Lene Mellemkjær

Women with estrogen receptor negative (ER‐) contralateral breast cancer (CBC) are 50% more likely to die of CBC compared to women with estrogen receptor positive (ER+) CBC. We comprehensively investigated risk factors for ER subtype‐specific CBC using the Women's Environmental Cancer and Radiation Epidemiology Study, a population‐based, case–control study of 1521 CBC cases under age 55 years and 2212 individually matched unilateral breast cancer controls. Information about breast cancer risk factors and ER status was obtained by interview and medical record abstraction and common genetic variants with GWAS. Multivariable rate ratios and 95% confidence intervals were estimated using conditional logistic regression models, adjusting for known and suspected risk factors. Among 2800 women where the matched case had known ER CBC status, chemotherapy, hormone therapy, and increasing number of full‐term pregnancies were associated with decreased ER+ CBC risk. Drinking, smoking, first breast cancer lobular histology, first‐degree breast cancer family history, a polygenic risk score (PRS) of 313 common variants, and an ER+ subtype‐specific PRS were associated with increased ER+ CBC risk. More than 2 years of breastfeeding was associated with decreased ER− CBC risk. While none of the PRSs were associated with ER− CBC risk, first‐ and second‐degree family history of breast cancer and second‐degree family history of ovarian cancer were associated with increased ER− CBC risk. We identified differential risk factors for ER subtype‐specific CBC which can inform individualized risk prediction models for future clinical impact. There remains an unexplained genetic component of ER− CBC.

Background: Despite improved treatments and survival, breast cancer (bc) remains a leading cause of cancer mortality in women. Thyroid dysfunction has been linked to bc risk, but its impact on bc survival is less clear. We investigated associations between pre-diagnosis thyroid-related biomarkers and all-cause and bc-specific survival. Methods: A prospective cohort study including 1,513 women with invasive bc from 7 European countries in the European Prospective Investigation into Cancer and Nutrition (EPIC) was conducted. We measured thyroid-stimulating hormone (TSH), free triiodothyronine (fT3), free thyroxine (fT4), and anti-thyroid peroxidase antibodies (Anti-TPO) in samples collected ∼8 years pre-diagnosis. We evaluated associations between these markers and all-cause and bc-specific mortality using Cox proportional hazards models adjusted for relevant covariates. Results: After mean follow-up postdiagnosis of 7 years, 223 deaths occurred, including 161 from bc. We observed no overall associations between circulating levels of TSH, thyroid hormones, thyroid autoimmunity, and mortality. However, in stratified analyses, TSH was inversely associated with all-cause (HR1-standard deviation (SD)=0.70; 95%CI = 0.52-0.94) and bc-specific-mortality (HR1-SD=0.61(0.44-0.84) in premenopausal women, while fT4 was associated with all-cause mortality in metastatic bc (HR1-SD=1.67(1.20 to 2.32), and with bc-specific mortality among users of menopausal/contraceptive hormones (HR1-SD=2.76(1.47-5.19). Anti-TPO-positivity was inversely associated with all-cause mortality in women with body mass index <25kg/m2 and in women with oestrogen receptor-negative tumours, separately. Conclusions: Thyroid function before diagnosis does not appear to play a major role in bc survival. Although subgroup-specific associations were identified, these findings should be interpreted with caution due to limited statistical power and warrant confirmation in independent populations.

