Herlev and Gentofte Hospital, Copenhagen University Hospital, Herlev, Denmark, ²⁹Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark, ³⁰Dr Margarete Fischer-Bosch-Institute of Clinical Pharmacology, Stuttgart, Germany, ³¹University of Tübingen, Tübingen, Germany, ³²German Cancer Consortium (DKTK), German Cancer Research Center (DKFZ), Heidelberg, Germany, ³³Division of Preventive Oncology, German Cancer Research Center (DKFZ) and National Center for Tumor Diseases (NCT), Heidelberg, Germany, ³⁴Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, MD, USA, ³⁵Department of Cancer Epidemiology, Clinical Sciences, Lund University, Lund, Sweden, ³⁶Department of Women’s Health, University of Tübingen, Tübingen, Germany, ³⁷Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of the Ruhr University Bochum, Bochum, Germany, ³⁸Department of Obstetrics and Gynecology, University of Heidelberg, Heidelberg, Germany, ³⁹Molecular Epidemiology Group, C080, German Cancer Research Center (DKFZ), Heidelberg, Germany, ⁴⁰Medical Oncology Department, Hospital Clínico San Carlos, Instituto de Investigación Sanitaria San Carlos (IdISSC), Centro Investigación Biomédica en Red de Cáncer (CIBERONC), Madrid, Spain, ⁴¹Genomic Epidemiology Group, German Cancer Research Center (DKFZ), Heidelberg, ermany, ⁴²Epidemiology Research Program, American Cancer Society, Atlanta, GA, USA, ⁴³Oncology and Genetics Unit, Instituto de Investigacion Biomedica (IBI) Galicia Sur, Xerencia de Xestion Integrada de Vigo-SERGAS, Vigo, Spain, ⁴⁴Research Group Genetic Cancer Epidemiology, University Cancer Center Hamburg (UCCH), University Medical Center Hamburg-Eppendorf, Hamburg, Germany, ⁴⁵Department of Genetics and Computational Biology, QIMR Berghofer Medical Research Institute, Brisbane, Queensland, Australia, ⁴⁶Division of Cancer Prevention and Control, Roswell Park Cancer Institute, Buffalo, NY, USA, ⁴⁷Westmead Institute for Medical Research, University of Sydney, Sydney, New South Wales, Australia, ⁴⁸Department of Oncology, Haukeland University Hospital, Bergen, Norway, ⁴⁹Section of Oncology, Institute of Medicine, University of Bergen, Bergen, Norway, ⁵⁰Department of Pathology, Akershus University Hospital, Lørenskog, Norway, ⁵¹Department of Breast-Endocrine Surgery, Akershus University Hospital, Lørenskog, Norway, ⁵²Department of Cancer Genetics, Institute for Cancer Research, Oslo University Hospital Radiumhospitalet, Oslo, Norway, ⁵³Department of Breast and Endocrine Surgery, Oslo University Hospital, Ullevål, Oslo, Norway, ⁵⁴Department of Research, Vestre Viken Hospital, Drammen, Norway, ⁵⁵Department of Tumor Biology, Institute for Cancer Research, Oslo University Hospital Radiumhospitalet, Oslo, Norway, ⁵⁶Institute of Clinical Medicine, Faculty of Medicine, University of Oslo, Oslo, Norway, ⁵⁷Department of Clinical Molecular Biology, Oslo University Hospital, University of Oslo, Oslo, Norway, ⁵⁸National Advisory Unit on Late Effects after Cancer Treatment, Oslo University Hospital Radiumhospitalet, Oslo, Norway, ⁵⁹Department of Oncology, Oslo University Hospital Radiumhospitalet, Oslo, Norway, ⁶⁰Department of Radiology and Nuclear Medicine, Oslo University Hospital Radiumhospitalet, Oslo, Norway, ⁶¹Oslo University Hospital, Oslo, Norway, ⁶²Department of Oncology, Oslo University Hospital Ullevål, Oslo, Norway, ⁶³Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN, USA, ⁶⁴Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, London, UK, ⁶⁵INSERM U1052, Cancer Research