SEARCH

SEARCH BY CITATION

Keywords:

  • oestrogen;
  • menopause;
  • cognition;
  • critical period

Abstract

  1. Top of page
  2. Abstract
  3. Evidence from basic neuroscience
  4. Animal behavioural studies
  5. Human studies
  6. Summary and conclusions
  7. Acknowledgements
  8. References

Although there is compelling evidence from small randomised controlled trials and cross-sectional studies indicating that oestrogen helps to protect against cognitive ageing in women, the findings of the large, Women's Health Initiative Memory Study failed to support the earlier findings. The attempt to resolve these discrepancies led to the formulation of the Critical Period Hypothesis which holds that oestrogen has maximal protective benefits on cognition in women when it is initiated closely in time to a natural or surgical menopause but not when treatment is begun decades after the menopause. This article reviews the evidence from basic neuroendocrinology, from animal behavioural studies and from human studies that supports the critical period hypothesis. In view of the promise of this hypothesis and its considerable clinical implications, a direct test of its validity is warranted.

The increase in female life expectancy during the past century is, in large part, due to the development of vaccines against infections diseases, to the reduction in maternal and infant mortality, to the advent of antibiotics, and to advances in public health measures and an increased standard of living. This increase in female longevity means that women in industrialised nations now live one-third of their lives after cessation of their reproductive capacity. Although for the most part, of course, this is a positive change, it has also led to an increase in the number of women who develop degenerative diseases in older age that restrict their normal functioning and severely diminish their quality of life. Impairment of cognitive function in the elderly can lead to a loss of independence while Alzheimer's disease (AD) inevitably leads to total disability. The challenge to medical scientists therefore, is to discover the aetiologies, effective treatments, and preventative strategies for these diseases to alleviate personal suffering and the growing economic burden of these disabilities for society.

A considerable amount of evidence exists to demonstrate that specific changes in cognitive functioning occur with normal ageing. For example, there is general agreement that declines in aspects of cognitive functions begin to occur after the age of 60 years and affect over 40% of people (1). Moreover, age-related cognitive changes are selective rather than diffuse. Although no changes were apparent in tests of language, visuospatial ability, and abstract reasoning, memory for the acquisition and retrieval of new information (explicit memory) becomes somewhat compromised with increasing age (2, 3). Working memory, the ability to hold information in memory while it is being manipulated, also declines with increasing age (4). Because explicit memory and working memory are subserved by the hippocampus and the prefrontal cortex, respectively, interest in cognitive ageing has largely focused on these two brain regions.

