1.1.1. Nomenclature

1.1.2. Structural and molecular formulae and relative molecular mass

1.1.3. Chemical and physical properties of the pure substances

From Budavari (1995), unless otherwise specified

1.1.4. Technical products and impurities

Tamoxifen in pharmaceutical formulations is invariably present as its citrate salt. Tamoxifen citrate is available as 15.2-, 30.4- and 45.6-mg (equivalent to 10, 20 and 30 mg tamoxifen base) tablets which also may contain carboxymethylcellulose calcium, croscarmellose sodium (type A) [a polymer of carboxymethylcellulose sodium], gelatin, hydroxypropyl methylcellulose 2.910, lactose, Macrogel 300, magnesium stearate, mannitol, polyvinylpyrrolidone (povidone), sodium carboxymethylstarch, corn starch or titanium oxide (Thomas, 1991; Farmindustria, 1993; Reynolds, 1993; British Medical Association/Royal Pharmaceutical Society of Great Britain, 1994; Medical Economics, 1996).

The impurities limited by the requirements of the European Pharmacopoeia include: (E)-2-(1,2-diphenylbut-1-enyl)phenoxy]ethyldimethylamine (the E-isomer of tamoxifen); 2-[4-(1-hydroxy-1,2-diphenylbutyl)phenoxy]ethyldimethylamine; 2-[4-(1,2-diphenylvinyl)phenoxy]ethyldimethylamine; 2-[4-(1,2-diphenylprop-1-enyl)phenoxy]dimethylamine; 2-[2-(1,2-diphenylbut-1-enyl)phenoxy]ethyldimethylamine; (Z)-2-[4-(1,2-diphenylbut-1-enyl)phenoxy]ethylmethylamine; and 1-(4-dimethylaminoethoxyphenyl)-2-phenylbutan-1-one (Council of Europe, 1995). The United States of America and British pharmacopoeias limit the E-isomer content to not more than 0.3% and 1%, respectively (British Pharmacopoeial Commission, 1993; United States Pharmacopeial Convention, 1994).

Trade names and designations for tamoxifen citrate and its pharmaceutical preparations include: Apo-Tamox; Citofen; Dignotamoxi; Duratamoxifen 5; Emblon; ICI-46 474; Jenoxifen; Kessar; Ledertam; Noltam; Nolvadex; Nourytam; Novofen; Oestrifen; Oncotam; Retaxim; Tafoxen; Tam; Tamaxin; Tamifen; Tamofen; Tamone; Tamoplex; Tamoxasta; Tamox-Gry; Tamoxigenat; Tamox-Puren; Taxfeno; Terimon; Valodex; Zemide; Zitazonium.

1.1.5. Analysis

Several international pharmacopoeias specify Potentiometric titration with perchloric acid as the assay for purity of tamoxifen citrate, and liquid chromatography (LC) or gas chromatography with flame ionization detection for determining levels of the E-isomer and other impurities and decomposition products. The assays specified for tamoxifen citrate in tablets use LC and ultraviolet/visible absorption spectroscopy with standards. An assay for heavy metal impurities is also specified (British Pharmacopoeial Commission, 1993; United States Pharmacopeial Convention, 1994; Council of Europe, 1995).

Tamoxifen and its metabolites can be analysed in biological fluids and tissues by thin-layer chromatography (Furr & Jordan, 1984), gas chromatography–mass spectrometry (MS) (Furr & Jordan, 1984) and high-performance liquid chromatography with ultraviolet, fluorimetric or electrochemical detection (Chamart et al., 1989; Berthou & Dréano, 1993; Lim et al., 1993; Fried & Wainer, 1994).

1.2.1. Production

Tamoxifen is prepared by reacting 4-β-dimethylaminoethoxy-α-ethyldesoxybenzoin with phenylmagnesium bromide or phenyl lithium to form 1-(4-β-dimethylaminoethoxyphenyl)-1,2-diphenylbutanol, which on dehydration yields a mixture of tamoxifen and its E-isomer that may be separated with petroleum ether. Tamoxifen is converted to the 1:1 citrate for pharmaceutical use (Gennaro, 1995).

Worldwide production of tamoxifen citrate has increased from approximately 7.0 tonnes in 1989 to 8.5 tonnes in 1991, 10.1 tonnes in 1993 and 103 tonnes in 1995.

1.2.2. Use

Tamoxifen was first synthesized by Bedford and Richardson in the United Kingdom (Bedford & Richardson, 1966). It was shown to be an anti-fertility agent in rats (Harper & Walpole, 1967a,b), but was soon found to induce ovulation in women and to have either oestrogenic or antioestrogenic effects, depending on species specificity and tissue and receptor status (Harper & Walpole, 1966, 1967a). The earliest clinical studies were carried out in postmenopausal women with advanceed breast cancer in Manchester, United Kingdom (Cole et al., 1971). It was soon appreciated that tamoxifen was a successful palliative therapy for advanced breast cancer, yielding response (see Glossary, p. 449) rates similar to those seen with other endocrine approaches while producing few side-effects (Ward 1973; O'Halloran & Maddock, 1974). It was approved for use as a pharmaceutical in the United Kingdom in 1973 (Jordan, 1988). By 1978, it was being widely adopted as first-line endocrine therapy for postmenopausal women with advanced disease, particularly after it was shown to be as effective as, and less toxic than, diethylstilboestrol, the previous standard in that setting (Ingle et al., 1981). It was tested in premenopausal women for therapy of metastatic breast cancer from the late 1970s (Manni et al., 1979; Pritchard et al., 1980; Planting et al., 1985; Buchanan et al., 1986; Ingle et al., 1986; Sawka et al., 1986) and shown to be 20–30% effective, but it has probably been far less widely used in this younger group, with chemotherapy or ovarian ablation remaining the more usual approaches (Sunderland & Osborne, 1991; Early Breast Cancer Trialists' Collaboration Group, 1992).

Since the early 1980s, tamoxifen has been widely accepted as first-line endocrine therapy for metastatic disease in postmenopausal women (Ingle, 1984). Most postmenopausal women who develop metastatic disease will at some point undergo at least one attempt at endocrine therapy, either as a palliative treatment or as first-line therapy for metastatic disease. Even those women who respond to tamoxifen for metastatic disease generally receive it for, on average, only 9–12 months (Muss, 1992).

In the mid- to late 1970s, many investigators became interested in using tamoxifen as adjuvant therapy in women at high risk for recurrence of breast cancer following surgery, because of the equivalence of this approach to other endocrine therapies and the drug's extremely low short-term toxicity profile. Several large trials of adjuvant tamoxifen therapy in mostly postmenopausal, axillary node-positive women demonstrated a small but statistically significant improvement (about 25%) in both recurrence-free and overall survival in these patient groups (Nolvadex Adjuvant Trial Organisation, 1983; Ribeiro & Swindell, 1985). Although the relative merits of tamoxifen and chemotherapy for use in this setting were widely debated, particularly between the United Kingdom and the United States, the final results from the Oxford Overview (Early Breast Cancer Trialists' Collaborative Group, 1988) convinced clinicians on both sides of the Atlantic of the benefit of tamoxifen not only in this patient group but in others as well (Breast Cancer Chemotherapy Consensus Conference, 1985; Glick et al., 1992).

From the early days of trials of adjuvant tamoxifen therapy, British clinicians in particular favoured the use of tamoxifen in node-negative (lower-risk) and premenopausal women as well as in postmenopausal, node-positive women. Several large trials soon supported the use of tamoxifen in this setting and showed that it improved overall and recurrence-free survival (Breast Cancer Trials Committee, 1987; Fisher et al., 1989a,b).

From the early days of these trials, dose and dosage varied from one country to another. In the United States and the United Kingdom, 20 mg daily given for one to two years was the early norm (Nolvadex Adjuvant Trial Organisation, 1983; Fisher et al., 1986; Nolvadex Adjuvant Trial Organisation, 1988), while in continental Europe doses of 30–40 mg daily for one to two years were more usual (Fornander et al., 1991; Mouridsen et al., 1988; Rutqvist et al., 1992; Rutqvist & Mattsson, 1993; Rutqvist et al., 1995). The wide variety of dosing is clear from the summary tables of the Early Breast Cancer Trialists' Collaborative Group (1988, 1992). In addition, data from rat models suggested that longer tamoxifen treatment would be advantageous (Jordan, 1978; Jordan et al., 1979, 1980) and trials examining two versus five years, five versus ten years and two or five years versus indefinite tamoxifen were undertaken. From the mid-1980s to the mid-1990s, tamoxifen therapy of five years or longer was used increasingly in many countries. Further, with reports of its effectiveness in node-negative women, the use of tamoxifen spread widely, to the point where it was sometimes used to treat even very small (< 1 cm) invasive carcinomas and carcinomas in situ for which it had never been tested in randomized trials. Trials of its use for very small invasive cancers and carcinoma in situ are now in progress: National Surgical Adjuvant Breast and Bowel Project, B21, B24.

Tamoxifen has been the adjuvant therapy of choice for postmenopausal, node-positive women and oestrogen receptor-positive or progesterone receptor-positive (see Glossary, p. 448) since the mid-1980s and for postmenopausal, node-negative and oestrogen receptor-positive or progesterone receptor-positive women since the early 1990s. It is also used in many cases in postmenopausal receptor-negative women and in premenopausal women with low-risk (node-negative) receptor-positive disease (Glick et al., 1992; National Institutes of Health Consensus Development Panel, 1992; Goldhirsch et al., 1995). In both pre- and postmenopausal women, it is also often given concurrently with or following chemotherapy as a type of adjuvant maintenance (Glick et al., 1992; Tormey et al., 1993; Goldhirsch et al., 1995). Thus, a high proportion (40–60%) of all women who undergo potentially curative surgery for breast cancer now receive adjuvant tamoxifen therapy for a period of some two to five years.

Most breast cancer patients with metastatic disease receive, at some time in the course of their treatment, cytotoxic chemotherapy as well as tamoxifen and often other hormonal therapies. Similarly, many women, particularly in the postmenopausal, axillary node-positive subset, receive cytotoxic chemotherapy as well as tamoxifen as part of their adjuvant therapy. Again, the Early Breast Cancer Trialists' Collaborative Group (1988, 1992) summary tables document the use of these drugs, which are mainly cyclophosphamide, methotrexate and 5-fluorouracil-based combinations, but also include other cytotoxic agents such as melphalan and adriamycin, many of which are documented carcinogens in their own right (see IARC, 1987a). There are nevertheless many randomized trials of tamoxifen adjuvant therapy versus no therapy, particularly in postmenopausal axillary node-positive and pre- and postmenopausal node-negative subjects, from which information relating to tamoxifen exposure without accompanying cytotoxic agents can be obtained (Early Breast Cancer Trialists' Collaborative Group, 1992).

Tamoxifen has also been commonly used, without surgery, as a primary therapy for breast cancer in elderly women who are considered poor candidates for surgery (Akhtar et al., 1991), although two randomized studies have suggested that surgical removal of the primary tumour plus tamoxifen treatment is preferable, leading to fewer problems with local recurrence (Bates et al., 1991; Mustacchi et al., 1994).

In the mid-1980s, because of its known cytostatic action and because of the reduction in new contralateral breast cancers observed in many clinical trials of tamoxifen (Breast Cancer Chemotherapy Consensus Conference, 1985; Nolvadex Adjuvant Trial Organisation, 1988; Early Breast Cancer Trialists' Collaborative Group, 1992), considerable interest arose in using tamoxifen as a preventive agent. At least three large trials are in progress in Italy (hysterectomized women), North America and the United Kingdom (Powles et al., 1990; National Surgical Adjuvant Breast and Bowel Project, 1992; Redmond et al., 1993; Vanchieri, 1993; Costa et al., 1996). The subjects are women believed to be at high risk for developing breast cancer. These so-called high-risk women include those at a risk as low as that of the average 60-year-old in at least one trial (National Surgical Adjuvant Breast and Bowel Project, 1992). Although the practice is discouraged by most investigators, it is known that some women are currently prescribed tamoxifen as a preventive agent outside of study, in various high-risk situations (such as women with lobular carcinoma in situ or a family history of breast cancer).

Tamoxifen has been widely adopted as the appropriate first-line therapy for hormone-responsive male breast cancer and is also used frequently as adjuvant therapy for men with oestrogen receptor- or progesterone receptor-positive breast cancer, in spite of the lack of any large randomized trials in this relatively small group of patients (Jaiyesimi et al., 1992).

Tamoxifen, at doses similar to those used in breast cancer treatment, has given response rates of 20–30% in several phase II trials in women with advanced endometrial cancer (Quinn & Campbell, 1989; Barakat & Hoskins, 1994; Lentz, 1994) and has been accepted as a second-line endocrine therapy for unresectable or recurrent endometrial cancer (Swenerton et al., 1984).

There have also been studies, but probably not wide use, of tamoxifen treatment for a variety of other malignancies, including hepatocellular carcinoma (Martínez Cerezo et al., 1994), carcinoma of the stomach (Harrison et al., 1989), renal-cell carcinoma (Yagoda et al., 1995), melanoma (McClay & McClay, 1994), adenocarcinoma of the pancreas (Taylor et al., 1993; Wong & Chan, 1993), carcinoma of the cervix (Vargas Roig et al., 1993), carcinoma of the ovary (Hatch et al., 1991), glioblastoma multiforme (Baltuch et al., 1993), carcinoma of the biliary tract (West et al., 1990), desmoid tumours (Brooks et al., 1992) and meningiomas (Goodwin et al., 1993). In addition, tamoxifen has been suggested to be useful in treating certain non-malignant conditions including retroperitoneal fibrosis, the POEMS syndrome (a multi-system syndrome consisting of polyneuropathy, organomegaly, endocrinopathy, serum M-band and skin changes) (Enevoldson & Harding, 1992), oligoospermia in men with incomplete androgen sensitivity (Gooren, 1989), idiopathic oligozoospermia (Sterzik et al., 1993), pulmonary lymphangioleiomyomatosis (Kitaichi et al., 1995), rectal polyps in patients with Gardner's syndrome (Parry et al., 1993), Peyronie's disease (Ralph et al., 1992), autoimmune progesterone dermatitis (Stephens et al., 1989), menstrual migraine (O'Dea & Davis, 1990) and painful idiopathic gynaecomastia (McDermott et al., 1990). It has also been shown to have anti-candidal activity (Beggs, 1993).

Tamoxifen has also been suggested to be useful as an adjuvant to cancer chemotherapy of various types, in that it displays synergy with drugs such as cisplatin (McClay et al., 1993). Tamoxifen has also been shown to cause apoptotic effects which appear to be independent of the oestrogen receptor status of the cells affected (Perry et al., 1985).

Interest has focused more recently on positive effects of tamoxifen on cardiovascular lipid profiles (Bruning et al., 1988; Love et al., 1990, 1991; Shewmon et al., 1994; Thangaraju et al., 1994; Grey et al., 1995a; Guetta et al., 1995; Gylling et al., 1995; Kenny et al., 1995), on cardiovascular events (Rutqvist & Mattsson, 1993) and cardiovascular deaths (McDonald & Stewart, 1991) and on bone density (Love et al., 1988; Wolter et al., 1988; Kristensen et al., 1994; Love et al., 1994a; Wright et al., 1994; Grey et al., 1995b; Kenny et al., 1995; Leslie et al., 1995). These apparently positive effects have made the drug even more attractive for long-term use in early breast cancer and as a preventive agent. Some authors have even hypothesized that it may have more beneficial preventive effects on mortality and morbidity from causes other than breast cancer, at least for women over 60 (Gray, 1993).

Thus, tamoxifen has become very widely used around the world. It is estimated that there have been over 7 million patient-years of treatment with tamoxifen since it was first approved in 1973. Tamoxifen is now registered for use in 97 countries. Most women and men with metastatic breast cancer receive it at some time in their therapy, and 40–60% of women may receive it as adjuvant therapy, in some instances for five years or longer. It is being increasingly tested as a preventive agent against the development of breast cancer. Studies are in progress to examine potential effects of its long-term use on risks for cardiovascular disease and osteoporosis.

2.1.1. Case reports

Following an initial report by Killackey et al. (1985) of three cases of endometrial cancer in women who received tamoxifen for breast cancer, a large number of case reports have been published. Those available to the Working Group are summarized in Table 1. One hundred and two cases of endometrial cancer were reported in 32 case reports, ranging from isolated cases to series of up to 20 cases. The reports concern 72 adenocarcinomas, including 8 mucinous, 3 clear-cell adenocarcinomas and 1 serous papillary adenocarcinoma, 14 Müllerian mixed tumours, 1 stromal sarcoma and 13 carcinomas not otherwise specified. [The Working Group noted that rare histological types of endometrial cancer are over-represented in these reports.]

Table 1

Case reports of gynaecological cancers in patients receiving tamoxifen for breast cancer.

2.1.2. Case series

Two groups of case series were available to the Working Group. The first comprises series of cases of endometrial cancer following breast cancer, among whom tamoxifen use was assessed (Table 2). In one of these series, 15 of 53 cases had received tamoxifen, and more of these had high-grade tumours (poorly differentiated endometrial carcinomas) (67%) than in the group not treated with tamoxifen (24%; p = 0.03) (Magriples et al., 1993). In another series of 73 cases, of whom 23 had received tamoxifen, the proportion of high-grade tumours was 23% in tamoxifen-treated patients and 19% in non-treated patients. The disease stage (according to the current International Federation of Gynecology and Obstetrics (FIGO) criteria) in this series did not differ between tamoxifen-treated and untreated cases (Barakat et al., 1994). The second group comprises series of cases of breast cancer treated with tamoxifen, among whom the development of endometrial cancer was assessed (Table 3).

