Induction of high incidence of mammary tumour in female Noble rats with a combination of 17β-oestradiol and testosterone


Breast cancer is the most common cancer and the second most frequent cause of cancer death in women. Despite extensive research, the precise mechanisms of breast carcinogenesis remain unclear. One of the reasons for this is due, at least in part, to a lack of a suitable animal model which can closely mimic the breast carcinogenesis in normal situations without using chemical carcinogens. We have developed an animal model of mammary gland carcinogenesis using a combination of oestradiol and testosterone, and succeeded in inducing a high percentage of female Noble rats to develop mammary cancer in a relatively short time (~6 months). The results showed that androgens might work as a promoter to shorten the latency time of mammary gland carcinogenesis. Histopathological examination revealed that hyperplasia and dysplasia were first observed 2 months after treatment, in situ carcinoma after 3 months, and fully developed carcinoma of various forms including cribriform, papillary and camedo types were observed from 5 to 6 months after hormone implantation. Animals implanted with oestrogen or testosterone alone also developed mammary cancers, though with a lower overall incidence than the two hormones combined. They ranged from well differentiated to poorly differentiated forms with predominantly infiltrating ductal carcinoma. We have also observed a case of secondary cancer in the uterus. In addition to the high incidence of carcinoma, there was also a peculiar unexplained ipsilateral correlation between the site of hormonal implantation and the location of tumours, and the highest incidence of carcinogenesis was found to be in thoracic mammary gland. The study showed that both oestrogens and androgens are important in mammary cancer development. The animal model would prove to be a useful model for analysis of the mechanism(s) of hormonal carcinogenesis.



Breast cancer is the most common cancer and the second most frequent cause of cancer death in women (1). The American Cancer Society has estimated that there will be 178 700 new breast cancer cases and ~16% of all cancer related mortality (43 500 cancer related deaths) in 1998. Unlike the common non-hormone-dependent adult cancers, breast cancer is exceptional before the age of 20 years and is rare below 30 years but then the incidence rises very steadily up to the age of 50 years, after which the rate of increase slows down, although the incidence rate continues to rise. This shows that a relationship must exist between ovarian hormones and breast cancer, although it has not been clarified which specific hormones are involved in the breast carcinogenesis (2).

The epidemiological evidence provides strong support for this concept (3). From among the various recognized risk factors, age of menarche, age of first pregnancy and age of menopause suggest that endogenous sex hormones may play the predominant role at all stages in the development of breast cancer (49). Because of their physiologic stimulatory actions on mammary gland, oestrogens, especially oestrone and oestradiol, have long been linked to the risk of breast cancer (10). Epidemiological evidence derives from investigating the effects of exogenous oestrogens on the incidence of breast cancer, such as long-term use of oral contraceptive pills (11) and oestrogen replacement therapy (12), supports this linkage. Furthermore, animal studies have shown repeatedly that oestrogens are able to induce and promote mammary tumours, and removal of ovaries, or administration of anti-oestrogenic drugs, achieves the opposite effect (13). However, the epidemiological evidence linking oestrogens with breast cancer risk is provocative but not conclusive for two reasons. Firstly, repeated full-term pregnancies at an early age, with the consequence of long-lasting increase of oestrogenic levels, are associated with a decreased risk of breast cancer. Secondly, the highest incidence of breast cancer is observed in old women when the oestrogenic level has been very low for decades (14).

Considerable efforts have been made in the past in attempting to show that women with high risk of breast cancer and/or those who subsequently develop the disease have abnormal endocrine profiles. However, comparative studies of high- and low-risk groups and familial studies have failed to yield consistent evidence that individuals' urinary and plasma oestrogen levels are associated with the increased risk (15). Interestingly, the evidence for urinary and plasma steroids is stronger for androgens than for oestrogens in predicting breast cancer (16).

Based on these observations, androgens have also been proposed as a possible carcinogenic factor in breast cancer (3). This proposal is mainly supported by the fact that the incidence of breast cancer is high in post-menopausal women when androgenic levels are high (17). Furthermore, administration of androgens for cystic disease has been shown to increase the risk of breast cancer (18). It has been reported that higher androgens and lower progesterone levels have been found in pre-menopausal women with breast cancer compared with controls (19). Moreover, Japanese women, who have a lower risk of breast cancer than their American and British counterparts, have lower plasma androgen levels. More recently, a number of researchers have demonstrated that, among all plasma steroids, the evidence for association of testosterone levels with breast cancer is strongest, although they could not determine whether this association is a cause or an effect of malignancy (20).