During the past 20 years, numerous randomised controlled studies and cross-sectional and longitudinal studies have been undertaken to investigate whether the administration of oestrogen to postmenopausal women could protect against the deterioration in explicit memory and in working memory that normally occur in a substantial proportion of ageing individuals. Although no attempt is made to review those findings here [for a review, see (5)], overall, these studies show that postmenopausal oestrogen-users tend to perform better on tests of explicit verbal memory compared to age-matched never-users of oestrogen. Having said that, it must be acknowledged that not every review and meta-analysis has reached this conclusion; the differences between studies in experimental design, in oestrogen compounds used, and in the characteristics of the women in the various samples (e.g. age, ethnicity, education, socioeconomic status, general health) complicate the effort to derive definitive conclusions. However, many studies have found a protective effect of oestrogen on the same two cognitive functions that deteriorate most with normal ageing in women, namely, explicit memory and working memory. However, this view was cast into doubt when the findings from the Women's Health Initiative Memory Study (WHIMS) were recently published. The WHIMS is an ancillary study to the Women's Health Initiative (WHI); it was designed to examine, in the context of a randomised trial, the effects of oestrogen-alone (E) (in hysterectomised women), of oestrogen plus progestin (E + P) or of a placebo on the incidence of probable dementia and, secondarily, on mild cognitive impairment (MCI) and global cognitive abilities (6). In the E + P study, there was a significantly higher risk of probable (all-cause) dementia in the drug group compared to the placebo group but no difference between the groups in the incidence of MCI, considered to be the precursor to AD (7), although it should be noted that WHIMS was not originally designed to test for the incidence of MCI. In the E-alone study (8), there were no significant differences in the incidence of probable dementia or MCI between the hormone and placebo groups, although there was a nonsignificant increase in probable dementia in the E-alone group [hazard ratio (HR) = 1.49; 95% confidence interval (CI) = 0.83–2.66, P = 0.18, for the oestrogen-alone group compared to placebo and the HR for the oestrogen plus progestin group was also nonsignificant (HR = 2.05; 95% CI = 1.21–3.48, P = 0.44)].The initially surprising failure of the WHIMS findings to support the predictions from earlier studies that oestrogen would protect against cognitive deterioration in postmenopausal women led to a reconsideration of all of the evidence in this area, especially in view of the fact that the WHIMS was the largest randomised controlled trial ever undertaken for the longest duration of time. A comparison of several characteristics of the WHIMS and other studies that had found a protective effect of oestrogen against cognitive decline led to numerous explanatory reasons for the differences. For example, almost 40% of the women in the WHIMS sample were either obese or morbidly obese, 55% had hypertension, 13% had a prior cardiovascular disease and 11% were diabetic compared to the healthier women who had participated in other trials. But perhaps the most striking difference between women who had participated in other randomised trials that found a protective effect of oestrogen on memory and the women in the WHIMS is age; whereas the most profoundly protective effect of oestrogen therapy (ET) had occurred in younger, surgically menopausal women (5), participants in the WHIMS were an average of 73 years of age at the time of their recruitment into the study and had been postmenopausal for approximately 21 years following their presumed spontaneous menopause at an average age of 51 years. This raised the issue of whether ET had protective effects on cognition only when it was initiated close in time to a natural or surgical menopause but not when initiated several decades after the menopause had occurred. The possibility that ET may have salutary effects on middle-aged brains and neutral, or even negative effects, on older brains led to the formulation of the critical period hypothesis, which holds that ET confers optimal benefits on cognition when initiated closely in time to the menopausal transition (9–11). Below, the available evidence is reviewed from basic neuroscience, from animal behavioural studies and from human studies that provide support for this hypothesis.

Evidence from basic neuroscience

  1. Top of page
  2. Abstract
  3. Evidence from basic neuroscience
  4. Animal behavioural studies
  5. Human studies
  6. Summary and conclusions
  7. Acknowledgements
  8. References

The finding that oestrogen increased the density of spines on neurones in the CA1 area of the hippocampus (12) provided a rationale for the clinical studies that had documented a protective effect of ET on aspects of memory in women. However, it was recently demonstrated that, whereas the administration of oestrogen to young ovariectomised rats increased axospinous synapse density, it was without effect on dendritic spine density in aged ovariectomised animals (13). This suggested that oestrogen-induced morphological plasticity is attenuated in aged rats compared with young animals. When young rats were ovariectomised and oestrogen treatment was initiated 4 days following surgery, synaptic density in three CA1 strata of the hippocampus was significantly greater than that in rats in whom ET had been withheld for 12 days following their ovariectomies (14). These findings suggested that delaying the initiation of ET may impair its ability to modulate synaptic density.

In ovariectomised rhesus monkeys, oestrogen treatment increased spinophilin-immunoreactive spines in the hippocampii of both young and aged animals (15, 16). However, it is important to note that the aged monkeys in those studies were actually pre- or peri-menopausal, corresponding to women in their early to mid-fifties and therefore these observations support the idea that ET increases spine density in the CA1 area of the hippocampus when treatment is initiated at the time of a spontaneous menopause or shortly thereafter.

Evidence also exists to show that the ability of oestrogen to increase the expression of brain-derived neurotrophic factor (BDNF) is dependant upon the age of the animal. When young adult rats were ovariectomised and treated with oestrogen, there was a significant increase in BDNF expression in the olfactory bulb as well as in its afferent, the horizontal limb of the diagonal band (17). However, when oestrogen was given to reproductively senescent animals, the expression of this neurotrophin actually decreased in both brain areas, suggesting once again that chronological ageing modulates brain responses to ET. Moreover, the ability of oestrogen to enhance the basal forebrain cholinergic function also declines with age (18).

