The Protective Effect of Pregnancy
It has been firmly established in epidemiological studies that early full-term pregnancy affords lifetime protection against the development of breast cancer. This phenomenon can be mimicked in rat and mouse models of mammary cancer in which the hormones estrogen and progesterone are given for 21 days. Carcinogen-induced proliferation is blocked as a consequence of hormone pretreatment. Among several genes implicated by molecular studies to be differentially expressed is the tumor suppressor gene p53. Both immunohistochemical and Western blot studies indicate that p53 protein expression is increased in hormone-pretreated mice and rats. The p53-regulated gene p21Cip1 is also increased concomitantly with p53. To test directly the causative role of p53 in conferring a protective phenotype, we examined the hormone-induced protective effect in BALB/c p53 null mammary epithelium. In the mammary epithelium, the absence of p53 gene expression abrogated the protective effect of prior pregnancy. The tumor incidence curves were superimposable in p53 null mammary epithelium that were treated with 7,12-dimethylbenzanthracene or pregnancy plus 7,12-dimethylbenzanthracene. These results demonstrate that p53 plays a pivotal role in hormone-induced protection and raises the question of the mechanisms by which the steroid hormones, estrogen and progesterone, functionally activate p53.
Breast cancer remains the major cancer among women in the United States with ∼190,000 new cases annually (1) . The annual incidence rate has leveled in the last decade to ∼111 cases/100,000 women, whereas the mortality rate has declined with the introduction of tamoxifen as an adjuvant and preventive therapy (2) . Nonetheless, the death of ∼40,000 women annually (1) attributed to breast cancer is a sobering fact and indicates the necessity to develop new approaches to prevention.
Breast cancer incidence rates are influenced by age, genetics, radiation, socioeconomic status, diet, ethnicity, and reproductive history. Reproductive history is the strongest and most consistent risk factor outside of genetic background and age (3) . Early menarche and total years of hormone exposure are risk factors for increased incidence, whereas early age of first pregnancy (≤20 years of age) is a strong protective factor (one-half risk compared with nulliparous women). The protective factor is especially observed in the postmenopausal years, the period of peak incidence rates. Parity-induced protection against breast cancer is principally dependent on the timing of the first full-term pregnancy rather than on its occurrence per se. Aborted pregnancies are not associated with decreased risk for breast cancer. The possible mechanistic interpretations are discussed below. The protective effect of early first pregnancy has been repeatedly demonstrated in numerous epidemiological studies and provides a physiologically operative model to achieve practical and affordable prevention of breast cancer in women. The logic and rationale for understanding the molecular basis for hormone-mediated prevention of breast cancer is based on the consistent observations in human epidemiological studies and the strong confirmatory experiments in rodent breast cancer models.
Modeling the Protective Effect.
There are two different experimental models that demonstrate parity or hormone-induced protection (Fig. 1)<$REFLINK> . In the first model, the animals undergo hormonal stimulation; the mammary gland is allowed to involute completely; then the carcinogen is administered to the hormone-treated and age-matched virgin animals. This model is termed the pretreatment model, and most of the experiments use this experimental protocol. In the second model, the animals are treated with a carcinogen, then exposed to hormone treatment for a specified time period. This model is referred to as the post-treatment model. Unless otherwise stated, comments herein refer to the pretreatment model. The distinction is important, as the underlying mechanisms for protection are likely to be different between the two models.
The protective effect of pregnancy is operative in both rats and mice (Table 1)<$REFLINK> . In outbred Sprague Dawley rats, and inbred Lewis and Wistar-Furth rats, there is often an 80% inhibition of the incidence of mammary carcinomas. In inbred C3H/Sm, BD2/fF1, and BALB/c mice, there is an ∼67% inhibition of mammary carcinomas. Prevention of mammary carcinogenesis can be attained by a single pregnancy; by a hormone regimen comprising the two main hormones of pregnancy, estrogen and progesterone; by estrogen alone; or by human chorionic gonadotropin (4) . The concentrations of estrogen (E) and progesterone (P) needed to confer protection yield circulating hormone levels found in mid-pregnancy. A pellet containing modest hormone doses (20 μg E and 20 mg P) and releasing hormones over 21 days is very protective and yields only a 10% tumor incidence with a mean latent period of 164 days. Lower doses (10 μg E and 10 mg P, and 5 μg E and 5 mg P) were partially protective, with 30% tumor incidence (latency 159 and 169 days, respectively). Thus, doses of E and P (10/10 and 5/5 E/P) that do not induce complete functional differentiation are partially protective (4) .