Table 2

Use of tamoxifen among endometrial cancer case series following breast cancer.

Table 3

Case series of endometrial cancer among breast cancer patients treated with tamoxifen.

2.1.3. Case–control studies

The case–control studies considered by the Working Group were those which compared tamoxifen use in women with breast cancer who did (cases) or did not (controls) subsequently develop endometrial cancer. A fundamental requirement for these controls is that they, like the cases, were at risk for developing endometrial cancer (namely, that they had an intact uterus). The failure to consider the hysterectomy status of controls may invalidate a case–control study of endometrial cancer. Other issues to be considered include factors which either may be determinants of risk for endometrial cancer (age, nulliparity, obesity, diabetes, hypertension, age at menopause and use of unopposed oestrogen therapy) or may influence the likelihood of tamoxifen prescription (calendar year of breast cancer diagnosis, menopausal status, and stage and oestrogen receptor status of the breast cancer). Determinants of risk for endometrial cancer are confounding factors in the studies discussed below only to the extent that they influence the likelihood of tamoxifen prescription. As in any case–control study, information and selection bias may also pertain. Finally, the possibility that endometrial cancer was diagnosed preferentially in women who had received tamoxifen constitutes a potential bias that is considered in greater detail in the introductory remarks to cohort studies and randomized trials (Sections 2.1.4 and 2.1.5).

Hardell (1988a) described a case series from a Swedish registry, covering 32 cases of endometrial cancer diagnosed at least one year after breast cancer. He then examined the same group in a case–control study (Hardell, 1988b), including 23 of these cases compared to 92 age- and breast cancer-matched controls. Eleven cases (48%) compared with 18 controls (20%) had received tamoxifen at 40 mg/day and nine (39%) cases compared with 10 (11%) controls had received pelvic irradiation. The odds ratio for patients receiving both treatments was 7.1 (95% confidence interval (CI), 2.3–22.1) compared to neither treatment, that for tamoxifen alone was 2.6 (95% CI, 0.7–9.6) and that for pelvic irradiation alone was 4.7 (95% CI, 0.8–27.3). [The Working Group noted that the presence of an intact uterus was not confirmed in the controls.]

van Leeuwen et al. (1994) conducted a large case–control study in the Netherlands in which 98 cases of endometrial cancer diagnosed at least three months following breast cancer were matched by age and date of diagnosis of breast cancer with 285 controls with breast cancer who survived with an intact uterus at least up to the time of diagnosis of endometrial cancer of the cases. Twenty-three cases and 58 controls had had treatment with tamoxifen. The relative risk associated with any use of tamoxifen was 1.3 (95% CI, 0.7–2.4). Statistically significant trends with duration and cumulative dose were found (p < 0.05 for both). The odds ratio for women who had taken tamoxifen for more than two years was 2.3 (95% CI, 0.9– 5.9). Most women were treated with doses of 40 mg/day (59%), 17% received 30 mg/day and 23%, ≤20 mg/day. The duration–response trends were similar with daily doses of 40 mg or 30 mg and less. No difference in stage or histology between the exposed and unexposed cases was found. [The Working Group noted that the statistical power to detect differences in risk associated with different dose intensities of tamoxifen was low.]

Cook et al. (1995) conducted a case–control study involving women under the age of 85 years registered in the Washington State Cancer Registry with a diagnosis of breast cancer between 1978 and 1990, who subsequently developed endometrial cancer at least six months after breast cancer. These were matched with controls without second primary cancer by age, year of breast cancer diagnosis and stage of cancer. Controls had to have an intact uterus and to have survived at least up to the time of diagnosis of endometrial cancer of their matched cases. Thirty-four endometrial cancer patients and 64 matched controls were analysed. All but two cases were postmenopausal. Tamoxifen use [mainly 20 mg/day] was more common in the controls (31% versus 26%). After adjustment for cytotoxic chemotherapy and duration of oestrogen replacement therapy, the matched odds ratio for any use was 0.6 (95% CI, 0.2–1.9). The mean duration of use was 14 months for cases and 21 months for controls. The odds ratio for more than one year's use was 0.2 (95% CI, 0.1–1.0) based on three cases and 16 controls. [The Working Group noted the inability of this study to address risks associated with long-term use of tamoxifen.]

Sasco et al. (1996) conducted a case–control study in Lyon and Dijon, France. Forty-three cases of endometrial cancer occurring at least one year after the diagnosis of breast cancer were matched with 177 controls for age, region, year of diagnosis of breast cancer, intact uterus and survival with breast cancer. The median dose was 20 mg/day; the median duration of treatment was greater in cases (63 months) than in controls (37 months); and tamoxifen was used in 67% of cases and 60% of controls (odds ratio, 1.4; 95% CI, 0.6–3.5). Information on duration of use was missing for 21% of exposed cases and 45% of exposed controls. The risk appeared to increase with duration of use, with a relative risk of 3.5 (95% CI, 0.9–12.7) for more than five years of use. [The Working Group noted that a difference in the percentages of cases and controls with unknown duration of treatment could have exaggerated the estimate of the effect of duration.]

2.1.4. Cohort studies

Detection bias may pertain to both cohort studies and randomized clinical trials, since tamoxifen is known to increase the frequency of symptoms such as vaginal bleeding or discharge which may lead to gynaecological evaluation. In addition, tamoxifen is known to induce benign gynaecological changes such as endometrial hyperplasia and polyps. Other changes include poorly defined thickening of the endometrium that may be revealed by ultrasound examination. Growth of leiomyomata may occur and provide a further opportunity for the diagnosis of endometrial cancer.

The longer survival of tamoxifen-treated patients may lead to greater duration of follow-up in which second cancers may occur. The appropriate methods of statistical analysis in this context are life table analysis or analysis of rates based on person-years at risk.

In an abstract, Champion et al. (1991) reported a breast cancer registry-based series from northern Alberta, Canada. Between 1953 and 1988, a total of 1874 women had taken tamoxifen, 20 mg/day for 22, 36 and 39 months and 8201 had not. Thirty-one women developed uterine cancer, three (two sarcomas and one adenocarcinoma) in the tamoxifen group and 28 in the other group. [The Working Group noted that no adjustment for period of diagnosis was made in this study and the results therefore could not be interpreted.]

Robinson et al. (1995) reviewed 586 eligible breast cancer patients without a previous hysterectomy in a medical centre series from Texas, United States. Of 108 patients who received tamoxifen (20 mg/day for at least one year), four developed endometrial adenocarcinoma and, of 478 breast cancer patients who did not receive tamoxifen, four developed endometrial cancer [odds ratio, 4.6; 95% CI, 1.3–16.0]. After adjustment for hypertension and diabetes mellitus, the odds ratio for development of endometrial cancer after tamoxifen use was 15.2 (95% CI, 2.8–84.4). [The Working Group noted the imprecision of the estimates and the difference between the crude and adjusted odds ratios, that the results were not adjusted for follow-up time and that the presence of an intact uterus during follow-up was not controlled for.]

Curtis et al. (1996) examined the effect of tamoxifen on risk for endometrial cancer in 87 323 women with breast cancer reported to the SEER (Surveillance, Epidemiology and End Results) Program in the United States. All women included in this study were diagnosed with early-stage (localized or regional) breast cancer between 1980 and 1992, were aged at least 50 years at diagnosis and had not been given chemotherapy as an initial treatment. For 14 358 women defined as the study group, the SEER database indicated that they had received hormonal therapy (which for over 90% was tamoxifen treatment). After a mean follow-up of [4.4] years, 73 cancers of the uterine corpus were observed, resulting in a standardized incidence ratio (SIR) (based on SEER incidence rates) of 2.0 (95% CI, 1.6–2.6). The SIR for women not known to have received hormones was significantly lower (1.2; 95% CI, 1.1–1.4). The differences in risk for endometrial cancer between hormone-treated women and women with no/unknown hormone treatment status were greater in five-year survivors (SIRs of 3.6 and 1.2, respectively). There was little difference in the severity of grade or stage of cancer of the uterine corpus according to initial therapy: in hormone-treated patients, 59% of the uterine cancers which developed were grade 1 or 2, 25% were grade 3 or 4 and for 16% the grade was unknown (versus 63%, 21% and 16% in no/unknown hormone treatment). The stage distribution was: localized, 78%; regional, 12%; distant, 4%; unknown, 6% (versus 76%, 11%, 8% and 5%, respectively). [The Working Group noted that no information on hysterectomy status was available and that misclassification of hormonal treatment in the study may have led to an underestimation of the difference in risk for cancer of the uterine corpus between the two groups.]

2.1.5. Randomized clinical trials (see Table 4)

Table 4

Endometrial cancers in patients treated for breast cancer: summary results of randomized clinical trials of adjuvant use of tamoxifen.

The design of randomized trials is such that they allow both known and unknown confounding factors to be distributed randomly between treatment groups. Therefore, these studies offer the best evidence regarding tamoxifen and the occurrence of second cancers. However, assessment of the risk for endometrial cancer was not the principal aim of the trials considered by the Working Group. None of the trials included procedures which would assure complete reporting of second primary tumours. Even in the trials (such as those in Scandinavia) that ascertained second primary tumours after linkage of the trial data to a population-based cancer registry, complete ascertainment of such tumours cannot be assumed, as second primary tumours are not reported systematically to cancer registries. However, there is no reason to believe that the reporting of such tumours would be biased by being related to tamoxifen therapy. Other biases which are relevant to randomized clinical trials are the same as those already discussed for cohort studies.

In a trial in Canada (Pritchard et al., 1987, updated by Nayfield et al., 1991), no case of endometrial cancer was found among 198 breast cancer patients treated with 20 mg/day tamoxifen for two years, but one occurred among 202 patients receiving no adjuvant therapy for breast cancer.

In a trial in the United States of tamoxifen therapy (20 mg/day) for two years for treatment of early breast cancer in women ≤ 75 years old versus no treatment, the NATO (Nolvadex Adjuvant Trial Organization) (1988) reported no case of endometrial cancer in 564 tamoxifen patients or in 567 patients receiving no further treatment following mastectomy during 1977–81 and followed up for eight years.

In a trial carried out in Denmark during 1975–78 with follow-up until 1988 of tamoxifen therapy (30 mg/day) for two years in women who were admitted for breast tumour (stage I, II and III), Palshof (1988) reported two endometrial cancers in 52 tamoxifen-treated postmenopausal patients (< 70 years old) and none in 52 patients in the placebo group. In premenopausal women, no case was found in either 112 tamoxifen-treated patients or 101 placebo patients.

In the IBCSG (International Breast Cancer Study Group) trial of tamoxifen (20 mg/day) plus low-dose prednisone (7.5 mg/day) therapy for one year, no endometrial cancer was observed in either 167 treated patients aged 66–80 years with operable breast cancer or 153 women randomized to observation (Castiglione et al., 1990). Subjects were entered into this trial during 1978–81 and were followed up for a median observation time of eight years.

A Danish study (Andersson et al., 1992) reported seven cases of endometrial cancer among 864 postmenopausal patients receiving radiotherapy and 30 mg tamoxifen for 48 weeks compared with two cases among 846 patients receiving radiotherapy alone. Eleven cases were reported among a third group of 1828 untreated, ‘low-risk’ patients. SIRs for endometrial cancer were computed from the incidence rates of the female Danish population. [The most valid comparison was between women with and without tamoxifen treatment in the radiotherapy group.] The SIRs for endometrial cancer were 1.9 (95% CI, 0.8–3.9) for patients who received radiotherapy and tamoxifen and 0.6 (95% CI, 0.1–2.1) for patients who received radiotherapy without tamoxifen (ratio of the SIRs, 3.3 (95% CI, 0.6–31).

After a maximal follow-up of 13 years, Ribeiro and Swindell (1992) reported one case of endometrial cancer among 282 postmenopausal patients ≤ 70 years old treated with tamoxifen (20 mg/day) for one year and one endometrial cancer among 306 untreated patients. There was no case of endometrial cancer among 199 premenopausal patients randomized to tamoxifen treatment or 174 patients randomized to irradiation-induced menopause.

In a trial in postmenopausal women, < 71 years old, in Sweden, of radiotherapy alone (236 patients), radiotherapy plus tamoxifen (30 mg/day) for one year (239 patients) or the same tamoxifen regimen alone (244 patients), two, four and five cancers of the corpus uteri were reported, respectively (Rydén et al. 1992). One endometrial cancer death was reported in the group receiving radiotherapy only. The median follow-up period was nine years.

In a Scottish trial of tamoxifen treatment (20 mg/day) for five or more years, Stewart and Knight (1989) found three uterine sarcomas in 539 tamoxifen-treated postmenopausal patients and two endometrial cancers in 531 untreated patients. In a subsequent set of 374 adjuvant tamoxifen-treated patients and 373 untreated patients, one endometrial cancer was observed in each group (Stewart, 1992). In another report of the same trial (Nayfield et al., 1991), four cases of endometrial cancer were reported among 661 tamoxifen-treated patients and two cases among 651 untreated patients.

The ECOG (Eastern Cooperation Oncology Group) in the United States open to accrual during 1978–82 (Cummings et al., 1993) found one case of endometrial cancer among 85 treated patients 65–84 years old (20 mg/day tamoxifen for two years) and another case among 83 placebo patients.

In a trial of the Cooperative Group for Chemohormonal Therapy of Early Breast Cancer (GROCTA) in Italy comparing tamoxifen (30 mg/day) for five years with six cycles of cyclophosphamide + methotrexate + 5-fluorouracil followed by four cycles of epidoxorubicin or a combination of the two, no case of endometrial cancer was seen among the 168 patients randomized to tamoxifen alone or among the 165 patients randomized to chemotherapy. One case was observed among 94 postmenopausal women who received both treatments. Women were 35–65 years old (Boccardo et al., 1994a).

In a large placebo-controlled trial, the National Surgical Adjuvant Breast and Bowel Project in Canada and the United States in 1982–88, Fisher et al. (1994) reported 15 cases of endometrial cancer among women with invasive breast cancer randomized to tamoxifen (20 mg/day) for five years compared with two cases among the placebo group. It was known at randomization that the proportion of women with hysterectomy was similar in the two groups. Both of the placebo patients with endometrial cancer had received tamoxifen for breast cancer relapse. One of the 15 patients who developed endometrial cancer was allocated to tamoxifen but never received it. One of the 15 endometrial cancers after review was found not to be a cancer. The annual hazard rate in the treated group was 1.6/1000 compared to 0.2/1000 in the placebo group. The latter rate was below that of the general population, from which about seven cancers would have been expected. The relative risk estimates were 7.5 (95% CI, 1.7–32.7) using the placebo control group and 2.2 [95% CI, 1.2–2.9] using population rates. The histological characteristics and grades of endometrial cancer were similar to those in patients who had not been treated with tamoxifen; 9/15 were grade 1 and 11 were stage I. Four patients allocated to tamoxifen died of endometrial cancer (among whom one never received the treatment). [The Working Group noted that the most valid comparison was between the group allocated to tamoxifen and the placebo control group. The rates used for calculating the expected numbers were from the United States SEER Program, whereas seven of the 12 major contributing centres were from Canada, where the rates for endometrial cancer are lower than in the United States.]

Kedar et al. (1994) in the United Kingdom recruited a randomized cohort of healthy postmenopausal women with a family history of breast cancer into groups given tamoxifen (20 mg/day) (61) or placebo (50) [duration of treatment not specified]. Fifty-five of the treated women had detectable serum levels of tamoxifen. No endometrial cancer was found in either group (median follow-up period, two years). [The Working Group noted the small numbers and short follow-up period.]

The SWOG (Southwest Oncology Group) in the United States (Rivkin et al., 1994) reported two cases of endometrial cancer among 295 postmenopausal patients treated with tamoxifen alone, three cases among 303 patients given tamoxifen and chemotherapy and none among 300 patients given chemotherapy alone during 1979–89.

Fornander et al. (1989) and Rutqvist et al. (1995) reported a study of endometrial cancer in the trial of treatment of postmenopausal women < 71 years old with tamoxifen (40 mg/day) for two or five years. ‘Low-risk’ patients (1774) and ‘high-risk’ patients (955) were randomized to tamoxifen or no tamoxifen; 678 of the ‘high-risk’ patients were also randomized to receive either radiotherapy or chemotherapy. Of the patients who received tamoxifen for two years, 809 were re-randomized to stop treatment or receive an additional three years of tamoxifen therapy. After a median follow-up time of nine years, 23 (one refused to take tamoxifen) endometrial cancers were found in 1372 patients randomized to tamoxifen versus four in 1357 untreated women (RR, 5.6; 95% CI, 1.9–16.2). Other corpus uteri cancers were reported: one in the tamoxifen-allocated group and three in the untreated group. A joint analysis of three Scandinavian trials (in Stockholm, Denmark (Andersson et al., 1992) and the south Sweden (Rydén et al., 1992)) presented in the same paper also showed a significant increase in the risk for endometrial cancer in tamoxifen-allocated patients (RR, 4.1; 95% CI, 1.9–8.9). Fornander et al. (1993) provided further details of 22 of the cases in the Stockholm trial, 17 of whom had actually received tamoxifen (out of 19 patients assigned to receive it). All cases were grade 1 or grade 2, and all but three were stage I. Endometrial cancer developed less than two years from the beginning of tamoxifen treatment in five of the 17 treated cases and after five years in four cases. Nine cases had received radiotherapy and two others had received chemotherapy. There were three deaths from endometrial cancer among the 17 cases receiving tamoxifen.

