Background 
Anastrozole, an aromatase inhibitor, has recently been approved as a preventive therapy for breast cancer in high-risk women based on findings from the IBIS-II trial. The trial observed that 5-year anastrozole use lowered the risk of breast cancer, reduced bone density and increased risk of hypertension. However, the long-term effects remain largely unclear. This study builds on the findings from IBIS-II, investigating on-target effects associated with prolonged aromatase inhibitor use, to explore the safety profile and repurposing opportunities of anastrozole. 

Methods 
The rs727479 variant in CYP19A1 gene, that mimics aromatase inhibition, served as a proxy to estimate effects on selected outcomes. Associations between genetically proxied aromatase inhibition and established effects of anastrozole observed in IBIS-II, including reduction in oestrogen-receptor positive (ER+) breast cancer risk and bone-mineral density (BMD), were tested for instrument validation. To evaluate effects of aromatase inhibition on other cancers beyond breast cancer, summary genetic association data for 5 cancers (colorectal, endometrial, lung, skin and ovarian cancer) were obtained from large-scale genome-wide association study consortia. We further examined potential adverse effects not measured in IBIS-II by performing a phenome-wide association study (PheWAS) of 449 disease outcomes and health-related traits in 162,360 postmenopausal women in the UK Biobank. Finally, we examined whether effects of long-term aromatase inhibition varied across clinically relevant patient subgroups, including body mass index and polygenic risk scores.
 
Results
Consistent with findings from the IBIS-II trial, genetically-proxied aromatase inhibition was associated with reduced risk of ER+ breast cancer (OR: 0.78, 95% CI: 0.67-0.92) and decreased heel bone mineral density (-0.32 SD change, 95% CI: -0.36 to -0.28). When evaluating the repurposing potential of aromatase inhibitors to other cancers, we found evidence that genetically-proxied aromatase inhibition was associated with reduced endometrial cancer risk (OR 0.34, 95% CI 0.26-0.44). The IBIS-II estimate for endometrial cancer risk (HR: 0.72, 95% CI: 0.18-2.65) was directionally consistent with our MR findings but imprecisely estimated likely due to limited statistical power. The PheWAS identified an association of genetically-proxied aromatase inhibition with six outcomes (P-FDR<0.05) including reduced risk of endometrial polyps (OR: 0.58, 95% CI: 0.45-0.74) and postmenopausal bleeding (OR: 0.67, 95% CI: 0.54-0.83). Subgroup analyses suggested that genetically-proxied aromatase inhibition was associated with greater reductions in risk of endometrial polyps (P-LRT=1.26×10−3) and postmenopausal bleeding (P-LRT=0.02) in women with higher body mass index.
 
Conclusions
The findings from this study reinforce associations observed in the IBIS-II trial, thereby demonstrating that Mendelian randomisation can serve as a valuable tool to recapitulate known effects of preventative therapy agents on cancer risk. Our results suggest a potential role for anastrozole in endometrial cancer prevention, warranting further investigation. Our PheWAS identified little evidence for adverse effects beyond the established effect on bone mineral density and subgroup analyses suggested that in women with higher BMI, aromatase inhibition may offer greater protection against endometrial conditions.