Mechanisms of Protection.
The ultimate mechanism through which early pregnancy protects the breast from cancer development remains largely unknown. Several hypotheses have been proposed to account for the hormone-driven protective effects of parity, using studies performed in an experimental animal model based on the induction of mammary carcinomas by administration of chemical carcinogen to rats. Whatever the mechanism, the effects must be nonreversible to explain why an early first pregnancy affords lifetime protection.
The main mechanisms proposed over the years are listed in Table 2<$REFLINK> . There is contradictory evidence regarding the importance of persistent phenotypic changes, the induction of differentiation per se, or decreased proliferative activity in the parous involuted gland, with the majority of the data suggesting that these properties are not intimately involved in hormone-induced protection. An increase in apoptotic activity has been implicated as an important mechanism in the post-treatment model (5) but is not implicated in the pretreatment model (6) . The role of apoptosis needs to be independently verified.
The two prevailing hypotheses, not mutually exclusive, are the induction of an altered systemic hormonal milieu and an altered cell fate (or persistent changes in intracellular regulatory circuits). The former hypothesis stresses the altered systemic levels of growth hormone and its downstream effectors as modifying the progression of chemical carcinogen-initiated mammary cells (7) . The latter hypothesis stresses that hormones induce a molecular switch in stem cells that results in cells with persistent changes in intracellular regulatory loops governing proliferation and response to DNA damage. Several recent publications by three laboratories have provided evidence in support of the latter hypothesis (8, 9, 10) . The remaining part of this article will focus on one potential gene alteration as a mediator of the protective effect, the p53 tumor suppressor gene.
A Role for p53 in Hormone-Induced Protection.
A role for p53 was suggested by two experiments. The first experiment examined the proliferative response of involuted and age-matched virgin (AMV) mammary gland to the chemical carcinogen methylnitrosourea (MNU). The frequency of apoptotic cells was low and comparable in the hormone-exposed and AMV gland at the time of carcinogen challenge, and it remained low for 8 days after MNU treatment. However, the frequency of proliferating cells was significantly different in the two treatment groups. The results showed that although the number of bromodeoxyuridine-labeled cells at the time of carcinogen challenge was low in both the AMV (1.8%) and hormone-exposed (0.8%) animals, when analyzed at 8 days after MNU treatment, cell proliferation in the AMV (5.7%) was significantly different from the parous involuted (1.2%) or the E/P-treated animals (1.5%). Therefore, it appears that hormone treatment results in persistent alterations in intracellular pathways governing proliferation responses to carcinogens, alterations that are most clearly evident at 8 days after carcinogen challenge rather than at the time of challenge. The difference in proliferation frequency disappeared by 20 days after carcinogen challenge. This differential response was observed in both rat and mouse mammary gland (4 , 6) .
Second, Kuperwasser et al. (11) discovered that the localization and functional activation of normal p53 is a result of pregnancy and hormone stimulation of the gland. Hormonal treatment is sufficient to induce alveolar proliferation and the differentiation activates p53 by unknown mechanisms. To determine whether the block in proliferation observed early after MNU treatment was associated with an E/P-induced change in the expression of p53, we compared the levels of p53 protein in AMV and E/P-treated glands from 0 to 8 days after MNU treatment by immunohistochemistry.
The results demonstrated a striking increase in both the level and nuclear localization of p53 in E/P-treated mammary glands compared with the AMV. The increase occurred as early as 3 days after E/P treatment and was sustained through the 21 days of stimulation and 28 days of involution. The expression persisted through 3 days after MNU treatment but decreased by day 8 post-MNU treatment. There was also a concomitant change in the protein levels of p21Cip1. We hypothesized that one mechanism of hormone-induced protection could be through the activation of p53, so that upon DNA damage (as induced by chemical carcinogen), p53 was functionally active to cause G1 arrest. We additionally suggested that this effect is post-transcriptional, because we had shown earlier that the p53 mRNA levels are comparable in the parous involuted and the AMV gland. This result has major implications. DNA damage leads to the accumulation of p53 protein in the nucleus, where p53 acts as a transcriptional activator for a group of genes involved in cell cycle arrest, DNA repair, or apoptosis. Recent results also demonstrate a dependence on p53 in global DNA repair of polycyclic aromatic hydrocarbon adducts (12) .
