[ Editorials | Departments and Series | Index ]

Guest Editorial FEBruary 2007

Just When You Thought You¡d Heard It All¸

Vivian M. Dickerson, MD

Few subjects in the field of menopausal health have been discussed more than the Women’s Health Initiative (WHI) and the ensuing publications. Everyone— from academician to Web-log “blogger”—is staunchly defending a given interpretation of the data, the quality of the study, and the clinical implications. What does this mean for our patients? On the one hand, it would seem that they are being well served by an informed group of caregivers. In reality, though, the messages from physicians have been less than cohesive. In some cases, all estrogen products have been unceremoniously withdrawn; in other cases, patients were reassured that the study was flawed and did not apply to them—ie, “business as usual.” The media seems interested in reporting only the bad news, and in sensationalizing the findings by ignoring absolute risk and focusing on the “sexier” relative risk.

As more data were revealed, it emerged that estrogen alone appeared to reduce the risk of breast cancer, whereas the estrogen-progestin combination appeared to increase that risk. Finally, as more professional societies weighed in with their opinions and time passed, most OB/GYNs became comfortable with the analysis from the American College of Obstetricians and Gynecologists—ie, that for very symptomatic women, the lowest effective dose of estrogen for the shortest duration was acceptable.1 However, in California, where I practice, there was still no unanimity among specialties. Cardiologists were stopping hormones for all patients, only to have them represcribed by OB/GYNs—perhaps at a different dose, with a different regimen or a different delivery system. Speculation about symptomatic women who had a uterus often centered around the possible causative or permissive role of progestin, the optimum way to administer it, and whether there was an effective estrogen dose for vasomotor symptoms that did not require a progestin.

Although patients remained the victims of warring factions, the uproar began to subside. At my institution, there was recently a call for physicians of various specialties to come together on a consensus regarding symptomatic women, so that we would speak with a more consistent voice. I thought that this was a positive step, but then BAM!! Ravdin et al2 presented data at a general session of the San Antonio Breast Cancer Symposium. The data are provocative, and the media response was predictable. However, what no one expected was a reprise of the 2002 panic over the WHI findings. What caused such an uproar?

The data in question are epidemiologic in nature, and come from the US Surveillance, Epidemiology, and End Results (SEER) public-use database. There were several remark-able findings:

• Between 1998 and 2003, the US breast cancer incidence decreased by 1% per year
• In 2003, there was a 7% decrease in 1 year for both in situ and malignant breast cancers
• The decline in 2003 was most evident in patients older than age 50 years
• The decline in estrogen- receptor-positive (ER+) cancers was 8%, versus 4% for ER-cancers
• In women aged 50 to 69 years, the difference in decline was 12% for ER+ and 4% for ER-cancers.

What are we to make of this, other than it sounds like good news for once? Although I did not attend this meeting or hear the ensuing discussion, my telephone began to ring. It became apparent that not only the media, but many physicians as well, had jumped to the conclusion that these data proved the link between estrogen and breast cancer—indeed, that it proved causation. There was a general conclusion that the sharp decrease in the number of women using estrogen (which began in late 2002) may be responsible for the SEER data. But shouldn’t we be asking some other questions? Aren't we forgetting something?

• The decrease in breast cancer began in 1998, long before WHI. What is that about? Is it related to the accelerated decrease in 2003?
• The WHI showed that estrogen alone did not increase (and indeed appeared to decrease) the risk of breast cancer. How do we reconcile these findings with the SEER data interpretation that estrogen is not only the culprit, but is definitely dangerous?
• If it takes years for tumors to become detectable, how could 6 to 12 months of reduced estrogen use be responsible for such findings?
• Why was there a decrease in ER-tumors, albeit less robust?
• Why, when the vast majority of women over age 70 years were not using hormones, do we see a 7% decrease in breast cancer among women in this age group?
• A subgroup analysis in WHI demonstrated trends toward protection against breast cancer in women aged 50 to 59 years who used estrogen-progestin or estrogen therapy, exactly the group that demonstrated the highest reduction (11%) in breast cancer for 2003. How do we make sense of these seeming contradictions?
• Have we forgotten that epidemiologic data cannot and do not prove causation, but can only highlight possible association?

I am not on the payroll of any pharmaceutical house, nor have I entirely made a decision regarding hormone therapy in all of my patients. I am certainly much more cautious and less likely to recommend long-term use than I was before WHI. However, I hope that we will learn, not only from our polarizing behaviors in 2002, but also from epidemiologists who caution against jumping to conclusions. One prominent epidemiologist has remarked “tongue-in-cheek” that, “In 2002, banana imports and consumption dropped in the United States. Does that mean bananas cause breast cancer?” Most definitely these data are intriguing, and give us pause. Until proved otherwise, we must consider all interpretations—eg, true, true, and unrelated; or, if related, not necessarily causal, but a marker for some other phenomenon. As with all good epidemiologic data, these help us to formulate the null hypotheses that can only be evaluated by well designed, randomized, placebo-controlled trials. “Oh no,” I hear you cry. “Here we go again!”

I suggest that for now, we use a modicum of restraint and a good measure of science as we assess these new data. We must study, evaluate, and place all data in context. As always, it behooves us to use our very best clinical judgment, weighing all of the issues for each
of our patients. They deserve no less.

back to top

References

1. The American College of Obstetricians and Gynecologists. Effective Summary. Hormone therapy. Obstet Gynecol. 2004;104(4 suppl): S1-S131.
2. Ravdin PM, Cronin KA, Howlander N, Chlebowski RT, Berry DA, Anderson MD; National Cancer Institute; Harbor UCLA Medical Center. A sharp decrease in breast cancer incidence in the United States in 2003; Abstract [5]. Presented at: Annual San Antonio Breast Cancer Symposium; December 14, 2006; San Antonio, Tex.

back to top