This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by López-Otín, C. Articles by Diamandis, E. P. Search for Related Content PubMed PubMed Citation Articles by López-Otín, C. Articles by Diamandis, E. P.

Breast and Prostate Cancer: An Analysis of Common Epidemiological, Genetic, and Biochemical Features¹

Departamento de Bioquímica y Biología Molecular (C.L.-O.), Facultad de Medicina, Universidad de Oviedo 33006, Oviedo, Spain; and Department of Pathology and Laboratory Medicine, Mount Sinai Hospital, Toronto, Ontario, Canada, M5G 1X5 and Department of Laboratory Medicine and Pathobiology, University of Toronto (E.P.D.), Toronto, Ontario M5G 1L5, Canada

	Abstract

Top Abstract I. Introduction II. Epidemiological Evidence... III. Incidence of Breast... IV. Genetic Abnormalities Common... V. Common Biochemical Features... VI. Growth Factors in... VII. Theories of Breast... VIII. Conclusions References

 

I. Introduction

II. Epidemiological Evidence Associating Breast and Prostate Cancer

III. Incidence of Breast and Prostate Cancer in Different Countries: Dietary Factors

IV. Genetic Abnormalities Common to Breast and Prostate Cancer

A. AR alterations in prostate cancer

B. AR alterations in breast cancer

C. BRCA1 and BRCA2 alterations in breast cancer

D. BRCA1 and BRCA2 alterations in prostate cancer

E. Other genes associated with breast or prostate cancer

V. Common Biochemical Features of Breast and Prostate Cancer

A. Prostate-specific antigen (PSA)

B. Apolipoprotein D (apoD)

C. Zn-₂-gp

D. Gross cystic disease fluid protein-15

E. Pepsinogen C

F. Other proteins

VI. Growth Factors in Breast and Prostate Cancer

A. AIGF

B. KGF

VII. Theories of Breast and Prostate Cancer Development: Role of Steroid Hormones

VIII. Conclusions

	I. Introduction

Top Abstract I. Introduction II. Epidemiological Evidence... III. Incidence of Breast... IV. Genetic Abnormalities Common... V. Common Biochemical Features... VI. Growth Factors in... VII. Theories of Breast... VIII. Conclusions References

 
BREAST cancer is the most common malignancy among females in North America. Statistics show that in the United States alone, about 200,000 new cases are diagnosed every year and about 50,000 women die annually from the disease. Worldwide, the approximate figures are about 10-fold higher (1). Overall, about 15% of all women will be diagnosed with breast cancer during their lifetime. Despite intense research efforts, which are increasing worldwide, the pathogenesis of the disease is still largely not well understood. Although diagnosis is now more effective through mammographic screening, mortality rates remain almost unchanged. Steroid hormones, including estrogens, androgens, and progestins, have long been implicated in the pathogenesis and progression of breast cancer. Early efforts to control breast cancer included either hypophysectomy or ovariectomy, which represent an attempt to remove the steroid hormones from the tumor environment. Such efforts, which are still used but with utilization of pharmacological agents instead of surgery, have clearly beneficial but mostly transient effects. A convincing finding that directly implicates steroid hormones in breast cancer development and progression is that women who have bilateral oophorectomy at an early age (<40 yr) are at markedly reduced risk of subsequently developing breast cancer; the earlier oophorectomy is done, the greater the risk reduction (1). Added to this finding are the well known modifying risks of breast cancer related to age at first full-term pregnancy, age of menarche and menopause, number of menstrual cycles in a lifetime, oral contraceptive use, number of pregnancies, etc. However, despite the wealth of literature on steroid hormone involvement in breast cancer, we do not as yet have definitive answers to even simple questions such as: Are steroid hormones carcinogens? Do steroid hormones control breast cancer cell proliferation and growth rates (2)? More recently, it has become evident that steroid hormones not only have direct actions on certain types of cells, but they can trigger additional effects through growth factors that are regulated by them; the latter act on neighboring cells in an autocrine/paracrine fashion (3). Lately, it has also been shown that steroid hormones are produced locally by cells that then use them intracellularly. Usually, the parent molecules are precursor steroids produced by the adrenals. This mode of hormone action is now referred to as ‘intracrine‘ and further expands the possible implications of steroid hormones in breast cancer pathogenesis in humans (4, 5).

In males, the breasts are rarely affected by breast cancer. The androgenic dominance over estrogens in males keeps the breasts underdeveloped throughout life. However, even for male breast cancer, the major contributing role of steroid hormones is evident from clinical observations. For example, a number of conditions have been established, almost all related to hypoandrogenism, that increase the risk of breast cancer in males. These include Klinefelter’s syndrome, testicular atrophy, orchitis, undescended testes, testicular trauma, infertility, and defects in androgen receptor (AR) genes (6).

Prostate cancer is the most common malignancy among males in North America. In the United States, about 200,000 new cases are diagnosed every year and about 45,000 men die annually from the disease. Overall, about one in every nine men will be diagnosed with prostate cancer during their lifetime. These numbers are strikingly similar to those already mentioned for breast cancer (7). Similarly to breast cancer, the pathogenesis of the disease is also obscure. Major risk factors include age, ethnicity, family history, and steroid hormones. While the rate of increase of breast cancer incidence declines postmenopausally in women (1), the rate of increase of prostate cancer incidence increases continually with age. This phenomenon is likely linked to the continuation of testicular function in males throughout life and the cessation of ovarian function during the female menopause. The involvement of steroid hormones in the pathogenesis and progression of prostate cancer has been suggested for many years. Huggins, as early as 1940, was able to achieve transient remission of prostate cancer with orchiectomy and with administration of estrogens (8, 9). Currently, pharmacological androgen ablation therapy is achieved either by blocking androgen production or activity by administration of antiandrogens or other agents. Males who have diminished androgen production due to castration, hypogonadism, or enzyme defects of androgen metabolism (e.g., 5-reductase) have minimal risk for prostate cancer.