To test the hypothesis that p53 was causally involved in hormone-induced protection of carcinogen-induced mammary tumorigenesis, we used the BALB/c p53 null mammary epithelium transplant model. In this model, p53 null mammary epithelium is transplanted into the cleared mammary pads of 3-week-old BALB/c p53 wild-type mice (13) . The mice were mated at 6 weeks of age, underwent a single pregnancy, nursed their pups for 7 days, and the mammary glands were allowed to involute for 21 days before receiving 1 mg DMBA/week for 4 consecutive weeks. DMBA induced an incidence of 65% mammary tumors in the p53 null mammary epithelium compared with 27% in untreated p53 null mammary cells in virgin mice (Fig. 2)<$REFLINK> . A single pregnancy did not significantly alter the incidence of mammary tumors in either the untreated p53 null mammary cells (15%) or in the DMBA-treated p53 null mammary cells (60%). Thus, the absence of p53 function completely abolished the protective effect of pregnancy against chemical carcinogen-induced mammary tumors that has been observed repeatedly in mice and rat models.
The parous mammary gland and the virgin mammary gland represent distinct developmental states of the gland, and information on the signal transduction pathways operative in these different developmental states is scarce. Whereas we know that estrogens and progestins are important in stimulating proliferation and differentiation, the specific receptor subpopulations, secondary messengers, effector loops, and growth factor responses operative in the two mammary cell populations are relatively unknown. It is impossible to effectively interpret the mechanisms underlying the refractory state without understanding the basic biological processes involved in regulation of cell growth and differentiation.
Recent experiments demonstrate the important role of the p53 tumor suppressor gene in the hormone-induced protection. Whereas the downstream effector proteins are partly known (i.e., p21Cip1), the mechanism by which hormones activate p53 function are unknown. Furthermore, it is likely the p53 pathway is only one of several pathways that are important for hormone-mediated protection. Gene expression profile experiments have demonstrated a consistent array of alterations in genes involving a number of growth factor pathways and gene transcription regulatory pathways in the parous mammary gland (8, 9, 10) .
Extensive data have shown the importance of parity as a significant preventive factor. The striking consistency of these data and the profound quantitative differences indicate the strongest need to understand the process at a molecular level. Because the effect of parity (hormones) is so dramatic and strong in both humans and rodents, it offers an approach to breast cancer prevention that is rational and attainable at the social, sexual, political, and scientific levels. The steady high incidence of human breast cancer is extensively documented and essentially unexplained. The economic, medical, and practical benefits of preventive approaches to solving and handling critical and widespread diseases are unarguable. Therefore, the ultimate goal of understanding the molecular mechanisms of hormone-induced protection has great potential for breast cancer prevention. If specific information can be generated on the mechanisms and protocols for hormonal prevention of breast cancer, it is possible that this information can be used to design a prevention protocol applicable for humans.
Dr. Steven Come: What is happening with oral contraceptives, which seem to slightly increase risk? Would you predict from this model that they would have a protective effect?
Dr. Medina: In this time window, if you gave hormonal contraceptives, I would predict a protective effect. No one has done that, using the specific hormonal combination in oral contraceptives in a short period of time. You have to distinguish between a short-term and a long-term treatment, and with oral contraceptives the use is probably long term. Continuous hormone treatment acts to promote because it is not changing initiation, the response to carcinogen. What it is changing is how the cell responds.
Dr. Richard Santen:It is interesting to think, then, about the morning after pill, which represents short-term use in younger women. It is likely that no epidemiological data are available regarding breast cancer risks after this mode of estrogen administration. If you had some mechanism of getting breast tissue, it would be of interest to study what effects on the breast would be occurring shortly after that exposure.