2.2.1. Case–control study

Cook et al. (1995) conducted the only case–control study that considered contralateral breast cancer. A total of 188 (18 receiving tamoxifen) < 85-year-old cases in Washington State, United States, were matched to 328 (58 receiving tamoxifen) controls without second primary cancer as described in Section 2.1.3. A 50% reduction in new breast tumours was observed for any use of tamoxifen (matched odds ratio, 0.5; 95% CI, 0.3–0.9) and an increased protection in women who used tamoxifen for more than one year (matched odds ratio, 0.4; 95% CI, 0.2–0.9). Odd ratios for any use of tamoxifen were somewhat larger in premenopausal than in postmenopausal women (0.7 versus 0.4), but were similar in pre- and postmenopausal women who used tamoxifen for more than one year (0.3 versus 0.4).

2.2.2. Cohort study

In the study by Curtis et al. (1996) (described in Section 2.1.4) of 87 323 breast cancer patients reported to the United States SEER Program, tamoxifen-treated patients had an SIR for contralateral breast cancer of 1.1 (95% CI, 1.0–1.3), compared with an SIR of 1.6 (95% CI, 1.5–1.7) for patients not known to have received tamoxifen. This represents a significant reduction [of approximately 30%] in the risk for contralateral breast cancer in tamoxifen-treated patients.

2.2.3. Randomized clinical trials

Results of trials of contralateral breast cancer following tamoxifen treatment are summarized in Table 5.

Table 5

Contralateral breast cancers: summary results of randomized clinical trials of adjuvant use of tamoxifen.

In Canada, Pritchard et al. (1987) (updated by Nayfield et al., 1991) found no difference in the risk of contralateral breast cancer: 3 cases in 198 tamoxifen-treated patients and 3 cases in 202 untreated patients.

In a trial of tamoxifen (20 mg/day) for two years versus no treatment, the NATO (Nolvadex Adjuvant Trial Organization) (1988) and Nayfield et al. (1991) reported 15 contralateral tumours in 564 tamoxifen-treated patients and 17 in 567 untreated patients.

A Danish study (Andersson et al., 1992) reported 8 cases of contralateral breast cancer occurring at least one year after the first primary cancer among 864 patients receiving radiotherapy and tamoxifen (30 mg/day) for 48 weeks compared with 10 cases among 846 patients receiving radiotherapy alone. A third group of 1828 untreated, ‘low-risk’ patients [with longer survival] experienced 10 cases in this period.

Baum et al. (1992), of the Cancer Research Campaign Breast Cancer Trials Group in the United Kingdom, in a 2 × 2 trial of 20 mg/day tamoxifen or perioperative cyclophosphamide found no overall effect of tamoxifen on contralateral breast tumours (RR, 0.9; 95% CI, 0.5–1.5). However, a reduction was seen in postmenopausal women (RR, 0.5; 95% CI, 0.2–1.1), but not in premenopausal women (RR, 1.4; 95% CI, 0.6–3.3).

After a maximal 13 years of follow-up, Ribeiro and Swindell (1992) reported seven cases of contralateral breast cancer in 282 postmenopausal patients treated with tamoxifen (20 mg/day) for one year and nine in 306 untreated patients.

In a trial in southern Sweden of radiotherapy alone (236 patients), radiotherapy plus tamoxifen (30 mg/day) for one year (239 patients) or the same tamoxifen regimen alone (244 patients), 15, 11 and 9 contralateral breast cancers were reported, respectively (Rydén et al., 1992).

In a Scottish trial of tamoxifen (20 mg/day) treatment for five or more years, Stewart (1992) found seven contralateral breast cancers in 374 tamoxifen-treated patients and 20 in 373 untreated patients.

The ECOG (Eastern Cooperative Oncology Group) group (Cummings et al., 1993) found one case of contralateral breast cancer among 85 patients treated with tamoxifen (20 mg/day) for two years and five among 83 untreated patients.

In a trial of the Cooperative Group for Chemohormonal Therapy of Early Breast Cancer (GROCTA) in Italy comparing tamoxifen (30 mg/day) for five years against chemotherapy or a combination of the two, four contralateral breast tumours were seen in the 165 patients randomized to chemotherapy, one contralateral tumour in the 171 patients who received both treatments and none in the 168 patients who received tamoxifen alone (Boccardo et al., 1994a).

In a placebo-controlled trial of tamoxifen (20 mg/day) for five years, Fisher et al. (1994) found 30 primary contralateral tumours in 1419 patients who received tamoxifen in the first five years of follow-up compared with 49 cases among 1424 patients receiving placebo. The reduction in incidence rate was 42%.

In a Swedish trial of treatment with tamoxifen (40 mg/day) for two or five years, Rutqvist et al. (1995) reported 40 contralateral tumours in 1372 women in the tamoxifen group and 66 in 1357 untreated patients. The risk ratio was 0.6 (95% CI, 0.4–0.9).

All of the studies mentioned above, and many other randomized trials, were included in an overview of data available up to 1990 on the occurrence of contralateral breast cancer in women allocated to tamoxifen treatment in trials (EBCTCG (Early Breast Cancer Trialists' Collaborative Group), 1992). In contrast to the analyses in some of the original reports, the life table method of analysis was used in this overview. Based on the occurrence of 122 contralateral breast cancers in 9128 tamoxifen-allocated women and 184 breast cancers in 9135 control patients, a 39% reduction in risk for contralateral breast cancer was observed (p < 0.00001). The effects were greater in trials using longer duration of tamoxifen treatment (reductions of risk were 26% for less than two years of tamoxifen treatment, 37% for two years of treatment and 53% for more than two years of treatment), although this was not statistically significant.

2.3.1. Case report

Johnstone et al. (1991) reported a single case of primary hepatocellular carcinoma in a woman who had received 20 mg tamoxifen daily and chemotherapy. The tumour developed six months after treatment began.

2.3.2. Cohort studies

There is probably some under-reporting of primary liver cancer in follow-up studies of breast cancer patients, because of confusion of these tumours with metastases from the breast.

Mühlemann et al. (1994) examined the incidence of hepatocellular cancer from 1974 to 1989 in white women aged 50 years or more with breast cancer in the United States and found no evidence of an increase after the introduction of tamoxifen in 1977.

Curtis et al. (1996) reported on liver cancer risk in 87 323 breast cancer patients reported to the United States SEER Program (see Section 2.1.4). In tamoxifen-treated patients, three cases of liver cancer were observed (SIR, 1.1; 95% CI, 0.2–3.2) and, in the larger group of patients not known to have received tamoxifen, eight cases were seen (SIR, 0.4; 95% CI, 0.2–0.7). [The Working Group noted that the low SIR for liver cancer in patients not known to have received tamoxifen may indicate underascertainment of liver cancers following breast cancer.]

2.3.2. Randomized clinical trials

In the Swedish trial (see p. 274), Rutqvist et al. (1995) reported three cases of hepatobiliary (two liver) cancer among subjects randomized to tamoxifen (40 mg/day) for two or five years and one case among untreated patients. Two of the three cases had been treated for 20 and 46 months, respectively (Rutqvist, 1993). Verification of the exposure was not reported for the third case.

The Danish study by Andersson et al. (1991) (see p. 271) reported one case of hepatobiliary cancer among 864 treated breast cancer patients (30 mg/day tamoxifen for 48 weeks + radiotherapy) and two cases among 846 radiation-treated breast cancer patients.

A study from southern Sweden (Rydén et al. 1992) reported two cases of hepatobiliary cancer among 244 patients who had received tamoxifen (30 mg/day) for one year, no case among 239 patients who had received tamoxifen plus radiotherapy and no case among 236 patients treated with radiotherapy alone.

No case of liver cancer was reported among 1419 tamoxifen-treated (20 mg/day for 5 years) patients and 1424 controls in the NSABP (National Surgical Adjuvant Breast and Bowel Project) B-14 trial (Fisher et al., 1994).

2.4.1. Case reports

Reports exist of one case of papillary mucinous endocervical adenocarcinoma, one in-situ and one invasive fallopian tube adenocarcinoma, one endometrioid carcinoma of the pelvic endometrium, one endometrioid carcinoma, one granulosa cell tumour of the ovary and one leiomyosarcoma of the myometrium uteri among women treated with tamoxifen (see Table 1).

2.4.2. Case–control study

Cook et al. (1995) reported on a population-based study of ovarian, endometrial and contralateral breast cancers following tamoxifen therapy of breast cancer patients. The results for endometrium and breast cancer have been reported above (pp. 267–268 and 274). For ovarian cancer, 34 cases (6 given tamoxifen) were compared with 89 controls who did not develop a second primary cancer, of whom 18 were given tamoxifen (matched odds ratio, 0.6; 95% CI, 0.2–1.8).

2.4.3. Cohort studies

Curtis et al. (1996) reported the risks for various second cancers in 87 323 breast cancer patients reported to the United States SEER Program (see p. 269). Between patients who had initial tamoxifen treatment and those not known to have received such treatment, there was no difference in the risks for ovarian cancer, digestive tract cancers or cancers at various other sites (other than endometrial cancer and contralateral breast cancer). The observed/expected ratios for all digestive system cancers combined were 1.0 (95% CI, 0.9–1.2) in tamoxifen-treated women and 0.9 (95% CI, 0.9–1.0) in women not known to have received tamoxifen treatment.

2.4.3. Randomized clinical trials

Two reports of randomized trials have been published which address the risks for second cancers other than those of the endometrium, breast and liver.

In a large trial (see p. 273), Fisher et al. (1994) found no significant difference between women allocated to tamoxifen and women allocated to placebo in the risk for second cancers of the colon, rectum, ovary or other sites.

In a joint analysis of three Scandinavian trials (see p. 274), Rutqvist et al. (1995) reported an excess risk for gastrointestinal cancer in tamoxifen-allocated women compared with those in untreated groups (RR, 1.9; 95% CI, 1.2–2.9). The relative risk for colorectal cancer was 1.9 (95% CI, 1.1–3.3) and that for stomach cancer was 3.2 (95% CI, 0.9–11.7). For other sites, no difference was observed between tamoxifen-allocated patients and untreated patients. [In view of the number of comparisons made, these results should be interpreted with caution.]

3.1.1. Mouse

In a study reported in a monograph, groups of 25 male and 25 female Alderley Park Strain 1 mice [age unspecified] were given 0 (control), 5 or 50 mg/kg bw tamoxifen [purity not specified] per day by gastric instillation for three months. The mice were then maintained for 12 months on a diet containing tamoxifen at concentrations to provide 0, 5 or 50 mg/kg bw tamoxifen, after which the experiment was terminated because of skeletal abnormalities in many of the exposed mice. Numbers surviving at 15 months were 16/25 control, 11/25 low-dose and 17/25 high-dose males and 15/25 control, 17/25 low-dose and 12/25 high-dose females. In males, interstitial cell tumours of the testes were found in 0/25 control, 2/25 low-dose and 21/25 high-dose animals. In females, granulosa-cell adenomas of the ovary were found in 0/25 control, 9/25 low-dose and 9/25 high-dose animals. Two other studies at lower doses were briefly described (Tucker et al., 1984). [The Working Group noted that the descriptions of these lower-dose studies did not provide sufficient information for evaluation.]

3.1.2. Rat

Groups of 51 male and 52 female Alderley Park Wistar-derived rats, five weeks of age, were given 5, 20 or 35 mg/kg bw tamoxifen [purity not specified] per day by gastric instillation in 0.5% hydroxypropyl methylcellulose for two years. A control group of 102 male and 104 female rats was given the vehicle alone. Moribund animals and those surviving to the end of the exposure period were killed and subjected to necropsy; all major tissues were examined histologically. Growth was reduced (by about 30% in females and 40% in males) in all tamoxifen-treated groups compared with controls. Survival was reduced in the groups given the two higher doses but increased in those given the lower dose. The reduced survival was attributed to early deaths from liver tumours and resulted in termination of the study at 87 weeks for the mid-dose group and at 71 weeks for the high-dose group. Hepatocellular adenomas occurred in 1/102 control, 8/51 low-dose, 11/51 mid-dose and 8/51 high-dose males and in 1/104 control, 2/52 low-dose, 6/52 mid-dose and 9/52 high-dose females (p < 0.0001 for trend). Hepatocellular carcinomas were found in 1/102 control, 8/51 low-dose, 34/51 mid-dose and 34/51 high-dose males and in 0/104 control, 6/52 low-dose, 37/52 mid-dose and 37/52 high-dose females [p < 0.001]. Hepato/cholangiocellular carcinomas were found in 0/102 control, 0/51 low-dose, 2/51 mid-dose and 5/51 high-dose males and in 0/104 control, 0/52 low-dose, 4/52 mid-dose and 5/52 high-dose females (p < 0.0001 for trend). No increase in the incidence of tumours was observed at any other site. Significant decreases in tumour incidence were observed in the pituitary and parathyroid glands of males and in the pituitary and mammary glands of females (see Table 6) (Greaves et al., 1993). [The Working Group noted that part of the reduction in tumour rates may have been related to decreased body-weight gain.]

Table 6

Incidence (%) of tumours in Alderley Park Wistar-derived rats exposed to tamoxifen.

In a study that also evaluated toremifene, groups of 57, 84 and 75 female Sprague-Dawley [Crl:CD(BR)] rats, approximately six weeks of age, were given 0, 11.3 and 22.6 mg/kg bw tamoxifen (99% pure) per day by gastric instillation in 0.5% carboxymethylcellulose on seven days per week for up to 12 months followed by a three-month recovery period. Nine rats from each group were killed at three and six months. At 12 months, 18 control, 36 low-dose and 24 high-dose rats were killed. The experiment was terminated at 15 months. All rats, including those found dead or moribund, were subjected to necropsy; organs examined histopathologically included liver, ovaries, uterus, mammary gland, adrenal glands, kidneys, tail bone, sternum, brain and pituitary. Weight gain in both groups receiving tamoxifen was less than that in controls. In the group killed at 15 months, uterine weights were significantly reduced in both high- and low-dose treated groups. A significant increase in the incidence of hepatocellular carcinomas occurred in both treated groups, compared with controls. At 12 months, the incidence of hepatocellular carcinomas was 0/18 control, 16/36 low-dose and 24/24 high-dose animals (p < 0.001); that at 15 months was 0/13, 13/21 and 8/9, respectively (p< 0.001). The incidence of hyperplasia in the mammary gland and the incidence of pituitary adenomas were decreased in tamoxifen-treated animals (Hard et al., 1993). [The Working Group noted the small numbers of animals, that the exposure was for one year and that the study was terminated at 15 months.]

In a study that also included toremifene, groups of 20 female Sprague-Dawley rats, six weeks of age, were given 0, 11.3 or 45 (maximum tolerated dose, MTD) mg/kg bw tamoxifen (purity > 99%) in carboxymethylcellulose per day by gastric instillation on seven days per week for up to one year. Five animals from each group were killed after 26 weeks and 52 weeks of treatment; all surviving rats were killed 65 weeks after the beginning of treatment. Weight gain was reduced in both tamoxifen-treated groups. No tumour was found in animals killed at 26 weeks. At 52 weeks, the incidences of hepatocellular carcinomas were 0/5 control, 0/5 low-dose and 3/5 high-dose animals; those at 65 weeks were 0/8, 0/8 and 5/6, respectively. Histopathological findings in tissues other than the liver were not reported (Hirsimäki et al., 1993). [The Working Group noted the small numbers of animals, that the exposure was for one year and that the study was terminated at 65 weeks.]

Groups of 55–57 female Sprague-Dawley [Crl:CD(BR) Charles River] rats, six weeks of age, were given 0, 2.8, 11.3 or 45.2 mg/kg bw tamoxifen [purity not specified] per day by gastric instillation in 0.5% carboxymethylcellulose on seven days per week for up to 12 months, followed by a three-month recovery period. Mortality was 5.3% in control, 9.0% in low-dose, 1.7% in mid-dose and 40% in high-dose animals. All tamoxifen-treated animals had weight gain depression and some developed alopecia. Among the seven high-dose rats examined at six months, five had adenomas and two had carcinomas. At 12 months of exposure, the incidences of hepatic adenomas were 5/10 mid-dose and 2/4 high-dose rats. The incidence of liver carcinomas in these groups was 1/10 and 3/4, respectively. During the three months of recovery, the occurrence of carcinomas in the mid-dose group increased to 5/11. No liver tumour was seen in control or low-dose groups (Williams et al., 1993). [The Working Group noted that exposure was for only one year and that the study was terminated at 15 months.]

In a study that included toremifene, groups of five female Sprague-Dawley rats, six weeks of age, were given 0, 11.3 or 45 (MTD) mg/kg bw tamoxifen (purity > 99%) per day by gastric instillation in 0.5% carboxymethylcellulose seven days per week for 12 months. Groups were killed at 12 months or after a further 13 weeks of recovery. The incidence of liver tumours (hepatocellular carcinomas) at 12 months was 0/5 control, 0/5 low-dose and 4/5 high-dose animals; that at 65 weeks was 0/5, 0/5 and 5/5, respectively (Ahotupa et al., 1994). [The Working Group noted the small number of animals and that the exposure was for only one year.]