Animal behavioural studies

  1. Top of page
  2. Abstract
  3. Evidence from basic neuroscience
  4. Animal behavioural studies
  5. Human studies
  6. Summary and conclusions
  7. Acknowledgements
  8. References

Animal studies also provide supportive evidence that oestrogenic effects on cognitive functioning may be dependant on the age of the animal or on the timing of the initiation of ET following the loss of ovarian function. When oestrogen or oestrogen plus progesterone (E + P) was initiated immediately or 3 months following ovariectomy in aged rats, they performed significantly better on a test of spatial memory than ovariectomised control rats and rats in whom ET was initiated after a 10 month delay following surgery (19). A more recent study provided confirmatory evidence. When E2 was administered to rats immediately following ovariectomy at either 12 or 17 months of age, a significant improvement in performance of acquisition and delay trials of a working memory task compared to the controls (20). Moreover, when E2 was administered 5 months following ovariectomy, no enhancements in memory performance were seen. Finally, while oestrogen enhanced basal forebrain cholinergic functions in middle-aged ovariectimised rats (irregular cyclicity), it was ineffective when administered to older ovariectomised rats (cessation of pro-oestrous) (18). These findings confirm that there may be a limited critical period, or window of opportunity, after the loss of ovarian function for the oestrogenic enhancement of brain functions that underlie memory to occur.

The findings regarding whether or not the time of initiation of ET following ovariectomy in aged rhesus monkeys is critical for a behavioural effect are inconsistent. Aged oestrogen-treated monkeys who had undergone ovariectomy approximately 12 years earlier performed better on a hippocampally mediated test of spatial memory (21) but not on a test of working memory compared to aged untreated controls (22). On the other hand, two studies found a substantial increase in performance on a test of working memory in oestrogen-treated compared to untreated aged monkeys (23, 24). Although the reason for these discrepancies is not clear, it is possible that, in rhesus monkeys, the hippocampus remains sensitive to oestrogen even after long periods of oestrogen deprivation whereas the prefrontal cortex loses its sensitivity to oestrogen when treatment is initiated following a considerable period of hormone deprivation.

Human studies

  1. Top of page
  2. Abstract
  3. Evidence from basic neuroscience
  4. Animal behavioural studies
  5. Human studies
  6. Summary and conclusions
  7. Acknowledgements
  8. References

There is accumulating evidence from cross-sectional studies of women to show that age of initiation of oestrogen treatment influences its protective effects on cognitive functioning in postmenopausal women. Significantly less cognitive decline occurred in elderly women who initiated ET at menopause compared to those with more recent exposures and those who were older when they initiated ET (25). Likewise, only the early initiation of hormone therapy (HT) around the time of menopause was associated with better cognitive function in a recent Australian sample of postmenopausal women (26).

The Women's Health Initiative Study of Cognitive Ageing recently reported on the conjugated equine oestrogen (CEE) + medroxypogesterone acetate (MPA) arm of the study (27). CEE + MPA had a negative effect on verbal memory but a positive effect on figural memory compared to the placebo group over time. However, the fact that MPA antagonises oestrogen-induced neuroprotection (28) make these findings unclear.

There has also been considerable interest as to whether ET or HT might be influential in the primary prevention of Alzheimer's disease (AD). Whereas findings from observational studies generally support the hypothesis that HT reduces AD risk in women, the results of the WHIMS found an increased incidence of AD in the group that received E + P (7) and a greater, albeit nonsignificantly increased risk of AD in women who had been randomised to E-alone (8). However, it is important to recall that the women who participated in WHIMS were all over the age of 65 years and over half of the sample was over the age of 70 years at the time of their recruitment. This raised the question of whether the early initiation of HT around the time of menopause might be beneficial whereas late initiation might even be potentially harmful with regard to the risk of AD. Indeed, findings from a prospective observational study in a homogeneous population of older women and men living in Utah, USA, provided some evidence that this may be true (29). In that study, prior users of ET who had initiated treatment earlier after the menopause had a significantly reduced risk of AD whereas current use of oestrogen did not affect the incidence of AD. Further support for the critical period hypothesis comes from the recent report of the Multi-Institutional Research in Alzheimer's Genetic Epidemiology Study which observed that, although HT was associated with reduced AD risk, the effect was apparent only among younger women (30). Finally, a follow-up study of women who had participated in randomised controlled trials of oestrogen and osteoporosis provides additional supportive evidence. Among women who were assessed approximately 15 years following their earlier participation in these trials, cognitive impairment was significantly less common among those who had been randomised to receive HT for 2–3 years, 15 years earlier, compared to those who had randomly been treated with placebo (31). These findings also imply that the administration of ET around the time of menopause endows enduring protective benefits against cognitive ageing. However, because of its considerable clinical implications, these finding need to be replicated.

