The analysis of the data collected within the E3N prospective cohort showed that postmenopausal women taking MHT have higher PMD and DA and lower NDA than never users of MHT and that this difference is observed already within the first year of MHT use. Stopping MHT leads to a decrease of PMD and DA whose levels reach those of never users after more than 8 years since stopping the therapy. No heterogeneity was observed between estrogen alone and estrogen plus progestogen MHT formulations. The effect of current use of MHT on BC risk was partially direct and partially mediated by PMD and BMI; the mediated effect was entirely restricted to the hormone receptor positive tumors.
The association between MHT use and MD has been previously reported. A randomized study estimated that after one year since starting therapy PMD increased of about ten percentage point in women in the estrogen-progestin arm, whereas no change was observed in women taking placebo 8; this longitudinal change was comparable to our estimate of a difference of 8 percentage points in PMD (estimated from the polynomial model) between a woman aged 56 years who took MHT for 1 year and a never user 55 years old. To our knowledge, our study is the first trying to model MD in a prospective cohort of postmenopausal women by pattern of MTH use including duration of use and time since last use. Our observation that women taking MHT for less than one year had higher MD levels than never users and that the difference was observed also in those who stopped treatment less than 8 years earlier is consistent with the findings about the association between MHT and BC risk. Previous analyses of the full E3N cohort reported an effect of MHT on BC risk already in the first two years after starting 16,17. A meta-analysis conducted on more than 100 000 BC cases from prospective studies estimated that the increased risk of BC associated with MHT in current users appeared in the first 5 years of use and almost doubled in the following 5–14 years of use; in past users, excess risk persisted even after 10 years since stopping 7.
According to our data, MD responds to MHT following a trend similar to that of BC risk, consistently with its role as mediator of the effect of MHT on BC risk. We estimated that PMD mediated between 34% and 50% of the effect of current use of MHT on MD, an effect comparable to the 31% effect reported in the Nurses’ Health Study 9, but much smaller than the 100% reported in the case-control study nested within the randomized Women Health Initiative 8.
In the meta-analysis by the Collaborative Group on Hormonal Factors in Breast Cancer, the increased risk of BC associated with MHT was observed for all type of MHT except vaginal estrogen and the risk was greater for estrogen plus progestogen than for estrogen alone; moreover, the risk was higher for ER positive than ER negative BCs 7. The analyses conducted on the entire E3N cohort identified heterogeneity of the effect on BC risk of the type of MHT preparation and provided evidence for a differential effect on risk by BC subtype 17. Our present analysis did not show evidence for heterogeneity of the effect of MHT on MD by formulation, although the number of women taking estrogen only was too small to provide an adequate power. As to the subtype of cancer, consistently with the previous observation on BC risk, we observed that PMD mediated the effect of MHT on BC risk only for hormone receptor positive BCs, but nor for hormone receptor negative BCs.
The main strengths of our study were its prospective design and the completeness of information about MHT pattern of use over a median period of time of 11 years. The richness of the information available from the E3N cohort allowed us to adjust for all known potential confounders of the relations between MHT, MD and BC risk, essential for an unbiased estimate of the mediated causal effect 18. One limitation of our study was the small proportion of women taking estrogen alone MHT, which did not allow to properly test for a difference between estrogen alone and estrogen plus progestagen. Also, not having repeated longitudinal measures of MD made it impossible to estimate intra-individual changes in MD over time.
The use of MHT in Western countries saw a strong increase in the nineties when studies suggested its beneficial effect on postmenopausal women health 19–22. After the first results from the Women Health Initiatives in 2002 23 reporting an excess risk of BC and cardiovascular diseases in the estrogen plus progestin arm compared to the placebo arm, the number of MHT consumers abruptly decreased. Subsequent reanalysis of the follow-up data of the same trial and independent studies suggested that the benefits of MHT taken over menopause overcome its negative effects, resulting in an overall improved survival 24–28. It has been estimated that in 2010 there were about 12 million users in Western countries 7. The meta-analysis by the Collaborative Group on Hormonal Factors in Breast Cancer published in 2019 provided new elements for the ongoing debate about the safety of use of MHT 7,29.