However, although quite a few studies have evaluated plasma androgen levels in relation to breast cancer risk, the results have been inconsistent. Testosterone has been positively associated with breast cancer in most (2126), but not all (27,28), previous studies. Furthermore, most studies suggested that increased levels of testosterone might have a modest but indirect association with breast cancer through its conversion to oestradiol (29,30).

The uncertainty of the roles of sex hormones, especially androgens, in breast carcinogenesis is, at least in part, due to a lack of a suitable animal model to enable researchers to carry out detailed investigations. The two most widely used experimental systems for the study of mammary cancer involve the use of chemical carcinogens, such as 1,2-dimethylbenz[a]anthracene and N-methylnitrosourea. These rodent models are excellent to study the early events of chemical carcinogenesis, malignant progression, and for carrying out in vivo signal transduction studies towards some oncogene-mediated events, such as ras-mediated events (31). However, these models, as well as other models induced by adenovirus and ionizing radiation, are not ideal for investigating mammary carcinogenesis, as they do not closely mimic the hormonal situation in normal animals.

Noble and his colleagues had conducted pioneering studies on the induction of mammary cancer by steroid hormones in a rat strain (Noble) (32). Because of their interests in prostate carcinogenesis, very little details about mammary carcinogenesis were given in the report. Furthermore, they used estrone, one of the most cytotoxic forms of oestrogen, which is present in high levels only in post-menopausal women. Key (33) has reviewed the relationship of hormones and cancer in humans and suggested that sex hormones affect both early and late stages in breast carcinogenesis. Oestradiol is the main form of oestrogen which affects the early stage of breast carcinogenesis.

We have developed a new animal model for mammary cancer based on Noble's method (32), with an important modification, using a combination of testosterone and 17β-oestradiol (E2), which more closely mimics breast cancer in human. Using this animal model, we have successfully induced the development of a high percentage of mammary cancer in female Noble rats. We have examined the expression of tumour growth factor-β and vascular endothelial growth factor and their receptors in mammary cancer (3436). We have established that an autocrine growth regulatory mechanism exists in this mammary cancer (B.Xie et al., submitted for publication).

We have also carried out a comparative study to further investigate the role of sex hormones, especially testosterone, in mammary carcinogenesis. We found that prolonged treatment with a combination of oestradiol and testosterone induces much higher incidence of mammary cancer in Noble rats, than either hormone alone. The histopathological patterns of the induced tumours and the number of tumours per tumour-bearing animal very closely resemble human breast tumours. Furthermore, the dosage of testosterone, although it does not influence the cumulative tumour incidence, does affect the tumour latency period. This suggests that testosterone may act as a promoter in the mammary tumorigenesis. Thus, the study: (i) provides the first in vivo evidence that both oestrogens and androgens are important cause for mammary tumorigenesis and that androgens may work synergistically as a promoter to shorten the latency period; and (ii) offers a useful animal model for exploring the mechanisms involved in human breast carcinogenesis, especially for the evaluation of the role of androgens in mammary tumorigenesis.

Materials and methods

Animals and animal care

The animals were housed in wire mesh suspended stainless steel cages, three to a cage, under standard conditions (22 ± 2°C, 40–70% relative humidity, 12 h light–12 h dark). Animals were fed with standard rat chow with tap water ad libitum. Body weights were recorded weekly. All surgical operation was done under pentobarbitone anaesthesia. The skin area involved was shaved and sterilized with 70% alcohol. A small incision was made and hormone-filled Silastic tubings were inserted into the subcutaneous tissue. Animals were killed at the end of experiments by cervical dislocation.