In a study that also included droloxifene, groups of 50 male and 50 female rats [strain and age not specified] were given to 0 (control), placebo [not specified] or 36 mg/kg bw tamoxifen [purity not specified] in the diet for 24 months. The incidences of hepatocellular carcinomas were: males — control, 0/50; placebo, 0/50; tamoxifen, 49/50; females — control, 0/50; placebo, 0/50; tamoxifen, 50/50 (Hasmann et al., 1994). [The Working Group noted the lack of experimental detail]

Groups of 10 female Fischer (344/Tox), Wistar (LAC-P) and Lewis (LEW Oka) rats, six weeks of age, were fed either basal diet or diets containing 420 mg/kg diet (ppm) tamoxifen (purity > 98%) until 50% mortality was reached, at which time all surviving animals were killed. In groups of five rats killed after 90 days of exposure, the incidence of hepatocellular altered foci was increased in the Wistar and Lewis animals (Table 7). In rats killed at 180 days, no liver tumour was present in controls, while, in treated rats, the incidence was 3/5 in Wistar, 1/5 in Lewis and 0/5 in Fischer rats. By 11 months, 50% of the tamoxifen-treated Wistar and Lewis rats developed palpable liver nodules or were in ill health and these animals were killed. In the Wistar and Lewis rats killed at or before 11 months, all 10 had multiple liver tumours of which one or more was a carcinoma. In Fischer rats, 50% mortality was reached at 20 months and the remaining animals were killed at this time. All 10 rats exhibited at least one hepatocellular carcinoma (Carthew et al., 1995a).

Table 7

Hepatocellular altered foci in tamoxifen-exposed rats at 90 days of treatment.

In a compilation of three experiments also including toremifene, reported in the proceedings of a meeting, groups of 109, 25 or 104 female Sprague-Dawley rats [age not specified] were given tamoxifen (> 99% pure) at 0, 11.3 or 45.0 mg/kg bw per day 7 days per week, for 20, 26 or 52 weeks with recovery periods of 12 or 13 weeks. In the high-dose group, squamous-cell metaplasia of the endometrium was found in 10 rats, dysplasia with metaplasia in three and squamous-cell carcinoma in two. The carcinomas were found after 20 or 26 weeks of dosing and a recovery period of 12–13 weeks. Among control or low-dose animals, no lesion of the uterus was reported (Mäntylä et al., 1995, 1996). [The Working Group noted the lack of study details.]

As part of a tumour-promotion study in which toremifene was also included, groups of 14–22 female Sprague-Dawley rats, weighing 130 ± 10 g [age not specified], were subjected to partial hepatectomy and two weeks later were exposed to tamoxifen [purity not specified] in the diet at concentrations of 0 (control), 250 or 500 mg/kg diet (ppm) for up to 18 months. Both exposures to tamoxifen suppressed body-weight gain. In animals killed at six months, uterine weights were reduced. At this time, increases were found in both the number and volume of hepatocellular altered foci identified by staining for γ-glutamyl transpeptidase. Incidences of hepatic neoplastic nodules were not increased. In the remaining animals killed at 18 months, hepatocellular carcinomas were found in 0/22 controls, 1/15 low-dose and 8/15 high-dose rats (Dragan et al., 1995).

3.2.1. Mouse

In a study using a strain of mouse with a high incidence of mammary tumours, groups of female C3H/OUJ mice, received a subcutaneous implantation of a silastic capsule containing 28 mg tamoxifen (release, 125 µg/day for at least six months) [purity not specified] on the back at two weeks (11 mice) or at five weeks (15 mice) after a pregnancy/weaning cycle (3.5 months of age). A control group of 11 mice received implantations of a placebo silastic capsule at the start of the experiment. After 15 months, the percentage of mice with mammary tumours was 100% in the controls, about 20% in the group exposed to tamoxifen at two weeks and about 50% in the group exposed to tamoxifen at five weeks (Jordan et al., 1990). [The Working Group noted that only the uterus and mammary glands were examined and only macroscopically.]

In a study using a strain of mouse with a high incidence of mammary tumours, two groups of 30 female C3H/OUJ mice, 2.5 months of age, were ovariectomized and two further groups of 30 females were left intact. Two weeks later, one ovariectomized and one intact group received an implantation of a silastic capsule containing 28 mg tamoxifen (release, approximately 125 μg/day for six months) [purity not specified]. The capsules were replaced every six months up to 17 months of treatment. Mice were observed up to 27 months of age. In the intact controls, the mammary tumour incidence was 100% by about 20 months of age. This was reduced to about 20% by tamoxifen. In ovariectomized mice, the incidence of mammary tumours was 50%, which was reduced to about 20% by tamoxifen. In a further experiment, four groups of 20 female C3H/OUJ mice, three months old, received a tamoxifen capsule implant for three, six or 12 months (capsule replaced at six months in this group). The incidence of mammary tumours in controls was 100% by 17 months, whereas all tamoxifen-treated groups had an incidence of about 25% (Jordan et al., 1991). [The Working Group noted that only the mammary gland was examined and only macroscopically.]

3.2.2. Rat

As part of a tumour-promotion study, groups of 10 female Sprague-Dawley rats, weighing 125–175 g [age not specified], were subjected to partial hepatectomy and, one week later, received subcutaneous implants of time-release tablets providing 0 or 50 µg/day tamoxifen. At four months, the body weights of treated rats were significantly lower than those of controls. The livers were examined for γ-glutamyl transpeptidase-positive altered hepatocellular foci. Controls had 0.07 ± 0.08 foci/cm², whereas tamoxifen-exposed rats had 1.11 ± 0.25 foci/cm² (Yager et al., 1986).

3.3.1. Mouse

Groups of 25–70 female Swiss mice, six to eight weeks of age, were ovariectomized or left intact and were divided into seven groups, one of which was left untreated; the others received a beeswax-impregnated thread inserted into the canal of the uterine cervix; subcutaneous injection of olive oil thrice weekly; an intracervical insertion of thread impregnated with 3-methylcholanthrene (MCA); an MCA-impregnated thread plus subcutaneous injections of 50 µg/kg bw tamoxifen in olive oil three times per week; a beeswax thread plus tamoxifen; or tamoxifen alone. The experiment lasted for at least 393 days. Vaginocervical smears were taken to monitor cervical dysplasia and carcinoma. In intact mice, the incidence of cervical carcinoma in mice exposed to MCA was 60 ± 8.2%, while, in mice receiving MCA and tamoxifen, it was 30 ± 6.8%. No other intact mouse developed a carcinoma. In ovariectomized mice, the incidence of cervical carcinoma in the MCA-treated group was 62 ± 7.2% and that in the mice also receiving tamoxifen was 27 ± 6.7%. No other ovariectomized mouse developed a carcinoma (Sengupta et al., 1991).

Groups of four to six male and female BALB/c/Bln mice, 8–12 weeks of age, were given intraperitoneal injections of diluted serum from leukaemic mice infected with the Rauscher murine leukaemia virus. Starting one day later, mice were given intraperitoneal injections of 0.2, 0.5 or 1 mg/animal tamoxifen [purity not specified] in dimethyl sulfoxide three times a week for three weeks (total doses, 1.6, 4 and 8 mg). At the end of treatment, the mice were killed and spleens were weighed as an index of leukaemic disease. Compared to the controls, the spleen weights of the tamoxifen-treated animals were about 60%, 45% and 20% in the low-, mid- and high-dose groups, respectively, indicative of reduced leukaemic activity (Sydow & Wunderlich, 1994).

3.3.2. Rat

A group of 36 male Fischer rats, approximately three months of age, was given a single gastric instillation of 100 mg/kg bw N-nitrosodiethylamine (NDEA), then fed a diet containing 200 mg/kg diet (ppm) 2-acetylaminofluorene (2-AAF) [duration not specified] and underwent partial hepatectomy. Six weeks after cessation of 2-AAF treatment, one group of 17 rats was maintained with no further treatment. The remaining rats were divided into three groups of seven, five and five animals that were given subcutaneous injections of 0.25, 1.0 and 2.5 mg/animal tamoxifen [purity not specified] in peanut oil twice a week. The experiment was terminated at 10 months. From graphic presentations, the incidence of liver malignancy [not further specified] was about 75% in controls and 10% in the low-dose, 25% in the mid-dose and 35% in the high-dose tamoxifen-treated groups (Mishkin et al., 1985).

Groups of 7–15 female Sprague-Dawley rats, 50 ± 2 days of age, were given 12 mg 7,12-dimethylbenz[α]anthracene (DMBA) per animal as a single gastric instillation in sesame oil. After six weeks, when mammary tumours had reached about 1 cm in diameter, groups were treated with 0, 1.0, 3.0 or 7.5 mg/kg bw tamoxifen (> 98% pure) per day by gastric instillation for at least five weeks. The numbers of new tumours per animal were 3.0 ± 2.6 in controls and 2.0 ± 0.6 in low-dose, 2.1 ± 1.4 in mid-dose and 0.3 ± 0.5 in high-dose tamoxifen-treated groups (Kangas et al., 1986) [The Working Group noted the small numbers of animals.]

Groups of 10 female Sprague-Dawley rats, weighing 125–175 g [age not specified], were given 25 mg/kg bw NDEA by intraperitoneal injection 24 h after partial hepatectomy (Yager & Shi, 1991). One week later, the animals received implants of a time-release tablet containing quantities of tamoxifen to give a daily release of 15 or 50 µg or no tamoxifen. At four months, liver samples were examined for γ-glutamyl transpeptidase-positive altered hepatocellular foci. The body weights of treated rats were substantially lower than those of controls. The incidence of altered foci per cm² was 0.9 ± 0.3 in controls, 7.1 ± 1.9 in low-dose and 4.9 ± 1.0 in high-dose tamoxifen-treated rats (Yager et al., 1986).

Male Sprague-Dawley rats, eight weeks of age, were given 200 mg/kg bw NDEA by intraperitoneal injection and were allocated to four groups two weeks later: four rats were fed olive oil in the diet; 12 rats were fed 1 mg/animal tamoxifen (analytical grade) in the diet daily; eight rats were fed 0.5 mg/animal diethylstilboestrol (DES) in olive oil in the diet daily; and 11 rats were fed both DES and tamoxifen. At eight months, the incidence of altered foci in the liver was quantified using γ-glutamyl transpeptidase as a marker. The numbers of foci per cm² [derived from graphic presentations] were about 10 in the NDEA-treated rats, about 20 in the NDEA/tamoxifen-treated rats, about 16 in the NDEA/DES-treated rats and about 22 in the NDEA/DES/tamoxifen-treated rats. The lesions in the NDEA/tamoxifen-treated group were larger than those in the group given NDEA alone, whereas, when tamoxifen was given with NDEA/DES, the lesions were smaller than with NDEA/DES alone (Kohigashi et al., 1988).

Groups of 20 female Sprague-Dawley rats, 50 days of age, were given 20 mg/animal DMBA in peanut oil by gastric instillation. Twenty-eight days later, one group was treated with 200 µg/animal tamoxifen per day by gastric instillation, the other one with peanut oil. At 100 days after DMBA administration, 75% of the rats given DMBA alone had developed mammary tumours, whereas, in the group also receiving tamoxifen, less than 20% had tumours. Cessation of tamoxifen treatment after four months led to the development of mammary tumours, so that by 4.5 months about 70% of animals had tumours (Robinson et al., 1988).

Groups of 12 female Sprague-Dawley rats, weighing 140–160 g [age not specified], and 12 female Fischer 344 rats, weighing 120–140 g [age not specified], were allocated to one of two protocols. To evaluate initiating activity, groups of Sprague-Dawley rats either remained untreated as controls or were fed diets containing 10 mg/kg diet (ppm) ethinyloestradiol, 400 ppm tamoxifen or ethinyloestradiol plus tamoxifen for six weeks. After seven days, all rats underwent partial hepatectomy and, on day 49 after a one-week recovery phase on basal diet, were fed a diet containing 200 ppm 2-AAF for two weeks; at the midpoint, they were given 1 mL [1.6 mg]/kg bw carbon tetrachloride by gastric instillation in corn oil. In order to evaluate promoting activity, groups of 12 Sprague-Dawley and 12 Fischer 344 rats were fed a diet containing 200 ppm 2-AAF for two weeks and, on day 7, were given 1 mL/kg bw carbon tetrachloride by gastric instillation in corn oil. On day 21, after a one-week recovery period, these groups were maintained as untreated controls or were fed diets containing 10 ppm ethinyloestradiol, 400 ppm tamoxifen or ethinyloestradiol plus tamoxifen for six weeks; on day 28 animals were subjected to partial hepatectomy. All treated rats showed reduced weight gain. Liver samples were examined for γ-glutamyl transpeptidase-positive altered hepatocellular foci. After the initiation protocol, the numbers of foci per cm² were 15.6 ± 6.5 in controls, 67.4 ± 10 in tamoxifen-treated rats, 68.2 ± 13.9 in ethinyloestradiol-treated rats and 140.7 ± 21.2 in ethinyloestradiol/tamoxifen-treated rats. In the promotion protocol, the numbers per cm² were 25.9 ± 6.9, 72.0 ± 13.0, 56 ± 10.4 and 98.5 ± 18.4 in the respective groups (Ghia & Mereto, 1989).

Thirty female Fischer 344 rats, weighing 130–200 g [age not specified], were subjected to partial hepatectomy and 24 h later were given 10 mg/kg bw NDEA by gastric instillation. After a two-week recovery period, nine rats were maintained on basal diet (controls), 11 rats were fed 250 mg/kg diet (ppm) tamoxifen [purity not specified] and 10 were fed 500 ppm tamoxifen in a semipurified diet for six months. At termination, livers were examined for altered hepatocellular foci with the help of a variety of histochemical markers. The numbers of foci per liver were 90 ± 30 in controls, and 5430 ± 310 in low-dose and 7280 ± 490 in high-dose tamoxifen-treated rats (Dragan et al., 1991a).

Groups of 6–12 female Fischer 344 rats weighing 125–150 g [age not specified] were subjected to partial hepatectomy and 24 h later were given 40 mg/kg bw tamoxifen [purity not specified] by gastric instillation. The animals were then maintained on a basal diet or a diet containing 500 mg/kg diet (ppm) phénobarbital. After six months, all rats were killed and slices of the three main lobes of the liver were used for enzymic histochemical demonstration of altered hepatocellular foci with the help of a variety of histochemical markers. The numbers of foci per liver were 100 ± 50 in controls, 130 ± 50 in those given tamoxifen alone and 370 ± 130 in those given tamoxifen followed by phénobarbital (Dragan et al., 1991b).

Three groups of 10 female Sprague-Dawley rats, seven weeks of age, were given 20 mg DMBA by gastric instillation in sesame oil; starting 92 days later, two of the groups were given 1.0 or 10.0 mg/kg bw tamoxifen by gastric instillation in 0.5% methylcellulose for seven days, while the third group received no further treatment. At 20 weeks, mammary tumours were found in 8/9 controls, 5/10 low-dose tamoxifen-treated rats and 7/9 high-dose tamoxifen-treated rats (Kawamura et al., 1991).

A group of 10 female Sprague-Dawley rats, weighing 200 g [age not specified], was given 0.5 mg/rat tamoxifen in peanut oil twice at a five-day interval by subcutaneous injection; one day after the last injection, the animals underwent partial hepatectomy and 24 h later received 10 mg/kg bw NDEA by gastric instillation. Another group of 10 rats was subjected to partial hepatectomy and NDEA treatment only (controls). At six weeks, the severity of altered hepatocellular foci [from graphic presentations] was graded as 1.8 in controls and 0 in tamoxifen-treated rats (Oredipe et al., 1992). [The Working Group noted that the evaluation of foci was semi-quantitative.]

Groups of female Sprague-Dawley rats [numbers not specified], 59 days of age, were given 5 mg/rat DMBA weekly by gastric instillation for four weeks. One group was also simultaneously given 10 mg/kg bw tamoxifen weekly by subcutaneous injection for four weeks. After 14 weeks, the incidence of mammary tumours was 100% and the multiplicity was 6.7 ± 0.8 tumours per animal in the DMBA-treated group. In the DMBA/-tamoxifen-treated group, the incidence was 49.5% and the multiplicity was 1.4 ± 0.2 tumours per animal. In a second experiment, two groups of rats were treated with DMBA by gastric instillation for four weeks. Beginning at week 9, one group was given 1 mg/kg bw tamoxifen twice daily by subcutaneous injection for three weeks. At 12 weeks, all animals were ovariectomized. A large percentage of tumours regressed. In the control group, 1.3 new tumours per rat appeared, whereas in the tamoxifen-treated rats 0.3 new tumours per rat developed. All new tumours that appeared in tamoxifen-treated rats were hormone-independent, whereas in controls, only 13% continued to grow (Fendl & Zimniski, 1992). It was subsequently reported that, following the cessation of tamoxifen treatment, some tumours resumed growth (Zimniski & Warren, 1993).

A group of 20 female Sprague-Dawley rats, 50 days of age, was given 50 mg/kg bw N-methyl-N-nitrosourea (MNU) as a single intravenous injection. When at least one mammary tumour had reached a diameter of 10 mm, the animals were given 6 mg/kg bw tamoxifen by gastric instillation on five days a week for four weeks. A positive control group of 50 female rats received the treatment with MNU alone. At the cessation of treatment (207 days), the multiplicity of mammary tumours was 1.1 in controls (median size, 10.2 cm³) and 1.25 in the tamoxifen-treated group (median size, 2.5 cm³) (Winterfeld et al., 1992).