We and others have hypothesized that breast and prostate cancer may represent, in some aspects, homologous cancers in females and males, respectively. Breast and prostate cancer are now the two most common cancers with a roughly equal lifetime risk. They are both influenced strongly by steroid hormones, gonadal removal reduces the risk dramatically in both sexes, and antiestrogens are beneficial and possibly preventive for breast cancer while antiandrogens are beneficial and possibly preventive for prostate cancer (10). Additionally, these two cancers have parallel incidence rates in various countries, and there is evidence suggesting that they are both influenced by the same dietary factors (e.g., fat consumption). Macklin, as early as 1954, provided evidence for a significantly higher frequency of prostate cancer among relatives of breast cancer patients and proposed for the first time that prostate cancer may be the male equivalent of some female breast cancers (11). In the last few years, additional epidemiological, genetic, and biochemical findings support the view that these two cancers have many similar features. Here, we review the current knowledge, focusing on common features, in an attempt to understand these malignancies better and possibly trigger some new thinking into their pathogenesis and progression. The reader, however, should be aware that there may be a bias in our presentation since we have selected literature that cites a connection between the two cancers. Other literature that either does not cite a connection or cites a connection of breast or prostate cancer with other cancers was not systematically reviewed since it falls outside the scope of our manuscript. This biased presentation may be specially relevant in the discussion of the putative genetic and biochemical abnormalities shared by breast and prostate cancer, since many of the associations described in the literature are not exclusive of these tumors and could merely reflect the general process of carcinogenesis. Thus, in the case of genetic abnormalities, it is well known that breast and prostate cancer, as well as other human carcinomas, result from the accumulation of genetic lesions in a variety of oncogenes and tumor suppressor genes. However, none of these genes is exclusively damaged in breast and/or prostate carcinomas, thus limiting their value in the context of this review. Consequently, we have focused the discussion on those few genes like AR or breast cancer susceptibility genes BRCA1 and BRCA2, which have a high degree of specificity for one of the two tumors (prostate or breast cancer, respectively), but whose involvement in the other tumor (breast or prostate cancer) has been suggested through epidemiological, biochemical, or mutational analysis. Nevertheless, it must be emphasized that the contribution of common genetic factors to the overall incidence of both tumor types may be low in quantitative terms and circumscribed to a specific subgroup of patients. A similar consideration must be done in the discussion of putative common biochemical features shared by breast and prostate carcinomas. In this case, the finding of commonalities in the expression pattern of diverse bio-markers associated with the development and progression of breast and prostate cancer may be only a consequence of general alterations of critical cell functions occurring during the malignant transformation of human cells, but not specifically of mammary or prostatic epithelial cells. Therefore, we have focused the discussion on those biochemical markers that may be of special interest for the biology of these two carcinomas because of their relative specificity of expression in breast or prostate carcinomas when compared with tumors from other sources, or by the occurrence of shared mechanisms of hormonal control mediating their up- or down-regulation in these two hormone-sensitive cancers. Likewise, the discussion of commonalities in the expression and regulation of growth factors associated with breast and prostate cancer may be of limited value because many growth factor pathways are universally altered in most human malignancies. Consequently, and as in the case of biochemical markers discussed above, we have focused our attention on those growth factors that may be of special relevance in the context of breast and prostate cancer by both the relative specificity of the alterations and the finding of common hormonal networks underlying their effects on these carcinomas. Taken together, we must conclude from these observations that, based on data of the few comparative analyses currently available, the existence of common factors in breast and prostate cancer is still speculative in many aspects. The next sections present a summary of available epidemiological, genetic, and biochemical data supporting associations between both tumors, with a special emphasis on describing the common hormonal aspects underlying the observed associations.

	II. Epidemiological Evidence Associating Breast and Prostate Cancer

Top Abstract I. Introduction II. Epidemiological Evidence... III. Incidence of Breast... IV. Genetic Abnormalities Common... V. Common Biochemical Features... VI. Growth Factors in... VII. Theories of Breast... VIII. Conclusions References

 
The first observations regarding a familial association between breast and prostate cancer were performed more than four decades ago by Macklin (11) who, in her pioneering work designed to look for the genetic basis of human breast cancer, found a significantly higher frequency of prostate cancer among relatives of women with breast cancer than among relatives of control groups. According to these results, she proposed that prostate cancer could be the male equivalent of at least some female mammary carcinomas. Since then, several genetic epidemiological studies performed by different groups in different populations have provided further support to this original proposal.

Thiessen (12), in 1974, after analysis of the familial incidence and distribution of all malignancies in a group of 145 breast cancer patients, compared with that of 139 randomized control patients, reported that significantly higher incidences of only uterine, prostatic, and breast cancer were found among both maternal and paternal relatives of the breast cancer patients. On this basis, he proposed that the mammary gland is part of an integrated genital organ system whose different parts share unique biological and pathological characteristics, including hormone responsiveness and cancer susceptibility. He also hypothesized the existence of some common etiological factor that could operate in the development of tumors in diverse reproductive organs, including breast and prostate. In 1982, Cannon et al. (13), in a study of genetic epidemiology of prostate cancer in a population from the Utah Mormon genealogy, showed a significant coaggregation of prostate cancer with breast cancer. More recently, in case-control studies based on anamnestic data, Andrieu et al. (14) and Rosenblatt et al. (15) did not find evidence of association between these two tumors. By contrast, Tulinius et al. (16) in a large cohort study including 1539 Icelandic women with breast cancer, reported that the risk of prostate cancer was significantly raised for all male relatives, as well as for first-degree relatives, and second-degree relatives of breast cancer patients. It is noteworthy that in this study the information concerning which family members had cancer was obtained from the Icelandic Cancer Registry, whereas genealogical trees were constructed by using information from records of the genetics committee of the University of Iceland, thus avoiding possible bias generated by directly asking the family members about the structure and cancer cases in their families. Similarly, Anderson and Badzioch (17) found that a family history of prostate cancer in male breast cancer patients resulted in a 4-fold increased breast cancer risk in first-degree female relatives compared with that in male breast cancer families with no history of prostate cancer. By contrast, a family history of lung cancer, colon cancer, or melanoma had no effect on increasing risk of breast cancer. Finally, a series of recent studies concerning the putative familial clustering of breast and prostate cancer have provided opposite results. Thus, Isaacs et al. (18) in a study of families selected because of the presence of prostate cancer did not find increased risks for cancer at other sites, such as breast, ovary, or endometrium. Similarly, Negri et al. (19) did not observe an elevated risk of prostate cancer in relatives of breast cancer patients. By contrast, Sellers et al. (20), in a large prospective cohort study of Iowa women, noted that a family history of breast and prostate cancers is a stronger risk factor for postmenopausal breast cancer than is a family history of breast cancer alone. The reasons for the discrepancies between the different epidemiological studies are unclear, although Anderson and Badzioch (21) have pointed out a number of potential explanations, including differences in the study populations, variability in the size of families, or some peculiarity of sampling. It is also possible that methodological aspects could influence the final results, since coaggregations between breast and prostate cancers were specially noted in those studies involving very large pedigrees in which only those relatives with medically documented tumors were considered eligible for the study.

Therefore, it seems clear that not all data on the potential association between breast cancer in females and prostate cancer in males are univocal. However, a number of studies performed by different groups in populations of different geographic origin appear to indicate that a family history of breast cancer may have a significant influence on prostate cancer risk and vice versa. This observed association between breast cancer and prostate cancer suggests that, at least in some cases, both tumors may share common factors, either genetic or epigenetic, that could finally lead to the development and progression of these malignancies. Among the different factors that can be shared by breast and prostate cancers, three of them deserve special attention. First, and considering that both carcinomas arise in hormonally regulated tissues, it is conceivable that common hormone alterations could play a role in the development or progression of both tumors. On the other hand, and since the above mentioned studies suggested a familial coaggregation of breast and prostate cancer in different populations from different origins, it seems clear that in addition to being hormonally related, these tumors may also share some genetic abnormalities that could contribute to the acquisition of the malignant phenotype by both mammary and prostatic epithelial cells. Finally, it is also possible that the coaggregation of these highly prevalent tumors may be also influenced by a number of lifestyle and environmental factors, including dietary factors, whose importance in the development of human cancer is becoming increasingly apparent.