Dr. Matthew Ellis: I’m thinking about the possibility of human studies with respect to the hormonal regulation of p53. There are several situations in which you might be able to get ductal epithelial cells from healthy virgin or at least nonparous human volunteers, with respect to either ductal lavage or ductoscopy. In fact, breast surgery is not that uncommon for fibroadenomas, for example, in young women. The question always with these very interesting animal studies is whether the mechanism happens in humans. Because if it does, p53 immunohistochemistry could be a nice surrogate for the protection effect, which you might be able to use as a platform to develop the appropriate “Pike’s pill.”
Dr. Medina: I don’t know how feasible it is, but I do think it is an important approach, to try to extend a hypothesis. We are rather trying to understand the mechanisms so that one can identify a target that one might modulate to achieve the same effect.
Dr. Jeffrey Green: Have you any idea on whether the hormone effect is direct or some effect is contributed by the stroma?
Dr. Medina: No one has tested the effect of the stroma. That is a hard experiment to do.
Dr. Santen: When you give them DMBA, is that a procarcinogen that has to be metabolized into a more active form? Does estrogen or progesterone or the p53 null genotype in any way alter the conversion of the procarcinogen to the carcinogen?
Dr. Medina: There are 2 carcinogens that have been used, DMBA which needs to be metabolized and MNU, which does not need to be metabolized. The Nandi group have looked at this with DMBA [Carcinogenesis (Lond.), 16: 2847–2853, 1995] and showed that you don’t change the rate of metabolism nor do you change the frequency of adducts that are formed as a consequence of hormone stimulation. So you don’t change the initiation.
Dr. Robert Nicholson: I guess I was brought up to believe that when you give carcinogens to rats and mice that the effects were primarily on the terminal end buds, but your studies on p53 showed p53 expression within the lobular units of the ducts. Have you looked specifically at the terminal end buds or have I got the story completely wrong?
Dr. Medina: What you say is true: the period of peak sensitivity in the rodent is the time when you have the developing gland and a maximum frequency of terminal end buds as the duct is elongating in the gland, so this is roughly 5 to 6 weeks of age. The terminal end bud represents a focus of proliferating cells. At 96 days the gland has grown out, and there are very few terminal end buds. The frequency of proliferation is much less, about 1 to 2%, as compared to 8 to 10% at 6 weeks of age. That is one reason why the frequency of transformation is much less. In most of these experiments when you give MNU or DMBA to a mature virgin you’re getting 60 to maybe 80% mammary tumors, 1 mammary tumor per animal. Whereas if you treat at an early time period when there are a lot of terminal end buds, it is 100%, shorter latency, with multiple tumors per animal. So there is a difference in the frequency of transformation, but there are susceptible cells in the mature gland.
Dr. Nicholson: So you think that the susceptible cells would be throughout the gland?
Dr. Medina: Yes, and that has been demonstrated. Beattie (14) did a really thorough study in the late 1980s, showing that decisive transformation occurred not only in terminal end bud, but in other sections of the mammary gland. Clearly in the adult mammary gland where there are no terminal end buds, it’s adverse, with premalignant lesions occurring throughout the mammary gland.
Dr. Myles Brown: Maybe this is actually the genotoxic effect of some estrogen metabolite messenger inducing p53, and it has nothing to do with the transcriptional or even nongenomic effects of ER. Is p53 induced in all the mammary epithelial cells or only the ER-positive cells?
Dr. Medina: p53 is in the majority of the epithelia, ER in a subset. We have never done the double labeling to see if they are mutually exclusive. I would suggest just from the numbers that they are not.
Dr. Adrian Lee: What do you think the mechanism is for the disruption of p53 levels?
Dr. Medina: We know that you get phosphorylation on serine 15 and serine 20 and acetylation on carboxy-N as a consequence of this. That is part of the stabilization and functional activity. What we don’t know is the kinase that is inducing the phosphorylation. We have several candidates that we will sort through and then test that hypothesis.
Presented at the Third International Conference on Recent Advances and Future Directions in Endocrine Manipulation of Breast Cancer, July 21–22, 2003, Cambridge, MA.
Grant support: Grant no. NCI-P01-CA64255.
Requests for reprints: Daniel Medina, Department of Molecular and Cellular Biology, Baylor College of Medicine, One Baylor Plaza, Houston, Texas 77030. Phone: (713) 798-4483; Fax: (713) 790-0545; E-mail: