Re: Effect of hormone replacement therapy on cardiovascular events in recently postmenopausal women: randomised trial
In response to: http://www.bmj.com/content/345/bmj.e6409
We are thankful for all of the interest in the above publication and hope to have responded to all relevant comments.
As stated in our article, the outcomes in the present study were secondary outcome measures in the original protocol. The primary outcome of the original study should be of no surprise to Dr. Schroll et al. or to others, and it was stated clearly in the manuscript (Methods).
The primary end-point of the present study was defined in the protocol as safety measures for cardiovascular events. These were outcomes defined before data analysis and it was decided to include only hard endpoints that are associated with poor prognosis: all-cause mortality and acute myocardial infarction and heart failure hospital admissions as these are also events validated in the National Danish Registers. There is a risk that clinical silent events are overlooked, but mortality is always captured. Thus, it was decided before data analyses not to include softer endpoints (driven by clinical decisions) that could also be associated with a reduction in mortality (angina, CABG or PCI) as pointed out by Dr Singh et al. For the same reason we did not report on intermediary markers of CVD such as inflammatory measures that are even softer outcome data.
Although randomization is the most critical aspect of any trial, we agree with Dr Schroll et al, Dr Turgeon et al and Dr Roussouw et al that performance and detection bias could be an issue in open-label trials, but the importance of these biases are reduced with endpoints driven by mortality. Further, it is well-understood that no HRT trial—including the WHI—conducted in women with a uterus is completely blinded or placebo-controlled [2;3]. Thus, our results are no less valid than any other randomized HRT trial, especially conducted in women with a uterus. As suggested by Dr Roizen, the benefits seen in DOPS with respect to cardiovascular effects could also have been larger if randomized treatment had continued for more than 10 years. In regards to this latter point, DOPS randomized treatment was stopped because beginning in 2002 it was believed that HRT had negative effects on cardiovascular health and thus, if that was the case, should have resulted in a higher detection rate of CVD in the subsequent post-randomization years of follow-up. However, this was not the case.
Whereas it is well-appreciated that detection bias likely played a role regarding breast cancer detection in the WHI-E+P trial as many women (and their physicians) were un-blinded to HRT treatment, and women were warned of the increased risk of breast cancer related to HRT [2;3], detection bias may have also played a role in breast cancer detection in DOPS. However, despite the likelihood of over-estimating breast cancer in the HRT group due to detection bias, breast cancer risk was in fact reduced by 42% (HR, 0.58; 95% CI, 0.27-1.27) relative to the control group after 10 years of randomized treatment in DOPS. Detection bias in regards to breast cancer is indeed something the Cochrane Center has been concerned about previously. [4;5]Data on actual numbers of cancer events are presented in the results, thus, we are uncertain which data Dr Abrahamson is missing.
As pointed out by several readers (Dr Schroll et al, Dr Abrahamsen, Dr, Turgeon et al, Dr Roussouw et al) and as stated in the limitations of the study, the power of our study is low, but this does not change our findings, as our results does not change the results of any other trial. We completely agree that power is low to detect breast cancer, but power is not a problem in relation to the significant findings driven by all-cause mortality.
We found a decreasing trend in breast cancer risk with HRT (consistent with the WHI-E trial) and we seek to understand our findings in the context of the included women. Thus, the timing hypothesis is a possible explanation.
Although the WHI trials were large, WHI lacks power in a population similar to ours, as acknowledged by Dr Roussouw et al. Unlike WHI, DOPS was a priori designed to “mimic the normal clinical situation”  in which perimenopausal and recently postmenopausal women are initiated on HRT. However, in post-hoc analyses from the WHI trials, the HR for CHD and total mortality for women less than 10 years-since-menopause when initiated on HRT was 0.76 (95% CI, 0.50-1.16) and 0.76 (95% CI 0.53-1.09), respectively. These post-hoc analyses are limited in that they were conducted in a sub-group of women from the WHI trials that reflect a much wider span of women relative to DOPS in which the maximum time-since-menopause was only 2 years. Significant reductions in CVD and total mortality have also been reported from meta-analyses of randomized trials of women who were initiated on HRT before age 60 years and within 10 years-since-menopause, 32% and 39%reductions, respectively, relative to placebo.
It is important to realize that DOPS and the WHI trials substantially differ in regards to HRT medications and regimens, age and time-since-menopause when women were randomized; WHI studied women who were on average 64 years old and >10 years-since-menopause when randomized which does not at all replicate clinical practice and therefore, cannot be generalized to the normal practice of initiating HRT at or near the time of menopause. However, DOPS tested the hypothesis that when initiated at or near menopause, HRT reduces CVD and does so safely with no increase in breast cancer or stroke, and importantly with a reduction in the primary endpoint including total mortality. We did in fact analyse the DOPS data separately for estradiol + sequential NETA and estradiol alone (labelled as intact uterus and hysterectomy in Figures 3-6 ); these data show that both HRT regimens had similar effects on all outcomes.
The results from DOPS are consistent with the more than 40 observational studies,[12;13]meta-analyses of randomized trials [10;11;14]and post-hoc analyses of the subgroup of young postmenopausal women <60 years of age and/or <10 years-since-menopause when initiated on HRTthat all consistently show that HRT reduces CVD and total mortality. Further analyses of the women <2 years-since-menopause in the WHI trials would be more similar to the DOPS cohort and would indeed be very welcome as it may add knowledge to the effect of HRT in recently postmenopausal women.