	III. Incidence of Breast and Prostate Cancer in Different Countries: Dietary Factors

Top Abstract I. Introduction II. Epidemiological Evidence... III. Incidence of Breast... IV. Genetic Abnormalities Common... V. Common Biochemical Features... VI. Growth Factors in... VII. Theories of Breast... VIII. Conclusions References

 
Another epidemiological element linking breast and prostate cancer is the incidence rate of these two cancers among different countries. Prentice and Sheppard published age-adjusted cancer incidence rates of males and females of ages 55-69 during years 1978-1982, in 21 countries with reputations for accurate cancer registries (22). We have plotted these data in simple linear regression formats to examine whether there is any correlation of the incidences of various cancers with those of breast and prostate cancer. The observed correlation coefficients are summarized in Table 1. The highest correlation between cancer incidences was observed between breast and prostate and breast and endometrial cancers (Fig. 1 and data not shown). The lowest incidence rates of both breast and prostate cancers were found in Japan and Hong Kong and the highest incidence rates were found in the USA and Canada. Similar data were obtained for endometrial cancer (not shown). These indirect findings, taken together with migrand studies, which suggest that cancer incidence rates change within two to three generations when low risk populations migrate to countries with high risk, suggest that common environmental factors may be responsible for these cancers.

View larger version (21K): [in this window] [in a new window]   Figure 1. Correlation between breast cancer and prostate cancer incidence rates in 21 countries listed in Table 1. Data are from Ref. 22. Lowest incidence rates for both breast and prostate cancer are found in Japan and Hong Kong and highest incidence rates are found in the United States and Canada.

Among all dietary factors, fat consumption has received the greatest attention (36). The connection between high-fat diet and increased cancer risk is supported by animal studies (37). In humans, breast cancer risk (22, 36) and prostate cancer risk (22, 38, 39) were found to increase with increased fat consumption. Although such associations are consistent between many studies, others question the validity of the data because of the presence of many confounders and the poor accuracy of obtaining food intake data (40, 41). It is expected that the role of dietary fat in the development of breast cancer will be further elucidated when a primary prevention trial among women age 50-79 is complete (1, 10). The Women’s Health Initiative is a randomized, placebo-controlled trial with three different interventions, one of which is dietary, aiming to reduce fat intake to 20% of total calories (from about 40% currently) and to increase intake of fruits and vegetables. In the same trial, another intervention includes vitamin D and calcium supplements (1). Other chemoprevention trials are underway in many countries (10). Prentice and Sheppard calculated, based on fat disappearance data, that a 50% reduction in fat consumption may reduce the relative risk of women of age 55-69 yr for breast cancer from 1.00 to 0.39 and in men for prostate cancer from 1.00 to 0.17. Such benefits, they postulate, may also be seen for endometrial, colon, rectal, and ovarian cancer (22). The biological basis of fat consumption and risk for breast and prostate cancer is not known, but there is evidence that fat intake alters steroid hormone concentration in serum. For example, it has been reported that plasma estradiol levels are reduced in postmenopausal women on low-fat diets (42). Also, there is evidence that a low-fat diet may reduce testosterone levels in adulthood (43) or may modify 5-reductase activity (39).

The association between breast cancer and fat consumption has recently been reviewed, and it was concluded that, in the absence of data from dietary intervention trials, the weight of available evidence suggests that the type and amount of fat in the diet is related to postmenopausal breast cancer (44). The associations between diet and breast and prostate cancer are also evident from migrant studies. Migrant groups usually adopt dietary patterns similar to those of their new country within a few years after migration. Statistical analysis has shown that dietary fat alone can provide an explanation for the major changes in cancer risk that followed Japanese migration to the United States. For example, Tominaga (45) reported relative risks (RR) of 3.5 and 5.7 for breast and prostate cancer, respectively, in Japanese migrants to the United States. The calculated higher risks, based on changes in fat consumption alone, are 2.9 and 7.2, respectively, in close agreement with the observed risks.

The current epidemiological data suggest that the epidemic of breast and prostate cancer may be partially attributable to increased fat consumption, increased caloric intake during growth, low fiber, vegetable, and fruit consumption, and other lifestyle factors including exercise, alcohol, and smoking (22, 41, 43). Refinements in our knowledge regarding fat consumption and its connection to cancer suggest that specific fatty acids (e.g., the n-6 polyunsaturated fatty acids) may be more potent tumor enhancers than other unsaturated or saturated fatty acids (46-49). Hopefully, the studies that are now underway will provide us with more insights that will be useful in designing successful prevention strategies.

	IV. Genetic Abnormalities Common to Breast and Prostate Cancer

Top Abstract I. Introduction II. Epidemiological Evidence... III. Incidence of Breast... IV. Genetic Abnormalities Common... V. Common Biochemical Features... VI. Growth Factors in... VII. Theories of Breast... VIII. Conclusions References

 
The epidemiological findings showing a potential association between breast and prostate cancers have prompted studies directed to search the putative molecular factors common to these two highly prevalent tumors. Similar to other tumors, a large number of factors, including oncogenes, tumor suppressor genes, or hormonal receptors, may be altered in breast and prostate carcinomas. In fact, acquired or inherited abnormalities in a wide variety of genes have been implicated in the pathogenesis of these tumors (reviewed in Refs. 50 -53). However, it should be emphasized that most of these genetic abnormalities, including those recently described in the PTEN/MMAC1 gene (54-56), are not exclusive of breast and prostate carcinomas and represent alterations in oncogenes or tumor suppressor genes commonly mutated in human tumors from different origins. Thus, until more data become available, our presentation on this issue should be regarded at present as speculative. Nevertheless, mutational studies on some genes, including the AR gene and those involved in hereditary breast cancer (BRCA1,BRCA2), have provided some results that may be of relevance in the context of putative genetic abnormalities common to breast and prostate cancer. The interest in AR as a potential factor common to both tumors arises from recent observations indicating that genetic abnormalities in this hormonal receptor are shared by these two hormonally dependent tumors, but admittedly in only a small proportion of patients (53-62). Similarly, genetic epidemiological data have suggested that some cases of prostate cancer could be linked to the recently described breast cancer-associated genes BRCA1 (63-68) and BRCA2 (67-74). Finally, and since it seems clear that these genes are not the only molecular factors that may be common to breast and prostate cancers, in the last part of this section, we examine other candidate genes that could contribute to establish associations between these two highly prevalent malignancies.

A. AR alterations in prostate cancer
The AR is a transcription factor that plays an essential role in a wide number of biological functions, from development and maintenance of male reproductive functions to modulation of immune responses or development of neural tissues (75). Like other nuclear receptors, AR exerts its biological effects after binding of circulating androgens mainly transported to target tissues by carrier proteins (76). Androgen binding induces a conformational change in the AR that facilitates receptor homodimerization, nuclear transport, and interaction with DNA. The binding of the AR to the hormone response elements (HRE) present in target genes results in the regulation of their transcriptional activity (77). The structure of the AR is also similar to that of the other members of the steroid-receptor family of ligand-dependent transcription factors, with an N-terminal transactivating domain (exon A), a central DNA-binding domain (exons B and C), and a C-terminal hormone-binding domain (exons D through H) (78).

