In December 2006 I discussed a paper presented at a national breast cancer meeting in San Diego which reported that there had been a dramatic decrease in the incidence of breast cancer between 2002 and 2003.
At the time, the authors indicated they thought the decrease was due to the fact that women had stopped taking their hormone replacement therapy medications in 2002 in response to a warning from the Women’s Health Initiative (WHI). This large, nationwide study reported that combined hormonal replacement therapy (HRT) with estrogen and progesterone in post-menopausal women increased a woman’s risk of developing breast cancer.
My response was that I wasn’t so certain that stopping HRT was the complete explanation for the observed decrease in breast cancer cases.
Now, with the publication of a more detailed paper in the current issue of the New England Journal of Medicine, I must admit that I am still confused as to all the factors that contributed to the decline.
Like any puzzle, this one may yet turn out to have many complexities that have to be solved before we get to the conclusion.
In doing an analysis of breast cancer cases reported in SEER registries—which monitor cancer incidence, stage and mortality in nine regions of the United States, and which are thought to reflect the incidence of cancer throughout the country—the authors of the New England Journal article found that in 2003 the incidence of breast cancer in the United States fell by 6.7% compared to 2002.
The decrease in breast cancer diagnoses began in mid-2002, coincident with the publication of the major report from the Women’s Health Initiative on July 17, 2002. The article, published in the Journal of the American Medical Association, concluded combination hormone replacement therapy increased a woman’s risk of developing breast cancer (see previous blog posting for more details of this report).
To bolster the case that this decrease in breast cancer cases—which appears to have leveled off in 2004 according to the SEER analysis—the authors were able to determine that the decline was present only in women age 50 and older (and therefore menopausal), and that the decline was limited to cancers which were estrogen-receptor positive (suggesting they hormones could accelerate their growth).
The total decrease in incidence, when comparing 2001 to 2004, was 8.6% which is certainly very dramatic.
Some additional important information was that the decrease was equivalent in women ages 50-69 and women age 70 and older.
The authors note—as also previously reported in this blog—that from 2000 to 2003 there had been a 3.2% decrease in mammography rates in this country. They further comment that as a result of women stopping their hormone therapy, they may see their doctors less frequently and that this too could have resulted in decreased adherence to a yearly screening mammogram schedule.
A final consideration that they considered to explain the decline was the fact that, like prostate cancer, there may come a time when fewer breast cancers will be found since many had already been discovered through mammography and early detection.
More simply, with effective screening, there should eventually be a decrease to a lower steady state incidence rate. However, the researchers indicated they thought that this was not a likely explanation for the current observation.
All of these comments are well taken, appropriate and backed up by evidence.
Nonetheless, the authors caution, it is still possible that this is a temporary observation and that at some time in the future, the incidence of breast cancer may once again increase.
Looking at the data, it is hard to ignore the relationship between the date the initial hormone study and editorial about the dangers of HRT appeared in the Journal of the American Medical Association (July 17, 2002) and the beginning of the decline in the incidence of breast cancer.
As reflected in the data in the current report, it is almost uncanny how the incidence of breast cancer began to decline in the third quarter of 2002, which would encompass July, August and September.
Although I was a skeptic that so many women stopped taking their hormones so quickly, maybe I was wrong in my assumption.
After all, as pointed out in the paper, 61 million prescriptions were written for hormone replacement therapy—both combination pills with estrogen and progesterone, and pills that contained only estrogen (for women who have had a hysterectomy)—in 2001, 47 million in 2002, and 21 million in 2004.
That implies that by 2004, only about 1in 3 women continued taking HRT compared to 2001 (this simplistic analysis doesn’t take into account the number of women who started taking HRT during that time).
My “resident expert” on gynecologic matters (my wife, who is a gynecologist and deals with these issues every day) indicated that these numbers reflected her own practice’s experience in terms of how long it took for the decrease in hormone use to run its course. (That’s not very scientific, but an anecdote from time to time is acceptable in a blog like this.)
