Assessing the effect of medically assisted reproduction on hormone-related cancers in women: an emulated target trial

1.ObjectiveTo determine the risk of hormone-related cancers following medically assisted reproductive treatment. DesignEmulated target trial. SettingAustralian health registries and administrative datasets. Participants1,748,927 women enrolled in Medicare, Australias universal health insurance scheme, aged 18 to 55 between 1 January 1991 and 31 December 2018. InterventionsThree exposures were defined from Medicare records of reimbursement for consultations, procedures and medications: assisted reproduction therapy (ART); intrauterine insemination/ovarian stimulation (IUI/OS); and ovulation induction with Clomiphene Citrate. Main outcome measuresHormone-related invasive cancers included breast, ovarian, uterine, thyroid, colorectal, and melanoma; in-situ cancers included breast and melanoma. Three cancers with no established hormonal links and high incidence - pancreatic, lung, and haematological - were included as negative controls. Flexible parametric survival models ascertained hazard ratios and cumulative marginal differences in incident cancers per 100,000 women. E-values assessed the risk of bias due to unmeasured confounding variables. ResultsThough most hormone-related cancers were elevated after MAR treatment (HRs: 1.09-1.64), E-value analysis suggested confounding due to underlying infertility conditions (endometriosis, polycystic ovarian syndrome) could account for this observed elevation for uterine, ovarian, and thyroid cancers. For any specific invasive cancer, fewer than 20 extra cancers per 100,000 women each year were predicted for treated versus comparator women. Emulated trials on the six hormone-related cancers showed increased cancer risk in the first years after treatment, however this was also observed for pancreatic and haematological cancers, suggesting detection bias. Haematological cancers showed an increased risk after treatment (HRs: 1.19-1.27), indicating uncontrolled confounding by ethnicity may account for the excess risk observed for haematological cancers and melanoma. Lung cancer risk was decreased after some treatments (HRs: 0.73-0.83). ConclusionsThough MAR showed an association with some hormone-related cancers, the absolute difference in the number of expected cancers was small and may be explained by unmeasured confounding and detection bias. 2. SUMMARY BOXESO_ST_ABSSection 1: What is already known on this topicC_ST_ABSO_LIThere is a biologically plausible link between medically assisted reproduction therapies and hormone-related cancers. C_LIO_LICurrent evidence suggests exposure to medically assisted reproduction does not increase the risk of most hormone-related cancers. C_LIO_LIAssessments of causation in this context are challenging, as women undertaking medically assisted reproduction have a different hormone-related cancer risk profile to those who do not. C_LI Section 2: What this study addsO_LIThis study is one of the largest examinations of womens cancer risk following medically assisted reproduction and the first to use both the emulated target trial framework and negative control cancers, and to quantify the potential for unmeasured confounding. C_LIO_LIMost hormone-related cancers show a small elevation in relative and absolute risk following medically assisted reproduction therapies, though this may be due in part or whole to detection bias and unmeasured confounding from infertility-related conditions like endometriosis or polycystic ovary syndrome, anovulation, obesity, and ethnicity. C_LIO_LIThis study provides a comprehensive assessment of the available evidence and guidance for improving future research on medically assisted reproduction and cancer. C_LI