Because of the essential participation of AR in the regulation of prostate growth and in the maintenance of prostatic function, over the last years many groups have tried to define the potential role of this hormone receptor in the development and progression of prostate cancer. The first studies in this regard were based on analysis of the AR functionality in prostate carcinomas by using ligand-binding activity assays and immunohistochemical techniques (79, 80). However, results of a series of structure-function relationship studies of mutated ARs have revealed that ligand binding or immunoreactivity are not the most appropriate indicators of AR functionality. Thus, investigators have described the occurrence of mutant ARs that do not bind androgens but are constitutively active, or receptors that bind steroids with high affinity but are nonfunctional as transcription factors (81, 82). As a consequence of these observations, more recent studies have examined the possibility that alterations in the integrity of the AR gene in prostate carcinomas could be a more accurate index of the AR functionality in these tumors (83) (Fig. 2). The first indication that structural changes in the AR could be important in the progression of prostate cancer was provided by the detection in LNCaP prostate cancer cells of a point mutation in the ligand-binding domain of this receptor (84). Interestingly, this mutation (Thr877Ala) leads to a decrease in steroid-binding specificity and completely reverses the effect of commonly used antiandrogens (84). After these findings in established cancer cell lines, several groups have attempted to demonstrate the putative occurrence of AR gene mutations also in tumor tissue specimens. The first description of an AR abnormality in human prostate cancer was done by Newmark et al. in 1992 (85). These authors found a point mutation (Val730 Met) in 1 of 26 early-stage prostatic carcinomas. Thereafter, other groups have reported that AR mutations may also occur in a small percentage of advanced cancers (86-92). By contrast, Ruizeveld de Winter et al. (93) did not detect mutations in AR genes from 18 patients with hormone-resistant, locally progressive prostate cancer. Although these studies appear to indicate that the frequency of AR mutations is low, even in advanced prostate cancer, recent work using improved strategies for mutational analysis of AR have found a higher proportion of genetic abnormalities in either latent prostatic carcinomas or in metastatic disease. Thus, Takahashi et al. (94) have found that a significant proportion of latent prostate carcinomas from Japanese men contain genetic alterations in the AR gene (18 of 79), while no such mutations were found in 43 latent carcinomas from American men. On the other hand, Gaddipati et al. (95) have shown the presence of the above described mutant Thr877Ala, in 6 of 24 prostatic tissue specimens obtained from patients with metastatic prostate cancer, providing the first evidence that a mutational hotspot may occur in the AR gene in a subset of these tumors. More recently, Taplin et al. (96) have shown the presence of AR gene mutations in metastatic cells from 5 of 10 patients with androgen-independent prostate cancer, which has led them to conclude that mutations in this gene are not as rare as previously considered by other authors. Consistent with this, Tilley et al. (97) have found somatic mutations in 44% of primary prostate tumors taken before initiation of androgen ablation therapy. The presence of AR amino acid substitutions was found not only in the hormone-binding domain, which is the region examined in most studies mentioned above, but also in the remaining functional domains of this protein. In fact, about 50% of the mutations found by Tilley et al. in prostatic tumors were within exon A of the AR, which encompasses 58% of the coding region of the gene, but whose integrity has not been examined in virtually any mutational study of the AR gene. These results demonstrate the need to examine the complete AR-coding region before any conclusion on the structural integrity of the AR gene in prostate carcinomas can be reached. It is also remarkable that Tilley et al. (97) have provided evidence that mutations found in AR are not a consequence of the generalized genetic instability inherent to different malignant processes, suggesting that they have functional relevance and do not simply reflect the neoplastic state. In fact, these authors have observed that the occurrence of the AR mutations in the studied prostatic carcinomas was associated with a rapid failure of subsequent hormonal therapies. Therefore, it seems that AR gene mutations may occur commonly in advanced prostate cancers before endocrine treatment, thereby contributing to the observed altered androgen responsiveness of these tumors, and finally leading to their progression to androgen independence. Finally, two germline point mutations in the 5'-untranslated region of the AR gene have been recently described in men with prostate cancer. It has been proposed that these mutations may contribute to the disease by altering rates of transcription and/or translation of this gene (98).

View larger version (16K): [in this window] [in a new window]   Figure 2. AR abnormalities identified in prostate and breast carcinomas. The eight exons (A to H) of the AR gene and the three functional domains (transactivation, DNA binding, and ligand binding) identified in the AR protein are represented in the figure. Alterations described in breast carcinomas including point mutations at residues Arg 607 and Arg 608 and deletion of exon C (del exC) are boxed. AR gene abnormalities identified in prostate carcinomas include point mutations, contraction of CAG trinucleotide repeats in exon A (Q24Y18), frameshifts (L547fr, P554fr, G743fr), and generation of premature stop codons (W741st, W751st, W796st).

In conclusion, it appears that AR gene alterations in prostate carcinomas and in metastatic tissue derived from these tumors are much more frequent than originally suggested. These genetic defects include point mutations, gene amplification, or variations in the length of trinucleotide repeats. At present, there is a lack of compelling evidence associating receptor variants with response to endocrine therapy, or clinical course of the disease. However, functional analysis of AR variants appear to indicate that they confer upon this receptor a broadening of ligand specificity, making it capable of activation by estrogens, progesterone, antiandrogens, and adrenal androgens in addition to testicular androgens (57, 84, 86, 90-92, 96, 97, 110). This is in marked contrast with the findings in other diseases involving abnormalities in the AR gene, such as androgen insensitivity syndromes or diseases generated by trinucleotide expansion in the CAG region of the AR, which are usually accompanied by loss of AR function (76). As a consequence of these gain-of-function mutations detected in prostatic carcinomas, tumor cells may proliferate in an androgen-deficient environment or during antiandrogen therapy. Therefore, these findings may explain why these carcinomas become refractory to endocrine therapy and could lead to the development of more effective hormonal therapeutic strategies as well as predictive tests for therapy failure.