Hormonal treatment

At 3 months of age, intact sexually mature female Noble rats, weighing 180–210 g, were randomly divided into six groups, according to the protocols listed in Figure 1. Rats of groups IL and IR were surgically implanted s.c. in the left and right inguinal regions, respectively, with four 2.0 cm Silastic tubings [(i.d. 1.6 mm, o.d. 3.2 mm) sealed with RTV-108 silicone rubber adhesive sealant (General Electric, Waterford, NY)], tightly packed with testosterone propionate (each 2 cm tubing = 30 mg, total 120 mg) (Sigma) and one 1.0 cm Silastic tubing containing E2 benzoate (22 mg) (Sigma, St Louis, MO). In groups II and III, rats were implanted s.c. in the left inguinal region with one 1.0 cm Silastic tubing of E2 benzoate (22 mg) and four 2.0 cm Silastic tubings of testosterone propionate (120 mg), respectively. Group IV received one 1.0 cm E2 benzoate (22 mg) and two 2.0 cm testosterone propionate (60 mg) tubings. In group V, rats received empty Silastic tubings sealed with the same sealant. All tubings were replaced at 3 month intervals.

In order to investigate the correlation between the number of testosterone implants and serum testosterone levels we also carried out an experiment of shorter duration, by treating the rats with two, four or six testosterone implants (2T, 4T and 6T, respectively).

Histopathological examination

The rats were palpated regularly for mammary tumours starting from 2 months after treatment. The rats were killed when they became moribund or when a tumour >2 cm in diameter was detected. A small number were killed from 2 months onward to detect any early changes in mammary glands even when no palpable lumps were detected. The remaining rats, including non-tumour bearing and those with tumours <2 cm in diameter, were killed at 12 months after the initiation of treatment. At autopsy, the mammary tumours were removed and fixed immediately in 10% phosphate-buffered formalin (pH 7.2), trimmed and embedded in paraffin. The remaining mammary glands, though without obvious palpable tumour development were also removed for further examination. Paraffin sections (4 μm in thickness) were prepared and stained with hematoxylin and eosin (H&E) for histopathological examination, using the criteria and classification of mammary tumours as outlined by Young and Hallowes (37) and Russo et al. (38).

Measurement of testosterone and E2 in serum

To determine the concentration of serum testosterone and E2, 5 ml of venous blood were collected in a plastic tube without additives at the end of experiment and after the animals had been killed. After clotting at room temperature, the tubes were centrifuged lightly and the serum was aspirated. The serum testosterone concentration was determined by ELISA method using EIAgen Testosterone Kit (BioChem ImmunoSystems, Rome, Italy), whereas the serum E2 concentration was measured by enzyme-linked fluorescent assay using VIDAS Oestradiol II Kit and mini VIDAS instrument (bioMerieux Vitek, Marcy L'Etoile, France).

Iball's index and statistical analysis

The Iball's index is defined as the ratio of incidence (%) to the average latency period in days, multiplied by 100 (39). Statistical analyses were carried out using Fisher's exact probability test for incidence. Data of body weight, serum testosterone concentration and serum E2 concentration were given as means ± SD. ANOVA was used to compare the means of the groups investigated. Student-Newman-Keuls was used as post-hoc test to determine the level of significance between two mean values. We used the statistical software program SPSS 7.0 to generate the analysis (40). A P-value <0.05 was considered to be statistical significant.


Quantitative analysis of sex hormones

We measured the serum testosterone and E2 concentration in different experiment durations (from 3 to 12 months) and with different treatments (i.e. 1E + 4T, 1E + 2T, 1E and 4T). The differences of serum testosterone concentration between groups IL and IV were very significant, whereas the differences between groups IL and III were not significant (Table I). The coefficients of variance (CVs) in the four experimental groups were very small, indicating the serum testosterone concentration was relatively constant during the period. In order to explore the correlation between the number of testosterone implants and the serum testosterone concentration, we further carried out an experiment of shorter (7 weeks) duration. The results showed that the serum hormonal concentration was very strongly positively correlated with the number of testosterone implants (Table II, R = 0.998). Furthermore, the serum testosterone concentration had shown no significant fluctuation between the time of 7 weeks and during the period of 3–12 months, indicating that the serum levels of testosterone in group IL were relatively constant during the period studied. A very small CV (7.97%) of concentration of testosterone in serum provides additional support for the stable serum levels of testosterone during the period studied.