However, the key point here is that when the HRT paper came out, it only dealt with women who were taking a combination replacement therapy, not estrogen alone. The recommendation to stop taking HRT was limited in that July paper and editorial to those women on the combination therapy. A similar recommendation was not made for women on estrogen alone, since that study was ongoing at the time.
The current paper doesn’t tell us how many of the women who stopped their HRT therapy were taking the combination pill (the one that increased breast cancer risk) compared to those who were taking the estrogen-only medication.
In fact, the estrogen-only arm of the study was continued and eventually reported that the women with a hysterectomy who took estrogen alone and were at average risk for developing breast cancer had a lower incidence of breast cancer when compared to women who took a placebo (or dummy pill).
In addition, the actual increased risk of developing breast cancer for women taking combination HRT compared to those taking a placebo was modest, about 8 additional cases of breast cancer each year for every 10,000 women taking the medication.
That amounts to about 0.8% fewer cases of breast cancer in a group of 10,00 women who theoretically were all taking combination hormonal therapy under the conditions of the WHI study (where, by the way, not every woman who was supposed to take the medication continued it for the duration of the study).
In 2001, the last year before HRT became suspect, the incidence of breast cancer in women age 50 and older in the 9 SEER registries was 38.3 cases/10,000 women. This rate fell to 33.8 in 2003, which is about 12% lower over that 2 year period of time.
(Another note: these data—as reported on the SEER website—are incomplete and do not accurately reflect the latest information, which apparently was available to the authors but has not yet been posted on the web for public view. It takes several years for the complete data to “trickle in,” allowing for delays in reporting.)
When you consider that not every woman took hormone pills, and that not every woman stopped her hormone pills, and finally that not every woman was taking the combination pill, you end up with an implied impact of stopping combination hormonal therapy on the risk of getting breast cancer that is nothing short of amazing.
Looked at another way, one would have expected even greater declines if every woman had stopped her medication.
And, yet, that’s not what the data from the WHI study suggested would happen, as noted above. That information suggested that, in real life with all of the considerations just noted, there would have been a less than 0.8% decline in breast cancer incidence if, in fact, combination HRT caused the breast cancers that "disappeared."
(One factor not considered in this discussion is that the increase in breast cancer incidence in the WHI study started only several years after women started taking HRT. It is plausible that if the women had continued the medication for many more years that their risk of getting breast cancer may have increased much more than that observed in the study.)
I will be the first to admit that I am not an epidemiologist, nor is this a scientifically rigorous discussion.
But this is not the first time that there has been a dramatic decrease in breast cancer incidence in the SEER registries.
In 1999, the incidence rate of breast cancer per 100,000 women age 50 and older in the United States was 395. In 2000, it was 378, a decrease of 17 cases per 100,000 women, or about 4.4%.
Much of this decrease appears to be due to a decrease in the incidence of breast cancer in black women age 50 and older, where the rate fell from 336 to 304.6, a decrease of 9.4%. In 2002, however, the rate of breast cancer in black women once again went up to 330.
From 2002 to 2003, a significant decrease in breast cancer incidence occurred in white women age 50 and older, where the rates went from 392.3 to 352.4, or a decline of 11.2%. (But, once again, I need to remind you that this may not be the latest, most complete set of data.)
When you look at the data over several years in women age 50 and older, what you begin to see is that there are fluctuations from year to year.
As the authors point out in their article, and as I mentioned previously here in the December blog, changing hormonal environments by the addition or withdrawal of hormones such as estrogen can influence the course of breast cancer.
For example, a long forgotten therapy for the treatment of pre-menopausal women with recurrent breast cancer was oophorectomy, or the removal of the ovaries. The goal was to substantially reduce the amount of estrogen in the body.
Before tamoxifen, when I was in the early years of my training and practice, this procedure was routinely used with occasionally excellent results in treating these young women.
If my memory serves me right, I recall some of the surgeons I worked with advising pre-menopausal women who were diagnosed with primary breast cancer to consider an oophorectomy as part of their initial treatment. This was prior to the tamoxifen era, which began in the mid-late 1970s.
