WHI follow-up study confirms combined HT risks outweigh benefits

Issue: April 2008

Women treated with combined hormone therapy in the Women’s Health Initiative trial experienced a greater risk of fatal and nonfatal malignancies after intervention. Their risk for cardiovascular disease decreased from intervention to post-intervention.

Researchers from various institutions across the nation conducted a study of the women enrolled in the WHI study to report health results three years after the HT intervention of estrogen and progestin ceased. The post-intervention phase, which began in July 2002, included 15,730 women, according to the study.

The women were monitored semi-annually and primary endpoints included invasive breast cancer and coronary heart disease. A global index was used to summarize the balance of risks and benefits.

Researchers found a decrease in cardiovascular disease from the time of intervention to post-intervention, but the HR for overall risk for malignancies increased from 1.03 (95% CI, 0.92-1.15) to 1.24 (95% CI, 1.04-1.48) post-intervention. In women randomly assigned to estrogen plus progestin, the global risk was 12% higher, compared with placebo.

Researchers also found more breast cancers diagnosed in the estrogen plus progestin group, compared with placebo, although the hazard ratio was slightly lower during the post-intervention follow-up. – by Stacey L. Adams

JAMA. 2008;299:1036-1045.

It is always of interest to learn more about the WHI. With a projected 164,500 women enrolled at 40 clinical centers in the United States, the WHI is perhaps the most massive and challenging program the National Institutes of Health has undertaken. It had a budget of $628 million from 1991-2007, and it’s estimated to have cost even more than that.

This study provides further follow-up data about patients randomized to 0.625 mg conjugated equine estrogens plus 2.5 mgm medroxyprogesterone acetate or placebo after the trial was stopped 5.6 years after its start due to an increase in more advanced breast cancers in those taking HT. Follow-up after the trial stopped was accrued under strict observance of the trial protocol for semi-annual end point ascertainment and verification, as well as annual mammography surveillance for those persons adherent to either arm.

The authors have essentially provided two case series—those continuing with the hormones and those with placebo. They have an advantage of being able to survey patients intensively, both during the randomized, controlled trial and after. The trial was designed to test the hypothesis that estrogen plus progestin in this dose prevented cardiovascular disease and reduced osteoporotic fractures. It did not prevent cardiovascular disease, but it did prevent fractures.

Examination of the Kaplan-Meier curves reveals that less than one half of those who signed up provided data after eight years. During the post-trial comparisons, none of the endpoints of interest were different “after intervention,” except breast cancer rates that were more in the active case series (HR 1.2 , 1.06-1.51). While it is a fun exercise to apply survival techniques for inference about randomized, controlled trial data for which it was planned, it is, in my opinion, very hazardous to combine clinical trial data with observational data (for which it was not planned). People that enter into clinical trials are by definition different than those who do not, or who refuse to participate. People who are compliant are different than those who are not, or switch to another strategy. It is unclear how many participants were continuing with other co-interventions for compliers or non-compliers. While the authors refer the reader to a publication to understand adjudication, in the reference cited I could not find that they discuss it. In a study of this expense we should demand that all hard endpoints were centrally adjudicated. Perhaps this will be revealed in time. I would have learned more if I would have had knowledge of outcomes and risk factors as they relate to the outcomes studied for those who were no longer being followed intensively.

The article does not change much, from my perspective. It might be useful as a hypothesis-generating effort to help us understand if hormones given orally in this dose are related at all to other cancers. This later finding was unexpected.

In general, the WHI is a randomized, controlled trial, and any randomized, controlled trial leaves us with potential problems of how to generalize. It is most applicable to those in the study. It is too bad that the WHI could not study alternative choices when hormones are needed for women with a uterus present. The arguments to use only one dose for the entire clinical trial was that this was necessary for purposes of obtaining adequate power and because this was the most widely used preparation in the United States at the time. I think the original trial has made it clear that oral estrogen plus progestin at this dose did not prevent heart disease for the women observed.

– Robert A. Wild, MD, PhD, MPH

Professor of Reproductive Endocrinology
Oklahoma University Health Sciences Center, Oklahoma City