Groups of female Sprague-Dawley rats, weighing 200–220 g [age not specified], received no treatment (control; 26 rats) or 1 mg/kg bw tamoxifen in propylene glycol by intraperitoneal injection 1 h (four rats) or 24 h (three rats) before a single intraperitoneal injection of 50 mg/kg bw NDEA. Other groups were subjected to partial hepatectomy (eight rats) or partial hepatectomy and 1 mg/kg bw tamoxifen (eight rats) 24 h before treatment with NDEA. Eight weeks after treatment, rats were killed and liver samples were examined for altered hepatocellular foci identified by glutathione S-transferase immunohistochemistry. Pretreatment for 1 h with tamoxifen increased the number of foci almost 2-fold and pretreatment for 24 h increased the number 3.4-fold (Servais & Galand, 1993).

Two groups of female OFA rats [number and age not specified], weighing 130–160 g, were given 20 mg/animal DMBA in sesame oil as a single gastric instillation. After seven weeks, one group was given 1 mg/kg bw tamoxifen in sesame oil or sesame oil alone by subcutaneous injection three times per week for six weeks. At 11 weeks, the mean number of mammary tumours in controls was 9.3 tumours per rat; in tamoxifen-treated rats, this was reduced by 17.2 ± 2.8% (Weckbecker et al., 1994).

Groups of 34–35 male Sprague-Dawley rats, aged 25 days, were given 20 mg/kg bw 1,2-dimethylhydrazine weekly by subcutaneous injection for 20 weeks and were fed control diet or diets initially containing 4 mg/kg diet (ppm) tamoxifen [purity not specified], which was reduced to 2 ppm at 29 days, 1 ppm at 56 days and to 0.5 ppm at 65 days. Most animals were killed 65 days after the last injection of 1,2-dimethylhydrazine. Animals receiving tamoxifen displayed reduced body weights. Mortality was comparable in control and treated animals. The total number of colon adenocarcinomas and their distribution in the proximal and distal portions did not differ between the groups (Gershbein, 1994).

Groups of 16–20 female Fischer 344 rats, 130–150 g [age not specified], were subjected to partial hepatectomy and 24 h later were given 10 mg/kg bw NDEA in tricaprilyn or vehicle alone by gastric instillation. Two weeks later, groups of NDEA- or vehicle-exposed rats were fed 250 mg/kg diet (ppm) tamoxifen free base [purity not specified] in a semi-purified diet. Treatment with tamoxifen reduced body weights by 16–24%. Tamoxifen increased the incidence of liver tumours at 15 months in rats given NDEA (3/8 hepatocellular carcinomas versus 0/6 in rats given NDEA alone) (Dragan et al., 1994).

In a study that also included toremifene, virgin Sprague-Dawley rats, aged 43 days, were randomized into groups of 20 and allocated to control diet or to diets containing 0.2 mg/kg diet (ppm) tamoxifen. Seven days later, groups were given either 50 mg/kg bw MNU or saline by intravenous injection. Animals were killed when moribund and the experiment was terminated at 180 days after MNU treatment. Tamoxifen did not affect the incidence, multiplicity or latency of mammary tumours compared with controls (Moon et al., 1994). [The Working Group noted that the dose of MNU, producing 100% of tumours in the MNU- and MNU/tamoxifen-treated groups, may have been too high to allow the detection of a protective effect.]

In a tumour promotion study, in which toremifene was also studied, groups of 14–22 female Sprague-Dawley rats, weighing 130 ± 10 g [age not specified], were subjected to partial hepatectomy and 24 h later were given a single dose of 10 mg/kg bw NDEA in trioctanoin by gastric instillation. Two weeks later, the rats were given either basal diet or diet containing 250 or 500 mg/kg diet (ppm) tamoxifen for 18 months. All exposures to tamoxifen suppressed body-weight gain. Uterine weights were suppressed at six months. The number of altered hepatic foci identified by any of four histochemical markers as well as the volume of liver occupied by foci was increased by tamoxifen. At six months, the incidence of hepatic neoplastic nodules was: controls, 8/15; low-dose, 13/15; and high-dose, 11/15. At 18 months, the incidence of hepatocellular carcinomas in all groups approached 100% and the incidence of renal carcinomas was slightly increased by tamoxifen [p = 0.008, Cochran-Armitage trend test] (Table 8) (Dragan et al., 1995).

Table 8

Incidence of tumours in female Sprague-Dawley rats exposed to tamoxifen after NDEA.

3.3.3. Hamster

Groups of 7–12 male Syrian hamsters, four to six weeks of age, received subcutaneous implants of one pellet containing either 25 mg 17β-oestradiol or 25 mg tamoxifen, or one pellet of each. All animals were killed after seven months. The number of kidney tumour-bearing animals was 3/3 oestradiol-treated hamsters examined, 2/8 oestradiol/tamoxifen-treated hamsters and 0/8 tamoxifen-treated hamsters. The multiplicity of kidney tumours was reduced from 6.5 with oestradiol to 1.0 in oestradiol/-tamoxifen animals; tamoxifen alone produced no kidney tumour (Liehr et al., 1988).

Two groups of 5–20 male and female Armenian hamsters (Cricetulus migratorius), two to three months of age, received subcutaneous implants of 36-mg pellets of zeranol on study days 0 and 94. One group was given 5 mg/animal of tamoxifen by subcutaneous injection twice a week up to day 32 and then once a week until day 202, when the experiment was terminated. Among 9 males and 7 females receiving zeranol only, the incidence of hepatocellular carcinomas was about 60%, whereas in four males and four females also receiving tamoxifen, the incidence was reduced to 1–2% (Coe et al., 1992).

4.1.1. Humans

The absorption, distribution, metabolism and excretion of tamoxifen have been reviewed extensively (Furr & Jordan, 1984; Buckley & Goa, 1989; Wiseman, 1994). All studies have involved oral administration of tamoxifen citrate unless stated otherwise.

Tamoxifen is well absorbed after oral administration and appears to be more than 99% bound to plasma proteins (mostly to albumin) (Lien et al., 1989). Tamoxifen absorption shows wide interindividual variation, which is probably due to differences in liver metabolism and differences in absorption in the gastrointestinal tract. Administration of 40 mg/day for two months to patients with breast cancer produced steady-state mean plasma concentrations of tamoxifen of 186–214 ng/mL. The single maximum plasma concentrations in this study were about 70 ng/mL tamoxifen and about 20 or 40 ng/mL N-desmethyltamoxifen (the different values being with Tamoplex® and Nolvadex®, respectively) (McVie et al., 1986). Administration of a single 20-mg dose of tamoxifen to six male volunteers resulted in peak plasma concentrations of 42 ng/mL tamoxifen and 12 ng/mL N-desmethyltamoxifen. Maximal levels were achieved approximately 5 h after administration and area under the curve (AUC) was 2606 ng × h/mL (Adam et al., 1980). The distribution half-life (i.e. initial t_½) of tamoxifen is 7–14 h. The mean terminal half-life was 111 h, somewhat shorter than the previously reported seven days or more (Fromson et al., 1973a). Steady-state concentrations of tamoxifen and of N-desmethyltamoxifen were reached after three to four weeks of 20 mg b.i.d. (40 mg/day) administration (McVie et al., 1986) and after 4–8 weeks of 20 mg/day administration (Lien et al., 1995).

The apparent volume of distribution for tamoxifen in humans is 50–60 L/kg (Lien et al., 1989), indicating that most of the drug (99.9%) is present in peripheral compartments, which is suggestive of extensive tissue binding (Lien et al., 1991).

In healthy male volunteers given 40 mg tamoxifen, the plasma elimination half-life of tamoxifen during the first day was 10 h. However, after 34 h, appreciable levels of tamoxifen and N-desmethyltamoxifen were still present, suggesting a lengthening of half-life with increasing study duration or the existence of multiple half-lives (Guelen et al., 1987). The pharmacokinetics of tamoxifen appear to be biphasic, with a distribution phase of 7–14 h and an elimination phase of about seven days (Fromson et al., 1973a). The elimination half-life of 7V-desmethyltamoxifen is around seven days and 4-hydroxytamoxifen has a shorter half-life than tamoxifen (Buckley & Goa, 1989).

Tamoxifen and its metabolites are mostly excreted via bile into faeces as glucuronides and other conjugates. Urinary excretion is a very minor route of elimination (Furr & Jordan, 1984).

In seven premenopausal breast cancer patients and nine postmenopausal women with non-neoplastic diseases treated with tamoxifen for 56 days, the serum levels of N-dides-methyltamoxifen were higher in the postmenopausal women (p < 0.02). A similar trend was observed for N-desmethyltamoxifen (p < 0.06) (Lien et al., 1995).

Administration of 40 mg/day tamoxifen (20 mg twice a day) to primary breast cancer patients for 15–940 days (Daniel et al., 1981) resulted in plasma concentrations of 27–520 (mean, 300) ng/mL tamoxifen, 210–761 (mean, 462) ng/mL N-desmethyltamoxifen and 2.8–11.4 (mean, 6.7) ng/mL 4-hydroxytamoxifen. Concurrent tumour biopsy concentrations were 5.4–117 (mean, 25.1) ng tamoxifen/mg protein, 7.8–210 (mean, 52) ng N-desmethyltamoxifen/mg protein and 0.29–1.13 (mean, 0.53) ng 4-hydroxytamoxifen/mg protein.

After 14 daily doses of 40 mg tamoxifen, concentrations in plasma and in breast tumour cell nuclear and cytosolic fractions were measured in three patients (Murphy et al., 1987). The results are shown in Table 9.

Table 9

Concentrations of tamoxifen and its metabolites in plasma and breast tumour cells of three patients receiving 40 mg/day for 14 days.

A number of metabolites (see Figure 1) have been identified in urine and plasma of human breast cancer patients by LC/MS/MS techniques. Plasma extracts contained tamoxifen, N-desmethyltamoxifen and tamoxifen-N-oxide (Poon et al., 1993). Glucuronides of four hydroxylated metabolites (4-hydroxytamoxifen, 4-hydroxy-N-desmethyltamoxifen, dihydroxytamoxifen and another monohydroxy- (possibly α-hydroxy-) N-desmethyltamoxifen) were detected in the patients' urine. In a more recent study, seven metabolites were identified in plasma (N-didesmethyltamoxifen, α-hydroxytamoxifen, 4-hydroxytamoxifen, tamoxifen-N-oxide, α-hydroxy-N-desmethyltamoxifen, 4-hydroxy-N-desmethyltamoxifen and 4-hydroxytamoxifen-N-oxide) (Poon et al., 1995).

Figure 1

Postulated metabolic pathways of tamoxifen

In biopsy and autopsy samples taken from 14 patients, levels of tamoxifen and its metabolites (N-desmethyl, N-didesmethyl, 4-hydroxy and 4-hydroxy-N-desmethyl) were 10- to 60-fold higher in tissues (liver, lung, pancreas, brain, adipose) than in serum, being particularly high in liver and lung. Nine of the patients were in steady-state (treatment for more than 35 days), three had received tamoxifen for 7–13 days and two had received tamoxifen for 3–3.5 years (but had been tamoxifen-free for 28 days and 14 months, respectively, at the time of tissue sampling). Tissue samples from all other patients were obtained within 4–60 h. Tissues from the pancreas, pancreatic tumour, primary breast cancer and metastatic breast cancer in brain also retained large amounts of the drug. The amounts of N-demethylated and hydroxylated metabolites were high in most tissues except in fat, and tamoxifen and some of its metabolites were also present in specimens of skin and bone (Lien et al., 1991). Post-mortem and biopsy analysis of liver from tamoxifen-treated patients showed the presence of tamoxifen (0.14–15 nmol/g), 4-hydroxytamoxifen and N-desmethyltarnoxifen (Martin et al., 1995).

4.1.2. Experimental systems

The kinetics, absorption, distribution, excretion and metabolism of tamoxifen in experimental animals have been reviewed (Furr & Jordan, 1984; Buckley & Goa, 1989; Wiseman, 1994).

In rats, mice, dogs and rhesus monkeys, tamoxifen is well absorbed following oral administration. Most of the dosed material appears in the faeces, but bile duct cannulation experiments with rats and dogs demonstrated that this was a result of biliary excretion (Fromson et al., 1973b).

In order to achieve similar plasma levels, much higher oral doses of tamoxifen are required by rats and mice than by human breast cancer patients: doses in rats over seven days of 3.0 mg/kg bw gave < 1 ng/mL and 200 mg/kg bw gave 1000 ng/mL; doses in mice over 7–10 days of 2.5 mg/kg bw gave < 10 ng/mL and 200 mg/kg bw gave 300 ng/mL; doses in human patients over 10 days of 4.9 mg/kg bw gave 1300 ng/mL (Robinson et al., 1991).

In rats given tamoxifen orally at 1 mg/kg bw per day for 3 or 14 days, essentially the same amounts of drug were found after three days and 14 days treatment (except for fat), suggesting that steady-state is obtained within three days. Concentrations of tamoxifen and its metabolites (N-desmethyl, N-didesmethyl, 4-hydroxy and 4-hydroxy-N-desmethyl) were 8–70-fold higher in tissues (brain, adipose, liver, heart, lung, kidney, uterus, testis) than in serum. The highest levels were found in lung and liver, but substantial amounts were also found in kidney and adipose tissue. Within one dosing interval (24 h), marked fluctuations in the tissue concentrations were observed in rats receiving the steady-state treatment, with maximum/minimum concentration (C_max/C_min) ratios for tamoxifen found in female rat lung and liver being 6.3 and 4.1, respectively (Lien et al., 1991).

Groups of female Fischer 344 rats were treated with a non-necrotic, subcarcinogenic dose of N-nitrosodiethylamine (10 mg/kg orally) and were given tamoxifen at 250 mg/kg of AIN-76A diet for 6 or 15 months (Dragan et al., 1994). Treatment with tamoxifen resulted in a decrease in body weight of 16–24% at serum levels comparable to the therapeutic level in humans. In serum, the ratio of tamoxifen/4-hydroxytamoxifen/N-desmethyltamoxifen was 1/0.1/0.5–1. Rat livers had 20–30 times more tamoxifen and 4-hydroxytamoxifen and at least 100 times more N-desmethyltamoxifen than the serum at both 6 and 15 months. The ratio in the liver after 6 or 15 months of continuous administration was 1/0.1/1.3–2.3.

Administration of [¹⁴C]tamoxifen to dogs, rats, mice and rhesus monkeys has shown that tamoxifen has a long half-life in all of these species: in rats, the distribution half-life in blood was 53 h but the elimination half-life was 10 days. The route of excretion in rats, mice, rhesus monkeys and dogs is predominantly faecal, with virtually none of the recovered material being unchanged tamoxifen (Fromson et al., 1973b).

In rats, tamoxifen is eliminated in urine to a significant extent as the acidic metabolites. In one study, in the period 0–24 h after dosing with [¹⁴C]tamoxifen, the radio-active components recovered (expressed as percentages of the administered dose) were: total radioactivity, 8.7%; tamoxifen acid, 1.02%; and 4-hydroxytamoxifen acid, 1.81%. In faeces, the corresponding values were 30.5%, 0.5% and 2.40%. In contrast to other tamoxifen metabolites, neither of these metabolites is excreted as glucuronic acid or glycine conjugates (Ruenitz & Nanavati, 1990).

Tamoxifen can be metabolized in vitro by both microsomal cytochrome P450 and flavin monooxygenase pathways to intermediates that bind irreversibly to microsomal proteins (Mani & Kupfer, 1991). Incubation of tamoxifen with rat liver microsomes yielded three major polar metabolites identified as the N-oxide, N-desmethyl and 4-hydroxy derivatives. Formation of the N-oxide was catalysed by flavin monooxygenase, while that of the N-desmethyl and 4-hydroxy metabolites was mediated by cytochrome P450. Tamoxifen N-demethylation appears to be catalysed in rats by CYP1A, CYP2C and CYP3A enzymes, while in man the evidence points to the CYP3A enzyme. However, these enzymes are not major contributors to the 4-hydroxylation of tamoxifen (Mani et al., 1993, 1994). Peroxidases may also metabolize tamoxifen to a reactive intermediate hat binds covalently with protein (Davies et al., 1995) and DNA (Pathak & Bodell, 1994; Pathak et al., 1995).

Lim et al. (1994) compared the metabolism of tamoxifen in microsomes from female human, rat and mouse liver. The major metabolites formed by rat liver microsomes were 4-hydroxytamoxifen, 4′-hydroxytamoxifen, N-desmethyltamoxifen and tamoxifen N-oxide. In addition, it was suggested that two previously unreported epoxide metabolites, 3,4-epoxytamoxifen and 3′,4′-epoxytamoxifen, and their hydrolysed derivatives, 3,4-dihydroxytamoxifen and 3′,4′-dihydroxytamoxifen, had been identified, but these conclusions were based only upon mass spectral data; no synthetic standards were available. Jarman et al. (1995) were unable to confirm the existence of these dihydroxy compounds in microsomal incubates containing tamoxifen or deuterated analogues of tamoxifen. Using tamoxifen and [ethyl-D₅]tamoxifen, they showed a large isotope effect in the formation of α-hydroxytamoxifen (see Section 4.4). They confirmed the presence of α-hydroxytamoxifen-N-oxide and identified a new metabolite, α-hydroxy-N-desmethyltamoxifen.