It must also be emphasised that not all research findings are in agreement with the critical period hypothesis. For example, in an RCT of 65 years old women, those who received E2 for 3 weeks had improvement in scores on tests of learning and memory (32) and, in another study, higher serum E2 levels in 69 years old women given E2 for 2 weeks were positively associated with scores on tests of verbal memory (33). In view of the fact that the critical period hypothesis would have predicted that women enrolled in these studies approximately 15 years following their menopause would not benefit from ET, other possible factors that may account for the discrepancies in this literature require serious consideration. In that vein, it is interesting to note that both of the positive studies of ET on aspects of cognition in women over the age of 65 years described above were achieved with the use of a transdermal E2 preparation. Therefore, future studies should attempt to evaluate the relative efficacy of different oestrogen compounds, different routes of administration and the role of progestogens on cognitive functioning in menopausal and in postmenopausal women.

Summary and conclusions

  1. Top of page
  2. Abstract
  3. Evidence from basic neuroscience
  4. Animal behavioural studies
  5. Human studies
  6. Summary and conclusions
  7. Acknowledgements
  8. References

It would seem that the critical period hypothesis, which holds that ET endows maximal cognitive benefits on ageing women when it is initiated during the perimenopause or the early postmenopause, has a considerable amount of support from disparate areas of scientific investigation. The findings from basic neuroendocrinology show that the ageing brain responds differently and less beneficently to oestrogen than a young or middle-aged brain. Although the causes that underlie these changes that occur with ageing are as yet unknown, their demonstrated occurrence provides one explanation for the results of the WHIMS. Moreover, findings from both rodent and nonhuman primate studies provide behavioural evidence that uphold these observations by showing that memory function in aged animals or in those in whom ET was initiated after a considerable period of time has elapsed since their spontaneous menopause or ovariectomies did not benefit from hormone treatment. Although it is more difficult, of course, to perform controlled experiments in humans to test the critical period hypothesis, evidence from studies that were not necessarily undertaken to test it, have nevertheless provided impressive, but not definitive, evidence in its support.

If it is true that ET provides enduring protective benefits against cognitive ageing and delays the onset of AD in people who are destined to develop it for reasons that are still unknown, then the responses of the brain to oestrogen would seem to have parallels to the response to oestrogen by other organ systems. For example, although the WHI was underpowered to detect an oestrogenic benefit for cardioprotection of women who initiated therapy under the age of 60 years, findings from the oestrogen-alone arm and from other studies suggest that the administration of oestrogen to women around the time of menopause might indeed be cardioprotective (34). The clinical implications of these findings take on increased importance in view of the fact that unopposed ET did not increase the risk of breast cancer for the first 10 years of its continuous use (35). If these data are eventually confirmed, taken together, they would suggest that the initiation of ET close in time to the menopause and taken for 3–4 years may endow lasting neuroprotection without increasing the risk for cardiovascular disease or breast cancer. However, it must be underlined that the evidence to support such a statement has been gathered from different studies of heterogeneous populations of women using different doses and routes of administration of a variety of oestrogen compounds which undermines, to some degree, confidence in its veracity. Because of the potential clinical importance of the critical period hypothesis for the health of ageing women, it is hoped that future studies will be undertaken to provide direct evidence to confirm it.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Evidence from basic neuroscience
  4. Animal behavioural studies
  5. Human studies
  6. Summary and conclusions
  7. Acknowledgements
  8. References

The preparation of this manuscript was supported by a grant from the Canadian Institutes of Health Research (No. # MOP-77773) awarded to B. B. Sherwin.