The results of serum E2 concentration are shown in Table III. The differences among groups IL, IV and II were not significant. Although the serum E2 concentration in group III was higher than that in control group, the difference was also statistically not significant. However, the serum E2 concentrations in groups IL, IV and II were significantly higher (~40-fold) than those in group III and control groups.

Body weight and hormonal protocols

The body weight of the hormone-treated rats was lower than that of controls, except for the group treated with testosterone alone (group III), which showed no significant difference to the control (Figure 2). The body weight of rats treated with oestradiol was maintained at a very stable level, while the weight of rats treated with testosterone alone increased steadily, although the rate of increase became slower after 20 weeks. By 40 weeks, there was no difference between the testosterone-treated and control groups. The weight of rats treated with a combination of testosterone and oestradiol increased only slightly.

Incidence of mammary carcinogenesis

The results of hormone-induced mammary carcinogenesis are summarized in Tables IV and V and in Figure 3. Spontaneous development of mammary tumours was not observed (in control group) during the experimental period of 12 months. Compared with other groups, the rats in group IL showed a significantly increased incidence of mammary cancer (52.78%), with a latency period of 5.82 ± 1.77 months, and with an average of just greater than one tumour per tumour-bearing rat (1.16 ± 0.37). Although multiple tumours had been observed in animals, most rats had only one tumour at the time of killing (Table V). Furthermore, most tumours induced were detected in thoracic mammary gland, only very rarely had tumours been observed in cervical, abdominal or inguinal mammary glands. The tumour incidence was 22.22% in group II (E2 alone) and 16.67% in group III (testosterone alone), which was not statistically significant between them. The highest Iball's index score for the overall development of mammary tumour was observed in group IL, which was 3-fold higher than that in group II and 4-fold higher than that in group III. The Iball's index of group IL was also significantly higher than that in group IV, which received one E2 and two testosterone capsules. As shown in Figure 3, the appearance of the first tumour occurred ~1–2 months earlier in the group IL than in groups II, III and IV. Significant differences in the cumulative incidence of tumours were observed between groups IL (44.44%) and II (11.11%), starting from 6.8 months. On the other hand, the difference in latency period between groups IL and IV was statistically significant, although there was no difference in cumulative incidence among them (52.78 and 53.33%, respectively; Figure 3 and Table IV).

The correlation between the location of tumours and the site of implantation of sex hormones was very strong. In group IL, >85% tumours appeared on the left side while in group IR all tumours (100%) were found on the right side amid a relatively small number of animals. The difference was highly significant (P < 0.01, Table V).


Mammary glands from age-matched control animals were characterized by a sparse cluster of epithelial tubules, embedded in a small amount of connective tissue, surrounded by a large fat pad. The epithelial ducts had small lumen, and darkly stained cuboidal epithelial cells (Figure 4). Mammary glands of all rats implanted with a combination of testosterone and oestradiol had a more extensively branched and dilated alveolar or ductal system against a background of loose connective tissue stroma. The alveoli or ducts showed a variable degree of epithelial dysplasia (Figure 5) after 2 months. Focal areas of dysplastic cells were seen interspersed among relatively normal glandular cells. Dysplastic cells were characterized by a loss of polarity and pleomorphic nuclear morphology. Irregular proliferation of epithelium within ducts were often accompanied by an apparent increase in the secretory activity. We also observed a marked dilation of ducts resulting from the accumulation of secretory material, foamy cell debris and macrophages in the lumen (Figures 5 and 6). Typically, patches of dysplastic and carcinomal cells were seen as focal thickening of epithelium extending into the lumen (Figures 6 and 7). Many alveoli or ducts were seen to be filled completely or partially by tumour cells. These were referred to as carcinoma in situ (Figures 6–8).

The fully developed carcinoma had variable histopathological features ranging from papillary (Figure 8), cribriform (Figure 9) or comedo (Figure 10) patterns, alone or in combination. The papillary pattern consists of bulbous, seemingly fragile, neoplastic epithelial projections protruding into the lumen of the ducts (Figure 8) were often seen in early development. These projections were composed of relatively uniform cells with small nuclei, arranged haphazardly or at right angles to the long axis of the papillae. The cribriform pattern was characterized by neoplastic cells which extended into the ductal lumen to form bridges of anastomosing arcades with many vacuole-like structure (Figure 9). Very often, tumours appeared as multilayered cells surrounding a central necrotic debris, resulting in comedo pattern (Figure 10). Central necrosis with dystrophic calcification was particularly common in this pattern. As the tumours enlarged further, they invaded the stroma, forming the typical infiltrating ductal or lobular carcinoma (Figures 11 and 12). Infiltrating ductal carcinoma also could develop into papillary, cribriform or comedo patterns.