Metabolites of tamoxifen were examined in human liver homogenate and a human hepatic G2 cell line treated with a mixture of tamoxifen and its deuterated analogues (Poon et al., 1995). In both the hepatic G2 cell line and the liver homogenate, α-hydroxytamoxifen, 4-hydroxytamoxifen, N-desmethyltamoxifen and tamoxifen N-oxide were detected. In the liver homogenate, N-didesmethyltamoxifen was also detected.

When primary cultures of human, rat and mouse hepatocytes were incubated with tamoxifen (10 µM) for 18–24 h, the concentration of α-hydroxytamoxifen in the medium was 50-fold lower in the human cultures (0.41 ± 0.55 ng/mL, two determinations) than in the rat (26.8 ±10.1 ng/mL, three determinations) and mouse (18.9 ± 13.5 ng/mL, four determinations) cultures (Phillips et al., 1996a).

4.2.1. Humans

The most reliable data regarding the association between tamoxifen and gynaecological symptoms come from randomized trials of tamoxifen versus placebo, in which some of these clinical data were collected prospectively.

The potential long-term toxicity of tamoxifen therapy has been reviewed. In large adjuvant trials, about 4% of recipients stop therapy because of side-effects (Love, 1989; Jaiyesimi et al., 1995). The most commonly reported side-effects of tamoxifen therapy are vasomotor symptoms, such as hot flushes and tachycardia, nausea and vomiting. A reduction over time in the vasomotor symptoms reported by patients receiving tamoxifen was observed in a randomized, double-blind, placebo-controlled clinical trial (Love & Feyzi, 1993). Atrophy is a common uterine response. Gynaecological adverse effects such as changes in vaginal discharge, bleeding, vaginal/external genitalia irritation, endometrial hyperplasia, polyps of the endometrium and, in premenopausal women, menstrual irregularities are summarized in Table 10.

Table 10

Tamoxifen-associated side-effects in the reproductive tract.

Some evidence exists that tamoxifen may be associated with thromboembolic events in patients with advanced breast cancer. Table 11 summarizes studies regarding the effects of tamoxifen on blood coagulation. In general, the effects of tamoxifen on clotting factors are not clinically significant and probably do not persist during chronic (more than six months) administration or after cessation of therapy.

Table 11

Tamoxifen-associated side-effects on blood coagulation.

In a randomized trial of tamoxifen (20 mg/day for five years) versus placebo in pre- and postmenopausal women with breast cancer, thromboembolic disease was reported in 0.2% of women who received placebo versus 0.9% of those who received tamoxifen (Fisher et al., 1989a).

In a randomized controlled study, morbidity due to cardiac and thromboembolic disease was assessed in 2365 postmenopausal breast cancer patients with (40 mg/day, two or five years) or without tamoxifen therapy. The median follow-up period was six years. Tamoxifen therapy was associated with a statistically significantly reduced incidence of hospital admissions due to any cardiac disease, with a relative risk of 0.7 for tamoxifen for two and five years versus control (95% CI, 0.5–1.0; p = 0.03). In the randomized comparison of five versus two years of tamoxifen treatment, there was a statistically significant reduction in risk with longer treatment (relative risk, 0.4; 95% CI, 0.2–0.9; p = 0.03). Although the trend of reduced risk in the tamoxifen group was also evident in the analysis of specific subgroups of cardiac diseases such as myocardial infarct and ischaemic heart disease, the results failed to reach significance. There was no association between tamoxifen treatment and relative risk for admission to hospital due to thromboembolic disease (1.1; 95% CI, 0.7–1.6) (Rutqvist et al., 1993). In another study, a total of 1312 women who had undergone mastectomy for breast cancer were randomized to receive either adjuvant treatment with tamoxifen (20 mg/day) or a placebo, with tamoxifen given only on first recurrence of disease. The maximal duration of tamoxifen treatment was 14 years. Use of tamoxifen was associated with lower rates of myocardial infarction, the relative risk in the control group being 1.9 (95% CI, 1.0–3.7) compared with women allocated to tamoxifen treatment (McDonald et al., 1995).

Studies of beneficial effects of tamoxifen upon blood cholesterol levels, and increased levels of sex hormone-binding globulin and thyroid-binding globulin (Dewar et al., 1992; Love et al., 1994a) are summarized in Table 12. Tamoxifen and 4-hydroxytamoxifen have also been shown to protect human low-density lipoproteins in vitro against copperion dependent lipid peroxidation, a model system that is relevant to events occurring within atherosclerotic lesions (Wiseman et al., 1993; Wiseman, 1994). Tamoxifen also lowers the levels of atherogenic amino acid homocysteine in humans (see Table 12) (Anker et al., 1995).

Table 12

Tamoxifen-associated side-effects on blood lipids, steroid hormone-binding globulin and thyroid function-associated parameters.

Some studies of the effects of tamoxifen on bone mineral density suggest that tamoxifen acts as an oestrogen agonist to preserve bone density in postmenopausal breast cancer patients; however, the effects are weak and not consistent among the different studies (Table 13).

Table 13

Effects of tamoxifen on bone mineralization.

Over 20 years of use have made tamoxifen one of the most studied anti-cancer drugs (Jordan, 1993), and it is associated with less toxicity than other current endocrine treatments for breast cancer (Muss, 1992). While the potential additional benefits of treatment in breast cancer patients in terms of blood lipid and cholesterol levels and bone mineral density remain to be fully established, it is worthy of note that two well-designed randomized trials of tamoxifen versus no adjuvant endocrine therapy (described above) have shown significantly reduced numbers of hospital admissions for cardiac disease and no difference in deaths due to cardiac or thromboembolic disease (Rutqvist et al., 1993) and significantly reduced risks of myocardial infarction and cardiac deaths (McDonald et al., 1995) in women receiving tamoxifen.

Tamoxifen has been associated in case reports with changes in liver enzyme levels in the serum (Hayes et al., 1995) and on rare occasions a spectrum of more severe events including fatty liver (Noguchi et al., 1987), cholestasis and hepatitis (Cortez Pinto et al., 1995) and a fatal case of hepatocellular damage and agranulocytosis (Ching et al., 1992).

Several in-vitro studies have demonstrated the antioxidant action of tamoxifen on microsomal and liposomal lipid peroxidation. The effects of tamoxifen on serum malondialdehyde and several antioxidant components were evaluated in 64 postmenopausal breast cancer patients after three and six months of treatment with 20 mg/day tamoxifen (Thangaraju et al., 1994). Serum malondialdehyde levels decreased significantly from 7.64 ± 1.2 nmol/dL before treatment initiation to 6.04 ± 0.95 nmol/dL (p < 0.001) after three months to 5.83 ± 0.91 nmol/dL (p < 0.001) after six months of tamoxifen administration. The levels of blood glutathione slightly increased from 2.61 ± 0.50 µmol/mL red cells to 2.86 ± 0.43 µmol/mL (p < 0.01) to 2.91 ± 0.47 µmol/mL (p < 0.01) and slight increases were also observed in serum levels of ceruloplasmin, uric acid, vitamins A, C and E and selenium, [Except for concentrations of malondialdehyde, the changes in all parameters were ≤ 10% of the initial value.]

Ocular toxicity has been reported following tamoxifen treatment, the first published reports of retinopathy and corneal changes being associated with particularly high doses of tamoxifen of at least 240 mg/day (Kaiser-Kupfer & Lippman, 1978). It is characterized by white refractile intraretinal deposits distributed mainly at the posterior pole. Ocular toxicity also appears to be an infrequent but serious complication of tamoxifen therapy at doses of 20–40 mg/day (Griffiths, 1987; De Jong-Busnac,1989). In the study by Pavlidis et al. (1992), four of 63 patients administered 20 mg tamoxifen per day displayed retinopathy and/or keratopathy 10, 27, 31 and 35 months, respectively, after the start of therapy. However, in another controlled study, no ocular toxicity was found in 79 breast cancer patients taking tamoxifen in conventional doses (10–20 mg two or three times a day) for an average of two years and three months (Longstaff et al., 1989). Ocular toxicity has not been reported in any of the major adjuvant breast cancer studies involving tamoxifen (Fisher et al., 1989a; Ribeiro & Swindell, 1992; Fornander et al., 1991). Where retinopathy was reported in patients, it was reversible on cessation of tamoxifen treatment if the condition was detected at an early stage (Ashford et al., 1988; Chang et al., 1992).

Leukopenia and neutropenia (Click et al., 1981; Boccardo et al., 1994b; Miké et al., 1994) have been reported on rare occasions following the administration of tamoxifen.

4.2.2. Experimental systems

In single-dose toxicity studies, the oral LD₅₀ of tamoxifen was approximately 3 g/kg for mice and 2.5 g/kg for rats (Furr & Jordan, 1984). Tamoxifen is well tolerated upon chronic administration to mice, rats and dogs at large multiples of the pharmacologically active dose (approximately 0.1 mg/kg), the pharmacological properties of tamoxifen (behaving as an oestrogen in some species and tissues and as an antioestrogen in others) accounting for many of the effects described in toxicology studies (Tucker et al., 1984). Tamoxifen is an antioestrogen with complex pharmacology encompassing variable species-, tissue-, cell-, gene- and duration of administration-specific effects from oestrogen-like agonist actions to complete blockage of oestrogenic action (Jordan & Robinson, 1987). In short-term laboratory assays, tamoxifen is usually classified as oestrogenic in mice but as a partial agonist/antagonist in rats. The concept that tamoxifen is oestrogenic in mice may not provide a complete description of the effects in this species, as prolonged administration to ovariectomized mice results in the uterus becoming refractory to oestrogen administration (Jordan et al., 1990). Whilst the initial response of the uterus to tamoxifen is oestrogen-like, as administration continues, uterine weight returns to initial values. In immature rats, administration of tamoxifen increases uterine weight in a dose-dependent manner without attaining the same maximal effect as obtained with 17β-oestradiol; administration of tamoxifen together with oestradiol provides a partial but incomplete antagonism of the uterotrophic action of 17β-oestradiol (Harper & Walpole, 1967a,b). The increase in uterine weight is largely due to hypertrophy of the luminal epithelium, with little change in the myometrium and stroma. This hypertrophic effect was not associated with any change in thymidine incorporation or cell division typical of uterine response to 17β-oestradiol treatment (Jordan et al., 1980). More recent studies have confirmed the hypertrophic effect of tamoxifen and other antioestrogens on the luminal and glandular epithelium of the rat uterus and contrasted this effect with that of oestrogens (Branham et al., 1993).

In mice, repeated administration of doses up to 50 mg/kg for 13–15 months caused atrophy of gonads and accessory sex organs, with cystic endometrial hyperplasia in the uterus, elongation of the vertebrae and a marked increase in bone density with resorption and new bone formation in irregular patterns. These changes are consistent with the pharmacological action in this species. In the liver, there were fatty changes and a swelling of the parenchymal cells (Tucker et al., 1984).

In rats, administration of tamoxifen for 6–24 months at doses between 35 and 100 mg/kg per day caused atrophy of the gonads and accessory sex organs, whilst, at lower doses (2 mg/kg per day), the uterine endometrium showed an absence of glands, flattening of the epithelium and occasional squamous metaplasia (Tucker et al., 1984; Greaves et al., 1993). After six months' treatment with 35 mg/kg per day, there was nodular hyperplasia in the liver (Greaves et al., 1993). The lysosomal lipidosis seen in a number of tissues, including retina and cornea, after chronic administration of 100–130 mg/kg per day is consistent with the cationic amphiphilic structure of tamoxifen (Lüllmann & Lüllmann-Rauch, 1981); the higher incidence of cataracts may be related to changes in sex hormone status (Greaves et al., 1993).

In dogs, repeated administration of up to 75 mg/kg tamoxifen per day for three months resulted in cessation of ovulation and hyperplasia of the germinal epithelium of the ovary and severe endometritis with squamous metaplasia, as well as biliary stasis with no other morphological change in the liver at the highest-dose level only (Tucker et al., 1984).

Six months' administration of tamoxifen at doses up to 8 mg/kg per day to marmosets produced a slight increase in ovarian follicular cyst weight and number, probably as a result of its antioestrogenic effect in this species (Furr et al., 1979; Tucker et al., 1984).

Rat, chicken and human microsomes exhibited low tamoxifen-binding activity compared with hamster and mouse microsomes (Mani et al., 1994). In another study (White et al., 1995), covalent binding of tamoxifen to microsomal proteins was observed with human, rat and mouse microsomes; the activity of mouse microsomes was highest (17–fold higher than human microsomes), while rat microsomes had intermediate activity (3.8-fold higher than human microsomes).

White et al. (1993) showed that, while the total hepatic microsomal cytochrome P450 content of rats given tamoxifen intraperitoneally (0.12 mmol/kg or 45 mg/kg per day for four days) was not increased (and, indeed, was transiently decreased), there were 30–60-fold increases in the metabolism of benzyloxy- and pentoxyresorufin; the metabolism of ethoxyresorufin was only slightly increased. Immunoblotting experiments revealed two-to three-fold increases in CYP2B1, CYP2B2 and CYP3A1 proteins. Induction of these proteins was centrilobular. None of these monooxygenase activities was induced in C57B1/6 mice and only small increases in benzyloxy- and pentoxyresorufin metabolism were seen in DBA/2 mice. There has been independent confirmation of the induction of CYP2B1, CYP2B2 and CYP3A in the liver of Fischer 344 rats (particularly females) administered tamoxifen orally for seven days. In addition, microsomal epoxide hydrolase was induced. These were selective inductions, there being no increased expression of CYP1A1, CYP1A2 or γ-glutamyl transpeptidase (Nuwaysir et al., 1995).

4.3.1. Humans

Tamoxifen is contraindicated during pregnancy (Vidal, 1995; Medical Economics, 1996).

Cullins et al. (1994) reported the birth of an infant with Goldenhar's syndrome (oculoauriculovertebral dysplasia) delivered by Caesarian section at 26 weeks to a 35-year-old woman with breast cancer who had received 20 mg/day tamoxifen throughout pregnancy. They noted that 50 pregnancies reported to the manufacturer had been associated with tamoxifen administration. There were 19 normal births, 8 terminations, 13 unknown outcomes and 10 associated with a fetal or neonatal disorder. Two infants had congenital craniofacial defects. [The Working Group noted that details of these 50 cases were not available in this secondary reference.]

Zemlickis et al. (1992) reported on three women with breast cancer who received treatment during the first trimester of pregnancy. One woman received tamoxifen with cyclophosphamide, methotrexate, fluorouracil and vincristine sulfate; the pregnancy ended in live birth, and the baby was alive and well at the time of follow-up. The other two women did not receive tamoxifen and the pregnancy ended in miscarriage. Two further women with breast cancer received chemotherapy during the third trimester. One received tamoxifen with fluorouracil, doxorubicin and cyclophosphamide, and a live-born infant was delivered who was alive and well at the time of follow-up. The other woman did not receive tamoxifen but also delivered a live birth with some intrauterine growth retardation; this child was well at the time of the follow-up.

Lai et al. (1994) reported the birth of a normal male infant with a birth weight of 3340 g at term following six months' daily therapy with 30 mg tamoxifen and 160 mg megestrol acetate for adenocarcinoma of the endometrium and three months' treatment with a combined oestrogen/gestagen oral contraceptive pill.

Two studies have been made of the effect of tamoxifen on reproductive parameters in healthy female volunteers. In the first, 16 women with regular menstrual cycles were followed during one control cycle, one treatment cycle during which women received 20 mg tamoxifen twice daily from cycle day 18 until day 30 or onset of menstruation, whichever came first, and one follow-up cycle (Swahn et al., 1989). The length of the treatment cycle (28.5 days ± 2.0 days) was significantly longer than that of the control (27.2 ± 2.0 days) or follow-up (27.5 ±1.9 days) cycles, owing to a prolonged luteal phase. Levels of follicle-stimulating hormone (FSH), progesterone, 17-hydroxyprogesterone, 20-dihydroprogesterone, oestrone, oestrone sulfate and 17β-oestradiol were significantly elevated during the treatment cycle compared with the control cycle. Levels of pregnanediol glucuronide remained unchanged during the treatment cycle, whereas the concentrations during the follow-up cycle were approximately double those of the control cycle. Prolactin levels decreased slightly during tamoxifen treatment, but the effect was not statistically significant. There was a positive correlation between plasma levels of tamoxifen and 17-hydroxyprogesterone (r = 0.72; p < 0.05), but not with plasma levels of the other hormones during treatment or follow-up cycles. The treatment did not cause any major disturbance of the bleeding pattern.

In the second study, Mäentausta et al. (1993) studied the effects of tamoxifen on endometrial 17β-hydroxysteroid dehydrogenase and progesterone and oestrogen receptors during the luteal phase of the menstrual cycle in 11 healthy female volunteers. The study included one control and two treatment cycles. During the first treatment cycle, the subjects received 200 mg mifepristone two days after the peak serum luteinizing hormone (LH) concentration. During the second treatment cycle, the subjects received 40 mg tamoxifen on the second and third days after the peak serum LH concentration. The time interval between these treatments was about a month, and the authors state that this ensured that the effects of each treatment modality had disappeared before the next treatment. 17β-Hydroxysteroid dehydrogenase and progesterone and oestrogen receptors were examined immunohistochemically in endometrial tissue specimens taken on the sixth to eighth day after the peak serum LH concentration. Tamoxifen did not have any significant effect on staining of 17β-hydroxysteroid dehydrogenase or the abundance of receptors. The authors state that serum concentrations of 17β-oestradiol, progesterone and LH were not significantly affected by the administration of tamoxifen.

