Customize your JAMA Network experience by selecting one or more topics from the list below.
Cobleigh MA, Norlock FE, Oleske DM, Starr A. Hormone Replacement Therapy
and High S Phase in Breast Cancer. JAMA. 1999;281(16):1528–1530. doi:10.1001/jama.281.16.1528
Context Prolonged postmenopausal hormone replacement therapy
(HRT) is associated with increased incidence of breast cancer and,
paradoxically, reduced breast cancer mortality. The biological
rationale for this discrepancy has not been explored.
Objective To compare the prognostic characteristics of cancers
arising in women who have used HRT with those in women who never have
Design Prospective cohort study from December 1989 to November
Setting Teaching hospital in a large midwestern metropolitan area.
Patients Cohort of 331 postmenopausal women who presented
consecutively with 349 invasive breast cancers.
Main Outcome Measures Estrogen receptor (ER) status (ER positive
vs ER negative) and S phase (low vs high) for current HRT users vs
Results The frequency of high S-phase fraction among cancers in
women who were using HRT was markedly increased compared with that in
women who had never used HRT (adjusted odds ratio [OR], 2.82; 95%
confidence interval [CI], 1.04-7.66). However, the greater frequency
of high S-phase fraction was limited to women with ER-positive cancers
(for HRT users vs never users, OR, 5.25; 95% CI, 1.36-20.28; for
ER-negative cancers in HRT users vs never users, OR, 1.08; 95% CI,
Conclusions Use of HRT appears to stimulate growth of ER-positive
but not ER-negative breast cancer as measured by S-phase fraction.
The prognostic significance of high S-phase fraction in current HRT
users who have ER-positive tumors is unknown.
Prolonged use of
hormone replacement therapy (HRT) correlates with increasing incidence
of breast cancer,1 which is counterbalanced by lower
mortality from the disease.2 We studied the influence of
HRT use on prognostic factors in primary invasive breast cancer in an
attempt to formulate a biological rationale for this paradox.
Beginning in 1989, postmenopausal patients with invasive breast
cancer who presented consecutively to a medical oncologist (M.A.C.) at
a teaching hospital in a large midwestern metropolitan area were
entered prospectively into a database. In 1996, all postmenopausal
subjects presenting consecutively to medical oncologists were
entered prospectively. Variables included birth date, diagnosis
date, pathologic tumor size (unknown when patients received
preoperative chemotherapy), number of involved nodes (unknown when
axillary dissection was not performed), last menstrual period,
gynecologic surgery, HRT use, estrogen receptor (ER) and progesterone
receptor (PR) values, ploidy, and percentage of cells in S phase
(measured by flow cytometry).
Receptor and S-phase cut points were taken from the laboratories
in which they were analyzed. About half the patients underwent primary
surgery at our hospital; flow cytometry was performed in our Clinical
Laboratory Improvement Act–approved laboratory. Other patients
referred from community hospitals usually had flow cytometry performed
in reference laboratories.
Hormone replacement therapy was defined as estrogen with or
without progestin. Menopause was defined as no menstruation
for 12 months, simple hysterectomy and patient older than 55 years, or
Categories of HRT included never users (no HRT exposure), current
users (HRT use within 1 month of diagnosis, with most using HRT at
diagnosis), prior users (HRT stopped ≥1 month before diagnosis, with
most having a remote history of use), and tamoxifen users (developed
contralateral breast cancer during adjuvant tamoxifen therapy).
Bivariate comparisons were assessed using a Yates corrected
χ2 test. Crude odds ratios were computed to represent the
likelihood of high S phase relative to the hypothesized high risk of a
prognostic factor (HRT use, tumor size >2 cm, ER-negative status,
PR-negative status, aneuploidy, and nodal involvement). Logistic
regression analysis was used to determine the multivariate likelihood
of high S phase among HRT users, adjusting for age at diagnosis, tumor
size, ER and PR status, ploidy, and nodal status.
Model fit was evaluated.3 Tetraploid tumors were excluded
because of small numbers. Analyses were performed using SPSS/PC+
software, Version 5.0 (SPSS Inc, Chicago, Ill). Probability values are
2-tailed, with P<.05.
There were 349 breast cancers among 331 women (bilateral cancers
included 4 synchronous and 14 metachronous). There were 142 current HRT
(40.7%), 165 never (47.3%), 38 prior (10.9%), and 4 tamoxifen users
(1.1%). Prior and tamoxifen users were excluded because of their small
Current users were significantly younger (mean [SD] age, 57
 years) than never users (mean [SD], 62  years;
P<.001) and more likely to have high S-phase cancers and
PR-positive status (Table 1). Controlling for age did not change the significance of these
associations. Other prognostic factors did not vary with HRT use.
S phase was evaluated by ER status and HRT use. Estrogen
receptor–positive cancers were affected by HRT; nearly half of current
users had high S-phase cancers compared with only about a fifth of
never users. Use of HRT did not affect the S phase of ER-negative
cancers; the majority had a high S phase (Table
Estrogen receptor–negative status, aneuploidy, and current use of HRT
correlated significantly with high S phase in the regression analysis (Table 3). When sorted by receptor
status, the model confirmed a significant association of current HRT
use with high S phase in ER-positive cancers only. Current users with
ER-positive cancers were 5 times more likely to have a high S phase
than never users (Table 4).
In this study, ER-positive cancers were more likely to have a high S
phase in current users of HRT. To our knowledge, this is the first
report of such an association. This result was expected based on
preclinical models. Estrogen causes proliferation of ER-positive but
not ER-negative human breast cancer cells in vitro4 and in
High S phase was independently and significantly predictive
of breast cancer recurrence in 8 trials of 9901 women, while 4 trials
including 1044 women found no such evidence.6- 17 History of
HRT use was not included in these models.
In a large database of 127,000 breast cancers, high S phase
correlated with aneuploidy and ER-negative and PR-negative status
(these results were not stratified by HRT use).18 Our
results support and extend those observations and suggest that high S
phase is also more common in ER-positive cancers among current HRT
Use of HRT has been correlated with low S phase19; however,
the results of that study were not stratified by receptor status. Use
of HRT was associated with low S phase in ER-positive cancers, but the
analysis did not control for other variables that affect S phase, such
as aneuploidy and nodal status.20
Others have evaluated the relationship between HRT and ER status of
primary cancers. Although occasional correlations have been
described,21,22 most studies found no significant
difference in ER profiles between users and nonusers of
HRT.19,23- 30 Our study supports this consensus.
Relationships between HRT use and other prognostic characteristics have
been reported. Although trends toward smaller tumors, negative nodal
findings, and diploid/tetraploid
cancers were observed, these were not statistically
significant in multivariate analysis.31 Others have
reported a high rate of well-differentiated tumors in HRT users, but
this observation was not subjected to multivariate
A strength of this study is the collection of information on HRT use at
the time of presentation, lessening the possibility of recall bias.
Limitations of our study include the fact that it is not population
based and that not all variables for every patient were known.
Most studies have reported reduced survival in
node-negative women with high S-phase cancers. Some use such
retrospective correlative data to recommend systemic adjuvant therapy,
even for women with small (<1 cm) node-negative cancers. However, it
is possible that high S phase may be iatrogenic in ER-positive cancers
diagnosed in women taking HRT.
Use of HRT may promote preexisting clinically occult cancers, bringing
them to light sooner in their natural biological history. This may
account for the better survival of breast cancer patients who have used
HRT. This cohort will be followed up and expanded to assess survival.
Hormone withdrawal may also prove therapeutic. Precedents for this have
been reported in studies of primary tumor regression32 and
metastatic breast cancer regression after HRT withdrawal.33