The well-differentiated type of infiltrating ductal carcinoma retained a tendency to form tubules and alveoli (Figure 11), while the moderately differentiated type was characterized by some features of glandular formation within the tumour masses. The poorly differentiated carcinoma was characterized by solid masses of poorly organized tumour virtually devoid of glandular formation (Figure 12). Very often, isolated islets or single tumour cells were seen infiltrating the connective tissue stroma (Figure 12). Although the breast tumours induced by testosterone and oestradiol were of all ranges, the predominant type was solid adenocarcinomas with moderate stroma and obvious evidence of local invasiveness. Close proximity of tumour cells to stromal areas were frequently observed (Figure 13). However, only one rat was found bearing a metastatic tumour in the uterus (Figure 15), while another one was believed to have invaded the underlying thoracic muscular layer (Figure 14). The possibility of primary uterine tumour was considered and excluded due to the fact that there was no tumour observed anywhere in the endometrium. The histopathology of tumours induced by oestradiol or testosterone alone was similar to those induced by a combination of testosterone and oestradiol.


Sex hormones and mammary cancer

The results of the present study show that in the rat, mammary tumours can be readily induced by a combination of E2 and testosterone. Over 50% of animals developed mammary cancer after sex hormone implantation, with an overall latency period of ~6 months. To the best of our knowledge, this is the first study to document in a systematic way, the successful induction of mammary carcinogenesis by a combination of female and male sex hormones, without addition of chemical carcinogens. We believe this is a far better model for in depth study of the mechanisms of mammary carcinogenesis, as it closely mimics the natural human breast cancer. The existing animal models for mammary cancer mostly involved the use of potent chemical carcinogens (4143). They do not mimic human mammary cancer well for at least three reasons. Firstly, the breast is a sex hormone target organ and has little chance to contact chemical carcinogens in any significant way under normal situations. Secondly, mammary cancers induced by chemical carcinogens are commonly multiple in number and the number of malignant tumours per rat shows a dose dependency (44). However, in clinical practice the finding of a double carcinoma in the same breast and of a simultaneous bilateral carcinoma is infrequent (14). Thirdly, certain oncogenes, such as the ras gene, are extremely rare in human breast cancers but are substantially expressed in chemically induced mammary cancers (31,45). In time, this will prove to be most suitable for detailed analysis of mammary carcinogenesis, especially for the role of these hormones in mammary cancer development.

Furthermore, most of the tumours induced by a combination of testosterone and oestradiol bear a striking resemblance, not only in growth patterns but also in histopathology, to their human counterparts. Although involvement of multiple mammary glands has been observed in three rats, the tumours usually occurred singly in a given rat. Examination of the expression of oestrogen receptor, some common oncogenes and growth factors as well as their receptors in these cancers (unpublished observation), have revealed that they behaved very similarly to human breast cancers. On the other hand, the low rate of distant metastases observed in our study is also consistent with observations in clinical situations in women. Human breast carcinoma frequently invades adjacent lymphatic vessels and blood vessels and metastasizes to axillary nodes, but distant metastases are rarely present at the time of first diagnosis of breast cancer (14). In our study, close proximity of cancer cells to vessels, suggesting potential vascular invasion, were frequently observed, but only one rat bearing metastatic tumour in uterus was observed with another case of invasion to the underlying muscular tissues. Another factor which may explain the relatively low rate of metastasis, the fact that animals were killed when tumours were still relatively small. Had we allowed tumours to proceed further in development, a higher metastatic rate might have resulted. This is being examined at the moment. Although the most common sites of distant metastases in human are lungs, bones and liver (14), breast carcinoma metastasizes to uterus have also been reported (46). Based on these observations, we consider that mammary carcinogenesis induced by testosterone and oestradiol is a model that mimics human breast cancer most closely. Thus, it can be used to elucidate the role and mechanism of sex hormones in mammary carcinogenesis.