References

  1. Top of page
  2. Abstract
  3. Evidence from basic neuroscience
  4. Animal behavioural studies
  5. Human studies
  6. Summary and conclusions
  7. Acknowledgements
  8. References
  • 1
    Hanninen T, Koivisto K, Reinikainen KJ, Helkala EL, Soininen H, Mykkanen L, Laakso M, Riekkinen PJ. Prevalence of aging-associated cognitive decline in an elderly population. Age Aging 1996; 25: 201205.
  • 2
    Small SA, Stern Y, Tang M, Mayeux R. Selective decline in memory function among healthy elderly. Neurology 1999; 52: 392396.
  • 3
    Driscoll I, Hamilton DA, Yeo RA, Brooks WM, Sutherland RJ. Visual navigation in humans: the impact of age, sex, and hormones. Horm Behav 2005; 47: 326335.
  • 4
    Salthouse TA. Mediation of adult age differences in cognition by reduction in working memory and speed of processing. Psychol Sci 1991; 2: 179183.
    Direct Link:
  • 5
    Sherwin BB. Estrogen and cognition in women. Endocrine Rev 2003; 24: 133151.
  • 6
    Espeland MA, Rapp SR, Shumaker SA, Brunner R, Manson JE, Sherwin BB, Hsia J, Margolis KL, Hogan PE, Wallace R, Dailey M, Freeman R, Hays J. Conjugated equine estrogens and global cognitive function in postmenopausal women: Women's Health Initiative Memory Study. J Am Med Assoc 2004; 291: 29592968.
  • 7
    Shumaker SA, Legault C, Rapp SR, Thal L, Wallace RB, Ockene JK, Hendrix SL, Jones BN III, Assaf AR, Jackson RD, Kotchen JM, Wassertheil-Smoller S, Wactawski-Wende J, for the WHIMS Investigators. Estrogen plus progestin and the incidence of dementia and mild cognitive impairment in postmenopausal women. J Am Med Assoc 2003; 289: 26512662.
  • 8
    Shumaker SA, Legault C, Kuller L, Rapp SR, Thal L, Lane DS, Fillit H, Stefanick ML, Hendrix SL, Lewis CE, Masaki K, Coker LH, for the Women's Health Initiative Memory Study. Conjugated equine estrogens and incidence of probable dementia and mild cognitive impairment in postmenopausal women. J Am Med Assoc 2004; 291: 29472958.
  • 9
    Resnick SM, Henderson VW. Hormone therapy and risk of Alzheimer disease: a critical time. J Am Med Assoc 2002; 288: 21702172.
  • 10
    Maki PM. Hormone therapy and cognitive function: is there a critical period for benefit? Neuroscience 2006; 138: 10271030.
  • 11
    Sherwin BB. Estrogen and memory in women: how can we reconcile the findings? Horm Behav 2005; 15: 3748.
  • 12
    Gould E, Woolley CS, Frankfurt M, McEwen BS. Gonadal steroids regulate dentritic spine density in hippocampal pyramidal cells in adulthood. J Neurosci 1990; 10: 12861291.
  • 13
    Adams MM, Shah RA, Janssen WGM, Morrison JH. Different modes of hippocampal plasticity in response to estrogen in young and aged female rats. PNAS 2001; 98: 80718076.
  • 14
    Silva I, Mello LEAM, Freymuller E, Haidar MA, Baracat EC. Onset of estrogen replacement has a critical effect on synaptic density of CA1 hippocampus in ovariectomized adult rats. Menopause 2003; 10: 406411.
  • 15
    Hao J, Janssen WGM, Tang Y, Roberts JA, McKay H, Lasley B, Allen PB, Greengard P, Rapp P, Kordower IH, Hof PR, Morrison JH. Estrogen increases the number of spinophilin-immunoreactive spines in the hippocampus of young and aged female rhesus monkeys. J Comp Neurol 2003; 465: 540550.
  • 16
    Hao J, Rapp PR, Leffler AE, Leffler SR, Janssen WGM, Lou W, McKay H, Roberts JA, Wearne SL, Hof PR, Morrison JH. Estrogen alters spine number and morphology in prefrontal cortex of aged female rhesus monkeys. J Neurosci 2006; 26: 25712578.
  • 17
    Jezierski MK, Sohrabji F. Neurotrophin expression in the reproductively senescent forebrain is refractory to estrogen stimulation. Neurobiol Aging 2001; 22: 311321.
  • 18
    Savonenko AV, Markowska AL. The cognitive effects of ovariectomy and estrogen replacement are modulated by aging. J Neurosci 2003; 119: 821830.
