ATLANTAA study of more than 400,000 postmenopausal women has found
no increased risk of fatal breast cancer with use of estrogen replacement
therapy (ERT). In fact, women who reported ever having used estrogen actually
had a 16% decreased risk of dying of breast cancer, Dawn Willis, PhD, MPH,
reported for the American Cancer Society (ACS) at a general session of
the San Antonio Breast Cancer Symposium.
The Society's Cancer Prevention Study II (CPSII) is a large prospective
cohort study that began in 1982 to study risk factors for various cancers.
"We have, at present, analyzed ERT in relationship to breast, ovarian,
and colon cancer," Dr. Willis said.
The cohort completed questionnaires in 1982; deaths were determined
from the National Death Index, and cause of death from death certificates.
During nine years of follow-up, there were 884 deaths among never-users
of estrogen and 585 among ever-users. For those using estrogen for one
year or less, the relative risk of breast cancer death was .85; for two
to 10 years of use, .78; and for 11 years or more, .93.
One surprising finding was a trend toward a greater decreased risk among
those who started estrogen at a younger age, Dr. Willis said. The relative
risk was .66 for those who began estrogen when less than 40 years of age;
.84 for those who started estrogen at age 40 to 49; and .89 for those who
started hormones at age 50 and above.
One possible explanation is that women who start to use estrogen early
have probably done so because their ovaries were removed. However, the
trend still held after adjusting for age at menopause and type of menopause
(natural or surgical).
The researchers also looked for interactions with potential confounders,
"to find out if estrogen acts differently in, say, women who have
a family history or who have had breast cysts," Dr. Willis said. The
data showed no interaction with family history and a possible interaction
with fibrocystic disease.
An unexpected finding, however, was a strong interaction with age of
menarche. "Women who reported menarche at 14 or older had the highest
degree of association of estrogen use with decreased risk of fatal breast
cancer," she said.
Unfortunately, Dr. Willis said, "epidemiology never provides definitive
answers; it only gives tantalizing clues." Thus, there are a number
of possible explanations for the findings.
She pointed out that there could be a selection bias. "Women who
take estrogen are well known to be generally healthier than women who don't;
they're thinner; exercise more, and are less likely to smoke," Dr.
Willis said. (Interestingly, she added, they are more likely to drink.)
A further selection bias may occur because physicians are perhaps not
inclined to put women at high risk for breast cancer on ERT. However, she
said, "in our population this was not true; women with family histories
and those with self-reported cysts, which might be considered high-risk
groups, were just as likely as others to have used estrogen."
Surveillance bias also might be present, since women who take hormones
are under a physician's care and generally receive yearly mammograms and
clinical breast examinations. Thus, any breast tumors that these women
had would be detected at an early stage with a better prognosis and less
risk of death.
A possible molecular biological explanation for the result is that estrogen
has been reported to increase the expression of the tumor suppressor gene
BRCA1, she added.
Overall, the researchers concluded from this study and other data that
six to 10 years of estrogen use will reduce a woman's risk of fatal breast
cancer by a small amount, increase her risk of ovarian cancer death by
a very small amount, and decrease her risk of colon polyps and colon cancer
"All of these cancer increases and decreases pale in comparison
to the decreased risk of death from cardiovascular disease with use of
ERT," she said, "and we cannot even factor in the improvements
in quality of life."
Dr. Willis called for more research into the pathology of tumors that
develop in women who are taking hormones, and for more basic science "to
explain this phenomenon so we'll know whether it really is a biological
difference in the tumors or a sociological difference in users."