The latest epidemiological evidence shows that increased levels of major sex steroid hormones (androgens and oestrogens) in the blood play an important role in the etiology of breast cancer in pre- and post-menopausal women (4750). However, the persisting controversies about the relationship between the hormone levels and the subsequent development of breast cancer still remain (51). There are at least two factors thought to be responsible for these discrepancies. Firstly, serum hormone measurements are subject to variation by many factors, including age, health status, laboratory variations and time of day when samples are taken. Furthermore, the underlying genetic differences in hormone metabolism among individuals may also introduce variables that may distort the results of studies of serum hormone levels. Secondly, although it is currently believed that androgens are strongly associated with breast cancer risk, it is still unknown whether androgens are a cause or an effect of breast carcinogenesis and whether androgens have a direct or an indirect effect on breast carcinogenesis (3,17). Our observation showed that testosterone implanted alone had the ability to induce mammary carcinogenesis in female Noble rats. Although the mechanism is not clear, it tends to support the hypothesis that androgens may have a direct role in mammary carcinogenesis, in addition to its role through oestradiol indirectly by conversion through aromatase activity. The results from our present study suggest that androgens may be a cause rather than an effect of mammary carcinogenesis in the rat.

The incidence and Iball's index in the rats treated with testosterone and oestradiol is significantly higher than those treated with either testosterone or oestradiol alone. Reduction in testosterone dosage while keeping the oestradiol level unchanged, increases the latency period, but not the cumulative incidence of mammary cancers. The results suggest that the two hormones act synergistically on the mammary gland, in mammary carcinogenesis, with testosterone playing a crucial role in the process.

Androgens and mammary cancer

A summary of the literature to date reveals the following: (i) androgens may act directly, by stimulating breast epithelial cell proliferation through binding to androgen receptors or by stimulating the synthesis of growth factors in the breast epithelium (52); (ii) androgens may act indirectly through conversion of androgens to oestrogens (53); (iii) androgens may also increase indirectly the risk of breast cancer by decreasing the fraction of oestradiol bound to sex hormone-binding globulin (SHBG), thereby increasing the unbound (free) oestradiol fraction, which is thought to be the fraction available to breast cells (54); and (iv) it has been suggested that testosterone inhibits hepatic secretion of SHBG (55), which could also result in a decreased fraction of oestradiol bound to SHBG. Our finding that the incidence of mammary cancer induced by a combination of the two hormones is higher than those induced by either hormone alone, indicates that the two hormones, as pointed out earlier, can act synergistically in the mammary gland tissue, to induce mammary carcinogenesis.

What might be the most likely way that androgens increase mammary cancer incidence? There is no definitive answer to this question for the time being. In this study, we have observed that, compared with controls, treatment with testosterone alone can also induce breast carcinogenesis. This result was surprising to us initially. We believe the real answer rests with synergism between testosterone and oestrogen. As has been established, testosterone can be converted to oestrogens by aromatase. Although the amount of testosterone implanted that eventually got converted into oestrogen is not known in this case, a scenario of coexistence of oestrogens (natural and converted) and testosterone can be established. We believe a synergistic effect of oestrogen and testosterone continues to exist in this situation. Similarly, compared with oestradiol-treated rats, treatment with a combination of testosterone and oestradiol shortens the latency period. When the number of tubings containing testosterone was reduced to half (1E2 + 2T) of those used in group IL (1E2 + 4T), no significant difference in cumulative incidence was observed but the latency period was significantly longer than in group IL. The rate of release of testosterone from Silastic tubings is correlated to the surface area of the tubings, reducing the number of the tubings would, therefore, result in decreasing the dosage of testosterone. On the other hand, treatment of intact male rats with testosterone alone does not induce mammary tumour, while a combination of testosterone and oestradiol does (data not shown), indicating that androgen is not a dominant factor in mammary carcinogenesis.