Center of Lyon, Lyon, France, ⁶⁶Sheffield Institute for Nucleic Acids (SInFoNiA), Department of Oncology and Metabolism, University of Sheffield, Sheffield, UK, ⁶⁷Academic Unit of Pathology, Department of Neuroscience, University of Sheffield, Sheffield, UK, ⁶⁸Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden, ⁶⁹Department of Clinical Genetics, Fox Chase Cancer Center, Philadelphia, PA, USA, ⁷⁰Center for Inherited Disease Research (CIDR), Institute of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, USA, ⁷¹Department of Non-Communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, UK, ⁷²Genomics Center, Centre Hospitalier Universitaire de Québec Research Center, Laval University, Québec City, QC, Canada, ⁷³Department of Biomedical Sciences, Faculty of Science and Technology, University of Westminster, London, UK, ⁷⁴Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA, ⁷⁵Cancer Sciences Academic Unit, Faculty of Medicine, University of Southampton, Southampton, UK, ⁷⁶Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA, ⁷⁷Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA, ⁷⁸Institute for Medical Informatics, Statistics and Epidemiology, University of Leipzig, Leipzig, Germany, ⁷⁹LIFE - Leipzig Research Centre for Civilization Diseases, University of Leipzig, Leipzig, Germany, ⁸⁰Division of Evolution and Genomic Sciences, School of Biological Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK, ⁸¹Manchester Centre for Genomic Medicine, St, Mary’s Hospital, Central Manchester University Hospitals NHS Foundation Trust, Manchester Academic Health Science Centre, Manchester, UK, ⁸²David Geffen School of Medicine, Department of Medicine Division of Hematology and Oncology, University of California at Los Angeles, Los Angeles, CA, USA, ⁸³Usher Institute of Population Health Sciences and Informatics, The University of Edinburgh Medical School, Edinburgh, UK, ⁸⁴,Cancer Research UK Edinburgh Centre, Edinburgh, UK, ⁸⁵The Breast Cancer Now Toby Robins Research Centre, The Institute of Cancer Research, London, UK, ⁸⁶Department of Breast Surgery, Herlev and Gentofte Hospital, Copenhagen University Hospital, Herlev, Denmark, ⁸⁷School of Public Health, Curtin University, Perth, Western Australia, Australia, ⁸⁸Genomic Medicine Group, Galician Foundation of Genomic Medicine, Instituto de Investigación Sanitaria de Santiago de Compostela (IDIS), Complejo Hospitalario Universitario de Santiago, SERGAS, Santiago de Compostela, Spain, ⁸⁹Moores Cancer Center, University of California San Diego, La Jolla, CA, USA, ⁹⁰Department of Medicine, McGill University, Montréal, QC, Canada, ⁹¹Division of Clinical Epidemiology, Royal Victoria Hospital, McGill University, Montréal, QC, Canada, ⁹²Department of Dermatology, Huntsman Cancer Institute, University of Utah School of Medicine, Salt Lake City, UT, USA, ⁹³Cancer & Environment Group, Center for Research in Epidemiology and Population Health (CESP), INSERM, University Paris-Sud, University Paris-Saclay, Villejuif, France, ⁹⁴Center for Hereditary Breast and Ovarian Cancer, University Hospital of Cologne, Cologne, Germany, ⁹⁵Center for Integrated Oncology (CIO), University Hospital of Cologne, Cologne, Germany, ⁹⁶Center for Molecular Medicine Cologne (CMMC), University of Cologne, Cologne, Germany, ⁹⁷Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA, ⁹⁸Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden, ⁹⁹Department of Health Sciences Research, Mayo Clinic, Rochester, MN, USA, ¹⁰⁰Molecular