B. AR alterations in breast cancer
Since, according to the above mentioned data, there was a significant body of epidemiological evidence suggesting an association between breast and prostatic cancer, it seemed likely that if AR gene alterations are important in the development of prostate cancer, similar abnormalities could also occur in some cases of breast cancer. The first indication that AR may also be altered in breast carcinomas was provided by Wooster et al. (58) who reported an AR germline mutation in two brothers with breast cancer and Reifenstein syndrome, a partial androgen insensitivity syndrome originally described as an X-linked familial syndrome of hypospadias, infertility, and gynecomastia in association with normal 17-ketosteroid excretion and high FSH levels (76). The mutation results in the substitution Arg607Gln, within the region encoding the DNA-binding domain of the receptor. More recently, Lobaccaro et al. (59, 60) identified another germline mutation in the AR gene, in a man with lobular carcinoma of the breast and partial androgen insensitivity syndrome. This mutation leads to an Arg608Lys substitution, also in the DNA-binding domain of the receptor and is identical to an alteration previously described in a patient with partial androgen insensitivity syndrome (111). Two main hypotheses have been proposed to explain breast cancer development linked to AR mutations (58-60). The first one involves the loss of a putative protective effect of androgens, which could explain the low incidence of breast carcinomas in males. However, in light of the above observations in prostate carcinomas, it is also possible that the development of breast carcinomas associated with AR mutations could be a consequence of the acquisition of additional properties by the mutated AR proteins. Thus, they could have an altered pattern of hormone responsiveness, including the acquisition of the ability to bind ligands other than testicular androgens, thereby extending their transactivating properties. In any case, elucidation of the potential role of AR in the development of some breast carcinomas will require the identification of additional cases and functional studies with the identified mutant receptors. Finally, it would also be of interest to look for the presence of somatic AR gene alterations in sporadic cases of breast carcinomas in both males and females. In this regard, Hiort et al. (112) have recently reported the absence of mutations in exons 2-8 of the AR gene in breast carcinomas from 11 males without clinical evidence of androgen insensitivity, suggesting that AR gene mutations do not play a major role in the development of sporadic male breast cancer. By contrast, one should note the recent identification of an exon 3-deleted splicing variant AR in breast cancer cell lines and tissues (62). This AR variant is expressed at high levels in some breast carcinomas (7 of 31), whereas in normal breast tissues its expression is undetectable. Also, recent immunohistochemical analysis in breast tumor specimens has suggested that structurally altered forms of the receptor, including amino-terminal truncated variants, may be present in a significant proportion of breast carcinomas (61).

In summary, a series of recent studies performed by different groups has revealed that inherited and acquired AR alterations may occur in breast carcinomas (58-62). The number of described AR abnormalities in breast cancer is low, suggesting that these would only affect a small subgroup of patients. However, it is also remarkable that this field has been largely unexplored and few studies have specifically addressed the role of AR mutations in breast cancer. Nevertheless, these preliminary mutational data, together with the finding of abnormalities in androgen levels in patients with breast cancer (113-115) and the widespread expression of AR in primary breast carcinomas (61, 116, 117), suggest that AR-mediated pathways may be of biological and clinical relevance in breast cancer. Further studies and functional characterization of AR variants in breast carcinomas, in a similar fashion to studies performed in prostatic carcinomas, will be required to clarify the putative contribution of AR to breast cancer cell growth and response or resistance to hormonal therapies.

C. BRCA1 and BRCA2 alterations in breast cancer
Evidence for a genetic component in breast cancer risk was first noted by Paul Broca more than one century ago, when he described four generations of breast cancer in his wife’s family (118). Since then, extensive epidemiological analyses of breast cancer cases that appear to be clustered in families have been reported. The results of these analyses suggest that about 5% of breast carcinomas may be explained by inherited mutations in one or more genes. Despite the genetic heterogeneity of breast cancer and the high prevalence of sporadic disease, several breast cancer susceptibility loci have been identified (119). The first of these genes, named BRCA1, was mapped in 1990 to chromosome 17q21 by genetic linkage analysis of large families that included many cases of early-onset breast carcinomas (120) and has been recently identified by Miki et al. (121) using positional cloning methods. BRCA1 is composed of 22 coding exons distributed over more than 100 kb of genomic DNA and encodes a 1863-amino acid protein, with two RING finger domains at its N-terminal part, which are thought to be involved in DNA-binding or in protein-protein interactions. In addition, BRCA1 shares a conserved region with 53bp1 (a p53-binding protein) and rad9 (a yeast protein involved in the control of the DNA damage-induced cell cycle arrest), which has suggested that BRCA1 is likely to function in the cell nucleus and may be involved in one or more cell cycle checkpoints (122). In marked contrast with this proposal, it has also been suggested that BRCA1 may play a role as a secreted protein, exhibiting properties of a granin (123). To date, the function of BRCA1remains unclear, although a recent study has shown that this protein inhibits the growth of breast epithelial cells (124). In addition, studies on the developmental pattern of BRCA1 expression in mice suggest that it is involved in the process of proliferation and differentiation in multiple tissues, notably in the mammary gland in response to ovarian hormones (125). Furthermore, analysis of BRCA1^-/- mutant mice has suggested that this protein may be a positive regulator of the cellular proliferative processes that occur in early embryonic development (126). On the other hand, Chapman and Verma (127) have recently reported that the carboxy-terminal fragment of BRCA1 acts as a strong transcriptional activator when fused to the GAL4 DNA-binding domain. In addition, this activity is completely abolished in sequences corresponding to four different mutations found in BRCA1-linked families, thus providing direct evidence for the possible function of BRCA1 as a transcription factor. Finally, a new insight into BRCA1 function has has been provided by the observation that it associates with the DNA-repair protein Rad51, suggesting that BRCA1 may be a component of the double-strand-break DNA repair pathway (128-130).

Mutations in the BRCA1 gene are thought to account for about half of the families susceptible to early-onset breast cancer and for at least 80% of families with clustered breast and ovarian cancers (131, 132). To date, germline BRCA1 mutations have been reported in more than 200 families from different geographic origins (131, 132). Germline BRCA1 mutations have also been found in young women with breast cancer who do not belong to families with multiple affected members (133). All classes of mutations are represented in these reported cases, including missense mutations, nonsense mutations, deletions, insertions, or intronic mutations, although the majority result in the production of a truncated protein. The finding of this large percentage of loss-of-function mutations is consistent with the hypothesis that BRCA1 acts as a tumor suppressor gene. It is also remarkable that most of the reported BRCA1 gene mutations have been identified in a single family, but a small number have been detected repeatedly. Of particular interest is a frameshift mutation caused by deletion of an AG dinucleotide (185delAG), which has been identified in more than 20 families of Ashkenazi Jewish descents and is estimated to occur at a frequency of about 1% in this population (134, 135).

The observation that less than half the families with multiple cases of breast cancer showed linkage to BRCA1 led to the proposal that there was at least an additional gene associated with breast cancer susceptibility. This result prompted another genomic linkage search and a second breast cancer susceptibility gene, named BRCA2, was located on chromosome 13q12 (136) and subsequently cloned (69, 137). BRCA2 is composed of 27 exons and encodes a protein of 3418-amino acid residues, which does not appear to be significantly similar to other proteins. Recent studies have shown that BRCA2 expression is coordinately regulated with BRCA1 expression during proliferation and differentiation in mammary epithelial cells, suggesting that both genes may act in the same pathway (138). Similarly to BRCA1, BRCA2 interacts with Rad51, providing additional support to the proposal that these proteins may be essential cofactors in the Rad51-mediated DNA repair of double-strand breaks (139). In fact, Connor et al. (140) have found evidence of a DNA repair defect in mice with a truncating BRCA2 mutation. Clinical studies have revealed that BRCA2 probably accounts for a proportion of early-onset breast cancer roughly equal to that resulting from BRCA1, and it may be of special importance in families with a high incidence of male breast cancer, but not in those with multiple cases of ovarian cancer. Mutational analysis of the BRCA2 gene in different populations has revealed that as in BRCA1, the identified mutations are widely distributed throughout the coding sequence of the gene, although evidence of some recurrent mutations has also been found (71, 141-144). Also of interest is the finding that BRCA2 mutations in families with the highest risk of ovarian cancer relative to breast cancer are clustered in a single exon of this gene (145). Finally, and also in common with BRCA1, diverse studies have shown that BRCA2 is a very infrequent target for somatic inactivation in breast and ovarian cancers (144-148).