We have measured serum testosterone and E2 concentrations. The results showed that there is no significant fluctuation in serum E2 levels in the presence of E2 implant, irrespective of whether the animals were treated with E2 alone or in combination with testosterone (i.e. 1E + 4T or 1E + 2T). There is also no significant difference in serum testosterone levels whether the treatment was by testosterone alone or in combination with E2 (i.e. 1E + 4T). However, the level of serum testosterone was reduced when the dosage of testosterone was reduced (i.e. 1E + 2T). These results show that the levels of serum testosterone or E2 are not significantly altered by the combination of these hormones. This indicates that the concentration of one hormone (e.g. testosterone) is not affected by another hormone (e.g. E2) in the blood and vice versa. However, the incidence of mammary carcinogenesis is substantially higher in rats treated with a combination of E2 and testosterone than either hormone alone, suggesting that some forms of synergism must have occurred. Our study further shows that a reduction of testosterone dosage by half (e.g. from 1E + 4T to 1E + 2T) does not result in a significant decrease of serum E2 level but a significant decrease of serum testosterone level. Given the fact that different dosage of testosterone affects the latency period of mammary carcinogenesis, but does not affect the cumulative incidence of mammary cancer, these observations lead us to speculate that testosterone may act as an endogenous promoter in mammary carcinogenesis and this role may be through stimulation of secretion of growth factors and oncogenes in mammary glands rather than stimulation of breast cell proliferation.

Site of hormone implants and mammary cancers

Our results further show that the site of hormone implantation has a significant influence on the origin of mammary cancers. This finding was surprising. In our preliminary study, all hormone implants were made subcutaneously in the left inguinal region. In these animals, >85% of tumours were found to occur on the left, especially the left thoracic mammary gland (rats mainly have four pairs of mammary glands i.e. cervical, thoracic, abdominal and inguinal mammary glands). We consequently added another group in which the hormones were all implanted in the right inguinal region. To our surprise, all tumours thus induced were on the right side, and predominantly occurred in the right thoracic mammary gland. The reason for this peculiar ipsilateral correlation of the site of hormone implantation to breast tumour development is unknown. If local diffusion, in addition to systematic blood circulation, was a factor, then the abdominal or inguinal mammary glands on the same side, which are nearer to hormonal implants in distance, should be affected more. The fact that most tumours were found in thoracic glands indicate that factors other than distance from the hormonal source might be involved. There is as yet no satisfactory explanation to this unexpected finding.

In an ongoing comparative study, we have measured the Ki-67 labeling index (proliferative index) and mammary gland density (MGD) to determine the proliferation of ductal epithelial and periductal stromal cells in response to hormone treatment. The data show that, in 1E + 4T group, the Ki-67 labeling indices between left and right thoracic mammary glands are not significantly different. However, the MGD is significantly higher in the left than the right thoracic mammary glands (1.0214 ± 0.079 and 0.9288 ± 0.065 g/cm3, respectively), indicating that the proliferation of perialveolar or periductal stroma is higher in the left than the right thoracic mammary glands (B.Xie et al., in preparation). Whether or not this has anything to do with the ipsilateral correlation of tumours remains to be elucidated.

Another unique feature was that most tumours were detected in the thoracic gland while much less frequently in other pairs, i.e. cervical, abdominal or inguinal glands. The reason for this is again not certain. Russo et al. (5658) have identified the terminal end bud (TEB) as the most likely site of the origin of breast carcinomas in this complex branching gland. They also have reported that tumour incidence in the thoracic mammary glands is higher than in the abdominal mammary glands. They ascribe this difference in carcinogenic response to the asynchronous development of glands in different topographic areas. The thoracic glands, which are lagging behind in development, retain a higher concentration of TEB and, thus, a higher tendency towards carcinogenesis. Our observations are consistent with Russo's reports.

In conclusion, the present study demonstrates clearly that prolonged treatment with a combination of oestradiol and testosterone induces much higher incidence of mammary cancer in female Noble rats, regardless of the dosage of testosterone used. The histopathological patterns of the induced tumours and the number of tumours per tumour-bearing animal bear close resemblance to human breast tumours. Furthermore, the dosage of testosterone, though it does not influence the tumour incidence, affects the tumour latency period. This suggests that testosterone may act as a promoter in the mammary tumorigenesis. Thus, our studies (i) provide in vivo evidence that both oestrogen and androgen are important causes of mammary tumorigenesis; (ii) offer a useful animal model for exploring the mechanisms involved in human breast tumorigenesis, especially for evaluation of the role of androgens in mammary tumorigenesis.