Background: Cardiovascular disease is a leading cause of death for long-term breast cancer survivors. We evaluated whether a polygenic risk score for coronary artery disease (CAD-PRS) was associated with the risk of incident CAD for survivors of unilateral or contralateral breast cancer. Methods: The study included 1307 women with breast cancer first diagnosed at younger than 55 years of age who participated in the Women’s Environmental Cancer and Radiation Epidemiology Follow-up Study. The CAD-PRS was based on a PRS developed and validated in a separate population. We modeled the association between incident CAD and the CAD-PRS, adjusting for age, CAD risk factors, first (and second) breast cancer treatment, study recruitment phase, and genetic population stratification. We also explored whether the risk of CAD depended on interactions between the CAD-PRS and cardiotoxic cancer treatment. Results: There were 66 incident CAD diagnoses reported at a median of 16 years after breast cancer diagnosis. Participants with CAD-PRS at or above the median had a 2.48-times increased risk of CAD (95% confidence interval [CI] = 1.44 to 4.29) relative to participants with CAD-PRS below the median. Anthracycline-based chemotherapy was associated with increased CAD risk (hazard ratio [HR] = 2.04, 95% CI = 1.04 to 3.98), and the association was not modified by the CAD-PRS. The association between incident CAD and left-sided radiation therapy (RT) was increased for those with CAD-PRS at or above the median (HR = 2.90, 95% CI = 1.26 to 6.68) but not for those with CAD-PRS below the median (HR = 0.96, 95% CI = 0.32 to 2.88). There was evidence of super-additive interaction between the CAD-PRS and left-sided RT (relative excess risk due to interaction = 2.06, 95% CI = 0.05 to 4.06). Conclusion: A genome-wide CAD-PRS was associated with nonfatal CAD risk for long-term breast cancer survivors, providing potential utility for personalized cardiovascular care, particularly after RT.

Breast cancer (BC) is a heterogeneous disease with subtypes based on receptor status (estrogen receptor [ER], progesterone receptor [PR], and human epidermal growth factor receptor 2 [HER2]), influencing prognosis and treatment. A higher socioeconomic position (SEP) is associated with an increased BC risk, but its relation to BC subtypes is less clear. This study analyzed 311,631 women from the EPIC cohort, focusing on the incidence of in situ and invasive BC (overall and by receptor status and subtype). Educational attainment was used as a proxy for SEP, and hazard ratios (HRs) were calculated using Cox regression models. Mediation analyses were performed to evaluate the extent to which selected risk factors explained the educational gradient. Over 14 years, 14,432 BC cases were identified, including 12,863 invasive cases. Lower education was associated with a reduced risk of both in situ and invasive BCs. The HRs for primary versus tertiary education were 0.61 (95% CI 0.49–0.73) for in situ and 0.81 (95% CI 0.75–0.87) for invasive BC overall, with similar reductions across ER‐positive, PR‐positive, HER2‐positive, Luminal A, BH−, and BH+. No significant association was found between education and ER‐negative, and HER2‐enriched BCs. Reproductive and lifestyle factors explained 20–40% of the educational differences in BC risk. While many of the risk factors through which education impacts the development of subtype‐specific BC were identified, others remain to be fully elucidated. Differences in screening attendance could partially explain the higher ER‐positive BC risk among highly educated; this study further contributes to the understanding of the complex nature of BC in terms of its social gradient and aetiology.

Aims: Observational studies have shown a U-shaped association between age at menarche (AAM) and cardiovascular disease (CVD). We assessed non-linearity of the observational association and the potential causal relationship between AAM and CVD using data from the European Prospective Investigation into Cancer and Nutrition—Cardiovascular Disease (EPIC-CVD) study and the UK Biobank. Methods and results: We included women without pre-existing myocardial infarction (MI) or stroke at baseline. We estimated hazard ratios for incident MI, ischaemic and haemorrhagic stroke later in life using Cox regression in observational analyses and conducted non-linear Mendelian randomization (MR) based on fractional polynomials and linear MR based on inverse-variance weighted regression. We analysed 283 210 women with a median AAM of 13 (IQR 12–14) years in both EPIC-CVD and the UK Biobank, of which 8468 experienced a MI, 5501 an ischaemic and 1887 a haemorrhagic stroke. The association between AAM and MI and ischaemic stroke was U-shaped with higher risks in women aged ≤12 and ≥16 compared with those aged 13 years at menarche. Our MR analyses found no evidence for non-linearity between genetically proxied AAM and any CVD endpoint later in life, but each year higher genetically proxied AAM was related to a lower risk of MI (hazard ratio 0.92 [95% CI 0.86–0.99]), but not to ischaemic and haemorrhagic stroke. Conclusion: This study supported non-linear observational associations between AAM and MI and ischaemic stroke. MR analyses suggested a causal relationship between higher AAM and risk of MI without an indication for non-linearity. There was no support for a potential causal link with ischaemic and haemorrhagic stroke.