We also know that in women who are at particularly high risk for breast cancer (carriers of the genetic abnormalities called BRCA 1 and 2), oophorectomy after completion of childbearing can decrease their risk of breast cancer.
In the midst of what must be to you, the readers, a bunch of confusing information, rests the real question of what is going on here.
Based on past experience, it is not unexpected that withdrawing a supplemental hormone pill may influence the rate of breast cancer.
It is hard to ignore the timing of the 2002 JAMA article and the resulting decrease in the number of breast cancers diagnosed in 2002 through 2004.
The problem in this “immediate decrease” scenario is that we know breast cancer takes many years to become detectable on a mammogram, and even more years to become palpable during a clinical examination.
It is unlikely that a cancer present in the breast suddenly disappears immediately or within months after hormones are stopped. That just doesn’t make sense. But that is one implication of this report.
A more likely scenario is that stopping hormones influenced the diagnoses of breast cancer in this situation, but are not responsible for causing most of the breast cancers that would have been expected between 2002 and 2004 and were not diagnosed.
We do know that hormones such as estrogen can influence the growth of breast cancer tumors. Perhaps there is some decrease in the size of a breast cancer after combination hormone therapy was stopped. But the likelihood of it going away completely, or simply stop growing forever, is biologically remote.
What else could explain this decrease in breast cancer incidence?
A factor which may play a role is the documented decrease in mammography rates. If you don’t look for a breast cancer, you won’t find it.
Maybe women are less likely to visit their doctors since they no longer have to get their prescriptions. If they don’t visit their doctors, they won’t hear the recommendation to get their annual mammogram.
Perhaps, as one reader pointed out in a comment on this blog, now that women are no longer taking hormones they mistakenly believe their risk of breast cancer has gone way, way down.
Or perhaps this is a statistical fluke, which we have seen before in these same registries—although not necessarily to the same degree—in years past, when HRT was not the major highlighted cause.
Maybe it is all of the above.
My sense is that in fact the decrease in the use of combination hormone therapy did contribute to this decline in breast cancer diagnoses, but does not account for the entire picture. The decline is simply too much too fast for me to believe that stopping combination HRT is the sole cause for this observation.
I also noted that this decline occurred equally in “younger” post-menopausal women and “older” post-menopausal women.
My sense is that the use of hormone replacement therapy was less frequent in women over 70 than in women in their 50s and 60s. That said, given the fact that rate of breast cancer is higher in women in their 70s when compared to younger women (the major risk factor for breast cancer, after all, is getting older), it may be that even a slight decrease in the number of 70+ year old women taking HRT may show a large impact in reducing the rate of breast cancer detection in that age group.
Even if stopping HRT explains almost all of this decline, it still could well be a temporary situation. If that is the case, once the cancers regain their footing we will see them grow once again. And that means we will see an increase in the numbers of diagnosed breast cancers in the years ahead.
If the decline in mammography has any role in this observation (and I believe it does) and if women aren’t going to their doctors for their regular check-ups (which may be the case) then we are headed for a serious problem in the not-too-distant future.
A recent report gives rise to further concern.
Once the WHI study was released, there was a storm of criticism from the gynecology community regarding the data analysis and the fact that the participants in the study were generally older post-menopausal women, and not those who are starting to have menopausal symptoms.
When the data was recently reanalyzed for younger post-menopausal women, the investigators found that when HRT was started closer to the onset of menopause, the risk of cardiovascular disease was reduced. Risk of stroke, however, was still elevated. (Breast cancer risk was not part of the new analysis.)
What that means is that we may find more women willing to start HRT to treat menopausal symptoms soon after they enter menopause than has been the case for the past several years. That, in turn, might further blur our understanding of what is happening as we analyze breast cancer data in the years ahead.
(The basic advice remains that women should use these medications for symptomatic treatment only, at the lowest dose for the shortest time possible.)
As I said at the beginning, the good news here is that the incidence of breast cancer is declining.
But the answer to the puzzle eludes us. We may have some of it solved, but there are those among us who believe the rest of story remains to be written.