  • 19
    Gibbs RN. Long-term treatment with estrogen and progesterone enhances acquisition of a spatial memory task by ovariectomized aged rats. Neurobiol Aging 2000; 21: 107116.
  • 20
    Daniel JM, Hulst JL, Berbling JL. Estradiol replacement enhances working memory in middle-aged rats when initiated immediately after ovariectomy but not after a long-term period ovarian hormone deprivation. Endocrinology 2006; 147: 607614.
  • 21
    Lacreuse A, Wilson ME, Herndon JG. Estradiol, but not raloxifene, improves aspects of spatial working memory in aged, ovariectomized rhesus monkeys. Neurobiol Aging 2002; 23: 589600.
  • 22
    Lacreuse A. Effects of ovarian hormones on cognitive function in nonhuman primates. J Neurosci 2005; 138: 859867.
  • 23
    Rapp PR, Morrision JH, Roberts JA. Cyclic estrogen replacement improves cognitive function in aged ovariectomized rhesus monkeys. J Neurosci 2003; 23: 57085714.
  • 24
    Tinkler GP, Voytko ML. Estrogen modulates cognitivre and cholinergic processes in surgically menopausal monkeys. Prog Neuro-Psychopharm Biol Psychiatry 2005; 29: 423431.
  • 25
    Matthews KA, Cauley J, Yaffe K, Zmuda JM. Estrogen replacement therapy and cognitive decline in older community women. J Am Geriatr Soc 1999; 47: 518523.
  • 26
    MacLennan AH, Henderson VW, Paine BJ, Mathias J, Ramsay EN, Ryan P, Stocks NP, Taylor AW. Hormone therapy, timing of initiation, and cognition in women aged older than 60 years: the REMEMBER pilot study. Menopause 2006; 13: 2836.
  • 27
    Resnick SM, Maki PM, Rapp SR, Espeland MA, Brunner R, Coker LH, Granek IA, Hogan P, Ockene JK, Shumaker SA. Effects of combination estrogen plus progestin hormone treatment on cognition and affect. J Clin Endocrinol Metab 2006; 91: 18021810.
  • 28
    Nilsen J, Brinton RD. Impact of progestins on estrogen-induced neuroprotection: synergy by progesterone and 19-norprogesterone and antagonism by medroxyprogesterone acetate. Endocrinology 2002; 143: 205212.
  • 29
    Zandi PP, Carlson MC, Plassman BL, Welsh-Bohmer KA, Mayer LS, Steffens DC, Breitner JC. Hormone replacement therapy and incidence of Alzheimer's disease in older women: the Cache County Study. J Am Med Assoc 2002; 288: 21232129.
  • 30
    Henderson VW, Benke KS, Green RC, Cupples LA, Farrer LA. Postmenopausal hormone therapy and Alzheimer's disease risk: interaction with age. J Neurol Neurosurg Psychiatry 2005; 76: 103105.
  • 31
    Bagger YZ, Tanko LB, Alexandersen P, Qin G, Christiansen C. Earlypostmenopausal hormone therapy may prevent cognitive impairment later in life. Menopause 2005; 12: 1217.
  • 32
    Duka I, Tasker R, McGowan JF. The effects of a 3-week estrogen hormone replacement on cognition in elderly healthy females. Psychopharmacology 2000; 149: 129139.
  • 33
    Yaffe K, Lui L-Y, Grady D, Cauley J, Kramer J, Cummings SR. Cognitive decline in women in relation to non-protein-bound oestradiol concentrations. Lancet 2000; 356: 708712.
  • 34
    Anderson GL, Limacher M, Assaf AR, Bassford T, Beresford SA, Black H, Bonds D, Brunner R, Brzyski R, Caan B, Chlebowski R, Curb D, Gass M, Hays J, Heiss G, Hendrix S, Howard BV, Hsia J, Hubbell A, Jackson R, Johnson JC, Judd H, Kotchen JM, Kuller L, LaCroix AZ, Lane D, Langer RD, Lasser N, Lewis CE, Manson J, Margolis K, Ockene J, O'Sullivan MJ, Phillips L, Prentice RL, Ritenbaugh C, Robbins J, Rossouw JE, Sarto G, Stefanick ML, Van Horn L, Wactawski-Wenda J, Wallace R, Wassertheil-Smoller S. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women's Health Initiative randomized controlled trial. J Am Med Assoc 2004; 291: 17011712.
  • 35
    Chen WY, Manson JE, Hankinson SE, Rosner B, Holmes MD, Willett WC, Colditz GA. Unopposed estrogen therapy and the risk of invasive breast cancer. Arch Int Med 2006; 166: 10271032.