Six women were treated for uterine fibroids for at least three months with 10 mg tamoxifen twice daily starting between days 1 and 3 of the menstrual period. Increased variability in the length of the menstrual and ovarian cycle was associated with significant lengthening of the luteal phase from 12.5 ± 1.5 days to 16.9 ±3.5 days (p < 0.02; Lumsden et al., 1989). A significant increase in the excretion of oestrone and pregnanediol glucuronide in the urine was associated with increased concentrations of 17β-oestradiol and progesterone in plasma, reflecting multiple follicular development and ovulation. A significant rise in the concentration of FSH occurred during the luteal phase of the cycle.

A number of studies of endocrine parameters have been carried out in pre- and postmenopausal women receiving tamoxifen therapy (for review, see Sunderland & Osborne, 1991). Most studies report that in postmenopausal women levels of gonadotropins, FSH and LH decrease with tamoxifen therapy, although remaining within the normal postmenopausal range. 17β-Oestradiol and progesterone levels do not change in postmenopausal women receiving tamoxifen. In contrast, in premenopausal women receiving tamoxifen, 17β-oestradiol and progesterone levels show a striking elevation, often to two or three times the normal level. The elevated hormone levels follow a pattern consistent with the normal menstrual cycle. Despite these supraphysiological levels of 17β-oestradiol, FSH and LH levels remain unchanged or only slightly increased. Many premenopausal women receiving long-term tamoxifen therapy continue to have regular ovulation and menstrual cycles, although as many as one third may develop temporary amenorrhoea or oligomenorrhoea.

In a study of the pregnancy outcome in the partners of men who had been treated with tamoxifen (n = 22) or clomiphene (n = 12) for oligozoospermia, no important difference in the course of pregnancy was found between the groups (Salata et al., 1993).

Gooren (1989) reported a patient with incomplete androgen insensitivity syndrome (hypospadias, unilateral cryptorchidism and pubertal gynaecomastia, all surgically corrected) and plasma FSH levels below the reference range for adult men. After treatment with 10 mg tamoxifen twice daily, his plasma FSH level rose, spermatogenesis improved and his wife conceived three times within a period of five years.

In a number of uncontrolled studies, treatment with tamoxifen has been reported to be associated with an increase in sperm count of men with idiopathic oligozoospermia. However, in addition to the fact that these studies were not randomized controlled trials, a limitation of the studies was that patients who would be considered fertile were included (Kotoulas et al., 1994). Krause et al. (1992) reported a randomized trial of tamoxifen in the treatment of idiopathic oligozoospermia. The sperm output and pregnancy rate was somewhat higher in the group of 39 patients who received 30 mg tamoxifen daily than in the 37 patients who received the placebo. [The numbers of subjects assigned to each group of the trial in the reporting of the pregnancy rates conflict with the numbers assigned to therapy recorded earlier in the paper.] Kotoulas et al. (1994) reported that 122 men randomly assigned to treatment with 10 mg tamoxifen twice daily for a period of three months had improved sperm density and number of live spermatozoa compared with 117 men who received the placebo therapy for the same period of time. The improvement in sperm density was more marked in the subgroups who were Oligozoospermic than in normozoospermic men. In addition, there was a statistically significant decrease in the number of abnormal sperm after tamoxifen treatment, but the authors comment that this decrease was observed in only 12 patients in the tamoxifen group compared with eight patients in the placebo group.

4.3.2. Experimental systems

(a) Effects on the fetus

Cunha et al. (1987) assessed the potential oestrogenicity and teratogenicity of tamoxifen in 54 genital tracts isolated from 4–19-week-old human female fetuses and grown from one to two months in untreated athymic nude mice. After grafting of the human fetal genital tracts, the hosts were given subcutaneous implants of 20-mg pellets of tamoxifen, clomiphene or diethylstilboestrol, or were sham operated. In all mice that received tamoxifen, the vaginal epithelia of the hosts were thickened and cornified, and the uteri exhibited cystic hyperplasia. In specimens of human genital tract grown to a gestational age equivalent of 15 weeks or less, the vagina and urogenital sinus were lined with an immature squamous epithelium, which was similar in drug-treated and untreated specimens. The authors noted that the absence of oestrogenic response in these specimens correlated with the apparent absence of oestrogen receptors. Two of the four tamoxifen-treated specimens grown to a gestational age equivalent of 16 weeks or more exhibited epithelial hyperplasia and maturation. Untreated specimens were unstimulated. Formation of endometrial and cervical glands proceeded in 13/15 (87%) control specimens grown to a gestational age equivalent to 13 weeks or more in untreated hosts. The results from tamoxifen-treated specimens were similar to those obtained with clomiphene-treated specimens, and therefore were pooled. Glands were present in only 6/13 (46%) age-matched specimens treated with tamoxifen or clomiphene. In the developing uterine corpus of untreated controls, the uterine mesenchyme segregated into inner (endometrial stroma) and outer (myometrial) layers, whereas, in specimens treated with tamoxifen, condensation and segregation of the mesenchyme were greatly impaired. The epithelium of the fallopian tubes of specimens treated with tamoxifen was hyperplastic and disorganized, and the complex mucosal plications characteristic of the fallopian tube were also distorted.

As tamoxifen is an antifertility agent in rats, it has proved difficult to conduct studies of possible teratogenic effects in this species (Furr & Jordan, 1984). Tucker et al. (1984) reported that 0.025 mg/kg bw tamoxifen was the highest dose that could be given throughout pregnancy in Alpk/AP rats without completely preventing implantation. At this dose, about 50% of matings gave rise to successful pregnancies. These authors also stated that in several teratogenic studies in rats, all doses above 2 mg/kg bw produced an incidence of irregular ossification of ribs in the fetus. They considered that this was secondary to a tamoxifen-induced reduction in the size of the uterus of the dam and noted that the effect disappeared in the early neonatal period. In rabbits, administration of 0.1 and 0.2 mg/kg bw tamoxifen throughout pregnancy did not produce any effect on implantation or on the fetus. In marmosets, doses of up to 10 mg/kg bw tamoxifen from days 25 to 35 of pregnancy did not produce any effect in the fetus. [The Working Group noted that few details of these studies were reported.]

Twenty adult female cynomolgus monkeys (Macaca fascicularis) with two spontaneous menstrual cycles of normal duration during an initial two- to three-month period of observation were randomly assigned to receive a low dose of tamoxifen (0.5 mg/kg bw per day; n = 6), a high dose of tamoxifen (3.0 mg/kg bw per day; n = 7) or lactose (n = 7) by gastric instillation daily for 12 days, starting four days after the mid-cycle 17β-oestradiol peak (Olive et al., 1990). The luteal phase of the menstrual cycle was prolonged in the groups receiving tamoxifen compared with the control group, but no difference between the groups receiving low-dose and high-dose tamoxifen was observed. No noteworthy difference in hormonal characteristics between the groups was observed. The authors concluded that administration of tamoxifen during the luteal phase did not alter pituitary gonadotropin secretion or corpus luteum function. They suggested that the prolongation by tamoxifen of the length of the luteal phase in a subset of monkeys was perhaps due to a direct effect on the endometrium. In a study of 26 female Macaca fascicularis with proven fertility and normal menstrual cycles, 13 were treated with a single oral dose of 5 mg/kg bw tamoxifen and 13 with vehicle only on post- ovulation day 4 (Tarantal et al., 1993). Serum progesterone and tamoxifen concentrations were evaluated on postovulation days 4, 8, 12, 16 and 18. There was no effect of tamoxifen on serum progesterone levels or on the fertility rate — 6/13 (46%) treated females and 4/13 controls (31%) became pregnant. In the centre in which the study was carried out, the conception rate for the Macaca fascicularis colony was approximately 50% per mated cycle. Among the six pregnant animals treated with tamoxifen, one aborted spontaneously on gestational day 40 (after detection of severe growth retardation and embryonic death on gestational day 38) and five delivered live births naturally during gestational days 160–163. Among the control females who became pregnant, one had an early embryonic loss (at or before gestational day 18), one had a still birth at gestional day 162 and two delivered live births. None of the live births exhibited any abnormality. Neither tamoxifen nor any of its metabolites was detected in maternal serum and urine. The authors postulated that either absorption of tamoxifen was negligible or its metabolism and excretion occurred at an extremely rapid rate.

Beyer et al. (1989) assessed the embryotoxic potential of tamoxifen by studying its effect on cultured whole rat embryos in the presence and absence of a NADPH-supple-mented post-mitochondrial supernatant fraction from Aroclor 1254-induced male rat liver (S9). Embryos were obtained from pregnant animals on gestational day 10. Only viable embryos, as determined by the presence of visible heart beat and active vitelline circulation, were evaluated. At the time of explantation, conceptuses were up to 10 ± 2 somite stage and were exposed to tamoxifen added directly to the culture medium at the beginning of the 24-h culture period. The results are presented in Table 14. In order to explore further the possible role of oestrogenicity, the interactive effects of tamoxifen with diethylstilboestrol and 17β-oestradiol were investigated. At 0.19 mM diethylstilboestrol or 0.1 mM 17β-oestradiol, in the presence of S9, a series of four tamoxifen concentrations ranging from 0.05 to 0.19 mM were added to the culture medium at the onset of the culture period. In all cases, the effect of tamoxifen appeared to be additive rather than antagonistic. Thus, the effect of tamoxifen was to exacerbate the embryotoxic/dysmorphogenic effects of both diethylstilboestrol and 17β-oestradiol.

Table 14

Embryotoxicity of tamoxifen to rat embryos.

In a preliminary experiment, groups of two or three rabbits were given 0.25, 0.5, 1.0, 2.0 or 4.0 mg/kg bw tamoxifen daily by gastric instillation from day 6 to day 18 of gestation (Esaki & Sakai, 1980). Abortions were observed in both rabbits treated with 4 mg/kg bw and one of the three rabbits treated with 2 mg/kg died. In a following experiment, groups of 10–14 rabbits were given 0.125, 0.5 or 2.0 mg/kg bw tamoxifen daily by gastric instillation from day 6 to day 18 of gestation. In the groups receiving 0.5 and 2.0 mg/kg doses, body-weight gain in the dams was suppressed. Abortions occurred in one animal in each of the lower-dose groups and in five animals in the group receiving 2.0 mg/kg bw. Fetal death rates, defined as the total number of resorption sites for dead embryos divided by the total number of implantations, were 11.6% in the control group, 16.7% in the low-dose group, 39.4% in the mid-dose group and 32.7% in the high-dose group. No effect of tamoxifen on the body weight of live fetuses was observed. On external observation of the live-born fetuses, one exencephaly was observed in the control group and one brain hernia [sic] in the group receiving 0.125 mg/kg and one abdominal hernia with brain hernia [sic] in the group receiving 2.0 mg/kg. On internal examination, four anomalies were observed in the control group, one in the group receiving the lowest dose of tamoxifen and two in each of the other groups. There were 11 instances of minor malformations of the skull, sternum, caudal vertebrae or ribs in the control group, 12 in the group receiving 0.125 mg/kg tamoxifen and 9 in each of the other two groups.

Groups of four pregnant rabbits were given 2 mg/kg bw tamoxifen per day orally, beginning on gestational day 10 or 20, or vehicle only (Furr et al., 1976). Administration of the drug from gestational day 10 resulted in considerable embryonic loss; only two rabbits gave birth and the average number of young born was 1.75 ±1.2 compared with 5.8 ± 1.6 for vehicle-treated controls. Since the number of implantation sites was similar, a major effect of the drug was to induce fetal resorption. Administration of the drug from day 20 caused premature parturition and abortion: the length of gestation in the group receiving tamoxifen was 26.8 ± 1.3 days compared with 32.5 ± 0.3 days in controls, and the percentage of young born alive was 65% as compared to 96%. Both of these effects were associated with a significant reduction in plasma progesterone concentration. The differences in plasma progesterone concentrations were not due to differences in the numbers of corpora lutea in the different groups.

Pregnant guinea-pigs (n = 6) of the Hartley albino strain were given 2 mg/kg bw tamoxifen by subcutaneous injection for three or six days (Gulino et al., 1984). Uteri of the fetuses (n = 10) were weighed 24 h after the last administration. Compared with control fetuses, a dose-related increase in uterine wet weight was observed, such that the uterine weight of fetuses of the dam that had received injections for a six-day period was 2.5 times that of controls. This uterotropic effect was associated with a significant increase in uterine DNA content. Tamoxifen also induced an increase in the size of the uterine stroma and myometrium in the fetus. Luminal epithelial cell height was increased by 67 ± 2% after six days of treatment of the dam and luminal epithelial cell number was increased by 160 ± 20%. Uterine epithelial downgrowths invading the stroma were observed in 26 ± 4% of tamoxifen-treated fetuses. In a subsequent study from the same laboratory, groups of 6–10 pregnant guinea-pigs of the Hartley albino strain were given 5 mg/kg bw tamoxifen per day by subcutaneous injection for 12 days or the vehicle alone (Pasqualini & Lecerf, 1986). The uterine epithelial cells of the fetal guinea-pigs were examined by transmission electron microscopy for ultrastructural changes. The fetuses were at 60–64 days of gestation at the time of necropsy. A moderate increase in the height of the fetal uterine epithelial cells and moderate effects on the Golgi system, the rough endoplasmic reticulum and mitochondria were observed. No change in the number of microvilli or cell degeneration was apparent. When the same dose of tamoxifen was administered in combination with 1 mg/kg bw 17β-oestradioI per day for 12 days, the effects on mitochondria were greater than when either compound was given on its own, while the increase in number of microvilli apparent when 17β-oestradiol was given on its own was reduced to the basal level. Therefore, although tamoxifen has antioestrogenic properties, in this species it acted as an agonist in the uterus during fetal development.

Pregnant pigs (sows) were fed diets containing 0 or 10 mg/kg diet (ppm) tamoxifen from gestational day 30 until weaning (Yang et al., 1995). No significant differences in sow body weight, litter size, live births per litter, piglet mortality, piglet sex ratio or piglet birth or weaning weight were observed between the groups. At the age of 21 days, female piglets exposed to tamoxifen in utero and during lactation had smaller ovaries and enlarged uteri compared with controls, but no histological abnormality. Corresponding male piglets had testes 15% lighter than those produced by sows fed the control diet; no consistent histological difference was observed between the groups. Subsequent breeding performance was not affected.

In two trials, approximately 500 eggs from single comb White Leghorn hens were injected on the day of set with either 100 µL tamoxifen (2 mg/µL) or vehicle (corn oil) into the albumen (Coco et al., 1992). Based on phenotypic sexing, the sex ratio of male to female chicks was 76: 24 for those exposed to tamoxifen compared with 47: 53 in those treated with vehicle only. When a subset of these chicks was sexed at three weeks of age by identification of gonadal type at necropsy, these ratios were corrected to 46: 54 and 52: 48, respectively. In the second trial, the sex ratio based on phenotypic sexing was 62: 38 for tamoxifen-exposed chicks and 45: 55 for vehicle-exposed chicks. Gonadal sexing corrected these ratios to 44: 56 and 45: 55, respectively. Thus, the genital sexing errors were 27% in the first trial and 18% in the second for tamoxifen-exposed chicks, significantly higher than those treated with vehicle (2 and 0.6%, respectively). Therefore, phenotypic genital sexual differentiation was altered by administration of tamoxifen.

(c) Effects of exposure of neonates and immature animals to tamoxifen on development of the reproductive system

Uterotropic effects of tamoxifen have been observed in mice, rats and guinea-pigs (Table 15). In rats, these effects are seen in neonatal rats only after short periods of treatment, prolonged treatment appearing to inhibit uterine development. In guinea-pigs, these effects are much stronger when the dose is administered to neonates rather than to immature animals (Gulino et al., 1984).

Table 15

Effect of administration of tamoxifen to neonatal or immature animals on subsequent development of the uterus.

Irisawa and Iguchi (1990) reported that neonatal treatment of mice with tamoxifen initially caused an increase in uterine weight and height of the uterine epithelium as well as an increase in the thickness of the vaginal epithelium. Tamoxifen treatment also caused poor formation of the uterine gland and development of the mesenchymal stroma. However, by 60 days of age, the weights of the uteri of the tamoxifen-treated mice were lower and the cell heights of the uterine luminal epithelium were smaller than those of controls. The number of uterine glands and the number of mice having a well developed tunica muscularis remained smaller than those in controls of the same age. Thus, neonatal treatment with tamoxifen resulted in a permanent alteration in both the uterine epithelial and stromal compartments. The critical period for induction of these uterine abnormalities was determined to be within three to seven days after birth.

