Abstract
‘…we consider it likely that a substantial number of women, perhaps 3% as a conservative estimate, have undetected breast cancers when starting MHT…’
Several observational studies over the past two decades have suggested that menopausal hormonal therapy (MHT) increases the risk of breast cancer [1]. The randomized Women's Health Initiative (WHI) trial of estrogen plus progestin supported this conclusion by reporting an increased relative risk (RR) of 1.26 for breast cancer [2]. Little attention has focused on the mechanisms responsible and, specifically, on whether this finding represents an increased rate of diagnosis of occult cancers or the initiation of completely new ones. The answer to this question is critically important for interpretation of published data.
How common are small, undiagnosed, occult breast cancers in the population of women who initially start on menopausal hormone therapy? Data on this question come primarily from autopsy studies conducted 20–30 years ago. Eight studies examined breast tissue at autopsy in 952 women dying from other causes and attempted to determine how many occult breast cancers were present [3,4]. The individual results varied but pooled data suggested that 5% of women have undiagnosed ductal carcinoma in situ at autopsy and another 1% have invasive breast cancer. The stringency of techniques utilized probably explains the variability in results. For example, some studies made serial sections throughout the breasts for histologic examination; others identified grossly visible lesions that were then sectioned and examined; and finally, radiographs were performed on excised breasts at autopsy with sections made from the areas with increased density. Pooling of results from these eight studies demonstrates a ‘reservoir’ of undiagnosed breast cancer in the population of women ranging from 50 to 80 years of age.
Critical interpretation of these ‘old’ data from a 2008 perspective would suggest that this reservoir is probably smaller now because of the more sensitive mammographic techniques currently used and the frequency of their use. However, this may not necessarily be true since a recent study using MRI of the opposite breast in high-risk women with their first breast cancer, demonstrated a 3% prevalence of contralateral breast cancer that had not been diagnosed by prior mammography or breast examination [5]. Based on these various studies, we consider it likely that a substantial number of women, perhaps 3% as a conservative estimate, have undetected breast cancers when starting MHT at the present time.
Further support for the hypothesis regarding the ‘reservoir of undiagnosed breast cancer’ comes from studies of tumor biology in women. It has been estimated that the doubling time of small breast tumors ranges from 50 to 100 days in younger women and up to 6 months in those over 70 years [6]. Breast cancers must start from a single cell and then grow to a billion cells before they reach a clinically detectable size (~1cm).
‘…breast cancers develop gradually over a woman's lifetime as the number of mutations in the breast accumulates.’
Based on the 50–100-day doubling time estimates, growth from initiation of the cancer to its clinical detection would take approximately 5–10 years. If ‘gompertzian’ kinetics would hold true where tumor doubling time continuously diminishes with increasing tumor size, a longer latent period might be expected [6]. A study of large existing tumors found a mean doubling time of 325 days, providing support for the presence of gompertzian kinetics [7].
Our current understanding of tumor genomics suggests that diagnosed breast cancers contain approximately 140 ‘passenger mutations’ and 10–15 ‘driver mutations’ [8,9]. The driver mutations are estimated to up- or down-regulate as many as 20 signaling pathways in a woman with breast cancer. A reasonable inference is that many years would be necessary for so many mutations to develop.
Epidemiologic data also suggest that development of breast cancer is a long-term process. The removal of the ovaries before the age of 35 years gradually reduces the lifetime risk of breast cancer and after 35 years, the decrease is on the order of 75% [10,11]. Alterations in birth weight can also affect breast cancer risk later in life [12]. These observations suggest that breast cancers develop gradually over a woman's lifetime as the number of mutations in the breast accumulates. Once progressing to a neoplastic state, tumors exhibit a prolonged doubling time (i.e., 50–100 days) when small and an even longer doubling time when larger [6,7]. These aspects of breast cancer biology would support the possibility that small tumors sojourn in the breast for a long period of time before diagnosis, resulting in a reservoir of undiagnosed breast cancer in the population [13].
What then is the magnitude of the reservoir? Based on a conservative estimate of 3% prevalence, there would be ten undiagnosed breast cancers per 1000 women in a 50–90-year age range. It is likely that the same percentage of women would harbor occult tumors when starting on MHT. Similarly, this should hold true for women entering into the WHI trials.
‘…small tumors sojourn in the breast for a long period of time before diagnosis, resulting in a reservoir of undiagnosed breast cancer in the population.’
How does this number of occult, pre-existing tumors compare with the absolute number of breast cancers newly diagnosed over time in the WHI estrogen plus progestin trial? Of the 8102 women on the placebo arm of this trial, 124 developed breast cancer over the 7 years of observation [2]. We estimate that the number of ‘occult undiagnosed breast cancers’ at the start of the study would have been 243 (i.e., 3% of 8102 women). This would suggest that approximately 50% of occult tumors (i.e., 124 of 243) had grown to a sufficient size to be diagnosed in the placebo group over the 7-year period. It should be noted that this would be consistent with a 50–100-day tumor-doubling time.
If one accepts the ‘reservoir hypothesis’, what implications would this have for women starting MHT? Based on the data of Hofseth et al. and others, estrogen plus a progestin would cause existing hormone-dependent tumors to grow more rapidly and to be diagnosed sooner than in the absence of hormonal therapy [14–16]. The RR of breast cancer associated with estrogen/progestin use in the WHI study was 1.26 [2]. The total number of women with cancer then increased to 166, with 42 excess tumors in the hormone therapy group. If the reservoir hypothesis were true, the WHI results could be explained exclusively by the more rapid growth of existing tumors and earlier diagnosis. Accordingly, this would represent a promotional effect and not tumor initiation. This conclusion is consistent with our knowledge of estrogen receptors and hormone-dependent breast cancer. Approximately 60% of the reservoir tumors would be ER-positive and half of these would be stimulated by hormone therapy. Accordingly, 65 of the 124 tumors would grow more rapidly in response to hormone therapy and a certain fraction of tumors (42 out of 124) would grow sufficiently large for clinical detection.