Genetics of Breast Cancer, German Cancer Research Center (DKFZ), Heidelberg, Germany, ¹⁰¹Cancer Genomics Research Laboratory (CGR), Division of Cancer Epidemiology and Genetics, Frederick National Laboratory for Cancer Research, Rockville, MD, USA, ¹⁰²Family Cancer Clinic, The Netherlands Cancer Institute - Antoni van Leeuwenhoek hospital, Amsterdam, The Netherlands, ¹⁰³Department of Medical Oncology, Family Cancer Clinic, Erasmus MC Cancer Institute, Rotterdam, The Netherlands, ¹⁰⁴Institute of Cancer studies, University of Manchester, Manchester, UK, ¹⁰⁵Department of Genetics and Pathology, Pomeranian Medical University, Szczecin, Poland, ¹⁰⁶Department of Gynecology and Obstetrics, University Hospital Ulm, Ulm, Germany, ¹⁰⁷Department of Epidemiology, Cancer Prevention Institute of California, Fremont, CA, USA, ¹⁰⁸Department of Health Research and Policy - Epidemiology, Stanford University School of Medicine, Stanford, CA, USA, ¹⁰⁹Stanford Cancer Institute, Stanford University School of Medicine, Stanford, CA, USA, ¹¹⁰School of Medicine, National University of Ireland, Galway, Ireland, ¹¹¹Department of Genetics and Fundamental Medicine, Bashkir State University, Ufa, Russia, ¹¹²Translational Cancer Research Area, University of Eastern Finland, Kuopio, Finland, ¹¹³Institute of Clinical Medicine, Pathology and Forensic Medicine, University of Eastern Finland, Kuopio, Finland, ¹¹⁴Imaging Center, Department of Clinical Pathology, Kuopio University Hospital, Kuopio, Finland, ¹¹⁵Program in Genetic Epidemiology and Statistical Genetics, Harvard T,H, Chan School of Public Health, Boston, MA, USA, ¹¹⁶VIB Center for Cancer Biology, VIB, Leuven, Belgium, ¹¹⁷Laboratory for Translational Genetics, Department of Human Genetics, University of Leuven, Leuven, Belgium, ¹¹⁸Epidemiology Program, University of Hawaii Cancer Center, Honolulu, HI, USA, ¹¹⁹Department of Epidemiology, University of Washington School of Public Health, Seattle, WA, USA, ¹²⁰Department of Cancer Epidemiology and Prevention, M Sklodowska-Curie Memorial Cancer Center & Institute of Oncology, Warsaw, Poland, ¹²¹German Breast Group, GmbH, Neu Isenburg, Germany, ¹²²Southampton Clinical Trials Unit, Faculty of Medicine, University of Southampton, Southampton, UK, ¹²³Research Centre for Genetic Engineering and Biotechnology “Georgi D, Efremov”, Macedonian Academy of Sciences and Arts, Skopje, Republic of Macedonia, ¹²⁴Unit of Medical Genetics, Department of Medical Oncology and Hematology, Fondazione IRCCS (Istituto Di Ricovero e Cura a Carattere Scientifico) Istituto Nazionale dei Tumori (INT), Milan, Italy, ¹²⁵Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA, ¹²⁶Department of Oncology - Pathology, Karolinska Institutet, Stockholm, Sweden, ¹²⁷Department of Medical Oncology, University Hospital of Heraklion, Heraklion, Greece, ¹²⁸Radiation Oncology, Hospital Meixoeiro-XXI de Vigo, Vigo, Spain, ¹²⁹Division of Gynaecology and Obstetrics, Technische Universität München, Munich, Germany, ¹³⁰Gynaecological Cancer Research Centre, Women’s Cancer, Institute for Women’s Health, University College London, London, UK, ¹³¹Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, ON, Canada, ¹³²Laboratory Medicine Program, University Health Network, Toronto, ON, Canada, ¹³³Department of Population Sciences, Beckman Research Institute of City of Hope, Duarte, CA, USA, ¹³⁴Department of Obstetrics and Gynecology, Helsinki University Hospital, University of Helsinki, Helsinki, Finland, ¹³⁵Leuven