D. BRCA1 and BRCA2 alterations in prostate cancer
As mentioned above, genetic epidemiological studies have provided evidence for clustering of prostate and breast cancer in some families. In addition, there is preliminary evidence that some plausible prostate cancer genes, like AR, may be altered in some breast tumors. Therefore, it seemed of interest to evaluate the possibility that genetic abnormalities in breast cancer susceptibility genes, such as BRCA1 and BRCA2, may also be associated with an increased risk of prostate cancer in men. The first of these studies was performed by Arason et al. (63) in seven large Icelandic breast cancer families, two of which showed evidence of linkage to BRCA1. These authors found that among presumed paternal carriers of mutant breast cancer gene alleles, 7 of 16 (44%) had developed prostate cancer, which led them to conclude that breast cancer genes may predispose to prostate cancer in male carriers (63). Additional evidence regarding the potential associations between BRCA1 and prostate cancer risk comes from an analysis of 33 BRCA1-linked families performed by Ford et al. (64). This analysis attempted to explore whether BRCA1 gene carriers are at increased risk of cancer at sites other than breast or ovary. According to the obtained results, there were statistically significant excesses of prostate cancer and colon cancer in BRCA1 carriers but not of cancer at any other sites. The maximum likelihood estimate of the relative risk of prostate cancer in BRCA1 carriers compared with the general population was 3.33. More recently, Gao et al. (65) in a study designed to establish the possible involvement in prostate cancer of BRCA1 and other potential tumor suppressor genes on chromosome 17q, have reported a high frequency of loss of heterozygosity at loci D17S856 and D17S855 (intragenic to BRCA1) in prostate cancer. These results suggest that BRCA1 and possibly other genes located within this region (149) may be important in this cancer.

Although these studies seemed to confirm the hypothesis that some connection could exist between breast cancer susceptibility genes and prostate cancer, very recent work performed by Langston et al. (66) has provided more definitive evidence. These authors, in a study aimed at directly examining the potential role of BRCA1 mutations in the etiology of prostate cancer, have screened for germ-line BRCA1 mutations in a subset of men with prostate cancer. The subgroup of cases selected included men in whom genetic factors were most likely to be relevant, including early-onset and family history of both breast cancer and prostate cancer. Interestingly, a total of seven germ-line alterations in a series of 49 cases were found. One of them corresponded to the above mentioned frameshift mutation (185delAG), which is the most common germ-line BRCA1 mutation reported to date (134, 135). In addition, five structural abnormalities were identified in six patients but not in the 145 population-based controls. One of them is a 12-bp insertion in intron 20, which was identified in two different cases, and which had previously been found in a woman diagnosed with cervical cancer and breast cancer (133) and also in a woman with a history of breast and ovarian cancer (150). Although the functional consequences of this genetic alteration are unknown, it seems likely that this 12-bp insertion may affect RNA processing. The remaining four sequence variants have not been reported previously and are located in both coding and noncoding sequences. The fact that none of the sequence variants was identified in DNA from the control population suggests that they may represent alleles predisposing to disease. Finally, Struewing et al. (151) have also detected a BRCA1 frameshift mutant (5256delG) in a male patient affected with both breast and prostate cancer.

In addition to these genetic alterations in the first breast cancer susceptibility gene, studies of families linked to BRCA2 have revealed that prostate cancer risk is significantly increased in these families (67-74). Further analysis of some of these families has shown that in three of four BRCA2-linked Icelandic families, all prostate cancers tested are carriers of a 5-bp deletion in exon 9 (999del5), which is a recurrent mutation in Icelandic patients (71). Interestingly, prostate cancer patients carrying this mutation have significantly worse survival, which suggests that the BRCA2 mutation may be a possible marker for an aggressive disease in prostate cancer patients (73). Taken together, these data appear to indicate that mutations in the BRCA2 gene may also confer some risk of developing other malignancies, including prostate cancer, although detailed BRCA2 mutational studies in prostate carcinomas need to be done before more definitive conclusions can be reached.

Although it is clear that the basis for the hypothesis of common genetic features between some breast and prostate cancers is still speculative, two recently published studies have provided new and interesting insights. Struewing et al. (67), in an extensive study of the risk of cancer in a large group of Ashkenazi Jews, found a significantly elevated estimated risk of prostate cancer among carriers of BRCA1 or BRCA2 mutations. According to these data, the authors suggest that prostate cancer is part of the phenotype for these carriers. Similarly, Khan et al. (68), after analysis of germline BRCA1 and BRCA2 mutations in prostate carcinomas from a different population of Ashkenazi Jews, have concluded that mutations in these breast cancer susceptibility genes may increase the risk of prostate cancer.

In summary, there are some data indicating that alterations in the structural integrity of breast cancer susceptibility genes may indeed occur in prostate carcinomas. Nevertheless, according to available information, it appears that the contribution of germline BRCA1 or BRCA2 mutations to the overall incidence of prostate cancer is very small. In addition, the genetic association between breast and prostate cancer, due to BRCA1 and BRCA2, seems somewhat diluted by the fact that mutations in these genes also play a role in other tumors, including ovarian (150-153) and pancreatic carcinomas (154). Further studies and identification of additional prostate cancer patients with genetic alterations of BRCA1 and BRCA2 will be necessary to clarify the putative involvement of these genes in at least some cases of prostate cancer.