Naturally, there were few events (including myocardial infarctions) in DOPS due to the relatively young age-group. We are however certain that the diagnoses of outcome events were captured by the Danish registers, and this is in fact a strength of the trial as in DOPS we had almost 100% ascertainment of events over the 10-year randomization period and over the total 16-year follow-up.
We wish to thank Dr McPherson for bringing smaller RCTs to our attention. We did not have the intention to discuss the vast HRT data, but these studies are relevant to our discussions.
In conclusion, DOPS is the only prospective longitudinal randomized trial designed to examine clinical outcomes among women who were specifically a priori randomized to HRT in the perimenopausal/early postmenopausal period. When randomized to DOPS, women were on average 50 years old and 7 months postmenopausal and there is no comparable cohort or outcome data available for comparison from any other randomized trial. Women were randomized for 10 years (~10,000 women-years of randomization), the longest randomized duration than any randomized trial examining clinical outcomes resulting from HRT in women on average 50 years old. Subsequent to randomized treatment, the women were followed for another 6 years for a total follow-up of 16 years (~16,000 women-years of follow-up). DOPS provides long-term longitudinal randomized trial data in a cohort of women, namely, the very women who are normally treated, for which no data previously existed and thus, directly informs women and health-care providers of the long-term effects of HRT after initiation in close proximity to menopause, especially in regards to breast cancer, CVD and total mortality. In addition, unlike other large-scale long-term randomized HRT trials of efficacy and safety, no participants were lost to follow-up (2 women were censored at the time of emigration) in DOPS that had nearly 100% ascertainment of outcome events over 10 years of randomized follow-up and 16 years of total follow-up using the Danish civil registration system and the Danish national hospital discharge register. Finding significant effects in small groups underscores the clinical relevance for efficacy in reducing CVD and total morality and lack of harmful effects on breast cancer and stroke.
Lastly, all relevant conflicts of interests are disclosed properly in the paper as well as in the conflict of interest forms.
(1) Schierbeck LL, Rejnmark L, Tofteng CL, Stilgren L, Eiken P, Mosekilde L, et al. Effect of hormone replacement therapy on cardiovascular events in recently postmenopausal women: randomised trial. BMJ 2012;345:e6409.
(2) Shapiro S. Risks of estrogen plus progestin therapy: a sensitivity analysis of findings in the Women's Health Initiative randomized controlled trial. Climacteric 2003 Dec;6(4):302-10.
(3) Shapiro S. Recent epidemiological evidence relevant to the clinical management of the menopause. Climacteric 2007 Oct;10 Suppl 2:2-15.
(4) Gotzsche PC, Nielsen M. Screening for breast cancer with mammography. Cochrane Database Syst Rev 2011;(1):CD001877.
(5) Jorgensen KJ, Zahl PH, Gotzsche PC. Breast cancer mortality in organised mammography screening in Denmark: comparative study. BMJ 2010;340:c1241.
(6) LaCroix AZ, Chlebowski RT, Manson JE, Aragaki AK, Johnson KC, Martin L, et al. Health outcomes after stopping conjugated equine estrogens among postmenopausal women with prior hysterectomy: a randomized controlled trial. JAMA 2011 Apr 6;305(13):1305-14.
(7) Hodis HN, Mack WJ. A "window of opportunity:" The reduction of coronary heart disease and total mortality with menopausal therapies is age- and time-dependent. Brain Res 2011 Mar 16;1379:244-52.
(8) Mosekilde L, Hermann AP, Beck-Nielsen H, Charles P, Nielsen SP, Sorensen OH. The Danish Osteoporosis Prevention Study (DOPS): project design and inclusion of 2000 normal perimenopausal women. Maturitas 1999 Mar 15;31(3):207-19.
(9) Rossouw JE, Prentice RL, Manson JE, Wu L, Barad D, Barnabei VM, et al. Postmenopausal hormone therapy and risk of cardiovascular disease by age and years since menopause. JAMA 2007 Apr 4;297(13):1465-77.
(10) Salpeter SR, Walsh JM, Greyber E, Salpeter EE. Brief report: Coronary heart disease events associated with hormone therapy in younger and older women. A meta-analysis. J Gen Intern Med 2006 Apr;21(4):363-6.
(11) Salpeter SR, Walsh JM, Greyber E, Ormiston TM, Salpeter EE. Mortality associated with hormone replacement therapy in younger and older women: a meta-analysis. J Gen Intern Med 2004 Jul;19(7):791-804.
(12) Grodstein F, Stampfer M. The epidemiology of coronary heart disease and estrogen replacement in postmenopausal women. Prog Cardiovasc Dis 1995 Nov;38(3):199-210.
(13) Grodstein F, Stampfer MJ. Estrogen for women at varying risk of coronary disease. Maturitas 1998 Sep 20;30(1):19-26.
(14) Salpeter SR, Cheng J, Thabane L, Buckley NS, Salpeter EE. Bayesian meta-analysis of hormone therapy and mortality in younger postmenopausal women. Am J Med 2009 Nov;122(11):1016-22.
Competing interests: Authors of the discussed research article.