What would happen when women with these reservoir tumors would stop estrogen/progestin therapy after a period of previous use? In the Collaborative Study on Hormones and Breast Cancer, the increased risk fell by 4 years after stopping MHT [17] and in the Million Women's Study [18], after 1 year. It is known that hormone responsive tumors frequently stop growing after withdrawal of estrogen and some even undergo apoptosis.
After publication of the initial WHI trial, nearly 80% of women stopped MHT. Based on the reservoir hypothesis, one would expect that the incidence of newly diagnosed breast cancers in the population would decrease following widespread cessation of hormone therapy. This is exactly what Peter Ravdin reported in the New England Journal of Medicine, namely a 6.7% reduction in breast cancer diagnosis in the USA in the 2 years following the drop in MHT use [19]. A California study provides even more convincing data on this relationship. For example, in the counties in California with the highest use of MHT, breast cancer incidence dropped by 22.6 versus 13.9% in the counties of intermediate use and 8.8% in those counties with the lowest use of MHT [20]. A further observation was that only ER-positive breast cancer incidence rates decreased, a finding expected if cessation of estrogen/progestin therapy affected a reservoir of ER-positive tumors [19]. If this conclusion were correct, one would predict there would be a rebound incidence in breast cancer later, as happened with prostate cancer several years after prostate-specific antigen testing was introduced.
A puzzling finding regarding MHT with estrogen alone in the WHI trial was the nonsignificant trend toward a 20% decrease in breast cancer incidence. A later, post hoc, subset analysis of the WHI estrogen-alone study demonstrated that this reduction in breast cancer risk was statistically significant in three subgroups:
Those continuing to stay on the assigned medication (RR: 0.67; 95% CI: 0.47–0.97)
Those with localized cancer (RR: 0.60; 95% CI: 0.51–0.95)
Those with ductal tumors (RR: 0.71; 95% CI: 0.52–0.99) [21]
A similar reduction in breast cancer risk with estrogen alone has also been seen in the Nurses Health Study with a 10% reduction in all women taking estrogen alone for 5–9 years (RR:0.90; 95% CI: 0.73–1.12) and a 26% reduction in obese women (RR: 0.74; 95% CI: 0.55–1.00) [22]. Lesser reductions were observed in the Lyytinen study (RR: 0.93; 95% CI: 0.80–1.04) [23], the Mission study (0.28% incidence with no hormone therapy vs 0.70% with an estrogen/progestin combination) [24], the Schairer study (5% decrease at 5 years) [25], and the Million Women Study (<1 year of estrogen alone RR: 0.81; 95% CI: 0.55–1.20) [18].
What could explain this paradoxical reduction of breast cancer incidence? Song et al. and Jordan et al. have shown that breast cancer cells that have been deprived of estrogen for a long period respond to estradiol administration with apoptosis [26–29]. This occurs because of upregulation of the Fas death receptor pathway, a reduction of the antiapoptotic protein, Bcl-2 and of NF
Our understanding of the physiology of estrogen-induced apoptosis is that long-term estrogen deprivation is required for this effect. Postmenopausal women have generally been deprived of estrogen for some time as a consequence of cessation of ovarian function. Of interest is the fact that if the women in the estrogen-alone arm of the WHI had previously taken exogenous estrogen, they did not have a reduction of breast cancer risk (e.g., prior estradiol, RR: 1.04; 95% CI: 0.7–1.50, no prior estradiol, RR: 0.65; 95% CI: 0.46–0.92) [21].
It is important to recall that estrogens have been used to treat menopausal women with breast cancer and this therapy can be highly effective [30,31]. An unexplained observation was that only those more than 5 years postmenopausal responded to this agent. Premenopausal women and those with menopause for less than 5 years did not respond [30]. These data would support the concept that long-term estradiol deprivation is needed for this hormone to induce apoptosis. Studies are now ongoing to see if estrogen does in fact induce apoptosis in women with postmenopausal breast cancer treated with aromatase inhibitors as a means of inducing long term estradiol deprivation [Matthew Ellis, Washington University, St Louis and Craig Jordan, Fox Chase Cancer Center, PA, USA, Pers. Comm.].
Against the reservoir hypothesis, one can argue that exogenous hormone therapy may actually be initiating new breast cancers. If atypical ductal hyperplasia is present in the breast at the time of starting MHT, this lesion may progress to cancer. The studies by Dupont in women with atypical ductal hyperplasia did not appear to confirm this [32]. Nonetheless, MHT could logically cause these atypical lesions to progress to cancer.
If the reservoir hypothesis were correct, what implications would this have for use of MHT in postmenopausal women in the future? Specifically, it would emphasize the need to develop methods of earlier detection of breast cancer before starting MHT. This could involve use of MRI, random fine needle aspiration around the nipple, nipple aspiration or lavage with proteomic markers or other methods. In addition, one would need to look carefully to determine whether there might be a later rebound in new breast cancer incidence rate in women stopping MHT over the next few years. Women treated with estrogen alone for menopausal symptoms might be alerted to a possible beneficial effect on breast [21–25]. Finally, the concern regarding MHT causing breast cancer would be somewhat tempered and would raise the issue of whether earlier diagnosis through estrogen/progestin stimulation might, in fact, result in longer overall survival. It has been suggested in the past that women with breast cancers diagnosed when they were receiving MHT lived longer than those not taking MHT [33].
Footnotes
The author has no relevant affiliations or financial involvementwith any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
No writing assistance was utilized in the production of this manuscript.