E. Other genes associated with breast or prostate cancer
In addition to the above described alterations in AR and BRCA genes, acquired or inherited abnormalities in other genes may occur in breast and prostate cancer. Analysis of reported alterations in oncogenes and tumor suppressor genes in both breast and prostate carcinomas reveals that somatic abnormalities are heterogeneous in terms of involved genes and mechanisms operating for their generation (reviewed in Ref. 50 -53). The class of genes that is altered during the progression of normal mammary or prostatic cells to hormone-independent or to highly aggressive metastatic cancer cells includes classic tumor suppressor genes (p53,RB1) and oncogenes (ras, myc, neu) (50-53, 155-159), as well as genes involved in other processes such as cell-cycle inhibition, cell-cell adhesion, angiogenesis, DNA repair, and apoptosis (160-167). The mechanisms underlying these alterations are also diverse and include point mutations, allelic deletions, high-level amplifications, or de novo DNA methylation (50-53, 155-160, 168-173). This heterogeneity is consistent with the idea, as originally proposed for colorectal cancer (174), that breast and prostate carcinomas result from the accumulation of genetic changes affecting a variety of genes associated with critical cell functions. However, it must be emphasized again that most of these genetic abnormalities are not exclusive to these tumors and have lesser value in the context of this review, which attempts to bring together factors common preferentially to breast and prostate cancer. Nevertheless, it is clear that new oncogenes and tumor suppressor genes important in the pathogenesis of these tumors are yet to be identified. In this regard, it is noteworthy that very recent studies from different groups have led to the identification in breast or prostate carcinomas of a series of candidate oncogenes, tumor suppressor, or metastasis suppressor genes, including H-cadherin (175), maspin (176), MDC (177), PCTA1 (178), PTI1 (179), MXI1 (180), PAC1 (181), KAI1 (182), and thymosin ß15 (183), whose relevance to the respective tumor processes has not as yet been definitively established. In this context, it will be interesting to explore the possibility that alterations in some of the new candidate genes associated with breast cancer may be also found to occur in prostate carcinomas and vice versa, thus helping to extend the genetic associations between these two hormone-sensitive tumors. It is also remarkable that the vast majority of genetic changes reported in both breast and prostate carcinomas arise in somatic cells but inherited defects may also predispose to both cancers. Interestingly, studies of familial aggregation in both diseases have revealed that the same percentage of breast or prostate cancers (5%) may be directly attributable to inherited cancer susceptibility alleles (50-53). Familial breast and prostate cancer genes have now been mapped and, in the case of breast cancer-susceptibility genes, some associations with prostate cancer have been reported (63-74). Therefore, it will also be of future interest to evaluate the possibility that alterations in the familial prostate cancer gene (HPC1), recently mapped to the long arm of chromosome 1 (184), may also occur in a subset of breast carcinomas. Of interest is the preliminary report of a modest increase in the occurrence of breast cancer in HPC1 families (185). Further studies directed to examine the putative genetic commonalities between breast and prostate cancers could provide better insights into the mechanisms of progression of these hormonally dependent tumors and generate novel ideas to improve therapeutic strategies.

	V. Common Biochemical Features of Breast and Prostate Cancer

Top Abstract I. Introduction II. Epidemiological Evidence... III. Incidence of Breast... IV. Genetic Abnormalities Common... V. Common Biochemical Features... VI. Growth Factors in... VII. Theories of Breast... VIII. Conclusions References

 
The accumulation of the above mentioned genetic lesions in mammary or prostatic epithelial cells could lead to uncontrolled cell proliferation, disruption of normal pathways of cell differentiation, hormone responsiveness or programmed cell death, and, ultimately, promotion of mechanisms that facilitate tumor invasion and metastasis. These functional alterations may be connected to the biosynthesis of specific proteins that could be very useful as biochemical markers of the respective tumor processes. In recent years, molecular and biochemical analyses of breast or prostate carcinomas have led to the identification of a number of proteins that could be useful for predicting the clinical course of these diseases or monitoring their response to hormonal therapy. Among the growing list of tumor markers of potential interest in these malignancies, we have noticed that some of them, including prostate-specific antigen (PSA), pepsinogen C, apolipoprotein D, Zn-₂-glycoprotein (Zn-₂-gp), and GCDFP-15, show a striking parallel expression in both breast and prostate cancers. Importantly, such expression is either very low or absent in other tumors. Thus, all of them are up-regulated or down-regulated in a significant percentage of tumors of both sites and in most cases, their production appears to be dependent of common regulatory hormonal mechanisms. This section summarizes the current evidence in the literature supporting our proposal that these five proteins may represent examples of biochemical similarities between breast and prostate cancer.

A. PSA
PSA was initially discovered in seminal plasma in the 1970s (186, 187). Purification was first achieved by Sensabaugh (188). PSA was found to be a prostatic protein in 1977 (189) and was identified in serum shortly afterward. Of paramount clinical importance were the findings that serum PSA is increased in patients with prostate cancer in comparison to normals and that changes of serum PSA concentration are associated with cancer metastasis, recurrence, response to treatment, and survival (190, 191). Currently, PSA is considered to be the most valuable tumor marker due to its tissue specificity and it is used widely for prostate cancer screening, diagnosis, and management. Several reviews examine these issues in detail (192-196).

PSA is a 30-kDa serine protease that shares significant protein and gene sequence homology with pancreatic/renal kallikrein (hK1) and glandular kallikrein (hK2). PSA is also known as hK3. The PSA gene has been extensively characterized (197). The 5'-untranslated region contains regulatory elements, two of which are androgen response elements (ARE I and ARE II), and the other is a strong enhancer (198, 199). PSA gene transcription in the prostate is known to be regulated by androgens through the action of the AR (197-200) (Table 2). In seminal plasma, in which PSA is present at very high amounts (1-2 g/liter), it appears that the role of PSA is proteolytic cleavage of the sperm motility inhibitor semenogelin, resulting in semen liquefaction post ejaculation (193, 194, 201). However, other substrates for PSA have been proposed including insulin-like growth factor binding protein 3 (IGFBP-3) (202), protein C inhibitor (203), transforming growth factor-ß (TGF-ß) (204), PTH-related peptide (205), and an unknown precursor protein that releases a putative vasoactive peptide (206). In male serum, PSA is present as a complex with ₁-antichymotrypsin (PSA-ACT), ₂-macroglobulin (PSA-A2M), and as free PSA (207, 208).

PSA has also been detected in the milk of lactating women (228) or women with prolactinoma (our unpublished data), in breast cyst fluid (229), and amniotic fluid (230). In female serum, PSA is present at levels approximately 1,000-fold lower than male serum (231). We failed to find any association between total serum PSA and clinicopathological features of breast cancer (232). However, PSA in serum increases during pregnancy (233). More recently, we were able to determine the molecular forms of PSA in female serum and concluded that: 1) in serum of normal women or women without breast pathology (e.g., hirsute women), the predominant form is always PSA-bound to 1-antichymotrypsin (PSA-ACT); 2) in presurgical sera from breast cancer patients, about half of them have free PSA as the major molecular form. Similar data were found for women with benign breast diseases (Ref. 234 and unpublished results). These data suggest that serum-free PSA, which is an enzymatically inactive form of PSA, is overexpressed in patients with benign or malignant breast disease. The mechanism of such changes is unknown but the data are in contrast to changes in prostate cancer where serum PSA-ACT increases and free PSA decreases in cancer patients in comparison to patients with benign prostatic hyperplasia (207). Finally, we identified some similarities between PSA expression in the breast and expression of the BRCA1 protein, which is believed by some to be a granin (123). Thus, we speculated that BRCA1 may be a substrate for PSA but as yet there is no experimental evidence for this proposal (235). Furthermore, a protein that appears to be immunologically identical to BRCA1 has been found in seminal plasma (236).

How is PSA regulated in the breast and in breast tumors? We have evidence that PSA is up-regulated by progestins in vivo (223) and in vitro (225, 226). Similar data exist for glucocorticoids (225, 237). In vitro, androgens up-regulate PSA at levels as low as 10^-11 M, similarly to progestins (226). We have also generated evidence that PSA up-regulation by androgens occurs in vivo because women with hyperandrogenic states have higher PSA than normal controls (238). Other evidence suggests that serum PSA changes during the menstrual cycle (239). The observation that some breast tumors bearing steroid hormone receptors do not produce PSA, while others that are receptor negative may produce high levels of PSA, led us to examine the sequence of all PSA exons and the 5'-regulatory region of the PSA gene in such tumors. No mutations were identified in any of the PSA exons, but we found deletions and point mutations in the 5'-flanking region in all of these tumors (240). This finding suggests that PSA expression is aberrant in at least some breast tumors.