Multidisciplinary Breast Center, Department of Oncology, Leuven Cancer Institute, University Hospitals Leuven, Leuven, Belgium, ¹³⁶Center for Clinical Cancer Genetics and Global Health, The University of Chicago, Chicago, IL, USA, ¹³⁷Department of Epidemiology, Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA, ¹³⁸Department of Genetics, Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA, ¹³⁹Laboratory of Cancer Genetics and Tumor Biology, Cancer and Translational Medicine Research Unit, Biocenter Oulu, University of Oulu, Oulu, Finland, ¹⁴⁰Laboratory of Cancer Genetics and Tumor Biology, Northern Finland Laboratory Centre Oulu, Oulu, Finland, ¹⁴¹Department of Gynecology and Obstetrics, Ludwig-Maximilians University of Munich, Munich, Germany, ¹⁴²Unit of Molecular Bases of Genetic Risk and Genetic Testing, Department of Research, Fondazione IRCCS (Istituto Di Ricovero e Cura a Carattere Scientifico) Istituto Nazionale dei Tumori (INT), Milan, Italy, ¹⁴³Division of Genetics and Epidemiology, The Institute of Cancer Research, London, UK, ¹⁴⁴Clalit National Cancer Control Center, Carmel Medical Center and Technion Faculty of Medicine, Haifa, Israel, ¹⁴⁵Medical Oncology Department, Hospital Universitario Puerta de Hierro, Madrid, Spain, ¹⁴⁶Hereditary Cancer Clinic, University Hospital of Heraklion, Heraklion, Greece, ¹⁴⁷Epidemiology Branch, National Institute of Environmental Health Sciences, NIH, Research Triangle Park, NC, USA, ¹⁴⁸Research Oncology, Guy’s Hospital, King’s College London, London, UK, ¹⁴⁹National Center for Tumor Diseases, University of Heidelberg, Heidelberg, Germany, ¹⁵⁰Division of Molecular Medicine, Pathology North, John Hunter Hospital, Newcastle, New South Wales, Australia, ¹⁵¹Discipline of Medical Genetics, School of Biomedical Sciences and Pharmacy, Faculty of Health, University of Newcastle, Callaghan, New South Wales, Australia, ¹⁵²Department of Pathology, The University of Melbourne, Melbourne, Victoria, Australia, ¹⁵³Cancer Control Research, BC Cancer Agency, Vancouver, BC, Canada, ¹⁵⁴School of Population and Public Health, University of British Columbia, Vancouver, BC, Canada, ¹⁵⁵The Curtin UWA Centre for Genetic Origins of Health and Disease, Curtin University and University of Western Australia, Perth, Western Australia, Australia, ¹⁵⁶Division of Breast Cancer Research, The Institute of Cancer Research, London, UK, ¹⁵⁷Epigenetic and Stem Cell Biology Laboratory, National Institute of Environmental Health Sciences, NIH, Research Triangle Park, NC, USA, ¹⁵⁸Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, NY, USA, ¹⁵⁹McGill University and Génome Québec Innovation Centre, Montréal, QC, Canada, ¹⁶⁰Department of Cancer Epidemiology, Clinical Cancer Registry, University Medical Center Hamburg-Eppendorf, Hamburg, Germany, ¹⁶¹Department of Surgery, Leiden University Medical Center, Leiden, The Netherlands, ¹⁶²Institute of Human Genetics, Pontificia Universidad Javeriana, Bogota, Colombia, ¹⁶³Department of Gynecology and Obstetrics, Helios Clinics Berlin-Buch, Berlin, Germany, ¹⁶⁴Department of Clinical Genetics, Erasmus University Medical Center, Rotterdam, The Netherlands, ¹⁶⁵Biostatistics Branch, National Institute of Environmental Health Sciences, NIH, Research Triangle Park, NC, USA and ¹⁶⁶Department of Medicine, Institute for Human Genetics, UCSF Helen Diller Family Comprehensive Cancer Center, University of California San Francisco, San Francisco, CA, USA

Conflict of interest: None declared.