The genital organs of female C57B1/Tw mice given five daily injections of 100 µg tamoxifen from the day of birth were examined at 5, 10, 15, 20, 30 and 60 days of age (Irisawa & Iguchi, 1990). Adenosis-like lesions were found in the vaginae of 5–30-day-old mice exposed to tamoxifen. These lesions were not detected at 60 days of age. The number of polyovular follicles containing two to four oocytes per follicle markedly increased from 10 to 15 days of age in mice exposed to tamoxifen, and the incidence was twice as high as that in age-matched controls. Corpora lutea were found in the ovaries of 60-day-old controls, whereas no corpora lutea were found in age-matched mice exposed to tamoxifen. In order to determine the critical period of induction by tamoxifen of abnormalities of the female genital organs, other groups of mice were also given five daily injections of 100 µg tamoxifen or of vehicle alone starting on the day of birth, or 3, 5, 7 and 10 days after birth. Tamoxifen injections starting within five days of birth caused a high incidence of polyovular follicles in the ovary and of aplasia of tunica muscularis in the uterus. Atrophy of the uterine luminal epithelium was also induced when the treatment was started within seven days of birth. However, in mice given tamoxifen from day 10, uterine weights were greater than in those given tamoxifen from zero to seven days, and the authors concluded that the postnatal limit of the critical period for the female genital organs lies within seven days of birth.

In female C57B1/Tw mice given five daily injections of 2, 20 or 100 µg tamoxifen starting on the day of birth, uterine hypoplasia, myometrial involution and suppression of uterine-gland genesis were found at 35 and 150 days of age (Iguchi et al., 1986). About half of the mice killed when 150 days old were ovariectomized at 90 days. Vaginal hypoplasia and hypospadia were observed in most treated animals at 150 days of age. Vaginal adenosis was found in 40% of the 35-day-old mice treated with 2 µg tamoxifen, 70% of those treated with 20 µg and 100% of those treated with 100 µg, whereas none of the controls showed these lesions. Adenosis was not observed in any of the 150-day-old mice. In both age groups, the weight of ovaries was significantly lower following injections of 20 or 100 µg tamoxifen than in controls. At 150 days of age, hernia of the urinary bladder, located under or below the symphysis pubis, was found in 56% and 100% of non-ovariectomized mice in the mid- and high-dose groups. The urinary bladders of the group receiving the lowest dose and the control group were situated normally in the abdominal cavity. The authors suggested that the tamoxifen treatment caused a long-lasting suppression of the development of the pubic bone and ligament, resulting in looseness of the symphysis pubis which allowed the bladder to descend through the extended subpubic space.

In NMRI mice treated neonatally with tamoxifen, adenosis-like lesions observed in the vagina and cervix were similar to those induced by neonatal treatment with diethyl-stilboestrol, but differed in that the tamoxifen-induced lesions regressed with time (Forsberg, 1985).

Fifteen male NMRI/Tg mouse pups were given 20 µg tamoxifen by subcutaneous injection daily for three days, starting approximately 24 h after birth. Ten control pups were given injections of saline. At three months of age, all mice were housed with normal females for two weeks to investigate reproductive capacities. All tamoxifen-exposed males were sterile and did not impregnate any female during the mating period, whereas all control males were able to impregnate one or both females with which they were housed. When killed at eight months of age, the exposed male pups had multiple reproductive tract lesions including testicular hypoplasia, undescended testes, epididymal cysts and squamous metaplasia of the seminal vesicle; two of the exposed mice had squamous metaplasia of the median prostate (Taguchi, 1987).

In addition to inhibiting uterine development, administration of 5 µg tamoxifen by subcutaneous injection to newborn female Sprague-Dawley rats on days 1, 3 and 5 of age produced early vaginal opening and absent cycles (Chamness et al., 1979). At four months of age, all ovaries were atrophic and the oviducts showed severe squamous metaplasia with abscess formation. Corporea lutea were uniformly absent except for a few that were found in one animal. The vaginal observations were stated to be unremarkable, except that one animal had a vaginal adenosis suggestive of that observed in women whose mothers received diethylstilboestrol during pregnancy.

Branham et al. (1993) gave 20–24-day-old rats five daily subcutaneous injections of tamoxifen (0.01–100 µg/rat/day) and then examined the uteri 2 h after the last dose. Uterine weights increased only slightly, whereas luminal epithelium hypertrophy increased 3-fold at 10 µg/rat and glandular epithelium hypertrophy increased 2-fold at the same dose. In contrast, oestrogens (17β-oestradiol, ethinyloestradiol, diethylstilboestrol) tested over the same dose range produced substantial increases in uterine weight and no glandular epithelium hypertrophy; the luminal epithelium response to 17β-oestradiol was similar to that of tamoxifen, while the other oestrogens required a 100-fold lower dose to elicit the same response. The greater hypertrophic response in luminal epithelium to both oestrogens and antioestrogens does not correlate with the oestrogen receptor content, which is greatest in the endometrial stroma and glandular epithelium in mouse, rat and macaque (Martin, 1980; Korach et al., 1988; Tse & Goldfarb, 1988; McClellan et al., 1984). It was suggested by Branham et al. (1993) that the high concentrations of tamoxifen (and other antioestrogens) required to elicit glandular epithelial hypertrophy is consistent with either a nonoestrogen-receptor-mediated response or a stromal mediation of epithelial responses.

Female rats of the T strain given single daily injections of 100 or 200 µg tamoxifen for five days beginning on the day of birth exhibited continued vaginal dioestrus when sacrificed on day 60, whereas vehicle-treated controls showed regular oestrus cycles (Ohta et al., 1989). Ovaries from the tamoxifen-treated rats were polyfollicular without corpora lutea, whereas ovaries from controls contained both follicles and corpora lutea. The vaginae of tamoxifen-treated rats ovariectomized on days 10 or 60 failed to respond to a three-day priming with 0.1 µg 17β-oestradiol, showing no oestrus smears. The authors concluded, therefore, that continued vaginal dioestrus in tamoxifen-treated rats may be accounted for by changed sensitivity of the ovary to gonadotropin and/or of the vagina to sex hormones. Lack of cyclicity and corpora lutea in the ovaries in rats was also reported by Irisawa and Iguchi (1990) and in NMRI mice by Forsberg (1985). Thus, tamoxifen appears to impair the female genital organs directly and/or indirectly through the hypothalamus.

In a study of the regulation of reproductive behaviour in rats (Ulibarri & Micevych, 1993), male rat pups were castrated, given sham surgeries or implanted with tamoxifen [dose not specified] within 3 h of birth. Female animals were implanted with a similar dose of tamoxifen or with empty capsules. All capsules were removed on postnatal day 10. Tamoxifen treatment was toxic to pups and only nine males and two females survived to surgery on day 90. Tamoxifen-treated females had large green ovarian growths and smaller growths throughout their peritonea. Tamoxifen-treated males had significantly smaller testes at adult surgery than sham-treated males. Tamoxifen-treated males and sham-operated males did not show appreciable lordosis behaviour. The authors noted that this finding differs from results with other oestrogen antagonists. The authors speculated that the antioestrogenic trans-isomer of tamoxifen may have been converted to the oestrogenic cis-isomer or that the trans-isomer of tamoxifen may have acted oestrogenically.

Pasqualini et al. (1986) reported that 100 µg tamoxifen administered subcutaneously to two-day-old guinea-pigs for two or 12 days substantially increased the weight and the protein and DNA content of the uterus and vagina; the weight increase was 3–4-fold in the group receiving prolonged treatment. Progesterone did not block this action. Tamoxifen caused a strong increase in the content of progesterone receptor in cytosol and nuclei of the cells of the vagina after prolonged treatment.

In immature, six-week-old pigs given 0.1 or 1.0 mg/kg bw tamoxifen per day for seven days, significant dose-related increases were seen in uterine weight, in the total content of uterine DNA, RNA and protein and in levels of progesterone receptor per milligram DNA (Lin & Buttle, 1991). The total duct area in the mammary glands increased about three-fold in the groups treated with either dose of tamoxifen, compared with vehicle-treated controls. The concentration of progesterone receptors in cytosol extracts of mammary tissue was very heterogeneous and independent of treatment with tamoxifen. Concurrent administration of tamoxifen with oestradiol benzoate induced significant increases in total uterine protein and in the concentration of progesterone receptors compared with treatment with oestradiol benzoate alone. However, concurrent administration partially inhibited the effect of oestradiol benzoate in stimulating an increase in mammary duct area. Thus, tamoxifen acts as an oestrogen agonist in the uterus of immature pigs, but as an antagonist in the mammary gland.

In rabbits, tamoxifen appears to act as an oestrogen antagonist when administered to sexually immature animals (Foster et al., 1993). Female New Zealand white rabbits were given 10 mg/kg bw tamoxifen per day by subcutaneous injection from day 22 of age for a total of 108 days, while control rabbits were given the vehicle only. Tamoxifen treatment impaired sexual development profoundly, as assessed by ovarian weight, diameter of growing follicles, diameter of anthral follicles and number of such follicles. There was profound suppression of pituitary gonadotropin levels in the group receiving tamoxifen, but no difference in the circulating levels of plasma 17β-oestradiol between control and tamoxifen-treated rabbits. These differences were associated with suppression of the developmental shift from smooth to rough gonadotropin-releasing hormone cell types in the hypothalamus.

4.4.1. Humans

No data were available to the Working Group.

4.4.2. Experimental systems (see also Table 16 for references and Appendices 1 and 2)

Table 16

Genetic and related effects of tamoxifen.

4.4.3. Metabolites of tamoxifen

4-Hydroxytamoxifen caused morphological transformation of Syrian hamster embryo (SHE) cells. Treatment of SHE cells with 10 µM 4-hydroxytamoxifen for 48 h resulted in the emergence of immortalized cells in 3 out of 5 flasks; these cells were able to form fibrosarcomas when injected into thymus-aplastic mice (Metzler & Schiffmann, 1991).

Treatment of rat primary hepatocyte cultures with α-hydroxytamoxifen resulted in 15- to 63-fold higher levels of adducts (and the same pattern of adducts detected by ³²P-postlabelling) than with comparable concentrations of tamoxifen (Phillips et al., 1994a; 1996a). A similar level of adducts (173.9 ± 4.1 adducts per 10⁸ nucleotides) was seen in mouse hepatocytes treated with α-hydroxytamoxifen at a concentration of 1 µM. Treatment of human cells with α-hydroxytamoxifen resulted in DNA adduct formation at levels (1.94 ± 0.89 and 18.9 ± 17.9 adducts per 10⁸ nucleotides at 1 αM and 10 αM, respectively) about 300-fold lower than those in rat hepatocytes. Concentrations of α-hydroxytamoxifen in the culture medium of cells incubated with tamoxifen were approximately 50-fold lower in experiments with human hepatocytes than in those with rat or mouse hepatocytes (Phillips et al., 1996a).

The hypothesis has been advanced that tamoxifen is activated to a DNA-binding species through oxidation at the α-position of the ethyl group (Potter et al., 1994). This proposal is supported by studies with the deuterated compound [ethyl-D₅]tamoxifen, which showed lower DNA-binding activity in rat liver in vivo than the non-deuterated compound (Phillips et al., 1994b). It also had a lower ability to induce micronucleus formation in MCL-5 cells. The magnitude of the reduction in genotoxicity (two- to three-fold) correlated well with the comparative rates of metabolism of the deuterated and non-deuterated compounds in rat liver microsomal incubations (Jarman et al., 1995). Thus, the reduced genotoxicity of [ethyl-D₅]tamoxifen is the result of its lower rate of oxidation due to the greater bond energy of the C–D bond compared with the C–H bond, implying that metabolic activation involves metabolism at the ethyl group of the molecule. This is supported by the observation of higher DNA-binding activity of α-hydroxytamoxifen in primary cultures of hepatocytes, in which DNA adducts were formed at between 25 and 49 times the level seen with equimolar concentrations of tamoxifen and which gave the same pattern of major adducts by ³²P-postlabelling (Phillips et al., 1994a).

Studies on the nature of the DNA adducts formed from tamoxifen in vivo and in vitro provide evidence for this and other pathways of activation of tamoxifen to form DNA-binding products. The demonstration that tamoxifen–DNA adduct formation in mice was altered by pretreatment with pentachlorophenol, a sulfotransferase inhibitor, suggested the existence of two pathways of activation (Randerath et al., 1994a,b). The levels of some of the major adducts were unaffected by pentachlorophenol, but levels of one major and several minor ones were increased 13–17-fold (Randerath et al., 1994a). However, another sulfotransferase inhibitor, 2,6-dichloro-4-nitrophenol, did not enhance the levels of these adducts (Randerath et al., 1994b). The metabolite 4-hydroxytamoxifen was found to give rise to adducts of the pentachlorophenol-inducible type (Randerath et al., 1994b; Moorthy et al., 1996).

α-Hydroxytamoxifen has low chemical reactivity towards DNA, but the synthetic compound α-acetoxytamoxifen is much more reactive. The DNA adducts formed with α-acetoxytamoxifen showed the same pattern on ³²P-postlabelling analysis as those from DNA treated with α-hydroxytamoxifen and those found in the DNA of rat hepatocytes treated with tamoxifen or of the livers of rats treated with tamoxifen in vivo. The major α-acetoxytamoxifen–DNA adduct was also isolated as a nucleoside and characterized by ultraviolet, mass and proton magnetic resonance spectroscopy and assigned the structure (E)-α-(N²-deoxyguanosinyl)tamoxifen, in which the α-ethyl position of tamoxifen is linked covalently to the exocyclic amino group of deoxyguanosine (Osborne et al., 1996).

When incubated with rat liver microsomal fractions or with peroxidase enzymes in the presence of DNA, the metabolite 4-hydroxytamoxifen forms DNA adducts. The principal adduct formed in each case co-migrated on thin-layer chromatography with a minor adduct formed in the liver of tamoxifen-treated rats (Pathak et al., 1995). In another study, two minor DNA adducts formed by rat microsomal activation of tamoxifen in the presence of DNA co-migrated in several thin-layer chromatography systems with the products of the reaction of (Z)-1,2-diphenyl-1-(4-hydroxyphenyl)but-1-ene (‘metabolite E’ see Figure 1) with DNA in the presence of silver(I) oxide (Pongracz et al., 1995).

4.5.1. Genotoxicity

The ability of tamoxifen to form DNA adducts in rodent liver cells in vivo and in vitro suggests a genotoxic mechanism for carcinogenicity in rat liver. This hypothesis is supported by the dose–response relationship for both tumour formation (in both male and female rats) and adduct formation. Although adducts are also formed in mouse and hamster liver, the levels are lower and these species have not been tested adequately for carcinogenesis by tamoxifen. Tamoxifen also possesses tumour-promoting activity in rat liver. An unusual aspect of tamoxifen–DNA adduct formation is that, in the rat, little or no adduct formation occurs in other tissues following oral administration. Although some information is available on the nature of the reactive intermediates of tamoxifen, the enzymes involved in their formation have not been clearly identified. The possibility that tamoxifen is inactive in most assays for mutagenic activity (gene mutations) because of failures of in-vitro metabolizing systems and/or transport of the reactive intermediate to the target site has not been excluded.

Studies on human hepatocytes indicate that they have a much lower ability to activate tamoxifen to DNA-binding products than those of rodents. In addition, limited information indicates that tamoxifen–DNA adducts are not formed in either the liver or endometrium of women taking tamoxifen as adjuvant therapy for breast cancer. These findings suggest that humans are less susceptible to the genotoxicity of tamoxifen than rodents.

4.5.2. Tamoxifen–oestrogen receptor interactions

Tamoxifen acts as an oestrogen agonist and/or antagonist by binding directly to the oestrogen receptor. In some tissues, including breast, tamoxifen exerts antiostrogenic effects by binding to the oestrogen receptor with high affinity. The tamoxifen–oestrogen receptor complex is incapable of binding to DNA-responsive elements, and therefore fails to induce normal transcriptional activity (Pasqualini et al., 1987). In other tissues, such as bone (Love et al., 1992b), uterus (Jordan & Prestwich, 1977) and liver, tamoxifen acts as a partial agonist, possibly because cells from those tissues contain a different array of DNA-binding sites, thereby leading to typical oestrogen-mediated changes in gene expression and subsequent biological effects on growth and differentiation.

Mechanistic considerations

Tamoxifen increases liver tumour incidence in rats, which may involve both DNA damage leading to increased numbers of initiated cells and oestrogen receptor-mediated clonal expansion of those initiated cells.

The available evidence suggests that tamoxifen is carcinogenic in rat liver by a genotoxic mechanism. Preliminary information from studies of human tissues suggests that humans are less susceptible to the genotoxicity of tamoxifen. Tamoxifen also possesses tumour-promoting activity in the rat liver.

Several studies have shown that the liver contains significant quantities of oestrogen receptor in hepatocytes, Kupffer cells and endothelial cells.

Tamoxifen acts as an oestrogen agonist and/or antagonist by binding directly to the oestrogen receptor. In some tissues, such as breast, tamoxifen exhibits antioestrogenic properties by binding to the oestrogen receptor with high affinity. The tamoxifenoestrogen receptor complex is incapable of binding to DNA-responsive elements. Thus, oestrogen receptor binding does not result in normal transcriptional activity. In other tissues, such as bone and liver, tamoxifen acts as a partial agonist, possibly because cells from those tissues contain a different array of DNA binding sites, thereby leading to typical oestrogen-mediated changes in gene expression and subsequent biological effects on growth and differentiation. Therefore, tissue-specific effects of tamoxifen–oestrogen receptor on gene expression may be involved in the ability of tamoxifen to increase or decrease tumour risk.

Overall evaluation

Tamoxifen is carcinogenic to humans (Group 1) and there is conclusive evidence that tamoxifen reduces the risk of contralateral breast cancer.