What is the physiological role of PSA in the breast? This is currently not known but based on the proteolytic activity of PSA, we speculate that this enzyme, regulated by steroid hormones in the female breast, must act upon a substrate to release other biologically active molecules. Others have already proposed that PSA may be a regulator of growth factors, cytokines, or PTH-related peptide, but the levels of PSA tested are much higher than those found in the breast (202, 204-206). In this regard, it is of interest that breast cancer cells secrete an IGFBP-3 protease with ability to release bound insulin-like growth factor-I (IGF-I), which can then act as a mitogen to stimulate breast cancer cell proliferation (241). Since IGFBP-3 is a substrate for PSA in seminal plasma (202), a similar role for PSA in breast carcinomas could be envisaged, although no data are available to support this hypothesis. On the other hand, the sequence homology of PSA to growth factors and growth factor-binding proteins suggests that this molecule may well be a growth factor in its own right (242). Also interesting is the proposal that PSA may act upon substrates to release vasoactive peptides, which could help in the expulsion of breast secretions, such as nipple aspirate fluid and milk, paralleling the semen liquefaction function of PSA in the prostate (206). Whatever the function of PSA is, the current evidence suggests that this molecule is a marker of differentiation and good prognosis in breast diseases, especially breast cancer. It is now very clear that this molecule, which wrongfully bears the name of a prostate-specific protein, is elegantly regulated by steroid hormones and is secreted at relatively high concentrations by breast epithelial cells. Notably, only prostate cells in males and breast cells in females produce appreciable amounts of PSA, the levels in other tumors being much lower (243).

B. Apolipoprotein D (apoD)
apoD is a protein component of the human plasma lipid transport system that was first identified and characterized by McConathy and Alaupovic (244). This glycoprotein is mainly associated with high-density lipoprotein particles and consists of a single polypeptide chain of about 30 kDa that exhibits sequence similarity to members of the lipocalin family of proteins, whose common function is to bind and transport small hydrophobic ligands in the plasma (245) (Table 2). The functional role of apoD in the metabolism of plasma lipoproteins remains elusive, but it has been proposed that it may be involved in transport of cholesterol or cholesteryl esters (246-248). In addition, recent studies from different groups indicate that apoD is able to bind and transport a wide variety of ligands other than cholesterol, including heme-related compounds (249), progesterone (250), arachidonic acid (251), or odorant substances (252), thus extending its potential functional significance to a number of different biological processes.

The unexpected connection between apoD and breast diseases arose after the observation that apoD accumulates to extremely high concentrations (1000-fold higher than in plasma) in cyst fluid from women with gross cystic disease of the breast (250), a benign condition associated with an increased risk of subsequent breast cancer (253, 254). The relationship of apoD to breast pathology was further supported by the finding of a certain type of breast carcinoma that is able to produce and secrete this glycoprotein (255-257). Analysis of putative correlations between apoD levels in breast carcinomas and clinical outcome of the disease has revealed that low apoD values are significantly associated with a shorter relapse-free and overall survival (257). A possible explanation as to why apoD confers a prognostic advantage to women with breast cancer is that its presence may reflect the existence of a complete hormone receptor pathway. To date, the hormonal stimuli potentially responsible for the expression of apoD by breast carcinomas are unclear, but several data suggest that androgens could play a major role in apoD overproduction. Thus, apoD is one of the few proteins that are up-regulated by androgens in human breast cancer cells (257-259). This stimulatory effect is blocked by the antiandrogen flutamide, indicating that the action of androgen is presumably mediated via an AR mechanism. Finally, apoD has been found to be produced by either normal or tumor prostatic cells, under androgen stimulation (260-262).

The first indication that apoD could also be a marker of steroid action in prostate cancer cells was provided by Simard et al. (260), who examined the regulation of apoD secretion by sex steroids in LNCaP cells, the most widely used in vitro model of human prostate cancer. According to their data, physiological concentrations of androgens exert a biphasic pattern of action on both apoD secretion and cell proliferation in LNCaP cells. Thus, low concentration of androgens stimulates proliferation of prostate cancer cells and inhibits apoD secretion, whereas higher concentrations of androgens increase the expression of apoD and inhibit cell proliferation. Interestingly, such an opposite action of sex steroids on apoD secretion and cell proliferation is in complete agreement with similar studies in breast cancer cells demonstrating that the action of androgens and estrogens on apoD secretion is inversely related to cell proliferation in breast cancer cells (258, 263). On the basis of these results, apoD has been proposed as a marker of hormone action in both breast and prostate cancer cells, which could be associated with inhibition of cell growth and tumor regression (262-264). This potential value of apoD as a marker of growth arrest, together with its specific pattern of hormone responsiveness in both breast and prostate cancer cells, may be of interest from the clinical point of view. Thus, quantitation of intratumor apoD values could help to identify subgroups of breast or prostate cancer patients with low or high risk for recurrence or death, and who could benefit from specific hormone therapies.

C. Zn-₂-gp
Zn-₂-gp was originally isolated from human plasma, and its name was derived from its ability to be precipitated by zinc acetate, its electrophoretic mobility in the ₂-region of the plasma globulins, and its high carbohydrate content (265). Amino acid sequence analysis of the protein purified from plasma has revealed that it consists of a single polypeptide chain of 276 amino acids with a high degree of similarity to class I antigens of the major histocompatibility complex (MHC) (266) (Table 2). The isolation and characterization of cDNA and genomic clones for human Zn-₂-gp have provided additional information on the relationship between this protein and transplantation antigens (267-269). Thus, the exon-intron organization and nucleotide sequence of the Zn-₂-gp gene are very similar to those of the first four exons encoding the signal peptide and the three extracellular domains characteristic of all class I MHC molecules. However, the Zn-₂-gp gene lacks the coding information for the transmembrane and cytoplasmic domains present in class I MHC genes, which explains its presence as a soluble protein in several human body fluids (270). The biological function of Zn-₂-gp is unknown but, according to its structural properties, this glycoprotein may play a role in the immune response as a soluble HLA adapted to bind and transport some nonpolymorphic substance in the plasma (271).

The potential interest of Zn-₂-gp in relation to breast cancer has arisen after the observation that, similar to apoD, this soluble HLA-like protein is accumulated at high concentrations in breast cyst fluid from women with gross cystic disease of the breast (255, 256). Furthermore, analysis of breast cancer tissues and secretions has revealed the existence of a significant percentage of mammary tumors (40%) that produce and secrete appreciable amounts of Zn-₂-gp (255, 256, 272-275). Interestingly, and also in agreement with data regarding apoD, higher levels of Zn-₂-gp were detected in histopathologically well differentiated tumors than in moderately or poorly differentiated tumors (273, 275), suggesting that this protein may be a marke